Lecture Notes in Civil Engineering

Sebastian Skatulla
Hans Beushausen Editors

Advances in
Information
Technology in
Civil and Building
Engineering
Proceedings of ICCCBE 2022 - Volume 2
Lecture Notes in Civil Engineering

Volume 358

Series Editors
Marco di Prisco, Politecnico di Milano, Milano, Italy
Sheng-Hong Chen, School of Water Resources and Hydropower Engineering,
Wuhan University, Wuhan, China
Ioannis Vayas, Institute of Steel Structures, National Technical University of
Athens, Athens, Greece
Sanjay Kumar Shukla, School of Engineering, Edith Cowan University, Joondalup,
WA, Australia
Anuj Sharma, Iowa State University, Ames, IA, USA
Nagesh Kumar, Department of Civil Engineering, Indian Institute of Science
Bangalore, Bengaluru, Karnataka, India
Chien Ming Wang, School of Civil Engineering, The University of Queensland,
Brisbane, QLD, Australia
Lecture Notes in Civil Engineering (LNCE) publishes the latest developments in
Civil Engineering—quickly, informally and in top quality. Though original research
reported in proceedings and post-proceedings represents the core of LNCE, edited
volumes of exceptionally high quality and interest may also be considered for publi-
cation. Volumes published in LNCE embrace all aspects and subfields of, as well as
new challenges in, Civil Engineering. Topics in the series include:
• Construction and Structural Mechanics
• Building Materials
• Concrete, Steel and Timber Structures
• Geotechnical Engineering
• Earthquake Engineering
• Coastal Engineering
• Ocean and Offshore Engineering; Ships and Floating Structures
• Hydraulics, Hydrology and Water Resources Engineering
• Environmental Engineering and Sustainability
• Structural Health and Monitoring
• Surveying and Geographical Information Systems
• Indoor Environments
• Transportation and Traffic
• Risk Analysis
• Safety and Security

To submit a proposal or request further information, please contact the appropriate

Springer Editor:
– Pierpaolo Riva at pierpaolo.riva@springer.com (Europe and Americas);
– Swati Meherishi at swati.meherishi@springer.com (Asia—except China,
Australia, and New Zealand);
– Wayne Hu at wayne.hu@springer.com (China).
All books in the series now indexed by Scopus and EI Compendex database!
Sebastian Skatulla · Hans Beushausen
Editors

Advances in Information
Technology in Civil
and Building Engineering
Proceedings of ICCCBE 2022 - Volume 2
Editors
Sebastian Skatulla Hans Beushausen
Department of Civil Engineering Department of Civil Engineering
University of Cape Town University of Cape Town
Rondebosch, South Africa Rondebosch, South Africa

ISSN 2366-2557 ISSN 2366-2565 (electronic)

Lecture Notes in Civil Engineering
ISBN 978-3-031-32514-4 ISBN 978-3-031-32515-1 (eBook)
https://doi.org/10.1007/978-3-031-32515-1

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature
Switzerland AG 2023

This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

The Proceedings of the 19th International Conference on Computing in Civil

and Building Engineering (ICCCBE) presents findings concerning current and
future applications of computing in a wide range of civil and building engineering
disciplines.
The conference was the latest in a sequence of ICCCBE International Conferences
being held for the first time in Africa where it took place in Cape Town, South Africa,
during 26–28 October 2022. It promoted innovation in civil engineering through
digital twin technology; geometric modelling; simulation and process modelling;
monitoring, information and communication technologies; big data, artificial intel-
ligence and machine learning software; robotics, automation and control; computa-
tional structural mechanics. The conference therefore provided a forum and a unique
opportunity for researchers and students from the African continent to interact with
members of the civil engineering community from around the world.
ICCCBE 2022 has received excellent support by researchers and practitioners all
across South Africa, with authors being drawn from numerous research and indus-
trial organisations. The Proceedings contains scientific contributions presented at the
conference, classified into a total of 16 themes split over two volumes. Only original
contributions were considered for inclusion. All papers submitted were subjected to
a full process of double-blind peer review. The review of manuscripts was under-
taken by the members of the Scientific Advisory Board and other identified leading
experts, acting independently on one or more assigned manuscripts. This invaluable
assistance, which has greatly enhanced the quality of the Proceedings, is gratefully
acknowledged.
Special acknowledgments are due to our supporters and partners in the industry:
• CIOB
• SSD
• SAICE

v
vi Preface

Finally, the editors wish to thank the authors for their efforts towards producing
and delivering papers of high standard creating a base for discussion and providing
suggestions for future development and research.

S. Skatulla
H. Beushausen
Contents

Information and Communication Technologies (IoT,

Crowdsourcing, Social Networks)
Trends and Recommendations for IoT-Based Smart City
Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Jose Joaquin Peralta Abadia and Kay Smarsly
Text Mining-Based Patent Analysis for Automated Rule Checking
in AEC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Zhe Zheng, Bo-Rui Kang, Qi-Tian Yuan, Yu-Cheng Zhou,
Xin-Zheng Lu, and Jia-Rui Lin
Unravelling the State of the Art of Blockchain Development
for Improved Infrastructure Delivery in the Built Environment:
A Bibliometric Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Motheo Meta Tjebane, Innocent Musonda, Adetayo Onososen,
and Molusiwa Ramabodu

Project Design, Construction, Planning, and Management

Organisational Leadership as a Driver for the Adoption of Digital
Technologies for Construction Project Delivery . . . . . . . . . . . . . . . . . . . . . . . 45
Matthew Ikuabe, Clinton Aigbavboa, Samuel Adekunle,
Babatunde Ogunbayo, and Rotshidzwa Mugaga
Modeling Drywall Construction Process Using a Spatiotemporal
Chronographical Scheduling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Adel Francis and Edmond Miresco
Towards Increased Situational Awareness at Unstructured Work
Zones: Analysis of Worker Behavioral Data Captured in VR-Based
Micro Traffic Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Julia Qin, Daniel Lu, and Semiha Ergan

vii
viii Contents

Resilient and Sustainable Urban and Energy Systems

How Can Digital Twins Support the Net Zero Vision? . . . . . . . . . . . . . . . . . 81
Eleni Papadonikolaki and Chimay Anumba
A Review of Smart City Maturity Assessment Models . . . . . . . . . . . . . . . . . 99
Pavan Kumar, Aritra Pal, and Shang-Hsien Hsieh
Challenges for the Implementation of Sustainable Construction
Practices in Developing Countries: A Bibliometric Review . . . . . . . . . . . . . 109
Motheo Meta Tjebane, Innocent Musonda, Adetayo Onososen,
and Molusiwa Ramabodu

Robotics, Automation, and Control

Structural Health Monitoring of Civil Infrastructure using Mobile
Robots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
Kay Smarsly, Kosmas Dragos, Jan Stührenberg, and Mathias Worm
Bottlenecks to the Implementation of Automation and Robotics
in the Construction Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Ayanda Boya, Opeoluwa Akinradewo, Clinton Aigbavboa,
Andrew Ebekozien, and Molusiwa Ramabodu

Simulation and Process Modeling

Integrating AEC Domain-Specific Multidisciplinary Knowledge
for Informed and Interactive Feedback in Early Design Stages . . . . . . . . . 153
Julia Reisinger, Shervin Rasoulzadeh, Balint Istvan Kovacs,
Peter Ferschin, Khrystyna Vasylevska, Michael Hensel, Iva Kovacic,
and Michael Wimmer
Comparative Application of Digital Image Processing
and Kuz-Ram Model in Blast Fragmentation Analysis: Case
of Shayona Cement Quarry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
Jabulani Matsimbe, Martin Shaba, Innocent Musonda,
and Megersa Dinka

Technology-Enriched Engineering Pedagogy

Proposal of a Collaborative Teaching Method for AEC Supported
by Additive Manufacturing Use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
Márcio Henrique de Sousa Carboni, Sérgio Scheer,
and Armando Luis Yoshio Ito
Development of a Safe and Anthropomorphic Drone
in an Interdisciplinary Research-Oriented Construction
Management Course . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Gilles Albeaino, Sebastian Paulis, Masoud Gheisari, and Raja R. A. Issa
Contents ix

The Integration of 4IR Technologies in Architectural Education

for Upskilling the Workforce in the Nigerian Built Environment . . . . . . . 227
Chika Okafor, Clinton Aigbavboa, John Aliu, and Ornella Tanga

Visualization (nD,VR, AR)

Interactive BIM-Based VR: A Case Study of Doors . . . . . . . . . . . . . . . . . . . 245
Sou-Han Chen and Fan Xue
Leveraging Virtual Reality for Improved Construction Health
and Safety Training . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
Damien Smuts, Ashvin Manga, and John Smallwood
Concept and Implementation of BIM-to-World Co-registration
for Mixed Reality Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
Marat Khairtdinov, Fulya Tasliarmut, and Christian Koch
Safety and Training Implications of Human-Drone Interaction
in Industrialised Construction Sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
Adetayo Onososen, Innocent Musonda, Molusiwa Ramabodu,
and Christopher Dzuwa
Task Performance to Understand the Effectiveness of Visualisation
Technology-Based Training for Human-Drone Interaction
Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
Adetayo Onososen, Innocent Musonda, Molusiwa Ramabodu,
and Christopher Dzuwa
Real-Time Inference of Temporal Emotional and Behavioral
Conditions of Learners in Problem Solving Within Mixed Reality
Environments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
Édouard Castets, Ivan Mutis, and Gady Agam

Other
Blockchain-Enabled-Trust for Construction Project Governance:
A Conceptual Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
Seongha Hwang, Mingzhu Wang, Mohamed Osmani, and Karen Blay
Blockchain Technologies and Building/City Information Modelling . . . . 345
Oluwatoyin O. Lawal and Nawari O. Nawari
Exploring Digitisation as a Solution to the Long-Term Insurance
Sector Efficiency Quagmire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
Linda Malifete, Samuel Adekunle, and Clinton Aigbavboa
x Contents

Space Partitions: An Alternative to Domain Assembly in

Geometric Modeling
A Computational Robust Method for Spatial Decomposition - Test
Case with Cadastral Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
Enrico Romanschek, Christian Clemen, and Wolgang Huhnt
Where is the End of the Wall: Decomposition of Air and Material
into Spaces and Building Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
Felix Gabler and Wolfgang Huhnt
Transforming Building Components into a Space Partition
to Identify Indoor and Outdoor Spaces in Digital Building Models . . . . . 403
Joanna Zarah Vetter and Wolfgang Huhnt
Space Partitioning as a Holistic Alternative to Traditional
Geometric Modeling Workflows in the AEC Industry . . . . . . . . . . . . . . . . . 415
Wolfgang Huhnt, Joanna Zarah Vetter, and Maximilian Sternal
Robust Modeling of Polyhedral Space Partitions . . . . . . . . . . . . . . . . . . . . . . 427
Maximilian Sternal and Wolfgang Huhnt
Feature-Based Decomposition of Architectural Spaces: Outline
of a Procedure and Research Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
Georg Suter

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 457

 Information and Communication
Technologies (IoT, Crowdsourcing, Social
Networks)
Trends and Recommendations
for IoT-Based Smart City Applications

Jose Joaquin Peralta Abadia and Kay Smarsly

Abstract Smart monitoring and wireless sensor networks have advanced the world-
wide implementation of smart city applications, improving the quality of life and
reducing negative effects on the environment. The Internet of Things (IoT) has
received special attention in recent years in the context of smart city applications,
providing the technological backbone necessary to perform real-time smart moni-
toring of the built and the natural environment. Nevertheless, due to the variety of IoT
elements and technologies used in IoT frameworks for smart city applications, new
implementations of IoT frameworks for smart city applications may have redundant
IoT elements or may use inappropriate technologies. This paper aims to summarize
and compare emerging IoT elements and technologies utilized in IoT frameworks
for smart city applications by means of a systematic review. As a result of this study,
a summary and comparison of trends in IoT frameworks for smart city applica-
tions are provided. Furthermore, recommendations for future implementations of
IoT frameworks for smart city applications are provided, based on the findings of the
systematic review. It is expected that the findings of the systematic review and the
recommendations for future implementations of IoT frameworks presented in this
study may serve as a guideline and a foundation for IoT framework implementations
in the context of smart city applications.

Keywords Smart cities · Internet of Things (IoT) · Wireless sensor networks · IoT
frameworks · IoT applications

J. J. P. Abadia
Faculty of Engineering, Mondragon University, Mondragon, Spain
K. Smarsly (B)
Institute of Digital and Autonomous Construction, Hamburg University of Technology, Hamburg,
Germany
e-mail: kay.smarsly@tuhh.de

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 3

S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_1
4 J. J. P. Abadia and K. Smarsly

1 Introduction

Industry 4.0, cyber physical systems, and artificial intelligence have brought rapid
advancements in smart cities and smart city applications, effectively replacing tradi-
tional civil engineering applications [1, 2]. Discussions regarding the constituents of
a smart city may be found in research, where other terms, such as “intelligent city”,
“digital city” and “ubiquitous city”, have also been used. However, common concepts
in the discussions may be identified, such as the use of information and communi-
cation technologies (ICT) [3]. Essentially, smart cities improve the quality of life by
means of user-friendly technologies and applications, i.e. smart city applications [4].
Smart city applications are usually supported by the Internet of Things (IoT), where
billions of interconnected “things” (IoT elements or devices) are coupled through
wide area networks and the Internet [5]. In this regard, IoT frameworks provide the
means for connecting IoT elements, serving as a basis for using IoT technologies.
Although IoT frameworks provide the means for connecting IoT elements, imple-
menting IoT framework for smart city applications may become complicated, costly,
and inefficient, in light of the variety of IoT technologies. Literature reviews of
IoT frameworks and smart cities have given emphasis on different aspects, such as
definitions of the term “smart city” and the domains smart cities are composed of
[6]. Other reviews have compared IoT technologies used in smart city applications
[7]. Furthermore, challenges and opportunities for smart cities have been given [8].
Nevertheless, a lack of focus on trends in the technologies used by IoT frameworks for
smart city applications has been identified. As such, this paper reviews the trends in
IoT frameworks for smart city applications. The remainder of this paper is structured
as follows. The next section presents the concepts of smart cities and IoT frameworks
required to perform the review. Then, the review of trends in IoT frameworks for
smart city applications is provided, followed by a discussion of the trends. Finally, a
short statement on conclusions and future work is given.

2 Concepts and Guidelines

This section presents the concepts of smart cities and IoT frameworks used as a basis
for the review of trends in smart city applications. Working definitions and domains
of smart cities are presented, followed by working definitions and architectures of
IoT frameworks.
When first coining the term “smart city”, it described an intelligent city, which
portrays a technopolis and hard infrastructure [9]. Subsequently, the term has been
defined in several ways. For example as territories with the capacity for learning and
innovation, building upon knowledge creation and creativity of their population, as
well as digital infrastructure [10]. In addition, research has attempted to define the
domains constituting smart cities, as described in [11]. From the aforementioned
review, two examples have been selected as a basis for this study. In [12], the authors
Trends and Recommendations for IoT-Based Smart City Applications 5

have listed the smart city domains as smart economy, smart people, smart governance,
smart mobility, smart environment, and smart living. Thereupon, in [13], the list has
been expanded to include smart buildings, while categorizing the smart city domains
into soft domains and hard domains. On the one hand, soft domains refer to intangible
characteristics of smart cities. On the other hand, hard domains refer to tangible char-
acteristics of smart cities. Considering the smart city domains mentioned above, this
study considers smart economy, smart governance, smart citizens (replacing smart
people), and smart living as soft domains of a smart city, and smart mobility, smart
environment, and smart infrastructure (replacing smart buildings) as hard domains
of a smart city.
As with the several definitions of the term “smart city”, IoT frameworks have been
defined in several ways. For example, IoT frameworks have been defined as a user-
friendly environment capable of supporting billions of IoT devices, facilitating the
development and testing of new functionalities [14]. Regarding the architecture of
IoT frameworks, the IEEE Internet Initiative (2015) has proposed a three-layered IoT
framework architecture constituted by an application layer, a networking and data
communications layer, and a sensing layer [15]. Additionally, a similar IoT frame-
work architecture has been proposed, consisting of an application layer, an integration
middleware layer, a gateway layer, a device layer, and a sensor and actuator layer
[16]. In this study, both architectures are merged into a three-layered IoT framework
architecture, composed of sensing layer (including gateways, devices, sensors, and
actuators), middleware layer (including networking and data communications), and
application layer. Based on the concepts presented thus far, the following section
identifies the trends in IoT frameworks for smart city applications, comparing the
smart city domains targeted by the IoT frameworks and the IoT elements and services
offered by the IoT frameworks.

3 Review of IoT Frameworks for Smart City Applications

This section reviews the trends in IoT frameworks for smart city applications. To make
this study brief, the application layer of the IoT frameworks, which is primarily used
for data visualization, has been omitted because it does not include IoT elements and
technologies to be reviewed.
The review methodology pursued in this study includes three phases. In the first
phase, data has been collected. In this phase, 42 studies between the years 2016 and
2020 and indexed in the Web of Science Core Collection, supplemented by confer-
ence papers indexed in the Scopus database, have been gathered. The search string
used for the data collection is (“smart cities” OR “smart infrastructure” OR “smart
living” OR “smart environment” OR “smart mobility” OR “smart economy” OR
“smart citizens” OR “smart governance”) AND (“Internet of Things” OR “IoT” OR
“IoT framework” OR “IoT platform”). In the second phase, the data has been orga-
nized and the IoT elements of each layer of the IoT frameworks have been tabulated,
based on the concepts presented in previous section. The focus on the sensing layer
6 J. J. P. Abadia and K. Smarsly

has been put on the IoT devices (hardware) used in the IoT frameworks. The focus
on the middleware layer has been put on the IoT services (software) provided by the
IoT frameworks. In the third and final phase, the data is analyzed, summarized and
compared. The results of the review, in accordance with the methodology, are the
listed as follows.
• Smart city domain: As presented in the previous section, smart cities are consti-
tuted by seven domains and IoT frameworks for smart city applications are usually
developed to be used in only one of the domains. However, several studies have
presented IoT frameworks that may be used in more than one smart city domain
[17]. Figure 1 presents the quantity of IoT frameworks that can be used in each of
the seven smart city domains. Hard domains are presented in yellow, soft domains
are presented in green stripes, and an extra category presented in white is used to
group the IoT frameworks that may be used in any of the seven smart city domains.
It may be observed that 39 IoT frameworks have been used in hard domains of
a smart city, while eight IoT frameworks have been used in soft domains. In
addition, eight IoT frameworks have been developed to be used in any smart city
domain.

• Sensing layer: As mentioned previously, focus in this study is put on the IoT
devices used in the sensing layer. Figure 2 presents the IoT devices used in the
sensing layer and the quantity of IoT frameworks that use the IoT devices. IoT
frameworks may be developed to accept any IoT device or specific IoT devices. It
has been found that 15 IoT frameworks have been developed to use any IoT device,
enabling interoperability. Contrarily, it has been found that 31 IoT frameworks
have been developed to use specific IoT devices. However, some IoT frameworks
have reported being able to use more than one specific IoT device type, as in [18].

16
14 15 Hard domain
Quantity of IoT frameworks

12 Soft domain

10
8 9 9
8
6
6
4
4
2 3 1
0

Fig. 1 Categorization of IoT frameworks into the smart city domains

Trends and Recommendations for IoT-Based Smart City Applications 7

Fig. 2 IoT device types 16

used in the sensing layer of

Quantity of IoT frameworks

14 15
IoT frameworks 12
10 11
8
8
6
6 6
4 5
2
0

In this regard, it has also been found that custom IoT devices are used 11 times,
Arduino-based IoT devices are used eight times, smartphones are used six times,
and Raspberry Pi-based IoT devices are used six times.

• Middleware layer: Opposing the sensing layer, focus in this study is put on the
IoT services provided in the middleware layer. Thus, Fig. 3 presents the services
that are provided by the middleware layer and the quantity of IoT frameworks that
provide the services. It has been found that the most-provided services are database
management and authentication services, which are used in 32 IoT frameworks,
and event and alert management, which has been provided in 27 IoT frameworks.
Following, resource management and resource discovery has been provided in 23
and 21 IoT frameworks, respectively, allowing the management and registration of
the IoT devices of the sensing layer. It has also been found that 12 IoT frameworks
provide machine learning functionality, performing data analytics and forecasting.
Furthermore, 11 IoT frameworks have provided encryption services, adding an
extra layer of security to the IoT frameworks. Finally, it has been identified that
the least provided service, offered in three IoT frameworks, is code management.

4 Discussion and Implications

This section discusses the results of the review of trends in IoT frameworks for smart
city applications presented in the previous section. Discussions of the trends of IoT
frameworks for smart city applications, in relationship to smart city domains and
sensing and middleware layers of IoT frameworks, are presented as follows.
• Smart city domains: On the one hand, it has been identified that hard domains
of smart cities are targeted by the majority of IoT frameworks, probably due to
ease of implementation and quantification of results, as hard domains refer to
tangible assets and characteristics of a smart city. On the other hand, soft domains
8 J. J. P. Abadia and K. Smarsly

35

30 32 32
Quantity of IoT frameworks
25 27

20 23
21
15

10 12
11
5
3
0

Fig. 3 Services provided by the middleware layer of the IoT framework

have rarely been targeted, where the results are harder to quantify, probably due
to the intangible nature of soft domains. Therefore, it has been noticed that smart
cities seem to prefer to invest more in smart city applications that provide fast and
quantifiable results.
• Sensing layer: It has been found that the usage of specific IoT devices has
been preferred by the majority of IoT frameworks. However, the preference for
specific IoT devices derives in a lack of interoperability and possible problems of
upgradeability, when new IoT technologies and IoT devices become available.
• Middleware layer: Representing basic services for the operation of the IoT
middleware layer, database management and authentication services have been
identified as the services that are mostly offered. Therefore, additional services
of interest, such as machine learning and data analytics services, emerge in IoT
frameworks, as large amounts of data are readily available. Nevertheless, it has
been observed that some IoT frameworks may be vulnerable to security breaches,
related mainly to the lack of data encryption.

5 Summary and Conclusions

In this study, definitions and concepts of smart cities and IoT frameworks, as well as a
review of trends in IoT frameworks for smart city applications, have been presented.
The IoT technologies and IoT devices have been summarized and compared. This
paper is a short version of a journal paper. This paper has exclusively been written
to present and discuss the systematic review with the ICCCBE audience. In the
Trends and Recommendations for IoT-Based Smart City Applications 9

corresponding journal paper, 86 sources have been incorporated, and details may be
found in [19].
Since IoT frameworks are usually based on a three-layered architecture, consti-
tuted by a sensing layer, a middleware layer, and an application layer, which can
distributedly be located in distant geographical positions, it becomes necessary to
guarantee secure communication between the layers. As a result of this study, it has
been identified that, in spite of most IoT frameworks offering authentication services,
there are possible security vulnerabilities in the IoT frameworks, as few offer encryp-
tion services. In addition, the lack of interoperability and code management has to
be targeted as well, thus enabling the use of future IoT devices and IoT technologies.
As a result of the findings of this study, it has been identified that IoT frameworks
for smart city applications should address decentralization, interoperability, security,
and modularity, as well as a need to address the use of the IoT frameworks in the
soft domains of a smart city.
Future research may focus on the development of new IoT framework concepts
for smart city applications, targeting the issues revealed in this study, i.e. a lack of
interoperability and decentralization of the services as well as security and modu-
larity. Furthermore, the development of IoT frameworks that can be used in soft
domains of a smart city may be addressed, thus giving focus to the quality of life of
the citizens.

Acknowledgements The authors of this study gratefully acknowledge the support offered by the
German Research Foundation (DFG) through grant SM 281/12 and grant SM 281/17. Any opinions,
findings, conclusions, or recommendations expressed in this paper are those of the authors and do
not necessarily reflect the views of the DFG.

References

1. Luckey D, Fritz H, Legatiuk D et al (August 18, 2020) Artificial intelligence techniques for
smart city applications. In: Proceedings of the International ICCCBE and CIB W78 Joint
Conference on Computing in Civil and Building Engineering 2020. Sao Paolo, Brazil
2. Fritz H, Smarsly K (June 7, 2020) A state-of-the-art review of nature-inspired systems for
smart structures. In: Proceedings of the European Workshop on Structural Health Monitoring.
Palermo, Italy
3. Hollands RG (2008) Will the real smart city please stand up? intelligent, progressive or
entrepreneurial? City 12:303–320
4. Law KH, Lynch JP (2019) Smart city: technologies and challenges. IT Professional 21:46–51
5. Gubbi J, Buyya R, Marusic S et al (2013) Internet of Things (IoT): A vision, architectural
elements, and future directions. Futur Gener Comput Syst 29:1645–1660
6. Silva BN, Khan M, Han K (2018) Towards sustainable smart cities: a review of trends,
architectures, components, and open challenges in smart cities. Sustain Cities Soc 38:697–713
7. Zhu H, Chang ASF, Kalawsky RS et al (October 29, 2017) Review of state-of-the-art wireless
technologies and applications in smart cities. In: Proceedings of the 43rd Annual Conference
of the IEEE Industrial Electronics Society. Beijing, China
8. Talari S, Shafie-khah M, Siano P et al (2017) A review of smart cities based on the internet of
things concept. Energies 10:421
10 J. J. P. Abadia and K. Smarsly

9. Heng TM, Low L (1993) The intelligent city: singapore achieving the next lap. Technology
Analysis & Strategic Management 5:187
10. Komninos N (July 5, 2006) The architecture of intelligent cities: integrating human, collective
and artificial intelligence to enhance knowledge and innovation. In: Proceedings of the 2nd
International Conference on Intelligent Environments. Athens, Greece
11. Albino V, Berardi U, Dangelico RM (2015) Smart cities: definitions, dimensions, performance,
and initiatives. J Urban Technol 22:3–21
12. Giffinger R, Fertner C, Kramar H et al (2007) Smart cities – ranking of european medium-sized
cities. Vienna University of Technology
13. Neirotti P, Marco A de, Cagliano AC et al. (2014) Current trends in smart city initiatives: some
stylised facts. Cities 38:25–36
14. Uviase O, Kotonya G (2018) IoT architectural framework: connection and integration
framework for IoT systems. Electronic Proceedings in Theoretical Computer Science 264:1–17
15. IEEE Internet Initiative (2015) Towards a definition of the Internet of Things (IoT). Revision 1
16. Guth J, Breitenbücher U, Falkenthal M et al (November 23, 2016) Comparison of IoT plat-
form architectures: a field study based on a reference architecture. In: Proceedings of the 2nd
Cloudification of the Internet of Things. Paris, France
17. Calderoni L, Magnani A, Maio D (2019) IoT manager: an open-source IoT framework for
smart cities. J Syst Architect 98:413–423
18. Hussain A, Wenbi R, da Silva AL et al (2015) Health and emergency-care platform for the
elderly and disabled people in the smart city. J Syst Softw 110:253–263
19. Peralta J, Walther C, Osman A, Smarsly K (2022) A systematic survey of Internet of Things
frameworks for smart city applications. Sustainable Cities and Society, 103949.https://doi.org/
10.1016/j.scs.2022.103949
Text Mining-Based Patent Analysis
for Automated Rule Checking in AEC

Zhe Zheng, Bo-Rui Kang, Qi-Tian Yuan, Yu-Cheng Zhou, Xin-Zheng Lu,
and Jia-Rui Lin

Abstract Automated rule checking (ARC), which is expected to promote the effi-
ciency of the compliance checking process in the architecture, engineering, and
construction (AEC) industry, is gaining increasing attention. Throwing light on the
ARC application hotspots and forecasting its trends are useful to the related research
and drive innovations. Therefore, this study takes the patents from the database of the
Derwent Innovations Index database (DII) and China national knowledge infrastruc-
ture (CNKI) as data sources and then carried out a three-step analysis including (1)
quantitative characteristics (i.e., annual distribution analysis) of patents, (2) identifi-
cation of ARC topics using a latent Dirichlet allocation (LDA) and, (3) SNA-based
co-occurrence analysis of ARC topics. The results show that the research hotspots
and trends of Chinese and English patents are different. The contributions of this
study have three aspects: (1) an approach to a comprehensive analysis of patents by
integrating multiple text mining methods (i.e., SNA and LDA) is introduced; (2) the
application hotspots and development trends of ARC are reviewed based on patent
analysis; and (3) a signpost for technological development and innovation of ARC
is provided.

Keywords Automated Rule Checking (ARC) · Patent analysis · Text mining ·

Social Network Analysis (SNA) · Latent Dirichlet Allocation (LDA) · Application
hotspots and future trends

1 Introduction

Authorities’ rules such as design guidelines, codes, standards, and international,

national, or local authorities’ laws stipulate the safety, sustainability, and comfort
of the entire lifecycle of a built environment [1]. For a long time, the regulatory
compliance of a building design was checked manually by domain experts. However,

Z. Zheng · B.-R. Kang · Q.-T. Yuan · Y.-C. Zhou · X.-Z. Lu · J.-R. Lin (B)
Department of Civil Engineering, Tsinghua University, Beijing 100084, China
e-mail: lin611@tsinghua.edu.cn

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 11

S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_2
12 Z. Zheng et al.

this process is not only immensely time-consuming but is also challenging for project
participants [2] due to the complexity and ambiguity of rules and regulations [3].
Therefore, automated rule checking (ARC, also known as ACC) has been extensively
studied by many researchers for nearly 70 years [4], and several rule-based ARC
systems have been established [5]. For example, CORENET (short for COnstruction
and Real Estate NETwork), which is the first large effort of ARC, was initiated in
1995 in Singapore [6]. A universal circulation network (UCN) is developed as a plug-
in based on the Solibri model checker (SMC) to check occupant circulation rules of
the US court design guide [7]. Other rule-based ARC systems include BCAider
[8], Jotne EDModelChecker [9], and so on. However, the above rule-based ARC
systems are costly to maintain and inflexible to modify and are often referred to as
black-box approaches [10]. Hence, to make the process more flexible and transparent,
attempts have been made to (1) automated interpretation of the rules based on natural
language processing (NLP) [11–13], (2) automated alignment of BIMs and computer-
interpretable regulations [14], and so on.
To promote future development in the field, many efforts have been devoted to the
analysis of the development trends. For example, Eastman et al. [5] examined and
surveyed five rule checking systems that assess building designs according to various
criteria in detail. Ismail et al. [15] reviewed the previous studies, which successfully
employed the appropriate techniques in interpreting the rules for checking purposes.
Fuchs [16] assessed the state-of-the-art of NLP for building code interpretation by
analyzing 42 research articles published since 2000. These works have pointed out the
key research areas and analyzed the current and ongoing research and development
(R&D) directions. However, the existing studies mainly focused on the existing
ARC systems or research articles, while seldom considering the patent analysis.
Patent analysis is valuable to throw light on the innovation and development of a
technological field over the course of time [17, 18].
Therefore, this work aims to explore the application hotspots and trends of compli-
ance checking by analyzing the related patents based on text mining methods. The
patents from the database of the Derwent Innovations Index database (DII) and China
national knowledge infrastructure (CNKI) are taken as data sources. Then the anal-
ysis is comprised of three steps. First, the quantitative characteristics and annual
distribution analysis of patents are carried out. Second, several topics of the ARC
patents are identified using a latent Dirichlet allocation (LDA). Third, to reveal the
relation between different topics, the identified ARC topics co-occurrence analysis is
carried out based on Social Network Analysis (SNA). Finally, based on the analysis
results, the application hotspots and trends are revealed. The findings are expected
to be valuable for the related stakeholders including researchers, ARC system devel-
opers, and companies. The introduced approaches to the analysis of ARC patents are
also useful for other scenarios.
Text Mining-Based Patent Analysis for Automated Rule Checking in AEC 13

2 Methodology

This study utilizes patents as data resources and integrates several text mining tech-
nologies including the LDA topic identification technology, and the SNA topic corre-
lation analysis to identify the application hotspots of ARC and forecast its trends.
Figure 1 shows the outline and workflow of this research. Meanwhile, the toolkits
used to implement the workflow are also displayed at the bottom of Fig. 1. The
methods deployed are detailly illustrated from Sect. 2.1 to Sect. 2.3.

2.1 Data Acquisition

In this work, the patent retrieval process consists of (1) screening by keywords and,
(2) screening manually by domain experts to filter the patents that are not relevant
to the rule checking. The Chinese and English patents are obtained from the CNKI
and DII database, respectively, and the retrieval time is from 2010 to 2021. For the
Chinese patents, the keyword is “design review”, and after the manual screening,
total 113 Chinese patents were collected for analysis. For the patents in English,
the keywords include “rule checking”, “compliance checking”, “design review”,
“automated compliance checking”. After the manual screening, finally, total 121
patents in English were collected for analysis.
The abstracts of the collected patents are used for further analysis. After the patent
retrieval process, the data preprocessing consists of tokenization, special characters
(e.g., !%$#& ∗ ?,/.;”\) removal, and stop words removal. For the Chinese patents,
the original texts are tokenized and removed stop words using the Python lib named
Jieba, while the English patents are processed using the Python lib named NLTK
(Natural Language Toolkit).

1 Data acquisition 2 Topic identification 3 Analysis and discussion

LDA-based topic
Patent search identification analysis Topic development
trend analysis
evaluation

Workflow Coherence
Data Cosine similarity SNA-based topic co-
preprocessing occurrence analysis
KL divergence

NLTK Gensim Networkx

Toolkits
Jieba Numpy Scipy

Fig. 1 Workflow of this research

14 Z. Zheng et al.

2.2 Topic Identification

After the acquisition and pre-processing of the patent data, the topic identification
is performed to mine the latent information of patents. Topic identification methods
(i.e., topic modeling methods) are typical text mining methods. The Latent Dirichlet
Allocation (LDA) model proposed by Blei et al. [19] is one of the most widely-used
topic modeling methods [18, 20–22]. LDA model is a generative probabilistic model
of corpus where each document is represented as random mixtures over latent topics,
and each topic is characterized by a distribution over words [19]. The basic idea of
the LDA model is shown in Fig. 2. The hyperparameter α is the Dirichlet prior on
the document-topic distributions and the hyperparameter β is the Dirichlet prior on
the topic-word distributions. LDA model assumes the following steps to generate a
document [18, 19].

1. For each document d, sample a topic proportion θ d from the Dirichlet distribution
Dir (α).
2. For each word wd, n to be generated in the document d:
a. Choose a topic zd, n ∼ Multinomial(θ d)
b. Choose a word wd, n ∼ Multinomial(ϕk|k = zd, n), where the word
proportion per topic ϕk is sampled from the Dirichlet distribution Dir (β).

To determine the optimal number of topics generated by the LDA model, the
coherence scores of the topics are used [24]. The coherence metric uses statistics
and probabilities extracted from the reference corpus to evaluate how well a topic is.
The larger the coherence scores are, the better the effect of LDA topic identification.
The coherence measure method consists of four main steps: (1) segmentation, (2)
probability calculation, (3) confirmation measure, and (4) aggregation [24]. The
coherence model in the genism toolkit is used to implement the evaluation process.
Besides, to evaluate the performance of the topic generation results, the topic
distances among the topics are measured via Kullback–Leibler (KL) divergence and
cosine similarity analysis. According to the LDA model, each topic is represented as
a set of words with proportional. So, based on KL divergence and cosine similarity,
the distance matrix should be constructed, where the element in the i row and the

Topic level

Document level

Fig. 2 Latent Dirichlet Allocation (LDA) model

Text Mining-Based Patent Analysis for Automated Rule Checking in AEC 15

j column represents the specific distance value between the number i topic and the
number j topic. For the KL divergence analysis, the distance matrix is an asymmetric
matrix. The larger the value of the non-diagonal elements of the distance matrix, the
greater the differentiation of words between two topics and the better performance of
topics generated by the LDA model. For the cosine similarity analysis, the distance
matrix is a symmetric matrix. The smaller the value of the non-diagonal elements of
the distance matrix, the greater the differentiation of words between two topics and
the better performance of topics generated by the LDA model.

2.3 Topic Co-occurrence Analysis

After the LDA analysis, the topic of each patent is recognized. To further explore the
topic relevance between the identified topics, the SNA-based topic relevance analysis
is performed. SNA is often used to explore the publication, citation, and cocitation
networks, collaboration structures, and other forms of social interaction networks
[25].
In this work, the nodes in the topic relevance graph are the identified topics.
Meanwhile, the link between two nodes means there is at least one article that contains
the two topics at the same time. According to the LDA model, each article consists
of a bag of topics, where the larger the percentage of topics, the more important the
topic is. Therefore, when a topic accounts for more than 10%, we consider this topic
to be the article’s main topic. If an article has multiple main topics at the same time,
these topics are called co-occurrence topics. Then a link will be generated for these
two topics. The attribute of each link means the number of articles that contains the
two topics. Therefore, the topic relevance graph is a typical undirected symmetric
network. After the construction of the topic relevance graph, Gephi [26] is then used
to perform the visualization of network analysis.

3 Results

3.1 Quantitative Characteristics of ARC Patents

The annual publication numbers of patents can reflect the development trend and
research interests. Therefore, after the patent data acquirement and preprocessing,
the trend analysis is conducted, as shown in Fig. 3. The first Chinese patent on rule
checking appeared in 2011, while the first English patent appeared in 2013. The trend
of the Chinese and English patents is similar. The number of patents grows slowly
until 2015. From 2016 to 2021, the growth rate of the related patents has significantly
accelerated, which indicates that the rule checking methods and systems are gaining
more and more interest in recent years.
16 Z. Zheng et al.

35
Chinese
30
English
Number of patents

25

20

15

10

0
2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021
Year
Fig. 3 The number of patents on rule checking

3.2 LDA-Based Topic Clustering of ARC Patents

To further analyze the application hotpots and development trends of the ARC
patents, the LDA model is utilized to explore the main topics and their corresponding
keywords from the collected ARC-related patents. First, the relatively proper topic
numbers are searched, as shown in Fig. 4. The coherence values are used to measure
the rationality of the topics. The best topic number is 5 and 7 for the Chinese patents
and English patents, respectively. The coherence value of the Chinese topics is higher
than that of the English topics, indicating that the quality of Chinese topics is higher
than that of English topics.
To further evaluate the performance of the topic generation results, the topic
distances among the topics are calculated via Kullback–Leibler (KL) divergence and
cosine similarity analysis. The distance matrixes of the topics are shown in Fig. 5. For
the Chinese patents, as shown in Fig. 5a, A13 (Aij means the element at the number
i row and the number j column) and A25 have the largest cosine similarity. A13 and
A25 have the smallest KL divergence, as shown in Fig. 5b. The results of the two
matrixes are consistent, indicating that topic 2 and topic 5 have the highest similarity,
and topic 3 and topic 2 have the second-highest similarity. For the English patents,
as shown in Fig. 5c, A25, A27, A56, and A67 have the largest cosine similarity. A52
and A76 have the smallest KL divergence, as shown in Fig. 5d. The results of the
two matrixes are consistent, indicating that the similarity between topics 2 and 5 and
the similarity between topics 6 and 7 are relatively high. In general, the similarity
between Chinese topics is lower than that between English topics, indicating that the
Text Mining-Based Patent Analysis for Automated Rule Checking in AEC 17

Fig. 4 Coherence values of 0.65

the topics constructed by
LDA
0.6 [5, 0.596]

Coherence
0.55

0.5

0.45
2 3 4 5 6 7 8 9 10 11 12 13
Number of topics
(a) Chinese patents
0.35
[7, 0.333]
0.33
Coherence

0.31

0.29

0.27

0.25
2 3 4 5 6 7 8 9 10 11 12 13
Number of topics
(b) English patents

quality of Chinese topics is higher than that of English topics. This is consistent with
the result of coherence analysis, as shown in Fig. 4.
Then all the patents are analyzed, and the identified topics according to the word
distributions are listed in Tables 1 and 2. The top 10 words and their corresponding
weight values (percentage) are also listed in Tables 1 and 2. For the Chinese patents,
the topics are topic 1 “codes database and auxiliary software”, topic 2 “drawings
and models information extraction”, topic 3 “CAD element identification modules”,
topic 4 “User interaction”, and topic 5 “Auxiliary devices and systems”. Topics 2
and 5 may share some keywords on drawings or model information; therefore, the
similarity is relatively high.
For the English patents, the topics are topic 1 “digital system and terminal”,
topic 2 “ BIM applications in design and construction”, topic 3 “ data integration
and management”, topic 4 “computer-executable rule and knowledge base”, topic
5 “BIM-based management platform”, topic 6 “Auxiliary tools for rule checking”,
and topic 7 “Rule Checking systems”. Topic 2 and topic 5 are about BIM-related
applications; therefore, the similarity between the two topics is relatively high. And
18 Z. Zheng et al.

(a) Cosine similarity for Chinese patents (b) KL divergence for Chinese patents

(c) Cosine similarity for English patents (d) KL divergence for English patents

Fig. 5 Distance matrixes of the topics

topics 6 and 7 are about auxiliary tools or systems for rule checking; consequently,
the similarity between the two topics is also relatively high, which is consistent with
the results of Fig. 5c and 5d.

3.3 Analysis and Discussion

After the topic construction via the LDA models, the topic-based quantification
analysis is conducted to explore the trends, and the SNA-based topic co-occurrence
analysis is performed to provide more insight into the combination and development
of the technologies.
Table 1 Distributions of the topic of the Chinese ARC-related patents
Topics Keywords and probabilities
Codes Design Rule module codes target system data database method region
database and review check
auxiliary 2.7 2.7 2.3 1.5 1.4 1.2 1.1 1.1 1.0 1.0
software
Drawings and drawings check component Design building method information application target acquirement
models’ review
information 7.7 2.8 2.8 2.6 1.8 1.5 1.3 1.3 1.2 1.1
extraction
CAD element data Design information module Construction model Rule check terminal method CAD
identification review drawings
modules 2.7 2.3 1.8 1.6 1.4 1.3 1.3 1.3 1.2 1.2
User Design interface display design change Phase method model files of click
interaction review diagram drawings
3.6 2.9 2.6 2.5 1.7 1.7 1.4 1.4 1.4 1.2
Auxiliary drawings target Design information module building project component application detection
Text Mining-Based Patent Analysis for Automated Rule Checking in AEC

devices and review

systems 3.1 2.6 2.3 1.7 1.6 1.5 1.3 1.2 1.1 1.1
Note: The numbers in the table represent the percentage of words in the topic
19
 20

Table 2 Distributions of the topic of the English ARC-related patents

Topics Keywords and probabilities
Digital system and design digital information drawing building review delivery terminal modeling description
terminal 3.3 2.9 2.9 2.8 1.8 1.5 1.3 1.3 1.2 0.9
BIM applications in model information BIM drawing software building construction design rtm 3D
design and 3.7 2.5 2.0 1.8 1.3 1.3 1.3 1.2 1.2 1.2
construction
Data integration and BIM information model drawing building system based object data unit
management 2.6 2.4 1.9 1.8 1.7 1.5 1.2 1.2 1.1 0.9
Computer-executable drawing BIM model module substation system building review information arch
rule and knowledge 2.1 1.9 1.9 1.8 1.4 1.4 1.2 1.2 1.1 1.1
base
BIM-based construction drawing BIM model design information data project according management
management 3.3 2.8 2.6 2.5 2.0 1.7 1.4 1.3 1.2 1.2
platform
Auxiliary tools for model BIM rule drawing component design construction description building set
rule checking 4.0 3.8 2.4 2.3 1.6 1.3 1.2 1.0 0.9 0.9
Rule Checking model BIM component information data building drawing design construction check
systems 3.5 2.7 2.2 2.1 2.1 2.0 1.8 1.6 1.4 1.0
Note: The numbers in the table represent the percentage of words in the topic
Z. Zheng et al.
Text Mining-Based Patent Analysis for Automated Rule Checking in AEC 21

As shown in Fig. 6a, for the Chinese patents, with the development of infor-
mation technology (e.g., natural language processing, deep learning), the patents
related to ARC have shown an increasing trend since 2016, and have grown substan-
tially since 2018. Topic 2 “Drawings and models’ information extraction” has the
highest growth rate and quickly became the most published topic. The result indi-
cates that researchers are more concerned about how to extract and integrate the
information from drawings and models. Topic 1 “Codes database and auxiliary soft-
ware” has received sustained attention in recent years. Topic 4 “User interaction”
and topic 5 “Auxiliary devices and systems” have also grown rapidly since 2018.
The result shows that good user interaction methods and more convenient auxiliary
tools are important to ARC systems. Topic 4 and topic 5 started late and developed
rapidly, which shows that making the ARC system more convenient may be the future
research direction.
As shown in Fig. 6b, for the English patents, the patents related to ARC have
shown an increasing trend since 2016. Topic 5 “BIM-based management platform”,
topic 6 “Auxiliary tools for rule checking”, and topic 7 “Rule Checking systems”
grew faster and were published the most. This shows that many researchers pay
attention to the development of platforms and systems. Among them, topic 5 contains
the most patents because it’s about the whole life cycle of BIM-based management
where automated rule checking may be used in different steps. Topic 4 “Computer-
executable rule and knowledge base” currently receives relatively little attention.
However, Ismail pointed out that rule interpretation is the most vital and complex
stage in the rule checking process [15]. To achieve a fully automated rule checking
system, more attention may be paid to the topic 4 related research. Topic 2 “BIM
applications in design and construction” has a downward trend, which may be a result
of many studies not only focusing on the design stage and the construction stage but
also focusing on the whole life cycle of the buildings.
It can be concluded from Fig. 6a and 6b that the content of Chinese patents
and English patents are different. Chinese patents pay more attention to the various
components of the ARC system. In contrast, the English patents focus more on
integrating the ARC system in various BIM-based applications.
For the co-occurrence analysis, the results are shown in Fig. 7. For the Chinese
patents, topic 1 “Codes database and auxiliary software” and topic 2 “Drawings
and models information extraction” have the most co-occurrences. The results show
that when the patent involves drawing and model information extraction, it will
also include other auxiliary tools. For the English patents, topic 5 “BIM-based
management platform” co-occurs with topic 3 “Data integration and management”,
topic 6 “Auxiliary tools for rule checking”, and topic 7 “Rule Checking systems”,
respectively. Because topic 5 mainly includes a full life cycle building management
platform, which involves data integration and other auxiliary tools.
22 Z. Zheng et al.

20
topic1
18 topic2
16 topic3
Number of patents

14 topic4
12 topic5
10
8
6
4
2
0
2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021
Year
(a) Chinese patents

16
topic1
14 topic2
12 topic3
Number of patents

topic4
10 topic5
8 topic6
topic7
6
4
2
0
2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021
Year
(b) English patents

Fig. 6 The number of patents on different topics constructed by LDA

Text Mining-Based Patent Analysis for Automated Rule Checking in AEC 23

(a) Chinese Patents (b) English patents

Fig. 7 Co-occurrence of the topics

4 Conclusions

ARC is expected to promote the efficiency of the compliance checking process in the
AEC industry. To analyze the trend development and hotpots of the ARC, this paper
first extracted the patents’ texts from the database of Derwent Innovations Index
database (DII) and China national knowledge infrastructure (CNKI). Subsequently,
a three-step analysis approach is carried out, which includes (1) quantitative char-
acteristics (i.e., annual distribution analysis) of patents, (2) identification of ARC
topics using a latent Dirichlet allocation (LDA), and, (3) SNA-based co-occurrence
analysis of ARC topics. The conclusions are as follows:
(1) The trend of the Chinese and English patents is similar. The number of pa-tents
grows slowly until 2015. From 2016 to 2021, the growth rate of the related
patents has significantly accelerated.
(2) The topic models are identified based on the LDA model, and the quality of the
constructed topics has been measured via three methods, including coherence
value, cosine similarity, and Kullback–Leibler (KL) divergence. The results of
the three methods are consistent.
(3) For the Chinese patents, the patents related to “drawings and models information
extraction” are gaining more attention. Besides, the results show that the number
of patents that focus on improving the user experience of the ARC system has
grown rapidly. Therefore, making the ARC system more convenient may be
the future research direction. For the English patents, the patents related to “
BIM-based management platform” are gaining more attention. While the patents
related to “Computer-executable rule and knowledge base “ received relatively
little attention currently. The rule interpretation is one of the most vital stages
in the rule checking process. Future research can pay more attention to the rule
interpretation to achieve a fully automated rule checking system.
24 Z. Zheng et al.

The limitations of this study are listed as follows:

(1) The performance of LDA analysis can be further improved for better topic
detection. Besides, more methods should be utilized to validate the results of LDA.
(2) Only the abstracts of the patents are used to perform the analysis, more
structured and unstructured data from patents is expected to use in the future.

Acknowledgement The authors are grateful for the financial support received from the National
Natural Science Foundation of China (No. 51908323, No. 72091512), the National Key R&D
Program (No. 2019YFE0112800), and the Tencent Foundation through the XPLORER PRIZE.

References

1. Nawari NO (2018) Building information modeling: automated code checking and compliance
processes. CRC Press, Boca Raton. https://doi.org/10.1201/9781351200998
2. Lee YC, Eastman CM, Lee JK (2015) Automated rule-based checking for the validation of
accessibility and visibility of a building information model. In: 2015 International Workshop
on Computing in Civil Engineering, pp. 572–579. https://doi.org/10.1061/9780784479247.071
3. Lee YC, Ghannad P, Dimyadi J, Lee JK, Solihin W, Zhang J (2020) A comparative analysis of
five rule-based model checking platforms. In: Construction Research Congress 2020: Computer
Applications, pp. 1127–1136. https://doi.org/10.1061/9780784482865.119
4. Guo D, Onstein E, La Rosa AD (2021) A semantic approach for automated rule compliance
checking in construction industry. IEEE Access 9:129648–129660
5. Eastman C, Lee JM, Jeong YS, Lee JK (2009) Automatic rule-based checking of building
designs. Autom Constr 18(8):1011–1033. https://doi.org/10.1016/j.autcon.2009.07.002
6. Lin TA, Fatt CT (2006) Building Smart-A strategy for implementing BIM solution in Singapore.
Synthesis Journal. Singapore, 117–124
7. Lee JM (2010) Automated checking of building requirements on circulation over a range of
design phases. Georgia Institute of Technology
8. Drogemuller R, Woodbury R, Crawford J (2000, June) Extracting representation from struc-
tured text: Initial steps. In: Proceedings of the 36th CIB W78 Conference on Information and
Communication Technologies, Reykjavik, Iceland, pp. 28–30
9. EDM ModelChecker: http://www.epmtech.jotne.com/index.php?id=512200
10. Nawari NO (2019) A generalized adaptive framework (GAF) for automating code compliance
checking. Buildings 9(4):86
11. Zhang J, El-Gohary NM (2015) Automated information transformation for automated regula-
tory compliance checking in construction. J Comput Civ Eng 29(4):B4015001. https://doi.org/
10.1061/(ASCE)CP.1943-5487.0000427
12. Zhou YC, Zheng Z, Lin JR, Lu XZ (2022) Integrating NLP and context-free grammar for
complex rule interpretation towards automated compliance checking. Comput Ind. https://doi.
org/10.1016/j.compind.2022.103746
13. Zheng Z, Lu XZ, Chen KY, Zhou YC, Lin JR (2022) Pretrained domain-specific language
model for natural language processing tasks in the AEC domain. Comput Ind. https://doi.org/
10.1016/j.compind.2022.103733
14. Zhou P, El-Gohary N (2021) Semantic information alignment of BIMs to computer-
interpretable regulations using ontologies and deep learning. Adv Eng Inform 48:101239.
https://doi.org/10.1016/j.aei.2020.101239
15. Ismail AS, Ali KN, Iahad NA (2017, July) A review on BIM-based automated code compliance
checking system. In: 2017 International Conference on Research and Innovation in Information
Systems (ICRIIS), pp. 1–6. IEEE
Text Mining-Based Patent Analysis for Automated Rule Checking in AEC 25

16. Fuchs S (2021) Natural language processing for building code interpretation: systematic
literature review report
17. McAleer M, Slottje D (2005) A new measure of innovation: the patent success ratio.
Scientometrics 63(3):421–429
18. Pan X, Zhong B, Wang X, Xiang R (2021) Text mining-based patent analysis of BIM application
in construction. J Civ Eng Manag 27(5):303–315
19. Blei DM, Ng AY, Jordan MI (2003) Latent dirichlet allocation. J Mach Learn Res 3(Jan):993–
1022
20. Wang Y, Xu W (2018) Leveraging deep learning with LDA-based text analytics to detect
automobile insurance fraud. Decis Support Syst 105:87–95
21. Zhong B, Pan X, Love PE, Sun J, Tao C (2020) Hazard analysis: a deep learning and text
mining framework for accident prevention. Adv Eng Inform 46:101152
22. Bastani K, Namavari H, Shaffer J (2019) Latent Dirichlet allocation (LDA) for topic modeling
of the CFPB consumer complaints. Expert Syst Appl 127:256–271
23. Lin JR, Hu ZZ, Li JL, Chen LM (2020) Understanding on-site inspection of construction
projects based on keyword extraction and topic modeling. IEEE Access 8:198503–198517
24. Röder M, Both A, Hinneburg A (2015, February) Exploring the space of topic coherence
measures. In: Proceedings of the eighth ACM international conference on Web search and data
mining, pp. 399–408
25. Otte E, Rousseau R (2002) Social network analysis: a powerful strategy, also for the information
sciences. J Inf Sci 28(6):441–453
26. Bastian M, Heymann S, Jacomy M (2009, March) Gephi: an open-source software for exploring
and manipulating networks. In: Proceedings of the international AAAI conference on web and
social media. vol. 3, no. 1, pp. 361–362
Unravelling the State of the Art
of Blockchain Development for Improved
Infrastructure Delivery in the Built
Environment: A Bibliometric Review
Motheo Meta Tjebane, Innocent Musonda, Adetayo Onososen,
and Molusiwa Ramabodu

Abstract Information management is critical to the smooth operation of infras-

tructure delivery. It is the process of gathering, storing, distributing, archiving, and
deleting/destroying data. Information and data management are critical to decision-
making toward achieving improved, resilient, and responsive infrastructure delivery.
While innovative tools abound for managing information in the built sector, barriers
such as technical challenges, expertise needed, and cost of procurement inhibit the
adoption of these technologies. In contrast, blockchain offers opportunities for acces-
sibility, lighter and faster transaction speed, and secured data processing. However,
research in this area is limited and slowly emerging. Therefore, this study adopts a
bibliometric review to avail current progress, critical discussions, and vital strategies
to advance the adoption of Blockchain development in the construction industry.
Blockchain has emerged as important to built environment innovation discourses
with energy management, payment processing, documentation, and asset lifecycle
management opportunities. This study is therefore important in improving extant
studies and availing future directions. This paper uses bibliometric analysis to
examine blockchain as a tool for infrastructure delivery in the built environment.
Empirical data from Scopus was analysed to establish the publication and citation
pattern, specifically publications published between 2011 and 2021. This also iden-
tifies research trends, common gaps, and best practices among researchers who use
Vosviewer. The most influential authors, institutions, journals, countries, and coop-
eration networks are then identified. This study presents the relevance of blockchain
in the built environment and its movement, adoption, and adaptation, which further
enables the spark of innovative ideas to realize and comprehend the current and future
technological transformation in infrastructure delivery.

M. M. Tjebane (B) · I. Musonda · A. Onososen · M. Ramabodu

Center of Applied Research and Innovation in the Built Environment (CARINBE), Faculty of
Engineering and the Built Environment, University of Johannesburg, Johannesburg, South Africa
e-mail: tjebane.motheo@mut.ac.za
M. M. Tjebane
Department of Construction Management and Quantity Surveying, Mangosuthu University of
Technology, Durban, South Africa

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 27

S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_3
28 M. M. Tjebane et al.

Keywords Built Environment · Blockchain · State of the art · Infrastructure ·

Resilient · Responsive

1 Background

The built environment faces numerous challenges and fragmentation, impacting the
infrastructure delivery rate. Infrastructure delivery processes in the built environ-
ment are highly fragmented, dispersed, and complex. As a result, massive efforts,
resources, and intermediaries are required to plan, monitor, and track these processes
under the schedule, cost, quality, and scope. Furthermore, delays and accidents
occur at multiple stages and varying degrees, affecting the project’s key objec-
tives [1]. Various management processes have been developed and integrated to
increase productivity and effectiveness while decreasing the likelihood of misalign-
ment [1]. However, most have seemed to fail over time as they cannot address all
issues simultaneously.
Although no technology or approach can address this, all the issues listed above,
blockchain technology and the circular economy, have revolutionized supply chain
management in several ways. Blockchain technology is a key component of digital
transformation, promising transparency, improved privacy, faster settlement, cost
savings, quality, and productivity through a distributed approach [2]. The use of
blockchain networks in such processes is part of that development. However, as the
fourth industrial revolution support market progresses, the built environment is exper-
imenting and developing innovative tools and technologies. One of the key issues
impeding industry modernization is the failure to adopt technological developments
that could have aided the industry’s achievements [3].
The blockchain is a digitized, decentralized public ledger of data, assets and
all pertinent transactions that have been executed and shared among participants
in the network. According to [4] blockchain technology allows multiple parties to
transact on a shared synchronous ledger. Each transaction is validated with a digital
signature to ensure its authenticity. Blockchain has several key features, including
decentralization, distributed computing, and consensus. While it is most associated
with digital cryptocurrencies such as Bitcoin, blockchain is an emergent technology.
Blockchain can revolutionize and transform the current digital operational landscapes
and business practices of finance, computing, government services, and virtually
every existing industry [5].
Previous studies have been conducted on blockchain in the built environment
[4]. conducted an exploratory literature review of blockchain in the construction
industry. [6] looked at blockchain-based security solutions with IoT applications
in the construction industry [7]. conducted a bibliometric review on the risks and
opportunities of blockchain in the built environment. Little study has focused on
blockchain in infrastructure delivery in the built environment.
Beginning in 2011, the literature on this topic consists of bibliometric research
articles or review articles. The bibliometric review aims to provide the reader with
general research directions, research groups, and gaps in the literature [7]. This article
Unravelling the State of the Art of Blockchain Development … 29

aims to provide a bibliometric overview of blockchain’s state of the art in improving

infrastructure delivery in the built environment. The article is as follows. Section 1
looks at the background and needs for the study. This is followed by a methodology
chapter, Sect. 2, which shows how the bibliometric was conducted. Section 3 provides
the results for the analysis. Section 4 is the conclusion.

2 Research Design and Method

To fulfill the research objectives, a quantitative bibliometric approach was used in

this study. a systematic approach was used in scoping and selecting publications.
The flowchart in Fig. 1 below shows the research method used.

Fig. 1 Research Method Flowchart

30 M. M. Tjebane et al.

The bibliometric analysis employs a hybrid technique to present or capture

research trends and features of publications by combining evaluative and descrip-
tive approaches [8, 9]. This study is the result of the current blockchain scenario for
infrastructure delivery, as it employs a systematic literature review with the assistance
of bibliometric network analysis to document the publication on the topic [10].
A thorough search is conducted on the Scopus database, which is a trustworthy
and unbiased source of reputable journal articles and fundamental requirement of
bibliometric analysis [11, 12].

2.1 Inclusion and Exclusion Criteria

Inclusion and exclusion was used to select publications for the bibliometric analysis.
The eligibility criteria address the inclusion and exclusion of studies for bibliometric
analysis [13]. In general, inclusion criteria form the basis for including articles that
provide the information required to answer the research question. While exclusion
criteria include the reason for excluding non-relevant articles, duplicate articles, and
articles with no full-text available, the inclusion criteria looked at the following:
• Publications made between 2011 and 2021
• Articles and review articles
• Published in the English language

2.2 Search Query

This study collects data from a Scopus dataset by searching the title, abstract, and
keyword search fields of articles published between 2011 and 2021. The query string
was blockchain AND built AND environment OR construction AND industry OR
infrastructure). three main segments were linked by the Boolean operator ‘AND’
each segment consisting of at least two fields linked by the ‘OR’ operator [14].

2.3 Data Analysis

VOSviewer is bibliometric network construction and visualization software. Jour-

nals, researchers, and individual publications can all be part of these networks
based on citation, bibliographic coupling, cogitation, or co-authorship relationships.
They also provide text-mining functionality for network construction, visualization
occurrences of terms extracted from scientific literature [15].
VOSviewer was chosen for this study to draw science mappings because it
has exceptional text-mining capabilities and is well-suited for dealing with larger
Unravelling the State of the Art of Blockchain Development … 31

300
285
250
220
200
150
100
82
50 14
0 0 0 0 0 0 0 1 4 5
0 0 0 0 0 0 0 0 0 10
2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
Year

Publications Citations

Fig. 2 Publication verses citation from 2011 to 2021

networks [16]. VOSviewer is currently used in construction industry research to

create science mappings for several topics.

3 Results

3.1 Descriptive Analysis

From 2011 to 2021, 28 documents on the role of Blockchain development for

improved infrastructure delivery in the built environment were published on Scopus.
The distribution of publications and citations by year. During the first seven years,
no publication was made concerning the subject matter. The first article was done in
2018, and the publication rate increased in 2019 to 4 articles. 2020 saw five articles
being published, with 2021 saw the highest number of publications with fourteen.
In terms of citations, by mid-2022, there was a citation of 285 for all twenty-eight
articles on the Scopus index (Fig. 2).

3.2 Country Analysis

The United Kingdom had the highest number of article publications; Australia and
South Africa followed this with four each. Singapore, Hong Kong, and China had
two publications each. Regarding citations, the United Kingdom had the highest with
377, Australia followed this with 120 citations, and United States with seventy-seven.
32 M. M. Tjebane et al.

The United Kingdom had the most publications and citations, while South Africa,
China, and Singapore had the most collaborations (Fig. 3 and Table 1).

Fig. 3 Country collaboration

Table 1 Country analysis

Country Documents Citations Total strength
United Kingdom 9 377 3
Australia 4 120 2
United states 1 77 0
Netherlands 1 24 1
Switzerland 1 24 1
Iran 1 22 0
Singapore 2 21 4
South Africa 4 20 4
Hong Kong 2 13 1
Italy 1 12 0
France 1 6 2
Germany 1 6 2
Japan 1 6 2
Sri Lanka 1 3 0
Belarus 1 2 3
China 2 2 4
Unravelling the State of the Art of Blockchain Development … 33

Table 2 Author analysis

Author Documents Citations Total Strength
Kassem M 2 179 3
Li J 2 179 3
Greenwood D 1 168 2
Nawari N.O 1 77 1
Ravindran S 1 77 1
Abrishami S 1 73 2
Elghaish F 1 73 2
Hosseini M.R 1 73 2
Edwards D 2 70 4
Parn E.A 1 70 1
Batty M 1 40 6
Engin Z 1 40 6
Lan T 1 40 6
Longley P.A 1 40 6
Penn A 1 40 6

3.3 Authorship

Table 2 shows the top authors in terms of the number of citations. From the list of
fifteen authors, two authors have 179 citations from publishing two articles. This
was followed by greenwood d. with 168 citations from the article. Furthermore,
three authors had ninety-nine citations from one published article. Newari N.O and
Ravindran S. had seventy-seven citations. Of the fifteen top authors, four authors had
the lowest number of citations, with forty. Figure 4 shows the lines of collaborations
among the different authors.

3.4 Journals

Table 3 shows the top seventeen journal articles in terms of article publications
relating to blockchain development for improved infrastructure delivery in the built
environment. The tables measure against the number of citations. Automation in
construction had the highest number of citations, with 262 from five published arti-
cles. It also has the most links with six, as shown in Fig. 5 This was followed by
the journal of building engineering with seventy-seven citations from one article.
However, from this list, three journals had no citations. Moreover, nine journals had
no links to other related journals.
34 M. M. Tjebane et al.

Fig. 4 Authorship collaboration network

Table 3 Journal source analysis

Journal Documents Citations Total strength
Automation in Construction 5 262 8
Journal of Building Engineering 1 77 0
Engineering, Construction, and Architectural 1 70 3
Management (ECAM)
Buildings 2 47 1
Sustainability (Switzerland) 3 43 0
Journal of Urban Management 1 40 1
IEEE Access 1 22 0
Journal of Cleaner Production 1 13 1
Frontiers in Energy Research 1 12 0
Future Internet 1 6 0
Internet of Things (Netherlands) 1 6 0
Journal of Facilities Management 1 3 0
IEEE Transactions on Computational Social Systems 1 2 0
Journal of Engineering, Design and Technology 1 1 1
African Journal of Science, Technology, Innovation and 1 0 3
Development
Building and Environment 1 0 2
Journal of Global Operations and Strategic Sourcing 1 0 0
Unravelling the State of the Art of Blockchain Development … 35

Fig. 5 Journal network

3.5 Clustering

A keyword frequency visualization graph of blockchain in infrastructure was gener-

ated by VOSviewer, as seen in Table 4 and Fig. 6. Among the most frequently
mentioned keywords were blockchain(o = 17, TlS = 97), smart contract(o = 8,
TlS = 66), architectural design (o = 7, TlS = 57), block-chain(o = 6,TlS = 48),
construction industry(o = 8,TlS = 48), built environment (o = 8, TlS = 47), infor-
mation management (o = 5, TlS = 41), smart contracts(o = 5,TlS = 34), building
information modelling(o = 4, TlS = 33) and project management (o = 5, TlS = 32).

4 Discussion

4.1 Cluster Analysis

Cluster one
This cluster consists of twelve keywords including circular economy, digitali-
sation, and industry 4.0, among others. Industry 4.0, which includes global stan-
dards, the digitisation of business processes, and cyber-physical systems that collab-
orate with supply chain and logistics networks, is fundamentally driven by Artificial
Intelligence (AI) [4]. [1] also noted AI, robotics, serverless computing, blockchain
networks, and the Internet of Things (IoT) are all considered to be components of this
trend. Many research efforts were focused on converting management procedures and
activities in the construction sector from manual processing to electronic processing
utilizing Information and Communication Technologies with the introduction of
Industry 4.0 and the expansion of automation (ICT).
Blockchain, smart contracts, BIM, IoT, blockchain-smart contracts, and
blockchain-information security are some of the current academic trends that are
attracting a lot of interest. These current study topics are implicitly influenced by
Industry 4.0 because IoT and blockchain-smart contracts have Industry 4.0 as a
common denominator [17].
36 M. M. Tjebane et al.

Table 4 Keyword cluster

Cluster one Cluster two Cluster three Cluster four Cluster five
Artificial Block-chain BIM architectural design/methodology/
Intelligence design approach
blockchain blockchain digital building information and
technology storage information communications technology
modelling
built construction distributed building risk assessment
environment industry ledger information
modelling
(BIM)
circular decentralisation future construction
economy applications sectors
construction digital twin hypeledger distributed
fabric ledger
technology
(DLT)
digital internet of things information project
technologies management management
digitalisation network network
architecture security
digitalisation smart city smart
contract
industry 4.0 smart contracts
procurement supply chains
processes
sustainability
sustainable
development

In the delivery of infrastructure, there is not much active research linking sustain-
ability with blockchain. Blockchain technology have received less attention in rela-
tion to sustainability initiatives including green procurement, sustainable procure-
ment, and circular economy. Furthermore, sustainability might be linked with current
blockchain technologies to develop a market for green construction materials [17].
The circular economy seeks to improve resource use throughout the whole construc-
tion process, from design through operation and destruction, using blockchain tech-
nology. It focuses on reducing waste by reusing materials, whether by removing
substances that would hinder reuse, recycling materials, or engaging in any other
activity that substitutes the idea of end-of-life [18].
Cluster two
This cluster consists of ten keywords including blockchain, construction industry,
decentralisation, and digital twin, among others. The blockchain’s transparency and
immutability make it an ideal platform for testing if smart contracts may be used to
automatically check the compliance of objects in digital models [4]. Smart contracts
Unravelling the State of the Art of Blockchain Development … 37

Fig. 6 Keyword cluster network

are among the most noteworthy features of blockchain for the construction industry.
They are self-executing pieces of code that conduct a contract’s provisions when
certain conditions are satisfied. As some components may still need human input
and control, smart contracts can alternatively be seen as regular contracts that can be
automated [19].
The prototype of an artefact that contains the data necessary to represent the
virtual model is referred to as its “digital twin.” This can be developed for opera-
tions, processes, and aircraft [20]. The concept of a digital twin is a broader phrase that
is evolving in tandem with the expanding deployment of IoT devices [21]. However,
there are still challenges of integrating blockchain with digital twin [22] Decentral-
ization, which distinguishes blockchain technology from the current, conventional,
centralized standard database system or platform, is its primary characteristic. Decen-
tralization guarantees that the contract’s terms are legitimate without the need for a
facilitator or central authority [3]. Additionally, [23] noted blockchain technology is
being integrated with other technologies, such as the Internet of Things (IoT) and Arti-
ficial Intelligence (AI), for use cases such as enabling IoT device interoperability in
a decentralized environment with fog computing acting as the orchestrating layer, on
a blockchain-based access control layer to the IoT data storage, and on decentralized
AI applications for enhanced data security and improved trust in AI.
Cluster three
This cluster consists of eight keywords including BIM, digital storage, distributed
ledger, and future applications, among others. Blockchain technology, offers a
distributed tamper-proof ledger with the agreement of many players to allay trust
issues. It facilitates reliable computing and processing automation for a variety of
applications and cross-organisational collaborations when used in conjunction with
smart contracts [24]. Although blockchain has already demonstrated effective uses
38 M. M. Tjebane et al.

in the financial sector, it is thought to be a disruptive technology that has the potential
to alter a wide range of industries, including supply chains and the healthcare sector.
Distributed ledger technology, or the use of Blockchain for the execution of
Smart contracts, may help to improve network integration among participants and
provide an appropriate means of monitoring the materials and waste cycle [25]. The
information stored on the Blockchain can be accessible at any moment to track the
proper management of building waste and its effects on the environment, while fully
transparently documenting all information or acts.
Implementing blockchain-based BIM is an administrative process that has evolved
significantly over time. This system uses blockchain technology to manage modelled
data on the BIM platform, keeping the data private while allowing access to all
stakeholders involved [3]. The Digital Twin is occasionally mistaken for BIM since
it could refer to a rich BIM produced by the designer prior to the construction phase
and utilized to create a building [20].
Cluster four
This cluster consists of six keywords including architectural design, construc-
tion sector and project management, among others. According to [4] blockchain is
currently used in smart contracts for effective project management. Project manage-
ment processes in the construction sector are extremely dispersed, fragmented, and
complex, particularly throughout the execution and monitoring phases and more
particularly while delivering infrastructure. To plan, monitor, and manage these
processes in accordance with schedule, cost, quality, and scope requires a great
deal of effort, resources, and intermediates [1].
Design an apartment layout using a blockchain-based decentralized application to
get consensus from participants, with voting rights based on reputation and tokens.
This is specific to architectural design. Additionally, blockchain helps automate archi-
tectural design using shape grammars to facilitate distribution, decision-making, and
collaboration [18].
Cluster five
This cluster consists of three keywords which are design/methodology/approach,
information and communications technology, and risk assessment. To design a cogent
approach to encourage blockchain adoption specifically in the construction industry,
it is critical to analyze the existing situation of blockchain in the built environment
and the construction sector [19]. Additionally, a methodology that offers a workable
and realistic resolution for its application to the delivery of infrastructure, particularly
in developing countries, needs to be developed [26].
The use of blockchain technology can significantly reduce administrative costs,
effectively safeguard intellectual property rights, remove onerous paperwork, manual
verifications, and contract execution. Evaluating and selling ideas and workflows
could potentially open a new source of income for design professionals [5].
Researchers in the ICT sector are creating a Blockchain platform to evaluate the
general effectiveness and performance of Blockchain-based project management
[27].
Unravelling the State of the Art of Blockchain Development … 39

5 Conclusion

Information management is critical to the smooth operation of infrastructure delivery.

It is the process of gathering, storing, distributing, archiving, and deleting/destroying
data. Information and data management are critical to decision-making toward
achieving improved, resilient, and responsive infrastructure delivery. While inno-
vative tools abound to managing information in the built sector, barriers such as
technical challenges, expertise needed, and cost of procurement inhibit the adoption
of these technologies. In contrast, blockchain offers opportunities for accessibility,
lighter and faster transaction speed, and secured data processing. However, research
in this area is limited and slowly emerging. Therefore, this study adopts a biblio-
metric review to avail current progress, critical discussions, and vital strategies to
advance the adoption of Blockchain development in the construction industry. This
study aimed to provide a state-of-the-art bibliometric review using Scopus data from
2011 to 2021.
The study’s results clearly show that the first publication was made in 2018, with
an increase of fourteen publications in 2021. Moreover, some of the top countries
leading in publications include the United Kingdom, Australia, and South Africa.
While the United Kingdom, Australia and the United States are leading in terms of
citations.
The leading journal in terms of the number of publications is automation in
construction, buildings, and sustainability (Switzerland). In terms of citations,
automation in construction is leading. This was followed by the journal of building
engineering and engineering, construction, and architectural management.
The most frequent words in the articles include blockchain, smart contract,
architectural design, blockchain, construction industry, built environment.
The bibliometric review of the selected articles revealed the progress of research
on Blockchain in improving infrastructure delivery. However, there were limits in
conducting the study. The study only looked at articles, and the data was sourced
from Scopus. Future studies should look at various kinds of scholarly publications
through various databases. The development of built infrastructure that scales well
and facilitates the interoperability of blockchain applications and services could be
a promising area of future research.

References

1. Adel K, Elhakeem A, Marzouk M (Sep. 2022) Chatbot for construction firms using scalable
blockchain network. Autom Constr 141. https://doi.org/10.1016/j.autcon.2022.104390.
2. Sadeghi M, Mahmoudi A, Deng X, Luo X (2022) Prioritizing requirements for implementing
blockchain technology in construction supply chain based on circular economy: Fuzzy Ordinal
Priority Approach. Int J Environ Sci Technol. https://doi.org/10.1007/s13762-022-04298-2
3. Akinradewo OI, Aigbavboa CO, Edwards DJ, Oke AE (2022) A principal component analysis of
barriers to the implementation of blockchain technology in the South African built environment.
40 M. M. Tjebane et al.

Journal of Engineering, Design and Technology 20(4):914–934. https://doi.org/10.1108/JEDT-

05-2021-0292. Jun.
4. Scott DJ, Broyd T, Ma L (Dec. 01, 2021) Exploratory literature review of blockchain in the
construction industry. Automation in Construction vol. 132. Elsevier B.V. https://doi.org/10.
1016/j.autcon.2021.103914
5. Nawari NO, Ravindran S (Sep. 01, 2019) Blockchain and the built environment: Potentials and
limitations. Journal of Building Engineering, vol. 25. Elsevier Ltd. https://doi.org/10.1016/j.
jobe.2019.100832
6. Singh P (2020) Blockchain based security solutions with IoT application in construction
industry. IOP Conf Ser Earth Environ Sci 614(1). https://doi.org/10.1088/1755-1315/614/1/
012052
7. Darabseh M, Martins JP (2020) Risks and opportunities for reforming construction with
blockchain: Bibliometric study. Civil Engineering Journal (Iran) 6(6):1204–1217. https://doi.
org/10.28991/cej-2020-03091541. Jun.
8. Umeokafor N, Umar T, Evangelinos K (2022) Bibliometric and scientometric analysis-based
review of construction safety and health research in developing countries from 1990 to 2021.
Saf Sci 156(August):10589. https://doi.org/10.1016/j.ssci.2022.105897
9. Tjebane MM, Musonda I, Okoro C, Onososen A (2023) Artificial Intelligence (AI) in sustain-
able construction management: a scientometric review. Lecture Notes in Civil Engineering
245:137–150. https://doi.org/10.1007/978-3-030-97748-1_12
10. Bhyan P, Shrivastava B, Kumar N (2022) Systematic literature review of life cycle sustainability
assessment system for residential buildings: using bibliometric analysis 2000–2020. Environ
Dev Sustain, no. 0123456789. https://doi.org/10.1007/s10668-022-02636-5
11. Arslan HM, Chengang Y, Komal B, Chen S (2022) Nexus between environmental disclosures
and top management team characteristics: a systematic review. Environmental Science and
Pollution Research, 0123456789. https://doi.org/10.1007/s11356-022-22615-7
12. Onososen AO, Musonda I, Ramabodu M (Aug. 01, 2022) Construction robotics and human–
robot teams research methods. Buildings. MDPI 12(8). https://doi.org/10.3390/buildings120
81192
13. Vong A, Rita P, António N (2021) Health-related crises in tourism destination management: a
systematic review. Sustainability (Switzerland) 13(24). https://doi.org/10.3390/su132413738
14. Marović I, Hanak T, Plaum S (2022) Performance management in civil engineering: a system-
-
atic literature review. Elektronički časopis gradevinskog fakulteta Osijek 13(24):47–58. https:/
/doi.org/10.13167/2022.24.5
15. Santana LS, Ferraz GAES, Teodoro AJ da S, Santana MS, Rossi G, Palchetti E (2021) Advances
in precision coffee growing research: a bibliometric review. Agronomy 11(8). https://doi.org/
10.3390/agronomy11081557
16. Tariq S, Hussein M, Wang RD, Zayed T (2021) Trends and developments of on-site crane layout
planning 1983–2020: bibliometric, scientometric and qualitative analyzes. Constr Innov. https:/
/doi.org/10.1108/CI-02-2021-0015
17. Yevu SK, Yu ATW, Darko A (Nov. 01, 2021) Digitalization of construction supply chain and
procurement in the built environment: Emerging technologies and opportunities for sustainable
processes. Journal of Cleaner Production 322. Elsevier Ltd. https://doi.org/10.1016/j.jclepro.
2021.129093
18. Li I, Kassem M (Dec. 01, 2021) Applications of distributed ledger technology (DLT) and
Blockchain-enabled smart contracts in construction. Automation in Construction, vol 132.
Elsevier B.V. https://doi.org/10.1016/j.autcon.2021.103955
19. Li I, Greenwood D, Kassem M (2019) Blockchain in the built environment and construc-
tion industry: a systematic review, conceptual models and practical use cases. Autom Constr
102:288–307. https://doi.org/10.1016/j.autcon.2019.02.005. Jun.
20. Sepasgozar SME (Apr. 01, 2021) Differentiating digital twin from digital shadow: elucidating
a paradigm shift to expedite a smart, sustainable built environment. Buildings vol. 11, no. 4.
MDPI AG. https://doi.org/10.3390/buildings11040151
Unravelling the State of the Art of Blockchain Development … 41

21. Engin Z et al (2020) Data-driven urban management: Mapping the landscape. Journal of Urban
Management 9(2):140–150. https://doi.org/10.1016/j.jum.2019.12.001. Jun.
22. Teisserenc B, Sepasgozar S (Dec. 2021) Adoption of blockchain technology through digital
twins in the construction industry 4.0: A PESTELS approach. Buildings 11(12). https://doi.
org/10.3390/buildings11120670
23. Yiu NCK (Apr. 2021) Decentralizing supply chain anti-counterfeiting and traceability systems
using blockchain technology. Future Internet 13(4). https://doi.org/10.3390/fi13040084
24. Xiao Z et al (2020) Blockchain and IoT for insurance: a case study and cyberinfrastructure
solution on fine-grained transportation insurance. IEEE Trans Comput Soc Syst 7(6):1409–
1422. https://doi.org/10.1109/TCSS.2020.3034106. Dec.
25. Pellegrini I, Campi S, Locatelli M, Pattini G, di Giuda GM, Tagliabue LC (Nov. 2020) Digital
transition and waste management in architecture, engineering, construction, and operations
industry. Front Energy Res 8. https://doi.org/10.3389/fenrg.2020.576462
26. Elghaish F, Abrishami S, Hosseini MR (Jun. 2020) Integrated project delivery with blockchain:
an automated financial system. Autom Constr 114. https://doi.org/10.1016/j.autcon.2020.
103182
27. Khalfan I, Azizi N, Haass O, Maqsood T, Ahmed I (May 2022) Blockchain technology:
potential applications for public sector E-procurement and project management. Sustainability
(Switzerland) 14(10). https://doi.org/10.3390/su14105791
Project Design, Construction, Planning,
and Management
Organisational Leadership as a Driver
for the Adoption of Digital Technologies
for Construction Project Delivery

Matthew Ikuabe , Clinton Aigbavboa, Samuel Adekunle,

Babatunde Ogunbayo, and Rotshidzwa Mugaga

Abstract The uptake of digital technologies for construction activities seek to

proffer solutions to some of the challenges plaguing the delivery of construction
projects. Organisational leadership players a vital role in the drive for the espousal
of digital technologies in construction management. This study evaluates the factors
affecting the pursuit of adopting digital technologies by construction organisational
leadership. The study utilised a quantitative technique by deploying questionnaire for
data collection using built environment professionals as the target respondents and
the study area was Gauteng province in South Africa. The methods of data analysis
used were Mean Item Score, Kruskal Wallis H-test, and SNK post hoc test. Findings
from the study presents influential factors to the adoption of digital technologies by
the leadership of construction organisations. The study contributes significantly to
the knowledge base on the call for digital transformation in the construction industry.

Keywords Construction · Digital Technologies · Drivers · Organisational

Leadership · Project Delivery

1 Introduction

Construction projects are characterised with dynamic series of planned and

unplanned tasks carried out to achieve goals that are often well established or envis-
aged ex ante [1]. With ever-increasing clients’ demands in delivering sophisticated
building designs, sustainable infrastructure, and projects within the ambits of stip-
ulated standards, the delivery of construction projects is hugely becoming more
challenging [2, 3]. These include health and safety challenges, delays in project
delivery, cost overrun, and underperforming projects amongst others [4–6]. Despite
the intense efforts to abate some of these challenges, not much has been attained.

M. Ikuabe (B) · C. Aigbavboa · S. Adekunle · B. Ogunbayo · R. Mugaga

CIDB Centre of Excellence, Faculty of Engineering and the Built Environment, University of
Johannesburg, Johannesburg, South Africa
e-mail: ikuabematthew@gmail.com

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 45

S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_4
46 M. Ikuabe et al.

Ballard [7] affirmed that some of these challenges are a result of the engaging process
peculiar to the construction industry which is quite different from other sectors due
to their monotonous system such as manufacturing.
Evidence has shown other industries such as retailing, banking, health, and manu-
facturing have embraced multifaceted approaches to the uptake of innovative tech-
nologies, while the construction industry is still trailing [8, 9]. The espousal of digital
technologies has been heralded as one that has the potential to yield process optimi-
sation, cost saving, reduction of task time, and enhanced service delivery [10, 11].
Hence, the propagation for the espousal of digital technologies is timely as it seeks to
proffer solutions to the aforementioned issues confronting the construction industry.
Moreover, Kamara et al., [12] opined that the construction industry is the right candi-
date for an upgrade considering the numerous challenges confronting it. However,
one significant feature in the drive for the adoption of digital technologies is the
role played by organisational leadership among construction organisations. Alade
and Windapo [13] noted that role of leadership of construction organisations is vital
in its pursuit of digital transformation. Furthermore, Markowitz [14] affirmed that
policy makers are significant determinants to propel the uptake of innovative tech-
nologies and its attendant benefits resulting. Based on this background, this study
seeks to explore the factors influencing the pursuit of adopting digital technologies by
construction organisational leadership, with a view to proffering enhancing measures
that can aid in propagating the uptake of innovative technologies for construction
project delivery.

2 Literature Review

The organizational structure is a system of roles, responsibilities, authority, and

communication links that are designed to assist an organization in accomplishing its
tasks and meeting its objectives [15]. Transformational leadership refers to adaptive
leadership behavior that has been shown to influence innovation, particularly the
tendency of organizations to innovate. The development of observable administra-
tive processes is referred to as organizational structure, whereas social interactions,
which are ethereal in nature, assist the coordination of organizational members [16].
The degree to which organization leaders completes tasks or achieves its objectives
is known as leadership effectiveness [17]. Hence, the hierarchal structure of organi-
sations must encourage the development of a business climate that strives to improve
the delivery mandates of the organisation and also push for the attainment of core
objectives set by the organisation.
According to Dall’omo [18], implementing new technologies comes at a high
cost. Therefore, the availability of the financial resource would play a vital role in
the choice of digital transformation by organisations. Also, the cost implication of
training the required personnel plays a determining role towards its uptake. Moreover,
cost of maintenance is a considerable influential factor in the deliberations of digital
transformations [19]. For construction organisations, the ability to properly define
Organisational Leadership as a Driver for the Adoption of Digital … 47

the potential benefits resulting from the uptake of a given innovative technology
would significantly influence the choice of its adoption [20]. Usually, this would
entail development strategies by engaging in analysis or assessment of its uptake.
Furthermore, organisations are undecided on the need to engage in research and
development. This is as a result of the uncertainty of the potential yields from such an
investment [21]. Investments in research and developments would potentially serve
as a driver for the espousal of innovative technologies by construction organisations.
Also, the return on investment serves as a considerable influence in the choice of
digitalisation by organisations [22, 23]. This is propelled by the need for organisations
to seek profitability in investments made.
Data security and protection is a considerable factor in the pursuit of innovative
technologies by organisational leadership of construction organisations. Qu et al.
[24] noted that construction organisations are particular about data protection and
cyber security in the pursuit of digital transformation. Moreover, the support of top
management of construction organisations is a significant determinant in pushing
the transformation pursuit through digitalisation. Ofori and Toor [25] stated that the
support of the top hierarchy of construction organisations would help expedite the
motive or agenda under focus. Moreover, the strategy employed by organisations is
influential to construction organisations’ digital transformation. Strategies employed
by organisational leadership is influential to the actualisation of digital transforma-
tion of the organisation [19]. The organisational culture attributed to construction
organisations plays a vital role in influencing the decision by organisational leader-
ship to embark on inculcating digital technologies for project delivery [26]. Also, the
strategies deployed by competitors serve as being influential towards determining the
need of digital transformation by construction organisations [27]. Since construction
is a competitive business, organisational leadership of construction organisations
would aim to keep tabs of competitors’ strategies in order to maintain a comparative
advantage for a favourable business environment.

3 Methodology

The study assessed the factors influencing organisational leadership of construc-

tion organisations to adopt digital technologies for construction projects. After the
review of extant literature, factors were identified which were subsequently presented
to construction professionals for rating based on their level of significance. This was
actualized with the use of a well-structured questionnaire which consisted of two
sections. The former focused on the demographic information of the respondents
while the latter was focused on the identified factors for rating by the respondents.
The target respondents of study which are construction professionals were architect,
quantity surveyors, construction managers, engineers and project managers. Conve-
nience and snowball sampling technique was adopted for the study, while the area
of study was Gauteng province of South Africa. A total of seventy-four responses
was retrieved and deemed appropriate for analysis. The methods of data analysis
48 M. Ikuabe et al.

employed for the study was mean item score, Kruskal-Wallis (K-W) H-test, and
Student Newman Kauls (SNK) post-hoc test. Mean item score was used in ranking
the identified factors as rated by the respondents, while Kruskal-Wallis H-test was
deployed to ascertain if there is a statistical difference in the opinion of the groups of
respondents based on their professional affiliation. While the SNK post-hoc test was
employed in differentiating the mean responses of the respondents based on their
professional designation.

4 Findings

A total of fourteen factors were identified after the review of extant literature and
thereafter presented to the respondents of the study for rating based on their signif-
icance. Table 1 presents the result of the ranking of the factors and also the result
of the analysis of K-W test conducted. The findings from the table indicates that the
most rated factors influencing construction organisational leadership in the adop-
tion of digital technologies are return on investment with a mean score of 4.34, top
management support having a mean score of 4.27, financial resource with mean
score of 4.19, organisational culture and definition of benefit with usage with mean
scores of 4.04 and 4.00 respectively. It is observed that all the factors have a mean
core above 3.00, thus implying their significance. Furthermore, the result from the
K-W test shows that there is no statistical significance difference in the opinions
of the group of professionals for thirteen of the identified factors. This is because
these thirteen factors all have a p-value above 0.05, thereby indicating that there is a
convergent view among the different professionals on these factors. However, there
is a discrepancy in views of the group of respondents on one of the factors (personnel
training) due to the derived p-value of the factor being less than 0.05. This might be as
a result of the different professional designations having different core professional
mandates in delivery of construction projects.
Table 2 portrays the result of the SNK post-hoc test (multiple comparison)
conducted for the study to showcase the mean difference of the group of respon-
dents. The findings of the test shows that there is a significant difference in the
factors influencing organisational leadership in adopting digital technologies among
two groups of the respondents. The first group is made up of quantity surveyors
and engineers with values of 2.7723 and 2.9148 respectively. While the other group
comprises of construction managers, project managers and architects with values of
3.1052, 3.4992, and 3.6782 respectively.
Organisational Leadership as a Driver for the Adoption of Digital … 49

Table 1. Factors influencing organisational leadership in adopting digital technologies

Factors X R K-W
X2 Sig
Return on investment 4.34 1 1.319 0.290
Top management support 4.27 2 5.836 0.411
Financial resource 4.19 3 3.227 0.537
Organisational culture 4.04 4 7.092 0.465
Definition of benefits with usage 4.00 5 6.295 0.071
Organisations’ objectives 3.95 6 4.003 0.173
Business environment 3.72 7 1.842 0.092
Market orientation 3.71 8 3.849 0.484
Organisation’s strategy 3.69 9 2.274 0.712
Personnel training 3.62 10 1.582 0.021*
Organisation’s corporate image 3.58 11 4.692 0.873
Competitive pressure 3.55 12 5.826 0.159
Research and development 3.42 13 8.379 0.585
Data security and privacy 3.37 14 6.662 0.231

Table 2. SNK Post Hoc Test

Groups N Subset for alpha = 0.05
1 2
Quantity Surveyors 14 2.7723
Engineers 14 2.9148
Construction Managers 14 3.1052
Project Managers 14 3.4992
Architects 14 3.6782
Sig 1.000 0.668
Means for groups in homogeneous subsets are displayed.
a. Uses Harmonic Mean Sample Size = 14.000.

5 Discussion of Findings

Main factors to organisational leadership of construction organisations to the adop-

tion of digital technologies was assessed in this study. Findings from the analysis
conducted shows that the most influential factor is return on investment. This finding
is corroborated by Oh et al., [22] who noted that the return on investment serves as
a considerable influence in the choice of digitalisation by organisations. Moreover,
organisations are propelled to commit funds for digital technologies when there is a
clear mandate that the investments would yield financial returns [23]. Generally, the
driving idea of engaging in digital transformation by organisational leadership is the
50 M. Ikuabe et al.

assurance of attaining financial advantage. Furthermore, the study reveals that top
management support is a significant factor in the quest for digitalisation by construc-
tion organisations. Ofori and Toor [25] stated that the support of the top hierarchy
of construction organisations is crucial to important decisions of construction organ-
isations. Hence, the support of top management of construction organisations is a
significant determinant in pushing the transformation pursuit through digitalisation.
Also, the financial resource at the disposal of the organisation is a considerable
determinant in the choice of digitalisation by organisational leadership. According
to Zhang et al. [21], implementing new technologies comes at a high cost. There-
fore, the availability of the financial resource would play a vital role in the choice
of digital transformation by organisations. Additionally, it is revealed that all the
group of professionals making up the respondents for the study all have a convergent
opinion on all the identified factors for the study except for personnel training. This
might be as a result of the different professional designations having different core
professional mandates in delivery of construction projects, thereby having different
obligations with the digitalisation process.

6 Conclusion

The study assessed the factors influencing the pursuit of the adoption of digital tech-
nologies by construction organisational leadership. Subsequent to the review of liter-
ature, the study identified fourteen factors which were presented to the respondents of
the study for rating with the use of a well-structured questionnaire. Retrieved data was
subjected to statistical analysis which revealed the most significant factors as return
on investment, top management support, and financial resource. Also, it was shown
that there was a divergent view among the group of respondents in only one of the
factors, which is personnel training; while other factors saw a convergent view among
the group of professionals. Based on the findings of the study, it is recommended
that organisational leadership of construction organisations should encourage digital
transformation of construction organisations since it guarantees effective construc-
tion processes, efficient service delivery and optimisation of resources is guaran-
teed. Consequently, yielding to a potential return on investment in the digitalisation
process. Also, the drive for the uptake of digital technologies should be one that
should get the support of top management of construction organisation since they
play a vital role in the management of organisations.

References

1. Hetemi E, Ordieres-Meré J, Nuur C (2020) An institutional approach to digitalization in

sustainability-oriented infrastructure projects: the limits of the building information model.
Sustainability 12(9):3893
Organisational Leadership as a Driver for the Adoption of Digital … 51

2. Ikuabe MO, Aghimien D, Aigbavboa C, Oke A, Thwala W (2021) Site accidents in the South
African construction industry: cleaning the Augean stables. In: Kantola J, Nazir S, Salminen
V (eds) Advances in human factors, business management and leadership. Proceedings of the
AHFE 2021 virtual conference, 25–29 Jul, pp 92–98
3. Simmons DR, McCall C, Clegorne NA (2020) Leadership competencies for construc-
tion professionals as identified by construction industry executives. J Constr Eng Manag
146(9):04020109
4. Aghimien DO, Aigbavboa CO, Meno T, Ikuabe MO (2021) Unravelling the risks of construction
digitalisation in developing countries. Constr Innov 21(3):456–475
5. Ikuabe MO, Aghimien DO, Aigbavboa CO, Oke AE (2020) Exploring the adoption of digital
technology at the different phases of construction projects in South Africa. In: International
conference on industrial engineering and operations management, Dubai, UAE, 10–12 Mar,
pp 1553–1561
6. Mustapha Z, Aigbavboa C, Thwala W (2016) Small and medium-sized enterprises contractors’
health and safety performance. In: International conference of socio-economic researchers,
Johannesburg, South Africa
7. Ballard G (2012) Should project budgets be based on worth or cost? In: Proceedings, 20th annual
conference of the International Group for Lean Construction, San Diego, CA, pp 761–770
8. Aghimien D, Aigbavboa C, Oke A (2018) Digitalisation for effective construction project
delivery in South Africa. In: Proceedings of the contemporary construction conference:
innovative and dynamic built environment, Coventry, UK, 5–6 Jul 2018
9. Ikuabe MO, Aigbavboa CO, Oke AE (2020) Cyber-physical systems: matching up its appli-
cation in the construction industry and other selected industries. In: International conference
on industrial engineering and operations management, Dubai, UAE, 10–12 Mar 2020, pp
1543–1542
10. Ikuabe MO, Aigbavboa CO, Anumba C, Oke AE, Aghimien LM (2022) Confirmatory factor
analysis of performance measurement indicators determining the uptake of CPS for facilities
management. Buildings 12(4):466
11. Sutton M (2018) Digitization of construction sector worth $1.7 trillion – ITP.net. 2018 [online].
ITP.net. Available at: http://www.itp.net/617479-digitization-of-construction-sector-worth-17-
trillion. Accessed 11 Mar 2021
12. Kamara M, Anumba C, Evbuomwan O (2000) Process Model for Client requirements
processing in construction. Bus Process Manag J 6(3):251–279
13. Alade K, Windapo A (2020) Developing effective 4IR leadership framework for construction
organisations. Eng Constr Archit Manag 28(5):1337–1396
14. Markowitz C (2019) Harnessing the 4IR in SADC: roles for policymakers, Occasional paper
303, South African Institute of International Affairs
15. Agbim KC (2013) The impact of organizational structure and leadership styles on innovation.
IOSR J Bus Manag 6(6):56–63
16. Ardi A, Djati SP, Bernarto I, Sudibjo N, Yulianeu A, Nanda HA, Nanda KA (2020) The
relationship between digital transformational leadership styles and knowledge-based empow-
ering interaction for increasing organisational innovativeness. Int J Innov Creat Change
11(3):259–277
17. Amirul SR, Daud HN (2012) A study on the relationship between leadership styles and
leadership effectiveness in Malaysian G.L.C.s. Eur J Bus Manag 4:193–201
18. Dall’omo S (2017) Driving African development through smarter technology. African
digitalization maturity report, pp 1–45
19. Demirkesen S, Tezel A (2021) Investigating major challenges for industry 4.0 adoption among
construction companies. Eng Constr Architect Manag, ahead of print. https://doi.org/10.1108/
ECAM-12-2020-1059
20. Ram J, Afridi NK, Khan KA (2019) Adoption of Big Data analytics in construction: develop-
ment of a conceptual model. Built Environ Proj Asset Manag 9(4):564–579. https://doi.org/10.
1108/BEPAM-05-2018-0077
52 M. Ikuabe et al.

21. Zhang L, Cao T, Wang Y (2018) The mediation role of leadership styles in integrated project
collaboration: an emotional intelligence perspective. J Project Manag 36:317–330
22. Oh JC, Yoon SJ, Chung N (2014) The role of technology readiness in consumers’ adoption of
mobile internet services between south Korea and China. Int J Mobile Commun 12(3):229–248
23. Hernandez-Ortega B, Serrano-Cinca C, Gomez-Meneses F (2014) The firm’s continuance
intentions to use inter-organisational CTs: the influence of contingency factors and perception.
Inf Manag 51(6):747–761
24. Qu T, Coco J, Rönnäng M, Sun W (2014) Challenges and trends of implementation of 3D point
cloud technologies in building information modeling (BIM): case studies. In: 2014 international
conference on computing in civil and building engineering, Orlando, Florida, 23–25 Jun 2014
25. Ofori G, Toor S (2012) Leadership and construction industry development in developing
countries. J Constr Dev Countries 1:1–21
26. Oliveira T, Thomas M, Espadanal M (2014) Assessing the determinants of cloud computing
adoption: an analysis of the manufacturing and services sectors. Inf Manag 51(5):497–510
27. Gangwar H, Date H, Raoot AD (2014) Review on it adoption: insights from recent technologies.
J Enterp Inf Manag 27(4):488–502
Modeling Drywall Construction Process
Using a Spatiotemporal Chronographical
Scheduling

Adel Francis and Edmond Miresco

Abstract The Gantt/Predictive Diagram is the most widely used model for plan-
ning building construction projects. This diagram focuses on defining activities and
constraints and then allocating resources to those activities. Space constraints are
generally neglected. This makes it difficult to represent the sequence of work, traffic,
and supply flow between different sites. Spatiotemporal models are considered more
appropriate as scheduling models. These models aim to linearize activities, resources,
and spaces at the same time. The objective of this research is to model drywall
construction process using a spatiotemporal Chronographical scheduling. Partition
drywall assembly involves many components. Each of these components has a wide
variety of types, sizes, or other technical specifications. Space planning models the
sequences of operations on the construction site by subdividing the construction site
into areas to define work zones. Although this type of division is well suited for
most construction stages of implementation, such as creation of spaces, systems,
finishing, closing of spaces, and outdoor facilities; it is less suited, for the envelope
and the division of spaces stages of implementation. Spatial divisions of the site must
therefore be adapted to this reality. The novelty of this publication is that it proposes
a new way of dividing spaces, and it demonstrates this through a practical example.

Keywords First Keyword · Second Keyword · Third Keyword · Forth Keyword ·

Sixth Keyword

1 Introduction

Traditional planning methods based on the Gantt/Precedence diagram focus on

defining activities and constraints, and then allocating resources to these activities.
Space constraints are usually neglected. Neglecting space may result in a congested

A. Francis (B) · E. Miresco

École de Technologie Supérieure, University of Quebec, 1100 Notre-Dame Street West, Montreal,
QC H3C 1K3, Canada
e-mail: adel.francis@etsmtl.ca

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 53

S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_5
54 A. Francis and E. Miresco

or relaxed site, negatively affecting project duration. In addition, allowing traffic to

occur haphazardly, without forethought or control, can produce traffic congestion
[5]. Resource management, without consideration of work location, is inefficient.
Resources cannot be used, even if available, beyond the capacity of the workplaces.
The planner must also ensure proper rotation of teams across spaces.
Over the years, many space-based scheduling methods have been developed. Most
of them rely on purely algorithmic solutions. These solutions define the model param-
eters considered important, but neglect others considered less important. Neglecting
parameters in order to do complex modeling could harm the results.
This paper supports the idea that graphical solutions are more suitable for
modeling complex construction sites. Graphical modeling acts as a decision support
system that guides planners and managers to make better decisions. In the past, some
researchers have selected graphical modeling for space-based planning. Winch and
North [11] defined spaces by their occupancy status. Riley and Sanvido [7] focus
on the division of work spaces and the sequencing of activities, Tommelein et al.
[10] show constrained spatial planning for multi-story buildings, Akinci and Fischer
[1] capture the spatial requirements of different trades, Francis [2–6] introduces the
spatio-temporal concepts of the Chronographical modeling to represent worksite
operations with various compatible graphical and tabular approaches, Winch and
North [11] developed the Critical Spatial Analysis (CSA) to link the spatial and
temporal aspects of a worksite, Seppänen et al. [8] use a location-based management
system (LBMS) to plan and control production.
The objective of this research is to model drywall construction process using a
spatiotemporal Chronographical scheduling. Space planning models the sequences
of operations on the construction site by subdividing the construction site into areas to
define work zones. Although this type of division is well suited for most construction
stages of implementation, such as creation of spaces, systems, finishing, closing of
spaces, and outdoor facilities; it is less suited, for the envelope and the division
of spaces stages of implementation. Spatial divisions of the site must therefore be
adapted to this reality. The novelty of this publication is that it proposes a new
way of dividing spaces into linear strips according to the different specialties of the
subcontractors, and it demonstrates this through a practical example.

2 Chronographical Modeling

To meet the needs of planners and address the shortcomings of current methods,
Francis [2] developed the Chronographical modelling. A graphical modeling for
scheduling the construction operations on the project site. This modelling considers
simultaneously the different elements of construction, namely, work, crews, mate-
rials, and spaces. Various compatible graphical modelling approaches, using graph-
ical means of distinction, association, scale and attributes, have been proposed. Each
approach demonstrates the necessary information to meet a certain need. The planner
Modeling Drywall Construction Process Using a Spatiotemporal … 55

can switch between these approaches by manipulating the different graphical param-
eters [3, 4]. The proposed tabular and graphical approaches allow grouping the
scheduling information according to the specialties of the subcontractors, the works
or the spaces. By modeling spaces, it is also easier to plan traffic and intermediate
stocks.
The Chronographical modeling can also presents different spatiotemporal
approaches in order to allow the modeling of site spaces in addition to activities
and resources. These modeling could allow:
i) to model the site spaces (floors and zones) as the background of the schedule and
demonstrates construction operations on the foreground (Francis [5]). Spatiotem-
poral approaches combine three linearities: resources, spaces and activities. These
approaches prioritize the critical space in the critical path of activities by defining
the optimal site occupancy rates. Graphical optimization as described in Francis
[5] became easier and more realistic. By combining graphical, procedural and
algorithmic aspects, planner could have better solution for building construction
sites without a need of complex calculations;
ii) to conversely represent the planning of construction operations on the project
site plan [6]. This second model is named a site-spatial-temporal scheduling. The
scheduling modeling process for this modeling define for each floor, the layers
that demonstrate the construction steps (creation of spaces, systems, divisions,
finishes and closing of zones) and divide each layer into zones that ensure a
Takt production. For each specialty, the model defines the sequence of work
by linking the activities of the specialty between the different areas, respecting
the predefined working direction. Figure 1, demonstrates an applied example
for the site-spatial-temporal scheduling. This figure shows at the same time, the
graphical site-spatial-temporal (in the upper part) and two tabular approaches
(in the lower part). The first is grouped by site spaces (Zones) and the second is
grouped by specialties (Teams).

3 The Design Model for Scheduling Drywalls Construction

The systematic graphical design model is used to create visual models for scheduling
drywalls construction. It involves breaking down a problem into its components and
analyzing each component using graphic elements to represent the structure of the
system. This technique allows planners to plan crew rotation on the site and helps
stakeholders to clearly understand how different parts interact with each other and
also encourages collaboration, making it easier for teams to make decisions in a
timely manner and reducing conflict on the site.
The design model is based on the process developed by Steenkamp and McCord
[9]. This process specifies the following steps:
1. The context of use: In previous publications, the Chronographic modeling used
areas to define work zones. Although this type of division is well suited for
56 A. Francis and E. Miresco

#### 14
#### 15
30% 16
30% 17
30% 18
30% 19
30% 20
70% 21
40% 22
40% 23
40% 24
70% 25
55% 26
55% 27
55% 28
55% 29
55% 30
45% 31
20% 32
20% 33
20% 34
55% 35
35% 36
35% 37
42% 38
49% 39
76% 40
60% 41
67% 42
74% 43
81% 44
93% 45
#### 46
#### 47
#### 48
#### 49
#### 50
#### 51
#### 1
#### 2

Structure ErectionVent. system V1 V3 D3 F11 Elevator

, ,

S1 Vent. system V2 D1 G1 D1 P1 C1 F1
F1
V2
10 Wall Stud 1 G2 D2 P2 C2 F2
F2
8
FSF-IA (days)

6 Pipes & wiring 1 F3 G3 D3 P3 C3 F3

4
2 Wall Stud 2 G4 D4 P4 C4 F4
F4
0
1 2 3
D2 Pipes & wiring 2 G5 D5 P5 C5 F5
Initial activities F5

V3 Vent. system V3 Wall Stud 3 G6 D6 P6 C6 F6

F6

Legend D3 Pipes & wiring 3 G7 D7 P7 C7 F7

F7
Technical relation Takt activities
Ressource relation G Gyprock & joints Wall Stud 4 G8 D8 P8 C8 F8
Takt Relation D Doors
Free Start Float for initial activities (FSF-IA) P Painting D4 Pipes & wiring 4 G9 D9 P9 C9 F9
F9
Earliest date of releasing of a layer C False ceiling

D1, 2, 3, 4
Zone ending depending on the start of F10 G10 P10 C10
F Flooring
Initial activity on the zone , ,

Fig. 1. Applied example Spatiotemporal Scheduling Optimization [5]

most construction stages of implementation, such as creation of spaces, systems,

finishing, closing of spaces, and outdoor facilities; it is less suited, for the envelope
and the division of spaces stages of implementation.
2. The need for new artifacts: For the envelope and the division of spaces stages of
implementation, the advancement of the teams is done linearly. Spatial divisions
of the site must therefore be adapted to this reality. The construction schedules
must also be compatible with these types of operations. The developed artifacts
show the dynamic evolution of the occupation of these types of works.
3. The structural specifications and the prototype: The conceptualization structure
and the prototype of these artifacts are based on the Chronographical approach.
The next section will explain the elements, and structures on a prototype based
on a practical example.
4. The implementation and the evaluation: The applied example presented will
demonstrate the implementation process. More complex examples have to done
in the future on real projects in order to validate the proposed design.

4 Modeling Drywall Construction Process

Partition drywall assembly involves many components, including gyprock, metal

(or woods) studs, insulation, electrical wiring, and telecommunication wiring. After
laying the drywall, the gyprock needs grouting and then a finish coat, such as painting,
wallpaper or ceramic. For the envelope, in addition to these components may be added
Modeling Drywall Construction Process Using a Spatiotemporal … 57

several exterior finishing works, including, the vapor barrier, sheathing (or bracing),
air barrier, battens, and sidings. Sidings also come in many types, including masonry,
glass curtain walls, aluminum, vinyl, wood, fiberboard, or fiber cement. Each of these
components has a wide variety of types, sizes, or other technical specifications.
While it is optimal to create typical partitions on the architectural drawings, it is
not the same for construction site operations. For the design phase, it is simpler and
more efficient to do groupings, which will lighten the plans and estimates of grades
C and D. Note that modelling software such as Revit also promotes this technique by
creating product families. For the construction phase, each component is carried out
by a different subcontractor. Construction contracts are based on materials (wood,
plastic, metal, etc.) and not on construction elements (walls, floors, doors, windows,
etc.). For example, wood, steel, and aluminum windows are not manufactured by
the same supplier. On the other hand, a supplier, who works with steel, will be able
to manufacture doors, windows, cabinets, and even stairs in steel. The same logic is
applied for wood, vinyl, and ceramic floors, for which different subcontractors will
be hired to do the installation on site. So, while groupings by type are excellent for
the design phase, they fail for site work planning.
The Chronographical method graphically models the sequences of operations on
the construction site by subdividing the construction site into zones. Areas over which
crews move to perform work in different emplacements of the site. To do this, the
modeling creates a dynamic hierarchy of spaces, a Location Breakdown Structure
(LBS). The LBS is structured into project components (different buildings); vertical
levels (floors); and horizontal spaces (zones). The LBS identifies, for the zones, the
seven (7) major successive stages in the implementation of the construction phase of a
building project, namely the creation of spaces (e.g., addition of new floors); systems
(e.g., ventilation ducts); envelope (e.g., exterior walls and roof); division of spaces
(e.g., partitions); finishing (e.g., painting); closing of spaces (e.g., carpeting); and
outdoor facilities and exterior landscaping. The Chronographical method graphically
models these stages by creating seven levels of graphic layers. Each is subdivided
into zones according to the stage of construction, creating a dynamic hierarchy of
spaces.
As mentioned before, for the envelope and the division of spaces, work is done
linearly. Spatial divisions of the site must therefore be adapted to this reality. The
novelty of this publication is that it proposes a new way of dividing spaces, and
it demonstrates this through a practical example. To date, the site decomposition
systems offer work surfaces, which are appropriate for many specialties, such as
floor installation. However, for linear work, such as partition installation, the site
breakdown into areas is an approximation that can hardly be used to establish a good
crew rotation. Thus, the new way of dividing the spaces into linear strips according
to the different specialties of the subcontractors, represents a good solution. The
principal strength of this method is related to its modelling ability that facilitates
communication of information between the site stakeholders on the construction
site. The dynamic representation of the occupancy of the site using adapted artifacts
will improve the visualization of the schedule, ensure a good balance of the use of
spaces throughout the project, and will act as a decision-making support system.
58 A. Francis and E. Miresco

5 Applied Example for Scheduling Drywalls Construction

The following example (Figs. 2, 3 and 4) shows the preparation work for drywall
assembly for a three-story building. The work includes metal studs, drywall
installation (Gyprock), insulation, electrical wiring and telecommunication wiring.
Grouting, and painting are not included in this example.
The graphical visual model designs the concept of artifacts that express the objects
on the two-dimensional plan. These artifacts demonstrate the modeling of walls
components and their sequences to facilitate team rotation, but not as an execution

0 4 8 12 16 4 8 12 16 20 12 16 20 24 8 12 16 20 24 28 Days

3
2 (10%) 3 (10%)
0 4 8 12 16

2 4 7 4 (10%)
1 (10%) 9 (20%) 5 (11%)
7 (12%) 8 (5%) 6 (12%)
24 20 16

1
Days 32 28 24 20 16 32 28 24 20 16 28 24 20 16 12

W1 W2 W3 W4 W5 W6

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
30%

30%

30%

30%

30%

40%

40%

40%

40%

40%

40%

40%

51%

61%

61%

61%

43%

45%

45%

45%

27%

27%

27%

22%

22%

22%

22%

22%

12%

12%

12%

Zone 1 Team 1 Team 3 Team

eam 6
Zone 2 Team 1 Team 3 Team
m5 Team 6
Zone 3 Team 3 Team 6
Zone 4 Team 1 Team 3
Te Team 6
Zone 5 Team
m1 Team 5 Team 6
Zone 6 Team 3 Team 5 Team 6
Zone 7 Team 1 Team 5
Zone 8 Team 3
Zone 9 Team 2 Team 4 Team 5 Team
m7
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

Team 1 Zone 1 Zone 2 Zone 4 Zone

ne 5 Zone 7

Team 2 Zone 9

Team 3 Zone 1 Zone 2 Zone 3 Zone

Z 4 Zone 6 Zone 8

Team 4 Zone 9

Team 5 Zone 9 ne 2
Zone Zone 5 Zone 6 Zone 7

Team 6 Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 Zone 6

Team 7 ne 9
Zone

Fig. 2. Applied example for the site-spatial–temporal scheduling [6]

Modeling Drywall Construction Process Using a Spatiotemporal … 59

weeks 1 2 3 4 5 6 7 8 9
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45
Floor 1
Gyprock Gyprock
Metal studs Metal studs
Insulation Insulation
Electrical wiring Electrical wiring
Telecommunication wiring Telecom
Floor 2
Gyprock Gyprock
Metal studs Metal studs
Insulation Insulation
Electrical wiring Electrical wiring
Telecommunication wiring Telecom
Floor 3
Gyprock Gyprock
Metal studs Metal studs
Insulation Insulation
Electrical wiring Electrical wiring
Telecommunication wiring Telecom

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45
Gyprock
Floor 1 Gyprock
Floor 2 Gyprock
Floor 3 Gyprock
Metal studs
Floor 1 Metal studs
Floor 2 Metal studs
Floor 3 Metal studs
Insulation
Floor 1 Insulation
Floor 2 Insulation
Floor 3 Insulation
Electrical wiring
Floor 1 Electrical wiring
Floor 2 Electrical wiring
Floor 3 Electrical wiring
Telecommunication wiring
Floor 1 Telecom
Floor 2 Telecom
Telecom
Floor 3

Fig. 3. Project planning grouped by floors and by specialties

Floor 1

1 2 3 4 1 2 3 4 1 2 3 4
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Floor 2

3 4 5 6 3 4 5 6 3 4 5 6
13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

Floor 3

6 7 8 9 6 7 8 9 6 7 8 9
26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

Fig. 4. Project planning and grouped by floors and by specialties.

Fig. 4. Project planning using site-spatial-temporal modeling

60 A. Francis and E. Miresco

weeks 1 2 3 4 5 6 7 8 9
Specialities days 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45
F01
Gyprock F02
F03
F01 F02
Metal studs F03
F01 F02
Insulation F03
F01
Electrical wiring F02
F03
F01
Telecommunication wiring F02 F03

weeks 1 2 3 4 5 6 7 8 9
Floors 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45
Gyprock
Metal studs
Floor 1 Insulation
Electrical wiring
Telecom
Gyprock
Metal studs
Floor 2 Insulation
Electrical wiring
Telecom
Gyprock
Metal studs
Floor 3 Insulation
Electrical wiring
Telecom

Fig. 5. Project planning and grouped by specialities or by floors

process. For the wall construction process, the model allows the planner to formulate
his execution strategies according to the site architecture. This strategy is usually
coordinated with the site workers in a Last Planner planning logic.
Figure 2 shows the planning of this project in two forms: i) The upper part shows
the specialties grouped by floor. This facilitates the coordination of the work on each
floor; and ii) the lower part shows the floors grouped by specialty. This shows the
sequence of each specialty on the different floors.
Figure 3 shows the same planning of the same project. The only difference with
Fig. 2 is that each floor is divided into two zones shown by the blue and pink colors.
Figure 4 shows this planning on the floors. It is a 2D+1 planning (a simulation
similar to the 4D planning, but with a 2D plan + time). In this figure, the evolution
of the construction is demonstrated, on each floor, on a weekly basis. Thus, the first
section of the figure (top left) shows the progress of the first floor during the first
week (days 1 to 5), while the section of the figure (bottom center) shows the progress
of the third floor during weeks 6 and 7 of the project (days 26 to 35).
Figure 5 shows the same groupings as Fig. 2, namely, specialties grouped by
floor, and floors grouped by specialty. The difference, as mentioned, is related to the
distinction between the work progress in each zone.

6 Conclusion

In conclusion, spatiotemporal models are considered more appropriate to model the

construction of building projects. These models model the sequences of operations
on the construction site by subdividing the site into zones to define the work areas.
Although this type of division is well suited to most stages of construction imple-
mentation, it is less suitable for the envelope and space division. The process of
building the envelope and drywall is done linearly. The spatial divisions of the site
Modeling Drywall Construction Process Using a Spatiotemporal … 61

must therefore be adapted to this reality. This paper develops artifacts that demon-
strate the linear and dynamic evolution of the occupation of these types of structures.
The use of these artifacts improves the visualization of the schedule and ensures a
good balance in the use of spaces. The novelty of this publication is that it offers a
new way of dividing spaces, and it demonstrates it through a practical example. Its
main asset is linked to its modeling capacity, which facilitates the communication of
information between the actors of the construction site on the construction site.

References

1. Akinci B, Fischer M (2000) An automated approach for accounting for spaces required by
construction activities. In: Construction Congress, vol IV. ASCE, Orlando, FL
2. Francis A (2004) The Chronographical method for construction project planning. Ph.D. Thesis
(20) École de technologie supérieure, Montreal, University of Quebec. http://espace.etsmtl.ca/
692/ (in French)
3. Francis A (2013) The chronographical approach for construction project modelling. In:
Proceedings of the Institution of Civil Engineers – Management, Procurement and Law 166(4).
ICE Publishing, pp 188–204
4. Francis A (2016) A chronographic protocol for modelling construction projects. Proc Instit
Civil Eng Manag Procure Law 169(4):168–177
5. Francis A (2019) Chronographical spatiotemporal scheduling optimization for building
projects. Front Built Environ 5:36. https://doi.org/10.3389/fbuil.2019.00036
6. Francis A (2020) Chronographical site-spatial-temporal modeling of construction operations.
Front Built Environ 6:67. https://doi.org/10.3389/fbuil.2020.00067
7. Riley DR, Sanvido VE (1997) Space planning method for multistory building construction. J
Constr Eng Manag 123(2):171–180
8. Seppänen O, Evinger J, Mouflard C (2014) Effects of the location-based management system
on production rates and productivity. Constr Manag Econ 32(6):608–624. https://doi.org/10.
1080/01446193.2013.853881
9. Steenkamp AL, McCord SA (2007) Approach to teaching research methodology for informa-
tion technology. J Inf Syst Educ 18(2):255–266
10. Tommelein ID, Levitt RE, Hayes-Roth B (1992) SightPlan model for site layout. J Constr Eng
Manag 118(4):749–766
11. Winch GM, North S (2006) Critical space analysis. J Constr Eng Manag 132(5):473–481.
https://doi.org/10.1061/(ASCE)0733-9364(2006)132:5(473)
Towards Increased Situational Awareness
at Unstructured Work Zones: Analysis
of Worker Behavioral Data Captured
in VR-Based Micro Traffic Simulations

Julia Qin , Daniel Lu , and Semiha Ergan

Abstract In 2020, the Federal Highway Administration (FHWA) reported 857 fatal-
ities due to crashes at roadway work zones, 117 of which were construction workers.
Current practices to protect the personal safety of workers, such as stationary sound
and light-based work zone intrusion alarms, are frequently disregarded since their
alarm characteristics (e.g., volume, duration, frequency) are not well suited to work
zone environments. Such alarm devices are also rarely deployed at unstructured (i.e.,
mobile and short term) work zones. There is a need to identify optimal alarm charac-
teristics that can effectively notify workers of incoming hazards at unstructured work
zones without alarm fatigue. This paper is part of a larger research vision, where an
integrated platform of virtual reality (VR), micro traffic simulations, and wearable
sensors is used to capture workers’ physiological and behavioral response to alarms.
This platform enabled analyzing behaviors during dangerous traffic scenarios that
are not feasible to test in real world settings. The captured data will be used as a
dataset to build reinforcement learning (RL) based calibration models to optimize
the frequency, modality, and duration of alarms that workers are more attentive to.
Towards this goal, this paper provides the results of initial data analysis on how
workers responded to alarms sent to wearable devices with different modalities (e.g.,
sound, vibration), durations, and frequencies. Effect of different alarms on worker
behavior and physiology has been measured with metrics that rely on workers’ posi-
tional changes, head movements and what they see in their field of views, heart rate
changes, and acknowledgement of alarms on wearable interfaces. Results indicate
that alarms with shorter durations and repeated less frequently can lead workers to
react faster by detecting vehicles and acknowledging the wearable warning device.
Findings from this study can serve as benchmarking data for the design of effective
alarm systems for work zones and be utilized as a dataset for RL models for training
agents for such systems.

J. Qin (B) · D. Lu · S. Ergan

New York University, Brooklyn, NY 11201, USA
e-mail: jq676@nyu.edu
D. Lu
e-mail: dbl299@nyu.edu

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 63

S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_6
64 J. Qin et al.

Keywords Wearable devices · Situation awareness · Virtual reality · Worker safety

1 Introduction

Latest figures from the United States (US) Federal Highway Administration (FHWA)
and Department of Labor report that vehicle crashes led to 857 fatalities, 117 of
which were construction workers, and resulted in 44,000 injuries at roadway work
zones in 2020 [1, 2]. These metrics have either increased or remained at similar
levels in the last five years. While roadway work zones employ a variety of safety
measures to alert vehicle drivers approaching a zone, including layout of traffic
control devices (i.e., cones, barrels) and signage, recent work zone intrusion alert
(WZIA) systems have been developed to warn workers with sound and light-based
alarms after sensors detect potentially hazardous vehicles [3, 4]. Research has begun
to understand how these alarm systems can impact workers’ safety behavior, such
as how responsive they are to alarm characteristics and their resultant situational
awareness (SA), the perception of their surrounding elements and ability to project
the status of safety hazards in the near future [4–6]. Most WZIA systems emit a
fixed set of light and sound-based alarms from a stationary location in the work zone.
Consequently, these fixed alarm characteristics (e.g., volume, frequency) are difficult
for workers to perceive, either due to environmental conditions (e.g., construction
noise, foggy weather, night time) or simply because workers are positioned farther
away from the alarm devices. Studies have shown that workers farther away from
alarm sources are less likely to notice the alarms and react slower to them, decreasing
their SA of hazardous vehicles [5].
Some WZIA systems attempt to counteract the effects of workers’ distance from
alarms with personal alarm devices carried by the worker [5]. This system feature
points to the potential of wearable alarm devices to improve worker’s perception of
alarm characteristics and improve their resultant SA of nearby traffic. Such wearable
alarm devices are especially useful for short-term and mobile roadway work zones,
where relatively few traffic control devices are used in layouts [5]. However, there is
limited information on what wearable alarm characteristics, such as modality (e.g.,
sound, vibration), are more effective at evoking a worker to physically react [8]. This
paper analyzes data obtained from virtual reality (VR) simulations of urban short
term and mobile roadway work zones to examine how workers react to varying alarm
characteristics emitted by wearable devices (i.e., a smartwatch).
Improvements in VR and wearable sensing technologies have enabled researchers
to more precisely analyze worker behaviors in roadway work zone hazard scenarios
without endangering human participants in physical testbeds and real work zones.
In order to determine which alarm characteristics more effectively notify workers
of incoming hazards, the research team developed a hardware-in-the-loop (HIL)
experiment platform integrating virtual reality (VR), calibrated micro-traffic simula-
tions, and wearable sensors [7, 9]. Participants were invited to a visualization lab to
perform tasks in virtual work zones while wearing a smartwatch with an application
Towards Increased Situational Awareness at Unstructured Work Zones … 65

that emitted a random pattern of sounds and vibrations every time a hazardous vehicle
(e.g., speeding, work zone collision) approached the work zone. The data obtained
from the integrated platform was analyzed to evaluate how varying an alarm pattern’s
characteristics impact different worker behaviors (e.g., acknowledging alarm, turning
head, positive body movement). This data also forms the basis for developing rein-
forcement learning (RL) controls for the alarm patterns in response to how quickly
workers react to the alarms and maintain their safety.

2 Related Work

2.1 Previous Research Studies on Understanding Worker

Reactions to Alert Systems

Numerous studies have examined the reactions of workers to alarms in other domains
such as firefighting and medical settings (e.g., nurses) [10]. Within the construction
industry, however, studies have mainly focused on how workers respond to proximity-
based alarms, where alerts are raised to a person when they walk too close to a
hazardous object on construction sites [10]. Relative to vertical construction sites
(e.g., high-rise buildings), horizontal roadway work zones face a unique risk from
passing vehicles. With the commercial development of various WZIAS, research
studies have sought to compare their effectiveness in preventing accidents in real-
world roadway work zones or physical testbeds involving predetermined vehicle
driving trajectories. For example, Gambatese et al. compared how well different work
zone intrusion alarms could be heard amidst heavy construction noise in real-world
construction work zones [4]. Both construction workers and research staff onsite were
given clicker remotes to record when they heard an alarm. Remote clicker reaction
times range from 2 to 6 s. Thapa and Mishra utilized physical testbed simulating
vehicles and workers in the real-world to evaluate the effects of different work zone
layouts and different WZIA systems on worker reaction times, defining reaction
specifically as when a worker started to move away from the hazardous vehicle.
Mean reaction times were observed to be around 2 s [5].
In these studies, reaction times were a key metric in evaluating WZIAS and were
measured using either a remote clicker for each worker to press after noticing an
alarm, or a research team member reviewing video recordings for when workers
physically moved away from vehicles. In the case of the remote clicker measurement
method, a reaction time can be quantified but does not capture specific aspects of
a worker’s physical reaction (e.g., head turn, body movement). In the case of video
recordings, researchers seemed to only focus on a specific physical reaction of body
movement, excluding other physical reactions and aspects of a worker’s SA that
ultimately improve their life safety. Did the worker turn their heads before moving
their body away from the vehicle? Did the worker notice the vehicle before the alarm?
Furthermore, none of the aforementioned studies employed biometric sensors (e.g.,
66 J. Qin et al.

heart rate, respiration sensors) to measure if workers’ physiology indicated that they
perceived either the alarms or the approaching hazardous vehicles. With the research
team’s previous work on an integrated VR, traffic simulation, and wearable sensor
platform, results discussed in this paper will discuss the many ways participants
reacted to the alarm, beyond when they acknowledged the alarm on a remote device
and physically moved their body. These forms of reaction times were all quantified
by wearable sensor measurements, rather than based on qualitative measures.

2.2 Previous Research Studies on Traffic Safety Using

Wearable Sensors and Virtual Reality

To analyze worker’s behavior and SA more closely, wearable sensors have been tested
in various field studies, examining workers’ visual attention (i.e., eye-tracking) and
physiological state on real-world construction sites [11, 12]. Meanwhile, many VR-
based approaches have been developed for the purposes of worker safety training,
but few focus on using virtual simulations for the purposes of testing alarm devices
[13]. Most notably, Kim, et al. evaluated risk habituation among human workers after
multiple exposures to backup alarms from construction vehicles, and a work zone
traffic accident [14]. Their user study protocol had participants wear VR and perform
construction tasks in two separate sessions. During each session, VR-based sensors
measured how often participants looked up towards a construction vehicle backing
up towards them. An accident occurred when participants frequently did not look up
at the construction vehicle, causing it to dramatically accelerate towards and collide
with the participant. Risk habituation was measured by counting how often workers
gaze direction was towards the construction vehicle, and how this “checking rate”
changed between the participants’ first and second sessions. This measure of worker
vigilance appeared to improve with more VR sessions and exposure to an accident.
Rather than focusing on how worker behavior is impacted by increased time spent in
the VR work zone environments, this paper examines the specific impact of wearable
alarm characteristics. Overall, this work focuses on how wearable alarm devices can
promote worker safety, rather than use of VR-based safety risk exposure methods.

3 Methodology

In order to analyze the effects of varying alarm characteristics on forms of worker

reactions, user studies were conducted on an integrated platform of VR, microscopic
traffic simulations, smartwatch applications, and wearable sensors. User study partic-
ipants wearing a VR headset performed simulated construction tasks near traffic flow.
Hazardous vehicles (e.g., speeding vehicles, vehicles colliding with the work zone)
Towards Increased Situational Awareness at Unstructured Work Zones … 67

approaching the work zone triggered alarms of random characteristics (e.g., dura-
tion, frequency, modality) on a smartwatch on the participant’s wrist. A suite of
wearable biometric and locomotion sensors recorded different forms of the partici-
pant’s response to the different alarms and dangerous vehicles: watch acknowledg-
ment time, body movement reaction time, vehicle detection reaction time, head turn
reaction time, and heart rate percent change. The following subsections describe the
aforementioned simulation and data collection methods in more detail.

3.1 Integrated VR-Traffic Simulation Platform

Previous work by the research team led to the development of an experiment plat-
form that integrated VR environments with traffic simulations to realistically model
work zones along urban streets [7, 9]. Real-world work zones were reconstructed as
VR environments using 3D laser scanning and Unity3D game engine, while traffic
patterns along the work zone were modeled with the open-source software Simu-
lation of Urban Mobility (SUMO). SUMO has been validated to replicate traffic
flow, helping to ensure realistic vehicular trajectories in the simulation of a roadway
work zone. To further increase the realism of the work zone traffic simulation,
a customized application programming interface (API) was created to enable bi-
directional communication of traffic control devices (e.g., cones, barrels) and vehicle
locations in both the traffic simulation and VR environment (Fig. 1). An ultrasonic
sensor was also placed around the perimeter of the VR experiment area, creating a
hardware-in-the-loop (HIL) system that connected external monitoring of the partic-
ipant’s physical location to the VR environment. Three virtual environments were
created on this platform to simulate a variety of work zone scenarios on different
urban roadway conditions: (1) placing cones to define a mobile work zone in an
urban street, (2) installing a traffic sensor on an urban highway, (3) surveying across
an urban intersection. These scenarios were developed for the platform after review
of traffic safety literature revealed mobile and short-term construction zones may be
more accident prone due to lack of traffic control devices [9]. This paper provides
preliminary analysis of how users of the VR-traffic simulation platform reacted to
smartwatch alarms in the first work zone scenario.
Traffic simulation also enabled modelling of hazardous vehicle behavior, such as
speeding vehicles and vehicle intrusions or collisions at the work zone perimeter (see
Sect. 3.3). When these hazardous events occur in traffic simulations, alarm data is
sent to a cloud server to pass to wearable devices, such as a smartwatch, to emit an
alarm for the worker (Fig. 1). When the VR participant’s physical location is also
near the edge of the work zone perimeter, the ultrasonic sensor’s RaspberryPi control
board would also send data to the server to trigger the wearable alarm. Overall, this
integrated VR-traffic simulation platform served as a virtual testbed for simulating
hazardous vehicle events around a work zone and collecting data on precise changes
in workers’ physiology and behavior when responding to different alarm patterns
and work zone scenarios.
68 J. Qin et al.

Fig. 1. Integrated VR and

traffic simulation platform
for evaluating smartwatch
alarm characteristics’ impact
on worker reactions

3.2 Smartwatch Alarm Characteristics

The smartwatch application for the Apple Watch was developed to emit different
alarm patterns of randomized characteristics during VR-traffic user studies. Due
to the gap in the literature on worker alarm responses, the research team tested a
wide variety of alarm characteristics to notify users of dangerous vehicle events. The
Apple Watch used in this study already had built-in alarms types of various sounds
and vibrations, each lasting around 0.8 s. Alpha tests of smartwatch alarms found
that Apple Watch’s built-in “Retry” notification sound and vibration was found to
be most noticeable and was therefore used to create different alarm patterns. Each
alarm pattern is defined by a combination of characteristics: (1) alarm durations,
ranging from 1.8 to 3.8 s (2) number of repetitions, ranging from one to three times,
(3) frequency of repetitions, ranging from 0.5 to 1 Hz and (4) modality, including
tactile vibrations and vibrations combined with sounds. In our study, “alarm dura-
tion” is defined as the total length of time of continuously emitted built-in alarms.
For example, a built-in alarm duration of 0.8 s that was emitted three consecutive
times was considered to have an effective alarm duration of 2.8 s for this study. The
continuous duration of built-in alarm could be repeated once or twice, and the repe-
titions could be spaced apart by a pause of one or two seconds (i.e., a frequency of
1 Hz or 0.5 Hz). The “total notification period” of an alarm was defined as the length
of time of the selected alarm pattern, or the sum of the associated alarm durations
and pause periods and is shown in Fig. 2 below. The resulting 30 combinations of all
possible alarm pattern characteristics had total notification periods ranging between
1.8 to 15.4 s.
In terms of alarm modality, the Apple Watch is able to emit alarms with vibrations
and sounds, or alarms with vibrations only. To switch between these two modalities,
it was necessary to enter the Apple Watch’s system-wide settings and turn on or off
Silent Mode. As a result, the smartwatch application was unable to change alarm
modalities automatically during a single VR user study trial. In order to test worker
responses to both alarm modalities, each participant in the VR user studies conducted
at least two trials of the same virtual work zone scenario with each alarm modality
Towards Increased Situational Awareness at Unstructured Work Zones … 69

Fig. 2. Illustration of an
alarm pattern with duration
and frequency between two
repetitions with respect to
total notification period

characteristic (sound with vibration alarms, or vibration only alarms) fixed during a
single trial.

3.3 VR Experiment Design and Procedure

In each trial of the VR user studies, a single participant was asked to establish a
work zone perimeter by moving six cones onto blinking white lights (Fig. 3). Cones
placed in the correct location turned the white light green and were registered by the
traffic simulation. Subsequent vehicles in the simulation treated the properly set up
cones as obstacles to avoid. Each trial concluded when all six cones were placed by
the participants in the correct location. At least two trials were conducted for each
user to monitor how each participant conducted this sequence of work tasks while
experiencing the two alarm modalities: vibration only, and sound with vibration
combined. The order of modalities was randomized among participants. New York
University’s IRB approved human subject experiments on the integrated VR-traffic
platform (IRB-FY2020-3946). Thirty-five (n = 35) participants were recruited for
user studies involving this scenario.

Fig. 3. Work zone scenario for participants in VR user studies

70 J. Qin et al.

Three possible events in the VR user study were able to trigger alarms on the
smartwatch during the trials: work zone collision vehicles, speeding vehicles, and
the participant leaving the work zone perimeter. A “collision vehicle” was preset
to ignore obstacles, thereby intrude the work zone and collide with the cones or
the worker. The Scenario 1 traffic simulation was set up to send a colliding vehicle
to the work zone 10 s after each trial began. Traffic simulation was also preset to
include three of five cars defined as “speeding vehicles” and were spaced such that
the speeding alarms would be raised at least 15 s apart in order for participants
to experience the full range of alarm total notification periods. Speeding vehicles
were defined as any vehicle which exceeded 25 miles per hour (12 m/s) in the area
surrounding the work zone. Participants could also trigger a smartwatch alarm by
passing the edge of the virtual work zone perimeter into the area for incoming traffic.
The ultrasonic sensors were set up to raise an alarm when the participant’s physical
location in the lab environment corresponded to the work zone perimeter in the VR
environment.
When a speeding vehicle, collision vehicle, or work zone perimeter event
occurred, the time stamp of each alarm event along with the smartwatch’s randomly
picked alarm pattern ID was recorded for analysis of participants’ reactions to the
alarm.
In order to measure how participants reacted to the alarm, various sensors were
worn by the participants during the VR user studies. Two wearable sensors, an Apple
Watch and a photoplethysmography (PPG) wristband (Empatica E4), were worn by
participants’ nondominant wrist to monitor their responses to the VR environment
and traffic simulation. The Apple Watch alarm application registered whether and
when a user chose to acknowledge or ignore the alarm. The PPG sensor monitored
physiological reactions by recording data on participants’ heart rate (HR) at a rate
of 1 Hz. The VR headset also served as a sensor for recording the participants’
head position, orientation at a rate of 50 Hz. together for statistical analysis. Custom
raycasting scripts in Unity3D were also written to record when a hazardous vehicle
(speeding or collision) was inside the participant’s VR field of view. Together with
timestamp records of alarm events’ pattern IDs, these sensor recordings served as
the basis for statistical analysis of worker reactions to alarm characteristics.

3.4 Measuring Workers’ Alarm Reactions

For the purpose of quantifying each user’s response to the different alarm patterns,
multiple aspects of worker reactions were examined: (1) acknowledging the alarm
on the smartwatch interface, (2) moving their body in a safe direction, (3) turning
their head, (4) detecting a hazardous vehicle in their field of view, and (5) heart rate
change. These reactions were analyzed with respect to the timestamp of each alarm
event. If the user tapped the watch screen to acknowledge the alarm, the time stamp
of the user’s response was recorded as well. In analyzing the user studies data, the
smartwatch response time, tack, was calculated as the difference between the alarm
Towards Increased Situational Awareness at Unstructured Work Zones … 71

and screen tap time stamps. The body movement reaction time, tbody, calculated
as the time elapsed between the alarm event and when the participant initiated the
maximum change in their perpendicular position on the road afterwards. The user’s
head movement was tracked by analyzing the orientation of the VR camera, recorded
as the 3D unit vector gaze direction of where the user is facing. The head turn reaction
time, thead, was defined as the time elapsed between the alarm event and when the
participant initiated the maximum angular change in gaze direction afterwards. When
a collision or speeding vehicle entered the participant’s VR field of view, the time
stamp was recorded as vehicle detection event. The vehicle detection reaction time,
tveh, was calculated from the time elapsed between the alarm event and when the
participant first saw the hazardous vehicle in their field of view. This study also
examined how participants’ physiological factors changed in response to alarms.
The PPG wristband sensor recorded heart rate (HR) in beats per minute (BPM). The
heart rate percent change was defined as the percent difference between the average
heart rate over 2.5 s time interval before and after each alarm. The time frame of
2.5 s was selected based on the literature review of general reaction times to alarms.

3.5 Evaluating Impact of Alarms on Worker Reactions

Each form of worker reaction was analyzed for average reaction times with respect
to alarm pattern characteristics (i.e., alarm modality, duration, repetitions, and
frequency). Average reaction times were analyzed with respect to each alarm pattern
characteristic independently for evaluating any positive or negative correlation
between a particular form of worker reaction’s speed and a specific alarm char-
acteristic. Nonparametric analysis of variation statistical tests were also conducted
to determine the impact significance of different alarm characteristics as sources of
variation, without having to assume the reaction times followed a particular family of
distributions in statistical tests. For each alarm characteristic in this study (modality,
duration, frequency, number of repetitions), a Kruskall-Wallis H-test was used to
determine if the distributions of sample groups of worker reaction times (e.g., head
turn reaction times grouped by duration = 1.8 s vs. 2.8 s vs. 3.8 s) for analysis
have significantly different distributions. The null hypothesis for each test is that all
sample groups were drawn from the same distribution, meaning that worker reac-
tion times are not significantly different with respect to different alarm characteristic
values. Statistical tests with significance levels of p < 0.05 would indicate the alterna-
tive hypothesis is true at 95% confidence, meaning that variation in a specific alarm
characteristic correlates with significantly different worker reaction times.
72 J. Qin et al.

4 Results and Discussion

The thirty-five (n = 35) participants recruited for the VR user studies consisted
of affiliated faculty and students. Only a select fraction (29%) of participants had
previous construction and road work experience, ranging between 2 months to
6 years. Results presented here are not representative of the roadway construction
workforce. That said, the task of placing cones in the VR environment was well
understood by all participants, and lack of construction experience was not expected
to impact the participants’ behavior in the VR traffic simulation. A total of 251
alarms were experienced by all participants. Only 3 of those alarms were triggered
by a participant stepping outside the work zone perimeter. Of the remaining 248
alarms, 179 resulted from speeding vehicles and 69 resulted from collision vehicles.
Results discussed in the following sections pertain only to speeding and work zone
collision vehicle alarms.

4.1 Worker Reaction Benchmarks

The average reaction time across worker response types for each alarm modality
(sound with vibration, or vibration only) are shown in Table 1. The sample sizes
used to determine mean reaction time for each response type and alarm modality
are also presented. Alarm reaction times two standard deviations beyond initial anal-
ysis of mean reaction time are omitted from these results. In terms of watch alarm
acknowledgement, Table 1 indicates that only around 22% of alarms were acknowl-
edged on watch by participants. This low response rate could be attributed to the
general inconvenience of tapping the watch screen rather than a sign that most people
did not notice the alarms. Sound and vibration alarms were nearly twice as likely
to be acknowledged as vibration only alarms. Additionally, participants were faster
(5.95 s) on average in acknowledging their watch in response to sound and vibration
alarms than in response to vibration only alarms (6.56 s). Given the low rate of watch
acknowledgement, the other forms of reaction may be more useful in evaluating
which alarm patterns may be more effective in evoking a physical reaction from the
worker. Table 1’s mean reaction times indicate that on average, participants may have
first seen a vehicle in their field of view before turning their heads or moving their
body. This could imply that most participants noticed the vehicle in their peripheral
vision after the alarm before physically moving out of the way. is may also explain
for the low rate of watch acknowledgement. If most participants quickly noticed the
hazardous vehicle and moved in some way within the first 3 s, they may not have
felt the need to acknowledge the watch alarm.
Mean and quartile reaction times are illustrated in box plots in Fig. 4, with indi-
vidual subplots showing a single form of worker reaction time compared with inde-
pendently varying a specific alarm characteristic. Subplots in Fig. 4 for body move-
ment and head turn reactions show mean reaction times generally hovering around
Towards Increased Situational Awareness at Unstructured Work Zones … 73

Table 1. Statistics for worker reaction times for each response type across all alarms
Sound + Vibration Alarms Vibration Only Alarms
n Mean N Mean
Reaction Time Reaction Time
Total Alarms 125 n/a 123 n/a
Watch Acknowledgement 20 5.95 13 6.56
Vehicle Detection 114 1.36 118 1.49
Head Turn 120 2.19 114 2.01
Body Movement 119 2.46 118 2.19

2 s regardless of variations of alarm characteristics of duration, number of repeti-

tions, and repetition frequency. On the other hand, plots of watch acknowledgment
time reveal that workers tended to acknowledge the watch alarm more slowly with
increasing durations of sound and vibration alarms (Fig. 4a, upper right). Similarly,
workers tended to acknowledge the alarm more slowly with increasing repetitions
and increasing frequency of repetitions. (Fig. 4b and c, upper right). Closer inspection
of vehicle detection time plots reveal that workers also tend to take more time to look
and see the hazard vehicle as sound and vibration alarm duration increases. While
these findings can change as more data is collected in future VR user studies, prelim-
inary results indicate that shorter durations and less frequently repeating alarms may
encourage workers to react faster.

4.2 Analysis of Variance

Table 2 shows the analysis of variance results for the different sources of variation
and measures of reaction. Each row is a different alarm characteristic or number of
previous alarms experienced in a single trial. Each measure of reaction column has
both the test statistic and p-value for the corresponding source of variation. Both p
< 0.05 and p < 0.10 levels of significance were considered in evaluation of the tests,
as indicated by the respective dark and light green highlighted boxes in Table 2.
Formally, the Kruskal–Wallis H-test analysis of variance statistical tests indicate
whether distributions of workers’ reaction times grouped by an alarm characteristic
belong to different populations. Informally, these statistics show if a worker exhibited
a significantly different reaction after experiencing different alarm characteristics,
without indicating whether that reaction was faster or slower relative to an increase
in an alarm characteristic value (i.e., suggests impact of an alarm characteristic with
no indication of positive or negative correlation). For example, the last row of Table
2 suggests that the most significant characteristic is the number of alarm repetitions,
as watch acknowledgement times grouped by alarm repetitions appear to belong to
significantly different distributions (H = 9.07, p = 0.011). In other words, workers
with two very different watch acknowledgement times likely experienced different
74 J. Qin et al.

Fig. 4. Plots of different forms of worker reaction time with respect to different alarm character-
istics: a) alarm duration, b) alarm repetitions, c) alarm repeat frequency
Towards Increased Situational Awareness at Unstructured Work Zones … 75

Table 2. Kruskal–Wallis H-test analysis of variance in different measures of reaction based on

alarm characteristics as potential sources of variation. Bold text in dark green boxes and italics text
in light green boxes denote significance levels of p < 0.05 and p < 0.10, respectively
Measures of Reaction
Watch Vehicle Body
Acknowl- Detection Head Turn Movement
edgement Reaction Reaction Reaction Heart Rate
Time Time Time Time Change
Alarm
Stat p Stat p Stat p Stat p Stat p
Characteristics
Alarm Modality 1.44 0.230 0.000 0.988 0.00 0.975 0.00 0.974 2.73 0.098
Alarm Duration 3.886 0.143 0.96 0.620 4.93 0.085 1.00 0.607 2.48 0.290
Alarm Frequency 3.65 0.161 0.344 0.842 0.94 0.626 0.81 0.666 1.47 0.479
Alarm Repetitions 9.07 0.011 1.27 0.529 1.54 0.464 0.30 0.863 1.79 0.408

alarm repetitions. At the higher significance threshold of p < 0.10, alarm duration
also appears to have a significant impact on head turn reaction times (H = 4.93,
p = 0.085). Alarm modality also appears to correlate with high variation in the
percent change in heart rate (H = 2.73, p = 0.085) immediately after alarms (i.e.,
different heart rate changes observed among participants are likely associated with
alarms of different modality). Overall, alarm characteristics of vehicle detection and
body movement appeared to have little impact on worker reactions. These results
indicate that alarm modality, alarm duration, and repetitions that have the potential
to significantly impact worker behavioral measures. These characteristics need to be
further investigated across the measures of reactions.

5 Challenges and Future Work

One of the first steps is to focus on the alarm characteristics that highly correlate with
response measures. Two additional VR work zone scenarios that were developed
by the research team will focus on these characteristics among all we identified.
Future work will also focus on the time sequence of physical reactions by workers
(i.e., turning their heads then moving body or some other sequence of reactions) in
evaluating how alarm systems can improve worker safety. As previous alarm-related
fatigue studies may suggest, further analysis may reveal that workers exhibit the
strongest reaction after their first alarm, with diminishing reactions thereafter. This
study did not employ a VR headset with built-in eye tracking, and was therefore
unable to assess whether people saw the vehicle in the center or periphery of field of
view.
Additional visual analysis methods will be employed in future studies to obtain
detailed data on workers’ situational awareness of hazardous vehicles. While alarms
appear to have little effect on participants’ heart rate changes, other physiological
76 J. Qin et al.

measures (e.g., EEG, respiratory rate) could be considered. More VR user studies
will be conducted for different scenarios (night-time highway sensor installation,
afternoon surveying on urban intersections) which can yield more data on worker
reactions and reveal the effects of different worker environment and nature of the work
performed. With more experiment data, a training environment can be developed for
a reinforcement learning model that can potentially optimize alarm characteristics
to improve worker attention and reaction to safety alarms.

6 Conclusions

This study presented an approach for measuring different forms of worker reaction
and evaluating how alarm characteristics affected those reaction times and types
with in an integrated VR traffic simulation platform. Preliminary analysis of average
reaction times indicate that workers tend to react faster at detecting vehicles and
acknowledging the alarms on the warning device when alarms are shorter in dura-
tion and repeated less frequently. Nonparametric statistical tests on data from 35
participants within a simulated urban mobile work zone indicate some variation in
worker’s watch acknowledgment, head turn, and heart rate reactions in response to
an alarm’s number of repetitions (p < 0.05), duration (p < 0.10), and modality (p <
0.10), respectively. Given the low rate of watch acknowledgement, this study indi-
cates that the use of remote clickers or similar devices may not be the best method
of evaluating worker reactions to alarms, especially when other forms of physical
reaction such as head turn and body movements are more important to a worker’s
situational awareness and overall safety. Second-order or time-dependent effects are
worth investigating, since the number of previous alarms experienced by participants
may induce alarm fatigue (i.e., slower responses, lack of response to alarms). Find-
ings from this study begin to contribute to the limited understanding of how alarm
system features can improve workers’ situational awareness in roadway construction
work zones. The approach presented here for analyzing the effects of varying alarm
characteristics on worker reactions will serve as a basis for future investigations in
reinforcement learning-based optimal control of alarm patterns on wearable devices.

References

1. National Work Zone Safety Information Clearinghouse. https://workzonesafety.org/work-

zone-data. Last accessed 1 Aug 2022
2. National Work Zone Safety Information Clearinghouse. https://workzonesafety.org/work-
zone-data/worker-fatalities-and-injuries-at-road-construction-sites. Last accessed 1 Aug 2022
3. Federal Highway Administration (2009) Manual on uniform traffic control devices for streets
and highways, 2009 edn. FHWA, Washington, D.C.
4. Work Zone Intrusion Alert Technologies: Assessment and Practical Guidance. https://rosap.
ntl.bts.gov/view/dot/32574. Last accessed 1 Aug 2022
Towards Increased Situational Awareness at Unstructured Work Zones … 77

5. Thapa D, Mishra S (2021) Using worker’s naturalistic response to determine and analyze work
zone crashes in the presence of work zone intrusion alert systems. Accid Anal Prev 156:106125
6. Endsley MR (1995) Measurement of situation awareness in dynamic systems. Hum Factors
37(1):65–84
7. Work Zone Safety. Behavioral Analysis with Integration of VR and Hardware in the Loop
(HIL). https://rosap.ntl.bts.gov/view/dot/61050. Last accessed 1 Aug 2022
8. Yadav N, Sadeghi N, Kang J (2022) On-body placement of wearable safety promotion devices
based on wireless communication for construction workers-on-foot: state-of-the-art review.
Sensors 22(9):3134
9. Ergan S, Zou Z, Bernardes SD, Zuo F, Ozbay K (2022) Developing an integrated platform to
enable hardware-in-the-loop for synchronous VR, traffic simulation and sensor interactions.
Adv Eng Inform 51:101476
10. Luo X, Li H, Huang T, Rose T (2016) A field experiment of workers’ responses to proximity
warnings of static safety hazards on construction sites. Saf Sci 84:216–224
11. Lee W, Lin K-Y, Seto E, Migliaccio GC (2017) Wearable sensors for monitoring on-duty and
off-duty worker physiological status and activities in construction. Autom Constr 83:341–353
12. Real-world Mapping of Gaze Fixations Using Instance Segmentation for Road Construction
Safety Applications. arxiv.org/abs/1901.11078. Last accessed 1 Aug 2022
13. Moore HF, Gheisari M (2019) A review of virtual and mixed reality applications in construction
safety literature. Safety 5(3):51
14. Kim N, Anderson BA, Ahn CR (2021) Reducing risk habituation to struck-by hazards in
a road construction environment using virtual reality behavioral intervention. J Constr Eng
Manag 147(11):04021157
Resilient and Sustainable Urban
and Energy Systems
How Can Digital Twins Support the Net
Zero Vision?

Eleni Papadonikolaki and Chimay Anumba

Abstract Net Zero relates to decarbonisation efforts to tackle climate change by not
adding new emissions to the atmosphere. Reaching Net Zero depends on datafica-
tion and digitisation as data on carbon emissions collected across assets’ lifecycle are
important for neutralising them. This study aims to understand how technological
ecosystems such as Digital Twins – that connect physical and digital artefacts – can
support a Net Zero vision in the built environment, in an industry-agnostic way to
transfer any relevant lessons from other sectors. The method is a systematic litera-
ture review that structures new knowledge on the topic. The study showed that envi-
ronment sustainability has been an overused idea in scientific literature, yet rarely
operationalised. A scarcity of studies utilising the potential of digital twins for Net
Zero was found. The emergent themes are: overreliance on technocratic solutions
at the expense of systems thinking, proliferation of renewable sources and misun-
derstandings regarding visualisation necessity, and showing pathways for future DT
system design for Net Zero.

Keywords Digital twins · model-based systems engineering · sustainability · net

zero · review

1 Introduction

In 2019, the United Kingdom (UK) became the first G7 country to legislate
for Net Zero, recommending a 2050 Net Zero carbon emissions target (CCC,
2019). Construction is a key contributor to emissions, as buildings and construc-
tion accounted for 36% of final energy use and 39% of process-related carbon

E. Papadonikolaki (B)
University College London (UCL), Gower Street, London WC1E 7HB, UK
e-mail: e.papadonikolaki@ucl.ac.uk
C. Anumba
University of Florida, Gainesville, FL 32611, USA
e-mail: anumba@ufl.edu

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 81

S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_7
82 E. Papadonikolaki and C. Anumba

dioxide (CO2) emissions in 2018. Digitalization has become a key trend in inno-
vating construction projects to meet Net Zero target and improve sustainability. For
example, digital transformation through the power of data science has the potential
to monitor and evaluate and, ultimately, cut emissions to facilitate a more sustainable
footing for the world economy.
Current and emerging digital technologies have the capacity to support sustain-
ability efforts and enable corporations and firms to meet their sustainability goals.
Technologies such as Building Information Modelling (BIM), wireless sensing,
Internet of Things (loT) and blockchain have shown their potential in connecting
with building Digital Twins (DTs) for making construction processes more efficient,
faster, safer, and cheaper, globally attracting attention. Similarly, DTs, originally
defined by Grieves (2014), pave the way to cyber-physical interaction and conver-
gence between the physical world and the cyber world of production. To this end,
DTs bring together data from all aspects of product lifecycle, laying the data foun-
dation that enables traceability and better control of sustainability objectives. This
paper draws upon a systematic literature review to identify the potential of digital
twins for sustainability and in particular meeting the Net Zero vision. Whereas this
paper is industry agnostic, it aims to provide some recommendations for the built
environment.
A recent study in digital technologies in the built environment (BE) found that the
main technologies discussed in the literature were: BIM, augmented reality, virtual
reality (VR), Internet of Things (IoT), cloud computing, and big data (Papadonikolaki
et al., 2022). Papadonikolaki et al. (2022) argued that the concept of digital evolves
toward connected technologies such as BIM and big data and smart cities/big data.
This trend shows that the BE finds solutions to its problems in the use of technologies
that rely on big data. Unlike other sectors where big data is available, the BE is
behind the curve in terms of asset digitization, usage, and labor (Agarwal et al.,
2016). However, the BE sector is slowly picking up.
This paper systematically first reviews policy and industry agendas, best practice
and various digital technologies that support digital twins to identify how they can
accelerate insight and enable the Net Zero vision. Afterwards, the paper is built
upon a Systematic Literature Review (SLR) aiming to document the state-of-the-art,
structuring disparate knowledge around Digital Twins and Net Zero and synthesize
a new perspective addressing the following Research Question (RQ):
How can Digital Twins support a Net Zero vision in the built environment and what lessons
are there to learn from other sectors?

This review aims to transfer lessons across sectors and enhance cross-disciplinary
and cross-industry learning beyond silos. The contribution of the paper is the provi-
sion of a practical guide to support the sustainability and corporate social respon-
sibility agendas of construction businesses by providing a summary and outlook of
how digital twins contribute to the Net Zero vision.
How Can Digital Twins Support the Net Zero Vision? 83

2 Theoretical background

2.1 Origins of Digital Twins

Originally developed to improve manufacturing processes, Digital Twins (DTs) were

defined in a white paper in the early 2000s as a “virtual representation of what
has been produced” (Grieves, 2014). Grieves (2014) argued that DTs consist of
three components: (a) physical asset, (b) virtual asset and (c) bi-directional data
connecting the physical and virtual asset. The term digital twin is used to describe
a digital replication of a physical asset and additionally, the process of seamlessly
transmitting data between the physical and the digital twin (El Saddik, 2018).
In theory, digital twins can update data in real time, so that virtual models can
undergo continuous improvement through comparing the virtual asset with physical
asset (Tuegel et al., 2011). Digital twins shorten the product development cycle,
improve the build efficiency and guarantee accuracy, stability, and quality. Further-
more, the use of digital twins promotes efficient synergies between the different
stages of a product lifecycle (Qi and Tao, 2018).

2.2 Types of DT technologies

There are three main types of DT technologies: (a) input technologies, for sending
data from the physical world to the machine, (b) processing technologies, for
analysing the data and (c) output technologies, for sending information from the
machine back to users and/or to the physical world. Therefore, DTs relate to a whole
interlinked ecosystem of technologies to deliver a DT.
First, during the input stage, DTs focus on modelling the behavior of the phys-
ical asset or product accurately over an entire lifetime. To do this, DTs use data
from input devices such as sensors installed on physical assets, e.g., through the
Internet of Things (IoT), a foundational technology for DTs, to follow their real-time
performance, operating conditions and changes over time.
Second, during the processing stage, DTs demonstrate the impact of live changes,
evaluating scenarios, external environmental conditions and other endless vari-
ables by incorporating multi-physics simulations, data analytics and Artificial
Intelligence (AI) including Machine Learning (ML) data analytics approaches.
This data-intensive stage completely eliminates the need for physical prototypes,
reduces modeling and simulation time but specifies the need for high-spec digital
infrastructures, including cloud and edge computing to deal with massive (big) data.
Third, as the DTs evolve, they are continuously updated to reflect changes to their
physical twins across the product lifecycle, feeding back through output technologies
into a virtual environment that enables users to monitor and continuously optimize
the product or asset. Visualization opportunities through Augmented Reality (AR)
84 E. Papadonikolaki and C. Anumba

and Virtual Reality (VR) usually make the top-most layer of a DT that combines the
data and insights to present, advise and interact with users or other machines.
DTs have various usages across the product lifecycle as they have the potential to
link production and product and its performance. According to Siemens, the appli-
cations of DTs depend on the product lifecycle, identifying three types of DTs: (i)
product, (ii) production and (iii) performance. When these three DTs are integrated,
they can support real-time and dynamic cyber-physical connectivity across all stages
of the product and production lifecycles.

2.3 Application areas of DTs

Aerospace: DTs originated from systems engineering and in particular in Model-

Based Systems Engineering (MBSE) (Madni et al., 2019). Various current initiatives
in the aerospace and defense industry are the continuation of work by NASA in the
1960s that later formed the basis of NASA Technology Roadmap (Shafto et al., 2012).
The NASA Apollo project entailed the creation of two identical space vehicles built
to allow mirroring the conditions of the space vehicle during the mission (Rosen et al.,
2015). After the one vehicle was launched, another vehicle remained on earth and
was called the ‘twin’ This ‘twin’ which was not yet digital but a physical prototype
was used extensively for training during flight preparation and afterwards, during
the mission to simulate real-time alternatives on the earth-based model, fed with
data on the flight conditions to assist the astronauts. In aircraft manufacturing DTs
are responsible for asset performance benefits of 30% increase in aircraft uptime
(Grieves and Vickers, 2017).

Automotive: The automotive industry has been traditionally credited for various
innovations in not only business, (see Taylorism, Fordism, Lean thinking) but also
production technology. Such a disruptive industry is leading the DT developments
from manually-operated cars to fully autonomous vehicles leveraging automated
control systems. Rudimentary developments of DTs in the automotive include simple
computer simulations of cars’ braking systems to simulate how they would perform
across various real-world scenarios and weather conditions (Madni et al., 2019).
Schleich et al. (2017) describe how through DT technologies, Tesla advances car
automation, through sensors, IoT, AI, ML and simulation tools. TESLA aims at
developing a DT for every car, hence enabling real-time data transmission between
car and factory. Other automotive manufacturers focus on AR and VR applications
for the vehicle’s DTs.

Plant Management: MBSE has been applied apart from aerospace and automotive
also in plant manufacturing, not only during design but also during the back-end of
the life cycle for diagnosis and optimized operations ((Rosen et al., 2015). In MBSE,
the role of modeling, simulation and optimisation are inter-connected as only after
accurate data is fed into the system to produce robust simulations, optimisation can
How Can Digital Twins Support the Net Zero Vision? 85

take place to increase the competitiveness of the product or asset as to energy and
resource efficiency, short production time-to-market and enhanced control. Relevant
developments in logistics and warehousing are also revolutionalised through industry
leaders such as Amazon and IBM for scenario-live-simulations and digital shop floor
management (Brenner and Hummel, 2017).
Medical: Digital twin technologies have also been pioneered in medical science. By
creating a human digital twin, clinicians can follow the performance of human and
artificial organs (Barricelli et al., 2020). Other applications include fitness trackers
that with very simple sensors can monitor patients’ health and performance and even
schedule medical tests when needed. In cardiovascular medicine, DTs can draw
diagnosis and prognosis to tailor future treatments based on health status and data
through cyber-physical synergies (Corral-Acero et al., 2020).

Infrastructure and Buildings: In construction, DTs are defined as “a realistic

digital representation of assets, processes or systems that is distinguished by the
data connection between the physical and virtual asset” (Bolton et al., 2018) and
their visual properties are accentuated. Whereas DTs can be developed at any project
stage, Sacks et al. (2020) argued that in order to have a fully mature DT, its imple-
mentation should start during the conceptual design and thereafter the information
from the design and construction stage of the physical asset, including as-planned
process, as-designed product, as-built product, as-preformed process information
about the physical asset, connected to the digital asset real-time to monitor the phys-
ical asset and start adding as maintained/renovated asset information. While DTs
may be used in isolation without the need to engage with other technologies, their
real value is realized when they are used for asset management. For example, Love
and Matthews (2019) found that DTs were linked to a suite of technologies such
as Building Information Modelling (BIM), Geographic Information Systems (GIS)
and supervisory control and data acquisition (SCADA) systems to enable a digital
twin during operations and maintenance. Akanmu et al. (2021) explore the develop-
ment of next-generation cyber-physical systems (CPS) and digital twins (DTs) for
the construction industry.

2.4 Sustainability

Triple Bottom Line (3BL): Sustainability in business has been defined as the
‘triple bottom line’ (3BL) of people, planet and profit (or societal, environmental
and economic sustainability). However, CEOs and business leaders predominately
focus on economic sustainability for business, profit and markets and less on envi-
ronmental and social sustainability. Elkington (2018), who developed the 3BL term,
notes that success or failure on sustainability goals cannot be measured only as profit,
but instead as wellbeing of billions of people and the health of our planet. Whereas
there have been improvements, our climate, water resources, oceans, biodiversity
86 E. Papadonikolaki and C. Anumba

and wellbeing are all increasingly threatened. There is a need to understand how
digital technologies can support people and planet targets.

Net Zero: Net Zero relates to decarbonisation efforts to tackle climate change by not
adding new emissions to the atmosphere according to the United Nations. There are
many similar terms around carbon, emissions and climate, such as: carbon neutral,
climate neutral, climate positive, zero emissions, and so forth. Net Zero carbon
emissions describe an activity that releases net zero carbon emissions back into the
atmosphere and is essentially similar to ‘Carbon Neutral’. Net Zero relates to carbon
or emissions. For example, when a building is powered by solar power using zero
fossil fuels its energy is labelled “zero carbon” and as no carbon was emitted from
it, no carbon needs to be captured or offset. Likewise, Net Zero emissions are when
there is a balance of the overall greenhouse gas emissions (GHG) produced and
taken out of the atmosphere. Net Zero Energy (NZE) building, is a building with
net zero energy consumption, that is the total amount of energy used by the building
on an annual basis equals the amount of renewable energy created on the site or by
renewable energy sources offsite. To this end, Net Zero describes manmade activities
that stop adding climate-heating gases in the atmosphere.

2.5 Synthesis of Research Gap

Digitalisation has been promoted as a means of achieving sustainability goals.

However, sustainability goals, especially as outlined in the United Nations’ Sustain-
able Development Goals (SDGs), are too broad and usually misplaced, given the
far-reaching implications of sustainability (Kirchherr, 2022). This study focuses on
narrowing down (mainly environmental) sustainability regarding reaching the Net
Zero vision as an end-goal of digitalisation efforts and, in particular, the deployments
of digital twins, which is the next frontier in intelligent data collection, analysis and
simulation aiming to provide real-time behaviour and performance information.
Any system that mirrors the operation of another different system is defined as
a model. A model in turn is an abstraction from the structure and processes that
define the mirrored system (Batty, 2018). In this sense, an immediate question is
to define the differences between a real system and any computer model of that
system, and thus it is worth noting the conundrums that this raises with respect to
the arguments about what constitutes a digital twin. The main research aim is to
understand how DTs can support a Net Zero vision in the built environment and
transfer lessons from other sectors to structure knowledge on the potential of Digital
Twins for delivering a Global Net Zero vision and provide insight informing policy
or practice, and developing a new theoretical model or framework.
How Can Digital Twins Support the Net Zero Vision? 87

3 Methodology

3.1 Systematic Literature Review

Drawing upon the research gap described above, this paper conducts a Systematic
Literature Review (SLR) to synthesize and compare findings from studies and answer
specific research questions (Klein and Müller, 2020). This study aims at synthesizing
a new perspective and structuring disparate knowledge around Digital Twins and Net
Zero. The main research aim is to structure knowledge on the potential of Digital
Twins for delivering a Global Net Zero vision.
SLR as a methodology was originally developed in the medical sciences to consol-
idate information from several sources in a transparent, rigorous and detailed manner
(Tranfield et al., 2003). SLRs can produce new knowledge (Tranfield et al., 2003)
or document the state-of-the-art (Lockett et al., 2006) providing a better under-
standing of the phenomenon in question for practitioners and scholars. Through
these research instruments, SLRs guide the building of theory accumulating knowl-
edge and evidence after analyzing a large number of studies and methods, thereby
increasing consistency in the results and conclusions (Akobeng, 2005).

3.2 Data Collection

The sample consists of research articles on the broader area of Net Zero, that is
sustainability and digital twins without any time or industry/sector restrictions, in
order to have a holistic view of the phenomenon. Most papers of this SLR were
published in 2018, that echoes the timing of the 2016 Paris Agreement. One of the
largest academic online databases, Scopus was used to sample articles and in the
future that might be complemented by Web of Science (WoS), although no signifi-
cant discrepancies are expected. The sampling was limited to refereed journal arti-
cles. Books and book chapters are typically excluded from SLRs, as they are often
categorized into grey literature (Adams et al., 2017) or not considered to be subject
of the robust review process journal articles go through (Clemens et al., 1995). We
excluded conference papers as many do not undergo peer review. The sampling string
on Scopus was as seen in Fig. 1 below:

( TITLE-ABS-KEY ( "digital twin*" ) OR TITLE-ABS-KEY ( "model-based

systems engineering" ) AND TITLE-ABS-KEY ( sustainab* ) OR TITLE-
ABS-KEY ( "net zero" ) ) AND ( LIMIT-TO ( SRCTYPE , "j" ) ) AND (
LIMIT-TO ( LANGUAGE , "English" ) )

Fig. 1. Keyword search in Scopus database

88 E. Papadonikolaki and C. Anumba

A key decision was using both ‘Sustainab*’ and ‘Net Zero’ in the sampling, as
the latter is a recent concept and only a few results were relevant to ‘Net Zero’
according to the indexing keywords, although more papers were relevant to it. The
term ‘Carbon Neutral’ was also added to show the relevance to ‘Net Zero’ visions.
This search returned 258 document results (September 20th , 2022).

3.3 Data Analysis

The data analysis contained three sequential main parts as seen below.
Step 1: Descriptive analysis: After collecting the sample of paper, the first step of
the analysis was descriptive in order to understand some initial trends appearing
in the sample.
Step 2: Content analysis: Afterwards, an extraction form was created in Excel
including various categories of additional categorical data. The abstracts of the
whole sample were read and categorised. (n = 258). This approach was preferred
to bibliometric approaches of analysing large bodies of literature because it could
give granularity to the findings and possibilities for customized analysis. The
chosen categories for this analysis part were:
• Technologies discussed (free code)
• Processes (if applicable, free code))
• Target improvement (free code)
• 3BL category (societal, environmental or economic sustainability)
• Net Zero vision applicability (Yes/No)
• Sector (free code)
• Sub-sector (if applicable, free code)
• Paper type (empirical, review, position)
Step 3: Thematic analysis: Finally, after evaluating the relevance to the RQ a
final sample of 16 final articles were isolated for full reading and deeper analysis.

4 Results

4.1 Descriptive Results

The resulting literature was a sample of 258 articles published between 2012 and
2023. Whereas there was no limit in the publication date, the sample consisted of
mostly recent literature that remained stable between 2012–2018 with 0–4 publi-
cations annually and a steep trajectory starting in 2019 (20 published documents).
This increased steep trajectory resulted in a marked increase in 2022 (111 published
documents).
How Can Digital Twins Support the Net Zero Vision? 89

Fig. 2. Description of the literature sample (n = 258) per year by source

Some publication outlets were more prolific than others. Key outlets are: (i)
Sustainability Switzerland (34 published documents), (ii) Applied Sciences Switzer-
land (12 published documents), (iii) Energies (10 published documents), (iv) Journal
of Cleaner Production (9 published documents) and (v) Buildings (7 published docu-
ments). Figure 2 shows in greater details how these trends are emerging the recent
years across these different outlets. Notably, outlets dedicated to energy sectors,
manufacturing and the built environment are most represented in this.
The literature was from a variety of research fields, revealing the inter-disciplinary
nature of the phenomenon. Most studies were multi-disciplinary and were categorised
in more than one subject area. In particular, although most literature stemmed from
engineering (144 published documents) and computer science (80 published docu-
ments) areas, these were followed by energy (79 published documents), environ-
mental science (64 published documents), social sciences (59 published documents)
and business, management and accounting (37 published documents) among others.
It is important to note that these categorisations are not representative of the appli-
cation areas and the industries that the empirical nature of applicable studies relates
to. Figure 3 below summarises the above areas and shows further areas relative to
the sample.

4.2 Content Analysis

The content analysis was based on screening (reading) all the abstracts of sample
(n = 258). This approach was preferred to bibliometric approaches because it could
give granularity to the findings and possibilities for customised analysis.

Triple Bottom Line (3BL): The first ‘code’ studied in this step of the analysis was
aiming to understand how sustainability was used in these studies and to what extent
the resulted literature related to the areas of the 3BL: societal, environmental and
90 E. Papadonikolaki and C. Anumba

Fig. 3. Description of the literature sample (n = 258) per year by subject area (based on the Scopus
database index)

economic sustainability. As expected, due to the keywords used, the majority of the
studies were related to the environmental sustainability (213 published documents),
followed by social sustainability (36 published documents) and economic sustain-
ability (10 published documents). Naturally, there were instances were more than one
aspect of 3BL was relevant to the study, nevertheless, the main idea is that sustain-
ability and Net Zero as well as their relation to digital technologies is approached
more from an environmental perspective, rather than socio-economic, with implica-
tions for organising and implementing relevant digital transformation initiatives. To
this end, there were little works on stakeholders, requirements and policy. Figure 4
shows the analysis of this data.

Relevance to Net Zero: As this study looks in particular at the links to Net Zero,
apart from general sustainability ideas, it was important to understand the nature of
the extant literature on the topic. Surprisingly, only 16 out of the 258 abstracts of the

4% 14%

Societal
Environmental
Economic

82%

Fig. 4. Content analysis of the literature review sample based on relevant to the triple bottom line
aspects
How Can Digital Twins Support the Net Zero Vision? 91

published documents reviewed were clearly related to the Net Zero vision. This obser-
vation reveals the popularity of the concept of sustainability but also the lack of its
operationalisation in research. Sustainability is typically seen as an all-encompassing
and transferable approach to show improvement from research and linkages to social
and environmental impact and benefit. However, most of the times these claims are
not substantiated. However, looking at Net Zero is a tangible approach to showcase
the said impact.
Associated Technologies and Processes: As the reviewed literature was already
captured from a scientific database (Scopus) it had already assigned author keywords.
To understand the sample better in a more systematic and coherent manner, additional
keywords related to associated technologies and processes, or initiatives discussed in
the studies were added. In total, a maximum of three keywords were assigned to reveal
new patterns in the sample. From these, most of the reviewed literature related to smart
manufacturing approaches (34 published documents), Industry 4.0 (27 published
documents), smart cities and lifecycle thinking (19 published documents each).
Regarding technologies, most of the studies related to Artificial Intelligence (AI)
(33 published documents), Internet of Things (IoT) (29 published documents) and
Building Information Modelling (15 published documents) among others. Figure 5
illustrates the most frequent of these associated technologies and processes. The
former is show in green in the bar chart and the latter in blue colour. Figure 6 shows
more keywords containing the associated technologies and processes in relation to
DTs for sustainability in a Word Cloud format.
Application Areas: Another important aspect of this initial analysis was focused on
identifying the application areas of the relevant literature as regards to industry and
sub-sector. The majority of the sample concerned applications in the manufacturing
industry (84 published documents). From this, the applications areas were either in
optimising machinery, machine tools and production methods, or less so looking at
optimising plants, industrial sites, creating smart factories, followed by a smaller
sub-set of studies where the DTs were used in optimising the performance and
sustainability of materials and material properties. This massive application area
was followed by applications in construction (33 published documents). Most of
the applications of the reviewed literature in the construction area were related to
housing and larger complexes in the form of smart campuses where DTs could
provide improvements in the management of energy, maintenance and performance.
Construction as an application area was purposely distinguished by urban plan-
ning applications and infrastructure, to provide greater granularity to the findings.
Infrastructure was another application area related to several studies (28 published
documents). These applications were mainly on water systems (8 published docu-
ments), rail (4 published documents), roads (3 published documents) and less so in
bridges and ports. Applications as a larger urban planning scale were also a signif-
icant number (24 published documents) with specific applications areas at neigh-
borhood levels. Other applications concerned energy (18 published documents) with
applications in energy production, power grids, battery optimisation, geothermal
and other sources of energy, as well as agriculture (14 published documents), e.g.,
92 E. Papadonikolaki and C. Anumba

Robotics
Smart transportation
Geographic Information Systems (GIS)
Safety
Digital transformation
Energy consumption
Climate change
Circularity
Data science
Building Information Modelling (BIM)
Lifecycle
Cyber-Physical Systems (CPS)
Internet of Things (IoT)
Smart manufacturing
0 5 10 15 20 25 30 35 40

Fig. 5. Associated technologies (in green) and processes or initiatives (in blue) in the reviewed
documents

Fig. 6. Word cloud with the associated technologies and processes to digital twins for meeting
sustainability visions goals
How Can Digital Twins Support the Net Zero Vision? 93

Fig. 7. Analysis of application areas of literature on digital twins for meeting sustainability

in farming, livestock management, food production and greenhouse. Figure 7 shows

the application areas.
Paper Type: Finally, the content analysis focused on the exact type of each piece of
the reviewed literature. Although bibliographic databases such as Scopus already
have categorisations of ‘document type’, the preliminary analysis in this study
showed that these categorisations are, in some instances, flawed. Therefore, in this
step, the abstracts of the published studies were reviewed and showed that only 168
published documents were empirical studies, followed by 47 review papers, 22 posi-
tion papers and 4 editorials. This shows that the topic is new and there is a lot of
interest in understanding the state-of-the-art of this novel field. This step is impor-
tant for understanding which studies have empirical basis before proceeding to the
thematic analysis. Figure 8 shows this information.

4.3 Thematic Analysis

The thematic analysis investigated closely the sub-set of the data on Net Zero, which
was a sample of 16 final papers mainly consisting of empirical (n = 14) and review
94 E. Papadonikolaki and C. Anumba

Empirical
Editorial
Review
Position
Unclear

Fig. 8. Analysis of document type in the literature review sample (n = 258)

(n = 2) papers. It is worthy to mention that some papers outside this sample studied
low-carbon emissions and efforts to reduce carbon emissions but not reaching Net
Zero and neutrality yet, despite being an effort towards that direction.

Renewable Sources: A considerable number of studies looked at Net Zero through

the lens of renewable systems, such as wind energy systems (Fahim et al., 2022),
Photovoltaic systems (PV) (Bartie et al., 2021) and geothermal energy systems
(Khanmohammadi et al., 2022). Even though a renewable energy system does not
have carbon dioxide emissions during the operation, it is marked with environmental
emissions embedded in their implementation. However, these were not covered in
these studies, nor any attempts were made to shift them.

Visualisation Capability: Another key theme emerging from the data was the
emphasis on the visualisation potential of the studied DTs for meeting Net Zero.
This trend was particularly present in studies coming from the wider area of the built
environment, such as construction of buildings and infrastructure, ranging existing
buildings (Kaewunruen et al., 2018), townhouses (Kaewunruen et al., 2019) and rail
station buildings (Kaewunruen and Xu, 2018). Equally, in this category of studies,
data derived from BIM, especially proprietary applications such as Autodesk Revit
was the norm.

Technocratic Solutions: One of the main themes of this investigation was the
linkages to data science and in particular Artificial Intelligence (AI) and Machine
Learning (ML) methods in order to deal with large sets of data points and make
predictions to reach carbon neutrality (Tariq et al., 2021). However, from these tech-
nical solutions, there was little consideration to wider workflows and policies. Most
of the studies showed a lack of systematic views of the problem of addressing Net
Zero.

Systems Thinking: Especially in the review papers, but also in some of the empirical
papers, there was a better linkage to system thinking and stakeholder engagement. In
these studies, there were attempts to link to policies and existing workflows, or create
How Can Digital Twins Support the Net Zero Vision? 95

new ones. These approaches considered both policy instruments and technological
methodologies. For example, in the effort to develop Positive Energy Districts (PED)
as energy-efficient and energy-flexible urban areas with surplus renewable energy
production and Net Zero greenhouse gas emissions, active information exchange and
analysis, as well as stakeholder engagement was seen as paramount for reaching Net
Zero (Zhang et al., 2021). Similarly, Howard et al. (2021) envisaged an ecosystem
of DTs in a distributed approach of “system of systems” interconnecting them in a
production facility of greenhouses to simulate all aspects of the production chain,
from production to supply.

5 Discussion and Conclusions

This study undertook a systematic literature review to answer the RQ of how Digital
Twins can support a Net Zero vision in the built environment, and showed, first, that
in terms of application areas, there is a lot of development in the manufacturing,
especially machinery and operations, and construction, especially at an urban level
and infrastructure. However, although most of the relevant advancements focus on
the environmental sustainability aspect of the 3BL, there is a scarcity of studies
on Net Zero and in general operationalisation of environmental sustainability for
neutralisation of carbon emissions. The investigation also showed that apart from
DT technologies, other associated technologies, such as AI, IoT and BIM and asso-
ciated processes, such as lifecycle thinking, and Industry 4.0 are complementing DT
solutions for environmental sustainability.
The thematic analysis of the sub-set explicitly targeting DTs for Net Zero goals,
showed several emergent themes. First, there was a proliferation of studies on renew-
able sources that are per se carbon neutral in their operation. However, the carbon
emissions during their implementation were not considered, signalling a need in
shifting these. Second, there were several studies focusing on visualisation proper-
ties of DTs that showed a superficial approach and missed opportunity in exploiting
heterogenous types of data. Finally, the study revealed an overreliance on techno-
cratic solutions in addressing the topic, but at the expense of systems thinking. In
such a complex topic with political implications, only a couple of studies approached
it from a systems thinking approach, attempting to connect DTs together in a “system
of systems” approach to meet Net Zero. These systems were of both hard (e.g., DTs)
and soft (e.g., stakeholders) nature, showing pathways for future DT system design
for Net Zero.
Whereas there is a popular belief that digitalisation and datafication can support the
green or sustainability transition in our society, this paper is the first study explicitly
connecting the developments in the field of Digital Twins with the potential to achieve
Net Zero. The practical implications of this work are on identifying key themes on
the topic, for further analysis and stepping stones for developing a framework or
roadmap for researchers, practitioners and policy-makers in the area. Naturally, this
96 E. Papadonikolaki and C. Anumba

work has some research limitations, especially located at the small data sample in this
niche field. Future studies in this direction can work towards revisiting the sample
and expanding it with more document types and broad definitions of environmental
sustainability so as to capture findings towards the preceding stages of research
investigations around the Net Zero vision.

References

Adams RJ, Smart P, Huff AS (2017) Shades of grey: guidelines for working with the grey literature
in systematic reviews for management and organizational studies. Int J Manag Rev 19:432–454
Agarwal R, Chandrasekaran S, Sridhar M (2016) Imagining construction’s digital future. Capital
Projects and Infrastructure. McKinsey & Company, McKinsey Productivity Sciences Center,
Singapore
Akanmu AA, Anumba CJ, Ogunseiju OO (2021) Towards next generation cyber-physical systems
and digital twins for construction. J Inf Technol Constr 26:505–525
Akobeng AK (2005) Understanding systematic reviews and meta-analysis. Arch Dis Child 90:845–
848
Barricelli BR, Casiraghi E, Gliozzo J, Petrini A, Valtolina S (2020) Human digital twin for fitness
management. IEEE Access 8:26637–26664
Bartie N, Cobos-Becerra Y, Fröhling M, Schlatmann R, Reuter M (2021) The resources, exer-
getic and environmental footprint of the silicon photovoltaic circular economy: assessment and
opportunities. Resour Conserv Recycl 169:105516
Batty M (2018) Digital twins. Env Plann B: Urban Anal City Sci 45:817–820
Bolton, A., et al. (2018) Gemini principles. Centre for Digital Built Britain (CDBB)
Brenner B, Hummel V (2017) Digital twin as enabler for an innovative digital shopfloor management
system in the ESB logistics learning factory at Reutlingen-University. Procedia Manuf 9:198–
205
Ccc (2019) Net Zero The UK’s contribution to stopping global warming. UK Committee on Climate
Change ed. UK Committee on Climate Change
Clemens ES, Powell WW, Mcilwaine K, Okamoto D (1995) Careers in print: books, journals, and
scholarly reputations. Am J Sociol 101:433–494
Corral-Acero J et al (2020) The ‘Digital Twin’to enable the vision of precision cardiology. Eur
Heart J 41:4556–4564
El Saddik A (2018) Digital twins: the convergence of multimedia technologies. IEEE Multimed
25:87–92
Elkington J (2018) 25 years ago I coined the phrase “triple bottom line”. Here’s why it’s time to
rethink it. Harv Bus Rev 25:2–5
Fahim M, Sharma V, Cao T-V, Canberk B, Duong TQ (2022) Machine learning-based digital twin
for predictive modeling in wind turbines. IEEE Access 10:14184–14194
Grieves M (2014) Digital twin: manufacturing excellence through virtual factory replication. White
Paper 1:1–7
Grieves M, Vickers J (2017) Digital twin: Mitigating unpredictable, undesirable emergent behavior
in complex systems. In: Kahlen F-J, Flumerfelt S, Alves A (eds) Transdisciplinary perspectives
on complex systems. Springer International Publishing, Cham, pp 85–113. https://doi.org/10.
1007/978-3-319-38756-7_4
Howard DA, Ma Z, Veje C, Clausen A, Aaslyng JM, Jørgensen BN (2021) Greenhouse industry
4.0–digital twin technology for commercial greenhouses. Energy Inform 4:1–13
Kaewunruen S, Rungskunroch P, Welsh J (2018) A digital-twin evaluation of net zero energy
building for existing buildings. Sustainability 11:159
How Can Digital Twins Support the Net Zero Vision? 97

Kaewunruen S, Sresakoolchai J, Kerinnonta L (2019) Potential reconstruction design of an existing

townhouse in Washington DC for approaching net zero energy building goal. Sustainability
11:6631
Kaewunruen S, Xu N (2018) Digital twin for sustainability evaluation of railway station buildings.
Front Built Environ 4:77
Khanmohammadi S, Musharavati F, Tariq R (2022) A framework of data modeling and artificial
intelligence for environmental-friendly energy system: application of Kalina cycle improved
with fuel cell and thermoelectric module. Process Saf Environ Prot 164:499–516
Kirchherr J (2023) Bullshit in the sustainability and transitions literature: a provocation. Circ Econ
Sustain 3:167–172
Klein G, Müller R (2020) Literature review expectations of project management journal®. SAGE
Publications Sage CA, Los Angeles, CA
Love PE, Matthews J (2019) The ‘how’ of benefits management for digital technology: from
engineering to asset management. Autom Constr 107:102930
Madni AM, Madni CC, Lucero SD (2019) Leveraging digital twin technology in model-based
systems engineering. Systems 7:7
Papadonikolaki E, Krystallis I, Morgan B (2022) Digital technologies in built environment projects:
review and future directions. Project Manag J 53:501–519
Qi Q, Tao F (2018) Digital twin and big data towards smart manufacturing and industry 4.0: 360
degree comparison. IEEE Access 6:3585–3593
Rosen R, Von Wichert G, Lo G, Bettenhausen KD (2015) About the importance of autonomy and
digital twins for the future of manufacturing. IFAC-Papersonline 48:567–572
Sacks R, Brilakis I, Pikas E, Xie HS, Girolami M (2020) Construction with digital twin information
systems. Data-Centric Eng 1:e14
Schleich B, Anwer N, Mathieu L, Wartzack S (2017) Shaping the digital twin for design and
production engineering. CIRP Ann 66:141–144
Shafto M et al (2012) Modeling, simulation, information technology & processing roadmap. Natl
Aeronaut Space Adm 32:1–38
Tariq R et al (2021) Artificial intelligence assisted technoeconomic optimization scenarios of hybrid
energy systems for water management of an isolated community. Sustain Energy Technol Assess
48:101561
Tranfield D, Denyer D, Smart P (2003) Towards a methodology for developing evidence-informed
management knowledge by means of systematic review. Br J Manag 14:207–222
Tuegel EJ, Ingraffea AR, Eason TG, Spottswood SM (2011) Reengineering aircraft structural life
prediction using a digital twin. Int J Aerosp Eng 2011:154798
Zhang X et al (2021) Digital twin for accelerating sustainability in positive energy district: a review
of simulation tools and applications. Front Sustain Cities 3:35
A Review of Smart City Maturity
Assessment Models

Pavan Kumar, Aritra Pal, and Shang-Hsien Hsieh

Abstract The urban population is expected to increase from 56.61% (2021) to 68%
(2050), hinting at an enormous population inflow in urban areas across the globe. As
governments worldwide toward a sustainable way of life with sustainable develop-
ment goals (SDG), today’s urban areas encounter challenges such as pollution, over-
crowding, housing, traffic, and poor infrastructure. Smart Cities intends to identify an
intelligent and efficient way of facing these challenges and reducing the complexity
of challenges in urban areas. Although the implementation of Smart-cities world-
wide is increasing with the support of policies, smart cities tend to mature in multiple
directions at their own pace at different levels, attributed to the uniqueness of the
cities in a particular geographical area. Consequently, tracking a trail of the smart
city’s holistic maturity advancement becomes crucial. Therefore, this paper reviews
the existing smart-city maturity assessment models and proposes an outlook of a
well-defined holistic smart-city maturity assessment model, which is expected to aid
the implementation and efficient management of smart cities across the world.

Keywords Smart city · maturity models · maturity frameworks · assessment

methods

1 Introduction

1.1 Smart city origin and definition

Urban spaces have always been a crucial dimension of civilization and cities across
the globe have grown at different rates in multiple directions. According to the
United Nations, Cities stage more than half of the world population, and further
growth in the urban population is predicted. With such an extraordinary growth, city

P. Kumar (B) · A. Pal · S.-H. Hsieh

Department of Civil Engineering, National Taiwan University, Taipei, Taiwan
e-mail: d10521028@ntu.edu.tw

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 99

S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_8
100 P. Kumar et al.

municipalities have always aimed to find a feasible solution to urban challenges.

Time-lapse of the growth of cities across the globe portrays the advancements in the
process of management of cities, as cities choose to inculcate novel methods in the
functioning of cities by replacing laborious ones.
Digitization initiatives by the local government and existing initiatives around
innovation and technology have become one way for cities to be called smart [1]
boosting the implementation of smart cities. Digitization of processes and services
in cities, and developing the capability to work with data in the city organization as a
core component has solidified the concept of smart cities. The inclusion of innovation
and experiments in pursuit of solutions to the challenges of urban spaces portrays
the growth of the smart city concept. Walravens N et al. [1] highlight the cities that
strongly embraced innovation and experiments with technology can often be titled
smart cities. The origins of Smart city concepts are attributed to the fundamental
improvement and offering of a better quality of life and enhance the ease of living
for its citizens, simultaneously improving the information system for citizens [2].
Maestre G et al. [3] discusses the emergence and the purpose of smart cities are to
fill the void of energy-efficient and sustainable urban spaces with better interaction
between inhabitants, city municipalities, and the environment.
Several attempts have been made to define smart cities in literature. Walravens
N et al. [1] identifies technology, service of citizens, efficiency, and effectiveness
as main goals in several smart city initiatives. Several studies find the integration
of information and communication technology (ICT) into existing cities’ structures,
enabling sustainable living in the city is a crucial characteristic of the smart city.
Although the smart city concept often tends to hover between several elements such
as ICT, economy, governance, mobility, environment, and living, there is obscurity
in the definition [4]. The void for a shared and holistic definition of smart cities can
be attributed to the wide diversity in cities across the globe.

2 Methodology

Keywords “smart AND city* AND maturity models AND “maturity frameworks”
AND “assessment” were used to accumulate the resources from the Scopus database.
The search resulted in 234 papers in total, Mendeley, and Spreadsheets were used
to sort and avoid duplication. Screening with a focus on a maturity assessment, 23
papers that discussed the maturity assessment of the smart cities, written in English
were filtered.
A Review of Smart City Maturity Assessment Models 101

3 Assessment Methods

Discussions on smart cities often raise questions regarding measuring the city’s
success in terms of its smartness, progress tracking, and timely human intervention.
Assessment plays an important role by providing useful insights into answering these
questions and assisting the implementation of policies. Assessment of the city aims
to guide city municipalities to identify crucial areas to be improved [5] and plan
and execute timely required interventions. Several organizations have put forward
different ways of assessing smart cities, such as ranking metrics, standards, and
assessment frameworks.
The use of a ranking system in the context of a smart city is widely used. Ranking
cities enables positioning cities in a hierarchy based on indicators, such as city
infrastructure and technology services to city inhabitants. Orlowski A [5] Lists the
following widely accepted smart city rankings mechanisms.
• IESE Cities in Motion Index
• Top 50 smart cities governments (2020)
• 2020 smart city index by the institute for management development, in collabo-
ration with the Singapore University for technology and design
• Smart cities index by 2019 by EasyParkGroup (Easypark 2019)
This method of putting cities to assessment not only impacts the inhabitant’s social
attention but also often simulates the discussion on city development strategies [6]
and their implementation. Although the ranking method is significant in logically
placing the smart cities in the hierarchy, it skips providing a holistic approach and
brings the city municipalities into ambiguity. For example, Orlowski A [5] attempts
to highlight the usual occurrence of different positions in multiple rankings at the
same time, this eventually impacts the decisions of city municipalities. Furthermore,
the usefulness of the information provided by the ranking system is limited as it
only succeeds in providing city municipalities with information for comparing smart
cities in the hierarchy and not in the city’s progression [5]
Other assessment methods of smart cities such as the use of standards also pose
challenges that are rightly pointed out by Orlowski A [5] as standards tend to impose
a rigid framework of indicators that often succeeds on specific cities. As mentioned
earlier, various cities around the globe have been using an assessment tool at city
levels, subsequently, helping city municipalities find efficient methods to counter the
challenges of cities and improve living. Maturity models are extremely popular and
many cities are put to use maturity frameworks to assess growth.
102 P. Kumar et al.

4 Smart City Maturity Models

The process of evolution of the smart city, from its concept to its implementation,
assesses cities in different aspects often these aspects are named differently such
as dimensions, elements, or field of action. Several maturity models have identified
various dimensions based on their requirements and purpose. Maturity models that
attempt a holistic assessment of maturity identify dimensions that are not necessarily
focused solely on Information and communication technologies, but rather consider
other crucial verticals of smart cities such as education, innovation, and governance.
For example, Afonso RA [7] identifies domains for a smart city such as water,
education, energy, governance, housing, environment, health, security, technology,
and transport, however, a few domains such as governance may tend to overlap with
other domains as governance can be measured in every other domain. Mohsin BS et al.
[4] highlighted two major domains of smart cities that can be noticed in literature,
the hard domain focuses on the significance of ICT, its importance in the functioning
of other smart city sectors, and the soft domain, where ICT application is not critical.
Nevertheless, technology and its applications remain the least unfocused subject
of research when it comes to smart cities. Few maturity models have considered
untapped specific areas of smart cities and attempted maturity assessments in those
domains. A few areas are worth mentioning as follows.

4.1 Strategy, Planning, and Citizen Participation

The introduction of technological interventions in the city certainly aids the better
functioning of smart cities. However, this technological advancement can turn into a
challenge without proper citizen involvement. Thus, posing the requirement of under-
standing the citizen’s level of maturity towards participatory planning. Bouzguenda I
et al. [8] identifies the level of maturity towards the introduction of digital participa-
tory planning (DPP) in small European cities and proposes a framework that would
allow local governments to assess their level of maturity of sustainable smart cities
through DPP. Lack of strategy before embarking on the smart city implementation
journey can be disastrous as a strategy provides a structure to follow. Kapkaev Y
et al. [9] identifies strategy void as a major problem in the journey of digital trans-
formation of territories. Korachi Z et al. [10] offers a methodology to formulate a
strategy for the implementation of a smart city. Understanding the requirement of
smart city transformation Identifying the stakeholders, and method of collaboration,
and defining strategic goals and challenges for the smart city are crucial milestones in
the development of smart city strategy. The novel smart city maturity model proposed
by Moustaka V et al. [11] stresses citizen involvement but claims real-time data and
privacy concerns are major challenges that must be addressed.
A Review of Smart City Maturity Assessment Models 103

4.2 Mobility

[12–14] discusses the maturity assessment of specific dimensions of smart cities

related to mobility which focuses on transportation and ICT. Warnecke D et al. [13]
proposes a web-based self-assessment tool called SCMAB that attempts to enable
stakeholders to measure the impact of smart city initiatives in the context of the
mobility dimension. The tool is based on 36 indicators spreading across six themes:
Policy and planning, ICT integration, Intermodal integration, public transport perfor-
mance, environmental impact, and social impact. SCMAB maturity levels are score
based, categorized into five stages with descriptions concerning the inclusion of ICT.
However, the SCMAB tool fails to approach the holistic assessment of the smart city
and claims the void of policy coherence across different levels of government as
one of the major challenges. Visan M et al. [15] proposed and defined a coopera-
tive intelligent transport system (C-ITS), a mobility-specific maturity model with
five stages of maturity characterized by description focused on processed-oriented
rather than outcome-oriented. Furthermore, it emphasizes the relevant technologies
to implement C-ITS.

4.3 Data Transparency and quality

Data in the context of a smart city is another area where several maturity assessments
often seem to take place. Data management becomes extremely crucial as cities
embrace technology as a major enabler in the transition into smart cities. Technology
is an inseparable element in making smart cities as every smart city initiative looks
toward technology deployment to carry out the existing services and unlock efficient
ways to perform activities. Nearly every maturity model proposed and used data
for maturity assessment of the smart cities, for example [5, 16, 17] used data as
a fundamental unit of the entire assessment framework. As studies like [5, 8, 18–
20] are based on qualitative and quantitative methodologies, they naturally tend to
lack the measure of data transparency and ignore the capability of an open data
ecosystem in smart cities. Lnenicka M et al. [21] stresses on assessing the level of
maturity of open data transparency in a smart city context, by examining 34 portals
representing 22 European cities. Factors such as data understandability, usefulness,
ease of user interaction, data quality, and accessibility impact the transparency of
the open data ecosystem. The study identifies five types of ecosystems for open
data each representing a variation of Open Government Data deployment for smart
cities. It defines an open data ecosystem as “systematic efforts to integrate ICT and
technologies into city life to deliver citizen-centric, better-quality services, solutions
to city problems with open data published through the data-centric and data-driven
infrastructure” In addition to the transparency of the open data ecosystem, the quality
of the data impacts the results of the assessment of a smart city, eventually reflecting
in smart city implementation initiatives. Data Quality Driven Smart Cities Maturity
104 P. Kumar et al.

Models (DQSC-MM) attempt to assess the maturity of smart cities with a focus
on the quality of data generated and used. The model identifies connectivity, data
centers, data analytics, application, and end users are five leverage domains that
are further broken down into key domain areas [20]. Unlike Jāmi‘at al-Bah.rayn
[21], Rotta MJR et al. [19] examines the maturity of eGov platforms as digital
commons and presents Municipal eGov Platform Assessment Model (MEPA) to
assess 903 municipal websites’ maturity to aid the understanding of the promotion
of commons and citizen participation and cites digital commons as a “resource based
on information and communication technology, shared by groups and integrated
into a value chain, under principles of equity, coproduction, and sustainability”. As
MEPA concerns factors like data openness, transparency, and civic engagement, it
also attempts to stage the maturity levels that map with the commons-principles.
MEPA adds to another specific dimension aiding the maturity assessment of smart
cities.
Smart city maturity models are established by various cities with diverse spec-
ifications and definitions. They often are staged at multiple levels with each level
portraying its characteristics. The characteristics often tend to differ from one matu-
rity model to another, hinting towards uncommonness in consideration of the funda-
mental definition of maturity, and sometimes acceptance of capability in the basic
assumptions in the maturity models. For instance, the maturity levels, provided by
Nuraeni A et al. [17] are characterized based on the output of the task with a descrip-
tion of various stages of achieving the task ranging from 0 to 100%. On the contrary,
maturity levels proposed by Bernal WN [18] are characterized by the nature of
performing the activity i.e., integrated, analytically managed, and automated opti-
mization. Although the formation of the maturity stages differs from the context
it rests in, the fundamental assumption of the definition of maturity and capability
remained a void to be filled. It can also be highlighted that the definition of matu-
rity and capability in the formation of maturity models in an attempt to assess the
smart city varies and remains unshared across maturity models of smart city maturity
assessment.
Maturity assessment of domains such as transport and digital implementation is
crucial, however, consideration of sustainability as a domain is rarely seen in the
literature. Shen L et al. [16] introduces a methodology to assess low-carbon cities
(LCC) using the capability maturity model. The model helps city municipalities to
determine the maturity of LCC. However, like several other attempts, the model is
specifically tested only in Beijing. Jāmi‘at al-Bah.rayn [22] highlights the concern
over the applicability of maturity models is limited.

5 Proposed Smart City Maturity Assessment

The proposed smart city assessment consists of multiple layers as illustrated in Fig. 1.
The top layer will account for domains or dimensions of the smart city and are char-
acterized by not being common with each other. For instance, the mobility domain is
A Review of Smart City Maturity Assessment Models 105

easily distinguishable from the city’s environment domain. City municipalities can
choose the domains of their priority-based consideration using the Lens layer, thus
allowing the maturity model to tend to be more holistic in nature.
The stack of layers below the lens layer represents the process and outcome-
oriented sectors whose characteristics highlight commonness among all domains.
For instance, governance is a sector that can be common in the mobility and the
environment domain of the smart city. These layers shall represent the smart city
path as it considers the processes required for selected domains. For example, digital
participator planning proposed by Bouzguenda I [8] might happen to be the first step
for the particular city to embark on the smart city journey followed by strategies for
smart city implementation as proposed by Korachi Z [10]. These layers are outcome-
oriented and answer the question of achievement of the task at multiple levels such
as incomplete, managed, and performed. The layers represent the capability of the
smart city and its status of achievement framework, each capability is measured to
assess the efficiency with which it is performed. Maturity can be staged with a focus
on each capability as structured, integrated, and optimized. Upon putting all the
elements together, the proposed framework presents a holistic approach to assessing
of maturity of the smart city, allowing city municipals to focus on required priority
domains, and simultaneously provide the smart city path and its maturity assessment.
This also characterizes the framework to be more dynamic in nature as it provides
the flexibility for city municipalities to follow their maturity assessment.

Fig. 1. Proposed conceptual smart city maturity assessment framework

106 P. Kumar et al.

6 Conclusions

To support the maturity assessment of the smart city, this study highlights the recent
developments in strategy, planning, citizen participation, mobility, data transparency,
and quality that aid the smart city assessment. Smart cities at their strategic phase
clearly require an agreeable understanding of the capabilities they are there to achieve
within a specified time frame. Maturity is the method to achieve those capabilities.
Clearly distinguishable consideration of the definition of capability and maturity in
the assessment framework is needed as they impact the staging strategy of maturity.
Considering the capability as the necessary task that smart cities can achieve and the
maturity as the effectiveness and efficiency with which the capability is achieved are
the definitions considered in the proposed smart city maturity assessment framework.
The presence of the lens layer makes the framework more flexible and dynamic in
nature for cities as it allows city municipalities to choose the required domain specific
to the city to achieve the capability for the city and higher maturity. Further, the details
of the framework are to be considered to be developed in future attempts to establish
a systematic maturity assessment model for the smart city.

Acknowledgment This research has been supported by the National Science Technology Council,
Taiwan with grants MOST 111-2923-E-002-010-MY3 and MOST 111-2621-M-002-008.

References

1. Walravens N, van Compernolle M, Waeben J, Colpaert P (2019) Being a smart city is not the
goal, but what is? a quantitative and qualitative analysis of smart city strategies in 13 cities in
Flanders, Belgium. In: Proceedings of 2019 CTTE-FITCE: Smart Cities and Information and
Communication Technology (CTTE-FITCE), pp 1–5
2. Torrinha P, Machado RJ (2017) Assessment of maturity models for smart cities supported by
maturity model design principles, In: Proceedings of IEEE international conference on the
smart grid and smart cities, pp 252–256
3. Maestre G, Astudillo H, Nieto W, Concha G (2018) Empirical evidence of Colombian
national e-government programs’ impact on local Smart City-Adoption. In: ACM international
conference proceeding series. Association for Computing Machinery, pp 517–525
4. Mohsin BS, Ali H, Alkaabi R (2019) Smart city: a review of maturity models, 2nd smart cities
symposium, 24–26 March 2019
5. Orlowski A (2021) Smart cities concept – readiness of city halls as a measure of reaching a
smart city perception. Cybern Syst 52:313–327. https://doi.org/10.1080/01969722.2020.187
1224
6. Giffinger R, Gudrun H (2010) Smart cities ranking: an effective instrument for the positioning
of the cities? ACE: Archit, City Env 4(12):7–26
7. Afonso RA, dos Santos Brito K, do Nascimento CH, et al (2015) Brazilian smart cities: using a
maturity model to measure and compare inequality in cities. In: ACM international conference
proceeding series. association for computing machinery, pp 230–238
8. Bouzguenda I, Alalouch C, Fava N (2020) Examining digital participatory planning: maturity
assessment in a Small Dutch city. J Cleaner Prod 264:121706. https://doi.org/10.1016/j.jclepro.
2020.121706
A Review of Smart City Maturity Assessment Models 107

9. Kapkaev Y, Kadyrov P (2021) Digital maturity of territories during digital transformation. E3S
Web Conf. 258:01001
10. Korachi Z, Bounabat B (2019) Towards a platform for defining and evaluating digital strategies
for building smart cities. In Proceedings – 2019 3rd international conference on smart grid and
smart cities, ICSGSC 2019. Institute of Electrical and Electronics Engineers Inc., pp 32–40
11. Moustaka V, Maitis A, Vakali A, Anthopoulos LG (2020) CityDNA dynamics: a model for
smart city maturity and performance benchmarking. In: The web conference 2020 – companion
of the world wide web conference, WWW 2020. Association for Computing Machinery, pp
829–833
12. Kijewska K, de Oliveira LK, Rocha O, dos Santos BV, Bertoncini S Iwan, Eidhammer O
(2021) Proposing a tool for assessing the level of maturity for the engagement of urban freight
transport stakeholders: a comparison between Brazil, Norway, and Poland. Sustain Cities Soc
72:103047. https://doi.org/10.1016/j.scs.2021.103047
13. Warnecke D, Wittstock R, Teuteberg F (2019) Benchmarking of European smart cities – a
maturity model and web-based self-assessment tool. Sustain Account, Manag Policy J 10:654–
684. https://doi.org/10.1108/SAMPJ-03-2018-0057
14. Strasser M, Weiner N, Albayrak S (2016) A maturity framework to evaluate smart cityservice
solutions. In: Proceedings of the 18th mediterranean electrotechnical conference. Cyprus, 18–
19
15. Visan M, Negrea SL, Mone F (2021) Towards intelligent public transport systems in Smart
Cities; Collaborative decisions to be made. In: Procedia computer science. Elsevier B.V., pp
1221–1228
16. Shen L, Du X, Cheng G, Wei X (2021) Capability Maturity Model (CMM) method for assessing
the performance of low-carbon city practice. Environ Impact Assess Rev 87:106549. https://
doi.org/10.1016/j.eiar.2020.106549
17. Nuraeni A, Munandar A, Sandhi Firmansyah H et al (2019) Smart city evaluation model in
bandung. West Java, Indonesia, pp 228–234
18. Bernal WN, Espitaleta KLG (2021) Framework for developing an information technology
maturity model for smart city services in emerging economies: (fsce2). Appl Sci (Switzerland)
11:10712. https://doi.org/10.3390/app112210712
19. Rotta MJR, Sell D, dos Santos Pacheco RC, Yigitcanlar T (2019) Digital commons and
citizen coproduction in smart cities: assessment of Brazilian municipal e-government platforms.
Energies (Basel) 12:2813. https://doi.org/10.3390/en12142813
20. Korachi Z, Bounabat B (2018) Data driven maturity model for assessing smart cities. In: ACM
international conference proceeding series. association for computing machinery, pp 140–147
21. Lnenicka M, Nikiforova A, Luterek M et al (2022) Transparency of open data ecosystems
in smart cities: definition and assessment of the maturity of transparency in 22 smart cities.
Sustain Cities Soc 82:103906. https://doi.org/10.1016/j.scs.2022.103906
22. Jāmi‘at al-Bah.rayn (2019) Systematic literature review of the smart city maturity model.
In: Proceedings of international conference on innovation and intelligence for informatics,
computing, and technologies, pp 1–9
Challenges for the Implementation
of Sustainable Construction Practices
in Developing Countries: A Bibliometric
Review
Motheo Meta Tjebane, Innocent Musonda, Adetayo Onososen,
and Molusiwa Ramabodu

Abstract Developing countries face the constant challenges of implementing

sustainable practices in their construction industry. However, a literature review is
still needed to bring together the disaggregated studies and determine the devel-
opment status of the sustainable construction research field. This study explores the
challenges the construction sector faces in developing countries in integrating sustain-
ability practices into infrastructure delivery. Through a bibliometric review, this study
analyses articles from the Scopus database published between 2011 and 2021. The
bibliometric analysis focused on analysing sustainable construction research publi-
cations and citation patterns, while content analysis was used to identify the primary
levels of analysis and topical focuses. Emerging trends in the sustainable construction
research field are reported. The study also determined the most productive authors
and collaboration among authors, most productive journals, and most active countries
and publications with the highest impact on sustainable construction research. The
study’s findings reveal the level of interest in sustainable construction by academics
and industry practitioners in developing countries. This also informs on the level
of awareness on sustainable development in developing countries. Challenges iden-
tified and discussed explore themes on institution support, stakeholders’ role, lack
of regulation and legislation, and building standards and guidelines. The results are
imperative for improving the understanding of challenges to sustainable develop-
ment, which is vital to policy formulation and decision-making. The discussions
also provide solutions to sustainable development challenges and offer strategies to
drive adoption.

M. M. Tjebane (B) · I. Musonda · A. Onososen · M. Ramabodu

Center of Applied Research and Innovation in the Built Environment (CARINBE), Faculty of
Engineering and the Built Environment, University of Johannesburg, Johannesburg, South Africa
e-mail: tjebane.motheo@mut.ac.za
M. M. Tjebane
Department of Construction Management and Quantity Surveying, Mangosuthu University of
Technology, Durban, South Africa

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 109
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_9
110 M. M. Tjebane et al.

Keywords Sustainable Development · Developing Countries · Resilient ·

Responsive · Infrastructure

1 Background

Sustainability has recently gained popularity due to its positive impact on environ-
mental, social, and economic issues [1]. Within the construction industry, projects
embrace sustainability to secure economic improvement while preserving biodiver-
sity and ecology [2]. This is because the construction industry consumes a significant
amount of energy and resources, including 16% water, 25% timber, 40% raw mate-
rials, and 32% total energy consumption. Furthermore, the industry is responsible
for 40% of waste generation, and one-third of carbon emissions [3, 4] noted that the
construction industry has a critical role to play in influencing the achievement of
Sustainable Development Goals (SDGs).
Even though the construction industry is the backbone of any economy, it still
contributes significantly to emissions by consuming energy and producing the
majority of waste [5]. Given the global climatic challenges, sustainable construc-
tion practices are the only solution. Many sustainable construction projects are being
built worldwide, mainly in developed countries, because of their enormous benefits
[6]. However, in developing countries, implementing sustainable construction poses
numerous challenges.
Various studies have been conducted on the challenges and barriers of sustainable
construction practices in various countries using different methods [7]. Looking at
green building barriers in Taiwan using a mixed approach of questionnaires and case
studies. [8]conducted a quantitative study on Barriers to Sustainable Management
of Construction and Demolition Waste in Developed and Developing Countries. The
study Ranking of the barriers was carried out using the Relative Importance Index
(RI), and the results were statistically analysed using Statistical Package for Social
Sciences (SPSS). [9] study looked challenges of sustainable construction delivery in
the United Arab Emirates (UAE). A quantitative approach questionnaire survey was
developed and distributed to construction professionals in the UAE to evaluate the
significance of sustainable construction project delivery challenges.
These studies provide insight into the challenges of sustainable construction. Few
have a quantitative impact regarding the amount of work conducted worldwide on
the challenges and barriers. Thus, this study aims to conduct a bibliometric analysis
of the challenges of sustainable construction practices in developing countries. This
is accomplished through a bibliometric review of articles published and indexed
on Scopus between 2011 and 2021. The following is how the article is structured:
Sect. 2 discusses the research design and methodology, while Sect. 3 discusses the
bibliometric findings. The findings are discussed in Sect. 4. Finally, Sect. 5 contains
the conclusions.
Challenges for the Implementation of Sustainable Construction … 111

2 Research Design and Method

This study used a quantitative bibliometric approach to fulfill the research objectives.
A systematic approach was used in scoping and selecting publications. The flowchart
in Fig. 1 below shows the research method used.
A bibliometric visualisation is a useful tool for visualising the domain of infor-
mation as well as the relationships between papers, and journals, among others [10].
Based on established literature, bibliometric mapping is used in this study to classify
knowledge domains and research patterns for challenges of sustainable construction
in developing countries. in the AEC industry. [11] noted that bibliometric survey
aids in understanding globally published research by storing information such as
publication year, country, author, journal of publication, and affiliation in a single
database. Meanwhile, a systematic analysis was conducted to include a holistic view
of current research to assess knowledge gaps and predict future research directions

Fig. 1. Research flowchart

112 M. M. Tjebane et al.

[12]. Scopus database was used as the platform to source publications. Scopus is
the most widely used database in science mapping due to the belief that it contains
higher-quality data [13]. Scopus, which includes over 23,500 peer-reviewed journals.
It has over 75 million records dating back to 1788 in various subject areas such as
[14].

2.1 Inclusion and Exclusion Criteria

The articles were selected an inclusion and exclusion criteria. This eligibility criterion
determines whether the titles, abstracts, years, and keywords of the publications read
are part of the required set for this review [15]. [16] noted that researchers avoid
bias by establishing exclusion and inclusion criteria in advance. In general, inclusion
criteria serve as the foundation for including articles that provide the information
needed to answer the research question. While exclusion criteria include the basis
for excluding non-relevant articles, duplicate articles, and articles with no available
full-text. The inclusion criteria looked at the following:
• Publications made between 2011 and 2021
• Articles and review articles
• Published in the English language

2.2 Search String

Because of the rapid advancement of technology, bibliometric analysis can now be

performed using various existing software. A list of keywords relevant to sustainable
construction practice was created. The query string was: sustainable AND construc-
tion AND developing AND countries AND challenges. The Boolean operator AND
ensured that every publication containing the keywords was extracted from Scopus
[15].

2.3 Data Analysis

Because of the rapid advancement of technology, bibliometric analysis can now

be performed using various existing software. VOSviewer is a free visualisation
software based on the Java platform that can be used to create accessible maps
using bibliographic data. It is far more effective and powerful in data processing and
map generation than other visualised tools. It allows researchers to quickly identify
hotspots and developments in their research field of interest [17]. VOSviewer was
chosen for this study to draw science mappings because it has exceptional text-
mining capabilities and is well-suited for dealing with larger networks. VOSviewer
Challenges for the Implementation of Sustainable Construction … 113

300
274
250
220
200

150
132
100
66
50 44
1 1 0 1 0 1 1 19 2 3 7 11
0 0 0 0 2 10 4
2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
Year

Publications Citations

Fig. 2. Publication verses Citation from 2011 to 2021

is currently being used in construction industry research to create science mappings

for various topics [15] and analyse the mapping analysis, bibliographic coupling,
keyword co-occurrences analysis, and co-authorship [18].

3 Results

3.1 Descriptive Analysis

From 2011 to 2021, 28 documents on sustainable construction practices in developing

countries were published on Scopus. The distribution of publications and citations by
year. This is shown in Fig. 2 below. Five papers on sustainable construction practices
in developing countries were published in the first seven years. The rate of article
publication has increased at a faster rate since 2018. Overall, the annual publications
on SEC in machining operations have been steadily increasing since 2020.

3.2 Country Analysis

China accounted for the greatest number of publications with six articles. Australia
followed this with five and the United Kingdom with four. Six countries had only
one publication each. In terms of citations, India had the highest with 298, this is
followed by China with 240 and Australia with 99. Ghana accounted for the lowest,
with 14. Figure 3 Australia has the most collaborations, followed by Iran and Saudi
Arabia (See Table 1).
114 M. M. Tjebane et al.

Fig. 3. Country collaboration network

Table 1. Country Analysis

Country Documents Citations Total Strength
India 1 298 0
China 6 240 4
Australia 5 99 8
united states 2 74 1
Germany 2 67 0
Denmark 1 64 1
Iran 3 39 6
Kazakhstan 1 27 0
New Zealand 1 27 0
Norway 1 27 0
Malaysia 2 26 0
Egypt 3 24 0
Saudi Arabia 1 20 0
United Kingdom 4 18 6
Ghana 3 14 0

3.3 Authorship Analysis

Table 2 shows the top authors in terms of number of citations. From the list of
16 authors, 6 authors have 298 citations from publishing 1 article. Furthermore, 3
authors had 99 citations from 1 published article. Figure 4 below shows the lines of
collaborations being clustered among authors.

3.4 Journals

Table 3 shows the top 22 most cited journal articles in sustainable construction
practice by total publications and citations. Construction and building materials had
the highest number of citations, with 298 citations from 1 article. The journal with the
second highest number of citations was the journal of environmental management,
with 99 citations from 2 published articles. The lowest citations came from advanced
sustainable systems, with seven citations.
Challenges for the Implementation of Sustainable Construction … 115

Table 2. Author Analysis

Author Documents Citations Total Strength
Ansari M 1 298 0
Dutta S.C 1 298 0
Joshi H 1 298 0
Kisku N 1 298 0
Nayak S 1 298 0
Panda S.K 1 298 0
Aslam M.S 1 99 4
Cui L 1 99 4
Huang B 1 99 4
Tam V.W.Y 2 82 5
Duan H 1 64 3
Liu G 1 64 3
Miller T.R 1 64 3
Manowong E 1 63 0
Kumanayake R 1 31 1
Luo H 1 31 1

Fig. 4 Author collaboration network

3.5 Keyword and Cluster Analysis

A keyword frequency visualisation sustainable practice research in contraction was

generated by VOSviewer as seen in the Table 4 and Fig. 5 Among the most frequently
mentioned keywords were developing countries (occurrence = 17), sustainable
development (occurrence = 15),construction industry (occurrence = 13), sustain-
ability (occurrence = 12), developing world (occurrence = 9), sustainable construc-
tion (occurrence = 9), surveys (occurrence = 5), waste management (occurrence =
5), construction (occurrence = 4), construction and demolition waste (occurrence =
4), developing country (occurrence = 4). Five clusters of 66 keywords were grouped.
116 M. M. Tjebane et al.

Table 3. Journal Analysis

Journal Documents Citations Total Strength
Construction and Building Materials 1 298 0
Journal of Environmental Management 1 99 1
Waste Management and Research 2 70 0
Waste Management 1 64 1
Sustainability (Switzerland) 3 32 0
Building and Environment 1 31 1
Environment, Development and Sustainability 1 27 0
Sustainable Cities and Society 1 26 0
Administrative Sciences 1 24 0
Engineering, Construction and Architectural 2 22 1
Management
Journal Of Engineering, Design and 1 20 0
Technology
Smart and Sustainable Built Environment 2 11 1
Case Studies in Construction Materials 1 10 0
Natural Resources Forum 1 10 0
Advanced Sustainable Systems 1 7 0

4 Discussion

4.1 Content Analysis

4.1.1 Cluster 1

This cluster comprised 16 keywords, including building industry, demolition, and

waste disposal, among others. From the content analysis, [19] the concept of sustain-
able cities is mainly about neutralising and conserving resources and maximising
economic benefits with minimal investment and pollution. This includes paying atten-
tion to waste management and material utilisation with all “new” buildings, regardless
of their class. In the case of “old” buildings, the highest priority is waste management,
use of materials, energy consumption, and sustainable use of ecosystems [3].
Construction and demolition waste, produced by rising demand for infrastruc-
ture and municipal growth, has expanded due to urbanisation’s effects on the popu-
lation and economy. With such a large amount of construction waste, environmental
challenges are created, such as resource depletion, loss of green space, increased air
and land pollution, and hazardous waste discharge [8]. The majority of developed
countries have environmental sustainability laws that direct infrastructure design
and force design teams to take environmental sustainability into account at the
design stage. Contrarily, insufficient ecological regulatory frameworks and a lack
Challenges for the Implementation of Sustainable Construction … 117

Table 4 Keyword cluster analysis

Cluster 1 Cluster 2 Cluster 3 Cluster 4 Cluster 5
Building Industry Affordable Architectural Article Assessment
housing design method
China barriers carbon construction Building
activities
circular economy challenges carbon dioxide construction Costs
materials
cities construction carbon construction life cycle
industry emissions waste materials
city construction construction construction life cycle analysis
sectors wastes
construction and decision design/ developing life cycle
demolition waste making methodology/ country assessment (LCA)
approach
demolition developing environmental economic and material selection
countries technology social effects
developing world drivers global warming economics scales (weighing
instruments)
economic framework green building environment social aspects
conditions
environmental green projects green buildings numerical model surveys
protection
municipal solid knowledge intelligent strategic
waste buildings approach
recycling project sustainable urbanisation
management development
refuse disposal stakeholders sustainable
practices
solid waste sustainability
waste disposal sustainable
construction
waste management

of standards for designers to follow regarding environmental sustainability goals are

prevalent in developing countries [20].

4.1.2 Cluster 2

This cluster comprised 15 keywords, such as construction sector, drivers, sustain-

ability, and stakeholders, among others. From the content analysis, [20] found the
infrastructure design stage in developing countries is where technical, functional,
and aesthetic factors take precedence over environmental considerations. Moreover,
118 M. M. Tjebane et al.

Fig. 5 Keyword cluster network

[2] found that the economic and regulatory-based hurdles more heavily influence
the failure of a change to sustainable practices in Iran. The major barriers to incor-
porating sustainability into construction projects are: (1) a lack of knowledge about
the potential benefits) a lack of collaboration between practitioners, research insti-
tutions, and environmental organisations, and a lack of a systematic strategy for
pursuing sustainability goals.
The difficulties in choosing acceptable materials for sustainable construction
projects are caused, among other things, by the high cost of construction materials
in developing nations like Ghana. This indicated the need for a proper selection of
sustainable construction materials that can promote a better quality of construction
structures, ensure faster construction solutions and foster new economic develop-
ment at better cost offers. The “sustainability” of energy efficiency and environmental
techniques for specific local situations, such as the social, economic, and ecological
components of sustainable practices, will be addressed in this choice [1]. Similarly
to a study by which found [21] most of the highly ranked barriers, such as “high-
interest rates,” “high inflation rate,” “high upfront cost of materials and sustainable
technologies,” and “high cost of serviced land.
However, developing countries can overcome this through government subsi-
dising small contractors to offset premiums, encouraging sustainable construction
in public projects, and hiring or engaging sustainability experts, among others [22].
Despite the well-known benefits of sustainable construction methods, adopting them
widely will take a lot of time and effortespecially in developing nations where the
presence of hurdles makes adoption difficult or unprofitable [3].
Challenges for the Implementation of Sustainable Construction … 119

4.1.3 Cluster 3

This cluster comprised 13, including architectural design, construction, intelligent

buildings, and sustainable development. Highlighting critical design factors for
infrastructure, such as functionality, aesthetics, and construction cost, is important
when designing with sustainability [20]. Project management key stakeholders must
proactively identify and overcome the adoption hurdles if sustainable practices are
to be incorporated into construction projects. This is due to the critical role that
project management plays in the planning, development, and execution of construc-
tion projects [2]. Also, sustainable construction practices should occur at every stage
or step. Better material selection is required to realise sustainable building prac-
tices for project delivery and achieve the desired sustainable results. This includes
using automation, energy utilisation and intelligent buildings to accelerate issues in
sustainability [23]

4.1.4 Cluster 4

This cluster comprises 12 keywords such as article, construction wastes, devel-

oping country and urbanisation [9]. The quantity and complexity of the construction
industry’s projects are growing, leading to increased environmental degradation and
greenhouse gas emissions. This has increased awareness and helped several nations,
even developing ones, act. This is due to increased construction-related issues brought
on by sustainable construction, making projects more complex. As a result, it’s impor-
tant to recognise and classify the difficulties in completing sustainable construction
projects.
The management of sustainability comes at an additional expense. Cost-saving
measures include finding the optimal routes, using waste transportation, recycling
resources, and improving quality to realise economic and environmental gains while
using less fuel [8]. [24] also noted that the primary force behind the guiding stan-
dard for decision-making in its development is the economy of sustainability. Where
the potential financial savings can be made—in the initial investment in sustain-
ability, later operation expenses, or reduced resource use—is still up for debate in
the economics of sustainable construction. These could all be influenced by the nation
and its stakeholders.
However, developing countries frequently struggle to achieve sustainable devel-
opment due to balancing economic development and environmental quality while
simultaneously enhancing social welfare. Because of benefits like lowering green-
house gas emissions and energy consumption, especially in developing countries
where environmental pollution has been a big issue, sustainability will undoubtedly
become increasingly popular in the future [7].
120 M. M. Tjebane et al.

4.1.5 Cluster 5

Cluster 5 comprises 10 keywords, including life cycle assessment (LCA), material

selection, scales (weighing instruments) and social aspects. [20] found the main
barrier to using sustainable building methods is their higher cost as compared to
conventional ones. Along with other elements like the absence of historical data or
information, the method used to choose a pricing strategy, the possibility of unex-
pected costs, the level of staff training and development, and others, this is one of
the contributing factors [1].
Infrastructure operation and maintenance will be more sustainable if a sustainable
material selection is used. This maximises the sustainability of the whole project life
cycle [25]. The necessity and value of sustainability reside in the fact that construc-
tion managers are the ones who make most choices throughout any infrastructure
project. Together with other experts, they choose the specific design and/or building
technique to apply. Therefore, recognising the larger, system-based picture is crucial
to successfully barriers and incorporating sustainability drivers into construction
projects [3]. Moreover, life cycle assessment is done mainly in old infrastructure and
not newly built. This is considering the high number of old buildings in urban areas
[26]

4.2 Implications

The construction industry is among the many sectors adopting sustainability practices
for economic, social and economic benefits. This article assessed the current research
status of sustainable construction practices in developing countries. Scientometric
methods were used to accomplish this, specifically downloading metadata informa-
tion from the Scopus index for more than 28 publications published between 2011
and 2021. The resulting dataset was examined in terms of the sources and research
areas associated with it and the affiliation between countries and international
collaboration.
The initial search results of 302 results to 28 articles after applying inclusion
and exclusion criteria. Moreover, within the ten years, the most significant jump in
publications was made in 2020, when seven articles were published. This shows little
research linked to sustainable construction practices in developing countries.
From the research on sustainable construction, the practice has grown within
the last four to 5 years, as shown in Fig. 2. Moreover, the analysis of the keyword
frequently mentioned words was developing countries, sustainable development,
construction industry, and sustainability. According to [27], to avoid subsequent
processes for managing construction projects in developing countries with limited
financial resources and limited construction management initiatives on sustainability.
Strategies should be established to take proactive steps and pay attention to existing
and anticipated conditions for construction sustainability practices. Moreover, the
sustainability criteria among developing and developed countries must be different.
Challenges for the Implementation of Sustainable Construction … 121

Regarding country analysis, China accounts for the most publications, with 6,
while India has the most citations, with 298 [28]. China and India appear to be in a
severe position from the viewpoint of environmental issues. [28] had 298 citations;
this accounted for the highest citations for country, authorship, and journal.
This article had some limitations. Only Scopus was used to gather the bibliometric
data. We also used inclusion and exclusion criteria which might have left out some
relevant articles.

5 Conclusion

This study examined the challenges of implementing sustainable construction prac-

tices in developing countries. This study provided a state-of-the-art bibliometric
review through bibliometric review. This was on the challenges of implementing
sustainable construction practices in developing countries. Future research should
expand the number of databases and publications despite the study’s limitations.
Moreover, the bibliometric review advances the body of knowledge on sustainable
construction practice.

Acknowledgements The work is part of collaborative research at the Centre of Applied Research
and Innovation in the Built Environment (CARINBE).

References

1. Kissi E, Sadick MA, Agyemang DY (2018) Drivers militating against the pricing of sustainable
construction materials: the Ghanaian quantity surveyors perspective. Case Stud Constr Mater
8:507–516. https://doi.org/10.1016/j.cscm.2018.04.003
2. Fathalizadeh A, Hosseini MR, Vaezzadeh SS, Edwards DJ, Martek I, Shooshtarian S (2021)
Barriers to sustainable construction project management: the case of Iran. Smart Sustain Built
Environ. https://doi.org/10.1108/SASBE-09-2020-0132
3. Tokbolat S, Karaca F, Durdyev S, Calay RK (2020) Construction professionals’ perspectives on
drivers and barriers of sustainable construction. Environ Dev Sustain 22(5):4361–4378. https:/
/doi.org/10.1007/s10668-019-00388-3
4. Mensah S, Nani G, Ayarkwa J (2021) Contractors’ adaptation to environmentally sustainable
construction: a micro-level implementation framework. Smart Sustain Built Environ. https://
doi.org/10.1108/SASBE-11-2020-0174
5. Norouzi M, Chàfer M, Cabeza LF, Jiménez L, Boer D (2021) Circular economy in the building
and construction sector: a scientific evolution analysis. J Build Eng 44:102704. https://doi.org/
10.1016/j.jobe.2021.102704
6. Maqbool R, Amaechi IE (2022) A systematic managerial perspective on the environmentally
sustainable construction practices of UK. Environ Sci Pollut Res 29(42):64132–64149. https:/
/doi.org/10.1007/s11356-022-20255-5
7. Shen W, Tang W, Siripanan A, Lei Z, Duffield C, Hui F (2018) Understanding the green
technical capabilities and barriers to green buildings in developing countries: a case study of
Thailand. Sustainability 10(10):3585. https://doi.org/10.3390/su10103585
122 M. M. Tjebane et al.

8. Al-Otaibi A et al (2022) Identifying the barriers to sustainable management of construction

and demolition waste in developed and developing countries. Sustainability (Switzerland)
14(13):1–17. https://doi.org/10.3390/su14137532
9. Ahmed S, El-Sayegh S (2022) The challenges of sustainable construction projects delivery–
evidence from the UAE. Architectural Eng Design Manag 18(3):299–312. https://doi.org/10.
1080/17452007.2022.2027224
10. Tjebane MM, Musonda I, Okoro C, Onososen A (2023) Artificial Intelligence (AI) in sustain-
able construction management: a scientometric review. Lect Notes in Civil Eng. 245:137–150.
https://doi.org/10.1007/978-3-030-97748-1_12
11. Hire S, Sandbhor S, Ruikar K (2021) Bibliometric survey for adoption of building information
modeling (BIM) in construction industry– a safety perspective. Arch Comput Methods Eng
29(1):679–693. https://doi.org/10.1007/s11831-021-09584-9
12. Manzoor B, Othman I, Pomares JC (2021) Digital technologies in the architecture, engineering
and construction (AEC) industry—a bibliometric—qualitative literature review of research
activities. Int J Environ Res Public Health 18(11):6135. https://doi.org/10.3390/ijerph18116135
13. Onososen AO, Musonda I, Ramabodu M (2022) Construction robotics and human–robot teams
research methods. Buildings 12(8):1192. https://doi.org/10.3390/buildings12081192
14. Amirkhani M, Martek I, Luther MB (2021) Mapping research trends in residential construction
retrofitting: a scientometric literature review. Energies (Basel) 14(19):6106. https://doi.org/10.
3390/en14196106
15. Souza L, Bueno C (2022) City information modelling as a support decision tool for planning and
management of cities: a systematic literature review and bibliometric analysis. Build Environ
207:108403. https://doi.org/10.1016/j.buildenv.2021.108403
16. Tariq S, Hussein M, Wang RD, Zayed T (2021) Trends and developments of on-site crane layout
planning 1983–2020: bibliometric, scientometric and qualitative analyzes. Constr Innov. https:/
/doi.org/10.1108/CI-02-2021-0015
17. Zhao M, Zhang H, Li Z (2022) A bibliometric and visual analysis of nanocomposite hydrogels
based on VOSviewer from 2010 to 2022. Front Bioeng Biotechnol 10(June):1–9. https://doi.
org/10.3389/fbioe.2022.914253
18. Dhiaf MM, Atayah OF, Nasrallah N, Frederico GF (2021) Thirteen years of operations manage-
ment research (OMR) journal: a bibliometric analysis and future research directions. Oper
Manag Res 14(3–4):235–255. https://doi.org/10.1007/s12063-021-00199-8
19. Kang W et al (2019) Building sustainable cities in China: experience, challenges, and prospects.
Chin J Urban Environ Stud 07(01):1940002. https://doi.org/10.1142/s2345748119400025
20. Zulu E, Zulu SL, Chabala M, Kavishe N, Chifunda C, Musonda I (2022) Infrastructure design
stage considerations for environmental sustainability in Zambia. J Eng Des Technol. https://
doi.org/10.1108/JEDT-12-2021-0742
21. Adabre MA, Chan APC (2021) Modeling the impact of barriers on sustainable housing in
developing countries. J Urban Plan Dev 147(1):1–19. https://doi.org/10.1061/(asce)up.1943-
5444.0000639
22. Hwang BG, Shan M, Lye JM (2018) Adoption of sustainable construction for small contractors:
major barriers and best solutions. Clean Technol Environ Policy 20(10):2223–2237. https://
doi.org/10.1007/s10098-018-1598-z
23. Ozturk GB (2018) The future of 3D printing technology in the construction industry: a
systematic literature review. Eurasian J Civ Eng Architect 2(2):10–24
24. Zhang Y et al (2019) A survey of the status and challenges of green building development
in various countries. Sustainability (Switzerland) 11(19):1–29. https://doi.org/10.3390/su1119
5385
25. Ismail S, Hon CKH, Crowther P, Skitmore M, Lamari F (2022) The drivers and chal-
lenges of adopting the Malaysia industrialised building system for sustainable infrastructure
development. Constr Innov. https://doi.org/10.1108/CI-05-2021-0088
26. Yılmaz Y, Seyis S (2021) Mapping the scientific research of the life cycle assessment in the
construction industry: a scientometric analysis. Build Environ 204:108086. https://doi.org/10.
1016/j.buildenv.2021.108086
Challenges for the Implementation of Sustainable Construction … 123

27. Manowong E (2012) Investigating factors influencing construction waste management efforts
in developing countries: an experience from Thailand. Waste Manage Res 30(1):56–71. https:/
/doi.org/10.1177/0734242X10387012
28. Kisku N, Joshi H, Ansari M, Panda SK, Nayak S, Dutta SC (2017) A critical review and
assessment for usage of recycled aggregate as sustainable construction material. Constr Build
Mater 131:721–740. https://doi.org/10.1016/j.conbuildmat.2016.11.029
Robotics, Automation, and Control
Structural Health Monitoring of Civil
Infrastructure using Mobile Robots

Kay Smarsly, Kosmas Dragos, Jan Stührenberg, and Mathias Worm

Abstract With the advancements in information, communication, and sensing tech-

nologies, structural health monitoring (SHM) has matured into a substantial pillar of
infrastructure maintenance. In particular, wireless sensor networks have gradually
been incorporated into SHM, leveraging new opportunities towards reduced installa-
tion efforts and enhanced flexibility and scalability, as compared to cable-based SHM
systems. However, wireless sensor nodes are installed at fixed locations and need to
be employed at high density to reliably monitor large infrastructure, which may cause
high installation costs. Furthermore, the limited power autonomy of wireless sensor
networks, installed at fixed locations for unattended long-term operation, still repre-
sents a significant constraint when deploying stationary wireless sensor nodes for
SHM. To resolve the critical constraints stemming from costly high-density deploy-
ment and limited power autonomy, a mobile structural health monitoring concept
based on legged robots is proposed in the feasibility study reported in this paper.
The study explores the feasibility of deploying legged robots for wireless SHM of
civil infrastructure, aiming to achieve insights into realizing the advantages of mobile
wireless sensor nodes in general and of legged robots in particular. As will be shown
in this paper, the legged robots, as compared to stationary wireless sensor nodes,
require a smaller number of nodes to be deployed in civil infrastructure to achieve
the same sensor information, entailing more cost-efficient – yet accurate – SHM. In
conclusion, this feasibility study represents a first step towards autonomous robotic
fleets advancing structural health monitoring.

Keywords Structural health monitoring (SHM) · quadruped robots · legged

robots · wireless sensor networks · infrastructure maintenance

K. Smarsly (B) · K. Dragos · J. Stührenberg · M. Worm

Institute of Digital and Autonomous Construction, Hamburg University of Technology, Hamburg,
Germany
e-mail: kay.smarsly@tuhh.de

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 127
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_10
128 K. Smarsly et al.

1 Introduction

The US Federal Highway Administration estimates a US$ 125 billion bridge repair
backlog in the United States [1]. The situation in other industrial countries is similar.
In Germany, almost 700 bridges are more than 100 years old, and more than 10%
of highway bridges are deficient [2]. The vulnerability of civil infrastructure due
to aging is being further exacerbated by climate change, thus highlighting the need
for timely structural maintenance, which has been increasingly relying on structural
health monitoring (SHM) [3].
Over the years, the SHM community has witnessed gradual transitions in strate-
gies and equipment adopted, following the advances in sensor technologies. Early
SHM case studies have been limited to civil infrastructure of high importance or of
strong academic interest, due to the dependence of SHM instrumentation on relatively
expensive cable-based sensors and data acquisition units. In recent years, however,
developments in sensor technologies have been offering cost-effective alternatives
for SHM, usually consisting of wireless sensor nodes [4]. Since the merits of wireless
technologies have been apparent from the very first “wireless” strategies in SHM,
practitioners have sought, on the one hand, to exploit the scalability of wireless
sensor networks towards deploying SHM systems of increasing density, in an attempt
to obtain spatially rich information on structural conditions [5]. On the other hand,
misuse of scalability of wireless SHM systems may result in particularly dense sensor
networks. Dense sensor networks, in turn, may raise questions regarding the eventual
cost-effectiveness of dense SHM deployment, the robustness of the SHM system,
and the conformity of the SHM strategy with aesthetic and operational requirements.
In recent years, representing the middle ground between obtaining spatially dense
information and utilizing the convenience of wireless technologies with minimal
deployment of equipment, the SHM practice has been drawing from the field of
robotics. At first, the notion of mobility in SHM systems had materialized in mobile
robots facilitating powering embedded remote sensors and retrieving data with close-
proximity telemetry, as reported by Huston et al. [6]. Furthermore, solutions based
on non-stationary wireless sensor nodes have sporadically been proposed in the
form of mobile SHM systems employing wheeled robots [7]. Examples include
the “robomote” introduced by Dantu et al. for augmenting static sensor networks
[8], and the prototype mobile SHM system, capable of navigating structures using
ferromagnetic materials, designed and validated by Zhu et al. [9]. The efficiency of
mobile robots for SHM lies in the capabilities of a minimum deployment of robots
(i) to scan relatively extensive areas on structures and (ii) to form ad-hoc wireless
networks for exchanging information to collaboratively analyze the structural condi-
tions [10]. However, state-of-the-art approaches on mobile SHM systems largely
rely on wheeled robots, which, in complex structures, may be unable to access areas
with impediments. In this regard, legged robots may offer an attractive alternative
for SHM, in terms of maneuverability and transversability, following up on recent
research advances in bionics [11].
Structural Health Monitoring of Civil Infrastructure using Mobile Robots 129

This paper reports on a feasibility study, proposing a mobile structural health moni-
toring concept based on legged robots, extending preliminary work published in [10]
and in [12]. Legged robots exhibit enhanced mobility, compared to wheeled robots,
and are able to access extensive locations on structures by traversing impediments,
such as stiffening plates, pipelines and electrical installations [13]. The feasibility
study essentially seeks to validate the mobile SHM system presented in the studies
mentioned above in real-world conditions, i.e. implemented in robust legged robots
and applied on a real-life pedestrian bridge. The mobile SHM system consists of
legged robots, featuring (i) a locomotion component, (ii) a processing component,
and (iii) a sensing component. The goal of the feasibility study is to showcase that the
mobile SHM system is capable of yielding the same level of information on struc-
tural conditions as a stationary wireless SHM system when applied on a real-life
pedestrian bridge. In the rest of the paper, Sect. 2 describes the mobile SHM system,
and Sect. 3 covers the validation tests conducted on a pedestrian bridge in Hamburg,
Germany. The paper concludes with a summary and an outlook on potential future
research efforts that may be undertaken to advance the concepts presented herein.

2 A Mobile Structural Health Monitoring System Based

on Legged Robots

Considering the refinement characterizing modern structural analysis methods, e.g.

using finite element models with fine meshes, it is reasonable to pursue SHM-derived
information that would at least spatially envelope structural analysis results. In this
context, the motivation behind the mobile SHM system, presented in this section,
is to maximize the information on structural conditions, while employing a SHM
strategy that is efficient, cost-effective, and least invasive. The capabilities of legged
robots to navigate extensive areas on structures and to reach locations that may be
hardly accessible to humans allow collecting elaborate fields of structural response
data that can help stakeholders make educated judgments on structural conditions.
In addition, elaborate fields of structural response data enable comparisons with
structural analysis results, which can be used for system identification and (finite
element) model updating.
From a budgetary perspective, for a mobile SHM system to be viable, only a
small number of legged robots should be used. Particularly considering the unit price
difference between legged robots and wireless sensor nodes, it follows that the cost-
effectiveness of a mobile SHM system can be ensured only if the number of legged
robots is kept to the minimum necessary for fulfilling the SHM objectives. Drawing
from common SHM practice, which largely involves analyzing acceleration response
data and computing experimental mode shapes, SHM objectives frequently require
simultaneous data acquisition from at least two locations. As a result, for the mobile
SHM system proposed in this feasibility study, two legged robots will be deployed.
Apart from collecting acceleration response data, the legged robots are capable of
130 K. Smarsly et al.

communicating with each other and of collaboratively analyzing the data to extract
information on the structure, e.g. in the form of experimental eigenfrequencies. In
what follows, the hardware and software specifications of the mobile SHM systems
are described.

2.1 Hardware Design and Implementation

The type of legged robots used for the mobile SHM system in this feasibility study
are “intelligent documentation gadgets” (IDOGs), built around Unitree’s A1 robot
model [14]. In accordance with the mobile SHM system requirements previously
described, the IDOGs are capable of multi-direction locomotion, of data acquisition,
and of on-board data processing. The hardware design implementation follows the
same reasoning as in [10] and is shown in Fig. 1.
The “locomotion component” of the IDOG, shown in Fig. 1, consists of four
legs driven by motors at three points (“hip”, “knee”, “thigh”), thus providing twelve
degrees of freedom for locomotion and ensuring advanced maneuverability. The loco-
motion component enables the IDOG to assume the “measuring posture”, essentially
lying down for conducting measurements, and the “walking posture”, i.e. standing up
and moving from one location to the other. The locomotion control is computed and
provided by the “locomotion board”, which resides in the “processing component”.
Apart from the locomotion board, an on-board computer, designated as “computing
board”, is included in the processing component for reading and analyzing data from
the sensors attached to the IDOG (included in the “sensing component”).
In this feasibility study, the sensing component encompasses a “light detection and
ranging” (Lidar) sensor that enables the IDOG to perceive its environment. In future

Fig. 1. Hardware components of the IDOG

Structural Health Monitoring of Civil Infrastructure using Mobile Robots 131

work, the Lidar sensor will enable the IDOG to autonomously navigate to measure-
ment locations. Moreover, a Bosch BMI055 inertial measuring unit (IMU) with an
integrated accelerometer is included in the sensing component for collecting acceler-
ation response data. The IMU measures at a range of ±4 g, at a sampling frequency
of 250 Hz, and with a minimum detectable acceleration of 0.98 mg [15]. As will be
shown in the validation tests, the measuring posture that the IDOG assumes ensures
that the measurements of the IMU correspond to the acceleration response data of
the structure being monitored. As a result, the acceleration response data collected
by the IDOGs are suitable for vibration-based SHM tasks, such as operational modal
analysis, using embedded software, which is described in the following subsection.

2.2 Software Design and Implementation

The IDOG leverages embedded software, flowcharted in Fig. 2 by means of the

process conducted by the mobile SHM system, designed around the “robot operating
system” (ROS) framework. The embedded software enables collecting, processing,
and analyzing acceleration response data. When beginning operation, a ROS process
is started that starts the accelerometer. Next, the IDOG assumes the measuring posture
at a measurement location, the acceleration response data is collected and stored in
a local file. Subsequently, the acceleration response data is analyzed by performing
fast Fourier transform (FFT) and peak picking, i.e. detecting “modal peaks” in the
frequency-domain representation of the acceleration response data that essentially
are frequency components exhibiting maximum amplitude in the FFT.
For perceiving the environment, two ROS processes are started. The first process
collects Lidar scan data in the form of a point cloud. The second process subscribes
to the scan data and executes a “simultaneous localization and mapping” (SLAM)
algorithm. In this study, the Google cartographer is used for creating a 2D grid map
[16]. The grid map is devised for visualization purposes, highlighting the positions of
the acceleration measurements. As discussed previously, in future work, the grid map
will be integrated into the mobile SHM system to provide perception for autonomous
locomotion of the IDOG.

3 Field Validation Tests at a Pedestrian Bridge

Validation tests within the framework of the feasibility study are conducted on a
pedestrian bridge. The purpose of the validation tests is to showcase that the mobile
SHM system is applicable in real-life conditions. Serving as a benchmark system for
comparison, a wireless SHM system comprising stationary wireless sensor nodes is
installed on the bridge. In the remainder of this section, the pedestrian bridge and the
experimental setups are described, and the test results are presented and discussed.
132 K. Smarsly et al.

Fig. 2. SHM process conducted by the embedded software designed for the IDOGs

3.1 Description of the Pedestrian Bridge and the Benchmark

SHM System

The pedestrian bridge used for the validation tests is the “Citadel Bridge” (German:
“Zitadellen-Brücke”), which spans the Lotse Canal at Hamburg, Germany. The
pedestrian bridge, shown in Fig. 3, is a relatively new structure, opened to traffic
in 2016. The main deck of the bridge comprises a welded steel trapezoidal hollow
“box” section of variable dimensions. The deck has a length of 45.5 m and a width
that varies from 3.25 m to 5.65 m. In the vertical direction, the main deck rests on
two reinforced concrete piers, and, at its ends, abuts the quay walls on each side of
the canal, essentially forming one middle span, between the two piers, and two side
spans, each between one pier and one abutment. The main part of the main deck
is capable of revolving around one of the piers, functioning as a swing bridge, to
facilitate traffic of canal vessels.
The benchmark system includes eight wireless sensor nodes of type Oracle
“Sunspots” [17]. The microcontroller of each sensor node features an ARM processor
running at 400 MHz, volatile memory of 1 MB, and flash memory of 8 MB. The
Structural Health Monitoring of Civil Infrastructure using Mobile Robots 133

Fig. 3. Views of Citadel Bridge, Hamburg, Germany

Lotseplatz

Kanalplatz
5.60
12.50 L1
L3
L5 L2
3.25

L6 L4
8.20 8.20
12.50 24.50

Fig. 4. Plan view of the measurement locations on the pedestrian bridge

sensing component of each sensor node integrates a 3-axial accelerometer, of type

MMA7455L, measuring at a selectable range between 2 and 16 g with a selectable
sampling rate up to 250 Hz. The analog-to-digital conversion supported by the
accelerometer is of 8-bit resolution. For the purposes of this study, the accelerome-
ters are placed along the longitudinal edges of the main deck, as shown in Fig. 4. In
particular, four sensor nodes are placed on the middle span, and two sensor nodes are
placed on each side span. The purpose of placing the sensor nodes along the edges
of the deck is to capture both translational and torsional vibrations. Due to the swing
function of the bridge, the left-hand side of the main deck, as shown in Fig. 4, is
separated by a full-depth gap from the rest of the deck, and, thus, exhibits indepen-
dent dynamic behavior from the rest of the bridge. Therefore, the measurements are
restricted to the middle part and the right-hand side of the deck. The wireless sensor
nodes are tasked to collect acceleration response data in the vertical (z) direction
with a measurement range of 2 g and a sampling frequency of f s = 125 Hz for a
duration of 120 s.

3.2 Validation Tests

The benchmark system and the mobile SHM system are compared on the basis of
information extracted from acceleration response data. As mentioned previously,
the information translates into “modal peaks” in the FFT amplitude spectra of the
134 K. Smarsly et al.

Fig. 5. Visualization of the pedestrian bridge using Lidar data

acceleration response data. For the benchmark system, the wireless sensor nodes
autonomously estimate modal peaks and communicate the respective frequencies
to a centralized server. For the mobile SHM system, a minimum deployment of
two IDOGs is adopted to progressively access all measurement locations of the
benchmark system in pairs and to collect acceleration response data. The reasoning
behind using at least two IDOGs is to ensure that the modal peaks detected in the
FFT amplitude spectrum are confirmed by two independent IDOGs from two separate
locations, thus avoiding false positive detections of modal peaks, given the output-
only nature of the SHM approach. Although the measurement locations are accessed
manually in this study, since autonomous navigation is part of future work, the
visualization of the pedestrian bridge by the SLAM algorithm showcases that the
IDOGs are capable of perceiving the surroundings of the structure through the Lidar
sensors. The measurement locations (L1…L6) are illustrated in Fig. 5. In each pair of
measurement locations, the IDOGs collect and analyze the acceleration response data
in the frequency domain to obtain modal peaks, which are wirelessly communicated
to a centralized station.
The IDOGs are positioned at the abutment on the right-hand side of the main deck.
Since the analysis of one of the IDOGs serves as confirmation of the modal peaks,
one of the measurement locations is selected as a “reference point”, against which
the analyses from the rest of the measurement locations are compared. Location L4
is selected as reference point based on engineering judgment, because most vibration
modes are expected to exhibit discernible amplitudes. One of the IDOGs (“IDOG
1”) is placed at the reference point, and the other IDOG (“IDOG 2”) progressively
accesses the rest of the measurement locations by assuming the walking posture,
shown in Fig. 6 (bottom left). Upon reaching a measurement location, the IDOG
switches to measuring posture, as shown in Fig. 6 (bottom right). Simultaneously,
the IDOG at the reference point switches to measuring posture, as shown in Fig. 6
(top), and both IDOGs collect acceleration response data for a duration of 120 s at
a sampling frequency of f so = 250 Hz. Next, each IDOG transforms its accelera-
tion response data into the frequency domain and detects modal peaks, using the
embedded FFT and peak-picking algorithm. Finally, the IDOGs exchange the modal
Structural Health Monitoring of Civil Infrastructure using Mobile Robots 135

Fig. 6. Simultaneous collection of acceleration response data at measurement locations L3 and L4

(top), walking posture of the IDOG (bottom left), measuring posture of the IDOG (bottom right)

peaks to confirm that the eigenfrequencies of the modal peaks detected by each
individual IDOG match.

3.3 Results and Discussion

The results of the embedded peak-picking algorithm, executed autonomously by

each IDOG, are summarized in Tables 1, 2 and 3. First, the results of IDOG 1 are
presented, next, the results of IDOG 2 are provided, and, finally, the results from
applying peak picking at the acceleration response data of the benchmark system are
tabulated for comparison purposes.
As can be seen from Tables 1, 2 and 3, the embedded peak-picking algorithm of
the IDOGs is capable of yielding 2–3 modal peaks at the same frequency components
as the benchmark system. Minor discrepancies between the modal peaks detected
by IDOG 1 and IDOG 2, as well as between the IDOGs and the benchmark system,
are attributed to interference of noise and to the approximative character of the FFT.
136 K. Smarsly et al.

Table 1. Peak picking results from IDOG 1

Setup Location Modal peaks (Hz)
1 2 3
1 L4 3.082 4.333 10.544
2 L4 3.067 4.318 10.590
3 L4 3.052 4.303 10.544
4 L4 3.082 4.288 10.437
5 L4 3.082 4.257 10.483

Table 2. Peak picking results from IDOG 2

Setup Location Modal peaks (Hz)
1 2 3
1 L1 3.113 4.303 -
2 L2 3.082 4.349 -
3 L3 3.113 4.303 10.620
4 L5 3.098 4.318 10.590
5 L6 3.113 4.303 10.590

Table 3. Peak picking results

Location Modal peaks (Hz)
from the benchmark system
1 2 3
L1 3.036 4.356 -
L2 3.029 4.333 -
L3 3.044 4.356 10.544
L4 3.036 4.341 10.597
L5 3.044 4.349 10.437
L6 3.036 4.326 10.597

Exemplarily, the FFT amplitude spectra computed by both IDOGs for setup 3 are
plotted in Fig. 7a. Furthermore, Fig. 7b illustrates the FFT amplitude spectra between
IDOG 1 and the corresponding sensor node of the benchmark system at location L3.
In both plots, the proximity between the FFT amplitude spectra is evident.
Structural Health Monitoring of Civil Infrastructure using Mobile Robots 137

60 60
IDOG 1 Benchmark

Amplitude (g)
IDOG 2 IDOG 2
Amplitude (g)

40 40

20 20

0 0
0 5 10 15 20 0 5 10 15 20
Frequency (Hz) Frequency (Hz)
(a) (b)

Fig. 7. Comparison between FFT amplitude spectra of IDOG 1 and IDOG 2 for setup 3 (a), and
comparison between FFT amplitude spectra of IDOG 2 and the benchmark system at location L3
(b)

4 Summary and Conclusions

The advantages of wireless sensor networks for structural health monitoring, in terms
of cost-efficiency and flexibility, may be easily nullified by designing particularly
dense wireless SHM systems consisting of stationary wireless sensor nodes. As alter-
natives, approaches proposing mobile wireless SHM system have been sporadically
proposed. Nevertheless, since the concept of mobile SHM is still in its infancy, this
feasibility study has focused on highlighting the applicability of mobile SHM in
real-world conditions. Specifically, building upon previous work of the authors, a
mobile SHM system, comprising legged robots, which can advantageously navigate
areas with impediments that are hard to reach by wheeled robots, has been tested
on a pedestrian bridge. The purpose of the feasibility study has been to investigate
the capability of the mobile SHM system to (i) perceive its surroundings and (ii)
yield information on the structural condition similar to the information obtained by a
benchmark system, consisting of stationary wireless sensor nodes. On the one hand,
the juxtaposition of the visualization of the pedestrian bridge with the actual geom-
etry of the bridge has showcased the capability of the legged robots to perceive the
surroundings. On the other hand, the comparison between the outcomes of the mobile
SHM system, in the form of eigenfrequencies obtained through peak picking, and
the corresponding outcomes of the benchmark system have proven the capability
of the mobile SHM system to provide rich information on the structural condition.
Future work will focus on using the mobile SHM system to obtain elaborate informa-
tion on the structural dynamic behavior, such as mode shapes, and on implementing
autonomous navigation of measurement locations.

Acknowledgements The authors would like to gratefully acknowledge the support offered by
the German Research Foundation (DFG) under grant SM 281/20-1. The authors would also like to
thank the Landesbetrieb Straßen, Brücken und Gewässer (LSBG) of Freie und Hansestadt Hamburg
(FHH), who generously authorized the field test at Citadel Bridge. Any opinions, findings, conclu-
sions, or recommendations expressed in this paper are those of the authors and do not necessarily
reflect the views of DFG and LSBG.
138 K. Smarsly et al.

References

1. American Society of Civil Engineers (2021) Infrastructure report card: Bridges. Available at:
https://infrastructurereportcard.org/wp-content/uploads/2020/12/Bridges-2021.pdf. Accessed
on 06 May 2022
2. Delhaes D (2021) Bridges at risk: The renovation backlog continues [Brücken in Gefahr:
Der Sanierungsstau hält an]. Available at: https://www.handelsblatt.com/politik/deutschland/
investitionen-bruecken-in-gefahr-der-sanierungsstau-haelt-an/27623112.html. Accessed on 06
May2022
3. Farrar CR, Worden K (2007) An introduction to structural health monitoring. Phil Trans R Soc
A 365(1851):303–315
4. Lynch JP, Loh KJ (2006) A summary review of wireless sensors and sensor networks for
structural health monitoring. Shock Vib Digest 38(2):91–130
5. Smarsly K, Petryna Y (2014) A decentralized approach towards autonomous fault detection in
wireless structural health monitoring systems. In: Proceedings of the 7th European Workshop
on Structural Health Monitoring (EWSHM) 2014. Nantes, France, 07/08/2014
6. Huston DR, Esser B, Gaida G, Arms S, Townsend C (2001) Wireless inspection of structures
aided by robots. In: Proceedings of the 6th Annual International Symposium on NDE for Health
Monitoring and Diagnostics. Newport Beach, CA, USA, 03/04/2001
7. Ramasamy V (2017) Mobile wireless sensor networks: an overview. In: Sallis PJ (ed) Wireless
Sensor Networks – Insights and Innovations. IntechOpen Ltd., London, UK, pp 3–19
8. Dantu K, Rahimi M, Shah H, Babel S, Dhariwal A, Sukhatme GS (2005) Robomote: enabling
mobility in sensor networks. In: Proceedings of the 4th International Symposium on Information
Processing in Sensor Networks. Los Angeles, CA, USA, 04/25/2015
9. Zhu D, Yi X, Wang Y, Lee K-M, Guo J (2010) A mobile sensing system for structural health
monitoring: design and validation. Smart Mater Struct 19(5):55011–55021
10. Smarsly K, Worm M, Dragos K (2022) Design and validation of a mobile structural health
monitoring system based on legged robots. In: Proceedings of the 29th International Workshop
on Intelligent Computing in Engineering (EG-ICE). Aarhus, Denmark, 07/08/2022
11. Zhong Y, Wang R, Feng H, Chen Y (2019) Analysis and research of quadruped robot’s legs: a
comprehensive review. Int J Adv Robot Syst 16(3):172988141984414. https://doi.org/10.1177/
1729881419844148
12. Smarsly K, Worm M, Dragos K, Peralta J, Wenner M, Hahn O (2022) Mobile structural
health monitoring using quadruped robots. In: Proceedings of the SPIE Smart Structures/NDE
Conference: Health Monitoring of Structural and Biological Systems. Long Beach, CA, USA,
03/06/2022
13. Biswal P, Mohanty PK (2021) Development of quadruped walking robots: a review. Ain Shams
Eng J 12(2):2017–2031
14. Unitree Robotics (2020) A1 robot. Available at: https://www.unitree.com/products/a1/.
Accessed on: 23 Jun 2022
15. Bosch Sensortec GmbH (2021) BMI055 Sensor data sheet. Available at: https://www.bosch-sen
sortec.com/media/boschsensortec/downloads/datasheets/bst-bmi055-ds000.pdf. Accessed on:
23 Jun 2022
16. Hess W, Kohler D, Rapp H, Andor D (2016) Real-time loop closure in 2D LIDAR SLAM. In:
Proceedings of the IEEE International Conference on Robotics and Automation. Stockholm,
Sweden 05/16/2016
17. Oracle Corp (2010) Sun SPOT eDEMO Technical Datasheet, 8th edn. Sun Labs, Santa Clara,
CA, USA
Bottlenecks to the Implementation
of Automation and Robotics
in the Construction Industry

Ayanda Boya, Opeoluwa Akinradewo, Clinton Aigbavboa,

Andrew Ebekozien, and Molusiwa Ramabodu

Abstract South Africa is increasing its use of technological innovation in the

construction sector. Given the recent advancements, understanding how tech-
nology affects the construction industry is crucial. Industry related concerns include
decreased quality and productivity, wastage, accidents, project delays, and unfavor-
able working conditions. A solution to these issues is the adoption of the implemen-
tation strategies of automation and robotics technologies, which have the potential to
increase productivity, safety, and quality in the construction industry. However, there
are barriers that hinder the progression of this implementation. This study aimed at
evaluating the challenges of the implementation strategies of automation and robotics
in the South African Construction industry. Corresponding literatures were reviewed,
and a quantitative research methodology was adopted to extract information about
the research question. To achieve this, a cluster sampling technique was used, and
a survey questionnaire was distributed to 111 construction professionals, who were
situated in the Gauteng province of South Africa. The data collected presumed that
the most impactful challenge to the implementation of automation and robotics in the
construction industry is the high costs of acquiring technologies. This could be based
on existing corroborations that SMME’s and other small organisations struggle to
obtain technologies due to lack of funding, and technologies are generally deemed
to be very costly. The study concluded that innovative technologies are generally
expensive and not easily attainable.

Keywords Automation · Construction Industry · 4IR · Robotics · Technology

A. Boya (B) · O. Akinradewo · C. Aigbavboa · A. Ebekozien

Cidb Centre for Excellence and Sustainable Human Settlement and Construction Research
Centre, Faculty of Engineering and the Built Environment, University of Johannesburg,
Johannesburg, South Africa
C. Aigbavboa
e-mail: caigbavboa@uj.ac.za
M. Ramabodu
Department of Construction Management and Quantity Surveying, Faculty of Engineering and the
Built Environment, University of Johannesburg, Johannesburg, South Africa
e-mail: molusiwar@uj.ac.za

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 139
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_11
140 A. Boya et al.

1 Introduction

Automation and robotics in construction is the mechanization and digitalization of

human skills to aid in productivity as well as to resolve issues related to labor scarcity,
safety factors etc. [1]. The rise of the industrial revolution 4.0 (4IR) has assisted in
making people more mindful of the convenience of implementing automation and
robotics in their daily tasks [2]. Aghimien [2] stated that the implementation of
automation and robotics in the building industry has been deemed non-progressive,
considering the fact that it has the ability to mitigate the dangers posed to human
health and safety. Without the use of building technologies, efficiency is minimal
[3]. Construction has been described as an industry with a higher rate of workplace
injuries and fatalities with building collapse and falls from great heights being the
leading cause of deaths. Furthermore, construction automation has shown to be neces-
sary for carrying large materials thus leading to injury prevention [4]. Thoroughly
embracing technologies such as robotics and building automation is the solution
to labour shortages, workplace health and safety, and a decline in efficiency and
productivity [5].
Conversely, robotics necessitates the ability of technicians to handle or monitor
the robots, so expertise of robots would be essential in order to work with them.
This implies that any time new equipment is introduced to the site; the operators will
need to be professionally trained to run the equipment. Automation is the process
of digitizing machines to execute certain tasks, thus academic knowledge would be
needed for the user to comprehend the equipment’s functionality [6]. Opposition to
robotics and building automation is also associated with a high installation cost, as
well as the cost of operating machinery. Construction expectations, such as improved
job efficiency, necessitate the operation of hefty materials or heavy-duty machines,
which, when influenced by the job description, can be quite costly [7]. Labourers fear
job loss as a result of the introduction of robotics and automation in the construction
industry [8]. As a result, automation becomes a deterrent because if robots are the
new construction technology, job loss is inevitable.
Although there are many benefits of automation and robotics, there are still
numerous implementation challenges. The main objective of this study is to inves-
tigate these implementation bottlenecks within the construction field. This research
will expose professionals and students to the barriers of implementing automation
and robotics in the South African construction industry. The findings that have been
presented will assist in creating strategies to mitigate these barriers.
Bottlenecks to the Implementation of Automation and Robotics … 141

2 Automation and Robotics in the Construction Industry

Automation and Robotics is the application of self-controlled mechanical and elec-

tronic machines with smart regulatory systems to automate building tasks and
processes [9]. The words and terminology are used interchangeably; and they
comprise systems ranging from human-operated mechanical machinery to partially
autonomous or autonomous equipment that are remote controlled, to automated
robotics with additional sensors and automatic monitoring functions. With regards
to this research, construction automation and robotics depicts a broad category of
machinery and applications used to automate construction-related tasks across the
project’s life cycle [9]. As maintained by some scholars, “construction automation”
depicts the use of designs assisted by computers and worksite technology handled
by robots to simplify all-inclusive tasks [10]. Currently the phrase “digital fabri-
cation” in substitution for “construction automation” is used when discussing the
customization of building construction. Construction robotics is explained as “a re-
programmable, multifunctional manipulator designed to move materials, parts, tools
or special devices through variable programmed motions for the performance of a
variety of tasks occurring in production processes of the building industry” [11].

2.1 Bottlenecks to the Implementation of Automation

and Robotics in the Construction Industry

Model Constraint
Due to automation still prevailing in the beginning stages of design and develop-
ment, there is a designer-operator divide in application [12], and there is a scarcity of
certified and relevant models that constitute the foundation of the structure of compu-
tative networks and hardware components. There is a lack of globally approved and
validated models that can be used to a variety of enterprises with diverse IT enabled
applications and operational capabilities [13]. The automation restrictions are deter-
mined by unstructured government mechanisms, non-existing government policies,
initiatives, and studies. Furthermore, models that are approved across a sector must
be reconditioned and adjusted in accordance with the types, stages, and degrees of
automation, i.e. including industry-approved manuals [14].

Money Constraint
The combination of hardware, software, and human-resource expenditures is referred
to as the monetary costs of automation. The fundamental expenditures of plan-
ning automation are lack of funding for automation or significant capital investment
for formalization, servicing, personnel training, professional partnership, specialist
teams for organizing periodic tasks and activities, computational (hardware) facili-
ties, and software development [15, 16]. In general, automation finance is too difficult
to grasp and implement due to the numerous expenses connected with automation
142 A. Boya et al.

testing [17]. There is resistance for building automation by entrepreneurs or small

businesses since they are not acknowledged by the market, meaning that stronger
businesses have a better chance of buying or hiring heavy equipment [18]. Some
studies go as far to say that the problems of robotic and automated installations
include expensive updates as technology improves as well as developing modern
technologies to complement the task since not all construction methods are the same
[19].

Management Constraint
There is a deficit in managerial initiative to bring about the breakthrough incor-
porated in the automation process from conception to application [20]. In addition,
managerial reluctance and openness towards change are the missing connection with
ever-changing businesses and competitive markets [21]. A few of the management-
oriented limitations of automation implementation include a lack of trust in automa-
tion systems, machine-managed operations imposed by political, legal, moral, and
societal pressures [22], privacy and security issues, lengthy difficulties of men-
machine task synchronization, and ever transpiring machine errors [23]. In contrast
with manual labour, construction automation and robotics cannot accept true owner-
ship and accountability for tasks, particularly in cases of incidents or error, which is
also a major management challenge, because of its “unstructured” and “uncontrolled”
characteristics [24].

Men-Machine Interaction
Operators must reskill in certain areas to transition from leading machines to working
on them as assistants or supervisors [25, 26]. The willingness of human operators
to employ the “learn-unlearn-relearn” methodology through comprehensive training
courses can minimize automation problems and further develop the performance
of the operator [27]. In addition, the connection between human and machines is
often mentioned as the main limitation accelerating automation system failures since
development engineers have never foreseen complicated system interactions [28].

Measurement Constraint
The lack of tools and methodologies for analyzing the effectiveness that auto-
mated processes, functionalities, and operations have on organizational workers and
revenues is a measurement restriction of automation. Several studies have studied
the costs assigned to automation stakeholders by automation equipment, employee
restructuring and training, software upgrades, as well as the contribution or benefits
each of these changers makes to companies [29]. The lack of effect measuring tools
against the starting automation process seems to be the most significant absent link
between automation idea and adoption, as well as post-implementation [30].

Mechanism Constraint
This constraint of automation leads us to the lack of literature on “who, when, where,
why, and how” following the automation process, as well as which processes must
be implemented for effective deployment or operation, which appears to be either
Bottlenecks to the Implementation of Automation and Robotics … 143

non-existent or unverifiable [31]. The lack of knowledge exchanging conferences,

forums, conventions on automation also limits business fraternity. Automation initia-
tives are very complicated, requiring planning and management on several levels,
including technological, administrative, clinical, social, and economical [32]. The
lack of support for government changes and cooperative machinery from the frame-
work of specialized automation stakeholders is even more concerning. The mecha-
nism for effectively testing and implementing automation is hampered by firms’ poli-
cies of concentrating their technological resources for operational efficiency alone
rather than generalizing them towards widespread benefits and influence [33].

Employment
According to James [34], the construction industry employs a substantial number
of people in the country. Majchrzak [8] stated that employees are afraid of job loss
potentially caused by the construction sector robots and automation, which makes it
a challenge to automation, because robots are the future construction technology they
would presumptuously create job losses. Full adaption of robotics in construction will
affect most jobs, since the accessibility of work will be reduced. Only occupations
of the people who perform robotic tasks will be lost, as machines may undertake
special tasks with the advantages of higher production and improved output. The
implementation of robots will substitute skilled people and employees impacted by
use of robots and automation will have to alter the occupation in which they are
familiar with, for instance with bricklayers, because it is envisaged that robots that
lay bricks will be used as a specialty in construction [35].

Robots and Automated Systems

Drones and autonomous vehicles
There are several obstacles that have to be confronted, such as i) high capital cost;
(ii) insufficient battery power which limits operations - for example, the majority
of drones have a flying time of not more than thirty minutes; (iii) sophisticated
hardware and software operational capability that requires considerable training and
increases expenses; (iv) apparent mistakes and accidents to the degree of precision
and acceptance; (v) strict restrictions that raise acquisition costs; and (vi) extra health
and safety hazards caused by drones [38, 39].

Exoskeletons
The perceived challenges to employing exoskeletons in the construction business
include i) concerns about safety and health, since exoskeletons might raise “catch”,
“snag”, and “fall” hazards; cleanliness difficulties and a false feeling of safety; (ii)
issues about usability, that include lifespan, robustness and adaptability; iii) The
failure to integrate additional protective equipment (PPE); (iv) the adoption’s capital
costs and the necessity for a speedy rate of return; (v) a key obstacle is the possibility
of poor worker acceptance rates [40].

New Work Methods

Most construction workers do not get advanced learning, they are taught to become
experts in particular craft on the job. The majorities of construction employees are
144 A. Boya et al.

trained in single craft and have been specializing in this craft for years, making
training difficult for employees because they have limited academic foundation.
Managers might consequently distrust the competence of personnel to manage new
robots and building automation [35]. Automation is all about computerization for
specific operations and the operator needs educational training to comprehend the
operation of the machine [6].

3 Research Methodology

The quantitative research approach was used for this study. A deductive logic
approach was implemented where surveys/questionnaires were gathered and issued
to collect data. The explorative technique was most suitable with comparing the
relationship between variables. The research was conducted in Gauteng, South
Africa. Construction organizations as well as professionals (Quantity Surveyors,
Construction Project Managers, Civil Engineers and Architects) in the engineering
and construction sector were targeted respondents. The type of sampling employed
for this research was random cluster sampling. Surveys were distributed to the
above-mentioned professionals in Gauteng. They represented organizations familiar
with automation and robotics in the construction industry. This random sampling
technique was used because it was not feasible to sample the whole population
of construction professionals in South Africa. A total of 111 valid responses were
obtained and the data collected was examined using Microsoft Excel and Statistical
Package for Social Science (SPSS). MIS – “Mean Item Score” and SD – “Standard
Deviation” were used to analyse the data that was collected. The study encountered
limitations due to the fact that the targeted number of professionals did not attend to
the survey and some individuals did not accurately complete the surveys. The relia-
bility of the results was tested using Cronbach’s Alpha coefficient, which amounted
to 0.903, meaning the findings of this study can be relied upon.

4 Findings and Discussion

4.1 Findings

The background data that was collected from the targeted respondents was analysed.
Results showed that 18.9% hold a post matric certificate or Diploma, 41.4% hold
a bachelor’s degree, 29.7% hold an honour’s Degree, 8.1% hold a master’s Degree,
and 1.8% hold a doctorate. Results also indicated that 8.1% are Architects, 49.5% are
Quantity Surveyors, 11.7% are Civil Engineers, 0.9% are Industrial Engineers, 10.8%
are Mechanical Engineers, 8.1% are Construction Managers, 8.1% are Construction
Project Managers, and 2.7% fall under other professions in the built environment.
Bottlenecks to the Implementation of Automation and Robotics … 145

Through the application of the 5-point Likert scale of ‘Strongly Disagree’ (SD) –
‘Strongly Agree’ (SA), the respondents were required to specify the extent at which
they agree with regards to the benefits of the strategies adopted in the implementation
of automation and robotics in the industry. Table 1 indicates the respondent’s rank-
ings of the challenges to the strategies adopted. The results reveal that “High costs
of acquiring technologies” was ranked first position, with a MIS of 4.50 and a SD of
0.712; “Risks of job loss caused by technologies” was ranked second position, with
a MIS of 4.35 and a SD of 0.931; “High costs of maintaining and updating technolo-
gies” was ranked third position, with a MIS of 4.30 and a SD of 0.848; “Resistance
to implementation of technologies by workers” was ranked fourth position, with a
MIS of 4.09 and a SD of 0.968; “Lack of funds required to implement technologies”
was ranked fifth position, with a MIS of 4.07 and a SD of 1.024. Additionally “Lack
of knowledge on how to operate technologies” was ranked sixth position, with a MIS
of 3.87 and a SD of 0.964; “Difficulty to acquire technologies due to unavailability”
was ranked seventh position, with a MIS of 3.77 and a SD of 1.128; “Difficulty to
understand and operate technologies” was ranked eighth position, with a MIS of 3.74
and a SD of 1.015; and lastly “Inadequate compatibility of technologies with current
practices” was ranked ninth position, with a MIS of 3.72 and a SD of 1.105.

Table 1. Bottlenecks to the implementation of Automation and Robotics

Challenges to the strategies adopted MIS SD R
High costs of acquiring technologies 4,50 0,712 1
Risks of job loss caused by technologies 4,35 0,931 2
High costs of maintaining and updating 4,30 0,848 3
technologies
Resistance to implementation of 4,09 0,968 4
technologies by workers
Lack of funds required to implement 4,07 1,024 5
technologies
Lack of knowledge on how to operate 3,87 0,964 6
technologies
Difficulty to acquire technologies due to 3,77 1,128 7
unavailability
Difficulty to understand and operate 3,74 1,015 8
technologies
Inadequate compatibility of technologies 3,72 1,105 9
with current practices
Mean Item Score (MIS); Standard Deviation (SD); Ranking (R)
146 A. Boya et al.

4.2 Discussion

On the basis of the previously mentioned findings, it can be concluded that ‘High cost
of acquiring technologies’ is the most significant challenge of the implementation
strategies. This conclusion is congruent with findings from [41], who according to
their literature ‘High cost of acquiring the construction robotics technologies’ was
ranked third out of a total of ten. The ‘Risks of job loss caused by technologies’
ranking is in agreement with literature presented by [8] that employees fear job
loss potentially caused by the construction robots and automation. ‘High costs of
maintaining and updating technologies’ is congruent with literature proposed by [19]
which relates the challenges to robot and automation installation as being expensive
to update as technology advances.
‘Inadequate compatibility of technologies with current practices’ was ranked least.
This corresponds with the literature of [41]. In this literature, the findings concluded
that “Workers are more prefer the traditional method and proven solution instead of
modern methods” was ranked least. It can be deduced that technologies are compat-
ible with current practices/modern methods. ‘Difficulty to understand and operate
technologies’ was ranked as one of the least significant challenges. This contradicts
literature from [36] that the lack of understanding material mechanical function is a
significant issue. [27] also stated that comprehensive training programs, which can
minimize the complexity of automation and increase the performance of the operator,
appear to be challenging and hypothetical.

4.3 Implication of Findings

The reason for ‘High cost of acquiring technologies’ being ranked first could be based
on existing corroborations that SMME’s and other small organisations struggle to
obtain technologies due to lack of funding and technologies are deemed to be very
costly. ‘Risks of job loss caused by technologies’ by employees may be a concern
because of the perception that their day-to-day tasks may be replaced by automation
and robotics technologies, and since their skills would not be required this could
ultimately result in retrenchment. ‘Difficulty to understand and operate technologies’
may be considered one of the least challenging strategies due to the probability that
employees have sufficient access to training in the workplace.

5 Conclusion and Further Research

This study’s research objective was established and met. The survey question-
naire demographic was restricted to professionals in the South African construc-
tion industry. As a result, the study’s findings reinforced methodological approaches
Bottlenecks to the Implementation of Automation and Robotics … 147

and offered valuable recommendations for automation and robotics implementation

approaches. The studies derived from the questionnaire based in Gauteng, South
Africa reveal that the challenges to the strategies adopted are: High costs of acquiring
technologies; Risks of job loss caused by technologies; High costs of maintaining
and updating technologies; Resistance to implementation of technologies by workers;
Lack of funds required to implement technologies; and Lack of knowledge on how
to operate technologies. A suggestion for further research or recommendation can
therefore be increase in exposure with regards to the benefits of using these tech-
nologies – further research needs to be implemented to provide more information
entailing the importance of implementing automation and robotics in the construction
industry.

References

1. Cai S, Ma Z, Skibniewski M, Bao S (2019) Construction automation and robotics for high-rise
buildings over the past decades: a comprehensive review. Adv Eng Inform 42:100989
2. Aghimien DO, Aigbavboa CO, Oke AE, Thwala WD (2019) Mapping out research focus for
robotics and automation research in construction-related studies: a bibliometric approach. J
Eng Des Technol 18:1063–1079
3. Bock T (2008) Construction automation and robotics. In: Balaguer C, Abderrahim M (eds)
Robotics and automation in construction. InTech
4. Chonnaparamutt W, Birk A (2006) Using rescue robots to increase construction site safety. In:
International Symposium of Automation and Robotics in Construction (ISARC), pp. 241–245
5. Kamaruddin SS, Mohammad MF, Mahhub R, Ahmad K (2013) Mechanisation and automation
of the IBS construction approach: a Malaysian experience. Asian Pac Int Conf Env Behav Stud
105:106–114
6. Kumar VSS, Prasanthi I, Leena A (2008) Robotics and automation in construction industry,
building integrated solution (ASCE), pp. 1–9
7. Mahbub R (2012) Readiness of a developing nation in implementing automation and robotics
technologies in construction: a case study of Malaysia. Civ Eng Archit 6(7):858–866
8. Majchrzak A (1988) The human side of factory automation. Jossey-Bass Publishers, San
Francisco, Calif
9. Mahbub R (2008) An investigation into the barriers to the implementation of automation and
robotics technologies in the construction industry. Queensland University of Technology
10. Keating S, Oxman N (2013) Compound fabrication: a multi-functional robotic platform for
digital design and fabrication. Robot Comput Integr Manuf 29(6):439–448
11. Willmann J, Knauss M, Bonwetsch T, Apolinarska AA, Gramazio F, Kohler M (2016) Robotic
timber construction — expanding additive fabrication to new dimensions. Autom Constr 61:16–
23
12. Parasuraman R, Riley V (1997) Humans and automation: use, misuse, disuse, abuse. Hum
Factors 39(2):230–253
13. Rovira E, McGarry K, Parasuraman R (2007) Effects of imperfect automation on decision
making in a simulated command and control task. Hum Factors 49:76–87
14. Parasuraman R, Sheridan TB, Wickens CD (2000) A model for types and levels of human
interaction with automation. IEEE Trans Syst Man Cybern-Part A: Syst Hum 30(3):286–297
15. Rajput PS, Gautam JN (2010) Automation and problems in their implementation: an
investigation of special libraries in Indore, India. Int J Libr Inform Sci 4:143–147
16. Boehm-Davis DA, Curry RE, Wiener EL, Harrison RL (2015) Human factors of flight-deck
automation: report on a NASA-Industry workshop. Ergonomics 26:953–961
148 A. Boya et al.

17. Hoffman D (1999) Cost benefits analysis of test automation. Star West 99:1–13
18. Kamar KAM, Hamid AZ (2009) Industrialised building system: the critical success factors and
way forward. In: 2nd CIDB.CREAM IBS Roundtable Workshop. CID, Kuala Lumpur
19. Sadique A, Mahesh G (2016) Barriers to mechanization and automation in Indian construction
industry. Int J Eng, Res, Sci Technol 3(1):11–14
20. Cascio R, Mariadoss BJ, Mouri N (2010) The impact of management commitment alignment
on salespersons’ adoption of sales force automation technologies: An empirical investigation.
Indus Market Manag 39:1088–1096
21. Barker RM, Gohmann SF, Guan J, Faulds DJ (2009) Why is my sales force automation system
failing? Bus Horiz 52:233–241
22. Wagner M (2014) The dehumanization of international humanitarian law: legal, ethical, and
political implications of autonomous weapon systems. Vanderbilt Journal of Transnational
Law, vol 47, pp 1371–1424
23. Honeycutt ED, Thelen T, Thelen ST, Hodge SK (2005) Impediments to sales force automation.
Indus Market Manag 34(4):313–322
24. Warszawski A, Navon R (1998) Implementation of robotics in building: current status and
future prospects. J Constr Eng Manage 124:31–41
25. Baldwin GB, Shultz GP (1955) Automation: a new dimension to old problems. Mon Labor
Rev 78:165–169
26. Agha SJ (1986) Constraints on library automation in Nigeria. Inform Dev 2:159–162
27. Nazir S, Kluge A, Manca D (2014) Automation in process industry: cure or curse? how
can training improve operator’s performance. 24th European Symposium on Computer aided
Process Engineering. Elsevier, pp 889–894. https://doi.org/10.1016/B978-0-444-63456-6.501
49-6
28. Bolton ML, Jimenez N, van Paassen MM, Trujillo M (2014) Automatically generating specifi-
cation properties from task models for the formal verification of human–automation interaction.
IEEE Trans Human-Mach Syst 44(5):561–575. https://doi.org/10.1109/THMS.2014.2329476
29. Poston R, Grabski S (2001) Financial impacts of enterprise resource planning implementations.
Int J Acc Inform Syst 2:271–294
30. Terwiesch P, Ganz C (2009) Trends in automation. In: Nof SY (ed) Springer handbook of
automation. Springer Berlin Heidelberg, Berlin, Heidelberg, pp 127–143. https://doi.org/10.
1007/978-3-540-78831-7_8
31. Cernetic J, Strmcnik S (1992) Automation success factors as seen from a developing country.
In: Forslin J, Kopacek P (eds) Cultural aspects of automation. Springer Vienna, Vienna, pp
34–41. https://doi.org/10.1007/978-3-7091-9220-7_9
32. Middleton SR (2000) Developing an automation concept that is right for your laboratory. Clin
Chem 46:757–763
33. Vishnoi SK, Tripathi A, Bagga T (2019) Intelligent automation, planning & implementation:
a review of constraints. Int J Emerg Technol 10(1a):174–178
34. James MW, Wong APC, Chan YHC (2006) The changing construction labour market: a case
of Hong Kong. J Eng, Des Technol 4(1):1–17
35. Navon R, Kelley PW, Fellow DW (1993) Human factors in introducing on-site construction
automation. J Constr Eng Manage 119(4):801–812
36. Wu P, Wang J, Wang X (2016) A critical review of the use of 3-D printing in the construction
industry. Autom ConStruct 68:21–23
37. Delgado Camacho D et al (2018) Applications of additive manufacturing in the construction
industry – a forward-looking review. Autom ConStruct 89:110–119
38. Li Y, Liu C (2018) Applications of multirotor drone technologies in construction management.
Int J Constr Manag 19:1–12
39. Pecoraro J, Harper C, Wang C (2017) Unmanned aircraft systems in construction and agricul-
ture: uses, benefits, challenges, and why companies choose to invest, KICEM. J Constr Proj
Manag 7:34–44
40. Kim S et al (2019) Potential of exoskeleton technologies to enhance safety, health, and perfor-
mance in construction: industry perspectives and future research directions. IISE Trans Occup
Ergon Hum Factors 7:185–191
Bottlenecks to the Implementation of Automation and Robotics … 149

41. Yahya MYB, Hui YL, Yassin ABM, Omar R, Anak Robin RO, Kasim N (2019) The challenges
of the implementation of construction robotics technologies in the construction. MATEC Web
Conf 266:05012
Simulation and Process Modeling
Integrating AEC Domain-Specific
Multidisciplinary Knowledge
for Informed and Interactive Feedback
in Early Design Stages

Julia Reisinger , Shervin Rasoulzadeh, Balint Istvan Kovacs,

Peter Ferschin, Khrystyna Vasylevska, Michael Hensel , Iva Kovacic ,
and Michael Wimmer

Abstract In the context of digitalization in the industry, a variety of technolo-

gies has been developed for system integration and enhanced team collaboration in
the Architecture, Engineering and Construction (AEC) industry. Multidisciplinary
design requirements are characterized by a high degree of complexity. Early design
methods often rely on implicit or experiential design knowledge, whereas contem-
porary digital design tools mostly reflect domain-specific silo thinking with time-
consuming iterative design processes. Yet, the early design stages hold the greatest
potential for design optimization. This paper presents a framework of a multidis-
ciplinary computational integration platform for early design stages that enables
integration of AEC domain-specific methods from architecture, engineering, math-
ematics and computer science. The platform couples a semantic integrative mixed
reality sketching application to a shape inference machine-learning based algorithm
to link methods for different computation, simulation and digital fabrication tasks.
A proof of concept of the proposed framework is presented for the use case of a
freeform geometry wall. Future research will explore the potential of the framework
to be extended to larger building projects with the aim to connect the method into
BIM-processes.

J. Reisinger (B) · S. Rasoulzadeh · I. Kovacic

Department for Integrated Planning and Industrial Building, TU Wien, Vienna, Austria
e-mail: julia.reisinger@tuwien.ac.at
B. I. Kovacs · P. Ferschin · M. Hensel
Department for Digital Architecture and Planning, TU Wien, Vienna, Austria
K. Vasylevska
Research Unit of Computer Graphics, TU Wien, Vienna, Austria
M. Wimmer
Institute of Visual Computing & Human-Centered Technology, TU Wien, Vienna, Austria

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 153
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_12
154 J. Reisinger et al.

Keywords Integrated Design · Early Design Stage · Mixed Reality Sketching ·

Shape Inference · Computational Design · Integration Platform · Digital
Fabrication

1 Introduction

Formalizing design intentions is a complex task. Design knowledge is ambiguous,

as it is based on both tacit (implicit) aspects that are difficult to codify and explicit
aspects that are codifiable (experience- or evidence-based) [1, 2]. Designers are
experts at implementing intuitive design knowledge in the process of synthesizing
“good design” – yet, it is often difficult to articulate such design decisions in explicit
terms [3]. An essential task in design science research is the representation of a design
problem and the generation and evaluation of its solutions [4]. Design processes
usually consist of two basic activities: Designing an artifact representing an improve-
ment for stakeholders (build), followed by an empirical assessment of the artifact’s
performance in a specific context (evaluate) [5]. An artifact’s performance is strongly
influenced by its environment. A lack of understanding that environment can lead
to inappropriately designed artifacts or artifacts that have unintended consequences.
The evaluation of an artifact can be complicated as its performance is dependent on
its intended use, and the intended use can include a wide range of tasks. A general
problem solving method can be applied to a wide range of problems, but its perfor-
mance varies considerably depending on the application domain and the criteria for
evaluation must be determined in advance [6]. The AEC industry is characterized by
interdisciplinary and multidisciplinary design teams. “Subjective” or non-formalized
criteria are often difficult to share with professionals from other disciplines, who do
not have the same explicit design knowledge and experience [7]. Explicit codified
knowledge has often been developed within a particular discipline and often oper-
ates within that discipline [8]. In computer-aided design, a designer’s creativity and
imagination are often limited by the capabilities of the software used, or designers
must construct workarounds or segment their process by switching between different
tools [9]. The aforementioned facts often lead to a lack of appropriate feedback
mechanisms in interdisciplinary design processes, consequently many promising
design ideas are ignored in early design stages. Yet, the early design stage holds the
greatest potential for optimization [10] and most crucial decisions are taken when
the architectural design is still in its rudimentary draft of a sketch [11].
For powerful communication at the early sketching phase, fast and easy-to-use
modeling methods are more important than the accuracy and detail of geometry.
Sketch-based design interfaces are an attractive solution for quickly modeling designs
to evaluate the overall concept before spending the effort on detailed design as they
are intuitive (pen-and-paper metaphor) [12]. Various researchers introduce novel
sketching methods for early stage 3D design exploration with sketching and gener-
ative design [13] and methods for immersive, freehand or hybrid 3D sketching in
virtual reality (VR) [14–17]. Other researchers address the sketch recognition [18]
Integrating AEC Domain-Specific Multidisciplinary Knowledge … 155

and sketch-based reconstruction and shape inference for 2D or 3D model creation

[12, 19, 20]. Such approaches enable the creation of 3D sketches followed by gener-
ation of 3D geometries, but do often not integrate interdisciplinary AEC relevant
computations or simulation methods to receive relevant performance feedback to
sketched designs.
Immediate feedback on the feasibility of a design alternative and the degree to
which design objectives are met requires constant computational reassessment of
constraints and objectives. Yet, holistic systems integration, connecting software
systems of cross-functional domains can be challenging due to numerous indepen-
dently evolving software systems [21]. Several research efforts address computa-
tional design methods for integrated design, performance optimization and deci-
sion support using parametric and generative [22–28] or BIM-based [29–31] design
tools to improve the economic, environmental or energy performance of buildings
at the early design stage. However, such computational methods are rarely suit-
able for early design sketching, as they require detailed information and modelling.
Various researches investigate the application of VR for assisted design exploration
in architectural design [11, 32], collaborative structural building and production
layout design [33] or present an iterative and interactive user-centered design system
for shared realities [34]. These approaches are remarkable but are mainly based on
systematic approaches that take into account limited knowledge about the require-
ments resulting from daily collaboration in interdisciplinary AEC design. Higher
efficiency of design processes could be achieved by extending and linking existing
methods for different design and simulation tasks at early design stage. To signifi-
cantly reduce lengthy and costly design and feedback cycles, designers should be able
to create and evaluate sketched design variants in real time to develop innovative and
well-performing solutions. The development of such a multidisciplinary sketching,
simulation and feedback platform, integrating AEC domain-specific knowledge, is
the aim of this presented research.
This research is conducted within the ongoing funded special research programme
(SFB) “Advanced Computational Design (ACD)”. ACD addresses the advancement
of design tools and processes through multi- and interdisciplinary basic research in
the areas of digital architecture, integrated building design, computer graphics and
virtual reality (VR), discrete and applied geometry, and computational mechanics.
The primary goal of ACD is integration of digital design tools and domain-specific
disciplinary knowledge from architecture, engineering, mathematics and computer
science, leading to an expansion of solution spaces and interactive feedback in early
design stages.
This paper presents a framework for a digital platform, based on a semantic
mixed reality sketching tool, allowing generation of geometry from a 3D sketch and
thus bi-directional data exchange with various simulation environments and models.
The platform, incorporating the mixed reality sketching app, enables 3D sketching,
which is a basic tool of early design; as well as simulation and real-time feedback
in the fields of geometrical form-finding, structural and material analysis, lighting
optimization, haptic feedback in VR and digital fabrication (see Fig. 1). The overall
goal is to provide immediate and informed design feedback in the sketching phase,
156 J. Reisinger et al.

Fig. 1. Diagram showing the main components of the multidisciplinary integration platform and
their relationship in a process that links design, simulations and feedback

in order to shorten design cycles and improve design quality at early design stages.
The paper provides the first proof of concept of the developed platform, performed
on the use case of a freeform geometry wall to verify the efficiency of the workflow,
identify the potential for system integration and initiate the next necessary steps.

2 Literature Review

A variety of technologies on integrated design and simulation methods for design-

space exploration and evaluation at early design stages have been developed in AEC
industry. The following summarizes the state of the art research on informed and
interactive design feedback in the early stage of sketching and highlights the new
contributions of this paper:
Three critical issues that affect the implementation of the building performance
optimization processes in the early design stage can be identified: (1) model inte-
gration, (2) real-time performance analysis, and (3) interactive optimization design
[23]. In current design practices, ideation is dominated by 2D sketches, while
disparate CAD-tools produce 3D concept models [20]. Various researchers have
introduced novel sketching methods. Barrera Machuca [14] presented Multiplanes,
a VR drawing system combining freehand sketching with the ability to draw accurate
shapes in 3D, generating snapping planes and beautification trigger points based on
Integrating AEC Domain-Specific Multidisciplinary Knowledge … 157

previous and current strokes and the current controller pose. Multiplanes automat-
ically beautifies a stroke in real-time while the user is drawing it. Kazi, et al. [13]
developed DreamSketch, a 3D design interface combining free form sketching, 3D
model generation and a generative design algorithm for computation. After sketching
a design context, the generative algorithm generates optimal solutions for the entire
design space using topology optimization. The authors highlight as limitation of the
work, that the topology optimization has not taken advantage of parallel or cloud
computing for simultaneous optimization. Arora, et al. [16] initiated SymbiosisS-
ketch, combining 3D interactions in the air with constrained sketching in 2D. Users
can create canvases and sketch on them using a tablet. The sketches are created in
the context of the physical objects in the scene, enabling quick post-processing to
seamlessly integrate the virtual objects into the real world. Drey, et al. [17] devel-
oped VRSketchIn, an immersive sketching application with unconstrained 3D air
sketching with a pen and constrained surface-based sketching on a tablet. Other
researchers address the reconstruction and shape inference integrated in sketching
methods. Yu, et al. [20] present CASSIE, where AR/VR technology is used to support
ideation and concept modeling in a shared immersive space. CASSIE combines
freehand sketching with the geometric and aesthetic constraints of 3D modeling to
produce 3D models out of surface user strokes with sufficient quality for subsequent
applications. Rosales, et al. [19] present SurfaceBrush, enabling to draw 3D strikes
using a VR brush. A surface process has been developed that converts such VR draw-
ings into user-intended manifold 3D free surfaces, providing a new approach to 3D
shape modeling. Mahoney [11] developed a system called V-Sketch, which employs
digital computation, translating from intuitive sketching to 3D geometric elements,
decoupling the sketch analysis from the creation of 3D forms by using machine-
learning techniques. The reconstruction function and a set of libraries containing
different geometric elements enable the designer to refine the solution space and
generate different results from the same sketch. A sketch-based system for quad
meshes is presented in Sketch2PQ [35], which aims to bridge the gap between
concept design and digital modeling of freeform roof-like shapes represented as
planar quadrilateral meshes.
Various research projects focus on real-time 3D environments for collaborative
design [36] not focusing on the integration of simulation and computation possibili-
ties directly in these environments. Leon, et al. [37] present a computational design
application for interactive surfaces within co-located collaborative design teams
making use of HCI. They provide an integration of different sensory modalities,
including visual and haptic. A computer-mediated environment (M.S.PixelSense)
was employed to be a facilitator for collaborative design. Full haptic feedback in
immersive VR environments, can enable teams to experience the spatial character-
istics of a room-scale design [38, 39]. Vonach et al. suggested a system utilizing
the robot actuation for haptics [40]. They showed that employment of the haptic
feedback contributes to the naturalness and more convincing experience of touch
and thus interaction with simulated virtual environment. Dumont, et al. [38] inves-
tigated real-time deformation simulation through haptic interfaces for the purpose
of the design evaluation of deformable mechanical parts in the process of industrial
158 J. Reisinger et al.

product lifecycle management. They develop an interactive mechanical simulation

framework for design improvement in virtual reality. Giunchi, et al. [39] present a
method for searching 3D model collections using free-form sketches within a virtual
environment as queries.
Optimization processes on free-form designs that are not determined a priori
by specified performance criteria often result in inadequate solutions. This can
often imply limitations in the realization process [41–43]. Regarding paneling of
architectural freeform surfaces, freeform panels generated from a set of simple-to-
manufacture geometric types have mainly been dealt with from an optimization point
of view [44]. A thorough treatment of discrete structures that underlie such arrange-
ments is investigated in various researches [44–46]. Recently, the study of mechanical
properties of sub-constructions [47] has added even more complexity to this problem.
Rigid-foldable discrete surfaces or origami is an active research area [48]. Integrating
such methods in a mixed reality sketchbook could make flexible quad-surfaces acces-
sible for designers. Tachi [49] presented a computational design method using a
restricted set of folding patterns. The web-based app “origami simulator” enables
to simulate how complicated origami crease pattern will fold in 3D [50], whereas
“crane” focuses on design, rigid folding simulation, form-finding and fabrication for
origami structures as a plug-in for Rhino/Grasshopper [51]. Recently, a framework
for the design of flat-foldable quadrilateral mesh origami approximating a general
surface was published [52]. Moreover, Jiang et al. [53] presented an optimization
technique to penalize an isometrically deformed surface with planar quads.
There is a scientific gap in linking computational methods of such geometry
optimizations to mechanical information of the underlying material as well as to
the structural behavior of the design. This has rarely been done yet, as considering
material nonlinearity leads to a sharp increase in the computational effort, and appro-
priate numerical methods [54] are still under development and often not sufficiently
stable for linking them with geometric design tools. Furthermore, the non-linear 3D
mechanical behavior of complex anisotropic material systems is not predictable to
the required extent for 3D form-finding tasks yet. Concerning the sketching systems
integrating structural or material analysis, interactive visualizations are being used to
a small degree. Murugappan and Ramani [55] present FEAsy, which is a sketch-based
interface integrating structural analysis in early design, enabling users to transform,
simulate and analyze finite element models quickly and easily through freehand 2D
sketching. A computational workflow for structural analysis feedback from architec-
tural design sketches is presented in Rasoulzadeh, et al. [56], using the Points2Surf
algorithm for shape inference [57]. Podkosova, et al. [33] present a multi-user VR
framework for early-stage collaboration, implementing a real-time bidirectional link
between a generative design model created in Rhino/Grasshopper that performs opti-
mized structural calculations of industrial building variants, and an immersive VR
environment in which the automatically calculated building variants with ecological
and economic performance feedback are visualized.
Interactive visualizations are increasingly employed in visual and haptic plan-
ning methods, mostly involving the use of immersive VR environments. Solutions
Integrating AEC Domain-Specific Multidisciplinary Knowledge … 159

allowing high-quality design interaction [58] are not widely adopted and not unbi-
ased due to the required simplifications. Furthermore, it is desirable to extend the
solution space, e.g., by optimizing light placement according to given criteria. This
has only been attempted for exterior lighting [59], while indoor lighting optimization,
which is more challenging due to diverse requirements, remains unresolved. Offering
design variations for editing operations has been shown in the domain of furniture
design [60], but not in lighting design. Haptic feedback can significantly improve task
performance and the perceived presence in a virtual environment [39]. An immer-
sive 3D sketching system in VR is provided by Oti and Crilly [15], however haptic
feedback to the 3D sketches is not included. Tactile interaction could support design
interaction through immediate haptic feedback. General-purpose haptic interfaces
proposed in literature have significant limitations for fully immersive large scale
VR systems, required for designing the extensive environments. In our presented
research we seek to improve the situation with a haptic platform employing a mobile
robot.
A similar integration goal to the one we take in our research is offered by
COMPAS [61], an open-source, python-based framework for computational research
and collaboration in architecture, engineering, digital fabrication and construction.
The framework provides geometry processing, data structures, topology, numer-
ical methods, robotics, the plugin mechanism, remote procedure calls etc. and can
be used independently of CAD systems in any environment that supports python
programming.
Several independently working software systems for 3D sketching, simulation,
calculation or digital manufacturing have been developed. However, to the best
of our knowledge, there is a lack of a cross-functional, multidisciplinary platform
that enables early design sketching, while integrating various AEC simulation and
computation methods and providing immediate performance feedback on sketched
designs.

3 Methodology

This research is conducted within the inter- and multidisciplinary research project
SFB – ACD. The SFB is structured into 8 scientific subprojects (SPs), connecting
researchers and knowledge from architecture, computer science, mathematics and
engineering. This study is conducted within subproject SP2 – Integrating AEC
domain knowledge – and explores a new design method that integrates different
methods into a digital platform for early design stages. In the following, a short
description of the research aim and goals of each subproject is given:
160 J. Reisinger et al.

Geometric form Finding Methods. Higher Order Paneling of Architectural

Freeform Surfaces (SP6) [46, 62] and Flexible Quad-Surfaces for Transformable
Design (SP7):

SP6. Investigation of methods for generation of smooth and discrete surfaces

that facilitate interesting panel shapes and/or allow the repetitive placement of
congruent as well as bendable panels.
SP7. The goal is to expand the solution space of flexible quad-surfaces and to make
it accessible for architectural applications by the development of computational
design tools.

Structural and Material Analysis Methods. Linking Mechanics to Form-Finding

of Plant-based Bio-Composite Structures (SP8) [63]:
SP8. Development of mechanical models for plant-based building materials, a
promising material category for future-proof buildings. The resulting compre-
hensive understanding of the mechanical performance of these materials should
increase the design options and enable efficient integration into design processes
through analytical models that can be evaluated quickly, as e.g. [64, 65].

Methods for Visual and Haptic Feedback. Lighting Simulation for Architectural
Design (SP4) and Large-Scale Haptic Feedback in Virtual Reality (SP5):

SP4. Development of fast differentiable global illumination solvers for modern

GPU hardware, enabling interactive visual feedback and automated lighting
design optimization, also aiming to investigate example-based and suggestive
lighting design methods.
SP5. This subproject focuses on the development of a natural haptic feedback in
virtual environment using a mobile collaborative robot (Cobot). The goal is to
provide a real-time haptics for simulation such as representation of mechanical
properties of simulated virtual structural elements and support haptic interaction
with various design elements at a large scale.

Digital and Rapid Prototyping Methods. Computational Immediacy (SP3) [66],

Material- and Structurally Informed Freeform Structures (SP9) [67, 68]:
SP3. The aim is to develop a hybrid workflow for physical/digital modeling of
architectural concept models based on 3D point clouds and machine learning. The
goal is to enable transfers of design features from learned point clouds via 2D
representations onto newly created models in creative design processes.
SP9. The goal is development of new geometrical definitions of freeform surfaces
based on their material and functional properties in order to develop strategies to
find cost-efficient freeform structures with stay-in- place and/or reusable form-
work. This will be achieved through an enrichment of computational geometrical
models with material and structural information.
Integrating AEC Domain-Specific Multidisciplinary Knowledge … 161

In this paper, we aim to answer the following research questions (RQ) regarding
the integration of the aforementioned subprojects:
RQ1: How can different design and simulation methods relevant for the AEC
from complex geometry building design, for example for panelization using spher-
ical panels or transformable surfaces made of planar or skew quadrilaterals, be
integrated in an early design stage tool to provide informed design feedback?
RQ2: How can a digital tool provide designers with an informed design feedback
in regards to computing lighting simulation for architectural design and large-
scale haptic feedback in VR so that the designer can still adapt the geometry to
provide solutions?
RQ3: How can shape-generation methods regarding geometric, material, mechan-
ical and structural constraints linking mechanics to form-finding of structures be
addressed in such digital design methods?
RQ4: How can experimental validation with real-world materials and rapid proto-
typing methods be integrated into the digital methodological approach to discover
unexpected mechanical or geometrical behavior, or verification of the properties
of structural simulations?
The methodology employed in this study is based on open-guided interviews and
a model and data analysis with the subproject researchers to analyze domain-specific
data structures and processes of each subdomain. The analysis served for identifica-
tion of discipline-specific interfaces and workflows and to express design goals and
constraints in a formal and structured way. Based on the interview results, a workflow
has been created that integrates knowledge and tools from the different computational
domains and enables the link of a semantic integration platform with feedback of
design automation, simulation, visualization and prototyping tools from the specific
research fields. The results are merged into the framework of the multidisciplinary
integration platform, coupling several concepts. The platform framework consists of
an interactive mixed reality sketching application for architectural form finding with
a shape inference module aided by machine learning enabling automated surface
mesh modelling from sketches and the created links to the specific computational
design tools. The proposed computational workflow, the integration platform and the
domain specific outputs are evaluated in the application area of theater-stage design,
using a basic element of a wall as a use case. A wall design is advantageous for
testing as it provides freeform structures with flexibility and small scale. The proof
of concept evaluates whether the application of the framework enables an adequate
data exchange and simulation with real-time feedback, verifies the efficiency of the
workflow for discipline-specific system integration and answers the above mentioned
research questions.
162 J. Reisinger et al.

4 Results – Multidisciplinary Integration Platform for AEC

This section presents the framework for a multidisciplinary integration platform that
enables simulation-based feedback on sketched designs in early stages. The inte-
grated methods address research on geometrical form finding, material and struc-
tural analysis, lighting optimization, haptic design interaction in VR and digital
fabrication.
The platform couples a semantic integrative mixed- reality sketching applica-
tion with a shape inference machine-learning based algorithm for interpretation of
geometry from sketches to link the specific computational tools and methods. The
goal is to achieve a bi-directional data exchange for real time feedback and design
optimization of sketches in future.
Figure 2 presents the framework of the multidisciplinary integration platform.
The core of the framework is the sketching application for mixed-reality design
and a machine learning-based algorithm called Strokes2Surface (S2S) for shape
inference of sketches. The specific research from the SFB subprojects providing
computer-aided simulation and feedback on the sketched designs can be catego-
rized as geometric form finding and material/structural analysis, visual and haptic
feedback, and computational immediacy through point-cloud computation.
An Integrative Mixed Reality Sketching application uses mixed reality interaction
and enables sketching directly in 3D space, capturing the process as temporal data,
and uses simulation feedback in form finding. The sketching application leverages
a tablet (Apple iPad Pro (2020)) and pen input (Apple Pencil 2) paradigm, and
the sketched designs can be used on systems with VR capabilities (Windows PC

Fig. 2. Framework of the multidisciplinary integration platform, integrating simulation feedback

from geometric form finding and material/structural analysis, visual and haptic feedback, and
computational immediacy through point-cloud computation
Integrating AEC Domain-Specific Multidisciplinary Knowledge … 163

with SteamVR support, Oculus/Meta Quest 2), further enhanced by hand-tracking

input modalities. Sketching is aided by geometric objects serving as canvases to
draw on. The sketching application forms the starting point of the design process
for later real-time simulation and collaboration. The integration platform is a set of
tools that integrate software applications or methods that are deployed in different
environments. The input data for the domain-specific computation comes from the
sketching application, while the computation-heavy tasks are mostly performed in the
software environment of the subproject. The resulting design calculation feedback
is then visualized in the sketching application (except for the digital fabrication).
Most of the other subproject’s computations and simulations rely on a structured
geometry, suggesting the necessity of translation of sketches to proper geometric
format through capturing the design intention. Thus, a shape inference algorithm
aided by machine learning employing various stylus-and-geometric-related attributes
has been conceptualized and coupled with the sketching tool, which derives a curve
network from the artist’s drawn polylines. In this setting, once the designer’s sketch
is finalized, it is converted into the desired 3D geometric format suitable for other
downstream processes.
With most of the subprojects, the communication form is geometry or 3D poly-
lines/curves. However, the model requirements vary for the different subproject
simulations. Subprojects dealing with lighting simulations and haptic feedback
in VR mainly deal with triangular meshes. Triangular meshes are also required
by the digital fabrication tools. On the other hand, there are some projects that
require other geometric formats for communication: For “Computational Imme-
diacy”, point clouds are required. Dealing with automated high-order paneling of
freeform surfaces, depends on a curve network comprising of the boundary and
paneling layout curves. Dealing with structural analysis and mechanics, volumetric
tetrahedral meshes computed from triangular meshes must be considered. To generate
and deform flexible foldable quad-meshes, three planar polylines must be drawn as
input in the sketching application – the profile, the trajectory and the prism curve.
The geometrical mesh is then computed in the computational software.
At the current state of the research, the assignment of material properties and
structural system requirements is manually conducted in the design workflow, thus
resulting in a semi-automated approach. Since most of the computational methods
and results of the different researches are highly dependent on the construction and
material types of designed sketches, a geometry- and computation-based material
ontology needs to be developed and integrated in the platform in future research to
automate the process and provide informed feedback.

5 Proof of Concept

Using a basic element of architecture – a wall – a proof of concept for the proposed
framework was conducted. In this particular use case, a wall of a theater stage design
setting was used due for several reasons: the small scale of space and short temporal
164 J. Reisinger et al.

duration offer the opportunity of testing the design in full scale utilizing the created
mixed-reality sketching tool; the absence of usual building regulations and procure-
ment procedures allows a greater design freedom and the exploration of new ways
of form finding.
A freeform wall was sketched in the mixed-reality sketching application and then
transferred to the specific subproject simulations for computation, simulation or
fabrication. Figure 3 describes the integration status achieved and the visualization
results of the individual subprojects for the wall use case.
The proof of concept demonstrated that data exchange to the different research
areas is possible starting from our sketching and integration platform. However,
manual intermediate steps are still required for post-processing or data exchange. At
the current state of the research, providing feedback in real time from the compu-
tational environment of every subproject is not possible. However, in the future,
feedback from the subprojects should be provided via a real-time data exchange
over the network. By answering the research questions in the following, the current
achievements and challenges for providing immediate feedback and automation of
calculations and simulations directly into the platform with steps for future research
are discussed.

Fig. 3. Results of the proof of concept. At the center, a sketched wall in the sketching tool and its
corresponding geometry inferred is shown, while other images in the figure depict latest integration
status with each sub-project’s specific simulations and computations
Integrating AEC Domain-Specific Multidisciplinary Knowledge … 165

The RQ1 could be answered by integrating simulation feedback of two different

geometrical form-finding methods into the integration platform. The first method is
a plug-in for transformable flexible quad-surface generation programmed in Rhino/
Grasshopper (SP7) which was integrated into the sketching application in Unity to
use the functionalities and feedback for transformable designs directly in the plat-
form. The proof-of concept demonstrates that the drawing of three polylines auto-
matically generates a transformable surface which can be manipulated directly in
the sketching application. The transformable surfaces can be used for generation of
various canvases to further sketching on. One interesting area to be investigated in
future is how different designed surfaces can be approximated by V- or T-surfaces,
which would be interesting for kinetic walls. The second simulation method finds
smooth and discrete surfaces with interesting panel shapes to sketched designs (SP6).
The proof of concept demonstrates that the enabled geometry exchange between the
integration platform and the respective simulation software of SP6 works, however,
automated immediate visualization feedback to the sketching application is not
implemented and manual geometry exchange is still necessary. In future research,
direct integration possibilities of the paneling simulation software and the generation
of different types of paneling layouts will be explored. In addition, the constructability
of the generated transformable surfaces and discrete paneled surfaces in combination
with the structural analysis is of importance in ongoing research.
To answer RQ2, a simulation method for optimization of lighting scenes in the
sketching application and a method for haptic feedback in VR to sketched designs
were integrated in the platform. A workflow has been created that allows auto-
matic lighting generation and optimization by SP4’s software system after sketching
lighting scenarios in the application by drawing lighter and darker strokes during
sketching. Thus, a way to get direct lighting feedback on sketched designs has
been achieved. To enable more accurate simulations in future, additional lighting
parameters and user input (e.g., light source position) need to be considered in the
sketching application for more accurate optimization. In addition, the integration
of material information to account for surface properties (e.g., reflectance) will be
critical for illumination simulation. The exchange with haptic feedback, SP5, is
currently in progress. The haptic subproject consists of two parts: a VR subsystem
based on Unity 3D game engine and a robotic standalone platform based on Robot
Operation System (ROS). The Unity application handles the visualization for the
user and communicates user’s actions to the autonomous mobile collaborative robot
(Cobot). The focus of the subproject is on the development of the safe and versatile
robotic platform. Based on the user’s actions, the Cobot adapts its behavior and when
needed provides the necessary haptic feedback. The role of the Unity application is
to define the parameters of interaction and communication with the robotic platform
for haptics that will be migrated to the sketching tool later on. Since both applica-
tions are developed in Unity, the integration should not be problematic. Currently, the
Unity subproject is used for testing and fine-tuning of the robotic platform in various
scenarios. The integration of the sketches and their structural deformation for haptic
feedback requires the following pipeline. The geometry of sketches is forwarded to
the structural analysis method for computation of all the possible deformations, and
166 J. Reisinger et al.

sent to SP5. One of the important challenges is to maintain the simulation in real-time
yet still get realistic results.
The RQ3 was answered by developing a workflow to receive structural feedback
to designed sketches. The users sketches a design, and receives visual structural
feedback. The structural analysis, however, is currently performed in an external
FEM software and not directly integrated, thus intermediate manual steps are neces-
sary. More parameter input possibilities such as definition of load application and
boundary condition will be necessary to integrate directly into the sketching applica-
tion for immediate structural analysis feedback. Two approaches can be distinguished
in relation to the incorporation of information about the mechanical performance
of materials into the design process. First, analytical models based on methods of
continuum micromechanics, e.g. [64, 65]. With such models, mechanical properties
of different materials can be determined very efficiently and real-time feedback on
the design process is possible. Second, numerical models, describing geometrically
more complex structures and non-linear mechanical processes [69, 70], can only
interact in real time if these simulations are restricted to a small number of degrees
of freedom and simple material behavior. An alternative is the pre-calculation of
the mechanical behavior and the generation of meta-models for certain design tasks.
This means that even more complex mechanical behavior can be integrated into the
early-stage design process in real time.
The RQ4 was answered by creating a workflow that sketched designs can be
directly printed by digital prototyping methods through geometry export out of
the platform. By integrating a novel material ontology in future research it will
be possible to also simulate and analyze some novel materials as composite clay-
mycelium within the platform. The hybrid workflow for physical/digital modeling
of architectural concept models based on 3D point clouds and machine learning does
not happen over the network, but as a functionality that is implemented and integrated
directly into the sketching application. The transfer of design features from learned
point clouds via 2D representations onto newly created models in creative design
processes will be explored in future research.
A limitation of the presented research is that currently the integration and simu-
lation feedback of every sub-project was investigated in a separate manner. The
presented workflow neglects the simultaneous presentation of simulations of several
subprojects, e.g. panelization of a surface with simultaneous structural analysis,
generation of foldable surfaces including illumination analysis, effects of haptics on
illumination optimization, haptic feedback from point clouds in VR, etc. A workflow
suitable for this purpose will be explored in the future.

6 Conclusion

This paper portrays the development of an innovative, digitally supported integrated

design framework, which allows capturing AEC-domain specific multidisciplinary
design requirements via a digital semantic sketchbook. An integrative mixed-reality
Integrating AEC Domain-Specific Multidisciplinary Knowledge … 167

sketching application transforms geometry inferred from sketches and obtained

through a machine learning-based algorithm to computable 3D geometry to enable
subdomain specific simulation, computation and fabrication. The provided frame-
work offers a multidisciplinary integration platform for design automation and simu-
lation tools, thus enabling interactive design feedback at an early design stage. The
main goal is to generate a platform in a way that promotes joint conversations that
designers need to have in the AEC industry, rather than providing a customized
solution.
The findings of the presented study show that different simulation aims need
different types of geometry and data inputs. This paper focused on geometry exchange
and the integration possibility of research in the field of geometrical form-finding,
structural and material analysis, lighting optimization, haptic feedback in VR and
digital fabrication. Discrete material information was neglected at this stage. In the
current state of the research, the integration of material information had to be added
manually in the design process and the respective computational tools. The devel-
opment and integration of a geometry and computation-related material ontology in
future research will enable the advancement of geometrical strategies for the defini-
tion of freeform surfaces based on their material and functional properties and the
automated enrichment of the computational geometrical models with material and
structural information, shortening the design process and fastening the feedback.
The proof-of-concept demonstrates that the developed design and integration plat-
form has the potential to shift early design workflows from a linear process (concept –
modeling – materialization) to a process of immediate feedback. Results reveal that
for some sub-domains, data exchange and simulation or computation currently work
semi-automatically. The presented haptic feedback VR simulation platform could
not be integrated and tested yet. Further research will focus on the improvement of
the data exchange, automation of the design process to provide immediate feedback
and the integration of the aforementioned material ontology. The multidisciplinary
design and simulation methodology will then be tested within a user study, testing
within an interdisciplinary student design studio to evaluate the developed workflows
and tools as well as the generated designs from a user perspective.
In future research, the methodology will be extended to the application to real
architectural buildings and models. In particular, it is aimed that the sketched objects
with their semantic attributes will be converted accordingly so that the multidisci-
plinary integration platform can be connected to current BIM environments.
In conclusion, the overall field of computational design addressed in the research
project SFB – ACD is vast, and the proposed subprojects represent only specific
challenges in this field. However, the developed platform paves the way for a new
direction of multidisciplinary integration and computational design tools, with a deep
impact on this research field. The proposed framework promotes joint conversations
that designers need to have in the AEC industry and can be applied to various tools
and customized solutions relevant for researchers in the fields of design visualization,
geometrical form-finding, structural and material analysis, lighting optimization,
haptic feedback and fabrication.
168 J. Reisinger et al.

Acknowledgements This work was funded by the Austrian Science Fund (FWF) project F77
(SFB “Advanced Computational Design”, SP2). In addition, the authors thank the following project
partners for their valuable feedback:
SFB, Subproject SP2. I. Erb, D. Daleyev
SFB, Subproject SP3. S. Rutzinger, K. Schinegger, T. Hell, M. Berger, S. Antholzer, M. Bank,
J. Schlusche, V. Sandor, R. Kraft
SFB, Subproject SP4. L. Lipp, P. Ecormier-Nocca, D. Hahn
SFB, Subproject SP5. H. Kaufmann, S. Mortezapoor, E. Vonach
SFB, Subproject SP6. M. Kilian, C. Müller, M. Raffaelli
SFB, Subproject SP7. I. Izmestiev, G. Nawratil, A. Rasoulzadeh, K. Sharifmoghaddam
SFB, Subproject SP8. J. Füssl, V. Senk, M. Königsberger, M. Lukacevic
SFB, Subproject SP9. M. Stavric, H. Vasatko, J. Jauk, L. Gosch

References

1. Polanyi M (1966) The tacit dimension. Routledge & Kegan Paul, London
2. Nonaka I, Takeuchi H (1995) The knowledge-creating company: how Japanese companies
create the dynamics of innovation. Oxford University Press, Oxford
3. Dabbeeru MM, Mukerjee A (2008) Discovering implicit constraints in design. In: Gero JS,
Goel AK (eds) Design computing and cognition ‘08: proceedings of the third international
conference on design computing and cognition. Springer Netherlands, Dordrecht, pp 201–220.
https://doi.org/10.1007/978-1-4020-8728-8_11
4. March S, Storey V (2008) Design science in the information systems discipline: an introduction
to the special issue on design science research. MIS Q 32(4):725. https://doi.org/10.2307/251
48869
5. Wieringa RJ (2014) Design science methodology for information systems and software engi-
neering. Springer Berlin Heidelberg, Berlin, Heidelberg. https://doi.org/10.1007/978-3-662-
43839-8
6. March ST, Smith GF (1995) Design and natural science research on information technology.
Decis Support Syst 15(4):251–266. https://doi.org/10.1016/0167-9236(94)00041-2
7. Cross N (2006) Designerly ways of knowing, vol 114. Springer-Verlag, London
8. van Aken JE (2003) On the design of design processes in architecture and engineering:
technological rules and the principle of minimal specification
9. Witt AJ (2010) A machine epistemology in architecture. Encapsulated knowledge and the
instrumentation of design. Candide J Architectural Knowl 3:37–88
10. Butterworth-Heinemann (2006) The design process—early stages. In: The ASHRAE Green-
Guide, Second edn. Butterworth-Heinemann, Burlington, pp 73–100. https://doi.org/10.1016/
B978-193374207-6/50008-6
11. Mahoney JM (2018) The v-sketch system, machine assisted design exploration in virtual reality.
In: Computer graphics. Cornell University, p 93. https://doi.org/10.7298/X4H1307J
12. Olsen D, Pitman N, Basak S, Wünsche B (2011) Sketch-based building modelling, pp 119–124
13. Kazi RH, Grossman T, Cheong H, Hashemi A, Fitzmaurice G (2017) DreamSketch: early
stage 3D design explorations with sketching and generative design. In: Proceedings of the 30th
annual ACM symposium on user interface software and technology. Association for Computing
Machinery, Québec City, QC, Canada, pp 401–414. https://doi.org/10.1145/3126594.3126662
14. Barrera Machuca M (2018) Multiplanes: assisted freehand VR sketching. https://doi.org/10.
1145/3267782.3267786
15. Oti A, Crilly N (2021) Immersive 3D sketching tools: implications for visual thinking and
communication. Comput Graph 94:111–123. https://doi.org/10.1016/j.cag.2020.10.007
Integrating AEC Domain-Specific Multidisciplinary Knowledge … 169

16. Arora R, Kazi RH, Grossman T, Fitzmaurice G, Singh K (2018) SymbiosisSketch: combining
2D & 3D sketching for designing detiled 3D objects in situ. In: Proceedings of the 2018 CHI
conference on human factors in computing systems, Association for Computing Machinery,
Montreal QC, Canada, p 185. https://doi.org/10.1145/3173574.3173759
17. Drey T, Gugenheimer J, Karlbauer J, Milo M, Rukzio E (2020) VRSketchIn: exploring the
design space of pen and tablet interaction for 3d sketching in virtual reality. In: Proceedings of
the 2020 CHI conference on human factors in computing systems. Association for Computing
Machinery, Honolulu, HI, USA, pp 1–14. https://doi.org/10.1145/3313831.3376628
18. Seddati O, Dupont S, Mahmoudi S. (2016) DeepSketch 2: deep convolutional neural
networks for partial sketch recognition. In: 2016 14th international workshop on content-based
multimedia indexing (CBMI). https://doi.org/10.1109/CBMI.2016.7500261
19. Rosales E, Rodriguez J, Sheffer A (2019) SurfaceBrush: from virtual reality drawings to
manifold surfaces. ACM Trans Graph 38(4):96. https://doi.org/10.1145/3306346.3322970
20. Yu E, Arora R, Stanko T, Bærentzen JA, Singh K, Bousseau A (2021) CASSIE: curve and
surface sketching in immersive environments. In: Proceedings of the 2021 CHI conference
on human factors in computing systems. Association for Computing Machinery, Yokohama,
Japan, p Article 190. https://doi.org/10.1145/3411764.3445158
21. Purao S, Bolloju N, Tan C-H (2018) A modeling language for conceptual design of systems
integration solutions. ACM Trans Manage Inf Syst 9(2):8. https://doi.org/10.1145/3185046
22. Feng K, Lu W, Wang Y (2019) Assessing environmental performance in early building design
stage: an integrated parametric design and machine learning method. Sustain Cities Soc
50:101596. https://doi.org/10.1016/j.scs.2019.101596
23. Lin B, Chen H, Yu Q, Zhou X, Lv S, He Q, Li Z (2021) MOOSAS – A systematic solution for
multiple objective building performance optimization in the early design stage. Build Environ
200:107929. https://doi.org/10.1016/j.buildenv.2021.107929
24. D’Agostino D, D’Agostino P, Minelli F, Minichiello F (2021) Proposal of a new automated
workflow for the computational performance-driven design optimization of building energy
need and construction cost. Energ Buildings 239:110857. https://doi.org/10.1016/j.enbuild.
2021.110857
25. Reisinger J, Kugler S, Kovacic I, Knoll M (2022) Parametric optimization and decision support
model framework for life cycle cost analysis and life cycle assessment of flexible industrial
building structures integrating production planning. Buildings 12(2):162
26. Hassan SR, Megahed NA, Abo Eleinen OM, Hassan AM (2022) Toward a national life cycle
assessment tool: generative design for early decision support. Energy Buildings 267:112144.
https://doi.org/10.1016/j.enbuild.2022.112144
27. Pan W, Turrin M, Louter C, Sariyildiz S, Sun Y (2019) Integrating multi-functional space
and long-span structure in the early design stage of indoor sports arenas by using parametric
modelling and multi-objective optimization. J Building Eng 22:464–485. https://doi.org/10.
1016/j.jobe.2019.01.006
28. Gerber DJ, Lin S-HE (2014) Designing in complexity: Simulation, integration, and multidis-
ciplinary design optimization for architecture. SIMULATION 90(8):936–959. https://doi.org/
10.1177/0037549713482027
29. Forth K, Abualdenien J, Borrmann A, Fellermann S, Schunicht C (2021) Design optimization
approach comparing multicriterial variants using BIM in early design stages. In: Proceedings
of the international symposium on automation and robotics in construction
30. Abrishami S, Goulding J, Rahimian F (2021) Generative BIM workspace for AEC conceptual
design automation: prototype development. Eng Constr Archit Manag 28(2):482–509. https://
doi.org/10.1108/ECAM-04-2020-0256
31. Abrishami S, Goulding JS, Rahimian FP, Ganah A (2014) Integration of BIM and generative
design to exploit AEC conceptual design innovation. J Inform Technol Constr 19:350–359
32. de Klerk R, Duarte AM, Medeiros DP, Duarte JP, Jorge J, Lopes DS (2019) Usability studies on
building early stage architectural models in virtual reality. Autom Constr 103:104–116. https:/
/doi.org/10.1016/j.autcon.2019.03.009
170 J. Reisinger et al.

33. Podkosova I, Reisinger J, Kaufmann H, Kovacic I (2022) BIMFlexi-VR: a virtual reality frame-
work for early-stage collaboration in flexible industrial building design. Front. Virtual Reality
3:782169. https://doi.org/10.3389/frvir.2022.782169
34. Krietemeyer B, Bartosh A, Covington L (2019) A shared realities workflow for interac-
tive design using virtual reality and three-dimensional depth sensing. Int J Archit Comput
17(2):220–235. https://doi.org/10.1177/1478077119852577
35. Deng Z, Liu Y, Pan H, Jabi W, Zhang J, Deng B (2022) Sketch2PQ: freeform planar quadrilateral
mesh design via a single sketch. IEEE Trans Vis Comput Graph 18 https://doi.org/10.48550/
arXiv.2201.09367
36. Du P, Song Y, Deng L (2016) A real-time collaborative framework for 3D design based on
HTML5. In: 2016 IEEE 20th International Conference on Computer Supported Cooperative
Work in Design (CSCWD). https://doi.org/10.1109/CSCWD.2016.7565991
37. Leon M, Doolan DC, Laing R, Malins J, Salman H (2015) Development of a computational
design application for interactive surfaces. In: 19th international conference on information
visualisation, IV 2015. Institute of Electrical and Electronics Engineers Inc. https://doi.org/10.
1109/iV.2015.90
38. Dumont G, Pontonnier C, Wang Z (2014) VES: virtual reality based on interactive mechanical
simulation for design improvement. ASME-ACIER (Adv Comput Inform Eng Res) 1:24
39. Giunchi D, James S, Steed A (2018) 3D sketching for interactive model retrieval in virtual
reality. In: Proceedings of the joint symposium on computational aesthetics and sketch-based
interfaces and modeling and non-photorealistic animation and rendering. Association for
Computing Machinery, Victoria, British Columbia, Canada, p Article 1. https://doi.org/10.
1145/3229147.3229166
40. Vonach E, Gatterer C, Kaufmann H (2017) VRRobot: Robot actuated props in an infinite virtual
environment. In: 2017 IEEE virtual reality (VR). https://doi.org/10.1109/VR.2017.7892233
41. Bechthold M (2008) Innovative surface structures – technologies and applications. Taylor &
Francis
42. Menges A (2016) Material performance – fibrous tectonics & architectural morphology.
Harvard University GSD, Cambridge
43. Kolarevic B, Klinger K (2008) Manufacturing material effects – rethinking design and making
in architecture. Taylor & Francis, London. https://doi.org/10.4324/9781315881171
44. Eigensatz M et al (2010) Paneling architectural freeform surfaces. ACM Trans Graph 29(4):45.
https://doi.org/10.1145/1778765.1778782
45. Jiang C, Rist F, Wang H, Wallner J, Pottmann H (2022) Shape-morphing mechanical
metamaterials. Comput-Aided Des 143:103146. https://doi.org/10.1016/j.cad.2021.103146
46. Pellis D, Kilian M, Wang H, Jiang C, Müller C, Pottmann H (2020) Architectural freeform
surfaces designed for cost-effective paneling through mold re-use
47. Kilian M, Pellis D, Wallner J, Pottmann H (2017) Material-minimizing forms and structures.
ACM Trans Graph 36(6):173. https://doi.org/10.1145/3130800.3130827
48. Lebée A (2015) From folds to structures, a review. Int J Space Struct 30:55–74
49. Tachi T (2010) Freeform rigid-foldable structure using bidirectionally flat-foldable planar
quadrilateral mesh. Springer Vienna, Vienna
50. Ghassaei A, Demaine ED, Gershenfeld N (2018) Fast, interactive origami simulation using
GPU computation. Origami 7:1151–1166
51. Suto K (2021) CRANE – food4Rhino [cited 2022 01.08.2022]. https://www.food4rhino.com/
en/app/crane
52. Dang X, Feng F, Plucinsky P, James RD, Duan H, Wang J (2022) Inverse design of deployable
origami structures that approximate a general surface. Int J Solids Struct 234–235:111224.
https://doi.org/10.1016/j.ijsolstr.2021.111224
53. Jiang C et al (2021) Using isometries for computational design and fabrication. ACM Trans
Graph 40(4):42. https://doi.org/10.1145/3450626.3459839
54. Belytschko T, Gracie R, Ventura G (2009) A review of extended/generalized finite element
methods for material modeling. Modell Simul Mater Sci Eng 17(4):24. https://doi.org/10.
1088/0965-0393/17/4/043001
Integrating AEC Domain-Specific Multidisciplinary Knowledge … 171

55. Murugappan S, Ramani K (2009) FEAsy: a sketch-based interface integrating structural anal-
ysis in early design. In: Proceedings of the ASME Design Engineering Technical Conference,
vol 2. https://doi.org/10.1115/DETC2009-87727
56. Rasoulzadeh S, Senk V, Kovacic I, Reisinger J, Füssl J, Hensel M (2022) Linking early design
stages with physical simulations using machine learning. In: 29th international workshop on
intelligent computing in engineering (EG-ICE). AU Library Scholarly Publishing Services,
Aarhus, Denmark, p 11. https://doi.org/10.7146/aul.455.c212
57. Erler P, Guerrero P, Ohrhallinger S, Mitra NJ, Wimmer M (2020) Points2surf learning implicit
surfaces from point clouds. In: Vedaldi A, Bischof H, Brox T, Frahm J-M (eds) Computer
vision – ECCV 2020: 16th european conference, Glasgow, UK, August 23–28, 2020, Proceed-
ings, Part V. Springer International Publishing, Cham, pp 108–124. https://doi.org/10.1007/
978-3-030-58558-7_7
58. Kufner T et al (2018) Towards a lifecycle oriented design of infrastructure by mathematical
optimization. Optim Eng 20(1):215–249. https://doi.org/10.1007/s11081-018-9406-5
59. Schwarz M, Wonka P (2014) Procedural design of exterior lighting for buildings with complex
constraints. ACM Trans Graph 33(5):166. https://doi.org/10.1145/2629573
60. Umetani N, Igarashi T, Mitra NJ (2012) Guided exploration of physically valid shapes for
furniture design. ACM Trans Graph 31(4):86. https://doi.org/10.1145/2185520.2185582
61. Van Mele TEA (2017–2021) COMPAS – A computational framework for collaboration and
research in Architecture, Engineering, Fabrication, and Construction [cited 2022 05.08.2022].
https://doi.org/10.5281/zenodo.2594510. https://doi.org/10.5281/zenodo.2594510
62. Pellis D, Wang H, Kilian M, Rist F, Pottmann H, Müller C (2020) Principal symmetric meshes.
ACM Trans Graph 39(4):127. https://doi.org/10.1145/3386569.3392446
63. Königsberger M, Lukacevic M, Füssl J (2023) Multiscale micromechanics homogenization of
plant fiber-based biocomposites: from fiber composition to composite stiffness. Mater Struct
56:13
64. Buchner T, Kiefer T, Königsberger M, Jäger A, Füssl J (2021) Continuum micromechanics
model for fired clay bricks: Upscaling of experimentally identified microstructural features to
macroscopic elastic stiffness and thermal conductivity. Mater Des 212:110212. https://doi.org/
10.1016/j.matdes.2021.110212
65. Buchner T, Königsberger M, Jäger A, Füssl J (2022) A validated multiscale model linking
microstructural features of fired clay brick to its macroscopic multiaxial strength. Mech Mater
170:104334. https://doi.org/10.1016/j.mechmat.2022.104334
66. Bank M et al (2023) Hybrid immediacy: designing with artificial neural networks through
physical concept modelling. In: Gengnagel C, Baverel O, Betti G, Popescu M, Thomsen MR,
Wurm J (eds) Towards radical regeneration: design modelling symposium berlin 2022. Springer
International Publishing, Cham, pp 13–23. https://doi.org/10.1007/978-3-031-13249-0_2
67. Jauk J, Vašatko H, Gosch L, Christian I, Klaus A, Stavric M (2021) Digital fabrication of
growth combining digital manufacturing of clay with natural growth of mycelium
68. Jauk J, Gosch L, Vašatko H, Christian I, Klaus A, Stavric M (2022) MyCera. Application of
mycelial growth within digitally manufactured clay structures. Int J Archit Comput 20(1):31–
40. https://doi.org/10.1177/14780771221082248
69. Pech S, Lukacevic M, Füssl J (2022) A hybrid multi-phase field model to describe cohesive
failure in orthotropic materials, assessed by modeling failure mechanisms in wood. Eng Fract
Mech 271:108591. https://doi.org/10.1016/j.engfracmech.2022.108591
70. Vida C, Lukacevic M, Eberhardsteiner J, Füssl J (2022) Modeling approach to estimate
the bending strength and failure mechanisms of glued laminated timber beams. Eng Struct
255:113862. https://doi.org/10.1016/j.engstruct.2022.113862
Comparative Application of Digital
Image Processing and Kuz-Ram Model
in Blast Fragmentation Analysis: Case
of Shayona Cement Quarry

Jabulani Matsimbe, Martin Shaba, Innocent Musonda, and Megersa Dinka

Abstract Blasting operation affects all downstream processes in the mining and
construction cycle hence the need to carry out an effective blast to give the
desired fragmentation. The present study compares the application of digital image
processing and the Kuz-Ram model in particle size distribution analysis at a lime-
stone quarry. Performance assessment utilized the root mean square error, correlation,
and regression analysis. Digital image processing utilized WipFrag V3.3 software
to analyze the images and benchmark improvements. One blast session was carried
out for each mine site. Particle size less than 10 mm, and greater than 300 mm were
treated as fines and oversize respectively. Results showed that the model predicted
2% fines and 2% oversize material while the observations on measured fragmentation
showed 16.5% fines and 0% oversize for the Chikoa blast. For the Livwezi blast, the
model predicted 3% fines and 1% oversize while the observations on measured frag-
mentation showed 0.27% fines and 42% oversize. The blast from Chikoa performed
optimally better than that from the Livwezi mine site. The Kuz-Ram model performed
well at predicting fragmentation for the Livwezi site with a 90.42% correlation while
for Chikoa it was an 85.39% correlation. The model performed with high correlation
values implying that it can be used on both sites to predict fragmentation during
preliminary blast design.

Keywords blasting · construction · computing · fragmentation · mining ·

photogrammetry · wipfrag

J. Matsimbe (B) · I. Musonda

Centre for Applied Research and Innovation in the Built Environment (CARINBE), Faculty of
Engineering and the Built Environment, University of Johannesburg, Johannesburg 2092, South
Africa
e-mail: jmatsimbe@poly.ac.mw
J. Matsimbe · M. Dinka
Department of Civil Engineering Science, Faculty of Engineering and the Built Environment,
University of Johannesburg, Johannesburg 2006, South Africa
J. Matsimbe · M. Shaba
Department of Mining Engineering, Malawi University of Business and Applied Sciences, P/Bag
303, Chichiri, , Blantyre 3, Malawi

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 173
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_13
174 J. Matsimbe et al.

1 Introduction

A carefully engineered blasting operation is important for the success of surface

excavations. As open pits become deeper, quality separation to avoid dilution/ore
losses and energy cost concerns should be given greater attention to optimizing the
entire mine-mill fragmentation system [1]. Blasting in civil and mining operations is
characterized by a sudden release of energy and fragmentation of a rock mass initi-
ated by explosives. The blasting process requires drilling, charging, and initiation [2].
Blasting is carried out to give the desired particle size distribution thereby optimizing
the overall mine economics [3]. Properly fragmented rock removes hiccups in the
downstream processes comprising loading, hauling, and crushing [4]. Therefore, it is
necessary to optimize and predict the rock fragmentation distribution. Fragmentation
models can be used to predict the probable particle size distribution dependent on a
set of blasting parameters and selecting those that give the optimum fragmentation
[5]. The response of rock mass to blasting energy is very complex due to its hetero-
geneous and anisotropic nature [6]. Therefore, it is difficult to obtain a mathematical
relationship between the rock fragmentation and blast design parameters broadly
categorized into controllable and uncontrollable [7]. Influential parameters on rock
mass fragmentation consist of explosive properties, blast geometry, and rock mass
properties [8]. The explosive properties and blast geometry are controllable parame-
ters while rock mass is an uncontrollable parameter. Some of the quantitative methods
for determining and predicting fragmentation size distribution include sieve anal-
ysis, direct observation, Swebrec function, digital image processing, artificial neural
network, and empirical models such as the Kuz-Ram, modified Kuz-ram, Kuznetsov–
Cunningham–Ouchterlony (KCO) and Julius Kruttschnitt Mineral Research Centre
(JKMRC) Model. Ref. [5] assessed the performance of the Kuz-Ram model, Modi-
fied Kuz-Ram, and KCO models using the root mean square error and correlation
and regression analysis to determine the most accurate model applicable to two mine
sites in Ghana. The fragmentation analysis showed a high quantity of fines with an
insignificant number of boulders from all the studied mines. All models had high
correlation coefficients (R) above 95%, the modified Kuz-Ram model performed
best at Mine A whilst the Kuz-Ram model performed best at Mine B. The Kuz-Ram
model performed better in fines prediction than the KCO and Modified Kuz-Ram
models.
Society’s need for a sustainable supply of natural resources requires improved
yield from blasting, crushing, and grinding, minimal transportation, and fewer
amounts of non-hazardous waste. The first step towards a higher yield in the
comminution process is blasting to specifications and producing fragmentation that
requires less energy during the crushing and grinding stages [9]. Mining costs for unit
operations such as loading, hauling, and crushing, can reduce with optimum rock
fragmentation sizes [6]. In Malawi, there is a huge technology gap in blasting oper-
ations mostly in assessing the particle size distribution of blasted rock and making
informed decisions on what/how to improve the blast design. There is a common
usage of direct observation as the main method for analyzing rock fragmentation
Comparative Application of Digital Image Processing and Kuz-Ram … 175

distribution in surface mines/civil works. This facilitates the use of a trial-and-error

approach in blast design. This practice is costly, time-consuming, and tedious since
a lot of resources are spent in the blast design stage to obtain optimal fragmentation.
Therefore, it is imperative to start using methods such as digital image analysis and
the Kuz-Ram model. Compared to other models, Ref. [10] attributed the preference
of the Kuz-Ram model for fragmentation analysis due to its ease of parameterization
and direct linkage between blast design and rock breakage. The empirical modeling
methods provide much better results than qualitative techniques as they are repeatable
and are not intrusive to production [9].
In hard rock mining, blasting is the most productive excavation technique applied
to fragment in-situ rock to the required size for efficient loading and crushing [6].
Rock fragmentation is the breaking down of rocks into small particles for easy end-
user handling. It is the ultimate measure of the efficiency of any production blasting
operations [11]. As over/under fragmentation both tend to increase the cost of mining,
the generation of fragment size in the desired range is necessary [12]. Quantification
of rock fragmentation has attracted a lot of significance over the years. Different
models have been used over the years to predict the size distribution resulting from
blast design [10, 13, 14]. Ref. [10] further divided the models into two approaches
consisting of empirical modeling which infers finer fragmentation from higher energy
input, and mechanistic modeling which tracks the physics of detonation and the
process of energy transfer in well-defined rock for specific blast layouts, deriving
the whole range of blasting results. The mechanistic approach is difficult and time-
consuming to use compared to empirical models hence it is rarely used. The Kuz-
ram model was established in the 1980s and it has been used up to the present. It
is used in estimating the mean fragment size resulting from a blast [5]. The Kuz-
Ram model measures fragmentation by estimating the 50% passing block size of a
muckpile [15]. According to [10], it has two adapted equations and one of them is
the Kuznetsov equation which was developed by Kuznetsov in 1973 [16] and is used
to calculate the mean particle size of a blast. It is given by Eq. 1:
  19
−0.8 − 16 115 20
X m = AK Q . (1)
RW S

where: Xm = particle mean size of muckpile (cm), A = rock factor/blastability index

(ranges from 0.8 to 22), K = powder factor (kg/m3 ), Q = mass of explosive in a blast
hole (kg), RWS = relative weight strength relative to ANFO. The blastability index
(or rock factor) in Eq. 1 is given by Eq. 2 originally developed by Lilly in 1986
[10]. It is used to modify the average fragmentation based on the rock type and blast
direction.

A = 0.06(R M D + R D I + H F) (2)

where: RMD is the rock mass description, RDI is the rock density influence, and HF
is the Hardness Factor. The Rosin–Rammler equation developed in 1933 by Rosin
176 J. Matsimbe et al.

and Rammler [10] is very useful in coming up with a logarithmic cumulative size
distribution. It is given by Eq. 3:
 
−0.693( XX m )
n
Rx = e (3)

where: Rx = mass fraction retained on screen x, n = uniformity index (ranges from

0.7 to 2).
Figure 1 shows the bench blasting design parameters where D blasthole diameter
(m), BH bench height (m), B burden (m), S blasthole spacing (m), SD blasthole
subdrilling (m), SL blasthole stemming length (m), C explosive column height (m).
Other parameters consist of the hp length of the column charge (m), hb length of the
bottom charge (m), Ip concentration of the column charge (kg), Ib concentration of
the bottom charge (kg/m), Qp weight of column charge (kg), Qb weight of bottom
charge (kg), Qe total weight of explosive used in a hole (kg), q powder factor (kg
of explosives/m3 of rock), g specific drilling (drilled meters/m3 of rock), blasthole
pattern, delay timing, and initiation sequence.
Figure 1 shows blast parameters used in the calculation of the uniformity index.
The Uniformity Index characterizes the particle size distribution from a blast and is
dependent on the blast geometry [10]. A high value of n signifies uniform fragmen-
tation size while a low value indicates higher proportions of both fines and boulders
[18]. It is given by Eq. 4:

Fig. 1. Controllable parameters considered in blast design [17]

Comparative Application of Digital Image Processing and Kuz-Ram … 177
┌
 |  | | 0.1
14B |
√ 1+
S
W | BC L − CC L |
| | + 0.1 L
n = 2.2 − B
1− | | (4)
d 2 B L H

where: B = Burden (m), S = Spacing (m), d = Hole diameter (mm), W = Standard

deviation of drilling precision, L = Charge length (m), BCL = Bottom charge length
(m), CCL = Column charge length (m), H = Bench height (m). Cunningham [19]
observed that the uniformity coefficient n varies bwn 0.8 and 1.5.
The field of digital image processing refers to processing digital images using
a digital computer [20]. One of the first applications of digital images was in the
newspaper industry when pictures were first sent by submarine cable between London
and New York around the 1920s [20]. The technique has been heavily applied in slope
and rock fragment analysis in the mining industry. The subject of these images can
be a muck pile, haul truck, leach pile, draw point, waste dump, stockpile, conveyor
belt, or any other location where clear images of rock fragments can be obtained
[11]. When the images have been acquired, they are input into software such as Split
Desktop, WipFrag, FragScan, PowerSieve, etc.,[15] which use their set of algorithms
to analyze the images and produce the Rosin–Rammler cumulative size distribution,
the rock characterization factor (A), the mean fragment size, a histogram, and the
blast geometry which can later be analyzed. The errors of the method can be reduced
by capturing many high-quality images for analysis [15].
Drilling and blasting influence the economy of open pit excavations since the
success of all downstream processes related to strata handling depend on the size
of the blasted rock mass. Therefore, the evaluation of fragments obtained from
blasting plays a very important role in monitoring compatibility with the handling
and crushing machinery. Optimum fragmentation is necessary to ensure sustainable
mining since both over/under fragmented rock tends to affect the cost of mining or
excavation. A blasting operation is optimized when both the blasting and handling
costs of fragmented rock are minimal [4]. Controllable parameters influencing the
fragmentation in a rock blast consist of powder factor, stemming column, bench
height, hole diameter, and burden. The analysis of fragmentation and blast design in
reference to the controllable parameters can enhance the performance of a rock blast;
therefore, the present study aims at conducting a digital image evaluation compared to
the Kuz-Ram model in monitoring/predicting the mean fragment size and percentage
of boulders. The adoption of the Kuz-Ram model and digital image processing in
blasting design can help optimize excavation works and make them cost-effective
leading to efficient downstream processes. This research is based on conventional
small diameter blasting performed in a quarry to obtain the desired fragmentation
with minimal cost.
178 J. Matsimbe et al.

2 Methodology

2.1 Study Area

The study area is in the Kasungu district found in the Central Region of the Republic
of Malawi in Southern Africa. The Limestone mine sites namely Livwezi, and Chikoa
at Shayona Cement are about 3 km apart. Figure 2 shows the Mine site with Livwezi
located at 12°52' 35.86'' S latitude and 33°43' 16.60'' E longitude, and Chikoa located
at 12°52 49.60 S and 33' 44' 33.75'' E longitude. The study area was chosen because
most of the data for blast geometry, rock mass properties, and explosive properties
are known.
The mine is an open pit and practices conventional drilling, blasting, loading,
and hauling. Malawi is comprised of mainly the Precambrian Basement Complex
[21]. These rocks underlie most of the Central region of Malawi where Kasungu
is located. Kasungu is mostly underlain by medium to high-grade metamorphic
rocks of the Mozambique belt. The Livwezi deposit elongates along a Southwest-
Northwest direction and is comprised of medium to coarse-grained marble within
intercalations of calci-silicates, granulite, and amphibolite. It is folded into isoclinal
folds whose foliations deep gently. The outcrop is associated with fractures without
rock or mineral infillings. In the Southeast of the deposit, shear zones are recorded
with a Northeast trend with infillings of very coarse crystals of calcite, quartz, and
feldspar. The Chikoa deposit, however, has the same trend as the Livwezi deposit
and its lithological sequence includes medium to very course-grained marble inter-
banded with granitoid quartz-feldspathic rocks and biotite gneisses.

2.2 Data Collection

The resulting fragmentation from the Quarry blast operations was evaluated using
WipFrag v3.3 digital image processing software and the Kuz-Ram empirical model.
WipFrag is an auto-image generation granulometry software that utilizes digital
images to evaluate fragment size distributions [15]. WipFrag uses automatic algo-
rithms to identify definite blocks and develop a mesh outline or netting using state-
of-the-art boundary or edge recognition techniques. It measures the 2D surface net of
a block assemblage, transforms, and reconstructs a 3D block size distribution using
geometric probability principles.
The Kuz-Ram model [10, 16] is an empirical relationship evaluating blast frag-
mentation through the incorporation of blast design parameters comprising blasting
geometry, characteristics of explosives, explosive quantity, and rock properties. The
Kuz-Ram model evaluates fragmentation through the estimation of the 50% passing
Comparative Application of Digital Image Processing and Kuz-Ram … 179

Table 1. Rock mass

Rock Description RMD value
description value for various
rock mass Friable 10
Massive 50
Vertically jointed Use Jointed Rock Factor

Table 2. JCF interpretations

Observation Interpreted value
from field observations
Tight joints 1.0
Relaxed Joints 1.5
Gouge filled joints 2.0

block size of a muckpile. To compare the two techniques, statistical analyses were
done.
i. Rock Mass Description
Table 1 shows the rock mass descriptor values used to describe its properties.
For the vertically jointed rock mass, the Jointed Rock Factor (JRF) is given
by Eq. 5:

JRF = (JCF ∗ JPS) + JPA (5)

where: JCF = Joint Condition Factor, JPS = Joint Plane Spacing factor, JPA = Joint
Plane Angle factor
ii. Joint Condition Factor (JCF)
From field observations, the joint condition factor will be interpreted and
determined from Table 2.

iii. Field tests

Rock mass properties that were obtained on-site include Rock Mass Description
(RMD) particularly Joint Plane Spacing and Joint Plane Angle factors, and Hardness
Factor (HF) for use in the computation of the rock factor for use in the Kuz Ram
model.
a. Joint plane spacing factor (JPS)
According to Cunningham [10], this factor is partly related to the absolute joint
spacing, and partly to the ratio of spacing to drilling pattern, expressed as the reduced
pattern P. Figure 3 shows the relationship between hole spacing and joint spacing on
blasting fragmentation.
The reduced pattern, P, is given by Eq. 6:
√
P= B×S (6)
180 J. Matsimbe et al.

Fig. 2. Location of mine site

Fig. 3. Relationship between hole spacing and joint spacing on blasting fragmentation [10]

In order to determine the JPS factor, a field test to determine the joint spacing in
the rock mass was carried out using a scan line of at least 3 m in length and calibrated
in millimetres. The test procedure was adapted from ISRM 2007. The procedure is
as described below:
• Whenever possible, the measuring tape should be held along the exposure such
that the surface trace of the discontinuity set being measured is approximately
perpendicular to the tape. If the tape is not perpendicular, the directional bias
corrections are required to obtain true spacing
• All distances (d), between adjacent discontinuities are measured and recorded
over a sampling length not less than 3 m (or the thickness of the rock unit being
Comparative Application of Digital Image Processing and Kuz-Ram … 181

observed if this is less than 3 m). The sampling length should preferably be greater
than ten times the estimated spacing. The distances (d) should be measured to
within 5% of their absolute values.
• The smallest angle, (α) between the measuring tape and the observed joint set is
measured with a compass to the nearest 5˚.
• The most common (modal) spacing is calculated from Eq. 7:

S = dm sin α (7)

where dm is the most common (modal) distance measured. It is helpful to

present the variation spacing using the histogram. When the joint spacing is deter-
mined, the mean value found was referred to in Table 3 together with the reduced
pattern parameter to determine the joint plane spacing factor.

b. Joint Plane Angle

Data for joint plane angle was collected using a clinometer to determine the joint
plane angle. The procedures for data collection followed ISRM standards 2007. The
procedure includes:
a. The maximum declination (dip) of the mean plane of the discontinuity is
measured with the clinometer and should be expressed in degrees as a two-digit
number e.g., 05˚ from 0–90°.
b. The azimuth of the dip (dip direction) is measured in degrees counted clockwise
from true north and expressed as a three-digit number for example 010° from
000°–360°.
c. The dip and dip direction should be recorded in that order, with the three-digit
and two digits number separated by a line
The collected data as shown in Table 4 was then used to determine the Joint Plane
Angle factor.

Table 3. Interpreted value of JPS from JS and P

Joint Spacing JPS Remark
<0.1 m 10 fine fragmentation will result from close joint
0.1–0.3 m 20 unholed blocks are becoming plentiful and large
0.3 to 95% of P 80 some very large blocks are likely to be left
>P 50 all blocks will be intersected

Table 4. Determining Joint

Dip out of face 40
Plane Angle
Strike out of face 30
Dip into face 20
182 J. Matsimbe et al.

iv. Hardness factor

For the hardness factor, data for the uniaxial compressive strength was estimated
by conducting a type L Schmidt rebound hammer test following ASTM C805. Later,
a specific gravity test using ASTM D792 was done to determine the unit weight of
the rock.
v. Blast sessions
One blast session was conducted for each mine site with a total of 6 images being
analyzed for each blast. Digital images which were taken for image analysis were
randomly sampled. To minimize errors, the camera lens was placed perpendicular
and very close to the surface of the blasted muckpile. However, there is a possibility
of deviation from this norm in actual practice. Therefore, the images were captured
from the viewpoint exactly in front of the muckpile with the horizontal axis of the
camera at some angle order than 90° to the face of the muck pile. Double scaling
objects of known dimensions were placed on the surface of the muck pile when
taking the images to provide scale and compensate for the slope of the muckpile [9].

3 Results and Discussion

Performance assessment of the Kuz Ram model using root mean square error was
done in Microsoft Excel 2013. Correlation and regression analysis for assessing
model performance was done using Minitab 20 statistical software. Digital image
processing utilized WipFrag V3.3 software to analyze the images.

3.1 Particle Size Distribution

Figures 4, 5, 6 and 7 show the analysis from the time the image is taken from a
muckpile containing two bamboo sticks 1.7 m in length used for scaling the image
and compensation for the slope of the muckpile until its particle size distribution is
plotted by the software.

3.2 Specific Density Test Results

The results of the specific density test for the rock samples from Chikoa and Livwezi
are presented in Table 5.
Comparative Application of Digital Image Processing and Kuz-Ram … 183

Fig. 4. Freshly shot image from the muck pile

Fig. 5. Auto delineated image

3.3 Uniaxial Compressive Strength

For the Schmidt hammer rebound test conducted on the field, the uniaxial compres-
sive strength (UCS) of the rock mass determined is shown in Table 6.
184 J. Matsimbe et al.

Fig. 6. Image after auto and manual delineations

Fig. 7. Final processed image showing the particle size distribution

3.4 Kuz-Ram Fragmentation Prediction

The utilized explosive was Ammonium Nitrate Fuel Oil (ANFO) with a density
of 0.8, bottom charge Explogel V8 32 × 270 mm (Watergel), and relative weight
strength. The rock mass and blast design parameters are shown in Tables 7 and 8,
respectively.
The predicted values by the model from the adapted Kuznetsov and Rossin-
Rammler equations are presented in Table 9 (Table 10).
Figures 8 and 9 shows the particle size distribution for the Chikoa and Livwezi
blasts predicted by the model.
Table 5. Results for specific density test
Description Livwezi Chikoa
Mass of bottle + rock + water m3 g 1546.59 1480.35 1476.95 1452.92 1439.4 1489.99
Mass of bottle + rock m2 g 592.3 515 511.23 461.96 447.76 509.91
Mass of bottles full of water m4 g 1337.14 1318.66 1318.71 1318.22 1318.33 1317.84
Mass of bottle m1 g 258.39 258.36 259.07 258.62 260.05 258.91
Mass of rock m2 -m1 g 333.91 256.64 252.16 203.34 187.71 251
Mass of water in a full bottle m4 -m1 g 1078.75 1060.3 1059.64 1059.6 1058.28 1058.93
Mass of water used m3 -m2 g 954.29 965.35 965.72 990.96 991.64 980.08
Volume of rock (m4 -m1 )-(m3 -m2 ) mL 124.46 94.95 93.92 68.64 66.64 78.85
m2−m1
Rock density (m4−m1)−(m3−m2) 2.68287 2.702896 2.684838 2.962413 2.816777 3.183259
Comparative Application of Digital Image Processing and Kuz-Ram …

Average density 2.690201 2.987483

185
186 J. Matsimbe et al.

Table 6. Uniaxial
Place UCS (MPa)
Compressive Strength (UCS)
of rock values Chikoa 89
Livwezi 54

Table 7. Rock mass

Parameter Chikoa Blast Livwezi Blast
parameters
Uniaxial compressive stress 89 53.5
(MPa)
Specific density (t/m3 ) 0.299 0.268
Unit weight (KN/m3 ) 29.9 26.8
Rock Mass Description 50 50
(RMD)

Table 8. Blast Parameters

Parameter Chikoa Blast Livwezi Blast
for both sites
Bench Height, H (m) 12 7.5
Hole diameter (mm) 100 100
Stemming height (m) 1.5 1.5
Sub drill (m) 0.5 0.5
Burden, B (m) 2.5 2.3
Spacing, S (m) 2.5 2.5
ANFO per hole (kg) 65.97 37.70
Number of holes 41 52
Total mass of ANFO required 2704.91 1960.35
(kg)
Blast volume (m3 ) 3075 2242.5
Powder Factor (kg/m3 ) 0.88 0.87
Drill Accuracy 95 96
Hole Deviation 0.125 0.1
Bottom Charge Length (m) 0.27 0.27
Column Charge Length (m) 10.23 5.73
Hardness Factor 10.6 15.4

Table 9. Predicted values

Parameter Predicted Value
from the Kuz-Ram model
Chikoa Blast Livwezi Blast
Uniformity Index 1.59 1.44
Rock Factor 4.25 3.81
Mean Particle Size (cm) 10.34 8.49
Comparative Application of Digital Image Processing and Kuz-Ram … 187

Table 10. Predicted Results by the Rosin–Rammler size distribution

Size Chikoa Predicted result Livwezi Predicted result
X(cm) Retained Rx % passing Retained Rx % passing
100 6.85276E-12 100 3.06502E-11 100
68.1 8.79464E-07 99.99991205 9.01428E-07 99.99990986
46.4 0.000515769 99.94842306 0.000332382 99.96676175
31.6 0.016465978 98.35340223 0.009997071 99.00029295
21.5 0.108149064 89.18509362 0.071036159 92.89638408
14.7 0.296951846 70.30481545 0.216681555 78.33184452
10 0.518149079 48.18509206 0.415623513 58.43764868
6.81 0.700021146 29.99788538 0.603616319 39.63836808
4.64 0.823975723 17.60242767 0.747909078 25.20909219
3.16 0.900296634 9.970336592 0.84617571 15.38242901
2.15 0.944698406 5.530159395 0.908544344 9.145565628
1.47 0.969422021 3.057797864 0.9460454 5.395459961
1 0.983323919 1.667608081 0.968659715 3.134028505
0.681 0.990919569 0.908043053 0.98185819 1.814180984
0.464 0.995060167 0.493983292 0.98952056 1.047944021
0.316 0.997317226 0.268277412 0.993960628 0.603937201
0.215 0.998545993 0.145400695 0.996527578 0.347242235
0.147 0.999206015 0.079398505 0.99799046 0.200954017
0.01 0.999988999 0.001100115 0.999958118 0.004188158
0.0068 0.999994047 0.000595344 0.99997597 0.002403028
0.0046 0.999996805 0.000319521 0.999986315 0.001368477
0.0032 0.999998207 0.000179293 0.999991887 0.000811348
0.0022 0.999999013 9.87367E-05 0.999995271 0.000472936
0.0015 0.999999463 5.36604E-05 0.999997276 0.000272399
0.001 0.999999719 2.81378E-05 0.999998481 0.000151897

3.5 Digital Image Processing Results

The particle size distribution analyses of the muckpile obtained from the three
different blasts using WipFrag program are shown in Figs. 10 and 11. The allowable
fragment maximum handled by the primary crusher is 1000 mm.
188 J. Matsimbe et al.

Chikoa Predicted Particle Size Distribution

110
100
90
80
70
60
50
40
30
20
10
0
0.01 0.1 1 10 100

Fig. 8. Chikoa mine blast particle size distribution

Livwezi Predicted Particle Size Distribution

110
100
90
80
70
60
50
40
30
20
10
0
0.01 0.1 1 10 100
-10

Fig. 9. Predicted size distribution for the Livwezi blast

3.6 Chikoa Blast

Figures 12 and 14 show the particle size distribution curve of the blast design deter-
mined from the Kuz-Ram model while Figs. 13 and 15 compares the Kuz-Ram curve
with the two results obtained from WipFrag. The particle size distribution from the
Comparative Application of Digital Image Processing and Kuz-Ram … 189

Fig. 10. WipFrag image analysis for Chikoa blast

Fig. 11. WipFrag image analysis for Livwezi blast

Kuz-Ram model deviates from that of the WipFrag program despite showing similar
trend. The model shows that all fragments are less than the 1000 mm benchmark
indicating a tolerable range of boulders. The differences in size distribution can be
attributed to the heterogeneous nature of the rock mass structural properties and
human error [15]. The mean particle size predicted by the model was 10.23 cm while
the actual particle mean size was 3 cm. The mean particle size gives a measure
of the average fragmentation of the muck pile. This in turn shows that the model
overpredicted the mean particle size for the muck pile.
From Fig. 13, the model predicted 2% fines (less than 10 mm) and 2% oversize
material (over 300 mm). The measured results show 16.5% fines and 0% oversize
material. This shows fines underprediction percentage error of 87% by the model
while the oversize material was overpredicted with an error of 50%. This finding
190 J. Matsimbe et al.

CHIKOA COMPARED BLAST FRAGMENTATION

Kuz Ram Fragmentaion analysis Actual Fragmetation result

110
100
90
80
70
60
50
40
30
20
10
0
0.001 0.01 0.1 1 10 100

Fig. 12. Comparison between the predicted and actual results

CHIKOA BLAST FRAGMENTATION

96%
83.5%

Predicted Fragmentation
Actual Fragmentation

16.5%
2% 0% 2%

30 10 1

Fig. 13. Percentage fines, range, and oversize for predicted and actual results

agrees with studies by [13–15] who found that the Kuz-Ram model underestimates
the number of fines in a mine blast. The percentage range from 10 mm mesh size to
300 mm was predicted at 96% while the measured percentage range was 83.5%.
Comparative Application of Digital Image Processing and Kuz-Ram … 191

Fig. 14. Comparative particle size distribution

LIVWEZI COMPARED FRAGMENTATION

95%

58%
Kuz Ram Prediction
42%
Actual Fragmentation

1% 3% 0.27%

30 10 1

Fig. 15. Comparison of fines and oversize for predicted and measured fragmentation
192 J. Matsimbe et al.

3.7 Livwezi Blast

The Livwezi blast produced a measured fragmentation mean particle size of 20.8 cm
while the model predicted a mean particle size of 8.49 cm. Figure 14 shows a compar-
ison of the particle size distribution predicted by the model against the measured one
with the line drawn perpendicular to the Y-axis at 50% crossing the two graphs
at their respective mean particle size values. The graphs also show that the model
underpredicted the overall particle size distribution as observed that its graph is to
the left of the measured distribution.
From Fig. 15, the model predicted 3% fines (less than 10 mm), a percent in the size
range of 95%, and an oversize (greater than 300 mm) of 1%. Measured results show
0.27% fines, a percent in the size range of 58%, and an oversize retained percentage of
42%. Similar findings on overprediction of fines were found by Ref. [5] in contrast to
underprediction of fines found by Ref. [14]. From the fine’s prediction, the model gave
a prediction error of 91% which is a very high error in fines prediction. Furthermore,
the model underpredicted the boulder size similar to findings by Ref. [5] at Mine B.

3.8 Model Performance

As shown in Table 11, the root mean square error (RMSE) calculated in Microsoft
Excel gave an error of 3.33 (Livwezi) and 23.85 (Chikoa). Linear regression (R2 )
between the model and the measured data shows 81.76% (Livwezi) and 79.92%
(Chikoa) blasts. Correlation coefficients (r) of 85.39% (for Chikoa) and 90.42% (for
Livwezi) were calculated. Livwezi blast had the least root mean square error and high
regression and correlation coefficient values which shows that the model performed
well on the Livwezi blast than on the Chikoa blast. This also shows that the model
performs well on certain sites than on others. The model had a weak performance
in predicting fines at Chikoa (2% vs 16.5%), and boulders at Livwezi (1% vs 42%)
but a slightly better performance in predicting fines at Livwezi (3% vs 0.27%) and
boulders (2% vs 0%) at Chikoa.

Table 11. RMSE, Regression, and correlation values

Mining Location Root Mean Square Error Regression (R2 ) Correlation coefficient (r)
(RMSE)
Chikoa blast 23.85 79.92% 85.39%
Livwezi blast 3.33 81.76% 90.42%
Comparative Application of Digital Image Processing and Kuz-Ram … 193

4 Conclusion

The research focused on reducing the technological gap in the mining industry in
Malawi by applying the Kuz-Ram model and digital image processing techniques to
analyze rock mass fragmentation. This can help in making better blast design deci-
sions to optimize mining processes and costs. Field tests, observations, and laboratory
tests were conducted to meet the objectives of the research. It was found that the Kuz-
Ram model predicted 2% fines (less than 10 mm) and 2% oversize material (greater
than 300 mm) for the Chikoa blast; and 3% fines (less than 10 mm) and 1% oversize
(greater than 300 mm) for the Livwezi blast. The measured fragmentation showed
16.5% fines and 0% oversize material for the Chikoa blast; and 0.27% fines and 42%
oversize for the Livwezi blast. The model performed well in predicting fragmen-
tation at the Livwezi mine site with a correlation of 90.42% while the correlation
for the Chikoa mine was 85.39%. Since the two blasts used a similar blast design
and explosives, the differences can be attributed to variations in rock mass structural
properties e.g., Compressive strength, groundwater, joint fillings, and aperture. Due
to high correlation and regression values, the model can be used on both mining
sites for preliminary blast design with sufficient prior knowledge of the rock mass
properties as an alternative to the trial-and-error approach. Ref. [15] observed that
the fragment size distribution obtained from Kuz-Ram vary from that of WipFrag but
followed a similar trend. Future research work should consider using other models
such as JKRMC, KCO, and Modified Kuz-Ram model on both mining sites to ascer-
tain the accuracy of the Kuz-Ram model considering that the model seems to work
best on certain sites and not on others.

Acknowledgements Opinions and conclusions are those of the authors and are not neces-
sarily attributable to ASIM. The work is supported and part of collaborative research at the Centre of
Applied Research and Innovation in the Built Environment (CARINBE). The authors are thankful
to Shayona Cement Corporation for providing access to their mining sites.
Funding This research is funded by the Intra-Africa Mobility Scheme of the European Union in
partnership with the African Union under the Africa Sustainable Infrastructure Mobility (ASIM)
scheme.

Conflict of Interest The authors declare no conflict of interest.

References

1. Hustrulid W, Kuchta M, Martin R (2013) Open pit mine planning and design, 3rd edn. CRC
Press
2. Seccatore J, Origliasso C, De Tomi G (2013) Assessing a risk analysis methodology for rock
blasting operations. Taylor & Francis Group, London
3. Michaux S, Djordjevic N (2005) Influence of explosive energy on the strength of the rock
fragments and SAG mill throughput. Miner Eng 18(4):439–448. https://doi.org/10.1016/j.min
eng.2004.07.003
194 J. Matsimbe et al.

4. Morin MA, Ficarazzo F (2006) Monte Carlo simulation as a tool to predict blasting fragmen-
tation based on the Kuz-Ram model. Comput Geosci 32(3):352–359. https://doi.org/10.1016/
j.cageo.2005.06.022
5. Kansake B, Temeng V, Afum B (2016) Comparative analysis of rock fragmentation models – a
case study. In: 4th UMaT biennial international mining and mineral conference, pp 1–11
6. Bowa V (2015) Optimization of blasting design parameters on open pit bench a case study of
Nchanga open pits. Int J Sci Technol Res 4(09):45–51
7. Hudaverdi T, Kuzu C, Fisne A (2012) Investigation of the blast fragmentation using the mean
fragment size and fragmentation index. Int J Rock Mech Min Sci 56:136–145. https://doi.org/
10.1016/j.ijrmms.2012.07.028
8. Thornton D, Kanchibotla SS, Brunton I (2002) Modelling the impact of rockmass and blast
design variation on blast fragmentation. Fragblast 6(2):169–188. https://doi.org/10.1076/frag.
6.2.169.8663
9. Ouchterlony F (2003) Influence of blasting on the size distribution and properties of muckpile
fragments, a state of the art review. Lulea University of Technology, Swebrec
10. Cunningham C (2005) The Kuz-Ram fragmentation model – 20 years on
11. Siddiqui F, Shah S, Behan M (2009) Measurement of size distribution of blasted rock using
digital image processing. J King Abdulaziz Univ-Eng Sci 20(2):81–93. https://doi.org/10.4197/
Eng.20-2.4
12. Singh BK, Mondal D, Shahid M, Saxena A, Roy PNS (2019) Application of digital image
analysis for monitoring the behavior of factors that control the rock fragmentation in opencast
bench blasting: a case study conducted over four opencast coal mines of the Talcher Coalfields,
India. J. Sustain. Min. 18(4):247–256. https://doi.org/10.1016/j.jsm.2019.08.003
13. Johnson N, de Klerk Q, Yeo W, Roux A (2012) Technical report and mineral resource and
reserve update for the nzema gold mine, Ghana, West Africa. West Perth
14. Cho SH, Kaneko K (2004) Rock fragmentation control in blasting. Mater Trans 45(5):1722–
1730. https://doi.org/10.2320/matertrans.45.1722
15. Shehu SA, Yusuf KO, Hashim MHM (2022) Comparative study of WipFrag image analysis
and Kuz-Ram empirical model in granite aggregate quarry and their application for blast frag-
mentation rating. Geomech Geoeng 17(1):197–205. https://doi.org/10.1080/17486025.2020.
1720830
16. Kuznetsov VM (1973) The mean diameter of the fragments formed by blasting rock. Sov Min
Sci 9(2):144–148. https://doi.org/10.1007/BF02506177
17. Nobel D (2010) Blasting and explosives quick reference guide. Kalgoorlie
18. Bergman P (2005) Optimisation of fragmentation and comminution at Boliden mineral, aitik.
University of Technology
19. Cunningham C (1987) Fragmentation estimations and the Kuz–Ram model – four years on
20. Gonzalez R, Woods R (2008) Digital image processing, 3rd edn. Pearson Hall, New Jersey
21. Malunga G (2018) The geology and mineral potential of Malawi
Technology-Enriched Engineering
Pedagogy
Proposal of a Collaborative Teaching
Method for AEC Supported by Additive
Manufacturing Use

Márcio Henrique de Sousa Carboni, Sérgio Scheer,

and Armando Luis Yoshio Ito

Abstract The Brazilian construction industry presents low productivity and limited
international competitiveness. In this scenario, two relevant aspects are identified:
academic education and technology adoption. In view of this context, two move-
ments stand out. The first is the reformulation of the Brazilian National Curriculum
Guidelines (DCNs) for Civil Engineering and Architecture undergraduate courses,
which seeks to improve professional qualification, with a greater focus on the expec-
tations of the industry of the 4th Industrial Revolution. The second is the emergence
of new professionals, capable of engaging in the AEC (Architecture, Engineering
and Construction) ecosystem to assist in the adoption of new technologies, linked to
Industry 4.0. Among various innovative technologies, this research focus on Additive
Manufacturing (AM), as it presents a great disruptive potential for civil construc-
tion industry, especially in education. Among the new workers who are joining the
AEC industry, Graphic Expression professionals, graduated from Federal Univer-
sity of Paraná (Brazil), are inducing agents for the adoption of new technologies.
For the effective incorporation of these elements in the AEC ecosystem, teaching
programs are required to provide the fundamentals of technology, allowing future
professionals to discover their capabilities. This paper presents a proposal of a collab-
orative teaching method for AEC, supported by AM use. This method, grounded on
constructivist pedagogy for teaching and learning, is the result of a Design Science
Research (DSR) approach. It puts together students from Civil Engineering, Archi-
tecture and Graphic Expression undergraduate courses, who then work on an inter-
disciplinary setting, studying a common theme. Our method uses Additive Manufac-
turing, Project-based Learning (PBL) and Collaborative Learning as teaching tools,
with an emphasis in motivation, always considering the best conditions for students.
In this paper, we establish important contents to be worked on in a hybrid teaching
model, which uses distance learning to teach theoretical concepts and practical labo-
ratory experience to solve problems collaboratively, ensuring the development of

M. H. de Sousa Carboni (B) · S. Scheer

Federal University of Paraná, Curitiba PR, Brazil
e-mail: mhcarboni@ufpr.br
A. L. Y. Ito
Federal University of Technology – Parana, Curitiba PR, Brazil

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 197
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_14
198 M. H. de Sousa Carboni et al.

skills and competences. Finally, the present paper indicates seven important steps
for the development of the proposed educational activities.

Keywords Additive Manufacturing · Project-based Learning · Collaborative

Learning · Graphic Expression · Teaching Method

1 Introduction

The Brazilian industry is in a position of low competitiveness and innovation in

the international market. According to the Global Innovation Index, Brazil lost 22
positions between 2011 and 2016, ranking 69th out of the 123 countries evaluated.
This low performance was mainly due to the score obtained in human resources and
research, especially in the area of engineering [1].
Compared to developed countries, Brazil presents a considerably retrograde
reality, since a large part of its national industry is not even in the Third Indus-
trial Revolution [2]. Several authors point out that, in the construction industry, this
scenario is even worse, since there are delays in the adoption of technologies and a
decline in the level of productivity in comparison to other industries [3–6].
Part of this situation is a result of the precarious training of professionals in the
area. According to [1], between 2015 and 2017, Brazil presented an occupancy rate in
Engineering undergraduate courses of less than 30%. The author also states that there
is evidence of a lack of interest in Engineering courses, beyond high dropout rate
among students – more than 50%, according to the National Council of Education
study CNE/CES n. 1/2019 [7].
Among other reasons, it stands out the unpreparedness of students, as a result of
weak basic education, as well as the unattractive organization of curricula, and other
disincentives that keep student away from professional practice [1, 8].
However, there are undeniable attempts to evolve, such as the recurring themes
of BIM, lean construction, modular construction, and digital manufacturing. In the
latter case, we can include Rapid Prototyping (RP) and Digital Fabrication (DF)
technologies, which make it possible to build physical elements from 3D models,
created with the support of CAD/BIM tools. In this scenario, we emphasize the use
of the additive processes known as Additive Manufacturing (AM), or 3D printing.
AM is the process of manufacturing physical models, using data from a virtual CAD/
BIM model, through the deposition of layers that will be manufactured successively,
with the goal of obtaining the complete geometry of the part [9–11].
Such technologies can be partly or fully inserted into the design-construction
process of architecture, since they can be applied in its different stages, from concep-
tual architectural design to the molds of construction elements. They are also appli-
cable in the manufacturing of final elements, which will then be ready for installa-
tion in the building. However, in Brazil, the use of DF as a constructive technique is
insipient, given the country’s small industrialization in AEC. For this reason, such
technologies are mainly addressed in higher education centers, in order to guarantee
Proposal of a Collaborative Teaching Method for AEC Supported … 199

the experimentation and training of professionals who are able to work with them
[12].
In the teaching–learning process, the support and benefits of 3D printing in the
most different areas is already a reality [13–17]. Such areas include arts, mathematics,
engineering, design, archeology and historical heritage, astronomy, architecture and
urbanism, medicine and anatomy.
Physical models can assist students in the spatial understanding of varied
phenomena. Consequently, they increase students’ involvement in classes. They
allow an appropriate language for learning, which facilitates and rationalizes the
process of learner’s thinking, in order to replace a naive view of reality with a more
critical and comprehensive posture [18]. In areas that demand creativity and notions
of design, they facilitate the understanding of design processes, allowing students to
transfer their ideas to reality and to learn about the weaknesses and virtues of their
projects. Thus, students are ablet experience the interactive cycle of design processes
[16, 19].
The evolution of digital technologies, especially since the end of the twentieth
century, has subjected society to profound changes in the processes of conceiving,
producing, disseminating, learning and applying knowledge, with social, cultural,
technical and educational implications [20]. In this context of rapid changes and
technological evolution, the role of the different professionals involved in the AEC
(Architecture, Engineering and Construction) industry is changing dramatically. New
operating niches are emerging specifically for the use of new technologies [12, 21].
In this context, two movements stand out. The first involves discussions about
the Civil Engineering (CE) and the Architecture and Urbanism (AU) curricula
of Brazilian universities, which culminated in the reformulation of the Brazilian
National Curriculum Guidelines (DCNs) for Engineering in 2019, and in the elabo-
ration of a proposal to change DCNs for Architecture and Urbanism (not yet approved
by the Ministry of Education). The second is related to the creation, in 2012, of the
first bachelor’s degree in Graphic Expression (CEGRAF), at Federal University of
Paraná (UFPR), which stems precisely from the perception of the great demand for
generalist professionals with focus on new digital technologies.
For the effective incorporation of such elements in the AEC ecosystem, educa-
tional programs are required to provide key technology fundamentals, allowing future
professionals to discover their capabilities, which must be aligned with the needs of
an education for the twenty-first century society.
This paper presents a proposal of a collaborative teaching method for AEC
supported by AM use, which puts together students from Civil Engineering, Architec-
ture and Graphic Expression (GE) undergraduate courses, to assure interdisciplinary
learning and studies on a common theme. Additive Manufacturing, Project-based
Learning (PBL) and Collaborative Learning are used as teaching tools. We also
consider motivation and appropriate conditions for all students, based on the MUSIC
academic motivation model, which presents 5 components for the design of a course:
eMpowerment, Usefulness, Success, Interest and Caring. We also establish impor-
tant contents to be included in a hybrid teaching model, with the distance learning of
theoretical concepts, combined with practical laboratory experience, so that students
200 M. H. de Sousa Carboni et al.

learn to solve problems collaboratively, ensuring the development of news skills and
competences. Finally, we indicate seven important steps for the development of the
proposed educational activities.

1.1 Brazilian National Curriculum Guidelines (DCNs)

for Civil Engineering and Architecture Undergraduate
Courses

In Brazil, the academic education of engineers does not focus enough on the work
environment. It is also disconnected from the expectations of the 4th Industrial Revo-
lution industry. Given this scenario, in 2008 a movement of large companies, called
Entrepreneurial Mobilization for Innovation (MEI), coordinated by the Brazilian
National Confederation of Industry (CNI), started to work for the improvement of
engineering education in Brazil, in line with the agenda of international trends.
In 2019, the work of MEI, together with ABENGE (Brazilian Association of
Education in Engineering), culminated in the reformulation of the 2002 Engineering
DCNs [22]. The new DCNs propose a new institutional organization for curriculum,
stimulating a diversity of learning forms. The idea is to extrapolate the traditional
classroom, in order to guarantee the interaction of teaching with research and exten-
sion activities, as forms of learning, and to promote active teaching methodologies
[1].
These guidelines seek to create an environment, in undergraduate courses, that
allows students to exercise entrepreneurship inside or outside companies, with a
greater potential to develop their talents, skills and abilities [2]. With this aim, some
suggested strategies are: to create flexible and student-centered curricula; to empha-
size the development of competences expected from graduates; to encourage the
adoption of active learning methodologies (applied knowledge); to ensure collabo-
ration between universities and companies, with the idea of bringing courses closer
to the market; to develop multidisciplinary programs; and to combine classroom and
distance activities [22].
The path in Architecture and Urbanism was similar. In 2013, meetings orga-
nized by the Brazilian Association of Architecture and Urbanism Teaching (ABEA)
promotes discussions on the rebuilding of the DCNs created in 2006 [23] and
amended in 2010 [24]. Later, after 2018, discussions continued in the Education
and Training Committees of the Brazilian Council of Architecture and Urbanism
(CAU/BR). In 2019, a new proposal for DCNs of AU undergraduate courses was put
forth.
As for Engineering, some of the main changes proposed for Architecture and
Urbanism seek to draw attention to the following aspects: the use of active teaching
methodologies, ensuring students’ active participation in the knowledge construc-
tion process; diversification of teaching and learning environments, so that students
have professional experience in the labor market and collaborative inter-professional
Proposal of a Collaborative Teaching Method for AEC Supported … 201

practices; and the possibility of curricular flexibility, which allows the search for
innovation inherent in the AEC area [25].
It should be noted that the concept of Industry 4.0 is understood as the current
approach to changes in production processes and business models that employ a
high degree of technology in the production chain. The goal is to guarantee increased
productivity and competitiveness, generating new values and services for both clients
and organizations [26]. In this context, technologies such as Internet of Things (IoT),
artificial intelligence, digital manufacturing, additive manufacturing, big data and
cloud computing are fundamental.
In the proposal of the new DCNs for architecture, for the first time, the “Digital
and Physical Modeling Processes and Tools” are mentioned as mandatory curric-
ular content. Therefore, the DCNs introduces the obligation of at least one digital
modeling and fabrication laboratory in undergraduate courses, emphasizing the
importance of new technologies, disseminated as fundamental concepts for Industry
4.0 and for all Architecture and Urbanism activity. This is reflected in the skills and
abilities that students are supposed to develop in AU undergraduate courses [25].
As for Engineering, training in Architecture and Urbanism should also be
reformulated, with a focus on professional practice, collaboration, active teaching
methodologies, beyond technological education.

1.2 Graphic Expression Professional

In 2012, a bachelor’s degree in Graphic Expression (CEGRAF) was created at Federal

University of Paraná (UFPR). Its creation was motivated by the identification, in
the industry, of a lack of communication and understanding among professionals
involved in the process of design and manufacturing. Lack of technical knowledge,
combined with the poor exchange of information between different professionals,
caused delays, with consequent financial losses to the business [27].
There are no specific DCNs for Graphic Expression, considering that the course
is currently unique in the country, as a bachelor’s degree in its specific approach. Its
objective is “to train professionals who work in the development of digital graphic
projects, in the range of existing functions between creation and production, who
participate in multidisciplinary teams” [28], in the areas of architecture, engineering
and industrial products, based on the concepts of digital design and prototyping. A
broad education is a requirement for such professionals, so that they can understand
important elements from several areas, to communicate with different professionals,
collaborating in project development processes.
Graphic Expression students were well received by the market, inside internships
in modeling and digital models, project compatibility through digital technologies,
and product production through digital manufacturing processes. In 2018, the evolu-
tion of labor market trends with new technologies reinforced the need to include new
content in the curriculum, in line with the principles of Creative Economy and Maker
Movement [29].
202 M. H. de Sousa Carboni et al.

Therefore, the Graphic Expression undergraduate course was restructured, to

ensure a greater focus on knowledge in digital technologies (such as digital modeling,
BIM, three-dimensional capture and digitization, digital fabrication and prototyping,
and virtual and augmented reality), beyond theoretical grounding in geometry
concepts, architecture, design, and engineering. These new professionals present
themselves to the market with the potential to be a part of collaborative projects with
other professionals, in the fields of architecture, engineering, arts and mathematics.
Specifically, considering the focus on the expertise acquired on digital modeling and
prototyping technologies, we can conclude that there is great potential for comple-
mentarity with Architecture and Engineering courses, which still have difficulties in
integrating such topics into their curricula.

2 Method

The proposal for a teaching method presented in this work is the result of a Design
Science Research (DSR) approach. The product of a DSR is an “artifact”, which is
something artificial, conceived by man, carried out with the aim of fulfilling a purpose
in a given environment [30]. According to [31] classification (constructs, models,
methods, instantiations), the artifact developed in this work is essentially a method
with a conceptual structure for using Additive Manufacturing as an integration tool
for students of AEC courses.
Other research methods were used to support problem awareness, data collection,
and artifact formulation. Among them, we used systematic literature review [32],
action-research, beyond a survey carried out by students in 2019. This artifact is part
of a doctoral thesis and will still be tested and evaluated.

3 Results and Discussion

In this section, we present and discuss the collaborative teaching method for AEC,
supported by the Additive Manufacturing use, considering its elements.

3.1 Interdisciplinary Learning for AEC

The collaborative teaching method for AEC, supported by AM use (Fig. 1), is
grounded on constructivist pedagogy for teaching and learning. It is also aligned
with the development of competencies and skills expected of AEC professionals in
the twenty-first century.
This method aims at ensuring the interdisciplinary learning of students of Archi-
tecture and Urbanism, Civil Engineering and Graphic Expression, in the light of the
Proposal of a Collaborative Teaching Method for AEC Supported … 203

Fig. 1. Collaborative teaching method for AEC supported by AM use

interdisciplinarity concept considered by Gatto [33], characterized by the following

aspects: (1) to have an object in common; (2) to be searched at the same time (simul-
taneity); (3) by specialists from different disciplines or areas; (4) in close cooperation
(collaborative work); (5) with constant exchange of information; (6) sharing the same
goals to achieve an integrated analysis of the common object or problem.
Initially, the topic of study, which will be approached through PBL activities,
must be defined. Such topic must be explored by the different undergraduate courses
(Architecture and Urbanism, Civil Engineering and Graphic Expression), using Addi-
tive Manufacturing as a teaching tool and as an integration and stimulus factor, so that
students develop their skills in communication, collaboration, problem solving, tech-
nological literacy, and autonomy, in addition to the technical knowledge necessary
for their professions.
Several topics can be addressed, from which we cite a few.
One possibility is Building Information Modeling (BIM). BIM is strategic for the
development of the AEC industry, considering the efforts of the Brazilian Federal
Government, which launched the National Strategy for the Dissemination of Building
Information Modeling in Brazil in 2019 [34]. The program seeks to promote a suitable
environment for the investment in BIM, ensuring its diffusion in the country. In
addition, it has great adherence to the innovative technologies of Industry 4.0 (such
as Additive Manufacturing), in search of modernizing the civil construction industry.
However, Brazilian undergraduate curricula still require rebuilding for the effective
204 M. H. de Sousa Carboni et al.

implementation of BIM. At UFPR, for example, the only undergraduate course with
a specific mandatory discipline about BIM is Graphic Expression.
Another possibility is “structural systems”. According to [35], the importance
of teaching the theory of the subject in Civil Engineering courses is settled, but
the focus is quite technicist, based on activities of calculation and on the analysis
of structural dimensioning. Engineering students end up having difficulties in the
previous stage, which encompass the structural conception and establishment of
structural arrangements, an activity that requires intuitive and creative skills.
On the other hand, the authors state that “the teaching of structures seems to
remain deficient in Architecture courses” [35] (p. 184), even though it is the subject
of recurrent research, publications, and even specific events, such as the National
Meeting on Teaching Structures in Architecture Schools (ENEEEA). In any case,
it is undeniable that the structural design is a fundamental stage for architectural
design. Therefore, architecture students need to master the use and functions of
structural systems. Although this is not a specific topic addressed in the Graphic
Expression undergraduate course, it leads to several approaches that involve knowl-
edge in the area, considering both the representation and the technical point of view
of calculation.
The physical experimental model is the most discussed and recommended
resource in seminars, meetings, lectures and published papers on the teaching of disci-
plines in the area of structures [35]. It allows a real visualization of the phenomena,
demonstrating concepts in an intuitive, reliable, and tangible way. This scenario
assures us that uniting such concepts with Additive Manufacturing is an appropriate
strategy to expand collaboration between the different courses we have exposed.
To make the application of the proposed method tangible, we sought a theme
that would work on the importance of understanding “construction systems”, so that
students could explore the concept of BIM in the future. We believe that it is essential
to understand how a building is built, so that it is possible to work well with BIM.
In conclusion, we chose “architectural details” as the central topic of learning.
It is a small piece of the whole, but it characterizes and defines the entire building.
"Indeed, detail is architecture at its smallest size” [36].
The connection of different materials leads to several problems that must be solved
by the designer. It is common in academia to approach such topic only with 2D draw-
ings, an abstract representation, but students need, first, to understand the function
of 3D elements in their context. By working with 3D printed models, students can
simulate how to build, understand the relationship of elements, and visualize them
spatially.

3.2 Conditions and Academic Motivation

To develop a successful method, firstly we need right conditions. It is necessary to

have an environment focused on students, in which they will have an active role
in their learning and will be able to act, discuss, problematize, and analyze their
Proposal of a Collaborative Teaching Method for AEC Supported … 205

actions. According to Elmôr Filho [37], an active learning environment should be

a common place where students and teachers are cognitively active, in order to
actively participate in reflective, interactive, collaborative and cooperative activities.
It requires the commitment of all students, but also of the teachers, to conduct the
activities in a way that all students feel valued and welcomed by the group.
Also, it is important to consider that getting students from different areas to solve
a problem within time constraints can be a stressful experience [38]. Thus, it is
important to ensure students’ motivation. Academic motivation is a process that
is perceived both by actions (e.g., attention, effort, persistence) and verbalizations
(e.g., “I like biology”), through which goal-directed physical or mental activities are
instigated and maintained [39].
All things considered, we used the MUSIC Model of Academic Motivation, devel-
oped by Jones [39], to support our method. This model is grounded on a social-
cognitive theoretical framework and seeks to increase students’ motivation, to allow
greater student learning. It consists of five components that ought to be considered
when designing a course: (1) empowerment, (2) usefulness, (3) success, (4) interest,
and (5) caring.

Empowerment. It considers the amount of perceived control that learners have over
their learning. It relates to students’ autonomy and the perception of both a high level
of freedom and a sense of choice over their actions during an activity. It varies from
student to student, since it depends on several variables, including content difficulty,
student’s skills, beyond the student’s previous experiences with the content. But the
essential factor is “that students must believe they have some control over some aspect
of their learning” [39].

Usefulness. Students must understand why the activities are useful for their educa-
tion and how they will use them in their professions and lives. Studies show that
students are more motivated when they have long-term goals in mind, compared to
students that have only short-term goals [39].

Success. Students must believe that their efforts will be valuable for their future. This
aspect is related to their self-perception of competence, that is, that their knowledge
and skills are sufficient to complete given tasks. If they believe they will succeed,
they will try harder and persist longer in the activity. They will also be more resilient
in the face of difficulties. It is important to make all course objectives clear, providing
clear and understandable assessment criteria, beyond accurate and honest feedback
[39].

Interest. Topics and activities should be interesting to students. The idea of interest
consists of two components: an affective component of positive emotion (the liking)
and a cognitive component of concentration (the engagement). This relates to atten-
tion, memory, comprehension, and deeper cognitive engagement. Hands-on activities
and contents that relate to students’ background knowledge are useful strategies. It
is also important to incorporate other components of the MUSIC model, which can
promote individual interest [39].
206 M. H. de Sousa Carboni et al.

Caring. This concept is related to the students’ perception that the instructors care
about their learning and that they desire to impact their lives positively. There are
several terms to refer to the concept of caring, such as sense of community, bonding,
attachment, involvement, and commitment. In this context, the instructor behavior
is determinant. It is important to listen to students’ opinions and ideas, devote time
to help students, and show concern and interest in students’ personal/professional
lives.

3.3 Teaching Tools for Active Learning

Teaching tools include everything that helps in the teaching–learning process. Espe-
cially active learning approaches demand a variety of tools in order to ensure success.
It is important to evaluate the goals of a given teaching activity and the type of learners
when choosing the most appropriate tool to meet specific learning goals [40].
Our method is based on the use of three teaching tools: Project-based Learning
(PBL), Collaborative Learning (CL), and Additive Manufacturing.
The PBL (Project-based learning) approach is a student-centered pedagogy
that involves interdisciplinary teaching–learning activities in an active-collaborative
learning environment. Students are tutored by qualified teachers, developing in-
depth knowledge by answering to real-world questions, problems, and challenges.
The classroom becomes an open space for ideas, knowledge, languages, and reali-
ties, stimulating conversations and generating bonds. The problem approach allows
students to grow individually and collectively in their learning experience, through
the construction of meanings that generate reflections, giving new significance to
their experiences. Through this method, students become managers of their learning
process [11, 41].
Teaching AM through the PBL approach is common practice. Authors [32] iden-
tify numerous works that use AM in higher education, by putting together students
from different undergraduate courses with the support of a PBL approach, with the
aim of promoting students’ engagement with their learning. In addition, the method
helps to bring teaching closer to students’ reality, culminating in more meaningful
learning.
Collaborative learning refers to an instruction/learning method in which students
work together, in small groups, towards a common goal [42]. It complements the
PBL approach, given the complementarity of latent knowledge of students from
three different courses (AU, CE, and GE). Students can teach concepts to each other
efficiently, while they work together to solve problems. In this context, Graphic
Expression students in an AEC group are important, because they have more in-depth
knowledge of digital technologies, such as Additive Manufacturing.
Additive Manufacturing, as we have already pointed out, has a special use in
education. It offers gains for the teaching–learning process, by combining PBL and
Proposal of a Collaborative Teaching Method for AEC Supported … 207

AM, such as greater student interest, aptitude to understand abstract concepts, stim-
ulus for the construction of student-centered and collaborative knowledge, and the
possibility of interdisciplinary integration [32].
Based on such foundations, the proposed method was organized so that students
build their own learning. In this context, Additive Manufacturing is a tool that keeps
them motivated to solve challenges.

3.4 Passive/Active Learning of AM

AM learning was used in both passive and active ways [43], according to the
undergraduate program characteristics and objectives.
Incorporating AM passively into courses means using AM as a support tool for
teaching. It makes sense to use this type of approach for AEC, since the main objective
of training these professionals is not the operation of digital fabrication equipment,
but the training of professionals who understand the technology, who know how to
take advantage of it to solve problems, and who are able to work in multidisciplinary
teams [32].
However, one of the barriers to AM adoption in teaching is the lack of famil-
iarity with the technology [33], both by students e teachers. This reinforces the
idea of including a more experienced and focused AM future professional in an
interdisciplinary AEC group, such as the Graphic Expression one. The proposed
method provides an active AM teaching way, in which the activities that are devel-
oped present an explicit focus on concepts related to 3D printing. Some important
didactic objectives for AM are:
• Learning and applying design principles to Additive Manufacturing
• Learning AM fundamentals and basic operating principles
• Stimulating creative competence using 3D printing
• Gaining experience in the operation of AM machines
• Identifying advantages and/or limitations of AM technologies

3.5 Hybrid Education

Although the specific study of AM is important, the main objective of the presented
method is the interdisciplinary integration of AEC students. In order to fulfil this
goal, hybrid education should be explored. In this context, theoretical and more
introductory contents could be addressed with extra-class activities at distance, using
video classes, beyond materials that support students in hands-on activities. This
allows Architecture and Civil Engineering students to delve deeper into AM, in case
they are interested. Most of class (or laboratory) time can be used to explore the
development of skills and knowledge through practical experiences in groups [32].
In Table 1, we present AM contents explored at distance and in laboratory.
208 M. H. de Sousa Carboni et al.

Table 1. Hybrid education

Approach AM Contents
of AM contents
Distance learning AM Fundamentals
Fundamentals in 3D printer operation
Design for manufacturing (DFM)
Laboratory Study common objects using AM as a tool
Stimulate creative competence
Gain experience in AM operation
Identify AM advantages/limitations

3.6 7-Step Pedagogical Model

To ensure the best development of the activities proposed in the collaborative teaching
method for AEC students, we recommend 7 steps (Fig. 2), based on the “Pedagogical
Model for Introducing 3D Printing Technologies” [44]. This pedagogical model is
grounded on a classic instructional design theory, the “Conditions of Learning” [45].
According to Gagne, there are nine instructional events that must be followed when
designing a course. Each instructional event is related to a specific cognitive process.
Also, we relate to the instructional events in which categories of academic moti-
vation [39] should be explored, considering the kind of hybrid education strategy
that can be adopted and the staff that should be involved in each step.
In this model, the subject of “constructive architectural details” was considered a
common object of study. However, it can be applied to any chosen topic, as previously
mentioned (e.g., BIM, structural systems, didactic models). As an example of an
activity, students should build 3D printed physical models that represent constructive
architectural details of iconic buildings.
Before analyzing the 7 steps, it is important to note that interdisciplinary groups
must be balanced in terms of number of members and technical background. It is
assumed that most architecture and civil engineering students have no knowledge
in 3D printing. Thus, graphic expression students should at least have knowledge of
the basic principles of additive manufacturing.
Another decisive factor is the need for knowledge in 3D CAD/BIM modeling to
work with AM. Therefore, it is recommended that instructors carry out an initial
questionnaire to assess students’ abilities, so that they create groups in which at least
one member has more experience or more advanced skills. These “advanced” students
will be able to share their knowledge with the groups (collaborative learning).
The first step is to get the attention of all participants. In order to fulfil this objective,
the instructor can present the course, with samples of what was developed in previous
editions. To present the importance and usefulness of the topic, a professional can
be invited to lecture on the common object chosen to be studied.
Next (step 2), the instructor must define the objectives and expectations of all
activities. This is important for students to realize they will be able to complete tasks
with the skills they already have and/or will develop in the process. To guarantee
Proposal of a Collaborative Teaching Method for AEC Supported … 209

Fig. 2. 7-step pedagogical model for collaborative teaching method for AEC supported by AM
use

the necessary theoretical knowledge, at this stage, recorded classes can be produced
for students to watch outside class hours. Students can absorb the content according
to their need and availability. When dealing with AM, we suggest that at least three
themes be explored: AM fundamentals, basic settings for 3D printing, and Design
for Manufacturing (DFM).
After completing this theoretical conceptualization phase, teams choose what they
will represent. Each team chooses at least three references. The instructor should
guide which ones could create more difficulty during the process and/or improve
results. Nonetheless, students must have some control over the tasks.
210 M. H. de Sousa Carboni et al.

In step 3, it is time for a considerable amount of teamwork. Students must

develop 3D models that represent their architectural details. They will need to: use
their knowledge to understand the functions and relationships of each constructive
element; conceive a three-dimensional visualization from, usually, two-dimensional
drawings; and plan their model considering DFM rules – that is, observing the needs,
facilities and difficulties of 3D printing and assembly.
Instructors, or guest professionals who can act as tutors for the teams, must accom-
pany the development of solutions, but never state whether something is right or
wrong. They should point out problems or difficulties that students may face, encour-
aging the rapid process of prototyping solutions and exploring the benefits brought
by the use of 3D printing.
In the next step (step 4), students are given a moment to review their proposals
and exercise collaborative learning and autonomy. They must review their solu-
tions and analyze the 3D models with a focus on Design for Manufacturing (DFM)
rules, retrieving the concepts presented in theoretical and distance classes. Espe-
cially, they should prioritize a large base for better adhesion to the printing bed and
avoid thin cross-sections in the lower section, as the fast movement of the 3D printer
can generate vibrations in the piece and its eventual break during the process. The
projecting of elements that are not supported by a previous layer should be designed
with a maximum angle of 45º to the vertical plane [46]. Advanced students can help
teams with a critical look.
When students encounter technical difficulties, they can seek assistance from
instructors, tutors, or monitors (step 5). Now, the instructor can guide the students,
indicating the best practices for the process of manufacturing, considering that, in
this phase, experience with 3D printing is highly important. Attention should be paid
to individual difficulties. The instructor ought to dedicate more time to the ones that
present greater difficulties. If instructors deem it necessary, they can invite students
to repeat step 4.
In step 6, students present their design proposal and their 3D printing planning
to instructors or laboratory manager/monitors, who evaluate their work in terms of
constructability and printability. Authors [47] refer to the “manufacturability analy-
sis” (or “constructability analysis”) for AM as the need to evaluate specific factors,
such as: build volume, minimum feature size, amount of support material, printing
time, estimation of surface roughness, and build orientation.
Finally (step 7), the responsible for the 3D printers in the laboratory receive the
models and generate the print planning for the teams, with the aim of grouping several
parts in a single print, in order to reduce printing time. Students should follow the
process in accordance with instructions given by monitors or lab managers. Later,
they must finish the assemblance of their physical models and present them to the
class.
The instructor conducts an evaluation of the results, together with the class, medi-
ating a discussion. This learning community should share different ways to solve
the same problem, listening to students’ opinions on the main difficulties faced
throughout the process, considering what they would do differently and what lessons
Proposal of a Collaborative Teaching Method for AEC Supported … 211

were learned. Also, it is important to share students’ feedbacks with everyone, due
to the importance of integrating students from different AEC undergraduate courses.

4 Conclusion and Future Work

We presented a proposal of a collaborative teaching method for AEC, supported

by Additive Manufacturing, that should be used by educators to integrate different
students, in order to stimulate educational experiences, as close as possible to the
professional practice.
Additive Manufacturing allows the approach of different contents in a practical
and tangible way. In addition to being an important technology to be included in the
civil construction area, with the precepts of industry 4.0, AM enables several gains,
allowing greater student engagement with their education.
In summary, our method is based on three major components: interdisciplinarity,
technology, and people. An important part of success relies on the fact that people
must be motivated to learn, to collaborate and to be active actors of their own educa-
tion. Given this context, we present ways to ensure that possibility, by working with
the seven steps of our method, with the MUSIC model of academic motivation.
Furthermore, we expect that collaborative educational experiences bring together
other professionals who are not used to being part of the AEC area, but who have
great potential to optimize processes, by including new technologies, such as Graphic
Expression professionals.
This work is part of a doctoral research. The next steps are the application of the
method in interdisciplinary courses, followed by its validation.

References

1. de Oliveira VF (2019) A engenharia e as novas DCNs: oportunidades para formar mais e

melhores engenheiros. LTC, Rio de Janeiro
2. Kern M, Sagazio G, Lourenção P, Pereira S, Miranda Z, Lopes A (2019) A mobilização
empresarial pela inovação (MEI) e a defesa da modernização do ensino de Engenharia. In: de
Oliveira VF (ed) A engenharia e as novas DCNs: Oportunidades para formar mais e melhores
engenheiros. LTC, Rio de Janeiro, pp 33–43
3. de Oliveira MR, Fabricio MM (2011) Projeto paramétrico e prototipagem rápida: casos em
instituições internacionais. In: Kowaltowski D, Moreira D de C, Petreche JRD, Fabricio MM
(eds.) O processo de projeto em arquitetura. Oficina de textos, São Paulo, pp 455–469
4. Eastman C, Teicholz P, Sacks R, Liston K (2011) BIM handbook: a guide to building infor-
mation modeling for owners, managers, designers engineers and contractors. Wiley, New
Jersey
5. NIST: Metrics and tools for construction productivity project. https://www.nist.gov/programs-
projects/metrics-and-tools-construction-productivity-project
6. Wu P, Zhao X, Baller JH, Wang X (2018) Developing a conceptual framework to improve
the implementation of 3D printing technology in the construction industry. Archit Sci Rev
61:133–142. https://doi.org/10.1080/00038628.2018.1450727
212 M. H. de Sousa Carboni et al.

7. BRASIL (2019) Conselho Nacional De Educação. Parecer CNE/CES no 1 de 23 de

janeiro de 2019. Diretrizes Curriculares Nacionais do Curso de Graduação em Engen-
haria. http://portal.mec.gov.br/index.php?option=com_docman&view=download&alias=109
871-pces001-19-1&category_slug=marco-2019-pdf&Itemid=30192
8. Godoy EV, Almeida ED (2020) Evasão nos cursos de engenharia: um olhar para os trabalhos
do COBENGE de 2000 a 2014. Rev. Bras. Ensino Ciência e Tecnol. 13:50–74. https://doi.org/
10.3895/rbect.v13n3.8583
9. Go J, Hart AJ (2016) A framework for teaching the fundamentals of additive manufacturing
and enabling rapid innovation. Addit Manuf 10:76–87. https://doi.org/10.1016/j.addma.2016.
03.001
10. Haavi T, Tvenge N, Martinsen K (2018) CDIO design education collaboration using 3D-desktop
printers. Procedia CIRP 70:325–330. https://doi.org/10.1016/j.procir.2018.03.277
11. Stern A, Rosenthal Y, Dresler N, Ashkenazi D (2019) Additive manufacturing: an education
strategy for engineering students. Addit Manuf 27:503–514. https://doi.org/10.1016/j.addma.
2019.04.001
12. Martins IL, Filho ZRP (2019) A produção acadêmica sobre a fabricação digital nas
escolas brasileiras de arquitetura e urbanismo. PARC Pesquisa em Arquitetura e Construção
10:e019007. https://doi.org/10.20396/parc.v10i0.8652734
13. Kolitsky MA (2014) Reshaping teaching and learning with 3D printing technologies. e-mentor
2014(56 (4)):84–94. https://doi.org/10.15219/em56.1130
14. Vaccarezza M, Papa V (2015) 3D printing: a valuable resource in human anatomy education.
Anat Sci Int 90:64–65. https://doi.org/10.1007/s12565-014-0257-7
15. Fukuda T, Tokuhara T, Yabuki N (2016) A dynamic physical model based on a 3D digital
model for architectural rapid prototyping. Autom Constr 72:9–17. https://doi.org/10.1016/j.
autcon.2016.07.002
16. Huleihil M (2016) 3D printing technology as innovative tool for math and geometry teaching
applications. J Phys Conf Ser 755:011001. https://doi.org/10.1088/1742-6596/755/1/011001
17. Ludwig PM, Nagel JK, Lewis EJ (2017) Student learning outcomes from a pilot medical
innovations course with nursing, engineering, and biology undergraduate students. Int J STEM
Educ 4:33. https://doi.org/10.1186/s40594-017-0095-y
18. Bassanezi CR (2015) Ensino aprendizagem com modelagem matemática: uma nova estratégia.
Contexto, São Paulo
19. Novak E, Wisdom S (2018) Effects of 3D printing project-based learning on preservice elemen-
tary teachers’ science attitudes, science content knowledge, and anxiety about teaching science.
J Sci Educ Technol 27:412–432. https://doi.org/10.1007/s10956-018-9733-5
20. Nery M de A (2016) O uso das tecnologias digitais da internet na Educação Superior:
representações docentes - entre o formal e o informal, as marcas da presença do possível
21. Pupo RT (2008) Ensino da prototipagem rápida e fabricação digital para arquitetura e construção
no Brasil: definições e estado da arte. PARC Pesquisa em Arquitetura e Construção 1:80. https:/
/doi.org/10.20396/parc.v1i3.8634511
22. BRASIL (2019) Conselho Nacional De Educação. Resolução CNE/CES no 2, de 24 de Abril
de 2019. Diretrizes Curriculares Nacionais do Curso de Graduação em Engenharia. https://
www.in.gov.br/web/dou/-/resoluÇÃo-no-2-de-24-de-abril-de-2019-85344528
23. BRASIL (2006) Conselho Nacional De Educação. Resolução No 6, de 2 de fevereiro
de 2006. Diretrizes Curriculares Nacionais do curso de graduação em Arquitetura
e Urbanismo. http://portal.mec.gov.br/index.php?option=com_docman&view=download&
alias=5649-rces06-06&category_slug=junho-2010-pdf&Itemid=30192
24. BRASIL (2010) Conselho Nacional De Educação. Resolução CNE/CES no 2, de
17 de junho de 2010. Diretrizes Curriculares Nacionais do curso de graduação em
Arquitetura e Urbanismo. http://portal.mec.gov.br/index.php?option=com_docman&view=
download&alias=5651-rces002-10&category_slug=junho-2010-pdf&Itemid=30192
25. ABEA (2019) Desafios do ensino de arquitetura e urbanismo no século XXI. In: XXXVII
Encontro Nacional sobre Ensino de Arquitetura e Urbanismo, XX Congresso da Associação
Brasileira de Ensino de Arquitetura e Urbanismo. p 515. Rio de Janeiro
Proposal of a Collaborative Teaching Method for AEC Supported … 213

26. Santos BP, Alberto A, Lima TDFM, Charrua-Santos FMB (2018) Indústria 4.0: desafios e
oportunidades. Rev. Produção e Desenvolv. 4:111–124
27. DEGRAF: Graduação em Expressão Gráfica homepage. http://www.exatas.ufpr.br/portal/ceg
raf/. Last accessed 20 Apr 2022
28. De Souza L.V., Costa DMB (2013) O curso de bacharelado em expressão gráfica da UFPR. In:
GRAPHICA 13. Florianópolis
29. DEGRAF (2018) Projeto pedagógico do curso de expressão gráfica, http://www.exatas.ufpr.
br/portal/cegraf/wp-content/uploads/sites/3/2019/09/PPC_Projeto_Pedagogico_de_Curso__
Expressao_Grafica.pdf
30. Lacerda DP, Dresch A, Antunes Junior JAV (2013) Design science research Método de Pesquisa
para Avanço da Ciência e Tecnolgia. Gestão Produção 20:741–761. https://doi.org/10.1590/
S0104-530X2013005000014
31. March ST, Smith GF (1995) Design and natural science research on information technology.
Decis Support Syst 15:251–266
32. Carboni MH de S, Scheer S (2021) Manufatura aditiva como ferramenta didática para a
formação de profissionais da AEC. In: Simpósio Brasileiro De Tecnologia Da Informação
E Comunicação Na Construção, 3, pp 1–14. ANTAC, Uberlândia
33. Gatto A, Bassoli E, Denti L, Iuliano L, Minetola P (2015) Multi-disciplinary approach in
engineering education: learning with additive manufacturing and reverse engineering. Rapid
Prototyping J 21(5):598–603. https://doi.org/10.1108/RPJ-09-2014-0134
34. BRASIL (2019) Decreto no 9.983, de 22 de agosto de 2019. Dispõe sobre a Estratégia Nacional
de Disseminação do Building Information Modelling e institui o Comitê Gestor da Estratégia do
Building Information Modelling. http://www.planalto.gov.br/ccivil_03/_ato2019-2022/2019/
decreto/D9983.htm
35. Gorgônio P, Da Nóbrega B, Da Nóbrega SHS (2020) Engenheiro civil x arquiteto: conflito no
aprendizado das estruturas. Rev de Ensino Eng 39(1):183–191. https://doi.org/10.37702/REE
2236-0158.v39p183-191.2020
36. Weber, P.D.: Beyond Bolts : architectural details, construction, meaning, http://hdl.handle.net/
1721.1/70861, (1991)
37. Elmôr Filho G, Sauer LZ, de Almeida NN, Villas-Boas V (2019) Uma nova sala de aula é
possível: Aprendizagem ativa na educação em engenharia. LTC, Rio de Janeiro
38. Corbacho AM, Minini L, Pereyra M, González-Fernández AE, Echániz R, Repetto L, Cruz
P, Fernández-Damonte V, Lorieto A, Basile M (2021) Interdisciplinary higher education with
a focus on academic motivation and teamwork diversity. Int J Educ Res Open 2–2:100062.
https://doi.org/10.1016/j.ijedro.2021.100062
39. Jones BD (2009) Motivating students to engage in learning: the music model of academic
motivation. Int J Teach Learn High Educ 21:272–285
40. Woods M, Rosenberg M (2016) Educational tools: thinking outside the box. Clin J Am Soc
Nephrol 11(3):518–526. https://doi.org/10.2215/CJN.02570315
41. Carbonell J (2016) Pedagogias do século XXI - bases para a inovação educativa. Penso, Porto
Alegre
42. Yokaichiya DK, Galembeck E, Brag DB, Torres BB (2004) Aprendizagem Colaborativa
no Ensino a Distância - Análise da Distância Transacional. In: Congresso Internacional de
Educação a Distância, 11, pp 1–11. ABED, Salvador
43. Ford S, Minshall T (2019) Invited review article: where and how 3D printing is used in teaching
and education. Addit Manuf 25:131–150. https://doi.org/10.1016/j.addma.2018.10.028
44. Chiu PHP, Lai KWC, Fan TKF, Cheng SH (2015) A pedagogical model for introducing 3D
printing technology in a freshman level course based on a classic instructional design theory.
Proc. – Front. Educ. Conf. FIE. 2014:1–6. https://doi.org/10.1109/FIE.2015.7344287
45. Gagne R (1985) The conditions of learning. Rinehart & Winston, New York
46. Junk S, Matt R (2015) New approach to introduction of 3D digital technologies in design
education. Procedia CIRP 36:35–40. https://doi.org/10.1016/j.procir.2015.01.045
47. Lynn R et al (2016) Toward rapid manufacturability analysis tools for engineering design
education. Procedia Manuf. 5:1183–1196. https://doi.org/10.1016/j.promfg.2016.08.093
Development of a Safe
and Anthropomorphic Drone
in an Interdisciplinary
Research-Oriented Construction
Management Course

Gilles Albeaino , Sebastian Paulis , Masoud Gheisari ,

and Raja R. A. Issa

Abstract This paper discusses a recent educational effort to develop interdisci-

plinary educational projects within the construction curriculum through a research-
oriented undergraduate honors course. Second semester freshman honors students
from different fields (e.g., construction, architecture, engineering, computer science)
developed a social drone featuring anthropomorphic characteristics and safety
features that could enable safer, user-friendlier, and improved onsite human-drone
collaboration and interaction in the future. First, a group of interdisciplinary under-
graduate students were asked to perform an exhaustive literature review to identify:
(1) drone applications in construction equipped with onboard sensors to perform data
collection; (2) safety-related challenges that drones could encounter on jobsites; and
(3) anthropomorphic characteristics (e.g., facial expressions) and features that have
been used in the fields of human-drone interaction (HDI) and human-robot interac-
tion (HRI). Based on the retrieved information, students were then asked to develop
a social drone design concept considering safety (overcoming typical safety chal-
lenges found with regular drones used in construction), anthropomorphism (making
the design more user-friendly), and capability to accomplish the same construction
tasks as its previous non-social models. Upon model conceptualization, students had
to rely on 3D modeling software packages and Virtual Reality (VR)-compatible game
engines to program and bring their designed social drone model to life within a VR

G. Albeaino
Department of Construction Science, Texas A&M University, 3137 TAMU, College Station,
TX 77843, USA
e-mail: gilles.albeaino@tamu.edu
M. Gheisari · R. R. A. Issa (B)
Rinker School of Construction Management, 573 Newell Dr, Gainesville, FL 32603, USA
e-mail: raymond-issa@ufl.edu
S. Paulis
Herbert Wertheim College of Engineering, University of Florida, 1949 Stadium Road,
Gainesville, FL 32611, USA

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 215
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_15
216 G. Albeaino et al.

environment. Through critical thinking, students were provided with an opportunity

to actively collaborate with peers from different STEM fields to evaluate informa-
tion gathered from the construction, HDI, and HRI literature, and apply the acquired
theoretical knowledge to design, develop and program a safe and anthropomorphic
social drone.

Keywords Drones · Safety · Construction education · Robot anthropomorphism ·

Interdisciplinary and collaborative work · Undergraduate research

1 Introduction

Interdisciplinary work takes into consideration diverse skills, topics, and tasks to
account for specific academic subjects while collecting related information across
different fields. Whether through structured classroom activities, capstone projects,
or competitions, educators have long realized the importance of interdisciplinary and
collaborative type of work and have been promoting to a certain extent, collaboration,
and group work in their day-to-day courses [5]. This allows students, particularly
those seeking Science, Technology, Engineering, and Mathematics (STEM) degrees,
to simulate what they are most likely to encounter or deal with upon graduation,
emphasizing the importance of knowledge retention, engagement, and scalability
towards the real world. Not only does interdisciplinary group work incentivize growth
in academic fields but it also increases awareness for new technologies that are rapidly
being integrated in the workplace [7]. Indeed, STEM graduates need to work together
to achieve one common goal: finding innovative solutions to complex real-world
problems.
Construction Management (CM), which involves stakeholders and professionals
from overlapping fields (e.g., construction, management, architecture, engineering)
and backgrounds, with commonalities and differences in their skills (e.g., communi-
cation, leadership, creativity, independence, trust), is one of those fields that benefits
from interdisciplinary academic efforts. For example, through an interdisciplinary
approach integrating Building Information Modeling (BIM) in the Construction
Management and Engineering (CME) curriculum, Demirdoven (2015) demonstrated
how the implementation of collaborative BIM education is associated with improved
overall student satisfaction and preparation for leadership roles in construction [1].
Another study combining BIM and interdisciplinary projects to achieve higher levels
of depth in academic trainings yielded positive results when observing how sustain-
able design options were analyzed earlier through student collaboration [4]. In
the engineering domain, Molina et al. (2014) have implemented Unmanned Aerial
Vehicle (also referred to as drone) projects with quadrotor limitations in curriculums,
providing students with hands-on experience and demonstrating the effectiveness of
collaboration in the classroom [6]. A very recent study [2] also implemented inter-
disciplinary efforts with modern technologies and projects, focusing on the student’s
engagement to reach greater levels of understanding and ability with “soft skills” for
Development of a Safe and Anthropomorphic Drone ... 217

a variety of fields. By prioritizing the professional experience in collaborative efforts

with many backgrounds, students were capable of translating their abilities into a
more real world simulated situation. This highlights the importance of combining
interdisciplinary work with academic teachings towards STEM fields in general, and
construction in particular.
The purpose behind this study is to reflect on the outcomes of a case study
conducted in a research-oriented undergraduate honors course with the aim of
promoting interdisciplinary and collaborative type of work among students of
different STEM disciplines. Many studies have explored collaboration and inter-
disciplinary work in construction education [1, 5], this project challenged a set of
diverse students to rely on their knowledge acquired from the tools and concepts
covered throughout the course, the findings of their own research, and their wide-
ranging yet overlapping backgrounds (e.g., construction, architecture, engineering,
computer science), and apply this knowledge to solve real-world problems while
meeting internal and external deadlines. This ultimately challenged their decision-
making and critical thinking capabilities and, through interdisciplinary information-
and knowledge-sharing, exposed them to a wide variety of new theoretical and prac-
tical tools and technologies, some of which have not been previously explored in
construction and could otherwise be too difficult to acquire in discipline-specific
courses. More specifically, the project involved developing a social drone with anthro-
pomorphic and safety characteristics that would enable safe human-drone collabora-
tion and interaction on construction jobsites. This required students to: (1) perform
literature search to identify current drone design limitations and understand different
configurations being used in construction; (2) conceptualize and design drone models
while taking into consideration the findings of the literature search and the goal of
the project; (3) develop the finalized platform design (including all the components
and features) using 3D modeling software; and (4) integrate the 3D drone model into
VR-compatible game engine and program it to simulate real-world platform move-
ment in the built environment, ultimately allowing them to further develop their skills
and accomplish their shared goals.

2 Methods

2.1 Course Description and Project Integration

The interdisciplinary project was integrated in BCN4905: Special Studies in

Construction, a dynamic and research-oriented undergraduate honors course that
provides second semester freshman honors students from different disciplines (e.g.,
architecture, engineering, construction, computer science) with the opportunity to
collaboratively find solutions to different problems. The course focuses on enhancing
students’ understanding of how advanced construction technologies can be used in
CM to improve the performance and productivity of construction processes. It also
218 G. Albeaino et al.

introduces students to modern automation techniques and applications within the

construction industry. The course is divided into two main parts. The first part,
which accounts for approximately 70% of the semester, was dedicated towards
direct exposure to topics and technologies used in construction. The covered topics
mainly include: (1) Building Information Modeling (architectural modeling); (2)
Simulation and Reality Capture (creating animations and 4D simulations, as well as
generating point clouds and conducting comparative analyses using laser scanning
and drone-mediated photogrammetry); (3) Virtual/Augmented/Mixed Reality—VR/
AR/MR (exposing students to a variety of applications); (4) computational design
(introducing generative and parametric design); as well as (5) robotics, automa-
tion, and 3D printing. Through these diversified technologies, the students received
introductory experiences with modern practices while learning about processes and
critical criteria when discussing the techniques and solutions used in construction.
This ultimately provided them with opportunities to explore their own connections
with respect to each technology/subject, promoting higher levels of growth in inter-
disciplinary skills. The other part of the course requires students to allocate a large
portion of their time towards a technical interdisciplinary project revolving around
and applying in-class topics and their diversified perspectives to solve a particular
problem within the construction industry.
The aim of the Spring 2022 student project was to explore drone anthropo-
morphism and safety in construction. More specifically, students were asked to
develop a social drone, featuring anthropomorphic characteristics and safety features
that would enable safe human-drone collaboration and interaction on construc-
tion jobsites. The increased adoption of drones in the Architecture, Engineering,
Construction, and Operations (AECO) domain, this technology has been associated
with several increased safety risks (i.e., physical risks, attentional costs, negative
psychophysiological effects) that could jeopardize its usage on jobsites [3]. The
goal was to develop virtual drones with anthropomorphic characteristics and safety
features to improve construction professionals’ perception of these aerial platforms
and potentially reduce the risks associated with their presence on jobsites. Students
were asked at first to collaboratively perform an exhaustive literature review to:
(1) identify all drone applications in construction together with the sensors used
to perform data collection; (2) identify all the safety-related challenges that drones
could encounter on jobsites (which potentially cause hazardous situations); and (3)
identify the anthropomorphic features/characteristics used to design drones from
the human-robot interaction (HRI) and Human-Drone interaction (HDI) literature.
Based on the identified information from the literature, students were then asked to
develop a drone design prototype that: (1) is capable of being fitted with any of the
data collection sensors typically used in construction (to accomplish same construc-
tion tasks as its previous non-social models); (2) is considered safe (overcoming
typical safety challenges found with regular drones used in construction); and (3)
has anthropomorphic characteristics (to make it social and more user-friendly). The
project criteria were broken down to several requirements, encouraging students
to implement different approaches, processes, and technologies to collaboratively
Development of a Safe and Anthropomorphic Drone ... 219

achieve the final product. The project steps included: Literature Review, Concep-
tual Design, 3D Model Development, Virtual Reality Implementation, and Project
Presentation. Through these five major steps in the timeline of the project, students
were given the opportunity to work in different environments, promoting the use
of different skills, including: discovery, comprehension, creativity, communication,
computer programming, 3D computer-aided design, art, and scientific writing. By
breaking down the structure of the project in this manner, students would not only
learn about the current status and requirements for drones in construction but would
also develop knowledge and strengths in more fields besides construction (e.g., in
critical thinking, innovation, physics, human-robot interaction, psychology, culture,
safety, feasibility versus theory, and perception).

3 Results

3.1 Literature Review

The literature review involved a variety of topics for the students to consider, e.g.,
drone applications in construction, data collection sensors mounted on drones, chal-
lenges associated with drone usage in construction, drone physics and configura-
tions (including software and hardware components and materials). The first step of
the literature review was to identify the different drone applications (i.e., progress
monitoring, site planning, site mapping and surveying, earthmoving, building inspec-
tion, safety management, building maintenance, post-disaster reconnaissance, aerial
construction, security surveillance, material handling, and site communication), and
sensors (i.e., Red-Green-Blue and Thermal cameras, Light detection and ranging
devices and Laser scanners) used to collect information (i.e., images, videos) on
construction jobsites. The safety challenges affecting drone usage in construction
were then identified, and all these factors (i.e., drone applications, drone sensors and
design, safety challenges, anthropomorphism) were taken into consideration, under
different drone tasks and scenarios, in the next steps of the drone design. For example,
in the case of drone collision, the aerial platform may descend rapidly on a civilian or
worker, requiring safety measures to reduce vertical collision forces; the drone may
collide horizontally with a civilian in the event of a malfunction, requiring safety
measures to reduce horizontal collision forces. In addition, the drone needs to have
the capability to interact with civilians and workers in some fashion to execute safe
maneuvers and promote healthy interactions and safety, keeping workers from feeling
threatened or in danger and ensuring safe human-drone interaction on jobsites.
This method of placing a high value behind the interaction between students of
different backgrounds and disciplines allowed them to develop a body of knowl-
edge that would support their future endeavors. The students read, interpreted, cited,
summarized, and sectionalized the literature review for each topic, encapsulating the
220 G. Albeaino et al.

unique aspects of the topics while considering the overlap between them. The litera-
ture review covered high level interactions between topics and their specific relevance
to each other. By performing the literature review, students were able to strengthen
their ability to analyze and synthesize large amounts of data in a short period of time
and coordinate actions for knowledge discovery amongst their group. These skills
not only benefited the literature review but were also used in other collaborative
tasks, like designing, modeling, virtual implementation, and presentation.

3.2 Conceptual Design

After performing the literature review and collecting background information, the
students listed their requirements and objectives for the drone and proceeded to iter-
atively sketch new designs with their criteria. Each set of sketches involved labels to
symbolize the function or purpose of the interconnected components and were visu-
alized from multiple perspectives. After every iteration, the students would discuss
amongst one another how different aspects of the current design could be enhanced
to better suit the problems. After internal discussion, they would present their most
recent version of their sketches to the instructors. By doing so, new perspectives and
points of clarity were brought up as part of the student-student and students-instructor
conversations. With each discussion, new changes were listed, and the students would
create new designs to fulfill their change logs. By performing this design and review
process, students were able to consider a wide variety of perspectives from peers and
literature context, as well as acquiring “soft skills” involving receptiveness and will-
ingness to change and adapt to new information and ideas. These soft skills impacted
the effectiveness of collaboration when progressing through future stages of the
project. This iterative process led to the final selection of the drone’s capabilities and
conceptual design, incorporating several safety and anthropomorphic features, which
are summarized in Table 1. The final conceptual design of the drone is shown in Fig. 1.
The overall design included, in addition to typical drone components, modular liga-
ments, parachutes, airbags, rotor guards, obstacle avoidance systems, modular sensor
mount, animalistic shell, as well as built-in input/output sensors for human interac-
tion. The required sensors had contextual support from computer science require-
ments, safety requirements, and anthropomorphic characteristics selected through
the interdisciplinary efforts.

3.3 3D Model Development

With thorough sketches, the students implemented the 2D representation of the drone
into a 3D environment using Autodesk® Fusion 360 as their selected software.
Through the support and context of both, the literature review as well as the concep-
tual design sketches, the students had all high-level resources to frame the general
Development of a Safe and Anthropomorphic Drone ... 221

Table 1 Anthropomorphic and safety features incorporated in the final drone design
Features Description
OLED Screen Convey Emotions and establish emotional connection
Speaker Increase understandable intention between drones and humans and improve
the communication between drones and humans
Bee-like Shell Ensure more natural interaction while relying on animal-like design
Camera Read gestures and reactions and react accordingly
Proximity Establish obstacle avoidance system to improve safety during operation
Sensors
Rotor Guards Increase safety and resilience of drone, pilot, and individuals nearby
Airbag Module Decrease probability of fatality from a drone falling and reduce damaged
objects in possible crashes
Parachute Decrease probability of fatality from a drone falling and reduce damaged
Recovery System objects in possible crashes
Nano-tech Wide operating thermal range, allowing for universal use in different
Batteries climates while not catching fire
Microphone/ Allows for remote communication between operator and people near the
Speaker Unit drone
LED Screen Allows for workers to see visual representations of facial patterns and voice
oscillations

parameters surrounding the model. To fulfill all environmental variables associated

with the model, the students had to set low-level requirements that would progress the
model to its final state. Therefore, they defined what commercial parts would serve
to fulfill the drone’s capabilities; each commercial part was either modeled in 3D
space or obtained from a model provided by the manufacturer. With all preliminary
parts modeled, the students imported all parts into a singular assembly to validate
the layout of all systems relative to the sketches. At this point, the main chassis could
be defined with all specific information like mounting holes, attachment mecha-
nisms, fittings, and tolerances. Finally, the animalistic shell was modeled to mimic
the defined sketches while encapsulating all components as necessary (Fig. 2). With
all parts and systems created and assembled, the drone was ready for virtual integra-
tion in a game engine. By performing this 3D modeling work, students were able to
better understand the steps required to prepare a project for a next phase and learned
better practices for project management when dealing with new environments. Also,
students got to gain first-hand experience with modern 3D modeling and design
technologies like Autodesk® Fusion 360.
222 G. Albeaino et al.

Top View With Shell Without Shell

Front View
Bottom View

Fig. 1 Conceptual drone design

3.4 Virtual Reality Implementation and Drone Integration

The students selected Unity® as their game engine to serve as their intermediary
between the digital model and VR. As new models were finalized, they were exported
as compatible files and given animations with controls in a virtual environment.
With a modeled drone, a realistic environment was best suited to symbolize a real-
world scenario of a drone on the built environment (Fig. 3). The students utilized all
resources?? to enhance the virtual implementation of the drone, interconnecting the
information from the literature review with all parameters defined in the sketches
and models as well as demonstrating programming and model skills alongside their
communication and organizational workflow. Through the integration of the environ-
ment with drone controls, simulations of accidents, actions, features, and achieve-
ments were recorded and visualized as the final project result to be presented at the
end of the semester.
Development of a Safe and Anthropomorphic Drone ... 223

Fig. 2 Developed 3D drone model

Fig. 3 Unity® game development and drone integration

224 G. Albeaino et al.

The students took a stepwise approach towards the implementation of the Unity®
environment with the previous developed material. Multiple aspects were considered
when producing the final product: models, physics, controls, automatic interactions,
and outputs. After importing the models, each model was colored, scaled, and config-
ured with their appropriate physical properties: collision, gravitational influence, and
custom scripts. Once physical collision metrics were established and tested, the drone
was programmed with controls for both manual and automatic functionality. Then,
features were added to provide visible and audible interpretation of the drone’s capa-
bilities: propeller and collision sounds, obstacle avoidance detection, drone check-
point system, and startup/shutdown procedures. By performing this virtual environ-
ment implementation, students were able to increase their ability to adapt their skills
to suit new technologies and how to transpose prior work to updated integrations.
They also gained experience using game engines like Unity® alongside sound gener-
ation software for virtual effects. Finally, simulations were produced with parameters
that would demonstrate the drone’s features for presentation purposes.

3.5 Project Presentation

To finalize the project, the students presented their work to the instructors of the
course. By bringing these ideas together, the presentation encapsulated all efforts
from the collaboration and interdisciplinary work. It included the iterative process,
starting from the literature review and ending with the Unity® implementation. Each
student presented their contributions, demonstrating the collaborative process and
dynamic interaction amongst the group. As with the collaboration, multiple topics
were implemented in the presentation: project purpose, required sensors, safety
features, anthropomorphic features, limitations, and future research. The limitations
and next steps section considered many facets, discussing the project management
side and technical skills/technologies required for future development. These wide-
ranging explanations demonstrated the efficacy of the interdisciplinary work and
collaboration in the classroom setting.

4 Lessons Learned and Conclusion

This paper reflects on an interdisciplinary project conducted in a research-oriented

undergraduate honors course with the aim of encouraging group work among second
semester freshman honors students from different STEM disciplines. With this imple-
mentation, the students demonstrated several notable factors influenced by the class
structure. The level of involvement and interest from the students to invest time
and effort towards topics like research, drones, virtual reality, game engines led
to more progress over a shorter period, exceeding expectations from initial outline
and allowing for more time committed towards smaller details in the process. With
Development of a Safe and Anthropomorphic Drone ... 225

several technologies involved and multiple perspectives from diverse students, they
considered information from multiple backgrounds in different topics influencing
several sections like the literature review, modeling software, virtual requirements,
factors that yielded higher quality results. The quality of work accounted for factors
or features that were out of scope before the implementation of diverse technologies
like digital versatility and edge case handling. This led to more efficient solution
integration: sketches were produced at a higher caliber and significantly assisted the
3D modeling process, the generated 3D models easily transitioned into the Unity®
virtual environment, and the simulations and renderings reflected initial expectation
due to the initial planning already accounting for such needs. Not only did collabora-
tion efforts influence the final deliverables, but the diverse topics also involved in the
process led to students finding support for their decisions through multiple facets,
bolstering their choices while developing a newer level of understanding surrounding
interdisciplinary work.
With the outcomes of the project defined, limitations impacting several aspects
were identified. Specifically, factors surrounding the study could be manipulated to
produce more results or change the accuracy: class size and project duration. The
class size consisted of only seven students, with four students selecting this project
and forming a singular group. This meant that all data and metrics produced and
analyzed resulted from the work of four individuals, which could have restricted the
extent of the underlying work. With a bigger class size, more groups and results
would be produced, assisting in the identification of trends, and leading to more
refined conclusions. The time associated with the project was limited to one half
of a semester (7–8 weeks). This short time period, although challenging students to
be more effective in decision-making and critical thinking, led students to become
concerned about the final product being influenced by the rapid progression of the
initial outline. With a longer project duration, more correlations and validation could
be generated between sections of the project with higher level of development of
the overall result, and students would have more time to explore the complexity of
individual topics, creating the possibility for deeper connections and understanding.
Future work could focus on these limitations. Replicating this study with more
students would allow for several groups and more results to compile and compare,
allowing for higher validation and accuracy. Extending the duration of the project
would result in better understanding of the impact of interdisciplinary work, poten-
tially reaching improved quality outcomes. In addition, having a higher level of
student diversification would enable the involvement of more topics, increasing
the interdisciplinary effort. This could be accomplished by, for example, including
art and psychology students, which allow for topics like anthropomorphism and
presentability to be enhanced by relying on more specialized backgrounds and expe-
rience. By manipulating the selected topics involved with drones in construction,
different subsets could be focused on and analyzed when discussing the effective-
ness of interdisciplinary work. Instead of focusing on building a drone specialized
for construction, the focus could be on topics surrounding the human interaction
aspect which would result from higher level of development in different topics like
social sciences and psychology.
226 G. Albeaino et al.

Acknowledgements We would like to thank Paige Anderson, Richard Huang, and Keith Lim for
providing the necessary data and metrics based on which this work was done.

References

1. Demirdoven J (2015) An interdisciplinary approach to integrate BIM in the construction manage-

ment and engineering curriculum. In: Proceedings of the BIMAS2015-9th BIM Academic
Symposium and Job Task Analysis Review. Washington, DC, pp 211–251
2. Huang Z, Kougianos E, Ge X, Wang S, Chen PD, Cai L (2021) A systematic interdisciplinary
engineering and technology model using cutting-edge technologies for stem education. IEEE
Trans Educ 64:390–397. https://doi.org/10.1109/TE.2021.3062153
3. Jeelani I, Gheisari M (2021) Safety challenges of UAV integration in construction: conceptual
analysis and future research roadmap. Saf Sci 144:105473. https://doi.org/10.1016/j.ssci.2021.
105473
4. Jin R, Yang T, Piroozfar P, Kang B-G, Wanatowski D, Hancock CM, Tang L (2018) Project-
based pedagogy in interdisciplinary building design adopting BIM. Eng Constr Archit Manag
25:1376–1397. https://doi.org/10.1108/ECAM-07-2017-0119
5. MacLaren AJW, Wilson M, Simmonds R, Hamilton-Pryde A, McCarthy J, Milligan A (2017)
Educating students for the collaborative workplace: facilitating interdisciplinary learning in
construction courses. Int J Constr Educ Res 13:180–202. https://doi.org/10.1080/15578771.
2016.1267667
6. Molina C, et al (2014) The use of unmanned aerial vehicles for an interdisciplinary undergraduate
education: solving quadrotors limitations. In: 2014 IEEE Frontiers in Education Conference
(FIE) Proceedings, pp 1–6
7. Underwood J, Khosrowshahi F, Pittard S, Greenwood D, Platts T (2013) Embedding building
information modelling (BIM) within the taught curriculum: supporting BIM implementation
and adoption through the development of learning outcomes within the UK academic context
for built environment programmes. http://www.bimtaskgroup.org/wp-content/uploads/2013/10/
BIM_June2013.pdf. Accessed 14 July 2022
The Integration of 4IR Technologies
in Architectural Education for Upskilling
the Workforce in the Nigerian Built
Environment

Chika Okafor, Clinton Aigbavboa, John Aliu, and Ornella Tanga

Abstract There is an increasing awareness and adoption of Fourth Industrial Revo-

lution (4IR) technologies in the Architectural, Engineering and Construction, Oper-
ations and Management (AECOM) industry. This has placed significant pressures on
academia to develop and produce graduates who are equipped with literacy skills to
seamlessly transit into the world of work after graduation. However, universities in
developing nations such as Nigeria have struggled to integrate 4IR technologies such
as visualization tools into their curricula. This study filled this gap by exploring the
various 4IR technologies that are relevant for teaching architectural designs and 3D
visualization in the architecture discipline. This study further explained how these
technologies can be employed in architectural education to improve students under-
standing of the design and building processes, thus equipping them for the world of
work. The objectives were achieved by reviewing existing theoretical literature on
how 4IR technologies can promote the upskilling of architectural students learning.
Findings from this study revealed that the integration of 4IR technologies in architec-
tural education can enhance students understanding of the design and construction
processes and upskill them for greater productivity in the industry. This study there-
fore recommended the integration of 4IR technologies in architecture education and
generally in AECO disciplines for improved learning outcomes and employability of
the students. This study also offers a reference point as well as directions for future
research in exploring other 4IR technologies for technology-enhanced architecture
pedagogy.

Keywords Architectural education · Pedagogy · Virtual reality · Employability ·

Digital skills · Digitalisation

C. Okafor (B) · C. Aigbavboa · O. Tanga

Centre for Excellence and Sustainable Human Settlement and Construction Research Centre,
Faculty of Engineering and the Built Environment, University of Johannesburg, Johannesburg,
South Africa
e-mail: chikanonyelum.okafor@unn.edu.ng
J. Aliu
Institute of Resilient Infrastructure Systems, College of Engineering, University of Georgia,
Athens, GA, USA

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 227
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_16
228 C. Okafor et al.

1 Introduction

Industry 4.0 has changed several important industries, including how work is done
and how students are educated. For instance, it has changed the nature of activities
that are currently performed manually by replacing them with tasks that are handled
by machines [10]. According to Zhao et al. [53] the one who owns the newest
knowledge, gets more understanding, develops the up-to-date knowledge, and creates
the use-value, including more expertise will be able to obtain a higher position in
the future. To create a workforce that is well prepared and qualified to work in
the current technologically driven period, the education system must provide its
current students with the essential and appropriate skills and competencies [10]
that will make them ready for the world of work. Maintaining speed with the rapid
developments in professional practice around the world is a problem for architectural
education [29]. Thus universities and professional organizations have a responsibility
to advance architectural education, support architectural careers, and improve the
abilities of practicing architects [39]. Likewise, academia must consider continuing
education, and also periodically evaluate and assess the prospective practitioners’
knowledge regarding the most recent developments in architecture-related areas.
These developments include the newest applications, construction techniques and
materials among others that are required in advancing the AECOM [34] industry.
This will motivate the students and practitioners to continue their education, improve
their abilities, and self-learn in order to meet the demands of the world of work [24].
Higher education institutions (HEIs) are perceived as a center for producing skilled
labour for social and international consumption. This is predicated on the idea that
instruction at this stage is essential for enhancing human capital, which expands
job options for graduates in the construction industry [3]. The knowledge-based
economic indicators in societies are one of the true markers of developing a profes-
sional job and making a livelihood using the knowledge acquired via university
education [24]. HEIs have the responsibility to continually evaluate the systems that
enable architects to effectively continue their education so that they can adapt to
changes in the field of architecture and in the Architectural, Engineering, Construc-
tion, Operations and Management sector as they occur. Educators frequently rely
on traditional methods to impart knowledge to students, which typically involve
lecturing students who are to pay attention and take in the information presented
[47: 554]. New technologies may not always be adopted in this context. Tarling
and Ng’ambi [47: 556] explained that new content (curriculum update) and abilities
(technological skills) must be developed to stay up to date with the changing tech-
nologies in the context of education. Tarling and Ng’ambi [47: 556] also described
the significance of sustainable advancement in terms of diverse pedagogies that may
and should be applied in varied circumstances to accomplish deep and meaningful
learning. Venter and van der Wath [49] highlight that there is a common good relation-
ship among professional and educational environments, as emerging technological
skills will need to be developed in studios if educators wish to train professionals who
The Integration of 4IR Technologies in Architectural Education ... 229

are proficient in these developing technologies. Demirkan [17] challenges educators

to be forward planning about how to educate, train, and excite the coming generation.
Technological application in the design studio does, however, enhance interaction,
active learning, and socialization, which are evident in students’ acquisition of design
knowledge, in certain settings like studio educational settings, according to Demirkan
[17]. The student is put at the center of the learning process due to substantial progress
of user modeling and personalization processes, according to Bacca et al. [12]. More
so, with the advancement in technology, students are now able to work together
and simulate or view the different stages of construction works in real time. The
development of software, video conferencing, use of emails, online learning and
other tools that facilitate better teamwork and communication during construction
projects have been beneficial in this [2]. This facilitates the dismantling of social
barriers between students as well as between them and academia. As a result of the
fact that technologies are constantly changing, students must become familiar with
the numerous types of technological and computer applications used in their fields in
order to analyze, design, and communicate effectively [36]. Despite its drawbacks,
emerging technologies can play a catalytic function in transforming pedagogy, as
much as they can also be utilized to support the transmission of pedagogies, as Tarling
and Ng’ambi [47: 270] remind educators. Technology has a big impact on how we
teach and learn, yet educators who do not trust new technologies can frequently make
them less effective. It is possible to criticize emerging technology as over- hyped and
insignificant [7], but their relevance in the current era is revolutionary and will define
how learning and the industry work henceforth.
The current study identifies the fourth industrial revolution technologies (4IR) that
are relevant for teaching architectural designs and 3D visualization in the architec-
ture discipline and how they can be employed in architectural education to improve
students understanding of the design and building processes, thus equipping them
for the world of work. This study is focused on architectural education and the fourth
industrial revolution technologies review, identifying the 4IR technologies’ that can
be integrated in architectural education for enhanced students’ learning outcome and
also for preparing architecture students for the world of work.

Method
Secondary data was adopted to accomplish the objective of this research study.
Keywords including architecture, education, built environment, and 4IR were used
to search for similar literature on the IEE, ISI Web of Science, Emerald, Taylor and
Francis, and SCOPUS databases. These databases were selected because they are the
most well-known and frequently utilized databases for scientific study [20]. After
the search, over 200 publications were found which were carefully assessed to deter-
mine their relevance to the subject of this study. A total of 65 articles were found
eligible as the search was restricted to journals articles and conference publications
between 1996–2022. Additionally, other 4IR publications such as [38, 53] and [26]
were adapted for this study. After thorough examination, a total of 37 articles were
230 C. Okafor et al.

finally considered relevant and significant for this study as they discussed the appli-
cation of 4IR tools in architecture education for improved student’s understanding
of the design and construction process that will equip them for the workforce.

2 The Fourth Industrial Revolution Technologies

in Architectural Education

Our world has undergone an enormous transformation because of the digital era.
Digital tools have completely changed how architectural designs are created, as
well as how structures are envisioned, optimized, evaluated, and built. However, the
current advancements in computer power are forcing design and architecture to deal
with new, smarter technologies and automated systems [46]. Today, digital technolo-
gies are being used more and more in architectural design. These new technologies
are employed in the design process for simulation, visualisation and clash detection
and in the construction phase for project monitoring and also in facility management,
among other varied uses. These technologies are reshaping our societal ideas, infusing
new values into our culture, and altering our way of life [46]. Demirkan [17] states
that using technologies in design studios enhances interaction, active learning, and
socialisation, that are reflected in students’ growth of design knowledge, in certain
settings, like studio educational settings. In addition, Soulikias et al. [46] stresses
that the form-finding procedures are made easier by the 4IR (digitization), as well as
designs which adapts to location and climate needs. A new era of computer technolo-
gies has been ushered in by the movement of media from the manual to the digital,
which has profoundly altered the perception of scale and the relationship between
the human body and architecture (anthropometrics) [46]. Therefore, to advance the
learning and appreciation of architectural design principles, educators should ensure
that students learn to draw using the traditional methods in their early undergraduate
years and fully employ the BIM and other Industry 4.0 technologies in their works
and projects in their second and subsequent years of study. The 4IR technologies
applicable to architectural education will be discussed in the following sections.

2.1 Virtual Reality (VR)

Virtual reality is a great tool which offers students a fresh perspective and real life
experience of what they are studying without even being there directly. According to
Mandal [30], VR is a computer-simulated environment that enables users to engage,
whether that environment is a representation of the real world or a fictional one.
Additionally, Pope [38] defines VR as the use of technology to create the illusion
of being in a non-existent environment. By providing information to several senses,
like sight and sound, the students’ brains are tricked into thinking they are perceiving
The Integration of 4IR Technologies in Architectural Education ... 231

a virtual object. The collaborative approach used in architecture education, which

comes from the contexts of design studios and apprenticeships, is well known [38].
As teaching architecture is a visually focused profession that mixes imagination, VR
technology promises an enhanced imaginative ability of the student. In the current
technological age, virtual reality has become one of the methods used in architecture
education in a variety of ways. VR can be employed to provide students with a form
of experienced based learning where virtual tours to non-existent sites can be taken
using the technology, where students are taught the construction processes without
having to visit an on-going real world construction site. The utilization of VR here
includes using virtual reality equipments such as headsets, and lenses and making a
tour of a building model in a virtual environment [40]. Architecture is one of the fields
that could profit from virtual reality as it enables visualizations of design models in
3D. Virtual reality technology was originally used to help architecture students with
its ability to visualize in three dimensions, which is undoubtedly the most difficult
component of architecture education [43]. This is one way that 3D software programs
(such as Revit and ArchiCAD) that have influenced design in the architecture school
curriculum can impact on VR (when integrated) in architectural education [40].
According to Mateu et al. [32], in the description of the virtual world, information is
acquired through learning by doing in an atmosphere that creates a sense of immersion
that seems real [32]. [1] opines that in architecture field, virtual reality has surpassed
all other methods as the “go-to” method for enhancing a building’s appearance.
Virtual Reality makes the process of design analysis significantly simpler and less
expensive [1].
Potential of Virtual Reality as a Paradigm for Collaborative Learning in Design
Studios

Virtual Design Studio: The design studio have earned recognition for promoting
teamwork up till today. By utilizing technologies like video conferencing, email,
internet publishing, Web3D, and digital modeling, students gain a greater under-
standing of new modalities of media integration and cooperation in design practices.
Virtual design studios will be beneficial for collaborative learning between insti-
tutions in various regions, with the original design studio’s teamwork orientation
[40]. Furthermore, students can learn without utilizing paper in these workshops.
This is accomplished by giving virtual-based software, headsets, and lenses to the
creators of the new virtual environments. Students can participate in designs, virtu-
ally walk through the models to access progress, review their designs’ outcomes,
flaws, and the details [9, 40]. Additionally, instructors use VR to access and grade
students’ projects. Commercial plugins are utilized to easily convert the model from
the simulation software (Revit) to the VR immersive environments [48]. According
to Maghool et al. [28], by simulating a building construction process for example,
students can better assimilate the knowledge gained through the virtual experience.
Maghool et al. [28] further opines that the LADUVR application is one of the inter-
active VR tools that is useful for teaching architectural detailing and the building
process. This tool has three main areas of application, one for depicting the system-
atic process of building construction, the next for clearly showing the architectural
232 C. Okafor et al.

details to learners, and the other for practice, Maghool et al. [28] explains. VR appli-
cation in architectural design would provide a visualisation in real life scale of the
model, thereby providing architecture students a clear view and understanding of the
model [51]. VR gives the full impression of the size of the space, the room height,
luminance of the space among others, affording the young designers an imaginary
feel of the space. This is an advantage over the traditional 2D plan view of drawings,
according to Williams III et al. [51]. In Bashabsheh et al. [13]’s case study at Jordan
University, the authors developed a software for teaching the building process using
VR. The process involved selecting a 2D drawing with structural details which were
developed into 3D models, exported to 3D max for rendering and then imported in
the VR game engine software for the VR experience. Students are able to view the
design from various angles including the aerial perspective by a click of a button.
According to Bashabsheh et al. [13], with a video button, the building process can
be replayed by the student or a pdf download of the process can be made using the
document button. Leal et al. [25] notes that the immersive environment in VR offers
learners the opportunity to assess the quality of the building work. Similarly, the
instructor can employ VR in simulating the building process such that the students
can have a near real site experience in the virtual world. Sirror et al. [45], also confirms
that designers are able to understand spatial relationships, make better choices of the
building finishes, colour and texture and the aesthetics of the building generally with
the use of VR. These are very relevant in architectural studies.

2.2 Augmented Reality (AR)

Augmented reality has the potential to dramatically improve interactive learning,

augmented reality (AR) is a cutting-edge learning tool that is growing in popularity
in educational settings. Because it may be utilized in addition to more traditional
learning tools like text, photos, video lessons, and the internet, augmented reality
has also made a big contribution to blended learning as its use can be combined with
the traditional learning methods such as books, graphical representations, video,
among others [22]. Venter and van der Wath [49] highlights that there is a common
good relationship among professional and educational environments as emerging
technological skills would need to be developed in studios if educators wish to foster
professionals who are proficient in these developing technologies. Demirkan [17]
challenges educators to be think inwardly on ways to educate, train, motivate and
excite the coming generation of professionals.
Architectural Education via Augmented Reality
Although AR is a relatively new technology (created in the 1960s), it is a dynamic
tool for training and instruction that is gaining a lot of pace. Bacca et al. [12] opines
that it can help students become more motivated to learn [12]. Similarly, Lee [26]
notes that AR is indeed a particularly effective technological advancement in higher
education since it may enhance students’ theoretical and practical understanding.
The Integration of 4IR Technologies in Architectural Education ... 233

Lee [26] further asserts that augmented reality (AR) can actively engage students
by incorporating rich contents from computer-generated) 3D surroundings and/or
models. Gurevych et al. [19] opines that AR provides the benefit of “self-study,”
which raises the interest of the students in the teaching materials and, as a result, helps
to stimulate a desire to use contemporary interactive media. Wu et al. [52] emphasize
that AR has the ability to enhance student knowledge and skills, by improving their
investigative skills and so they get a more precise understanding of the subject.
According to Lee [26], AR education lends itself more to constructivist learning,
where students can generate and manage the building of their own knowledge by
leveraging both their prior experience and their own initiative. This can result in a
genuine form of education and training where a context exists and what is intended
by contextualization could be redefined by AR technologies, which can blur the
lines between informal and formal environments as well as inside and outside of the
classroom Wu et al. [52]. The use of AR in architecture design studios can promote
learning as the incorporation of machine-generated 3D environments or models can
be engaging, promoting understanding of the content as well as motivating students
to learn, thereby bridging the gap between theory and practice. Riera et al. [42]
developed an AR application that can be utilized on mobile phones. The authors
asserts that the AR application is able to show several 3D models simultaneously,
thereby making it easy for learners to make a comparison of many design options.
This tool offers the flexibility of switching and flipping the model displayed. Megahed
[33] opines that AR promotes interactive and collaborative learning in the architecture
discipline. With AR, students are able to visualize their design in AR setting by the
use of special AR HHDs and HMDs displays. The interface provides students the
avenue for viewing 3D models which are incorporated in real life settings in a more
imaginative way. Leal et al. [25] suggests the use of an AR application like the
Skope to take the students on a site visit that mirrors a real life site visit where an
explanation of building construction mechanisms and the structural components of
a building are made. The AR magic book can also be employed to enhance learning,
Leal et al. [25] asserts. According to Billinghurst [15], this technology will enable
students to view the magic book as 3D computer-generated models popping out of
the book pages. This technology can assist architecture students in visualizing in
3D. Hajirasouli and Banihashemi [21] put forward that AR can be adapted as a form
of experiential learning where students apply theoretical knowledge (the building
construction) in an AR environment. Therefore, architecture students in this way get
a practical knowledge of the design and building process without necessarily being
engaged in the industry.

2.3 Mixed Reality (MR)

Another technology that can be integrated in architectural education is mixed reality.

MR essentially refers to a sliding immersion scale where, on one end, the observer is
viewing reality through a machine with minimal layouts of virtual information and
234 C. Okafor et al.

augmentation, on the other end, virtual reality, where there is a total immersion into
the digital world with little to no reality [22]. Typically, academics refer to virtual
items as superimposing, coexisting, or merging [12]. It is important to remember that
augmented reality and virtual worlds are not the same thing [11]. It should be clear that
AR is not like a virtual environment, in which the user is fully submerged in a created
environment, but rather enhances (rather than replacing) reality [11]. AR enables
interaction with virtual items that are two- or three-dimensional and merged into the
real world [22]. With the help of the mixed reality technology known as Virtual Touch,
students may create 3D virtual collaboration spaces and actively take part in problem
solving. Utilizing this technology, students are able to set up workgroups where
they can produce and share materials. Furthermore, the manipulation of real-world
things can have an impact on the virtual environment because of the Virtual Touch’s
tangible interfaces. Students can interact and experiment in a learning environment
using physical manipulations so that they may visualize the immediate effects of
their actions [32, 40]. Hosny and Kader [23] opines that MR not only involves the
visualisation and the auditory sense but it also arouses the student’s sense of touch
and feel for an all emcompassing design experience and learning. According to
Hosny and Kader et al. [23] architecture students can sit round a table and are able
to visualize 3D models of their design collaboratively, utilizing transparent headsets.
Therefore, MR promotes collaborations and students’ interactions on a project,

2.4 Building Information Modeling (BIM)

BIM is the design and simulation tool which is key to architecture education in
the technologically advancing age. The implementation of BIM has various bene-
fits for the educational sector. The easiest way to visualize building systems and
components in three dimensions is learning-through-building. Digital portfolios with
BIM projects are used by graduating architectural students when they apply for jobs
or graduate programs as it gives them an advantage and better chances of being
recruited or admitted respectively [27]. Additionally, the built environment industry
employers value graduates who are knowledgeable in the most recent BIM tools
and can utilize these techniques to address integrated design difficulties. Moreover,
increased employment rates are one indicator of BIM’s influence on BIM ready
graduates [27]. In order to improve the built environment industry as a whole, BIM
instruction in architecture programs strives to give the aspiring AECOM leaders
learning programs that can better equip them to deal with cutting-edge technology
and procedures [27]. A successful design workflow depends on being able to switch
back and forth between hand sketching and BIM. At an even higher level, a BIM
model’s completion serves as a starting point for more complex analysis software,
construction management methodologies and post-occupancy as well [27]. BIM inte-
gration in architecture curricula is cost-effective, universities cannot accomplish the
change necessary to conduct efficient BIM training programs without investment
and cooperation between the industry and government. According to Casasayas et al.
The Integration of 4IR Technologies in Architectural Education ... 235

[16], AECOMs must be actively involved in a collaborative culture shift and establish
relationships with HEIs. This will make the investment necessary for architecture
graduates to gain entry to the workforce with the skill sets needed for the BIM-related
employment of the future. Additionally, government support is also necessary [16].
Since BIM is an environment that allows many kinds of simulations, it can also be
used as a teaching tool to teach project management using real world scenarios. It is
very effective, especially when compared to more traditional methods of designing in
the field [35]. Mutai and Guidera [37] points out that with the global BIM trend, it can
be anticipated that there will be an expectation among AECOM professionals that
construction curriculums will be integrating BIM skill development so that graduates
joining the industry will have developed competencies in BIM technology.
Vinsova et al. [50] proposed a framework for fully integrating BIM in the archi-
tecture curriculum of Czech technical university in Prague. The author proposed that
architecture students utilize BIM for the presentation as well as working drawings
and detailing. Knowledge of BIM application for simulations, analysis and collabo-
rations as well as for building services and structures were also recommended by the
author. Similarly, Leal et al. [25], proposes the use of BIM in architectural education
for teaching structural analysis and building materials properties. Furthermore, the
authors opine that BIM can be utilized in assessing the performance of structures,
therefore, determining the best method to be used in construction of buildings. BIM
can also be used to relate the different stages of construction to students by showing
the assembling of the different parts of the parametric model [25]. An advanced
level of BIM learning was recommended for architecture students by Vinsova et al.
[50] such that the students learn to utilize all BIM features owing to its many bene-
fits. According to Aksamija [4], the application of the BIM levels of detail (LOD)
established by the American Institute of Architects is recommended in architectural
education. The LOD 100 for massing, simulation of overall dimensions and volumes,
LOD 200 for simulating the building elements as a whole, LOD 300 for simulating the
building elements with their precise shapes and sizes as parts for assembly, LOD 400
for modeling building elements as parts that can be assembled with precise dimen-
sions, locations in the building and number, giving clear information for their details,
fabrication and assembly and L0D 500 for modeling building elements as completed
structures with precise shapes, sizes and orientation. Aksamija [4], further proposes
the integration of BIM courses into core compulsory architecture courses across
architecture departments due to the benefits that it offers. Similarly, Schroeder [44]
proposes the integration of BIM across the architecture curriculum. Schroeder [44]
stresses that learning of building components using a modular approach (creation of
precise representation of a building prefabricated components) will improve students’
as well as educators’ conception of building materials and their assembly. The use of
node based modeling software such as Dynamo for Revit in architecture schools was
recommended by Schroeder [44] to analyze difficult geometries, minimize students
errors and save student’s time through the automation of repetitive procedures (revi-
sions of sheets). Additionally, Leal et al. [25] asserts that BIM 4D, 5D and 6D videos
can be used to teach students, scheduling, costing and site planning.
236 C. Okafor et al.

2.5 Internet of Things (IOT)

In the building industry, the practice of architecture involves other professions for
decision-making, including civil engineering, mechanical engineering, and electrical,
sanitary, or environmental engineering. However, the education of architects lacks a
systematized foundation to connect to knowledge in fields other than the conventional
ones of construction [18]. As a result, to advance in this area, knowledge integration
(KI) processes is advocated. KI demands the integration of unique IoT hardware,
software and devices for architectural design where user inputs are displayed within
the design space. The IoT gadgets’ low cost makes it possible for them to be used
in many different institutions, and their quick adoption enables students to learn via
discovery [8]. One of the design challenges have been the use of geometrical variables
and abstractions, without taking the surroundings, constraints, and limitations of
the spaces it occupies into account. A conceptual model that has the potential to
transform education and move it toward intelligent education is the use of IoT as a
teaching tool [18]. The IoT paradigm in architecture pedagogy will make it easier to
customize devices for academic purposes and guide the users through each stage and
will represent a significant advancement in the teaching of architecture. The BIM
parametric modeling tool that is frequently used in architecture is the focus of this
implementation in architecture education, as it lowers the implementation complexity
[41]. IoT can be applied in the design development of smart homes and smart cities
according to Bayani et al. [14]. This will enable better data flows and the control of the
building. Furthermore, students in general including architectural students can apply
IoT on a daily basis as it can be employed in monitoring the classroom and design
studios air-conditioning, it helps better remote communication between students and
lecturers and it is employed in promoting campus security (web cameras). IoT is also
applied in providing digital campuses as it was especially utilized during the COVID-
19 pandemic. The technology also improves information access for all students for
a better learning environment [5].

3 Lessons Learned

The mode of teaching in architecture in Nigeria universities is still the tradi-

tional passive system, where students pay attention to the instructor and assimilate
the information passed on without the infusion of technological tools. The adoption
of new technologies will promote active learning whereby interactions and social-
izations can take place and this will result in a more efficient learning of the design
process and the acquisition of technical as well as collaborative and communica-
tion skills. Furthermore, with technology integration in architecture education in
Nigeria, students can learn to collaborate on design projects, perceive and under-
stand the different phases of construction as in real site construction work processes.
The simulation softwares, video conferencing applications, use of email, e-learning
The Integration of 4IR Technologies in Architectural Education ... 237

and other project collaborative and communication tools has been instrumental in
achieving this in the developed country context. Some school of thoughts in academia
opine that in order not to lose the sense of application and to have an appreciation
of the architectural design principles, architecture educators still advocate for the
learning of hand-drafting and sketching in the early design years of architecture
students’ education and the uptake of technological design and simulation tools in
the following years to ensure that HEIs produce well-grounded architects for the
twenty-first century Architectural, Engineering and Construction, Operations and
Management industry. At the technological level, BIM, VR AR and IOT can be
applied for enhanced learning outcomes as well as in preparing architecture students
for the digitally advancing world.
As teaching 3D visualizations with 2D representations is often difficult for educa-
tors and understanding of the concept difficult for undergraduate students in their
early study years, VR integration in architectural design studio will be beneficial.
VR application in architecture design studios creates the impression of a virtual
studio, which is paperless, where students can try several design options without
the use and discarding of several drawing sheets as in the traditional design studios.
Presently, the industry is in high demand of BIM ready architecture graduates who
can effectively utilize BIM technology in the design and construction processes,
therefore, creating opportunities for architects who are BIM proficient for employ-
ment after graduation from HEIs. BIM is not only a design and construction tool
in the AECOM industry but also a useful teaching tool. It is considered a powerful
tool for teaching design, construction processes, scheduling, building materials and
building services as well as for structures. IOT is also relevant for teaching building
services which involves the collaborations and inputs from other design professionals
like the structural, electrical, mechanical (plumbing), HVAC engineers and also the
specialty contractors. Therefore, architecture students will have a grasp of the work
processes in the discipline as well as a versatile knowledge of allied professionals’
input, for better coordination of the design and construction works.
HEIs therefore, have a role to play in upskilling architecture students, the prospec-
tive industry practitioners in the digital era in order for them to excel in practice in
the industry. Students are tasked to be efficient in the use of the emerging technolo-
gies in their disciplines for design and construction processes in order to maintain
their relevance in the industry. Professional practice and education are inter-related
therefore, to promote the use of the 4IR technologies in the industry, the applica-
tions first need to be implemented in architectural studies in HEIs. There is need for
industry-academia-professional bodies collaboration as the application of the 4IR
technologies needs to cut across all HEIs in Nigeria, moreover, their adoption and
integration in architectural education has cost implications which require support
from the industry and the government.
238 C. Okafor et al.

4 Conclusions

The integration of 4IR technologies in architectural education will be beneficial to

the students, the educators and the industry. Many students especially undergraduates
find the comprehension of geometries, visualizations of buildings in 3D, massing,
proportion, and balance challenging. Therefore, concept formation and innovative
designs becomes a difficult task for them especially for students with no prior tech-
nical orientations. This results in most students copying designs from the internet, past
students and even their colleagues as the jury date approaches and they have neither
concept nor the basic design drawings required for presentation. The integration
of the 4IR technologies will greatly improve understanding and visualizations that
will lead to creative and innovative designs. Furthermore, the use of VR in teaching
design will help students visualize their proposed concepts and try out different design
options in immersed environment, thereby being more original in their design. Simi-
larly, AR actively engages students in the design by the application of rich-contents
from the digitalized 3D environment or models that would excite their imaginations
and boost their creativity [26]. BIM technology, the simulation, visualization and
modeling tool can multi-functionally aid students in understanding and undertaking
design, trying out different design options with different concepts, geometries and
materials. It also helps in achieving proportion and massing, the understanding of
building components and the construction processes during the modelling process
and it also increases the speed of drafting as opposed to the traditional hand drawing
and or two-dimensional (2D) CAD drafting.
As architectural design is not complete without the involvement of other building
professionals; the structural engineer, mechanical engineer with specialty in heating
ventilation and air-conditioning (HVAC), electrical engineer, sanitary engineer, in
the architectural education context, knowledge integration model using IoT special-
ized devices for architectural design where users’ actions are integrated and inter-
acted within the digitized design environment is beneficial [18]. This aids the design
students to understand the processes and the building services required to be imputed
for a complete construction drawing and documentations preparation. The adop-
tion and integration of 4IR technologies in architectural education will greatly
enhance learning, creativity, output and employability of architectural graduates.
These 4IR technologies which are applicable in the built environment industry should
be applied to architectural education for enhanced learning outcomes, innovative-
ness and upskilling of architectural students for the work force in the Nigerian built
environment.

References

1. Abdelhameed WA (2013) Virtual reality use in architectural design studios: a case of studying
structure and construction. Procedia Comput Sci 25:220–230. https://doi.org/10.1016/j.procs.
2013.11.027
The Integration of 4IR Technologies in Architectural Education ... 239

2. Ahn E, Kim M (2016) BIM awareness and acceptance by architecture students in Asia. J Asian
Archit Build Eng 15(3):419–424. https://doi.org/10.3130/jaabe.15.419
3. Aigbavboa C, Aliu J (2017) Requisite skills for graduate success: perceptions of the Nigerian
construction industry. Socioeconomica 6(11):21–42. https://doi.org/10.12803/SJSECO.61102
4. Aksamija A (2015) Innovation and BIM in architectural education: teaching advanced digital
technologies to beginner designers. In: Proceedings of the 31st National Conference on the
Beginning Design Student 2015, pp 28–35
5. Aldowah H, Rehman SU, Ghazal S, Umar IN (2017, September) Internet of Things in higher
education: a study on future learning. J Phy Conf Ser 892(1):012017
6. American Institute of Architects. AIA document E202: the American institute of architects
building information protocol exhibit (2008)
7. Amiel T, Reeves TC (2008) Design-based research and educational technology: rethinking
technology and the research agenda. Educ Techn Soc 11(4)
8. An educational application based on virtual reality technology for learning architectural details:
challenges and benefits. Archnet-IJAR Int J Archit Res. https://doi.org/10.26687/archnet-ijar.
v12i3.1719
9. Anderson L, Esser J, Interrante V (2003) A virtual environment for conceptual design in
architecture. In: Eurographics Workshop on Virtual Environments
10. Anshari M, Hamdan M (2022) Understanding knowledge management and upskilling in fourth
industrial revolution: transformational shift and SECI model. VINE J Inf Knowl Manag Syst
52(3):373–393. https://doi.org/10.1108/VJIKMS-09-2021-0203
11. Azuma RT (1997) A survey of augmented reality. Presence Teleoper Virtual Environ 6(4):355–
385. https://doi.org/10.1162/pres.1997.6.4.355
12. Bacca J, Baldiris S, Fabregat R, Graf S (2014) International forum of educational technology &
society augmented reality trends in education: a systematic review of research and applications.
Sour J Educ Technol Soc 17(4):133–149. https://doi.org/10.2307/jeductechsoci.17.4.133
13. Bashabsheh AK, Alzoubi HH, Ali MZ (2019) The application of virtual reality technology in
architectural pedagogy for building constructions. Alex Eng J 58:713–723
14. Bayani M, Leiton K, Loaiza M (2017) The Internet of Things (IoT) advantages on e-learning
in the smart cities. Int J Dev Res 7(12):17747–17753
15. Billinghurst M (2001) The magic book, moving seamlessly between reality and virtuality. IEEE
Comput Graph Appl https://doi.org/10.1109/38.920621
16. Casasayas O, Hosseini MR, Edwards DJ, Shuchi S, Chowdhury M (2021) Integrating BIM in
higher education programs: barriers and remedial solutions in Australia. J Archit Eng 27(1).
https://doi.org/10.1061/(asce)ae.1943-5568.0000444
17. Demirkan H (2016) An inquiry into the learning-style and knowledge-building preferences of
interior architecture students. Des Stud 44:28–51. https://doi.org/10.1016/j.destud.2015.12.009
18. Ferreira LCBC, Branquinho OC, Chaves PR, Cardieri P, Fruett F, Yacoub MD (2019) A PBL-
based methodology for IoT teaching. IEEE Commun Mag 57(11):20–26. https://doi.org/10.
1109/MCOM.001.1900242
19. Gurevych R, Silveistr A, Mokliuk M, Shaposhnikova I, Gordiichuk G, Saiapina S (2021) Using
augmented reality technology in higher education institutions. Postmod Open 12(2). https://
doi.org/10.18662/po/12.2/299
20. Guz AN, Rushchitsky JJ (2009) Scopus: a system for the evaluation of scientific journals. Int
Appl Mech 45(4):351
21. Hajirasouli A, Banihashemi S (2022) Augmented reality in architecture and construction
education: state of the field and opportunities. Int J Educ Technol High Educ 19(1):1–28
22. Hendricks D (2022) Applications of augmented reality as a blended learning tool for
architecture education. Sun South 6(1):79–94
23. Hosny SS, Kader SMA (2004) Integrating intelligent mixed reality in architectural education:
a theoretical model
24. Laal M, Laal A, Aliramaei A (2014) Continuing education; lifelong learning. Procedia Soc
Behav Sci 116:4052–4056. https://doi.org/10.1016/j.sbspro.2014.01.889
240 C. Okafor et al.

25. Leal BMF, Salgado MS, Silvoso MM (2018, February) Impact of the fourth revolution in
architecture undergraduate course. In: ZEMCH 2018 International Conference, Melbourne,
Australia, 29 January–1st, pp 1–12
26. Lee K (2012) Augmented reality in education and training. Tech Trends 56(2):13–21
27. Leite F, Brooks G (2020) Integrating an architectural engineering undergraduate program
with building information modeling. J Archit Eng 26(2). https://doi.org/10.1061/(asce)ae.1943-
5568.0000389
28. Maghool SAH, Moeini SH, Arefazar Y (2018) Archnet-IJAR 12(3):246–272. Regular Section
29. Mamdouh A, Abdelkader M, Samir T (2022) Analysis of the gap in architects’ skills of lifelong
learning. Ain Shams Eng J 13(6). https://doi.org/10.1016/j.asej.2022.101805
30. Mandal S (2013) Brief introduction of virtual reality & its challenges. Int J Sci Eng Res 4(4).
http://www.ijser.org
31. Martínez-Ventura J, De-Miguel-arbonés E, Sentieri-Omarrementería C, Galan J, Calero-
Llinares M (2021) A tool to assess architectural education from the sustainable development
perspective and the students’ viewpoint. Sustainability 13(17). https://doi.org/10.3390/su1317
9596
32. Mateu J, Lasala MJ, Alamán X (2015) Developing mixed reality educational applications: the
virtual touch toolkit. Sensors 15(9):21760–21784. https://doi.org/10.3390/s150921760
33. Megahed NA (2014) Augmented reality based-learning assistant for architectural education.
Int J Adv Educ Res 1(1):35–50
34. Megahed G, Elshater A, Afifi SMZ (2020) Competencies urban planning students need to
succeed in professional practices: lessons learned from Egypt. Archnet-IJAR 14(2):267–287.
https://doi.org/10.1108/ARCH-02-2019-0027
35. Muller M, Loures E, Mendes R, Canciglieri O, Freitas MDCD, Frederico GF (2016) Developing
BIM culture in a university - past and future steps. Adv Transdiscip Eng 4:358–367. https://
doi.org/10.3233/978-1-61499-703-0-358
36. Mumtaz S (2000) Factors affecting teachers’ use of information and communications tech-
nology: a review of the literature. J Inf Technol Teach Educ 9(3):319–342. https://doi.org/10.
1080/14759390000200096
37. Mutai A, Guidera S (2010) AC 2010-1694: building information modeling in construction:
current practices and their implications for construction engineering. Am Soc Eng Educ
38. Pope H (2018) Virtual and augmented reality in libraries. Libr Technol Rep 54(6). https://doi.
org/10.5860/ltr.54n6
39. Rashida Ng (2019) What will it take?: reflections on diversity, equity, and inclusion in archi-
tectural education. J Archit Educ 73(2): 142–144). https://doi.org/10.1080/10464883.2019.163
3189
40. Latif Rauf H, Shareef SS, Othman N (2021) Innovation in architecture education: collaborative
learning method through virtual reality. J High Educ Theory Pract 21(16):33–40
41. Reategui JL, Herrera PC (2021, 25 August) Customized IoT devices for the architectural
education future in connectivity for smart cities. In: 2021 2nd Sustainable Cities Latin America
Conference, SCLA 2021. https://doi.org/10.1109/SCLA53004.2021.9540124
42. Riera AS, Dominguez ER, Escudero DF (2012, 2 November) Developing and augmented reality
application in the framework of the architecture degree. In: UXeLATE 2012, pp 37–42
43. Sala N (2006) Multimedia and virtual reality in architecture and in engineering education.
Paper presented at the Proceedings of the 2nd WSEAS/IASME International Conference on
Educational Technologies, Bucharest, Romania.
44. Schroeder R, Dean W (2019) Comprehensive BIM integration for architectural education using
computational design visual programming environments. Build Technol Educ Soc 2019. https:/
/doi.org/10.7275/7gw6-dt36
45. Sirror H, Abdelsattar A, Dwidar S, Derbali A (2021) A review on virtual reality for archi-
tecture education. In: Proceedings of the 11th Annual International Conference on Industrial
Engineering and Operations Management Singapore, 7–11 March 2021, pp 994–950
46. Soulikias A, Cucuzzella C, Nizar F, Hazbei M, Goubran S (2021) We gain a lot…but what are
we losing? A critical reflection on the implications of digital design technologies. Open House
Int 46(3):444–458. https://doi.org/10.1108/OHI-02-2021-0041
The Integration of 4IR Technologies in Architectural Education ... 241

47. Tarling I, Ng’ambi D (2016) Teachers pedagogical change framework: a diagnostic tool for
changing teachers’ uses of emerging technologies. Br J Educ Technol 47(3):554–572. https://
doi.org/10.1111/bjet.12454
48. Tayeh R, Issa RRA (2021) Developing construction information systems courses based on
collaboration between industry and academia. J Archit Eng 27(3). https://doi.org/10.1061/(asc
e)ae.1943-5568.0000476
49. Venter T, der Wath V (2014) Industry and university cooperative design: a retrospective critique
on a rapid prototyping workshop. In: ICERI2014 Conference, pp 1474–1483
50. Vinsova I, Achten H, Matejovska D (2015) Integrating BIM in education: lessons learned. In:
CAAD Education – Applied, eCAADe 33, vol 2, pp 127–131
51. Williams III JE, Orooji F, Aly SJ (2019) Integration of virtual reality (VR) in architectural
design education: exploring student experience. Paper ID #27354, 126th Annual Conference
and exposition, American Society for Engineering Education
52. Wu H, Lee SW, Chang H, Liang J et al (2013) Current status, opportunities and challenges of
augmented reality in education. Comput Educ 62(2013):41–49. https://doi.org/10.1016/j.com
pedu.2012.10.024
53. Zhao F (2021) Research on the construction of engineering management course system based
on BIM tech. In: ICCREM 2021, pp 479–485
Visualization (nD,VR, AR)
Interactive BIM-Based VR: A Case
Study of Doors

Sou-Han Chen and Fan Xue

Abstract The adoption of virtual reality (VR) in the construction industry provides
an immersive experience for users to view and interact with Building information
modeling (BIM). However, in the current practice, the VR experience is created
manually, which is time-consuming and does not refer to other validated data. This
study proposes an automatic BIM to interactive VR method based on Revit and
Unreal Engine in three steps. First, a bounding box-based segmentation is conducted
in Revit through Dynamo for classifying the motion-bearing components. Then the
BIM along with the segmentation results are imported to Unreal Engine for compo-
nents mapping and interactive behavior selection. Finally, Oculus Quest presents
an interactive BIM-based VR experience to users. A case study of BIM doors was
conducted to validate the proposed method. In the output VR models, users can
control and interact with swing doors through the touch controller of the VR headset.
Four interactive behaviors of the door were realized for interactive and experiencing
user interactions. The findings confirmed the feasibility as well as automation of the
proposed method, for faster BIM-to-VR content creation.

Keywords Building information modeling · Virtual reality · Interactive ·

Motion-bearing components · Revit · Unreal engine

1 Introduction

The architecture, engineering, construction, and operation (AECO) sector is seeking

an efficient visualization approach to enhance the interpretation of construction
projects. The rapid growth of urbanization has fastened and increased the construc-
tion complexity in architectural features, construction methods, and smart facilities
management [1]. With such a high complexity, the traditional design approach grad-
ually cannot meet the users’ expectations of visualization and interaction. While all

S.-H. Chen (B) · F. Xue

The University of Hong Kong, Pokfulam, Hong Kong SAR, China
e-mail: souhan@hku.hk

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 245
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_17
246 S.-H. Chen and F. Xue

the information and 3D geometry are presented on the 2D screen, users sometimes
have difficulties visualizing the special environment of the project, especially for
complex projects [2]. As a result, design faults and inappropriate space utilization
[3] still occur and lead to a low reputation in the sector.
To reduce and alleviate the problems caused by visualization, virtual reality (VR)
technology was introduced and adopted throughout the project lifecycle. Academic
research and industrial applications have been conducted and validate the VR imple-
mentation. For example, facilitating the design process [4–6]; providing immer-
sive visualization [7–10], safety training [11], and providing a virtual environment
for users to interact with the building elements [12, 13]. However, the VR content
creation process is resource-intensive and requires designers to have both AECO
and VR knowledge. Without sufficient domain knowledge, the VR content might be
unrealistic and cannot provide a good experience and leading to the failure of the
VR applications. Therefore, an efficient approach to generating VR content is vital.
Building information modeling (BIM) has been increasingly adopted in the AECO
sector due to its integrational capability. BIM is a comprehensive digital representa-
tion of the physical and functional characteristics of a facility [14]. The benefits and
impacts of BIM have been widely agreed upon by researchers and industries [15–17]
and have become mandatory in many regions [18]. BIM provides a systematic infor-
mation hub for all the involved parties to collaborate at the same project pace. With
the capability of storing accurate, semantic, and detailed data, BIM can represent
a digital twin of reality, where architects, engineers, contractors, and end-users can
simulate and manipulate the design and construction details. Although BIM can be
considered to be a solid foundation of VR content, users cannot have any interaction
with the objects as there is no mechanism and physic engine inside BIM software.
Therefore, there is still a gap between BIM and digital twin applications.
In addition, gaming technology has demonstrated its ability to produce high-end
visualization of architectural features and to provide various interactive scenes for
users to experience the virtual built assets [19–22]. Moreover, many gaming compa-
nies provide great tools for users in the AECO sector to import and link their BIM
projects to the game engines. For example, Epic Games provides an add-on, Data-
smith, for automatically transferring and synchronizing BIM with Unreal Engine
(UE) projects [23]; Unity provides Unity Reflect for real-time collaboration across
platforms and applications [24]. Although such tools can minizine the informa-
tion lost during data transformation, the transferred information is not automatically
utilized for further applications. After importing BIM into the game engine, designers
still need to spend effort on setting the interaction of the objects and configuring the
parameters. This process is time-consuming and not suitable for large-scale projects.
As a result, an automatic workflow for the interaction settings is desired.
This research aims to present a preliminary method to exploit the potential of using
BIM and gaming technologies for automatically facilitating VR content and expe-
rience. The proposed approach is designed to (1) facilitate the VR content creation
process, (2) explore the interactive ability of BIM, and (3) provide a realistic and
diverse VR experience to the users through the game engine.
S.-H. Chen and F. Xue 247

Fig. 1 Method of the proposed three-step automatic BIM to interactive VR process

2 Methodology

2.1 Overview

The proposed method is described below (Fig. 1) and includes three major steps: (1)
object segmentation in BIM software, (2) interaction setting in the game engine, and
(3) VR configuration and deployment.

2.2 Step 1. BIM Object Segmentation

The first step of the method is to classify the motion-bearing components which
will be used as the physic constraints, also known as hinges, in the game engine.
Since BIM can be considered a well-structured information hub [25], the object
data and parameters can be retrieved and analyzed in a systematic approach. For
example, Dynamo, a Revit plug-in for managing and extracting data, can be used
to identify the elements that contain certain text, tags, or under specific categories.
Therefore, users can easily classify the motion-bearing components and stored the
information in a CSV file. However, since the properties of the BIM objects might
not be comprehensive, a bounding box-based segmentation approach can be used to
identify the motion-bearing components.
The bounding box is an invisible cube that contains all model elements, anno-
tation elements, and datum elements [26]. The motion-bearing component can be
identified by checking if its bounding box is intersecting with two or more connected
components. Using the door as an example, hinges can be identified correctly as their
bounding boxes intersect with the panel’s bounding box and the frame’s bounding
box, while the frame will not be classified as motion-bearing components as its
248 S.-H. Chen and F. Xue

bounding box contains the panel’s bounding box. This essential approach is favor-
able for segmenting the BIM objects into different parts as objects in the AECO
sector are generally uncomplicated and more typical.

2.3 Step 2. Interaction Setting in the Game Engine

The target of this step is to transfer BIM objects to interactive objects with the option
that designers can select the interactive behaviors of the objects (Fig. 2). Unlike the
traditional mesh models in OBJ or FBX files, which mainly focus on the geometry of
the model, BIM contains more metadata, such as category names, custom labels, and
unique BIM IDs. Through mapping metadata with the segmentation results generated
in Step 1, key motion-bearing components and other parts of the BIM objects can be
recognized by the game engine and the interactive behaviors can be applied to the
corresponding components. There are two advantages of adopting this pre-defined
interactive behavior, applicable to large-scale transformation and can provide users
with a diverse interactive experience.
Firstly, the interactive behaviors can be custom-built and applied to all the same
object types. For example, once the designers define an interactive behavior of a
push/pull window, all the push/pull windows can be applied to the same settings and
designers don’t need to configure them one by one. Even though the shape or the
size of each push/pull window might be different, the game engine can still use the
segmentation results to recognize the motion-bearing components along with other
parts of the objects and perform the correct behaviors of the objects.
Secondly, through applying different interactive behaviors, the same BIM objects
can perform differently. For example, a BIM furniture can be lifted either by one
touch controller or two touch controllers; a BIM swing door can be opened either by
a simple push or need a key to open it. With the ability to quickly change the behavior
of the objects, designers can effortlessly try out different settings for providing the
most realistic experience to the users.

Fig. 2 General framework of transferring BIM objects to interactive objects

S.-H. Chen and F. Xue 249

2.4 Step 3. VR Environment Settings and Deployment

To successfully develop and deploy the VR content inside the game engine, proper
VR configuration should be set. Generally, there are three essential settings needed to
be configurated sequentially before packing all the content into an Android Package
(APK) file: (1) VR operating system configuration; (2) touch controller settings; (3)
interactive game object settings.
First of all, since different VR companies have their own VR operating system,
designers need to determine which VR headset will they use to provide the interactive
experience. Based on different systems, some plug-ins, packages, or software devel-
opment kits needed to be installed in advance. Secondly, based on the VR operation
system, designers need to define the action of each bottom on each touch controller.
The actions can be classified into two categories: locomotion system and hand inter-
action system. Typically, the locomotion system can be set to either be continuously
moving or teleporting, while the hand interaction system contains various actions.
For example, holding/releasing game objects, controlling game objects to perform
specific behavior, and pausing the game. Finally, the priorities of the interactive game
objects should be modified to match the touch controller settings so that the objects
can behave correctly based on the users’ input.

3 Results

3.1 Case Study of BIM Door

Doors are one of the most important object types for fire safety checking during the
preliminary design of building works. Besides the appearance of the door, deciding
its opening direction and planning the exit route are crucial works for the designer.
Although the adoption of VR technology can provide an immersive experience for
the user to interact with doors, the interactive behavior is still manually designed. As
a result, this research used a BIM door from the UK National Building Specification
(NBS) BIM Library [27] as a case study for validating the proposed method in Fig. 1.
Table 1 lists the software used in this case study.

3.2 Bounding Box-Based Segmentation Results

Figure 3 shows the Dynamo scripts for door segmentation. Based on the general struc-
ture of the door, all the components can be segmented into one of the following cate-
gories, Door_Panel, Door_Hinge, or Door_Frame. To correctly segment the deco-
rations and handle on the panel to Door_Panel, the bounding box of the panel was
extended on the y-axis. Then, by comparing each bounding box of all the components
250 S.-H. Chen and F. Xue

Table 1 Software list

Software Version Description
Revit 2022 Loading BIM door in Revit family document (.rfa) and the supporting the
environment for Dynamo
Dynamo 2.12 Performing the bounding box-based segmentation and exporting the
results in CSV file
Unreal 4.27.2 Creating interactive behavior of the BIM objects; setting VR environment;
Engine Packing the VR model into APK file
(UE)
SideQuest 0.10.32 Installing the APK file into Oculus VR

with the bounding box of the panel, their relationships with the panel can be known.
Figure 4 shows the segmentation results of the test BIM door. The first three rows
show the segmentation results and list out the unique ID of each component, while
the rest rows show the locational data of all the hinges, including the X-axis, Y-axis,
and Z-axis. The unit is in centimeters.

Fig. 3 Dynamo script of bounding box-based segmentation

Fig. 4 Segmentation results. (a) Test BIM door; (b) Illustration of the segmentation results; (c)
segmentation results in CSV file
S.-H. Chen and F. Xue 251

Besides the segmentation results, the BIM object was exported as a Datasmith
file, which is a file type created by Epic Games and used for importing BIM into
Unreal Engine (UE).

3.3 Interactive Behaviors

Unreal Engine provides a visual system, called Dataprep, for designers to pre-process
the Datasmith file before the actual BIM object is imported to the scene. Figure 5
shows the Dataprep system used in this case study. The core of this Dataprep system
is shown in Fig. 5d. These blocks enable designers to select and modify the compo-
nents. For example, designers can select all the components with native BIM material,
Glass_-_Avanti_-_Laminated_Glass, and change their material to native UE mate-
rial, which is more realistic and detailed. In this case study, the Dataprep system
is formed by four main steps: (1) substituting materials; (2) merging components
based on the segmentation results; (3) adding the interactive behavior to the door;
(4) deleting unwanted data. With such a pre-defined Dataprep system, designers can
effortlessly filter unwanted data and reorganize the data structure.
By changing the interactive behavior setting, users can experience different
interactions with the door. These interactive behavior settings were pre-defined in
Blueprint, a visual gameplay scripting system in UE. Designers can make use of the
Blueprint system to custom-build users’ interactive experiences. In this case study,

Fig. 5 Dataprep system for transferring a BIM door to an interactive game object. (a) original BIM
door data hierarchy in Datasmith file; (b) interactive BIM door preview; (c) interactive BIM door
data hierarchy preview; (d) data processing blocks for transferring BIM door to interactive BIM
door
252 S.-H. Chen and F. Xue

four interactive behaviors were created in advance (Table 2). The Blueprint of each
interactive behavior is provided in Fig. 6.
Once all the VR models are properly placed in the scene and the VR configurations
are done, UE can pack the contents into an APK file, which can be used for installing
VR models into Oculus VR. In this case study, SideQuest is used for the APK
installation. The final interactive BIM-based VR is presented to the users through
Oculus Quest 2. Figure 7 shows two screenshots of the interactive BIM-based VR.
Figure 7a shows the initial status of the VR model, while Fig. 7b shows the scene
after 7 s with some doors already performing different interactive behaviors. This
clearly reveals the great potential of the interactive ability of BIM.

Table 2 Interactive behaviors description

Name Trigger Description
Physic door Physic The door contains physic priorities. Once there is any external
force applied to the door, the physic engine will stimulate its
movements and behaviors automatically
Automatic Timer The door will follow the pre-set movements and behaviors based
door (1) on the internal timer
Automatic Collision The door will open automatically once there is any object with
door (2) the collision priorities entering the specific area and will close
once the object leaves
VR locking VR settings The door will behave based on the pre-defined conductions. The
door door can only be opened once the key is in a specific location

Fig. 6 Blueprint for different interactive behavior settings. (a) Physic-based interaction; (b) Time-
based interaction; (c) Collision-based interaction; (d) VR-based with locking function interaction
[28]
S.-H. Chen and F. Xue 253

Fig. 7 Screenshots of the interactive BIM-based VR. (a) The initial scene; (b) Scene after 7 s

4 Limitations

In contrast to the manual VR creation approaches, the case study has validated the
feasibility of the proposed automatic method. However, there are three limitations
in the proposed method that shall be overcome in future studies.
First, this method only works for Revit family documents. Due to the Revit system
data hierarchy, the bounding box-based segmentation approach cannot perform
correctly as detailed components inside family documents will not be passed to
the project document. For example, door hinges might be invisible inside the Revit
project.
Secondly, the bounding box-based segmentation relies heavily on the accuracy
of the model. If the geometric and the location of objects and components are not
accurate, the segmentation results would be affected, and the final VR experience
might not be realistic.
Last but not least, in the current practice, the major panel of the door must be
selected manually. Since the appearance of the door might be diverse, the door
panel cannot be accurately identified by its geometry. Some small experiments were
conducted in Dynamo to automatically identify the panel. For example, based on
the largest volume; based on the largest surface area; based on the components with
nearby components. However, all the segmentation results are still not ideal and not
general.

5 Conclusion

Virtual reality technologies have provided the AECO sector with an immersive expe-
rience of visualizing and interacting with the construction project. However, the
process of VR content creation is time-consuming and requires great domain knowl-
edge in both VR and ARCO knowledge. This study presents an automatic BIM to
interactive VR method with the adoption of BIM and gaming technology. A case
study of BIM door was implemented and validated the proposed method. The results
254 S.-H. Chen and F. Xue

showed the proposed method can facilitate and fasten the VR content creation process
and provide a more interactive VR experience.
To improve the process of BIM to interactive VR, several research directions can
be explored. First, the adoption of the Revit project document can improve the scal-
ability of the VR experience. Furthermore, a more efficient method of segmentation
and auto-BIM-detailing can be developed and investigated. Also, a standardization
of the interactive behaviors can be explored and shared across both academia and
industry.

Acknowledgements The work present in this paper was financially supported by Hong Kong
Research Grants Council (RGC) (No. 27200520).

References

1. Luo L, He Q, Jaselskis EJ, Xie J (2017) Construction project complexity: research trends and
implications. J Constr Eng Manag 143(7):04017019
2. Potseluyko L, Rahimian FP, Dawood N, Elghaish F, Hajirasouli A (2022) Game-like interactive
environment using BIM-based virtual reality for the timber frame self-build housing sector.
Autom Constr 1(142):104496
3. Özmen C, Ünay Aİ (2007) Commonly encountered seismic design faults due to the architectural
design of residential buildings in Turkey. Build Environ 42(3):1406–1416
4. Gan VJ, Liu T, Li K (2022) Integrated BIM and VR for interactive aerodynamic design and
wind comfort analysis of modular buildings. Buildings 12(3):333
5. Davila Delgado JM, Oyedele L, Beach T, Demian P (2020) Augmented and virtual
reality in construction: drivers and limitations for industry adoption. J Constr Eng Manag
146(7):04020079
6. Kim M, Wang X, Love P, Li H, Kang SC (2013) Virtual reality for the built environment: a
critical review of recent advances. J Inf Technol Constr 2013(18):279–305
7. Roach DM, Demirkiran I (2017) Computer aided drafting virtual reality interface. In: 2017
IEEE/AIAA 36th Digital Avionics Systems Conference (DASC), 17 September 2017. IEEE,
pp 1–13
8. Wolfartsberger J (2019) Analyzing the potential of virtual reality for engineering design review.
Autom Constr 1(104):27–37
9. Pratama LA, Dossick CS (2019) Workflow in virtual reality tool development for AEC industry.
In: Advances in Informatics and Computing in Civil and Construction Engineering. Springer,
Cham, pp 297–306
10. Elgewely MH, Nadim W, ElKassed A, Yehiah M, Talaat MA, Abdennadher S (2021) Immersive
construction detailing education: building information modeling (BIM)–based virtual reality
(VR). Open House Int.
11. Sacks R, Perlman A, Barak R (2013) Construction safety training using immersive virtual
reality. Constr Manag Econ 31(9):1005–1017
12. Wong MO, Du J, Zhang ZQ, Liu YQ, Chen SM, Lee SH (2019) An experience-based interactive
lighting design approach using BIM and VR: a case study. In: IOP Conference Series: Earth
and Environmental Science, 1 February 2019, vol 238, no 1. IOP Publishing, p 012006
13. Alizadehsalehi S, Hadavi A, Huang JC (2021) Assessment of AEC students’ performance using
BIM-into-VR. Appl Sci 11(7):3225
14. National Institute of Building Sciences. National Building Information Modeling Standard.
Version 3. https://www.nationalbimstandard.org
S.-H. Chen and F. Xue 255

15. Chan DW, Olawumi TO, Ho AM (2019) Perceived benefits of and barriers to building infor-
mation modelling (BIM) implementation in construction: the case of Hong Kong. J Build Eng
1(25):100764
16. Ghaffarianhoseini A et al (2017) Building information modelling (BIM) uptake: clear bene-
fits, understanding its implementation, risks and challenges. Renew Sustain Energy Rev
1(75):1046–1053
17. Xue F, Wu L, Lu W (2021) Semantic enrichment of building and city information models: a
ten-year review. Adv Eng Inform 1(47):101245
18. Lee G, Borrmann A (2020) BIM policy and management. Constr Manag Econ 38(5):413–419
19. Sepasgozar SM (2020) Digital twin and web-based virtual gaming technologies for online
education: a case of construction management and engineering. Appl Sci 10(13):4678
20. Ma YP (2022) Improved interaction of BIM models for historic buildings with a game engine
platform. Appl Sci 12(3):945
21. Duan H, Li J, Fan S, Lin Z, Wu X, Cai W (2021) Metaverse for social good: a university
campus prototype. In: Proceedings of the 29th ACM International Conference on Multimedia
17 October 2021, pp 153–161
22. Ma YP (2021) Extending 3D-GIS district models and BIM-based building models into
computer gaming environment for better workflow of cultural heritage conservation. Appl
Sci 11(5):2101
23. Boost your pipeline with Datasmith. https://www.unrealengine.com/en-US/datasmith.
Accessed 28 Sept 2022
24. Unity Reflect. https://unity.com/products/unity-reflect. Accessed 28 Sept 2022
25. Heaton J, Parlikad AK, Schooling J (2019) Design and development of BIM models to support
operations and maintenance. Comput Ind 1(111):172–186
26. Autodesk about Bounding Boxes. https://knowledge.autodesk.com/support/revit/learn-exp
lore/caas/CloudHelp/cloudhelp/2021/ENU/Revit-StructEng/files/GUID-C88AAF36-7F89-
4599-8C57-CD9B02390C27-htm.html. Accessed 29 Sept 2022
27. NBS National BIM Library. https://www.nationalbimlibrary.com/en/. Accessed 22 Sept 2022
28. Locked Door in Virtual Reality - Unreal Engine VR Tutorial. https://www.youtube.com/watch?
v=DqRnOC8rylc. Accessed 29 Aug 2022
Leveraging Virtual Reality for Improved
Construction Health and Safety Training

Damien Smuts, Ashvin Manga, and John Smallwood

Abstract The International Labour Organisation estimates, that on average, 60 000

fatal accidents occur on construction sites annually. The importance of developing
and providing health and safety (H&S) training that allows trainees the best oppor-
tunity to understand and appreciate the hazards and risks present on construction
sites as well as to understand and implement the policies and procedures in place to
promote a healthy and safe working environment are fundamental to H&S training
practices. Virtual reality (VR) is an emerging technology with practical applications
to significantly improve understanding and appreciation of H&S training. This study
employed an experimental quantitative approach to evaluate the effectiveness of VR
H&S training by conducting VR training sessions with H&S practitioners within the
construction industry. The VR training software was provided by research and inno-
vation company 3M and included 3 different training sessions. The training sessions
included fall protection and harness inspection, working from heights awareness, and
general construction site H&S induction. A total of 14 construction and H&S practi-
tioners participated in the study including H&S agents, and site agents. Results from
the study indicate that participants found the training experience engaging and real-
istic aiding their learning experience. Additionally, participants experienced limited
difficulty interacting with the platform and navigating within the VR environment
with no participants experiencing any discomfort from the experience.

Keywords Construction · Health and safety · Training · Virtual reality

1 Introduction

Commercial construction constitutes a broad range of shareholders working collec-

tively on a single project often under constraints inclusive of cost, time, and quality.
This unique manufacturing process has recognised the construction sector as an

D. Smuts · A. Manga (B) · J. Smallwood

Nelson Mandela University, Port Elizabeth 6031, South Africa
e-mail: ashvin.manga@mandela.ac.za

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 257
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_18
258 D. Smuts et al.

industry in which health and safety is paramount to project success [1]. Several
health and safety (H&S)-related research projects have emphasised that most acci-
dents on sites could have been mitigated or prevented by implementing the correct
and consistent H&S management processes and a programme that includes planning,
education and training, and inspection [1]. H&S training can leverage visualisation
technology such as virtual reality (VR) to significantly improve the effectiveness of
H&S training. Furthermore, traditional on-site training often interferes with construc-
tion activities thereby reducing productivity and ultimately costing the client more
[2]. A VR approach to construction H&S training allows for improved comprehen-
sion and understanding by providing information in a visual format and offering
virtual off-site hands-on training, mitigating the risks of on-site training and loss in
productivity [2].
Given the status of H&S in South African construction, and that limited, if any
research has been conducted relative to the impact of VR delivered construction H&S
training in South Africa, a study was conducted, the aim of which was to assess the
potential of adopting VR H&S training as an alternative to that of traditional H&S
training. The objectives of the study were to determine the:
• effectiveness of traditional H&S training programmes;
• effectiveness of VR H&S training programmes;
• effectiveness VR H&S training compared to traditional H&S training, and
• extent to which six forms of discomfort were experienced during the VR training
applications, and
• potential to implement VR H&S training.

2 Review of the Literature

2.1 History of Virtual Reality (VR)

In 1957 Morton Heilig designed and patented the Sensorama, a device that allowed
users to mimic riding a bicycle. The user sat in the machine, which exhibited a
3D city, and could hear city noises, feel wind, and seat vibrations, and even smell
individual odours. The Sensorama was the first move toward the commercialisation
of VR [3]. The first head-mounted display was the Philco HMD. It did not use a
simulated world, but instead used a video displayed at a distance creating the effect
of telepresence, a VR application that is still in use today [3]. The next VR related
technology worth mentioning is the development of haptic and gesture feedback by
using wired gloves. The first one developed was the Sayre glove in 1977, by Tom
DeFanti and Daniel J. Sandin, which functioned courtesy of fiber-optics. Thomas G.
Zimmerman worked in the early 80 s to develop the Data-glove, which significantly
influenced other devices, such as the Power Glove. It was manufactured by Mattel for
the Nintendo Entertainment System, available at an affordable price, although it was
considered a commercial failure [4]. The Oculus Rift was the first headset focused on
Leveraging Virtual Reality for Improved Construction Health ... 259

gaming. It provided an extended field of view of 110°, stereoscopic vision, and fast
head tracking. It does that by processing data that passes through a 3-axis gyroscope,
accelerometer, and magnetometer, giving the user a fast image update, meaning no
visual latency. To date, several companies have entered the VR space, most notably
Meta investing significantly into the VR space.

2.2 Virtual Reality Training

Workplace conditions and practices have deteriorated because of industrialisa-

tion, owing to inexperienced, unskilled, and uneducated workforces [5]. As a
result, workers may learn the proper approach to interact with machines and their
working environment through training modules that employ technologically sophis-
ticated methodologies. Furthermore, smart industry 4.0 deployments have added
a new dimension to productivity by providing operators with cost-effective and
time-effective off-site H&S training [6].
Workers in industries receive on-site and off-site H&S training to gain the neces-
sary hands-on skills and practical experience. This leads to an increase in workers’
skill levels, which helps to avoid high-risk accidents in the workplace. However,
statistics suggest that 30–40% of industrial accidents are due to inexperienced and
unskilled operators [6]. This is because current H&S training techniques fail to: repli-
cate the dynamic interaction in a socio-technical system; increase cognitive perfor-
mance learning that is more engaged and knowledge that is retained for a longer
amount of time and provide an interactive platform for determining a strategy for
effective preventative measures [7]. Furthermore, due to variables such as time and
space constraints, organisational costs, and a lack of expert knowledge, these tech-
niques failed to deliver instruction in a real-world environment [8]. With these limi-
tations in mind, VR based training, one of the major dimensions under the umbrella
of Industry 4.0, has emerged as a potential solution due to its ability to provide real-
time visualisation, dynamic accident causation modelling, risk-free human-machine
interaction, and an increased learning rate.
VR-based H&S training has provided a pervasive and persistent platform for
developing perceptual experience and cognitive capacity for problem-solving and
decision-making in complicated stressful circumstances without exposing employees
to real-time hazardous or environmental conditions [9]. Furthermore, H&S training in
an immersive VR environment will aid in the enhancement of process comprehension
and awareness, the emergence of unanticipated issues, the evaluation and anticipation
of errors in decision making, and the detection of workplace dangers. The usage of
VR-based training has replaced traditional training techniques with a user-friendly
and adaptable training platform [10].
Construction workers can now be trained without risk of injury by leveraging
virtual reality technology. Furthermore, when compared to traditional training
methods that rely on movies and handouts, VR is a significantly more immersive
and realistic alternative [11]. As the construction industry is a high-risk industry
260 D. Smuts et al.

for occupational injury, VR H&S training has been targeted with interest by both
academia and industry [11]. A published review of the literature regarding the effec-
tiveness of VR training in 2017 indicated that the technology showed significant
potential, however, the main pitfall regarding the use of VR technology according
to the researchers was access to VR content [12]. Furthermore, it is notable that
in only four years the cost barrier to obtain and develop VR content has decreased
significantly and will continue to decrease as companies such as NVIDIA and Meta
are making massive investments in bringing the metaverse into fruition [13]. The
successful use of VR to improve H&S training is evident across other industries
that require extensive scenario-based training. Then, research concluded that VR
‘drills’ had improved the self-preparedness and self-efficacy of nursing professionals
dispatched to respond to chemical disasters [14]. Research has determined that there
is a cost barrier preventing companies that have limited funds for research and
development to invest in VR [15].

2.3 Learning with Virtual Reality

Recent research provided substantial evidence to suggest that the visual fidelity of a
head mounted display did not impact the users’ learning experience as significantly
as hypothesised by the researcher. Instead, the results indicate that the degree of
interaction and seriousness of the VR experience significantly increased the partici-
pants’ H&S knowledge [16]. The findings from this research are significant as there
is evidence to support the use of affordable VR solutions such as Google Cardboard
which retails in South Africa for approximately R400.00. educators should focus on
the development of interaction and problem-solving type VR environments that can
be experienced using affordable VR displays [16]. One of the drawbacks of using
VR as an educational tool is that users may experience various symptoms that are
similar to motion sickness, disorientation, nausea, pallor, sweating and headaches
leading to vomiting. These symptoms are related to a conflict within the biology of
the brain [17]. However, results provided by researchers [18] indicate that immer-
sive and interactive learning environments increase learning motivation. A notable
finding is that learning motivation plays a substantial role in learner engagement and
retention of the content.
To uncover the usefulness of VR resources in engineering studies, researchers
developed varied surveys that were circulated during seven academic courses to
determine the opinion of students enrolled in diverse engineering degrees at different
universities [19]. The results from a database of 200 students surveyed considered
that the most important features in VR application are interactivity, realism (including
immersion), motivation, ease of use, and educational usefulness. From these results,
several conclusions are listed below:
• Interactivity and realism are the most important features for motivating students
to engage with VR learning;
Leveraging Virtual Reality for Improved Construction Health ... 261

• The students were impressed by the realism of the VR applications developed

by the researchers. However, students demand the use of VR resources as much
as possible to improve their learning experience, according to survey results.
Thus, collaboration between experts in a specific subject and VR technicians are
necessary for designing a useful and attractive VR environment for students, and
• VR is not useful as a stand-alone learning aid, therefore, revised learning method-
ological approaches must be developed to maximise the effectiveness of VR
learning.

3 Research

The aim of the study was to determine the effectiveness of VR H&S training compared
to that of traditional H&S training within the construction industry. The purpose
of VR H&S training is to provide employees with realistic workplace experiences
and allow them to learn from taking risks while working in difficult conditions. The
empirical study was conducted using an experimental quantitative design using a self-
administered pre and post questionnaire designed to determine the impact of the VR
training. The experiment was conducted in a large lecture room at Nelson Mandela
University. Participants partook in three VR applications, which were supplied by
3M™, who created and developed the VR H&S training applications. The three
VR environments included H&S aspects that are common hazards and risks on
a construction site, which included the fundamentals of fall protection, working at
heights awareness, and fall protection harness inspection. Participants were informed
before the start of the VR training that they might experience some discomfort during
the training, namely dizziness, sweating, or a headache, and that the experiment can
be terminated at any stage should the participant experience any discomfort.
The participants spent about 20 min wearing a VR headset while receiving the
H&S training. The participants’ performance in the VR environment was recorded
with their knowledge, but the footage of the actual participants was not captured.
After the VR session was completed, the participants were then given a self-
administered questionnaire to complete, which entailed a series of questions relative
to the objectives of the study.
A total of 14 participants were subjected to the experiment and completed the
questionnaire. The questionnaire was developed in Google forms and used open-
end questions together with Likert Scale and check boxes. In terms of e industry
experience, 6/14 (42.9%) participants had ‘1–10 years’, 42.9% had ‘11–20 years’, and
2/14 (14.2%) had ‘21–30 years’. In terms of injury experience, 71.4% of respondents
indicated they have not been injured on site, while 28.6% have been and 85.7% of
participants had no experience with VR, while 14.3% had.
262 D. Smuts et al.

In response to the question ‘Is there a need for an improvement in H&S training?’
all the participants responded in the affirmative. Table 1 indicates the effectiveness
of traditional H&S training in terms of realising compliance with H&S legislation
and regulations in terms of percentage responses to a scale of 1 (Very ineffective)
to 5 (Very effective), and a MS ranging between a minimum value of 1.00 and a
maximum value of 5.00. Given that all the MSs are >2.60 to ≤3.40, the effectiveness
of traditional H&S training in terms of realising compliance with H&S legislation
and regulations can be deemed to be between less than effective to effective/effective.
Table 2 indicates the participants’ learning experience from training application
(TA) 1 ‘PPE Inspection’, TA 2 ‘working at heights’, and TA 3 ‘harness and fall
protection’ in terms of percentage responses to a scale of 1 (Not at all) to 5 (Substan-
tially), and a MS ranging between a minimum value of 1.00 and a maximum value
of 5.00. All the MSs are >3.00, which indicates that the participants’ learning expe-
rience was between somewhat to substantial. However, given that all the MSs are
>3.40 to ≤4.20, the respondents’ learning experience from the TAs can be deemed
to be between somewhat to a great extent/great extent.

Table 1 The effectiveness of traditional H&S training in terms of realising compliance with H&S
legislation and regulations
Aspect Response (%) MS R
Unsure Very ineffective…….Very effective
1 2 3 4 5
Workers wearing PPE 0.0 11.1 11.1 0.0 66.7 7.1 3.36 1=
Workers identifying hazards 0.0 11.1 11.1 0.0 66.7 7.1 3.36 1=
& risks
Workers operating plant and 0.0 0.0 22.2 33.3 33.3 11.1 3.33 3
equipment
Workers following safe work 0.0 0.0 33.3 33.3 22.2 11.1 3.11 4=
procedures
Workers committing unsafe 11.1 0.0 11.1 33.3 33.3 11.1 3.11 4=
acts
Workers reporting unsafe 0.0 11.1 33.3 33.3 11.1 11.1 2.78 6
conditions

Table 2 Extent to which participants learnt from the three types of VR H&S training
Training type Not at all……………..Substantially MS R
1 2 3 4 5
Working at heights 0.0 0.0 28.6 28.6 42.9 4.14 1=
Harness and fall protection 0.0 0.0 21.4 42.9 35.7 4.14 1=
PPE inspection 0.0 0.0 28.6 42.9 28.6 4.00 3
Leveraging Virtual Reality for Improved Construction Health ... 263

Table 3 Extent to which the site felt realistic

Training type Response (%) MS R
Not at all………………..Extremely
1 2 3 4 5
Working at heights 0.0 0.0 0.0 35.7 64.3 4.64 1
(Cherry picker)
PPE inspection 0.0 0.0 0.0 42.9 57.1 4.57 2
Harness and fall 0.0 7.1 21.4 21.4 50 4.14 3
protection

Table 3 indicates the extent to which the participants felt the training application
(TA) 1 ‘PPE inspection’, TA 2 ‘working at heights’, and TA 3 ‘harness and fall protec-
tion’ sites felt realistic in terms of percentage responses to a range of 1 (Not at all) to
5 (Extremely), and a MS ranging between a minimum value of 1.00 and a maximum
value of 5.00. All the MSs are >3.00, which indicates that the participants felt that
the respective training type scenarios were between somewhat to extremely realistic.
However, given that all the MSs are >3.40 to ≤4.20, the respondents’ learning expe-
rience from the TAs can be deemed to be between somewhat to a great extent/great
extent.
78.6% of respondents opine that it is easier to learn using VR as a tool, while
14.3% responded in the negative, and 7.1% were unsure. Respondents were required
to indicate whether VR training is more beneficial than traditional training. 71%
responded in the affirmative, 14.3% in the negative, and 14.3% were unsure.
Table 4 indicates the extent to which various contentions apply in terms of
percentage responses to a scale of 1 (Minor) to 5 (Major), and a MS ranging between
1.00 and 5.00. All the MSs are >3.00, which indicates the potential is more major
than minor. However, four of the MSs are >4.20 to ≤5.00, which indicates the extent
to which the contentions apply is between a near major to major/major extent—
‘learning is easier using VR as a tool’, ‘there is a need for an improvement in
H&S training’, ‘the VR training application simulated a real site’ and ‘VR training
enhances appreciation of workplace risks’. Given that the MS for ‘VR training
assists in terms of understanding hazards’ is >3.40 to ≤4.20, the extent to which
the contention applies is between some extent to a near major/near major extent.
Table 5 indicates the degree of ease experienced in terms of navigating around the
VR site in terms of percentage responses to a scale of 1 (Not at all) to 5 (Extremely),
and a MS ranging between 1.00 and 5.00. Given that the MS is >4.20 to ≤5.00,
the respondents experience can be deemed to be between very easy to extremely/
extremely easy to navigate around the site.
264 D. Smuts et al.

Table 4 Extent to which various contentions apply

Aspect Response (%) MS R
Minor………………………..Major
1 2 3 4 5
Learning is easier 0.0 0.0 0.0 27.3 72.7 4.73 1
using VR as a tool
VR training is more 0.0 0.0 9.1 18.2 72.7 4.64 2
beneficial than
traditional training
There is a need for an 0.0 0.0 14.3 28.6 57.1 4.43 3
improvement in H&S
training
The VR training 0.0 7.1 7.1 35.7 50 4.29 4
application simulated
a real site
VR training enhances 0.0 7.1 14.3 28.6 50.0 4.22 5
appreciation of
workplace risks
VR training assists in 0.0 7.1 21.4 28.6 42.9 4.07 6
terms of
understanding
hazards

Table 5 Degree of ease experienced in terms of navigating around the VR site

Response (%) MS
Not at all ……………….Extremely
1 2 3 4 5
0.0 0.0 14.3 28.6 57.1 4.43

Table 6 Extent to which the respondents were immersed by the VR surroundings

Response (%) MS
Not at all …………………..Major
1 2 3 4 5
0.0 0.0 14.3 28.6 57.1 4.43

Table 6 indicates the extent to which the respondents were immersed by the
surroundings in the VR training application in terms of percentage responses to a
scale of 1 (Not at all) to 5 (Major), and a MS ranging between 1.00 and 5.00. Given
that the MS is >4.20 to ≤5.00, the extent of immersion in the surroundings can be
deemed to be between a near major extent to major/major extent.
Leveraging Virtual Reality for Improved Construction Health ... 265

Table 7 indicates the extent to which six forms of discomfort were experienced
during the VR training applications in terms of percentage responses to a scale of
1 (Not at all) to 5 (Substantially), and MSs ranging between 1.00 and 5.00. All the
MSs are <3.00, which indicates the extent to which discomfort-related factors were
experienced by the respondents during or after the VR simulation is negligible as
opposed to substantial. Sweating achieved the highest MS (1.64), followed closely by
eyestrain (1.43). All the MSs are ≥1.00 to ≤1.80, which indicates that the experience
of the six forms of discomfort during or after the VR simulation are between not at
all to very little.
Table 8 indicates the potential for the use of VR for training in the construction
industry in terms of percentage responses to a scale of 1 (Minor) to 5 (Major), and a
MS ranging between 1.00 and 5.00. The MS is >3.00, which indicates the potential is
more major than minor. However, given that the MS is >4.20 to ≤5.00, the potential
for the use of VR training in the construction industry can be deemed to be between
near major to major/major.

Table 7 Extent to which six forms of discomfort were experienced during the VR training
applications
Form Response (%) MS R
Unsure Not at all…………………Substantially
1 2 3 4 5
Sweating 0.0 50.0 35.7 14.3 0.0 0.0 1.64 1
Eyestrain 0.0 64.3 28.6 7.1 0.0 0.0 1.43 2
Fatigue 0.0 92.9 7.1 0.0 0.0 0.0 1.07 3
Headache 0.0 100.0 0.0 0.0 0.0 0.0 1.00 4=
Nausea 0.0 100.0 0.0 0.0 0.0 0.0 1.00 4=
Dizziness 0.0 100.0 0.0 0.0 0.0 0.0 1.00 4=

Table 8 The potential for the use of VR for training in the construction industry
Response (%) MS
Minor………………………..Major
1 2 3 4 5
0.0 7.1 7.1 28.6 57.2 4.36
266 D. Smuts et al.

4 Discussion

This research was conducted in 2021, a year in which the world was recovering from
a global pandemic in which nationwide lockdowns impacted businesses across South
Africa. The construction sector was no exception to the economic turmoil Covid-
19 brought and subsequently most construction entities were under operational and
financial strain limiting their willingness to participate in academic research. Further-
more, the research ethics clearance process introduced time constraints and tight time-
lines. However, the convenience sample adopted for the study managed to garner 14
active construction H&S practitioners, therefore, the findings can be deemed indica-
tive as opposed to representative. A notable finding is that 85.7% of the participants
had never experienced any form of VR or VR delivered training. Additionally, not
a single participant experienced discomfort of any sort. This indicates the improved
comfort head mounted display (HMD) VR headsets have made in improving user
comfort which was once a significant issue for HMD’s. As stated in the literature the
main benefit of VR training is exposing users to immersive, dynamic and dangerous
scenarios in a safe and secure environment. This affords the trainee the opportunity to
make mistakes and experience the consequence of their actions. This level of immer-
sion promotes appreciation of the dangers on a construction site, understanding of
the procedures established to keep workers safe and holistic awareness of the task.
New findings from this research indicate that VR technology within the construction
health and safety training sector is in its infancy in terms of adoption.

5 Conclusions

Given the effectiveness of traditional H&S training in terms of realising compliance

with H&S legislation and regulations relative to six aspects, it can be concluded that
traditional H&S training is not effective as it should be. This conclusion is reinforced
by the unanimous affirmative response relative to the need for an improvement in
H&S training. Given the extent to which the participants learnt from the VR H&S
training applications, the extent to which VR sites felt realistic, the benefits of VR
H&S training, the extent to which various contentions apply, and the potential for
the use of VR for H&S training in the construction industry, it can be concluded that
VR H&S training is realistic, preferable to traditional H&S training, and VR H&S
training has the potential to improve H&S compliance, and H&S performance.
Visual fidelity did not have a significant impact on the learning experience as
per the findings of previous research [16]. However, the degree of interaction and
seriousness of the VR experience had a significantly greater impact on learning
retention and understanding [16]. The findings of this study corroborate with the
findings of other studies reported on in the review of the literature confirming the
importance of interactivity and realism in positively influencing the VR learning
experience [14]. Furthermore, the findings of this research underscore the findings
Leveraging Virtual Reality for Improved Construction Health ... 267

of previous research [14] in that subject matter and VR specialists are required
to develop impactful VR learning experiences to enable the research methodology
adopted, which entailed collaboration with 3 M™ who developed the VR training
used in this study.

6 Recommendations

VR assisted H&S training should be adopted in construction as it transforms the

presenter’s role from information deliverer to facilitator, assisting trainees in their
exploration and learning. This allows trainees to explore a virtual world where they
may learn experientially and at their own pace, avoiding the situation where trainees
are left behind during the training session, and spend the remainder of the session
trying to catch up. The resources available to create virtual environments are signif-
icantly more abundant. The development of VR environments is a skill that should
be present within a AEC business as vertical integration of digital skills is easy to
accomplish compared to supply chain integration.
Employer and professional associations, and statutory councils in the construction
industry should raise the level of awareness with respect to Industry 4.0 technologies
and promote the adoption thereof. Stakeholders that adopt industry 4.0 technologies,
VR assisted H&S training included, should interrogate, and quantify the benefits of
such adoption. Tertiary education institutions should collaborate with such stake-
holders, in addition to spearheading such research, and exposing students to VR
H&S training during their studies. In closing, previous research opens the door for
future research in terms of the development of affordable VR learning solutions such
as Google Cardboard, which places an emphasis on interaction and seriousness,
averting the procurement of expensive high-fidelity HMD [11].

References

1. Park CS, Kim HJ (2013) A framework for construction safety management and visualization
system. Autom Constr 33(1):95–103
2. Guo H, Yu Y (2017) Visualization technology-based construction safety management: a review.
Autom Constr 73(1):135–144
3. Rojas-Sanchez MA, Palos-Sanchez PR, Folgado-Fernandez JA (2022) Systematic literature
review and bibliometric analysis on virtual reality and education. Educ Inf Technol 27(1):1–38
4. Goncalves YA, Boas V (2022) Overview of virtual reality technologies. Comput Sci 1(1):1–6
5. Rossini M, Costa APS, Tortorella G (2019) Industry 4.0 and lean production: an empirical
study. IFAC-PapersOnLine 55(13):42–47
6. Manca S, Ranieri M (2013) Is it a tool suitable for learning? A critical review of the literature on
Facebook as a technology-enhanced learning environment. J Comput Assist Learn 29(6):487–
504 (2013)
7. Laberge EM, Calvet B (2014) Why are occupational health and safety training approaches not
effective? Understanding young worker learning processes using an ergonomic lens. Saf Sci
68(1):250–257 (2014)
268 D. Smuts et al.

8. Hou L, Chi HL, Chai J, Panuwatwanich K, Wang X (2017) A framework of innovative learning
for skill development in complex operational tasks. Autom Constr 83(1):29–40
9. Hajiseyedjavadi F, Zhang T, Agrawal R, Knodler M, Fisher D, Samuel S (2018) Effectiveness
of visual warnings on young drivers hazard anticipation and hazard mitigation abilities. Accid
Anal Prev 116(1):41–52
10. Dhalmahapatra K, Maiti J, Krishna OB (2021) Assessment of virtual reality based safety
training simulator for electric overhead crane operations. Saf Sci 139(1):105241
11. Rokooei S, Shojaei A, Alvanchi A, Azad R, Didehvar N (2023) Virtual reality application for
construction safety training. Saf Sci (157):105925
12. Guo H, Yu Y, Skitmore M (2017) Visualization technology-based construction safety
management: a review. Autom Constr 1(73):135–144
13. Omniverse platform for Virtual Collaboration (n.d.) NVIDIA. https://www.nvidia.com/en-us/
omniverse/. Accessed 29 Jan 2023
14. Chang C-W et al (2022) Effectiveness of the virtual reality chemical disaster training program
in emergency nurses: a quasi experimental study. Nurse Educ. Today (119):105613
15. Manjrekar S, Sandilya S, Bhosale D, Kanchi S, Pitkar A, Gondhalekar M (2014) Cave:
an emerging immersive technology – a review. In: 2014 UKSim-AMSS 16th International
Conference on Computer Modelling and Simulation
16. Buttussi F, Chittaro, L (2018) Effects of different types of virtual reality display on presence
and learning in a safety training scenario. IEEE Trans Vis Comput Graph (2):1063–1076
17. Liagkou V, Salmas D, Stylios, C (2019) Realizing virtual reality learning environment for
industry 4.0. Procedia CIRP 79(1):712–717
18. Huang HM, Liaw SS (2018) An analysis of learners’ intentions toward virtual reality learning
based on constructivist and technology acceptance approaches. Int Rev Res Open Distance
Learn 19(1):91–115
19. Vergara D, Rubio MP, Lorenzo M, Rodríguez S (2020) On the importance of the design of
virtual reality learning environments. Adv Intell Syst Comput 1007(1):146–152
Concept and Implementation
of BIM-to-World Co-registration
for Mixed Reality Applications

Marat Khairtdinov, Fulya Tasliarmut, and Christian Koch

Abstract Mixed reality tools have witnessed a surge in meaningful applications

within the construction industry. A critical prerequisite for these tools is the proper
alignment or co-registration of the real world with the virtual and digital content,
a process referred to as co-registration at the beginning, and tracking during users’
movement. Notably, the majority of showcased use cases focus primarily on super-
imposing innovative virtual design content onto tablets (AR) and see-through devices
(MR) that have already been accurately registered and aligned, without providing
insights into the registration process. Although most of the available MR devices
perform well in tracking the device’s position and orientation during user movement,
the co-registration procedure is usually manual, tedious, and constitutes a significant
shortcoming in the overall user experience and MR tool adoption. This paper presents
a distributed client-server concept for a co-registration process for MR devices. The
server process is bifurcated into a semi-automated rough registration part that uses
minimal user input, and a fully automated fine registration part employing the Iter-
ative Closest Point algorithm. The client part runs the projection, visualization, and
tracking process utilizing the initial co-registration data. The presented concept is
implemented and tested on the Microsoft HoloLens as the MR-client and a software
server that runs a dedicated Python script. Experiments and results demonstrate that
the presented concept and implementation effectively support the co-registration
process with minimal user input, offering reasonable accuracy and latency, even
when the corresponding part of the real world has not been captured previously.

Keywords Augmented reality · Mixed reality · Building information modeling ·

Co-registration · HoloLens

M. Khairtdinov · F. Tasliarmut · C. Koch (B)

Chair of Intelligent Technical Design, Bauhaus-Universität Weimar, Weimar, Germany
e-mail: c.koch@uni-weimar.de

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 269
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_19
270 M. Khairtdinov et al.

1 Introduction

The concept of Building Information Modeling (BIM) implies the creation, integra-
tion, and continuous use of digital data throughout the entire life cycle of a facility
including design, building, operation, and maintenance [1]. Often visually repre-
sented with the MacLeamy curve, the BIM workflow aims to shift the effort and cost
related workload from the later stages of construction to its design phase [2]. While
this enables enhanced architectural, structural, and financial planning, the application
of BIM to on-site construction management is still in the developmental stage. Next
to digitization, automation is one of the most important aspects of the BIM work-
flow. Within the context of the Architecture, Engineering and Construction (AEC)
industry, many solutions rely on extended reality (XR) technologies including virtual,
augmented, and mixed reality (VR, AR, MR). The integration of XR tools with BIM
not only play an important role in terms of better spatial understanding through real-
time data visualization, they also contribute to improving the communication on-site
and to enhancing decision-making processes [3].
Despite advancements in the sector, progress monitoring remains labor-intensive
and time-consuming. To this date, it is more often performed by manually acquiring
data on-site for later evaluation. To automate progress monitoring, the most straight-
forward approach is to compare geometric structures between the as-modeled 3D data
with the as-built shape through superimposition. Incorporating digital information
into the real-world surroundings of a user is the core of MR experiences. For real-
time inspection and monitoring, MR devices such as the head-mounted Microsoft
HoloLens or hand-held tablets can be utilized. To overlay as-planned models as
holographs on the corresponding as-built environment, an accurate and precise posi-
tioning of the virtual and the physical objects is essential. The process of aligning
a set of captured points from a MR device to a corresponding set of points from
a given as-planned model defines the task of co-registration. This process usually
involves either the detection of pre-defined markers in the captured environment
or considerable user input to estimate a suitable initial guess for the registration
algorithm.
In this work, we demonstrate our approach to conduct a semi-automated co-
registration of the as-built and as-planned BIM data. We employ the Microsoft
HoloLens hardware and a performance-efficient client-server architecture to offload
the computation-intensive task of initial pre-registration and fine-registration of the
environment with the as-planned building information model. We commence with
an overview of current registration methods and related work, followed by an outline
of the methodology of our proposed approach. Furthermore, we verify the concept
on experiments by creating synthetic and world point cloud data of a real facility.
Concept and Implementation of BIM-to-World Co-registration … 271

2 Related Work

Point set registration defines the task of aligning two (i.e., pairwise or co-registration)
or more (i.e., group-wise) set of points. An overview of registration algorithms has
been shown in the works of Zhu et al. [4]. The objective of point set registration is to
detect correspondences between point sets and calculate the transformation needed
to map one point set onto another. Registration can be performed between images,
from a 2D image to a 3D point cloud or between point clouds. Raw data for 3D point
clouds are often obtained from sensors like 3D laser scanners, RGB-D cameras,
Time-of-flight (ToF) cameras, or from photogrammetric reconstruction algorithms
such as Structure from Motion (SfM). The computation of a rigid transformation
between sets of points can be decomposed into a translation and a rotation, while
the calculation of non-rigid transformations poses an arduous challenge including
affine transformations (e.g., scaling, shearing) [5]. Depending on the use case, the
registration process is typically divided into two main stages: (i) the coarse and (ii)
the refined alignment [6].
The traditional way to coarsely register a scanned point cloud to a reference
model is by manually adjusting the point set to be registered onto the physical coun-
terpart until a sufficiently accurate alignment is achieved. The process is complicated,
time-consuming, and there exists a chance to introduce inaccuracies due to human
error. For the initial co-registration of a point cloud, corresponding points are either
manually selected or 3D descriptor-based or descriptor-free methods are applied [7].
After extracting salient key points through feature detection algorithms, the interest
points are computed with incorporation of their local neighborhoods, which can
then be matched through a similarity metric. Robust 3D descriptor methods include
Fast Point Feature Histograms (FPFH) [8] and Signature of Histograms of Orienta-
tions [9]. One prominent example by Mahmood et al. combined FPFH as a feature
descriptor with the Random sample consensus (RANSAC) algorithm to filter out false
correspondences during the pre-registration process [6]. Descriptor-free methods aim
to perform a global co-registration through the use of Principal Component Analysis
(PCA) with which the rotation for the alignment is estimated [7]. The translation is
then calculated by computing the centroids of the respective point clouds.
Another approach is to utilize markers that are attached to other objects within the
device camera’s field of view and apply pose estimation algorithms to calculate the
relative distance between markers and the camera. However, marker-based methods
are not robust when occlusion or varying lighting conditions are introduced [10]. In
particular, overcoming viewpoint and pose deformations within different models as
well as filtering out noise in the form of occlusion and outliers is notably cumbersome
with markers.
Following the initial coarse registration, a refinement stage can further improve
the accurate superimposition. One of the most widely used methods for fine point set
co-registration is the Iterative Closest Point (ICP) algorithm [11]. The optimization
approach calculates the rigid transformation in an iterative manner to align an unregis-
tered point cloud to a reference cloud. In an initial step, corresponding data points are
272 M. Khairtdinov et al.

identified under the assumption that the closest points correspond. The aim is to mini-
mize the objective function, which defines the distance between two corresponding
points in their respective point sets. By computing the center of mass, the scanned
point set is shifted to the target point set and subsequently rotated by using singular
value decomposition (SVD). After each iteration, the correspondence is updated and
the procedure is repeated either until a maximum number of iterations or a threshold
is reached. Being sensitive to the initialization, one of the main shortcomings of the
ICP algorithm is that it can get stuck in local minima under adverse conditions [12].
In addition, due to the explicit calculation of the closest points, the time complexity
scales quadratically with the number of the corresponding points [13]. Many vari-
ants of the ICP algorithm have been developed to overcome its limitations. One often
utilized metric for better performance of ICP is the point-to-plane distance coupled
with the least squares approach instead of the point-to-point measurement [14].

3 Methodology

The task of co-registering as-built data with as-modeled data can be expressed
as aligning two 3D shapes under the rigidity constraint. Given two 3D shapes in
arbitrary positions within the same Euclidean coordinate system, we aim to find a
transformation of the first shape that aligns it with the second.
The task of co-registration of as-built data with as-modeled data using a portable
MR-device implies that the system needs to obtain the shape of the environment
(as-built data) and the initial transformation of this shape. As depicted in Fig. 1, in
the proposed method, the process of co-registration is divided between a client and
a server application that share the overall tasks as follows.
The client application deployed on the MR device (Microsoft HoloLens) recon-
structs the 3D-shape of the real room (world point cloud) and performs tracking of

Mobile MR device (Client)

As-designed
BIM
Initial World Co-
Create world Apply trans-
Real trans- point registered
point cloud formation
room form. cloud BIM

BIM Final
As-designed Create synth. Rough co- Fine co-
point trans-
BIM point cloud registration registration
cloud form.

Desktop workstation (Server)

Fig. 1 The process of semi-automated co-registration divided between the client and the server
application
Concept and Implementation of BIM-to-World Co-registration … 273

its camera inside of this environment. It also contains the 3D shape of the corre-
sponding model (e.g. as-designed BIM) that has to be transformed to align it with
the environment (co-registered BIM).
On the server side, the shape of the as-designed BIM is used to create a synthetic
point cloud (BIM point cloud) that is, firstly, manually and roughly aligned using
an initial transform of the MR device, and secondly, automatically and accurately
aligned during the fine co-registration process using the world point cloud. The
latter process results in a final transform that is sent to the MR device to align the
as-designed BIM with the world environment (co-registered BIM).

3.1 Creation of Synthetic Point Cloud

3D solids have surface boundaries, which represent the geometry of the as-designed
BIM. These surfaces can be transformed into 3D meshes that consist of points or
vertices that are grouped into polygons, for example, triangles. This process is called
tessellation. The result of the tessellation is a synthetic point cloud of the as-designed
BIM. In order to create a certain compatibility between the world and the synthetic
point cloud in terms of point density, the synthetic point cloud must be up-sampled.
The process of up-sampling is depicted in Fig. 2, and is performed on the server.

(a) (b) (c)

Fig. 2 The up-sampling process of a triangular mesh with the corresponding effects in the 3D point
cloud. (a) Original resolution of the point cloud. (b) Point cloud up-sampled by two times. (c) Point
cloud up-sampled by four times
274 M. Khairtdinov et al.

Fig. 3 Spatial mapping of the environment using the HoloLens results in a mesh (a), or point cloud
(b), respectively

3.2 Creation of World Point Cloud

The creation of the world point cloud can be performed using MR devices. In case of
the HoloLens, this process is called spatial mapping (Fig. 3a) and results in a mesh
that represents the inner surface of the environment, e.g. a real room. Accordingly,
the vertices or points of this mesh form the world point cloud (Fig. 3b).

3.3 Co-registration

As mentioned above, the co-registration process is a semi-automated process as

it is divided into a manual client-based rough alignment and a server-based fine
registration procedure.
The idea is that the user enters the construction site and obtains the 3D scan of
the environment using the device’s built-in sensor as described in Sect. 3.2. This 3D
scan lies in the coordinate system defined by the device: the position of the device
at the initial moment of tracking is the origin of this coordinate system (see Fig. 4).
After the user has entered the first room, they must initiate registration by scanning
the environment to obtain the 3D world point cloud. Then, the user informs the system
of their location and the pre-defined direction (e.g., north) they are facing. Knowing
both the direction (Fig. 4b) and the room on the floor (Fig. 4c), the system aligns the
center of the corresponding room with the position and orientation of the camera,
assuming the user to be at the center of the room. The position and orientation can
be easily tracked in the real environment by the MR device. The obtained initial
transform is then sent to the server and provides the basis for the fine registration
process.
The process of fine registration is carried out on the server using the ICP algorithm
[11], which takes two point clouds of similar shape and density at arbitrary positions
as input. In our case, it strives to align or co-register the world point cloud with the
synthetic BIM point cloud.
Concept and Implementation of BIM-to-World Co-registration … 275

Fig. 4 The process of manually and roughly co-registering the as-designed model with the real
world environment on the client MR device

4 Implementation

This section describes the actual implementation of the proposed method. In order
to share the computational load of scanning and co-registering, a client-server
architecture has been chosen (Fig. 5).

Fig. 5 Software design for the server running a Python script (a) and the client running a Unity3D
program (c). User interface of the server (b) and the MR client (d)
276 M. Khairtdinov et al.

16 bit

16 bit
32 bit 32 bit 32 bit Packet size Residual size

Residual
Number of

size
Packet

packets
size

Residual
packet

packet
Regular
Command

Data type

Fig. 6 Structure of the data transfer protocol for sending data between client and server

4.1 Client-Server Architecture

The co-registration tasks are separated as follows. The server app performs the ICP
algorithm (Registration manager) based on the initial transform and the world point
cloud (Fig. 5a), and the client app runs the spatial mapping and the initial registration
based on user input (Fig. 5c). For that purposes, corresponding input data handlers
and input/output gates have been implemented. The server runs a dedicated Python
script (Fig. 5b), whereas the client is based on a Unity3D program using the MR
HoloLens library (Fig. 5d).

4.2 Data Transfer

Figure 6 depicts the structure of the data protocol used for the transfer of the transfor-
mation matrices (initial transform from client to server; final transform from server
to client) and the world point cloud data that is sent from the client to the server.

5 Case Study and Results

This section describes the case study that was used to test the proposed method as
well as the results of the co-registration process.

5.1 Case Study

The experiments were conducted on the Bauhaus-Universität Weimar campus, as

shown in Fig. 7. A specific room was selected for this purpose; in this case, room
number 2 in Fig. 7.
Concept and Implementation of BIM-to-World Co-registration … 277

Fig. 7 Floor in a university building used in this case study, highlighting the specific test room
(number 2)

Figure 8 gives an impression on both the user interface of the HoloLens app
(Fig. 8a) as well as the qualitative results of co-registration (Fig. 8). In Fig. 8b, it
can be observed that the real and virtual objects of both the cupboard and the table
appear to be aligned and co-registered. However, a slight misalignment can also be
recognized, which is in this particular case, also related to the imperfect as-designed
modeling dimensions.

Fig. 8 (a) User interface of the HoloLens MR device. (b) Qualitative result of the co-registration
process showing the superimposed as-designed BIM model onto the real world scene
278 M. Khairtdinov et al.

6 Results

Figure 9 illustrates the exemplary results of the manual, initial, and rough registration
process that, as previously described, is based on user input. It can be observed that
a simple user input, here room and orientation, can be used to roughly co-register
virtual and real content, and then serve as qualified input for the next step.
The results of the fine registration process are quantified as the distance error
between world scan points and surface planes in the as-designed BIM based on
Eq. (1)
┌
| n
|1 ∑ )2
R M S E point− plane =√ n t,i • ( ps,i − pt,i (1)
n i=1

where nt represents the normal unit vector of the target point cloud (BIM surface
plane) associated with its point pt , and ps represents the closest point of the source
point cloud (world) while n represents the number of points.
Figure 10 highlights the results of the fine registration process, while distin-
guishing a different number of rooms on the same floor. The rough registration
is based solely on the scan of one room. However, if more than one room is scanned,
the fine registration can be performed on the respective point clouds of more than
one room, e.g. one room (Fig. 10a), two neighboring rooms (Fig. 10b) or even three
neighboring rooms (Fig. 10c). It can be observed that the overall co-registration error
does not improve with an increasing number of rooms or point clouds, respectively,
used within the ICP algorithm.

Fig. 9 Results of the manual, initial and rough registration process

Concept and Implementation of BIM-to-World Co-registration … 279

Fig. 10 Results of the fine registration process, using (a) One room. Point-to-plane RMSE =
0.114 m. (b) Two rooms. Point-to-plane RMSE = 0.109 m. (c) Three rooms. Point-to-plane RMSE
= 0.112 m

7 Conclusions

This paper presents a concept and an implementation of a semi-automated BIM-

to-World co-registration process for mixed reality applications using the Microsoft
HoloLens as the MR device. This process is divided into a client and a server process
in order to balance the computational load. While the rough, initial and manual
registration is performed on the MR device, a dedicated Python server application
implements the ICP algorithm that takes the initial transformation of the scene as
input. A specialized case study demonstrates that the proposed method provides
feasible and practical results.

References

1. Borrmann A, König M, Koch C, Beetz J (2018) Building information modeling: why? What?
How? In: Borrmann A, König M, Koch C, Beetz J (eds) Building Information Modeling.
Springer, Cham
2. MacLeamy P (2004) Collaboration, integrated information, and the project lifecycle in building
design, construction and operation. In: WP-1202, The Construction Users Roundtable
3. Nassereddine H, Hanna A, Veeramani D, Lotfallah W (2022) Augmented reality in the construc-
tion industry: use-cases, benefits, obstacles, and future trends. Front Built Environ 8. https://
doi.org/10.3389/fbuil.2022.730094
4. Zhu H, Guo B, Zou K, Li Y, Yuen KV, Mihaylova L, Leung H (2019) A review of point set
registration: from pairwise registration to groupwise registration. Sensors 19:1191. https://doi.
org/10.3390/s19051191
5. Tam GK, Cheng ZQ, Lai YK, Langbein FC, Liu Y, Marshall D, Martin RR, Sun XF, Rosin
PL (2013) Registration of 3D point clouds and meshes: a survey from rigid to nonrigid. IEEE
Trans Vis Comput Graph 19(7):1199–1217. https://doi.org/10.1109/TVCG.2012.310
6. Mahmood B, Han S, Lee DE (2020) BIM-based registration and localization of 3D point clouds
of indoor scenes using geometric features for augmented reality. Remote Sens 12(14). https://
doi.org/10.3390/rs12142302
7. Persad RA, Armenakis C (2017) Automatic co-registration of 3D multi-sensor point clouds.
ISPRS J Photogramm Remote Sens 130:162–186. https://doi.org/10.1016/j.isprsjprs.2017.
05.014
280 M. Khairtdinov et al.

8. Rusu RB, Blodow N, Beetz M (2009) Fast point feature histograms (FPFH) for 3D registration.
In: 2009 IEEE International Conference on Robotics and Automation, pp 3212–3217. https://
doi.org/10.1109/ROBOT.2009.5152473
9. Tombari F, Salti S, Di Stefano L (September 2010) Unique signatures of histograms for local
surface description, vol 6313, pp 356–369. https://doi.org/10.1007/978-3-642-15558-126
10. Heinrich F, Schmidt G, Hansen C (September 2019) A novel registration method for optical
see-through augmented realty devices and optical tracking data
11. Besl P, McKay H (1992) A method for registration of 3-D shapes. IEEE Trans Pattern Anal
Mach Intell Pattern Anal Mach Intell 14:239–256. https://doi.org/10.1109/34.121791
12. Mejia-Parra D, Lalinde-Pulido J, Sánchez JR, Ruiz-Salguero O, Posada J (2019) Perfect spatial
hashing for point-cloud-to-mesh registration. In: Casas D, Jarabo A (eds) Spanish Computer
Graphics Conference (CEIG). The Eurographics Association. https://doi.org/10.2312/ceig.201
91202
13. Aoki Y, Goforth H, Srivatsan RA, Lucey S (2019) PointNetLK: robust & efficient point cloud
registration using PointNet. CoRR abs/1903.05711. http://arxiv.org/abs/1903.05711
14. Chen Y, Medioni G (1991) Object modeling by registration of multiple range images. In:
Proceedings of the 1991 IEEE International Conference on Robotics and Automation, vol 3,
pp 2724–2729. https://doi.org/10.1109/ROBOT.1991.132043
Safety and Training Implications
of Human-Drone Interaction
in Industrialised Construction Sites

Adetayo Onososen , Innocent Musonda, Molusiwa Ramabodu,

and Christopher Dzuwa

Abstract Over the past decade, the development of construction robots, such as
drones, has increased, given the need to apply emerging technologies and innovations
in enhancing occupational health and safety on construction sites. The usage of these
innovations is not without safety concerns for human-robot interaction. Previous
studies have highlighted the importance of training in enabling a safe collaboration
between human workers and robots in the construction environment. However, the
cost of acquiring these novel innovations makes training cost expensive. This study,
therefore, examines the safety and training implications of human-drone interactions
based on the simulation of drones for material handling in a virtual environment.
The findings of this study would help construction stakeholders and designers iden-
tify critical factors towards enabling an effective virtual reality training approach to
building safe adoption of robots and ensuring workers’ health is assured.

Keywords Human-drone interaction · Human-robot collaboration · Human-robot

interaction · Construction safety · Collaborative robots · Training

1 Introduction

Drones, also known as Unmanned Aerial Vehicles (UAVs), have increased in usage
and adoption in the Built environment and on construction sites, given their merits
in progress monitoring, health and safety monitoring, security, progress tracking,
surveillance, and inspection, aerial data collection etc. [1–4] These efforts at digital-
ising the sector through the adoption of drones, however, is not without its demerits

A. Onososen (B) · I. Musonda · C. Dzuwa

Centre for Applied Research and Innovation in the Built Environment (CARINBE), Faculty of
Engineering the Built Environment, University of Johannesburg, Johannesburg, South Africa
e-mail: adetayoo@uj.ac.za
M. Ramabodu
Department of Construction Management and Quantity Surveying, Faculty of Engineering and the
Built Environment, University of Johannesburg, Johannesburg, South Africa

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 281
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_20
282 A. Onososen et al.

[5]. Among these are safety issues in human-drone interaction, ergonomics, workers’
privacy and the cost of training to facilitate adoption [6, 7]. Previous studies have
mentioned that while drones are important and would be more adopted, they nonethe-
less bring in a wider spectrum of unwarranted risks and hazards [8–10]. While training
to ensure safe adoption has been advocated by previous studies [11, 12], the cost of
training arising from crashes of drones limits investment and discourages adoption
[13, 14].
Therefore, recent advances have seen virtual reality as a more effective platform
for human-drone training. Virtual Reality (VR) mimics real-life scenarios and helps
train users on drone operations, and it has been identified as a highly effective training
approach [11, 12]. Experts have employed VR in the built environment to help
visualise city operations, buildings, and design since roughly the 1990s [15, 16].
Users may evaluate prototypes realistically, beginning with the earliest design phases,
thanks to the concept of virtual prototyping or virtual design review. Several firms
perform design reviews to find errors or areas of improvement in their innovations
before the actual product is built [17]. Studies have also mentioned that Virtual
reality (VR) has demonstrated its capacity to improve communication, early project
planning and coordination in several sectors [18, 19]. It creates a unique environment
where stakeholders can intuitively visualise and interact with 3D representations of
a product before it is manufactured [20].
While VR has these merits, its use in construction is limited. For effective human-
drone interaction in industrialised construction sites where all workers will not have
a similar degree of understanding and involvement with drones on the construction
site, their training implementation is inevitable [21]. Adoption of drones in itself
can be a safety risk as previous studies mentioned that while a drone user with a
background in civil engineering can understand certain drone actions fitted for their
expertise, an artisan might otherwise be concerned about the advent of drones and
could get distracted and interpret it as spying tools [21]. This unfamiliarity with the
operation of drones on-site causes concern with non-users and close communities.
While drones are operated around them, this may bring discomfort and anxiety to
their working environment [22, 23].
While drone collision deaths and injuries are rare, safety risks will always be a
concern. Also important is a collision with wildlife, properties, distraction, violation
of privacy, or trespassing. All of these speak to the need for a VR-Based drone training
approach to ensure adoption safety and lower training costs. Therefore, studies on
appropriate safety and privacy management to accompany drone integration on-site
are important to facilitate adoption and improve training efforts.
A human-computer interaction in which the computer generates a sensory expe-
rience is known as virtual reality, a setting that reacts to and is interactively modified
by user behaviour [24, 25]. Experiences in virtual reality settings come from three
different places: immersion, interaction, and multi-sensory feedback. With immer-
sion, the user is completely engulfed in the setting, giving them the impression of
being real in the world portrayed. At the same time, interactivity can be seen as the
Safety and Training Implications of Human-Drone Interaction ... 283

ability to control events in the simulation environment by using human body move-
ments, which initiate responses in the simulation environment as a result of these
movements [24].
The safety implications of drones in construction have recently received much
attention from researchers and practitioners. The ability of drones to be airborne and
be controlled remotely or autonomously has only increased the safety implications of
utilising such technology in a construction environment. Therefore, while adoption
benefits are immense, the safety implications limit the construction environment
sector adoption. Therefore, the challenge remains aiding drone users with knowledge
and expertise on underlying safety risks associated with flying drones and eliminating
these risks through repetitive training in a virtual environment [26, 27]. The presence
of onboard mounted cameras continues to raise reasonable suspicions and privacy
concerns in most environments and therefore has implications on training to ensure
drone users keep to strict ethical conduct during drone flying. UAV technology is
fairly new in construction, and the full scope of the negative outcomes has yet to be
addressed and further investigated [8, 28, 29].
Without adequate training, workers may exhibit unsafe actions, leading to
dangerous outcomes (i.e., catastrophic accidents). As stated in previous studies,
unsafe work behaviour has been mentioned as one of the main influences of workplace
hazards in construction [8]. Privacy issues such as the invasion of drone hovering
workers can startle workers and lead to loss of focus which can cause trips, falls and
other potential accidents, hence the need for training on the safe usage of drones
on-site [22, 30].
Hence, this study examines the safety and training implications of human-drone
interactions based on the simulation of drones in material handling in a virtual envi-
ronment. The phases involved in the study are presented based on a pilot study’s
output. The study provides insights on users’ behaviour in the environment which
are critical to inform training and education of users in using VR, and also on how
people perform given tasks in the virtual environment.

2 Method

The research method involved designing and developing a virtual reality environment
with human participants interacting with drones in an immersive environment. The
first stage in the method approach consisted of 3D modelling of the assets necessary to
set up the scenario for the virtual environment. The assets modelled in 3D included the
construction site, construction vehicles and materials on construction sites, Drones,
and tasks to be carried out. In the second stage, the 3D models were transformed into
assets.
The drone simulator was implemented in Unity® 2022.1.16 using the C# program-
ming language. Following the import of the model into unity®, design parameters,
i.e., weight, gripping mechanism joint friction, rotation, retraction and expansion
speed, were assigned to the drone. The flight environment was a construction site
284 A. Onososen et al.

with four uncompleted buildings. Other objects commonly found on construction

sites were also added to the scene. The third stage involves integrating the different
assets to create the whole environment and integrate them. Furthermore, the lighting
system ensured a good degree of photorealism.
The fourth stage involved interaction programming using C++ to ensure the
Oclus® touch controllers and thumbsticks can help users navigate the environment,
perform tasks and interact with other humans in the environment. Task manage-
ment was designed to allow the participants to pick materials (bricks in this case)
between a lower point on site and deliver them at a much higher level (On the top
of a structure). After the design and development of the drone, a pilot study was
conducted with thirteen built environment professionals from diverse disciplines.
The users began by learning to control the drone, navigate it without weight, pick the
material and deliver it through vertical and horizontal flight patterns. To study what
users were doing in the environment and provide insight for further development and
improvement, an automatic data collection system was integrated into the application
to provide indices on the user’s performance based on different factors. Each interac-
tion in the environment is further supported by associating sounds with actions in the
environment to improve the overall realism; before the tests, participants were taken
through a training session that involved guiding users on the controls to navigate the
environment and manually fly the drone (Fig. 1).
For the experiment, the Oculus® Quest 2 was adopted. Participants were required
to sign ethical consent forms before participating in the experiment. Afterwards, the
workings of the Oculus® quest 2 were briefly explained to the participants. Partici-
pants were advised to adjust the HMD to their comfort. Participants calibrated their
eye gaze to ensure accurate eye-tracking data. The study began after a successful
eye calibration was achieved. Participants were immersed in the learning environ-
ment, where they completed tasks involving the delivery of construction materials

Fig. 1. 3D model-conceptual view

Safety and Training Implications of Human-Drone Interaction ... 285

(bricks in this case). The use of delivery drones as a means of transporting goods
is becoming increasingly common in today’s society. By incorporating the use of
delivery drones in the task stated above, we were able to simulate a practical and
meaningful real-world scenario. The task of material pickup and delivery using a
drone allowed us to assess participants’ ability to navigate the drone, manipulate it,
and deliver a package to a specified location. This task also enabled us to evaluate the
effectiveness and efficiency of the drone as a delivery vehicle. Furthermore, the use
of delivery drones in the task allowed us to investigate the potential benefits and limi-
tations of using such technology in the transportation industry. Overall, the inclusion
of delivery drones in our study provided a relevant and practical application for the
use of drones in the real world. This procedure was repeated in the same order for
all participants. The data for the study was collected by the virtual application based
on each participant’s behaviour in the environment. The parameters used to retrieve
the data were Payload, Collision object, Time, Drone speed, Drone Position, Drone
Orientation, Safe Distance, Pilot Position, Command, Drone Range and Battery life
(Figs. 2, 3 and 4).

Fig. 2 Flight environment with drone

286 A. Onososen et al.

Fig. 3 Participants in the

study

Fig. 4 Participants in the

study

3 Results and Discussion

3.1 Altitude Distribution and Velocity Distribution

The frequency of the altitude and velocity distribution of the experiments are shown
in Fig. 5 and Fig. 6. The frequency distribution along an altitudinal gradient revealed
that most of the participants were flying at low altitudes. This is not unusual for
first-time drone users who, at initial flight operations, focus on being able to control
the drone. However, the altitude of flight is critical to human-drone interaction. It
indicates users’ ability to maneouver around different building heights, fly at heights
that will not cause noise and distraction to site workers, and fly drones to deliver
materials at a given height. This is critical to inform drones of virtual training in
understanding the behaviour of participants in the environment and how this transfers
to developing skills in real-life scenarios.
Furthermore, the velocity distribution of the experiment also revealed very low
flight speeds of 1–2 m/s. While this indicates users’ understanding of the drone’s
control, the graph further shows participants improved on speed as they improved on
controls of the drone. Hence, for safety training in virtual drone environments, the
speed and altitude that participants can fly safely depend on the skill level associated
Safety and Training Implications of Human-Drone Interaction ... 287

Fig. 5 Altitude distribution of experiments performed by participants

Fig. 6 Velocity Distribution of experiments performed by participants

288 A. Onososen et al.

with drone control. While drones are expensive and prone to crashes when used
in training, a virtual environment can aid users in honing their skills to understand
drones’ directional behaviours before being placed with drones in real-life scenarios.
This helps the confidence level of trainees, improves skills level before physical
flights and reduces the cost of replacing drones.

3.2 Collisions Per Experiment

The collision per experiment for the study was generated from the application and
indicated the frequency of collisions experienced by users in the virtual environ-
ment. As shown in Fig. 7, the reveals a high number of collisions by participants
in the environment. Collisions are an important indicator to measure the risk of
losing drones and the safety of flying drones without risk to humans around the site.
While the results are not unusual given that the participants are new to flying, the
ability of the virtual environment to integrate object and human contact avoidance
in participants, especially in adverse weather conditions, is important for training.
This addresses salient issues raised on the safety of integrating drones with humans
present on site. Furthermore, a virtual platform allows users to repetitively conduct
training on collision avoidance before uptaking physical drone flights.

Fig. 7 Frequency of collisions in the Experiment

Safety and Training Implications of Human-Drone Interaction ... 289

3.3 Collisions Frequency Based on Time

The collision frequency based on time indicates the chances of colliding during the
experiment based on periods. Figure 8 revealed that time periods less than 200 m/s
had more collisions compared to other time periods. This indicates that the collision
occurred while the users were just starting. This is expected as users struggle to
control the drone when starting and could fly at high speeds, colliding with objects
while learning the controls. This is further justified by the results showing that the
collisions reduced as participants progressed in the study. This implies that more
training hours are needed for drone training. Given the ability of VR to aid repetitive
training without the risk of losing the drone, it is imperative for training guides to
ensure users learn drone operations control firstly in a VR environment before actual
deployment on-site.

3.4 Collisions Frequency Based on Velocity

The collision frequency based on velocity indicates how active the participants are
in the environment and could also indicate the level of idleness. Figures 9 reveal that
most collisions happened at about 4 m/s. At the initial stage, most of the collisions
happened at lower velocities. This could indicate that users collided with objects
while attempting to pick the materials.

3.5 Drones Command

To further confirm the users’ activities in the environment with the previously
discussed results, the command directed to the drone was analysed using word
cloud and revealed that at various instances in the virtual environment, users did
not send any command to the drone. This indicates idleness for most participants.
Discussions with participants after the experiments revealed some likely reasons,
such as users pausing to check out structures in the environment and having virtual
walkthroughs. To improve supervised learning and reduce idleness in the virtual
environment, further development of the virtual training platform will have multiple
users collaborating in the same environment.
290 A. Onososen et al.

Fig. 8 Collision Frequency based on time

3.6 Collision Frequency Based on Height

The collision frequency based on height revealed that most collisions happened under
7 m, as shown in Figs. 10. This indicates that the users often collided at this range
due to the challenge of picking the bricks to be delivered in the task. The challenge
of picking the objects is also attributed to understanding how to control the drone.
As previously stated, this could be improved through repetitive virtual environment
training.
Safety and Training Implications of Human-Drone Interaction ... 291

Fig. 9 Collisions Frequency Based on Velocity

Fig. 10 Drone command word cloud

292 A. Onososen et al.

Fig. 11 Collision frequency based on height

3.7 Percentage Out of Boundary

Privacy issues are critical barriers to drone adoption, as drone operations can capture
areas beyond the target point. This has led to various litigations, which have impacted
the interest and adoption level of drones in construction. To train participants on
ensuring drones stay within controlled bounds and to ensure users do not spy on
neighbouring structures, the study collected data on the boundaries incursion tenden-
cies of the users and discovered that for most parts of the experiment, people were
within the boundaries with only one user straying away as shown in Fig. 11. Pilots are
liable to pay for damages and their flight license suspended over privacy issues, so
this training is critical to eliminating privacy issues associated with drone operations
on site (Fig. 12).
Safety and Training Implications of Human-Drone Interaction ... 293

Fig. 12 Percentage out of boundary

4 Conclusion

The study designed and developed a virtual reality training platform to examine
drones’ safety and training implications for material handling on construction sites.
While the study revealed the high number of collisions, this was explained as usual
with initial training experiences with physical drones and demonstrated the impor-
tance of virtual environments as a testing platform before the physical deployment
of robots with humans. This critically reveals that virtual environments offer users
the ability to understand skills level, identify likely safety and performance errors
able to be committed in real-life scenarios, utilise virtual reality to eliminate those
risks and improve performance before physical interaction with robots on site. For
construction organisations and training institutions, this helps appreciate the reduced
initial capital investment achieved when robot such as drones are adopted with virtual
training platforms. The study also revealed the issues associated with the idleness of
participants in the virtual environment and how virtual reality can help correct unsafe
behaviour and privacy concerns associated with drones application on construction
sites. Further design and development options are raised and discussed in the study.
The use of delivery drones is identified as a limitation of the study and future studies
will incorporate other drones use cases.

Acknowledgements The drone was modelled by Mr. Samuel Moyo Ndovie, a University College
of Engineering, Osmania University student. This research is funded by the National Research
Foundation, NRF (Grant Number- 129953). Opinions and conclusions are those of the authors and
294 A. Onososen et al.

are not necessarily attributable to the NRF. The work is supported and part of collaborative research
at the Centre of Applied Research and Innovation in the Built Environment (CARINBE).

References

1. Gheisari M, Esmaeili B (2016) Unmanned aerial systems (UAS) for construction safety
applications masoud. In: Construction research congress 2016, pp 2642–2650
2. Golizadeh H, Hosseini MR, Martek I, Edwards D, Gheisari M, Banihashemi S, Zhang J
(2019) Scientometric analysis of research on “remotely piloted aircraft”: a research agenda
for the construction industry. Eng Constr Archit Manag 27(3):634–657. https://doi.org/10.
1108/ECAM-02-2019-0103
3. Kas KA, Johnson GK (2020) Using unmanned aerial vehicles and robotics in hazardous
locations safely. Process Saf Prog 39(1):1–13. https://doi.org/10.1002/prs.12066
4. Mohsan SAH, Khan MA, Noor F, Ullah I, Alsharif MH (2022) Towards the unmanned aerial
vehicles ( UAVs ): a comprehensive review. Drones 6(147):1–27
5. Lee YY, Falahat M, Sia BK (2021) Drivers of digital adoption: a multiple case analysis among
low and high-tech industries in Malaysia. Asia Pacific J Bus Adm 13(1):80–97. https://doi.org/
10.1108/APJBA-05-2019-0093
6. Chigbu BI, Nekhwevha FH (2020) The collaborative work experience of robotics and human
workers in the automobile industry in South Africa. African J Sci Technol Innov Dev. https://
doi.org/10.1080/20421338.2020.1837446
7. Behzadi A (2016) Using augmented and virtual reality technology in the construction. Am J
Eng Res 5(12):350–353
8. Khalid M, Namian M, Behm M (2021) Safety implications of using UAVs in construction: an
ethical perspective. In: Proceedings of the joint CIB W099 & W123 international conference
2021
9. Wang R, Nakhimovich D, Roberts FS, Bekris KE (2021) Chapter 5 robotics as an enabler of
resiliency to disasters: promises and pitfalls. In: LNCS, vol 12660. Springer, Cham, pp 75–101
10. Onososen A, Musonda I (2021) Research trends of human-robot teams/robotics in construc-
tion: a scientometric analysis. In Musonda I (ed): Building smart, resilient and sustainable
infrastructure in developing countries, DII-2022, Johannesburg, pp 398–412
11. Hafsia M, Monacelli E, Martin H (2018) Virtual reality simulator for construction workers. In:
LAVAL, France, no VRIC 2018, 4–6 April 2018. Association for Computing Machinery
12. Froehlich MA, Azhar S (2016) Investigating virtual reality headset applications in construction.
In: 52nd ASC annual international conference proceedings, pp 1–8
13. Sakib N, Asce SM, Chaspari T, Behzadan AH, Asce M (2021) Physiological data models to
understand the effectiveness of drone operation training in immersive virtual reality. J Comput
Civ Eng 35(1):1–13. https://doi.org/10.1061/(ASCE)CP.1943-5487.0000941
14. Sakib N, Asce SM, Chaspari T, Behzadan AH (2021) Physiological data models to understand
the effectiveness of drone operation training in immersive virtual reality. J Comput Civ Eng
35(1):1–13. https://doi.org/10.1061/(ASCE)CP.1943-5487.0000941
15. Manuel J, Delgado D, Oyedele L, Demian P, Beach T (2020) A research agenda for augmented
and virtual reality in architecture, engineering and construction. Adv Eng Inform 45:101122.
https://doi.org/10.1016/j.aei.2020.101122
16. Wolfartsberger J, Zenisek J, Wild N (2019) Supporting teamwork in industrial virtual reality
applications. Procedia Manuf 2020(42):2–7. https://doi.org/10.1016/j.promfg.2020.02.016
17. Hasan SM, Lee K, Moon D, Kwon S, Jinwoo S, Lee S (2021) Augmented reality and digital
twin system for interaction with construction machinery. J Asian Archit Build Eng 00(00):1–12.
https://doi.org/10.1080/13467581.2020.1869557
Safety and Training Implications of Human-Drone Interaction ... 295

18. Wen J, Gheisari M (2020) Using virtual reality to facilitate communication in the AEC domain:
a systematic review. Constr Innov, 509–542. https://doi.org/10.1108/CI-11-2019-0122
19. Rahimian FP, Ibrahim R, Rahmat RWBOK, Abdullah MTB, Jaafar MSBH (2011) Mediating
cognitive transformation with VR 3D sketching during conceptual architectural design process.
Archnet-IJAR 5(1):99–113
20. Ibrahim R, Pour Rahimian F (2010) Comparison of CAD and manual sketching tools for
teaching architectural design. Autom Constr 19(8):978–987. https://doi.org/10.1016/j.autcon.
2010.09.003
21. Albeaino G, Eiris R, Gheisari M, Issa RRA (2021) Development of a VR-based drone-mediated
building inspection training environment. Comput Civ Eng, 1401–1408
22. Namian M, Khalid M, Wang G, Turkan Y (2021) Revealing safety risks of unmanned aerial
vehicles in construction. J Transp Res Board
23. Masullo M, Pascale A, Adina R, Ruggiero G, Maffei L (2019) Virtual reality overhead
crane simulator. Procedia Comput Sci 2022(200):205–215. https://doi.org/10.1016/j.procs.
2022.01.219
24. Abdelhameed WA (2013) Virtual reality use in architectural design studios : a case of studying
structure and construction. Procedia Procedia Comput Sci 25:220–230. https://doi.org/10.1016/
j.procs.2013.11.027
25. Dunston PS, Wang X (2005) Mixed reality-based visualisation interfaces for architecture,
engineering, and construction industry. J Constr Eng Manag 131(12):1301–1309. https://doi.
org/10.1061/(asce)0733-9364(2005)131:12(1301)
26. Srivastava S, Gupta S, Dikshit O, Nair SA (2019) Review of UAV regulations and policies in
India. Springer, Cham. https://doi.org/10.1007/978-3-030-37393-1
27. Eiris R, Benda B, Faris R (2020) Indrone: visualizing drone flight patterns for indoor building
inspection tasks. In: Enabling the Development and Implementation of Digital Twins - Proceed-
ings of the 20th International Conference on Construction Applications of Virtual Reality, 30
September–2 October 2020, no 3, pp 281–290
28. Doyle-Burke D, Haring KS (2020) Robots are moral actors: unpacking current moral HRI
research through a moral foundations lens. In: 12th international conference, ICSR 2020,
Golden, CO, USA, November 14–18, 2020, Proceedings, vol 12483. LNAI. Springer, Cham,
pp 170–181. https://doi.org/10.1007/978-3-030-62056-1_15
29. Onososen A, Musonda I (2022) Perceived benefits of automation and artificial intelligence in
the AEC sector: an interpretive structural modeling approach. Front Built Environ, 61. https://
doi.org/10.3389/fbuil.2022.864814
30. Moud HI, Flood I, Shojaei A, Zhang Y, Zhang X, Tadayon M, Hatami M (2019) Qualitative
assessment of indirect risks associated with unmanned aerial vehicle flights over construction
job sites, 83–89. https://doi.org/10.1061/9780784482445.011
Task Performance to Understand
the Effectiveness of Visualisation
Technology-Based Training
for Human-Drone Interaction Learning

Adetayo Onososen , Innocent Musonda, Molusiwa Ramabodu,

and Christopher Dzuwa

Abstract Drones’ adoption in the Architecture, Engineering and Construction

(AEC) sector is increasing with the need for digitalizing construction processes
and workflows. However, the investment cost does not match the cost of replacing
drones from crashes and accompanying safety and privacy challenges resulting in
litigation. Therefore, this study examines visualization technology-based training to
study human-drone interaction for learning. To achieve the objectives, drones were
modelled in 3D alongside a virtual construction site for participants in the experi-
ment to perform tasks. The attached tasks gave indicators of the effectiveness of such
learning platforms in drone training. The results noted that while virtual reality is
excellent as a form of training, the low number of delivered items indicates the need
for more training time. The findings of the study are useful for those in academia,
policy-making and industry buy-in.

Keywords Human-drone interaction · Human-robot collaboration ·

Visualisation · Virtual reality · Construction safety · Task performance · Learning

1 Background to Study

Unmanned Aerial Vehicles (UAVs) or drone applications have the potential to provide
value-driven solutions for a wide range of applications that need a variety of capa-
bilities [1–3]. Recent technological developments in unmanned aerial vehicles have
significantly aided in improving the diverse application of drones in construction

A. Onososen (B) · I. Musonda · C. Dzuwa

Centre for Applied Research and Innovation in the Built Environment (CARINBE), Faculty of
Engineering the Built Environment, University of Johannesburg, Johannesburg, South Africa
e-mail: adetayoo@uj.ac.za
M. Ramabodu
Department of Construction Management and Quantity Surveying, Faculty of Engineering and the
Built Environment, University of Johannesburg, Johannesburg, South Africa

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 297
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_21
298 A. Onososen et al.

[5]. Drones in construction are increasingly being adapted for site security, thermal
imaging, progress monitoring, building inspection, building surveys, equipment
tracking, photography and material handling, amongst others [4, 7]. Firms are begin-
ning to maximise opportunities associated with digitalising processes and workflow
in the sector through drone adoption. Still, they are constrained by a lack of training
resources given the high initial investment required to acquire drones [6, 8, 9].
To solve this problem, recent innovations are pushing the boundaries of visualiza-
tion to provide training for users of digital technologies in an immersive environment
mimicking real-life scenarios and equipment [10]. The term “virtual reality” (VR)
describes the artificially produced three dimensions using a computer system to
create an environment with the use of sensors [11]. This use case extends to areas
such as; VR in various safety contexts, including identifying risks, educating the
workers, transferring skills, and ergonomics. With the introduction of VR, the use of
sophisticated VR tools to produce a visualization of difficult working circumstances
and risk mitigation Knowledge and education have increased [12–14]. Rokooei et al.
[12] highlight that the use of VR in training is gaining momentum with examples
including virtual reality (VR) as a teaching tool for roofers since it poses no danger
and can replicate high-altitude scenarios real hazards of falling.
This advantage has placed VR at the centre of the emerging nature of training for
workers in the built environment, especially for high-risk tasks. With the increasing
adoption of technologies, virtual reality training absorbs the risk of losing or
damaging digital tools, which comes with high investment costs. Also important,
virtual training provides users accessibility, availability and affordability to repet-
itive learning with no danger and the ability to self-learn. Therefore, while large
construction firms can invest in these digital innovations, virtual training is still
required to ensure investment costs are maximized, and loss is eliminated. Platforms
for training, monitoring, and control are provided by VR technology, enhancing the
safety of building projects and fewer accidents at building sites [15, 16].
The technologies of augmented reality (AR) and virtual reality (VR) in the Archi-
tecture, Engineering, and Construction (AEC) industries are important since the built
environment is fundamentally reliant on information related to three-dimensional
(3D) space and a strong dependence on images for communication and progress
monitoring [17]. Constructivist learning is supported by VR’s immersive character
[18]. As mentioned by Abdelhameed [18], the theory of constructivism is a learning
approach whereby people create knowledge by taking lessons from their experiences.
Human learners get knowledge from the world via experiences and the use of their
senses. VR uses computer-generated sensory input in place of real-world sensory
input simulation to facilitate learning by giving users a setting where they may
explore situations rather than just envisioning things. Therefore, this study inves-
tigated the use of visualization-based technology with a specific focus on virtual
reality training to examine effectiveness of task performance in human-drone inter-
action learning to improve safe adoption of drones in industrialised construction
sites. The overarching research question in this study was to explore how effective
VR is in training users on drone flight operations with an additional focus on aerial
Task Performance to Understand the Effectiveness of Visualisation Technology ... 299

material transport on construction sites. Results based on the data generated by the
virtual training platform is presented.

2 Research Method

The main objective of this study was to determine the efficacy of virtual reality (VR)
training for drone flying operations measured based on task performance in the virtual
environment. The research technique entailed creating a virtual reality environment
where people may interact with drones in a realistic setting. The elements required
to set up the scenario for the virtual environments were modelled in 3D as the first
step in the technique approach. The construction site, construction trucks, supplies
on building sites, drones, and jobs to be completed were among the 3D-modelled
assets. The 3D models were turned into assets in the second stage. The drone simu-
lator was developed using C# as a programming language with Unity 2022.1.16. The
design characteristics, such as weight, grasping joint mechanism friction, rotation,
retraction, and expansion speed, were assigned to the drone. The flying environment
was four unfinished structures on a construction site. The training scene was supple-
mented with other items frequently seen on building sites. In the third stage, the
various assets are combined to form the overall environment and are integrated.
Furthermore, a good level of photorealism was produced using the lighting system.
For the Oculus touch controllers and thumbsticks to assist users in navigating the
scene, carrying out activities, and interacting with other people in the environment,
this level of interaction was achieved using C++ . The purpose of task management
was to enable participants to choose materials (in this example, bricks) between a
lower point on the job site and deliver at a much higher level (On the top of a structure).
A pilot study with experts in the built environment was carried out after the design
and development of the drone. The users first learned how to fly the drone vertically
and horizontally, control it without any weight to pick up the material, and deliver it.
An automatic data gathering system was included in the program to offer indices on
the user’s performance depending on various parameters. This allowed researchers
to investigate what users were doing in the environment and provide information
for future development and improvement. Associating sounds with environmental
actions further supports each interaction to increase the level of realism. Before the
testing, participants had a training session that involved instructing users on how to
operate the controls to fly the drone and navigate through the area manually.
The Oculus Quest 2 was chosen for the test. Before participating in the trial, partic-
ipants had to sign paperwork requesting their ethical consent. After that, the partici-
pants quickly explained how Oculus Quest 2 functions. Participants were advised to
adjust the HMD to their comfort. To ensure precise eye-tracking data, participants
calibrated their eye gaze. After a successful ocular calibration, the trial got underway.
Participants were immersed in the environment with activities requiring the delivery
of building materials while fully immersed in the learning environment (bricks in
this case). For everyone, this process was performed in the same sequence. Based
300 A. Onososen et al.

on each participant’s behaviour in the setting, the virtual program collected the data
for the research. The parameters used to retrieve the data were Payload, Collision
object, Time, Drone speed, Drone Position, Drone Orientation, Safe Distance, Pilot
Position, Command, Drone Range and Battery life.
Tasks given to users in the VR environment involved (1) learning the controls
to understand how to fly the virtual drones, (2) flying the drone in the environment
within the boundaries of the site to ensure privacy, (3) safe flying of the drone to avoid
collisions with objects and humans (4) using the drone to transport materials vertically
and horizontally without dropping the materials (5) learning safe navigation of the
drone with materials handling at high altitude (6) controlling the drone in adverse
weather conditions to simulate real-life scenarios (7) maintaining contact with the
drone by ensuring it is not lost (8) safe and effective delivery within a given period
(9) beyond visual line of sight learning to adopt skills required for the drone in out
of sight or night operations. To validate the virtual environment developed, it was
tested by industry experts such as drone pilots in practice and drone design experts
to ensure the applicability and correctness of embedded components and required
actions.

3 Results and Discussion

3.1 Number of Task Deliveries in the Experiment

As shown in Fig. 6 the frequency of deliveries in the environment by participants

using drones for material handling was low. In testing the task performance of the
experiments, most could not successfully deliver 3 tasks within the specified 10-
min duration. While this is surprising, it is not unusual for first-time drone users
who are inclined to initially understand the drone controls and virtual settings before
performing the required tasks. However, the successful deliveries revealed that virtual
environments could efficiently train users in learning operations of drones for material
handling. The low number of deliveries also indicates the level of rigour ensured in
demonstrating the drone mimics real-life scenarios. This was further supported by
a post-experiments discussion with licensed drone pilots who participated in the
experiment (Figs. 1, 2, 3, 4 and 5).
Task Performance to Understand the Effectiveness of Visualisation Technology ... 301

Fig. 1. 3D
model-conceptual view

Fig. 2 Image showing VR

environment

Fig. 3 Image showing

materials in VR

3.2 Delivery Distribution Based on Time-Taken

The delivery distribution based on time-taken, as shown in Fig. 7, revealed that the
tasks performed took about 40–60 s for most of the tasks performed successfully.
Given the high number of materials transported daily on the construction site and the
battery span for most commercial drones in operations, this was considered critical
to be measured. It is important as it can justify the cost-benefit use of adoption drones
for material handling on-site against current traditional approaches. With the current
drone battery span of an average of 20 min, the number of tasks to be performed
302 A. Onososen et al.

Fig. 4 Participants

Fig. 5 Participants

before recharging is vital to convincing stakeholders of the cost-benefit of adoption

and returns on investment that integrating drones in daily construction operations
bring. However, collisions were also measured to ensure, in an attempt to maximise
the use of drones in delivering tasks, that this is done in a safe and healthy approach
with no risk of hazards to humans, animals, aircraft and other objects.

3.3 Mean Drone Velocity Distribution

Mean drone velocity distribution, as highlighted in Fig. 8, is vital to measure the

successful flight behaviour of users in the environment and indicate what velocity
levels are adequate for successful and safe flight operations. While there are standard
speed levels for drone flights, with drones handling materials on construction sites,
a safe velocity level at which the materials can be well handled and tasks delivered
Task Performance to Understand the Effectiveness of Visualisation Technology ... 303

Fig. 6 Number of tasks delivered successfully in the experiment

Fig. 7 Frequency of delivery distribution based on time taken

304 A. Onososen et al.

Fig. 8 Mean drone velocity distribution.

successfully is vital to training insights in directing users and ensuring the safety of
human-drone operations on site. This is important for drones carrying payloads for
non-military uses in a construction environment, given the presence of humans and
properties at close proximities.
As shown in Fig. 9, the low-velocity distribution at which unsuccessful experi-
ments flew reveals that velocity was not a major criterion in the ability to deliver
tasks but depends on the users’ skill to control the drone. Hence, this reveals that
virtual environments are important in drone flight operations to increase skills before
interacting with drones in a real-life setting.

3.4 Mean Drone Altitude Distribution

Figure 10 reveal the frequency of the mean drone altitude distribution of unsuccessful
experiments, while Fig. 11 reveal the mean drone altitude distribution of successful
experiments. They revealed that altitude is not a significant factor in the ability to
deliver tasks in the environment. They also show that depending on the altitude
level, tasks are delivered successfully or unsuccessfully based on the participants’
skills level in using the drone to execute the task given. As identified previously, this
reveals visualisation’s effectiveness in building skills before physical human-drone
interaction on site.
Task Performance to Understand the Effectiveness of Visualisation Technology ... 305

Fig. 9 Frequency showing mean drone velocity distribution of unsuccessful experiments

Fig. 10 Frequency showing mean drone altitude distribution of unsuccessful Experiments

306 A. Onososen et al.

Fig. 11 Frequency showing mean drone altitude distribution of successful Experiments

3.5 Delivery Altitude Distribution

Figure 12 reveals that most of the delivery made were done at 35.5 to 40 m height.
At the same time, some deliveries were also made at heights lower and higher than
this. This further supports the previous discussion stating that pilots can deliver tasks
at specified heights when the drone operator’s skill level is ensured. Further testing
on this will examine the difference in delivery altitude of users using a visual line of
sight in drone operations and those using beyond visual line of sight (BVLOS). This
is further important to improve the usage of drones to deliver tasks and materials at
any altitude. The success of this would be critical for deploying drones in hazardous
construction sites beyond physical reach [19, 20].

4 Delivery Velocity Distribution

Figure 13 revealed that the delivery velocity for the experiments was under 2.5 m/s.
This is not surprising, given that users can fly at much higher speeds until they pick
up objects or materials with the drone. Therefore, the materials reduce the velocity
of the drone. This is critical to two factors. Firstly, it demonstrates the rigour of the
virtual training platform to mimic real-life scenarios, which is important to justify
the ability of virtual environments to precede physical training exercises. Secondly,
it informs drone operators and first-time users on what to expect concerning velocity
Task Performance to Understand the Effectiveness of Visualisation Technology ... 307

Fig. 12 Frequency showing delivery altitude distribution

in using drones with payload and those without payload. This is important, especially
for BVLOS operations and flying drones in adverse weather conditions. Users can
adjust flight plans based on understanding the difference in velocity between a loaded
drone and a non-loaded drone. This knowledge is essential to not losing drones and
payloads during adverse flight situations and to ensuring safe and successful flight
task performance [20–22].

5 Typical Flight Paths (Altitude)

Graphical representations of typical flight paths (altitude) for successful and unsuc-
cessful experiments are revealed in Fig. 14. The results revealed that successful users
had a progressive flight pattern by first learning the drone control at lower altitudes
and progressively flying to higher altitudes. However, this is different from unsuc-
cessful flights, which revealed users were starting out at high altitudes and coming
down to low altitudes. The implication of this shows that when users fly at high
altitudes at the initial commencement of the drones, they find it difficult to under-
stand the drone’s controls, affecting the experiments’ success. This seriously impacts
flying safety and collision with aircraft, animals and properties. This provides critical
insight for training guides in directing users’ learning [23, 24].
308 A. Onososen et al.

Fig. 13 Frequency showing delivery velocity distribution

Fig. 14 Graphical representation of typical flights paths (altitude) for Successful experiments
Task Performance to Understand the Effectiveness of Visualisation Technology ... 309

6 Conclusion

While drones have been considered for material handling in other sectors such
as logistics, Health, manufacturing etc., they are underexamined in construction
environments. Using drones to handle construction materials is an emerging drone
research area. They are important for risky situations such as the delivery of robotic
systems/equipment in hazardous environments, which makes using drones in mate-
rial handling on construction sites inevitable. However, they are also prone to health
risks from dropping objects, colliding with properties and humans, and mishandling
materials transported. Therefore, this study examined virtual reality’s use for safe
drone flight operations through task performance in an immersive environment.
With time, due to the latest technological innovation, drones are becoming cheaper
and more accessible, but not without accompanying safety and privacy concentra-
tions. This work demonstrates the effectiveness of using VR as a training platform
for drone usage on construction sites. The study revealed that VR is an effective plat-
form in task performance training involving drones such as robots. Also, the study
identified safe flight behaviours and patterns based on altitude and velocity.

Acknowledgements The drone was modelled by Mr. Samuel Moyo Ndovie, a student at the Univer-
sity College of Engineering, Osmania University. This research is funded by the National Research
Foundation, NRF (Grant Number-129953). Opinions and conclusions are those of the authors and
are not necessarily attributable to the NRF. The work is supported and part of collaborative research
at the Centre of Applied Research and Innovation in the Built Environment (CARINBE).

References

1. Liu D, Chen J, Hu D, Zhang Z (2019) Dynamic BIM-augmented UAV safety inspection for
water diversion project. Comput Ind 108:163–177. https://doi.org/10.1016/j.compind.2019.
03.004
2. Martinez JG, Gheisari M, Alarcón LF (2020) UAV integration in current construction safety
planning and monitoring processes: case study of a high-rise building construction project in
chile. J Manag Eng 36(3):5020005. https://doi.org/10.1061/(asce)me.1943-5479.0000761
3. Khalid M, Namian M, Behm M (2021) Safety implications of using UAVs in construction: an
ethical perspective. In: Proceedings of the joint CIB W099 & W123 international conference
2021
4. Munawar HS, Ullah F, Heravi A, Thaheem MJ, Maqsoom A (2022) Inspecting buildings using
drones and computer vision: a machine learning approach to detect cracks and damages. Drones
6(5)
5. Mendes E, Albeaino G, Brophy P, Gheisari M, Jeelani I (2022) Working safely with drones: a
virtual training strategy for workers on heights. In: Construction research congress 2022, vol
3–C, pp 964–973. https://doi.org/10.1061/9780784483978.098
6. Eiris R, Benda B, Faris R. (2020) Indrone: visualizing drone flight patterns for indoor building
inspection tasks. In: Enabling the development and implementation of digital twins - proceed-
ings of the 20th international conference on construction applications of virtual reality, 30
September–2 October 2020, no 3, pp 281–290
7. Onososen AO, Musonda I, Ramabodu M (2022) Construction robotics and human – robot
teams research methods. Buildings 12(1192):1–33
310 A. Onososen et al.

8. Khan NA, Jhanjhi NZ, Brohi SN (2020) Emerging use of UAV’S: secure communication
protocol issues and challenges. Elsevier Inc. https://doi.org/10.1016/B978-0-12-819972-5/000
03-3
9. Mohsan SAH, Khan MA, Noor F, Ullah I, Alsharif MH (2022) Towards the unmanned aerial
vehicles (UAVs): a comprehensive review. Drones 6(147):1–27
10. Badamasi AA, Aryal KR, Makarfi UU, Dodo M (2021) Drivers and barriers of virtual reality
adoption in UK AEC industry. Eng Constr Archit Manag https://doi.org/10.1108/ECAM-09-
2020-0685. No 2016
11. Huang Y, Shakya S, Odeleye T (2019) Comparing the functionality between virtual reality and
mixed reality for architecture and construction uses. J Civ Eng Archit 13:409–414. https://doi.
org/10.17265/1934-7359/2019.07.001
12. Rokooei S, Shojaei A, Alvanchi A, Azad R, Didehvar N (2022) Virtual reality application for
construction safety training. Saf Sci 2023(157):105925. https://doi.org/10.1016/j.ssci.2022.
105925
13. Froehlich MA, Azhar S (2016) Investigating virtual reality headset applications in construction.
In: 52nd ASC annual international conference proceedings, pp 1–8
14. Zhang Z, Pan W (2021) Virtual reality supported interactive tower crane layout planning for
high-rise modular integrated construction. Autom Constr 130:103854. https://doi.org/10.1016/
j.autcon.2021.103854
15. Masullo M, Pascale A, Adina R, Ruggiero G, Maffei L (2019) Virtual reality overhead crane
simulator. Procedia Comput. Sci. 2022(200):205–215. https://doi.org/10.1016/j.procs.2022.
01.219
16. Wen J, Gheisari M (2020) Using virtual reality to facilitate communication in the AEC domain:
a systematic review. Constr Innov, 509–542. https://doi.org/10.1108/CI-11-2019-0122
17. Delgado D, Manuel J, Oyedele L, Ajayi A, Akanbi L, Akinade O, Bilal M, Owolabi H (2019)
Robotics and automated systems in construction: understanding industry-specific challenges
for adoption. J Build Eng 26:100868. https://doi.org/10.1016/j.jobe.2019.100868
18. Abdelhameed WA (2013) Virtual reality use in architectural design studios: a case of studying
structure and construction. Procedia Procedia Comput Sci 25:220–230. https://doi.org/10.1016/
j.procs.2013.11.027
19. Kim JI, Li S, Chen X, Keung C, Suh M, Kim TW (2021) Evaluation framework for BIM-based
VR applications in design phase. J. Comput. Des. Eng. 8(3):910–922. https://doi.org/10.1093/
jcde/qwab022
20. Zhang Y, Liu H, Kang S, Al-hussein M (2020) Virtual reality applications for the built environ-
ment: research trends and opportunities. Autom Constr 118:103311. https://doi.org/10.1016/j.
autcon.2020.103311
21. Lilibeth JR, Kolloffel B (2021) The effect of task complexity and mental workload on task
performance in an immersive virtual reality training Lilibeth, University of Twente
22. Zdenek T, Jiri T, Knoflí R, Blecha P, Bradá F (2014) The process simulation using by virtual
reality 69:1015–1020. https://doi.org/10.1016/j.proeng.2014.03.084
23. Zhang M, Shu L, Luo X, Yuan M, Zheng X (2022) Virtual reality technology in construction
safety training: extended technology acceptance model. Autom Constr 135:104113. https://
doi.org/10.1016/j.autcon.2021.104113
24. Vosniakos EMG (2015) Design of a virtual reality training system for human – robot
collaboration in manufacturing tasks. https://doi.org/10.1007/s12008-015-0259-2
Real-Time Inference of Temporal
Emotional and Behavioral Conditions
of Learners in Problem Solving Within
Mixed Reality Environments

Édouard Castets, Ivan Mutis , and Gady Agam

Abstract Using new affordances of Machine Learning (ML) and Mixed Reality
(MX) technology, the presented research looks at how learners’ decision-making
in problem-solving can increase with the technology interaction. The approach
empowers users for enhanced problem-solving outcomes by enabling them to
perform adapted actions in an MX environment. For example, present an adapted
user’s view of 3D design components in the immersive space for any action (locking
and moving) in problem-solving (e.g., design interpretations). The aim is to incorpo-
rate ML to tailor the user’s ability to problem-solve complex 3D design components
by capitalizing on the flexibility of MX immersive environments. The system uses
a collection of physiological and pupillometry data to create real-time inferences of
the user’s state (emotional and physiological conditions). This paper focuses on the
first phase of developing the MX and ML system for the augmented technology-user
interaction to allow inference of the data in real time.

Keywords Spatial and temporal information · Spatial and temporal ability ·

Problem-solving

É. Castets
Department of Mechanical, Materials, and Aerospace Engineering,
Illinois Institute of Technology, Chicago, IL 60616, USA
I. Mutis (B)
Department of Civil, Architecture, and Environmental Engineering,
Illinois Institute of Technology, Chicago, IL 60616, USA
e-mail: imutissi@iit.edu
G. Agam
Department of Computer Science, Illinois Institute of Technology, Chicago 60616, USA

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 311
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_22
312 É. Castets et al.

1 Introduction

Significant limitations in the current pedagogy in construction engineering (CE)

include ineffective application, analysis, and synthesis of any form of design repre-
sentations to situational and physical contexts. The abundance of variable [1] in
“real-life” construction sites impacts learners’ abilities to effectively manage and
comprehend significant amounts of spatial (how to design components or resources
are related to one another in the 3D space) and temporal (the logic in a process, such
as the order, sequences, and hierarchies of the resources within a construction task
information) information [2]. Limited or no ability to process spatial and temporal
information (i.e., lack of spatial and temporal cognitive ability [3–7]) hinders the
understanding of designs and management of the varying local conditions (e.g.,
unplanned conditions) that transpire during construction processes. The limitation
leads to a suboptimal integration and internalization of complex spatial and temporal
information, hindering decision-making tasks in problem-solving. Research chal-
lenges emerge in the processing and integration of complex spatial information
effectively from image-based visualizations [8–12] effectively.
This research addresses these challenges by focusing on the capability of new tech-
nologies to enhance users’ cognitive processes. The approach explores the users’ use
of cognitive and physiological responses during problem-solving by examining how
learners’ decision-making in problem-solving can increase with technology inter-
action. The approach empowers users for enhanced problem-solving outcomes by
enabling them to perform adapted actions in a mixed reality (MX) environment—
e.g., present an adapted user’s view of 3D design components in the immersive
space for any action (locking and moving) in problem-solving (e.g., design interpre-
tations). The aim is to incorporate ML to tailor the user’s ability to the complexity
of 3D design components in problem-solving by capitalizing on the flexibility of
MX immersive environments. The system uses a collection of physiological and
pupillometry data to create real-time inferences of the user’s state (emotional and
physiological conditions) (Fig. 1).
This paper focuses on the first phase of developing the MX and ML system for
augmented technology-user interaction. The design incorporates methods to collect

Fig. 1 MX experience with

engineering design
components
Real-Time Inference of Temporal Emotional and Behavioral Conditions … 313

and treat data that could be used to build models for inference of the user state
(temporal emotional and behavioral conditions) in problem-solving (e.g., the user’s
realization of difficulty—or moments of impasse when problem-solving). The state
changes are characterized by physiological stimulation (symptoms, e.g., rapid heart-
beat, muscle tension) or behavioral indicators (e.g., rapid hand movements) as the
users (physical bodies) face a situation (e.g., difficulty in addressing a solution in
problem-solving). The states change as the situation from an activity fade—they are
temporal conditions.
The presented approach develops a ML model to predict and infer the users’ states
as users (learners) may experience different states in response to various triggers
about a particular situation or activity. The model will enable manipulation of the
temporal conditions, thereby facilitating decision-making in problem-solving.
For this first phase of the approach, the designed system is an enhanced OST-HMD
through sensors (visual sensory-motor and bio-metric sensors) and accompanying
computing power for an adaptive human-computer interaction (HCI). For instance,
the system will capture the user’s activity in the MX environment during simulations
using wearable sensing technology and eye-tracker devices in real time and predicts
the users’ state while manipulating MX reality objects. It addresses the technical
challenge of modeling the users’ physiological and behavioral responses by taking
advantage of one of the users’ free body motions in the MX environment when
manipulating MX objects. This system will be used for the problem-solving task
using MX to address the users’ cognitive ability in real-time. The contributions
of this paper are (i) method for data collection with unobtrusive and open-source
technology; (ii) framework for inference in real-time to assess the user’s state during
problem-solving.
The organization of paper this paper arranges the presentation with a summary of
the main contributions to the collection and use of biometric data, a description of
the MX problem-solving task, and the technology architecture and experimentation
outputs.

2 Related Work

Many studies have been conducted to assess or observe subjects’ abilities and psycho-
logical states based on physiological signals. There is a lack of studies on the explo-
ration of studying the user and augmented, virtual, or mixed reality technology inter-
action using physiological data. The interaction is founded on tracking physiological
signals together with eye features signals (i.e., pupillometry).
In 2018 P. Schmidt et al. [13] introduced the most extensive open-source dataset
related to stress using physiological features. Their study aimed to measure subjects’
physiological features (Table 1) as they were exposed to different affective stimu-
lation (stress, resting, and amusement). The outcome provided data to classify the
emotional state of the subject. Their data was collected using unobtrusive collec-
tion wearable sensing devices to measure the Blood Volume Pulse, Electrodermal
314 É. Castets et al.

Table 1 Physiological
Feature Device position Frequency
features
Electro-cardiogram Chest 700 Hz
Electrodermal activity Chest 700 Hz
Electro-myogram Chest 700 Hz
Body temperature Chest 700 Hz
Acceleration XYZ Chest 700 Hz
Respiration Chest 700 Hz
Acceleration XYZ Wrist 32 Hz
Blood volume pressure Wrist 64 Hz
Electrodermal activity Wrist 4 Hz
Body temperature Wrist 4 Hz

Activity, body temperature, and the 3-axis acceleration. The study presents their clas-
sification results on their data, with up to 80% accuracy with a 3-class classification
problem (amusement, stress, neutral state) and up to 93% with a 2-class classification
problem (no stress, stress).
Pupillometry has proved itself useful for the classification of emotional states. In
2014 Pedrotti [14] and his team presented a method based on wavelet transformation
on pupillometry signal to predict with a neural network the psychological stress of
a subject. They realized a data collection during a driving session on a simulator
where the subjects had to drive normally or in a stressful situation. They tracked
pupil diameter and electrodermal activity, and after training their neural network,
they predicted the psychological stress level with a precision of up to 79%.
Yamada et al. [15] propose to detect mental fatigue while watching a video from
eye tracking. The study presents a method based on a neural network with feature
selection to predict mental fatigue from data collected (pupil diameter and eye gaze
saccades features) during an experiment. Their final model succeeds in predicting
mental fatigue state with an accuracy of up to 91%.
One of the challenges of data collection system is to make the data usable for
inference in real time. For example, it appears that on some wearable sensing devices
for data collection, such as Empatica E4 used in the Philip Schmidt paper [16], the
motion artifact is interfering with the Blood volume pulse signal, resulting in a
noisy blood volume pulse signal. An important aspect of the blood volume pulse is
that a clean signal contains most of the electrocardiogram features, including heart
variability, which has been revealed to be an important feature in the Smith’s and
colleagues’ study [13]. Two notable articles tackle the question of cleaning the blood
volume Pulse signal. The first one is a study from Reiss [17]. Their study presents a
large dataset of physiological signals recorded during various activities with motion
artifacts. The data included acceleration, blood volume pulse, and electrocardiogram.
Moreover, their study presents activities based on a convolutional neural network to
Real-Time Inference of Temporal Emotional and Behavioral Conditions … 315

predict the heart rate from time-frequency spectra of the noisy BVP signal and 3-
axis acceleration with a mean average error performance of 7.65±4.2 bpm, which
outperforms many other methods in 2019.
In 2021, Chang et al. [17] provided a study where they proposed to train a convo-
lutional neural network to reconstruct a clean BVP signal from a noisy signal. Then
they offer a method to predict heart rate with a frequency spectra analysis. Their final
model predicts the heart rate with an average mean error of 1.65 bpm.

3 Problem-Solving Tasks Using MX and ML Ecosystem

The design is a problem-solving task that incorporates spatial and temporal abilities.
The problem-solving task is an experiment that enables the collection and treatment
of data to be used to build models of the ML inference of the user state (behav-
ioral conditions. An example of the behavioral conditions is the user’s realization of
difficulty—or moments of impasse when problem-solving (as shown in Fig. 2). As
the behavioral conditions analyzed on data collected on problems with simple MX
visualizations features, the same conditions should easily translate to more complex
CE problems with MX visualizations. The conditions will be used to indicate the
users’ spatial and temporal ability performance.
The approach allows MX users to experience different states in response to various
triggers about a complex task on problem-solving. The state changes are character-
ized by behavioral indicators (e.g., rapid hand movements) as the users (physical
bodies) face a situation (e.g., difficulty in addressing a solution in problem-solving).
The states change as the situation fade—they are temporal conditions.
The problem-solving design of the experiment consists of a cube puzzle (similar
to Tetris game features), where the users are required to move pieces into a grid
to complete a whole cube. Users must move around to select, move, and fit pieces
into the grid to complete the puzzle. Records of each user’s movements for spatial
and temporal information (e.g., time to relate MX objects to one another in the
MX space) for each user are collected for further analysis. Figure 2 illustrates the
sequence the user followed for the cube puzzle, each scene indicating the embodiment
(moving pieces, walking in the physical space) for the MX objects in the associated
problem-solving.
316 É. Castets et al.

Fig. 2 Foundational experiment for problem-solving on spatial and temporal cognition: individuals
move and lock MX objects (actions) as they experience different states in a process (from start to
finish) in response to various triggers about a complex task in problem-solving

4 System Design

For this first phase of the approach, the system design consists of an enhanced OST-
HMD through sensors (visual sensory-motor and bio-metric sensors) for human-
computer interaction (HCI). The system will capture the user’s activity in the MX
environment during simulations using wearable sensing and eye-tracker devices in
real-time for the prediction of the users’ state while manipulating MX reality objects.
The first device used for data collection is the wearable sensing technology device
to collect physiological feature data. (Table 2).
The system design incorporates an OST-HMD as an MX Headset (MS Hololens2).
The MX device is a sophisticated ecosystem with multiple sensing and tracking
technology functioning in a system to make high-quality MX experiences possible.
Although the MX device incorporates advanced technology to engage the user in
scenarios with a needed degree of awareness in the MX experience, user-computer
interaction research is a challenging activity due to still limited sensing, computing
capabilities, and the narrow field of view—as factors of portability (untethered
device) and affordability may come into place for this type of market-ready device.
The researchers address this challenge by incorporating additional computational
and sensing capabilities without reducing the ability of the user to move freely within
the physical environment with unbearable weight, heat, and battery life for lab activi-
ties. The researchers were able to build a new MX technology ecosystem for ML and
MX user-interaction research, facilitating activities such as data collection methods

Table 2 Initial collected

Feature Device
features
Blood volume pulse Empatica E4
3 axis acceleration Empatica E4
Body temperature Empatica E4
Electrodermal activity Empatica E4
Real-Time Inference of Temporal Emotional and Behavioral Conditions … 317

Fig. 3 Wireless communication framework between Pupil Core mounted on OST-HMD device
and a computer

to data treatment, such as using an eye tracker with the use of open-source data (see
Fig. 3).
The technology ecosystem was designed using the following specification factors:
(1) portability (e.g., maintaining untethered OST-HMD feature); and (2) stability
(e.g., no- demand for intensive computations to the OST-HMD). (3) reliability (e.g.,
no-demanding peripheral performance in the OST-HMD).
The developed ecosystem consisted of wearable sensing technology devices
connected via Bluetooth Low Energy protocol. The ecosystem included eye trackers
connected to a companion computer (e.g., Raspberry PI 4) powered by a low-
consumption battery. The design was conceived to permit the functionality of addi-
tional fix eye tracking cameras on the OST-HMD. The system is depicted in Fig. 3
and Fig. 4.
One of the major challenges was obtaining full wireless mode operations in each
of the sensing devices. For example, the pupil tracking software is robust and accurate
but uses many resources, including a GPU, to compute the eye tracking features in
real-time. Hence a wireless mode can’t use a companion computer to run the pupil
tracking software. Thus, the researcher chose to make the companion computer
redirect the video stream of the cameras (see Fig. 4) via a TCP protocol.
Important technical design considerations for the ecosystem were: (i) the addi-
tional eye tracker sensors must function at the same time as the OST-HMD; (ii) addi-
tional wearable sensing technology should be synchronized with the OST-HMD as
data will be collected in real-time and wirelessly for inference and real-time-feedback
to the OST-HMD.
318 É. Castets et al.

Fig. 4 Prototype OST-HMD + Core Device

5 Data Collection and Treatment

A summary of the collected data (features) is summarized in Table 3. The collected

data will be treated to feed the ML models.
Some of the features, like pupil diameter, can differ from one person to another.
Thus, the researchers developed a baseline moment during the collection to compare
the value from one moment to another.
The eye gaze velocity was computed from the Cartesian coordinates (x, y, z) of
the eye gaze point converted into spherical coordinates (r, θ, ψ), as expressed from
Eqs. 1, 2, 3 and 4.
√
r= x2 + y2 + z2 (1)
(y)
θ = ar cos (2)
r

Table 3 Collected features

Feature Device
Blood volume pulse Empatica E4
3 axis acceleration Empatica E4
Body temperature Empatica E4
Electrodermal activity Empatica E4
Pupil diameter Pupil Device
Eye gaze velocity Pupil Device
Real-Time Inference of Temporal Emotional and Behavioral Conditions … 319

ψ = (→
u 1 , u→2 ) (3)

where u→1 = (1, 0) and u→2 = (z, x)

√
dψ 2 +dθ 2
ω= (4)
dt
with ω, the gaze velocity
There was the need to pre-process the eye-tracker data. Firstly, as the pupil size
is very different from person to person, the pupil diameter signal at time Pd(t) was
compared to the mean pupil diameter during baseline, as in Eq. 5.

Pd(t)
Pdr elative(t) = (5)
Pd meanbaseline

The eye tracker provided eye-tracking data, including pupil diameter and eye gaze
position, even during blinks. Sensing issues occurs such as the no detection of pupil
data during blinking moments and, at other time, due to unusual eye position. Hence
these data need to be corrected during these moments. The employed method to
address this challenge was discarding the outliers and realizing a linear interpolation
for the missing data.
The pupil tracking software associates a confidence index between 0 and 1 for
each recorded data. Hence the retained data were data with a confidence index above
0.85. For the pupil diameter, the data that must be discarded was data where the
pupil diameter drops to 0 due to a lack of detection of pupil data. The remaining
data that was in the 2nd percentile was discarded. Also, for the eye gaze velocity, the
remaining outliers that must be discarded were high-velocity values due to a lack of
detection of the pupil. When the position of the eye gaze was moving with a high
velocity, the values not included in the 98th percentile were discarded.
On the clean blood volume pulse signal, it was possible to compute the heart rate
and the heart rate variability, which are features that have demonstrated their impact
on the classification of human emotional states [13]. The researchers considered that
the motion artifact strongly affects the blood volume pulse signal. To extract the heart
rate from the blood volume pulse signal, the researchers developed a framework on a
Deep Neural Networks model. The model will inform the different states of emotional
status (e.g., higher levels of stress) based on the physiological signal while the user
solves the problem (see problem-solving task in Sect. 3). This network model was
320 É. Castets et al.

Fig. 5 Prototype Hololens 2 + Core Device

based on the work of Reiss, et al. [16]. The model was trained on the PPG-DaLiA
dataset, the largest dataset focusing on electrocardiogram, acceleration, and blood
volume pulse in various activities (see Fig. 5). The input was 4-time frequency spectra
(1 for blood volume pulse, 1 per acceleration axis), and the label is the heart rate on
the sequence.

6 Results

6.1 Signal Collection with System

The study uses the problem-solving tasks in the MX explain Sect. 3. The outcome are
each feature collection and associated signal treatments: pupil diameter signal (see
Fig. 6), eye-gaze velocity (see Figs. 7), electrodermal activity and body temperature
signals (see Fig. 8), and blood volume pulse (BVP) and 3-axis acceleration signals
(see Fig. 9). It also important to annotate that were collected and treated in the
MX technology ecosystem in real-time in. Note that the pre-processing of the eye
data permitted the removal of all the outliers due to errors in pupil detection. The
framework provides a solution to collect data and proceed with inference in online
mode and in real-time.
Real-Time Inference of Temporal Emotional and Behavioral Conditions … 321

6.2 Output of the Deep Learning Model for Heart Rate

Prediction

At the phase of development, the first deep learning model to predict heart rate was
trained on the PPG-DaLia dataset with cross-validation training with early stopping.
Figure 10 represents the mean absolute error in beat per minute of each best model
for each fold on each validation set. The evaluation metrics of each fold’s model on
the validation set are given in Fig. 11.
The mean error of close to 18 beats per minute can be explained because subject
15 was part of the validation set of this fold, and subject 15 has a significantly higher
heart rate than other subjects. This model was fully trained on the validation set to
observe its performance on the testing set. However, the model has been performing
on the testing set with a mean error of 10.9 bpm and a standard deviation of the error
of 1.18 bpm. Yet with this low standard deviation, the heart rate variability should be
computed with a result close to reality. Future work includes measuring the ground
truth of the heart rate using a new sensor to be used as a comparative-based line for
the model.
322 É. Castets et al.

Fig. 6 Pupil diameter signal treatment

Fig. 7 Eye gaze velocity signal treatment

Real-Time Inference of Temporal Emotional and Behavioral Conditions … 323

Fig. 8 Electrodermal activity and body temperature signals

Fig. 9 Blood volume pulse and 3-axis acceleration signals

324 É. Castets et al.

Fig. 10 Validation loss over epoch for each fold’s training

Fig. 11 Mean and standard deviation of the error for each fold’s best model
Real-Time Inference of Temporal Emotional and Behavioral Conditions … 325

7 Conclusion and Discussion

This study presented the first phase of developing the MX and ML system for the
augmented technology-user interaction to allow inference of the data in real time.
The approach is a construction of a technology platform that incorporates wearable
sensing technologies and OST-HMD into a system. The ecosystem allows the collec-
tion of physiological and pupillometry signals in real-time using an online protocol.
The developed system aims to gain a highly effective user interaction with the MX
when problem-solving—by collecting users’ data to feed the ML models, assessing
the estate of the user, and getting feedback from the system to the user in real time.
The study also proposes a treatment of the data methodology on eye trackers,
as the reduction of errors on data to build inferences using ML models is a major
technology challenge. The method identifies and drops values due to errors in the
eye tracker. Examples of this outcome are given in the data collection system. The
system is ready to provide real-time data collection and inference for collected data.
The future work will be to improve the deep neural network(s), which predicts
the heart rate, and use a device to measure the ground truth of heart rate to eval-
uate the efficiency of this neural network data collected by the ecosystem system.
Secondly, data collection will be realized by creating a dataset to build and train a
deep learning model for user’s estate assessment. The researchers expect to improve
the data processing layer, especially concerning the blood volume pulse data, as the
heart rate prediction could be enhanced by improving the training and the archi-
tecture of the neural network. The researchers anticipate that the approach will be
successful in assessing the cognitive process states in problem-solving in Construc-
tion Engineering and Management tasks by treating physiological and pupillometry
signals collected with this system.

Acknowledgements This material is based upon work supported by the National Science
Foundation under Grant No. (2044444) and Grant (2040422).
Any opinions, findings, conclusions, or recommendations expressed in this material are those
of the author(s) and do not necessarily reflect the views of the National Science Foundation.

References

1. Jonassen DH (2011) Learning to solve problems: a handbook for designing problem-solving

learning environments. Routledge
2. Mutis I, Antonenko P (2022) Unmanned aerial vehicles as educational technology systems
in construction engineering education. J Inf Technol Constr 27:273–289. https://doi.org/10.
36680/j.itcon.2022.014
3. Antonenko P, Mutis I (2017) Using unmanned aerial systems to bring STEM field experiences to
the classroom. presented at the national association for research in science teaching (NARST),
San Antonio, TX. http://www.narst.org/annualconference/2017conference.cfm
4. Antonenko PD, Mutis I (2017) Empowering learning through remote visualizations using
unmanned aerial systems: perspectives of education and industry experts. presented at the
326 É. Castets et al.

2017 American educational research association, San Antonio, TX. http://www.aera.net/Pub

lications.
5. Mutis I, Issa RA (2014) Enhancing spatial and temporal cognitive ability in construction educa-
tion through augmented reality and artificial visualizations. In: Issa R, Flood I (eds) The inter-
national conference for computing in civil and building engineering (ICCCBE) and CIB-W78,
2014, Orlando, FL. ASCE, p 8
6. Mutis I (2015) Enhancing spatial and temporal cognitive ability in construction education
through the effect of artificial visualizations. In: O’Brien W, Caldas C (eds) 2015 ASCE
international workshop on computing in civil engineering, Austin, TX. ASCE, p 8
7. Mutis I (2018) Spatial-temporal cognitive ability: coupling representations to situations and
contexts for coordinating activities in the construction project environment. In: Mutis I,
Fruchter R, Menassa C (eds) Transforming engineering education through innovative computer-
mediated learning technologies Arlington, VA. ASCE, American Society of Civil Engineers,
ch 2, p 14
8. Glick S, Porter D, Smith C (2012) Student visualization: using 3-D models in undergraduate
construction management education. Int J Constr Educ Res 8(1):26–46. https://doi.org/10.
1080/15578771.2011.619247
9. Litzinger T, Lattuca LR, Hadgraft R, Newstetter W (2011) Engineering Education and the
Development of Expertise. J Eng Educ 100(1):123–150. https://doi.org/10.1002/j.2168-9830.
2011.tb00006.x
10. Reisslein M, Moreno R, Ozogul G (2010) Pre-college electrical engineering instruction: the
impact of abstract vs. contextualized representation and practice on learning. J Eng Educ
99(3):225–235. https://doi.org/10.1002/j.2168-9830.2010.tb01058.x
11. Sorby SA (2009) Educational research in developing 3-D spatial skills for engineering students.
Int J Sci Educ 31(3):459–480. https://doi.org/10.1080/09500690802595839
12. Towle E, Mann J, Kinsey B, O’Brien EJ, Bauer CF, Champoux R (2005) Assessing the self-
efficacy and spatial ability of engineering students from multiple disciplines. In: Proceedings
frontiers in education 35th annual conference, FIE 2005, pp S2C–15. https://doi.org/10.1109/
fie.2005.1612216
13. Schmidt P, Reiss A, Duerichen R, Van Laerhoven K (2018, October) Introducing WeSAD, a
multimodal dataset for wearable stress and affect detection. In: ICMI 2018 - Proceedings of
the 2018 International Conference on Multimodal Interaction, pp 400–408. https://doi.org/10.
1145/3242969.3242985
14. Pedrotti M, Mirzaei MA, Tedesco A, Chardonnet JR, Mérienne F, Benedetto S, Baccino T
(2014) Automatic stress classification with pupil diameter analysis. Int J Hum Comput Interact.
30(3):220–236. https://doi.org/10.1080/10447318.2013.848320
15. Yamada Y, Kobayashi M (2017, August) Fatigue detection model for older adults using eye-
tracking data gathered while watching video: evaluation against diverse fatiguing tasks. In: 2017
IEEE international conference on healthcare informatics (ICHI). IEEE, pp 275–284. https://
doi.org/10.1109/ICHI.2017.74. http://ieeexplore.ieee.org/document/8031157/
16. Reiss A, Indlekofer I, Schmidt P, Van Laerhoven K (2019) Deep PPG: large-scale heart rate
estimation with convolutional neural networks. Sensors 19(14). https://doi.org/10.3390/S19
143079.
17. Chang X, Li G, Xing G, Zhu K, Tu L (2021) DeepHeart: a deep learning approach for accurate
heart rate estimation from PPG signals. ACM Trans Sens Netw 17(2). https://doi.org/10.1145/
3441626
Other
Blockchain-Enabled-Trust
for Construction Project Governance:
A Conceptual Framework

Seongha Hwang, Mingzhu Wang, Mohamed Osmani, and Karen Blay

Abstract Construction project governance (CPG) acts as a ‘steering wheel’ that

keeps a construction project steady in a challenging external environment. CPG
frameworks can be divided into two types: control-based hierarchical governance and
trust-based relational governance. To increase cooperation and networking between
the participants in a project, effective CPG depends on well-combined control and
trust mechanisms. However, the current CPG framework, which is control-based,
is biased toward hierarchical governance and can lead to a lack of trust in the
construction project. These issues have resulted in several studies that have focused
on blockchain technology (BT) as a potential enabler for addressing the low levels
of trust in CPG. However, existing research in this area has failed to address the
nature of the relationship between trust as a key CPG challenge and BT. Hence, this
paper aims to identify trust as a major CPG challenge and potential BT capabilities
to reveal the relationship, through a state-of-the-art review. The findings show that
the effective use of three decentralized BT capabilities (decentralized data storage,
validity, and access) can positively influence the three relational norms (mutuality,
flexibility, and solidarity), which are the functional tools of relational governance.
Thus, a more flexible trust-based CPG framework is proposed by establishing rela-
tional governance with blockchain technology. Ultimately, this study is expected to
enhance the trust between the key stakeholders in construction projects and enable
more efficient collaboration and networking between them.

Keywords Blockchain technology (BT) · Construction project governance

(CPG) · Hierarchical governance (HG) · Relational governance (RG) · Trust

S. Hwang (B) · M. Wang · M. Osmani · K. Blay

School of Architecture, Building and Civil Engineering, Loughborough University,
Loughborough LE11 3TU, UK
e-mail: S.Hwang3@lboro.ac.uk

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 329
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_23
330 S. Hwang et al.

1 Introduction

The complexity of construction projects can increase due to external environmental

factors (such as the technological development and fragmentation of the construc-
tion industry) and internal factors (such as stakeholder management issues, the
complex scope of a project, and the range of project interfaces) [1, 2]. This increasing
complexity can result in low project performances [3, 4]. Accordingly, the respon-
siveness and resilience characteristics of governance are emphasized for steering and
decreasing construction project complexity [5]. Construction project governance
(CPG) provides appropriate structures and processes with decision-making tools
and systems [6, 7]. In particular, CPG aligns construction project objectives with
the overall organizational strategies developed to benefit a diverse range of stake-
holders [8]. However, the type of construction project governance (CPG) that is being
currently practiced—hierarchical CPG—has failed to respond to these two charac-
teristics (responsiveness and resilience) because of the low levels of trust across
construction project networks [9, 10]. Indeed, many construction projects based on
a hierarchical governance framework face low performance such as time delay and
cost overrun due to governance failures that arise from a low level of trust [11, 12].
In response to this, various industries have begun to consider blockchain technology
(BT) as an enabler for a more flexible trust-based CPG due to its transparency charac-
teristic [13, 14]. BT as a distributed database or digital ledger provides more flexible
access and validity by storing records of transactions and events in blocks [15, 16].
However, existing research in this area has failed to address the nature of the corre-
lation between CPG and BT [13, 17–19]. Accordingly, this research aims to identify
trust as a major CPG challenge and potential BT capabilities to reveal the relation-
ship. To achieve this aim, this study conducted a state-of-the-art literature review on
blockchain and trust in construction project governance. Based on the review, this
paper proposes a new type of CPG framework: the BT-Enabled-Trust Framework
for CPG.

2 Research Methodology

The state-of-the-art review is adopted in this research because it is to provide infor-

mation on the level of scientific development in the field and offers new perspectives
on the relationship between trust as a key CPG challenge and BT [20]. The Scopus
and Web of Science academic databases were searched to obtain high-quality publi-
cations [21]. Details of the inclusion criteria and search terms are given in Table
1.
Figure 1 describes the procedure of the state-of-the-art review conducted in this
study. Specifically, Step 1 performed scoping and identification of relevant literature
by using keywords and inclusion criteria in the selected databases. In Step 2, the
screening process was performed based on the abstract review. Finally, through the
Blockchain-Enabled-Trust for Construction Project Governance ... 331

Table 1 Details of a state-of-the-art review

Category Descriptions
Databases Scopus and Web of Science
Inclusion Academic journal papers including the following keywords:
criteria project governance, blockchain, trust, relational governance,
contractual governance, hierarchical governance, relational norm, construction
project
Search terms Term 1: (“project governance” OR “governance” OR “relational
governance” OR “contractual governance” OR “trust” OR “relational norm”)
AND “construction project”
Term 2: (“distributed ledger” OR “smart contracts” OR “trust”) AND
“blockchain”

eligibility process based on the full-text review, a total of 60 papers were selected
for the final analysis. Ultimately, the state-art-review revealed that existing research
has failed to examine the relationship between BT and relational CPG.

Fig. 1 The state-of-the-art review procedure in this study (devised by authors)

332 S. Hwang et al.

Table 2 CPG theories (devised by authors based on the literature) [23–25]

Governance theory Assumption Characteristic
Agency theory Agent’s opportunism Transparency, Accountability,
Responsibility
Transaction costs Transactional costs Transparency, Accountability,
economics (TCE) Responsibility
Stakeholder theory Stakeholders’ benefit Fairness
Contingency theory Organizational effectiveness Responsiveness
Network theory Network efficiency Resilience

3 Construction Project Governance

3.1 Characteristics of Construction Project Governance

Many governance theories are used in CPG research; these provide the foundation for
conceptualizing and understanding CPG [22]. In general, governance theories can be
divided into three major categories [23]: agency theory, transaction cost economics
(TCE), and stakeholder theory. In addition to these general governance theories,
contingency theory and network theory are both emphasized in CPG research and
provide additional information about the characteristics of a construction project [24].
As can be seen from Table 2, CPG is characterized by six distinct factors—trans-
parency, accountability, responsibility, fairness, responsiveness, and resilience—
arising from five CPG theories [23–25]. Various assumptions based on CPG theories
provide contextual information about the current practices and challenges of CPG
[12].

3.2 Challenges of Construction Project Governance

Governance frameworks can be divided into two types: hierarchical governance (HG)
and relational governance (RG), details of which are shown in Table 3 [26].

Table 3 Governance frameworks (devised by authors based on the literature) [27, 28]
Governance framework Also known as: Power Binding Mechanism
relations
Hierarchical Contractual or formal Static Legal Control
governance (HG) governance
Relational Cooperative or Flexible Psychological Trust
governance (RG) network governance
Blockchain-Enabled-Trust for Construction Project Governance ... 333

In general, CPG adopts a HG framework to support a better response to the

complexity of construction projects [22, 29]. Specifically, the decision-making
process for the current CPG framework is the control mechanism (not the trust
mechanism) in the governance context, whereby future outcomes are perceived as
negative [30, 31]. Accordingly, the current CPG framework has a relatively high focus
on HG through its utilization of many control-based project governance tools, such
as procurement methods, construction contracts, and relevant laws and regulations
[11].
However, the detailed contractual specifications and rigorous monitoring
processes of the current hierarchical CPG framework lead to a lack of trust among the
stakeholders within the construction project [32]. This lack of trust has been deemed
a critical CPG challenge and results in reduced responsiveness to the internal and
external construction project environments and less resilience in the network [33, 34].
In other words, the current hierarchical CPG framework is designed to only respond
to the characteristics of transparency, accountability, responsibility, and fairness.
Effective CPG relies on well-combined control and trust mechanisms [35].
However, the current control-based CPG framework, with its lack of trust, is biased
towards HG [30]. Consequently, several CPG studies have converted the hierarchical
control-based CPG to a more flexible trust-based governance, which is founded on
the principles of relational governance.

4 Trust in Relational Construction Project Governance

The current CPG framework can be converted into a more flexible trust-based CPG
framework by establishing relational governance (RG) [11, 27]. RG is defined as
governing inter-organizational relationships through informal structures and shared
norms [36]. RG aims to avoid opportunistic behavior by exploiting participants’
willingness to preserve their own and the organization’s reputation and develop long-
term relationships [37]. Specifically, RG leads to reliable transactional relationships,
by using the concept of relational norm as a governance tool [38]. Relational norm
theory describes a social process that regulates relationship behaviors [39], and there
are three underlying dimensions (see Table 4) [40, 41]. Ultimately, relational norms
are the core components of relational governance [39].

Table 4 Three relational norm dimensions (devised by authors based on the literature) [40, 41]
Dimension Description
Mutuality Willingness to mutually improve the current situation compared to a previous
situation
Flexibility Willingness to accept adjustments and contract modifications
according to changed circumstances
Solidarity Willingness to maintain and stabilize partnership relationships
334 S. Hwang et al.

Table 5 Three types of trust (devised by authors based on the literature) [42, 43]
Trust type Description
System-based trust Aims to strengthen the communication channels between the parties in a
transaction by establishing a formalized and procedural system
Cognition-based Refers to the knowledge-based trust between the parties in a
trust transaction by establishing a formalized and procedural system
Affect-based trust Establishes a sentimental framework to strengthen the emotional bonds
between individuals and parties

There are three types of trust in a construction project, including system-based,

cognition-based and affect-based trust, as shown in Table 5. Ultimately, RG can
positively affect trust through the establishment of different relational norms at the
project level [42–44].
However, RG requires a considerable amount of time and resources to develop,
which often limits its implementation and development [37]. This has led to several
studies [14, 19, 45] that have focused on blockchain as a potential enabler for
establishing RG, which are discussed in the next section.

5 Blockchain Technology

5.1 Characteristics of Blockchain Technology

Blockchain Technology consists of four elements: peer-to-peer (P2P) networks,

hashing algorithms, cryptography, and consensus algorithms (see Table 6) [46].
Specifically, BT’s blocks are connected with cryptographic hashes with a timestamp
to form a connected P2P network by using consensus algorithms [14].

Table 6 Blockchain elements (devised by authors based on the literature) [18]

Element Description
Peer-to-peer A P2P network is a type of distributed system that keeps copies of project files
(P2P) network on each node, which eliminates the need for a server to act as a central
administrator
Hashing Hashing algorithms connect multiple blocks and the multiple
algorithms transactions within a block by creating an encryption link between each block
Cryptography Cryptography secures blockchain data, such as user privacy and
transaction information, by using symmetric and asymmetric
encryption and decryption
Consensus The consensus algorithm aims to reach an agreement on a single piece of data
algorithm
Blockchain-Enabled-Trust for Construction Project Governance ... 335

Table 7 Blockchain layers (devised by authors based on the literature) [18, 48]
Layer Sub-layer Description
Foundation Infrastructure Blockchain sets a decentralized database through P2P
layer networks
Data The data layer processes the blockchain data through its
hashing algorithms and cryptography
Network The network layer maintains an effective state through
communication, propagation, and synchronization
between blockchain nodes
Consensus The consensus layer draws consensus among all nodes within
the blockchain network through various consensus algorithms
Application Application The application layer is based on user application
layer programming interfaces (APIs) and frameworks

Blockchain Technology forms a layered architecture that consists of two main

layers: (1) the foundation layer, which is the underlying architecture of the blockchain
database, and (2) the application layer [47]. In particular, the foundation layer
comprises four sub-layers: infrastructure, data, network, and consensus. Ultimately,
the two main layers and five blockchain sub-layers represent BT. Each layer is
described in Table 7.
Such a layered architecture leads to seven crucial characteristics of BT, including
immutability, transparency, anonymity, security, decentralization, consensus and
programmability [49–51]. Details of each characteristic are explained in Table 8.
These characteristics indicate the capability of BT to improve trust in construction
project governance.

5.2 Capabilities of Blockchain Technology

Based on the seven blockchain characteristics outlined in Table 8, there are three
blockchain capabilities: decentralized data storage, decentralized data validity, and
decentralized data access [54]. Specifically, the relationships between the blockchain
capabilities and characteristics are illustrated in Table 9.
The three blockchain use cases – Blockchain 1.0, 2.0, and 3.0—are associated
with the different blockchain architecture layers and capabilities (see Table 10).
Specifically, Bitcoin is the world’s first blockchain application (Blockchain 1.0) [19].
Blockchain 2.0 introduce various financial applications such as lending platforms,
decentralized exchange and stable coins by using smart contracts [18]. Blockchain
3.0 describes the autonomous industrial services in numerous industries including
banking, public sectors, healthcare, insurance, manufacturing and supply chains [59].
336 S. Hwang et al.

Table 8 Blockchain characteristics (devised by authors based on the literature) [50, 52, 53]
Layer Characteristic Description
Foundation Immutability Transactions stored in the BT cannot be cancelled once they
layer have been added to the blockchain
Transparency The BT network participant can view and trace records
within any node because all transactions are recorded and
verified with a timestamp
Anonymity By using public and private keys within network
transactions, blockchain can maintain anonymity for its
users and protect personal information
Security Information stored in the BT is encrypted through
asymmetric public-key cryptography to prevent fraud
Decentralization BT is a distributed ledger system that does not have a
central data storage mechanism
Consensus All relevant parties agree to verify transactions within the
blockchain network
Application Programmability BT provides computer programming capabilities that
layer automate business algorithms

Table 9 Blockchain capabilities (devised by authors based on the literature) [55–58]

Capability Description
Decentralized data A blockchain-enabled decentralized database guarantees immutability,
storage transparency, anonymity, security, decentralization, and consensus.
Accordingly, a blockchain network leads to a reliable and standardized
decentralized data storage system
Decentralized data The decentralized data storage system provides the decentralized data
validity validity function by using cryptography and consensus algorithms to
prevent fraud and achieve decentralized trust in the blockchain
Decentralized data Programmability on the blockchain application layer optimizes
access decentralized data access within the blockchain network. In particular,
smart contracts and DApps (decentralized applications) maximize the
utilization of decentralized data accessibility on the blockchain by
providing autonomous services based on predetermined business
algorithms and user-friendly interfaces

Table 10 Blockchain use cases (devised by authors based on the literature) [15, 18]
Layer Capability Use Case Application
Foundation layer Decentralized Data Storage Blockchain 1.0 Cryptocurrencies
Decentralized Data Validity
Application layer Decentralized Data Access Blockchain 2.0 Smart contracts and financial
applications
Blockchain 3.0 Industry applications
Blockchain-Enabled-Trust for Construction Project Governance ... 337

Fig. 2 The capabilities of blockchain technology (devised by authors)

Figure 2 shows the relationships between the BT elements, architecture, character-

istics, use cases, and capabilities. Specifically, BT’s four different elements including
P2P networks, hashing algorithms, cryptography, and consensus algorithms form
the layered architecture consisting of infrastructure, data, network, consensus and
application layer [46]. The layered architecture of BT creates seven distinct charac-
teristics including immutability, transparency, anonymity, security, decentralization,
consensus and programmability [49–51]. Ultimately, three BT capabilities (decen-
tralized data storage, decentralized data validity, and decentralized data access) based
on the seven characteristics triggered Blockchain 1.0, 2.0 and 3.0 era [18, 54].

6 Blockchain as a Trust Enabler for Relational

Construction Project Governance

The three decentralized blockchain capabilities (data access, data validity and data
storage) support the three relational norm dimensions—mutuality, flexibility and
solidarity, as explained in Table 11. This support indicates that blockchain capabilities
have the potential to assist relational governance and increase trust at the project level
[14].
338 S. Hwang et al.

Table 11 Blockchain capabilities and the relational norm dimensions (devised by authors based
on the literature) [27, 41, 60–62]
Relational norm Blockchain capabilities
dimension
Mutuality Two decentralized blockchain capabilities (data storage and data
validity) improve mutuality. The decentralized database architecture is used
for remote interaction and information-sharing between participants in the
construction project network through the P2P network. Also,
decentralized data validity enables collaboration between non-trustworthy
parties through the decentralized network
Flexibility All three decentralized blockchain capabilities (data storage, data validity
and data access) are involved in improving flexibility. The decentralized
storage system secures the flexibility of the system controls. In addition,
decentralized data access determines the level of data access within the
construction project network. It also enables real-time data sharing and rapid
integration of data into the network to cope with the rapidly changing
external environment through decentralized data access
Solidarity All three decentralized blockchain capabilities (data storage, data validity
and data access) enhance solidarity. Construction information-sharing in
real-time leads to better collaboration between participants in an
established partnership. Accordingly, information transparency (gained
through decentralized data storage and decentralized data validity) and
stable access to information (gained through decentralized access) can
increase participants’ solidarity

7 A Framework for Blockchain-Enabled-Trust

in Construction Project Governance

From the literature review, it was found that the existing CPG system only focused on
hierarchical CPG based on procurement methods, construction contracts and project
assurance. Therefore, a BT-Enabled-Trust framework was proposed in this study to
provide an integrated solution by establishing a practical relational CPG system and
integrating it with the existing CPG system (see Fig. 3).
Specifically, effective CPG depends on well-combined control and trust mech-
anisms [35]. However, the current control-based CPG framework, with its lack of
trust, is biased toward HG [30]. In other words, the current approach to CPG has very
little balance between the control and trust governance mechanisms (i.e., between
hierarchical and relational governance). Meanwhile, the control governance mecha-
nism is an essential element of CPG [63, 64]. The current control mechanism cannot
be ignored because it uses essential governance tools such as procurement methods,
contracts, and project assurances to make construction projects more predictable [65].
Due to its characteristics, BT can be an enabler for a more trust-based construction
project governance by establishing RG. Therefore, a BT-enabled trust framework is
proposed for CPG. Specifically, three BT capabilities (decentralized data storage,
validity and access) positively impact three relational norms (mutuality, flexibility
and solidarity) of RG. Such BT-enabled relational norms can raise the levels of
Blockchain-Enabled-Trust for Construction Project Governance ... 339

Fig. 3 A proposed BT-Enabled Trust Framework for CPG (devised by authors)

system-based, cognition-based and affect-based trust in construction projects. Conse-

quently, this BT-enabled trust enables responsiveness to the internal and external
construction project environment, and resilience for networking within the construc-
tion. Ultimately, BT-enabled trust, gained through RG equipped with blockchain,
offers a cornerstone for building a more effective CPG framework.

8 Conclusions

This paper sets out to identify trust as a major CPG challenge and potential BT
capabilities to reveal the relationship, through a state-of-the-art review. A balanced
CPG of the control and trust mechanisms in a construction project is needed to
cope with the complexity of the internal and external environments the project sits
within. However, the current CPG framework is unbalanced because it is biased
toward control-based hierarchical governance. Establishing and strengthening rela-
tional governance is essential for providing a more flexible trust-based CPG. The
findings of this review suggest that BT could be a crucial enabler for establishing
RG. It has the ability to strengthen the three relational norm dimensions, which are
essential tools of relational governance, based on three crucial BT capabilities: decen-
tralized data access, data validity and data storage. Ultimately, a BT-enabled trust
framework is proposed based on BT-enabled relational governance, which operates
340 S. Hwang et al.

Fig. 4 A proposed BT-Enabled Trust System for future research (devised by authors)

within the current control-based hierarchical CPG framework, and can effectively
control the complex construction project environment by satisfying the six distinct
CPG factors.
To establish the practicalities of using this new framework, future research could
further investigate how existing project governance systems function as gover-
nance and explore appropriate BT infrastructures, data, networks, consensus, and
applications that meet the requirements of relational CPG (see Fig. 4).
Through the establishment of a BT-enabled trust system, the new proposed BT-
Enabled Trust Framework for CPG provides the opportunity to allow for a rapid
response to a complex project environment without changing the existing project
governance system, which is likely to be based on construction contracts, procure-
ment methods, and project assurances. Ultimately, BT-Enabled Trust Framework for
CPG has the potential to act as a steering to develop the practical BT-Enabled Trust
system that keeps a construction project steady through more efficient collaboration
and networking between the key project stakeholders including clients, consultants,
and contractors.
Blockchain-Enabled-Trust for Construction Project Governance ... 341

References

1. Kermanshachi S, Dao B, Shane J, Anderson S (2016) Project complexity indicators and

management strategies – a Delphi study. Procedia Eng 145:587–594
2. Luo L, He Q, Jaselskis EJ, Xie J (2017) Construction project complexity: research trends and
implications. J Constr Eng Manag 143:04017019. https://doi.org/10.1061/(ASCE)CO.1943-
7862.0001306
3. Dao B, Kermanshachi S, Shane J, Anderson S, Hare E (2016) Identifying and measuring project
complexity. Procedia Eng 145:476–482. https://doi.org/10.1016/j.proeng.2016.04.024
4. Luo L, He Q, Xie J, Yang D, Wu G (2017) Investigating the relationship between project
complexity and success in complex construction projects. J Manag Eng 33:04016036. https://
doi.org/10.1061/(ASCE)ME.1943-5479.0000471
5. ul Musawir A, Serra CEM, Zwikael O, Ali I (2017) Project governance, benefit manage-
ment, and project success: towards a framework for supporting organizational strategy imple-
mentation. Int J Proj Manag 35:1658–1672 (2017). https://doi.org/10.1016/j.ijproman.2017.
07.007
6. Winch GM (2014) Three domains of project organising. Int J Project Manag 32:721–731.
https://doi.org/10.1016/j.ijproman.2013.10.012
7. Levie F, Burke CM, Lannon J (2017) Filling the gaps: an investigation of project governance in
a non-governmental organisation’s response to the Haiti earthquake disaster. Int J Proj Manag
35:875–888. https://doi.org/10.1016/j.ijproman.2016.10.007
8. Sirisomboonsuk P, Gu VC, Cao RQ, Burns JR (2018) Relationships between project governance
and information technology governance and their impact on project performance. Int J Proj
Manag 36:287–300. https://doi.org/10.1016/j.ijproman.2017.10.003
9. Too E, Le T, Weaver P, Bourne L (2017) The core functions of project governance. Presented at
the AUBEA 2017: Australasian universities building education association conference. https:/
/doi.org/10.29007/bxg5
10. Too EG, Weaver P (2014) The management of project management: a conceptual framework
for project governance. Int J Proj Manag 32:1382–1394. https://doi.org/10.1016/j.ijproman.
2013.07.006
11. Yan L, Zhang L (2020) Interplay of contractual governance and trust in improving construction
project performance: dynamic perspective. J Manag Eng 36:04020029. https://doi.org/10.1061/
(ASCE)ME.1943-5479.0000791
12. Zwikael O, Smyrk J (2015) Project governance: balancing control and trust in dealing with
risk. Int J Proj Manag 33:852–862. https://doi.org/10.1016/j.ijproman.2014.10.012
13. Demagistris PE, Petruzzi S, Pampaloni R, Šmigić M, De Marco A, Khan W, Ottaviani FM
(2022) Digital enablers of construction project governance. https://doi.org/10.48550/ARXIV.
2205.05930
14. Lumineau F, Wang W, Schilke O (2021) Blockchain governance—a new way of organizing
collaborations? Organ Sci 32:500–521. https://doi.org/10.1287/orsc.2020.1379
15. Kiu MS, Chia FC, Wong PF (2020) Exploring the potentials of blockchain application in
construction industry: a systematic review. Int J Constr Manag 0:1–10. https://doi.org/10.1080/
15623599.2020.1833436
16. Wang Z, Jin H, Dai W, Choo K-KR, Zou D (2021) Ethereum smart contract security research:
survey and future research opportunities. Front Comput Sci 15:152802. https://doi.org/10.1007/
s11704-020-9284-9
17. Akgiray V (2019) The potential for blockchain technology in corporate governance. https://
doi.org/10.1787/ef4eba4c-en
18. Perera S, Nanayakkara S, Rodrigo MNN, Senaratne S, Weinand R (2020) Blockchain tech-
nology: is it hype or real in the construction industry? J Ind Inf Integr 17:100125. https://doi.
org/10.1016/j.jii.2020.100125
19. Zwitter A, Hazenberg J (2020) Decentralized network governance: blockchain technology and
the future of regulation. Front Blockchain 3:12. https://doi.org/10.3389/fbloc.2020.00012
342 S. Hwang et al.

20. ASCE: Publishing in ASCE journals: a guide for authors. American Society of Civil Engineers,
Reston, VA (2022). https://doi.org/10.1061/9780784479018
21. Zhai Z, Shan M, Darko A, Le Y (2020) Visualizing the knowledge domain of project gover-
nance: a scientometric review. Adv Civ Eng 2020:1–15. https://doi.org/10.1155/2020/681
3043
22. Guo F, Chang-Richards Y, Wilkinson S, Li TC (2014) Effects of project governance structures
on the management of risks in major infrastructure projects: a comparative analysis. Int J Proj
Manag 32:815–826. https://doi.org/10.1016/j.ijproman.2013.10.001
23. Derakhshan R, Turner R, Mancini M (2019) Project governance and stakeholders: a literature
review. Int J Proj Manag 37:98–116. https://doi.org/10.1016/j.ijproman.2018.10.007
24. ul Musawir A, Abd-Karim SB, Mohd-Danuri MS (2020) Project governance and its role
in enabling organizational strategy implementation: a systematic literature review. Int J Proj
Manag 38:1–16. https://doi.org/10.1016/j.ijproman.2019.09.007
25. Biesenthal C, Wilden R (2014) Multi-level project governance: Trends and opportunities. Int
J Project Manage 32:1291–1308. https://doi.org/10.1016/j.ijproman.2014.06.005
26. Haq SU, Gu D, Liang C, Abdullah I (2019) Project governance mechanisms and the perfor-
mance of software development projects: moderating role of requirements risk. Int J Proj Manag
37:533–548. https://doi.org/10.1016/j.ijproman.2019.02.008
27. Benítez-Ávila C, Hartmann A, Dewulf G, Henseler J (2018) Interplay of relational and contrac-
tual governance in public-private partnerships: the mediating role of relational norms, trust and
partners’ contribution. Int J Proj Manag 36:429–443. https://doi.org/10.1016/j.ijproman.2017.
12.005
28. Colombelli A, Paolucci E, Ughetto E (2019) Hierarchical and relational governance and the
life cycle of entrepreneurial ecosystems. Small Bus Econ 52:505–521. https://doi.org/10.1007/
s11187-017-9957-4
29. Caldwell ND, Roehrich JK, Davies AC (2009) Procuring complex performance in construction:
London heathrow terminal 5 and a private finance initiative hospital. J Purch Supply Manag
15:178–186. https://doi.org/10.1016/j.pursup.2009.05.006
30. Ruuska I, Ahola T, Artto K, Locatelli G, Mancini M (2011) A new governance approach for
multi-firm projects: lessons from Olkiluoto 3 and Flamanville 3 nuclear power plant projects.
Int J Proj Manag 29:647–660. https://doi.org/10.1016/j.ijproman.2010.10.001
31. Das TK, Teng B-S (2001) Trust, control, and risk in strategic alliances: an integrated framework.
Organ Stud 22:251–283. https://doi.org/10.1177/0170840601222004
32. Kadefors A (2004) Trust in project relationships-inside the black box. Int J Proj Manag 22:175–
182. https://doi.org/10.1016/S0263-7863(03)00031-0
33. Müller R, Andersen ES, Kvalnes Ø, Shao J, Sankaran S, Turner JR, Biesenthal C, Walker
D, Gudergan S (2013) The interrelationship of governance, trust, and ethics in temporary
organizations. Proj Manag J 44:26–44. https://doi.org/10.1002/pmj.21350
34. Zhai Z, Ahola T, Le Y, Xie J (2017) Governmental governance of megaprojects: the case of
EXPO 2010 Shanghai. Proj Manag J 48:37–50. https://doi.org/10.1177/875697281704800103
35. Edelenbos J, Eshuis J (2012) The interplay between trust and control in governance processes:
a conceptual and empirical investigation. Adm Soc 44:647–674. https://doi.org/10.1177/009
5399711413876
36. Liu B, Shao Y-F, Liu G, Ni D (2022) An evolutionary analysis of relational governance in an
innovation ecosystem. SAGE Open 12:215824402210930. https://doi.org/10.1177/215824402
21093044
37. Cao Z, Lumineau F (2015) Revisiting the interplay between contractual and relational gover-
nance: a qualitative and meta-analytic investigation. J Oper Manag 33–34:15–42. https://doi.
org/10.1016/j.jom.2014.09.009
38. Zhou Y, Zhang X, Zhuang G, Zhou N (2015) Relational norms and collaborative activities:
Roles in reducing opportunism in marketing channels. Ind Mark Manag 46:147–159. https://
doi.org/10.1016/j.indmarman.2015.01.014
39. Ott CM, Ivens BS (2009) Revisiting the norm concept in relational governance. Ind Mark
Manag 38:577–583. https://doi.org/10.1016/j.indmarman.2009.04.004
Blockchain-Enabled-Trust for Construction Project Governance ... 343

40. Valta KS (2013) Do relational norms matter in consumer-brand relationships? J Bus Res 66:98–
104. https://doi.org/10.1016/j.jbusres.2011.07.028
41. Müller R, Martinsuo M (2015) The impact of relational norms on information technology
project success and its moderation through project governance. Int J Manag Proj Bus 8:154–176.
https://doi.org/10.1108/IJMPB-04-2014-0036
42. Wong WK, Cheung SO, Yiu TW, Pang HY (2008) A framework for trust in construction
contracting. Int J Proj Manag 26:821–829. https://doi.org/10.1016/j.ijproman.2007.11.004
43. (Alice) Qian X, Papadonikolaki E (2020) Shifting trust in construction supply chains through
blockchain technology. ECAM 28:584–602. https://doi.org/10.1108/ECAM-12-2019-0676
44. Zheng X, Yuan J, Guo J, Skibniewski M, Zhao S (2018) Influence of relational norms on
user interests in PPP projects: mediating effect of project performance. Sustainability 10:2027.
https://doi.org/10.3390/su10062027
45. van Pelt R, Jansen S, Baars D, Overbeek S (2021) Defining blockchain governance: a framework
for analysis and comparison. Inf Syst Manag 38:21–41. https://doi.org/10.1080/10580530.
2020.1720046
46. Yang R, Wakefield R, Lyu S, Jayasuriya S, Han F, Yi X, Yang X, Amarasinghe G, Chen S (2020)
Public and private blockchain in construction business process and information integration.
Autom Constr 118:103276. https://doi.org/10.1016/j.autcon.2020.103276
47. Roth U, Ngne Djoua T (2018) Message exchange on base of a blockchain-based layered
architecture. IT Inf Technol 60:253–261. https://doi.org/10.1515/itit-2017-0037
48. Shi R (2021) Blockchain network as a platform: conceptualising its adapted layered architecture
design. JBBA 4:1–6. https://doi.org/10.31585/jbba-4-2-(1)2021
49. Wang Z, Wang T, Hu H, Gong J, Ren X, Xiao Q (2020) Blockchain-based framework for
improving supply chain traceability and information sharing in precast construction. Autom
Constr 111:103063. https://doi.org/10.1016/j.autcon.2019.103063
50. Mahmudnia D, Arashpour M, Yang R (2022) Blockchain in construction management: appli-
cations, advantages and limitations. Autom Constr 140:104379. https://doi.org/10.1016/j.aut
con.2022.104379
51. Yaqoob I, Salah K, Jayaraman R, Al-Hammadi Y (2022) Blockchain for healthcare data
management: opportunities, challenges, and future recommendations. Neural Comput Appl
34:11475–11490. https://doi.org/10.1007/s00521-020-05519-w
52. Treiblmaier H (2019) Toward more rigorous blockchain research: recommendations for writing
blockchain case studies. Front Blockchain 2:3. https://doi.org/10.3389/fbloc.2019.00003
53. Viriyasitavat W, Hoonsopon D (2019) Blockchain characteristics and consensus in modern
business processes. J Ind Inf Integr 13:32–39. https://doi.org/10.1016/j.jii.2018.07.004
54. Bauer I, Zavolokina L, Leisibach F, Schwabe G (2020) Value creation from a decentralized car
ledger. Front Blockchain 2:30. https://doi.org/10.3389/fbloc.2019.00030
55. Zahed Benisi N, Aminian M, Javadi B (2020) Blockchain-based decentralized storage
networks: a survey. J Netw Comput Appl 162:102656. https://doi.org/10.1016/j.jnca.2020.
102656
56. Du Y, Duan H, Zhou A, Wang C, Au MH, Wang Q (2021) Enabling secure and efficient
decentralized storage auditing with blockchain. IEEE Trans Dependable Secure Comput, 1.
https://doi.org/10.1109/TDSC.2021.3081826
57. Cai C, Weng J, Yuan X, Wang C (2021) Enabling reliable keyword search in encrypted decen-
tralized storage with fairness. IEEE Trans Dependable Secure Comput 18:131–144. https://
doi.org/10.1109/TDSC.2018.2877332
58. Yao Z, Pan H, Si X, Zhu W (2020) Decentralized access control encryption in public blockchain.
In: Zheng Z, Dai H-N, Tang M, Chen X (eds) Blockchain and trustworthy systems. Springer
Singapore, pp 240–257. https://doi.org/10.1007/978-981-15-2777-7_20
59. Di Francesco Maesa D, Mori P (2020) Blockchain 3.0 applications survey. J Parallel Distrib
Comput 138:99–114. https://doi.org/10.1016/j.jpdc.2019.12.019
60. Zou W, Kumaraswamy M, Chung J, Wong J (2014) Identifying the critical success factors for
relationship management in PPP projects. Int J Proj Manag 32:265–274. https://doi.org/10.
1016/j.ijproman.2013.05.004
344 S. Hwang et al.

61. Kim J-S, Shin N (2019) The impact of blockchain technology application on supply chain
partnership and performance. Sustainability 11:6181. https://doi.org/10.3390/su11216181
62. López-Pintado O, Dumas M, García-Bañuelos L, Weber I (2022) Controlled flexibility in
blockchain-based collaborative business processes. Inf Syst 104:101622. https://doi.org/10.
1016/j.is.2020.101622
63. Das T, Teng BS (1998) Between trust and control: developing confidence in partner cooperation
in alliances. Acad Manag Rev 23:491–512. https://doi.org/10.5465/AMR.1998.926623
64. Gallivan MJ (2001) Striking a balance between trust and control in a virtual organization: a
content analysis of open source software case studies. Inf Syst J 11:277–304. https://doi.org/
10.1046/j.1365-2575.2001.00108.x
65. Addae-Boateng S, Wen X, Brew Y (2015) Contractual governance, relational governance, and
firm performance: the case of Chinese and Ghanaian and family firms. Am J Ind Bus Manag
05:288–310. https://doi.org/10.4236/ajibm.2015.55031
Blockchain Technologies and Building/
City Information Modelling
A Conceptual Framework to Facilitate Urban Asset
Management Using Distributed Ledger Technology

Oluwatoyin O. Lawal and Nawari O. Nawari

Abstract The advent of smart technologies and their application in the Architec-
tural, Engineering, and Construction (AEC) domain has been extensively docu-
mented; however, the development of Blockchain technologies (BCT) is a concept
that is relatively new to many fields. In design and construction, there is a plethora of
literature but very few practical use cases. Following an exploration of substantive
scholarly contributions on relationships between Building Information Modelling
(BIM) and blockchain technologies and between Building Information Modelling
and City Information Modelling (CIM), this research investigates the potential of
a tripartite interface between BCT, BIM, and CIM using Distributed Ledger Tech-
nology (DLT). The study proposes a conceptual framework to facilitate geo-specific
value exchanges and asset management within the urban space akin to the exchange
of cryptocurrency within the financial space. This framework is validated with simu-
lated data. With the use of a permissioned blockchain, the suggested framework
offers streamlined management of physical assets by leveraging the existence of
digital models and representations of these assets. Moreover, the paper concludes
with a discussion of the benefits and limitations of the research.

Keywords Blockchain technologies · Distributed ledger technologies · BIM ·

CIM · AEC · Asset management

O. O. Lawal (B) · N. O. Nawari

University of Florida, Gainesville, FL, USA
e-mail: o.lawal@ufl.edu
N. O. Nawari
e-mail: nnawari@ufl.edu

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 345
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_24
346 O. O. Lawal and N. O. Nawari

1 Introduction

Cities are the largest, most complex, and most dynamic man-made systems [1]. Para-
doxically, their large size and rapid growth testify to their superiority as a technique
for exploiting the earth, yet they provide a poor local environment for man [2]. The
automation of numerous industries, through the exploration of artificial intelligence,
blockchain, robotics, Internet of Things is at the core of industry 4.0, but the use
of blockchain in architectural design and construction has seen an abundance of
research proposing a conceptual framework with very little use cases on the actual
implementations [3].
The cybercity has been conceived as a speculative idea of immaterial electronic
imagery of the city, which exists in parallel with reality and its potential to affect
a material existence [4]. As with many urban networks of the previous era, such as
cybernetics, ubiquitous computing, Internet of Things, blockchain technology (BCT)
promises to be yet another disruptive innovation in urban studies. Since the success of
bitcoin, there has been an increased focus on studying the application of blockchain
in a broad range of contexts, leveraging its decentralized and distributive attributes
to address pre-existing issues [5]. This distributed setup, the driving force behind
Bitcoin and other cryptocurrencies, is now being piloted and implemented across
many other domains [6].
This paper proposes a conceptual framework for improving urban management
and governance. Ultimately, the trust mechanism is explored with respect to the
acquisition, ownership, control, and management of urban features in a decen-
tralized space. A consensus is developed through Building Information Modelling
(BIM)—City Information Modelling (CIM)—blockchain integration in urban devel-
opment and administration. The interoperability of BCT and BIM at the city scale is
attained through Geographic Information System (GIS) by deploying decentralized,
immutable distributed ledgers throughout the lifecycle of all urban assets within the
blockchain. This guarantees holistic and auditable management of assets.
The paper begins with a systematic literature review around thematic areas such
as blockchain data structure, BCT, BIM and CIM integration, value exchanges within
the urban environment, and the future of blockchain technology in the AEC industry.
This is followed by an illustration of the proposed framework and its mode of oper-
ation with respect to its participants. The study closes with a real-life use case in
urban blockchain to derive conclusions.
The study investigates the efficacy of BCT in facilitating the management and
exchange of value and valuable urban assets. The applications and potentials of
blockchain as it pertains to the BIM workflow is evident [7]; however, the applica-
tion of CIM and the Distributed Ledger Technology (DLT) of blockchain towards
improved asset exchanges in cities is yet to be explored. Advancement in CIM as the
integration of models of varying semantic details for urban components, yet having a
semantically rich model already gravitates towards a more distributed system of urban
Blockchain Technologies and Building/City Information Modelling ... 347

information management. Thus, this paper proposes a framework for a blockchain-

based CIM that leverages the immutability of historical data and the interaction data
linking these components to facilitate effective urban asset management.

1.1 Research Methodology

Relevant literature was selected for review of scholarly work to address the following
thematic areas; Blockchain Technology, Building Information Modelling, City Infor-
mation Modelling, Urban Assets Management, and Value Exchanges. This paper
begins by defining the underlying concepts of blockchain, such as DLT and its data
structure, consensus mechanisms, BIM, and CIM. A system is proposed for effective
value exchange within the urban space by first discussing the current challenges of
the traditional approach to urban asset management. The framework development
is in two parts; first, the study expatiates on the mode of integration of BIM data
of building models with CityGML data of geographic information systems to create
a City Information Model. Secondly, a suitable distributed ledger is introduced to
record activities within the digital assets in the CIM.

2 Literature Review

2.1 Blockchain—Definition and Data Structure

Blockchain gained its current popularity after the creation of Bitcoin in January
2009; however, research into its application is relatively recent. Bibliometric studies
by Darabseh & Martins (2020) showed the most recent scholarly work in blockchain
research in construction dating back to 2017 [8]. Blockchain is a shared ledger
containing records of transactions in a given network that is locked cryptographically.
The decentralized nature of the network database and the interconnection of one block
to another makes it immutable. Many scholars define blockchain as a decentralized
ledger system; however, Zheng et al. (2019) classified blockchains based on three
different degrees of decentralization, namely; private, consortium or permissioned
and public blockchains. The difference in these types lies in the consensus approach.
The degree of decentralization increases from private to consortium blockchains to
public blockchains [9].
Decentralization in the blockchain is in data storage and recording of data. It is not
a homogenous technology but a network of technology, thereby creating a new form
of distributed record keeping within a network. As Sundararajan (2016) highlighted
in The Sharing Economy, and in the context of this research, a distributed platform’s
first critical infrastructure element is the shared ledger. The second is the distributed
348 O. O. Lawal and N. O. Nawari

hash table, to index what is available on the network, while the third is the contract
which acts as the mediator between two parties [10].
The fundamental unit of the blockchain is the “block.” The block refers to a
record of a category of transactions or a sequence of transactions hash-chained with
the previous one in a blockchain [11]. Nodes are outlets on the network connected
to hardware which may be a computer or any electronic device. Specific hardware is
required in order to connect to a node.
Consensus in a peer-to-peer network requires no centralized server or authority
to validate incoming blocks. Communication between nodes is not controlled by a
single node, and the data security is assured by various consensus mechanisms such
as Proof of Authority, Proof of Space, Proof of Work or Proof of Stake [8]. The
hash function maintains the security, integrity and immutability of the blockchain by
linking blocks together in chronological order [12]. The hash function is a unique
identifier of a data block and a mathematical function that converts text input into an
encrypted code, connecting each block to its previous block [13].

2.2 Blockchain, BIM and CIM

Building Information Modelling is the process of generation and management of the

physical representation of building components. It includes transactions at the data,
information, and knowledge semantic levels [14]. The City Information Model is
often simply referred to as a digital representation of urban components and infor-
mation or the urban equivalent of a Building Information Model. This definition may
suggest that CIM is an enlarged version of BIM against CIM as a digital proliferation
of BIM models and other models that make up the city fabric.
One of the ways to derive a city information model with semantically rich infor-
mation is the integration of BIM with GIS as a suitable method to organize city
information [15], as a suggestive tool for integrating the numerous fragmentary
urban definitions, discourses, and ideological positions [16]. Similarly, combining
a digital platform with a spatial database to provide storage, interoperability of file
formats, access to information, centralization of data, and an urban environment data
model which provides the level of detail of the urban components can form a spatial
data model. This spatial data model is described as the backbone of the CIM [17].
BIM and CIM have to be unified for geolocation. Yu & Liu (2016) proposed inte-
grating BIM and GIS to analyze the 3DGIS valuation model in effectively computing
property values [18]. The blocks and territories in cities are always referenced by rela-
tional geography, therefore CIM is conceived as an evolution of GIS, with emphasis
on relationships between “blocks” and “territories” [16].
Blockchain Technologies and Building/City Information Modelling ... 349

2.3 Value Exchange in the Urban Context

The land is a ubiquitous urban asset, therefore, administration and management of

land and all above and within it is crucial towards achieving urban smartness. This
follows the argument that future cities are only smart when investments in human
and social capital coupled with urban infrastructure fuel innovative sustainable
development [19].
Faniran & Olaniyan (2016) suggested that the adoption of GIS-based applications
in land administration by sub-national governments across the country can ultimately
reduce the number of slum dwellers [20]. The adoption of BCT by the public sector
is no longer new, with land administration being one of the earliest adoptions of
the BCT and DLT [21]. BCT was first used to solve the problem of irregularities
and counterfeit land registration in Honduras between 2016 and 2017. Since then,
other areas where blockchain-based solutions have been deployed in land registration
include Brazil, Ukraine, Sweden, and India [22].

2.4 The Future of Blockchain Technology in the AEC

Industry

As of 2019, research on blockchain applications in environment, construction, and

smart cities was significantly lower than studies of its application in other subtopics
such as ownership, security and privacy, and supply chain management [23]. This
is in resonance with Sarah Underwood (2016) regarding blockchain’s data privacy
issue, as reported by Microsoft. [24]. Inter-blockchain communications need to be
enabled and improved upon. A few industrial solutions to BCT interoperability exist
within cryptocurrency, however, interoperability amongst other major blockchain
platforms by connecting heterogenous blockchains will be a massive paradigm shift
and is key to unlocking its diverse applications [25].

3 Proposed Frameworks

This framework streamlines increasingly complex blockchain concepts for specific

applications in urban studies. Three interaction components across two distinct inter-
action levels are identified. The interaction components are the BIM, GIS, and DLT
components. The interaction levels are the blockchain-BIM (bcBIM) building level
framework and the BIM-GIS city level framework.
Two frameworks for the application of blockchain and CIM are presented and
coupled, to ensure a geo-referenced solution at building level. The bcBIM building
level framework focuses on design, construction and lifecycle management processes
and workflow. As Sreckovic et al. (2021) proposed, this framework uses BC and
350 O. O. Lawal and N. O. Nawari

Fig. 1 bcBIM building level and BIM-GIS city level frameworks

SC for traceable documentation and transaction across the design and construction
ecosystem [26]. Based on the hybridized bcBIM model, the city-level framework
digitizes the relationship between buildings and their geographical context (Fig. 1).
Earlier studies have proposed a Common Data Environment (CDE) for secure data
storage of digital assets, interdisciplinary coordination, management, and versioning
of information containers [26–28]. This study proposes a double-layered CDE in
the form of the Building Level Data Environment (BLDE), which is a subset of a
larger City Level Data Environment (CLDE). BLDE is used in the building-level
framework, while CLDE is used in the city-level framework.
The two frameworks use contrasting standards, yet they are not mutually exclu-
sive. Hence, coupling both concepts is required to complete the urban blockchain
loop. The building level and city level frameworks employ a private consortium
blockchain; however, a public blockchain may be more suitable once the adoption
of this framework becomes more widespread. In both concepts, smart contracts are
service solutions deployed to automate agreements and payments when are where
necessary.

3.1 The bcBIM Building Level Framework—The

‘BIMCHAIN’ Approach

As recently as 2021, despite the high volume of scholarly publications within the field
of BIM and blockchain integration, perhaps the most common known commercial
product that integrates BIM and blockchain is BIMchain [29]. The BIMchain soft-
ware is developed by Lucetium SAS in Paris, France. For a holistic blockchain-based
BIM solution, it is essential to understand how the various processes are designed
Blockchain Technologies and Building/City Information Modelling ... 351

Fig. 2 Stakeholders workflow wheel

and configured for DLT and smart contracts (SC) and how they can be adapted for
multiple workflows and procurement methods. The key participants in this frame-
work are the project stakeholders in the form of the client or owner, Project Manager
(PM), Technical Consultants (TC), General Contractor (GC), and Facility Managers
(FM), and the activities include the exchange of digital assets in varying forms.
Figure 2 shows the stakeholder workflow where the client or owner is at the start and
end of the process.
In the Building Level framework, Blockchain and Smart Contracts (SC) are
service-level solutions that ensure the interconnectivity of all activities within the
workflow and allow for transparency and scalability. SC serves as a mechanism for
retrieving details and provenance records of any information within the blockchain.
The workflow (Fig. 3) begins with the client commissioning project consultants.
The paper-based consultancy engagement is stored in the BLDE. The client and PM
further engage other domain-specific consultants. The PM role includes interpreting
the client’s brief, developing and coordinating the complete design package, and
managing other technical consultants and an intermediary between the client and
project team, depending on the conditions of engagement.
The client defines the project requirement and provides all relevant information,
forming the basis of the design. For effective provenance audit of decisions and
outcomes, the basis of design is stored as a paper-based document in the BLDE. All
digital assets are evaluated by the client and PM based on the stored brief and basis
of design. The client gives the approval to proceed to the next stage once the design
is satisfactory. The architect develops the BIM model and uploads it to the BLDE.
Other consultants synchronize with the BLDE-hosted BIM model and develop their
domain-specific design information. The architect is however responsible for periodic
352 O. O. Lawal and N. O. Nawari

Fig. 3 The building level framework

review and audit of the BIM model to ensure data integrity. The architect’s verdict
on any audit or coordination of the BIM model is communicated to the PM, who
then deploys SC to document all changes to the model.
The GC also falls under domain-specific professionals. The exact workflow
between the PM and TC in the pre-construction stage is adopted between the GC and
subcontractors in the construction stage. All professionals update the BIM model
throughout the project lifecycle and all updated information is stored in the BLDE.
The completion of the contract triggers an SC between the client and FM. The FM
also continually updates the BIM model throughout the project lifecycle and stores
all maintenance records in the BLDE. All revised versions of digital assets supersede
previous ones but will always be in reference to the superseded versions. This ensures
transparency, traceability, and immutability of building information.

3.2 The BIM-GIS City Level Framework

The framework proposes a solution that allows peer-to-peer interactions of asset

owners. The city-level framework investigates the relationship between the asset and
its locale. Owners of non-digital assets or digital assets that are not blockchain-based
may also participate at City Level, but the total value of the solution is derived when
digital assets incorporate BC and SC.
The key participants of this framework are the client or owner and the municipal
authorities. The BLDE is now expanded into a City Level Data Environment (CLDE)
by integrating the bcBIM model with GIS data by merging the Industry Foundation
Blockchain Technologies and Building/City Information Modelling ... 353

Fig. 4 The city level framework

Class (IFC) data structure of Revit with CityGML data structure of ArcGIS. First, the
asset owner provides the traditional paper-based title documents, which is reviewed
by the municipal authorities before uploading to the CLDE. This BIM model is
then integrated into CityGML data and hosted with other ownership documents. The
municipality is the administrator of a private consortium urban blockchain. They
register asset owners to the network, using SC to record inputs of new registrants.
The algorithmic workings of the urban blockchain are enumerated in the next chapter
(Fig. 4).

4 Use Case

4.1 Project Overview

In previous sections, the integration of blockchain technology in various workflows

and scenarios has been enumerated. A Build-Operate-Transfer (BOT) procurement
model for a four-storey student apartment building in Orlando, Florida is used for
demonstration. This concept is further expanded in order to deconstruct the workflow
and illustrate the real-life application of BIM, BC, and SC.
The Build-Operate-Transfer type arrangement is a popular option amongst many
procurement routes for its high efficiency considering the risks to be undertaken by
a concessionaire whose responsibility is to design, finance, construct, operate and
transfer infrastructure within stipulated terms of the agreement [30].
354 O. O. Lawal and N. O. Nawari

4.2 Application

Using the above-coupled framework, transferring the concession to a new stake-

holder will trigger a multi-layered series of inquiries that will require data audit
and provenance in a tamper-proof database. The design phase is as described under
the building level framework wherein the client engages the PM and the rest of the
project team, who develop a synchronized BIM model of the apartment building.
After necessary reviews, coordination, and approvals, all activated by SC, the model
is shared with the municipal authorities for planning permit. Upon approval from the
authorities, a smart contract triggers the commencement of tendering process and
the engagement of a GC.
GC signs a contract with the client. The contract clauses and obligations are
codified and self-enforced by the SC, thereby protecting the interest of all parties
involved by automating the next step once conditions are met and as a gatekeeper
when conditions of the contract are breached. All contract information is stored in the
CDLE. The GC’s contractual relationship with the subcontractors is also maintained
with SC. The BIM model will be updated by the contractor for the development
of shop fitting drawings and an as-built model. The PM, however, ensures that the
CLDE is constantly synchronized with the most recent domain-specific models.
Upon completion of the contract, an SC prompts the client to commence opera-
tions through the engagement of FM. The workflow between FM and other vendors
is a replica of the workflow between GC and the subcontractors. Once FM is
engaged, the PM cedes administrative authority over the BIM model to the FM,
who now ensures that the as-built model is constantly synchronized with all main-
tenance records. Peer-to-peer, decentralized asset exchanges within this network are
described in Sects. 3.1, 3.2

5 Conclusion

The research proposes two integrated frameworks, bcBIM building level framework
and BIM-GIS city level framework, seeking to answer the question of how efficient
urban administration and management will be using DLT.
In response to the research question, the suggested BCT has to be designed to be
highly robust to accommodate changes in workflow. Furthermore, the study proposes
reconfiguring traditional procurement workflows. The proposed cohesive integration
of BCT and SC in BIM for asset management requires a highly digitized process and
auditable digitization of the entire workflow in a project lifecycle. The contribution
of the presented research is twofold; First, the paper demonstrates the feasibility of a
BC, BIM, and GIS integration to develop an immutable CIM. Secondly, it proposes
a system to facilitate the peer-to-peer exchange of value as a substitute for traditional
paper-based processes with a high level of third-party involvement.
Blockchain Technologies and Building/City Information Modelling ... 355

Future studies will focus on (1) Virtual Planning Permitting Process where a
Common Data Environment, a blockchain repository, and municipality participation
can allow for a completely automated permit application process. And (2) Auto-
mated Building Regulation and Control, whereby the deployment of Smart Contracts
prevents unauthorized transition to a subsequent phase of work provided conditions
for progress are yet to be fulfilled.

References

1. Schmitt G (2012) A planning environment for the design of future cities. In: Arisona SM,
Aschwanden G, Halatsch J, Wonka P (eds) Digital urban modeling and simulation. Springer,
Berlin, pp 3–16
2. Harris C, Ullman E (1945) The nature of cities. Ann Am Acad Polit Soc Sci 242(1):7–17
3. Dounas T, Lombardi D, Jabi W (2020) Framework for decentralised architectural design BIM
and blockchain integration. Int J Archit Comput.
4. Boyer MC (1996) CyberCities: visual perception in the age of electronic communication.
Princeton Architectural Press
5. Aggarwal S, Chaudhary R, Aujla GS, Kumar N, Choo K-KR, Zomaya A (2019) Blockchain for
smart communities: applications, challenges and opportunities. J Netw Comput Appl 144:13–
48
6. Nofer M, Gomber P, Hinz O, Schiereck D (2017) Blockchain. Bus Inf Syst Eng, 183–187
7. Nawari N, Ravindran S (2019) Blockchain and the built environment; potentials and limitations.
J Build Eng 25
8. Darabseh M, Martins J (2020) Risks and opportunities for reforming construction with
blockchain: a biblio. Civ Eng J.
9. Zheng R, Jiang J, Hao X, Ren W, Xiong F., Ren Y (2019) bcBIM: a blockchain-based big data
model for BIM modification audit and provenance in mobile cloud. Math Probl Eng.
10. Sundararajan A (2016) The sharing economy: the end of employment and the rise of crowd-
based capitalism. The MIT Press, Cambridge, London
11. Bernardini M, Pennino D, Pizzonia M (2019) Blockchains meet distributed hash tables:
decoupling validation from state storage. arXiv preprint arXiv:1904.01935
12. Ali S, Wang G, White B, Cottrell RL (2018) A blockchain-based decentralized data storage
and access framework for pinger. In: 17th IEEE international conference on trust, security
and privacy in computing and communications/12th IEEE international conference on big data
science and engineering (TrustCom/BigDataSE), pp 1303–1308
13. Raharja U, Hidayanto AN, Lutfiani N, Febiani DA, Aini Q (2021) Immutability of distributed
hash model on blockchain node storage. Sci J Inform 8(1):137–143
14. Succar B, Sher W, Aranda-Mena G, Williams T (2007) A proposed framework to investigate
building information modelling through knowledge elicitation and visual models. In: Confer-
ence proceedings of the Australasian universities building education association, Melbourne,
pp 4–5
15. Xu X, Ding L, Luo H, Ma L (2014) From building information modelling to city information
modelling. J Inf Technol Constr
16. Stojanovski T (2013) City information modeling (CIM) and urbanism: blocks, connec-
tions, territories, people and situations. In: Proceedings of the symposium on simulation for
architecture & urban design, pp 1–8
17. Jil G, Almeida J, Duarte JP (2011) The backbone of a city information model (CIM):
implementing a spatial data model for urban design
18. Yu H, Liu Y (2016) Integrating geographic information system and building information model
for real estate valuation. In: Proceedings of the FIG working week
356 O. O. Lawal and N. O. Nawari

19. Sabri S, Rajabifard A, Ho S, Namazi-Rad M-R, Pettit C (2015) Alternative planning and land
administration for future smart cities. IEEE Technol Soci Mag 34(4): 33–73
20. Faniran S, Olaniyan K (2013) From slums to smart cities: addressing slum-dwelling in Nigeria
through e-land administration. In: Proceedings of the 7th international conference on theory
and practice of electronic governance. Association for Computing Machinery, New York, pp
189–189
21. Konashevych O (2021) GoLand registry’ case study: blockchain/DLT adoption in land admin-
istration in Afghanistan. In: DG.O2021: the 22nd annual international conference on digital
government research. Association of Computing Machinery, New York, pp 489–494
22. Mendi A, Cabuk A (2020) Blockchain applications in geographical information systems.
Photogramm Eng Remote Sens 86(1):5–10
23. Zeadally S, Abdo JB (2019) Blockchain: trends and future opportunities. Internet Technol Lett
2(6):e130
24. Underwood S (2016) Blockchain beyond bitcoin. Commun ACM, 15–17
25. Qasse IA, Abu Talib M, Nasir Q (2019) Inter blockchain communication: a survey. In:
Proceedings of the ArabWIC 6th annual international conference research track, pp 1–6
26. Sreckovic M, Sibenik G, Sigalov K, Ye X, Konig M, Reitmayer K (2021) Upkeeping digital
assets during construction using blockchain technology. In: Proceedings of the conference CIB
W78, 2021, Luxembourg, pp 11–15
27. Wang J, Wu P, Wang X, Shou W (2017) The outlook of blockchain technology for construction
engineering management. Front Eng Manag, 67–75
28. Pishdad-Bozorgi P, Yoon JH, Dass N (2020) Blockchain-based information sharing: a new
opportunity for construction supply chains. EPiC Ser Built Environ 1:274–282
29. Kuperberg M, Geipel M (2021) Blockchain and BIM (building information modeling): progress
in academia and industry. arXiv preprint arXiv:2104.00547
30. Wang W, Dai D-S (2010) Research on the concessionaire selection for build-operate-transfer
projects. In: 2010 international conference on management and service science. IEEE, pp 1–4
Exploring Digitisation as a Solution
to the Long-Term Insurance Sector
Efficiency Quagmire

Linda Malifete, Samuel Adekunle, and Clinton Aigbavboa

Abstract The insurance industry is necessary for the global economy, the stability
of the economic system, and its sustainability. However, the industry continues to
face challenges related to customer service, market competition and re-engineering
of processes. The insurance industry is no different to other industries, it is also
affected by the forever changing customer requirements and demands and this fast
growing digitalization. Almost every organization wants to be part of the new revo-
lution because of its ability to bring company improvement in process efficiency,
productivity and performance. Just like industrial revolution did more than a century
ago, it continues to reshape the way humans live and perform their daily duties. As
a result, organizational processes, customer expectations and the adoption of new
channels, products, and services have all undergone significant change, forcing a
reconsideration of business models. Digital technology has been embraced by many
businesses because it has the potential to transform business operations, much as it did
for other industries, and it has the potential to change the insurance industry as well.
The sparse use of digital technology is the main hindrance to the South African insur-
ance sector’s success. The main goal of this paper was to examine the extent in which
digitalization has been explored in the South African long-term insurance industry
compared to other industries. This was accomplished by doing a thorough PRISMA-
based critical analysis of the current literature. Sixty-two accredited journal articles
were assessed to understand the contribution that has been made by digital models
in the previous research and to identify gaps and be able to contribute to the body
of knowledge by introducing effective techniques to close the identified gaps. The
first finding was that digitalization has not been thoroughly explored in the insurance

L. Malifete (B)
Department of Operations and Quality Management, University of Johannesburg, Johannesburg,
South Africa
e-mail: linda.malifete@gmail.com
S. Adekunle · C. Aigbavboa
Cidb Centre of Excellence, University of Johannesburg, Johannesburg, South Africa
e-mail: sasamuel@uj.ac.za
C. Aigbavboa
e-mail: caigbavboa@uj.ac.za

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 357
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_25
358 L. Malifete et al.

industry and there is dearth of research about digitalization in this industry, partic-
ularly in the African continent. Secondly, the reviewed literature revealed that the
application of Internet of Things and big data has the potential to enhance customer
database in insurance business, because these digital tools can help insurers obtain
customer data in real-time. This has been highlighted as a solution to the problems
the long-term insurance sector is currently experiencing.

Keywords Company performance · Digitalization · Insurance · Long-term

insurance

1 Introduction

Companies are working on solutions to keep up to date with new technology to stay
relevant in the business. The term digitalization is increasingly getting recognition
in the current era. This is the latest strategy of manufacturing centered on scientific
notions of Internet of Things (IoT), Internet of Services (IoS), automation, cyber-
physical systems, and robotics [59]. German has a rich digitalization history and is
known worldwide as an international leader in manufacturing equipment sector and
for having the best competitive manufacturing businesses [59]. This is an exciting
time for all industries and an opportunity to be smarter than the competitors, hence it
cannot be overlooked. Digitalization can change the way in which people work, learn,
and think about production or industrial manufacturing and expertise needed in the
job market. It is difficult to overlook the notion of digitalization because young and
older people are affected by this revolution. We live in a world whereby almost every-
thing is digitalized, with the help of the latest technologies and automation of most
occupations has become a neccesity. Digital revolution is an important word associ-
ated with the information revolution. The whole world is going through an immense
transformation because of the digital revolution [49]. Schroeder [60] further adds that
the digital revolution is often referred to as the ability to advance the technology from
equivalent electronic expedients to digital machinery. It is therefore understood as
the conversion from equivalent gesture to a numerical form and it allows the gesture
to be processed by computers. Zhong et al. [61]; Brettel et al. [47] added that a lot of
industries have adopted the digitalization, these include, healthcare, communication,
administration, finance, customer service, entretainment, hospitality, manufacturing,
training and education etc. The initial three revolutions were basically about mech-
anization, electrical energy, and IT. The Fourth wave came about because of the
presentation of the IoT and IoS into the industrial environment [54]. This means that
for the first time, we can network resources, data, items, and individuals to generate
the IoT and Services.
Exploring Digitisation as a Solution to the Long-Term Insurance Sector ... 359

2 Overview of the Insurance Industry

The insurance industry is one of the most fascinating and important sectors. The
insurance business is vital to the security and sustainability of the economic struc-
ture and the worldwide economy [6]. It is known for its rich antiquity that dates back
to the Middle Ages. The insurance sector matured in the nineteenth century, conse-
quently, the life and non-life insurance products were established [57]. According
to Willet [32], insurance is a social mechanism for building up resources to prevent
capital loss due to unforeseen situations. It does this by transferring risks from indi-
viduals to a community of people. In accordance with NWAFOR [16], insurance is a
contract whereby one person, known as the insurer, makes a commitment to compen-
sate another person, known as the insured, for a set amount of money, known as the
premium, which is paid to the insured in the event of a disaster/loss. The insured
person passes the risk of a potential loss to the insurer in exchange for a mone-
tary payment known as the premium as a risk management strategy (NWAFOR,
2018). Governments have indorsed the transaction of insurance products because
they provide protection against financial loses and they promote long term savings.
This is because of the insurance’s ability to provide policyholders with the way of
securing their future well-being and that of their loved ones (NWAFOR, 2018).
All South Africans will be capable of safeguarding themselves from unforeseen
circumstances if they have access to insurance. This is crucial for households with
low incomes, as job losses can lead them to sink further into poverty. They can avoid
falling further and further into poverty by purchasing insurance, which compensates
the loss. Increasing insurance availability can support the nation’s growth strategy.
One of the industries that contributes the most to GDP is financial services. The insur-
ance sector has undergone gradual change, which is not consistent with South African
reality. Building more teams with different backgrounds and skill sets is necessary
because they stimulate business innovation. The insurance industry is searching for
new markets, business strategies, and business models to expand, build customer
relationships, and develop a functional infrastructure. The industry requires sophis-
ticated organizational infrastructure and highly trained human capital. According to
Kazancoglu and Ozkan-Ozen [55] digitalization doesn’t just influence the industrial
organisations, there is an extensive impact on the idea of work, which adjusts the
desires from work personel in all sectors as well, including the insurance industry.
Kazancoglu and Ozkan-Ozen [55] further added that we cannot run away from the
fact that the digital technologies are anticipated to change the occupation profiles of
workers in various ways and it is important to explore all the areas. This means that
the need for labor in an old-fashioned manner that needed man power falls gradu-
ally as a result of latest technology machineries which are able to interconnect and
communicate between one another to control themselves [55]. For many years, the
insurance industry has been using old fashioned technologies such as web applica-
tions, mobile, online insurance portals, and call centers. However, there are various
insurers that are already using innovative tools such as big data, IoT, advanced sensors
with the aim to retain more customers, improve profitability of services and optimize
360 L. Malifete et al.

Fig. 1 Insurance Value

Chain (Author’s
compilation)

pricing. The new technologies have a potential to help insurers to make better risk
calculations using modern predictive techniques to project customers that are high
risk. We now live in a digital era where people’s data can be easily retrieved from
various platforms. This means all activities done by customers can become data.
This therefore brings new technology called Internet of Things (IoT), because of its
ability to connect database from different types of objects that can all be connected to
the internet. This type of database is valuable for any kind of business and if utilized
correctly, can give a competitive advantage.
The current insurance value chain includes the below six stages moving from
product development to asset management as shown on the below diagram labelled
as Fig. 1.
The studies done by Gatteschi et al. [50] suggested incorporation of Blockchain
technology in the above value chain with the aim to provide innovative user product
based on a usage model, improved analysis of the customer activities pattern, usage
of machine learning techniques like deep learning to evaluate information kept on
Blockchain. Because customers want to oversee the management of the insurance
policies, therefore insurers need to offer products that are more transparent. For this
reason, it is thus vital for insurers to have an innovative value chain. Previously,
customers would communicate more with brokers and sales advisors, but as the
digitalization is gradually being adopted by companies, there has been a decrease in
interaction between customers and insurance advisors [50]. However, the adoption
of the new technology requires building of new ICT systems, making alterations to
business model, changing the business culture (Gatteschi et al.).
By preventing losses for consumers and governments, insurance may play a crucial
part in creating a sustainable economy. In South Africa’s financial economy, the
insurance industry is a crucial pillar. The largest insurance market in Africa is in
South Africa. The insurance industry can be crucial in addressing some of the prob-
lems South Africa is currently experiencing, including the consequences of climate
Exploring Digitisation as a Solution to the Long-Term Insurance Sector ... 361

change, COVID-19, the current state of economic uncertainty, a prospective reces-

sion, a credit rating reduction that is rapidly approaching, and more. The insurance
sector is a critical one in any economy. Its services are advantageous to individuals
and business entities, thus it caters to a wide range of industries, including construc-
tion, health, retail, manufacturing, agriculture, and so forth. However, the industry
in South Africa has faced numerous difficulties, including disputes, legal matters,
revenue losses, and customer dissatisfaction, among others. This has been attributed
to many causes, among which are perceived unfair premium calculation, and data
management challenges, among others.
This study looked at how emerging technologies could be adopted in the long-
term insurance industry to address these issues. The study also looked at how these
emerging technologies have improved customer service, performance, and produc-
tivity in other sectors. Several technologies were investigated, and their potential
for resolving issues in the insurance industry was carefully examined. This study
is important because it has relevance to the current industrial revolution and assists
stakeholders in the insurance business in being more productive, efficient, and digi-
tally aligned. In addition, it provides an insight into ways to have a digital matured
long term insurance sector in South Africa.
Assuming that an insurer is acting solely in his or her own self-interest when
it comes to refusing to pay up on a disputed claim is not sufficient. The insur-
ance industry is one of the industries that are strictly regulated, there aren’t many
other sectors that are this strictly regulated. According to Wang et al. [31] insurance
companies still use statistical factors like age, gender, and profession to determine a
customer’s premium and the likelihood of unforeseen events like accidents, illnesses,
deaths, and other situations, even though they have specialized departments like actu-
arial. Furthermore, Wang et al. [31] noted that it doesn’t seem like this calculation is
fair for all parties. This is because a client who lives a healthy lifestyle may be required
to pay the same premium as a client who lives an unhealthy lifestyle just because
they share the same age, gender, or profession. Insurance companies may experi-
ence financial hardship if medical issues are not disclosed. The study has looked
more closely at customer complaints that were filed with the long-term insurance
Ombudsman (OLTI), specifically cases where insurance companies rejected claims
because the customers failed to disclose pre-existing medical conditions. The graph
below shows the number of long-term insurance cases that the Ombudsman marked
in the complainants’ favor from 2017 to 2021 (Fig. 2).
The above graph shows that there is a significant disagreement between customers
and insurers throughout the claim stage because of the policyholders’ failure to
provide significant data. This is because insurers solely rely on the information
provided by customers during the insurance application stage, even when this
information is sometimes misleading. The administration and engagement of the
ombudsman in settling disputes between dissatisfied clients and insurers is very
expensive, because the insurers are required to pay more than R4000 for each standard
case submitted to the Ombudsman.
362 L. Malifete et al.

Fig. 2 Cases ruled in favor

of complainants at the W/P based on non-disclosure
ombudsman (OLTI Annual
Reports 2017–2021) 30% 28%
24% 25%
25% 22% 22% 22%
19% 20%
20% 17% 16%
15%

10%

5%

0%
2021 2020 2019 2018 2017

Life Disability

Table 1 Ombudsman costs (Ombudsman Annual Reports: 2017–2021)

Ombudsman costs
Year cost per standard case Compensation Total expenses Recovered value for
awarded complainants
2021 R4,406 R948 592 R33.634 m R200.7 m
2020 R4,387 R817 970 R30.761 m R177.9 m
2019 R 4,086 R874 286 R29.49 m R200.4 m
2018 R 3,629 R632 737 26.04 m R185.8 m
2017 R 3,707 R531 429 R24 406 m R193.3 m

The costs of each case submitted to the Ombudsman, total client compensation
paid, total insurer expenses paid, and total amounts recovered for complainants are
all shown in the table below (Table 1):

3 General Overview Of Digitalization

Earlier studies have been conducted to help organizations realize and comprehend
the entire potential of artificial intelligence, the findings showed that all examined
industries could benefit significantly from artificial intelligence [22]. The value that
artificial intelligence adds to industrial organizations has been previously measured
and the results have revealed that most industrial organizations are undergoing a
Exploring Digitisation as a Solution to the Long-Term Insurance Sector ... 363

digital transformation with the goal of enhancing the usage of computers and automa-
tion through intelligent, autonomous systems that are driven by data and machine
learning. This transformation has been broadly embraced in industries thanks to the
introduction of digital machines, which have helped sectors like construction tran-
sition to a technology-driven industry and keep up with other sectors [46]. In the
construction industry the potential for AI methods like neural networks to ensure
accurate cost estimating is very high. AI technology has been used to estimate the
expenses related to the construction sports fields and the earlier studies proposed a
neural network-based approach for calculating construction expenses [53]. IoT-based
algorithms have been developed by several authors to improve the rate of produc-
tivity and performance. IoT algorithms were found to play a huge role to detect
when temperatures in work environments are too high and signal an emergency [26].
IoT was also explored by Raiyn [25] when he proposed a technique to compute
trip time by first determining the fastest route based on previous data. The research
showed that this strategy supports independence and proposed that it might be used
in Internet of Things-compatible smart cities. According to research, IoT is crucial
to future activity prediction. Previous research has provided IoT based models to
assess IoT cyber risk and estimate IoT cyber risk in the future [24]. A trustworthy
model that can identify and avoid cybersecurity threats and weaknesses has been
previously recommended with the aim of predicting future activity occurrence times
using an IoT-based sensor model [13] and [30]. Without any human involvement,
this approach may computerize the procedures for sharing crucial real-time data
across machines. Because this model gets rid of operational management tools while
enhancing security on the Internet of Things environment. In the Agriculture environ-
ment, IoT based models have been developed track drought, monitor soil moisture,
humidity, and temperature [9] and [10]. The models also provide a 10-day forecast
to assist farmers in successfully planning their farming activities. This model was
developed in the hope that it will help the government address the problems caused
by climate change, which have long had an impact on the economy and society. The
outcomes of the IoT-based system that was created showed excellent dependability
and efficacy. Other industries are faced with difficulties of managing the database of
a digital twin product properly and securely in terms of data storage, data sharing,
and access because there are many competitors involved in the product lifecycle.
A Blockchain-based platform to manage digital twins of products was created to
address this challenge [51]. In the health sector, digitalization has been thoroughly
investigated to enhance the performance of the health sector, as seen in studies by
Tanwar et al. [28], and Murugan et al. [14]. In their studies, they suggested using
the blockchain technology to share healthcare data as a solution to the issue with
patient access to hospital and medical records. Technology has proven to the health
sector that it would help by gathering data for further insurance claims and research
purposes. Authors like Jamal et al. (2020) have employed artificial intelligence in the
medical field to predict drug resistance in particular genes. AI approaches have been
recommended to lessen the workload that is continually growing in the healthcare
industry, they allow doctors to offer patients a wide range of personalized treatment
364 L. Malifete et al.

and detect low blood sugar levels in healthy individuals [45, 52, 58]. Artificial Intel-
ligence frameworks to predict the patients’ overall survival time and the progression
of a virus have been suggested in the medical unit [48]. In testing the efficiency of the
suggested models, they provided the machine with data such as age, historical back-
ground, to produce the final prediction result [56]. The experimental results show that
this AI framework beats standard methods, achieving a reliability of 90.66%. The
modern technology tools have been extensively explored in the healthcare industry
and it has been validated by several authors as a solution to many of their problems
and a great way to improve their services.

3.1 The Status Quo of Digitalization in the Insurance

Industry

The term digitalization is not new, and it has been broadly defined in a variety of ways
by different scholars. In their study, Eling and Lehmann [6] defined digitalization as
the integration of the analog and digital worlds with modern technologies that can
enhance customer interaction, data accessibility, and operational processes. In the
context of the insurance sector, Eling and Lehmann [6] analyzed various definitions
of digitalization from various authors. Digitalization was described by Łyskawa et al.
[44] as the adoption of Information and Communication Technologies (ICT) in the
insurance sector to lower premium costs and increase profits. Digitization in the
insurance sector will offer an opportunity for a meaningful innovation.

3.2 Method

The next section analyzes the digital technology models that were employed in the
earlier journal papers through a systematic review of the literature. This review’s
objective was to compare the insurance business to other industries in terms of
how the use of digital technology has been investigated. To understand how prior
academics interpreted digitalization, existing studies were reviewed. Due to its
capacity to reduce bias associated with the article selection process, systematic litera-
ture reviews are widely employed in all fields [21, 29]. PRISMA (Preferred Reporting
Items for Systematic Reviews and Meta-Analyses) was used in this paper since it is
an evidence-based collection of articles designed to help researchers report a wide
variety of systematic reviews and meta-analyses [21] (Fig. 3).
Using Scopus database, the study first looked for journal articles using the
following keywords: Insurance 4.0, digitalization, big data, industry 4.0, artificial
intelligence, internet of things, and blockchain. Thousands of publications were
produced by the search, which was limited to only complete peer-reviewed academic
articles. The papers were then filtered based on subject, language, publication year,
Exploring Digitisation as a Solution to the Long-Term Insurance Sector ... 365

Fig. 3 Data retrieval process using a Prisma Approach (Author’s compilation)

and abstract to select the most relevant studies. More elimination was done by
checking the most cited papers. All authors’ self-citations were not included in
the study. According to the graph below, marked Fig. 4, the fields of engineering,
computer science, business management, and transdisciplinary studies were those
where the most relevant research was found.
The next chart shows how various nations have examined and embraced the
evaluated literature on the adoption of digitalization (Fig. 5).

Fig. 4 Documents by
subject area (Author’s Documents by subject area
compilation) 50.00% Enginnering,
43.90%
45.00% Computer
40.00% Science, 37.70%
35.00%
30.00%
Business
25.00% Management,
20.00% 17.20%
15.00%
10.00%
Multidisciplinar
5.00% y, 1.10%
0.00%
Multidisciplinary Enginnering Business Computer Science
Management

Multidisciplinary Enginnering Business Management Computer Science

366 L. Malifete et al.

Fig. 5 Illustration of the

countries and sectors from Illustration of the countries and sectors
the literature (Author’s from the literature
compilation)
1
2
2
UK 1
1
SWEDEN 1
1
1 2
SINGAPORE 1
1
1
LITHUANIA 1
1
1 3
1
IRAN 1
1
1
1 2 3 5
ICELAND 1
1
2 3
ENGLAND 1
1 2
1 3 4
AUSTRALIA 1
1
0 1 2 3 4 5 6

Other Manufacturing Health Insurance

The illustration in the previous figure shows how digitalization has been investi-
gated in various countries. India and China are the top two countries in the adoption of
digital techniques, according to the 62 research produced from the Scopus database.
Despite the fact that these two nations have published numerous articles on digi-
talization, their research focuses more on other sectors, such as the manufacturing,
health, and other sectors. Only 21 percent of the 62 publications examined focused on
the insurance industry, and none of these studies were from the continent of Africa.
This demonstrates the need for scholars to investigate digitalization in Africa.

3.3 The Adoption of Digitalization

The selected papers are discussed in detail in the following section, highlighting
the proposed technological models, the explored digitalization tools, and the impact
they are making and will make in the insurance industry, the potential to improve
productivity, performance, and customer service in the insurance industry.

4 Discussion

(a) Fraud in the insurance industry

Fraud is experienced by all types of insurance including the long-term insurance.
The fight against fraud has been slowly getting popular to reduce insurance rates
[34]. Insurance fraud is an issue that affects the entire world, not just the continent
of Africa [8]. According to Viaene and Dedene [34], fraud in the insurance industry
has existed since the establishment of the industry. The study further highlights that
in addition to jeopardizing the insurer’s ability to generate a profit, insurance fraud
and, abuse of insurance may have a severe negative impact on economic frameworks
Exploring Digitisation as a Solution to the Long-Term Insurance Sector ... 367

as well as its value chain [34]. In the insurance industry, fraud is defined in the
study done by Goleiji and Tarokh [35] as: setting up a false claim with the intent
to intentionally mislead; reporting a loss or damage to be greater than it is, and/or
using any other techniques to obtain funds above what the insured party is legally
entitled to receive. Rawashdeh and Singlawi [36] defines it as any action taken with
the purpose to mislead the insurance system to attain a misleading outcome. This
could happen if a claimant tries to receive benefits to which they are not otherwise
entitled. Insurance fraud is a significant issue for the insurance industry, a serious
crime that weakens the entire system and affects insurers, customers, businesses,
society, and the economy [36].
To reduce fraud in the insurance industry, studies have been conducted by different
authors like Hargreaves and Singhania [2], Dhieb et al. [4], Bohnert et al. [8], and
Łyskawa et al. [44]. These studies were conducted to develop a safe and digital
insurance framework that would be able to reduce human contact, protect financial
loss for the insurance processes, notify and inform about customers with high risk,
and identify fraudulent claims. The findings show that digitalization can increase
the accuracy of risk assessment [4]. The cost of manually managing fraud in the
insurance sector is very high, data analytics can be used in the insurance industry to
assist insurers in detecting fraud [8]. A powerful method for being more proactive
in the battle against fraud is data analytics, which may be used to spot transactions
that point to fraudulent activity or a higher risk of fraud. Studies have asserted that
the insurance sector can detect fraudulent claims with accuracy and that this method
has the potential to reduce costs for insurers and increase their profitability [4] and
[8]. To understand the development of financial technology through big data, Liu
et al. [12], conducted a study to enhance big data’s usage in financial technology.
By combining big data with space-time data, the industry can reduce costs, manage
risks, and use both internal resources and those generated from a variety of products.
Big data, according to Liu et al. [12] is crucial for the insurance industry’s product
development, advice, pricing, marketing, and fraud detection.
(b) Real time data in the insurance industry
According to Sagiroglu and Sinanc [42] Big Data refers to highly bigdata sets that
may be analyzed to reveal patterns and trends. This indicates that there is a variety
of information accessible to calculate risk, allowing for more accurate and detailed
pricing as well as speedy and risk-specific underwriting. For insurance companies,
it is a resource that, if used appropriately, can provide the company a competitive
advantage [31]. However, Liu et al. [43] contend that the insurance business lacks
sufficient sample data, real-time data, and internal data, which is a problem that must
be addressed. The insurance sector is one of the many developing sectors that heavily
rely on data to regulate some aspects in their business. The adoption of digital models
by the insurance industry would have a significant positive impact on the economy
of South Africa because the country is home to many of the largest international
insurance companies, like Sanlam and Old Mutual. To address the issue that was
raised by Liu et al. [43] about the lack of sufficient sample data, real-time data,
and internal data in the insurance industry, Wang et al. [31] carried out research
368 L. Malifete et al.

to improve the availability of client data in the insurance sector. Wang et al. [31]
expanded on the claim made by Liu et al. [43], that the insurance sector depends more
on historical customer data to decide key operational elements, including product
design and pricing. According to Wang et al. [31], insurance businesses can collect
accurate and reliable client data in real time with the help of advanced business
tools. By using big data, the insurance companies can analyze the risk posed by their
customers. The premiums can be priced fairly and accurately when an insurance
provider is aware of the risk posed by the customer. After discovering this challenge,
Wang et al. [31], recommended that the Internet of Things and Big Data will help
the insurance industry to understand their customer’s risk up front because these
technological tools are able to gather information on the activities and health of the
customer in real-time. This research will help improve the transparency and fairness
of the insurance industry. This issue has also been a challenge in the short term-
insurance industry. The inconsistent and unreasonable premium rates have been a
long-standing complaint of customers. Machine learning and telematics have also
been suggested to eliminate unfair premium pricing in the short-term insurance sector
[27]. The models that are currently used to determine premiums in the field of auto
insurance were mentioned as being unappealing to customers. Just like Wang et al.
[31] proposed the use of Internet of Things and Big Data in the long-term insurance for
accurate and fair pricing, Reddy and Premamayudu [27] suggested the combination
of Blockchain technology with Deep learning, Bayesian network and fuzzy logic
techniques to improve the calculations of the premiums, mostly based on the user’s
driving style. As the premiums would depend on the customer’s driving style, Reddy
and Premamayudu [27] evaluated the model’s effectiveness and believed that it would
offer fairness to all auto insurance customers.
(c) Automated underwriting
Insurance underwriting is ready for automation. It entails obtaining and analyzing
data in organized, unstructured, or mixed formats from a diverse range of sources to
assess the risks. It is an extremely drawn-out and lengthy process that, on average,
takes 3 to 4 weeks to finish in the life insurance industry [37]. A health examination
is required for applicants, and the insurance company requests any applicable docu-
mentation. It is time-consuming and error-prone to manually gather medical records
[38]. The underwriter then carefully assesses the profile to establish the risk involved
and evaluate if the application may move forward with additional processing. Then,
premiums are determined [38]. The study also shows that approving wrong appli-
cants for insurance coverage happens when clients do not provide complete and
honest information, leading to insurers approving high-risk scenarios. According to
Biddle et al. [39] all parties involved can benefit from automation of the underwriting
process, which can help in a variety of ways. It is possible to speed up the underwriting
process drastically, decrease instances of human error, and eliminate any confusion or
knowledge gaps among the underwriters. Research on the importance of risk assess-
ment in the long-term insurance industry were conducted by Biddle et al. [39] and
Boodhun and Jayabalan [3] as the topic of risk prediction becomes more important
Exploring Digitisation as a Solution to the Long-Term Insurance Sector ... 369

and a center of attention. Biddle et al. [39] and Boodhun and Jayabalan [3] empha-
sized the importance of the underwriting process because underwriting is an essential
stage in the insurance industry because it helps to make informed decisions about
the acceptance of customers and pricing of insurance products. The development of
data analytics has made it simple for insurers to automate the underwriting process.
Their research sought to provide recommendations for using predictive analytics to
enhance risk assessment in the insurance industry. To predict the individuals’ level
of risk, previous research used ML algorithms. Results showed that this method is
far superior to the extensive, complicated actuarial formulas that were previously
employed for risk assessment, and it will be able to benefit the insurance business.
According to Liu et al. [12] with data analytical solutions, insurers will be able to
calculate the risk assessment faster and get much better results.
(d) Storage of confidential information
Data is essential to the insurance industry because it enables the industry to assess
risks, charge for them, and provide the necessary protection to their clients. Data is
essential to remaining competitive because insurance businesses must gather huge
amounts of information in our growing, data-rich world to maximize performance,
mitigate risk, and satisfy consumers’ increasing expectations [33]. Blockchain-based
insurance storage system are needed in the insurance industry to give customers a
high level of trust due to Blockchain’s strong resilience to manipulation and its
capacity to keep a virtual record of all data and transactions safe [40] and [33]. Zhou
et al. [40] and Xia et al. [33] contended that because of the decentralized nature
of the blockchain based solutions, users would be able to interact with one another
without the intervention of a third party. According to Zhou et al. [40] and Xia
et al. [33] to meet the insurer’s requirements, the designated servers will be able to
apply cryptography algorithms to the data and produce results. In case of hospital
medical records, insurer will be able to retrieve the proper patient’s spending data if
they are able to gather a certain number of accurate answers [33]. Additionally, the
blockchain-based solution would be able to carry out user functions more quickly
and do public verification [40]. Consumer’s willingness to pay for blockchain tech-
nology and smart contracts have been studied and the findings suggested that using
the qualities of potential clients to build insurance products would be beneficial [15].
It has been highlighted that blockchain is incredibly safe by design and can handle a
lot of data well. As a result, smart contracts created on the Blockchain can automate
all processes and have a self-executing nature Nam [15]. The use of machine learning
(ML) in the insurance sector, particularly has been evaluated because the ongoing
global revolution is having an impact in the insurance sector. The adoption of the
technological tools associated with the fourth industrial wave comes with signifi-
cant changes to company culture, data-driven results, and increased competition, all
of which may lead to higher levels of customer satisfaction [7]. The importance of
creating an AI organization based on a theoretical framework for fusing artificial and
human intelligence, was also explored. It was established that many organizations
are dealing with difficulties brought on by the urge to replace human intelligence
with artificial intelligence. Lichtenthaler [11] concluded that businesses require the
370 L. Malifete et al.

three types of intelligence: meta, artificial, and human. He went on to say that the
idea of a meta-intelligence, which benefits from the reintroduction and recombining
of artificial and human intelligence, facilitates the reconciliation of numerous find-
ings from previous research. Without meta-intelligence, most artificial intelligence
programs will be difficult to sustain, which could have positive outcomes but also
have a chance of falling short of expectations [11]. Therefore, organizations will
require intelligence made up of meta-intelligence, artificial intelligence, and human
intelligence.

5 The Gap in Literature Review

From the papers reviewed, 79% of the studies that were examined had a focus on
different industries, such as health, agriculture, manufacturing. Studies pertaining to
the insurance sector made up just 21% of all studies. It seems that many different
industries have studied technology like blockchain, the internet of things, and artifi-
cial intelligence. The investigations’ findings revealed significant advancements and
a bright future for each of these models. It is evident that research must focus on
examining how the insurance industry is becoming more digital. This is a result of
the dearth of studies looking into modern technologies like the Internet of Things
and Artificial Intelligence that have been done in the insurance sector. The authors
have suggested using IoT and big data to improve client databases in the insur-
ance industry. This is because IoT can assist insurers in gathering real-time client
data. In turn, this will make it easier for insurers to assess client risks, consumer
behavior, and health status. Authors like Albrecher et al. [1], Pousttchi and Gleiss
[23] and Eckert and Osterrieder [5], have made important theoretical contributions
to the body of knowledge in insurance while investigating how much digitalization
is being implemented in the insurance business. All of them emphasize how the
digital age is compelling traditional insurance companies to change their strategies,
operational procedures, and organizational structures to incorporate AI tools into the
insurance value chain. The commonality of these studies is that they offer a thor-
ough evaluation and significance of digital tools in the workplace. The findings of
these research indicate that all industries, including the financial services sectors,
stand a considerably greater possibility of benefiting from the digitalization. But like
many studies carried out in the insurance sector, they review the body of knowl-
edge. To investigate these modern digital technologies that are part of digitalization,
empirical studies in the insurance industry are required. Modern technologies enable
communication between companies and their customers. These technologies enable
companies to collect real-time customer data to identify customer patterns. Because
the insurance sector uses previous customer data to inform business choices like
accepting a customer, calculating premiums, and deciding whether to pay or reject a
claim, this database is more beneficial to such an industry. As a result, big data and
internet of things technologies can help insurers collect real-time client data and to
make informed decisions. Due to the changes brought about by digitization, insurers
Exploring Digitisation as a Solution to the Long-Term Insurance Sector ... 371

must act in the form of opening their minds and choosing their own path to digital
transformation, which would need a significant investment in skills and financial
resources. It is necessary to restructure IT infrastructure, consider contracts with
significant technology manufacturers, reconsider business processes in engineering
labs, and create internal innovation and digitization teams. Think of partnering with
FinTech or InsurTech startups as well as players from other sectors who can give
client data. Adapting their human resource practices, such as creating or hiring a new
class of tech-savvy and data-savvy employees, understanding what it means to act in a
customer-centric way in the digital age, and making sure that the equitable insurance
agenda is considered for underserved client categories are all crucial. The existing
technologies contribute to the industry’s transformation and present opportunities
that could become increasingly significant in the face of competition. Claims costs
and expenses can be reduced by using insurance use cases such as automated claims
processing, effective underwriting with better data use, and more robust fraud detec-
tion. In addition, insurers could improve customer experience management while
also strengthening client retention and satisfaction.

6 Conclusion

This paper presented a broad overview of the insurance industry, its purpose, and
the contribution it plays to human lives. Not only it is good for individuals, but it is
also essential to the economic system’s sustainability, stability, and the worldwide
economy. The insurance industry continues to face challenges such as fraudulent
claims, brand reputation, disputes due to factors like non-disclosure of the existing
health conditions and so on. The financial costs and administration efforts related
to these issues are hindering the insurance industry. The study looked at how the
insurance business may use the digitalization to address these problems. The idea
of digitalization is not new; it has been broadly examined by various researchers in
several ways, and it has already transformed business models in areas like healthcare,
tourism, and other industries. To compare how digital technology models have been
studied in the insurance business to those in other industries, the study conducted a
systematic literature review adopting the PRISMA approach. The study also looked
at prior research’s findings to determine how academics in the past had perceived digi-
talization and majority of the studies focused on other industries, including health,
manufacturing, agriculture, retail, food, etc. The findings of the studies indicated a
substantial advancement and a promising future for all the evaluated digital models.
The examination of the digitalization in the insurance industry must be the focus of
research. This is due to the lack of research done in the insurance industry regarding
the latest technology like the Internet of Things and artificial intelligence, particu-
larly in South Africa. The study proposes enhancing client databases in the insurance
sector with IoT and big data. This is because IoT can help insurers obtain client
data in real-time. Insurance companies will then find it simpler to evaluate client
372 L. Malifete et al.

risks, consumer behavior, and health status. Due to digitization, traditional insur-
ance companies are being compelled to transform their organizational structures,
techniques, and processes to integrate AI tools into the insurance value chain. The
survival and success of the insurers merely depend on the integration and execution
of the new technologies into the business strategies. This is because the main benefit
of the use of the latest technology is the access to real-time policyholder database
and consumer patterns.

References

1. Albrecher H, Bommier A, Filipović D, Koch-Medina P, Loisel S, Schmeiser H (2019)

Insurance: models, digitalization, and data science. Eur Actuar J 9(2):349–360
2. Bohnert A, Fritzsche A, Gregor S (2019) Digital agendas in the insurance industry: the
importance of comprehensive approaches. Geneva Pap Risk Insur Issues Pract 44(1):1–19
3. Boodhun N, Jayabalan M (2018) Risk prediction in life insurance industry using supervised
learning algorithms. Complex Intell Syst 4(2):145–154
4. Dhieb N, Ghazzai H, Besbes H, Massoud Y (2020) A secure AI-driven architecture for auto-
mated insurance systems: fraud detection and risk measurement. IEEE Access 8:58546–58558
5. Eckert C, Osterrieder K (2020) How digitalization affects insurance companies: overview and
use cases of digital technologies. Zeitschrift für die gesamte Versicherungswissenschaft, pp
1–28
6. Eling M, Lehmann M (2017) The impact of digitalization on the insurance value chain and the
insurability of risks
7. Grize YL, Fischer W, Lützelschwab C (2020) Machine learning applications in nonlife
insurance. Appl Stoch Model Bus Ind 36(4):523–537
8. Hargreaves CA, Singhania V (2015) Analytics for insurance fraud detection: an empirical
study. Am J Mob Syst Appl Serv 1(3):223–232
9. Hoang VP, Nguyen MH, Do TQ, Le DN (2020) A long range, energy efficient Internet of
Things based drought monitoring system. Int J Electr Comput Eng 2088–8708:10
10. Kiani F, Seyyedabbasi A (2018) Wireless sensor network and internet of things in precision
agriculture. Int J Adv Comput Sci Appl
11. Lichtenthaler U (2020) Beyond artificial intelligence: why companies need to go the extra step.
J Bus Strategy
12. Liu R, Gailhofer P, Gensch CO, Köhler A, Wolff F (2019) Impacts of the digital transformation
on the environment and sustainability
13. Minor BD, Doppa JR, Cook DJ (2017) Learning activity predictors from sensor data:
algorithms, evaluation, and applications. IEEE Trans Knowl Data Eng 29(12):2744–2757
14. Murugan A, Chechare T, Muruganantham B, Kumar SG (2020) Healthcare information
exchange using blockchain technology. Int J Electr Comput Eng 10(1):421
15. Nam SO (2018) How much are insurance consumers willing to pay for blockchain and smart
contracts? A contingent valuation study.Sustainability 10(11):4332
16. Nwafor MC (2018) Effect of insurance business on economic growth and development in
Nigeria. Int J Res Manag 6(7)
17. Ombudsman (2017) Ombudsman for long term insurance Annual Report 2016, South Africa.
https://www.ombud.co.za/publications/annual-reports. Accessed 19 Sept 2020
18. Ombudsman (2018) Ombudsman for long term insurance Annual Report 2017, South Africa.
https://www.ombud.co.za/publications/annual-reports. Accessed 19 Sept 2020
19. Ombudsman (2019) Ombudsman for long term insurance Annual Report 2018, South Africa.
https://www.ombud.co.za/publications/annual-reports. Accessed 19 Sept 2020
20. Ombudsman (2020) Ombudsman for long term insurance annual report 2019
Exploring Digitisation as a Solution to the Long-Term Insurance Sector ... 373

21. Parris DL, Peachey JW (2013) A systematic literature review of servant leadership theory in
organizational context. J Bus Ethics 113(3):377–393
22. Plastino E, Purdy M (2018) Game changing value from Artificial Intelligence: eight strategies.
Strategy Leadersh.
23. Pousttchi K, Gleiss A (2019) Surrounded by middlemen-how multi-sided platforms change
the insurance industry. Electron Mark 29(4):609–629
24. Radanliev P, De Roure DC, Nicolescu R, Huth M, Montalvo RM, Cannady S, Burnap P (2018)
Future developments in cyber risk assessment for the internet of things. Comput Ind 102:14–22
25. Raiyn J (2017) Road traffic congestion management based on a search-allocation approach.
Transp Telecommun J 18(1):25–33
26. Rao SVRK, Devi MS, Kishore AR, Kumar P (2018) Wireless sensor network based industrial
automation using Internet of Things (IoT). Int J Adv Trends Comput Sci Eng 7(6):82–86
27. Reddy T, Premamayudu B (2019) Vehicle insurance model using telematics system with
improved machine learning techniques: a survey. Ingénierie des Systèmes d Inf 24(5):507–512
28. Tanwar S, Parekh K, Evans R (2020) Blockchain-based electronic healthcare record system
for healthcare 4.0 applications. J Inf Secur Appl 50:102407
29. Tranfield D, Denyer D, Smart P (2003) Towards a methodology for developing evidence-
informed management knowledge by means of systematic review. Br J Manag 14(3):207–222
30. Vijayakumaran C, Muthusenthil B, Manickavasagam B (2020) A reliable next generation cyber
security architecture for industrial internet of things environment. Int J Electr Comput Eng
2088–8708:10
31. Wang G, Alamas N, Anggraeni M (2019) The use of internet of things and big data to improve
customer data in insurance company. Int J Emerg Trends Eng Res 7(12):756–761
32. Willet A (1901) Economic theory of risk and insurance, 14th edn. The Columbia Univer-
sity Press, London. http://shodhganga.inflibnet.ac.in/bitstream/10603/90414/3/03_%20chap
ter%201.pdf
33. Xia Q, Sifah EB, Smahi A, Amofa S, Zhang X (2017) BBDS: blockchain-based data sharing
for electronic medical records in cloud environments. Information 8(2):44
34. Viaene S, Dedene G (2004) Insurance fraud: issues and challenges. Geneva Pap Risk Insur
Issues Pract 29(2):313–333
35. Goleiji L, Tarokh MJ (2015) Identification of influential features and fraud detection in
the insurance industry using the data mining techniques (case study: automobile’s body
insurance). Majlesi J Multimedia Process 4(3)
36. Al Rawashdeh FM, Al Singlawi O (2016) The existence of fraud indicators in insurance
industry: case of Jordan. Int J Econ Financ Issues 6(6):168–176
37. Wuppermann AC (2017) Private information in life insurance, annuity, and health insurance
markets. Scand J Econ 119(4):855–881
38. Srivastava A, Kumar A, Damle M (2020) Hyperautomation in transforming underwriting
operation in the life insurance industry. PalArch’s J Archaeol Egypt Egyptol 17(6):4928–4944
39. Biddle R, Liu S, Tilocca P, Xu, G (2018, May) Automated underwriting in life insurance:
predictions and optimisation. In: Australasian database conference. Springer, Cham, pp 135–
146
40. Zhou L, Wang L, Sun Y (2018) Mistore: a blockchain-based medical insurance storage system.
J Med Syst 42(8):149
41. Geneva Papers on Risk and Insurance: Issues and Practice, 1–38. https://doi.org/10.1057/s41
288-017-0073-0
42. Sagiroglu S, Sinanc D (2013, May) Big data: a review. In: 2013 international conference on
collaboration technologies and systems (CTS). IEEE, pp 42–47
43. Liu Y, Peng J, Yu Z (2018, August) Big data platform architecture under the background of
financial technology: in the insurance industry as an example. In: Proceedings of the 2018
international conference on big data engineering and technology, pp. 31–35
44. Łyskawa K, K˛edra A, Klapkiv L, Klapkiv J (2019, May) Digitalization in insurance compa-
nies. In: International scientific conference “contemporary issues in business, management and
economics engineering
374 L. Malifete et al.

45. Agarwal Y, Jain M, Sinha S, Dhir S (2020) Delivering high-tech, AI-based health care at Apollo
Hospitals. Glob Bus Organ Excell 39(2): 20–30
46. Berawi MA (2020) Managing artificial intelligence technology for added value. Int J Technol
11(1):1–4
47. Brettel M, Friederichsen N, Keller M, Rosenberg M (2014). How virtualization, decentraliza-
tion and network building change the manufacturing landscape: an Industry 4.0 perspective.
Int J Inf Commun Eng 8(1):37–44
48. Chen DB, Xiao R, Zeng A (2014) Predicting the evolution of spreading on complex networks.
Sci Rep 4(1):1–6
49. Collins A, Halverson R (2018) Rethinking education in the age of technology: the digital
revolution and schooling in America. Teachers College Press
50. Gatteschi V, Lamberti F, Demartini C, Pranteda C, Santamaría V (2018) Blockchain and smart
contracts for insurance: is the technology mature enough? Future Internet 10(2):20
51. Huang S, Wang G, Yan Y, Fang X (2020) Blockchain-based data management for digital twin
of product. J Manuf Syst 54:361–371
52. Lizuka O, Kanavati F, Kato K, Rambeau M, Arihiro K, Tsuneki M (2020) Deep learning
models for histopathological classification of gastric and colonic epithelial tumours. Sci Rep
10(1):1504
53. Juszczyk M, Leśniak A, Zima K (2018) ANN based approach for estimation of construction
costs of sports fields. Complexity (2018)
54. Kagermann H, Helbig J, Hellinger A, Wahlster W (2013) Recommendations for implementing
the strategic initiative industrie 4.0: securing the future of German manufacturing industry;
final report of the industrie 4.0 working group. Forschungsunion
55. Kazancoglu Y, Ozkan-Ozen YD (2018) Analyzing workforce 4.0 in the fourth industrial revolu-
tion and proposing a road map from operations management perspective with fuzzy DEMATEL.
J Enterp Inf Manag
56. Nie D, Lu J, Zhang H, Adeli E, Wang J, Yu Z, Liu L, Wang Q, Wu J, Shen D (2019) Multichannel
3D deep feature learning for survival time prediction of brain tumor patients using multi-modal
neuroimages. Sci Rep 9(1):1103
57. Pearson R (1997) Towards an historical model of services innovation: the case of the insurance
industry, 1700–1914. Econ Hist Rev 50(2):235–256
58. Porumb M, Stranges S, Pescapè A, Pecchia L (2020) Precision medicine and artificial intel-
ligence: a pilot study on deep learning for hypoglycemic events detection based on ECG. Sci
Rep 10(1):170
59. Rüßmann M, Lorenz M, Gerbert P, Waldner M, Justus J, Engel P, Harnisch M (2015) Industry
4.0: the future of productivity and growth in manufacturing industries. Boston Consult Group
9(1):54–89
60. Schroeder W (2016) Germany’s industry 4.0 strategy. Friedrich Ebert Stiftung, London
61. Zhong RY, Xu X, Klotz E, Newman ST (2017) Intelligent manufacturing in the context of
industry 4.0: a review. Engineering 3(5):616–630
 Space Partitions: An Alternative
to Domain Assembly in Geometric
Modeling
A Computational Robust Method
for Spatial Decomposition - Test Case
with Cadastral Data

Enrico Romanschek , Christian Clemen , and Wolgang Huhnt

Abstract This paper presents a case study of a novel spatial decomposition algo-
rithm in the field of Geographic Information Systems (GIS). Real estate cadastral
data of a district, consisting of parcels and buildings, are used as a test data set.
A cadastral map is a set of parcels and buildings, or more generally GIS features.
Each feature is geometrically represented as a polygon. The legally binding nature
of the cadastral map (rights, restrictions and responsibilities on land) requires that
the polygons do not overlap. There must also be no gaps between the parcels.
Based on the test cases, it is shown that the computationally robust space decom-
position presented here with a complete, gapless and overlap-free two-dimensional
topology can be used very well in this domain. With the added benefit that the results
provided are completely error-free and reliable, and spatial queries can be easily
formulated using set operations. The foundations of the algorithm were presented in
previous research papers, and are summarized shortly in this paper.
While a previous paper provided a proof of concept using artificial datasets, this
paper now uses real cadastral data. The fully automated procedure transforms OGC
simple features in a space decomposition model. For validation purposes, all cases are
additionally tested with a geospatial ETL-software, namely the Feature Manipulation
Engine (FME).

Keywords Space decomposition · GIS · DE-9IM · robust geometry

E. Romanschek (B) · C. Clemen

Faculty of Spatial Information, Hochschule für Technik und Wirtschaft Dresden University of
Applied Sciences, 01069 Dresden, Germany
e-mail: enrico.romanschek@htw-dresden.de
C. Clemen
e-mail: christian.clemen@htw-dresden.de
W. Huhnt
Faculty VI Planning Building Environment, Institute of Civil Engineering, Chair in
Bauinformatik, TU Berlin, Technische Universität Berlin, 13353 Berlin, Germany
e-mail: wolfgang.huhnt@tu-berlin.de

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 377
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_26
378 E. Romanschek et al.

1 Introduction

A feature is an abstraction of a real-world phenomena [6]. Geometrically and topo-

logically consistent features are essential in the field of Geographical Information
Systems (GIS) and especially in the real estate cadastre storing legal data. How-
ever, commonly used GIS data structures such as OGC-Simple features1 are purely
geometrical, e.g. polygons. The data structure does not have inherent logical set
operations for checking the topological validity of the feature itself nor the relation
between two features.
In practice OGC-Simple features are validated and used in a wide range of
commercial and open-source GIS software products, libraries and spatial databases
including GEOS library2 , GDAL library3 , PostGIS4 , QGIS5 , ArcGIS6 , Civil/Map
3D7 as well as in the spatial Extract Load and Transform (ETL) software FME 8 .
Since the coordinates of the Simple Features are represented by floating point
numbers, the calculation of spatial relationships is not always unambiguous. Floating
point numbers have a limited precision. Unavoidable rounding errors on floating
point coordinates may lead to contradictions between topological and geometrical
assertions.
An additional weakness of purely feature based approaches is that only known
objects are represented. If there are gaps within a set of polygons, they are not
conceptualized as objects in their own right. This means that gaps may only be
detected indirectly, using workarounds to model the surrounding of a feature.
To tackle these problems with geospatial geometries, a novel approach to space
decomposition was developed [8]. With this approach, it is guaranteed that the
affected area is represented completely, without gaps and no overlaps. This is
achieved by the following principles:
– Implementing a data structure that maps the topology as space decomposition.
– Modeling even empty spaces as objects.
– Exclusive use of integer coordinates for location tests and position calculations.
– Representing intersections only as positions on edges and not by coordinates.
– A special algorithm for reconstructing features in a mesh.
This approach allows unambiguous statements to be made about the spatial valid-
ity and adjacency of GIS features. In addition, once the space partition is created,
the queries for spatial relations are fast and efficient, since they are based on set
comparisons only and do not require any geometric calculations.

1 https://www.ogc.org/standards/sfa.
2 https://libgeos.org.
3 https://gdal.org.
4 https://postgis.net.
5 https://qgis.org.
6 https://www.esri.com.
7 https://autodesk.com.
8 https://www.safe.com.
Robust Spatial Decomposition - Test Case with Cadastral Data 379

The purpose of this paper is to show that this approach has advantages compared
to established methods. Firstly, the results are absolutely reliable and secondly they
can be obtained efficiently and flexibly.
To prove this, small test datasets as well as a German real estate cadastre dataset
is examined with the presented method. The simple features are checked first for
validity and then used for spatial queries. To validate the results of the test, the same
input data were conducted with the well established ETL-software FME.

2 Related Research

The fundamental concepts of the approach used for the presented testbed are based
on [5]. The relevance to GIS the general usability for space decomposition with
geodata was discussed in [8]. Especially the importance of the DE-9IM matrices for
the description of spatial relations was shown in the previous work of the authors.
The practical problem of inconsistencies in geometric computations using floating
point numbers are well described in [7]. In general, floating point numbers have only
a limited number of significant digits, so calculations, such as intersections, may
deliver wrong results. An overview of the problem area on methods of exact geometric
computation with geospatial data is provided in [12]. This research shows, that integer
coordinates and space decomposition are good concepts for robust calculations also
for geodata.
The Dimensionally Extended 9-Intersection Model (DE-9IM) is a well establish
mathematical concept to describe topological relations and is founded on the research
by Clementi and Eggenhofer [3, 4]. A diversity of topological data models has been
published and found to be important for many use cases, e.g. [11, 13]. Current
research extends theses concepts of topological queries [9, 10].
However, to the best knowledge of the authors, there is no extensive research
literature on implementation details, for the particular use case “executing queries
on spatial relations, using set operations” that focuses on robust calculations. A first
description can be found in [2], but there the whole subject is considered in the
context of database systems, mainly focusing on creating spatial indexes for fast
queries on object-relational geo-databases.

3 Data Structure and Algorithm

The objective of the algorithm is to create a spatial decomposition for a given set
of simple features. The resulting decomposition is complete, gap-less and does not
contain any overlaps.
The algorithm reads simple features (Point, Polygon), creates a Delaunay
triangulated mesh with integer numbers, reconstructs the features with HalfEdges
in a space decomposition and eventually creates dimension-based feature sets
380 E. Romanschek et al.

Fig. 1 Datastructure of the

model

(Dim0Set, Dim1Set, Dim2Set), that allow robust DE-9IM set queries on the
given simple features.
The topological data structure for the space decomposition is based on half-edges
(Fig. 1) and consists of the following types:
HalfEdge: Half-edges have a starting point (Orig), a following half-edge (Next),
a previous half-edge (Prev), a twin (the adjacent opposite half-edge Twin)
and the associated face (RefFace). As a special feature, each half-edge also
has an associated edge (RefEdge). The starting point is not a Vertex object
but a relative position on the associated edge. To specify the orientation of the
half-edge, it also contains a flag (InEdgeDirection) which is true if the
half-edge is in the direction of the reference edge.
Edge: Each Edge starts at a Vertex A and ends at Vertex B. To make these edges
unique, the endpoints are sorted with respect to their coordinate values. Moreover,
each edge contains a list of all intersecting edges (Intersections), sorted by
their relative position on the edge. The intersection position is a proper fraction
between . 01 (start point) and . 11 (end point). In addition, each edge contains another
list with associated HalfEdges and a set with Ids of the associated features.
Vertex: The endpoints of the Edges are Vertex objects. These have an associated
2D coordinate represented by Point, a set of outgoing (Edges) and a set of
associated features (Ids).
Face: The Face object contains a reference to a half-edge of the boundary
(RefHalfEdge) and the set of associated feature Ids.
This data structure provides a representation of the simple features geometries,
that is consistent with respect to geometry and topology. To generate this space
decomposition, the following steps are necessary:
Robust Spatial Decomposition - Test Case with Cadastral Data 381

Fig. 2 Essential steps of the algorithm

1. A list of simple features of type Point, LineString, MultiLineString

or Polygon with associated ids, are read from a given data set.
2. All points from the imported features are collected (Fig. 2a). From this, a bound-
ing box is formed and the .scale factor and the translation .xmin and . ymin for
converting the coordinates into integers are determined.
( ) [( ) ( )]
xint x f loat xmin
= scale − (1)
yint y f loat ymin

The .scale converts all given coordinates to integers without rounding. Since
the given geospatial features the have millimeter precision, the resulting scaling
factor(.scale) is 1000.
3. With the scaled integer coordinates a Delaunay triangulation is computed
(Fig. 2b). This mesh forms the basic data structure and guarantees that the model
is gap-less and overlap-free. The outside half-edges along the boundary of the
entire model are assigned to the EXTERIOR facet.
4. Reconstruction of the simple feature edges.

(a) Intersection of the features-edges with the triangulation-edges.

(b) Intersection of the inserted feature-edges, that are inside of triangles (Fig. 2c).
This sub step avoids geometric calculations, by only using the order of inter-
secting edges within a mesh-triangle.
(c) Once all edges inside a triangle are reconstructed, new faces can be created.

5. Assignment of the newly created faces to the original simple features.

6. Clean up of the finished model from unnecessary edges and intersections that
come only from triangulation and do not belong to any feature (Fig. 2d).

After these steps, the space decomposition is complete and will not be changed
in the following topological queries. Each Vertex, Edge or Face contains the
corresponding ids and types (parcel, building, ...) from the original geospatial dataset
and these can be grouped into dimension-based feature sets Dim1Set, Dim2Set
382 E. Romanschek et al.

Table 1 The dimension-based feature sets Dim0Set, Dim1Set and Dim2Set are linked to the
space decomposition Vertex, Halfedge and Face
0-Dimension Vertex 1-Dimension HalfEdge 2-Dimension Face
Interior Dim0Set
Dim1Set Dim1Set
Dim2Set Dim2Set Dim2Set
Boundary Dim1Set.a
Dim2Set Dim2Set
.
a The Boundary 0-Dimension attribute of closed Dim1Set objects is an empty set.

and Dim3Set according to the ids. These new types form the basis for the analysis
of spatial relationships using DE-9IM matrices.

Dim0Set: A Dim0Set corresponds to an original Point feature. And thus consists

only of a Vertex, which forms the 0-dimensional interior of the Point feature.
A Dim0Set has no boundary and everything except the Vertex itself is the
exterior.
Dim1Set: A Dim1Set consists of vertices and half-edges. It is a 1-dimensional
geometry object and is the equivalent of the LineString and MultiLine
String features. The vertices between the half-edges form the interior of dimen-
sion 0, the half-edges form the interior of dimension 1.
Dim2Set: A Dim2Set is the equivalent of the original polygon features. Its
interior consists of the faces, their common half-edges and common vertices. All
half-edges and their vertices adjacent to unrelated faces form the boundary in the
dimensions 0 and 1.
These three sets only use the indices of the respective elements (Vertex,
HalfEdge, Face) of the space decomposition. Specifically, the attributes
Interior and Boundary are used to link the space decomposition to given fea-
tures as shown in Table 1. With this data structure, the spatial relationships of the
original features can be mapped by simple set operations.
Only the Dim0Set, Dim1Set and Dim2Set are used to determine spatial rela-
tionships of features or to check the validity using DE-9IM matrices. Such matrices
have the advantage to represent complex questions about spatial relations of features
in a simple and expressive way. Two feature sets .a,.b can be queried, if they fulfill
the topological relations, declared with a DE-9IM-Matrix.
⎡ ⎤
dim(I (a) ∪ I (b)) dim(I (a) ∪ B(b)) dim(I (a) ∪ E(b))
D E9I M(a, b) = ⎣dim(B(a) ∪ I (b)) dim(B(a) ∪ B(b)) dim(B(a) ∪ E(b))⎦
dim(E(a) ∪ I (b)) dim(E(a) ∪ B(b)) dim(E(a) ∪ E(b))
(2)
The dimensions for the Interior . I , Boundary . B and Exterior . E are 0, 1 or 2. All
operations to calculate the query are set based and do not require any geometrical
calculations.
Robust Spatial Decomposition - Test Case with Cadastral Data 383

Fig. 3 Map of a local subdistrict

4 Test Design

The purpose of the testbed is to demonstrate the functionality and robustness of the
new approach. For this task, simple test files and a data set from the German real
estate cadastre was used, which covers the parcel and building data of a subdistrict as
shown in Fig. 3. These data were originally available in the gml-based NAS exchange
format, which is common for German geo-data infrastructure (GDI). It was converted
with QGIS into two simple CSV-files containing the geometry of the simple-features
as well known text (WKT) and the corresponding ID of the feature9 .
All parcels and buildings are exported separately into a single file, so that an
assignment to the categories can be made via the file name during import. In this
way, each feature is assigned to a two-part Id, with category and object Id. The tests
arise from the following fundamental questions:
1. Do the features meet the requirements for OGC simple features?
2. Are there overlapping parcels?
3. Are there gaps between parcels?
4. On which parcels is a building located?
Question 1 is related to geospatial data quality for a single feature, question 2 and
3 for a set of features. Question 4 exemplifies queries on heterogeneous data sources.

9 https://github.com/dd-bim/cadastralTestData.git.
384 E. Romanschek et al.

The testbed also ensures that the implementation can detect invalidities. Several
possible defects are modeled as shown in Table 2. The CSV-files contain single simple
feature geometries as WKT with invalid geometries. The test succeeds (true negative)
if the simulated defect is detected.

4.1 Test 1: Do the Features Meet the Requirements for OGC

Simple Features?

This test serves to exemplify the basic prerequisite for geospatial analysis in general.
It checks whether the input geometry can be transformed to a valid space decompo-
sition.
In [1] the types and their properties of the OGC simple features are described. In
the testbed only the OGC types Point and Polygon are used.
Simple feature polygons are valid if they fulfill OGC/ISO-standardized topologi-
cal conditions [1, sec. 6.1.11]. In addition, the assertions for LinearRings [1, sec.
6.1.7] have to be true.
Lines 1 to 13 in Table 2 shown which particular defects are simulated in the
testbed. Column “Processing step” describes, whether these defects are detected
during import of simple features, or while converting to the features to space decom-
position, or at the final query level.

4.2 Test 2: Are There Overlapping Parcels?

This test stems from the legal requirement that parcels are disjunctive areas of land
ownership, surrounded by well defined boundaries. This test is performed on the
space decomposition, not on features.
The basic functionality is checked by the simple test from Table 2, line 14, then
the implementation is done on the real estate cadastre data. All dimension-based
feature sets of the category “parcel” are mutually tested. If the following equation
holds, then the two tested parcels (.a, .b) overlap.
⎛ ⎞
2∗∗
r elate(a, b) = ⎝∗ ∗ ∗⎠ (3)
∗∗∗

With this DE-9IM matrix given, the presented algorithm checks, whether the
Interior2 attributes of the parcels are a non-empty intersection. This implies for
any parcel (.a) compared to any other (.b), there is an overlap if the following equation
holds:

2 = dim(I nterior (a) ∩ I nterior (b)) (4)

Robust Spatial Decomposition - Test Case with Cadastral Data 385

Table 2 Constructed defects for checking the test cases

# Defect Processing step WKT or figure
1 consecutive identical POLYGON ((0 0,2 0,2 0,...))
points
2 too few points POLYGON ((0 0))
3 not closed POLYGON ((0 0,2 0,1 1,0 1))
feature / import
4 point used twice

5 overlapping edges POLYGON ((0 0,0 1,0 0))

6 empty polygon POLYGON ()

7 wrong orientation

8 inner polygon is out-

side

9 more than one vertex

connected to the in-
ner polygon

10 self intersection decomposition /

generation
11,12,13 inner face shares
outer edge

14 Test 2 subsection 4.2 decomposition /

(overlap) query

15 Test 3 subsection 4.3

(gap)

4.3 Test 3: Are There Gaps Between Parcels?

An important condition in the real estate cadastre is that the concerned region is
completely covered. This means that there must be no gaps between the parcels.
First, the simple test from (Table 2, row 15) is performed to be sure that defects are
determined correctly. With the completed space decomposition it is possible to check
whether the entire set of Dim2Set objects contains objects that do not belong to
the category “parcel”.
This is where the advantage of space decomposition becomes particularly appar-
ent, as the topology covers an area without gaps. With feature-based approaches, an
empty area is not recognized as a separate object and can therefore only be determined
indirectly via geometrical queries on “nearby” objects.
386 E. Romanschek et al.

4.4 Test 4: On Which Parcels Is a Building Located?

This test serves to demonstrate the advantages of the space decomposition. It shows,
that simple set queries can be used to reproduce complex use cases.
To determine on which parcels a building is located, one can proceed as in Subsect.
4.2. In this test however, only the objects of category “parcel” with those of the
category “building” are compared. If the matrix results from Eq. 3, then the building
and the parcel have a common area. Alternatively, it is again possible to compare the
interior2 attributes, as in Eq. 4.
Moreover, it tests if buildings edges cover edges of a parcel. For this purpose, when
comparing the “parcel” (.a) with the “building” (.b) objects, the following equation
based on a DE-9IM matrix must hold to show that both have a common edge:
⎛ ⎞
∗∗∗
r elate(a, b) = ⎝∗ 1 ∗⎠ (5)
∗∗∗

Expressed as a set operation, the following equation must be true:

1 = dim(Boundar y(a) ∩ Boundar y(b)) (6)

This means the two objects have a common edge if the intersection of the
Boundary1 attributes of both objects is not empty.

5 Test Results and Comparison

This section covers the results of the presented test cases. In addition, the same tests
were performed with the software FME. Both software systems have identical results
indicating that the developed approach is valid.

5.1 Results Test 1 (Valid Feature)

The analysis of this type of defects (lines 1 to 13 Table 2) is done during the import
procedure when the model is created. When importing the artificial test data, as
expected, the tests failed with the defective data set. Also the FME detected invalid
features with the “GeometryValidator”-module.
The presented approach is especially useful for large datasets, one can identify
the defective data from the error messages and possibly repair it afterwards, and still
have a space decomposition from the remaining data (if any).
Nevertheless, no defects were found in the data from the real estate cadastre.
Robust Spatial Decomposition - Test Case with Cadastral Data 387

5.2 Results Test 2 (Overlap)

In this test, the artificial data set (Table 2, line 14) was imported first. The subse-
quent search for overlapping areas, according to the procedure in Subsect. 4.2 was
successful. Thus, the overlapping area was found.
Subsequently, the data set from the real estate cadastre was imported and searched
for overlapping parcels. This search was unsuccessful, meaning that the dataset does
not contain any overlapping parcels.
The same results were obtained with FME with the modules “SpatialRelator” and
“SpatialRelation” using the parameter “OGC_overlap”.

5.3 Results Test 3 (Gap)

After the artificial test (Table 2, line 15) data was imported, it was possible to directly
detect the gap.
As can be seen in the magenta polygons in Fig. 4, there are three gaps between the
parcels. The gaps have been correctly found by the software as well as in FME. How-
ever FME required some advanced transformation modules like “Dissolver” to create
a surface, that contains all parcels and then “DonutHoleExtractor” to conceptualize
the holes.

Fig. 4 Gaps (magenta) between parcels

388 E. Romanschek et al.

Fig. 5 Examples from the results of test 4

5.4 Results Test 4 (Parcels and Buildings)

This test is not made to detect invalid geometries, but serves as an example how to
use the concept of space decomposition with heterogeneous geodata. The testbed
determines on which parcels a building is located and which building has edges
located on a parcel boundary.
A Dim2Set-Dim2Set result of the query “On which parcels is a building
located?”, is shown in Fig. 5a. Here, a connected block of small townhouses is located
on several narrow parcels.
An example of a building with edges on a parcel boundary, is shown in Fig. 5b.
There, the building is located with two sides on one parcel boundary each.
The algorithm presented in this paper and the FME gave the same results. However
FME required some advanced transformation modules like “Topologybuilder” with
some hidden parameters (e.g. “Maximum Coords Per Edge”) and additional filter on
the number of intersections.

6 Discussion and Outlook

In this paper it could be shown that an alternative approach, using a robust space
decomposition with integer numbers only, can be successfully used to check data
of the real estate cadastre. The results of the presented algorithm are more reliable,
as there can be no ambiguities due to errors in the topology or due to floating point
rounding errors. The tests were conducted with simulated and real data from German
cadastre. By receiving equal results of basic topological queries with the ETL tool
FME the implementation of the presented alternative approach has been verified.
Robust Spatial Decomposition - Test Case with Cadastral Data 389

With the presented implementation, topological queries on spatial features become

more flexible, because proper DE-9IM matrices are supported, whereas FME only
offers topological predicates. The conceptual key of the presented approach is, that
topological queries are pure set operations. No computational expensive and error
prone geometric calculations are necessary.
The parcel polygons data had been quality assured by the authorities before being
published within the German geo-data infrastructure (GDI). Further research will be
undertaken on more, possible inconsistent geodata like the open street map (OSM).
The developed concepts are also applicable for two-dimensional CAD models. There-
fore, the presented data structure and queries must be extended for additional seman-
tic properties. These properties like color, line type, layer or block attributes might
be extracted from CAD file.
The presented algorithm assumes purely deterministic input values. In the future,
stochastic input data, such as measuring points from laser scanning, could also be
included in the spatial analysis. The algorithm would then have to carry out sig-
nificance tests on the nodes, edges and regions and thus check whether points are
identical, edges have no significant length or regions have no significant area. The
Statistical tests would then delete the non-significant elements and thus simplify the
topological structure of the space decomposition.
Originally it was planned to carry out the spatial decomposition in geospatial and
surveying applications in three-dimensional. This next step is non-trivial and can
only be approached once the concepts, algorithms, data structures and use cases in
the two-dimensional space are sufficiently understood and robustly implemented.

References

1. OpenGIS implementation standard for geographic information - simple feature access - part
1: common architecture (2011) OpenGIS Implementation Standard OGC 06-103r4, Open
Geospatial Consortium. https://www.ogc.org/standards/sfa
2. Clementini E, Sharma J, Egenhofer MJ (1994) Modelling topological spatial relations:
strategies for query processing. Comput Graph 18(6):815–822. https://doi.org/10.1016/0097-
8493(94)90007-8
3. Egenhofer M (1990) A mathematical framework for the definition of topological relations. In:
Proceedings of the fourth international symposium on spatial data handing, pp 803–813
4. Egenhofer MJ, Sharma J, Mark DM, et al (1993) A critical comparison of the 4-intersection and
9-intersection models for spatial relations: formal analysis. In: AUTO-CARTO 11 proceedings.
ASPRS, pp 1–11
5. Huhnt W (2018) Reconstruction of edges in digital building models. Adv Eng Inform 38:474–
487. https://doi.org/10.1016/j.aei.2018.08.004
6. ISO 19101-1: geographic information – reference model – part 1: fundamentals (2014)
Standard, International Organization for Standardization, Geneva, CH. https://www.iso.org/
standard/59164.html
7. Li C, Pion S, Yap C (2005) Recent progress in exact geometric computation. J Log Algebraic
Program 64(1):85–111. https://doi.org/10.1016/j.jlap.2004.07.006
8. Romanschek E, Clemen C, Huhnt W (2021) A novel robust approach for computing DE-9IM
matrices based on space partition and integer coordinates. ISPRS Int J Geo-Inf 10(11):715.
https://doi.org/10.3390/ijgi10110715
390 E. Romanschek et al.

9. Salleh S, Ujang U, Azri S (2022) Representing 3d topological adjacencies between volumes

using a 36-intersection model. Geomat Environ Eng 16(2):127–155. https://doi.org/10.7494/
geom.2022.16.2.127
10. Shen J, Chen M, Liu X (2018) Classification of topological relations between spatial objects
in two-dimensional space within the dimensionally extended 9-intersection model. Trans GIS
22(2):514–541. https://doi.org/10.1111/tgis.12328
11. Strobl C (2008) Dimensionally extended nine-intersection model (DE-9IM). In: Shekhar S,
Xiong H (eds) Encyclopedia of GIS. Springer, Boston, pp 240–245
12. Thompson RJ (2007) Towards a rigorous logic for spatial data representation. NCG, Neder-
landse Commissie voor Geodesie
13. Zlatanova S, Rahman AA, Shi W (2004) Topological models and frameworks for 3D spatial
objects. Comput Geosci 30(4):419–428. https://doi.org/10.1016/j.cageo.2003.06.004
Where is the End of the
Wall: Decomposition of Air and Material
into Spaces and Building Components

Felix Gabler and Wolfgang Huhnt

Abstract It is the state of the art to use point clouds in architecture and civil engi-
neering. Lots of algorithms have been developed to create lines and polygons in 2D
or patches and polyhedrons in 3D from point cloud data. The goal of this extensive
research is an automatic generation of a building model. Specific subtasks in this
field are supported sufficiently. Examples are object recognition or the identification
of windows or doors. However, the internal decomposition into building compo-
nents is still a challenge because only visible surfaces are detected and processed.
This paper proposes a decomposition of the material into building components. The
result is a decomposition into convex sub-objects. The algorithm supports the user
by the finest decomposition but does not execute any unions. The algorithm identifies
and computes the shape of each component. It does not classify. Furthermore, the
algorithm can be used to decompose the space. The paper shows examples for both:
the decomposition of material and air into building components and spaces. The
paper focuses on a 2D approach. The outlook addresses the applicability to 3D. The
algorithm presented in this paper is embedded in a sequence of algorithms which
use space partitioning as the basic concept for all computations: the identification of
material and space. The paper presents the actual state of research in this field. The
concept has been worked out. Artificial examples give excellent results. The transfer
to real point clouds as input is in progress.

Keywords Scan-to-BIM · Indoor environment reconstruction · Building

components · Decomposition · Space partitioning

Supported by German Research Foundation.

F. Gabler (B) · W. Huhnt

Technische Universität Berlin, Straße des 17. Juni 135, 10623 Berlin, Germany
e-mail: f.gabler@tu-berlin.de

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 391
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_27
392 F. Gabler and W. Huhnt

1 Introduction

In architecture and civil engineering the scan-to-BIM approach is a rapidly grow-

ing field of research in the past years [5]. Various application possibilities during
the design- and construction-process and the operation- and maintenance-phase are
present [15]. However, current approaches in the industry still require labor-intensive
manual work. It is therefore not surprising that multiple semi-automated and auto-
mated approaches have been presented in the recent years, comprehensive overviews
of current approaches are given in [5, 11, 15]. Nevertheless, a complete automation
of deriving as-built models from point cloud data remains an open challenge [4].
The research presented in this paper makes use of an approach developed by [13].
The location of the scanner plays a significant role. Triangles are the result of point
cloud processing, describing a region of air that starts at the scanner and ends at
the surface of material. This intermediate result is mapped onto a space partition.
Furniture and circulating elements are removed. In addition, doors and windows are
identified and removed as well so that the resulting model consists of regions of air,
regions of material and unknowns.
The complete pipeline is addressed in the common research project ReconTOP
which is executed with partners from TU Wien, Austria and TU Berlin, Germany.
This paper addresses as a part of that project the decomposition of material and air. It
is focused on geometry and topology. Semantic information are not addressed except
the identification of spaces and material.
This present paper is structured in such a way it presents the concept using artificial
examples. The transfer in using point clouds as an input is far driven and explained
in parallel.

2 Related Research

A comprehensive overview of automated and semi-automated as-is BIM reconstruc-

tion techniques in the field of indoor environment reconstruction is given by [5]. Good
results are computed if the geometry is parallel to the coordinate system (Manhat-
tan assumption). Other geometries are still challenging. A summary of the present
point cloud processing methods including registration, down-sampling and semantic
segmentation methods are given by [12]. Lots of problems such as an automated
registration are solved. The article points out that “a fully automated process for the
extraction of semantics from the raw data in BIM remains a challenge”.
Multiple different approaches for the reconstruction of an indoor environment
from point cloud data have been presented in the last years. [1] subdivides present
approaches in “wall-based” and “room-based” approaches. At the end, these
approaches address both, walls and rooms. However, they always start either with
walls or rooms. There is no integrated approach identifying both, walls and rooms, in
Decomposition of Air and Material into Spaces and Building Components 393

a single step. The approaches presented do not address the decomposition of material
into components.
The identification of geometry can be structured in region-growing and random
sample consensus (RANSAC)-based methods [5], applicable both in 2D and 3D.
Presenting an approach for the automatic reconstruction of a floorplan, [3] makes
use of a region-growing method for the detection of vertical planes, followed by a
projection on the x-y plane. The result is a set of polygons, each of them represent-
ing the boundary of a room. A decomposition of material is not addressed. For an
automatic extraction of structural elements, [14] makes use of a decomposition in
2D cells and region-growing for the determination of slabs and walls. Afterwards,
a voxel-based region-growing method in 3D is used to derive columns and beams.
The authors apply their method to structural components from concrete. They do not
address any kind of rooms. In its present state the method can not be applied for an
existing building (as-is BIM).
A known and often addressed challenge in indoor reconstruction is occlusion.
Furniture may be a reason for occluded surfaces of building elements. An approach
presented by [10] makes use of “pending” walls and an energy minimization problem
for extracting a watertight 2D floorplan. The method is applicable for geometries
which fulfill the Manhattan assumption.
Making use of region-growing combined with a conditional random field algo-
rithm, [1] presents a method to derive wall elements from a 3D point cloud. In
combination with the convex hull, room objects are generated. Not intersecting,
best-fitting walls are calculated and topological connections of walls are generated
afterwards. The approach is focused on walls as defined in IFC. The approach has
several limitations. The authors write: “A manual operator should therefore still val-
idate the model and some manual adjustments may be in order to complete the BIM
walls.”
Reconstructing structural elements, such as walls, floors and ceilings, including
openings, is identified as a major field in indoor reconstruction. However, due to
the fact, that only geometric information of visible surfaces of building components
can be obtained by using a laser scanner [14], a lack of approaches addressing the
decomposition of structural elements seems to be present. One approach taking non-
visible contact faces into account and reconstruct wall intersections volumetrically
is presented by [8]. In point cloud data detected surfaces are interpreted as infinite
planes and intersected with each other. The result is 3D-space-segmentation into
convex polyhedrons, followed by a determination of room, wall/slab and outside
cells. With this approach building element intersections are modeled volumetrically.
This approach seems to be far driven and seems to have the option to really result
in a automated reconstruction of building models. The goal is not yet achieved. For
instance, slanted walls can not be processed.
A basic assumption of our approach is the existence of exactly two conditions
“known” and “unknown” or “air” and “material”. The voxel-based growing method
presented by [9] makes use of the same concept, the presence of “void” and “non-
void” areas. Growing void-voxels results in cuboid 3D-cells, each of them known
to be a room. The space between them, is identified as walls or more general as the
394 F. Gabler and W. Huhnt

building structure. However, the method is only applicable for geometries following
the Manhattan assumption. In addition, a decomposition of material is not addressed.
In computational geometry, polygon decomposition is still a field of research
with many applications including pattern recognition, collision detection and skele-
ton extraction. An extensive overview for planar polygons is given by [7]. Existing
methods are classified by [6] in regards of the type of the input polygon, the allowance
of Steiner-points during the decomposition and the desired sub-polygons. Many cal-
culations, e.g. testing if a point is inside a polygon, are much easier to perform
on convex polygons. An algorithm for the optimal convex decomposition is pre-
sented by [2], decomposing a polygon into a minimum number of convex parts. The
algorithm presented in this present paper decomposes simple polygons into convex
sub-polygons. Holes are allowed. This algorithm does not contribute to the research
field of optimal or approximate convex decomposition.

3 Concept, Algorithm and Examples

3.1 Concept

Focus of this research is the decomposition of the building structure and of spaces.
The concept is presented in this section. It is structured in several steps. In the
2-dimensional approach, triangles are generated from point clouds in a first step.
These triangles are known to be from air. They have a line which separates air from
material. The triangles are not ordered and form a triangle soup. In a second step, the
triangles are mapped onto a space partitioning. This step identifies contiguous areas
from air. All other areas of the space partitioning are potential candidates to be from
material. The identification of the areas from material takes place in step three. The
decomposition is step four. Results can be graphs and closed polygons in 2D.
In the 3-dimensional approach, all steps are equivalent. Tetrahedrons from air
are generated from points clouds. They have a triangle which separates air from
material. Results after the execution of all steps are closed polyhedrons. The following
subchapters are focused on the 2-dimensional approach.

3.2 The Triangle Soup

Figure 1 shows two artificial examples of triangles from air. The first example has
two locations of the laser scanner in the interior of a building, each in a separate
room. The second example has four locations of the laser scanner, one in the interior
and three outside the building.
The triangles shown in Fig. 1 describe the complete surface of the building without
any gaps or overlaps. This is not true for real laser scans. Figure 2 shows an example
Decomposition of Air and Material into Spaces and Building Components 395

s1
v17 v11

v6 v5
v15 v16
v5 v4
v8 v7
s0 v3 v4
s0
v13 v12
v2 v3

v9 v8 s3
v2 v14
s1 v0 v1
v0 v1 v6 v7
v10 v9
s2

Fig. 1 Two artificial examples

Fig. 2 Triangles from air detected from point clouds

of a simulated laser scan with several locations inside and outside. The triangles
overlap; and not all points of the surface of building components are detected. The
generation of these triangles is shown in [13].

3.3 Identification of Areas from Air

The triangle soup is mapped on a space partition. The result is a mesh of triangles.
In this mesh, a breadth-first-search determines sets of connected triangles from air.
Each breadth-first-search starts with a triangle behind an edge which is known to be a
396 F. Gabler and W. Huhnt

Fig. 3 Identification of Areas from Air of the two examples shown in Fig. 1

Fig. 4 Identification of Areas from Air of the Example shown in Fig. 2

boundary of an original triangle from air. The search ends at edges with the identical
information.
Figure 3 shows the results for the two examples shown in Fig. 1. Figure 4 shows
the result for the example shown in Fig. 2.

3.4 Identification of Areas from Material

All triangles which are not identified as triangles from air are the input of the iden-
tification of areas from material. The first artificial example shown in Fig. 1 has no
laser scans of the exterior of the building. In this example, the triangles not from
Decomposition of Air and Material into Spaces and Building Components 397

Fig. 5 Triangles not from Air of the Examples shown in Fig. 1

Fig. 6 Triangles not from Air of the Example shown in Fig. 2

air are potentially from material. The second artificial example shown in Fig. 1 is
constructed in such a way that all triangles are from material which are not from air
shown in Fig. 5.
Figure 5 shows examples which are special cases. In general, some boundaries of
building components are not detected. This results in areas of unknowns. In addition,
one boundary of a building component can be detected several times from different
laser scanner positions at different locations due to imperfections of the scanning
device and the scanning procedure. Figure 6 shows the areas which are identified to
be not from air of Example 2 (see Fig. 2).
The areas not from air are investigated. Two topological conditions are applied:

– A triangle which starts at a laser scanner position must be from air.

398 F. Gabler and W. Huhnt

Fig. 7 Identified Areas of Material

– A set of triangles with no boundary between air and material which is surrounded
by triangles from air must also be from air.

The result are sets of triangles in which material must be located. In a last step, a
geometric condition is applied to the outer polygon of each connected set of triangles:
Consider two polygon edges e1 and e2 which are known to separate air from
material. Consider only two edges e1 and e2 where between e1 and e2 are edges
which do not separate air from material. p1 and p2 are start and end points of e1; p3
and p4 are start and end points of e2. Compute the intersection point p of the straight
lines defined by e1 and e2. Substitute all edges between e1 and e2 by the edges p2
and p and p and p3 if p exists and if p lies behind p2 on e1 and in front of p3 on e2.
Figure 7 shows the result of the application of the three conditions on the example
shown in Fig. 2.

3.5 Decomposition

The proposed decomposition algorithm focuses on splitting both, identified areas

from material and identified areas from air into several convex sub-objects, leading
to a convex decomposition of both, the building structure and the spaces.
Each connected set of triangles from either air or material is handled individually.
First, the boundary polygon of a given set of connected triangles is determined, which
may lead to polygons with holes. While an object can have multiple boundary loops,
it is ensured that every boundary is a closed edge loop.
Starting from a random point, the boundary will be traversed counterclockwise,
at every vertex the internal angle will be investigated. If the internal angle is greater
Decomposition of Air and Material into Spaces and Building Components 399

than 180.◦ , a vertex making the given object non-convex is found and stored for the
next processing step.
In the first example given in Fig. 1, investigating the area known not to be air, v2
is a vertex fulfilling the previously mentioned criteria. Oriented counter-clockwise,
the previous edge is from v1 to v2, the next edge from v2 to v3.
For all reflex vertices, the algorithm determines two direction vectors: one in the
positive direction of the previous edge and a second in the negative direction of
the next edge. For both vectors the closest intersection point with another boundary
edge is calculated. To avoid round-off errors, the intersection point will be stored as
a fraction as described in [13]. For the example in Fig. 1 this will lead to intersection
points with the edges v4-v5 and v5-v0. Two new edges from v2 are created and
stored as new interior boundary edges. For every edge the corresponding twin will
be generated too.
After all edges have been created for vertices fulfilling the criteria, sub-objects will
be found with a breadth-first search. The result is a set of multiple convex objects
for every given initial object. The information about being “air” or “material” is
consistent through the whole process.

input: n objects with m triangles each

forall objects do
get boundary of object
forall vertices v of boundary do
calculate internal angle
if concave vertex then
calculate two direction vectors u,
one in positive direction of previous edge,
one in negative direction of next edge
forall vectors u do
forall edges e of boundary do
calculate intersection point p
if amount of intersection points >1 then
choose closest intersection point
create interior boundary edge from v to p
create interior boundary edge from p to v

Algorithm 1: The Decomposition Algorithm

3.6 Results

Figure 8 and Fig. 9 show the results of the presented decomposition. Both, geometric
and topological data are available. Geometric data results from the geometry of
the triangles describing each identified object from air or material. For better display
purposes, only the boundary of each identified object is shown. Edges of the triangles
in the interior of each identified object have been removed.
400 F. Gabler and W. Huhnt

Fig. 8 Results of the Decomposition of the first Example from Fig. 1

Fig. 9 Results of the Decomposition of the second Example from Fig. 1

Topological data results from the connected sets of triangles describing each object
and the neighboring relation to other objects.
Topological data can be mapped onto each data structure describing the graph
which now consists of objects from air which are spaces (circles) and objects from
material which describe the internal structure of the building components (squares).

4 Summary, Discussion and Outlook

This paper presents research in progress. The overall goal is an automatic decom-
position of a building into building components from point clouds. The presented
results specifically focuses on the decomposition of areas into convex subareas. This
is applied to both, spaces and material.
Decomposition of Air and Material into Spaces and Building Components 401

Artificial examples shown that the proposed concept results in acceptable results.
The application to real scanned point clouds is still in progress. Two steps are still
under development, the determination of points which are used to decompose an area
and the determination of the two directions for the decomposition. The results of real
scanned point clouds are boundary polygons with points with coordinates which of
a stochastic nature. As a consequence, a point and the directions for a decomposition
must be determined as best fitting points and best fitting directions. Algorithms from
adjustment technique are at present time investigated.
The principle structure of all concepts and algorithms is set up in such a way
that it can be transferred to the three dimensional space. In the three dimensional
space, a mesh of tetrahedrons partitions the space. Input are tetrahedrons from air
which start at the location of a laser scanner. Triangles separate air from material.
The decomposition of air and material results in polyhedrons.
At present time, the research in the application to the three dimensional space is in
progress. An algorithm for the partitioning of space into tetrahedrons has already been
developed. Future results will show its runtime behavior for this field of application.

Acknowledgements The results presented in this paper are funded by the Deutsche Forschungs-
gemeinschaft (DFG, German Research Foundation) - Project number 454008779. Our project is
a part of ReconTOP which consists of two additional projects. The authors thank all members of
the ReconTOP research project, Georg Suter, Timo Hartmann, Abdullah Elsafty, Amir Ziaee and
Elsa Kronke for the fruitful discussions. The authors thank also Enrico Romanschek and Christian
Clemen from HTW Dresden, Germany, for supporting the presented research with the data of the
third example.

References

1. Bassier M, Vergauwen M (2020) Unsupervised reconstruction of building information model-

ing wall objects from point cloud data. Autom Constr 120:103338. https://doi.org/10.1016/j.
autcon.2020.103338. https://www.sciencedirect.com/science/article/pii/S0926580520309183
2. Chazelle B, Dobkin DP (1985) Optimal convex decompositions. In: Machine intelligence and
pattern recognition, vol 2. Elsevier, pp 63–133. https://doi.org/10.1016/B978-0-444-87806-9.
50009-8. https://linkinghub.elsevier.com/retrieve/pii/B9780444878069500098
3. Fang H, Lafarge F, Pan C, Huang H (2021) Floorplan generation from 3D point clouds: a space
partitioning approach. ISPRS J Photogram Remote Sens 175:44–55. https://doi.org/10.1016/
j.isprsjprs.2021.02.012. https://linkinghub.elsevier.com/retrieve/pii/S0924271621000538
4. Jung J, Stachniss C, Ju S, Heo J (2018) Automated 3D volumetric reconstruction of multiple-
room building interiors for as-built BIM. Adv Eng Inform 38:811–825. https://doi.org/10.1016/
j.aei.2018.10.007. https://linkinghub.elsevier.com/retrieve/pii/S1474034618300600
5. Kang Z, Yang J, Yang Z, Cheng S (2020) A review of techniques for 3D reconstruction
of indoor environments. ISPRS Int J Geo-Inf 9(5):330. https://doi.org/10.3390/ijgi9050330.
https://www.mdpi.com/2220-9964/9/5/330
6. Lien JM, Amato NM (2006) Approximate convex decomposition of polygons. Comput Geom
35(1-2):100–123. https://doi.org/10.1016/j.comgeo.2005.10.005. https://linkinghub.elsevier.
com/retrieve/pii/S0925772105001008
402 F. Gabler and W. Huhnt

7. Mark Keil J (2000) Polygon decomposition. In: Handbook of computational geometry. Elsevier,
pp 491–518. https://doi.org/10.1016/B978-044482537-7/50012-7. https://linkinghub.elsevier.
com/retrieve/pii/B9780444825377500127
8. Ochmann S, Vock R, Klein R (2019) Automatic reconstruction of fully volumetric 3D
building models from oriented point clouds. ISPRS J Photogram Remote Sens 151:251–
262. https://doi.org/10.1016/j.isprsjprs.2019.03.017. https://linkinghub.elsevier.com/retrieve/
pii/S0924271619300863
9. Pan Y, Braun A, Borrmann A, Brilakis I (2021) Void-growing: a novel scan-to-BIM method
for manhattan world buildings from point cloud. In: Proceedings of the 2021 European con-
ference on computing in construction. Computing in construction, vol 2. ETH, Online Con-
ference, pp 312–321. https://doi.org/10.35490/EC3.2021.162. https://ec-3.org/publications/
conference/paper/?id=EC32021_162
10. Previtali M, Díaz-Vilariño L, Scaioni M (2018) Indoor building reconstruction from occluded
point clouds using graph-cut and ray-tracing. Appl Sci 8(9):1529. https://doi.org/10.3390/
app8091529. http://www.mdpi.com/2076-3417/8/9/1529
11. Pătrăucean V, Armeni I, Nahangi M, Yeung J, Brilakis I, Haas C (2015) State of research in
automatic as-built modelling. Adv Eng Inform 29(2):162–171. https://doi.org/10.1016/j.aei.
2015.01.001. https://linkinghub.elsevier.com/retrieve/pii/S1474034615000026
12. Rashdi R, Martínez-Sánchez J, Arias P, Qiu Z (2022) Scanning technologies to build-
ing information modelling: a review. Infrastructures 7(4):49. https://doi.org/10.3390/
infrastructures7040049. https://www.mdpi.com/2412-3811/7/4/49
13. Romanschek E, Clemen C, Huhnt W (2020) From terrestrial laser scans to a surface model
of a building; proof of concept in 2D. In: Ungureanu LC, Hartmann T (eds) EG-ICE 2020
workshop on intelligent computing in engineering, pp 432–442
14. Truong-Hong L, Lindenbergh R (2022) Extracting structural components of concrete build-
ings from laser scanning point clouds from construction sites. Adv Eng Inform 51:101490.
https://doi.org/10.1016/j.aei.2021.101490. https://www.sciencedirect.com/science/article/pii/
S1474034621002391
15. Wang Q, Guo J, Kim MK (2019) An application oriented scan-to-BIM framework. Remote
Sens 11(3):365. https://doi.org/10.3390/rs11030365. http://www.mdpi.com/2072-4292/11/3/
365
Transforming Building Components
into a Space Partition to Identify Indoor
and Outdoor Spaces in Digital Building
Models

Joanna Zarah Vetter and Wolfgang Huhnt

Abstract Modeling indoor and outdoor spaces is an important base in different

application fields. In state-of-the-art modeling software products, spaces are often
modeled as virtual solids. Nevertheless, these software tools suffer from unreliable
identification algorithms for these virtual solids. This paper highlights different appli-
cation fields and explains why current modeling techniques for indoor and outdoor
spaces are not sufficient for those application fields. Examples of problematic cases
and situations are given using two software products, Autodesk Revit and ArchiCAD
by Graphisoft. Furthermore, the paper proposes the usage of space partition in digital
building models as an alternative to overcome the illustrated problems. The bound-
ary representations of building components are transformed into a space partition.
Consequently, empty spaces in the interior and exterior of a building are an integral
part of this partition and are identified automatically. It is shown that the resulting
identified indoor and outdoor spaces of the partition are more reliable in problematic
situations compared to the virtual solids created by conventional modeling software
products. The advantages of using the concept of space partition to model indoor
and outdoor spaces in digital building models are shown and discussed.

Keywords Room Generation · Space Partition · Empty Space · Topological

Relations · 3-Dimensional Triangulation

1 Introduction

The developments of the last decades have enabled digital building models to become
a centralized starting point for various fields of application. The functional scope of
software tools has increased while the compatibility between software tools has

This research is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foun-
dation) – Project number: 424641234.

J. Z. Vetter (B) · W. Huhnt

Technische Universität Berlin, Berlin, Germany
e-mail: j.vetter@tu-berlin.de

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 403
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_28
404 J. Z. Vetter and W. Huhnt

improved through standardized file formats such as the Industry Foundation Classes
(IFC). Conventional modeling software tools for digital building models focus on
modeling solids. Therefore, some parts are treated rather poorly during the modeling
process, considering their significance in various use cases. By only modeling solids,
modeling software tools insufficiently cover the empty indoor and outdoor spaces in
building models. To compensate for missing information of the empty spaces, rooms
or zones are often created as virtual solids in existing modeling software products.
The generation of rooms based on such an approach tends to suffer from different
problems. Additionally, modeling software products often do not consider outdoor
spaces.
Nevertheless, different application fields rely on robust knowledge about empty
spaces in digital building models. This paper suggests the alternative approach of
using space partition to identify rooms and zones automatically. There are two differ-
ent ways to construct a digital building model based on a space partition: developing
a tool to model new building components based on a space partition or transforming
a given model into a space partition. This paper uses the latter one. The input of
the transformation process is a given boundary representation of solids. The output
is a tetrahedral mesh in which a set of tetrahedrons describes each building object
or each empty space. Identifying neighboring relations is possible without calcula-
tions of high effort by using resulting data structures that explicitly store topological
relations.
The following paper presents relevant application fields that use indoor and out-
door spaces in digital building models in Sect. 2. Section 3 shows that the current
modeling software products Revit and ArchiCAD cannot meet the requirements of
different application fields. Section 4 presents the suggested transformation process
of a given boundary representation into a space partition and an overview of previous
research in this field. At the end of this paper, Sect. 5 describes relevant examples
that show the difficulties of current modeling software tools. These examples also
show the advantages of using a space partition to identify indoor and outdoor spaces.

2 Demand for Modeling Indoor and Outdoor Spaces

This section gives an overview of the current demand for modeling indoor and outdoor
spaces in building models in different fields of application.
Indoor and outdoor spaces in building models play an essential role in different
application fields. Digital building models play an increasing role in all processes in
the AEC industry. Steps like quantity take-off calculation require detailed informa-
tion about the geometry of objects and the empty spaces in building models. Current
practices are often based on rooms in digital building models and may require high
amounts of manual adjustments to ensure the correctness of volume or area com-
putation. Recent discussions on social media platforms like LinkedIn showcase the
demand for modeling indoor spaces and that workarounds are the constant compan-
ions of architects or civil engineers, as seen in Fig. 1.
Identifying Indoor and Outdoor Spaces in Digital Building Models 405

Fig. 1 Different options of placing spaces in a model. From left to right: automatic placement of
rooms; addition of a separate room for the empty space; usage of separation lines to divide the lower
space into two rooms; manual remodeling of the automatically created rooms [7].

Further application fields rely on information about indoor spaces in the AEC
industry, e.g., indoor route planning and energy simulation processes. While indoor
route planning requires information about topological relations between building
objects, rooms and potential openings in building components, energy simulations
often require second-level space boundaries. The Industry Foundation Classes (IFC)
are a widely used standard to transfer building information models between software
tools. They also provide necessary structures such as the entity IFCRelSpaceBound-
ary [4] to transfer information about topological relations and second-level space
boundaries. Several papers cover the export of given indoor spatial information in
IFC files to the OGC standard IndoorGML or other graph-based data structures for
indoor route planning [19, 20, 22]. Additionally, an open-source tool to convert
IFC files to the IndoorGML standard was developed by Diakité et al. [6]. Even
though many papers depend on the space boundaries exported into IFC files, they
are often incorrect or completely missing. Export functions of space boundaries
in modeling software tools often require the accurate modeling of indoor spaces.
Therefore, incorrectly modeled indoor spaces result in incorrectly exported space
boundaries. Various approaches exist for computing spatial relations between indoor
spaces directly from the geometry in IFC files, e.g. [14, 15], or to identify and fix
errors in exported second-level space boundaries in IFC files [13, 23]. This paper
follows another approach aiming to close the gap between digital building models
and the automatic identification of indoor spaces in those models.
Next to indoor spaces in digital building models, outdoor spaces also need more
focus. Outdoor spaces of building models could be the connector between BIM and
GIS data. Lee and Zlatanova [10] developed a framework to map IFC files into the
CityGML format, a commonly used standard for GIS data. They highlight that the
outer surfaces of exterior walls are not explicitly saved in IFC files, even though
a wall can be marked as an exterior building element in an IFC file. Therefore, the
geometric computation of the outer surface is necessary for successfully mapping IFC
files into the CityGML standard. Different approaches were developed to calculate
this exterior surface from IFC files to overcome this issue [1, 5, 8, 11]. Donkers
et al. [8] suggest improving the standardization and integration of exterior spaces
406 J. Z. Vetter and W. Huhnt

Fig. 2 Extruded rooms

cannot be widened above
walls that are not reaching
up to the ceiling [3].

in the IFC standard. Arroyo Ohori et al. [1] listed guidelines for better conversion
and integration of BIM and GIS data. One suggestion was to carefully and correctly
model indoor spaces without gaps or intersections. However, this is not adequately
supported by existing tools as described above.

3 State-of-the-Art Modeling Software Products

This paper investigates two state-of-the-art modeling software tools to identify which
problems exist regarding the creation and export of indoor and outdoor spaces. For
this paper, Autodesk Revit 2023 and ArchiCAD 25 by Graphisoft are used to show-
case current problems. Revit and ArchiCAD are two of the most used modeling
software tools in the AEC industry. It is assumed that investigating and comparing
Revit and ArchiCAD covers the majority of users all around the world.
In Revit, spaces are called rooms. In ArchiCAD, spaces are referred to as zones.
Indoor spaces are created the same way in both software products. A polygonal
boundary can be created manually or automatically using the bounding building
objects. This polygonal boundary is extruded until a vertically bounding element is
reached. This general approach causes problems while creating rooms or zones for
complex geometries. The extruded solid can be modified so that slanted roofs and
slabs can be determined as bounded elements and crop the zone accordingly.
Nevertheless, it is impossible to widen a created zone horizontally after its extru-
sion in the upper part while preserving the lower boundary polygon. Figure 2 shows
a simple example modeled in Revit suffering from this problem. The different
workarounds created for a room with a gallery in Fig. 1 result from the same problem.
Revit additionally implemented a tolerance for creating rooms, making it difficult
or even impossible to place rooms in tiny spaces. ArchiCAD does not have such a
tolerance, making creating smaller rooms in ArchiCAD easier than in Revit. In Revit,
the room creation algorithm uses an artificially increased wall width of 1.004 ft to
detect spaces where it is possible to place a room and identify the room-bounding
elements. This process is shown in Fig. 3. Once the bounding elements are identified,
the actual room boundaries are calculated by reverting the artificially increased wall
widths and using the actual wall measurements. A discussion [2] on the Autodesk
Knowledge Network summarizes the impact and the issues caused by this implemen-
Identifying Indoor and Outdoor Spaces in Digital Building Models 407

Fig. 3 Setting the wall

widths artificially to 1.004 ft
to determine all areas where
it is possible to place a room
[2].

tation. Another impact of this implementation is that spaces between building objects
placed closer than 1.004 ft to each other will not be included in the surrounding room.
Issues regarding indoor spaces result in incorrectly calculated space boundaries
in an IFC file because rooms and zones are the base for calculating first- and second-
level space boundaries in Revit and ArchiCAD. Additionally, ArchiCAD cannot
differentiate between first- and second-level space boundaries. It is only possible to
export second-level space boundaries during an IFC export. In Revit, it is possible
to select whether first- or second-level space boundaries are exported. However, it is
not possible to export both.

4 Suggested Solution

The suggested approach in this paper is to transform a given building model into a
tetrahedral space partition. Neighboring relations between the building components
and the indoor spaces in the model are stored explicitly. Additionally, the outdoor
spaces are an integral part of the space partition. Topological relations can be derived
automatically without any geometric calculations once the space partition is created.
This section summarizes the development of this approach from similar approaches
in the literature and gives an overview of the transformation process.

4.1 Related Research

Using a tetrahedral space partition to represent digital building models is not a com-
pletely new concept. Penninga [16] used a tetrahedral mesh to model 3D topographic
data. He uses a lightweight data structure in which tetrahedrons save topological rela-
tions. As an input, he uses piecewise linear complexes without any self-intersections.
Kraft [12] used existing approaches and combined them to transform boundary rep-
resentations of building models into a tetrahedral space partition. It is necessary to
insert so-called Steiner points to generate a tetrahedral mesh from a given set of
408 J. Z. Vetter and W. Huhnt

triangles describing the boundary. In some cases, uncontrollable refinement occurs

and an unacceptable amount of points must be inserted to ensure robustness. The sec-
ond problem results from the inaccuracy of floating point numbers which cannot store
each real number. Huhnt [9] developed the first steps of an algorithm for creating a
tetrahedral space partition of a given building model with integer values to overcome
the problem of uncontrollable refinement. Romanschek et al. [17] implemented this
algorithm in the two-dimensional space and used it for a scan-to-BIM approach.
Romanschek et al. [18] also showed that this algorithm can robustly compute spa-
tial relations between two geometric features. Exact computation based on rational
numbers was used to ensure robustness. A previous research [21] showed that using
this approach in the three-dimensional space is inefficient and suggested the usage
of rounded integer coordinates. This previous paper introduced an implementation
in three-dimensional space and investigated the accuracy of transforming a given
model into a tetrahedral space partition. This implementation is also the suggested
approach in this paper. The following section summarizes the presented algorithm.

4.2 Theoretical Aspects

A given building model is the input of the transformation process. Objects of this
model need to be represented via triangulated boundary representations. All surface
triangles are plane and must be oriented, each object forming a water-tight and
orientable solid.
The overall process consists of three significant steps once a boundary representa-
tion is given, illustrated in Fig. 4. In the first step, an initial mesh is created with all the
given points of the model and additional boundary points. The additional boundary
points ensure that all newly inserted and rounded points during the refinement pro-
cess lie inside the space partition. Everything outside of the initial mesh represents
the unbounded space.
The initial mesh includes all given points after the first step but does not preserve
the surfaces of the objects. Therefore, the next step is the refinement of the initial
mesh. For a three-dimensional model, all intersection points between the given tri-
angles of the input boundary representation and the edges of the tetrahedral mesh are
computed and inserted into the mesh. The validity of the tetrahedral mesh in each
step is ensured by a special treatment which is not the focus of this paper.
After the tetrahedral mesh refinement, the last step is assigning every tetrahedron
to an object, multiple objects or no object at all. The result is a valid tetrahedral parti-
tion in which a set of mesh tetrahedrons represents each object. Indoor spaces can be
identified automatically by collecting all tetrahedrons which are not part of building
objects. Every disconnected component forms a different indoor space. The outdoor
space is identified automatically as well. All neighboring relations or touching faces
of this outdoor space to any building objects can be computed without further geo-
metric computations. Contact faces of the outdoor space with the unbounded space
are also saved.
Identifying Indoor and Outdoor Spaces in Digital Building Models 409

Fig. 4 Transformation process of a given boundary representation in 2D (upper left) into an ini-
tial mesh (upper right) in which all intersections are computed (lower right) and all objects are
represented by a set of triangles (lower left) [21].

Fig. 5 Left: floor plan of the first example of a small room in Revit. Right: The room that is
recognized by Revit. The space behind the column along with the smaller room is not recognized.

5 Validating Examples

In this section, two given models are transformed into tetrahedral space partitions.
The purpose of the examples is to show the basic concepts of the suggested approach.
Therefore, they are simple and small. The first example in Fig. 5 shows the problem-
atic situations in Revit, which were described in Sect. 3. Revit cannot recognize the
small room bounded by the inner walls in this example. The part of the room behind
the column is not recognized either.
The boundary representation of this example was exported by using the Revit
API. The triangulated input for the transformation process is shown on the left side
of Fig. 6. After the transformation, sets of tetrahedrons representing the building
410 J. Z. Vetter and W. Huhnt

Fig. 6 Left: the triangulated input for the transformation process. Right: the resulting surfaces of
the computed sets of tetrahedrons which represent the building components and the indoor spaces.

Fig. 7 Left: boundary representation of a small house and its openings. Right: the automatically
extracted indoor spaces of the space partition.

objects can be collected. Only the surfaces of these resulting sets are displayed in
the following pictures. Even though the underlying mesh is used to collect the sets
of tetrahedrons, the mesh is not displayed to guarantee an optimal visualization.
The second example is a small house with complex indoor space geometry, win-
dows and doors. Figure 7 shows the example’s triangulated and exported boundary
representation. The input is then transformed into the tetrahedral space partition. All
indoor spaces can be extracted automatically using the resulting mesh, as shown in
Fig. 7. Even indoor spaces with complex geometry can be extracted shown in Fig. 8.
It would not have been possible to create all room geometries of the shown example
with Revit or ArchiCAD.
The outdoor space of this example can be also extracted but is not visualized
in the following pictures. Nevertheless, from the automatically extracted outdoor
space, all touching faces with the actual building can be extracted without further
geometric computations resulting in the outer shell of the building. All different types
of touching faces can be determined. Therefore, the parts of the outer shell resulting
Identifying Indoor and Outdoor Spaces in Digital Building Models 411

Fig. 8 Left: cut through the example shown in Fig. 7. The red highlighted space cannot be created
as a single room or zone in ArchiCAD or Revit. Right: the automatically extracted indoor space of
the space partition.

Fig. 9 Left: The outer shell of the building extracted from the space partition. Right: Topological
relations between the extracted indoor spaces and all opening elements.

from an opening can be distinguished from parts resulting from walls or the roof.
The underlying tetrahedral mesh can be used to extract topological relations between
indoor and outdoor spaces. In Fig. 9, all resulting topological relations between indoor
spaces and openings are visualized as an example. These topological relations can be
used to create a connectivity graph of the building and its indoor and outdoor spaces.

6 Discussion, Conclusion and Outlook

This paper suggested transforming a given boundary representation of a model into

a tetrahedral space partition to model indoor and outdoor spaces in building models.
The resulting tetrahedral mesh stores topological relations explicitly. These topo-
logical relations can be a basis for extracting valuable information for various use
cases.
412 J. Z. Vetter and W. Huhnt

This paper also showed current problems in state-of-the-art modeling software

tools considering the representation of indoor and outdoor spaces. It is shown that
these issues do not exist while using the suggested transformation approach and a
tetrahedral space partition.
The suggested approach is a holistic way to detect indoor and outdoor spaces for
input models. However, correctly detecting these spaces relies on correctly modeling
the input boundary representations. Once transformed, the spaces are detected by
collecting a set of connected tetrahedrons that are not assigned to any building objects
for each indoor space. Therefore, the approach requires the indoor spaces to be
separated (e.g., by boundary representations of opening elements). If the indoor
spaces are not properly separated in the input model, they cannot be detected correctly.
Additionally, only one set of connected tetrahedrons represents each indoor space.
The partition of these indoor spaces into smaller subsets is possible but requires the
addition of separation planes as input constraints.
Although this paper suggested an approach to model indoor and outdoor spaces
in building models, usable output format exports are missing. The main focus of this
paper was to validate the use of the suggested approach by identifying problems in
software tools and showing how to overcome these problems. This research lays a
foundation for the following research to export the collected information stored in
the tetrahedral mesh to connect the suggested approach with the presented use cases.
As stated in Sect. 5, the presented examples only show why the presented approach
overcomes the problems of the state-of-the-art modeling software tools. As a next step
in validating the suggested approach, more extensive examples must be transformed
to show the approach’s applicability for realistic building models.

References

1. Arroyo Ohori K, Diakité A, Krijnen T, Ledoux H, Stoter J (2018) Processing BIM and GIS mod-
els in practice: experiences and recommendations from a GeoBIM project in The Netherlands.
ISPRS Int J Geo-Inf 7(8):311. https://doi.org/10.3390/ijgi7080311
2. Autodesk (2022) Remove minimum room dimension. Autodesk Knowledge Net-
work. https://forums.autodesk.com/t5/revit-ideas/remove-minimum-room-dimension/idc-p/
7843358/highlight/true. Accessed 11 July 2022
3. Autodesk (2022) Situations That Can Affect Room Volume Computations. Autodesk
Knowledge Network. https://knowledge.autodesk.com/support/revit/learn-explore/caas/
CloudHelp/cloudhelp/2021/ENU/Revit-ArchDesign/files/GUID-052DFFB9-5076-4CE6-
BC8B-420C8D03ACF0-htm.html. Accessed 11 July 2022
4. buildingSMART (2022) IFCRelSpaceBoundary. buildingSMART International Stan-
dards Server. https://standards.buildingsmart.org/IFC/DEV/IFC4_3/RC1/HTML/schema/
ifcproductextension/lexical/ifcrelspaceboundary.htm. Accessed 13 Sept 2022
5. Deng Y, Cheng JC, Anumba C (2016) Mapping between BIM and 3D GIS in different levels of
detail using schema mediation and instance comparison. Autom Constr 67:1–21. https://doi.
org/10.1016/j.autcon.2016.03.006
6. Diakité AA et al (2022) IFC2INDOORGML: an open-source tool for generating IndoorGML
from IFC. Int Arch Photogram Remote Sens Spatial Inf Sci XLIII-B4-2022:295–301. https://
doi.org/10.5194/isprs-archives-XLIII-B4-2022-295-2022
Identifying Indoor and Outdoor Spaces in Digital Building Models 413

7. Dilhas S (2022) Yesterday’s post about my suggestions for modeling double-height

spaces in BIM tools was controversial. [...] [Image attached][Post]. LinkedIn. https://
www.linkedin.com/posts/simon-dilhas_yesterdays-post-about-my-suggestions-for-activity-
6943063745497604096-s1S1/. Accessed 13 Sept 2022
8. Donkers S, Ledoux H, Zhao J, Stoter J (2016) Automatic conversion of IFC datasets to geo-
metrically and semantically correct CityGML LOD3 buildings: automatic conversion of IFC
datasets to CityGML LOD3 buildings. Trans GIS 20(4):547–569. https://doi.org/10.1111/tgis.
12162
9. Huhnt W (2018) Reconstruction of edges in digital building models. Adv Eng Inform 38:474–
487. https://doi.org/10.1016/j.aei.2018.08.004
10. Isikdag U, Zlatanova S (2009) Towards defining a framework for automatic generation of
buildings in CityGML using building information models. In: Lee J, Zlatanova S (eds) 3D
geo-information sciences. Springer, Heidelberg, pp 79–96. https://doi.org/10.1007/978-3-540-
87395-2_6
11. Kang TW, Hong CH (2018) IFC-CityGML LOD mapping automation using multiprocessing-
based screen-buffer scanning including mapping rule. KSCE J Civil Eng 22(2):373–383. https://
doi.org/10.1007/s12205-017-0595-9
12. Kraft B (2016) Ein Verfahren der Raumzerlegung als Grundlage zur Prüfung von Geometrie
und Topologie digitaler Bauwerksmodelle. PhD thesis, TU Berlin
13. Lilis GN, Giannakis GI, Rovas DV (2017) Automatic generation of second-level space bound-
ary topology from IFC geometry inputs. Autom Constr 76:108–124. https://doi.org/10.1016/
j.autcon.2016.08.044
14. Lin WY, Lin PH (2018) Intelligent generation of indoor topology (i-GIT) for human indoor
pathfinding based on IFC models and 3D GIS technology. Autom Constr 94:340–359. https://
doi.org/10.1016/j.autcon.2018.07.016
15. Pang Y, Zhang C, Zhou L, Lin B, Lv G (2018) Extracting indoor space information in complex
building environments. ISPRS Int J Geo-Inf 7(8):321. https://doi.org/10.3390/ijgi7080321
16. Penninga F (2008) 3D Topography: A Simplicial Complex-based Solution in a Spatial DBMS.
PhD thesis, Technische Universiteit Delft
17. Romanschek E, Clemen C, Huhnt W (2020) From terrestrial laser scans to a surface model
of a building; proof of concept in 2D. In: Ungureanu LC, Hartmann T (eds) EG-ICE 2020
workshop on intelligent computing in engineering, pp 432–442. https://depositonce.tu-berlin.
de/handle/11303/11088
18. Romanschek E, Clemen C, Huhnt W (2021) A novel robust approach for computing DE-9IM
matrices based on space partition and integer coordinates. ISPRS Int J Geo-Inf 10(11):715.
https://doi.org/10.3390/ijgi10110715
19. Teo TA, Yu SC (2017) The extraction of indoor building information from BIM to OGC
IndoorGML. Int Arch Photogram Remote Sens Spatial Inf Sci XLII-4/W2:167–170. https://
doi.org/10.5194/isprs-archives-XLII-4-W2-167-2017
20. Teo TA, Cho KH (2016) BIM-oriented indoor network model for indoor and outdoor combined
route planning. Adv Eng Inform 30(3):268–282. https://doi.org/10.1016/j.aei.2016.04.007
21. Vetter JZ, Huhnt W (2021) Accuracy aspects when transforming a boundary representation
of solids into a tetrahedral space partition. In: Abualdenien J, Borrmann A, Ungureanu LC,
Hartmann T (eds) EG-ICE 2021 workshop on intelligent computing in engineering, pp 320–
329. https://depositonce.tu-berlin.de/handle/11303/13226
22. Xu M, Wei S, Zlatanova S, Zhang R (2017) BIM-based indoor path planning considering
obstacles. ISPRS Ann Photogram Remote Sens Spatial Inf Sci IV-2/W4:417–423. https://doi.
org/10.5194/isprs-annals-IV-2-W4-417-2017
23. Ying H, Lee S (2021) A rule-based system to automatically validate IFC second-level space
boundaries for building energy analysis. Autom Constr 127:103724. https://doi.org/10.1016/
j.autcon.2021.103724
Space Partitioning as a Holistic
Alternative to Traditional Geometric
Modeling Workflows in the AEC
Industry

Wolfgang Huhnt , Joanna Zarah Vetter , and Maximilian Sternal

Abstract The need for data is indisputable in all fields of architecture and engi-
neering. It is also state-of-the-art to differentiate kinds of data, e.g. geometric data
and semantics. Modern engineering workflows are characterized by their high inter-
operability between different fields of application. Almost all architecture and civil
engineering tasks have in common that they need geometric data. Task-specific data
formats are defined proprietorially in each software tool and standardized for differ-
ent fields of application such as Industry Foundation Classes, CityGML or gbXML.
Thus, model transformation is a frequently performed activity. State-of-the-art mod-
eling tools focus on representing object shapes and their transformation in space.
Neighboring relations, which are treated in topology, are relevant in many engineer-
ing applications but neglected in many modeling tools and data formats. Another
approach in geometric modeling, named space partitioning, focuses on the complete
space. Space partitioning itself is not a new approach. However, its application to
architecture and civil engineering geometry and shape representations has not been
investigated in detail. This paper discusses fundamental aspects of space partitioning
as an alternative approach to model geometry in architecture and civil engineering.
The goal is to overcome existing challenges and to provide the basics for the stream-
lined alignment of domain models using space partitioning as central geometry.

Keywords Geometric Modeling · Geometric Models · Space Partitioning ·

Topology

1 Introduction

The geometric representation of building objects is a central task in the AEC industry.
Building Information Modelling (BIM), Computer-Aided Design (CAD) and Engi-
neering (CAE) workflows and software tools focus on modeling building objects.

W. Huhnt (B) · J. Z. Vetter · M. Sternal

Technische Universität Berlin, Berlin, Germany
e-mail: wolfgang.huhnt@tu-berlin.de
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 415
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_29
416 W. Huhnt et al.

A common way to describe volumetric elements and their shapes is the definition of
their boundary. This boundary constitutes separating the interior and exterior points
of the represented object. It can consist of planar or curved faces, edges and points in
the 3-dimensional space. While all boundary representations have an inner topology
in common, topological neighborhood relations between objects are conventionally
not considered. Shapes are created using local coordinate systems and transformed
to a specific position in space. Manual issue management and heavy geometric algo-
rithms are the consequence to overcome this problem.
The concept of space partition is a way to represent the overall topology of a model.
Space partition is used to describe and divide the complete space by using different
subspaces of various dimensions. These domains are conventionally called vertices,
edges, faces and volumes in 3-dimensional applications. From a mathematical point
of view, a topology classifies the space: (1) Each domain is not empty, (2) each pair
of different domains do not intersect and (3) the union of all domains is the complete
space. Typical data structures often used in computer graphics for partitioning space
are the Octree or the Binary Space Partition. Different examples of applications of
these data structures exist in the civil engineering field, Solihin et al. [13] using Octree
decomposition of space for fast spatial queries in building models and Sacks et al.
[12] propose R-tree space indexing as preprocessing for rule inference procedure.
While space partition can be used for indexing and fast search in space, its charac-
teristics can also be combined with the boundary description of objects. A connection
of both concepts, the representation of the geometry of objects and the integration
of topology in the complete space, has the potential to function as a holistic way to
represent building models. Fundamental research in this field has been conducted,
for instance, by Walter Nef [8], who described the theory for polyhedron shapes and
provided a first data structure considering Boolean operations of set intersection and
set complement. Different research exists that combines the use of space partition and
geometric representation of objects in civil engineering and architecture applications.
Weiler introduced the Radial Edge data structure [17], which addressed the topology
of the complete space and can be considered the groundwork for many Half-Edge
data structures such as Partial Edge [7] or Dual-Half-Edge [2]. While these concepts
are based on polyhedrons, Penninga [10] used tetrahedrons as topological manifolds
to reach a space partition to represent city models in combination with Triangulated
Irregular Networks (TIN). While this work thoroughly investigates the application
of this concept for city models, the integration of building models is not covered
sufficiently.
Creating a valid space partition for digital building models can be differentiated
into two different approaches. The first approach is transforming a given boundary
representation into a space partition. Examples of this approach are given in the field
of computer graphics, such as by Hu et al. [4]. This example creates a tetrahedral space
partition by transforming a given triangle soup. While this approach has been proven
to be robust for single objects, the application of this concept to building models is
not yet evaluated. Alongside transforming geometry into a space partitioning, the
need to model and modify a valid partition arises. Therefore, the second approach
for creating a space partition for digital building models is to develop operators for
Space Partitioning as a Holistic Alternative to Geometric Workflows 417

the step-by-step modeling process of building objects embedded in a partition of the

complete space. A similar concept is implemented by Jabi [5] in the “topologic.app”
kernel. While this tool can represent dimensionally reduced geometries for room
topologies with a wide application for architects, modeling building objects has
not been investigated in-depth. Furthermore, the unbounded space and multiply-
connected subspaces are not explicitly defined. Thus, it does not constitute a complete
space partition.
The different mode of looking at geometry has huge consequences. A lot of
research addresses model transformation. Different application fields and problems
of the transformation of geometric models are discussed in Sect. 2. Section 3 presents
characteristics of space partition models to explain the presented approaches. In
Sect. 4, examples illustrate the power of space partitions and it is shown how the
application fields introduced in Sect. 2 could benefit by using space partition instead
of a transformation approach. The paper ends with a discussion, conclusions and an
outlook in Sect. 5 and 6.

2 Transformation as the Approach to Reuse Data

Creating and passing geometric descriptions of a specific design is a key factor

in engineering communication and essential in every planning process. While a
physical object only has one geometry, a modeled geometry of an object might have
an abstracted, simplified geometric form constructed for the unique purpose of the
specific domain model. While the level of detail is a general factor for all kinds of
geometric models, domain models often use different domain-specific data structures
that fulfill the intrinsic requirements of the intended engineering application, such
as energy calculation, structural analysis or indoor and outdoor navigation and city
modeling.
To pass on geometric data to other designers and applications, geometric descrip-
tions must be transformed during export and import processes, leading to expensive
algorithms, robustness issues and information loss. Standardized data formats such
as the “Industry Foundation Classes” (IFC), which are based on the general “Stan-
dard for the Exchange of Product model data” (STEP) offer a vast number of possible
geometric data structures, both implicit, e.g. Constructive Solid Geometry (CSG),
and explicit, e.g. triangulation or Boundary Representation (BRep), but consistency
or transformation functionality heavily lies with the domain software implementa-
tions of this huge standard that additionally features rich semantic data. Therefore,
bidirectional interfaces are common practice in many workflows. On the other hand,
vendor-specific and open source cloud platforms for exchanging geometry are on
the rise recently, such as Nemetschek “BIMPlus”, Autodesk “FORGE” [1, 9] or
open source alternatives such as “SPECKLE” or the “Buildings and Habitats object
Model” (BHoM) [14, 16]. While these systems provide great features for interop-
erability and outsource some of the geometric transformation needed in workflows,
their geometric kernels and data structures use the same principles as provided with
418 W. Huhnt et al.

standards such as IFC and STEP and do not provide capabilities to reduce transfor-
mations and increase robustness significantly.
Future requirements for circular economy and data management in the opera-
tional phase of a building or infrastructure asset (“digital twin”) demand a holistic
view of building data in general and geometric data in particular. Therefore, a vast
number of projects in recent years discussed aligning building and urban models as
well as indoor and outdoor models [18] and cadastral data [6, 15]. A common and
reliable geometric description is essential for interoperability in these application
fields. A high number of transformations is detrimental to the robustness and the
consistency of geometric project information. The utilization of space partitioning
can provide a general and comprehensive platform for a vast number of geomet-
ric applications, thereby reducing error-prone transformation processes and making
topological relations available, which are key factors for many engineering tasks.

3 Space Partitioning as an Alternative Approach

In order to understand the advantages and particularities of the space partition-

ing approach for representing geometric models, this section describes fundamen-
tal basics of space partitioning in general and the characteristics of the proposed
approach in particular. Well-known concepts are compared to the presented approach
and important properties are highlighted and explained.

3.1 Basics

The term space partition is sometimes used for specific data structures such as Octree,
Quadtree, Binary Space Partition or kd-tree. These data structures allow efficient
access to geometric objects so that, for instance, neighbors can be identified with a
small number of operations. Nevertheless, the representation of geometric shapes is
not of interest in the typical applications of these data structures.
Space partitioning, as treated in the context of this paper, requires a different mode
of thinking. It addresses the complete space. Internally, data structures such as the
Half-Edge and Half-Face data structures can store neighboring relations explicitly.
The consequence is that the topology is stored by pointers. Topology is explicitly
available and can be processed efficiently.
Figure 1 shows a rectangle in the 2-dimensional space. A boundary representation
of the rectangle stores four vertices and four edges. The face of the rectangle is
determined by the sequence of the points on the boundary and a rule that the interior
of the represented element is on the left side. Looking to the same figure from a space
partitioning point of view results in ten domains due to the fact that the unbounded
exterior of the rectangle and the face of the rectangle are the 2-dimensional domains.
Space Partitioning as a Holistic Alternative to Geometric Workflows 419

Fig. 1 Left: boundary representation of a rectangle. Right: space partition of the complete space
including a rectangle

Fig. 2 Example of a manifold face left figure and a non-manifold face right figure in the 2-
dimensional space

As a prerequisite, all domains of dimension .n must have a manifold boundary of

dimension .n − 1. Figure 2 shows two geometric models of a face. The right model
is not manifold. The connection to the opening violates manifoldness. Manifoldness
ensures that the partition is a topology and fulfills the mathematical conditions of a
classification. The left face is 1-manifold.
Tetrahedral meshes and triangular meshes always fulfill the condition of mani-
foldness. A tetrahedron is a convex 3-dimensional simplex; a triangle is a convex
2-dimensional simplex. Meshes are widely used in architecture and civil engineering
for visualization purposes and as a basis for numerical methods. However, a mesh
of an object typically does not include the exterior. Meshes that include the exterior
fulfill the conditions of space partitioning. While using polyhedrons in meshes, the
manifoldness must be ensured.

3.2 Characteristics

Topology: The prerequisite concerning manifoldness, introduced in Sect. 3.1, ensures

that the partition classifies the space:
– Each domain is a subset of the space and not empty
– Each pair of different domains do not intersect
420 W. Huhnt et al.

Fig. 3 A 2-dimensional example represented by a space partition with its connectivity graph

– The union of all domains is the complete space

The topology is the fundamental part of the introduced concept of using space
partition. This topology can be used to extract information from the space partition
and the represented shapes. Each partition can be mapped onto a graph. In the 3-
dimensional space, volume elements can be regarded as vertices and their neighboring
relations where two volume elements share a common face as edges of the graph.
In 2-dimensional spaces, faces are vertices of the graph. Their neighboring relation,
where two faces share a common edge, are the edges of the graph. An example is
shown in Fig. 3.
Bounded and Unbounded Domains: It is necessary to introduce bounded and
unbounded edges, faces and volume elements. Unbounded domains are required
to consider the complete space.
In the literature, modeling concepts sometimes use bounding boxes for modeling
the exterior. The term axis aligned bounding box (AABB) describes a bounding
box that is parallel to the faces of the coordinate system. The disadvantage of such
concepts is that they must be treated in special ways. For instance, a special treatment
of the boundaries of such artificial objects is necessary to avoid reflections in the
computation of wave propagation in soil mechanics.
In addition, the unbounded exterior of buildings is of interest. During the construc-
tions of infrastructure like railways, there are often restrictions regarding the space
above these railways. For instance, in Germany it is not allowed to work above rail-
ways on which trains operate. Additionally, the unbounded space above landmarks
is of interest in geodesy.
Convex and Non-convex Domains: The geometry of buildings is not always convex.
Faces are often triangulated, and volumes are often modeled as sets of tetrahedrons.
These meshes consist of convex domains to describe the geometry of non-convex
objects. The algorithms for triangles and tetrahedrons are simple compared to non-
convex domains. On the other hand, a set of triangles and/or tetrahedrons is necessary
instead of a single polygon or polyhedron for the description of a single object.
Space Partitioning as a Holistic Alternative to Geometric Workflows 421

Fig. 4 Geometries of
clashes in the given input
model of Fig. 6 identified by
using space partition

Fig. 5 Geometries of
touches between the building
objects in the given input
model of Fig. 6 identified by
using space partition

Singly- and Multiply-Connected Domains: The geometry of buildings requires the

possibility to model multiply connected domains. The wall in Fig. 3 is an example of
a multiply-connected face. The window in Fig. 3 is an example of a singly-connected
face.

4 Fields of Application

This section describes fields of application. All applications are based on the funda-
mental property of the space partition that it classifies the space and that all neigh-
boring relations are explicitly available, as described in Sect. 3.
Clashes and Touch Detection: Conventional clash detection is reached by testing
surface triangles of boundary representations against each other geometrically. In
contrast, clashes and touches can be easily identified once the model is transformed
into a space partition. Additionally, information about the geometry of a clash or a
touch can be derived from the space partition as well.
Figures 4 and 5 show examples, calculated in an ongoing research project called
ROF that is funded by the German Research Foundation (Project-No.: 424641234)
about transforming given boundary representations into tetrahedral space partitions.
These figures show identified contacts and clashes of the given input, which was
transformed into a space partition.
422 W. Huhnt et al.

Fig. 6 Input boundary

representations of a storey in
an example model

Fig. 7 Automatically
identified indoor spaces in
the given input of Fig. 6 of a
space partition

Empty Space: Another advantage of representing digital building models with a

space partition data structure is the integration of the empty spaces in the model.
While only building objects are part of traditional geometric representations, empty
spaces are an integral part of a space partition. Empty spaces can be rooms that are
not represented by a geometric object in the interior of a building, unintended gaps
or the exterior. This advantage can be used for the automatic recognition of indoor
and outdoor spaces in building models. Two different ways of dealing with indoor
spaces in models can be identified. Figures 6 and 7 show an example which also
results from the ROF project. On the one hand, it is possible to automatically detect
indoor spaces in space partitions from scratch. On the other hand, it is possible to
validate existing modeled spaces by finding any leftover gaps in a given model by
transforming this model and all modeled spaces into a space partition.
Once correctly identified, different application fields can benefit from the knowl-
edge of indoor and outdoor spaces in building models. The explicit storage of topo-
logical information in the space partition enables the extraction of space boundaries.
Energy modeling data structures like gbXML require these first- and second-level
space boundaries. These different types of boundaries are shown in Fig. 8.
Detection of Air and Solids from Laser Scans: Point cloud processing can be
improved by the information that there is air between the scanning device and
a point, which reflects the laser [11]. This information can be processed so that
in 2-dimensional space, a set of triangles describes identified areas from air. One
vertex of each triangle is the location of the scanning device. The opposite edge
Space Partitioning as a Holistic Alternative to Geometric Workflows 423

Fig. 8 First- left and second-level space boundaries right as defined in IFC [3]

separates air from material. In 3-dimensional space, a set of tetrahedrons describes

identified air.
Connectivity and Navigation: Topology can be interpreted in different spaces. One
typical example for different topologies is the theory of duality, where e.g. rooms in
a building are reduced to 1-dimensional vertices, building a connectivity graph with
edges, connecting different rooms. While such graph data structures must be addi-
tionally modeled with classical geometric descriptions, graph data is easily inferable
from a space partitioning model. An inferred graph generated from a space parti-
tioning prototype implementation of a building complex is shown in Fig. 9. Typical
examples of connectivity graphs in architecture and engineering are found in energy
modeling and indoor and outdoor navigation. Data structures such as IndoorGML
explicitly store topological graph data, while primal geometries of objects and spaces
are optional.
Furthermore, semantic web graphs of connectivity can be represented using the
Building Topology Ontology (BOT), offering a minimal data structure to represent
rooms and elements in regard to their neighborhood relations. An exported and
visualized BOT graph, showing room and wall connectivity, is shown in Fig. 10.
Because unbounded space is explicitly modeled in the space partitioning approach,
it can easily deal as a link model for inside and outside information and feature
cadastral and city models. Parcels in cadastral models can be modeled unbounded

Fig. 9 Generated dual

connectivity graph from a
space partitioning model
424 W. Huhnt et al.

Fig. 10 Exported topology

as semantic web graph in
sparql-visualizer

to include air space above a building or infrastructure to fit the needs of modern
integrated 3D cadaster.

5 Discussion

There are several advantages in describing buildings in space partitioning models

with explicit neighboring relations. First of all, each geometrical and topological
request can be answered in an efficient way. Navigation is simple and free from any
inaccuracy. Each volume element can be extracted by extracting its boundary. Its
selection is simple and does not require time-consuming calculations. Clashes and
voids are avoided by the nature of the space partitioning approach. Each contact
between domains is modeled once. There is no gap and no overlap in such a model.
Unbounded objects are necessary to fulfill specific requirements as described
above. In addition, dimensionally reduced models such as skeletons are possible
using this approach. The building components can be modeled by 1-dimensional
and 2-dimensional domains. Spaces are modeled as cells. Such models are used, for
instance, in energy calculations.
It is possible to model parts of a space partition in different level of details. This
can be achieved in a simple way if the interior of a domain is detailed keeping its
boundary unmodified. Such a model can be regarded as hierarchical graph where the
hierarchies describe different levels of detail.
A challenge facing the acceptance of this approach is that space partition models
require another mode of thinking and modeling. Architects and civil engineers are
not trained in modeling faces which separate their objects of interest. However, this
disadvantage can be avoided by suitable user interfaces.
Another mode of modeling always requires the implementation of new model-
ing functionalities. This is true for space partitioning as well. An ongoing research
project called PPSpace, funded by the German Research Foundation (Project No.
429900376), tries to close this gap and develops specific modeling operators and spa-
tial queries for navigation. The goal is the implementation of the core functionality
required for a robust modeling of a polyhedral space partition.
Space Partitioning as a Holistic Alternative to Geometric Workflows 425

6 Conclusion and Outlook

This paper presented basic principles of space partitioning as a holistic way of mod-
eling geometries relevant in the architecture and civil engineering industry. Because
many fields of application demand specific characteristics of geometrical models,
geometric transformation has been highlighted as a central task in modern workflows.
A crucial feature missing is the topology, describing spatial and neighborhood rela-
tions within a model. Furthermore, empty spaces are often ignored or even produced
by transformation. These disadvantages can be overcome by a space partitioning
approach.
Digitization of authorities and many workflows in planning and operating building
and infrastructure assets increase the requirements for geometric models. In the
future, data must be linked and consistent. Data silos in the form of legacy data
formats such as IFC will prevent seamless cooperation, especially when it comes to
geometrical information. Explicitly modeled topology will enhance any query to a
building model substantially. Furthermore, the growing trend of separating geometric
and semantic information using more generic data formats such as semantic web
technologies strengthens the need for connected, complete and correct geometric
data models.
Space partitioning models can support this necessity and can be applied when
transforming, modeling or scanning geometries is needed. 3-dimensional cadaster to
manage land use and the alignment of geo and building information modeling can
benefit from the use of geometric representation in the form of space partitioning.
However, extensive research is still necessary to benefit from the advantages
of space partitioning models. Suitable user interfaces are necessary for modeling
geometries based on this approach. Existing models need to be transformed. A sim-
ple standard is necessary to access space partitioning models. A link to semantic
data is necessary so that geometric domains can be enriched by semantics. In the
end, products such as windows, doors, etc. must be available in a space partitioning
model so that users can easily model their buildings using data from built-in compo-
nent vendors at any required level of geometry. This list of required research is not
complete. We are in front of fascinating tasks to apply space partitioning as a holistic
approach for the architecture and civil engineering industry.

References

1. AUTODESK: Autodesk forge (2022). https://forge.autodesk.com/developer/documentation.

Accessed 29 Sept 2022
2. Boguslawski P, Gold C (2016) The dual half-edge-a topological primal/dual data structure and
construction operators for modelling and manipulating cell complexes. ISPRS Int J Geo Inf
5:19. https://doi.org/10.3390/ijgi5020019
3. buildingSMART: IfcRelSpaceBoundary. buildingSMART international standards
server (2022). https://standards.buildingsmart.org/IFC/DEV/IFC4_3/RC1/HTML/schema/
ifcproductextension/lexical/ifcrelspaceboundary.htm. Accessed 29 Sept 2022
426 W. Huhnt et al.

4. Hu Y, Schneider T, Wang B, Zorin D, Panozzo D (2020) Fast tetrahedral meshing in the wild.
ACM Trans Graph 39(4):117:1-117:18. https://doi.org/10.1145/3386569.3392385
5. Jabi W, Chatzivasileiadi A (2021) Topologic: exploring spatial reasoning through geometry,
topology, and semantics. In: Eloy S, Leite Viana D, Morais F, Vieira Vaz J (eds) Formal methods
in architecture. Advances in science, technology & innovation. Springer, Cham, pp 277–285.
https://doi.org/10.1007/978-3-030-57509-0_25
6. Jaljolie R, Riekkinen K, Dalyot S (2021) A topological-based approach for determining spatial
relationships of complex volumetric parcels in land administration systems. Land Use Policy
109:105637. https://doi.org/10.1016/j.landusepol.2021.105637
7. Lee SH, Lee K (2001) Partial entity structure: a compact non-manifold boundary representation
based on partial topological entities. In: Proceedings of the sixth ACM symposium on solid
modeling and applications, SMA 2001. Association for Computing Machinery, New York, pp
159–170. https://doi.org/10.1145/376957.376976
8. Nef W (1978) Beiträge zur Theorie der Polyeder: mit Anwendungen in der Computergraphik,
vol 1. Herbert Lang, Bern
9. NEMETSCHEK: Allplan Bimplus Bauprojekte erfolgreich durchführen (2022). https://www.
allplan.com/de/produkte/allplan-bimplus/. Accessed 29 Sept 2022
10. Penninga F (2008) 3D topography: a simplicial complex-based solution in a spatial DBMS.
No. 66 in publications on geodesy, Nederlandse Commissie voor Geodesie, Delft
11. Romanschek E, Clemen C, Huhnt W (2020) From terrestrial laser scans to a surface model
of a building; proof of concept in 2D. In: Ungureanu LC, Hartmann T (eds) EG-ICE 2020
workshop on intelligent computing in engineering, pp 432–442. https://depositonce.tu-berlin.
de/handle/11303/11088
12. Sacks R, Ma L, Yosef R, Borrmann A, Daum S, Kattel U (2017) Semantic enrichment for
building information modeling: procedure for compiling inference rules and operators for
complex geometry. J Comput Civ Eng 31(6). https://doi.org/10.1061/(ASCE)CP.1943-5487.
0000705
13. Solihin W, Eastman C, Lee YC (2017) Multiple representation approach to achieve high-
performance spatial queries of 3D BIM data using a relational database. Autom Constr 81:369–
388. https://doi.org/10.1016/j.autcon.2017.03.014
14. Speckle: About (2021). https://speckle.systems/about/. Accessed 29 Sept 2022
15. Tekavec J, Čeh M, Lisec A (2021) Indoor space as the basis for modelling of buildings in a 3D
Cadastre. Surv Rev 53(380):464–475. https://doi.org/10.1080/00396265.2020.1838761
16. The BHoM: the buildings and habitats object model (2022). https://bhom.xyz. Accessed 29
Sept 2022
17. Weiler K (1988) The radial-edge structure: a topological representation for non-manifold geo-
metric boundary representations. Geometric modeling for CAD applications. North-Holland
18. Yan J, Zlatanova S, Diakité A (2021) A unified 3D space-based navigation model for seamless
navigation in indoor and outdoor. Int J Digit Earth 14(8):985–1003. https://doi.org/10.1080/
17538947.2021.1913522
Robust Modeling of Polyhedral Space
Partitions

Maximilian Sternal and Wolfgang Huhnt

Abstract This paper introduces polyhedral space partition as a representation of

geometric objects in digital building models. Operators to modify such a partition
model are presented and strategies to guaranty robustness are explained. The topol-
ogy of a polyhedral space partition can be stored explicitly by pointers in a digital
environment. However, coordinates are mapped to numbers which are unavoidably
imprecise in a digital environment. The partition model is called consistent, if its
imprecise geometric attributes do not contradict its exact topological attributes. The
platform is called robust if the imprecision in typical applications does not make
the model inconsistent. This paper presents two novel concepts that enhance robust
modeling: focused model construction and topologically controlled work steps. Con-
ventionally, each object is constructed individually and added to the existing set of
objects. Geometric imprecision leads to collisions of and gaps between objects. The
algorithms that detect collisions or gaps are not focused on a subset; and they are
expensive because they must consider all existing objects. Focused transformations
affect only a small number of domains. A work step is topologically controlled, if
modifications on boundary elements of the domain must be performed in preced-
ing work steps. Errors are reported and corrected before the construction proceeds.
Topological checks are performed on the manifolds based on a rank concept. It is
tested geometrically that all domains that are added to a domain lie inside the existing
domain using the novel concept of anchors and their clients for robustness. Trun-
cation and round-off errors are treated with a single value subjected to the machine
epsilon.

Keywords Geometric Modeling · Robustness · Topology · Space Partitioning ·

Building Information Modeling

The authors thank em. Prof. Dr. Dr. h.c. mult. Peter Jan Pahl, TU Berlin, Germany for his exten-
sive and excellent contributions to the space partitioning project PPSpace. The research is funded
by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) - Projektnummer
429900376.

M. Sternal (B) · W. Huhnt

Technische Universität Berlin, Berlin, Germany
e-mail: maximilian.sternal@tu-berlin.de

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 427
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_30
428 M. Sternal and W. Huhnt

1 Introduction

Digital geometric models constitute a basis for many tasks and analyses in architec-
ture, engineering and construction. Therefore, a vast number of different modeling
techniques and data models have emerged to create and store digital geometry since
the invention of the first computer-aided drafting and design tools. Traditionally,
many of these systems focus predominantly on the description of single objects such
as building components, which are specified by parameters, transformed and placed
in space. The description of their topology is often neglected or restricted to the
intrinsic topology of one component. Space partitioning, on the other hand, focuses
on the description of the complete space. Unbounded Euclidean space is chosen as
the complete space that contains all objects. The operators are designed in such a
way that each work step transforms a valid space partition into another valid space
partition. If errors are detected, they are reported and the work step is aborted.
This paper presents robust modeling operators for a polyhedral partition of
Euclidean space. The extrinsic topological relations between objects are described
explicitly in the digital model with exact object names. The geometry of the partition
is specified with the necessarily imprecise coordinates of its nodes and the direc-
tion vectors of its unbounded edges. Because the topological model is exact and the
geometric model is imprecise, the two models are necessarily inconsistent. The mod-
elling approach is called robust if the imprecision that occurs in typical applications
does not make the topology and the geometry of the digital model inconsistent.
The model construction is called focused, if each work step modifies only a small
number of the model domains. In order to achieve focus, a rank equal to its dimen-
sion is assigned to each domain. The two basic operations for model construction
are split and merge. In a split, a single domain of the partition is replaced by exactly
two domains of the same rank. In a merge, two domains of equal rank are replaced
by a single domain of the same rank. The conditions that are imposed on rank lead to
topological control of the modelling. The atomic topological and geometric opera-
tions on the model in a work step are restricted to the attributes of the single domain
that is split, or the two domains that are merged. All higher modeling procedures are
implemented with a topologically controlled sequence of split and merge operations.
In the space partitioning approach, the input commands are performed on domains
of a user model, which consists of nodes as well as edges, faces and cells. The
domains can be bounded or unbounded. The operations that are performed on the
user model are arranged in work steps. The domains in the user model are not oriented.
Internally, a core model automatically constructs and stores oriented arrows, facets
and voxels (volume elements). The user does not come into contact with the core
model, but indirectly makes use of its functionality for model construction and for
navigation in the model. The partition concept includes efficient checks of the location
of neighboring domains. Topological checks are performed on the manifolds. It is
tested geometrically that the domains, which are added to split a domain, lie inside the
existing domain. In order to reduce the effects of numerical inaccuracy, conventional
Robust Modeling of Polyhedral Space Partitions 429

methods that require solutions of sets of equations are replaced by the computation
of distances and the novel concept of anchors and their clients.
The theory and the algorithms for polyhedral partitions are designed to treat non-
convex, multiply-connected and unbounded domains without user assistance. Topo-
logical errors are detected, allowing only closed and connected manifolds. Geometric
errors, such as intersections and gaps, are recognized and reported. The implemen-
tation makes use of floating-point variables to store coordinates. Thus, numerical
imprecision occurs. The concept introduces a single value dependent on the machine
epsilon and on the overall model size in order to describe the geometric surrounding
of domains. All input domains, which are intended to be inside another domain, must
lie in the surrounding of that domain, while no other domain is allowed to lie in the
surrounding. The geometric checks make use of the concepts of anchors and client
zones to guarantee robustness in all dimensions.
Section 2 introduces basic components of the polyhedral space partitioning
approach that is the platform for the robust operators. The approach is compared with
existing edge-based data structures and important features are highlighted. Subse-
quently, the operators for split and merge operations on domains are introduced in
Sect. 3, before the concept of robust modeling in the context of topologically con-
trolled work steps is introduced. The concepts for robust geometric modeling are
evaluated in Sect. 4 by comparing their theories and their implementations. Check-
ing procedures are presented separately for topological and geometrical checking in
Sect. 5.
The paper contains some examples that illustrate the functioning of the operators.
It ends with a discussion of the benefits of the proposed approach compared to other
approaches in robust geometric modeling and an outlook to fields of application for
operators that model space partition.

2 Polyhedral Partition of Space

2.1 Existing Space Partition Concepts

Space partitioning divides a space into subspaces. Intersection tests and other geom-
etry queries on large models are performed faster by preprocessing with tree-based
methods and data structures such as binary space partitioning and octrees. Every
point in space can be uniquely allocated to one leaf in the tree. These features have
been utilized in various projects for point cloud processing [9] or efficient database
access [14] and rule-based geometric queries [2].
In general, the partition of a metric space is called a classification of space, if no
subset is empty, none of the subsets intersect and the union of all subsets is the entire
space. Paul [10] describes in detail, how the mathematical foundation of topology
and continuous functions can enhance the description of geometric shapes using
space partitioning.
430 M. Sternal and W. Huhnt

Fig. 1 Domains in a polyhedral space partition. Two cells share one common face (blue). The face
itself is bordered by edges that share common nodes

2.2 Existing Topological Data Structures

Most data structures designed for boundary representation store topological infor-
mation about the boundary of a single object only. In addition, data structures have
been developed to describe neighborhood relations between objects placed in space.
In the past, many data structures for the topological properties of planar, polygonal
geometries have been proposed, spanning from radial-edge [16] and partial-edge
[7] data structures up to the more recently proposed dual half-edge [1] and compact
abstract cell complexes (CACC). Commonly, such edge-based data formats use topo-
logical pointers to traverse along the (co)edges of a boundary. Common examples
for such pointers are “next”, “previous” in a loop and “mate” or “twin” to walk along
the surface of a cell or switch to the opposite side of one (twin) faces. In addition,
“radial” pointers are used to reference different cells around an edge. While some of
these structures have been proposed many years ago, traversal in space is still not a
common feature for exporters of modern modeling systems and exchange formats.
The approach presented in this paper builds upon topological pointers and offers
enhancements of modeling capabilities regarding features of a complete partition of
space. In particular, the treatment of multiply-connected and unbounded domains is
similar to that of simply-connected bounded domains.
Separating model entities for geometry and topology is a convenient way to struc-
ture a shape model. The approach is often encountered in data structures. Weiler [16]
describes the connectivity among faces, edges, and vertices introducing the “-use”
concept for the oriented objects. Lee and Lee [7] use “partial” entities equivalent to
the radial edge. Separating geometry and topology has already led to enhancements
in the usability and clarity, but the concept can be extended even further [4]. The
extended aim is to keep the two models consistent at any work step as achieved with
the user and core model structure in this paper. Figure 2 shows the oriented arrow
bundles stored in the core model and used in the polyhedral space partition approach.
The oriented arrows and twin-facets in this approach are based on existing half-edge
concepts.
Robust Modeling of Polyhedral Space Partitions 431

Fig. 2 A 2-dimensional face is embedded in the 3-dimensional space. Each edge is complemented
by a oriented arrow bundle, describing polygons. Each face f has to sides (facets) with normal
vectors pointing outside

2.3 Properties of a Linear Partition of Space

A linear partition of space is an alternative to the conventional approach of accu-

mulating objects in space. In the novel approach, the point set of the unbounded
Euclidean space is the initial model. During the model construction, subsets of the
point set of the space are defined as user model domains of type node, edge, face
and cell, as shown in Fig. 1. The domain types have rank 0 to 3 respectively. Because
the defined geometric domains are linear, the partition is called a linear partition of
space. The work steps of the partitioning are chosen such that the resulting model is
a mapping of the original and of the space in which it is contained.
While the domains of the user model have some of the characteristics of a topology,
navigation must be supported by orientation of the domains and explicit pointers for
efficient spatial querying. Therefore, the described specified user model domains of
the point set topology are complemented by a core model, where oriented boundary
polygons are defined, surfaces are oriented by their normal vectors and the relation-
ships between the faces at edges is described with dihedral cycles, as demonstrated
in Fig. 2. Additionally, pointers for inner polygons in facets and inner surfaces in vol-
ume elements explicitly store information for multiply-connected domains without
the use of supporting edges or faces as found in established data structures.
Modelling of the entire space requires a data structure that contains both bounded
and unbounded domains. The boundary of a bounded domain consists of the point sets
432 M. Sternal and W. Huhnt

Fig. 3 Imaginary nodes are marked as triangle right. Edge e7 does not have a point-set, but the
topological properties to enable navigation

of bounded domains of lower rank. The boundary of an unbounded domain contains

the point sets of bounded and unbounded domains of lower rank. The unbounded
domains are related by topological elements called imaginary domains, which do not
have a point set. Figure 3 shows a bounded and unbounded face. Edge e7 and nodes
n5 and n6 are imaginary domains without point-sets.

3 Modeling Operators

3.1 Atomic Operators

Geometric kernels include core functionality to create and modify geometric models.
Usually, models are created component-wise, with a focus on the boundary geometry
of each object. Modeling operators are provided to the user. These are commonly
based on parametric primitives and higher-level operators such as extrude, sweep
or slice. While modern modeling kernels are able to calculate and represent a large
number of different shapes with boundary representations, connectivity between
objects is often not considered. Furthermore, partitions of space such as octrees are
mostly calculated from existing boundary representations rather than created from
scratch using operators.
This paper describes a step-wise construction process for a polyhedral space par-
tition. Starting with a simple initial partition of the complete space, the partition is
modified by repeated application of a small set of operator types. [18] gives a detailed
overview of manipulation operators for both, manifold and non-manifold cell com-
plexes. An important group amongst these operators are Euler-operators [3], which
are commonly used in edge-based data structures [1, 16]. While the combination
of parametric modeling and Euler operators are state-of-the-art in various modeling
systems, a model constructed by partition of the complete space requires a different
set of atomic operators, as described in this chapter.
Robust Modeling of Polyhedral Space Partitions 433

Fig. 4 Different splits

showing single and
multiply-connected faces
before and after split

Fig. 5 Following the rank

concepts, boundary edges e1
and e2 are split first before
the face can be split

Various categories of operators can be distinguished according to Comic and

de Floriani [18]: initialization, topology preserving and topology modifying opera-
tors. This paper introduces an initialization with a base decomposition of the entire
space, complemented by only two types of topology modifying operators: “split” of
a domain or “merge” of connected domains of equal rank. These two types of atomic
operators are related to the frequently described operators “glue” and “splice” for
manifold cell complexes [18]. The topology modifying operators “split” and “merge”
are implemented for domains with rank 1 to 3. Any higher-level operator is a sequence
of split and merge operations. Like the basic Euler operators such as “MAKE”,
domains are created only by a split operation and deleted only by a merge oper-
ation. Figure 4 shows various split operations for faces. New domains are marked
red. It is shown that multiply-connected domains can be created or modified in this
approach without the need of supporting edges. A valid operation always leads to a
valid partition of the complete space.
434 M. Sternal and W. Huhnt

3.2 Sequences of Work Steps

A sequence of work steps is always sorted according to the rank of the affected
domains. Figure 5 shows an additional split of a face. Before the interior of the face
can be subdivided in work step 3, all boundary domains must be prepared. Therefore,
step 1 and 2 splits the interior of edges e1 and e2 to introduce new nodes, before
edge e7 is added to the partition in a split of the interior of the face.
User interfaces for geometric modeling in engineering practice require higher
order operators, build from sequences of atomic operators. An example is a move
operation of a box from cell A to cell B. Several work steps have to be performed to
create a new box in cell B and remove the old one box from cell A. Following the
rank concept, cells must be merged first, before edges and nodes in their boundaries
can be deleted by subsequent merge operations. Boundary domains may only be
deleted if they do not bound other domains in addition to A and B. Nodes and edges
are introduced in the boundary of cell B by splits of edges and faces, before the new
box is created by a cell split. This completes the sequence of work steps for the move
operation.
During the execution of sequences of work steps, a number of checks must be
performed to guarantee the consistency of the model after the step sequence. Various
errors due to incorrect topology or geometry and due to numerical imprecision can
occur. The concepts of robustness and of checking are outlined in the following
chapters. If any kind of error is found during the checking of a work step or a
sequence of work steps, the construction is aborted and the preceding work steps in
the sequence are reversed. This operation is straightforward, because pairs of split
and merge operations are counterparts that are readily derived from each other.

4 Robustness

4.1 Definition

A variety of definitions and interpretations of robustness exists in the field of computa-

tional geometry. Keyser [6] describes different implementation characteristics, where
robustness is one characteristic besides “efficiency” and “accuracy” and increased
robustness would be often a trade-off regarding run-time or storage efficiency. Tra-
ditionally, “robustness problems” are often discussed due to precision problems and
specific “degeneracies”, which can occur during computations [12]. Due to this
lack of a clear definition, some developments claim to “improve robustness”, while
others claim that their approach is “robust”. Different degeneracies in polyhedral
computations are outlined by [15], who outlines two general strategies: multiple pre-
cision arithmetic or high priority for topological consistency. Hoffmann [5] gives
an overview of robustness in solid modeling and justifiably calls it a “catastrophic
event”, if inconsistency occurs between highly structured topological data and the
Robust Modeling of Polyhedral Space Partitions 435

geometry of the model. Generally speaking, logical facts, or predicates, such as inci-
dence, are deduced in algorithms based on numerical calculations, but imprecision
of floating-point arithmetic makes the conclusions, inferred from such predicates,
unreliable.

4.2 Inexact Computation

In the past, several approaches for exact geometric computations have been proposed
addressing the robustness problem [17]. An implementation of an algorithm in which
the arithmetic calculations are carried out exactly is guaranteed to avoid problems of
consistency, since it will always construct a result with exactly the correct topology.
In fact, correct topology is guaranteed if the predicates are evaluated exactly, irrespec-
tive of how the real values concerned are represented. Libraries such as LEDA and
CGAL provide implementations of “exact” predicates using arbitrary-precision inte-
gers based on the arithmetic of Shewchuk [11]. While this solution seems straightfor-
ward, it can slow down computation substantially. Graphic-processing units (GPUs)
are tested to accelerate the calculations [8], but this concept is very hardware depen-
dent. To tackle this problem, floating-point filters can be used, providing an adaptive
evaluation and calculating a strict bound on the extent of any numerical error, based
on the known precision of individual arithmetic operations. Dynamic filters can com-
pute a closer error bound than static filters, but also come with higher computation
cost. Another specific form of filter is the use of interval arithmetic, in which upper
and lower bounds are maintained by a switch of rounding up and down for each bound
respectively compared to standard “round to nearest” in floating-point operations.
This approach is inefficient.
While these implementations of exact computation may lead to a solution in every
case, high computational costs may come with this certainty. Therefore, different
approaches for “inexact computation” have been proposed recently, especially in the
field of Boolean intersections. Smith [13] extensively describes different approaches
for handling degeneracies, which guarantee consistency between the geometrical
and the topological description. One of these strategies is the topologically-oriented
approach and an extension of this approach, topologically controlled work steps, is
used in the operators for polyhedral space partitioning presented in this paper.

4.3 Robust Modeling by Topologically Controlled Work Steps

A work step in the construction of a linear complex is called topologically controlled

if a single domain is replaced by exactly two domains of the same rank in the work
step, or if exactly two domains of equal rank are merged to form a single domain of
the same rank. The intended topological change in the work step is known explicitly.
In the beginning of a topologically controlled work step, it is checked whether the
436 M. Sternal and W. Huhnt

input data do in fact lead to the intended topological change. If they do not, the error
is reported and must be corrected before the construction proceeds. Uncontrolled
topological changes cannot occur if this procedure is adopted.
Because of the atomic operators, the topology-oriented approach is robust. Geo-
metric degeneracies are detected. Because each operation is focused on the neigh-
borhood of the modified domains, the number of geometric checks required when
introducing new domains is significantly reduced from the number required for estab-
lished procedures. The method does not require tree-based decomposition such as
binary space partitioning or octrees.

5 Checking of Specified Input Data

5.1 Affected Domains

The input data are checked for every atomic work step. Work steps are ordered by
rank. Interior point-sets can only be split after their boundaries have been split. If an
error is detected, the work step is aborted and the error is reported. Errors can occur
due to improper topological configuration of the input, or due to deviations from
the real coordinates of the nodes. Each operation is focused, because the domain, in
which a subdomain will be created, is explicitly specified in the commands. Because
of the explicit topology in the core model, location test must only be performed on
a minimal subset of geometric domains in the direct neighborhood of the modified
domain.
The novel approach is structured so that the tools for topological and geometrical
tests on domains are largely independent of the rank of the affected domains. The
geometric tests are similar for bounded or unbounded domains. Because topological
tests are performed by traversing the graph structure of the core model, they are
faster than geometric tests that calculate floating point coordinates, distances and
predicates to determine whether a point is in the interior of a domain. Topology is
tested before geometry in the approach of topologically controlled work steps.

5.2 Topological Checking

There are four categories of topological errors that can occur when merging domains
or splitting a domain and introducing new subdomains:

– Unbounded domains that intersect bounded domains

– Domains that are not closed
– Non-manifold domains
– Unconnected domains
Robust Modeling of Polyhedral Space Partitions 437

Fig. 6 Topological errors that are checked in the split of a face, not closed boundary left, non-
manifold boundary middle and multiple not connected boundaries right

Because of the decomposition of the complete space, outer domains are bordered
by imaginary nodes, edges and faces, which describe the topological relations of the
unbounded edges, faces and cells. To ensure a correct partition of the unbounded
space, unbounded edges and faces are not permitted to intersect with each other and
are improper in the interior of a bounded domain. The configuration of the domain
is checked at the beginning of a work step.
All domains of a model must be closed topological manifolds, homeomorphic to
a Euclidean space. An example of a 2-manifold is a face with a 1-dimensional closed
boundary as shown in Fig. 6. While each face can have multiple inner boundaries,
each boundary by itself must be homeomorphic to a circle. This can easily be checked
based on the topological graph structure of the input. The affected domain is evaluated
independent of its surrounding. Lastly, each merge operation must be performed on
exactly two connected domains of the same rank, resulting in one manifold output
domain. Equally, a split operation must result in exactly two connected topological
manifolds. If a decomposition into more domains is intended such as seen in Fig. 6
(right), several splits must be performed in a sequence of work steps.

5.3 Geometrical Checking

If the topological relations of the input data are valid, geometric checks based on the
coordinate values are the next step in the check algorithm before any modification
of the space partition takes place in the atomic work step. Geometric imprecision
must be considered. Base geometries such as planes and lines can be analyzed to
evaluate the distance of a point in space to the domain and to decide, whether a
specific point in space lies inside a geometric domain. Figure 7 shows a simple face
with four nodes in a plane. Due to imprecise coordinates, the nodes may not lie
exactly in a plane. Therefore, a global accuracy parameter .ε is introduced to limit
the maximum distance of a node from the plane that approximates the exact plane
that topologically contains all four nodes. This accuracy parameter .ε is derived as a
static filter from the model dimensions .si ze = 10s and upper bound on the relative
approximation error due to rounding in floating point arithmetic .eps = b ∗ 10m :
438 M. Sternal and W. Huhnt

Fig. 7 Base geometry of a plane is shown. A face (blue) with 4 nodes is demonstrated. The point
is inside the domain, if the distance is smaller than a accuracy parameter

.ε = 10m+s+1 (1)

For instance, the machine epsilon for double-precision is .eps = 2.22 × 10−16 . In
case of a model size of .si ze = 106 units, a suitable accuracy parameter would be
chosen as .ε = 10−9 . While planes and lines in space are unbounded, the boundary of
an unbounded geometric domain in a space partition contains at least one bounded
domain. Predicates, to determine whether a point lies in the interior of a domain, must
be calculated for geometric checks. To reduce number of geometric test types to a
minimum, bounded and unbounded domains are treated in a similar manner. Because
model size .s and the accuracy .ε are directly correlated, temporary coordinates for
imaginary nodes can be chosen at the border of the model. These coordinates are used
like coordinates of real nodes to calculate distances and predicates for unbounded
domains in the same manner as for bounded domains.
As described in Sect. 4, computational strategies for imprecise values are often
implemented for robust geometric computations. In space partitioning, the computa-
tion is reduced to three general procedures: (1) calculating the distance of a domain
to another domain, (2) projecting a domain of lower rank, e.g. a node, into another
domain such as a straight line or a plane and (3) computing the predicate of a point in
relation to a specific domain using the orientation of arrows in polygons and facets
in surfaces. Ray casting or segmentation algorithms, which are used to determine the
point-in-polyhedron predicate in established procedures, are not required in the space
partitioning approach. For a 3-dimensional cell in the polyhedral space partition, all
boundary domains can be sorted according to the distance of their base geometry to
a point in space. By projecting the point into the lower dimensional base geometries,
the anchor on the boundary of a cell can be determined effectively.
Anchor and Client Zones. In the methods that determine the point on a boundary
of a domain that is nearest to a given point in space and calculate location predicates
for domains of different dimensions, the conventional concept of localization with
simultaneous inequalities is replaced by the concept of localization with anchor zones
and associated client zones. In order to localize a point relative to a domain A, the
boundary of the domain is decomposed into open subdomains called anchor zones.
Each anchor zone is associated with a client zone. The client zone contains the points
Robust Modeling of Polyhedral Space Partitions 439

Fig. 8 Anchor and Client Zones in the 2-dimensional space. Client zones of nodes are green. Inside
predicates for grey and orange zones are determined using orientation of the core model

in space, which are as near or nearer to the anchor zone than to any other anchor
zone. The client zones are a decomposition of space.
Different types of condition are formulated for client zones of different rank. A
point of a client zone is an inner, outer or boundary point of a domain, depend-
ing on which of the conditions for the client zone the location vector of the point
satisfies. Figure 8 shows different client zones and anchors for clients (points) on
a plane surrounding a face. Red anchors H have the projected coordinates of the
closest boundary domains, in this case edges and nodes. Using the orientation of
the core model, predicates for all domains of rank larger than 0 can be determined
by calculating the scalar product or the vector product of two vectors in space. This
principle is independent of the tested domain or the client zone. Normal vectors of
faces are stored explicitly. Normal vectors of edges and nodes are calculated from
the normal vectors of all adjacent faces. If the angle .a between the normal and the
vector from an anchor to its client (Fig. 8) is smaller than pi/2 the point is determined
to be outside, and inside if it is bigger than pi/2. This principle is applied for all
predicates needed in the geometric tests regardless of its dimension, thus leading to
a systematic and convenient minimal set of calculations based on explicitly available
topology and the theory of client zones.
Contact and Intersection. While the concept of anchors and clients in combination
with oriented domains is a consistent way to determine distances and predicates if
the affected domains are inputs, geometric errors can also occur due to unintentional
contacts or intersections of domains. Therefore, a structured method of testing for
intersection of existing domains and input domains is needed.
Figure 9 shows various types of intersections between domains. Because space
partitioning provides all neighboring domains, no further calculation is needed to
determine which geometries must be checked for intersections. Intersections of base
geometries are calculated and the same theory of anchors and client zones is utilized
in order to determine erroneous geometric input. The model topology and geometry
440 M. Sternal and W. Huhnt

Fig. 9 Different types of

intersections, checked during
every atomic work step

are modified only if the checking does not reveal any errors. In this manner, it is guar-
anteed that the polyhedral partition is proper at the end of every atomic work step.

6 Conclusion

6.1 Summary

Various methods to treat topology in digital building models and to ensure robustness
in computational geometry have been reviewed and compared to a novel approach
of modeling a polyhedral space partitioning. Bounded and unbounded domains
have been introduced and investigated for a complete partition of space. Imagi-
nary domains have been introduced to topologically represent relations between
unbounded domains. While the imaginary entities do not have a geometric point-set,
they permit navigation in the complete space in the same manner as in a bounded
space. Many applications such as energy simulations require information about the
exterior surfaces of building components and opening. This information can be
deduced using the presented unbounded domains.
For the construction of polyhedral space partitions, a minimal set of operators that
split and merge domains of equal rank have been presented. Sequences of atomic
operators are structured with a rank concept, ensuring that only the interior of a
point-set is modified during an operation. Boundary domains of lower rank must be
modified in preceding work steps split in the case of a split or removed in succeeding
work steps merge in the case of a merge. Sequences of work steps for higher level
operators such as move, which combine a sequence of split and merge operations,
are described in Sect. 3.2 by an example.
Robust Modeling of Polyhedral Space Partitions 441

The concept of space partition requires all domains to be manifolds. To ensure

robustness in every work step, topologically controlled work steps are introduced
and compared to existing approaches for exact and imprecise robust modeling. By
checking topology and geometry in every atomic work step and using the charac-
teristics of focused operators, efficient and robust work steps are implemented. The
operators are called focused because the subset of domains to be tested is explicitly
known due to the inherent properties of the space partitioning concept.
The presented concept of anchor and client zones provides a consistent theory for
all geometric checks that are required. Distances and predicates are applied to base
geometries to determine the location of points relative to domains of any rank. A
global accuracy parameter is introduced to handle numerical imprecision, analogical
to static floating-point filters used in imprecise computations. Because the topology
is mapped exactly to the digital model, the system is robust if no discrepancy is
detected between its topology and its geometry. Otherwise, the operation is aborted
and the partition is not modified and related preceding work steps in a sequence of
work steps are reversed using the complementary operators.

6.2 Outlook

The presented operators are key components for a future space partitioning modeling
platform for geometric models in architecture and civil engineering. The robustness
of the atomic operators is a prerequisite for modeling functionality of higher order,
such as extrude or sweep. Robustness is treated independent of the source of the
geometric data, and is ensured by the operators themselves. Future user interfaces
will calculate coordinates and generate input for split and merge operations, whose
consistency in the model is automatically checked and guaranteed by the basic oper-
ators.
Further performance tests have to be conducted to evaluate the usability of the
topologically controlled work step approach for modeling polyhedral space parti-
tions. First verification results show the general applicability of the robustness theory.
In order to generate acceptance for this approach, interfaces for existing parametric
modeling interfaces should be provided when implementing higher order operators.
Many fields of application can benefit from consistent and topologically enriched
geometric data models. Transformation and issue management are two main tasks in
engineering workflows today, especially in the coordination of architectural models,
structural design and energy modeling. The alignment of indoor and outdoor mod-
eling calls for robust operators that handle bounded and unbounded domains and
provide means to create graph models for connectivity and navigation. First results
in generating graph data structures such as semantic web graphs highlight the ability
to model geometry and topology consistently. The novel approach provides a gen-
eral basis for modeling that aligns different fields of application around one common
geometrical model. Polyhedral partitions fulfil the requirements of most engineering
applications and map the complete space.
442 M. Sternal and W. Huhnt

References

1. Boguslawski P, Gold C (2016) The dual half-edge-a topological primal/dual data structure and
construction operators for modelling and manipulating cell complexes. ISPRS Int J Geo-Inf
5(2):19. https://doi.org/10.3390/ijgi5020019. https://www.mdpi.com/2220-9964/5/2/19
2. Daum S, Borrmann A (2014) Processing of topological BIM queries using boundary represen-
tation based methods. Adv Eng Inform 28(4):272–286. https://doi.org/10.1016/j.aei.2014.06.
001. https://linkinghub.elsevier.com/retrieve/pii/S1474034614000391
3. Eastman C, Weiler K (1979) Geometric modeling using the Euler operators. Institute of Physical
Planning, pp 248–262
4. Galishnikova V, Huhnt W (2021) Polyhedral space partitioning as an alternative to component
assembly. In: ECPPM 2021 – eWork and eBusiness in architecture, engineering and construc-
tion. CRC Press, p 6
5. Hoffmann CM (2001) Robustness in geometric computations*. J Comput Inf Sci
Eng 1(2):143–155. https://doi.org/10.1115/1.1375815. https://asmedigitalcollection.asme.
org/computingengineering/article/1/2/143/450161/Robustness-in-Geometric-Computations
6. Keyser J (1997) Robustness issues in computational geometry. Duke University, p 19
7. Lee SH, Lee K (2001) Partial entity structure: a compact non-manifold boundary representation
based on partial topological entities. In: Proceedings of the sixth ACM symposium on solid
modeling and applications - SMA 2001. ACM Press, Ann Arbor, pp 159–170. https://doi.org/
10.1145/376957.376976. http://portal.acm.org/citation.cfm?doid=376957.376976
8. Menezes M, de Magalhães SVG, de Oliveira MA, Randolph Franklin W, Coelho B (2022)
Employing GPUs to accelerate exact geometric predicates for 3D geospatial processing. In:
Spatial gems, 1st edn., vol 1. Association for Computing Machinery, New York, pp 97–110.
https://doi.org/10.1145/3548732.3548744
9. Park S, Ju S, Yoon S, Nguyen MH, Heo J (2021) An efficient data structure approach for BIM-
to-point-cloud change detection using modifiable nested octree. Autom Constr 132:103922
10. Paul N (2013) Applications of continuous functions in topological CAD data. arXiv:1308.0256
[cs]
11. Richard Shewchuk J (1997) Adaptive precision floating-point arithmetic and fast robust
geometric predicates. Discrete Comput Geom 18(3):305–363. https://doi.org/10.1007/
PL00009321
12. Schirra S (1998) Robustness and precision issues in geometric computation, p 36
13. Smith JM (2009) Towards robust inexact geometric computation, p 186
14. Solihin W, Eastman C, Lee YC (2017) Multiple representation approach to achieve
high-performance spatial queries of 3D BIM data using a relational database. Autom
Constr 81:369–388. https://doi.org/10.1016/j.autcon.2017.03.014. https://linkinghub.elsevier.
com/retrieve/pii/S0926580517302388
15. Sugihara K (2001) Degeneracy and instability in geometric computation. In: Kimura F (ed)
Geometric modelling. Springer, Boston, pp 3–16. https://doi.org/10.1007/978-0-387-35490-
3_1
16. Weiler KJ (1986) Topological structures for geometric modeling (Boundary representation,
manifold, radial edge structure)
17. Yap CK (1997) Towards exact geometric computation. Comput Geom 21 (1997)
18. Čomić L, De Floriani L (2012) Modeling and manipulating cell complexes in two, three and
higher dimensions. In: Brimkov VE, Barneva RP (eds) Digital geometry algorithms: theoret-
ical foundations and applications to computational imaging. Lecture notes in computational
vision and biomechanics. Springer, Dordrecht, pp 109–144. https://doi.org/10.1007/978-94-
007-4174-4_4
Feature-Based Decomposition
of Architectural Spaces: Outline
of a Procedure and Research Challenges

Georg Suter

Abstract Space modeling is a key activity in building design that is supported

by BIM authoring systems. However, decomposing spaces with specific properties,
such as spaces with non-convex volumes, into disjoint partial spaces, or sub-spaces,
is a manual and non-trivial task. This study identifies the need for automated 3d
space decomposition methods. Informed by a review of related space and shape
decomposition methods, the problem scope for feature-based space decomposition
is defined. A procedure is outlined that addresses this problem. The procedure recur-
sively decomposes a space into sub-spaces using cutting operations derived from
recognized features. The decomposition of a (sub-)space ends if it meets decompo-
sition goals or if no features are recognized. The procedure is divided into space
evaluation, feature recognition, and space cutting phases. The recursive decomposi-
tion of a space may be represented as a tree where nodes correspond to iterations of
the procedure and edges to sub-spaces. Research challenges related to architectural
feature taxonomies, feature recognition, and cutting operations are identified.

Keywords Building information modeling · Space modeling · Feature

recognition · Volume decomposition

1 Introduction

Building information modeling (BIM) authoring systems support the modeling of

architectural spaces. For several space modeling tasks, such as visualization, clas-
sification, and geometry simplification, there is a need to decompose spaces with
specific properties into disjoint partial spaces or sub-spaces. Examples are spaces
with non-convex volumes, multi-functional spaces, large spaces, and spaces with
partitions or built-in furnishing elements. Currently, space decomposition is a manual

Supported by Grant Austrian Science Fund (FWF): I 5171-N.

G. Suter (B)
Design Computing Group, TU Wien, Vienna, Austria
e-mail: georg.suter@tuwien.ac.at

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 443
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1_31
444 G. Suter

task that is time-consuming and error-prone. It is non-trivial because the boundaries

between adjacent sub-spaces are usually virtual rather than physical. Boundaries may
be defined in multiple ways. Automated space decomposition could improve user
productivity as well as the consistency and data quality of space models.
The scope of most known space decomposition methods in BIM is limited to 2d
geometries and specific application domains, such as indoor navigation and evacua-
tion modeling [22]. By contrast, more general 3d geometry decomposition methods
exist in the areas of computational geometry and computer-aided engineering [7, 29,
32, 35]. These methods may be reused and adapted for space decomposition.
This study aims to i) identify the need for 3d space decomposition methods; ii)
review related methods; iii) define the problem scope for space decomposition based
on features; iv) outline a decomposition procedure that addresses the problem; and v)
identify research challenges. The novelties of the proposed decomposition procedure
include coverage of 3d space geometries and the use of goals and feature types to
guide the search for decomposition solutions that meet the diverse needs of space
modeling applications.

2 Motivation

Space decomposition needs are identified for space visualization, space classification,
and space geometry simplification. These tasks are relevant to multiple applications,
such as floor area measurement or energy analysis. A space layout of an apartment
unit is used to illustrate these needs (Fig. 1). Two alternative decompositions of a
large space .s (Fig. 1a) into partial spaces, or sub-spaces .s1 , s2 , . . . , sn , are shown in
Fig. 1b and 1c.

2.1 Space Visualization

Besides the large, multi-functional space .s, the apartment unit consists of five small
service spaces (Fig. 1a). Space volumes are convex except for .s and the bathroom.
A volume is convex if all interior angles between adjacent, planar face pairs are less
than 180.◦ . A property of a convex volume is that its geometric center, or centroid, is
always located in its interior. For example, this property is useful for placing space
labels in visualizations. There is at least one concave edge on the boundary of a non-
convex volume. An edge is concave if the interior angle between its incident face pair
is greater than 180.◦ . The centroid of a non-convex volume is located on its interior
or exterior. Thus, a label placed at the centroid may be incorrectly associated with
an adjacent space, or it may be placed near the space boundary. One way to achieve
an improved visualization would be to decompose a non-convex space into a small
number of convex sub-spaces, and to place the label at the centroid of the volume of
the largest sub-space. All spaces in the second decomposition example have convex
Architectural Space Decomposition 445

S2 S6

S5
S S1 S3 S3 S4 S2
S1
A A
S4 S7

A-A

S S1 S3 S1 S5 S3 S4
S2

(a) (b) (c)

Fig. 1 Space decomposition examples. a Space layout of an apartment unit with space.s, b Example
for a non-convex decomposition of .s into sub-spaces .s1 , s2 , s3 and .s4 , c Example for a convex
decomposition of .s into sub-spaces .s1 , s2 , . . . , s7 .

volumes (Fig. 1c). Accordingly, the label for space .s (Fig. 1a) would be placed at
the centroid of the volume of sub-space .s4 . The first decomposition is non-convex
because the volumes of sub-spaces .s1 and .s3 are non-convex (Fig. 1b). Both would
need to be further decomposed until all sub-space volumes in the decomposition
meet the convexity condition for label placement.

2.2 Space Classification

Space .s in Fig. 1a supports multiple activities or functions, including living, cooking,

sleeping, and circulation. The function and shape of each sub-space must be deter-
mined for accurate floor area measurement or energy analysis. This could be done in
two steps. First, .s could be decomposed into functional sub-spaces based on its fea-
tures. The latter include a vertical hole, two window niches, and a kitchen element.
A functional sub-space is a space that can accommodate people or equipment. In a
second step, sub-space functions could be classified based on their features, other
geometric properties and connectivity to other sub-spaces or space elements [3, 33].
The first decomposition consists of four sub-spaces (Fig. 1b). All space volumes
are large enough to be classified as functional spaces. While the two circulation spaces
.s2 and .s4 have convex volumes, the sleeping and living space volumes .s1 and .s3 are
non-convex because of window niches. A decomposition goal could be defined such
that spaces in a decomposition do not need to be convex. Similarly, the bathroom
includes a shallow niche that extends from ceiling to floor. It could be considered as
being nearly convex. The decomposition procedure may end when only minor con-
cave features remain. Again, near-convexity could be defined as a goal.
Cooking and living spaces .s3 and .s4 are modeled as separate spaces in the second
decomposition (Fig. 1c), as are two window niches .s1 and .s2 . The latter are shallow
extensions of adjacent spaces .s4 and .s5 and may be considered non-functional spaces
446 G. Suter

because they do not accommodate people or equipment. Allowing for non-functional

spaces may be defined as a goal. The cooking space.s3 is derived relative to the kitchen
element contained in space .s. More generally, decomposition may be based not only
on space volumes but also on the shapes of related space elements.
In both decompositions, the two narrowest functional spaces are located on oppo-
site sides of the vertical hole. Based on their size and dimensions, these may be
classified as circulation spaces. This illustrates how space geometry may correlate
with function, and thus the former may be inferred from the latter. Both decomposi-
tions differ regarding the number of spaces, space volumes, related space elements,
and accessibility. This may affect space classification. In the first decomposition, the
circulation space .s4 only connects sleeping and living spaces .s1 and .s3 . The corre-
sponding circulation space .s7 is additionally accessible from the unit entrance door
in the second decomposition. It may thus be classified as an entrance space.

2.3 Space Geometry Simplification

Space geometry simplification is relevant for visualization, planning, or simulation

environments that support models with multiple levels of detail [34]. It involves
transforming or generalizing a model with a higher level of detail into one with a
lower level of detail while preserving salient geometric properties of the original
model, such as parallelism or orthogonality.
Simplified space volumes are required for certain applications, such as energy
analysis. Thus niches, protrusions, openings, or other detailed features may need to
be isolated and removed from a space volume. In other applications, the types and
shapes of these features must be determined. An example is floor area and volume
measurement, where niche spaces need to be considered dependent on their geomet-
ric properties in relation to adjacent spaces [24, 36]. In the second decomposition
example, the smaller niche space .s1 is elevated, while the larger niche .s2 extends
from the floor (Fig. 1c). In the latter case, the niche area is considered an extension
of the living area of .s4 . This example shows that distinguishing different niche types
requires 3D rather than 2D space geometries. Moreover, in scan-to-BIM applica-
tions, source data are 3D geometry data with high levels of detail from which space
models are created [25].

3 Decomposition Methods

3.1 Space Decomposition

Space decomposition methods exist for several domains. In indoor navigation and
emergency evacuation modeling, indoor spaces are decomposed to derive navigation
Architectural Space Decomposition 447

models with desirable properties, such as paths that avoid obstacles or model people’s
movement [22]. Common decomposition methods are based on planar grid [21] or
irregular cells. Examples for the latter include 2d or 3d triangulations [2, 6] as well
as point-based [4] and line-based (or skeleton) Voronoi diagrams of space polygons
[11, 20]. The decomposition method developed by Jeong and Ban [18] generates
convex sub-spaces for space access analysis [17]. Candidate sub-space polygons
are generated from a given space polygon. The ‘fattest’ polygons with the highest
area-perimeter ratios are included in a decomposition.
Each reviewed method is limited concerning at least one space decomposition
need (Sect. 2). In most methods, space geometries are modeled as 2d polygons instead
of 3d volumes. Decomposition of space volumes by 3d triangulation would, in most
cases, result in many small sub-spaces. These are undesirable for visualization, clas-
sification, and geometry simplification. Similarly, the number of convex sub-spaces
created from Voronoi diagrams is neither minimal nor near minimal. Geometric prop-
erties, in particular angles between adjacent edges or faces, are not well preserved
by most reviewed methods.

3.2 Feature Recognition and Shape Decomposition

Decomposition methods are relevant for simulation, engineering design, and pro-
duction planning. In finite element analysis, features such as nuts and bolts may
be isolated and removed from a volumetric model of a part in order to reduce the
number of mesh elements. This speeds up calculations without reducing accuracy
[27]. In a volume simplification method developed by Thiemann and Sester [34],
which extends work by [27], building volumes are decomposed (or segmented) into
sub-volumes. Examples of segmented elements are chimneys, balconies, doors, and
windows. At first, protrusion and holes are detected by one or more cutting planes
that are derived from the volume boundary. Next, recognized features are subtracted
from the volume by half-spaces that are created from cutting planes. This results in
a simplified building volume with a lower level of detail than the original volume
while preserving geometric properties, such as orthogonality or parallelism between
volume boundary faces. A building volume may be decomposed in multiple ways.
Feature quality values are computed to identify and remove the most relevant fea-
tures.
Feature recognition is relevant for integrating mechanical engineering design and
production planning tools [35]. Slots, holes, steps, and pockets are examples of
machining features. A feature may be defined as a “region of interest on the surface
of a part” [26]. Features may be recognized automatically in a part design using
recognition methods. This facilitates the re-interpretation of a design for produc-
tion planning [35]. Most feature recognition methods assume that part designs are
modeled as solid (or volume) Breps.
Topological, heuristic, and volumetric are common feature recognition methods
[29]. Topological methods are based on face adjacency graphs that are created from a
448 G. Suter

volume Brep. An example is the attributed adjacency graph (AAG), where nodes and
edges correspond to faces and edges in the Brep [19]. AAG edge attributes indicate
if an edge is convex or concave. Edges labeled as convex are first removed from
the AAG. Next, features modeled as AAGs are matched against the modified AAG
using a subgraph search. While topological methods work well for isolated features,
recognizing features that interact with each other is difficult [29].
Volumetric methods involve the decomposition of a given volume into sub-
volumes or cells. For example, in volume decomposition and recomposition, a non-
convex volume is decomposed into ‘atomic’ convex cells; cells are merged into
‘macro’ volumes, and the latter are classified as features [29]. Cells may be created
by half spaces that are derived from faces that are incident on a concave edge [30].
The method by Sakurai [28] computes maximal convex cells instead of ‘atomic’
cells. Features are generated by subtraction of maximal cells.
The approach to decompose a volume into convex sub-volumes by cutting planes
that extend faces adjacent to concave edges is adopted in a recursive decomposition
method by Eftekharian and Campbell [10]. At least one concave edge disappears with
each cutting operation, which ultimately results in convex sub-volumes. A volume is
partitioned recursively into convex sub-volumes with planar and cylindrical cutting
surfaces. The decomposition can be represented as a tree where the nodes correspond
to complex or simple volumes and the edges to cutting directions. There are two
cutting directions for each concave edge. The given volume is at the root, and simple
volumes are at the leaf nodes. The latter are typically convex. The depth of the
decomposition tree corresponds to the number of concave edges. There are between
.n + 1 and .2 sub-volumes, where .n is the number of concave edges. Since the
n

exhaustive exploration of potentially large solution spaces is impractical, concave

edges are classified and ordered according to a ranking strategy. The most promising
cutting direction for the top-ranked edge is identified based on sub-volume metrics.
A common issue in decomposition methods is remote and local over-cutting [23].
In remote over-cutting, extending a face by a cutting plane may cause undesirable
cutting in remote locations. As a result, a decomposition may have more sub-volumes
than necessary. In local over-cutting, more cutting planes than necessary are used to
create a convex sub-volume locally. Remote over-cutting may be avoided by using
selected volume faces to constrain the cutting locally. Multiple cutting planes may
be aggregated into a single cutting unit or operation to avoid local over-cutting [23].
Applying feature recognition and volume decomposition methods to architectural
space decomposition is promising for two reasons. First, their scope is volumes rather
than polygons. Second, these methods are flexible. That is, they may be adapted to
meet domain-specific decomposition needs.

4 Problem Formulation

The problem for architectural space decomposition using feature recognition and
volume decomposition methods is formulated as follows:
Architectural Space Decomposition 449

Given goals, feature types, and data about a space .s, including its volume .v,
boundary elements, and space elements, find a decomposition of .s into .n sub-spaces
.s1 , .s2 , · · · , sn (.n ≥ 2) based on features that are recognized in the boundary of .v or
space elements. Sub-space volumes must not overlap and .v = v1 ∪ v2 · · · ∪ vn .
Decomposition goals guide the search for decomposition solutions. Goals may
include convexity of sub-space volumes [17], finding a minimal set of convex vol-
umes [8, 16], finding a decomposition with the largest convex volume [28], or opti-
mizing the differences between volumes [10]. A feature type is a rule or pattern
to recognize features in spaces. Examples of feature types are niches or openings.
Similar to decomposition goals, feature types influence the search for decomposition
solutions.
It is assumed that source building data, which include space data, are prepared
by a designer or planner using a commercial BIM authoring system and exported as
an IFC file. Space data are extracted from the IFC file as described by Suter [33].
Semantic and geometric space properties are separated, which allows the reuse of
computational geometry and solid modeling libraries for implementation [5, 31]. In
addition to geometric properties, semantic ones may also be considered to recognize
features.
Space volumes are modeled as boundary representation (Brep) solids with planar
faces. Brep data structures facilitate feature recognition and the application of existing
volume or cellular decomposition methods [7, 32]. In the latter, a solid is decomposed
into internally connected cells. The decomposition of spaces with curved faces,
particularly curved concave faces, is considered out of scope as it is significantly
more challenging than spaces with planar faces.
A space is bounded by several virtual or physical space boundary elements. Virtual
space boundary elements exist between adjacent sub-spaces. There is a one-to-one,
uni-directional relationship between space boundary elements and the faces in a
space volume Brep. Space elements are either space enclosure elements, such as
doors and windows, or elements that are contained in a space, such as furnishing,
sanitary, or technical elements. They are related to space boundary elements and
spaces, respectively. These spatial relationships must be derived for each sub-space
that is created in the decomposition process.
A space is typically decomposed based on features detected in its volume’s bound-
ary. Space element features may be used for decomposition as well. For example, the
cooking space in Fig. 1c is created from features detected in the shape of the kitchen
element.
Decomposition solution spaces may be large. For the convex decomposition
method by Eftekharian and Campbell [10], the solution space size is .n in the best
and .2n in the worst case, where .n is the number of concave edges in a given volume.
According to the method, the number of solutions for decomposing space.s in Fig. 1a,
which includes 11 concave edges, would be between 11 and 2048. An exhaustive
exploration of a potentially large number of decomposition solutions is generally
impractical. Thus decomposition procedures are desirable that explore only a subset
of these alternatives. They should be able to identify promising alternatives that best
meet given decomposition goals.
450 G. Suter

Fig. 2 Flowchart for feature-based space decomposition.

5 Decomposition Procedure

Decomposition-recomposition and recursive decomposition approaches could

address the architectural space decomposition problem defined above. For spaces
with many features, the former approach may result in a large number of fragmented
cells after decomposition, and hence there would be many solutions for recomposi-
tion. Both steps would be computationally expensive. Cell fragmentation is less of
an issue in recursive decomposition since only one cutting operation is executed at a
time. The effectiveness of this approach depends on identifying promising features
and cutting operations. Steps in such a recursive decomposition process are outlined
in the following (Fig. 2). It can be divided into space evaluation, feature recognition,
and space cutting phases.
In the space evaluation phase, a given (sub-)space is evaluated against decompo-
sition goals. Evaluation relies on basic geometric space properties, such as height,
floor area, volume, or concave edge count. Semantic properties may be considered as
well, such as the number of doors or windows. Decomposition goals may be defined
in terms of desirable values for space properties. An example of a goal is that all sub-
spaces in a decomposition must be convex. The procedure ends if the space meets
decomposition goals.
In the first step of the feature recognition phase, feature types are used to recognize
features in the space. If no features are recognized, decomposition of the space ends.
Next, promising candidate features are selected among recognized features using a
ranking strategy that is based on feature properties. Examples are a feature’s type,
concave edge counts or lengths. One or more of the highest ranked candidate features
are selected for further processing.
Architectural Space Decomposition 451

In the space cutting phase, candidate cutting operations are first created from
selected features. A cutting operation consists of one or more planar cuts. Multiple
candidate cutting operations are typically created for each feature. For example,
there may be two cut directions for a concave edge that are extensions of its incident
faces. Additional cut directions may be co-planar with pairs of parallel or co-planar
edges. All candidate cutting operations are evaluated and ordered according to a
ranking strategy. Finally, the highest ranked candidate cutting operation is executed
to decompose the space into sub-spaces. The latter are passed to the next iteration of
the decomposition procedure.
Two approaches are conceivable to evaluate candidate cutting operations. In the
first approach, each candidate cutting operation is executed [10]. The highest ranked
set of candidate sub-volumes is retained to create corresponding sub-spaces while
other candidate sub-volumes are discarded. Examples for sub-volume properties are
volume difference and number of concave edges [10]. A benefit of this approach is
that actual sub-volumes are evaluated. However, the computational cost is already
high in the case of only a few candidate cutting operations. This is because solid
modeling operations that are used to execute cuts are computationally expensive. In
the second approach, cutting operations are evaluated based on cut rather than volume
properties. An example of a cut property is the cut length, which may be defined as
the shortest distance between a midpoint on a concave edge and its projections on
boundary faces that intersect with a cut plane. A ranking strategy may prefer cutting
operations with short distances.
A tree that models the decomposition of the large space .s in the example (Fig. 1a)
is shown in Fig. 3. The goal is a convex decomposition, that is, to decompose the
.s into convex sub-spaces (Fig. 1c). Each tree node represents an iteration of the
decomposition procedure (Fig. 2). Each edge corresponds to a sub-space created by
the selected cutting operation. For simplicity, all cutting operations are assumed to
be selected and executed by a user. A ‘C’ label indicates a non-leaf cut node where a
cutting operation is executed. The type of the selected feature from which the cutting
operation is derived is shown next to the node. An ‘E’ label indicates a leaf node
where the decomposition of a sub-space ends. All sub-spaces that are passed to leaf
nodes are convex, and thus the decomposition goal is achieved. The ranking strategy
for feature selection is by feature type. Feature types [‘Niche’, ‘Space Element’,
‘2 Vertical Concave Edges’] may be recognized, where ‘Niche’ is the top-ranked
feature type. If a feature type is detected multiple times, then the ranking strategy
considers additional feature properties, such as the total concave edge length. For
example, small niches may be preferred.
Space .s is passed to the root node for recursive decomposition. ‘Niche’ features
are the highest ranked recognized features at the root and its child cut node. Cutting
operations that consist of single planar cuts are executed. The output of each node is
a small niche and a copy of .s where the niche space is subtracted. Niche spaces .s1
and .s2 are convex and thus not decomposed further. A convex cooking space .s3 that
contains the kitchen element is created at the second-level cut node where a ‘Space
Element’ feature is highest ranked. The corresponding cutting operation consists of
four cut planes which are derived from the shape of the kitchen element. A ‘2 Vertical
452 G. Suter

Fig. 3 Decomposition tree for space .s in Fig. 1a. The type of the feature selected to create cutting
operations is shown next to cutting nodes. Sub-spaces .s1 , s2 , . . . , s7 that are passed to leaf nodes
correspond to sub-spaces in Fig. 1c.

Concave Edges’ feature is selected at the fourth-level cut node. In this feature, a pair
of vertical concave edges bound the same face. The cutting operation consists of a
single cut plane and creates two spaces .s4 and .s4' . The smaller is the convex living
space. At the fifth-level cut node, a second feature of the same type is selected. In
this case, the cutting operation consists of two cut planes which create three convex
spaces, including a sleeping and two circulation spaces (.s5 , .s6 and .s7 ).
Architectural Space Decomposition 453

6 Research Challenges

6.1 Feature Taxonomies

The effectiveness of feature-based space decomposition methods is partly dependent

on comprehensive architectural feature taxonomies. Therefore, features relevant to
architectural space decomposition need to be systematically identified. Potential
sources for features include professional and research publications on architectural
design and heritage [1, 9, 15]. Vertical or horizontal concave edges in a space volume
boundary are examples of basic geometric features. Such features may be grouped.
Examples are loops of convex or concave edges in a face [12] or a pair or vertical
concave edges that bound the same face. If known, features may be further specialized
according to the classes of related space elements. For example, a vertical concave
edge whose adjacent faces are related to walls may be classified as a wall corner.
Some features are generally useful, such as openings and wall niches, while others
are more domain specific. For example, steps and ramps are mostly relevant for the
circulation domain. Granular feature sub-types may be required for certain domains.
For example, a wall or window niche may extend from a floor or parapet to a lintel
or ceiling. These distinctions are relevant for floor area measurement [36]. Features
are not necessarily restricted to spaces but may also cover space elements. Examples
include furnishing or kitchen element corners (Fig. 1).
As the examples illustrate, architectural features may range from generic to spe-
cific. While the former are geometric, the latter combine semantic and geometric
properties. As a result, architectural feature taxonomies are expected to be signifi-
cantly larger than those for machining features.

6.2 Feature Recognition

Architectural space features need to be formally defined to be recognized. Definitions

of such feature types or patterns may include topological, geometric, and semantic
properties. The topology may be represented as a feature sub-graph matched against
the AAG of a space volume [19]. Other feature representations are concave or convex
edge loops or edge lists. The latter may capture patterns of concave and convex edges.
Similar to regular expressions [14], these patterns could be matched with edge loops
in faces. An example of a pattern is an edge loop consisting of a pair of vertical
concave edges separated by one or more edges.
Feature interaction is an issue that limits the effectiveness of existing feature
recognition methods, in particular topological methods [13]. Two features interact if
the intersection of their volumes is non-empty. For example, consider a volume with
a hole and a slot feature. The two features interact if their intersection is non-empty,
e.g., the slot may split the hole into two parts [29]. The occurrence of such feature
interaction in architectural spaces needs to be studied.
454 G. Suter

6.3 Cutting Operations

A rich set of features would be complemented by flexible cutting operations to

decompose a volume in multiple ways, as illustrated by the example in Fig. 1. Cutting
planes as face extensions appear promising for architectural space decomposition
because convex angles in the original space volume would typically be preserved,
and the number of sub-spaces would either be minimal or near minimal. The latter
is relevant for functional space decomposition (Sect. 2). Neither of these properties
is preserved in triangulations or skeletons of space polygons (Sect. 3.1). Ideally,
cutting operations may consist of single or multiple planar or curved cuts. Solid
modeling libraries support single planar cuts. An example of such a library is the
cellular topology component in the ACIS solid modeler [32]. Multiple simultaneous
cuts [23] are currently not supported. Implementing single cuts, followed by post-
processing to merge cells due to local and remote over-cutting, is computationally
expensive. Thus algorithms that process multiple cuts and are easy to implement are
desirable.

7 Conclusions

Feature-based architectural space decomposition is a non-trivial problem due to the

need to navigate potentially large solution spaces and advanced geometry processing.
A decomposition procedure was outlined that uses goals and feature types to guide the
search for decomposition solutions. It was shown how the recursive decomposition
of a space could be modeled as a tree structure where nodes correspond to iterations
of the decomposition procedure and edges to spaces.
Several research challenges related to feature taxonomies, feature recognition,
and cutting operations need to be addressed to apply and extend existing approaches
and methods to space decomposition. Among these, the development of cutting
operations that support multiple planar cuts while avoiding local and remote over-
cutting appears particularly relevant as a foundation for future work in this promising
area.

References

1. Alexander C, Alexander P, Ishikawa S, Silverstein M, Jacobson M, Center for Environmental

Structure, Fiksdahl-King I, Shlomo A (1977) A pattern language: towns, buildings, construc-
tion. Center for Environmental Structure Berkeley, Center for Environmental Structure Series,
OUP USA, California
2. de Berg M, van Kreveld M, Overmars M, Schwarzkopf O (2000) Computational geometry:
algorithms and applications, 2nd edn. Springer, Heidelberg
Architectural Space Decomposition 455

3. Bloch T, Sacks R (2018) Comparing machine learning and rule-based inferencing for semantic
enrichment of BIM models. Autom Constr 91:256–272. https://doi.org/10.1016/j.autcon.2018.
03.018
4. Boguslawski P, Mahdjoubi L, Zverovich V, Fadli F (2016) Automated construction of variable
density navigable networks in a 3D indoor environment for emergency response. Autom Constr
72:115–128. https://doi.org/10.1016/j.autcon.2016.08.041
5. CGAL open source project: computational geometry algorithms library 5.5 (2022). https://doc.
cgal.org/5.5. Accessed 20 Sept 2022
6. CGAL open source project: computational geometry algorithms library 5.5 - CGAL 5.5 - 3D
triangulations (2022). https://doc.cgal.org/5.5. Accessed 20 Sept 2022
7. CGAL open source project: computational geometry algorithms library 5.5 - convex decom-
position of polyhedra (2022). https://doc.cgal.org/5.5. Accessed 20 Sept 2022
8. Chazelle B (1984) Convex partitions of polyhedra: a lower bound and worst-case optimal
algorithm. SIAM J Comput 13(3):488–507. https://doi.org/10.1137/0213031
9. Davies N, Jokiniemi E (2012) Architect’s illustrated pocket dictionary. Taylor & Francis
10. Eftekharian A, Campbell M (2012) Convex decomposition of 3D solid models for automated
manufacturing process planning applications. In: ASME design engineering technical confer-
ence, Chicago, IL, USA, vol 2, pp 727–735. https://doi.org/10.1115/DETC2012-70278
11. Fu M, Liu R, Qi B, Issa RR (2020) Generating straight skeleton-based navigation networks
with industry foundation classes for indoor way-finding. Autom Constr 112:103057. https://
doi.org/10.1016/j.autcon.2019.103057
12. Gadh R (1994) Feature mapping and feature recognition in geometric design generation. In:
Shah JJ, Mäntylä M, Nau DS (eds) Advances in feature based manufacturing, manufacturing
research and technology, vol 20. Elsevier, pp 107–128. https://doi.org/10.1016/B978-0-444-
81600-9.50011-0
13. Gao S, Shah J (1998) Automatic recognition of interacting machining features based on min-
imal condition subgraph. Comput Aided Des 30(9):727–739. https://doi.org/10.1016/S0010-
4485(98)00033-5
14. Goyvaerts J, Levithan S (2012) Regular expressions cookbook: detailed solutions in eight
programming languages. O’Reilly Media
15. Heckmann O, Schneider F (2017) Grundrissatlas Wohnungsbau. Birkhäuser
16. Hershberger J, Snoeyink J (1998) Erased arrangements of lines and convex decompositions of
polyhedra. Comput Geom 9(3):129–143. https://doi.org/10.1016/S0925-7721(97)00024-2
17. Hillier B, Hanson J (1989) The social logic of space. Cambridge University Press, Cambridge
18. Jeong SK, Ban YU (2011) Developing a topological information extraction model for space
syntax analysis. Build Environ 46(12):2442–2453. https://doi.org/10.1016/j.buildenv.2011.05.
024
19. Joshi S, Chang T (1988) Graph-based heuristics for recognition of machined features from a 3D
solid model. Comput Aided Des 20(2):58–66. https://doi.org/10.1016/0010-4485(88)90050-
4
20. Lin WY, Lin PH (2018) Intelligent generation of indoor topology (i-GIT) for human indoor
pathfinding based on IFC models and 3D GIS technology. Autom Constr 94:340–359. https://
doi.org/10.1016/j.autcon.2018.07.016
21. Lin YH, Liu YS, Gao G, Han XG, Lai CY, Gu M (2013) The IFC-based path planning for 3D
indoor spaces. Adv Eng Inform 27(2):189–205. https://doi.org/10.1016/j.aei.2012.10.001
22. Liu L, Li B, Zlatanova S, van Oosterom P (2021) Indoor navigation supported by the Indus-
try Foundation Classes (IFC): a survey. Autom Constr 121:103436. https://doi.org/10.1016/j.
autcon.2020.103436
23. Lu Y, Gadh R (2000) Constrained and aggregated half space volume decomposition: gener-
ating cutting patches with “internal” and “external” extending. In: Proceedings of the eighth
Pacific conference on computer graphics and applications, pp 262–271. https://doi.org/10.
1109/PCCGA.2000.883949
24. Magistratsabteilung 25, Gruppe Miet- und Nutzwertberechnung: Leitfaden der MA 25 zur
Berechnung der Nutzfläche nach dem MRG/WEG igF
456 G. Suter

25. Ochmann S, Vock R, Wessel R, Klein R (2015) Automatic reconstruction of parametric building
models from indoor point clouds. Comput Graph 54:94–103. https://doi.org/10.1016/j.cag.
2015.07.008. Special issue on CAD/Graphics 2015
26. Pratt M, Wilson P (1988) Requirements for support of form features in a solid modelling
system. Computer Aided Manufacturing-International
27. Ribelles J, Heckbert P, Garland M, Stahovich T, Srivastava V (2001) Finding and removing fea-
tures from polyhedra. In: International design engineering technical conferences and computers
and information in engineering conference, volume 2A: 27th design automation conference,
pp 561–570. https://doi.org/10.1115/DETC2001/DAC-21068
28. Sakurai H (1995) Volume decomposition and feature recognition: part I-polyhedral objects.
Comput Aided Des 27(11):833–843. https://doi.org/10.1016/0010-4485(95)00007-0
29. Shah JJ, Anderson D, Kim YS, Joshi S (2000) A discourse on geometric feature recognition
from CAD models. J Comput Inf Sci Eng 1(1):41–51. https://doi.org/10.1115/1.1345522
30. Shen Y, Shah J (1994) Feature recognition by volume decomposition using half-space partition-
ing. In: American society of mechanical engineers, design engineering division (publication)
DE, proceedings of the 1994 ASME design technical conferences. Part 1, vol 69-1. ASME, pp
575–583
31. Spatial Technologies, Inc. (2019) 3D ACIS modeler 2019.1.0.2. http://doc.spatial.com.
Accessed 5 Dec 2020
32. Spatial Technologies, Inc. (2019) 3D ACIS modeler 2019.1.0.2, cellular topology data structure.
http://doc.spatial.com. Accessed 5 Dec 2020
33. Suter G (2022) Modeling multiple space views for schematic building design using space
ontologies and layout transformation operations. Autom Constr 134:104041. https://doi.org/
10.1016/j.autcon.2021.104041
34. Thiemann F, Sester M (2004) Segmentation of buildings for 3D-generalisation. In: Proceedings
of the ICA workshop on generalisation and multiple representation, Leicester, UK, pp 20–21
35. Verma AK, Rajotia S (2010) A review of machining feature recognition methodologies. Int J
Comput Integr Manuf 23(4):353–368. https://doi.org/10.1080/09511921003642121
36. ÖNORM (2011) ÖNORM B 1800, Ermittlung von Flächen und Rauminhalten von Bauwerken
und zugehörigen Außenanlagen
Author Index

A F
Abadia, Jose Joaquin Peralta, 3 Ferschin, Peter, 153
Adekunle, Samuel, 45, 357 Francis, Adel, 53
Agam, Gady, 311
Aigbavboa, Clinton, 45, 139, 227, 357
Akinradewo, Opeoluwa, 139 G
Albeaino, Gilles, 215
Gabler, Felix, 391
Aliu, John, 227
Gheisari, Masoud, 215
Anumba, Chimay, 81

B H
Blay, Karen, 329 Hensel, Michael, 153
Boya, Ayanda, 139 Hsieh, Shang-Hsien, 99
Huhnt, Wolfgang, 391, 403, 415, 427
Huhnt, Wolgang, 377
Hwang, Seongha, 329
C
Castets, Édouard, 311
Chen, Sou-Han, 245
Clemen, Christian, 377 I
Ikuabe, Matthew, 45
Issa, Raja R. A., 215
D Ito, Armando Luis Yoshio, 197
de Sousa Carboni, Márcio Henrique, 197
Dinka, Megersa, 173
Dragos, Kosmas, 127 K
Dzuwa, Christopher, 281, 297 Kang, Bo-Rui, 11
Khairtdinov, Marat, 269
Koch, Christian, 269
E Kovacic, Iva, 153
Ebekozien, Andrew, 139 Kovacs, Balint Istvan, 153
Ergan, Semiha, 63 Kumar, Pavan, 99

© The Editor(s) (if applicable) and The Author(s), under exclusive license 457
to Springer Nature Switzerland AG 2023
S. Skatulla and H. Beushausen (eds.), Advances in Information Technology in Civil and
Building Engineering, Lecture Notes in Civil Engineering 358,
https://doi.org/10.1007/978-3-031-32515-1
458 Author Index

L Reisinger, Julia, 153

Lawal, Oluwatoyin O., 345 Romanschek, Enrico, 377
Lin, Jia-Rui, 11
Lu, Daniel, 63
Lu, Xin-Zheng, 11 S
Scheer, Sérgio, 197
Shaba, Martin, 173
M Smallwood, John, 257
Malifete, Linda, 357 Smarsly, Kay, 3, 127
Manga, Ashvin, 257 Smuts, Damien, 257
Matsimbe, Jabulani, 173 Sternal, Maximilian, 415, 427
Miresco, Edmond, 53 Stührenberg, Jan, 127
Mugaga, Rotshidzwa, 45 Suter, Georg, 443
Musonda, Innocent, 27, 109, 173, 281, 297
Mutis, Ivan, 311
T
Tanga, Ornella, 227
N Tasliarmut, Fulya, 269
Nawari, Nawari O., 345 Tjebane, Motheo Meta, 27, 109

V
O
Vasylevska, Khrystyna, 153
Ogunbayo, Babatunde, 45
Vetter, Joanna Zarah, 403, 415
Okafor, Chika, 227
Onososen, Adetayo, 27, 109, 281, 297
Osmani, Mohamed, 329
W
Wang, Mingzhu, 329
Wimmer, Michael, 153
P Worm, Mathias, 127
Pal, Aritra, 99
Papadonikolaki, Eleni, 81
Paulis, Sebastian, 215 X
Xue, Fan, 245

Q
Qin, Julia, 63 Y
Yuan, Qi-Tian, 11

R
Ramabodu, Molusiwa, 27, 109, 139, 281, Z
297 Zheng, Zhe, 11
Rasoulzadeh, Shervin, 153 Zhou, Yu-Cheng, 11