Available online at www.sciencedirect.

com
Available online at www.sciencedirect.com

ScienceDirect
ScienceDirect
Available online atonline
Available www.sciencedirect.com
at www.sciencedirect.com
Procedia CIRP 00 (2019) 000–000

ScienceDirect
Procedia CIRP 00 (2019) 000–000
www.elsevier.com/locate/procedia
ScienceDirect www.elsevier.com/locate/procedia

Procedia CIRP 00 (2017)

Procedia 000–000
CIRP 84 (2019) 650–655
www.elsevier.com/locate/procedia

29th CIRP Design 2019 (CIRP Design 2019)

29th CIRP Design 2019 (CIRP Design 2019)
Cost
Cost Estimation
Estimation Method
Method
28th
for Gas
Gas Turbine
forConference,
CIRP Design Turbine in
in Conceptual
Conceptual
May 2018,
Design Phase
Nantes, France Design Phase
Irene Martinellia*, Federico Campib, Emanuele Checcaccia, Giulio Marcello Lo Prestia,
A new
Irenemethodology to analyze
Pescatoriaa,the
Martinellia*, Federico
Francesco
Campib, Emanuele functional
Antonio
Checcacci
Pumoaa, Michele and a
physical
Germanibb architecture of
, Giulio Marcello Lo Prestia,
Francesco Pescatori , Antonio Pumo , Michele Germani
existing products forBHGE,
anaassembly oriented
GE Company, Via Felice
a
Matteucci, 2, product family identification
50127, Firenze, Italy
b a
BHGE,
Department of Industrial Engineering and a GE Company,
Mathematical ViaUniversità
Sciences, Felice Matteucci, 2, 50127,
Politecnica Firenze,Via
delle Marche, ItalyBrecce Bianche 12, 60131, Ancona, Italy
b
Department of Industrial Engineering and Mathematical Sciences, Università Politecnica delle Marche, Via Brecce Bianche 12, 60131, Ancona, Italy
Paul Stief *, Jean-Yves Dantan, Alain Etienne, Ali Siadat
* Corresponding author. Tel.: +39-055-423-2168 . E-mail address: irene.martinelli@bhge.com
* Corresponding author. Tel.: +39-055-423-2168 . E-mail address: irene.martinelli@bhge.com
École Nationale Supérieure d’Arts et Métiers, Arts et Métiers ParisTech, LCFC EA 4495, 4 Rue Augustin Fresnel, Metz 57078, France

*Abstract
Corresponding author. Tel.: +33 3 87 37 54 30; E-mail address: paul.stief@ensam.eu
Abstract
Introduction of new gas turbine machines on market is a complex project that requires optimization of different performance parameters such as
Introduction of newmaintenance
power, efficiency, gas turbine plan, machines
producton market
cost andis life.
a complex project
The ability that requires
to control optimization
cost and impact onof different performance
performances parameters
and life strongly such as
decreases
power, efficiency, maintenance
from conceptual to detailed design phase. plan, product cost and life. The ability to control cost and impact on performances and life strongly decreases
Abstract
from conceptual
Actually, 80 % oftoproduct’s
detailed design
cost and phase.
performances are committed based on decisions made in conceptual design.
Actually,
This Paper % of product’s
80describes a cost and
systematic performances
procedure are committed
to estimate based on decisions made in conceptual design. design phase, comparing
InThis
today’s
Paperbusiness
describes environment,
a systematic the trend towards
procedure morethe
to estimate the
cost ofvariety
product multiple
cost of multiple anddesign
design
alternatives
customization
alternatives
during conceptual
is unbroken. Due to this
during conceptual development,
design the need of
phase, comparing
different
agile cross sections
and reconfigurable for gas
production turbine solutions.
systems emerged to cope with various products and product families. To design and optimize production
different cross sections for gas turbine solutions.
Examples of parametric costing tool for part family will be described, to show the approach that allows to estimate costs in conceptual design
systems
Examplesas of
well as to choose
parametric costing the optimal product
familymatches, product analysis methods are needed. Indeed, most of the known methods aim to
phase, when detailed design has tool for part
not been developed will
and be described,
lack to show
of information is the approach
a daily topic. that allows to estimate costs in conceptual design
analyze
phase, a product
when or one product family on the physical level. Different product families, however, may differ largely in terms of the number and
The idea is todetailed
be able design
to readhas not been
design developed
information and part
of each lack from
of information
an enhanced is across
dailysection
topic. and enter parametric costing tool to have a
nature
The of
idea components.
is to be able toThis
read fact impedes
design
preliminary cost estimation in conceptual phase. an
information efficient
of comparison
each part from and
an choice
enhanced of appropriate
cross section product
and family combinations
enter parametric costing toolfor to the
haveproduction
a
system. A new
preliminary
Doing that cost
for methodology
estimation
each inis conceptual
part or module proposed
present,tophase.
analyze
it will beexisting
possibleproducts in view
to estimate totalof their
cost of functional
the product. and physical architecture. The aim is to cluster
these
Doingproducts
that isfortoin
The scope newpart
each
create assembly
anorinternal
module oriented product
present,
database it will
where families
be whole
the forknow-how
possiblethetooptimization
estimate and ofcost
total
best existing
of theassembly
practices product.
are lines
stored. and
This the creation
database can be ofexamined
future reconfigurable
in early
The scopesystems.
assembly
program is to create
stages, toBased an on
reduce internal
Datum
time to database
Flow and
market where
Chain,avoid thepursuing
the whole know-how
physical structure that
solutions andthe
of best practices
products
would not be are stored.
is viable
analyzed. This database
Functional
or convenient, can be
subassemblies
in a sort examined
are twin
of digital inapproach.
early and
identified,
aprogram
Another stages,
functional toaspect
analysis
positive reduce time toand
is performed.
pursued market and avoid
Moreover,
presented, pursuing
aishybrid
the solutions
functional
positive impact andonthat wouldarchitecture
physical
engineering notproductivity,
be viable or convenient,
graph directlyinreflects
(HyFPAG)
that isa the
sortoutput
of
on digital twindepicts
which
program approach.
the
Another
similarity positive
developmentbetweencost.aspect
productpursued
families and bypresented,
providing is the positive
design impact
support to on
both,engineering
production productivity, that directly
system planners reflectsdesigners.
and product on program An illustrative
development
example cost.
of a nail-clipper is used to explain the proposed methodology. An industrial case study on two product families of steering columns of
thyssenkrupp Presta France
© 2019 The Authors. is then
Published bycarried
Elsevier outB.V.
to give a first industrial evaluation of the proposed approach.
© 2019
© 2019 The
The Authors.
Authors. Published
Publishedby by
by Elsevier B.V.
Elsevier B.V.
©Peer-review
2017 The Authors.
under
Peer-review under Published
responsibility
under responsibility
responsibility of ofElsevier
the
of the B.V.
scientific
scientific committee
the scientific committee of of the
the CIRP
CIRP Design
Design Conference
Conference 2019 2019.
Peer-review
Peer-review under responsibility of the committee
scientific committee of
of the
the CIRP
28th Design
CIRP Conference
Design Conference2019 2018.
Keywords: new product introduction, gas turbine, target cost, conceptual costing, parametric cost estimation
Keywords:Assembly;
Keywords: new product introduction,
Design method; gas turbine,
Family target cost, conceptual costing, parametric cost estimation
identification

1. Introduction and State of Art higher, up to 60 % [3]. Proceeding in design development, cost
1.1.Introduction
Introduction and State of Art higher,
of
of the up increases
to 60range
product
change % [3]. Proceeding
asand in [4].in design
characteristics
stated development,
manufactured cost
and/or
New Product Introduction is quite a complex procedure assembled of change
Should inincreases
cost as
thisatsystem. stated
conceptual in
In this [4].
context,
design is slowthe and
mainunreliable
challengedue in
whileNew
Due Product
to
dealing withIntroduction
the fast
gas turbine is quite in
development
product. a complex
the domain procedure of to aShould
modelling
lack of cost
and at conceptual
a analysis
specific is now
tool, design
andnotthisis slow
only to and
limits the unreliable
cope due
with single
evaluation of
while dealing
communication
A lot with
andgasanturbine
of performance ongoing
targetsproduct.
trendbe of
shall met,digitization
from powerand to a lackalternatives.
to products,
design aoflimited
a specific tool,range
product and or
Development thisexisting
limits the evaluation
product
and validation of ofa
families,
A lot offrom
digitalization,
efficiency, performance
manufacturing targets
manufacturability shall
enterprises beare met,
to maintenance. from
facing power
important
Each to but
single design
also toalternatives.
configurationbe able Development
to analyze
at conceptual and torequires
stage andmonths,
compare validation
products of thea
to define
limiting
efficiency,
challenges from manufacturability
in today’s
choice is a trade-off studymarket
between to maintenance.
environments:
different scenarios, Each
a continuing single
to select newconfiguration
product of
exploration at conceptual
families. stage requires months, limiting
It can be observed that classical existing
design alternatives. the
choice
tendency is a trade-off
optimizedtowards
solution study
reduction between
of product
for a specific different scenarios,
development
application. Cost astimes to select
well and exploration
can product
Timefamiliesof design
constraints alternatives.
are regrouped
prevent in functioncost
effective of clients or features.
optimization in
optimized
shortened
be considered solution
product for a specific
anlifecycles.
internal In
target application.
addition, there Cost
to match, an as
isand itswell
increasing can However,
related Time constraints
conceptual assembly
design orientedprevent
phase, effective
asproduct
stated costand
families
in [5] optimization
are hardly to find.
[6], since in
no
be
demandconsidered
businessofcase an internal
customization,
for a specificbeing target to match, and
at the same time in a global
NPI project. its related conceptual design
On the product
automation phase,
familyfor
is available as stated
level, in
costproducts [5] and [6], since
differ mainly checks.
versus performance in twono
business
competition casewith
As explained forin
acompetitors
specific NPIall
[1] and [2], project.
over80the
about % world.
of cost This trend,
is locked-in automation
main is available
Valuecharacteristics:
estimation for
(i) the
is used becost
to number versus performance
of components
a qualitative procedure checks.
andbased
(ii) the
on
which As explained
is inducing
at conceptual in [1]
design the and [2], about
development
stage, when 80
project % ofmacro
from
teams cost is to
freeze locked-in
themicro Value
cross type of estimation
components
intuition of an expertis (e.g.
used toonbeanalogies
a qualitative
ormechanical, procedure
electrical, based on
electronical).
of similar products, as
at conceptual
markets,
section the design
ofresults new stage, when
in machine
diminished evenlotproject
sizesteams
though freeze
dueknowledge
the to thegap
augmenting cross
is intuition
Classical
described of an expert
[7]. or onconsidering
bymethodologies
Niazi analogies mainly
of similar products,
single products as
section of the new machine even though the
product varieties (high-volume to low-volume production) [1]. knowledge gap is described by Niazi [7].
or solitary, already existing product families analyze the
To cope with this augmenting variety as well as to be able to product structure on a physical level (components level) which
identify
2212-8271 possible
© 2019 The optimization
Authors. Publishedpotentials in the existing
by Elsevier B.V. causes difficulties regarding an efficient definition and
2212-8271
Peer-review©under
production 2019responsibility
system,The it
Authors. Published
is important
of tobyhave
Elsevier
the scientific B.V.of the
a precise
committee knowledge comparison
CIRP Design Conference 2019 of different product families. Addressing this
Peer-review under responsibility of the scientific committee of the CIRP Design Conference 2019

2212-8271©©2017
2212-8271 2019The
The Authors.
Authors. Published
Published by Elsevier
by Elsevier B.V. B.V.
Peer-review
Peer-review under
under responsibility
responsibility of scientific
of the the scientific committee
committee of the of theCIRP
28th CIRP Design
Design Conference
Conference 2019.
2018.
10.1016/j.procir.2019.04.311
 Irene Martinelli et al. / Procedia CIRP 84 (2019) 650–655 651
2 Author name / Procedia CIRP 00 (2019) 000–000

Ability to reduce cost decreases while cost accuracy Results on this specific case of study will be discussed and
increases with ongoing time in design process as described in extended to other items to complete the parametric cost
Figure 1. mapping of all the possible items. The aim is to be able to
evaluate at least all the scale and mix of scenarios already
produced in factory.

Nomenclature

BOM Bill of Material

GT Gas Turbine
LCC Life Cycle Cost
NPI New Product Introduction
PBS Product Breakdown Structure
PMI Product and manufacturing information
PO Placed orders
SC Should Cost
Fig. 1. Impact on Cost Vs. NPI program stage - © 2019 Baker Hughes, a GE TDT Total Drilling Time
company, LLC - All rights reserved. TGT Total Grinding Time
The paper objective is to describe a systematic approach to
2. Materials and Methods
compare different design options and predict cost implications
through parametric cost models developed internally.
2.1. Parametric Cost Estimation for GTs
The idea is to leverage parts cost drivers to generate quick
cost assessments for each item [8] and [9]. By doing that for
The aim is to leverage GT parts cost drivers to generate
each item of the bill of material, and adding the assembly
quick cost assessments, doing that in an accurate but also quick
procedures cost contribution, it is possible to evaluate total
way during conceptual or preliminary in design, when lack of
product cost. Xu et all, in [10], cover several key areas in Cost
information is inherent to the stage of the project.
Engineering, from design to manufacturing cost, from
At Step 1 a sketch, so called Cartoon, is generated: Cartoon
operating to life cycle cost (LCC) and risk costs. Gu et all, in
is a preliminary enhanced cross section, enriched of
[11], deal with implicit costs and quality cost efficiency, giving
information and used in this stage to compare different
a wider overview to product total cost using analytic method.
conceptual solutions.
On the contrary, Cicconi et all, in [12], present a parametric
Cartoon generation is related to product structure, from a
approach to cost estimation during embodiment design phase.
Cartoon it is possible to generate a PBS (step 2), as described
The novelty of this paper is related to the specific field of
also in [15] and [16], and vice versa when a PBS is selected it
interest (gas turbines). No literature cost models where
is possible to create a Cartoon of the gas turbine.
available, due to the high complexity of the product, but also to
The Cartoon is created by a software tool called Cartoon
the fact that cost models cannot leave aside the supply chain
Generator and it will be composed of simplified model with
used to manufacture all the components, relying both on
preliminary attributes and features: attributes and PMIs
external suppliers and on internal capabilities, and pushing to
necessary to provide information for Cost Estimation. Cartoon
optimize balance between cost and lead time.
Models are completely parametric and associative. Step 3 is
This is for sure the first goal achieved, while a depth internal
related to simplify model with preliminary attributes regarding
knowledge on cost assessment is under construction inside
geometry and PMIs for cost estimation.
design engineering teams.
From step 3 it is possible to enter parametric cost estimation
This article will deal with the method of parametric
at step 4 for each item included into PBS.
estimation, describing how it is related with the parametric
Parametric cost estimation will provide components cost
simplified model with preliminary attributes and how it fits
using configured gas turbine database as input. See fig.2 to
specific product structures, in a feature-based approach as
understand flow information.
described in [13].
This procedure allows to estimate the costs of multiple
A case of study will be presented: Axial Compressor' Disks
design alternatives during conceptual design phase, comparing
cost models for both forging and machining procedures will be
different Cartoon cross sections for gas turbine solutions.
deeply analyzed, describing which are the main contributions
to the cost achieved and how we can predict value of a new
design.
Manufacturing and assembly procedure are properly
evaluated to understand if some criticalities will lead to cost
increase [14].
652 Irene Martinelli et al. / Procedia CIRP 84 (2019) 650–655
Author name / Procedia CIRP 00 (2019) 000–000 3

Step 2 is the first attempt to generate a cost model: a

preliminary supposition of main cost drivers is made based on
experience and previous cost models realized and then to find
correlation and identify cost drivers (step 2). Some items and
parts, even if not properly of the same family type, have been
considered together because of material type communality and
machining cycles similarities.
As a first attempt, screening and data analysis looks for
proportionality between cost and simple parameter such as
weight and preliminary parametric cost regression curves are
calculated in Step 3. In Step 4 R2 is calculated as described in
[17] and in case regression fit data points within acceptable
Fig. 2. Parametric Cost Estimation Structure - © 2019 Baker Hughes, a GE accuracy the process is completed.
company, LLC - All rights reserved.
For R2 higher that 0.9 a satisfying correlation is already met.
R values between 0.6 and 0.9 suggest proceeding with a deeper
2
2.2. Parametric Cost Models Development analysis of data, to identify more cost drivers. R 2 values lower
than 0.6 are not acceptable and no correlation can be used for
The idea is to be able to read design information of each part
further studies.
from an enhanced cross section and enter parametric costing Step 5 deals with database maintenance over years. Curves
tool to have a preliminary cost estimation in conceptual phase. shall be updated during time to take into account new data
Doing that for each part or module present, it will be possible
points introduction while new orders have been placed.
to estimate total cost of product. Assembly cost shall be Sometimes more complex cost drivers shall be considered
considered as well to correctly estimate total product cost. and a deeper drill down analysis is needed, causing an iteration
Input are read by CAD or automatically imported from
loop. Statistical analysis can be needed to identify outliers and
geometrical features or manual input at the very beginning of refine cost models. Only when a good correlation is reached,
the project and elaborated through parametric cost curves to the parametric cost model can be considered viable to be
estimate part value.
included into parametric costing tool.
The flow chart of the following steps is described in fig.3. An important aspect is related to supply chain identification,
In step 1, all the orders of items belonging to a same family in fact a major difference in terms of supply chain is between
group are collected together. To populate structure and database
full buy item, purchased by external suppliers, and make
of parametric cost estimation first of all it is necessary to collect components, for which internal capability is used.
data from historical PO. Each individual machining phase of a cycle have been
described in terms of the following base operations and
evaluated through a specific parametric equation.
Typical base operations are drilling, turning, milling,
grinding, painting, balancing and assembly.
For each machining or operation type, set up hours as well
as machining time and cost related to inspections, pack and
shipment shall be considered and properly estimated.
Cost parametric function can be queried in an automatic way
or, while the tool development is still in progress, to maximize
and anticipate benefits of the tool itself, even with a manual
approach, using a manual input interface.
Parametric costing tool and equations for a specific part
family (disks) will be described in the next paragraph to show
the approach that allows to estimate costs in conceptual design
phase, when detailed design has not been developed yet and
lack of information is a daily topic.

3. Case Study

Rotor disks total cost can be estimated by two main

contributions. Raw material cost is the cost of the forged part,
supplied by external partners (farm-out). On the contrary,
rough and finish machining of disks is manufactured with a
Fig. 3. Parametric Regression Curves Flow Chart - © 2019 Baker Hughes, a
farm-in approach. Raw material parametric curves have been
GE company, LLC - All rights reserved. calculated by dividing POs into clusters, depending on material
 Irene Martinelli et al. / Procedia CIRP 84 (2019) 650–655 653
4 Author name / Procedia CIRP 00 (2019) 000–000

of disk. That means that the very first cost driver is the material Internal bore present yes=1/no=0 Groove int diameter [mm]
selected. Afterwards, for forged parts is it possible to find a Internal bore diameter [mm] Groove depth [mm]
very good correlation between part weight and item cost. Many
Number of tire rod [#] Number of blades [#]
curves have been defined, one for each material available in our
database selection, and entering the proper material curve, with Tie rod bore diameter [mm] Width of blades base [mm]
the expected part weight, it is possible to have a raw material Overstock material [mm] Depth of blades base [mm]
cost estimation.
Groove features yes=1/no=0 Grinding surface height [mm]
In figure 4, parametric costing curves are shown for three
different materials A, B and C. Parametric equations for two based operation are described
As shown by R2 values, some time is it possible to find a below: drilling and grinding.
good correlation, some other correlation is lower. For drilling and grinding phases we have the following
A lower correlation can have different root causes. inputs:
One of that is the presence of outlier points (due for example
to higher cost payed in case of urgency in procurement or lower Table 2. Drilling Phase Technological and Geometrical Inputs - © 2019
Baker Hughes, a GE company, LLC - All rights reserved.
cost payed in case of bundle buy volumes). In case an outlier
point is identified it will be removed from the data point group. Technological Input Description Unit
Another possible explanation is the presence of other tdrilling Drilling time [s]
important cost drivers not addressed yet. In that case a deep treversal Reversal time [s]
dive into cost model generation and definition is needed to
guarantee more accurate prediction. Only in case of real tload/unload Load/Unload time [s]

business urgency also costing curves with lower fitting can be Nreversal Number of reversal [#]
used, adding a proper contingency value to consider variability Nload/unload Number of load/unload [#]
of the prediction.
tsetup Set up time [s]

tOnline inspection Online inspection [s] = 0.16

tOperator physiological Operator physiological time [s] = 0.05

T Thickness [mm]

dbore Bore diameter [mm]

lbore Bore length [mm]

h height of grinding surface [mm]

O Overstock [mm]

Parametric cost equation for drilling contribution is the

following:

𝑡𝑡𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 = 𝑎𝑎 + 𝑏𝑏 ⋅ 𝑙𝑙𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 + 𝑐𝑐 ⋅ 𝑑𝑑𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 (1)

Fig. 4. Disks Raw Material Cost Vs. Weight - © 2019 Baker Hughes, a GE
Where a, b, and c are constant calculated from regression,
company, LLC - All rights reserved.
Drilling Time is measured in seconds and bore length and
For what concern machining cost, a first attempt has been diameter in mm. Bore length is calculated as follows:
made trying to correlate cost with delta weight between forged 𝑙𝑙𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 = 𝑇𝑇 + 𝑂𝑂 (2)
design and machined one of the same component.
The result is a leak correlation among numbers which lead
to a deep dive analysis, looking for a better prediction. Total Drilling Time TDT is then calculated as follows:
First we defined the minimum set of geometrical parameters
needed to describe item geometry of the machined part as TDT = [(treversal ⋅ NReversal )
mentioned in Table 1. + (tload/unload ⋅ Nload/unload )
+ tdrilling ] + k[ tload/unload (3)
Table 1. Minimum set of geometrical parameters for disks - © 2019 Baker
Hughes, a GE company, LLC - All rights reserved. ⋅ Nreversal + ( 𝑙𝑙𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 ⋅ Nload/unload )
Parameter Units Parameter Units ⋅ tdrilling ] + 𝑡𝑡𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
External Diameter [mm] Groove quantity [#]
Wherby:
Thickness [mm] Groove ext diameter [mm] 𝑘𝑘 = 𝑡𝑡𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 + 𝑡𝑡𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂 𝑝𝑝ℎ𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦 (4)
654 Irene Martinelli et al. / Procedia CIRP 84 (2019) 650–655
Author name / Procedia CIRP 00 (2019) 000–000 5

For the entire grinding time of the component, the equation effort is the positive impact on engineering productivity, that
is as follows: directly reflects on program development cost.

𝑇𝑇𝑇𝑇𝑇𝑇 = [(treversal ⋅ NReversal ) + (tload/unload 5. Conclusions

⋅ Nload/unload ) + (𝑎𝑎 ⋅ ℎ + 𝑏𝑏) ]
+ 0.05 [(treversal ⋅ NReversal ) The major aim is to compare multiple design options and
(5)
+ (tload/unload ⋅ Nload/unload ) predict cost implications through design tools and cost model
+ (𝑐𝑐 ⋅ ℎ + 𝑑𝑑) + 0.16 (𝑒𝑒 ⋅ ℎ integration into each design phase, since early stages.
+ 𝑓𝑓) + 𝑡𝑡𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 This example on axial compressor rotor shows that the
approach is a strategic milestone even if a drawback consists in
Where a, b, c, d, e and f are constants calculated through parametric cost equation maintenance effort. Actually,
regressions, as described in [16] and relative mean absolute database shall always be updated considering new PO placed
error and root mean squared error are calculated with the aim and scouting of new suppliers.
to predict accuracy of interpolation made using these equations. A further development in cost prediction and optimization
The database we are building contains BHGE proprietary can be reached through co design with suppliers, automated
information data and cannot be shared in its completed form, tools linked to the CAD system and for sure enlarging items
but method described is general and example given for disk part database adding all the possible families for each applicable
family is representative of the approach followed. size in production with the scope is to create an internal
Drilling and Grinding are only two factors of the whole database where the whole know-how and best practices are
parametric equation for disk machining. Each feature shall be stored.
realized on a specific machine and with a specific cost
associated and time needed to perform the operation, and the References
scope of all these parametric equation is to describe as more
[1] Budiono HDS, Kiswanto G, Soemardi TP.Method and Model Development
accurate as possible the machining cycle to predict total lead for Manufacturing Cost Estimation during the early design phase related to
time and cost. For the disk case of study, the gap between total the complexity of the machining processes. International Journal of
cost estimated with parametric method and with analytic ones Technology 2014; 2: 183-192.
is always lower than 3 %. [2] Hooshmand Y, Köhler P, Korff-Krumm A. Cost Estimation in Engineer-
to-Order Manufacturing. Open Engineering formerly Central Europea n
Journal of Engineering 2016; 6:22-34.
4. Results and Discussion [3] Lukic D, Milosevic M, Borojevic S, Durdev M, Vukman J, Antic A.
Manufacturing Cost Estimation in the Conceptual Process Planning.
As described in previous paragraphs, the scope is to create Machine Design 2016; 8(3):83-60.
an internal database where the whole know-how and best [4] Ehlhardt H. Computer Aided Cost Estimating.16th International Conference
practices are stored. This database will be examined in early on Engineering and Product Design Education 2014; DS78.
https://www.researchgate.net/publication/270896291, September 2014
program stages, to reduce time to market and avoid pursuing [5] Saravi, M., Newnes, L., Mileham, A., & Goh, Y-M. (2008). Estimating
solutions that would not be viable or convenient. Cost at the Conceptual Design Stage to Optimize Design in terms of
Total product cost is obtained considering the cost estimated Performance and Cost. 123-130. Paper presented at Proceedings of the 15th
for each item or operation from the bill of material, and this ISPE Int Conf on Concurrent Engineering (CE2008), Queens Belfast, .
means that a specific cost model for assembly operation shall https://doi.org/10.1007/978-1-84800-972-1_11
[6] BARONE, G. and FRANGOPOL, D.M., 2014. Life-cycle maintenance of
be finalized as well. deteriorating structures by multi-objective optimization involving
As already stated, cost assessment is very important since reliability, risk, availability, hazard and cost. Structural Safety, 48, pp.40-
early project stages, where the impact on total product cost can 50.
be maximized. On the other and, a trade-off between cost [7] Niazi, A ; Dai, J S ; Balabani, S ; Seneviratne, L. / Product cost estimation:
estimation availability and accuracy shall be met. In fact, at the Technique classification and methodology review. In: JOURNAL OF
MANUFACTURING SCIENCE AND ENGINEERING-
very beginning of the project not all the information is TRANSACTIONS OF THE ASME. 2006 ; Vol. 128, No. 2. pp. 563-575
available, and some assumptions are needed. Not all the cost [8] Filippazzo G. Parametric Cost Modeling of an Activity Based Proposal for
contributions have the same accuracy: in fact, for example, a System of Systems. IEEEAC paper #1123, Version 2, December 15,
machining cost are easy to estimate, leveraging experience of 2004.
internal manufacturing capabilities. All the outsourced scope of [9] Ardiansyah R, Sutopo W, Nizam M. A Parametric Cost Estimation Model
to Develop Prototype of Electric Vehicle based on Activity-based Costing.
work, i.e. forged parts or some specific coatings, will have a 2013 IEEE. 978-1-4799-0986-5/13 .
higher uncertainty in cost estimation. [10] Xua Y, Elghb F, Erkoyuncua JA, Bankolea O, Gohc Y, Cheungd WM,
A mitigation of this aspect is obtained engaging suppliers Baguleya P, Wange Q, Arundachawata P, Shehaba E, Newnesf L, Roya R.
since the early stage of the program. The idea is both to have Cost Engineering for manufacturing: Current and future research.
an accurate cost estimation, but also leverage their experience International Journal of Computer Integrated Manufacturing. 2011, 1–15
[11] Huijuan Lin, Yanglin Li, Wanxin Li. An Empirical Analysis of Activity
and know-how and implement positive feedbacks from their Based Costing in Chinese Enterprises. Journal of Finance and Accounting.
experience. Co-design approach with suppliers will help to Vol. 4, No. 5, 2016, pp. 301-309. doi: 10.11648/j.jfa.20160405.17
meet the target as well as to predict cost in early stages in an [12] Cicconi, P., Germani, M., & Mandolini, M. (2010). How to Support
accurate way. Another positive aspect pursued with all this Mechanical Product Cost Estimation in the Embodiment Design Phase. In
 Irene Martinelli et al. / Procedia CIRP 84 (2019) 650–655 655
6 Author name / Procedia CIRP 00 (2019) 000–000

P. Jerzy, F. Shuichi, & J. Salwìnski (Eds.), New world situation: new [15] Panchenko Y, Moustapha H, Mah S, Patel K, Dowhan MJ, Hall D.
directions in concurrent engineering (Vol. 1, pp. 465-477). Cracow. Preliminary Multi-Disciplinary Optimization in Turbomachinery Design.
http://doi.org/10.1007/978-0-85729-024-3_45 RTO-MP-089
[13] ] Sajadfar, Narges, Luis Campos Triana and Ma Y.-S. "Interdisciplinary [16] Gu Y, Guo L, Sun Q. The research of Quality Cost Efficiency with
semantic interactions within a unified feature model for product cost Considering the Implicit Cost Based on Non-parametric Analysis. 2009
estimation." International Journal of Mechanical Engineering and Second International Symposium on Electronic Commerce and Security.
Mechatronics (2014): 10-19 ISECS.2009.244
[14] Roy, R., et al., Detailed cost estimating in the automotive industry: Data [17] Kavitha S, Varuna S, Ramya R. A Comparative Analysis on Linear
and information requirements.International Journal of Production Regression and Support Vector Regression. 2016 Online International
Economics (2011), doi:10.1016/j.ijpe.2011.05.018 Conference on Green Engineering and Technologies (IC-GET). 1-5.
10.1109/GET.2016.7916