https://doi.org/10.3311/PPch.

23166
Creative Commons Attribution b |195
Periodica Polytechnica Chemical Engineering, 68(2), pp. 195–202, 2024

Evaluation of the Open Source Process Simulator DWSIM for

Bioprocess Simulation

Siddhi Sreemahadevan1*, Krishnapriya Velusamy Sivakumar1, Menjith Palanisamy1

1
Department of Biotechnology, PSG College of Technology, Avinashi Road, Peelamedu, Coimbatore, Tamil Nadu 641004, India
*
Corresponding author, e-mail: sid.bio@psgtech.ac.in

Received: 09 August 2023, Accepted: 16 February 2024, Published online: 22 May 2024

Abstract
In the recent times, the application of process simulation software has found its way to the simulation of bioprocesses. Presently,
bioprocesses are simulated and technoeconomic analysis is performed in commercially available software for such as SuperPro
Designer®, Aspen Plus® etc. These softwares are not freely available and therefore, there exists a need for an open source software for
the simulation of bioprocesses. Furthermore, the chemical process softwares show compatibility issues, when used for bioprocesses.
More recently, an open source process simulator called DWSIM was introduced for the simulation of chemical processes. However,
the applicability and compatibility of DWSIM for complex bioprocess simulation involving microbial biomass has not been reported
yet. Therefore, the present study evaluates and reports the simulation results of the bioethanol production process developed by
NREL (National Renewable Energy Laboratory, USA), performed in DWSIM. The simulator results were compared to those of existing
commercial software, and the results showed good agreement with the literature. This suggests that DWSIM could be a promising
alternative for bioprocess flowsheeting and simulation, and further technoeconomic analysis.
Keywords
flowsheeting, bioprocess simulation, DWSIM, bioethanol, open source software, technoeconomic analysis

1 Introduction
Biotechnologists have always tried to identify computational Software-based simulation plays a key role in the design
methods available for efficient practical applications. With of bioprocess plants. There are numerous commercially
the evolution of both hardware and software for computing, accepted software options available, such as Aspen Plus®,
bioprocess engineers are attracted to the recent advances in Aspen HYSYS®, ProSim Plus®, SuperPro Designer®
the same for use in bioprocessing. Bioprocesses are devel- and UniSim®. These software are used to simulate chem-
oped through a combination of media optimization, pro- ical and bioprocesses [6–9], but their high cost makes
cess optimization, process design, and control. This can be them difficult to acquire. Most of the chemical process
a tedious task if performed manually, but becomes easier softwares also present compatibility issues with biologi-
when software is used [1]. The significance of computational cal processes. To address this, the present study focuses
tools lies in their ability to solve complex problems in biopro- on using open-source software like DWSIM, which is
cessing that would otherwise be difficult to address [2]. There freely accessible. Open-source software provides distinct
are three interrelated areas of particular importance in bio- advantages in modeling dynamic processes with its user-
processing, namely process simulation, process monitoring, friendly interface. This accessibility makes it a preferred
and process analysis. The major contribution of bioprocess choice for researchers and professionals. The collabora-
software is in simulating bioprocesses to maximize profits tive and transparent nature of open-source software fos-
by predicting and operating the most efficient processes [3], ters innovation, serving as a valuable resource for diverse
i.e. technoeconomic analysis. This involves demonstrating applications [10]. In contrast to commercial simulators,
how laboratory and pilot plant data can be captured and used open-source alternatives offer benefits for simulating mul-
to build a flowsheet model, estimate process economics, and tidisciplinary plants, allowing access and modification
identify priorities for process improvement [4, 5]. of source code. They are particularly accessible to rural

Cite this article as: Sreemahadevan, S., Sivakumar, K. V., Palanisamy, M. "Evaluation of the Open Source Process Simulator DWSIM for Bioprocess
Simulation", Periodica Polytechnica Chemical Engineering, 68(2), pp. 195–202, 2024. https://doi.org/10.3311/PPch.23166
196|Sreemahadevan et al.
Period. Polytech. Chem. Eng., 68(2), pp. 195–202, 2024

communities. While commonly used for dynamic simu- Energy Laboratory, USA) report [18] was used for simula-
lations, they have untapped potential for steady-state sim- tion (Fig. 1). The process involves dilute acid pretreatment
ulations [11]. Use of open-source software, hence, can aid of corn stover biomass, enzymatic saccharification of the
in promoting the transfer of technology and dissemina- cellulose and fermentation of the six and five carbon sugar
tion of information in both educational and trade sectors, compounds to ethanol. The major auxiliary processes
enabling a better understanding of various processes [12]. involve handling and storage of feedstock, purification of
Medeiros, a software developer, created the open- product, wastewater treatment and lignin combustion.
source chemical process simulator, DWSIM. As it is
freely available, it allows users to understand the work- 2.2 Software input
ings of chemical process systems. The advantage of using The Aspen flowsheet of NREL process (Fig. 1) consists of
DWSIM is its compatibility with various platforms such 9 areas viz. A100 (feed handling), A200 (pretreatment and
as Windows, Linux, macOS, Android, and iOS, enabling conditioning), A300 (enzymatic hydrolysis and fermenta-
the user to perform steady-state and dynamic simulations tion), A400 (enzyme production), A500 (distillation dehy-
with thermodynamic models and unit operations. The user dration solid separation), A600 (wastewater treatment,
interface is user-friendly and features advanced property WWT), A700 (storage), A800 (burner/boiler turbogener-
packages. The latest version of DWSIM includes a new ator) and A900 (utilities). For simulation in DWSIM, the
feature called the dynamic property, which enables users major steps that involve the biological processes in areas
to integrate Python scripts into the flow sheet to calculate A200 (pretreatment and conditioning), A300 (enzymatic
relevant parameters [13]. DWSIM introduced Version 7, hydrolysis and fermentation) and A500 (distillation dehy-
featuring DWSIM Pro with extensions like additional unit dration solid separation) were only considered. The omis-
operations and property packages. It allows modification sion of other areas was made up by introducing appropri-
of individual units and entire calculation routines in the ate recycle streams or input streams.
flowsheet. DWSIM supports customization of communi-
cation between unit operations and the flowsheet, as well 2.3 Reactants
as between operations. Notably, it excels in creating entire All the reactants involved were given as input in DWSIM
unit operations as custom plugins [14]. This enhances the (version 7.5) [19], either from the existing list (Chemsep) or
applicability of the software for a wide range of processes, after creating them using the Compound Creator Wizard
including bioprocesses. Several studies have reported the available in the software. This creation involved the entry
application and comparison of DWSIM with other process of molecule ID, constant properties, temperature depen-
modeling softwares for chemical processes [15, 16]. More dent properties and molecular structure in the Compound
recently, Elavazhagan et al. [17] have reported the applica- Creator Wizard window manually or by importing data
tion of DWSIM for the simulation of biodiesel production from JSON files available in online repositories such as
by transesterification. However, the same has not been Pubchem [20]. Data for the manual entry of the molec-
reported yet for complex biological processes involving ular properties were obtained from the online reposi-
microbial systems such as fermentations, which require tory Chemeo [21]. Table 1 gives the details of the exist-
creation of biological compounds, adoption of relevant ing compounds and the added compounds in the software.
property packages and creation of unit operations compat- Furthermore, the details of the major properties of the
ible for such biological processes. Therefore, the present compounds added new to the software are summarized in
study was conducted with the following objectives: Part D of the Supporting Information.
• Create components and custom unit operations in
DWSIM for a well-known ethanol production bio- 2.4 Property package and unit operations
process involving microbial fermentation. Following the recommendation of the thermodynamic
• Simulate the bioethanol production process in decision tree discussed by Carlson [22], Peng Robinson
DWSIM and validate with the literature. thermodynamic property package was chosen for the
simulation. The software calculates the thermodynamic
2 Methodology properties of the compounds as explained in Section 2.1.1
2.1 Ethanol production process of the DWSIM user guide [23]. Subsequently the flow
The process of ethanol production from lignocellulosic sheet was generated by adding unit operations and mate-
biomass as described in the NREL (National Renewable rial streams. The units for mixer, cooler, heater, heat
 Sreemahadevan et al.
Period. Polytech. Chem. Eng., 68(2), pp. 195–202, 2024 |197

Fig. 1 Ethanol production process reprinted with permission from the National Renewable Energy Laboratory [18]

Table 1 Reaction components used for simulation

Compound Additional details*
Existing compounds in the software
Ethanol MW: 46
Water MW: 18
Cellulose MW: 163
CO2
Ammonia
Sulphuric acid
O2
Furfural
Added compounds in the software
Glucose MW: 180
Xylose MW: 150
Xylan MW: 132; Formula C5H8O4 (monomer)
Lignin Modelled as vanillin C8H8O3
Other sugars (arabinose, mannose, galactose, cellobiose, Arabinose, mannose and galactose modelled as glucose; cellobiose modelled
sucrose) as sucrose
MW of glucose oligomers: 162; MW of xylose oligomer: 132; Modelled as
Sugar polymers (glucose oligomers & xylose oligomers)
glucose and xylose
Protein MW: 22; Formula: CH1.57O0.31N0.29S0.007; Modelled for corn protein, cellulose
MW: 23; Formula: CH1.64O0.39N0.23S0.0035; Modelled for WWT sludge, Zymomonas
Biomass
mobilis from fermentation, Trichoderma reesei from cellulase production
*
MW – molecular weight
198|Sreemahadevan et al.
Period. Polytech. Chem. Eng., 68(2), pp. 195–202, 2024

exchangers, rectifiers etc. already existed in the DWSIM Where a is the mass flow rate (kg/h) obtained from ASPEN
library. However, the units for the reactions of saccharifi- simulation and b is the mass flow rate (kg/h) obtained from
cation, fermentation, and distillation were user-defined (as DWSIM simulation.
described in Section 2.5) to simulate the process. The pro-
cess flow sheet also included a few recycle water streams 3 Results and discussion
viz. rectifier bottom to pretreatment tank, saccharified 3.1 Simulation results
slurry to inoculum, ethanol water vapour to rectifier bot- The results of the mass flow rates of the process reaction
tom, scrubber bottom to distillatory vessel. components after DWSIM simulation, as well as the values
available in the literature are compared in Fig. 3. The per-
2.5 Reactions and reactors centage of error in the overall mass flow rates ranged
The reactions and their rates of conversions were obtained between 1 and 8.6, as summarized in Table 2. The high-
from the NREL report for pretreatment hydrolysis, enzy- est error was observed for pretreated liquid, contributed
matic hydrolysis, seed train reactions and co-fermenta- by ethanol. Ethanol's contribution to the error percentages
tion reactions (summarised in Part A of the Supporting was reflected consistently in other streams as well. A devi-
Information). The inlet flow rates of the components were ation of the simulation results was expected to occur due
obtained from the NREL report for simulation. Since the to the omission of certain areas in the original process.
default conversion reactors and CSTRs available with Therefore, the recycle stream from the distillation that was
DWSIM had restrictions with the number of input and out- directly routed to the pretreatment has led that error to
put streams, to suit the requirement of pretreatment tank, show an effect on other areas.
saccharification and fermentation tank, seed fermenter and Because of the omission of the cellulase enzyme pro-
distillation column, custom unit operations with python duction step in DWSIM simulation, the saccharified slurry
script were used to simulate the same. To perform this, produced at the end of the simultaneous saccharification
the Dummy Unit Operation was selected from the Object and fermentation step is not recycled for cellulase seed
Palette in the flowsheet window of the software. After production. This has resulted in an error of around 13% in
renaming the new unit operation in the Information tab, the protein flow rate in saccharified slurry.
the Python Script Editor window was opened to input the
necessary Python script. The scripts are detailed in Part B 3.2 Comparison between Aspen and DWSIM
of the Supporting Information). The details related to the The report detailed the simulation of biochemical con-
size of these units are also mentioned in the NREL report. version of lignocellulosic biomass (corn stover) to etha-
nol through a sequential reaction process using DWSIM
2.6 Simulation and comparative study and a comparison with ASPEN Plus. The simulation pro-
The overall flow sheet of the bioethanol production process cess involved the pretreatment of lignocellulosic biomass
that was simulated in DWSIM is given in Fig. 2. The step- to separate the cellulose from lignin. And then, the result-
by-step flowsheet is presented in Part C of the Supporting ing product from the pretreatment reaction was subjected
Information for better visibility. The streams considered to hydrolysis and fermentation to convert into the desired
for the analysis were the properties of the outputs from product ethanol. Simulation of the same process resulted in
pretreatment (pretreated liquid, blow down slurry, flash a minor deviation in DWSIM in comparison with ASPEN
tank, hydrolysate, saccharified slurry, fermented liquid, Plus. This is primarily because of the omissions of certain
distilled ethanol vapor, rectifier ethanol vapor and dehy- reactions from the main process, a few differences in soft-
drated ethanol. Once the flowsheet was solved, the out- ware and their facilities to simulate. Although DWSIM is
let mass flow rates of each component was obtained from a chemical process simulator, because of its latest feature
each section and compared with the ASPEN simulated of the dynamic property, it was able to simulate any pro-
values available in the literature [18]. Further, the differ- cess using Python script editor [13]. Unlike other chemical
ence between the values was calculated using Eq. (1) and process software where users have trouble inserting bio-
reported as error percentage and used for analysis. reactors and biological compounds, in DWSIM, through
a b Python codes, the user can insert the customized reactors
%error  (1)
100 and compounds to simulate the process of interest.
 Sreemahadevan et al.
Period. Polytech. Chem. Eng., 68(2), pp. 195–202, 2024 |199

Fig. 2 Overall flowsheet of bioethanol production

200|Sreemahadevan et al.
Period. Polytech. Chem. Eng., 68(2), pp. 195–202, 2024

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

(j) (k) (l)

(m) (n) (o)

Fig. 3 Comparison of mass flow rates of bioethanol production process in DWSIM and ASPEN: (a) CO2, (b) water, (c) ethanol, (d) xylose, (e) xylan,
(f) protein, (g) lignin, (h) glucose, (i) cellulose, (j) biomass, (k) furfural, (l) sulphuric acid, (m) other sugars, (n) sugar polymers, (o) ammonia
 Sreemahadevan et al.
|201
Period. Polytech. Chem. Eng., 68(2), pp. 195–202, 2024

Table 2 Error of mass flow rates of each stream

Stream (values are in %)
Sl. Component Pretreated Blowdown Flash Saccharified Fermented Distilled Rectifier Dehydrated
Hydrolysate
No. name liquid slurry tank slurry liquid ethanol vapor ethanol vapor ethanol
1 CO2 0.0 0.0 0.0 0.0 0.0 1.5 4.2 4.2 4.2
2 Water 3.6 0.4 0.4 2.6 0.1 0.1 0.9 0.8 0.7
3 Ethanol 13.9 2.5 14.5 14.5 9.6 0.3 1.2 1.1 1.1
4 Xylose 0.0 0.0 0.0 0.0 0.5 0.3 0.0 0.0 0.0
5 Xylan 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
6 Protein 0.0 0.0 0.0 0.0 13.5 1.3 0.0 0.0 0.0
7 Lignin 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
8 Glucose 0.0 0.0 0.0 0.0 0.5 0.7 0.0 0.0 0.0
9 Cellulose 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
10 Biomass 0.0 0.0 0.0 0.0 0.4 0.5 0.0 0.0 0.0
11 Furfural 0.0 0.0 0.0 0.0 0.0 0.3 0.4 0.0 0.0
Sulphuric
12 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
acid
13 Other sugars 0.6 0.6 0.6 0.6 0.0 0.6 0.0 0.0 0.0
Sugar
14 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
polymers
15 Ammonia 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Total mass
8.6 6.2 6.5 7.1 4.8 5.8 1.0 1.1 1.1
flow rate

In the present study, the flow rate deviations are less analysis etc. can only be performed in commercial pro-
than 10% and therefore the DWSIM simulation results are cess simulation software.
acceptable. It has been proven from this study that free
and open-source chemical process simulator DWSIM was Statements and declarations
able to complete tasks similar to the commercial software, Funding
Aspen Plus. This study also reports that DWSIM can sim- This research did not receive any specific grant from fund-
ulate not only chemical processes but also bioprocess. ing agencies in the public, commercial, or not-for-profit
sectors.
4 Conclusion
The study was performed out of the necessity of finding an Conflict of interest
open-source software for bioprocess simulation. Presently, The authors declare that they have no conflict of interest.
there are no free open source software available, exclu-
sively for bioprocess simulation. But, this could be com- CrediT Author statement
pensated for, by utilizing the freely accessible DWSIM. Siddhi Sreemahadevan: Conceptualization, Methodology,
Although DWSIM is used only for chemical process simu- Writing – Original Draft Preparation, Writing, Reviewing
lation, the same is compatible for bioprocess simulation, as and Editing, Visualization, Supervision. Krishnapriya
is evident from the study. However, the use of DWSIM is Velusamy Sivakumar: Software, Validation, Data
restricted to flowsheeting and material and energy balance Curation, Investigation. Menjith Palanisamy: Software,
calculations. Presently, advanced simulations viz. equip- Validation, Data Curation, Investigation.
ment specification, technoeconomic analysis, sensitivity
202|Sreemahadevan et al.
Period. Polytech. Chem. Eng., 68(2), pp. 195–202, 2024

References
[1] Chéruy, A. "Software sensors in bioprocess engineering", Journal [12] Tangsriwong, K., Lapchit, P., Kittijungjit, T., Klamrassamee, T.,
of Biotechnology, 52(3), pp. 193–199, 1997. Sukjai, Y., Laoonual, Y. "Modeling of chemical processes using
https://doi.org/10.1016/S0168-1656(96)01644-6 commercial and open-source software: A comparison between
[2] Bungay, H. R. "Computer Applications in Bioprocessing", In: Aspen Plus and DWSIM", IOP Conference Series: Earth and
Fiechter, A. (ed.) History of Modern Biotechnology II, Springer, Environmental Science, 463(1), 012057, 2020.
2000, pp. 109–138. ISBN 978-3-540-67792-5 https://doi.org/10.1088/1755-1315/463/1/012057
https://doi.org/10.1007/3-540-44965-5_6 [13] Hachhach, M., Akram, H., Achak, O., Chafik, T. "Simulation of
[3] Vikash, P. V., Shastri, Y. "Conceptual design of a lignocellulosic The Synthesis Route of MoS2 at Laboratory Scale: Comparison
biorefinery and its supply chain for ethanol production in India", Between Simulated and Experimental Results", In: 2017
Computers & Chemical Engineering, 121, pp. 696–721, 2019. International Renewable and Sustainable Energy Conference
https://doi.org/10.1016/j.compchemeng.2018.11.021 (IRSEC), Tangier, Morocco, 2017, pp. 1–3. ISBN 9781538628485
[4] Petrides, D., Cooney, C. L., Evans, L. B., Field, R. P., Snoswell, M. https://doi.org/10.1109/IRSEC.2017.8477367
"Bioprocess simulation: An integrated approach to process [14] Aziaba, K., Jordan, C., Haddadi, B., Harasek, M. "Design of a
development", Computers & Chemical Engineering, 13(4–5), Gas Permeation and Pervaporation Membrane Model for an Open
pp. 553–561, 1989. Source Process Simulation Tool", Membranes, 12(12), 1186, 2022.
https://doi.org/10.1016/0098-1354(89)85038-0 https://doi.org/10.3390/membranes12121186
[5] Li, K., Kirkland, S., Yeo, B.-L., Tubbesing, C., Bandaru, V., [15] Jain, R., Nayak, P., Rahul, A. S., Dalve, P., Moudgalya, K. M.,
Song, L., Holstege, L., Hartsough, B., Kendall, A., Jenkins, B. Naren, P. R., Wagner, D., Fritzson, P. "Implementation of a Property
"Integrated economic and environmental modeling of forest bio- Database and Thermodynamic Calculations in OpenModelica
mass for renewable energy in California: Part I – Model develop- for Chemical Process Simulation", Industrial and Engineering
ment", Biomass and Bioenergy, 173, 106774, 2023. Chemistry Research, 58(18), pp. 7551–7560, 2019.
https://doi.org/10.1016/j.biombioe.2023.106774 https://doi.org/10.1021/acs.iecr.8b05147
[6] Okoro, O. V., Nie, L., Podstawczyk, D., Shavandi, A. [16] Nayak, P., Dalve, P., Sai, R. A., Jain, R., Moudgalya, K. M., Naren,
"Technoeconomic and Environmental Assessment of Alternative P. R., Fritzson, P., Wagner, D. "Chemical Process Simulation Using
Biorefineries for Bioenergy and Polyphenolic Production from OpenModelica", Industrial and Engineering Chemistry Research,
Pomace Biomass", BioEnergy Research, 16(3), pp. 1639–1653, 58(26), pp. 11164–11174, 2019.
2023. https://doi.org/10.1021/acs.iecr.9b00104
https://doi.org/10.1007/s12155-022-10530-1 [17] Elavazhagan, E., Sowmiya, A., Sivalingam, S. "Comprehensive
[7] van Rijn, R., Nieves, I. U., Shanmugam, K. T. Ingram, L. O., analysis and process simulation of biodiesel production from bio-
Vermerris, W. "Techno-Economic Evaluation of Cellulosic mass sources", Indian Journal of Chemical Technology (IJCT),
Ethanol Production Based on Pilot Biorefinery Data: A Case Study 30(5), pp. 623–633, 2023.
of Sweet Sorghum Bagasse Processed via L+SScF", BioEnergy https://doi.org/10.56042/ijct.v30i5.5203
Research, 11(2), pp. 414—425, 2018. [18] Humbird, D., Davis, R., Tao, L., Kinchin, C., Hsu, D., Aden, A.,
https://doi.org/10.1007/s12155-018-9906-3 Schoen, P., Lukas, J., Olthof, B., Worley, M., Sexton, D., Dudgeon,
[8] Lee, S. C., Oh, H. W., Woo, H. C., Kim, Y. H. "Energy-efficient D. "Process Design and Economics for Biochemical Conversion of
bioethanol recovery process using deep eutectic solvent as Lignocellulosic Biomass to Ethanol: Dilute-Acid Pretreatment and
entrainer", Biomass Conversion and Biorefinery, 13(17), Enzymatic Hydrolysis of Corn Stover", [pdf] National Renewable
pp. 15815–15826, 2023. Energy Laboratory, Golden, CO, USA, Technical Report NREL/
https://doi.org/10.1007/s13399-021-02213-2 TP-5100-47764, 2011. Available at: https://www.nrel.gov/docs/
[9] Rocha, C., Soria, M. A., Martins, F. G., Madeira, L. M. "Techno- fy11osti/47764.pdf [Accessed: 29 April 2022]
economic analysis of the olive oil mill wastewater steam reform- [19] DWSIM "DWSIM, (Version 7.5)", [computer program] Available at:
ing process: A case-study", Chemical Engineering Research and https://sourceforge.net/projects/dwsim/files/DWSIM/DWSIM%20
Design, 184, pp. 277–290, 2022. 7.5/ [Accessed: 01 January 2021]
https://doi.org/10.1016/j.cherd.2022.05.048 [20] National Library of Medicine "PubChem", [online] Available at:
[10] Chantasiriwan, S. "Simulation and Optimization of Vapor https://pubchem.ncbi.nlm.nih.gov/ [Accessed: 01 January 2021]
Absorption Refrigeration System Using Dwsim", Chemical [21] Céondo GmbH "Cheméo: High Quality Chemical Properties",
Engineering Transactions, 100, pp. 613–618, 2023. [online] Available at: https://www.chemeo.com/ [Accessed:
https://doi.org/10.3303/CET23100103 01 January 2021]
[11] Sigue, S., Abderafi, S., Vaudreuil, S., Bounahmidi, T. "Design and [22] Carlson, E. C. "Don't Gamble With Physical Properties For
steady-state simulation of a CSP-ORC power plant using an open- Simulations", Chemical Engineering Progress, 92(10), pp. 35–46,
source co-simulation framework combining SAM and DWSIM", 1996.
Thermal Science and Engineering Progress, 37, 101580, 2023. [23] Oliveira de Medeiros, D. W. "DWSIM - Process Simulation,
https://doi.org/10.1016/j.tsep.2022.101580 Modeling and Optimization User Guide, version 1.7", 2010.
[online] Available at: https://www.yumpu.com/en/document/
read /19143280/dwsi m-process-si mulat ion-modeli ng-and-
optimization-user-guide [Accessed: 20 January 2024]