Investigating the

feasibility and
effectiveness of
artificial intelligence
within structural
engineering.

EG6104 INDIVIDUAL RESEARCH PROJECT

Shuaib
U1708936
Abstract
Technological advancements have drastically altered the ways in which the
industry has moved forward. These developments grow increasingly popular as
evolutionary technologies continuously evolve, supporting various fields within
the industry. This project is based on the feasibility and effectiveness of
technologies such as artificial intelligence (AI) within structural engineering. This
project will also inform an overall insight of all positive and negative outcomes
as well as improvements, that could be made, by investigating various methods
using artificial intelligence. The different AI techniques such as, machine
learning (ML), deep learning (DL), and data mining (DM) are established as
intelligent and innovative methods for use in structural engineering. The
objective of this study is to summarise recently developed techniques of the
noted AI methods in structural engineering over the last decade. A data mining
software called orange, is the focal area in which I will be investigating
throughout this project, by inputting data from an excel sheet, where the shear
strength of reinforced concrete beams is found, displaying the formulas used to
calculate this data, in order to receive the predicted results. This is calculated by
the software detecting and identifying a pattern within the database. These
results shall then be further discussed to determine the accuracy of the artificial
intelligence software, concluding the feasibility and effectiveness of AI within
structural engineering.

i
Acknowledgments
I firstly wish to express my sincerest gratitude to my supervisor Dr. Ali Abbas
Senior Lecturer in Civil Engineering at the University of East London, as he
spared no effort to help me throughout my project. Ali’s approach to teaching
has been truly inspiring, and I’ll forever be thankful to him. Additionally, I would
like to express my thankfulness to my parents and family, who have always
strived to help me be the best version of myself alongside teaching me the true
values of life. I truly feel blessed to have you in my life. My gratitude also
extends to my friends, for your continuous support; especially Abid-al-Aziz. I
would also like to take the time and thank my other friends, who although never
made it to university, were determined to see me succeed and be the first one
out of us to get a degree. lastly, I’d like to thank everyone who has been a part
of this blessed journey and been there through the struggle.

ii
Contents

Abstract................................................................................................................ i

Acknowledgments................................................................................................ii

Contents..............................................................................................................iii

List of Figures......................................................................................................v

List of Tables....................................................................................................... v

List of Equations..................................................................................................v

Glossary and Abbreviations...............................................................................vii

1 Introduction.................................................................................................. 1

1.1 Background............................................................................................1

1.2 Aims....................................................................................................... 1

1.3 Objectives.............................................................................................. 1

1.4 Project structure.....................................................................................2

2 Literature......................................................................................................3

2.1 What is Artificial Intelligence?.................................................................3

2.1.1 Methods of Artificial intelligence......................................................3

2.1.2 Machine learning.............................................................................3

2.1.3 Supervise learning...........................................................................4

2.1.4 Unsupervised learning.....................................................................4

2.1.5 Deep learning..................................................................................5

2.1.6 Data mining..................................................................................... 5

2.1.7 Application of AI methods within structural engineering industry.....5

2.1.8 Growth of Artificial intelligence.........................................................7

2.2 BIM 360 machine learning......................................................................8

2.3 Orange data mining................................................................................9

iii
 2.3.1 Using orange to predict shear strength within a reinforced concrete
beam 10

2.4 Effect of artificial intelligence on structural engineers...........................12

2.5 Benefits of artificial intelligence............................................................12

2.5.1 Improving workplace safety...........................................................12

2.5.2 AI will address labour shortages....................................................12

2.5.3 Automate site activity monitoring...................................................13

2.5.4 Make sense of big data with machine learning and data analytic..13

2.5.5 Saving time via design phase optimization....................................13

2.5.6 Years of experience can be easily used and attained...................14

2.6 Disadvantages of artificial intelligence.................................................14

2.6.1 Can be very expensive..................................................................14

2.6.2 Lacks creativity..............................................................................14

2.6.3 Unemployment may rise................................................................15

2.6.4 Robots do not get better with experience…yet..............................15

2.7 Conclusion........................................................................................... 15

3 Methodology...............................................................................................16

3.1 Calculation of capacity of shear reinforcement/links in a reinforced

concrete beam............................................................................................... 16

3.2 Calculations of capacity of section due to diagonal concrete strut.......17

3.3 Changing data within excel to obtain Nemours shear capacity values.17

3.3.1 shear capacity of the reinforcement data.......................................19

3.3.2 Values to be altered to get shear capacity of reinforcement..........19

3.3.3 shear capacity of the section due to diagonal concrete strut data. 20

3.3.4 values to be altered to get shear capacity of the section due to

diagonal concrete strut...............................................................................21

3.4 Inputting data within orange data mining..............................................21

4 Results....................................................................................................... 25

4.1 Shear capacity of reinforcement in a reinforced concrete beam..........25

iv
 4.2 Shear capacity of section due to diagonal concrete strut.....................25

4.3 Orange data mining predications.........................................................26

5 Discussion..................................................................................................29

5.1 Results.................................................................................................29

5.1.1 Comparing the predicted results to the expected results...............29

5.1.2 Critical analysis of data mining targets..........................................30

5.2 Analysis of the project..........................................................................30

5.2.1 Aims and objectives.......................................................................30

5.2.2 Limitations..................................................................................... 31

5.2.3 Repeatability..................................................................................32

5.3 Engineering application findings...........................................................32

6 Conclusion................................................................................................. 33

6.1 Main conclusion................................................................................... 33

6.2 Recommendations for further research................................................34

References........................................................................................................35

List of Figures
Figure 1. Shows the process of machine learning...............................................4

Figure 2. Research publications by Hadei Salehi on the use of pattern

recognition and machine learning (Engineering structures,2018).......................6

Figure 3. showing the rise in Artificial intelligence papers and studies

(Scopus,2015).....................................................................................................7

Figure 4. showing the growth and estimates in revenues for Artificial intelligence
(Statisa, 2018).....................................................................................................8

Figure 5. Showing the use of AI within BIM.........................................................9

Figure 6. Showing the shear design of links (The concrete centre,2015)..........11

Figure 7. Showing the use of a drone at a construction site..............................13

Figure 8. Displaying the file widget and how to change the attributes of the
training data.......................................................................................................22

v
Figure 9. displaying how the orange model should look....................................23

List of Tables
Table 1. Showing the keys in excel sheet 1......................................................18

Table 2. Showing the constant values within the excel sheet 5.........................18

Table 3. Excel sheet 1, showing the dependant and independent variables

needed to find the capacity of the shear reinforcement.....................................19

Table 4. Excel sheet 3, showing the values needed to be changed to find the
capacity of the section due to diagonal concrete strut.......................................20

Table 5. Excel sheet 5, showing some of the training data that was input into
orange to find shear capacity at A and B...........................................................21

Table 6. Excel sheet 6, showing the input data for the output data we need to
predict for the shear capacity at links A and B..................................................22

Table 7. Excel sheet 7, Showing the training data for the capacity of the section
due to diagonal concrete strut...........................................................................24

Table 8. Excel sheet 8, Showing the input data for the outputs we need to
predict for the shear capacity of the diagonal concrete strut.............................24

Table 9. Excel sheet 2, showing shear capacity of reinforcement values that

shall be obtained............................................................................................... 25

Table 10. Excel sheet 4, showing the values of capacity of section due to
diagonal concrete strut to be obtained..............................................................26

Table 11. Prediction of shear capacity at link A in N.........................................26

Table 12. Expected results of shear capacity at link A in N and Kn..................27

Table 13. prediction of shear capacity at link B in N..........................................27

Table 14. Expected results of shear capacity at link B in N and Kn..................28

Table 15. Prediction of capacity of section due to diagonal concrete strut in N 28

Table 16. Expected results of capacity of section due to diagonal concrete strut
in N and Kn........................................................................................................29

vi
List of Equations
Equation 1......................................................................................................... 10

Equation 2......................................................................................................... 10

Equation 3......................................................................................................... 11

Equation 4......................................................................................................... 11

Equation 1......................................................................................................... 17

Equation 2......................................................................................................... 18

Equation 3......................................................................................................... 18

Equation 4......................................................................................................... 18

Glossary and Abbreviations

SHM (structural health management)

AI (Artificial intelligence)
DL (Deep learning)
DM (Data mining)
ML (Machine learning)
CNNs (convolutional neural networks)
RC (Reinforced concrete)

vii
1 Introduction
This segment will illustrate the background, Aims, and Objectives, as well as
project structure.

1.1 Background
Artificial intelligence is advancing on a global scale, in 2020, the global AI
software market is expected to grow approximately 54 percent year-on-year
(Global AI software market growth 2019-2025 | Statista, 2021), therefore
bringing many benefits along with it, enhancing the production of work output
within structural engineering.

Methods used to collect and process data is ever-growing and the uprise of
artificial technology software’s is growing increasingly popular. These software’s
generate immediate results by deciphering numeric data and identifying
patterns that supports the final results. Within this report, I will also be taking an
insight into various artificial intelligent influences such as 3d machines.

The future of structural engineering could potentially be greatly impacted by

technological developments as software’s, devices and machines are gradually
changing the way in which factors are being calculated. As the industry is
adjusting and shifting towards utilizing artificial intelligent technologies,
exploring into the feasibility and effectiveness of current and upcoming
software’s shall provide a better understanding of the impact it will have on the
future of structural engineering.

1.2 Aims
This case study will aim to investigate the feasibility and effectiveness of
artificial intelligence within structural engineering by carrying out data analysis,
thus providing a critical overview of the impact it can have on the future of the
construction industry.

1
1.3 Objectives
1. Describe Artificial intelligence and Investigate various methods of
artificial intelligence.
2. To examine how different artificial intelligence software can improve
proficiency in structures.
3. Look into the application of artificial intelligence.
4. Research on how Orange data mining and other Artificial intelligence are
being used within structural engineering.
5. To use orange data mining to solve structural problems and analyse if it
is a viable solution to be used in structural engineering.

1.4 Project structure

This report is constructed of seven chapters to differentiate the different stages of
research, development, and conclusions. The first chapter creates an overview of
the project and establishes the aims and objectives. The second chapter analyses
current literature surrounding artificial intelligence within structural engineering,
current artificial intelligence software while exploring the performances, as well as
studying the history and potential future of artificial intelligence. The third chapter
classifies how data extracted from the literature review has been used to govern
the results of this study. The fourth chapter follows the methodology in chapter
three, using the data from chapter two, to determine whether artificial intelligence is
effective and feasible within structural engineering. This chapter also calculates
data using orange to show the prediction of shear strength within a reinforced
concrete beam. In the fifth chapter, the results of the investigation have been
discussed, with the original aims and objectives of the project revised, and any
complications that come across have been identified. Conclusions and future
recommendations have been made in chapter six and references to the work of
others in chapter seven.

2
2 Literature
In this chapter of the report, relevant literature was found concerning the feasibility
and effectiveness of artificial intelligence within structural engineering. Different
aspects of Artificial intelligence were researched and the impact they have on
structural engineering.

2.1 What is Artificial Intelligence?

As stated by Franz J. Kurfess, in (Encyclopaedia of physical science and
technology,2003) Artificial Intelligence is a division of science which infuses
decision-making abilities similarly to human-like features, making them
somewhat reliable. Examples of AI machines include Image Processing,
Speech Recognition, Problem-solving, etc. Data mining, deep learning and
machine learning are all needed to create an AI product.

2.1.1 Methods of Artificial intelligence

Within artificial intelligence, there are different methods used by AI such as data
mining, machine learning, and deep learning these are amongst the new
artificial intelligence methods that are increasingly emerging as reliable and
efficient tools in the field of structural engineering. This section provides
technical background on the noted methods and an understanding regarding
the use of algorithms for structural engineering problems.

2.1.2 Machine learning

Machine learning is a method for developing complex algorithms that process
large data and provides results to its users. It uses complex programs that learn
from experience to make predictions. Regular input of training data allows the
algorithm to be enhanced. Machine learning aims to use and understand data to
make models from the data that may be understood by humans, The term
Machine Learning was created by Arthur Samuel (1959), who also stated that,
‘It gives computers the ability to learn without being explicitly programmed.’

3
Machine learning is classified into two types:
 Supervised Learning
 Unsupervised Learning

2.1.3 Supervise learning

The relationship between the input and output variable is known. With the input
data, the ML algorithms shall predict the outcome and it will be compared to the
outcome expected. If any errors occur, they will be corrected, and this step will
continuously be performed until an acceptable level of performance is met.
‘Supervised Learning is like a teacher and student learning’ (Haykin 1998).

2.1.4 Unsupervised learning

Unsupervised learning depends on trained data sets in order to predict the
outcomes. However, by using direct techniques, Unsupervised learning problems
can be further grouped into clustering and association problems (Jason
Brownlee,2016). Trained data sets are sets of data for which the input and output
are known. `

Figure 1. Shows the process of machine learning.

4
2.1.5 Deep learning
Deep learning is an AI role that simulates the function of the human brain when
data processing and making patterns, to be used within decision making. Deep
learning is a subset of ML in artificial intelligence that has networks skilled at
learning unsupervised from data that is unlabelled or unstructured. This is
known as deep neural learning or deep neural network. Shakir Mohammed has
used deep learning algorithms to see what makes the best concrete
(GoogleDeepmind,2015).

2.1.6 Data mining

Data mining, also known as knowledge discovery in a database, within
computer science, is a method of using large volumes of data to discover
interesting and useful patterns and their relationships. This sector joins methods
from statistics and AI (such as neural networks and ML) with database
management that will allow an analysis of big digital collections, known as data
sets.

2.1.7 Application of AI methods within the structural engineering industry

Below, the relevant literature on the application of methods of AI within
structural engineering shall be reviewed.

2.1.7.1 Machine learning application

Machine learning (ML) methods have been widely utilised over the last decade
to model real-world difficulties that concern structural engineering. This is due to
their vast capacity to capture relations amongst input and output data that are
complex to formulate mathematically or nonlinear. Stone J states that some of
the first uses of ML, methods within structural engineering have been used to
resolve issues such as developing management tools to sort issues within
structural safety (Civ Eng Syst,1989) and information attainment for the design
of steel members (Arciszewski T,1987). In general, ML methods have been
used for damage identification and SHM, optimisation, structural reliability,
reliability assessment, performance evaluation, and structural parameter
identification, with these, SHM and concrete property modelling are the uses to

5
get the most attention within the last decade which can be seen in the figure
below

Figure 2. Research publications by Hadei Salehi on the use of pattern

recognition and machine learning (Engineering structures,2018)

2.1.7.2 Deep learning application

The use of deep learning has increased within the last few years as applications
such as, convolutional neural networks (CNNs) for structural engineering,
mainly in structural health management (SHM) has created simpler ways to
capture the structure.. The use of CNNs is very recent in the field of damage
detection and SHM. CNNs within the context of SHM is defined as learning and
extracting optimal features and classification using learned features. CNNs are
primarily designed for two-dimensional signals (e.g., images, video frames,
etc.), thus leading to an effective image recognition method. Therefore, CNNs
are categorised and used as vision based SHM techniques in which datasets
are images captured at various stages of the structure being monitored.

The first use of CNNs in structural engineering was conducted by Sarkar et al.
(Deep Learning for Structural Health Monitoring,2017) for characterising crack
damage on composite materials. Further, Cha et al (Comput Civ Infrastruct Eng,
2017) presented a deep learning network to find concrete cracks in tunnels

6
without the need for computing defect features. These were some examples of
the application of deep learning within structural engineering.

2.1.7.3 Data mining application

Analysis and data sensing helps in monitoring the conditions of infrastructure
above and below the ground (Nilesh Modak,2019). The data from these sensors
are stored and is then used to predict the ground conditions from the stored
data. Methods of correction can also be applied. The data for the ground
conditions can help structural engineers to recognise the type of foundation
needed, for example, what type of soil strength must be used to support the
structure.

2.1.8 Growth of Artificial intelligence

The charts below provide an insight into AI’s rapid growth:

The number of Computer Science academic papers and studies has

soared by more than 9X since 1996. (Scopus,2015)

Figure 3. showing the rise in Artificial intelligence papers and studies

(Scopus,2015)

7
 Global revenues from AI for enterprise applications are projected to grow
from $1.62B in 2018 to $31.2B in 2025 attaining a 52.59% compound
annual growth rate in the forecast period. (Statisa, 2018)

Figure 4. showing the growth and estimates in revenues for Artificial intelligence
(Statisa, 2018)

2.2 BIM 360 machine learning

BIM software companies recently began using artificial intelligence (AI) to
benefit the potential and efficiency of their programs. BIM software such as BIM
360 can now use machine learning to study data and detect patterns. Based on
the detections, decisions are made from the improvement and automation of
model building process, which can be made independently.

BIM software gathers sufficient data, which is explored through AI as it is a

faster way in finding the optimum solution. This does not only make the process
quicker but it also decreases human errors, as a result, this will improve safety
on sites.

8
With the rapid growth of AI, we will most likely be seeing more AI-assisted BIM
over time. ‘Later, one can conceive of a building design system supporting
automated building code checking in city hall or the architect's office’ (Eastman,
1975).AI will push BIM to the next level. The construction industry has not
adapted much as technology has been implemented in most industries,
however has yet to evolve within the construction industry as it has been rather

traditional throughout time. AI will bring faster and safer construction, also
bringing ease and cost-efficiency. The use of AI within construction will expand
over time.

Figure 5. Showing the use of AI within BIM

2.3 Orange data mining

Orange is a component-based visual programming software package

for machine learning, data visualisation, data analysis, and data mining (orange
data mining, 2016).

Orange components are called widgets and they range from simple data
visualisation, pre-processing, and subset selection, to empirical evaluation of
learning algorithms and predictive modelling.
9
Visual programming is implemented through an interface in which workflows are
created by linking predefined or user-designed widgets, while advanced users
can use Orange as a Python library for widget alteration and data manipulation.

2.3.1 Using orange to predict shear strength within a reinforced concrete beam
We shall use variable strut inclination method with a varying range of inputs that
shall be adjusted so we can calculate the outputs on excel and see how the
inputs affect the output for the shear capacity of the diagonal concrete strut and
the shear capacity at the links, the input shall be used to predict an output which
will be compared to the output in excel. (Eurocode 2, 2004)

The cross-sectional area of the links

no . of . legs ( π d2 )
A sw=
4
Equation 1

Asw = Cross-sectional area of the legs

No of legs= Amount of legs in each link

d= Diameter of shear links

The capacity of shear reinforcement/links

A sw f ywd z ( cotθ )
V Rd , s=
s
Equation 2

Vrd,s= Resistance of member governed by the yielding of shear reinforcement

Asw = Area of the shear reinforcement

fyk = characteristic strength of the links

fywd = design yield strength = fyk/1.15

z= inner lever arm. In the shear analysis of reinforced concrete without axial
force, the approximate value z = 0.9d may normally be used. d being the
hogging and sagging value.

10
θ = Angle of the strut

s = Link/reinforcement spacing

The capacity of the section due to the diagonal concrete strut

V Rd ,max =a cw v 1 f cd b w z ( cosθ ) ( sinθ )

Equation 3

Vrd,max = Resistance of member governed by the crushing of compression struts

acw= 1.0 Coefficient for stress in compression chord

v1 = strength reduction factor concrete cracked in shear ν1 = ν = 0.6(1-fck/250)

fcd = design compressive strength = αccfck/1.5 = fck/1.5 (αcc = 1.0 for shear)

bw= is the minimum width

z= inner lever arm. In the shear analysis of reinforced concrete without axial
force, the approximate value z = 0.9d may normally be used. d being the
hogging and sagging value.

θ = Angle of the strut

Assuming the stress in the shear reinforcement is greater than 0.8f yk

v 1=0.6 1−[ f ck
250 ]
Equation 4

v1 = strength reduction factor concrete cracked in shear

fck= characteristic cylinder strength of concrete

The image below shows how you can vary the range of the strut angle and
number of links, in order to receive varied values of the shear capacity on the
links.

11
 Fig ure 6.
Showing the shear design of links (The concrete center,2015)

2.4 Effect of artificial intelligence on structural engineers

The construction industry has expanded by 1 percent per year in the past few
decades. In comparison to 3.6 percent for manufacturing and 2.8 percent for the
world economy, productivity or the total output economically per worker has
remained flat in construction. Productivity has grown by 1500 percent within,
retail, agriculture, and manufacturing since 1945 (Mckinsey,2017). The reason
for this is that the construction industry is one of the least digitised industries
because it is challenging to adopt new technologies.
Learning new technologies can be difficult and time consuming for teams to
adjust. However, artificial intelligence (AI) and machine learning increase
productivity at job sites. AI solutions have made an impact on many industries
and now are starting to emerge within the structural and civil engineering
industry.
2.5 Benefits of artificial intelligence
The benefits of artificial intelligence within structural engineering shall be
discussed below.

12
2.5.1 Improving workplace safety
Risk is inevitable when it comes to construction projects, the larger the project
the greater the risk, you can use AI and machine learning tools to prioritise risks
and monitor on a job site. Subcontractors can be rated by using AI-based on a
risk score therefore prioritising certain issues can be assigned so the project
manager can work closely with high – risk teams to avoid risks. Additionally, job
sites are also being automated, with wearable sensors that can detect the
location of workers and provide alerts if any of them slip or fall. (team
invonto,2018).

2.5.2 AI will address labour shortages

Labour shortages boost the construction/ structural engineering industry as it
forces construction firms to bring artificial intelligence and data science into their
sector of work. Construction firms could boost productivity by as much as 50
percent through real-time analysis of data (McKinsey,2017). AI can help with
the lack of structural engineers within the industry as algorithms solve and
addresses complex problems and can help structural engineers get their jobs
done faster.

2.5.3 Automate site activity monitoring

By automating your activity monitoring, a business can make smarter decisions
based on hard data. With AI, you can track, capture, and classify anything that
enters and exits a job site that could be from people, vehicles, or equipment.

Additional insight can be gained by tacking what’s happening on a worksite,

rapid responses can be provided by using this data. This is vital to prevent
safety hazards and to monitor cost, materials, and ensuring employee
productivity. Stakeholders can access data for real-time updates.(Invonto,2018)
Team Invonto states that Drones can be used by construction companies for
data collection, to analyse a site, and gain any information that may come in
handy for a construction team.

13
 Figure 7. Showing the use of a drone at a construction site

2.5.4 Make sense of big data with machine learning and data analytic
Endless amounts of data machine learning (ML) systems are exposed to help
workers learn. Data sources could potentially gather statistics from every job
site. More data becomes available over time, machine learning makes the data
analytics systems “smarter.” ML and AI can help make sense of data gathered
from mobile devices, images captured via drones, and security sensors, for
example, to glean insights that will help companies operate more efficiently.

2.5.5 Saving time via design phase optimisation

Structural engineers gain data and information via AI, without having to
physically go to the construction site. This benefits structural engineers working
from abroad. AI identifies everything that is needed for the structural engineer to
be able to construct the building. This is time-efficient as all analysis can be
done from any part of the world, as well as being easily sent worldwide. This
makes collaborative work most efficient as assisting in projects can be easily
accessible to anyone.

2.5.6 Years of experience can be easily used and attained

Many corporate projects oversee budgeting and tend to overspend on
employment. Real life training materials can be accessed by Al to assist staff in
enhancing their chosen skills and knowledge in a shorter period of time.
Budgeting is a vital area in engineering which must be assessed accurately.
Artificial neural networks provide the accuracy by predicting expenditures and
overrun costs based on factors such as contract type and project size. Historical

14
data and timelines such as planned start and end dates are used by AI to
envision realistic timelines for projects in the future.

2.6 Disadvantages of artificial intelligence

The disadvantages of artificial intelligence within structural engineering shall be
discussed below.

2.6.1 Large expenditures

Maintaining robots and machines can be costly as it is stated(Web/fx,2021) that
a custom AI solution can cost between $6000-300,000, as time is also needed
for repairs and maintenance. Softwares needs to be updated continuously to
meet the needs of an ever-changing environment. It’s relatively expensive to
receive machines within a workplace, hence only large companies are able to
afford them.

2.6.2 Lacks creativity

Robots can only do as much as they are programmed to, programming is stored
in their internal circuits and firmware. Robots lack creativity in comparison to
humans, computers are unable to produce innovative designs, drawing,
building, or completing a task on a construction site. A machine can simply not
‘think outside the box, by coming up with new ideas.

2.6.3 Unemployment may rise

The ability for artificial intelligence to mimic human behaviour has led to job
losses, which has affected the workforce within structural and civil engineering.
According to leading design and consultancy firm Arcadis (Arcadis,2018), a
shortfall in human workers has led to an increase of AI in the construction
industry. Arcadis stated that the construction sector needs to recruit over
400,000 workers each year to keep up with demands. Additionally, they said
this figure doesn’t include the shortage of workers after Brexit. Research done
by international construction company Mace (Mace, 2017) predicts that 600,000
employees could potentially loose their job, within the UK construction industry
out of the 2.2 million workers. It is predicted that the industry will be taken over
by artificial intelligence by the year 2040. Unfortunately, A major impact on the
brick layer is said to occur from 73,000 workers expected to drop to 4,300 with

15
the rapid growth of AI. After conducting a public survey, many have suggested
that the future will evolve into a human-free construction industry.

2.6.4 Robots do not get better with experience…yet

Unlike humans, Artificial intelligence does not get better with experience. Machines
may be able to store huge amounts of data but are not as effective as humans as
over time machines can become faulty and could possibly loose data. Machines
are able to store large amounts of data, however data can easily be deleted in
comparison to relieving information from an individual, as the way in which data is
accessed and stored is different to humans.

2.7 Conclusion
In section 2.1 current literature was reviewed regarding the effectiveness and
feasibility of artificial intelligence within structural engineering. The artificial
intelligence industry is established to will grow from $1.62B in 2018 to $31.2B in
2025 attaining a 52.59% compound annual growth rate in the forecast period.
The number of scientific papers has increased by 9 times since 1996 (Statisa,
2018).

Throughout the literature review I have determined how different types of

artificial intelligence methods work from data mining, machine learning, and
deep learning, I have also established how these methods work hand in hand
within the construction industry and the effects they may have on structural
engineers stating the benefits and disadvantages within section 2.6 of this
review.

Finally, a new and upcoming software called orange data mining was discussed
along with its functionality and how it can be used to benefit the construction
industry, further within this report by using the widgets and data visualisation
within orange, I will include the output for the shear capacity on reinforced
concrete beams, this shall be compared to the expected output.

16
3 Methodology
This part of the study will be aimed at using the material from the literature
review to meet the objectives. This will be done by using formulas from
Eurocode 2 to calculate shear capacity at links of a reinforced concrete beam,
and the shear capacity of a diagonal concrete strut in the beam. An excel
spreadsheet with varying input values shall be made for different output data. I
will also attach an excel sheet with multiple data sets that will be put into the
orange data mining software to learn from. The widgets within the software shall
use predictive models to predict an output with this data to see if the output data
on orange data mining matches the output of the excel spreadsheets.

3.1 Calculation of capacity of shear reinforcement/links in a

reinforced concrete beam
To calculate the capacity of the shear reinforcement, independent variables
were needed such as strut angle θ, the characteristic yield strengths for the
links and concrete, spacing of the links, the diameter of the links this led to
dependant variables such as design yield strength (fywd), area of the links
(asw), and inner lever arm (z).

For Equation 1, there shall be independent variables for the area of the links
such as their diameter, then the no of legs which shall vary throughout to give
different data sets, they shall be input to find the area of the links (asw). This
value shall then be input into Equation 2.

no . of . legs ( π d2 )
A sw=
4
Equation 1

17
For Equation 2, there will be independent variables for the capacity of shear
reinforcement such as spacing of the links (s) at A and B, area of the links
(asw), the angle of the strut (θ), and dependant variables such as the inner
lever arm(z) which will change according to the diameter(d) of the struts, and
design yield strength fywd which shall depend on the characteristic yield
strength (fck) of the material. This will then give the value of the capacity of
shear reinforcement/links at A and B.

A sw f ywd z ( cotθ )
V Rd , s=
s
Equation 2

3.2 Calculations of the capacity of section due to diagonal

concrete strut
To calculate the capacity of section due to diagonal concrete strut, constant
variables were needed such as strut angle θ, the coefficient for stress in
compression cord (acw) was taken as 1.0, minimum width (bw) cos θ, sin θ the
value of inner lever arm (z) this led to some independent variables, design
compressive strength (fcd), the strength reduction factor for concrete (v1).

For equation 3, there will be independent variables such as θ, lever arm(z),

coefficient of compression cord (acw), the minimum width (bw), and then some
dependant variables such as strength reduction factor (v1) calculated in
equation 4, finally the design compressive strength (fcd).

V Rd ,max =a cw v 1 f cd b w z ( cosθ ) ( sinθ )

Equation 3

For equation 4, I needed to input the value for characteristic cylinder strength of
concrete (fck) this can then be plugged into equation 3 to give the capacity of
the section due to the diagonal concrete strut.

v 1=0.6 1−[ f ck
250 ]
Equation 4

18
3.3 Changing data within excel to obtain Nemours shear
capacity values
In this section, the steps taken to create excel spreadsheets, that will provide
the values of the shear capacity, at links at A and B and the capacity of the
section due to the diagonal concrete strut shall be shown. Nemours data within
the excel sheet shall be changed to get different values of shear capacity, this is
done so the varying data sets can be used as input for orange data mining to be
trained and predict the values of missing data, which shall be compared to the
data in excel so see if these predictions were correct.

The keys we shall follow for the data that needs to be changed within the excel
sheet. Is shown below.

Table 1. Showing the keys in excel sheet 1.

The constant values within the excel spreadsheet that can be found within
Eurocode 2, is shown below. For the factor of safety of steel and concrete.

Table 2. Showing the constant values within the excel

sheet 5.

19
3.3.1 shear capacity of the reinforcement data
Below, we have in yellow the data that we will need to vary to find the capacity
of shear at A and B via equations 1&2.

Table 3. Excel sheet 1, showing the dependant and independent variables

needed to find the capacity of the shear reinforcement.

3.3.2 Values to be altered to get shear capacity of reinforcement

θ = Angle of the strut

The angle of theta will range from 22 °< θ<45 ° with increments of 0.2 to create
multiple data sets. This will be repeated 4 times, to obtain results for 464 RC
beams in total, by changing values of no of legs, hogging and sagging, spacing
at A and B, shear link diameter and strength of the link each, the variation of
values will be stated below. This data can be used to feed into orange for
predictions.

No of legs= Amount of legs in each link

The no of legs shall be 2,3,4,6 in order to observe the effect of these on the
shear capacity.

d= hogging and sagging

20
The hogging and sagging values shall be taken as 650,670,700,750 in mm.

s = Link/reinforcement spacing at A and B

The spacing at A will range from 100,150,200,250 in mm.

The spacing at B will range from 200,250,300,350 in mm.

Shear link diameter

The range for the diameter of shear links after looking into the Eurocode 2
designers manual shall be 10,12,16,20 in mm.

Strength of the links

The strength of the links shall vary from 400,500,600,700 in N/mm2 this
correlates to the diameter of the links.

3.3.3 shear capacity of the section due to diagonal concrete strut data
Below, we have in yellow the data that we will need to vary to find the shear
capacity of the section due to the diagonal concrete strut via equations 3 and 4.

Table 4. Excel sheet 3, showing the values needed to be changed to find the
capacity of the section due to diagonal concrete strut.

21
3.3.4 values to be altered to get shear capacity of the section due to
diagonal concrete strut.
We will also create 4 data sets that will range between 22 °< θ<45 ° each set and
change the values of the following below.

fck= characteristic cylinder strength of concrete

The optimum design values of fck have been decided by looking into Eurocode 2
designers manual these are as follows 35,40,20,24 N/mm2.

bw= is the minimum width

The values of bw shall be 300,310,320,350mm.

3.4 Inputting data within orange data mining

Step 1; Shows the training data created using equations 1 and 2 to give the
values of the shear capacity at links A and B which were added to excel sheet 1
from excel sheet 2 which is shown in the results section. Excel sheet 1 shall
then be input into the file widget within orange. Making the shear capacity at A
the target variable and all other data a feature, The feature statistics widget
provides a quick way to inspect and find interesting features in a given data set
(Orange,2017). However, the beam number shall be a meta attribute, meta
attributes are not used in modelling, but several widgets can use them as
instance labels. (Orange,2017) After finding the prediction of shear capacity at
A by following the other steps below, this shall be repeated for the shear at B
making it the target and the shear at link A a feature.

22
 Table 5. Excel sheet 5, showing some of the training data that was input into
orange to find shear capacity at A and B.

Figure 8. Displaying the file widget and how to change the attributes of the
training data.

Step 2; Below is the data that an output shall be predicted for, another file
widget shall be placed in orange by following step 1 and keeping the shear
capacity at A as the target variable, predictions for shear capacity at A can be
made after connecting the rest of the widgets to these two files which shall be
shown how to do in step 3.Then the same can be done for the shear at B which
is stated how in step 1.

Table 6. Excel sheet 6, showing the input data for the output data we need to
predict for the shear capacity at links A and B.

23
Step 3; orange data mining has been used to predict outputs by feeding 80% of
both input and outputs in the training data set to Orange to train the model.
Feeding only the 20% values of inputs for the data, to be predicted, this data is
shown in step 1 and 2, by using classification tree. It will connect the training
data and split the data into nodes, which then can be viewed with the
classification tree viewer widget (the input data for which we shall be predicting
the output can be seen via the data table). The prediction widget within orange
shall be connected to the classification tree, the training data and the data
needed to predict an output for, a prediction of the outputs shall be made, once
we click on the predictions tab it will show the results for the missing values of
Shear capacity at A. Then vice versa this can be done for shear capacity at B.

Figure 9. displaying how the orange model should look.

The data for the capacity of the section due to diagonal concrete strut is shown
below in which data cos θ , sin θ , z and the values for capacity of section, due to

24
diagonal concrete strut as it was added from excel sheet 1 and 4 to make a
training data set, for which results shall be predicted. Again, there are two files
with training data and the data we need to predict, following the steps above
from 1 to 3, predictions can be made using the same widgets in orange.
However, only the capacity of the section due to the diagonal concrete strut
needs to be made the target variable this time.

Table 7. Excel sheet 7, Showing the training data for the capacity of the section
due to diagonal concrete strut.

Table 8. Excel sheet 8, Showing the input data for the outputs we need to
predict for the shear capacity of the diagonal concrete strut.

4 Results
Data from the Literature Review, Section 2, will be used following the methodology
set out in Section 3, to determine the feasibility and effectiveness of orange data
mining within structural engineering calculations.

25
4.1 Shear capacity of reinforcement in a reinforced concrete
beam
After inputting the values from excel sheet 1 to make excel sheet 2 for the
values of shear capacity at link A and B which can be seen below, these values
were fed into excel sheet 5 and 6 to make it easier for orange to read the data
as the input and output data was in one place.

Table 9. Excel sheet 2, showing shear capacity of reinforcement values that

shall be obtained.

4.2 Shear capacity of section due to diagonal concrete strut

After inputting the values from excel sheet 3 to make excel sheet 4, shown
below, the values for the capacity of the section due to diagonal concrete strut
shall be obtained. As stated in the methodology there shall be 464 different
values for RC beams to give orange more data to learn from.

26
 Table 10. Excel sheet 4, showing the values of the capacity of section due to
diagonal concrete strut to be obtained.

4.3 Orange data mining predications

The data predictions for the shear capacity at link A for a random selection of
data of RC beams between 349-365, are shown below. The results are in
newtons shown after clicking the predictions widget displayed in the
methodology. The output required in grey can be seen side by side with the
predictions of the data to the left.

Table 11. Prediction of shear capacity at link A in N

Below are the expected results for the shear capacity at A for the RC beams
349-365, that we found in excel sheet 2.

27
 Table 12. Expected results of shear capacity at link A in N and Kn

Table 13. prediction of shear capacity at link B in N

Below are the expected results for the shear capacity at B for the RC beams
400-416, that we found in excel sheet 2.

28
 Table 14. Expected results of shear capacity at link B in N and Kn

The table below, shows the data predictions for the capacity of a section due to
diagonal concrete strut for a random selection of data of RC beams between
396-412 the results are in newtons, shown after clicking the predictions widget
displayed in the methodology. The output required is in grey and is side by side
with the data predictions on the left.

Table 15. Prediction of the capacity of section due to diagonal concrete strut in
N

Below are the expected results for the capacity of the section due to the
diagonal concrete strut for the RC beams 396-412, that we found in excel sheet
4.

29
 Table 16. Expected results of the capacity of section due to diagonal concrete
strut in N and Kn.

5 Discussion
In this chapter, the findings of this investigation will be discussed.

5.1 Results
The results of the parametric case study are discussed.

5.1.1 Comparing the predicted results to the expected results

In section 2.3 and 2.3.1 of the literature review, states how orange data mining
can be used to help structural engineers, within section 3 and 4 this information
was used to calculate the shear capacities within a reinforced concrete beam.
Looking into the results chapter the values that have been obtained by orange
for the shear capacity at links A and B, and the capacity of the section due to
diagonal concrete strut, were then compared to the expected results, it is
evident that the predictions were correct, This result justifies the study of the
feasibility and effectiveness of artificial intelligence within structural engineering
as the artificial intelligence software orange has been used to successfully
predict results of the shear capacities within an RC beam.

5.1.2 Critical analysis of data mining targets

Data mining has become a big scope within structural engineering and is used
in almost every aspect of life it is used for determining helpful things which are
then used to make decisions and identifications. The most extensive data
mining techniques which are used in structural engineering are classification
and regression. They have been used to analyse landslide susceptibility,
accident severity prediction, suspended sediment loading, concrete property
estimation, and to use data to find ground conditions. These methods of data
mining can support decision-making and provide new insight into the data for
structural and civil engineers, which inevitably involves experts from specialised
civil engineering and data analytics, as they can read the data. Data mining
algorithms can enhance the quality of predictions, diagnosis, and classification

30
of data. The major goal is to evaluate different techniques of data mining and
the effect they have on structural engineers.

5.2 Analysis of the project

5.2.1 Aims and objectives
This project aimed to investigate the feasibility and effectiveness of artificial
intelligence within structural engineering, and to investigate an artificial
intelligence software that can be used for structural engineering. This aim was
successfully evident using the five objectives in chapter 1 of this report.

In the literature review, different methods of artificial intelligence such as

machine learning, deep learning, and data mining were reviewed alongside the
use of such artificial intelligence within the structural engineering industry, also
BIM artificial intelligence technologies have been looked into alongside orange
data mining reviewing and how they can improve the proficiency and
productivity of construction.

In the results section 4, The shear capacities in reinforced concrete beams

which include the shear capacity at links A and B and the capacity of the section
due to the diagonal concrete strut were calculated using the formulas found in
the literature review.

The values of the shear capacities were then compared, and it was established
that artificial intelligence could be used to find missing values of the shear
capacities as long as it had enough training data to learn from, this finding
means that data mining the sector of AI can be used in numerous ways to
generate results from data within structural engineering problems.

In the discussion, section 5 the contribution that the results of this study could
make towards finding the feasibility and effectiveness of artificial intelligence in
structural engineering. And the findings were then analysed.

5.2.2 Limitations
This report only mainly covers the feasibility and effectiveness of artificial
intelligence within structural engineering, and it disregards research in detail
about applications of AI within civil engineering, as civil engineering has a wide
scope with many branches such as geotechnics, traffic, and highways,
31
surveying, project management, etc. Within the report, the findings from orange
data mining were for only a few formulas within structural engineering and when
the complexity of the formulas increase it will make it harder for orange data
mining to recognise patterns within data, however, structural engineering is
more than just formulas and this report only briefly covers how AI can help with
3d designs analytics. This is also a major part of structural engineering so this
report is focused and narrowed in what was investigated, in order to research
the effect and feasibility of artificial intelligence within civil engineering.

Artificial intelligence methods that were mentioned in the literature review, such
as Machine learning and deep learning could have also been tested using
software with their involvement.

The findings in the literature review are a few years old and there are not many
recent data, making the data slightly unreliable as data on the topic of artificial
intelligence within structural engineering is not widely available due to the fact
the construction industry has not made many technological advancements. AI
has only recently been introduced to the industry, allowing more space for error
or broadening in 1 specific area. For instance, orange data mining is a fairly
new software however it is mainly used for categorical data and can be used for
image recognition, etc. It is not viable when gathering data from random places
as the data needs to be structured in order for it to understand, which can by
time-consuming.

The possibility for orange to solve other things within structural engineering has
not been looked into and the findings based on one software using a few
formulas cannot determine the full effect artificial intelligence has on structural
engineering. As the literature review mentions many different forms of AI and
there uses but only data mining is looked into within this report.

5.2.3 Repeatability
The results of this study, as discussed in section 5.2.2, are dependent on the
formulas used and found within the literature review, however, only three
equations have been used that find a similar output so, therefore, other
formulas could be introduced and only one of the widgets were used to predict
the output so various widgets within orange could be used for better data
analytics.

32
5.3 Engineering application findings
The results of this report show a strong bond between artificial intelligence and
structural engineering and how they come together and work hand in hand. The
use of artificial intelligence within structural engineering is far from little and can
have many advantages and plays a big role in cost and time-efficiency within
engineering as the risks are minimised with algorithms that can solve complex
problems and gather the most critical information within data sets in a matter of
seconds.

This study proves that artificial intelligence will have a big future within the
structural engineering industry and looking at the charts within the literature
review, there will be a 52.59% compound annual growth rate (Satistsa,2018)
and that the number of Computer Science academic papers and studies has
soared by more than 9X since 1996. (Scopus,2015) The reason for such high
growth is that AI has had a positive effect within every industry and the research
in the literature backs this statement, it also shows the use case scenarios of AI
in structural engineering, although autonomous it will enhance the work output
of structural engineering firms and help structural engineers rather than put
them out of a job.

The findings of this study can, therefore, contribute to proving that AI is effective
and feasible in the structural engineering industry.

6 Conclusion
This section summarises the findings that were drawn from the discussion,
Section 5 of this report, and recommends any further research that could be
done, looking at the results of this study to help achieve the goals within the
report.

6.1 Main conclusion

This project aims to investigate the feasibility and effectiveness of artificial
intelligence within civil engineering by finding relevant literature and analysing it
also using orange data mining software to prove that AI can be effectively used
in structural engineering.

33
This aim was accomplished using the 5 objectives in section 1 of the report,
current literature was critically reviewed to provide research and data for the
project. In section 4 the results of the shear capacities within an RC beam were
compared, and the contribution these results had on helping to find how
effective and feasible AI is in structural engineering was discussed in section 5.

It was established that by using 80% of the data input and output we could
calculate the output for the remaining 20% data that was input by using the
widgets within orange data mining. The results for the predictions of these
outputs were indeed accurate and proved the main purpose of this report.

Based on the findings in the literature review, it’s evident that AI will soon be
taking over the world and the development of such software’s will enhance
every aspect of structural engineering. Potentially, saving many lives and big
money with automated health and safety detection and damaged crack
detection, showing any flaw the human eye may not see, reducing fatalities and
errors in calculations.

6.2 Recommendations for further research

As discussed in section 5.2.2, the results of this study are dependent on all the
literature harnessed in the literature review, although the data sets were large
enough for orange to learn from, I used one software out of the many artificial
intelligence software’s available and we only used one classifying widget within
orange data mining. However, we could have used other features such as linear
regression and neural networks to get more out of the data. This way the
information can be used differently as orange can use these widgets to learn
more about the training data allowing for various outputs. As a result, we may

34
be able to see the correlation each independent variable has on the results of
the shear capacities within the RC beam.

Another recommendation for further research, based on the limitations

discussed in 5.2.2, would be to complete the same research, however for a
different formula instead of the shear capacity in an RC beam. I could have
used, deflection and bending formulas to provide further data, to analyse how
the orange widgets would be able to extract valuable information from this data
that can help validity of the aims of this report.

Also, BIM artificial intelligence could have been used to provide more statistical
data by using data from pre-existing 3d models, to help generate a new model
based on requirements without structural engineers having to spend hours on
the design. As it could be done automatically with powerful algorithms in BIM
artificial intelligence, That could have added to show the effectiveness and
feasibility of artificial intelligence within structural engineering.

References
(Artificial Intelligence and 3D printing: future of manufacturing, 2020)

(Artificial Intelligence and 3D printing: future of manufacturing, 2020)

Artificial Intelligence Applications in Civil Engineering, (tayfun dede, May 2019)

Available at: <https://medium.com/@venkat34.k/artificial-intelligence-and-3d-printing-future-of-

manufacturing-d84fb94b1c7d#:~:text=3D%20printing%20is%20a%20game,applications%20of
%20additive%20manufacturing%20technology.> [Accessed 8 November 2020].

Computational Intelligence in Civil and Hydraulic Engineering, fei keng, December 2012

[Cha Y, Choi W, Buyukozturk O. Deep Learning‐Based Crack Damage Detection Using

Convolutional Neural Networks. Comput Civ Infrastruct Eng 2017;32:361–78.]
35
(How artificial intelligence will reshape civil engineering, 2020)

Haykin, S. (1998). Neural networks: A comprehensive foundation (2nd ed.). Upper Saddle
River: Prentice Hall. ISBN 0-132-73350-1.

Institution of Civil Engineers (ICE). 2020. How Artificial Intelligence Will Reshape Civil
Engineering. [online] Available at:
<https://www.ice.org.uk/news-and-insight/ice-thinks/infrastructure-transformation/how-artificial-
intelligence-will-reshape-civil-eng> [Accessed 8 November 2020].

Iranarze.ir. 2021. [online] Available at: <http://iranarze.ir/wp-content/uploads/2019/01/E10801-

IranArze.pdf> [Accessed 18 April 2021].

Iranarze.ir. 2021. [online] Available at: <http://iranarze.ir/wp-content/uploads/2019/01/E10801-

IranArze.pdf> [Accessed 18 April 2021].

Iranarze.ir. 2021. [online] Available at: <http://iranarze.ir/wp-content/uploads/2019/01/E10801-

IranArze.pdf> [Accessed 18 April 2021].

Iranarze.ir. 2021. [online] Available at: <http://iranarze.ir/wp-content/uploads/2019/01/E10801-

IranArze.pdf> [Accessed 18 April 2021].
Iranarze.ir. 2021. [online] Available at: <http://iranarze.ir/wp-content/uploads/2019/01/E10801-
IranArze.pdf> [Accessed 18 April 2021].

Invonto. 2021. Smart Construction: Build Smarter with Artificial Intelligence. [online] Available at:
<https://www.invonto.com/insights/smart-construction-artificial-intelligence/> [Accessed 19 April
2021].

Investopedia. 2021. Machine Learning. [online] Available at:

<https://www.investopedia.com/terms/m/machine-learning.asp> [Accessed 19 April 2021].

Investopedia. 2021. Machine Learning. [online] Available at:

<https://www.investopedia.com/terms/m/machine-learning.asp> [Accessed 17 April 2021].

Lib.dr.iastate.edu. 2020. [online] Available at: <https://lib.dr.iastate.edu/cgi/viewcontent.cgi?

article=1033&context=ccee_conf> [Accessed 8 November 2020].

(Machine Learning, 2021)

Macegroup.com. 2021. Next industrial revolution set to transform more than half a million
construction jobs. [online] Available at: <https://www.macegroup.com/media-centre/171031-
next-industrial-revolution-set-to-transform-more-than-half-a-million--construction-
jobs#:~:text=Depending%20on%20the%20pace%20of,4.0%3A%20A%20skills%20revolution'.>
[Accessed 19 April 2021].

Medium. 2020. Artificial Intelligence And 3D Printing: Future Of Manufacturing. [online]

Medium. 2021. Artificial Intelligence and 3D printing: future of manufacturing. [online] Available
at: <https://medium.com/@venkat34.k/artificial-intelligence-and-3d-printing-future-of-

36
manufacturing-d84fb94b1c7d#:~:text=3D%20printing%20is%20a%20game,applications%20of
%20additive%20manufacturing%20technology> [Accessed 18 April 2021].

[Sarkar S, Reddy KK, Giering M, Gurvich MR. Deep Learning for Structural Health Monitoring:
ADamage Characterization Application n.d.]

(Software, 2020)

Software, D., 2020. RFEM - FEM Structural Analysis Software. [online] Dlubal. Available at:
<https://www.dlubal.com/en/products/rfem-fea-software/what-is-rfem#:~:text=RFEM
%20provides%20deformations%2C%20internal%20forces,designs%20according%20to
%20various%20standards.> [Accessed 17 November 2020].

(Stone J, Blockley D, Pilsworth B. Towards machine learning from case histories. Civ Eng
Syst1989;6:129–35.)

UK Construction Online. 2021. The benefits of AI in construction - UK Construction Online.

[online] Available at: <https://www.ukconstructionmedia.co.uk/features/benefits-ai-construction/>
[Accessed 19 April 2021].

Webfx.com. 2021. AI Pricing | How Much Does Artificial Intelligence Cost in 2020? [online]
Available at: <https://www.webfx.com/internet-marketing/ai-pricing.html> [Accessed 19 April
2021].

2021. [online] Available at: <https://www.mckinsey.com/~/media/McKinsey/Business

%20Functions/Operations/Our%20Insights/Reinventing%20construction%20through%20a
%20productivity%20revolution/MGI-Reinventing-Construction-Executive-summary.pdf>
[Accessed 18 April 2021].

2021. [online] Available at: <https://www.pbctoday.co.uk/news/planning-construction-news/ai-

construction-industry/46121/> [Accessed 19 April 2021].

2021. [online] Available at: <https://www.arcadis.com/media/3/4/0/%7B3403FC9B-FA2C-47C0-

9011-C1B71DB9C3AF%7DArtificial-Intelligence-in-the-AEC-Industry-a-Code-of-
Practice_002.pdf> [Accessed 19 April 2021].

37