Site layout optimization using novel hybrid ant lion optimizer (alo) algorithm

FEBRINA VIENNA SOULISA SITE LAYOUT OPTIMIZATION USING NOVEL HYBRID ANT LION OPTIMIZER ALO ALGORITHM TỐI ƯU HÓA BỐ TRÍ MẶT BẰNG THI CÔNG SỬ DỤNG THUẬT TOÁN LAI GHÉP KIẾN SƯ TỬ - ALO MỚI

FEBRINA VIENNA SOULISA

SITE LAYOUT OPTIMIZATION USING NOVEL HYBRID

ANT LION OPTIMIZER (ALO) ALGORITHM

TỐI ƯU HÓA BỐ TRÍ MẶT BẰNG THI CÔNG SỬ DỤNG THUẬT TOÁN LAI GHÉP KIẾN SƯ TỬ - (ALO) MỚI

Major: CONSTRUCTION MANAGEMENT

Major code: 8580302

MASTER’S THESIS

HO CHI MINH CITY, DECEMBER 2022

THIS RESEARCH IS COMPLETED AT:

UNIVERSITY OF TECHNOLOGY – VNU – HCM CITY

Instructor:

PhD/Associate Prof Pham Vu Hong Son

Master’s thesis is defended at HCM city University of Technology, HCM City on December 7th, 2022

VNU-The board of the Master’s VNU-Thesis Defense Council includes:

1 President: Dr Do Tien Sy

2 Secretary: Assoc Prof Dr Tran Duc Hoc

3 Counter-argument member 1: Assoc Prof Dr Luong Duc Long

4 Counter-argument member 2: Dr Nguyen Van Tiep

5 Council member: Dr Nguyen Thanh Viet

Verification of the Chairman of the Master’s Thesis Defense Council and the Dean of faculty of Civil Engineering after the thesis being corrected (If any)

Examiner 1: Assoc Prof Dr Luong Duc Long

Examiner 2: Dr Nguyen Van Tiep

VIETNAM NATIONAL UNIVERSITY

HO CHI MINH CITY

HO CHI MINH CITY UNIVERSITY

OF TECHNOLOGY

SOCIALIST REPUBLIC OF VIETNAM Independence – Freedom - Happiness

THE TASK SHEET OF MASTER’S THESIS

Full name: FEBRINA VIENNA SOULISA Student code: 2070288

Date of birth: 03/02/1994 Place of birth: Bandung, Indonesia Major: Construction Management Major code: 8580302

I THESIS TOPIC: Site layout optimization using novel hybrid ant lion

optimizer (ALO) algorithm Tối ưu hóa bố trí mặt bằng thi công sử dụng

thuật toán lai ghép kiến sư tử - (ALO) mới

II TASKS AND CONTENTS: Site layout optimization using artificial

intelligence in Construction Management

III TASKS STARTING DATE: March, 2022

IV TASKS ENDING DATE: December, 2022

V INSTRUCTOR: PhD/Associate Prof Pham Vu Hong Son

ACKNOWLEDGEMENT

I want to express my gratitude to all the lecturers (both Vietnamese and Japanese), admins (both academic and OISP), and classmates, especially my fellow CEP students (John, Zei, and Poe) of the International Master Program at the Ho Chi Minh City University of Technology for the continuous support and help given during the program

The completion of this research is possible thanks to my respectful and kind thesis instructor and advisor, Ph.D./Associate Prof Pham Vu Hong Son, for his guidance and assistance The constructive feedback, advice, suggestions, and shared knowledge helped me a lot during the process of making this thesis research

I would like to send my special thanks to the Japan International Cooperation Agency Project for ASEAN University Network / Southeast Asia Engineering Education Development Network (JICA project for AUN-SEED-Net) for the scholarship of Collaborative Education Program (CEP) that allows me to pursue my master's degree and have the opportunity to broaden both my knowledge and connection with people within the construction industry

Finally, I would like to express my deepest gratitude to my family members (Reviyani Malinda Soulisa, Acung, Genu) and friends, especially Addinda, Precelia, Alver, and Aurora, for their continuous support I am also forever thankful for my favorite characters Suna Rintarou, Miya Atsumu, Miya Osamu, and Kokonoi Hajime, for being my source of motivation

ABSTRACT

The layout of a construction site must be properly planned to guarantee safe and effective operations Planning the construction site's layout has a significant impact on a project's productivity, budget, and timeline An effective process for optimization models has been developed using a number of different algorithms Artificial intelligence-based solutions, such as metaheuristic algorithms, have been studied in depth for the site planning problem in the construction industry However, the Ant Lion Optimization (ALO) algorithm has limited research in solving site layout optimization problems although proven consistent for problem that require an optimization approach Moreover, a novel hybrid ALO algorithm is necessary to produce a more optimal solution with better runtime Thus, this study developed a novel hybrid Ant Lion Optimizer (ALO) algorithm and applied it to a case study representing the site layout problem to determine the optimal site layout and travel distance Overall, the proposed algorithm outperformed other algorithms in terms of consistency and efficiency and can be used to solve site layout problem

Keywords: Construction, Site Layout, Optimization, Novel Hybrid Ant Lion Optimizer (ALO), Algorithm

TÓM TẮT LUẬN VĂN THẠC SĨ

Bố trí mặt bằng thi công của một công trường phải được lên kế hoạch chính xác để đảm bảo các hoạt động diễn ra an toàn và hiệu quả Việc lên kế hoạch sắp xếp mặt bằng thi công của một công trường ảnh hưởng quan trọng đến hiệu suất, ngân sách và thời gian thực hiện dự án Các thuật toán trí tuệ nhân tạo đã được sử dụng để tạo lập mô hình tối ưu hóa mặt bằng thi công Các giải pháp dựa trên trí tuệ nhân tạo, chẳng hạn như thuật toán metaheuristic, đã được nghiên cứu chuyên sâu cho việc bố trí bề mặt công trường trong ngành xây dựng Tuy nhiên, các nghiên cứu

về thuật toán tối ưu kiến sư tử (ALO) rất hạn chế trong việc giải quyết các vấn đề tối

ưu hóa bố cục mặc dù đã được chứng minh là nhất quán đối với vấn đề tối ưu hóa toàn cục Hơn nữa, một thuật toán ALO lai ghép mới là cần thiết để tạo ra một giải pháp tối ưu hơn với thời gian chạy tốt hơn Do đó, nghiên cứu này đã phát triển một thuật toán lai ghép kiến sử tử mới (ALO) áp dụng vào một nghiên cứu điển hình đại diện cho vấn đề bố cục lại bề mặt công trường xây dựng để tối ưu hoá khoảng cách

di chuyển Nhìn chung, thuật toán được đề xuất vượt trội so với các thuật toán khác

về tính nhất quán, hiệu quả và có thể được sử dụng để giải quyết vấn đề tối ưu bố trí mặt bằng thi công trên công trường

Từ khóa: Xây dựng, Bố trí mặt bằng thi công, Tối ưu hóa, Thuật toán lai ghép kiến

sư tử mới (ALO)

AUTHOR’S COMMITMENT

The undersigned below:

Place and date of born : Bandung, Indonesia, February 3, 1994

With this declaring that the master thesis entitled “Site Layout Optimization Using Novel Hybrid Ant Lion Optimizer (ALO) Algorithm” is done by the author under

supervision of the instructor All works, ideas, and material that was gain from other references have been cited in the corrected way

Ho Chi Minh City, December 14, 2022

Febrina Vienna Soulisa

TABLE OF CONTENTS

THE TASK SHEET OF MASTER’S THESIS i

ACKNOWLEDGEMENT ii

ABSTRACT iii

TÓM TẮT LUẬN VĂN THẠC SĨ iv

AUTHOR’S COMMITMENT v

TABLE OF CONTENTS vi

LIST OF FIGURES ix

1 INTRODUCTION 1

1.1 Problem Statement 1

1.2 Research Objectives 3

1.3 Scope of Study 3

1.4 Research Methodology 4

1.5 Academic and Practical Significances 5

1.5.1 Academically 5

1.5.2 Practically 6

2 LITERATURE REVIEW 7

2.1 Construction Site Layout Optimization 7

2.2 Ant Lion Optimizer (ALO) Algorithm 8

2.3 Related Studies 11

3 DEVELOPMENT AND APPLICATION 15

3.1 Application of Ant Lion Optimizer (ALO) 15

3.2 Novel Hybrid Ant Lion Optimizer (NH-ALO) 18

3.3 Tournament selection 20

3.4 Opposition-based learning (OBL) 20

3.5 Mutation and crossover 21

4 CASE STUDIES 24

4.1 Objective Function and Parameter 24

4.2 Case Study 1 24

4.3 Case Study 2 30

4.4 Case Study 3 34

4.5 Case Study 4 39

5 CONCLUSION 44

5.1 Conclusion 44

5.2 Future Research 45

REFERENCES 46

APPENDICES 52

Appendix 1 MATLAB Coding 53

ALO.m 53 caseStudy1.m 56

caseStudy2.m 57

caseStudy3.m 58

caseStudy4.m 59

CrossOverU.m 60

graphResult.m 60

initialization.m 60

main_cs1.m 61

main_cs1.m (2) 62

main_cs2.m 63

main_cs3.m 64

main_cs4.m 65

mutation.m 66

Random_walk_around_antlion.m 66

show_result.m 68

TournamentSelection.m 69

Appendix 2 MATLAB Command Window Result 70

Case study 1 command window result 70

Case study 1 command window result (2) 71

Case study 2 command window result 76

Case study 3 command window result 77

Case study 4 command window result 78

Appendix 3 MATLAB Excel result 79

Case study 1, 2, 3, & 4 excel result 79

Case study 1 (2) excel result 80

LIST OF FIGURES

Figure 1-1 Time-Cost-Quality in project management 2

Figure 1-2 Research objectives 3

Figure 1-3 Scope of study 3

Figure 1-4 Research Flowchart 4

Figure 2-1 Predatory behavior of antlions [34] 9

Figure 2-2 Ant's random walk inside the trap [30] 10

Figure 2-3 Scopus report on ALO application [36] 11

Figure 3-1 Ant lion predation [28] 15

Figure 3-2 Illustration of three ants' random walk [30] 16

Figure 3-3 Flowchart of novel hybrid ant lion optimizer (NH-ALO) algorithm 19

Figure 3-4 Theoretical illustration of OBL in one-dimension optimization [49] 21

Figure 3-5 Theoretical illustration of mutation [50] 22

Figure 3-6 Conceptual illustration of crossover [51] 23

Figure 4-1 Pre-arranged site layout of case 1 25

Figure 4-2 Convergence curve NH-ALO for case 1 27

Figure 4-3 NH-ALO site layout arrangement of case 1 28

Figure 4-4 Convergence curve NH-ALO for case 1 (200 iterations) 29

Figure 4-5 The convergence curve comparison 30

Figure 4-6 Pre-arranged site layout of case 2 30

Figure 4-7 Convergence curve NH-ALO for case 2 33

Figure 4-8 NH-ALO site layout arrangement of case 2 34

Figure 4-9 Pre-arranged site layout of case 3 35

Figure 4-10 Convergence curve NH-ALO for case 3 37

Figure 4-11 NH-ALO site layout arrangement of case 3 38

Figure 4-12 Pre-arranged site layout of case 4 39

Figure 4-13 Convergence curve NH-ALO for case 4 42

Figure 4-14 NH-ALO site layout arrangement of case 4 43

LIST OF TABLES

Table 2-1 List of related studies 14

Table 4-1 Pre-arranged site layout information of case 1 25

Table 4-2 Distance between locations of case 1 (meters) 26

Table 4-3 Trip frequencies of case 1 26

Table 4-4 Total traveling distance comparison of case 1 27

Table 4-5 Site layout arrangement for case 1 28

Table 4-6 Total traveling distance case 1 (200 iterations) 29

Table 4-7 Pre-arranged site layout information of case 2 31

Table 4-8 Distance between locations of case 2 (meters) 31

Table 4-9 Trip frequencies of case 2 32

Table 4-10 Total traveling distance comparison of case 2 33

Table 4-11 Site layout arrangement for case 2 33

Table 4-12 Pre-arranged site layout information of case 3 35

Table 4-13 Distance between locations of case 3 (meters) 36

Table 4-14 Trip frequencies of case 3 36

Table 4-15 Total traveling distance comparison of case 3 37

Table 4-16 Site layout arrangement for case 3 38

Table 4-17 Pre-arranged site layout information of case 4 40

Table 4-18 Distance between locations of case 4 (meters) 40

Table 4-19 Trip frequencies of case 4 41

Table 4-20 Total traveling distance comparison of case 4 42

Table 4-21 Site layout arrangement for case 4 43

1 INTRODUCTION

- - -

- - -

This chapter presents the research problem which describes the

importance of well-planned site layout and brief explanation about

optimization of site layout (Section 1.1), research objectives (Section

1.2), scope of study (Section 1.3), research methodology (Section 1.4)

The academic and practical significances (Section 1.5) also presented

in the last part of this chapter

1.1 Problem Statement

It is known that construction and project planners often aim to accomplish maximum quality by considering minimal cost and time The most common challenge is maximizing resource utilization to strike a balance between various and frequently conflicted aspects of project management Thus, quality, cost, and time are intricately related as these three objectives are broadly discussed for their practical relevance in project accomplishment [1] The layout of the construction site also has

a significant impact on the project’s cost, productivity, safety, and other aspects [2] However, the site layout plan is frequently created without consideration for the project plan itself and, occasionally, no specific site layout plan in practice [3] As a result, it leads to long project duration, poor quality, and expensive costs on the construction

Figure 1-1 Time-Cost-Quality in project management

A construction site has a space that is considered a limited resource in construction projects compared to other resources such as material and equipment Layout planning of construction sites often interacts with construction management (i.e., planning, scheduling, and cost estimating) [2], [4] A study mentioned that site layout planning should be carried out successfully prior to starting the main work A well-planned layout will contribute to saving time and site congestion Moreover, minimize travel distance, material handling effort, and operational cost [5], [6] This process of organizing facilities, known as the production factors, is also considered part of operational strategies to produce a more efficient system [7] Besides, workers and site personnel spend most of their time on construction sites Therefore, if they can move around the site easily and quickly, it will improve productivity [8]

Layout planning for each construction site is unique as it depends on variables such as workspace and interaction between each location [9], [10] In order to optimize workspace utilization and minimize construction conflicts, the constraints

of each construction project must be considered [8], [11], [12] Hence, optimization techniques have been applied to find the solution to site layout problems [13] An improved model of the optimization algorithm is also faster than manually determining the information [14] Moreover, generating optimal solutions contribute

to reducing material handling cost by about 10-30% due to better material flow of the site [15], in other words, proving the optimization model's performance

The Ant Lion Optimization (ALO) algorithm has limited research in solving site layout optimization problems In other studies that require an optimization approach, using ALO is proven consistent It also has simplicity and guaranteed convergence to generate high-quality solutions [16] However, a novel hybrid ALO algorithm is necessary to produce a more optimal solution with better runtime The convergence level increased by adding and replacing methods for the original ALO algorithm Hence, this study focuses on developing a novel hybrid ALO algorithm and its implementation in four site layout optimization case studies

1.2 Research Objectives

This research has four objectives, which are depicted in Figure 1-2 Those objectives are:

• To develop a novel hybrid Ant Lion Optimizer (ALO) algorithm,

• To assess the performance of the developed novel hybrid Ant Lion Optimizer (ALO) algorithm in three case studies,

• To produce an applicable novel hybrid Ant Lion Optimizer (ALO) algorithm,

• To apply the produced novel hybrid Ant Lion Optimizer (ALO) algorithm in

a new case study to generate optimum site layout arrangement

Figure 1-2 Research objectives

1.4 Research Methodology

The overall research flowchart starts by conducting a literature review to define the problem statement related to site layout optimization Next, developing the novel hybrid Ant Lion Optimization (ALO) algorithm In this research, several methods were used in order to generate better performance than the original ALO algorithm via MATLAB Once the performance of the developed algorithm is confirmed to be more consistent compared to algorithms used in the case study, it is then applied to a new case Last, the conclusion and suggestions were made based on the findings

Figure 1-4 Research Flowchart

Therefore, this study consists of the following content:

• Introduction

This chapter describes the problem statement by explaining the importance of construction site layout planning, research objectives, the scope of study, the methodology used, and the significance of this study

• Literature review

This chapter summarizes construction site layout optimization, the ALO algorithm, and findings through related studies

• Development and application

This chapter introduces the proposed novel hybrid ALO algorithm model and the methods applied to increase the original ALO capability further and enhance its efficacy

• Case Study

This chapter explains the application and evaluation of novel hybrid ALO algorithm performance in four case studies The evaluation consists of a comparison between the algorithm with other algorithms' results

• Conclusion

This chapter concludes the results and summarizes the study Recommendation for future research is also suggested in this chapter

1.5 Academic and Practical Significances

The research contributes to the construction industry's academic and practical fields by providing the optimal solution for site layout arrangement and increasing productivity by minimizing total traveling distance

1.5.1 Academically

Due to limited study utilizing the Ant Lion Optimizer (ALO) for site layout problem, the academic significances of this study are as follows:

• Expand the Ant Lion Optimizer (ALO) algorithm application for site layout optimization through iterative computations related to specified criteria instead of making excessive hypotheses about the optimization problem

• Modifying the ALO algorithm and developing a novel hybrid version to improve the ability to produce an optimal solution

• Open new possibilities for improvement of the ALO algorithm and suggestions by implementing new methods

• Proving the capability of the ALO algorithm and its potential for an optimization-related problem, especially for layout planning

• Encourage further study and implementation of the ALO algorithm

1.5.2 Practically

The developed novel hybrid Ant Lion Optimizer (ALO) algorithm is expected

to become a useful decision instrument to generate an optimal solution for the site layout arrangement of the actual construction site with minimum total traveling distance Thus, the practical significances are:

• Create an optimal site layout with optimal maximum travel distance between each facility

• Help provide better workspace for workers and site personnel to move during construction by reducing the possibility of site congestion

• Save more time by considering the distance and frequency of travel Therefore, reducing unnecessary material handling effort and expenses

• Lessen the risks of accidents and incidents on site

• Help create a better project system and management

• Provide a useful decision tool for the contractor to use and encourage the application of AI for their project

• The ALO algorithm is also applicable to transportation-related problems and the layout of manufacturing-related problems

2 LITERATURE REVIEW

- - -

- - -

This chapter consists of a summary of the literature review and its

findings Section 2.1 describes construction site layout optimization

and its importance, while section 2.2 describes the Ant Lion Optimizer

(ALO) algorithm In addition, section 2.3 listed the studies of

construction site layout optimization that implement heuristic

approaches and the application of the ALO algorithm in

construction-related problems, its advantages, and drawbacks This chapter further

emphasizes the need for a novel hybrid ALO algorithm and the

references used for performance evaluation of the proposed algorithm

2.1 Construction Site Layout Optimization

There are several obstacles that must be overcome in the engineering and management of construction projects due to the existence of numerous workers and resources As operational efficiency rises, associated benefits like decreased costs and shortened project timelines become more apparent [17]–[19] Construction sites are dynamic environments that pose numerous hazards, and the importance of a site layout plan cannot be overstated A well-thought-out site layout plan can save down

on expenses and time spent on a project while also fostering efficiency, safety, and a better workflow [20]–[24] More precisely, a site layout design can save operational expenses by 20% to 50% through minimizing the risk of potential conflicts with material handling, congestion on the site, and travel distances [25]

An effective process for optimization models has been developed using a number of different algorithms Artificial intelligence-based solutions, such as metaheuristic algorithms, have been studied in depth for the site planning problem in the construction industry Optimal site layout was investigated in a 2018 Prayogo study, which compared the efficacy of the Particle Swarm Optimization (PSO), the Artificial Bee Colony (ABC), and the Symbiotic Organisms Search (SOS) algorithm [26] by optimizing workers' travel distance given their trip frequency Additionally,

the 2018 hybrid WOA-CBO algorithm [27] and the 2020 hybrid SOS with Local Operators (HSOS-LO) algorithm [25] improved upon their predecessors in terms of consistency and efficiency The total distance traveled is calculated as follows [25]–[27]:

(2-1)

Subjected to:

(2-2)

(2-3) (2-4)

where n represents the number of facilities, TD is the travel distance, and f jk and d jl

represent frequency and distance between locations i and j, respectively

The computationally expensive ALO algorithm [28], [29] outperformed PSO,

GA, SMS, BA, FPA, CS, and FA It is necessary to conduct further research to improve the ALO algorithm as well as other random walk algorithms [30] In an effort

to increase accuracy, convergence, and computational efficiency, some research has substituted the TS technique [31], [32] for the roulette wheel approach and the Laplace distribution and OBL to increase the exploration area [33]

2.2 Ant Lion Optimizer (ALO) Algorithm

Predatory activity in antlions is typically directed at unsuspecting passing ants

As a larva, it would crawl in a circle to excavate a conical pit with its gigantic jaws, and then it would empty the pit of sand to entice passing prey The antlion will then wait as they hide further inside the trap This behavior inspires the Ant Lion Optimizer (ALO) algorithm as it depicts the behavior through a model Figure 2-1 represents the predatory behavior of antlions [34]

Figure 2-1 Predatory behavior of antlions [34]

When Mirjalili first introduced it in 2015, the ALO algorithm simulated five crucial larval hunting processes Those are the random walk of ants, development of traps, capture of ants, sliding ants towards antlion, predation and rebuilding of the pit, and elitism [30]

2.2.1 Random Walk of Ants

It is difficult to identify the optimum method for a randomly changeable problem At each stage of optimization, the ants randomly relocate Considering that every search space is constrained by some boundary Hence, min-max normalization

is used to confine the ants' random travel to the confines of the search space The

position update of an ant with respect to an antlion is modeled in [35]

2.2.2 Developing Traps

An antlion is chosen using the roulette wheel operator for each ant Each ant can only enter one antlion trap every iteration in the ALO algorithm The antlion chosen by the roulette wheel for each ant is the one that has captured the ant T the roulette wheel operator favors solutions with higher fitness functions similar to an antlion with a bigger trap that can catch more ants [34]

2.2.3 The Capture of Ants

The Ant Lion has taken a position at the bottom of the pit, where it is patiently awaiting its food Roulette wheels are utilized by ALO as their operator Antlions are picked using a roulette wheel, which is an operator that takes into account the antlion fitness value The Ant Lion's fitness level determines the size of the pit it digs to trap its prey In this model, the effect of antlion traps on the random

paths of ants is simulated [35]

2.2.4 Sliding ants towards antlion

In order to prevent an ant from escaping its trap, an antlion will begin tossing sand outward from the pit's center For the purpose of mathematically modeling this phase, the radius of the ants' random walks hyper-sphere is reduced in

an adaptive manner [35] The figure below shows an ant's random path as an antlion traps it

Figure 2-2 Ant's random walk inside the trap [30]

2.2.5 Predation and rebuilding the pit

During the final phase of antlion hunting, an ant slips into the trap and is captured by the antlion's jaw The ant is dragged into the sand, where the antlion devours it The ALO algorithm determines that an ant has successfully captured its prey when its fitness function has become superior to that of its corresponding antlion

The antlion then shifts its position with the hunted ant [34]

2.2.6 Elitism

By definition, elitism saves the finest solutions from each generation Therefore, the fittest antlion that has been created is kept and treated as an elite that influences all ants' movement per iteration The roulette wheel's randomly selected antlion also affects the ant’s movement Elitism helps evolutionary algorithms locate

the optimal solution at any optimization stage [35]

2.3 Related Studies

Various study has been developing many algorithms to solve construction site layout problem Moreover, discussing the importance of the site layout itself Meanwhile, according to Scopus, ALO algorithm application in the engineering field

is 25%, which is second place after computer science with 33.7% [36]

Figure 2-3 Scopus report on ALO application [36]

For non-construction related studies that applied ALO emphasize that it balances exploration and exploitation with global and local searches The ALO approach can swiftly find a global optimum and explore the search space due to its rapid convergence, effective exploration employing random walks and random selection of search agents, and precise exploitation using adjustive trap limits [37] In

2018, a summary of the ALO algorithm and its implementation concluded that In contrast to other advanced swarm intelligence optimization algorithms and conventional algorithms, its method for dealing with optimization issues in the real world is substantially more applicable [35] However, although ALO provides excellent solutions for various engineering optimization challenges, it has a few drawbacks The run time of the ALO algorithm is particularly slow, which is primarily attributable to the random ant walking model This is the method's primary shortcoming [38]

Several studies related to civil engineering have implemented the ALO algorithm A recent review paper summarizes the ALO application, its variants, and hybrids [36] For civil engineering related, the application of ALO for structural damage detection using vibration data [39] and novel multi-objective ALO for detecting structural damage [40] are listed In addition, a study implemented the algorithm to produce an optimum skeletal structures design [41] in 2016 A recent study develops an ALO algorithm to produce an optimal design of high-performance concrete mixes [42] However, few studies have explored the ALO algorithm's application to the site planning problem

Many studies have used various decision tools to aim for effective and efficient construction In the case of construction site layout, several studies are focusing on applying artificial intelligence to find the optimal solution these past ten years Metaheuristic algorithms are often used to find the solution [22] A study by Prayogo

in 2018 was conducted to compare the performance of three algorithms The Particle Swarm Optimization (PSO), Artificial Bee Colony (ABC), and Symbiotic Organisms Search (SOS) algorithms in optimizing site layout arrangement of actual case studies [26] The model was used to determine the optimum layout by minimizing workers’

traveling distance between each location given the traveling frequency In addition, a hybrid Whale Optimization Algorithm (WOA) – Colliding Bodies Optimization (CBO) algorithm was implemented with the same objective by Kaveh in 2018 [27] There is yet a study utilizing a hybrid ALO algorithm related to construction site layout arrangement However, it proves its applicability and potential to contribute and provide an optimal solution as there are many studies of different scopes utilizing the ALO algorithm [16]

Multiple benchmarks are used to demonstrate the efficacy of the ALO algorithm Successful applications of ALO have been made across 19 benchmark functions and 4 traditional engineering challenges [43] As such, the algorithm might

be viewed as a potential nature-inspired alternative method for addressing difficult engineering problems [44] Even so, it is recommended to hybridize the ALO algorithm to handle practical problems, as its long run time is due to the random walking process [45] Hence, this study developed a novel hybrid ALO algorithm 1) implements TS to increase the convergence level of each iteration and 2) implements OBL and MCS to improve performance The proposed algorithm is expected to serve

as a decision tool with higher stability and better performance than other algorithms

in solving the construction site layout problem

The list of related studies to construction site layout problems and the ALO algorithm for the construction-related problem is listed in Table 2-1

Table 2-1 List of related studies

3 DEVELOPMENT AND APPLICATION

- - -

- - -

This chapter presents the development and application of a novel

hybrid ALO algorithm model Section 3.1 explain the application of ALO

algorithm while section 3.2 shows the flowchart of the novel hybrid ALO

algorithm’s scheme Sections 3.3, 3.4, and 3.5 describe the methods

implemented in the novel hybrid ALO algorithm Those methods are

opposition-based learning in section 3.4 and mutation and crossover in

section 3.5 Moreover, the roulette wheel methods were replaced by

tournament selection in section 3.3

3.1 Application of Ant Lion Optimizer (ALO)

As an insect, antlions belong to Myrmeleontidae, the Neuropteran family, and possess predatory behavior Predation has been a part of its lifestyle ever since it was

a larva when it would travel in a circle to dig a conical pit with its massive jaws and then remove the sand from the pit to lure passing prey [46] The antlion larvae will then wait as they conceal themselves further within the trap The passing ants is a typical prey The antlion predation is shown in figure below

Figure 3-1 Ant lion predation [28]

The ALO algorithm mimics the ants passing and moving through the search space and antlion predatory behavior by using pit traps Ants will naturally wander around randomly in search of food Consequently, its random movement can be represented by the equation shown below [34]:

(3-1)

where cumsum represents the cumulative summation; n represents the maximum number of iterations; t represents the iteration index Meanwhile, the random function, r(t), is as follows:

(3-2)

for this function, the rand repressents a randomly generated number in [0,1], and

Figure 3-2 illustrates the result of three ants’ random walk through 500 iterations From the figure, the random walk shows significant deviation from its starting position, which is depicted by red, the upsurge by black, and the downturn by blue

Figure 3-2 Illustration of three ants' random walk [30]

𝑟 𝑡 = 1 𝑖𝑓 𝑟𝑎𝑛𝑑 > 0.5

0 𝑖𝑓 𝑟𝑎𝑛𝑑 ≤ 0.5

Each ants’ position is saved and then used throughout the optimization in the matrix in Eq 3-3

(3-3)

where:

M Ant = matrix that stores the position of each ant,

A i,j = position of the ith ant in the jth dimension,

n = number of ants in the population,

d = number of variables (the number of dimensions)

In addition, each ant’s position refers to the parameter for a particular solution

During the optimization process, the M Ant matrix was considered to store all ants’ positions (variables of all solutions)

The fitness feature is used throughout the optimization process to evaluate

each ant The matrix that stores the fitness value of every ant is as follows:

(3-4)

where:

M OA = matrix to store the fitness function of each ant,

A i, j = value of the jth dimension of the ith ant,

n = number of ants in the population,

f = objective function

Moreover, the antlions are storing their positions and power value by hiding somewhere within the search space The following matrices is used for this behavior

(3-5)

where:

M Antlion = matrix to store the position of each antlion,

AL i, j = value of the jth dimension of the ith antlion,

n = number of antlions,

d = number of variables

(3-6)

where:

M OAL = matrix to store the fitness of each antlion,

AL i, j = value of the jth dimension of the ith antlion,

n = number of antlions,

f = objective function

3.2 Novel Hybrid Ant Lion Optimizer (NH-ALO)

Inspired by Antlion's hunting behavior, the first version of the Ant Lion Optimizer (ALO) was developed Methods like opposition-based learning, mutation, and crossover were used In addition, tournament selection is implemented to produce

a hybrid version The efficiency and accuracy will increase by adding and replacing the methods during random walks Figure 3-3 displays the algorithm's scheme with the initial current iteration equal to 1 The parameter is set similarly to the data from references in order to make a comparison In this case, the population size, maximum iteration, and objective function that includes frequencies and travel distance from each location of facilities The process of elitism will be repeatedly done As the maximum iteration is finished, the hybrid model expects the optimal solution

Calculate fitness functions of initial ants and antlions

Determine elite antlion

For each ant select antlion using tournament selection

Apply OBL, MCS, and replace antlion positions with corresponding ant, if fitness is better

Update elite antlion

3.3 Tournament selection

Tournament selection replaces the roulette wheel method to develop a hybrid version that improves efficiency and reduces lengthy runtime in the optimization process [31] It has been demonstrated that the tournament selection approach is superior to the roulette wheel method in terms of solving minimization challenges It

is advised that this method be utilized for a variety of models in order to solve optimization-related problems [33] The tournament selection method produces k elements randomly and selects those with better values for the objective function based on comparison [47] Thus, improving the competence to obtain the optimal value

For this study, the value of k = 5 It indicates that the likelihood of discovering

a suitable candidate is increased five times The process of sampling and selecting is the main emphasis of the selection

3.4 Opposition-based learning (OBL)

Compared to OBL, more than fifty percent of expected solutions deviate from the globally optimal solution based on probability theory The notion of opposition-based learning is to develop a solution opposed to the original In addition, this method is applicable for the initial and new solutions generated by the algorithm until the optimum solution is reached Initiating the opposite prediction to accelerate the convergence [48] The theoretical example of OBL in one-dimensional optimization

is shown in Figure 3-4 Choosing a superior solution between locations x and x*shortens the search area Eventually increasing the effectiveness of the search and bringing x closer to x*

Figure 3-4 Theoretical illustration of OBL in one-dimension optimization [49] 3.5 Mutation and crossover

Mutation and crossover are two common processes used in optimization and

performed at different stages The mathematical model for a single vector with n

dimensions for each 𝑥𝑖 = {𝑥𝑖1, 𝑥𝑖2, , 𝑥𝑖𝑛}

Step 1: Mutation

The mutation algorithm selects components from vectors x a , x b , x c 𝑎 ≠ 𝑏 ≠ 𝑏 ≠

𝑖 randomly in order to generate a mutation vector u i The model consists of F as

a random number representing various sizes of mutation with the range of (0;1)

The mutation probability factor (pm) for this study is pm = 0.01 The formula is

as follows:

The theoretical illustration of a mutation is shown in Figure 3-5, where it modifies one or more genes to keep population variety and prevent similar solutions from emerging

Figure 3-5 Theoretical illustration of mutation [50]

The probability factor (p c) controls the population’s diversity and reduces the

localized optimum risk For this study, the determined value of p c = 0.6

Meanwhile, j 0 represents an index [1,2,3, ,n] that guarantees vector v i to, at least,

inherit an element from the mutant of vector u i The conceptual illustration of

crossover is shown in the Figure 3-6, where the crossover point is randomly selected

Figure 3-6 Conceptual illustration of crossover [51]

4 CASE STUDIES

- - -

- - -

This chapter presents the result of four case studies Case studies 1, 2,

and 3, were obtained from references The result of the proposed

algorithm for those case studies is compared to other algorithms While

case study 4 is an additional case This chapter is divided into five

sections to discuss the objective function and parameter, result, and its

meaning further Those sections are Case Study 1 (Section 4.2), Case

Study 2 (Section 4.3), Case Study 3 (Section 4.4), and Case Study 4

(Section 4.5)

4.1 Objective Function and Parameter

By referring to Eq 2-1 for the objective function, the input data for each case

study before running the program are:

• The initial position of each location

• Travel distance between each location

• Frequency of trips between each location

Furthermore, the value for tournament selection is k = 5, and the probability

factor for crossover is pc = 0.6 In addition, the probability factor for mutation is pm

= 0.01 Aside from the parameter mentioned above, the population size (popsize) and maximum iteration (maxiter) are required Therefore, to compare the result and

evaluate the performance, those two parameters should be set similarly to the references and will be mentioned accordingly

4.2 Case Study 1

First case [26] contains 11 facilities for 11 respective locations The permanent locations are the site gate (SG) and main gate (MG) Both facilities are permanently located in the first and tenth locations The pre-arranged layout is shown in Figure 4-

1 as follows:

Figure 4-1 Pre-arranged site layout of case 1 The location information is shown in Table 4-1 with the information on the listed facilities, location numbers, and notes on the permanent locations

Table 4-1 Pre-arranged site layout information of case 1

The information on traveling distance, in meters, between each of the 11 locations is shown in Table 4-2

Table 4-2 Distance between locations of case 1 (meters)

Trip frequencies between each of the facilities are shown in Table 4-3

Table 4-3 Trip frequencies of case 1

For this study, the parameter for population and maximum iteration are:

• popsize = 30 population

• maxiter = 30 iterations

After 30 iterations, the outcome (Figure 4-2) from NH-ALO algorithm is compared with the outcome from other algorithms in Table 4-4 NH-ALO is able to have a similar minimum (12546 m), lowest maximum (12578 m), and average total

traveling distance (12549.20 m) with a standard deviation value of 9.764 In this case, the proposed algorithm has the lowest standard deviation, which proves its consistency

Figure 4-2 Convergence curve NH-ALO for case 1 Table 4-4 Total traveling distance comparison of case 1

The site layout arrangement for each algorithm is shown in Table 4-5, while the new arrangement from NH-ALO algorithm is shown in Figure 4-3