See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/352539622

OPTIMAL CREW ROUTING FOR LINEAR REPETITIVE PROJECTS USING GRAPH

THEORY AND ENTROPY MAXIMIZATION METRIC

Thesis · February 2015

DOI: 10.13140/RG.2.2.36341.91360

CITATIONS READS

0 15

1 author:

Ahmed Gouda Mohamed

The British University in Egypt
5 PUBLICATIONS 21 CITATIONS

SEE PROFILE

All content following this page was uploaded by Ahmed Gouda Mohamed on 19 June 2021.

The user has requested enhancement of the downloaded file.

 School of Sciences and Engineering

OPTIMAL CREW ROUTING FOR LINEAR REPETITIVE PROJECTS

USING GRAPH THEORY AND ENTROPY MAXIMIZATION METRIC

A Thesis submitted to the

Department of Construction and Architectural Engineering

In partial fulfillment of the requirements for the degree of

Master of Science

With specialization in:

Construction Engineering

By

Eng. Ahmed Gouda Mohamed

Under the supervision of:

Dr. Ossama Hosny Dr. Khaled Nassar

Professor, Construction and Architecture Associate Professor, Construction and

Engineering Department Architecture Engineering Department

The American University in Cairo, Egypt The American University in Cairo, Egypt

January 2015

|PageI
 DEDICATIONS

I dedicate my thesis work to my family, to my father, to my beloved

mother, to my beautiful wife and my daughter, for their love,

understanding, encouragement, and support.

| P a g e II
ACKNOWLEDGEMENTS

First and foremost I would like to thank Allah for his gracefulness for providing me

the enough patience, courage and wisdom for finishing my masters.

I would like to thank my thesis advisors, Dr. Osama Hosny and Dr. Khaled Nassar,

whom led me into my masters’ study and their continuous support that granted me

opportunities to gain valuable experience in both the academic and the professional

worlds.

I have to thank Dr. Ahmed Al Hakim for his helpful pieces of advice throughout my

work with him to whom I owe a debt of gratitude for his assistance. I have also to

thanks Dr. Dorra el-Khayam for his supportive pieces of information and his

assistance in achieving my work.

I owe my deepest gratitude to my father "Dr Gouda Ghanem", your motivation and

support has always enlighten my way. You are being my role model and have

inspired me to try something I would have never thought possible.

To my mother, you have always encouraged me through my life and usually standing

beside me. I am eternally grateful to you. To my brother, sister, and her husband. I am

really fortunate to have you as my family.

Leaving the best for last always, I would like to thank my wife and my beautiful

daughter "Lara" whom I owe everything. They are my muses, my power and my main

passion. They have been with me at each step and for that I am eternally grateful.

| P a g e III
 The American University in Cairo

OPTIMAL CREW ROUTING FOR LINEAR REPETITIVE PROJECTS

USING GRAPH THEORY AND ENTROPY MAXIMIZATION METRIC

By:

Ahmed Gouda Mohamed

Thesis Advisor: Dr. Ossama Hosny and Dr.Khaled Nassar

ABSTRACT
Construction projects that contain several identical or similar units are usually known

as repetitive or linear projects. Repetitive projects are regarded as a wide umbrella

that includes various categories of construction projects and represents a considerable

portion of the construction industry, and contain uniform repetition of work. CPM has

been proved to be inefficient in scheduling linear projects because CPM does not

address two key aspects, which are maintaining crew work continuity, and achieving a

constant rate of progress to meet a given deadline. Line-of-balance (LOB) is a linear

scheduling methodology that produces a work schedule in which resource allocation

is automatically performed to provide a continuous and uninterrupted use of resource.

The fundamental principles of LOB have several shortfalls that raise many concerns

about LOB, which need to be attuned and improved in order to suit the nature of

construction projects. Hence, this thesis proposes a hybrid approach for scheduling

linear projects that stresses on the limitation of LOB scheduling technique. To meet

the physical limitation of resources in linear projects, this study presents a flexible

optimization model for resolving resource constraint dilemma in linear scheduling

problems .The proposed model utilizes a MATLAB code as the searching algorithm

to automate the model formulation. The novelty of this model is supplementing a new

optimization engine and a decision supporting system that formulate the optimal

| P a g e IV
crews routing between different activities in different units and guarantee the optimal

crew distribution for cost efficiency. This model investigates the mechanics of

allocating a multi- task skilled workforce between different activities in different units

that can lead to increased productivity, flexibility, and work continuity; besides, this

model has the capability of accurately identifying the critical path in linear projects.

Furthermore, to avoid day-to-day fluctuation in resource demands, this study

encompasses a simulation model for handling the resource leveling in linear

construction projects. The proposed model was implemented using crystal ball ribbon

based on an entropy maximization metric. The entropy-maximization method

accounts for such possibility of allowing activity duration to be stretched or crunched

relying on activity type without affecting total completion date of a project and

provides more optimized resource allocation solutions. A case study for a 4-km

sewage pipeline is used to demonstrate the capability of the proposed models, which

illustrates the implementation of the proposed models in construction projects.

|PageV
 TABLE OF CONTENTS

CHAPTER 1 - INTRODUCTION ...................................................................... 1

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

1.2 Line of Balance………………………………………………………………..4

1.3 Statement of the problem. …………………………………………………….5

1.3.1 Limitation in LOB ………………………………..………….…..…..…..6

1.3.2 Leveling of Resources in LOB……………………………….…..…..…..7

1.4 Objectives and Scope…………………………………..………………....…..8

1.5 Research Methodology. …………………………………..…..……………...9

1.6 Layout of the Thesis. …………………………………..…..…..…………....10

CHAPTER 2 – LITERATURE REVIEW ...................................................... 12

2.1 Line-of-Balance…………..…..…..…………………………………………13

2.1.1 Historical Review about LOB. …………..…..…..………………….….13

2.1.2 Basic Procedures undertaken to perform LOB …………..…..…..….…14

2.1.3 Challenges and enhancements to improve LOB for Construction Industry.15

2.1.4 Concept of ‘Criticalness in LOB. …………..…..…..……….………... 18

2.1.5 Learning Curve Effect on LOB. …………..…..…..………….………..19

2.1.6 CPM and Line of Balance integration Methodology…………..……….20

2.1.6.1 Calculation procedures of integrated CPM and LOB…………..… 21

2.1.6.2 Previous research effort of integrated CPM and LOB …………….23

2.2 Resource Management in repetitive projects. …………..…..…..……….…25

2.2.1 Introduction. …………..…..…..…………………………………..……25

2.2.2 Previous research in the field of resource management of linear projects.26

2.2.3 Leveling in Linear repetitive schedule Projects …………..…..…..……..28

2.2.4 Heuristic approaches used for leveling Resources. …………..…..…..… 28

| P a g e VI
 2.2.5 Optimization approaches used for leveling Resources. …………..…..…30

2.2.6 Metaheuristic approaches used for leveling Resources. …………..…….31

2.3 Entropy Maximization …………………………………….……..…..…..….33

2.3.1 State of the Art…………..…..…..……………………………………....33

2.3.2 The Application of Entropy Theory in Construction Management……..34

2.4 Graph Theory………………………………………………………………..36

2.4.1 State of the Art…………………………………………………………..36

2.4.2 Graph theory and Sparse Matrices………………………………………37

2.5 Discussion of literature findings…………………………………………….39

CHAPTER 3 – RESEARCH METHODOLOGY AND MODEL

DEVELOPMENT.............................................................................................. 44

3.1 Introduction …………..……………………………………………………..44

3.2 Areas Requiring Enhancement ……………………………………………...46.

3.2.1 A developed schedule approach for managing and manipulating repetitive

construction projects…………………………………………………………….46

3.2.2. Clearly identifying the critical path of linear construction projects…… 47

3.2.3. Minimizing the maximum number of crews utilized to perform a repetitive

activity by changing their routing.………………………………..……………. 47

3.2.4. Enhancing the resource leveling in LOB scheduling technique………. 49

3.3. The Proposed Representation of LOB Scheduling Technique …………….49

3.3.1. The Representation of Activities in LOB after Breaking-down Activities

……………………………………………………………………………………49

3.3.2. Proof of the concept ………………………………………………….…50

3.4. A proposed algorithm (Model) for minimizing number of Crews utilized in LOB

Scheduling Projects and allocating crews among different activities…………... 52

| P a g e VII
 3.4.1. Introduction ………………………………………………………….….52

3.4.2. Process outline of the proposed optimization Model…………….…..… 52

3.4.3. Main attributes and parameter associated with the proposed model…… 52

3.4.4. Description of Optimization Model Formulation ……………………….54

3.4.4.1 Proof of Concept …………………………………………………………54

3.4.5. Optimization Model Framework …………………………………………..54

3.4.6. Description of Crew Diagramming Approach ………………………...…65

3.4.7. Optimization tool………………………………………………………... 66

3.4.8. Model Development…………………………………………………….. 67

3.5 The proposed Simulation Model for resource leveling dilemmas in LOB using

Entropy Maximization ……………………………………………... ……………..69

3.5.1. Introduction ……………………………………………………………..…69

3.5.2. Overview of the Simulation Model……………………………………..… 69

3.5.3. Main attributes and parameter associated with the proposed model……... 70

3.5.4 Description of Simulation Model formulation …………………………….71

3.5.5. An overview of the proposed simulation Model .................................……72

3.5.6. Optimization tool ……………………………………………….…………76

3.5.7. Output module …………………………………………………………….77

3.5.8. Model Development ………………………………………………………77

3.6. Summary and Closure of the Chapter………………………………………… 78

CHAPTER 4 – VERIFICATION AND VALIDATION..................................... 79

4.1. Introduction…………………………………………………………………… 79

4.2. The Developed Strategy for the proposed Approach………………….……… 79

4.3. Description of the Case Study ………………………………………………...81

4.4. Traditional Manner for solving the Case Study……………………………… 82

| P a g e VIII
 4.4.1. CPM calculations for a single unit……………………………………….. 82

4.4.2. Basic LOB Representation using Integrated CPM-LOB Model…………. 82

4.5. The Proposed approach for Scheduling Multi Repetitive Construction Projects.84

4.5.1 Basic LOB Representation using Integrated CPM-LOB Model………..… 85

4.6. The implementation of the developed framework of the Optimization model using

MATLAB on the case study…………………………………………….……….. 89

4.6.1. Step-1: Input module ……………………………………………………..89

4.6.2. Step-2: Defining Initial Inputs…………………………………………… 90

4.6.2.1. Application on MATLAB……………………………………………. 90

4.6.3. Step-3: Forecasting the successor and predecessor activities depending on the

input variables…………………………………………………………………….. 90

4.6.4. Step-4: Sparse Matrix Module ……………………………………………90

4.6.4.1. Application on MATLAB……………………………………….…… 91

4.6.5. Step 5-Graph Theory Module ………………………………………….…92

4.6.5.1. Application on MATLAB ………………………………………….…92

4.6.6. Step 6- Structural Array Module …………………………………………94

4.6.6.1. Application on MATLAB ……………………………………………94

4.6.6.2. Output Crews routing after implementing the optimization model…. 95

4.6.6.3. Output Crews routing after implementing the optimization model... 97

4.6.6.4. Output Crews routing after implementing the optimization model ..99

4.6.7. Step-7: Crew Diagramming Approach (CDA) …………………………101

| P a g e IX
 4.6.7.1 Application of Crew Diagramming Approach on activity C………... 101

4.6.8. The assignment of multi-task skilled workforce in LOB ………………103

4.7. Clearly Identifying Critical Path in LOB schedules……….……………… 106

4.8. Summary and Conclusions of this section………………………………… 108

4.9. The proposed Simulation Model for resource leveling in LOB using Entropy

Maximization Metric with stretching or crunching activity duration…….……. 109

4.9.1. Resource Leveling using Minimum Moment Algorithm……………… 110

4.9.1.1. Application of Resource Leveling using Minimum Moment Algorithm on

Repetitive Activity "C"………………………………………………………… 111

4.9.1.2. Application of proposed optimization model on activity "C" after leveling

with minimum moment algorithm ………………………………………………113

4.9.2. Implementation of proposed simulation model for leveling resources using

entropy maximization on activity "C" …………………………………………..114

4.10. Summary and Conclusions of this section ………………………………….118

CHAPTER 5 – CONCLUSIONS AND RECOMMENDATIONS..................... 119

5.1. Research Overview………………………………………………………… 119

5.2. Conclusion and research contribution ……………………………………...122

5.3 Recommendation for Future Research…………..…………………………….. 123

REFERENCES……………………………..…………………………………..124

APPENDICES…………………………………………..…………………..….129

Appendix A- Calculation Procedures of integrated CPM and LOB………...…..130

A.1. Calculation of desired rate of delivery……………………………...…..131

A.2. Calculation of number of crews required…………………………...…..131

A.3. Calculation of actual production rate…..……………………….…...…..131

A.4. Calculation of each activity duration…...…………………………...…..131

|PageX
 A.5. Specifying logical relationships………………………...…………...…..131

A.6. Performing a time scheduling calculations ………………..…………….132

A.7 Calculation Procedures of integrated CPM and LOB for the case study using

conventional approach ……………………………….………………...……….132

A.8. Calculation Procedures of integrated CPM and LOB for the case study using the

new approach ………………………………………………………….………..133

Appendix B- The optimization model using MATLAB Code……………….... 135

B.1. Defining initial input values for activity C (Start, finish date& Act. No.)…136

B.2. Defining initial input values for activity D (Start, finish date& Act. No.) …137

B.3. Defining initial input values for activity E (Start, finish date& Act. No.) …138

B.4. Defining initial input values for activity F (Start, finish date& Act. No.) …139

B.5. Defining initial input values for activity G4, K and L (Start, finish date& Act.
No.) ……………………………………………………………………………….140

B.6. Defining initial input values for activity E (Start, finish date& Act. No.) after
leveling using minimum moment algorithm……………………………………….141

B.7. Code or algorithm of MATLAB optimization model……………………….. 142

Appendix C- Resource leveling using minimum moment algorithm…………..…145

C.1. Computational procedures of the improvement factors for activity C…..…..146

Appendix D- Resource leveling using entropy maximization metric…………….148

D.1. Assumptions and Forecasts summary reported from the simulation model using

crystal ball ribbon…………………………………………………………………..149

D.2. Defining the project completion date as a forecast…………………………..149

| P a g e XI
D.3. Defining the total system entropy as a forecast………………………………150

D.4. Defining the total number of resources as a forecast……....……….………..151

D.5. Resource assignment per day is defined as an assumption with a discrete uniform

distribution (minimum and maximum value) based on the total float of each

activity…………………………………………………………....……….………..152

| P a g e XII
 LIST OF FIGURES

CHAPTER 1 – INTRODUCTION

Figure 1.1: Thesis Organizational Chart………………………………………..…..11

CHAPTER 2 – LITERATURE REVIEW

Figure 2.1: CPM Representation of repetitive Networks with Multiple-Distributed

Sites (Hakeem, 2001). …………………………………………………...……..…..14

Figure 2.2: Example of Critical path in LOB (Arditi et al, 2002)……………..……19

Figure 2.3: Example of LOB scheduling with learning curve effect (Arditi et al, 2002)

…………………………………………………………………………..………..…20

Figure 2.4: Desired Project rate of delivery (Ammar, 2013). ……………..….……21

Figure 2.5: Duration of a repetitive activity along all units (Ammar, 2013)…….... 22

Figure 2.6: SS Relationship between Activities (Ammar, 2013)………………..…23

Figure 2.7: FF Relationship between Activities. (Ammar, 2013)………….……….23

Figure 2.8: A diagram for the problem solved by Euler using Graph theory (Dickson,

2006). …………………………………..……………………………………..……36

Figure 2.9: The adjacency graph of a sparse matrix (Dompierre, 2010)……….…..37

Figure 2.10: Undirected graphs with non-oriented edges (Dompierre, 2010)………38

CHAPTER 3 – RESEARCH METHODOLOGY AND MODEL

DEVELOPMENT

Figure 3.1: Steps of achieving Research Methodology Framework………………….……..44

Figure 3.2: Traditional LOB scheduling technique ………………………………….……..47

| P a g e XIII
Figure 3.3: Crew Circulation in LOB scheduling technique ………………………………..48

Figure 3.4: Methodology of Breaking-down activities ……………………………………..51

Figure 3.5: LOB representation for the simple example ……………………………….…..51

Figure 3.6: Optimization Model Flowchart…………………………………………………53

Figure 3.7: Case Study to show model efficiency………………………………………..…54

Figure 3.8: Data grouping for the input module……………………………………….....…55

Figure 3.9: A Snap shot for the function used in MATLAB for step 1&2. …………......…56

Figure 3.10: Defining Input Data of the Model. ………………………………..…….....…56

Figure 3.11: Identification of successor activities. ……….………………………..….....…57

Figure 3.12: A Snap shot for the function used in MATLAB for step 3. ……….……....…58

Figure 3.13: A Snap shot for the function used in MATLAB for step 4. ………..…........…60

Figure 3.14: A Sparse Matrix Formulation Diagram. ……………………….…..….....……60

Figure 3.15: A Snap shot for the function used in MATLAB for step 5..…..........………....62

Figure 3.16: Graph Theory Network Diagram..……………………………..........………....62

Figure 3.17: Structural Array Module..…………………………………..........………....64

Figure 3.18: Simple representation of an activity in CDA …………………..........………....65

Figure 3.19: Crew Diagramming Approach………………………………….............……....66

Figure 3.20: A Snap shot for the function used in MATLAB to figure out number of activities……. 67.

Figure 3.21: A Snap shot for the function used in MATLAB to plot the graph theory……..68

| P a g e XIV
Figure 3.22: A Snap shot for the function used in MATLAB to formulate a decision making

criteria. ………………………………………...……………………………..........………....68

Figure 3.23: A Snap shot for the function used in MATLAB to Conduct Output data………68

Figure 3.24: A Flowchart for Simulation using Entropy Maximization. ……………….……71

Figure 3.25 The main functionalities of the activity and resource flow template……72

Figure 3.26: Activity and Resource Flow Excel Template ………………………….….……73

Figure 3.27: Bar Chart Excel Template. …………………………………………….….……75

Figure 3.28: Maximum Entropy analogy Template…………………………………………..76

CHAPTER 4 – VERIFICATION AND VALIDATION

Figure 4.1: Implementation of Proposed Approach based on the case study………..80

Figure 4.2: Traditional CPM calculations for a single 4-K.m Pipeline installation....82

Figure 4.3: Traditional CPM time analysis of the hypothetical proposed case study.83

Figure 4.4: CPM network for a single unit after splitting activities…………………84

Figure 4.5: CPM time analysis of the hypothetical proposed case study after splitting

activities. ……………………………………………………………………………85

Figure 4.6 LOB of activities C1, C2, C3 & C4 before relaxing production rate..……86

Figure 4.7: LOB schedule for activities C1-C2-C3-C4…………………..…………87

Figure 4.8: LOB schedule for activities D1-D2-D3-D4…………………………….87

Figure 4.9: LOB schedule for activities E1-E2-E3-E4. …………………………….88

Figure 4.10: LOB schedule for activities F1-F2-F3-F4. ……………………..…….88

| P a g e XV
Figure 4.11: LOB schedule for activities G1-G2-G3-G4. ………………………….89

Figure 4.12: A screen shot for input variable data on MATLAB. ………………….09

Figure 4.13: A screen shot for sparse matrix application on MATLAB for activity-C…….09

Figure 4.14: A screen shot for sparse matrix application on MATLAB for activity-D……..09

Figure 4.15: A screen shot for sparse matrix application on MATLAB for activity-E……..92

Figure 4.16: A screen shot for graph theory network application on MATLAB for

activity-C………………………………………………….……………………….93

Figure 4.17: A screen shot for graph theory network application on MATLAB for

activity-D. ………………………………………..……….……………………….93

Figure 4.18: A screen shot for graph theory network application on MATLAB for

activity-E………………………………………………….……………………….94

Figure 4.19: Number of crew for activity-C using traditional method. ……..……..95

Figure 4.20: LOB graph after implementation new crews routing using proposed

model output "C………………………………….……….……………………….96

Figure 4.21: Developed Approach Vs. Traditional Approach for Activity "C"…..96

Figure 4.22: Number of crew for activity-D using traditional method…………….97

Figure 4.23: LOB graph after implementation new crews routing using proposed

model output "D"………………………………….……….……………………….98

Figure 4.24: Developed Approach Vs. Traditional Approach for Activity "D"……98

Figure 4.25: Number of crew for activity-E using traditional method……………..99

| P a g e XVI
Figure 4.26: LOB graph after implementation new crews routing using proposed

model output "E"………………………………….……….………………………100

Figure 4.27: Developed Approach vs. Traditional Approach for Activity "E"…....100

Figure 4.28: Developed CDA for first three crews in activity "C"..……………….101

Figure 4.29: Developed CDA for crew 4 & 5 in activity "C'………..…………….102

Figure 4.30: Developed CDA for crew 6, 7, 8 & 9 in activity "C"………………..102

Figure 4.31: A screen shot for graph theory network application on MATLAB for

activity C & D. ………………………………….……….………………………104

Figure 4.32: LOB graph after implementation multi-skilled crews using proposed

model output C&D……………………………….……….………………………105

Figure 4.33: Longest path (critical path of the case study) pinpointed using the

developed algorithm…………..………………….……….………………………106

Figure 4.34: Graph theory network for all possible critical paths in the project…107

Figure 4.35: One of the critical paths presented in LOB format………….………107

Figure 4.36: Sequential procedures performed to test the doable of the proposed

model ………………………………….………………….………………………109

Figure 4.37: LOB diagram for activity C before leveling the resources …………110

Figure 4.38: Resource leveling histogram after utilizing minimum moment algorithm

on activity C………………………………….……..…….………………………112

Figure 4.39: LOB representation of Activity "C" after leveling using minimum

moment algorithm …………………………………..…….………………………112

Figure 4.40: Graph theory network of Activity "C" after leveling using the

optimization model ……………..…………………..…….………………………113

| P a g e XVII
4.41. LOB graph for activity C after implementation new crews routing using

optimization model ………………………………....…….………………………114

4.42. A screen shot for the activity and resource flow template of activity "C"….115

4.43. A screen shot for bar chart template of activity "C"……………...…………116

4.44. A screen shot for simulation outcomes of an activity "C"…………………..117

4.45. Daily resource usage after implementing entropy maximization……….…..117

4.46. LOB graph of activity C after leveling using entropy maximization metric with

different production rates among each activity……………………………….…...118

| P a g e XVIII
 LIST OF TABLES

CHAPTER 2 – LITERATURE REVIEW

Table 2.1- Summary for the previous research efforts developed for scheduling and

managing repetitive Projects. …………………………………………………..…...40

CHAPTER 3 – RESEARCH METHODOLOGY AND MODEL

DEVELOPMENT

Table 3-1 shows the major attributes and parameters associated with proposed

optimization model………………………………………..……………………..…...52

Table 3-2: Major attributes of the Simulation model………………………………..70

CHAPTER 4 – VERIFICATION AND VALIDATION

Table 4.1- Summary for the activities code, name, description, duration and

precedence relations of a single 4-K.m pipeline installation

project………………………………………………………………………………………………..81

Table 4.2. Project related information……………………………………………….89

Table-4.3 Activities enrolled under each crew based on the model output "C"....…..96

Table-4.4 Activities enrolled under each crew based on the model output "D"….…98

Table-4.5 Activities enrolled under each crew based on the model output "E"….…100

Table-4.6 Activities enrolled under each multi-skilled crew for activity C & E……….104

| P a g e XIX
 CHAPTER- 1

INTRODUCTION

This chapter starts with a general overview on the current scheduling

techniques for multi repetitive projects. Then, it highlights the problem statement,

need for the research, scope, objectives and outcomes of the thesis. The chapter also

tackles the use of the line of balance (LOB) in scheduling repetitive construction

project and discusses the resource leveling approaches applied in linear construction

projects.

1.1 Background

It’s crucial, but hard, to separate good process from good outcomes. Often it’s

assumed that any good outcome, must reflect a good managerial process and vice

versa. Implementation of any construction project involves unique environments,

challenges, and project management needs that have to integrate to achieve a

successful outcome. Construction management had undergone a significant ratio of

project failure and overrun over the last years. The underlying reasons beyond the

failure in construction projects are inappropriate methods used in planning

construction projects, mismanagement of resources, and lack of efficient project

management tool (Russell and Radtke, 1991).

There are different examples of projects that succeeded in meeting perception

of success, but failed to meet the project management success. An example of this

projects is the Sydney Opera House that is arguably one of the most well-known and

fascinating buildings in the world in its functional and architectural design. Yet, from

a project management perspective, it was a spectacular failure. The construction of

the project initiated in 1959, it was planned to be completed in 1963 with an

estimated cost of $7 million. However, it was completed in 1973 for over $100

|Page1
million (a cost overrun of 1300% and a schedule overrun of 250%). Another example

is the project constructed when a paper manufacturing company with five plants

across North America decided to increase its manufacturing capacity by embarking

on a de-bottlenecking program. The project estimated completion period was 18

months at a cost of $26 million. The project eventually spanned five years with an

extra $40 million as a result of mismanaging the project (Enos and Rogers, 2002).

Formulation of a project scheduling is not only perceived as a simple matter

of pointing out the sequence and timing of activities within a project, but also it has

to cope with a number of constraints and simultaneous considerations of several

issues. Time, cost and targeted quality are considered the most essential concerns and

considerations that have to be taken into consideration in any construction project to

ensure the feasibility of a construction project process throughout its whole lifecycle.

To maintain this objective, the development of a practical and doable

schedule approach in terms of time, cost, and resource utilization for the project is a

key of concern. Several tools and techniques are used for scheduling of construction

projects, these methods include network diagrams, bar charts, matrix schedules,

check lists, linear scheduling methods and others. Linear scheduling methods are

concerned with the planning and scheduling of repetitive projects, which are

characterized by sequential activities that are repeated for substantial number of

times.

Repetitive projects are regarded as a wide umbrella that includes various

categories of construction projects that consist of a series of repetitive activities. One

main objectives in scheduling repetitive projects is the capability of ensuring the

continuation of work without any interruptions. Interruptions or idle time is the

duration where a resource completes a task and being hold up without performing a

|Page2
new task till the complete of the predecessor task .Accordingly, repetitive activities

should be scheduled in a way to omit or reduce interruption time. Linear projects are

sub-divided into two types, which are horizontal projects as highways tunnels and

pipeline networks, and vertical ones such as high rise buildings.

A survey was conducted by Russell and Radtke (1991) to identify the factors

that contribute to failure of construction projects. These factors include underbidding,

insufficient cash flow, external difficulties, and lack of experience in estimating and

monitoring costs. These factors indicate lack of proper project management, which is

in part due to the drawbacks associated with critical path method (CPM). One of the

major drawbacks of using the CPM technique for repetitive projects is the complexity

to represent and to visualize different relationships between repetitive activities.

Moreover, the CPM network technique does not address two key aspects, which are

maintaining crew work continuity, and achieving a rate of progress to meet a desired

deadline. Maintaining the crew work continuity in linear projects is considered a

main objective to permit the crews to move from one site to the next without any

interruption (Lumsden 1968 and Laramee 1983). CPM is inappropriate for repetitive

scheduling as it rules out the requirement of creating work continuity to obtain

maximum crew utilization (Selinger, 1980). In addition, CPM is a time based

scheduling approach not a resource based one, as resources availability is assumed to

be unlimited during scheduling any construction project (Lutz and Halpin 1994).

Many attempts have been proceeded to develop scheduling methods and

tools for planning and scheduling linear repetitive construction projects. The most

common methods are as follows.

1. Line of Balance (LOB).

2. Vertical Production Method (VPM)

|Page3
3. Linear Scheduling Method (LSM)

4. Repetitive Scheduling Method (RSM).

This research will focus on and tackle one of the abovementioned linear

scheduling techniques, which is LOB. The drawbacks of LOB in scheduling

repetitive projects will be highlighted. A framework adopts into LOB scheduling

technique to enhance its efficiency as a linear scheduling tool.

1.2 Line of Balance

The LOB scheduling technique is deliberated as an effective scheduling

technique for planning and scheduling linear projects. It was originally derived from

the manufacturing industry and was developed by the U.S. Navy department for the

programming and controlling of repetitive projects in the early 1950s (Neale and

Neale 1989). Afterwards, it was later developed by the National Building Agency in

the United Kingdom for repetitive housing projects to show that LOB is a resource-

oriented scheduling methodology and to announce that a schedule approach that was

based on resource utilization from the scratch is more realistic than activity

dominated scheduling (Trimble, 1984).

LOB is a display of the project's profile and its situation by representing the

cumulative completions of activities associated with a level of planned units' number

at a given point in time. LOB shows the delays in the schedule that requires a

deviation from the planned schedule due to the actual unforeseen progress of

activities in a graphical way, and enables the assessment of this deviation (Khisty

1970). Repetitive projects are composed of a series of repetitive activities that need

crews working and moving from one unit to another. These crews are required to

repeat the same work in different areas in the project. Line of Balance (LOB) ensures

the ability of maintaining resources working continuously without any interruption.

|Page4
Consequently, LOB in scheduling repetitive activities is a way that reduces

interruption and idle time of resources, and incorporates resource constraints into the

schedule.

Many attempts have been established to combine benefits of CPM as a

duration driven scheduling method and benefits of LOB as a resource driven

scheduling technique in scheduling linear repetitive construction projects, where both

logic dependency and resource continuity constraints are taken into account during

scheduling process. (Suhail and Neale, 1994, Hegazy and Wassef, 2001, Ammar and

Mohieldin, 2002, Ammar, 2003, Ammar, 2012, and Jung et al, 2013).

This thesis is aiming to introduce a new schedule framework for planning and

scheduling multi repetitive projects where LOB scheduling tool will be utilized. First

and foremost, it discusses the key aspects of LOB technique in scheduling repetitive

projects, and highlights different concepts imbedded into LOB to improve its

performance in scheduling linear construction projects. The thesis also provides a

new ideology for maintaining crew circulation to optimize its number, and tackles the

dilemma of resource leveling in repetitive construction projects. Finally, an approach

to level resources utilization is developed in a manner to minimize its daily usage and

smoothing its fluctuation with different production rates.

1.3 Statement of the problem

Despite the long history and expanding use of LOB in planning and

scheduling linear projects, the literature indicates that LOB has a number of

limitations that raise concerns about its use in the construction industry. The reasons

for the limitation of LOB in scheduling linear projects are described as follows.

|Page5
1.3.1 Limitation in LOB

1. In LOB scheduling technique, a single unit project is scheduled with its activities

and those activities are repeated gradually with the number of desired repetitive

units. This may result in rendering the process with same production rates and all

activities could become critical with no float, which is required for relaxing the

production rate in case of resource limitation.

2. A limitation of this methodology is that it assumes that production rates are

linear and constant. To account for practical assumption, the amount of work

within a repetitive activity along the entire units is identical and productivity of

resources is constant. Due to the stochastic nature of construction processes, the

assumption that production rates of construction project activities and tasks are

linear may be erroneous and unrealistic (O'Brien, 1985; Russell and Wong

,1993; Yang, 2002).

3. Despite being an essential tool for project scheduling, LOB technique is

inadequate in identifying criticalness and floats; thus, LOB method has to be

developed to single out critical, non-critical repetitive activities, and floats to

precisely detect the critical activities and path in LOB (Arditi et al, 2002).

4. From a resource utilization context, the assumption of linear, constant

production rates hinder LOB in restricting each activity to be attained by the

same crew in a linear sequential manner and eliminating the likehood of

allocating resources among entire project units (Yang, 2002).

5. Construction projects involve circumstances that mandate the use of visual color

coded diagrams or graphs for crews various circulations to facilitate pointing out

each crew cycle, mapping out it's overlapping between different activities and to

easily figure out concurrent activities at same periods (Arditi & Albulak, 1986).

|Page6
1.3.2 Leveling of Resources in LOB

Resource planning and management is a key aspect for successfulness and

profitability of any construction project (Karaa and Nasr 1986). Resource limitations

should be taken into account, otherwise schedules will be conducted in an erroneous

way (Hinze, 2004). Resource leveling in linear scheduling projects continues to grasp

the motivation of researchers who expand its capabilities and applications. Linear

scheduling methods rely mainly on resource allocation but do not implement any

leveling on resources. Some researchers have been analyzing the problem of single-

resource leveling in linear schedules (Georgy 2008; Lucko 2011), but known of the

above-mentioned researches have taken into consideration the following parameters:

1. Among the well-documented factors that contribute to the difficulty of providing

more improvement in resource usage, is the absence of different allocation or

movement of crews between different activities in different units that is not

previously considered during resource leveling in linear scheduling.

2. In addition, previous research do not consider the stretching or crunching of an

activity duration in LOB to reach different and optimum crew sizes, they only

considered activities with constant durations and crew sizes.

3. In this thesis, resource leveling in LOB scheduling technique is seamlessly

extended to be implemented by applying entropy maximization metric using a

simulation model.

In this thesis, a developed framework will be adopted into LOB scheduling

technique, where the single unit project will be broken down into sub-activities with

different floats before performing LOB calculation, which assist in relaxing sub-

activities production rates. Afterwards, LOB deals with this sub-activities as a multi

repetitive project, where each repetitive activity will be split down into equal

|Page7
repetitive sub-activities with different production rates, and consequently different

number of crews are required. This method will assist LOB scheduling technique to

have the tendency to incorporate a new concept for crew allocation from one activity

to another in different units not only in a vertical linear direction, but also in all

directions (upwards, downward, left and right) with no interruption between crews,

which aims to find the optimum use of crews' number within repetitive activity. In

addition, the visualization of crew circulation is enhanced and is presented using a

color graphics for each crew route to easily identify crew number and its movement;

besides, each crew movement is presented in a graph like a time scaled bar chart with

its specified color to facilitate the process of pointing out the crew circulation within

different units and to observe its starting and finishing dates.

Moreover, in this thesis, a new manner for single resource leveling of

repetitive activities is implemented, where the number of crews required to perform a

repetitive activity is the main purpose; besides, the possibility of ensuring different

resource allocation during leveling between different activities in different units.

Moreover, this research will take into account the likelihood of stretching or

crunching duration of repetitive activities, as an approach to have an extensive

enhancement in the resource histogram profile based on several conditions.

1.4 Objectives and Scope

The main challenges of this research stem from the desire to create an

innovative framework and scheduling approach that overcome shortfalls of LOB in

scheduling linear projects. This schedule approach will better handle schedule and

resource constraints like project deadline, resource limit, minimizing indirect cost;

besides it will adequately identify the critical path, which is difficult to be forecasted

in LOB. The following are the major objectives of this research:

|Page8
 Identifying the practical areas of potential enhancement that can improve the LOB

scheduling tool by accounting for the breaking down of activities into sub-

activities in order to formulate more efficient linear schedules

 Supplementing the repetitive scheduling literature with a new optimization model

and an algorithm that can provide different and optimal crews routing between

different activities in different units and serve as a decision support system for

crew's circulation, and guarantee the optimal crew distribution for cost efficiency.

 Minimizing indirect costs by decreasing maximization in daily resource usage and

allowing for using multi-task skilled crews between different activities in different

units in repetitive projects; in addition to the new representation of different crews

paths using Crews Diagramming Approach.

 Conducting a new resource leveling technique based on a simulation model that

can capture the resource leveling dilemma by permitting for activity duration

stretching and crunching, in which activity durations can be differentiated along

the repeated units to ensure enhancing and smoothing resource fluctuations over

time.

1.5 Research Methodology

The following steps are undertaken to achieve the research objectives:

Step-1. An extensive review of LOB drawbacks in scheduling linear

repetitive projects is carried out in order to examine the existing LOB procedures and

to identify its limitations.

Step-2. An optimization model is conducted by creating an algorithm on

MATLAB using Sparse Matrix, Graph theory, and structural array. This model is

considered a decision making tool to opt most optimum path and formulation for

each crew and to allocate crews within activities in different units.

|Page9
 Step-3. A Crew Diagramming approach is rendered as a layout to map crews

various routing from one site to another, and to figure out overlapping of crew

circulation between different activities in various units.

Step-4. A spreadsheet simulation model is performed where the entropy-

maximization metric will be used for leveling resources which allows for activity

stretching and crunching to provide different resource allocation solutions and to

achieve better resource leveling profile using Crystal Ball Ribbon software.

Step-5. Validation: A Case Study is used in order to validate the

abovementioned and demonstrate its functionality and usefulness in scheduling

multiple repetitive projects.

The case study shows the development of a new representation of crews' movement

and routing while maintaining resource continuity and availability.

The case study shows the ability of LOB to accurately define the critical path.

The case study overviews an obvious enhancement and smoothness in resource usage

profile.

1.6 Layout of the Thesis.

This thesis consists of five chapters. Chapter 1 – Introduction that introduces

the research study and gives a generic overview of the thesis problem statement,

research scope and objectives. Furthermore, it discusses the research methodology

applied to reach the main objective of this thesis.

Chapter 2 – Literature Review tackles an extensive review for the previous

researches that cover different concepts and challenges of line of balance (LOB) in

scheduling repetitive projects; besides it introduces different approaches used for

| P a g e 11
resource leveling in linear repetitive construction projects, entropy theory and graph

theory.

Chapter 3 – Research Methodology and model development demonstrates the

methodology and the interference between the research themes.

Chapter 4 – Verification and Validation implements a hypothetical case study

for a repetitive project of a sewage pipe line installation in order to illustrate the

proposed framework. In addition, it tackles and analyzes the results obtained from the

applied case study.

Chapter 5 – Conclusions and Recommendations highlights the summary,

conclusions, limitations, and recommendations of this research.

Figure 1.1: Thesis Organizational Chart

| P a g e 11
 CHAPTER -2

LITERATURE REVIEW

One of the main purposes of this research is to support and to guide planners

in precisely planning, scheduling and monitoring construction projects that are

characterized by sequential repetitive activities. Hence, this literature review

evaluates and discusses the pertinent theory and previous research conducted in the

fields of LOB scheduling. Afterwards, it focuses broadly on overviewing previous

research endeavors associated with enhancing LOB scheduling technique and its

resource leveling. Finally, the chapter previews various efforts done in the field of

implementing entropy maximization and graph theory in construction industry.

In this chapter, several topics will be introduced and discussed in details. The

key topics, in which the chapter will focus on, could be divided into six main

sections as follows:

1. Basic concept of LOB in repetitive projects.

2. Challenges and enhancement to improve LOB scheduling technique.

3. CPM and LOB integration Methodology.

4. Leveling in linear repetitive schedule Projects

5. Entropy Maximization Theory

6. Graph Theory

| P a g e 12
2.1 Line-of-Balance

This section demonstrates the Line of balance technique in depth, where it

firstly begins with a historical review about Line of Balance. Then, it tackles the

procedures undertaken to perform a Line of Balance calculations. Thenceforward, it

deliberates the implementation of LOB in construction industry showing different

challenges and enhancements proceeded to improve LOB technique. Finally, a

generic summary about LOB and its relationship with this study will be emphasized.

2.1.1 Historical Review about LOB

The LOB methodology was originally introduced by the members of a group

headed by George E. Fouch. During 1941, the Goodyear Tire & Rubber Company

monitored production rate using LOB technique. LOB was developed by the US

Navy in the early 1950s, as it was applied to the production planning and scheduling

of the huge Navy mobilization program of World War ll. LOB has been implemented

to industrial manufacturing and production adjustment to evaluate a production line

flow rate of completed products and has been expanding across a whole spectrum of

activities ranging from research and development through job shop and process flow

operations. The line-of-balance technique is based on an assumption that the rate of

production for an activity is uniform, where time is plotted on the horizontal axis and

units of an activity is plotted on the vertical axis. The production rate of an activity is

the slope of the production line (Johnston 1981).

Research have showed that LOB scheduling technique is more precise than

CPM in scheduling linear construction projects. CPM has been criticized extensively

in the literature for their inability to schedule repetitive projects (Selinger, 1980;

Reda, 1990; and Russell and Wong, 1993). The size of the network of a CPM

schedule for a repetitive project is relatively huge which lead to miscommunication

| P a g e 13
 among construction management team. Furthermore, CPM was designed for

minimizing project duration rather than concerning with resource constraints of

repetitive projects. LOB scheduling technique was claimed to have the tendency to

conduct a crew circulation from one unit to another with less interruption or idle time

for workers and equipment (Arditi et al, 2002)

Figure 2.1: CPM Representation of repetitive Networks with Multiple-Distributed Sites (Hakeem, 2001).

2.1.2 Basic Procedures undertaken to perform LOB.

LOB scheduling technique depends on a required number of units that have to

be completed and delivered in a specific period of time. The production rate of each

activity, and its duration relies mainly on the targeted rate of delivery and are

expected not to be less than this target rate (Lumsden, 1968). The rate of outcome

that a crew of optimum size will be able to generate is called the ‘‘natural rhythm’’ of

the activity. Any rate of outcome that varies from a multiple of the natural rhythm

will result in the existence of an idle time for resources. Psarros (1987) has developed

an algorithm to conduct the number of needed crews by an activity such that the rate

of outcome, a multiple of the natural rhythm is as close to the target rate of delivery

of the project as much as possible (Psarros, 1987).

| P a g e 14
 The number of crews required to perform a repetitive activity and the actual

rate of output are calculated to plot the LOB diagram, where the number of repetitive

units are drawn against time. Two oblique or parallel lines, whose slope is equal to

the actual rate of outcome represents the start and finish date of each activity in all

unit from the first one to the last one, as it is calculated as follows (Lumsden,

1968).

M= Qj - Qi / Tj – Ti , I < J…………………………………………………….(1).

Where M is the rate of production (unit of production per unit of time); Qi, Qj

equal number of units started; Ti, Tj equal time elapsed between the start of the

project and the start of I and J units. The slope of line of balance connecting finish

time of repetitive activities in each unit is equal to m. If the duration of the activity is

known and the actual rate of outcome is limited to a multiple of natural rhythm, then

the equation will be (Lumsden, 1968).

Q = P/d……………………………………………………………………….. (2).

Where P is the number of crews of an activity and d is the duration of an

activity in one unit. The starting time of an activity is calculated as follows:
Ti= T1 + (1/m) * (Qi-1)……………………………………………………… (3).

Where ti is the starting time of an activity in i- th unit; t1 is the starting time of an

activity in the first unit; m is the rate of production and Qi is the number of units
produced (Lumsden, 1968).

2.1.3 Challenges and enhancements to improve LOB for Construction Industry

LOB scheduling technique have been implemented in the construction

industry as a planning and scheduling technique in Finland in the 1980s (Harris and

McCaffer, 1989), in which it is considered a scheduling methods that permits the

balancing of construction projects and ensures activities to be performed

continuously. The major benefit of the LOB techniques is providing production rate

| P a g e 15
and duration of each repetitive activity in an easily graphical format. The LOB graph

also allows the detecting of delays with the progress of an activity and permits the

likelihood of production rate of different repetitive activities to deviate. Moreover,

the LOB lets on the movement of resources and its flow in a smooth and efficient

manner with less time consumed (Arditi and Albulak 1986).

LOB permits the adjustment of the production rate of activities to maximize

resource utilization by eliminating idle time. This process of monitoring production

rate is known as “balancing production rates.” LOB provides means of selecting crew

size in order to minimize inefficiency and waste in resource usage. Furthermore,

balancing production rates keeps all activities working at the same pace which may

minimize project duration (Lutz 1990).

An early attempt was developed to schedule repetitive construction projects

using LOB techniques with a computer application called “System for Repetitive

Unit Scheduling” (SYRUS). It was developed to assist planners in scheduling linear

projects in the construction industry. It is a menu driven application that relies on

mixing both network and LOB techniques. This application failed to deal with many

dilemmas encountered during scheduling projects (Arditi and Psarros 1987).

LOB computer applications were developed to enhance inefficiencies of the

LOB technique and to improve its application. Hegazi et al (1993) proposed a

computer program called BAL to schedule and to control linear projects with uniform

sequential activities. This program is regarded as a user-friendly software that has the

ability to generate schedules based on calendar and working days, and to come up

with the desired rate based on the specified deadline. Moreover, this program

previews possible updates for meeting deadlines and presents the resource usage

throughout the project life cycle (Hegazi et al, 1993).

| P a g e 16
 Senouci and El-din (1996) developed a non-serial dynamic programming

technique for the scheduling linear projects with non-serial sequential activities. This

method applies time/cost trade-off for linear projects, and manipulates the project

durations and its costs efficiently. Furthermore, this technique overviews different

crew formations, production rates and lag durations (Senouci and El-din, 1996).

Wang and Huang (1998) introduced a new scheduling approach that tackles

the disadvantages of LOB to control the interval times. This method started with

applying multistage linear scheduling method (MLS) based on the multistage

decision making concept. The MLS overcomes the dilemma of LOB to monitor the

interval times between activities in repetitive projects. In this approach, interval times

are regarded as a function of the total number of repeated units, difference in time

between the construction of adjacent activities and the order of activities. This

method reduces the total project duration without reducing the duration of each

activity at each unit (Wang and Huang, 1998).

Arditi et al (2002) proposed a computerized system for the implementation of

LOB to imitate the concepts of a multi-level LOB diagram and to generate flexible

unit network for scheduling high-rise buildings. Although RUSS and ALISS showed

enhancement in satisfying deadline constraints in resources, they do have some

shortcoming, as they do not guarantee the optimal crew distribution. Moreover, they

neglect the distribution of production rates in the decision-making process, and the

opting of consecutive activities with same production rates is not allowed because the

algorithms only choose one activity in each iteration (Arditi et al, 2002).

Tokdemir et al (2003) maintained a computerized system of LOB called

ALISS, which have the capability of accelerating the project for a specified deadline

| P a g e 17
or milestone by increasing the number of crews of selected activities (Tokdemir et al,

2003).

Agrama (2011) introduced a spreadsheet algorithm, where LOB based

scheduling technique was used. This novel model identifies start and finish times in

an easy analytical way. It maintained crew work continuity without allowing any

interruption, and achieving logical constraints and relationships between consecutive

activities. In this study, non-typical activities were assumed, in which activity

durations can be varied along the repeated units. The model was presented in three

steps: first was the spreadsheet data; second was the model calculations; whereas

third was LOB diagram (Agrama, 2011).

2.1.4 Concept of ‘Criticalness in LOB

There is a huge difference between the concept of criticalness in LOB and the

concept of criticalness in CPM network. This difference was proved by identifying

the controlling activity path in case of using repetitive scheduling method (Harmelink

and Rowings, 1998). LOB technique do not have the ability to identify criticalness

and floats, so LOB method should be developed to single out critical, non-critical

activities, and floats to accurately detect the criticalness in LOB. LOB is well known

by having different production rates for activities, so the critical activities identified

after LOB calculation may or may not coincide with the critical activities identified

after analyzing the CPM network. The critical path in the LOB analysis may become

noncritical if the production rate of an individual activity is changed by changing the

number of crews required by an activity due to resource constraints (Arditi et al,

2002).

| P a g e 18
 Arditi, et al. (2001) claimed that at least one critical path occurs while using

CPM, and activities which are located on this critical path should be started and

finished with their assigned dates in order not postpone the project completion date.

On the other hand, the LOB scheduling methods criticality was relied on time and

resources, unlike CPM which was only relied on time. As a result of the different

rates of production of individual activities, critical activities in the CPM network may

or may not agree with the critical activities in the LOB schedule as shown in figure

2.2.

Figure 2.2: Example of critical path in LOB (Arditi et al, 2002).

2.1.5 Learning Curve Effect on LOB.

In LOB the relationship between time and the number of units produced is

assumed linear with a constant rate of production over time, which opposes the real

circumstances as the more times an operation is performed, the shorter will be the

time required to perform it. This phenomenon is called the learning curve effect. The

effect of the learning is incorporated into a repetitive project schedule to reflect the

real circumstances of the project (Arditi et al. 1999).

The learning trend cannot be directly implemented in the LOB method

because the LOB requires the consistency in the rate of production of each activity

during repetitive unit construction. Therefore, the learning rate of each activity
| P a g e 19
should be done and then converted into man-hour estimates. The activity durations in

each unit has to be calculated individually because the rate of production of each

activity will not be constant. The resulting production curves will plotted be in a LOB

format which represent the start and finish times of each activity in a non-parallel

lines as shown in figure 2.3 (Arditi et al. 2002).

Figure 2.3: Example of LOB scheduling with learning curve effect (Arditi et al, 2002).

2.1.6 CPM and Line of Balance integration Methodology.

LOB is regarded as a resource controlling tool that conduct a balanced

resource utilization in an efficient way. The major benefit of LOB scheduling

technique is the visualization of production rate and duration of the repetitive

activities in an ease graphical format with the applicability of controlling the progress

rate to meet project duration deadlines, with preserving work continuity of resources

(Yang and Ioannou 2004).

Many efforts have been attempted to integrate the benefits of CPM and LOB

techniques as a new generation of planning and scheduling structures. The CPM/

LOB appears to be more convenient because it has advantages of a graphical basis

and a mathematical one, where (1) resource allocation is done due having a deadline,

(2) the determination of a critical path is easy like CPM, (3) Its application results in
| P a g e 21
smoothing the fluctuation of resources usage, (4) it guarantees work continuity in

linear projects, (5) finally CPM/LOB allows more than one crew to be assigned to an

activity concurrently (Suhail and Neale ,1994).

2.1.6.1 Calculation procedures of integrated CPM and LOB

The process of integrating CPM with LOB is composed of four steps which

are as Follows:

Step-one is LOB calculations which initiate with formulating a strategy for

meeting a project deadline through a desired rate of delivery.

Rd= N-1/Tp-T1+Tf, Where N= number of repetitive units, Tp is the desired

Completion date of the whole project, T1 is the completion date of one single unit,

and Tf is the total float of noncritical activities to reduce the number of crews

employed and to relax production rate of an activity (Suhail and Neale 1994).

Figure 2.4: Desired Project rate of delivery (Ammar, 2012).

Then, the number of crews (Cdi) needed to maintain the desired rate of

delivery of activity i can be calculated as follows:

 Cdi= di * Rdi, where di is the duration of an activity in one single unit

 Actual number of crews to be used (Cai) will be Cdi rounded up.

 Rai= Cai / di .Where Cai is the actual number of crew needed to perform a

specific repetitive activity, and Rai is the actual rate of progress of an activity

| P a g e 21
 Step -two is calculating the duration from the start of an activity in first unit

till the finish of the same activity in last unit , in which STiN = start time of last unit;

STi1 = start time of first unit; and Di= duration along all units of activity i. Di =di +

STiN − STi1 = di+ (N−1) / Ra.

Figure 2.5: Duration of a repetitive activity along all units (Ammar, 2012).

Step-three is specifying logical relationships using overlapping activities by

comparing the actual progress rate of predecessor and successor activities. If the

actual progress rate of Predecessor is greater than actual progress rate of successor,

then a start to start relationship will be applied plus buffer time. If the actual progress

rate of Predecessor is less than actual progress rate of successor, then a finish to

finish relationship will be applied plus buffer time. Finally, if the actual progress rate

of Predecessor is equal to actual progress rate of successor, then a start to start

relationship will be applied plus buffer time (Ammar, 2012).

| P a g e 22
 The fourth step is performing a time scheduling calculations using CPM.

Forward pass calculations are conducted to determine early timings of activities,

whereas late timings of activities are determined in the backward pass calculations.

Figure 2.6: SS Relationship between Activities (Ammar, 2012). Figure 2.7: FF Relationship between Activities. (Ammar, 2012).

2.1.6.2 Previous research effort of integrated CPM and LOB

Suhail and Neale (1994) developed a model that determine the number of

crews needed to meet a project duration deadline. The model presents a formulation

for determining crews required to meet the project due date. Activities’ total float are

utilized to relax non-critical activities without influencing the total project duration.

However, the model works well only when the calculated number of crews is not

rounded to integer numbers and the crews’ availability is not considered.

Hegazy and Wassef (2001) developed a model to minimize total construction

cost (direct cost, indirect cost, interruption cost, incentives and liquidated damages)

using integrating LOB and CPM method. The model uses genetic algorithms to

outcome the optimum construction methods, number of crews, and interruptions for

each repetitive activity. Nevertheless, the model performs time-cost trade-off analysis

| P a g e 23
considering a project deadline. One its defects is limitation of the number of

predecessors and successors for each activity to three only.

Ammar and Mohieldin (2002) developed a CPM-based repetitive scheduling

model that utilizes the benefits of CPM to schedule repetitive activities in an easy non

graphical approach. This model accounts for only the most significant resource for

each activity.

Ammar (2003) developed a model for determining different types of floats for

non‐serial repetitive activities, in which the time float and rate float are extended to

repetitive activities. Float analysis is performed in a manner similar to CPM analysis,

without the need for graphical presentation. The analysis was based on a repetitive

scheduling method, which utilizes CPM network of a typical unit, and overlapping

between activities are used to model repetitive activities. A constant activity

production rate is assumed and resource continuity is maintained.

Ammar (2012) proposed an integrated CPM and LOB model to schedule

repetitive projects in an easy non graphical way, considering both logic dependency

and resource continuity constraints. Overlapping activities of a single typical unit are

used to indicate duration and logical relationships of repetitive activities. The

proposed model consists of four steps. In the first step, LOB calculations are

performed to ensure resource continuity. In the second step, activity duration of

repetitive units is calculated. In the third step, overlapping activities are used to show

logical relationships between repetitive activities. Finally, CPM is performed to

specify activities’ starting dates, finishing dates and floats.

Jung etal (2013) proposed a study that examined the simplified CPM/LOB

methodology in construction scheduling planning and controlling to achieve accurate

schedules. This study improves scheduling calculations of CPM/LOB to simplify

| P a g e 24
existing scheduling method, straightforward design of the discharge procedure and

enhances its applicability through CPM/LOB to provide CPM scheduling repetitive

for the use of project resources (Jung etal, 2013).

Reza etal (2014) presented a methodology to enhance CPM/LOB scheduling

method for deadline constraint satisfaction. A heuristic approach called “heuristic

line of balance (HLOB) is introduced for scheduling projects with serial activities,

which is composed of four heuristic rules and their combination in addition to

seeking CPM/LOB initial calculation. Furthermore, the search-based heuristic line of

balance (SHLOB) algorithm is presented in case extending the project size.

Eventually, a LOB model in the form of integer nonlinear programming is

implemented for verify proposed models (Reza etal, 2014).

2.2 Resource Management in repetitive projects.

2.2.1 Introduction.

Management of resources is considered a crucial key that should be taken into

account for competitiveness and profitability in developing reliable and accurate

construction schedules (Karaa and Nasr, 1986). One of the main reasons beyond the

delay in construction projects is not well considering resource management during

planning phase; besides, the resource availability, resource allocation and resource

fluctuation are not taken into consideration that increase the probability of delaying

the completion date of the project. Thus, resource limitations should not be ignored,

in order not to generate unrealistic schedule (Hinze, 2004).

There are two categories of project scheduling, resource-constrained and

time-constrained. Resource constraints in repetitive construction projects are

composed of two types, resource availability constraints and resource continuity

constraints. Resource availability constraints represent the limitations in the number

| P a g e 25
of crew available for performing a specific repetitive activities. Resource continuity

constraints stipulate the requirement of crew's movement and circulation in a

consistent way without being interrupted. On the other hand, time-constraint

resembles the completion of the project in the specified and desired duration.

Resource leveling is regarded as the one of the approaches used for the

management of resources. Resource leveling in linear repetitive construction is

required to maintain the encountered constraints during the project construction

lifecycle. Resource leveling deals with minimizing fluctuations in resource utilization

depending on the total float of non -critical activities without changing project

duration (Son and Skibniewski 1999; Leu et al. 2000; Hegazy and Ersahin 2001;

Senouci and Adeli 2001; Doulabi et al. 2011; Hariga and El-Sayegh 2011).

2.2.2 Previous research in the field of resource management of linear projects.

For linear construction projects, it has been known that resource utilization is

important in improving work efficiency. However, few existing scheduling

techniques can satisfy needs for solving such issues. Hence, this section will show

several attempts that have been developed to tackle this dilemma.

Senouci and Naji (2006) implemented a computerized system for scheduling

non- serial linear projects and optimizing total cost of the project using a genetic

algorithm model. This model minimized the project total cost by comprising direct

cost, indirect cost, and interruption cost. This system determined the optimum crew

formations and reduced interruption days to decrease overall project cost (Senouci

and Naji, 2006).

Hyari and El rayes (2006) introduced a computation model that include three

computational modules which are scheduling, optimization, and ranking modules. (1)

The scheduling module used a resource-driven scheduling algorithm to conduct

| P a g e 26
practical schedules for repetitive construction projects, (2) The optimization module

used a multi objective genetic algorithm to generate doable construction plans and to

establish optimal tradeoffs between project duration and crew work continuity. (3)

Finally the ranking module used a multi attribute utility theory to rank the produced

plans to ease the selection and execution of the best plan for the project (Hyari and El

rayes, 2006).

Liu and Wang (2007) established an optimization novel model for resolving

linear scheduling dilemmas involving different objectives and resource assignment

tasks. The proposed model adopted constraint programming (CP) as the searching

algorithm for model formulation to create the flexibility for optimizing total cost and

project duration. Additionally, the concept of outsourcing resources is used to

enhance project performance (Liu and wang.2007).

In light of resource-constrained problem in repetitive projects, Hsie etal

(2009) implemented a scheduling model to find the optimal set of production rates

for crews in different time periods to remedy the issue of having the same production

rate along the entire periods. This model took into consideration limited availability

of resources and addressed work continuity while maintaining lead-time and lead-

distance between operations. In this model, the optimization problem was solved

using an evolutionary strategy algorithm (Hsie etal, 2009).

Gafy (2011) proposed an Ant Colony Optimization model based on the

foraging behavior of ants to allocate resources in repetitive construction schedules

that are constrained by an activity precedence and multiple resource limitations. The

algorithm is used to optimally assign resources to repetitive activities to minimize

interruption days (Gafy, 2011).

| P a g e 27
 A proposed optimization model was maintained to integrate single/multiple-

skilled crews to enhance work performance. Constraint programming (CP) was used

to improve the efficiency of problem solving by handling sophisticated combinatorial

scheduling issues and by engaging several heuristic rules. The CP-based optimization

model minimized project duration considering the usage of both single-skilled and

multi-skilled crews (Liu and Wang, 2012).

2.2.3 Leveling in Linear repetitive schedule Projects

This section tackles the resource leveling in linear repetitive schedules in

depth, in which it discusses notable effort and different approaches done for leveling

linear projects, which are subdivided into three approaches: Heuristic approaches,

Optimization approaches and Metaheuristic approaches.

2.2.4 Heuristic approaches used for leveling Resources.

A proposed model was developed to tackle the important activities for

multiresource leveling in linear construction project schedules. The model generates

a linear project schedule and applies multiresource leveling that uses a resource

weight and causes a deviation in daily resource utilization, and conducts a combined

resource leveling histogram. In spite of the main contribution of this model in

leveling resources, it was based on critical path method (CPM) calculations as it uses

free floats instead of using the production rates of the activities, and uses a modified

version of the minimum moment algorithm that does not produce better leveling than

the original algorithm developed by Harris in 1978 (Dubey, 1993).

Minimum Moment algorithm is regarded as one of the most crucial

techniques used for leveling resources, which was developed by Harris (1978). In

Minimum Moment algorithm, activities are assumed to be uninterrupted, and once an

activity starts, it continues till it is completed. Moreover, the technique assumes that
| P a g e 28
the resource utilization is constant over the activity duration and only one activity can

be move at a single time. In this leveling technique, the non-critical activities are only

leveled by shifting respecting their total float durations. The main concept behind this

algorithm is concentrated on an improvement factor which is considered the only

discussion maker to moving any activity with its float. The improvement factor is

calculated for all activities, and the activity with the largest improvement factor has

the highest priority to be shifted within its float.

A multi heuristic hybrid model was introduced that combines the local

optimizer with simulated annealing to remedy resource leveling problems. This

model allows the probability of splitting activities to make it near optimum to real

construction projects. The results shows that these heuristic models generate viable

remedies for the resource leveling dilemmas in repetitive construction projects (Son

and Skibniewski, 1999).

A developed model modified the two-stage procedure for single resource

leveling which was developed by Liu (1999) to solve the multiresource leveling

problem. The first stage deals with resource allocation and scheduling using a

heuristic algorithm. The second stage uses the results of the first stage for

multiresource leveling through simulated annealing. This model has the advantage of

handling multiresource leveling, as it runs the multiresource leveling procedure only

if different resource allocation alternatives generate the same project duration.

Moreover, this model reveals that the resource leveling may not provide any

improvement in resource usage, because it assumes a linear relationship between

crew size and productivity, whereas this relationship is actually not linear due to the

principle of optimum crew size (Yen’s ,2005)

| P a g e 29
 The resource leveling problem of linear repetitive projects was described as a

constraint satisfaction problem where a constraint programming (CP) technique was

used by Tang etal (2014) to remedy the resource leveling dilemma. A two-stage

scheduling system for resource leveling of linear construction projects was created,

which uses the rate float of the activity to obtain a more optimal schedule (Tang etal,

2014).

2.2.5 Optimization approaches used for leveling Resources.

Mattila and Abraham (1998) implemented an integer linear programming

model by assigning weights while combining the multiple resources for multiresource

leveling. This model relies on the linear scheduling model that was developed by

Harmelink and Rowings (1998). The model provides an algorithm to determine the

controlling activities of a project (same as critical activities in CPM), non-controlling

activities (same as noncritical activities in CPM), rate floats, and activity floats. The

model has the ability to find an optimum solution using integer linear programming,

however it may be inefficient on large-scale projects with many variables as it will be

hard to find the optimal solution. This model do not calculate the start/finish times of

the activities and generate the linear schedule diagram and resource histograms

automatically. In this model, the linear schedule diagram and resource histograms

have to be manually generated by the scheduler before and after multiresource

leveling is performed (Mattila and Abraham, 1998).

An automated optimization model was implemented which uses dynamic

programming formulation and incorporates a scheduling algorithm and an

interruption algorithm to follow the generation of interruptions during scheduling.

This transforms the consideration of interruption in optimizing resource utilization,

from an unbounded dilemma to a doable one (El-Rayes etal, 2001)

| P a g e 31
 Finally, Hariga and El-Sayegh (2011) developed an optimization model for

resource leveling that permits activity splitting and minimizes its associated costs.

The objective of the model is to level resources to provide a tradeoff between the

extra cost of acquiring and releasing resources versus the extra cost of activity

splitting. The model can be used to determine the level where splitting an activity is

beneficial and recommended (Hariga and El-Sayegh, 2011).

2.2.6 Metaheuristic approaches used for leveling Resources.

Further studies have been conducted for resource leveling in linear projects. A

model was introduced that considered the need of an activity in its entirety for

resource adjustments. The limitation of the controlling path was eliminated in the

optimization process and the concept of changing buffer was introduced to realize

high flexibility. Simultaneously, a genetic algorithm was used for solving the model.

The possibility of obtaining optimal solutions could be improved because solving

was based on many initial feasible solutions. However, this approach still has some

limitations. The schedule quality of the genetic-algorithm-based model could not be

guaranteed owing to the characteristics of the genetic algorithm (Georgy, 2008).

A new methodology of modeling linear schedules was implemented using

singularity functions, which have been used for identifying criticality of an activity

and performing the float analyses. The model extracts one flexible equation for the

complete resource profile of a schedule, including any changes in the resource

production rates of activities. A subsequent equation describes the first moment of

area of the resource profile. Minimizing the moment of daily resource usage is the

objective function for leveling the resource profile. A genetic algorithm with inverse

ranking is performed to conduct successive iterations. Chromosomes contain

different permutations resource rates at which the activities can be performed.

| P a g e 31
Probabilistic reproduction, crossover, and mutation steps. The calculations illustrate

the precise of singularity functions in generating a model that integrates the linear

schedule with its resource profile and facilitates the overall optimization process

(Lucko etal, 2011).

A genetic algorithm (GA)-based multiresource leveling model was developed

for linear repetitive projects scheduled using LOB. The model shows that the

production rate and an activity duration are controlled by the resource that requires

the longest duration in completing a unit. Once the LOB schedule is created, resource

leveling is performed according to the principle of optimum crew size that makes use

of a utility data curve and shows the suffering in the productivity rate if the crew size

is different than the optimum crew size. The principle of natural rhythm is

implemented that permits shifting the start dates of an activity forward or backward

at different units of production by changing the number of crews employed. The

activity duration and the precedence relationships between activities do not change

during performing of leveling process. (Damci etal, 2013).

A backward-forward hybrid genetic algorithm (BFHGA) was presented for

producing an optimal scheduling of a resource-constrained multi project scheduling

problem (RCMPSP). This approach combines complementary strengths of the

backward-forward scheduling method, genetic algorithms, and simulated annealing to

remedy resource-constrained multi project scheduling problem (Sonmez and Uysal,

2014).

| P a g e 32
2.3 Entropy Maximization

This section introduces the concept of entropy as a measure of a system

smoothness and degree of spontaneity and shows its implementation in construction

industry. This section begins with defining the theory of entropy in physics.

Afterward, this section points out the preceding application of entropy theory in

construction industry.

2.3.1 State of the Art.

Entropy, in physics, is a measure of the unavailability of a system's energy to

do work, and it is central to the second law of thermodynamics which deals with

physical processes and the degree of spontaneity in their occurrence (spontaneous

changes occur with an increase in entropy). Entropy is a measure of smoothness and

changes of the different systems and it has been associated with the amount of order,

disorder and the amount of wasted energy in the transformation from one state to

another. Entropy relations provide the means for expressing the total amount of

disorder (SD) and order (SO) in the system (Landsberg, 1984).

SD = CD / CI

SO = 1− CO / CI

Where CD is the disorder capacity of the system, CI is the information

capacity and CO is the order capacity of the system. Entropy (Hx) is also defined as

the product of the probability distribution (PX) of a variable x, times the natural

logarithm of the inverse of that probability distribution Eq.

Hx = PX ln (1/px)

| P a g e 33
 There are two main properties of entropy that are associated with

subadditivity and maximality. Subadditivity concept is a function’s property which

means that the function’s value for the sum of two elements is less than or equal to

the sum of the function’s values for each element (Landsberg, 1984)

In entropy terms, if a system consists of two subdomains having n and m

components, the total system entropy is less than or equal to the sum of the

subdomains’ entropy.

Whereas, the concept of maximality implies that the entropy function, (H p1,

p2...pn), takes the greatest value when all results have equal probabilities.

H (v11 ,v12, ... , v1 m ,v21 , ... ,v2 m , ... , V n1 , ... ,Vn m)

2.3.2 The Application of Entropy Theory in Construction Management.

The concept of entropy has not only been tackled in physics and chemistry, it

has also been implemented in construction management. The concept of the long-

term entropy and profitability change of public construction firms was introduced in

the United States. This concept showed that a firm's entropy is the weighted sum of

related and unrelated diversification that can be expressed on the basis of the standard

entropy equation, which is defined as the sales share of the business segment of the

industry group's total sales. The major outcome of this research was formulating a

linkage between entropy and profitability (Choi and Russell, 2005).

A fuzzy and entropy-based mathematical model was developed to solve the

weighting dilemma of in overall cash-flow of the project. The model was applied in

multi project cash flow diagram, which acts as a decision making tool for investing in

a new construction project (Tang et al, 2006).

| P a g e 34
 An optimum model was established for the design scheme of construction

relying on the concept of entropy. This model was applied on indices of different

schemes to build up a non-fuzzy judgment mathematical matrix to calculate entropy,

then these indices were implemented in calculating entropy weight. Afterwards, the

attribute matrix can be calculated to point out the ideal point (Ruan et al, 2009).

A new method for resource allocation and scheduling of resource-constrained

construction projects was pointed out using entropy measurement. The entropy is

considered an indicator of a project's ability to progress out orders and change into a

chaotic condition to predict a project's progress approach. The entropy was used

based on the resource assignments per activity (required vs. assigned resource units)

(Christodoulou et al, 2009).

Another model proposed the use of entropy principal properties which are

sub-additivity and maximality to revisit the minimum moment method for resource

leveling. The entropy-maximization method presented permits the stretching of an

activity and provides resource allocation solutions to enhance the overall resource

usage fluctuation (Christodoulou et al, 2010).

| P a g e 35
2.4 Graph Theory.

2.4.1 State of the Art.

General speaking, a Graph theory is the study of graphs, which are

mathematical structures used to model pairwise relations between objects. A "graph"

in this context is made up of "vertices" or "nodes" and lines called edges that connect

them. A graph may be undirected, meaning that there is no distinction between the

two vertices associated with each edge, or its edges may be directed from one vertex

to another

Graph Theory started with Leonhard Euler in his study of the Bridges of

Konigsberg problem (Dickson, 2006). The city of Konigsberg was located on the

Pregel River in Prussia. The river divided the city into four separate landmasses,

where these four regions were linked by seven bridges as shown in figure 2.8.

Dwellers of the city asked if it were possible to leave home, cross each of the seven

bridges exactly once, and return home. Leonhard Euler (1707-1783) thought about

this problem and the method he used to solve this issue is the graph theory. Since

Euler solved this problem in Graph Theory, it has become one of the most crucial

areas of applied mathematics. Graph Theory is cross-disciplinary between Math,

Computer Science, Electrical Engineering and Operations Research (Dickson, 2006).

Figure 2.8: A diagram for the problem solved by Euler using Graph theory (Dickson, 2006).

| P a g e 36
 The key to Euler's solution was in a very simple abstraction of the puzzle.

Each land masses in the city of Konigsberg is represented as a vertex and each bridge

is imitated as an edge connecting the vertices corresponding to the land masses. A

graph was drawn that encodes the necessary information. The problem reduces to

ending a "closed walk" in the graph which traverses each edge exactly once

(Dickson, 2006).

2.4.2 Graph theory and Sparse Matrices.

Graph theory is a fundamental tool in sparse matrix techniques. A graph G is

defined as the pair G= (V, E) where V is the set of vertices V= {v1, v2... Vn} and E

is the set of edges between vertices .e. E= {e1, e2... ek} consists of pairs ek= (vi, vj)

where Vi and vj are elements of V (Dickson, 2006).

The adjacency graph of a sparse matrix is a graph G= (V, E) whose N vertices

in V represent the N unknowns. Its edges represent the binary relations established by

the equations in the following manner: There is an edge from vertex i to vertex j

when aij not equal 0. This edge will therefore represent the binary relation “equation i

involves unknown j.” We assume that equation i will always involve unknown i, i.e.

there is a loop at each vertex as shown in figure 2.9 (Dickson, 2006).

Figure 2.9: The adjacency graph of a sparse matrix (Dickson, 2006).

| P a g e 37
 When a matrix has symmetric non zero pattern, i.e. when a Ij and A ji are

always non zero at the same time, then the graph is undirected. Undirected graphs can

be represented with non-oriented edges (Dickson, 2006).

Figure 2.10: Undirected graphs with non-oriented edges (Dickson, 2006).

Principle of sparse matrix techniques: Store only the non- zero elements of A

and try to minimize computations and storage.

The following part shows list of problems and situations, in which the Graph Theory

can be implemented (Dickson, 2006).

1- Optimization Problems using Graphs theory: Graph theory treats problems of

optimization like a road network that attempt to maximize the flow along that

network while minimizing costs.

2- Topological Graph Theory: Defining methods that imbed graphs into

topological spaces, where certain properties are maintained. For example, it can

identify the probability of drawing a graph on a plane in such a way so that no two

edge cross like the case of the bridges of Konigsberg graph.

3- Graph Coloring: This part tackles the identification of numbers of colors needed

to color each vertex (or edge) of a graph so that no two adjacent vertices have the

same color.

| P a g e 38
4- Algebraic Graph Theory: Is the application of abstract algebra (sometimes

associated with matrix groups) to graph theory. Many results can be proved about

graphs when using matrices and other algebraic properties.

2.5 Discussion of literature findings.

Many approaches have been developed to remedy the dilemmas of scheduling

repetitive projects since 1960. The Line-Of-Balance technique (LOB) was introduced

to the construction industry as one of the major linear scheduling graphical

techniques. Many researchers have adopted the concept of LOB, and focused on

balancing crew production and maintaining continuity of work flow. After the advent

of the personal computer and the improvement in computing capabilities, many

mathematic approaches have been established to solve repetitive project scheduling

problems. Sophisticated mathematic models and algorithms were developed to

optimize resources required and to smoothen the fluctuation of resources profile.

Advancement of these models was established to mathematically model complicated

and realistic construction operations.

While reviewing the literature, it has been realized that none of the above

mentioned research have tackled the unrealistic assumptions of LOB that assumes

constancy in production rates of each repetitive activity. Moreover, Considerable

research has been developed to ensure work productivity and continuity. However,

most of these studies neglect the flexibility of altering crew movement between

different activities in different unit simultaneously. Besides, the availability of using

multi-task skilled crews in linear construction projects between different activities in

different units. Therefore, there is a need for developing a tool that tracks and

monitors the above-mentioned drawbacks in a dynamic manner by ensuring the

possibility of having different production rates within an activity. Besides, there is a

| P a g e 39
need for applying a multi-task skilled crew between different activities in different

units. Moreover, there is a need for an algorithm that changes linear crews routing and

provides several and different formulation for crews' circulation to generate optimal

crews' allocation for resolving resource constraints problems; besides this algorithm

precisely identifies the critical path in linear projects. Furthermore, the daily resource

usage needs to be minimized by using the concept of stretching or crunching of

activities duration with varying duration in linear projects based on activity type and

the location of the project.

Table 2.1- Summary for the above -mentioned research efforts developed for
scheduling and managing repetitive Projects.

Reference Technique Considerations Comments

Models of Line of Balance Scheduling

Technique.
Arditi and Computerized Integrating both network system and LOB Linear scheduling
Psarros 1987 System (SYRUS) technique to schedule linear projects. projects
Computerized Generate schedule based on calendars and
Hegazi et al, System (BAL) working days to conduct desired rates based Linear scheduling
1993 on deadlines. projects

Senouci and El- Non serial dynamic A time cost tradeoff analytical model for Non- Serial linear
din ,1996 programming linear projects. projects.

Wang and MLS decision Monitoring and controlling interval times Multi stage linear
Huang ,1998 making tool between activities in repetitive projects. scheduling
Computerized Imitate the concept of multi-level LOB Scheduling high
Arditi et al , System (RUSS) diagram to satisfy deadline constraints in rise buildings.
2002 resources.

Tokdemir et al, Computerized Accelerating the project for specified Linear scheduling
2006 System (ALISS) deadline or milestone. projects

Agrama, 2011 Spreadsheet Define relationship controlling start time of Non- Typical
Algorithm. activities at first unit to maintain resource linear project.
without interruption.
CPM/LOB - Models

| P a g e 41
 Suhail and CPM/LOB Determining number of crews to meet Crew availability
Neale ,1994 Model. desired project duration where total float not considered
relax non-critical activities.
Hegazy and
Wassef ,2001 CPM/LOB & Minimizing direct, in-direct, interruption Predecessor and
Genetic algorithm. costs and liquefied damages using G.A and successor are only
CPM/LOB. three
Ammar &
Mohieldin,2002 CPM/LOB CPM based repetitive schedule model for Non-graphical
Model. significant resources only, not multiple approach
ones.

CPM/LOB Determining different types of floats for Constant

Ammar ,2003 Model. non-serial repetitive projects. production rate

CPM/LOB Overlapping between activities are utilized Non-graphical

Ammar ,2012 Model. to indicate duration and logic relationship approach
between activities.
Jung etal, 2013 CPM/LOB Identifying The FS relationship formula Using FS
Model. derivation and application of the simplified relationship only
CPM/LOB method.

Heuristic Lien of Lob model in the form of integer non-linear Deadline

Reza etal, 2014 Balance Model. programming is used for deadline constraint constraint
satisfaction. Satisfaction
Models of Resource Optimization in Linear Projects
Senouci and Genetic Determining optimal number of crews to Non-serial
Naji, 2006. Algorithm. reduce overall project cost. repetitive projects

Hyari and El Computational -Using resource algorithm. Linear scheduling

rayes, 2006. Model. -Multi objective G.A for tradeoff analysis projects

Liu and Constraints Linear scheduling

wang.2007. Programming Optimizing total cost and project duration projects
Algorithm.
Chang and Evolutionary Finding optimal set of production rates. Consider
Huang, 2009. strategy Resource
Algorithm. Limitation
Gafy, 2011. Anty Allocate resources to reduce duration and Consider Multiple
Colony. interruption days. Resource
Limitation
Liu and Wang, Constraints Integrate single/multiple skilled crews to Engaging several
2012. Programming enhance work continuity. heuristic ways
Algorithm.

Models of Resource Leveling in Linear Projects

1. Heuristic Approaches.

| P a g e 41
 Min. Moment Shifting non-critical activates based on total Not applied in
Harris ,1978 Algorithm. float using an improvement factor. linear projects

Modified Min. Multi resource leveling depending on Multi resource

Dubey, 1993 Moment Algorithm. resource weight. linear projects

Son and
Skibniewski, Multi Heuristic Combining local optimizer with simulated Leveling Linear
1999 Hybrid Model annealing depending on splitting of projects
activities.

Modified Two Using a heuristic algorithm to allocate Multi resource

Yen’s , 2005 Stage Method resources, and then leveled using simulated linear projects
annealing.

Constraint Two stage scheduling system for leveling Leveling Linear

Tang etal ,2014 Satisfaction resources using rate float of activities. projects
Problem

2. Optimization Approaches.
Integer Linear Detect controlling activities by Leveling Linear
Mattila & Programming assigning weights while combining the projects
Abraham, 1998 Model multiple resources

Dynamic Considering interruption in optimizing Leveling Linear

El-Rayes et al, Programming resource utilization. projects
2001 Formulation

Hariga and el- Mixed binary - Minimizes he costs of shutdown Leveling Linear
integer - Restarting an activity if it is split projects
Sayegh, 2011 programming - Minimize costs resulting from the
variations of the resource usage.

3. Meta-Heuristic Approaches.

George,2008 Genetic Algorithm Introducing concept if changing buffer to Leveling Linear

achieve high flexibility in resource leveling. projects

Lucko et al, Singularity Integrating linear schedule with resource Leveling Linear
2011. Function. profile through minimizing moment of projects
resources.

Damci et al, Genetic Algorithm Resource leveling is performed using Multi resource
2013. optimum crew and natural rhythm linear projects
principals.

| P a g e 42
 Sonmez and Backward-Forward Integrating forward and backward Resource
Uysal, 2014. Hybrid G.A calculations with simulation annealing. constraint multi
project scheduling
problem

The Theory of Entropy

Choi and Mathematical The concept of the long-term entropy and Profitability
Russell, 2005 Model profitability analysis

Tang et al, A fuzzy and Solve weighting problem of in overall cash-

2006 entropy-based flow of the project Cash-flow
mathematical model analysis

Non-fuzzy Established the design scheme of Decision making

Ruan et al, judgment construction project depending on the tool
2009 mathematical concept of entropy
matrix

Resource allocation and scheduling of

Christodoulou Mathematical resource-constrained construction projects Resource
et al, 2009 Model was pointed out using entropy measurement allocation

Christodoulou Mathematical Using sub-additivity and maximality Resource leveling

et al, 2009 Model principles to revisit the minimum moment
method for resource leveling

| P a g e 43
 CHAPTER 3 – RESEARCH METHODOLOGY AND MODEL DEVELOPMENT

In order to achieve the research main targets and objectives, which were

demonstrated in chapter one, a clear research methodology have to identified and

stated. Therefore, this chapter highlights the proposed research methodology for

achieving the main scopes of this thesis.

3.1 Introduction

In this chapter, the research methodology is declared and discussed in details.

The chapter initiates with stating the main research problem and listing the research

scope and objectives. Subsequently, the research methodology is proposed to single

out the main scheme that have to be followed to pinpoint the research main

objectives. This chapter will discuss the following main topics:

Figure 3.1: Steps of achieving Research Methodology Framework

Thusly, the above-mentioned points are deliberated deeply with a main target of

attaining the research objectives.

| P a g e 44
 Based on the literature review represented in Chapter 2, five areas in which

the current LOB scheduling technique needs improvements are identified.

1. Enhancing and providing a hybrid schedule approach for managing and

monitoring repetitive construction projects in case of time and resources.

2. Clearly identifying critical and non-critical activities in LOB.

 Accurately identify the critical path of a linear construction project.

3. Developing a decision support tool for minimizing resource usage required to

attain a repetitive activity by altering the crews' itinerary.

 The feasibility of applying multi-task skilled crews between different

activities in different units to execute a repetitive construction project.

4. Improving the visualization of crews' movement and overviewing their circulation

among repetitive activities.

5. Improving the resource leveling in LOB scheduling technique.

The aforementioned points are the key areas that require enhancement, in

addition to the areas highlighted in Chapter 2. It is essential to figure out that

resolving the dilemma in these fields needs a careful and a precise accounting for all

the areas combined and necessitates a departure from traditional LOB technique. The

following sections include a full description of these areas of enhancement, followed

by the presenting and analyzing a scheduling scheme for repetitive projects that can

address these needed improvements in a collective manner.

| P a g e 45
3.2 Areas Requiring Enhancement.

3.2.1 A developed schedule approach for managing and manipulating repetitive

construction projects.

Often, linear construction projects involve circumstances that mandate the

occurrence of a float or non-critical repetitive activities to accommodate the dilemma

associated with resource constraints in construction projects. In LOB scheduling

technique, the production rate of a repetitive activity is assumed to be constant along

the entire number of units of a repetitive project, where the quantity of work within

this repetitive activity is identical and productivity of resources is steady.

Furthermore, LOB scheduling technique is generally incapable to portray the

repetitive activities as partly critical, mainly because of the assumption that each

repetitive activity is a single undivided parallelogram with a specific duration. This

hypothesis is unrealistic as the repetitive activity is composed of sub-activities or

stations as a result of breaking down of an activity, which may have different

production rates, and may acquire different crew's usage due to the probability of

having a float that relax the sub-activities production rate.

Figure 3.2 illustrates a simple case study of a repetitive construction project

that is composed of three repetitive activities I, J and K respectively. CPM

calculations of the three activities in the single unit is implemented that show the

criticality of the three activities with no total float, which consequently lead to the

similarity and constancy in the work performed and crews productivity rate of the

three activities along the repetitive units as shown on LOB diagram.

| P a g e 46
 Equal Production
rate (zero total float).

Figure 3.2: Traditional LOB scheduling technique

3.2.2. Clearly identifying the critical path of linear construction projects.

There is an obvious variety between the concept of criticalness in the LOB

scheduling technique and the concept of criticalness in CPM networks. LOB

scheduling technique do not have the ability to identify the accurate criticalness and

floats; hence, an improvement in LOB scheduling techniques should be developed to

point out critical, non-critical activities, and floats to precisely identify the critical

path in LOB.

3.2.3. Minimizing the maximum number of crews utilized to perform a repetitive

activity by changing their routing.

In LOB scheduling technique, each individual repetitive activity is considered

as a single longitudinal bar where its length is equal to the number of units in the

project. Besides, each individual linear activity has its own number of crews that are

required to repeat the same work in different sectors and to perform this repetitive

activity along the number of units of the project. Moreover, the relationship between

crews in an individual activity is a linear vertical relationship only (Linear nature

since one crew follows another sequentially), as there is no any route or path for

crews circulation inside a repetitive activity. LOB scheduling technique assumes that

| P a g e 47
the rate of production for an activity is uniform where the amount work attained by

crews in any individual repetitive activity is equal and constant. This assumption may

be unrealistic in real life construction projects due to the occurrence of unforeseen

conditions and force majeure that may confront any construction project leading to

amending the production rate in a specific unit. Consequently, this results in the

appearance of an ideal time between crews and eventually interrupts the crews work

continuity and extends the total project duration.

Figure 3.3 shows a linear project composed of three repetitive activities. Each

repetitive activity need two crews to be achieved, in which each crew linearly pursue

another in a vertical sequence. The total number of crews required to complete these

three repetitive activities will be six different crews (two per each activity). As well

as , LOB do not have the capability to execute different tasks by applying multi-task

skilled crews that can maneuver and can circulate between different activities in

different units to improve productivity and to reduce indirect costs.

Note: Crew circulation is a

linear vertical circulation
with equal amount of work.

Activity-1 Activity-2 Activity-3

Figure 3.3: Crew Circulation in LOB scheduling technique

| P a g e 48
 3.2.4. Enhancing the resource leveling in LOB scheduling technique.

The majority of linear construction projects are significantly impacted by

circumstances that involve a resolving tool for resource limitation problems. For

resource-constrained projects, the backward pass calculation may proceed incorrect

total floats calculations because the sequence of some activities depends not only on

the logical relationships but also on resource dependencies (Kim and de la Garza

2003). One of the key issues with existing LOB techniques is the fact that each

repetitive activity is regarded as a single parallelogram having a constant production

rate. While the literature describes several techniques that have been developed in

order to remedy resource limitation dilemmas, none of the techniques has been

devoted to consider the principle of stretching or crunching activities duration (

depending on activity type) to get a better resource fluctuation profile and to comply

with actual situations.

3.3. The Proposed Representation of LOB Scheduling Technique.

The following subsections tackle the measurable procedures undertaken to

enhance the representation of a project and activity progress in LOB, which addresses

the areas of required improvement in an ease, practical approach and within a

consolidated framework.

3.3.1. The Representation of Activities in LOB after Breaking-down Activities.

Since the representation of activities and their durations are the basic

parameters for schedule calculations; hence, the precise representation of the

activities would remedy many issues accompanied with the abovementioned

limitation of LOB. In contrary to the traditional manner of overviewing activities as

parallelogram that span a specific number of units and period of time, repetitive

| P a g e 49
activity is represented as a number of separate sectors or stations, but connected time

sectors that add up to the total duration of the repetitive activity. For example, an

activity of any linear project with a three kilometer distance and a duration of six

days is broken down into three equal separate time station or sector with and a length

of one kilometer and a duration of two days.

This method permits the representation of any logical relationship between

the broken-down activities using only a Finish-to-Start (FS) relationship. The

breaking down of activities into separate equal stations or sectors provides more

accurate calculations that result in creating a partly critical and non-critical activities,

unlike the traditional representation of activity, and consequently generate a more

precise critical path.

On converting the above-mentioned methodology into LOB calculations, each

activity is decomposed into separate sub-activities with different production rates due

to the creation of critical and non-critical sub-activities within a repetitive activity.

Accordingly, this reflects the stochastic nature of construction processes and do not

account for practical assumption of having constant production dates while using

LOB scheduling technique.

3.3.2. Proof of the concept

To illustrate the feasibility of the abovementioned ideology in providing

better representation and analysis than the traditional LOB, the following simple case

study evaluates the applicability of the concept. The simple example is composed of

three different activities, each with a three kilometer distance. Nevertheless, these

three activities are broken-down into separate sectors, each with a one kilometer

distance and equal durations. The breaking down of activities results in increasing

ability to represent the internal relationships between the activities, which

| P a g e 51
 consequently creates partly critical and non-critical sectors with an activity. The

breaking down of activities is rendered into a LOB format, which emanates in

breaking down an activity into sub activities with different production rates (activity

relaxation as a result of an activity float and different crews' requirement as shown in

figure 3.4 and 3.5 respectively.

Repetitive activities are relaxed as a result of an activity float

Note: This example was

previously solved without
breaking down, in which
all activities are critical.

Figure 3.4: Methodology of Breaking-down activities

Figure 3.5: LOB representation for the simple example

| P a g e 51
3.4. A proposed algorithm (Model) for minimizing number of Crews utilized in

LOB Scheduling Projects and allocating crews among different activities.

3.4.1. Introduction.
The allocation of resources to repetitive activities of a construction project is

regarded as a difficult managerial issue, since inaccurate allocation of crews in LOB

scheduling technique can result in extension of the project time and increase in the

project overall cost. This proposed optimization model presents a platform to

minimize the total resource usage of a project, while meeting the requirements for

work continuity and the target deadline of each activity. Moreover, the proposed

model have the capability of accurately identifying the critical path in LOB

schedules.

3.4.2. Process outline of the proposed optimization Model.

The following sections will discuss the outlined procedures in detail covering

required input data for model formulation and the nature of computations within them.

First, an overview of the main attributes and parameters associated with the proposed

model and their relevance will be fully presented. This will be followed by a detailed

description of the sequential processes and steps needed to construct the model skeleton

and realistically to generate the expected outputs. Finally, the output modules of the

model will be displayed. MATLAB R2013a is used to run the proposed optimization

model.

3.4.3. Main attributes and parameter associated with the proposed model

Table 3-1 shows the major attributes and parameters associated with proposed

optimization model, in which the main objectives, variables and constraints accompanied

with the proposed algorithm or model is illustrated in details.

| P a g e 52
 Attribute Description

 Minimize the number of resource usage.

 The assignment of multi task skilled crews in LOB among
Objective Functions different activities in different units.
 Accurately identifying the critical path in LOB schedules

 Number of crews.
 Crews' formulation.
Variables
 Circulation or routing of each Crew.

 Start and finish date of each activity.

 Activity Production rate.
Constraints
 Overall project completion date.
 Logic relationships among activities.

Figure 3.6 illustrates sequential processes and steps undertaken to construct the

proposed algorithm or optimization model skeleton by which the data is utilized to

generate the needed outcome module.

3.4.4. Description of Optimization Model Formulation

Figure 3.6: Optimization Model Flowchart.

| P a g e 53
3.4.4.1Proof of Concept
To demonstrate the ability of this model to provide better results and analysis

than the traditional LOB, the following example is introduced throughout the model steps

to ensure and test the capabilities of this optimization model. Figure 3.7 illustrates the

simple example, which is composed of three repetitive activities with different

production rate. The three activities i1, i2 and i3 require number of crews equal to 5,

3, 2 respectively to meet the desired completion date. The total crews' number using

traditional LOB is ten, needed to complete the five units in the specified date.

Figure 3.7: Case Study to show model efficiency

3.4.5. Optimization Model Framework

Step-1: Input module

This module is used by the scheduler for inputting data related to the project. Figure

3.8 shows the gathering data required for the input module

| P a g e 54
 Figure 3.8: Data grouping for the input module

 Project related information. Information related to the project such as the

project name, project code, number of repeated units, start and completion date of

the project.

 Activities. Descriptions of activities and their codes.

 Activity Start and finish dates. The start and finish date of each activity should

be specified to identify relationship between activities.

The first step in model formulation is calculating the total number of activities

(NT) of each repetitive activity, where (NT) is equal to the number of repetitive sub-

activities (Ns) multiplied by number of repetitive unit (N).

Step -2: The second step is defining the initial inputs of the model that are

activities start date, finish date and activities numbers as shown in figure 3.9 , where

the first activity ( start node) is defined as X1 followed by X2, X3, X4…Till the final

activity that is expressed as X (NT-1). Each activity is represented by two variables,

variable one is the start point (event) and the second variable is the finish point.

Segment 1………………… X1.

Segment 2………………… X2.

Segment final……………....X-final.

| P a g e 55
 Application of step 1&2 on MATLAB.
1- The total number of activities = number of units (5) multiplied by number of sub-
sub-activities per unit (3), which is equal to 15.
 Functions used in MATLAB to calculate number of activities.
Numactivity=A (size (A, 1), 3).

 Where A is the input values (start date, finish date and number of activities).

 1 implies the start dates of activities

 3 represents the number of activities

Figure 3.9: A Snap shot for the function used in MATLAB for step 1&2.

2- Defining the initial inputs.

NO
X-1
X-2
X-3
Activities start and X-4

finish dates are extracted

from LOB graph of the

Case study in figure 3.7

X-14
Figure 3.9: The initial input data for the Model
X-15

Figure 3.10: Defining Input Data of the Model.

| P a g e 56
 After inserting all of the abovementioned parameters shown in figure 3.10,
these inputs are used in step 3 to determine predecessor and successor activities.

Step -3: The third step is identifying the successor and predecessor activities

depending on the input variables. Moreover, a table is tabulated that include two

columns, the left hand side column (A) that contain the initial segment (X1), while on

the right hand side column (B) the successor activities of the initial one is specified.

The finish date of each activity of array A is compared for equality with the

corresponding start date of activities of array B. If A and B are equal then activity in

column B is successor to that on columns A, but if A and B are not equal then

activity in column B is not successor to that in column A. Afterwards the previous

steps are iterated until all successor activities are identified. Eventually, a final index

will be displayed showing successor activities as show in figure 3.19.

Column-A Column-B

Figure 3.19: Identification of successor activities.

| P a g e 57
Application of step 3 on MATLAB.
 Functions used in MATLAB for the illustrative example are shown in Figure 3.12.

- i=1: size (A, 1)-1

- Indexr=find((A(:,1)==A(i,2))| A(:,1)==A(i,2));
- Index1 (:, 2) =A (Indexr,3);
- Index1 (:, 1)=A(i,3);
- Index= [Index; Index1];
 Where A is the input values (start date, finish date and number of activities).

 1 implies the start dates of activities

 2 represents finish dates

 3 represents the number of activities

Figure 3.12: A Snap shot for the function used in MATLAB for step 3.
Step 4-Perform Sparse Matrix Module:

This module is accountable for performing a (NT-by-NT) matrix composed of

zero and one entries (number of columns = number of rows = number of activity

segments), this module uses the data generated from previous step to perform the

adjacent matrix. This matrix is achieved by observing the predecessor and successor

activities presented in step three. If there is a relation between an activity and another

one, then the intersected entry between the row and column will be 1 (non-zero

entry). On the other hand, if there is no relation between an activity and another one,

then the intersected entry will be zero (zero entry) as shown in figure 3.13.

With a relationship

(X1, X4) A one entry will be acquired

Without a relationship

(X2, X3) A zero entry will be acquired

| P a g e 58
Application of step 4 on MATLAB.
 Functions used in MATLAB for the illustrative example are shown in Figure 3.13.
The sparse matrix function is generated using MATLAB, in which S = sparse

(A) converts a full matrix to sparse form by squeezing out any zero elements. A

sparse matrix was created from a list of nonzero elements using the sparse function

with five arguments.

S = sparse (i, j, s, m, n)………………………………… Input arguments.

 i the row indices of the nonzero elements;

 j the column indices of the nonzero elements;
 s the values of the nonzero elements;
 m the number of rows in the matrix;
 n the number of columns in the matrix;
 nz_max the maximum number of nonzero elements in the matrix;
 nz_max is omitted, in which case its value is taken from the length of s.

Sparse Matrix Functions in MATLAB

S=sparse(Index(:,1),Index(:,2),ones,Numactivity,Numactivity);

 Where Sparse: Create a sparse matrix or convert full matrix to sparse matrix.

 Index (:,1): The column that includes all number of activities.

 Index (:, 2): The column that includes predecessor and successor activities.

 Numactivity : Implies the total number of activities (NT)

To generate a Sparse Matrix, the following function is used

[S, Names]=graphattempt (A);

 Where S=sparse(Index(:,1),Index(:,2),ones,Numactivity,Numactivity);

 Names: Means the activities names (X1, X2, X3, X4,………..).

 Graph attempt: The Name of the file incorporating the code

 A is the initial inputs (start date, finish date and number of activities).

| P a g e 59
 Figure 3.13: A Snap shot for the function used in MATLAB for step 4.

Figure 3.14 demonstrates the results of implementing the sparse matrix as a

result of sets of data obtained in step three. This figure points out data extracted

from Matlab, in which the sparse matrix is developed.

Figure 3.14: A Sparse Matrix Formulation Diagram.

| P a g e 61
Step 5-Graph Theory Module:

Graph theory is a mathematical way of representing connections or

relationships between activities. Graph theory pairwise relations between objects

using vertices and edges. The abovementioned sparse matrix is responsible for

establishing a graph theory network (GTN) by identifying number of activities that

are represented as nodes (vertices) and number of lines that are called edges, which

connect the nodes according to assigned relationships between activities as

mentioned earlier in step three and four. The main objective of the graph theory

network is classifying and catering out the possible and total number of paths of each

crew throughout the entire network. The arrangement and allocation of the nodes is

defined according to the relationships among activities as shown in figure 3.16.

Application of step 5 on MATLAB.

 Functions used in MATLAB for the illustrative example are shown in Figure 3.15.

The graph theory functions work on sparse matrices. The only restriction is
that the matrix should be square. In graph theory there are two variables: 1) S, a
sparse matrix, and 2) Names, a list of the names of the nodes (activities) of the graph.
Visualizing the Graph
The first function used in visualizing the graph is the biograph object,
which is a way of indicating the number of nodes and edges
1- gObj = biograph(S,Names)
 Where S is the Sparse Matrix

 Names: List of activities

The result generated from function 1 is Biograph object with 15 nodes and 10 edges
(In which nodes are number of activities and edges are the number of arrows
representing relationships between activities).

The second function is the view method, which lays outs the graph and
displays it in a figure or network format.

| P a g e 61
 Figure 3.15: A Snap shot for the function used in MATLAB for step 5.

Figure 3.16 shows the graph theory network rendered from Matlab, which is

generated by identifying number of Nodes ( activities) and number of Arrows

(relationship between activities).

2- gObj = view(gObj);

Figure 3.16: Graph Theory Network Diagram

Step 6- Structural Array Module:

This step is proceeded through an integration between the sparse matrix

conducted from step four and the graph theory network genertaed from step five to

develop a structure array module that include and list out all the possible paths for

each crew routing. The procedures of this step is as follows:

| P a g e 62
Application of step 6 on MATLAB.

 Functions used in MATLAB for the illustrative example.

1) This step begins with the crew path containing just the start node (X1).

MATLAB code for this step: lastpath = [startnode];

2) A matrix is created with a row for each possible path, in which each crew
path is accompanied by a cost equal to the number of activities included in
this path.

MATLAB code for this step: nextmove = wt (lastpath (:, i - 1), :) ~= 0;

3) The first crew is assigned to the path (crossed nodes) with the highest cost,
which is the longest path in the graph theory network.

4) Zero out the crossed nodes already visited by the first crew from the list of
paths.
MATLAB code for this step: d = diag(1:size (lastpath,1));
nrows = d * ones(size(lastpath));
inds = sub2ind(size (nextmove), reshape(nrows,[],1), reshape (lastpath,[],1));
nextmove(inds) = false;
5) The second crew cross out the nodes (activities) with the highest cost, the
crossed node is eliminated from the list of available paths
MATLAB code for this step: nextmoverow = d * nextmove;
nextmovecol = nextmove * diag(1:N);
rowlist = reshape(nonzeros(nextmoverow),[],1);
collist = reshape(nonzeros(nextmovecol),[],1);
nextpath = [lastpath(rowlist,:), collist];
inds = sub2ind([N,N],nextpath(:, i-1),nextpath(:,i));
costs = costs(rowlist) + wt(inds);
reachedend = nextpath(:,i) == endnode;
paths = [paths; {nextpath(reachedend, :)},{costs(reachedend)}];
lastpath = nextpath(~reachedend, :);
costs = costs(~reachedend);
6) The previous step is iterated until there are no more available moves.

| P a g e 63
 Figure 3.17 illustrates the steps undertaken to assign crews for the available

activities and to minimize their usage by changing their routing while maintaining

work continuity. This figure shows the procedures of structural array module

achieved to reach the objective function of this model.

All available crew The Assignment of

paths as a result of crews after eliminating
structural array redundant activities

Figure 3.17: Structural Array Module

Step-7: Crew Diagramming Approach (CDA).

In case of multi repetitive construction projects more than one crew is used to

execute the repetitive activities of a construction project. The above-mentioned

proposed model assign crews to different activities, where these activities may exists

in different units and in different sub activities. Hence, a new representational system

for different crews circulation need to be performed that focuses on a good

acknowledgment of the relationships between various activities in different unit that

| P a g e 64
acquires the same crew while maintaining the crew work continuity. This system is

called Crew Diagraming Approach (CDA), which enhance planning of crews'

circulation and avoid the problems associated with the confusion that occurs due to

increasing number of crew paths from one unit to others. Besides, it is considered a

networking technique that offers the simplest possible scheme to enable the visual display

of planning and circulation of different crew among different units. Moreover, this method

allows project managers, superintendents, key subcontractors and other stakeholders to

have a hands-on, planning tool that utilizes a variety of simple logic ties to convey crew

circulation and relationships.

3.4.6. Description of Crew Diagramming Approach.

The crews diagram approach is composed of a time scale on the horizontal

direction, and number of units required on the vertical direction. CDA denotes

activities as a bar between two nodes, which are start and finish nodes, and these two

nodes are connected to each other by a bar that resemble duration of the activity. The

embedded nodes connect dependent activities and permits only finish to start

relationship to ensure the maintaining of crew work continuity. This bar is

highlighted by a segment code which includes the activity and unit number as shown

in figure 3.18.

Figure 3.18: Simple representation of an activity in CDA

| P a g e 65
 Figure 3.19 shows a series of arrows that are used to follow the passage of the

crews from one unit to the other in an upward and downwards direction. Each crew is

highlighted by a color to distinguish between crews and to assist on the mapping out

crew's circulation from one unit to the other one. The CDA contain a legend with the

different crew names and various crews' colors.

Figure 3.19: Crew Diagramming Approach

3.4.7.Optimization tool

MATLAB R2013a is the optimization tool used, which is considered a high-

level language and interactive environment for numerical computation, visualization,

and optimization. In MATLAB, the user can analyze data, develop algorithms, and

create models and applications. In this model, codes and mathematical function for

optimization were built-in to enable exploring multiple approaches and reaching a

| P a g e 66
solution faster and more precise than that generated using spreadsheets or traditional

programming languages. Functions used in MATLAB are written as follows:

Function [B] = C (A), where C is the name of the function that accepts

inputs A and returns outputs B. This declaration statement must be the first

executable line of the function.

3.4.8.Model Development

The model is composed of four different files, each is accountable for

performing a mathematical function, in which the function's output is utilized by the

successor file. Each function code is saved in a text file with a .m extension. The

name of the file should match the name of the first function in the file. Valid

function names begins with an alphabetic character. The model files is as follows:

1- Graph attempt.m (Figure 3.20): Responsible for pinpointing number of

activities, predecessor and successor of each activity to execute the sparse matrix.

Figure 3.20: A Snap shot for the function used in MATLAB to figure out number of activities.

 Function used: function[S, Names] =graphattempt (A).

 Where S refers to the sparse matrix.
 Name refers to the activities names (X1, X2, X3,………).
 A represents the input data (start date, finish date and activity number).
2- Attempt 4.m (Figure 3.21): Accountable for plotting the graph theory network

based on data picked out from the previous file (sparse matrix). Besides in this file

a structural array is performed relying on the extracted information from the graph

theory network, in which all possible crews paths and longest crew path are

identified.

| P a g e 67
 Figure 3.21: A Snap shot for the function used in MATLAB to plot the graph theory.

 Function used: function [pathf, lengthf,longestpath]=attempt4(A)

 Where path f stands for all possible paths for crews.
 Length f refers to length of each path to determine the longest crew path.
 Longest path is the longest crew path as a result of structural array
 A represents the input data (start date, finish date and activity number).
3- All paths.m (Figure 3.22): Responsible for generating a decision making criteria

for assigning the crews for different paths by computing cost of each path and

zero out the redundant activities in each crew path.

Figure 3.22: A Snap shot for the function used in MATLAB to formulate a decision making criteria.

 Function used: function [paths] = allpaths (wt, startnode, endnode).

 Where all path stands for all possible paths for crews.
 Wt refers to weight or cost of the path.
 Start node represents is first activity in each path
 End node represents is final activity in each path.
4- Final attempt.m (Figure 3.23): Accountable for generating the output by

assigning crews and determining the routing of each crew.

 Function used: function [crew] = Final attempt4(A)

Figure 3.23: A Snap shot for the function used in MATLAB to Conduct Output data.

| P a g e 68
3.5 The proposed Simulation Model for resource leveling using Entropy
Maximization.

3.5.1. Introduction.

To meet the physical limitation of construction resources, to avoid day-to-

day fluctuation in resource demands, and to maintain the flow of implementation for

construction resources, resource leveling is needed in the construction industry. This

section deliberates the employment of different strategies for resource allocation and

assignment per activity per period through simulation to single out an optimum

solution for resource constraint problems. Moreover, this method guarantees the

satisfaction in respecting the logic relationships, resource continuity and resource

availability

3.5.2. Overview of the Simulation Model.

Simulation modeling using entropy maximization for leveling linear repetitive

construction projects has not been applied either in LOB scheduling or any other

linear scheduling technique. As discussed previously, there are two main properties

of entropy that are associated with subadditivity and maximality. Subadditivity

concept is a property which means that the function’s value for the sum of two

elements is less than or equal to the sum of the function’s values for each element.

On the other hand, the concept of maximality implies that the entropy function takes

the greatest value when all results have equal probabilities. With respect to the

concept of entropy, a higher percentage of a resource-based entropy optimization

should achieve a well-executed project in term of resource utilization. The dilemma

of resource leveling in LOB technique can be translated to an entropy maximization

problem, where the number of resource needed by an activity per period should be

diverted to increase its entropy while taking into account the total resource

| P a g e 69
availability of an activity. This section tackles employing a simulation model

templates using entropy maximization to level linear construction projects in

compliance with discrete-event simulation. To level linear repetitive projects

successfully using entropy maximization through simulation modeling, the following

parameters must be employed when establishing the simulation model.

3.5.3. Main attributes and parameter associated with the proposed model

Table 3-2: Major attributes of the Simulation model.

Attribute Description

Objective Function  Minimizing the daily resource usage

 Smoothing the fluctuation in resource utilization profile

 Start and Finish dates of activities.

Variables
 Activities duration.
 Daily resource assignment.
 Resource usage per period.
 Value of daily resource entropy.
 Resource continuity constraints.
Constraints
 Resource availability constraints.
 Total available number of resource per each activity.
 Overall project completion date.
 Total float of each activity.

In accordance with the abovementioned attributes of the simulation model,

three templates are used. 1) Activity and resource flow template, 2) Bar chart

template and 3) Maximum entropy analogy template are designed to systematically

level resource in LOB scheduling technique. Activity and resource flow template is

designed for modeling crew's different paths, variability in activity durations, early

start, early finish and total float of each activity, resource availability constraints,

total available number of resources per each activity, and total project duration

constraint.

| P a g e 71
 Additionally, the bar chart template is used for modeling different scenarios

for the number of resources used per activity per period, while taking into account the

total number of resources per activity, which consequently reflects on the total

number of resources per day and on the resource profile. On other hand, maximum

entropy analogy template is for showing the best scenarios accompanied with the

maximum entropy and the allocation of resources per period for each activity.

To demonstrate the implementation of abovementioned three templates used

in the simulation model, an extensive detailed description is introduced showing the

usability, flexibility, and extensibility of the three templates as shown in figure 3.24.

3.5.4 Description of Simulation Model formulation

Rt is the total number of

resources per activity

Ra is the daily resource

assigned for an activity

Figure 3.24: A Flowchart for Simulation using Entropy Maximization.

| P a g e 71
3.5.5. An overview of the proposed simulation Model.

1- Activity and Resource Flow Template

Main functionalities of the activity and resource flow template are shown in figure 3.25.

 Maintaining crew's different paths.

 Identifying early start, early finish and total float of each activity.

 Pinpointing the available number of resources per period for each activity.

 Figuring out the total sum of number of resources per each activity.

Figure 3.25 the main functionalities of the activity and resource flow template.

Following are the steps to be followed for implementing this template.

1) The first step in implementing the activity and resource template is listing out the

different crews or resources paths obtained from LOB, and determining the

number of activity per each path.

2) The second step is specifying the predecessor and successor of each activity with

respecting each crew circulation in order to calculate ES, EF &TFof each activity.

3) The third step is representing the available number of resource per each activity

per period (the number which will be adjusted to reach the most appropriate

allocation to minimize resource fluctuation). Furthermore, resource assignment

| P a g e 72
 per period is an input variable defined as an assumption using a discrete uniform

distribution probability.

4) The final step is calculating the total available number of resource per each

activity by multiplying the activity duration by the available number of resources

per period for each activity.

Figure 3.26 shows the spreadsheet activity and resource flow template that is

composed of six main columns which are activity duration, predecessor, and

successor, Resource assignment per period, activities calculations and total resource

per activity. This template includes start and finish date of the project.

Defined as a Defined as an
Forecast assumption

Figure 3.26: Activity and Resource Flow Excel Template

2- Bar Chart Template.

Based on the abovementioned data extracted from activity and resource flow

template, a total of 500,000 simulation cycles are run on Crystal Ball Ribbon

Simulation Software. Resource assignments for each activity per period are defined

as an assumption that are randomly selected from discrete uniform distributions,

with a minimum and a maximum values to generate resource assignment

combinations by means of simulations, subject to a limited resource availability

| P a g e 73
based on activity type that was obtained from expertise and keep records of worker

hours, crew sizes, and daily working hours in previously completed projects Several

assumption should be taken into consideration to be well performed like:

1) All assigned resource units have the same productivity rate.

2) The total project duration is a constraint
3) There is no restriction in extending the activity duration.
4) The resource assignment per period is an integer value not a fractional one.
The total entropy of the system, the total resources of the activity and project

duration are defined as forecasts. The outputs are classified into two parts, which are

variable output and constant output. The variable output is the total entropy of the

system, while the constant output is the total resources of the activity and the project

duration (driven from possible resource assignments combination). Moreover, the

resulted simulation is filtered for values with a summation equal to the total number

of available resource for the all activities. Besides, the entropy of each resource

usage per period is computed to come up with the total project entropy. The resource

usage per period entropy can be calculated as follows: HRu = Ru ln (Ru/Rt), where

HRu is the daily resource entropy, Ru is the resource usage per period, and Rt is the

total number of resources.

Whereas the total project entropy can be calculated as follows:

In which, ri=number of resources assigned to the project on period i;

nt=number of total time-units in the project, and rT=total number of resource

required to complete the project. The aforementioned entropy framework aims to

maximize the project’s total entropy, with respect to resource constraints.

Figure 3.27 shows the bar chart template designed for the simulation model.

This template uses the data generated from the previous template, in which each
| P a g e 74
 activity include number of cells where the resource assignment is diverted and

extended or shrunken among all activities. The stretching of activities is constrained

with the premise of increasing or decreasing resource assignment (constrained by a

limited value based on activity type) per period till the maximum entropy is reached.

Defined as
constant
Forecast

Activity
Splitting
Criteria

Defined as
constant Figure 3.27: Bar Chart Excel Template.
Forecast

3- Maximum Entropy analogy.

In this step, a template is tabulated that includes all activities, and the best

scenarios having the highest entropy value, using simulation the best feasible

entropy-based and resource-leveled cane be achieved. Figure 3.28 shows different

scenarios and their associated total entropy, project duration and maximum resource

assignment. Consequently, the scenario attaining the maximal entropy value and

maintaining the total project duration with the least resource usage is opted.

| P a g e 75
 Figure 3.28: Maximum Entropy analogy Template.

3.5.6. Optimization tool

Once having variables, constraints and the objective function ready, the

simulation tool is used. There are several simulation applications available, one of

these is Crystal Ball ribbon. Crystal Ball ribbon is a program created by oracle and

run on several versions of Microsoft excel. Crystal Ball ribbon is a graphical,

analytical oriented forecasting and risk analysis program that takes the uncertainty

out of decision-making. Crystal ball represents the relationships between input and

output variables using functions, formulas, and data, in which it closely matches the

behavior of a real project circumstances. Crystal Ball ribbon uses Monte Carlo

simulation to overcome limitations encountered with traditional spreadsheet analysis,

where the user can describe a range of possible values with Monte Carlo simulation.

Moreover, Crystal Ball displays results in a forecast chart that shows the entire range

of possible outcomes and the likelihood of achieving each one. In addition, Crystal

Ball keeps track of the results of each scenario for the user.

| P a g e 76
3.5.7. Output module
To substantiate the feasibility of the proposed model, its outputs are

represented graphically using LOB graph. This LOB graph is drawn using the data

generated from the simulation model (start date, finish date and production rate of

each activity). The LOB has an X-axis representing the unit numbers and a Y-axis

representing the duration. Due to the applicability of this model, LOB is drawn with

different production rates for each activity per each unit (stretching and extending of

the repetitive activities along the units' number).

3.5.8. Model Development

A computerized model was implemented using the abovementioned

framework of the proposed model, in which a spreadsheet modeling tool (Microsoft

Excel) is utilized. The process of the proposed model is composed of three main

stages, which are overviewed as follows:

1) Establishing a spreadsheet model that describes an uncertain variables of the

daily assignment of resource usage, which is defined as an assumption with a

probability distribution (uniform discrete distribution). The simulation process

calculates numerous scenarios of the model by repeatedly picking values from

the probability distribution for the uncertain variables (resource assignment).

2) Running a simulation based on the input values entered and stored in the

assumption cell.

3) Analyzing the results through keep tracking the forecasts of each assumption.

For each forecast, crystal ball represents the cell value for all the trials, and after

hundreds or thousands of trials, the user can view sets of values, the statistics of

| P a g e 77
 the results (Output values for daily resource assignment), and certainty of each

value.

3.6. Summary and Closure of the Chapter.

This chapter has overviewed areas of required improvements in LOB

scheduling tool which are: (1) providing a schedule approach for scheduling

repetitive projects in case of time and resources; (2) Clearly identifying critical and

non-critical activities; (3) Minimizing the number of crews usage to perform a

repetitive activity by altering the crews routing and applying multi-task skilled

crews; (4) Enhancing the visualization of crews' movement and circulation between

repetitive activities, and; (5) Improving the resource leveling in the LOB scheduling

technique.

First and foremost, this chapter highlighted the research framework in depth,

where it presented the optimization model for minimizing number of crews utilized

by changing the criteria of their routings. A small case study has been presented

throughout the model in order to demonstrate the feasible approach that can reduce

the number of crews assigned without influencing the activity or total project

duration.

Moreover, this chapter introduced the methodology of the proposed

simulation model and its procedures for resource leveling with activity stretching or

crunching using entropy maximization theory, in which all the templates designed to

run the model are extensively discussed in details to obtain a near optimum

integration of the variable leading to the reliable outputs. In order to show the doable

implementation of the proposed models and validate its ability to generate near

optimum results, the following chapter shows the model validation by applying a

real case study of a pipeline installation project from the literature review.

| P a g e 78
 CHAPTER 4 - VERIFICATION AND VALIDATION

4.1. Introduction

In this chapter, validation and practical implementation of the proposed

methodology for scheduling multi repetitive projects in Chapter 3 – Research

Framework and methodology is explained. In order to ensure that the proposed

models for the new approach is capable of successfully improving the resource

constraint problems associated with linear projects, a case study is conducted that

demonstrates the effectiveness of proposed models for meeting constraints and for

enhancing the efficiency of resource usage. Nevertheless, this chapter represents the

model verification process in details and the computational methods which is

applied on a real case study obtained from the literature.

4.2. The Developed Strategy for the Proposed Approach.

This chapter shows the implementation of the tremendous efforts done to

improve the scheduling of linear projects through LOB; accordingly, the first section

in this chapter deals with the assumption that production rates of repetitive activities

are linear and constant which is erroneous and unrealistic due to the stochastic

nature of construction processes. The second section tackles applying the developed

decision support tool (algorithm) that aims for minimizing the number of crews

employed to attain a repetitive activity by altering the crews' itinerary. This model

applies the usage of multi-task skilled crews to execute a repetitive construction

project, and accurately identify the critical path in LOB. Eventually, the final section

addresses improving the resource leveling in LOB scheduling technique via a

simulation spreadsheet model based on entropy maximization metric, in which

parameters of the simulation model (i.e. assumptions, decisions and forecasts) are

presented as they are defined within the spreadsheet model. Figure 4.1 shows the

| P a g e 79
flowchart diagram attained to validate the proposed approaches by measuring their

feasible application on a real case study from the literature.

Figure 4.1: Implementation of Proposed Approach based on the case study.

| P a g e 81
4.3. Description of the Case Study.

In order to demonstrate and to evaluate the feasibility of the proposed

framework models, a real case study is undertaken to validate and to test the

developed framework. Since the approach developed works best at horizontal

infrastructure projects, a 4-Km pipeline installation and control room project is the

repetitive case study that consists of 5 repetitive activities. These activities, in their

order of precedence are identified, and all precedence relations are finish to start,

with no lag time as summarized in table 4-1.The project is divided into 10 parallel

relocated repetitive pipelines of equal lengths, each is 4 km length.

Table 4.1- Summary for the activities code, name, description, duration and

precedence relations of a single 4-K.m pipeline installation project.

Activity Code Activity Name Duration Predecessor

A Move in 2 ---

B Survey & Layout 1 A

C Locate & Clear (4km) 16 B

D Excavate ( 4Km) 20 C

E Prepare Pipes (4 KM ) 16 C

F Lay Pipes ( 4 KM ) 36 E,D

G Local Test for ( 4 KM ) 4 F

H Foundation for Control Room 8 B

I Finishing Control Room 17 H

J Installing Control Equipment 5 I

K Testing Control with Pipe Line 4 G,J

L Clean up 2 K

| P a g e 81
4.4. Traditional Manner for solving the Case Study.

4.4.1. CPM calculations for a single unit.

A time analysis (CPM) is performed, in which forward pass calculations are

conducted to determine early timings of activities, whereas late timings of activities

are proceeded in the backward pass calculations. The forward and backward pass

calculations result in having critical and non-critical activities. Critical activities are

A, B, C, D, F, G, K, and L respectively, while non-critical activities are E, H, I, J

respectively. Work breakdown of activities and estimated total project duration of a

single 4-Km pipeline are calculated as 86 days according to CPM calculations as

shown in figure 4.2. The desired contract date to complete installing 10 identical

pipeline is 95 days.

Critical Activities

Non-Critical Activities

Critical Path

Figure 4.2: Traditional CPM calculations for a single 4-Km Pipeline installation.

4.4.2. Basic LOB Representation using Integrated CPM-LOB Model.

The main objective of LOB formulation is to attain a resource-balanced

schedule by identifying the number of crews to be employed in each repetitive

activity. The integrated CPM and LOB model developed by Ammar (2012) is

utilized to schedule repetitive projects in an easy manner, in which logic

| P a g e 82
dependency and resource continuity constraints are taken into account.

Furthermore, overlapping activities of a single typical unit is used to figure out

duration and logical dependency relationships among different activities that can

be specified according to the selected rate of progress of each activity. To specify

such relationships, the actual progress rate of each activity is compared with that

of its successors, in which overlapping repetitive activities with start to start

relationship is specified when the rate of predecessor activity is faster than that of

the successor one. On the other hand, finish to finish relationship is specified

when the rate of predecessor activity is slower than that of the successor one. The

calculation steps and procedures (LOB calculations, calculating activity duration

along all repetitive units, specifying logical relationships using overlapping

activities, and CPM time analysis) undertaken to schedule the case study are

discussed in details in Appendix-A. Figure 4.3 represents the CPM time analysis

achieved as a result of applying the CPM-LOB integrated model.

Critical Activities

Non-Critical Activities

Critical Path

Figure 4.3: Traditional CPM time analysis of the case study.

| P a g e 83
4.5. The Proposed Approach for Scheduling Repetitive Construction Projects

To better represent practical instances in actual construction projects, the

abovementioned activities are being split into equal separate sector (sub -activities),

in which each broken-down activity is considered a separate flexible activity with a

1-Km length and a duration equals total activity duration divided by 4. This

approach intents the creation of inner relationship between detached activities that

produce more practical and realistic schedules.

Figure 4.4 shows the implemented case study after splitting each activity into

equal separate sub-activities, where activity "C,D,E & F" is broken-down into four

equal separate 1-Km distance ( for instance, activity C is sub-divided into C1, C2,

C3, & C4). In contemplating the abovementioned network with the one conducted in

figure 4.2, it is clearly noticeable the reduction in single unit duration by 30 days.

Not only the optimization in duration is the only benefit, but also the activity C, D

and G that were regarded as critical activities using traditional method in figure 4.2,

includes partly non-critical activities as in figure 4.4. For instance in activity C, C1

is critical activity while C2, C3 and C4 are non-critical activities. Another benefit

outcomes from splitting activities is the precise detection of the critical path.

Critical Activities

Non-Critical Activities

Critical Path

Figure 4.4: CPM network for a single unit after splitting activities

| P a g e 84
4.5.1 Basic LOB Representation using Integrated CPM-LOB Model.

Appendix-A shows calculation steps and procedures (LOB calculations,

calculating activity duration along all repetitive units, specifying logical relationships

using overlapping activities, and CPM time analysis) undertaken to schedule the

presented case study and to generate CPM time analysis as shown in figure 4.5.

Figure 4.5 shows the early and late time after accounting for logical dependency

relationships among different activities that are specified according to the rate of

progress of each activity. Activities C1, C2, C3 & C4 have different durations because

the total float of noncritical activities (C2, C3 & C4) is utilized to reduce the number

of crews employed and to relax activity production rate. For instance, C1, C2, C3&

C4 originally requires 5 crews, which is reduced into 4, 3 & 2 crews for activities C2,

C3 & C4 respectively as summarized in LOB calculation procedures in Appendix-A

and shown in figure 4.6. The resulting project duration is 71 days, in which the timing

of activities determined in figure 4.5 is graphically plotted on a LOB format with the

number of crews required as shown in figure 4.7, 4.8, 4.9, 4.10 and 4.11

Critical Activities

Non-Critical Activities

Critical Path

Figure 4.5: CPM time analysis of the hypothetical proposed case study after splitting activities.

| P a g e 85
 Figure 4.6 shows LOB graph of activities C1, C2, C3 and C4 with constant

production rates before using total float to relax activity production rate and to reduce

crews employed, in which each activity employs 5 crews to be achieved. The crew is

assumed to work in a single activity not multiple (linear sequential route).

Figure 4.6: LOB of activities C1, C2, C3 & C4 before relaxing production rates

| P a g e 86
Figure 4.7: LOB schedule for activities C1-C2-C3-C4.

Figure 4.8: LOB schedule for activities D1-D2-D3-D4.

| P a g e 87
 Figure 4.9: LOB schedule for activities E1-E2-E3-E4.

Figure 4.10: LOB schedule for activities F1-F2-F3-F4.

| P a g e 88
 Figure 4.11: LOB schedule for activities G1-G2-G3-G4.

4.6. The implementation of the developed framework of the Optimization model

using MATLAB on the case study.

4.6.1. Step-1: Input module.

Table 4.2. Project related information

Project Name Pipeline Installation Project

Project Code

Number of repeated units Ten Repetitive Pipeline Installation

Project start date 0

Completion date of the project. 71

 Total number of activities (NT) of each repetitive activity, where (NT) is

equal to the number of repetitive sub-activities (Ns) multiplied by number of

repetitive unit (N) = 4 (Number of sub-activities) * 10 (Repetitive Pipelines)

= 40 activity.

| P a g e 89
4.6.2. Step-2: Defining Initial Inputs.

This step deals with identifying the initial inputs of the model by tabulating

the start date, finish date and activities numbers of repetitive activities as showing in

Appendix-B.

4.6.2.1. Application on MATLAB.

Figure 4.12 shows the translation of the input data

into code or label ranging from X1 to X40, in
accordance with the number of activities.

Figure 4.12: A screen shot for input variable data on MATLAB.

4.6.3. Step-3: Forecasting the successor and predecessor activities depending on

the input variables.

The start and finish date of each activity are arrayed and compared for

equality with the entire available corresponding activities. A final index is

conduced and displayed by applying this step on MATLAB.

4.6.4. Step-4: Sparse Matrix Module

Using the abovementioned function for this module discussed on Chapter-3

Research Methodology, sparse matrix for each activity is tackled as shown in

figure 4.13, 4.14, and 4.15 respectively.

| P a g e 91
4.6.4.1. Application on MATLAB
1- Activity-C

Figure 4.13: A screen shot for sparse matrix application on MATLAB for activity-C1, C2, C3, C4.

2- Activity-D

Figure 4.14: A screen shot for sparse matrix application on MATLAB for activity-D1,D2, D3, D4.

| P a g e 91
2- Activity-E

Figure 4.15: A screen shot for sparse matrix application on MATLAB for activity-E1, E2, E3, E4.

4.6.5. Step 5-Graph Theory Module:

The graph theory functions work on sparse matrices generated from step-4, in

which a graph theory network is rendered and displayed using MATLAB function by

identifying number of activities that are represented as nodes (vertices) and number

of lines that are called edges, which connect the nodes according to assigned

relationships between activities. The outcome shows catering out the possible and

total number of paths of each crew throughout the entire network.

4.6.5.1. Application on MATLAB

1- Activity-C1, C2, C3, C4

Figure 4.16 shows conducted graph theory functions for activity C1,C2,

C3, C4 based on the abovementioned data required, where number of nodes is 40,

whereas number of vertices is 56.

| P a g e 92
Figure 4.16: A screen shot for graph theory network application on MATLAB for activity-C1,C2,C3,C4.

2- Activity-D1, D2, D3, D4

Figure 4.17 shows conducted graph theory functions for activity D1, D2,

D3, D4 based on the abovementioned data required, where number of nodes is 40,

whereas number of vertices is 51.

Figure 4.17: A screen shot for graph theory network application on MATLAB for activity-D1,D2,D3,D4.

| P a g e 93
3- Activity-E1, E2, E3, E4

Figure 4.18 shows conducted graph theory functions for activity E1, E2,

E3, E4 based on the abovementioned data required, where number of nodes is 40,

whereas number of vertices is 69.

Figure 4.18: A screen shot for graph theory network application on MATLAB for activity-E1,E2,E3,E4.

4.6.6. Step 6- Structural Array Module:

As mentioned previously, this step is accountable for proceeding an

integration between the sparse matrix conducted from step four and the graph theory

network genertaed from step five to develop a structure array module that includes

and lists out all the possible paths for each crew routing.

4.6.6.1. Application on MATLAB.

The procedures undertaken for implementing a structural array module to

attain the objective function of this model is illustrated in chapter-3 Research

Methodology. The optimization model features a MATLAB code as a solution

| P a g e 94
algorithm with an objective of forecasting an optimum number of crews and

allocation required to maintain the crews work continuity without impacting total

project duration. Hence, this model targets minimizing the resource usage employed

to attain activities.

1- Activity-C (Crew assignment and routing based on the proposed model)

Figure 4.19 shows the total number of crews required

to attain activity C in accordance with traditional
crews' assignment methodology used in LOB. Crews
are assumed to work in a single activity not multiple.

Figure 4.19: Number of crew for activity-C after implementing the optimization model.

4.6.6.2. Output Crews routing after implementing the optimization model.

After running the optimization engine, the model resulted in a near optimum

crews' number with each crew path relying on the production rate of each activity

generated from LOB calculations. The crews number usage is minimized into 9

crews' instead of 14 crews by applying crews to work in multiple activities while

maintaining crew work continuity to meet the desired project completion date. The

path of each crew can be visualized in table 4.3. Furthermore, figure 4.20

demonstrates the codes imbedded for each activity to indicate different crew's paths

and to outcome all crews routing rendered on LOB format based on the output

generated from MATLAB.

| P a g e 95
 Table-4.3 Activities enrolled under each crew based on the model output "C".

Figure 4.21 includes two diagrams, the left

Longest Crew Path
hand side one includes the number of crews

employed after performing the optimization

model, and number of activities enrolled

under each crew. While the right hand side

Total number of crews
one includes the number of crews assigned
assigned is 9 crews
before and after relaxing production rate,

and number of crews employed after

applying the optimization model.

Codes imbedded into activities

Output crews routing

from MATLAB

Figure 4.20: LOB graph after implementation new crews routing using proposed model outputs

Number of Activities

Crews Number

Figure 4.21: Developed Approach Vs. Traditional Approach for Activity C1-C2-C-C4

| P a g e 96
 2- Activity-D (Crew assignment and routing based on the proposed model)

Figure 4.22 shows the total number of crews required

to attain activity D in accordance with traditional
crews' assignment methodology used in LOB. Crews
are assumed to work in a single activity not multiple.

Figure 4.22: Number of crew for activity-D after implementing the optimization model.

4.6.6.3. Output Crews routing after implementing the proposed optimization

model.

After running the optimization engine, the model resulted in a near optimum

crews' number with each crew path relying on the production rate of each activity

generated from LOB calculations. The crews number is minimized into 11 crews'

instead of 18 crews by applying crews to work in multiple activities while

maintaining crew work continuity to meet the desired project completion date. The

path of each crew can be visualized in table 4.4. Furthermore, figure 4.23

demonstrates the codes imbedded for each activity to indicate different crew's paths

and to outcome all crews routing rendered on LOB format based on the output

generated from MATLAB.

| P a g e 97
 Table-4.4 Activities enrolled under each crew based on the model output "D"

Figure 4.24 includes two diagrams, the left Longest Crew Path

hand side one includes the number of crews

employed after performing the optimization

model, and number of activities enrolled

under each crew. While the right hand side

one includes the number of crews assigned

Total number of crews
before and after relaxing production rate,
assigned is 11 crews
and number of crews employed after

applying the optimization model.

Codes imbedded into activities

Output crews routing From MATLAB

Figure 4.23: LOB graph after implementation new crews routing using proposed model output "D"

Number of Activities

Crews Number

Figure 4.24: Developed Approach Vs. Traditional Approach for Activity "D"

| P a g e 98
3- Activity-E (Crew assignment and routing based on the proposed model)

Figure 4.25 shows the total number of crews

required to attain activity E in accordance with
traditional crews' assignment methodology
used in LOB. Crews are assumed to work in a
single activity not multiple.

Figure 4.25: Number of crew for activity-E after implementing the optimization model.

4.6.6.4. Output Crews routing after implementing the proposed optimization

model.
After running the optimization engine, the model resulted in a near optimum

crews' number with each crew path relying on the production rate of each activity

generated from LOB calculations. The crews number is minimized into 9 crews'

instead of 13 crews by applying crews to work in multiple activities while

maintaining crew work continuity to meet the desired project completion date. The

path of each crew can be visualized in table 4.5. Furthermore, figure 4.26

demonstrates the codes imbedded for each activity to indicate different crew's paths

and to outcome all crews routing rendered on LOB format based on the output

generated from MATLAB.

| P a g e 99
 Table-4.5 Activities enrolled under each crew based on the model output "E"

Figure 4.27 includes two diagrams, the left Longest Crew Path

hand side one includes the number of crews

employed after performing the optimization

model, and number of activities enrolled

under each crew. While the right hand side

one includes the number of crews assigned Total number of crews

before and after relaxing production rate, assigned is 9 crews
and number of crews employed after

applying the optimization model.

Codes imbedded into activities

Output crews routing

Figure 4.26: LOB graph after implementation new crews routing using proposed model output "E"

Figure 4.27: Developed Approach vs. Traditional Approach for Activity "E"

| P a g e 111
4.6.7. Step-7: Crew Diagramming Approach (CDA).
The above-mentioned proposed model assigns crews to different activities in

different units. Hence, a new representational method called Crew Diagraming

Approach (CDA) is developed to realistically enhance planning of crews' circulation

and to overcome the dilemmas accompanied with the confusion due to different

crew's passing through activities in different units. CDA consistently displays an

interactive graph for each crew evolvement and route through different units

highlighted by a color to differentiate between crews and to facilitate the mapping

out of each crew passage.

4.6.7.1 Application of Crew Diagramming Approach on activity C.

Crews'1, 2 &3 path

in LOB

Activities enrolled under crew 1, 2 & 3

Figure 4.28: Developed CDA for first three crews in activity "C"

| P a g e 111
 Crews' 4 &5 path in LOB

Activities enrolled under crew 4 & 5

Figure 4.29: Developed CDA for crew 4 & 5 in activity "C"

Crews' 6, 7, 8 &9 path in

LOB

Activities enrolled under crew 6, 7, 8 & 9

Figure 4.30: Developed CDA for crew 6, 7, 8 & 9 in activity "C"

| P a g e 112
 Figure 4.28, 4.29 and 4.30 shows the interactive graph of CDA for activity

"C", in which each of the nine crews of activity "C" is continuously displayed that

enables the visualization of each crew profile as an intrinsic part of planning. Far too

often, CDA breakthrough provides a renewed opportunity for visualizing each crew

passage as a time-scaled calendar that allows each crew to move spontaneously

rather than sequentially. Furthermore, each figure includes the activities enrolled

under each crew path and the spontaneous circulation of each crew displayed on a

LOB format.

4.6.8. The assignment of multi-task skilled workforce in LOB.

Another main objective of the developed optimization model using

MATLAB is introducing the capability to assign crews in a more flexible and

efficient manner. This strategy aims to optimize the total number of crews used on

the project by increasing their employment duration on the project life. Moreover,

this strategy has been shown to improve productivity by the participation and

allocation of a crew in more than one activity in different units. These objectives

reflect the challenges of planning and scheduling in the field, and achieving those

results benefit to both the workers and the project. The capabilities of the proposed

model to assign multi-task skilling crew on LOB was feasibly implemented on

activity C and D. Figure 4.31 demonstrates the graph theory network applied on two

different activities ( C & D), where a multi task skilled crews are assigned to achieve

those two different activities. Figure 4.32 shows the rendering of the generated paths

or routes of each multi-task skilled crew as mentioned on table 4.6, which

consequently results in implementing 20 multi skilled crew to attain activity C and D

in accordance with their completion dates.

| P a g e 113
Table-4.6 Activities enrolled under each multi-skilled crew for activity C & D

Activity C&Dِِ Activity C&Dِِ

Total number of multi-

task crews assigned is 20

crew

Figure 4.31: A screen shot for graph theory network application on MATLAB for activity C & D.

| P a g e 114
 Repetitive Activity "C"

Codes imbedded into activities for

model formulation

Repetitive Activity "D"

Codes imbedded into activities for

model formulation

Figure 4.32: LOB graph after implementation multi-skilled crews using proposed model output C&D

| P a g e 115
4.7. Clearly Identifying Critical Path in LOB schedules
An additional objective of the developed optimization engine and algorithm

using MATLAB is precisely forecasting the critical path in LOB schedules

compared to conventional LOB due more accurate calculations. This algorithm can

identify any partly critical activity by figuring out the all possible longest paths in

the entire project (different critical paths in the project). This step is attained by

inserting the input values, which are start dates, finish dates and number of all

critical activities as shown in Appendix- D "The optimization model using

MATLAB algorithm". The capabilities of the proposed algorithm was implemented

using the hypothetical case study, in which critical activities are A,B, C1, D1, F1,

F2, F3,F4, G4, K and L as shown in figure 4.33.

Figure 4.33: Longest path (critical path of the case study) pinpointed using the developed algorithm

Figure 4.34 shows the graph theory network for the critical activities

rendered using the proposed algorithm that displays all possible critical paths in the

project and identifies clearly each critical activity where all possible critical paths

move through. Moreover, figure 4.35 shows one of the critical paths applied in LOB

graph of the proposed case study.

| P a g e 116
Figure 4.34: Graph theory network for all possible critical paths in the project

Figure 4.35: One of the critical paths presented in LOB format

| P a g e 117
4.8. Summary and Conclusions of this section
This section shows a new approach for scheduling repetitive projects is

proposed, in which each repetitive activity is being simulated as separate activities

with equal distances and durations. A case study of a repetitive pipeline installation

project from the literature is used to reveal the ability of this approach to offer a

detailed schedule, to generate a precise critical path, and to optimize time needed to

schedule pipeline installation by allowing the presentation of the intent relationship

between split activities

Furthermore, an proposed optimization model was introduced through

promoting an algorithm or code on MATLAB to search for the feasible crews

routing and to produce optimal crew formulation by permitting the accessibility of

crews moving in a spontaneous direction not only a linear one. A case study was

presented in order to demonstrate the ability of this model to achieve a better

allocation of crews and minimizing the number of crews utilized to schedule

activities in repetitive linear projects

| P a g e 118
4.9. The proposed Simulation Model for resource leveling in LOB using Entropy
Maximization metric with stretching or crunching activity duration.

This part deals with the developed simulation model for leveling resource using

entropy maximization theory with permitting extension of activities duration and

allowing for different production rates per each activity. As previously mentioned, this

simulation model is composed of three spreadsheet templates. 1) Activity and resource

flow template, 2) Bar chart template and 3) Maximum entropy analogy template that

are designed to systematically level resource in LOB scheduling technique. Activity

and resource flow template is designed for modeling crew's different paths,

variability in activity durations, early start, early finish and total float of each activity,

resource availability constraints, total available number of resources per each activity,

and total project duration constraint.

The implementation of the proposed simulation model is feasibility addressed

on the repetitive activity C on the previous hypothetical case study used above-

mentioned. Figure 4.36 shows the further procedures undertaken before applying the

developed model to verify the workability and applicability of this engine.

Figure 4.36: Sequential procedures performed to test the doable of the proposed model

| P a g e 119
 Figure 4.37 demonstrates the rendering of LOB graph for activity "C", LOB

calculations are transformed into a CPM format where each activity has its start and

completion date to perform resource leveling. Each activity is represented by a code

indicated as follows: For instance activity C 41, in which C is the activity name, 4 is the

unit number and 1 defines the 1st kilometer. The starting date of each activity is

conducted by dividing the activity duration with the number of crews needed for

performing the repetitive activity obtained from LOB calculation as summarized in

Appendix-C "Resource Leveling using Minimum Moment Algorithm".

Figure 4.37: LOB diagram for activity C before leveling the resources

4.9.1. Resource Leveling using Minimum Moment Algorithm.

This method assumes that the moment of the daily resource usage about the

horizontal axis of a resource histogram is a good measure of the resource utilization

and that an optimal resource allocation exists when the total moment is at a

minimum. The total resource moment Mx is calculated by summing up the individual

resource moments about the time axis (Harris, 1978).

| P a g e 111
 This method targets the reduction of the daily fluctuations in resource demand by

shifting non critical activities with their float. The overall project duration is definitely

defined as a constraint. The minimum moment method relies on the following

assumptions:

 The activities are time-continuous and, thus, once started they cannot be
interrupted;
 The resource assignments for each activity are assumed constant.
 The duration of each activity remains as originally planned;
 The logic relationship between activities and overall project duration is fixed

A resource improvement factor is computed for all activities and the activity

producing the largest positive improvement factor is shifted, this process is iterated. The

improvement factor is given by

Where m=minimum between the time-units by which an activity is to be shifted

and the activity duration; R=resource rate; x=sum of daily resources to which m daily

resource units R are to be deducted; and w=sum of daily resources to which m daily

resource units R are to be added.

4.9.1.1. Application of Resource Leveling using Minimum Moment Algorithm on

Repetitive Activity "C".

As mentioned earlier LOB graph of repetitive activity C is translated into a CPM

format to level activities using above mentioned procedures of the minimum moment

algorithm. The procedures for implementing minimum moment algorithm on activity C

can be overviewed in Appendix-C "Resource Leveling using Minimum Moment

Algorithm". Figure 4.38 shows the enhancement in the resource usage profile compared

to that before performing leveling using minimum moment algorithm.

| P a g e 111
 Furthermore, the number of crews demanded to attain repetitive activity C is

reduced from 14 to 12 crews by assuming that crews work in a single activity not

multiple one in LOB as displayed in figure 4.39.

Figure 4.38: Resource leveling histogram after utilizing minimum moment algorithm on activity C

Figure 4.39: LOB representation of Activity "C" after leveling using minimum moment algorithm

| P a g e 112
4.9.1.2. Application of proposed optimization model on activity "C" after

leveling with minimum moment algorithm.

Figure 4.40 shows graph theory network obtained after implementing the

optimization engine using MATLAB. Graph theory network presents all possible paths

of crews, which consequently results in generating and allocation each crew route as

previewed in figure 4.41.

Figure 4.40: Graph theory network of Activity "C" after leveling using the optimization model

| P a g e 113
Fig. 4.41. LOB graph for activity C after implementation new crews routing using optimization model

4.9.2. Implementation of proposed simulation model for leveling resources using

entropy maximization on activity "C".

1- Activity and resource flow template

In this template, the following inputs are to be provided to the model.

 Different crews routing and activities enrolled in each crew as shown in figure
4.42.

 Early start, early finish of each activity as pointed out from LOB graph as shown
in figure 4.42.

 Available number of resources per period for each activity, which is 5 resources
per day as shown in resources column in figure 4.42.

 The total sum of number of resources per each activity, which is 20 resource per
each activity (daily resource multiplied by activity duration).

 Indicating the total float for each activity to define the permeable stretching and
splitting duration per each activity.

| P a g e 114
 Figure 4.42 illustrates the activity and resource flow template used in the

simulation process. The discrete uniform distribution of each activity with their

parameters is explained deeply in Appendix-D.

Resource assumption are defined

as an assumption with a discrete
uniform distribution

Fig. 4.42. A screen shot for the activity and resource flow template of activity "C"

2- Bar chart template

Figure 4.43 shows the bar chart template representing the stretching or

crunching of each activity duration according to the assumed resource assignment

that are randomly selected from discrete uniform distributions with a minimum and

maximum value in accordance resource type and based on experience. Nevertheless,

in this template, daily resource usage is displayed accompanied with its entropy,

which is calculated as follows:

 Entropy of daily resource usage = Rd * Ln ( Rd / RT )

 Total Entropy of an activity = Summation of total activities entropy * (-1/ RT)

| P a g e 115
 Where Rd is the daily resource usage; RT is the total number of resource of an

activity. The total project duration, total activity entropy and total number of resource

of an activity are defined as a forecast probability, this is extensively illustrated in

Appendix-D. A total of 500,000 simulation cycles are run on Crystal Ball Ribbon

Simulation Software.

Figure 4.43. A screen shot for bar chart template of activity "C"

3- Maximum entropy analogy template

In this template, two types of an output are generated, the variable output

which is the total entropy of the system, and the constant output that is the total

resources of the activity and project duration. Moreover, the simulation results are

filtered for values with a summation equal to the total number of available resources

for the all activities and planned total activity duration as previewed in figure 4.44.

Figure 4.45 shows daily resource usage profile before leveling, after leveling using

minimum moment algorithm, and after leveling using entropy maximization metric.

Figure 4.46 renders the application of the resource usage leveling using entropy

maximization metric in LOB graph.

| P a g e 116
Entropy values Total resource of an activity Activity completion date Resource assignment /day

Figure 4.44. A screen shot for simulation outcomes of an activity

"C"

Figure 4.45. Daily resource usage after implementing entropy maximization

| P a g e 117
 Figure 4.46. LOB graph of activity C after leveling using entropy maximization metric with different
production rates among each activity
4.10. Summary and Conclusions of this section
A recap on the minimum moment method was presented, that was

implemented to level, to fluctuate resource usage profile and to minimize number of

crews utilized. This study presents a flexible simulation model for handling the

resource leveling dilemmas for linear construction projects. The proposed model was

constructed using crystal ball ribbon based on an entropy-maximization approach.

The entropy-maximization method accounts for such possibility of allowing activity

to be stretched without affecting total completion date of a project. From the resource

leveling perspective, results generated using minimum moment algorithm and

proposed model are compared that reveal an enhancement in the resource usage

profile, and a reduction in the maximum resource usage per day from 40 to 32 using

the proposed simulation model. Moreover the total entropy of the system before

applying the developed model is 3.51, where as it became 3.6 after applying the

system which reveals the smoothness of the resource usage profile and minimizes

daily resource usage per day in case of the occurrence of limitation in resources.

| P a g e 118
 CHAPTER 5 - CONCLUSIONS AND RECOMMENDATIONS

This is the last chapter of the write-up, which summarizes research outcomes

reached up to this point. It recaps the limitations that was proposed and was

attempted to be remedied in this thesis. Then, it outlines the main findings of this

thesis and shows the main contributions to the knowledge. Finally, it highlights

doable future research that relates to the topic.

5.1. Research Overview

Scheduling of repetitive projects represents a great challenge for construction

scheduler and planner. Traditional CPM analysis does not suit properties of

repetitive projects. Although the LOB scheduling technique has been widely used

for scheduling linear repetitive construction projects, the fundamental principles of

LOB have several shortfalls that raise many concerns about LOB, which need to be

attuned and improved in order to comply with the nature of construction projects.

The limitations of LOB and solutions are described below.

 Assuming the constancy, identically and linearity of the production rates and

amount of work of an activity along the entire units, which is did not correlate

well with real life projects.

 Regardless of being an essential tool for project scheduling, LOB technique is

imprecise in detecting and pinpointing criticalness and floats; thus, LOB

method requires to be developed more to single out critical, non-critical

activities and floats.

 In LOB scheduling technique, the perception that crews continuously move in

a linear vertical way only may be impractical. Linear repetitive projects

encounters resource-constrained problem when there are limits on the

availabilities of resources. Hence, LOB needs the occurrence of a mechanism

| P a g e 119
 or an algorithm as a decision making tool to provide an optimum crews

routing and distribution and to implement multi-task skilled crews between

different activities in different units.

 Construction projects involve circumstances that need the use of visual color

coded diagrams or interactive graphs for rendering each crew path to easily

pinpoint different crews allocation and to single out concurrent activities

attained at same instance.

 The application of the entropy maximization as a metric used to level

resources in linear projects has not been previously taken into account.

As a result, there is a need for a practical and effective solution that

guarantees a proper road map to accommodate the recent discussed limitation in

LOB. The main rationale behind this, is developing a complete coherent framework

for planning and scheduling repetitive projects where LOB scheduling technique is

utilized. This proposed approach will better handle the following:

 Identifying the practical areas of potential enhancement that can improve the

LOB scheduling tool by accounting for the breaking down of activities into

sub-activities in order to formulate more precise linear schedules

 Supplementing the repetitive scheduling literature with a new algorithm and a

decision supporting system that permits the evolvement and allocation of

crews in a spontaneous routing between different activities in different units

rather than sequential, and further, the crews working may be switched from linear

vertical to spontaneous.

 The flexibility of implementing multi-task skilled crew in linear repetitive

construction projects between different activities in different units that can

lead to increased productivity, flexibility, and work continuity.

| P a g e 121
  Conducting a new resource leveling tool based on entropy maximization

metric using a simulation model temple by permitting for activity duration to

be extended or crunched to ensure minimizing resource fluctuations over time.

In addition, the methodology of breaking down of activities into sub-activities

in order to formulate more efficient linear schedules that guarantees a proper

criticalness and float analysis in LOB scheduling technique. Furthermore, the

proposed framework is capable of minimizing resource usage, identifying optimum

crews' route, different crews' allocation, and applying multi-task skilled using

MATLAB R2013a, in which an algorithm as an optimization engine was built-in to

attain the main objectives of the proposed model. The developed framework was

illustrated using a real case study of installing ten identical sewage pipelines. A

comparison was done between conventional LOB scheduling and proposed approach

in order to track the different in crews' number, their formulation and to improve the

efficiency in crews' utilization. The application of the model has shown a tremendous

reduction in crews' utilization by approximately 36% from overall required number of

crews in term of using traditional method, and representing a new formulation for

continuous crews' usage in a spontaneous manner. Moreover, the model allowed the

usage of multi-task crew in LOB; hence, cost efficiency of the entire project was

enhanced.

Moreover, a computerized model is developed to minimize the total resource

usage of a project by allowing the expanding or shrinking of activity durations and

having different crews' size. A spreadsheet modeling tool is established using three

template (activity and resource flow template, bar chart template, and entropy

maximization template). A simulation engine was run using crystal ball ribbon

| P a g e 121
software. The same abovementioned case study was applied to figure out and test the

validation of the proposed model

5.2. Conclusion and research contribution.

The new approach for scheduling multi repetitive projects has the potential to

revolutionize the way linear schedules are generated and managed. Based on the

current development, this research makes a number of contributions:

 Better understanding of linear construction scheduling needs: This study has

provided an in-depth review of the research reported in literature with respect

to the theoretical and practical drawbacks of LOB. Based on these shortfalls,

areas in LOB scheduling procedures that require enhancement have been

identified.

 Better representation of the intention of activity relationships; as a result, a

better identification of criticalness and floats in LOB are attained, and precise

critical path is identified.

 Introducing a decision-making, supporting, optimization engine that deals with

different crew allocation and formulate the optimum crews routing to attain

near optimum crew usage and allocation.

 The usability and extensibility of implementing multi- task skilled crews in

linear construction projects between different activities in different units.

 Minimizing the total resource usage of a project, while meeting the

requirements for work continuity and the target deadline of each activity by

promoting the stretching or crunching of an activity duration and altering

crew's sizes.

| P a g e 122
Recommendation for Future Research

Several aspects of the proposed approach could be improved through further

research. The following areas are recommended for further study in order to

enhance the capability of proposed approach in scheduling linear projects so that

it will be more practical for real construction projects:

 Resource leveling using entropy maximization metric could be converted into

a heuristic algorithm that can be implemented without the use of simulation.

 The effects of multiple crew assignments per activity on productivity loss and

duration.

 The impact of weather and other global factors on crews' productivity could be

incorporated in the entropy model in order to be more realistic, and more

effective.

 An algorithm could be developed to accurately forecast the minimum and

maximum daily resource usage without out affecting the productivity.

 Features related to financing decisions and project control could be

incorporated into the proposed algorithm using MATLAB, such as cash flow

analysis, earned value, earned schedule, and productivity analysis.

| P a g e 123
References
Agrama.F, 2011. Linear projects scheduling using spreadsheets features. Alexandria
Engineering Journal .Volume 50, Issue 2, June 2011, Pages 179–
185.dx.doi.org/10.1016/j.aej.2011.01.018

Ammar, M., and Mohieldin, Y. (2002). “CPM/RSM: CPM-Based repetitive

scheduling method.” III Middle East Regional Civil Engineering Conf., Egypt
Section, ASCE, Reston, VA.

Ammar, M. (2003). “Float analysis of non-serial repetitive activities.” J. Constr.

Manage. Econ., 21(5), 535–542.

Ammar, M. (2012). "LOB and CPM Integrated Method for Scheduling Repetitive
Projects". Journal of Construction Engineering & Management. DOI: 10.1061/
(ASCE) CO.1943-7862.0000569

Arditi, D., and Albulak, M. Z. (1986). "Line-of-balance scheduling in pavement

construction." J. Constr. Engrg. and Mgmt., ASCE, 112(3), 411-424.

Arditi, D., Tokdemir, O. B., and Suh, K. ~(1999)!. ‘‘Effect of learning on line-of-
balance scheduling.’’ Int. J. Proj. Manage., 15~5!, 265–277.

Arditi, D., Sikangwan, P., & Tokdemir, O. (2002). Scheduling System for High Rise
Building Construction. Construction Management and Economics, 20. Retrieved
November 15, 2007, from the Business Source Premier.

Choi, J., and Russell, J. S. _(2005)_. “Long-term entropy and profitability change of
United States public construction firms.” J. Manage. Eng., 21_1_, 17–26.

Christodoulou, S., Ellinas, G., and Aslani, P. _(2009)_. “Disorder considerations in

resource-constrained scheduling.” Constr. Manage. Econom., 27_3_, 229–240.

Christodoulou, S., Ellinas, G., Aslani ,p. (2009). Entropy-based scheduling of

resource-constrained construction projects. Automation in Construction.
doi:10.1016/j.autcon.2009.04.007

Christodoulou,S., Ellinas, G., Kamenou, M. (2010). "Minimum Moment Method for

Resource Leveling Using Entropy Maximization". Journal of Construction
Engineering and Management. DOI: 10.1061/_ASCE_CO.1943-7862.000014

D. Arditi, O.B. Tokdemir, K. Suh, Scheduling system for repetitive unit construction
using line-of-balance technology, Eng. Constr. Archit. Manag. 8 (2) (2001) 90– 103.

Dickson, A. (2006). "Introduction to Graph Theory". University of Utah.

| P a g e 124
Damci, A., Arditi, D., Polat, Gul. 2013." Multiresource Leveling in Line-of-Balance
Scheduling". J. Constr. Eng. Manage .DOI: 10.1061/(ASCE)CO.1943-7862.0000716

Doulabi, S. H. H., Seifi, A., and Shariat, S. Y. (2011). "Efficient Hybrid Genetic
Algorithm for Resource Leveling via Activity Splitting." Journal of Construction
Engineering and Management, 137(2), 137-146.

Dubey, A. (1993). “Resource leveling and linear scheduling.” M.Sc. thesis, Dept. of
Civil Engineering, Univ. of British Columbia, Vancouver, Canada.

El-Rayes, K., and Moselhi, O. (2001) “Optimizing Resource Utilization for Reptitive
Construction Projects”, Journal of Construction Engineering and Management, ASCE,
127(1), 18-27.

Elwany, M. H., Korish, I. E., Barakat, M. A., and Hafez, S. M. (1998). “Resource
smoothing in repetitive projects.” Comput. Ind. Eng., 35(3–4), 415–418

Georgy, M. E. (2008).“Evolutionary resource scheduler for linear projects.”Automate.

Constr., 17(5), 573–583.

Hariga, M. and El-Sayegh, S (2011). " Cost Optimization Model for the Multiresource
Leveling Problem with Allowed Activity Splitting ". (doi: 10.1061/(ASCE)CO.1943-
7862.0000251)

Harmelink, D. J., and Rowings, J. E. ~1998!. ‘‘Linear scheduling model:

Development of controlling activity path.’’ J. Constr. Eng. Manage.,124~4!, 263–268.

Harris, F., and McCaffer, R., Modern Construction Management, Crosby Lockwood
Staples, London, U.K., 1977.

Harris, F. and McCaffer, R.(1989) Modern construction management, London.

Harris, R. B., and Ioannou, P. G. (1998), “Scheduling Projects with Repeating

Activities”, Journal of Construction Engineering and Management, ASCE, 124(4),
269-278.

Hegazy, T., Moselhi, O., and Fazio, P., ~1993!. ‘‘BAL: An anlgorithm for scheduling
and control of linear projects.’’ AACE Transactions, Morgantown,
W. Va., 8.1–8.14.

Hegazy, T. and Ersahin, T. (2001). ”Simplified Spreadsheet Solutions. I:

Subcontractor Information System.” J. Constr. Eng. Manage., 127(6), 461–468.

Hegazy. and Wassef , N., 2001.Cost Optimization in Projects with Repetitive Non
serial Activities. Journal of Construction Engineering and Management.127 (3), pp.
183-191

Hinze, J.2004 .Construction planning and scheduling, 2nd Ed., Prentice-Hall, Upper
Saddle River, N.J
| P a g e 125
Hyari,K., and El rayes,K.,2006. Optimal Planning and Scheduling for Repetitive
Construction Projects. Journal of Management in Engineering. Volume 22, Issue 1
(January 2006). dx.doi.org/10.1061/(ASCE)0742-597X(2006)22:1(11).
Johnston, D.W. (1981), “Linear Scheduling Method for Highway Construction”,
Journal of the Construction Division, ASCE, 107(CO2), 247-261.

Jung.S. Lin.Y., and Wang, L., 2013. Methodology for Construction Scheduling
Management. Hok kaido University Collection of Scholarly and Academic Papers:
HUSCAP

Karaa, F., and Nasr, A. 1986. “Resource management in construction.” Journal

Constr. Eng., 1123, 346–357

Khisty, C. J. (1970). "The application of the 'Line of Balance' technique to the

Construction industry." Indian Concrete J., (July), 297-320.

Landsberg, P. _1984_. “Is equilibrium always an entropy maximum?” J. Stat. Phys.,

35, 159–169.

Leu, S., Yang, C. and Huang, J., (2000). “Resource Leveling in Construction by
Genetic Algorithm-Based Optimization and Its Decision Support System
Application”, Automation in Construction, Vol. 10, pp. 27-41.

Liu, S. and Wang, C. (2007). "Optimization model for resource assignment problems
of linear construction projects." Automation in Construction, 16(4), 460-473.
Online publication date: 1-Jul-2007
Liu, S. and Wang, C. (2012). "Optimizing linear project scheduling with multi-skilled
crews." Automation in Construction, 24, 16-23. Online publication date: 1-Jul-2012.
Lucko. (2011)." Integrating Efficient Resource Optimization and Linear Schedule
Analysis with Singularity Functions". Journal of Construction Engineering &
Management. DOI: 10.1061/_ASCE_CO.1943-7862.0000244

Lumsden, P., The Line-of Balance Method, Pergamon Press, London, U.K., 1968

Lutz, J. D. (1990), Planning of Linear Construction Projects using Simulation and

Line Of Balance, Ph.D. Thesis, Purdue University, West Lafayette, IN.

Lutz, J.D., and Halpin, D.W. (1994), “Simulation of Learning Development in

Repetitive Construction”, Journal of Construction Engineering and Management,
ASCE, 120(4), 753-773.

M. Hsie, C.J. Chang, I.T. Yang, C.Y. Huang, Resource-constrained scheduling for
continuous repetitive projects with time-based production units, Automation in
Construction 18 (7) (2009) 942 –949.

| P a g e 126
Mattila, K., and Abraham, D.1998. “Resource leveling of linear schedules using
integer linear programming.”J. Constr. Eng. Manage., 1243, 232–24

Mohammad A. Ammar (2012). INTEGRATED LOB AND CPM METHOD FOR

SCHEDULING REPETITIVE PROJECTS. Journal of Construction Engineering and
Management. accepted April 5, 2012.

Neale, R. H., and Neale, D. E. (1989). Construction planning. 1st Ed., Thomas
Telford Ltd., London, England.

O’Brien, J.J. (1985), “VPM Scheduling for High-Rise Building”, Journal of the
Construction Division, ASCE, 101(CO4), 895-905.

Psarros, M. E. ~1987!. ‘‘SYRUS: A program for repetitive projects.’’ Master’s thesis,

Dept. of Civil Engineering, Illinois Institute of Technology, Chicago.

Reda, R. M., (1990), “RPM: Repetitive Project Modeling”, Journal of Construction

Engineering and Management, ASCE, 116(2), 316-330.

Resource Allocation for Repetitive Construction Schedules: An Ant Colony

Optimization Approach Mohamed A. El - Gafy, Ph.D., P.E., M.A.I.I llinois State
University, Illinois

Ruan, Y., Wang, J., Dong, W., and Jie, M. (2009) Optimum Choice for the
Construction Management Plan Based on Entropy Weight. ICCTP 2009: pp. 1-7. doi:
10.1061/41064(358)16

Russell, A. D., and Wang, W. C. M. (1993). "New generation of planning structures."

J. Constr. Engr and Mgmt., ASCE, 119(2), 196-214.

Russell, J. S., and Radtke, M. W. (1991). "Subcontractor failure. Case history."

AACE International Transactions, E.02.1-E.02.6

S. Christodoulou, “Scheduling Resource-Constrained Projects with Ant Colony

Optimization Artificial Agents,” Journal of Computing in Civil Engineering, vol. 24,
no. 1, January 2010.

Selinger, S., "Construction Planning for Linear Projects," Journal of the Construction
Division, ASCE, Vol. 106, No. C02, June, 1980, pp. 195-205.

Senouci, A. B., and Eldin, N. N. ~1996!. ‘‘Dynamic programming approach to

scheduling of nonserial linear project.’’ J. Comput. Civ. Eng., 10~2!, 106–114.

Senouci, A. and Adeli, H. (2001). ”Resource Scheduling Using Neural Dynamics

Model of Adeli and Park.” J. Constr. Eng. Manage., 127(1), 28–34.

| P a g e 127
Senouci, A., and Naji,K.,2006. A Computerized System for Scheduling and Cost
Optimization of Non-serial Linear Projects. Surveying and Built Environment Vol
17(2), 49-62 December 2006 ISSN 1816-9554

Son, J., and Skibniewski, M. 1999. “Multiheuristic approach for resource leveling
problem in construction engineering: Hybrid approach.” J. Constr. Eng. Manage. , 125
1, 23–31.
Sonmez, R. and Uysal, F. (2014). "Backward-Forward Hybrid Genetic Algorithm for
Resource-Constrained Multiproject Scheduling Problem." J. Comput. Civ. Eng. ,
10.1061/(ASCE)CP.1943-5487.0000382 , 04014072.

Suhail, S.A., and Neale, R.H., (1994) “CPM/LOB: New Methodology to Integrate
CPM and Line of Balance”, Journal of Construction Engineering and Management,
ASCE, 120(3), 667-684.

Tang, C., Leung, A., and Lam,K., (2006)." Entropy Application to Improve
Construction Finance Decisions". Journal of Construction Engineering and
Management.DOI:10.1061/_ASCE_0733-9364_2006_132:10_1099_

Tang, Y., Liu, R., and Sun, Q. (2014). ”Two-Stage Scheduling Model for Resource
Leveling of Linear Projects.” J. Constr. Eng. Manage., 140(7), 04014022.

Trimble, G. (1984). "Resource-oriented scheduling." Proj. Mgmt. J., Project

Management Institute, 2(2), 70-74

Wang, C., and Huang, Y. ~1998!. ‘‘Controlling activity interval times in LOB
scheduling.’’ Constr. Manage. Econom., 16~1!, 5–16.

Yang, I. (2002), Repetitive Project Planner Resource-Driven Scheduling for

Repetitive Construction Projects, Ph.D. Dissertation, University of Michigan, Ann
Arbor, MI.

Yang, I; Ioannou, G (2004). "Scheduling system with focus on practical concerns in

repetitive projects". Construction Management and Economics". DOI:
http://dx.doi.org/10.1080/01446190310001649065

Yen's C.I. (2005) simulated annealing for optimizing linear scheduling projects with
multiple resource constraint. PhD thesis, Purdue University

| P a g e 128
APPENDICIES

| P a g e 129
Appendix A- Calculation Procedures of

integrated CPM and LOB

| P a g e 131
A.1. Calculation of desired rate of delivery

Rd= N-1/Tp-T1+Tf, Where N= number of repetitive units, Tp is the desired

Completion date of the whole project, T1 is the completion date of one single unit,

and Tf is the total float of noncritical activities to reduce the number of crews

employed and to relax production rate of an activity (Suhail and Neale 1994).

A.2. Calculation of number of crews required

 Cdi= di * Rdi, where di is the duration of an activity in one single unit

 Cai =rounded up Cdi

A.3. Calculation of actual production rate

 Rai= Cai / di .Where Cai is the actual number of crew needed to perform a

specific repetitive activity, and Rai is the actual rate of progress of an activity

A.4. Calculation of each activity duration

Computing the duration of an activity i over all units (Di), in which STiN =

start time of last unit; STi1 = start time of first unit; and Di= duration along all units

of activity i. Di =di + STiN − STi1 = di+ (N−1) / Ra.

A.5. Specifying logical relationships

Specifying logical relationships using overlapping activities by comparing the

actual progress rate of predecessor and successor activities. If the actual progress rate

of Predecessor is greater than actual progress rate of successor, then a start to start

relationship will be applied plus buffer time. If the actual progress rate of Predecessor

is less than actual progress rate of successor, then a finish to finish relationship will

be applied plus buffer time. Finally, if the actual progress rate of Predecessor is equal

to actual progress rate of successor, then a start to start relationship will be applied

plus buffer time (Ammar, 2013).

| P a g e 131
A.6. Performing a time scheduling calculations

Performing a time scheduling calculations using CPM. Forward pass

calculations are conducted to determine early timings of activities, whereas late

timings of activities are determined in the backward pass calculations.

A.7 Calculation Procedures of integrated CPM and LOB for the hypothetical
case study using conventional approach

Table A-1. LOB calculation of the proposed case study using conventional approach.

Table A-2. Computing the duration of repetitive activities of the case study.

| P a g e 132
A.8. Calculation Procedures of integrated CPM and LOB for the hypothetical
case study using the new approach
Table A-3. LOB calculation of the proposed case study using new approach

| P a g e 133
Table A-4. Computing the duration of repetitive activities of the case study.

| P a g e 134
 Appendix B- The optimization

model using MATLAB Algorithm

| P a g e 135
B.1. Defining initial input values for activity C (Start, finish date& Act. No.)

| P a g e 136
B.2. Defining initial input values for activity D (Start, finish date& Act. No.)

| P a g e 137
B.3. Defining initial input values for activity E (Start, finish date& Act. No.)

| P a g e 138
B.4. Defining initial input values for activity F (Start, finish date& Act. No.)

| P a g e 139
B.5. Defining initial input values for activity G4, K and L (Start, finish date&
Act. No.)

| P a g e 141
B.6. Defining initial input values for activity E (Start, finish date& Act. No.) after
leveling using minimum moment algorithm.

| P a g e 141
B.7. Code or algorithm of MATLAB optimization model

Screen shot for sparse matrix code on MATLAB

| P a g e 142
 Screen shot for graph theory code on MATLAB

Screen shot for structural array code on MATLAB

| P a g e 143
Screen shot for crew formulation code on MATLAB

| P a g e 144
Appendix C- Resource leveling using

minimum moment algorithm

| P a g e 145
C.1. Computational procedures of the improvement factors for activity C

| P a g e 146
An overview for the bar chart used in leveling resources using minimum moment algorithm

| P a g e 147
Appendix D- Resource leveling using

entropy maximization metric

| P a g e 148
D.1. Assumptions and Forecasts summary reported from the simulation model using crystal
ball ribbon

D.2. Defining the project completion date as a forecast

| P a g e 149
D.3. Defining the total system entropy as a forecast

| P a g e 151
D.4. Defining the total number of resources as a forecast

| P a g e 151
 D.5. Resource assignment per day is defined as an assumption with a discrete uniform
distribution (minimum and maximum value) based on the total float of each activity

| P a g e 152

View publication stats