improving hoist performance during the up-peak of tall ... · pdf fileimproving hoist...

123
Improving Hoist Performance during the Up-Peak of Tall Building Construction by Mohamed Kamleh A thesis submitted in conformity with the requirements for the degree of Master of Applied Science Civil Engineering University of Toronto © Copyright by Mohamed Kamleh 2014

Upload: dinhdang

Post on 17-Mar-2018

215 views

Category:

Documents


1 download

TRANSCRIPT

Improving Hoist Performance during the Up-Peak of Tall Building Construction

by

Mohamed Kamleh

A thesis submitted in conformity with the requirements for the degree of Master of Applied Science

Civil Engineering University of Toronto

© Copyright by Mohamed Kamleh 2014

ii

Improving Hoist Performance during the Up-Peak of Tall Building

Construction

Mohamed Kamleh

Master of Applied Science

Civil Engineering

University of Toronto

2014

Abstract

Purpose: With the increased demand for tall buildings, it has become crucial to study current

construction methods with respect to this emerging construction environment. The increase in

height of buildings produces difficulties in the vertical delivery of resources. This study will

examine the hoist’s (a temporary construction elevator) performance and its impact on worker

delays.

Approach: First, a discrete event simulation model using Simphony.Net software was developed

to represent the morning delivery of workers. Data from site observations, manufacturers’ data,

and expert opinion were collected and incorporated. The model was verified and validated. Then

alternative strategies for hoist management were studied.

Findings: A combination of staggered arrivals and zoning for hoist operations have shown to

provide hoist performance improvements by reducing the waiting time of workers.

Value: This research tests new methods to decrease the waiting time of workers, and improve

the hoist’s performance.

iii

Acknowledgments

I would like to take this opportunity to graciously thank Professor Brenda McCabe for her

constant support, guidance and mentorship. I would like to acknowledge her dedication to this

project and my education, and her endless patience.

I would like to extent to thank my parents and siblings for their support and love. They have

helped me stay on track and provided me with the utmost support during times of frustration. I

would most certainly like to extent a special thanks to my sister, May, for her tremendous

support.

I would like to extend a special thanks to Nelly Pietropaolo and the entire student services staff

for making my experience remarkable, for their help and support throughout this endeavor.

Sincere appreciation goes to those who helped make this research as success: Dr. Simaan

AbouRizk at University of Alberta for allowing free use of his simulation modeling software,

Simphony.Net; Sam, Gokul, and Steve at Daniels Group for their enthusiastic support and time.

Finally, I would like to thank my friends for being there for me throughout my education. They

helped guide and motivate me and were there for me when I needed them.

iv

Table of Contents

Acknowledgments .......................................................................................................................... iii

Table of Contents ........................................................................................................................... iv

List of Tables ............................................................................................................................... viii

List of Figures ................................................................................................................................. x

List of Appendices ....................................................................................................................... xiii

Chapter 1 Introduction .................................................................................................................... 1

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

1.2 Motivation ........................................................................................................................... 4

1.3 Research Objective ............................................................................................................. 5

1.4 Scope ................................................................................................................................... 5

1.5 Research Contributions ....................................................................................................... 6

1.6 Study Methodology ............................................................................................................. 6

1.7 Thesis Organization ............................................................................................................ 7

Chapter 2 Elevator and Hoist Planning ........................................................................................... 9

2.1 Elevator planning ................................................................................................................ 9

2.1.1 Elevators vs. Hoists ............................................................................................... 11

2.2 Hoist Operation during Up-Peak ...................................................................................... 12

2.2.1 Factors Affecting Hoist Performance ................................................................... 13

2.3 Hoist planning methods .................................................................................................... 15

2.3.1 Summary of Limitations of Current Methods ....................................................... 20

2.4 Summary of factors impacting hoist operation ................................................................. 21

Chapter 3 Analysis Method .......................................................................................................... 22

3.1 Numerical methods ........................................................................................................... 22

3.1.1 Linear Models ....................................................................................................... 22

v

3.1.2 Regression ............................................................................................................. 23

3.1.3 R-squared .............................................................................................................. 24

3.2 Simulation ......................................................................................................................... 25

3.2.1 Monte Carlo Simulation ........................................................................................ 25

3.2.2 Discrete Event Simulation .................................................................................... 26

3.3 Characteristics of Methods ............................................................................................... 26

3.4 Selected Method: Discrete-Event simulation .................................................................... 27

3.4.1 DES in Construction ............................................................................................. 28

3.4.2 Limitations of DES ............................................................................................... 29

3.4.3 DES Software in Construction .............................................................................. 29

3.4.4 Applicability of Method to Hoist Operation ......................................................... 29

Chapter 4 Development of the Proposed Model ........................................................................... 31

4.1 Introduction ....................................................................................................................... 31

4.2 Simphony.Net ................................................................................................................... 31

4.2.1 Main Interface ....................................................................................................... 32

4.2.2 Modelling Elements .............................................................................................. 33

4.2.3 Simphony’s modelling distributions ..................................................................... 36

4.3 Model Uses ....................................................................................................................... 38

4.3.1 Current operation of the hoist ............................................................................... 39

4.3.2 Alternative strategy for improving hoist performance .......................................... 40

4.3.3 Comparison of the arrival time functions ............................................................. 41

4.4 Model description ............................................................................................................. 41

4.4.1 User-Input component .......................................................................................... 44

4.4.2 Arrival of workers ................................................................................................. 46

4.4.3 Loading of the Hoist ............................................................................................. 49

4.4.4 Hoist operation ...................................................................................................... 52

vi

4.4.5 Output generation .................................................................................................. 61

4.5 Decisions during Model Development ............................................................................. 63

4.6 Model Factors ................................................................................................................... 65

4.6.1 Factors as user-inputs ............................................................................................ 66

4.6.2 Factors built into the model .................................................................................. 66

4.7 Model Output .................................................................................................................... 67

4.8 Model Scenarios ................................................................................................................ 68

4.9 Scenario components ........................................................................................................ 69

4.9.1 Model entities ........................................................................................................ 69

4.9.2 Model resources .................................................................................................... 69

4.9.3 The model programming ....................................................................................... 69

4.9.4 Model variables ..................................................................................................... 70

4.10 Model Verification and Validation ................................................................................... 71

4.10.1 Model verification ................................................................................................. 71

4.10.2 Validation .............................................................................................................. 72

4.11 Planning Options ............................................................................................................... 73

4.11.1 Site characteristics ................................................................................................ 73

4.11.2 Hoist characteristics .............................................................................................. 73

4.11.3 Stage of construction ............................................................................................. 73

4.11.4 Worker schedules .................................................................................................. 73

4.12 Chapter Summary ............................................................................................................. 73

Chapter 5 Impact of Model Inputs ................................................................................................ 75

5.1 Distribution of the arrival of workers ............................................................................... 75

5.2 Inter-arrival rates of workers ............................................................................................ 76

5.3 Impact of Hoist Characteristics ......................................................................................... 78

5.4 Impact of the number of workers and number of floors ................................................... 79

vii

Chapter 6 Using the Model to Improve Hoist Performance ......................................................... 84

Chapter 7 Conclusion and Recommendations .............................................................................. 91

7.1 Conclusions ....................................................................................................................... 91

7.2 Limitations of the study .................................................................................................... 92

7.3 Recommendations ............................................................................................................. 93

Bibliography ................................................................................................................................. 95

Appendices .................................................................................................................................. 101

viii

List of Tables

Table 1: Sample hoist speeds and capacities ................................................................................ 14

Table 2: Simple formulaic method for hoist planning .................................................................. 15

Table 3: Factors affecting hoist operation .................................................................................... 21

Table 4: Characteristics of Analysis Methods .............................................................................. 26

Table 5: Description of Simphony.Net modelling elements ......................................................... 33

Table 6: Example of alternative strategy schedule using step function ........................................ 40

Table 7: Difference between arrival of workers in the model scenarios ...................................... 41

Table 8: Example algorithm for input assignment ........................................................................ 45

Table 9: Example algorithm for assigning durations according to step function ......................... 47

Table 10: Sample algorithm for assigning hoist ID ...................................................................... 55

Table 11: Sample algorithm for directing the hoist ...................................................................... 55

Table 12: Sample algorithm for counting hoist cycles ................................................................. 56

Table 13: Sample algorithm for assigning travel durations .......................................................... 56

Table 14: Sample algorithm for tracking hoist travel ................................................................... 58

Table 15: Sample algorithm for checking if it is the final stop .................................................... 58

Table 16: Sample algorithm for checking if there is a stop .......................................................... 59

Table 17: Sample algorithm for resetting variables ...................................................................... 61

Table 18: Sample algorithm for trace data generation .................................................................. 62

Table 19: Input variables categories ............................................................................................. 65

ix

Table 20: Values of factors built into the model ........................................................................... 67

Table 21: Definition of model variables ....................................................................................... 70

Table 22: Validation through case studies .................................................................................... 72

Table 23: Summary of impact studies .......................................................................................... 75

Table 24: Summary of R-Squared values for different fits ........................................................... 81

Table 25: Inputs used in the analysis ............................................................................................ 84

Table 26: Arrival schedule details ................................................................................................ 85

Table 27: Cumulative delay for 50% and 80% of workers ........................................................... 90

x

List of Figures

Figure 1: Number of high-rise constructed per year in Toronto ..................................................... 3

Figure 2: Construction crane (left) and hoist (right) ....................................................................... 4

Figure 3: Decision variables (left) and planning method (Right) survey. .................................... 16

Figure 4: Simphony.Net’s main interface components ................................................................. 33

Figure 5: Uniform distribution ...................................................................................................... 37

Figure 6: Normal distribution ....................................................................................................... 37

Figure 7: Exponential distribution layout ..................................................................................... 38

Figure 8: Step function layout ....................................................................................................... 38

Figure 9: Flow chart of method used for modelling the hoist operation ...................................... 39

Figure 10: Hierarchy of the model ................................................................................................ 42

Figure 11: Sample scenario layout ................................................................................................ 43

Figure 12: Current operation scenario model inputs ..................................................................... 44

Figure 13: Procedure for defining model inputs ........................................................................... 45

Figure 14: Alternative strategy scenario model inputs ................................................................. 46

Figure 15: Model elements representing arrival of workers ......................................................... 49

Figure 16: Model elements representing loading of one hoist ...................................................... 50

Figure 17: Model elements representing loading of two hoists .................................................... 50

Figure 18: Model elements describing the loading of one hoist ................................................... 51

Figure 19: Model elements describing the loading of two hoists ................................................. 51

xi

Figure 20: Setting attributes for entities after hoist loading ......................................................... 52

Figure 21: Complete hoist cycle ................................................................................................... 53

Figure 22: Model elements describing initial launching of the hoist ............................................ 54

Figure 23: Release and capture of hoist ........................................................................................ 56

Figure 24: Branch elements directing hoist stops ......................................................................... 57

Figure 25: Return cycle of hoist .................................................................................................... 60

Figure 26: Allowing hoist loading ................................................................................................ 60

Figure 27: Capturing the output data ............................................................................................ 62

Figure 28: Elements allowing generation of graphical output ...................................................... 62

Figure 29: Example of graph Delay per Worker .......................................................................... 68

Figure 30: Example of graph Cumulative Distribution of Delay ................................................. 68

Figure 31: Examination of arrival distributions ............................................................................ 76

Figure 32: Study of the impact of the arrival rate on the average waiting time ........................... 77

Figure 33: Worker Delay graph using inter-arrival rates .............................................................. 78

Figure 34: Cumulative distribution showing effect of inter-arrival rates ..................................... 78

Figure 35: Examination of hoist speeds and capacity on the performance ................................... 79

Figure 36: Examination of the impact of the number of workers. ................................................ 80

Figure 37: Examination of the impact of the number of floors .................................................... 80

Figure 38: Study of linear fit as a model for number of floors ..................................................... 82

Figure 39: Study of quadratic fit as a model for number of floors ............................................... 83

xii

Figure 40: Study of power fit as a model for number of floors .................................................... 83

Figure 41: Results showing worker arrival cases using the delay graph. ..................................... 86

Figure 42: Results showing worker arrival cases using the cumulative distribution. ................... 87

Figure 43: Highlighting the impact of the Zoning on the hoist performance. .............................. 88

Figure 44: Highlighting the impact of the alternative strategy over one hour arrival time. ......... 89

xiii

List of Appendices

Appendix 1: References on Artificial Intelligence ..................................................................... 101

Appendix 2: Model programmed algorithms ............................................................................. 102

1

Chapter 1 Introduction

1

This chapter will present an overview of the thesis. Firstly, it will begin by providing a general

background of construction productivity. Secondly, the reader will be provided with the

objectives and scope of the study. Finally, the motivation and thesis organization will be

presented.

1.1 Background

The construction industry is one of the largest industries in North America. Construction in

Canada is a $171 billion industry, providing 1.24 million jobs (Zuppa 2014) and accounts for

about 7% of the country’s GDP (Statistics Canada 2014). It consumes 40% of the country’s

energy and 50 % of its primary resources (Zuppa 2014). Governments have traditionally sought

to invest in infrastructure construction to enhance the economy and provide jobs.

The construction industry is known for being highly competitive and risky. Contractors are

always seeking ways to reduce their costs and increase their chances of being the lowest bidder.

To remain competitive in construction, more has to be produced for each dollar spent (Dozzi and

AbouRizk 1993).

One of the most well-known and studied ways to reduce construction cost is improving

productivity. Construction productivity is a measure of how much output is produced for each

input of a resource, such as equipment, materials and labour. Project managers aim to improve

productivity by ensuring that the resources are allocated when and where they are needed. A

typical constraint to accomplishing a project as planned is the availability of resources

(Christodoulou et al. 2009).

Construction productivity can be expressed as production rate, unit person-hour rate, and

performance factor (Dozzi and AbouRizk 1993). Many factors impact productivity including

project scope, layout, weather and construction methods (Soekiman et al. 2011). Another factor

that impacts productivity is the time workers spend waiting for materials, equipment, or

instructions that are required for them to continue their work (Thomas 1991).

2

Over the past century, there has been a dramatic shift toward city living (Brown and Newbold

2012). To reduce urban sprawl, there has also been an increase in the number of high-rise

buildings, providing affordable living and efficient supply of services. There has been a steady

increase worldwide in the construction of high-rise buildings and this trend is projected to

continue (CTBUH 2011).

Several definitions have been used to distinguish the buildings in terms of height. The

nomenclature “tall building” may be used to describe a high-rise depending on: location,

proportion, and technologies used (CTBUH 2014). The location of a building provides the

context of how tall the building is in comparison to the surrounding structures. Proportion refers

to its height to width ratio. For example, a building with a height to width ratio of 10:1 is more

likely to be considered tall than a building with a height to width ratio of 1:1, even if the

buildings are of the same height. Finally, to be considered tall, the building must use the type of

technologies used in tall buildings, such as structural wind bracing or dampers. Supertall and

Megatall buildings are more specific in their definitions with minimum heights of 300 and 600

metres respectively.

Recently, the City of Toronto experienced a sharp increase in the construction of high-rise

buildings, as shown in Figure 1. The increase is expected to continue as the number of high-rises

under construction or proposed increases as demonstrated ( Figure 1). Furthermore, there has

been a shift towards high-rise residential buildings. In 2001, 96% of tall buildings were non-

residential (CTBUH 2012). In contrast, 88% of buildings being constructed in 2012 were

residential (CTBUH 2012). There are 15 buildings taller than 150 metres under construction in

Toronto, more than any other city in the western hemisphere (CTBUH 2012). Toronto is

projected to have 45 buildings taller than 150 metres by 2015, about a 3.5 times increase since

2005 (CTBUH 2012).

3

Figure 1: Number of high-rise constructed per year in Toronto (Skyscraper Center 2014)

This increase in high-rise construction in Toronto has introduced new challenges for project

managers. One of the issues project managers face as the buildings get higher is the efficient

vertical delivery of materials and labour, which are typically achieved using the tower crane and

hoist. A hoist is a temporary elevator that moves vertically along a mast structure that is erected

on the outside of a building. As the building progresses in height, the mast is extended.

Figure 2 provides a picture of a crane and a hoist.

There are several reasons for the challenges in the vertical delivery of resources as the buildings

get higher.

The increase in the distance a hoist must travel with tall buildings poses a time challenge

in moving labour and material efficiently.

The increased wind speeds at greater heights and poor weather affecting visibility limit

the crane operations, thereby increasing the demand on the hoist for delivering materials.

Construction in Toronto usually means working in a limited space, which often also

limits the number of hoists that can be installed. Therefore, improving the productivity of

a single hoist is essential.

0

5

10

15

20

25

30

35

40

45

50

192

9

193

1

196

5

196

7

196

8

196

9

197

2

197

3

197

4

197

5

197

6

197

8

197

9

198

1

198

3

198

4

198

5

198

9

199

0

199

1

199

2

199

3

200

3

200

5

200

6

200

7

200

8

200

9

201

0

201

1

201

2

201

3

201

4

Pro

po

sed

Nu

mb

er o

f B

uil

din

gs

Co

mp

lete

d

Year

Toronto's High-rise Construction

4

Hoist operation is dependent upon neighborhood bylaws, labour regulations, and

collective agreements.

An increase in the number of trades and other personnel required in tall buildings leads to

bottle necks at peak arrival and departure times.

Figure 2: Construction crane (left) and hoist (right)

1.2 Motivation

This research began with discussions with developers in the Greater Toronto Area (GTA)

focused on the difficulties and challenges that arise when constructing high-rise buildings. One

of the challenges faced by the industry is productivity losses due to the time required to move

people and materials where they are needed. The operation of hoists is a key element on site

(Cho et al. 2013), and an inefficient hoist can cause worker delays and productivity losses (Shin

et al. 2011). Furthermore, a hoist may have a direct influence on the overall project schedule

depending on the number, location and operation method (Cho et al. 2010). In contrast to crane

planning for construction, the issues in hoist planning have hardly been studied (Hwang 2009).

So, the need for innovative ways to plan hoist operations became apparent.

5

1.3 Research Objective

The primary objective of this study is to improve hoist operations for high-rise building

construction, enabling an efficient delivery of workers during the morning peak.

A secondary objective is to compare operational strategies, such as staggered arrivals and

zoning, on the effectiveness of the hoist. Finally, the output will introduce a bench mark for

comparing alternative hoist management strategies in the future.

1.4 Scope

The scope of this study will be limited to high-rise residential building construction by

developers within the Greater Toronto Area (GTA). The focus on residential presents a worst-

case scenario in that office building interiors are often left unfinished when turned-over to the

user. Interior finishing is resource intensive and can have a major impact on the number of

construction hoists needed on a project.

The scope is also limited to the movement of labour during the morning peak. The morning peak

occurs when the workers arrive to site at the beginning of the shift, and are transported to

different floors in the building. This study will assume that during this morning peak only

workers will be transported in the hoist. This assumption has been validated using site

observations and expert opinion. The reasoning for this assumption is that most construction sites

use the hoist to solely deliver workers during the morning peak. Once the workers have been

delivered, the hoist will begin its mixed-used function of delivering both workers and materials.

This intense activity, of moving workers during the morning peak, is more complex than that of

materials because material delivery could be organized in advance and be scheduled during the

night or weekends (Park et al. 2013). Getting the workers to their work location faster during the

morning peak reduces delays and provides the hoist with more time for material delivery. Thus,

it has a direct influence on the overall project success.

Finally, although productivity is important, this research will focus on the efficiency and

effectiveness of hoists. Hoists are not in themselves a production unit but are instead service

equipment that allows all other work to be performed. Therefore, their productivity is not as

important as their ability to facilitate the real production units. So, throughout this thesis, hoist

efficiency and effectiveness will be used instead.

6

1.5 Research Contributions

This study provides the three contributions in the field of construction hoist operation

improvement:

A hoist model that reduces the number of user-inputs relative to previous models found in

the literature. As such, the model is more user-friendly.

An examination of the impact of changing the workers’ arrival to site on the hoist

efficiency has been conducted. This examination has not been previously conducted for

hoist operations.

Novel ways to present hoist effectiveness and efficiency for alternative management

strategies.

1.6 Study Methodology

The steps that were undertaken in this study were:

The problem was identified through discussions with experts in the field

A literature review was conducted and the limitations of previous efforts were identified

Factors affecting hoist operations were acknowledged

Data were collected to understand and formulate the operation of the hoist

DES was selected as the method of analysis to address the complex uncertainties in the

cyclic operation of a hoist.

Several iterations of the DES model were developed, verified and validated using data

from site and expert opinion. The iterations considered the impact of the factors and the

output

The final model is proposed and presented

7

The impact of the alternative approach, a combination of zoning and staggered arrivals of

workers, is demonstrated through an example.

1.7 Thesis Organization

This dissertation is organized according to the following sections:

Chapter 1-Introduction: This section provides an overview of the problem being studied.

First, an introduction to the need to study the productivity of construction methods in tall

buildings is presented. Second, the objective, scope and motivation of this research are

discussed. Finally, the approach and steps undertaken for this analysis are summarized.

Chapter 2-Hoists and Elevator Planning: A literature review of the research undertaken in

hoist and elevator planning is presented. This section defines the hoist operation

problem, the data collected, and the factors used for the analysis.

Chapter 3-Analysis Method: This section presents the selected analysis method for hoist

planning. In addition, it provides the reader with a description of the mathematical

models used for the analysis of the results. A description of discrete-event simulation and

regression is included.

Chapter 4-Development of the Proposed Model: This section describes the discrete-

event-simulation model. It will begin with the process undertaken in the development of

the model. Subsequently, it will outline the different components of the model and the

methods used for modeling the hoist operation.

Chapter 5-Impact of Model Inputs: An analysis of the model variables and output is

conducted. The study of the impact of each of the inputs of the model on the hoist

performance is included.

Chapter 6 –Using the Model to Improve Hoist Performance: An example of how the

method may be used for hoist planning is also presented through a hypothetical case

study of a tall building.

8

Chapter 7- Conclusion and Recommendations: This section concludes the dissertation.

Limitations and recommendations are presented for guiding future research of hoist

planning and performance improvements.

9

Chapter 2 Elevator and Hoist Planning

2

Due to the scarcity of previous work on hoist planning, a review of elevator planning studies was

undertaken to supplement this research. The similarities in the function of hoists and elevators

allow for a better understanding of how the study on hoist operation could be facilitated.

However, the differences between their operations prevent the use of elevator planning methods

for hoists. This chapter will begin by summarizing previous efforts in elevator planning. Then, it

will describe the similarities and differences between hoists and elevators. Finally, a review of

the research on the operation of construction hoists will be presented.

2.1 Elevator planning

Elevator planning methods range from numerical analysis to simulation techniques. These

methods aim to find an appropriate elevator configuration to serve the traffic during the

operating life of a high-rise building (Tervonena et al. 2008).

Elevator planning is dependent on the characteristic traffic profiles of different building types

(Benmakhlouf and Khator 1993). Office buildings typically have up-peak traffic at the start of

the work day, two-way or inter-floor traffic during the day, and down-peak traffic at the end of

the work day (Tervonena et al. 2008).

The up-peak is most commonly used for the design of elevator systems in office buildings

because it has the most demanding traffic. For this reason, analytical methods for calculating the

up-peak handling capacity and interval are most commonly used (Tervonena et al. 2008).

Early studies in elevator planning focused on developing formulas to estimate the average

waiting time and the round-trip time for elevator users, and these measures became the primary

indicators for elevator performance (Tervonena et al. 2008). The average waiting time (AWT) is

calculated by taking the mean of the actual time prospective passengers wait after registering a

hall call (or entering the waiting queue if a call has already been registered) until the responding

elevator doors begin to open (Cortés et al. 2004). The round-trip time (RTT) is defined as the

time between a passenger’s call for the elevator and when they reach the destination floor. The

10

calculation of round trip time is shown by Equation 1, and average wait time is shown by

Equation 2.

Equation 1: Round-trip time for elevators (Barney and Santos 1975)

𝑹𝑻𝑻 = 𝟐𝑯𝒕𝟏 + (𝑺 + 𝟏)𝒕𝟐 + 𝑷𝒕𝟑

𝑅𝑇𝑇 − 𝑅𝑜𝑢𝑛𝑑 𝑡𝑟𝑖𝑝 𝑡𝑖𝑚𝑒

𝐻 − 𝐻𝑖𝑔ℎ𝑒𝑠𝑡 𝑟𝑒𝑣𝑒𝑟𝑠𝑎𝑙 𝑓𝑙𝑜𝑜𝑟∗

𝑆 − 𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑐𝑎𝑟 𝑠𝑡𝑜𝑝𝑠 𝑎𝑏𝑜𝑣𝑒 𝑡ℎ𝑒 𝑙𝑜𝑏𝑏𝑦∗

𝑃 − 𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑐𝑎𝑟 𝑙𝑜𝑎𝑑

𝑡1 − 𝑖𝑛𝑡𝑒𝑟𝑓𝑙𝑜𝑜𝑟 𝑡𝑟𝑎𝑛𝑠𝑖𝑡 𝑡𝑖𝑚𝑒

𝑡2 − 𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑛𝑔 𝑡𝑖𝑚𝑒

𝑡3 − 𝑝𝑎𝑠𝑠𝑒𝑛𝑔𝑒𝑟 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟 𝑡𝑖𝑚𝑒

∗ −𝑻𝒉𝒆𝒔𝒆 𝒇𝒂𝒄𝒕𝒐𝒓𝒔 𝒄𝒐𝒖𝒍𝒅 𝒃𝒆 𝒂𝒕𝒕𝒂𝒊𝒏𝒆𝒅 𝒇𝒓𝒐𝒎 𝒂 𝒈𝒓𝒂𝒑𝒉

Equation 2: Average wait time for elevator (CIBSE 1993)

𝑨𝑾𝑻 = 𝟎. 𝟒𝑰𝑵𝑻 𝒇𝒐𝒓 𝒄𝒂𝒓 𝒍𝒐𝒂𝒅𝒔 𝒐𝒗𝒆𝒓 𝟓𝟎%

𝐼𝑁𝑇 − 𝐼𝑛𝑡𝑒𝑟𝑣𝑎𝑙 𝑎𝑟𝑟𝑖𝑣𝑎𝑙 𝑡𝑖𝑚𝑒 (𝑚𝑖𝑛𝑢𝑡𝑒𝑠)

The inter- arrival rate and the capacity of an elevator are factors that determine the performance

of an elevator. In an office building, it is suggested that the up-peak interval between arrivals of

passengers is 20-30s (Tervonena et al. 2008). Another suggestion is to use an up-peak arrival

pattern that follows a Poisson distribution with a mean of 23 persons/minute (Ladany and Hersh

1979). An alternative possibility is to use discrete arrivals of 10-15 persons every 5 minutes

(CIBSE 1993).

Once the arrival pattern is determined by the planners, it is used to calculate the average waiting

time for different elevator capacities and combinations. Depending on the acceptable average

waiting time decided by either the owner or designer, a suitable elevator design is selected.

With the increase in mixed-use buildings, the uncertainty in the behaviour of passengers

increases. Because the future behaviour of passengers is unknown, numeric approaches are not

11

always suitable. As such, simulation models become viable. But in practice, the simulation

methods have not progressed much from their initial development in the 1960s (Tervonena et al.

2008). Using simulation to study the morning peak period, it was concluded that: (i) it is

essential to determine the capacity of elevators and approximate a passenger’s waiting time

(Nagatani 2003), and, (ii) that the number of elevators is a major factor when accessing the up-

peak period (Nagatani 2004). Other important factors in determining an elevator’s performance

are the elevator’s speed, acceleration, door types, and the control algorithm (Tervonena et al.

2008).

One of the most commonly used simulation models in the elevator planning and design industry

is Elevate TM

(Peters 2014). This software simulates the performance of an elevator for different

types of buildings. However, through discussions with the developer, it was concluded that

elevator simulation models could not effectively be used for hoist planning. The differences in

the operation of the hoist and elevator limit the application of elevator simulation tools to its

intended purpose - elevators.

2.1.1 Elevators vs. Hoists

The operation of hoists and elevators is similar in concept; however, there are some differences

which render the elevator planning methods unsuitable for that of the hoist. Assumptions in

elevator planning software are (Cortés et al. 2004) :

Each call for the elevator is responded to by only one elevator, even if there are a lot of

people waiting.

The capacity of the elevator constitutes the maximum number of passengers transported.

The elevator stops at a floor only if there is a request to stop on that floor. Furthermore, an

elevator stops at a floor if there is a call, even if it is at capacity.

The elevator calls are consecutively arranged and met depending on the elevator trip

direction.

Elevators that are occupied cannot change the trip direction.

12

One of the major differences between elevators and hoists is in the operation of the system.

Elevators operate automatically within a fixed zone in accordance to a pre-set algorithm while

hoists are operated manually by an operator (Hwang 2009). Furthermore, the nature of

construction makes the hoist operation more dynamic than that of an elevator. The number of

stories, number of workers and working hours change as the construction progresses. With a

hoist, there is more flexibility in controlling the passenger traffic, by, perhaps, changing or

restricting the workers’ schedule. Although an elevator could not change direction if it is

occupied, this does not apply to a hoist. When a hoist is called, there is no pre-determined

destination programmed into the system.

Some similarities may be drawn between the operation of a hoist and an elevator in an office

building. The peaks for both situations occur at the beginning and the end of the work shift if the

transportation of people is considered only. Also, the arrival of workers is similar. It could be

assumed that the elevator and hoist are both bound by the capacity of the car. Furthermore, the

loading and unloading of workers is similar. When materials are not taken into consideration, the

hoist is unloaded similarly to that of an elevator in that both are dependent upon the opening and

closing times of the doors and the movement of workers in and out.

2.2 Hoist Operation during Up-Peak

Meetings with industry experts in the construction of tall buildings and hoist operators were

conducted to study the hoist operation. Construction site visits and recordings of hoist operations

were done to supplement the understanding of the problem at hand. During the meetings, the

project managers described the operation of the hoist and the factors that affect the hoist

operation. An investigation of hoist characteristics has been conducted using manufacturer’s

data.

The hoist operation is cyclic in nature. Typically, a hoist operation follows four steps during the

morning period:

1. Workers arrive to the construction site and wait for the next available hoist.

2. Workers board the hoist and relay their destination to the hoist operator.

3. The operator stops at the called floors and allows the workers to exit.

13

4. Hoist returns to the base of the building for further loading until all workers have been

delivered.

Note that this cycle only occurs in the delivery at the beginning of the work shift. After the

delivery of workers is complete, the hoist’s operation can be shifted to inter-floor transportation

and material delivery. This study will focus on the morning period. Once the method is tested,

the simulation could be expanded to other functions of the hoist.

2.2.1 Factors Affecting Hoist Performance

Worker Arrivals: The arrival of workers to a construction site can be similar to that of workers

to office buildings. However, depending on the scheduling of work by the project manager,

construction workers may take longer to arrive and their arrival can be more dispersed. For

example, some construction companies require their employees to gather at the start of their shift

and sign a safety form. Therefore, their arrival tends to be as one group; while others are

transported to the location as soon as they arrive to the site.

Number of Workers: The number of workers on site in a day is reliant on the project and

project stage. For example, typically at the final stage of the project the number of workers is

less than half that of the construction stage. However, the progress of the construction also plays

a role in the number of workers. If the project is behind schedule, more workers may be needed.

Hoist Departure Decisions: Another factor that affects the hoist operation is loading. As

workers arrive to site, they board the hoist. Typically, the hoist will transport all the workers that

are present even if it is not at capacity. Thus, the time the hoist waits for others to arrive directly

affects the total time of worker delivery.

Hoist Speed and Capacity: While hoist speeds provided by the manufacturers are accurate in

ideal situations, the operational speed is usually less, depending on the project, project stage,

number of stops and the experience of the operator. Table 1 provides examples of the hoist’s

speeds and capacity as provided by the manufacturers.

Hoist Cycle Time: The time it takes the hoist to complete a cycle is dependent on the number of

stops, building footprint, travel distance (number of stories and height of each storey) and speed

of the hoist. The number of workers is directly related to the footprint of the building.

14

Table 1: Sample hoist speeds and capacities

Hoist Type Single or

Double

Capacity

(persons)

Speed

(m/min) Source

Champion US-60-1R Single 27 45

(Metro Elevator 2013)

Champion US-60-2R

Twin Double 27 45

Champion US-6002-

1R Single 27 90

Champion US-6002-

2R Double 27 90

Champion US-80-1R Single 35 45 or

90

Champion US-80-2R Double 35 45 or

90

Hercules F 2200 SLS Both 27 15

(Bigge Crane and Rigging

Company 2014)

Hercules F 6000 SLS Both 30 40

Hercules F 7000 SHS Both 30 45

Hercules F 7000 SLS Both 25 100

PEGA 2832 TD VFC Both 25 100

PEGA 2840 TD VFC Both 20 100

Scando 650 Single 27 50

US-60-1Rx Single 27 45 or

90 (McDonough Elevators 2013)

US-60-2Rx Double 35 45 or

90

HS 80 Both 30 150

(USA Hoist 2014)

USA 7000 Both 27 100

Number of Hoists: In residential construction, the developer wants to minimize the number of

hoists because hoists delay the completion of the suites to which they are attached. This prevents

15

the developer from turning these suites over to the buyers, thereby reducing their cash flow.

Therefore, operating two hoists to their full potential, or adopting alternative strategies to

improve hoist performance is more desirable than adding additional hoists.

2.3 Hoist planning methods

One method for planning hoist operations uses empirical formulas based on experience from

construction projects (Cho et al. 2010). Table 2 shows an example of an existing hoist planning

method. This method is prone to error because it is dependent on the planner’s assumptions (Shin

et al. 2011). When this method is used for tall buildings, the experience is not sufficient to make

valid assumptions. Moreover, this method is used to estimate the required number of hoists;

however, it does not provide insight on how to improve a hoist’s productivity. Thus, the use of

the formulas has been minimal in industry.

Table 2: Simple formulaic method for hoist planning (Shin et al. 2011)

Phase description Formula Description

Transportation Frequency (Ft) Ft=a x b

a: Transportation frequency per

unit area based on historical

data of similar project(s)

b: Gross area of actual project

Transportation frequency per

day (Fd) Fd=Ft/n

n: Total construction duration,

days

Average height of

transportation(Ha) Ha=H x (1+C)

H: Height of Building

C: Charged rate for handicap

Cycle time of Transportation

(Tc) Tc=T1+T2+T3+T4

T1:Loading of hoist time

T2:Unloading of hoist time

T3:Time for lifting up

T4:Time for lifting down

Available Transportation

Frequency per day (Ta) Ta=(Tw/Tc)x d

Tw: Work time per day

D: Operation ratio of hoist

The adequate number of

temporary hoists(Nh) Nh=Fd/Ta

In 1996, a Scalable simulation model was used for elevator operations in construction of high-

rise buildings (Ioannou and Martinez 1996). The cyclic nature of the elevator operation is

modeled through a scalable code which allows the user to predetermine the number of stories.

This was the first attempt at modelling elevator operations in the context of construction.

However, the algorithm had been designed for an elevator as opposed to a hoist.

16

A survey was conducted with construction practitioners inquiring about their experience with

hoists, situations with inappropriate hoist planning, variables needed for hoist decision-making,

and methods used in projects for hoist planning (Hwang 2009). It was noted that almost 60% of

respondents reported cases and consequences of an inappropriate plan. Figure 3 demonstrates

some of the responses gathered in the survey. It is evident that the mathematical formulas are not

commonly used as a planning method for hoists. Also, the survey provided insight on the factors

that affect a hoist’s operation. Using the assembled data, the author proposed a discrete event

simulation model for the analysis of an effective plan for temporary hoists (Hwang 2009).

Simulation allows for the stochastic consideration of all the different factors inherent in hoist

operations. Furthermore, it may provide insight on ways to improve hoist performance.

Figure 3: Decision variables (left) and planning method (Right) survey (Hwang 2009).

A DES simulation model of a hoist operation was developed to calculate the hoist’s cycle time

based on alternative demands (Cho et al. 2010). The hoist cycle time was then calculated using

Equations 4 to 7 and verified using the simulation and data collected from construction sites.

These equations are important for understanding the variables needed for simulating the

operation of the hoist. It is evident that the loading, unloading, acceleration and deceleration

times are required for the simulating the cycle time of the hoist.

17

Equation 3: Cycle time of hoist- version I

𝑻 = 𝑻𝒎 + 𝑻𝒍

𝑇 − 𝐶𝑦𝑐𝑙𝑒 𝑡𝑖𝑚𝑒 (𝑚𝑖𝑛𝑢𝑡𝑒𝑠)

𝑇𝑚 − 𝐿𝑖𝑓𝑡𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 (𝑚𝑖𝑛𝑢𝑡𝑒𝑠)

𝑇𝑙 − 𝐿𝑜𝑎𝑑𝑖𝑛𝑔/𝑢𝑛𝑙𝑜𝑎𝑑𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 (𝑚𝑖𝑛𝑢𝑡𝑒𝑠)

Equation 4: Lifting time of hoist

𝑇𝑚 = 𝑇𝑎𝑠 + 𝑆1 + 𝑆2

𝑇𝑜𝑠 − 𝐿𝑖𝑓𝑡𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 𝑎𝑡 𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑠𝑝𝑒𝑒𝑑 (𝑚𝑖𝑛𝑢𝑡𝑒𝑠)

𝑆1 − 𝐴𝑐𝑐𝑒𝑙𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑡𝑖𝑚𝑒 (minutes)

𝑆2 − 𝐷𝑒𝑐𝑒𝑙𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑡𝑖𝑚𝑒 (minutes)

Equation 5: Lifting time at operation speed of hoist

𝑇𝑜𝑠 = 𝑇𝑣1 + 𝑇𝑣(𝑛−1) + 𝑇𝑣

𝑇𝑣1 − 𝐿𝑖𝑓𝑡𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 𝑜𝑛 𝑓𝑖𝑟𝑠𝑡 𝑓𝑙𝑜𝑜𝑟 (𝑚𝑖𝑛𝑢𝑡𝑒𝑠)

𝑇𝑣(𝑛−1) − 𝑙𝑖𝑓𝑡𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 𝑏𝑒𝑓𝑜𝑟𝑒 𝑑𝑒𝑐𝑒𝑙𝑒𝑟𝑎𝑡𝑖𝑜𝑛 (𝑚𝑖𝑛𝑢𝑡𝑒𝑠)

𝑇𝑣 − 𝑝𝑎𝑠𝑠𝑖𝑛𝑔 𝑓𝑙𝑜𝑜𝑟𝑠 𝑙𝑖𝑓𝑡𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 (𝑚𝑖𝑛𝑢𝑡𝑒𝑠)

Equation 6: Lifting time for individual floors

𝑇𝑣1 =𝐻

𝑉, 𝑇𝑣(2) =

𝐻2

𝑉, 𝑇𝑣 =

∑ 𝐻𝑖𝑖=𝑛𝑖=1

𝑉

𝐻 − 𝐻𝑒𝑖𝑔ℎ𝑡 of story (meters)

𝑉 − 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 of story (meters

minutes)

Equation 7: Loading/unloading time of hoist

𝑇𝑙 = 𝑇𝑑𝑜 + 𝑇𝑑𝑐 + 𝑇𝑙𝑜

𝑇𝑑𝑜 − 𝐷𝑜𝑜𝑟 𝑜𝑝𝑒𝑛𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 (minutes)

𝑇𝑑𝑐 − 𝐷𝑜𝑜𝑟 𝑐𝑙𝑜𝑠𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 (minutes)

𝑇𝑙𝑜 − 𝑙𝑜𝑎𝑑𝑖𝑛𝑔/ 𝑢𝑛𝑙𝑜𝑎𝑑𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 (minutes)

18

The inputs required of the simulation are (Cho et al. 2010) :

Total number of workers

Hoist capacity

Number of floors

Loading/unloading time

Door open and close time

Hoist speed

Number of hoists

A schedule of the workers and the destination floor.

While the simulation was verified to provide accurate cycle time calculations, several aspects

limit its use. First, providing a large amount of data has proven to be a difficult task in many

studies (Cho et al. 2010). A simulation model that eliminates the need for many input variables,

without jeopardizing the accuracy of the results, would be more user-friendly. Second, the output

of the cycle time of the hoist is not an effective indicator of its productivity. As the building gets

higher, the cycle time would naturally increase due to the travel time. However, the time lost

while workers are in a queue or waiting for a hoist is not reflected in this study.

A discrete-event simulation incorporating genetic algorithms was proposed to assist in optimal

hoist planning (Shin et al. 2011). The study focused on developing genetic algorithms for the

peak times of personnel and material hoisting. The proposed method addresses the long time

needed to use simple formulas or simulation in planning hoists (Shin et al. 2011). The formula

used for the cycle time of the hoist is represented in Equation 8. This equation demonstrates how

the operational efficiency and wait time are incorporated into the cycle time of the hoist, which

are factors that were not considered in equations 4-7.

Equation 8: Cycle time of the hoist- version II

𝑻𝒋(𝒋𝒕𝒉 𝒄𝒚𝒄𝒍𝒆 𝒕𝒊𝒎𝒆) = 𝒘𝒋 + 𝒍𝒋 + ∑ ((𝒇𝒌 − 𝒇𝒌−𝟏)𝒉

𝒔𝒆+ 𝒅𝒋𝒌)

𝒏

𝒌=𝟏

+ (𝑴𝑨𝑿(𝒇𝒌)𝒉

𝒔𝒆)

𝑤 − 𝑤𝑎𝑖𝑡𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 (𝑚𝑖𝑛𝑢𝑡𝑒𝑠)

19

𝑙 − 𝑙𝑜𝑎𝑑𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 (𝑚𝑖𝑛𝑢𝑡𝑒𝑠)

𝑓 − 𝑑𝑒𝑠𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛 𝑓𝑙𝑜𝑜𝑟 𝑜𝑓 𝑒𝑎𝑐ℎ 𝑟𝑒𝑠𝑜𝑢𝑟𝑐𝑒 (𝑖𝑛𝑡𝑒𝑔𝑒𝑟)

ℎ − 𝑢𝑛𝑖𝑡 ℎ𝑒𝑖𝑔ℎ𝑡 𝑝𝑒𝑟 𝑠𝑡𝑜𝑟𝑦 (𝑚𝑒𝑡𝑒𝑟𝑠)

𝑠 − 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 𝑠𝑝𝑒𝑒𝑑 (𝑚𝑒𝑡𝑒𝑟𝑠

𝑚𝑖𝑛𝑢𝑡𝑒)

𝑒 − 𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑎𝑙 𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑐𝑦 (%)

𝑑 − 𝑑𝑢𝑚𝑝𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 (𝑚𝑖𝑛𝑢𝑡𝑒𝑠)

This model requires eleven inputs and provides the waiting time for the hoist and the queue

length as outputs (Shin et al. 2011). Although the outputs provided a better estimate of the hoist

performance in comparison to previous studies, it needed a larger number of inputs from the

user.

A combination of simulation and a Branch and Bound (B&B) algorithm was used to compare the

times to deliver workers by changing the delivery routes for the hoists (Cho et al. 2013). While

this tool is capable of providing the best hoist plan with respect to travel time, it falls short in

several areas. First, the destination of the workers is determined mathematically, which may be

inaccurate in real-life applications as worker destination is usually reliant on the type of work

and the construction schedule. Second, the operation of the hoist is optimized based on the total

time needed to deliver the workers. This is a direct function of the height of the building and

does not represent the time lost waiting for the hoist.

While hoist optimization has primarily focused on hoist selection, its operation is equally

important. The only operational practice suggested in the literature to improve hoist performance

was the concept of zoning. Zoning is an elevator operation strategy where cars are limited to

serving specific floors. A simulation model to evaluate the optimal zoning configuration for

hoists over the construction phase was developed (Park et al. 2013), but some challenges were

identified by the authors. First, the process of manually estimating the lifting-demand was time

consuming. Second, the zoning configuration of the hoists had to be adjusted as the construction

progressed. Updating the zoning too frequently may cause confusion to the workers while

updating less frequently may cause a mismatch between the lifting demand and frequency.

Further investigation is needed. While demand-based zoning could reduce the total hoist time by

approximately 40% (Park et al. 2013), the zones were optimized based on the total hoisting time

20

and the wait time by workers for the hoist was not considered. Therefore, the study does not

provide insight on productivity losses due to workers waiting for the hoist.

2.3.1 Summary of Limitations of Current Methods

While the previously developed methods for planning hoist operations have advanced this area,

their limitations make them difficult to use.

The numerical methods used to estimate the number of hoists required are rarely used in

industry. They are ineffective when the user does not provide the appropriate assumptions, and

they typically require a large number of input variables. Furthermore, they do not provide

suggestions to improve hoist effectiveness other than increasing the number of hoists, which as

already discussed, causes other challenges. Therefore, it is more effective to improve the

performance of the hoist.

None of the methods considered the arrival of the workers to the site as a factor. Elevator

productivity studies have shown that the arrival of passengers has a major impact on the

operation. Furthermore, most methods used the cycle time and the total time to deliver workers

as the output variable. However, this is not a complete indicator of the performance of hoists as it

does not indicate how much time is lost waiting for the hoist. The wait time, commonly used in

evaluating elevators, provides a better indication of how much time is lost.

While simulation was used to model the operation of a hoist, only one study presented an

operational management strategy to improve the efficiency of the operation.

21

2.4 Summary of factors impacting hoist operation

Using past research in both hoist operation and elevator planning, factors affecting the

performance of hoists have been compiled. These factors will be either input variables or

inherited within the model, depending on the factor type.

Table 3: Factors affecting hoist operation

Factor Source

(Hw

ang 2

009)

(Cho e

t al

. 2010)

(Shin

et

al. 2011)

(Par

k e

t al

. 2013)

(Lad

any a

nd H

ersh

1979

)

(Ter

vonen

a et

al.

2008

)

(CIB

SE

1993)

Number of Stories X X X

Number of Workers X X X

Arrival Distribution X X X

Mean arrival Rate X X X

Capacity of Hoist X X X X

Number of Hoists X X X X

Time per floor X X X

Loading/unloading Time X X X

Door open/Close Time X X X

Acceleration/Deceleration Time X X X

Speed of Hoist X X X

Number of stops X X X

22

Chapter 3 Analysis Method

3

This chapter will begin by providing a brief description of analysis methods. After an

introduction to and comparison of numerical methods and simulation, the chosen method of

analysis, discrete-event simulation (DES), is described in detail. It will also provide insight to

how DES has been used in the construction industry and why it is effective for studying hoist

operations.

Artificial intelligence (AI) methods, such as Bayesian belief networks and fuzzy logic, are

considered powerful tools for studying construction operations. After careful review, however, it

was decided that these tools will not be used in this research and therefore will not be discussed.

Appendix 1 provides a list of references that describe these methods for the reader’s interest.

3.1 Numerical methods

Numerical analyses solve problems in a numerical form (Gautschi 2012). These approaches are

used to solve problems with many variables and constraints and directly produce optimum points

(Arora 2012). Numerical models could be linear or non-linear, such as polynomials and

differential equations.

3.1.1 Linear Models

The simplest deterministic mathematical relationship between two variables is a linear form

(Devor 2009). The linear model is described in Equation 9.

Equation 9: Linear equation model

𝒚 = 𝜷𝟎 + 𝜷𝟏𝒙𝟏 + 𝜷𝟐𝒙𝟐 + 𝜷𝟑𝒙𝟑 … 𝜷𝒏𝒙𝒏

The x-variables are independent variables which influence the dependent variable y. One of the

methods for developing a linear model is regression. The following section provides an

introduction to regression.

23

3.1.2 Regression

Regression was introduced in 1869 by Sir Francis Galton in his book Hereditary Genius-An

enquiry into its laws and consequences (Bingham and Fry 2010). Regression has evolved since,

and many methods have been developed for the estimation of the equation’s parameters. Fitting

models by minimizing the sum of squares is the most commonly used method (Monahan 2001).

In a regression, it is assumed that the data points are collected independently and are randomly

distributed. Using the sum of squares, the vertical distance of the data points to the regression

line is minimized. The sum of squared vertical deviations from the points to a line is presented in

Equation 10.

Equation 10: Sum of squared vertical deviation (Devor 2009):

𝑓(𝑏0, 𝑏1) = ∑[𝑌 − (𝑏0

𝑛

𝑖=1

+ 𝑏1𝑥𝑖)]2

Using this, the coefficients of the variables are computed. The following equations are used to

calculate the slope and intercept of the linear equation.

Equation 11: Slope of linear equation

𝑏1 =∑(𝑥𝑖 − 𝜇𝑥)(𝑦𝑖 − 𝜇𝑦)

∑(𝑥𝑖 − 𝜇𝑥)2

𝑊ℎ𝑒𝑟𝑒 𝜇 𝑟𝑒𝑝𝑟𝑒𝑠𝑒𝑛𝑡𝑠 𝑡ℎ𝑒 𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒

Equation 12: Y-intercept of linear equation

𝑏0 =∑ 𝑦𝑖 − 𝑏1 ∑ 𝑥𝑖

𝑛

𝑊ℎ𝑒𝑟𝑒 𝑛 𝑖𝑠 𝑡ℎ𝑒 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑑𝑎𝑡𝑎 𝑝𝑜𝑖𝑛𝑡𝑠

24

3.1.3 R-squared

The R-Squared value also referred to as the coefficient of determination, is a value between zero

and one that measures of the variation of the observed dependent variable to the value attained

by the model. A value of 1 indicates a perfect fit of the data to the model (Nelson et al. 2003).

Equations 13 through 15 demonstrate how the R-Squared value is computed.

Equation 13: Total variation of data (SStotal)

𝑆𝑆𝑇𝑜𝑡𝑎𝑙 = ∑(𝑦𝑖

𝑛

𝑖=1

− 𝜇𝑦)2

𝑆𝑆𝑇𝑜𝑡𝑎𝑙 − 𝑇𝑜𝑡𝑎𝑙 𝑣𝑎𝑟𝑖𝑎𝑡𝑖𝑜𝑛 𝑎𝑠 𝑆𝑢𝑚 𝑜𝑓 𝑆𝑞𝑢𝑎𝑟𝑒𝑠

𝑦𝑖 − 𝑂𝑏𝑠𝑒𝑟𝑣𝑒𝑑 𝑌 𝑣𝑎𝑙𝑢𝑒𝑠

𝜇𝑦 − 𝑀𝑒𝑎𝑛 𝑜𝑓 𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑑 𝑣𝑎𝑙𝑢𝑒𝑠

When calculating the total variation, the sum of the square difference between the observed

value and the mean of the data is calculated.

Equation 14: Data residuals (SSe)

𝑆𝑆𝑒 = ∑ (𝑦𝑗 − 𝑦𝑖

𝑛

𝑗=1,𝑖=1

)2

𝑆𝑆𝑒 − 𝑅𝑒𝑠𝑖𝑑𝑢𝑎𝑙 𝑣𝑎𝑟𝑖𝑎𝑡𝑖𝑜𝑛

𝑦𝑗 − 𝑌 𝑣𝑎𝑙𝑢𝑒𝑠 𝑐𝑎𝑙𝑐𝑢𝑙𝑎𝑡𝑒𝑑 𝑏𝑦 𝑚𝑜𝑑𝑒𝑙

𝑦𝑖 − 𝑂𝑏𝑠𝑒𝑟𝑣𝑒𝑑 𝑌 𝑣𝑎𝑙𝑢𝑒𝑠

Equation 15: R-Squared calculation

𝑅2 =𝑆𝑆𝑇𝑜𝑡𝑎𝑙 − 𝑆𝑆𝑒

𝑆𝑆𝑇𝑜𝑡𝑎𝑙

25

3.2 Simulation

Simulation provides a tool to study the behavior of systems that are too complex to study using

analytical methods (Halpin and Riggs 1992). Simulation models contain mathematical and

logical formulations that describe the behavior of systems over time (Naylor et al. 1996).

Simulation models may be characterized with respect to the method that they operate. The

classifications of simulation models can be described as (Rubinstein and Kroese 2007):

Static versus Dynamic Models: Static models are time invariant and tend to represent steady state

systems while dynamic models characterize systems that change with time increments.

Deterministic versus Stochastic Models: Deterministic models have pre-set relationships and

produce the same output given a set of initial conditions while stochastic models contain

randomness and therefore produce variable outputs even with the same initial conditions.

Continuous versus Discrete Models: In discrete models, the model is updated when events or

state variables change. These changes typically occur at uneven time intervals. Continuous

systems, on the other hand, are used to model continuous processes and are typically evaluated at

equal time steps.

3.2.1 Monte Carlo Simulation

Monte Carlo is a static and stochastic simulation which describes a variety of approaches that

conduct “what-if” analyses. It is a simulation method that relies upon the generation of random

numbers and statistical analysis to compute the results (Raychaudhuri 2008). The basic steps for

the Monte Carlo procedure are (Mooney 1997):

1. Develop a computer system to generate data by identifying a pseudo-population to generate

samples.

2. Sample from the population such that the data is representative of the statistical situation

being studied.

3. Calculate a desired characteristic in the sample and collect the information in a vector.

4. Redo Steps 2 and 3 for a predetermined number of trials.

5. Develop a frequency distribution of the results from the trials. This is estimate of

the sampling distribution of the characteristic under the circumstances specified.

26

3.2.2 Discrete Event Simulation

Discrete-event simulation (DES) models possess a state of data that captures the salient variables

of the system at any point in time (Altiok and Melamed 2007). In very general terms, the

methodology of DES systems is the following (Halpin and Riggs 1992):

1. Fix the simulation time to an initial value.

2. Create events, their connections and set their duration.

3. If the event list is empty, the simulation run is terminated. Otherwise, find the awaiting event

and unlink it from the event list.

4. Progress the simulation time to that of the most impending event, and execute it

5. Redo steps 2 and 3 until the simulation is terminated.

3.3 Characteristics of Methods

The use of the appropriate analysis method to represent a problem is important. Table 4

summarizes when a method is used and the advantages and limitations of each method.

Table 4: Characteristics of Analysis Methods

Type Method Use Advantages Disadvantages

Numerical

methods

Regression Used to represent the

effect of independent

variables on a

dependent variable

Provide insight on

the effect of each

variable on the

output

Very flexible

Model easily used

and modified

Underlying

assumptions must be

validated

Not effective in

modelling stochastic

variables

Simulation Monte Carlo Uses random number

generation to study

different alternatives

and impacts

Probabilistic

results of the

outcome

Allows for

multiple scenario

analysis

Effectively model

stochastic systems

Results depend on the

number of runs

Time-consuming to

develop and conduct

Difficult to validate

Not very user friendly

Discrete

Event

Uses events with

discrete durations to

represent a process

In many instances, a hybrid of more than one method was used to model a solution for a

problem. Hybrid methods usually complement each other’s weaknesses, thus producing a better

and more effective model.

27

3.4 Selected Method: Discrete-Event simulation

This section will introduce in more detail discrete-event simulation and why it is suitable for

modelling hoist operations.

A discrete-event simulation model is a representation of a system. The key elements in a DES

model are variables and events. There are three variable types in a DES model (Ross 2013):

1. Time variables: amount of simulated time that has elapsed.

2. Counter variables: keep track of the number of times an event has occurred.

3. System state variables: describe the state of the system at a certain time.

Discrete-event simulation is a representation of discrete events through which entities flow.

Whenever entities pass through an event, the values of the model variables change (Ross 2013).

The major aspects of DES systems are described next.

Model: A model is the representation of the operation being studied. A model is limited by

its boundaries and assumptions. A simulation model is typically created using nodes that

represent an action or activity. Nodes are connected by arrows that represent the flow of

entities (e.g. people, equipment or materials) from node to node.

Event: An event is a time-constrained occurrence that changes the system. An event in the

model, also referred to as a task, is a time or occurrence that delays the entities for a period of

time. For example, the time it takes a truck to travel could be represented in the model as the

“travel” event with the associated time. The duration of an event could be fixed, drawn from

a distribution, or depend upon other model variable states.

Entities: Entities represent an object that moves in the model and may represent workers,

equipment, materials, or any other entity in the system.

Attributes: Attributes are variables assigned to describe entity characteristics.

Resources: A resource is an object that provides a service to entities with a pre-determined

function. A resource is captured and used. For example, a resource could be an excavator that

loads trucks when they arrive at a loading event. The resource can be limited to a specific

28

quantity, so if a resource is not available, the entities wait for one to become available before

continuing into the event.

3.4.1 DES in Construction

Construction sites are unique in nature and therefore require a production plan catered to its

individual characteristics. Therefore, construction planning methods must be flexible to

accommodate the variable environment (Cho et al. 2011). The cyclic nature of construction

activities and the ability to easily evaluate construction alternatives are also challenging for

traditional methods (Chen and Huang 2013); (Levitt et al. 1999).

DES provides an alternative to traditional methods and has therefore been widely used to

represent the variable cyclic operations in construction. Built-in flexibility in DES allows for the

study of unique characteristics of a construction site and can easily represent the cyclic nature in

construction operations (Martinez J. C. 2010).

A very useful technique for the quantitative examination of operations and processes that occur

during the life cycle of a constructed project (Martinez J. C. 2010), simulation can be used to

examine and compare the performance of different construction approaches (Ioannou and

Martinez 1996). The steps for building a DES model in the construction engineering and

management field are (Martinez J. C. 2010):

1. Determine if DES is the appropriate method of analysis by understanding how the model can

be used to understand the system and evaluate performance.

2. Understand what questions need to be answered by the model and limit the scope and the

boundary accordingly.

3. Define how detailed the model will be and the features of the operation are. This is attained

by defining the elements, logic and model components will be used.

4. Collect data of the operation being modelled. The data should describe the probabilistic

assumptions and distribution fits that will be fed into the model.

5. Verify the model by ensuring the results and function is as expected. Validate the model

using data from the system.

6. Assess the output of the model for a single run.

7. Design simulation experiments and test the output to determine the performance of different

options.

29

8. Document the results and use them for determining the favorable options.

3.4.2 Limitations of DES

The assumption in DES is that the time between events is negligible; however, real world

activities are continuous; nothing is innately discrete (Puri and Martinez 2013). Therefore, a

modeller must ensure that the discretization assumptions are valid in the situation being studied.

Also, activity durations must represent the “real-world” scenario. Discretization, transforming

continuous processes into discrete-events, provides a significant improvement in the

computational performance during simulation (Puri and Martinez 2013).

DES is used to make decisions prior to implementation of the proposed changes in the field.

Therefore, it is not possible to completely validate the model by comparing it to real-world

output (Martinez J. C. 2010), but it is possible to model existing operations for validation

purposes.

3.4.3 DES Software in Construction

The use of DES in construction is credited to the development of CYCLONE by Haplin in 1977

(Hajjar and AbouRizk 2002). CYCLONE is a general-purpose modelling interface which allows

the user to use embodied functions to model a scenario. Since the development of CYCLONE,

many construction simulation interfaces have been developed such as INSIGHT (Paulson 1978),

RESQUE (Chang 1987), UM-CYCLONE (Ioannou P. 1989), CIPROS (Tommelein and Odeh

1994), STROBOSCOPE (Martinez J. C. 1996).

3.4.4 Applicability of Method to Hoist Operation

Typically, the problems that are well suited to DES include (Martinez J. C. 2010), (Ruwanpura

and Ariaratnam 2007):

1. Situations with significant uncertainties in the time required for an event and/or resource

quantities, operation and organization.

2. Situations that are logistically complicated with numerous dynamic rules and decisions that

change according to the context of the situation. Simulation provides an alternative to study

the behaviour of these complex systems. It offers a method of direct and detailed

observations. Using Simulation allows for the development of an approximate solution.

30

3. Problems that have interdependent components subject to variable event start-up conditions

that includes many resources that must collaborate in a complex organization.

4. Simulation provides an alternative to modelling problems which are difficult to model with

other mathematical methods.

5. If it is difficult to conduct a physical experiment, simulation provides an alternative to

conducting studies and observing the results.

Hoist operation is dependent upon many factors, and thus uncertainties lie in its tasks. The

arrival of workers, hoist breakdowns, stage of construction, and changes in schedule are

examples of the types of factors that produce uncertainty in a hoist’s operation. The operations of

a hoist and the decisions for locating and scheduling are complex in nature. The inter-linked

dependence of the hoist and construction schedule makes it very difficult to optimize the

operation of a hoist alone without considering its impacts on other activities. Furthermore, when

tall buildings are being studied, the complexity increases. As such, these characteristics of a hoist

operation are well suited for DES modeling.

31

Chapter 4 Development of the Proposed Model

4

The purpose of this chapter is to present the process used to develop the simulation model of the

hoist operation, along with a description of the model itself. First, the simulation modeling

environment software is introduced to the reader. Second, the iterations and factors of the model

are outlined. Finally, the model components and strategy are described.

4.1 Introduction

To ensure that the model meets the objectives of the study, several elements have been

considered. First, flexible designs enable managers to respond easily and cost-effectively to

changing circumstances (de Neufville and Scholtes 2011). To enable the representation of

different projects, project stages and characteristics, the model should allow the user to easily

change the inputs of the model. The model should also enable future expansion to other peak

times and to include the delivery of material.

Second, the model outputs should provide useful feedback on the different strategies under study

and allow direct comparisons between strategies. Moreover, the output must reflect not only the

operations of the hoist but of the labour delays caused by the hoist as well.

Third, the model must be developed for easy use. Thus, it must limit the number of inputs needed

while not compromising the modelling accuracy. Previous models required many user inputs,

which made their use time-consuming and less attractive to industry users. As such, this model

will automate as many inputs as possible.

Finally, using the model to study alternative operational strategies for hoists is important. This

will enable the planner to test ways to improve the hoist operations.

4.2 Simphony.Net

Simphony.Net (AbouRizk S. 2014) is a well-established DES software in both the research

community and the construction industry. Developed at University of Alberta, it has been

32

graciously provided for this research. Along with its user-friendly environment, it provides

support for (AbouRizk 2014):

Working in a user-friendly integrated interface environment

Modeling complicated scenarios by enabling modular and hierarchical structures

Storing and recovering frequently used simulation model layouts

Creating and using special purpose templates such as earthmoving, along with traditional

simulation templates such as CYCLONE

Writing user code to compute functions beyond the standard elements

Customizing the output with tools such as charts and tables

Producing automated data outputs in a standard form

4.2.1 Main Interface

The main interface of Simphony.Net is comprised of several components (Figure 4) as briefly

described next. For a more detailed description, the reader is directed to the Simphony.Net’s user

guide (AbouRizk S. 2014).

Modelling Space is the workspace for developing the graphical simulation model. Independent

modelling spaces can be created for each model scenario.

Template area contains the standard modelling elements.

Ribbon bar contains the file, view, run and help menus.

Model explorer window allows the user to view the files that comprise the structure of the model

and select scenarios and composite elements.

Property window lists the properties of model components, including scenarios and elements.

Properties are modifiable by the user and include probabilistic distributions to define activity

durations, model controls, or physical attributes,

33

Trace window provides the user with the text specified using a trace element or the errors

attained during modelling.

Figure 4: Simphony.Net’s main interface components

4.2.2 Modelling Elements

To supplement the reader’s understanding of Simphony.Net, a summary of selected modelling

elements is included in Table 5. This summary has been developed using the template manual

provided by the software’s developers. For a more detailed description of the elements, the

reader is directed to the general template manual (AbouRizk S. 2014).

Table 5: Description of Simphony.Net modelling elements

Element Symbol Description

Comment

Allows the user to add comments as text in the model.

Modelling Space

Ribbon Bar

Template

Area

Model

Explorer

Property

Grid

Trace Window

34

Element Symbol Description

Create

Creates entities at different quantities that leave the

element at user specified time distributions.

Task

Used to represent an activity. An entity is delayed for a

specific period of time. The task completion is also

dependent upon the availability of a server.

Counter

Increments a user-specified quantity each time an entity

passes through the element.

Set Attribute

Allows the user to assign attributes to the passing

entities.

Destroy

Deletes any entities that arrive at the element.

Execute

Allows the user to customize an action using user-

written code every time an entity passes through the

element.

Composite

A child group of elements that are grouped within a

scenario.

Trace

Allows the user to track the progress of the model by

outputting user-specified text.

Branch

Directs the entity to a different path based on either a

user –specified probability or condition.

35

Element Symbol Description

Generate

Used to clone a quantity of entities from an original

entity.

Consolidate

Merges a quantity of entities.

Batch

Merges a quantity of entities that could be unmerged

using an Unbatch element.

Unbatch

Unmerges a group of batched entities.

Resource

Identifies a resource in the model.

File

Tracks the waiting of entities for resources.

Capture

Allows the entities to capture resources.

Release

Allows the entities to release resources.

Preempt

Captures a resource with priority over a capture

element.

Statistic

Used to compute statistics on parameters in the model.

36

Element Symbol Description

StatisticCollect

Adds observations to a statistic element.

Valve

Stop or allow the entities to pass.

Chart

Displays data in a user-specified chat format.

ChartCollect

Receives an entity and collects the data points.

4.2.3 Simphony’s modelling distributions

Symphony.net has built in statistical distributions that allow for probabilistic modelling. The

following describes the distributions that have been used in the model.

The Constant distribution is a deterministic user-defined value.

The Uniform distribution describes a scenario when all values between a and b have an equal

probability of occurrence, as shown in Figure 5. The uniform distribution is commonly used for

random number generation.

37

Figure 5: Uniform distribution

The Normal distribution is one of the most commonly used distributions in modelling. It

describes many observed phenomena such as heights and weights. As a symmetric distribution,

the average has the highest probability of occurring. Figure 6 demonstrates the shape of the

normal distribution.

Figure 6: Normal distribution

The Exponential distribution (Figure 7) is shaped such that the probability increases faster as x

increases. For example, it can describe the arrivals of workers to office buildings in the morning.

The distribution is demonstrated.

Uniform Distribution

a b

P(x)

Pro

bab

ility

X-values

Normal Distribution

38

Figure 7: Exponential distribution layout

The Step function (Figure 8) describes occurrences where a group of x values share the same

probability. For example, the first 5 individuals that buy a product have an equal probability of

getting a certain price, and the probability that the price increases changes with each 5 buyers.

Figure 10 provides an example of how the step function could be displayed. Note that the step

function is not built in Simphony.Net’s modelling elements. However, this function has been

utilized by programming it into the model.

Figure 8: Step function layout

4.3 Model Uses

This section will describe the two components to modelling hoist operations: (i) modelling the

current operation of the hoist, and (ii) providing an alternative method for improving the

operation of the hoist.

Pro

bab

ility

X-values

Exponential Distribution

Pro

bab

ility

X-Values

Step Function

39

4.3.1 Current operation of the hoist

In this component of the model, the operation of the hoist is based on descriptions found in the

literature, through site observation, and from expert knowledge. Two scenarios are studied,

representing a single or double hoist. Figure 9 provides the flow chart of the method. Each of the

colors in this figure represents a component in the model which will be described in Section 4.4

Figure 9: Flow chart of method used for modelling the hoist operation

40

4.3.2 Alternative strategy for improving hoist performance

Two operational strategies have been included in the model to allow the user to improve the hoist

effectiveness by 1) staggering arrivals in which the workers arrive at different times, and, 2)

using zoning concepts to schedule the destination floors.

Zoning is a concept widely used in elevators and refers to predetermining the destination floors.

Many office and residential tall buildings have elevators that only travel to specific floors to

reduce elevator waiting times. While elevators are designed to only travel to specific floors,

hoists could be managed in a similar way. In this strategy, workers arrive at different times

depending on their destination floors or “zones”.

The staggered arrivals concept has been studied in transportation and elevator planning in office

buildings. Staggering arrivals reduces the queue that occurs. While this concept theoretically

achieves lower waiting times for elevators, it is usually difficult to implement due to the lack of

control of the arrivals. However, because the project manager is responsible for all personnel on

site, it is possible to schedule different trades to start work at different times. The distribution of

arrivals in these scenarios is the step function as shown in Table 6.

As an example, the first group of workers could be scheduled to arrive at 7:00 am and are

transported to floors 60-70 (zone 1), while the following group of workers are scheduled to

arrive at 7:15 am and are transported to floors 40-59.

Table 6: Example of alternative strategy schedule using step function

Group Number of Workers Time between Group Arrivals (mins) Destination floors

1 40 15

60-70

2 60 40-59

15

3 50 1-39

The impact of using this planning strategy on the hoist will be demonstrated through an example

in Chapter 6.

41

4.3.3 Comparison of the arrival time functions

The distribution used in the arrival time for the current operation and alternative strategy is the

main difference between the scenarios. Table 7 highlights the difference in the arrival of workers

and assignment of floors for workers.

Table 7: Difference between arrival of workers in the model scenarios

Operation Arrival Distribution Destination

Current Operation Constant inter-arrival time between

workers

Randomly assigned to any floor in

building

Alternative

Strategy

Step function between group of

workers

Zone assigned by user for each

group

4.4 Model description

This section will describe the model, which has four scenarios in a hierarchy structure as shown

in Figure 10. Within each scenario, the composite modelling element has been used to group a

sub collection of elements that come together for a specific purpose.

To describe the model for the reader, it has been divided into four main sections: (i) variable

input, (ii) arrival of workers and loading of hoist, (iii) operation of hoist, and, (iv) generation of

output. All these components work in tandem to provide a functional model. Figure 11 shows the

overall model. Each of these model components will be described including the components for

the current hoist operation and the alternative strategy.

42

Level 1: Model Level 2: Scenario Level 3: Composite

Pro

po

sed

Mod

el

S1: Single Hoist Current Operation

Arrival of workers

Wait for Hoist

Global Variables

S2: Single Hoist Alterative Strategy

Arrival of workers

Wait for Hoist

Global Variables

S3: Double Hoist Current Operation

Arrival of workers

Wait for Hoist

Global Variables

Hoist Selection

S4: Double Hoist Alterative Strategy

Arrival of workers

Wait for Hoist

Global Variables

Hoist Selection

Figure 10: Hierarchy of the model

43

Figure 11: Sample scenario layout

1-V

ari

ab

le I

np

ut

4-G

ener

atio

n o

f O

utp

ut

3-O

per

ati

on

of

Hois

t

2-A

rriv

al

of

work

ers

an

d l

oad

ing o

f h

ois

t

44

4.4.1 User-Input component

The first component of the model allows the user to input the variables that describe the project.

The input options are different depending on the scenario.

4.4.1.1 Current Operation Scenario Inputs

The current operation scenarios contain five inputs as shown in Figure 12.

Figure 12: Current operation scenario model inputs

Once the input variables are defined by the user, the values are transferred to global variables

through the composite component called “Global Variables”. This composite module consists of

create, execute and destroy elements. Once created, the entity passes through an execute function

where all the input variables are defined according to the user input. Figure 13 demonstrates the

model elements used for this process.

45

Figure 13: Procedure for defining model inputs

The execute elements runs the following algorithm (Table 8) to equate the variables in the model

to the values inputted by the user.

Table 8: Example algorithm for input assignment

Element: Execute Name: Assign array and

random variables

Scenarios:

1 hoist, current operation

2 hoists, current operation

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Execute) As System.Boolean

Redim context.Scenario.Ints(600)

Dim a as Comment = context.Scenario.GetElement(of Comment)("1")

Dim c as Comment = context.Scenario.GetElement(of Comment)("3")

Dim e as Comment = context.Scenario.GetElement(of Comment)("5")

Dim f as Comment = context.Scenario.GetElement(of Comment)("6")

dim g as Comment = context.Scenario.GetElement(of comment)("7")

context.Scenario.ints(203)=Cint(a.Text)

Context.Scenario.floats(2)=(cdbl(c.text))

context.Scenario.Ints(213)=Cint(e.Text)

Context.Scenario.floats(1)=cdbl(f.Text)

Context.Scenario.Ints(205) = Cint(g.Text)

Return true

End Function

End Class

4.4.1.2 Alterative strategy scenarios inputs

The alternative strategy scenarios allow the user to select the number of workers for different

groups, with the time between the groups and the destination for these groups. An example from

the model is displayed in Figure 14. The elements used in the model are same as the ones

presented in Figure 13 for the current operation scenarios.

46

Figure 14: Alternative strategy scenario model inputs

4.4.2 Arrival of workers

Once the global variables are assigned, the arrival of workers begin according to the distribution

and rate assigned by the user input. A composite function contains the elements used to model

the arrival of workers.

First, a large number of entities are created at Time=0 representing the workers. To model the

inter-arrival rate, the entities pass through a task, whose duration is dependent on the scenario

being used. For the current operation scenarios, the interval of the task is a constant distribution

with the value provided by the user. An inter-arrival duration of zero represents the workers

arriving all at the same time whereas and value greater than zero separates their arrivals. In the

alternative strategy scenarios, the task’s duration is based on the schedule inputted by the user.

This has been programmed as shown in Table 9. A “Pseudo” resource, not representing an actual

resource, is used to ensure that one entity passes the task at a time.

Once the entities arrive, they are randomly assigned a floor between 0 and the number of stories

completed. Also, the time of arrival of each worker is recorded as an attribute of the entity. This

will allow tracking of the assigned destination and time of arrival of each worker individually.

47

Finally, the number of workers is maintained according to the user input. This is done through a

branch element which ensures only the required number of workers pass through to load the

hoist.

Figure 15 provides a sample of the elements used in the model.

Table 9: Example algorithm for assigning durations according to step function

Element: Execute Name: Assign array and

random variables

Scenarios:

1 hoist, Alterative strategy

2 hoists, current operation

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.Modeling.Task(Of Simphony.Simulation.GeneralEntity)) As

System.Double

Dim N1 as Comment = context.Scenario.GetElement(of Comment)("N1")

Dim N2 as Comment = context.Scenario.GetElement(of Comment)("N2")

Dim N3 as Comment = context.Scenario.GetElement(of Comment)("N3")

Dim N4 as Comment = context.Scenario.GetElement(of Comment)("N4")

dim N5 as Comment = context.Scenario.GetElement(of comment)("N5")

Dim T1 as Comment = context.Scenario.GetElement(of Comment)("T1")

Dim T2 as Comment = context.Scenario.GetElement(of Comment)("T2")

Dim T3 as Comment = context.Scenario.GetElement(of Comment)("T3")

Dim T4 as Comment = context.Scenario.GetElement(of Comment)("T4")

Dim L1 as Comment = context.Scenario.GetElement(of Comment)("L1")

Dim L2 as Comment = context.Scenario.GetElement(of Comment)("L2")

Dim L3 as Comment = context.Scenario.GetElement(of Comment)("L3")

Dim L4 as Comment = context.Scenario.GetElement(of Comment)("L4")

dim L5 as Comment = context.Scenario.GetElement(of comment)("L5")

Dim H1 as Comment = context.Scenario.GetElement(of Comment)("H1")

Dim H2 as Comment = context.Scenario.GetElement(of Comment)("H2")

Dim H3 as Comment = context.Scenario.GetElement(of Comment)("H3")

Dim H4 as Comment = context.Scenario.GetElement(of Comment)("H4")

dim H5 as Comment = context.Scenario.GetElement(of comment)("H5")

Dim W as Counter = context.Scenario.GetElement(of

Counter)("workers")

Select Case W.Count

'1

Case is <= Cint(N1.text)

Context.CurrentEntity.Ints(0)=

Cint(uniform.sample((Cint(L1.text)),(Cint(H1.text))))

Return 0

Case (Cint(N1.text)+1)

Context.CurrentEntity.ints(0)=

Cint(uniform.sample((Cint(L2.text)),(Cint(H2.text))))

Return Cdbl(T1.text)

48

'2

Case is <= (Cint(N2.text)+Cint(N1.text))

Context.CurrentEntity.Ints(0)=

Cint(uniform.sample((Cint(L2.text)),(Cint(H2.text))))

Return 0

Case (Cint(N2.text)+Cint(N1.text)+1)

Context.CurrentEntity.ints(0)=

Cint(uniform.sample((Cint(L3.text)),(Cint(H3.text))))

Return Cdbl(T2.text)

'3

Case is <= (Cint(N3.text)+Cint(N2.text)+Cint(N1.text))

Context.CurrentEntity.Ints(0)=

Cint(uniform.sample((Cint(L3.text)),(Cint(H3.text))))

Return 0

Case Cint(Cint(N3.text)+Cint(N2.text)+Cint(N1.text)+1)

Context.CurrentEntity.ints(0)=

Cint(uniform.sample((Cint(L4.text)),(Cint(H4.text))))

Return Cdbl(T3.text)

'4

Case is <=

(Cint(N4.text)+Cint(N3.text)+Cint(N2.text)+Cint(N1.text))

Context.CurrentEntity.Ints(0)=

Cint(uniform.sample((Cint(L4.text)),(Cint(H4.text))))

Return 0

Case (Cint(N4.text)+Cint(N3.text)+Cint(N2.text)+Cint(N1.text)+1)

Context.CurrentEntity.ints(0)=

Cint(uniform.sample((Cint(L5.text)),(Cint(H5.text))))

Return Cdbl(T4.text)

'5

Case is <=

(Cint(N5.text)+Cint(N4.text)+Cint(N3.text)+Cint(N2.text)+Cint(N1.text))

Context.CurrentEntity.Ints(0)=

Cint(uniform.sample((Cint(L5.text)),(Cint(H5.text))))

Return 0

End select

Return nothing

End Function

End Class

49

Figure 15: Model elements representing arrival of workers

4.4.3 Loading of the Hoist

Once the workers arrive for scenarios with two hoists, an algorithm embedded in nested branch

elements determine which hoist is available for loading. The algorithm asks the following

questions and directs the workers accordingly:

Is Hoist 1 available?

o If yes, is Hoist 2 also available?

If both hoists are available allow worker to load any hoist.

If only hoist one is available, load hoist one.

If no, is hoist two available?

If yes, load hoist two.

If no, wait for a hoist to be available.

50

Figure 16 and Figure 17 demonstrate a sample from the model for the selection of either one or

two hoists depending on the scenario.

Figure 16: Model elements representing loading of one hoist

Figure 17: Model elements representing loading of two hoists

The first entity to enter a hoist is passed into a task that has a duration of one minute. The hoist is

launched through a valve element in two cases, if the hoist has been loaded up to capacity or the

51

first entity has completed the one minute task. This will ensure that the hoist does not wait for

more than one minute.

If no hoist is available, the entities are redirected to a composite element called “wait for hoist”.

In this element the entities, wait for a hoist to become available.

This is presented in Figure 18 and Figure 19 from the model.

Figure 18: Model elements describing the loading of one hoist

Figure 19: Model elements describing the loading of two hoists

After a hoist is launched, data on the workers that have loaded the hoist is collected. The number

of stops on each floor and the final stop for that trip is assigned to global variables.

52

Figure 20: Setting attributes for entities after hoist loading

4.4.4 Hoist operation

Once the hoist is loaded and launched, an entity representing the hoist is released. The entity is

then assigned an ID attribute indicating whether it is hoist one or two. Depending on the hoist,

the appropriate resource is captured.

Once the hoist is captured, the cycle of travel begins. A task representing the hoist’s travel

duration per floor is used. As the hoist travels a floor, an algorithm keeps track of which floor the

hoist is currently on, and a series of branch elements direct the hoist entity. If the hoist has a stop

on the current floor, it is directed to a task that accounts for the time of loading, unloading,

deceleration and acceleration. Then it is redirected to travel another floor in the cycle. If the hoist

does not have a stop, it is directed to travel another floor in the cycle without the added time of

stopping.

Once the hoist reaches the final stop, it is unloaded, returned to the beginning of the cycle,

resource is released, and the waiting workers are allowed to enter the hoist. During the return of

the hoist entity, the variables that represent the characteristics of the trip for the hoist such as the

final stop are reset to zero.

This cycle is repeated until all the workers have been delivered. A description of this process is

presented along with the model elements used. The description will follow the path of the hoist

entity along the cycle. Figure 21 demonstrates the entire hoist cycle. However, to better present

to the reader each of the model elements used in the analysis, following figures will provide a

closer look at each of the different components.

53

Figure 21: Complete hoist cycle

54

A create element is used to produce two entities, each representing a hoist. The entities pass

through an execute function which sets an attribute for each entity to identify it as either hoist

one or two. A branch is then used to direct each entity to its path. Once the entities are directed,

they wait at a valve element which is opened by having the hoist at either capacity or if it has

been waiting more than one minute, which occurs in the loading of the hoist cycle. The

algorithms used in the execute and branch elements are displayed in Tables Table 10 and Table

11, respectively.

Figure 22: Model elements describing initial launching of the hoist

55

Table 10: Sample algorithm for assigning hoist ID

Element: Execute Name: Assign Hoist

Number

Scenarios:

All

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Execute) As System.Boolean

context.scenario.ints(211)= context.scenario.ints(211)+ 1

Context.currententity.Ints(2)= context.scenario.ints(211)

Return true

End Function

End Class

Table 11: Sample algorithm for directing the hoist

Element: Branch Name: Hoist 1 or 2 Scenarios:

All

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Branch) As System.Boolean

If context.CurrentEntity.Ints(2) = 1 then

Return true

end if

Return False

End Function

End Class

After either valve is opened, the entity captures the associated hoist resource. An execute

function is then used to count the number of cycles that is taken by each hoist. The entity is then

passed through a task which has a duration that represents travelling a single floor by the hoist.

This task takes the time per story inputted by the user and reduces the amount provided by 80%.

This aligns with the observations from site and expert opinion. This is demonstrated in Figure

23, Table 12 and Table 13.

56

Figure 23: Release and capture of hoist

Table 12: Sample algorithm for counting hoist cycles

Element: Execute Name: Number of Trips

for H1

Scenarios:

All

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Execute) As System.Boolean

Context.Scenario.Ints(207)=context.Scenario.Ints(207)+1

Return true

End Function

End Class

Table 13: Sample algorithm for assigning travel durations

Element: Task Name: Travel 1 Floor Scenarios:

All

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.Modeling.Task(Of Simphony.Simulation.GeneralEntity)) As

System.Double

Return (Context.Scenario.floats(1)*0.8)

End Function

End Class

57

Once an Entity has traveled a floor, an execute element tracks the floor the hoist is on. Then a

group of branch elements are used to direct the hoist based on the value of the counter according

to the following logic:

Is there a stop on this floor?

o If yes: Is it the final stop?

If yes, go to the return cycle

If no, pass through a task which takes accounts for the acceleration,

deceleration, loading and unloading time components and then return to

travel another floor.

o If no, then go back to traveling another floor

Figure 24 demonstrates the model elements used for this process. Table 14, Table 15 and Table

16 will also demonstrate the algorithms used in these elements to model this operation.

Figure 24: Branch elements directing hoist stops

58

Table 14: Sample algorithm for tracking hoist travel

Element: Execute Name: Floor Counter Scenarios:

All

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Execute) As System.Boolean

Dim H as Integer

H = context.CurrentEntity.Ints(2)

Select case H

Case 1

context.Scenario.Ints(202) = context.Scenario.Ints(202)+1

Case 2

context.Scenario.Ints(302) = context.Scenario.Ints(302)+1

End Select

Return true

End Function

End Class

Table 15: Sample algorithm for checking if it is the final stop

Element: Branch Name: More Floors to go? Scenarios:

All

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Branch) As System.Boolean

Dim H as Integer

H = context.CurrentEntity.Ints(2)

Select case H

Case 1

If context.Scenario.Ints(201) = Context.Scenario.Ints(202) then

Return false

End if

Case 2

if context.Scenario.Ints(301) = context.Scenario.Ints(302) then

Return False

End If

End Select

Return true

End Function

End Class

59

Table 16: Sample algorithm for checking if there is a stop

Element: Task Name: Stop on Current

Floor?

Scenarios:

All

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Branch) As System.Boolean

Dim H as Integer

H = context.CurrentEntity.Ints(2)

Select case H

Case 1

If context.Scenario.Ints(Context.Scenario.Ints(202))> 0

then 'there is a stop

context.Scenario.Ints(Context.Scenario.Ints(202))=0

return true

End if

Case 2

If context.Scenario.Ints(400+Context.Scenario.Ints(302))> 0

then 'there is a stop

context.Scenario.Ints(400+Context.Scenario.Ints(302))=0

return true

End if

End Select

'There is no stop

Return false

End Function

End Class

Finally, once the entity has completed all the stops of the hoist, it is sent through the return cycle.

In this cycle, a task is used to represent the return trip of the hoist. Furthermore, the hoist

resource is released, and the variables are reset to zero. This will allow the hoist to be reloaded.

Thus, a valve controlling the loading of the hoist is opened by passing the entity through an

activator element. And the release of the hoist is restricted by another activator, this will ensure

that the hoist is only launched by the loading of the hoist operation, which is based on capacity

or waiting time. Figure 25 and Figure 26, present the modelling elements used for the return

cycle. Table 17 demonstrates the algorithm used for resetting the variables.

60

Figure 25: Return cycle of hoist

Figure 26: Allowing hoist loading

61

Table 17: Sample algorithm for resetting variables

Element: Execute Name: Reset variables Scenarios:

All

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Execute) As System.Boolean

Dim H as Integer

H = context.CurrentEntity.Ints(2)

Select case H

Case 1

Context.Scenario.Ints(200)=0

Context.Scenario.Ints(201)=0

Context.Scenario.Ints(202)=0

For A as Integer = 0 to 150

Context.Scenario.Ints(A) = 0

Next

context.Scenario.Ints(215) = 0

Case 2

Context.Scenario.Ints(301)=0

context.Scenario.Ints(300)=0

Context.Scenario.Ints(302)=0

context.Scenario.Ints(216)=0

End Select

Return true

End Function

End Class

4.4.5 Output generation

As the final worker is being loaded into the hoist, the outputs are computed. The waiting time of

each worker is stored as an attribute to the entity. Chartcollect and statisticscollect elements are

used to track the delay of each worker and the statistics of that delay. These elements provide the

user with a graph of the delay of each worker in order of arrival along with the histogram and

cumulative delay of the waiting times. The user could display these graphs for a single run or all

runs by selecting the statistics and chart elements shown in Figure 27.

Finally, a branch checks whether it is the final worker. Once the final entity passes, the average

waiting time is outputted to the user in a trace element. The algorithm used for the output is

shown in Table 18.

62

Figure 27: Capturing the output data

Figure 28: Elements allowing generation of graphical output

Table 18: Sample algorithm for trace data generation

Element: Execute Name: Data Scenarios:

All

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As Simphony.General.Trace)

As System.String

Dim numofhoist as Integer

Select case context.Scenario.Name

Case "1 Hoist-CS"

numofhoist = 1

Case " 2 Hoist-CS"

numofhoist= 2

End select

Return " |Senario| "& context.Scenario.name & " |numofhoists| "

&Cstr(numofhoist) &" |no. of workers-Actual| "&

Cstr(Context.Scenario.Ints(203))& " |Average Arrival interval| "&

63

Cstr(context.scenario.floats(2))&" |no. of stories| "&

Cstr(Context.Scenario.Ints(213))&" |timeperfloor| "&

Cstr(context.Scenario.floats(1))&" |Capacity| "&

Cstr(context.Scenario.Ints(205))&" |Avg Wating time| "&

Cstr(context.Scenario.floats(6))&" |Max Wating time| "&

Cstr(context.Scenario.floats(3))

End Function

End Class

4.5 Decisions during Model Development

This section summarizes the modeling decisions and iterations that have been undertaken with

details provided in later sections.

1. Determination of Inputs: When the model was first developed, all the factors affecting the

hoist (see Table 3) were user-defined inputs. However, to reduce the number of user-defined

inputs, the impact of each input on the model output was studied, resulting in the number of

user-defined inputs being reduced from 12 to five. This was achieved by embedding the other

variables in the model and using typical values. The results have been compared to those

from other studies and similar results were obtained. Section 4.6 outlines the specific factors

and the impact analysis of the input variables is presented in Chapter 5.

2. Determination of output/decision variables: Several output statistics were tested for their

ability to adequately describe hoist productivity. These include:

The hoist’s cycle time

Total time for the delivery of workers

Percentage of workers waiting a certain time

Average waiting time of workers

It was decided that the hoist’s cycle time and total time for the delivery were not meaningful

indicators of productivity because they are both dependent on the height of the building.

Further, the time lost by workers waiting for the hoist is not reflected in these factors. The

percentage of workers waiting greater than 5, 10 and 15 minutes was also considered. While

64

this factor provides how many workers wait in the queue, the total time lost waiting for the

hoist is not reflected.

The average waiting time has been widely used in the planning of elevators, but not for the

planning of hoists. While the average waiting time provides an overall picture of the

operation, it is limited in the ability to provide feedback on the times that the hoist

effectiveness is reduced. For example, very high waiting times by a few workers may not be

reflected in the average. Therefore, several graphs have been added as outputs.

First, a worker delay graph provides an overview of the delay of each worker and when the

delays are high. Second, a cumulative delay graph provides the percentage of workers

waiting a specific duration and enables the user to make decisions based on their tolerance

for delay. For example, the project manager will be able to decide whether 10% of the

workers waiting for more than 15 minutes is acceptable. Section 4.7 contains a detailed

description of the model output.

3. Modelling the arrival of workers: The arrival of workers is a significant factor that has not

been previously considered in hoist operation studies. Using observations from site and

expert opinion, it was noted that the rate of worker arrival is dependent on and specific to the

site. Therefore, three different distributions were tested: exponential, uniform and a constant

interval. Surprisingly, the average waiting time output of the model was reasonably similar

for all three. While the distribution did not change the average waiting time for the workers,

the time between arrivals of workers has a major impact on the worker delays. Thus, this

factor must be input by the user. The step function was introduced to allow the user to better

organize the arrival of workers. While for the current operation analysis a constant inter-

arrival rate has been used, the modeller could select an alternative distribution as required.

Site observations indicated that workers arrive slightly before the beginning of the shift and

gather at a trailer. Once the shift begins, the workers collectively make their way to the hoist

to be transported. Therefore, for this analysis, a constant distribution has been used. This

allowed for the analysis of all workers arriving at once. To provide a distinction for the

reader, other inter-arrival rates within the constant distribution have also been considered.

4. Categorical vs. Specific Inputs: The use of categorical inputs as opposed to specific inputs

was studied. Table 19 displays how the inputs were categorized for the modelling process.

65

Instead of inputting the actual value for a variable, a category is selected. However, when the

output was examined, it was evident that the categorical inputs did not provide the required

accuracy. The averaging of the results over a range of values provided an output which was

not indicative of the true delay observed on site. Therefore, the model inputs have been

designed to be a specific value dependent on the project.

Table 19: Input variables categories

Case Number of

Workers

Mean time

between

arrivals (min)

Number of

stories

Time per

story (min)

Capacity

0 30-200 0.08-0.1 0-100 0.01-0.3 20-40

1 30-60 0.08-0.1 0-20 0.01-0.05 20-25

2 61-90 0.125-0.167 20-40 0.05-0.1 25-30

3 91-120 0.25-0.5 40-60 0.1-0.15 30-35

4 121-150 60-80 0.15-0.2 35-40

5 151-180 80-100 0.2-0.25

6 181-200 0.25-0.3

5. Planning Interface: Once the hoist operation was modelled and the results were verified, a

planning interface was developed to simplify the process of describing alternative strategies

to the hoist operation. For example, the user may choose to schedule the worker arrivals at

different start times and the floors to which they are scheduled to be transported. This

proposed strategy for minimizing delays will be tested in Chapter 6.

4.6 Model Factors

Using the literature review of hoist and elevator operation, along with expert opinions, the

factors affecting the up-peak hoist operation and the decision variables have been categorized as:

Factors as user-inputs

Factors built into the model

66

4.6.1 Factors as user-inputs

The five user-input variables reflect the situation being modelled, and may change with project

type, project stage, or project management strategy.

1. Number of Workers is an input variable because it represents the project size and stage. For

example, the user may determine that for a given project the number of workers is high only

toward the end of the project and that the lower performance of the hoist for this time may be

acceptable.

2. Inter-arrival rate is the average time between each worker. The project manager could use

his/her experience to determine the average arrival time for their project. This may also be

controlled by the project manager by staggering the start times and therefore arrival of

workers.

3. Number of Stories is naturally dependent on the project, but it is also dependent on the stage

of construction. The user may want to examine hoist operations part way through the project

to determine when an additional hoist might be needed.

4. Average Travel Time per Story is dependent on the speed of the hoist and average height of

the story. Therefore, it depends on the project under construction and the hoist model. The

user must perform the following calculation for this input:

Equation 16: Calculation of average travel time per story

𝑻𝒓𝒂𝒗𝒆𝒍 𝑻𝒊𝒎𝒆 𝑷𝒆𝒓 𝑺𝒕𝒐𝒓𝒚 =𝑨𝒗𝒆𝒓𝒂𝒈𝒆 𝒉𝒆𝒊𝒈𝒉𝒕 𝒑𝒆𝒓 𝒔𝒕𝒐𝒓𝒚(𝒎)

𝑺𝒑𝒆𝒆𝒅 𝒐𝒇 𝑯𝒐𝒊𝒔𝒕 (𝒎

𝐦𝐢𝐧)

5. Capacity of the hoist reflects the number of workers the hoist could deliver per trip, and can

be approximated by the size of the hoist and the total load it could carry as provided by the

manufacturer.

4.6.2 Factors built into the model

Six factors have been built into the model and their values have been approximated using the

literature and expert opinion. One can see that the variability of the values is not significant

relative to the travel time, so their impact is small. Furthermore, the amount of time required for

67

the completion of these tasks is minimal in comparison to cycle time of a hoist. Table 20

provides a summary of the built-in factors and their approximated values.

Table 20: Values of factors built into the model

Factor Value

Loading/unloading Time 10-15s

Door open/Close Time 10-15s

Acceleration/Deceleration

Time

20-30s/stop

Number of stops Randomly generated, depending on the number of workers and

arrival distribution and rate

Number of Hoists Each scenario represents a different number of hoists, either

single or double.

Distribution Type constant inter-arrival or step function

4.7 Model Output

The main decision statistic used to portray the hoist performance is the average waiting time of

workers.

Average waiting time of workers represents the average time a worker waits for the hoist at the

start of the day.

Graphical Output presents data from which the user may make decisions about the operation of

the hoist.

The delay per worker (Figure 29) shows the delay of each worker in order of their arrival. This

graph shows how the hoist is performing through the morning rush and reflects when the hoist

has high delay times. In this case, the delay begins to increase for a large number of workers

after the hundredth worker arrives. This graph could be used as a basis of comparing the

performance of different hoists.

68

Figure 29: Example of graph Delay per Worker

The second graph, cumulative distribution of delay shown in Figure 30, allows the user to

determine the percentage of workers waiting more than a specific duration. This graph is more

flexible than delay per worker because it allows the user to make decisions based on their

tolerance for delay.

Figure 30: Example of graph Cumulative Distribution of Delay

4.8 Model Scenarios

The model contains four scenarios, which model four situations:

-5

0

5

10

15

20

25

30

0 50 100 150 200 250

Wai

tin

g Ti

me

(m

ins)

Worker in order of arrival

Delay per Worker

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0 20 40 60 80 100 120

Per

cen

tag

e o

f w

ork

ers

Waiting Time (mins)

Cumulative Distribution of Delay

69

Single hoist, current hoist operation

Single hoist, alternative scheduling strategy

Double hoist, current hoist operation

Double hoist, alternative scheduling strategy

This will allow the modelling of either one or two hoists for both the current and alternative hoist

operations. Thus, a decision could be made on the number of hoists provided the inputs selected

by the user. Also, this makes the model flexible to expansion to more hoists, if required.

4.9 Scenario components

This section is a description of how model components, such as entities and resources, have been

used in the context of the model development.

4.9.1 Model entities

There are three types of entities in the model. First, one entity is released at the start of the

simulation to activate global variables. This entity does not have a physical representation, but is

used to activate an execute element. Second are the entities that represent workers. Each entity

represents a single worker with attributes that define its arrival time and floor destination. The

third type of entity is to represent the movement of the hoists.

4.9.2 Model resources

Two resources were used in the model. First, when a hoist is launched, it is captured as a

resource and travels with the hoist entity described in the previous section. Second, resources

were used to ensure that one worker at a time entered an event by making them capture a

“Pseudo” resource before entering the event and releasing it after completion. Since the other

workers have to wait for the “Pseudo” resource to be available, only one worker enters the event

at a time.

4.9.3 The model programming

To model the operations of the hoist accurately, algorithms programmed using the Visual Basic

.Net language were used. These algorithms were essential for model components such as:

70

The assignment of global variables according to user input,

The launch of the hoist once it is at capacity, and,

The output of data once all workers have been transported.

Sample algorithms have been selected and presented in Appendix 2 to provide insight to the

modelling method.

4.9.4 Model variables

To model the hoist operation, several global and local arrays were used, as shown in Table 21.

Simphony.Net allows the user to assign variables (attributes) specific to the scenario (Global) or

entity (local).

Table 21: Definition of model variables

Scenario (Global) Variables

Type Number Description

Integer

1 to 190 Number of Stops on each floor

401-590

200

Counters

300

211

215

216

201 Last stop for hoist

301

202 Current floor that hoist is on

302

203 Number of workers

205 Capacity of hoist

207 Number of Cycles by hoist

208

212 Last stop modelled

213 Number of Floors

Floats /double

1 Time per floor

2 Mean arrival rate

3 Maximum waiting time

5 Cumulative delay

6 Average delay

71

Entity (local) Variables

Entity Type Number Description

Worker

Integer 0 Floor number

Floats /double

0 Time of arrival

1 Waiting time

2 Time of first loading of hoist

Hoist Integer 2 Hoist ID

4.10 Model Verification and Validation

To ensure that the model is functioning as intended, a rigorous verification process has been

undertaken. Furthermore, validation of the model was conducted using results from comparable

studies and site data.

4.10.1 Model verification

The model components and their interactions were verified. First, each component was run

separately to ensure that the component provides the desired task. An output with the intended

function of that component was used to verify each component separately. For example, the

output of the arrival of workers was the:

Number of workers

Time of Arrival of each worker

Floor assigned to each worker

Interval time

Capacity of hoist

Loading of hoist: Time hoist waits for loading and number of workers loaded

Time the first worker loads the hoist for each cycle

Time of hoist launch

Waiting time per worker

The algorithms in each component are functioning as intended. Furthermore, the interactions

between the components have been verified using the same procedure. For example, when a

hoist is at capacity, the output was that the hoist is launched and when the hoist returns, the

output was the start of the loading of the hoist.

72

During the development of the model, the trace element was used to track each algorithm. The

trace element tracked variables, entity number, and time during a run in each step of the model to

ensure that all the items perform in accordance to their intended use. By tracking multitudes of

possibilities, the author believes that the algorithms are functioning as expected.

4.10.2 Validation

Three data sources were used to validate the model: two case studies from the literature and one

with data collected from a current site. The three scenarios compare three different aspects of the

hoist’s operation, the total time to deliver the workers, the cycle time of the hoist and the average

waiting time of the workers. Each of the results provided values similar to those from site data as

shown in Table 22.

Table 22: Validation through case studies

Case Study 1:

(Cho et al. 2010)

Case study 2:

(Park et al. 2013)

Case study 3:

Site observations

Number of workers 180 205 46

Hoist capacity 20 18 30

Hoist speed 100 m/min 70 m/min 90 m/min

Number of hoists 1 1 1

Number of floors 62 51 23

Time to deliver

workers

74 min 27 min 5 min 30 sec

Model results 80 min 25 min 5 min 6 sec

73

4.11 Planning Options

This section reviews the inputs that can be changed by the user to vary site characteristics, host

characteristics, stage of construction, and worker schedules.

4.11.1 Site characteristics

The project characteristics that can be varied include the number of stories, number of workers,

and worker arrival rate.

4.11.2 Hoist characteristics

The user chooses the hoist model using the manufacturer’s model or by providing general

parameters, such as speed and capacity. The model has been programmed to provide the solution

for both single and double hoists so that the user can easily compare the performance of each.

4.11.3 Stage of construction

As the construction of a high-rise progresses, demand for the hoist may change. For example,

the maximum demand for the hoist may occur just before the building is topped off and the

elevators are put in service. By varying the number of stories, number of workers, and worker

arrival rates, the demand for the hoist could be evaluated throughout the project’s stages.

4.11.4 Worker schedules

The modeller may alter worker scheduling to affect the arrival rate to improve the hoist’s

performance. Scheduling strategies could be selected for different project stages.

4.12 Chapter Summary

This chapter demonstrated the development of the model used for the analysis of the hoist

performance. First, a description of the modeling software, Simphony.Net, was reviewed.

Second, the model was presented by defining its components and their functions. The model has

four main components: (i) variable input, (ii) arrival of workers and loading of hoist, (iii)

operation of hoist, and, (iv) generation of output. Furthermore, it has four scenarios to represent

one or two hoists and the current or alternative operation.

The model has 5 inputs required by the user:

74

Number of Workers

Inter-arrival rate of workers in the morning

Number of Stories

Average Travel Time per Story

Capacity of the hoist

Other factors have been built-in to the model, including the acceleration /deceleration of the hoist

and the time to open and close the doors. This allows for the reduction of the number of inputs

required by the user.

In addition, the model provides three output statistics for the user:

Average waiting time

Delay by Worker Graph

Cumulative distribution of delay

75

Chapter 5 Impact of Model Inputs

5

In this chapter, the analysis that was conducted to study the performance of the hoist will be

discussed. The impact of changes will be demonstrated through graphs and sensitivity analysis.

Finally, a case study will demonstrate the effect of using the alternative strategy for planning the

operation of the hoist.

This section examines the impact of the factors on the performance of the hoist. For this analysis,

a single factor has been changed in each case while the other factors have been controlled. Table

23 summarizes the cases and provides the reader with the analysis section.

Table 23: Summary of impact studies

Case No. of

Hoists

Distribution

of Arrival

No. of

Workers

Rate of

Arrival

(mins)

No. of

Floors

Time per

story(mins)

Hoist

Capacity

(persons)

1

(Section

5.1)

2 Changed 200 2.5 70 0.04 30

2

(Section

5.2)

2 Constant 200 Changed 70 0.04 30

3

(Section

5.3)

2 Constant 200 2.5 70 Changed Changed

4

(Section

5.4)

2 Constant Changed 2.5 Changed 0.04 30

5.1 Distribution of the arrival of workers

Using the data collected and expert opinion, it was discovered that the way in which workers

arrive to site is dependent on the site and company. Figure 31 provides a sample analysis by

76

which all the factors, except the distribution type, are constant so that an understanding of the

impact of the distribution type on the average waiting time by the workers can be achieved.

Figure 31: Examination of arrival distributions

The results show that for the three distributions, exponential, uniform and constant, the output of

the model is similar although the variance is different. This provides the reader with insight on

the difference in the results depending on the selected distribution.

5.2 Inter-arrival rates of workers

To study the impact of changing the inter-arrival rate of workers, all other variables were kept

constant. In this case, workers arrive individually at evenly spaced time intervals. The average

waiting time (Figure 32) clearly shows that as the inter-arrival time increases, the average

waiting time decreases. This decrease seems to reach an asymptotic minimum value. The slight

decrease and then increase at the lower values of the function is due to workers arriving before

the first hoist cycle is complete. When workers arrive after the first cycle has completed, there is

a slight increase. Finally, as the inter-arrival rate increases further, AWT continues to decrease

until it reaches the minimum value. This examination shows that the inter-arrival time has a

0

2

4

6

8

10

12

14

16

18

Ave

rage

wai

tin

g ti

me

Impact of Distribution Type on Average waiting Time

Exponential Uniform Constant Average

77

major impact on the hoist, and that there is an opportunity to improve hoist performance by

controlling worker arrivals.

Figure 32: Study of the impact of the arrival rate on the average waiting time

The second output was the Worker Delay graph (Figure 33). Four inter-arrival rates were studied

(0, 0.3, 0.6, and 0.9 minutes). When all the workers arrive at once i.e. inter-arrival rate = 0, the

delay graph reflects a step function with each step coinciding with the hoist’s cycle time. This

investigation demonstrates how the inter-arrival rates impact the delay the workers are subjected

to. Furthermore, the delay does not have a steady increase.

The final output is a cumulative distribution of the delay (Figure 34). It is evident that the delay

of workers is drastically decreased by increasing the average inter-arrival time. This is an

indication of the importance of this factor on accurately predicting the performance of the hoist

and the use of this factor as an input to the model.

0

5

10

15

20

25

30

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Ave

rage

Wai

tin

g ti

me

Time between Workers (mins)

Impact of Average Inter-Arrival Rate on the Average Waiting Time

78

Figure 33: Worker Delay graph using inter-arrival rates

Figure 34: Cumulative distribution showing effect of inter-arrival rates

5.3 Impact of Hoist Characteristics

To study the impact of the hoist speeds and capacity, all the other variables are constant. Figure

35 displays the impact of the hoist characteristics on the average waiting time. The three lines

0

10

20

30

40

50

60

70

80

0 20 40 60 80 100 120 140

Wai

tin

g Ti

me

(min

s)

Worker in order of arrival

Worker Delay for Inter-Arrival Rates (min)

0

0.3

0.6

0.9

0.00%

20.00%

40.00%

60.00%

80.00%

100.00%

120.00%

0 10 20 30 40 50 60 70 80

Per

cen

t o

f W

ork

ers

Waiting time

Cummulative Delay by Inter-Arrival Rate (min)

0

0.3

0.6

0.9

79

represent the hoist’s speed measured in meters per minute. Each line connects three points of

hoist capacity for that speed.

Figure 35: Examination of hoist speeds and capacity on the performance

It is evident that the hoist speed has a greater impact on the performance of the hoist than the

capacity. However, both variables affect the average waiting time.

5.4 Impact of the number of workers and number of floors

Figure 36 presents the change in the average waiting time with a change in the number of

workers for buildings of 20, 60 and 100 floors. As the number of workers increases, the average

waiting time increases. For buildings of 20 stories, at some point the change in the number of

workers reaches a maximum average waiting time due to the lower cycle time of the hoist.

A similar analysis was conducted for the change in the number of floors to that of workers. The

change of the average waiting time due to the change of the number of floors is represented for

90, 150 and 210 workers in Figure 37. Note that for buildings with fewer stories, the wait

times are similar without regard to the number of worker, which is consistent with Figure

Figure 36. The increased hoist cycle time due to height directly impacts the performance of the

0

5

10

15

20

25

30

35

40

15 20 25 30 35 40

Ave

rage

Wai

tin

g Ti

me

Capacity (persons)

Impact of Hoist Speed (minutes/metre) on AWT

45

90

140

80

hoist. This reaffirms the importance of improving the hoist operation as the buildings increase in

height.

Figure 36: Examination of the impact of the number of workers.

Figure 37: Examination of the impact of the number of floors

0

2

4

6

8

10

12

14

16

18

20

0 50 100 150 200 250

Ave

rage

Wai

tin

g ti

me

(min

s)

Number of Workers

Impact of Number of Workers by Number of Workers

100

60

20

0

5

10

15

20

25

30

0 20 40 60 80 100 120 140 160

Ave

rage

wai

tin

g ti

me

Number of Floors

Impact of Number of Floors

210

150

90

Number of

Workers

Number of

Floors

81

Figure 37 also suggests a linear function, so this analysis is provided to supplement

understanding of the impact of the number of floors on the average waiting time. Three

functions were investigated: linear, quadratic and power and they have the general formulas:

Equation 17 : Linear trend-line equation

𝑌 = 𝐴𝑥 + 𝐵

Equation 18: Quadratic trend-line equation

𝑌 = 𝐴𝑥2 + 𝐵𝑥 + 𝐶

Equation 19: Power trend-line equation

𝑌 = 𝐴𝑥𝐵 + 𝐶

Figures Figure 38, Figure 39 and Figure 40 demonstrate the fit of these functions. Using the R-

Squared parameter to compare the fits, it is evident that any of the three functions could

reasonably represent the relationship. However, by observing the trends, the linear function is not

a good representation for the lower values of the number of floors. Furthermore, the power

function deviates at the higher number of floors. The quadratic function provides the best fit for

the representation of this relationship. This suggests the large impact that the number of floors

has on the hoist productivity. Table 24 summarizes the R-Squared results with the best fit per

worker category bolded. The quadratic functions provide the best overall fit based on the R-

squared and on examination of the graphs.

Table 24: Summary of R-Squared values for different fits

Function R-Squared

90 workers 150 workers 210 workers

Linear 0.9532 0.9733 0.9764

Quadratic 0.9863 0.9806 0.9533

Power 0.9707 0.9655 0.9691

82

While each equation provides the relationship between the number of floors and the average

waiting time, the different equations represent the change due to the number of workers. It is

evident that the coefficients of the equations are higher as the number of workers increase. This

shows that with a higher number of workers, the impact of the number of stories on the average

waiting time increases.

Note that the lines on the linear graph cross at about 20 floors. This is explained by the inter-

arrival rate selected for generating these graphs. At about 20 floors, the cycle time of the hoist is

in tandem with the arrival of workers, thus all the lines cross.

Figure 38: Study of linear fit as a model for number of floors

For 210 workers y = 0.1891x - 2.1775

R² = 0.9764

For 150 workers y = 0.1288x - 0.9293

R² = 0.9733

For 90 wokers y = 0.0686x + 0.3854

R² = 0.9532

-5

0

5

10

15

20

25

30

35

0 20 40 60 80 100 120 140 160

Ave

rage

wai

tin

g ti

me

Number of Floors

Impact of Number of Floors With Linear Fit

83

Figure 39: Study of quadratic fit as a model for number of floors

Figure 40: Study of power fit as a model for number of floors

For 210 workers y = 0.0005x2 + 0.1138x - 0.34

R² = 0.9863

For 150 workers y = 0.0003x2 + 0.0853x + 0.1718

R² = 0.9806

For 90 workers y = 2E-05x2 + 0.0663x + 0.4406

R² = 0.9533

-5

0

5

10

15

20

25

30

0 20 40 60 80 100 120 140 160

Ave

rage

wai

tin

g ti

me

Number of Floors

Impact of Number of Floors With Quadratic Fit

For 210 workers y = 0.0931x1.1084

R² = 0.9707

For 150 workers y = 0.126x0.9724

R² = 0.9655

For 90 workers y = 0.1775x0.7992

R² = 0.9691

0

5

10

15

20

25

30

0 20 40 60 80 100 120 140 160

Ave

rage

wai

tin

g ti

me

Number of Floors

Impact of Number of Floors With Power Fit

84

Chapter 6 Using the Model to Improve Hoist Performance

6

This section provides the analysis of a hypothetical project to test the impact of two strategies on

the performance of the hoist. A tall building with 90 stories is considered with 230 workers and a

hoist capacity of 30 workers.

The hoist operating strategy will look at three schedules. Schedule A is where workers arrive all

at once at the start of the work day. Schedule B is where workers arrive individually throughout

one hour. For simplicity, this was modeled with equal interarrival times. For example, 230

workers arriving over one hour means that one worker arrives every 15.65 seconds or 0.26

minutes. In reality, this might represent cases where subcontractors first meet with their workers

for safety talks before going up to work, thereby delaying their arrival at the hoist. Finally,

Schedule C represents zoning and staggered starts, where the floors are separated into 5 zones

and workers who are scheduled to work on the higher zones are taken their first (zoning) or

arrive earlier (staggered starts). The groups represent the work underway at each level such as

formwork, electrical and plumping. The number and destination of the workers was developed

using expert opinion. Table 25 summarizes the variables and Table 26 summarizes the arrival

schedules.

Table 25: Inputs used in the analysis

Case 1 2 3 4 5 6 7

Arrival Schedule (described in Table 26) A A B B C

0

min

C

15

min

C

20

min

Hoist Travel Time per Floor (min.) 0.04 0.025 0.04 0.025 0.04 0.04 0.04

Number of Floors 90

Number of Workers 230

Hoist Capacity (people) 30

85

Table 26: Arrival schedule details

Arrival Type A: Workers arrive together at once

Arrival Type B: Workers arrive individually over 1 hour

Arrival Type C – Zoning & Staggered Starts

Workers Going to floors Case 5:

Arrive at

Case 6:

Arrive at

Case 7:

Arrive at

50 81-90 07:00 07:00 07:00

30 51-80 07:00 07:15 07:20

65 26-50 07:00 07:30 07:40

35 1-25 07:00 07:45 08:00

50 1-25 07:00 08:00 08:20

Cases 1 and 2 are the same except that case 2 has a faster hoist. Likewise, cases 3 and 4 have the

same arrivals, but case 4 has a faster hoist.

Based on the model outputs, a Worker Delay graph (Figure 41) and a Cumulative Delay graph

(Figure 42) were produced. These results will be used as the means of comparing the impact of

the alternative schedules to delay of workers.

In Figure 44, the change in speed does not have as much impact as changing the arrival schedule.

For example, the reduction of wait time for the 230th

worker between case 1 and 2 is just 10%

whereas the reduction between cases 1 and 3 is 18%. This is also observed in the cumulative

distribution function (Figure 42).

86

Figure 41: Results showing worker arrival cases using the delay graph.

Case 1- arrive together and slower hoist Case 5- arrive together and zoning

Case 2-arrive together and faster hoist Case 6- staggered starts (15 minute intervals)

Case 3-arrive individually and slower hoist Case 7- staggered starts (20 minute intervals)

Case 4- arrive individually and faster hoist Cases 5, 6, and 7 use the faster hoist

-20

0

20

40

60

80

100

120

0 30 60 90 120 150 180 210

Wai

tin

g Ti

me

(m

ins)

Worker in order of arrival

Delay per worker

1

2

3

4

5

6

7

87

Figure 42: Results showing worker arrival cases using the cumulative distribution.

Case 1- arrive together and slower hoist Case 5- arrive together and zoning

Case 2-arrive together and faster hoist Case 6- staggered starts (15 minute intervals)

Case 3-arrive individually and slower hoist Case 7- staggered starts (20 minute intervals)

Case 4- arrive individually and faster hoist Cases 5, 6, and 7 use the faster hoist

Scheduling workers according to their destination is a novel strategy in hoist planning. In the

cases where the workers arrive together at the beginning of the shift, there is an improvement

attained from just zoning. However, this improvement is not as significant as the cases where the

workers from each zone are scheduled to arrive at different times. Case 5 demonstrates that while

zoning has a 20% reduction in the waiting time of the 230th

worker in comparison to case 1, the

reduction attained from case 6 (combination of staggered arrivals and zoning) has a reduction of

about 85%. Figure 43 highlights these reductions in waiting times.

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

90.00%

100.00%

110.00%

0 20 40 60 80 100

Pe

rce

nt

of

Wo

rke

rs

Waiting Time (mins)

Percent Waiting

1

2

3

4

5

6

7

88

Figure 43: Highlighting the impact of the Zoning on the hoist performance.

Case 1- arrive together and slower hoist Case 4- arrive individually and faster hoist

Case 6- staggered starts (15 minute intervals and faster hoist )

In the cases where the workers arrive at once and over one hour (schedules A and B), there is an

increasing trend of waiting time. While cases 3, 4 and 6 all have the workers arrive within one

hour, scheduling staggering starts as in case 6 greatly reduces the waiting times of the workers.

This comparison is highlighted in Figure 44.

-20

0

20

40

60

80

100

120

0 30 60 90 120 150 180 210

Wai

tin

g Ti

me

(m

ins)

Worker in order of arrival

Delay per worker

1

4

6

89

Figure 44: Highlighting the impact of the alternative strategy over one hour arrival time.

Case 3-arrive individually and slower hoist Case 5- arrive together and zoning

Case 6- staggered starts (15 minute intervals and faster hoist )

Table 27, demonstrates the benefit of the proposed scheduling strategy by comparing the delay

of 80% and 50% of workers in the cumulative graph.

Staggered starts in 15 min intervals (case 6) as opposed to having workers start at once (case 1)

reduces the 80th

percentile wait by over one hour and 45 minutes by the 50% percentile. The

specific schedule would depend upon the stage of construction, the local practices and/or

regulations, and the ability of the site manager to affect the working schedule of the

subcontractors. However, the opportunity to reduce wait times should be very appealing to

subcontractors.

-10

0

10

20

30

40

50

60

70

80

0 30 60 90 120 150 180 210

Wai

tin

g Ti

me

(m

ins)

Worder in order of arrival

Delay per worker

3

5

6

90

Table 27: Cumulative delay for 50% and 80% of workers

Case Cumulative delay for the 50th

percentile of workers (mins)

Cumulative delay for 80th

percentile of workers (mins)

1 51 79

2 45 71

3 29 46

4 21 36

5 38 58

6 6 11

7 0 4

In conclusion, this study demonstrates the opportunity of using the proposed staggered starts

strategy for the improvement of the hoist productivity.

91

Chapter 7 Conclusion and Recommendations

7

This chapter summarizes the contributions of this research, its limitations, and recommendations

for future research.

7.1 Conclusions

Tall buildings are becoming more common in North America, including in the Toronto area. The

challenges of moving resources vertically for tall buildings are based upon the increase of wind

speeds, longer travel times and limited space for the construction in the city. This research

examined the effectiveness of hoist operations and its effect on worker delays.

A comprehensive overview of the previous efforts to study hoists was presented. However, due

to the scarcity of studies that looked at the hoist, research that studies elevators was also

reviewed. It was observed that previous efforts to study hoists had limitations. First, a large

number of inputs are needed by some of the models. Second, the measures of hoist performance

do not necessarily reflect the domino effect of hoist operations. Finally the way in which workers

arrive to the site has not been studied, which has been a common factor in elevators.

A discrete-event simulation (DES) model was developed for this project. DES provides a

mechanism to model complex systems and of accounting for statistical variability.

To develop the model, data from site observations and expert opinion were collected to represent

the transportation of workers during the morning up-peak. The inputs required by the model

were limited to factors that have a high impact on the output. Also, the way in which workers

arrive to the site in the morning was studied and included in the model. Finally, the model

provides three outputs: average waiting time, delay graph and cumulative delay distribution to

represent the performance of the hoist. The average waiting time is a common decision variable

that has been used in elevator planning, but has not yet been adapted to hoists. The model was

verified and validated. The results show that the model’s output is reflective of the operation of

the hoist.

92

An analysis of the impact of the factors on the performance of the hoist has been conducted. It is

evident that each of the factors has an influence on the hoist’s performance. Furthermore, an

application for the model to improve hoist performance was undertaken to study the impact of

the alternative scheduling strategy on the delay of workers. It was concluded that scheduling

staggered starts for the workers in accordance to their location of work provides a significant

improvement to the hoist’s productivity.

The object of this study was to optimize hoist performance for high-rise building construction,

enabling an efficient delivery of workers during the morning peak. The impact of staggered

arrivals and zoning on the performance of the hoist provides a solution and meets the objective

of the study.

7.2 Limitations of the study

Provided the scope and assumptions of this study, a few limitations arise. The limitations of this

study are as follows:

This study is limited to the delivery of workers. However, the hoist may be used for the

delivery of materials. Therefore, in situations where the hoist is being used for both

materials and workers, the model is not reflective of the hoist operation.

The wind speeds as a building gets higher may cause malfunctions in the hoist. The hoist

might operate at lower speeds or completely shut-down. This study did not take these

situations into consideration.

Some site activities during the morning peak may not be reflected in the model. For

example, if there is inter-floor travel or if workers are using the hoist to return to the

ground floor during the morning peak.

The hoist operator’s characteristics, such as skill or experience, have not been considered

in this study.

The different hoist technologies have not been taken into account. A self-leveling hoist or

hoists with an automatic control system may produce productivity improvements.

93

The staggered starts strategy has not been validated using data collected from site as it

has not been tried. Therefore, the performance of the hoist in these idealistic situations

may be altered by factors which have not been taken into consideration.

The operation of the building’s elevators has not been taken into account as it is assumed

that they are not yet available for use.

7.3 Recommendations

This study provides a starting point for many possible future endeavours to further research hoist

operations and improving hoist productivity. Some of the aspects that future researchers may

undertake are:

The model focused on the delivery of workers in the morning. While this aspect is

commonly used in elevators, the delivery of material is specific to hoists. Therefore, the

model could be expanded to include the delivery of materials. Furthermore, the model

may also be extended to different times of day and to model construction progress

throughout the project. This requires the addition of the inter-floor travel, travel during

breaks and the end of the shift, along with the delivery of materials.

Aspects that have an influence on the performance of the hoist include new hoist

technologies, location of the hoist and platform design. Future research may study the

impact of these factors on the performance of the hoist. The use of a self-leveling hoist,

for example, may be an aspect which improves the hoist performance. Additionally, the

location of the hoist and the platform design may influence the loading and unloading

times of the hoist.

The study of the hoist operation using different methods may provide more insight on

ways of improving the hoist’s performance. The analysis of this study suggests that a

linear regression may be used to represent the hoist’s operation. Moreover, artificial

intelligence methods could be an alternative method of analysis.

An automated method for inputting the information into the model, and providing an

output may be developed to make the model more user-friendly.

94

For the purpose of decision-makers, graphical representation in the form of design graphs

for different situations may be created. This would make the decision making process

easier and faster. The graphical format is more industry-friendly and does not require

knowledge in DES.

95

Bibliography

AbouRizk, S. (2014). NSERC Industrial Chair, Construction Management and Engineering,

Software. Retrieved from University of Alberta:

http://irc.construction.ualberta.ca/en/Research/ChairProfile/Software.aspx

AbouRizk, S. (2014). "Simphony.Net: An integrated enviroment for construction simulation."

NSERC IRC in Construction Engineering and Managmenet-Simulation:Simphony, 12.

Ahuja, H., Dozzi, S., and AbouRizk, S. (1994). "Project Management Techniques in Planning

and Controlling Construction Projects." New York: Wiley & Sons Inc.

Altiok, T., and Melamed, B. (2007). "Simulation Modeling and Analysis with ARENA." Elsevier.

Arora, J. S. (2012). "Introduction to Optimum Design, Third Edition." San Diego, San Fraciso,

Singapore, Sydney, Tokyo: Elsevier.

Banks, J. (1998). "Handbook of Simulation: Principles, Methodology, Advances, Application and

Practice." New York: John Wiley & Sons, Inc.

Barney, G. C., and Santos, S. M. (1975). "Improved Traffic Design Methods for Lift Systems."

Building Science, 277-285.

Belegundu, A. D., and Chandrupatla, T. R. (2011). "Optimization Concepts and Applications in

Engineering (2nd Edition)." New York: Cambridge University Press.

Benmakhlouf, S. M., and Khator, S. K. (1993). "Smart Lifts: Control Design and Performance

Evaluation." Computers and Industrial Engineering, 175-178.

Bigge Crane and Rigging Company. (2014). "Construction Hoist Rental." Retrieved from Bigge

Crane and Rigging Co.: http://www.biggetowercrane.com/construction-hoists.html

Bingham, N. H., and Fry, J. M. (2010). "Regression-Linear Models in Statistics." London:

Springer-Verlag.

96

Brown, W. M., and Newbold, K. B. (2012). "Cities and Growth: Moving to Toronto –Income

Gains Associated with Large Metropolitan Labour Markets." Statistics Canada-Economic

Analysis Division.

Chang, D. (1987). "RESQUE PhD Thesis." University of Michigan: Ann Arbor, Michighan

Chen, H.-M., and Huang, P.-H. (2013). "3D AR-Based Modeling for Discrete-Event Simulation

of Transport Operations in Construction." Automation in Construction, 123-136.

Cho, C.-Y., Cho, M.-Y., Kim, Y.-S., Kim, J. Y., Lee, J. B., and Kwon, S.-W. (2010).

"Simulation Method of Construction Hoist Operating Plan for High Rise Buildings

Considering Lifting Heights and Loads." 27th International Symposium on Automation

and Robotics in Construction, 22-28.

Cho, C.-Y., Kwon, S., Shin, T.-H., Chin, S., and Kim, Y.-S. (2011). "A Development of Next

Generation Intelligent Construction Liftcar Toolkit for Vertical Material Movement

Management." Automation in Construction, 14-27.

Cho, C.-Y., Yoosub, L., Cho, M.-Y., Kwon, S., Shin, Y., and Lee, J. (2013). "An Optimal

Algorithm of the Multi-Lifting Operating Simulation for Super-Tall Building

Construction." Automation in Construction, 595-607.

Christodoulou, S., Ellinas, G., and Aslani, P. (2009). "Entropy-Based Scheduling of Resource-

Constrained Construction Projects." Automation in Construction, 919-928.

CIBSE. (1993). "CIBSE Guide D,Transportation Systems in Buildings." The Chartered

Institution of Building Services Engineers.

Cortés, P., Larrañeta, J., and Onieva, L. (2004). "Genetic Algorithm for Controllers in Elevator

Groups: Analysis and Simulation During Lunchpeak Traffic." Applied Soft Computing,

159-174.

CTBUH . (2014). CTBUH Height Criteria. Retrieved from Council on Tall Buildings and Urban

Habitat:

http://www.ctbuh.org/HighRiseInfo/TallestDatabase/Criteria/tabid/446/language/en-

GB/Default.aspx

97

CTBUH. (2011). "2013: A Tall Building Review." CTBUH Journal, 1-2.

CTBUH. (2012). "Canada Grows Taller." CTBUH Journal, 2-3.

de Neufville, R., and Scholtes, S. (2011). "Flexibility in Engineering Design." Cambridge,

Massachuetts, London: MIT Press.

Devor, J. L. (2009). "Probability amd Statistics For Engineering and the Sciences." Belmont:

Brooks/Cole.

Dozzi, S., and AbouRizk, S. (1993). "Productivity in Construction." Ottawa: National Research

Council Canada.

Gautschi, W. (2012). "Numerical Analysis, Second Edition ." New York,Dordrecht,

Heidelberg,London: Springer Science+Business Media.

Hajjar, D., and AbouRizk, S. (1999). "Simphony: An Environment for Building Special Purpose

Construction Simulation Tools." Proceedings of the Winter Simulation Conference,

Phoenix, Arizona, 998-1006.

Hajjar, D., and AbouRizk, S. (2002). "Unified Modeling Methodology for Construction

Simulation." Journal of Construction Engineering and Managment, 174-185.

Halpin, D., and Riggs, L. (1992). "Planning and Analysis of Construction Operations." Canada:

John Wiley & Sons, Inc.

Hauser, J. R. (2009). "Numerical Methods for Nonlinear Engineering Models." Dordrecht:

Springer Science+ Business Media.

Hwang, S. (2009). "Planning Temporary Hoists for Building Construction." Construction

Research Congress, 1300-1307.

Ioannou, P. (1989). "UM_CYCLONE." Dept. of Civil Engineering, University of Michigan: Ann

Arbor, Michigan

98

Ioannou, P. G., and Martinez, J. C. (1996). "Comparison of Construction Alternatives Using

Matched Simulation Experiments." Journal of Construction Engineering and

Management, 231-241.

Ioannou, P. G., and Martinez, J. C. (1996). "Scalable Simulation Models for Construction

Operations." Proceedings of the 1996 Winter Simulation Conference , 1329-1336.

Ladany, S. P., and Hersh, M. (1979). "The Design of an Efficient Elevator Operating." European

Journal of Operational Research, 216-221.

Levitt, R. E., Thomsen, J., Christiansen, T. R., Kunz, J. C., Jin, Y., and Nass, C. (1999).

"Simulating Project Work Processes and Organizations: Toward a Micro-Contingency

Theory of Organizational Design." Management Science, 1479-1495.

Martinez, J. C. (1996). "STROBOSCOPE: State and Resource Based Simulation of Construction

Processes." Ph.D. Dissertation, University of Michigan: Ann Arbor, Michigan

Martinez, J. C. (2010). "Methodology for Conducting Discrete-Event Simulation Studies in

Construction Engineering and Management." Journal of Construction Engineering and

Management, 3-16.

McDonough Elevators . (2013). "Construction Hoist Rental ." Retrieved from McDonough

Elevators : http://www.mcdelevators.com/construction-hoist-rental.html

Metro Elevator. (2013). "Construction Hoists for lease ." Retrieved from Metro Elevator:

http://metroelevator.com/construction-hoists-lease

Monahan, J. F. (2001). "Numerical Methods of Statistics." Cambridge: Cambridge University

Press.

Mooney, C. Z. (1997). "Monte Carlo Simulation." Thousands Oaks, CA: SAGE publications,

Inc.

Nagatani, T. (2003). "Complex Behavior of Elevators in Peak Traffic." Physica A, 556-566.

Nagatani, T. (2004). "Dynamical Transitions in Peak Elevator Traffic." Physica A, 441-452.

99

Naylor, T. J., Balintfy, J. L., Burdick, D. S., and Chu, K. (1996). "Computer Simulation

Techniques." New Work: John Wiley & Sons.

Nelson, P. R., Coffin, M., and Copeland, K. A. (2003). "Introductory Statistics for Engineering

Experimentation." Online : Elsevier.

Park, M., Ha, S., Lee, H.-S., Choi, Y.-k., Kim, H., and Han, S. (2013). "Lifting Demand-Based

Zoning for Minimizing Worker Vertical Transportation Time in High-Rise Building

Construction." Automation in Construction, 88-95.

Paulson, B. J. (1978). "Interactive Graphics for Simulating Construction Operations." Journal of

Construction Div., 69-76.

Peters, R. (2014). Elevate. Retrieved from Peters Research: https://www.peters-

research.com/index.php/elevate

Puri, V., and Martinez, J. C. (2013). "Modeling of Simultaneously Continuous and Stochastic

Construction Activities for Simulation." Journal of Construction Engineering and

Management, 1037-1045.

Raychaudhuri, S. (2008). "Introduction to Monte Carlo Simulation ." Proceedings of the 2008

Winter Simulation Conference, 91-100.

Ross, S. M. (2013). "Simulation (Fifth Edition)." Amesterdam, Boston, Hidelberg, London, New

York, Oxford, Paris, San Daiego, San Francisco, Singapore, Sydney, Tokyo: Elsevier Inc.

Rubinstein, R. Y., and Kroese, D. P. (2007). "Simulation and the Monte Carlo Method, Second

Edition." Hoboken, New Jersey: John Wiley & Sons, Inc.

Ruwanpura, J. Y., and Ariaratnam, S. T. (2007). "Simulation Modeling Techniques for

Underground Infrastructure Construction Processes." Tunnelling and Underground Space

Technology, 553-567.

Shi, J. (1997). "A Conceptual Activity Cycle-Based Simulation Modeling Method." Proceedings

of the 1997 Winter Simulation Conference, 1127-1133.

100

Shin, Y., Cho, H., and Kang, K.-I. (2011). "Simulation Model Incorporating Genetic Algorithms

for Optimal Temporary Hoist Planning in High-Rise Building Construction." Automation

in Construction, 550-558.

Skyscraper Center . (2014). Interactive Data. Retrieved from Skyscraper Center- The Global

Tall Building Database of the CTBUH: http://skyscrapercenter.com

Soekiman, A., Pribadi, K., Soemardi, B., and Wirahadikusuman, R. (2011). "Factors Relating to

Labor Productivity Affecting the Project Schedule Performance in Indonesia." Procedia

Engineering, 865–873.

Statistics Canada. (2014, 04 30). "CANSIM." Gross domestic product at basic prices, by

industry (monthly). Canada. Retrieved 05 12, 2014, from http://www.statcan.gc.ca/tables-

tableaux/sum-som/l01/cst01/gdps04a-eng.htm

Tervonena, T., Lahdelmabr, R., and Hakonen, H. (2008). "Elevator Planning with Stochastic

Multicriteria Acceptability Analysis." Omega, 352-362.

Thomas, H. R. (1991). "Labor Productivity and Work Sampling: The Bottom Line." Journal of

Construction Engineering and Management, 423-444.

Tommelein, I., and Odeh, A. (1994). "Knowledge-Based Assembly of Simulation Networks

Using Construction Designs, Plans and Methods." Proceedings of the 1994 Winter

Simulation Conference, 1145-1158.

USA Hoist. (2014). "Construction Hoisting ." Retrieved from USA Hoist : www.usahoist.com

Zuppa, D. (2014, 01 07). Construction . Retrieved from National Research Council Canada :

http://www.nrc-cnrc.gc.ca/eng/rd/construction/index.html

101

Appendices

Appendix 1: References on Artificial Intelligence

1. Comprehensive Materials processing by Dobrzanski, Trzaska, Dobrzanska-Danikiewicz

2. Artificial intelligence Foundations of Computational Agents by David Poole and Alan

Mackworth

3. Artificial Intelligence and Expert Systems by Y. Leung

4. Artificial Intelligence; A Modern Approach by Stewart J. Russell and Peter Norvig

102

Appendix 2: Model programmed algorithms

Element: Execute Name: Assign array and random

variables

Scenarios:

All imports System

imports Simphony.General

imports Simphony.Mathematics

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Execute) As System.Boolean

Redim context.Scenario.Ints(600)

Dim a as Comment = context.Scenario.GetElement(of Comment)("1")

Dim c as Comment = context.Scenario.GetElement(of Comment)("3")

Dim e as Comment = context.Scenario.GetElement(of Comment)("5")

Dim f as Comment = context.Scenario.GetElement(of Comment)("6")

dim g as Comment = context.Scenario.GetElement(of comment)("7")

context.Scenario.ints(203)=Cint(a.Text)

Context.Scenario.floats(2)=(cdbl(c.text))

context.Scenario.Ints(213)=Cint(e.Text)

Context.Scenario.floats(1)=cdbl(f.Text)

Context.Scenario.Ints(205) = Cint(g.Text)

Return true

End Function

End Class

End Class

Element: Task Name: Task 1 Scenarios:

1 hoist, Current

Strategy

2 hoists, Current

strategy imports Simphony.Mathematics

imports Simphony.General

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.Modeling.Task(Of Simphony.Simulation.GeneralEntity)) As

System.Double

Return context.Scenario.Floats(2)

End Function

End Class

Element: Branch Name: Workers Selection Scenarios:

All imports Simphony.Mathematics

Public Partial Class Formulas

103

Public Shared Function Formula(ByVal context As

Simphony.General.Branch) As System.Boolean

context.Scenario.Ints(204)= context.Scenario.Ints(204)+1

if context.Scenario.ints(204)<= context.Scenario.ints(203)then

return false

end if

return true

End Function

End Class

Element: Branch Name: Is hoist one available? Scenarios:

All Imports Simphony.General

imports Simphony.Mathematics

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Branch) As System.Boolean

Dim e As Resource = context.Scenario.GetElement(Of Resource)("Hoist1")

If e.Available > 0

Return true

end if

Return false

End function

End Class

Element: Branch Name: Is hoist two Available Scenarios:

2 Hoists, Current

Strategy

2 Hoists, Alternative

Strategy Imports Simphony.General

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Branch) As System.Boolean

Dim e As Resource = context.Scenario.GetElement(Of Resource)("Hoist1")

Dim f As Resource = context.Scenario.GetElement(Of Resource)("Hoist2")

if f.Available> 0

Return true

end if

Return false

End function

End Class

Element: Branch Name: No Hoist Available Scenarios:

2 Hoists, Current

Strategy

2 Hoists, Alternative

Strategy Imports Simphony.General

104

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Branch) As System.Boolean

Dim e As Resource = context.Scenario.GetElement(Of Resource)("Hoist1")

Dim f As Resource = context.Scenario.GetElement(Of Resource)("Hoist2")

if f.Available> 0

Return true

end if

Return false

End function

End Class

Element: Branch Name: Capacity Reached? Scenarios:

All Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Branch) As System.Boolean

if context.CurrentEntity.Floats(2)-context.Engine.TimeNow > 1

then

return false

end if

context.Scenario.Ints(200) = context.Scenario.Ints(200) + 1

if context.Scenario.Ints(200) < (context.Scenario.Ints(205)*0.8)

then

Return true

End if

Return False

Element: Execute Name: Number of Stops Scenarios:

All Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Execute) As System.Boolean

Context.Scenario.Ints(context.CurrentEntity.Ints(0))=Context.Scenario.I

nts(context.CurrentEntity.Ints(0))+1

Return true

End Function

End Class

Element: Execute Name: Last Stop Scenarios:

All Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Execute) As System.Boolean

If context.CurrentEntity.Ints(0)> context.Scenario.Ints(201) then

105

Context.Scenario.Ints(201) = context.CurrentEntity.Ints(0)

End if

Return True

End Function

End Class

Element: Execute Name: Calculation Scenarios:

All Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Execute) As System.Boolean

context.Scenario.Floats(5) = context.Scenario.Floats(5)+

context.CurrentEntity.Floats(1)

if context.CurrentEntity.Floats(1)>

context.Scenario.Floats(3) then

context.Scenario.Floats(3)= context.CurrentEntity.Floats(1)

end if

If context.CurrentEntity.Ints(0)> context.Scenario.Ints(212) then

Context.Scenario.Ints(212) = context.CurrentEntity.Ints(0)

End if

Return true

End Function

End Class

Element: Chart Collect Name: ChartCollect Scenarios:

All

X-Value imports Simphony.General

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.ChartCollect) As System.Double

Dim e as Counter = context.Scenario.GetElement(of

Counter)("numberofworkers")

Return e.Count

End Function

End Class

Y-Value Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.ChartCollect) As System.Double

Return context.CurrentEntity.Floats(1)

End Function

End Class

106

Element: Statistic

Collect

Name: StatisticsCollect Scenarios:

All

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Collect) As System.Double

Return context.CurrentEntity.Floats(1)

End Function

End Class

Element: Branch Name: Branch 2 Scenarios:

All

imports Simphony.General

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Branch) As System.Boolean

Dim a as Counter = context.Scenario.GetElement(of

Counter)("numberofworkers")

If a.count< context.Scenario.Ints(203) then

return true

End if

Return false

End Function

End Class

Element: Execute Name: Average Waiting Time Scenarios:

All

imports Simphony.Mathematics

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Execute) As System.Boolean

Context.Scenario.Floats(6) =

(Context.Scenario.Floats(5)/context.Scenario.Ints(203))

Return true

End Function

End Class

Element: Trace Name: Data Scenarios:

All

Public Partial Class Formulas

107

Public Shared Function Formula(ByVal context As Simphony.General.Trace)

As System.String

Dim numofhoist as Integer

Select case context.Scenario.Name

Case "1 Hoist-CS"

numofhoist = 1

Case "2 hoists-CS"

numofhoist= 2

End select

Return" |Senario| "& context.Scenario.name & " |numofhoists| "

&Cstr(numofhoist) &" |no. of workers-Actual| "&

Cstr(Context.Scenario.Ints(203))& " |Average Arrival interval| "&

Cstr(context.scenario.floats(2)/2)&" |no. of stories| "&

Cstr(Context.Scenario.Ints(213))&" |timeperfloor| "&

Cstr(context.Scenario.floats(1))&" |Capacity| "&

Cstr(context.Scenario.Ints(205))&" |Avg Wating time| "&

Cstr(context.Scenario.floats(6))&" |Max Wating time| "&

Cstr(context.Scenario.floats(3))

End Function

End Class

Element: Execute Name: Assign Hoist Number Scenarios:

All

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Execute) As System.Boolean

context.scenario.ints(211)= context.scenario.ints(211)+ 1

Context.currententity.Ints(2)= context.scenario.ints(211)

Return true

End Function

End Class

Element: Branch Name: Hoist 1 or 2 Scenarios:

All

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Branch) As System.Boolean

If context.CurrentEntity.Ints(2) = 1 then

Return true

end if

108

Return False

End Function

End Class

Element: Execute Name: Number of Trips Scenarios:

All

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Execute) As System.Boolean

Context.Scenario.Ints(208)=context.Scenario.Ints(208)+1

Return true

End Function

End Class

Element: Task Name: Travel 1 floor Scenarios:

All

Imports Simphony.Mathematics

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.Modeling.Task(Of Simphony.Simulation.GeneralEntity)) As

System.Double

Return (Context.Scenario.floats(1)/0.8)

End Function

End Class

Element: Execute Name: Floor Counter Scenarios:

All

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Execute) As System.Boolean

Dim H as Integer

H = context.CurrentEntity.Ints(2)

Select case H

Case 1

context.Scenario.Ints(202) = context.Scenario.Ints(202)+1

Case 2

context.Scenario.Ints(302) = context.Scenario.Ints(302)+1

End Select

Return true

End Function

End Class

Element: Branch Name: More Floors to go? Scenarios:

All

109

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Branch) As System.Boolean

Dim H as Integer

H = context.CurrentEntity.Ints(2)

Select case H

Case 1

If context.Scenario.Ints(201) = Context.Scenario.Ints(202) then

Return false

End if

Case 2

if context.Scenario.Ints(301) = context.Scenario.Ints(302)

then

Return False

End If

End Select

Return true

End Function

End Class

Element: Branch Name: Stop on current floor? Scenarios:

All

Imports Simphony.General

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Branch) As System.Boolean

Dim H as Integer

H = context.CurrentEntity.Ints(2)

Select case H

Case 1

If context.Scenario.Ints(Context.Scenario.Ints(202))> 0

then 'there is a stop

context.Scenario.Ints(Context.Scenario.Ints(202))=0

return true

End if

Case 2

If context.Scenario.Ints(400+Context.Scenario.Ints(302))> 0

then 'there is a stop

context.Scenario.Ints(400+Context.Scenario.Ints(302))=0

return true

110

End if

End Select

'There is no stop

Return false

End Function

End Class

Element: Execute Name: Reset Variables Scenarios:

All

Public Partial Class Formulas

Public Shared Function Formula(ByVal context As

Simphony.General.Execute) As System.Boolean

Dim H as Integer

H = context.CurrentEntity.Ints(2)

Select case H

Case 1

Context.Scenario.Ints(200)=0

Context.Scenario.Ints(201)=0

Context.Scenario.Ints(202)=0

context.Scenario.Ints(215) = 0

Case 2

Context.Scenario.Ints(301)=0

context.Scenario.Ints(300)=0

Context.Scenario.Ints(302)=0

context.Scenario.Ints(216) = 0

End Select

Return true

End Function

End Class