Professional Documents
Culture Documents
A thesis presented to
the faculty of
the Russ College of Engineering and Technology of Ohio University
In partial fulfillment
of the requirements for the degree
Master of Science
Julian Gonzalez
June 2007
by
JULIAN S. GONZALEZ
Janet K. Yates
Professor of Civil Engineering
Dennis Irwin
Dean, Russ College of Engineering and Technology
Abstract
GONZALEZ, JULIAN S., M.S., June 2007, Civil Engineering
USING FUZZY MATHEMATICAL MODELS FOR CONSTRUCTION PROJECT
SCHEDULING WITH TIME, COST, AND MATERIAL RESTRICTIONS (171 pp.)
Director of Thesis: Janet K. Yates
This research investigation evaluated the viability of using fuzzy mathematical
models for determining the duration of construction schedules and for evaluating the
contingencies created by schedule compression and delays due to unforeseen material
shortages. The research also used heuristic material allocation and sensitivity analysis to
test five cases of material constraints. Material constraints increase the cost of
construction and delay the finish of projects. Mathematical models allow the
multiobjective optimization of project schedules, considering constraints such as time and
cost defined in the project, and unexpected materials shortages to determine the fuzzy
aspiration levels of Decision Makers.
Approved:
Janet K. Yates
Professor of Civil Engineering
Acknowledgements
I would like to express my gratitude for the many people involved in supporting me
while I wrote my thesis: to my parents Santiago Gonzalez and Raquel Gonzalez and
my sister Monica Gonzalez who encouraged me even though they are far away;
especially my brother, Carlos A. Gonzalez, who always supported me and whose
dreams live on through my work here. To my host family Tom Walker, Anne Walker,
Emilie Walker and Jimmy Walker whose constant confidence sustained me. To my
professors, Drs. Daniel Castro-Lacouture, Grsel A. Ser, Janet Yates, James
Thompson, Ludwig Figueroa, and Vardges Melkonian who guided me through this
process with their knowledge and clear instruction. Without their support and
collaboration with my close friends, this work would not have been possible.
Table of Contents
Page
Abstract ............................................................................................................................... 3
Acknowledgements............................................................................................................. 4
List of Tables ...................................................................................................................... 7
List of Figures ..................................................................................................................... 8
1
CHAPTER ONE: INTRODUCTION....................................................................... 10
1.1
Research Hypothesis......................................................................................... 10
1.2
Objectives of the Research ............................................................................... 11
1.3
Research Justification ....................................................................................... 12
2
CHAPTER TWO: DEFINITIONS ........................................................................... 14
2.1
Construction Materials Management................................................................ 14
2.1.1
Construction Projects.................................................................................... 15
2.1.2
Activity Networks......................................................................................... 15
2.1.3
Critical Path Method (CPM) Scheduling...................................................... 19
2.2
Mathematical Models ....................................................................................... 20
2.1.2
Linear Programming ..................................................................................... 21
2.3
Mathematical Models in Construction.............................................................. 27
2.4
Fuzzy Mathematical Models (FMM)................................................................ 28
2.4.1
Fuzzy Mathematical Programming Models.................................................. 28
2.4.2
Membership Functions ................................................................................. 28
2.5
Material Restrictions......................................................................................... 33
3
CHAPTER THREE: LITERATURE REVIEW ....................................................... 34
3.1
Construction Network Scheduling Approach and Variations with CPM ......... 35
3.1.1
Modified CPM as a Tool for the Decision Making Process ......................... 36
3.2
Optimization Models in Construction Networks. ............................................. 37
3.3
State-of-the-Art in Construction Scheduling in the US .................................... 39
3.4
The Effects of Applying Pressure to the Final Quality on Network Projects... 40
3.5
Approach of Using Optimization for Construction Management .................... 41
4 CHAPTER FOUR: PROBLEM DEFINITION ........................................................ 42
4.1
Background....................................................................................................... 42
4.2
Material Delivery Uncertainties in Construction Project Networks ................. 45
5
CHAPTER FIVE: METHODOLOGY ..................................................................... 47
6
CHAPTER SIX: RESULTS ..................................................................................... 58
6.1
General Results ................................................................................................. 58
6.2
Time-Cost Trade-off Calculations with Material Constraints .......................... 59
6.3
Allocation with Material Restrictions in Five Different Cases......................... 68
6.4
Sensitivity Analysis .......................................................................................... 84
6.5
Fuzzy Mathematical Model Analysis ............................................................... 98
7 CHAPTER SEVEN: CONCLUSIONS AND RECOMMENDATIONS ............... 108
7.1
Future Research and Recommendations......................................................... 112
REFERENCES: .............................................................................................................. 114
Appendix A.
Abbreviations...................................................................................... 118
Appendix B.
Definitions........................................................................................... 119
6
Appendix C.
Calculations and Graphical Representations ...................................... 122
Appendix D.
Mathematical Models.......................................................................... 128
Optimization Programming Language (OPL) codes .................................................. 128
Minimum Completion Time ....................................................................................... 130
Fuzzy Mathematical Model for the 60 activity project............................................... 133
Mathematical Models for the 20 Activities Case Study ............................................. 137
Minimum Completion Time ....................................................................................... 137
Minimum Crashing Cost for the 20 Activities Case Study ........................................ 142
Cost Sensitivity Analysis for the Case Study ............................................................. 149
Fuzzy Mathematical Models for the 20 Activities Case Study .................................. 152
List of Tables
Page
Table 2.1
Table 2.2
Table 2.3
Table 2.4
Table 2.5
Table 5.1
Table 6.1
Table 6.2
Table 6.3
Table 6.4
Table 6.5
Table 6.6
Table 6.7
Table 6.8
Table 6.9
Table 6.10
Table 6.11
Table 6.12
Table 6.13
Table 6.14
Table 6.15
Table 6.16
Table 6.17
Table 6.18
Table 6.19
Table 6.20
Table 6.21
Table 6.22
Table 6.23
Table 6.24
Table 6.25
Table 6.26
Table 6.27
Table 6.28
Table 6.29
Table C.1
List of Figures
Page
Figure 2.1
Figure 2.2
Figure 2.3
Figure 2.4
Figure 4.1
Figure 4.2
Figure 5.1
Figure 5.2
Figure 5.3
Figure 5.4
Figure 5.5
Figure 6.1
Figure 6.2
Figure 6.3
Figure 6.4
Figure 6.5
Figure 6.6
Figure 6.7
Figure 6.8
Figure 6.9
Figure 6.10
Figure 6.11
Figure 6.12
Figure 6.13
Figure 6.14
Figure 6.15
Figure 6.16
Figure 6.17
Figure 6.18
Figure 6.19
Figure 6.20
Figure 6.21
Figure 6.22
Figure 6.23
Figure 6.24
Figure 6.25
Figure 6.26
Figure 6.27
Figure C.1
Figure C.2
Figure C.3
Figure C.4
Figure C.5
9
Project Network (Final Presentation)...................................................... 123
Project Network (Continued Final Presentation).................................... 124
Manually CPM Calculations (Final Presentation) .................................. 125
Manually CPM Calculations (Continued Final Presentation) ................ 126
Project Network(Original Presentation) ................................................. 127
10
1.1
Research Hypothesis
The research hypothesis investigated during the research project was the
Chapter Organization
This chapter includes the following: (1) the Objectives of the Research, (2) the
11
Research Justification, and (3) the limitations of the study.
1.2
models for construction project scheduling and it included using 1) heuristic methods
along with manual calculations; (2) fuzzy mathematical models; (3) and Primavera
Project Management (P5). Simulations in special environments (different constraints)
were used to identify which techniques are more suitable for a specific arrangement of
activities and what types of complexities should be taken into account.
Research Objectives:
The objectives of the research were the following:
1. To analyze a Construction Network using Different Methods
2. To build a Time and Cost Optimization Model for a Network.
3. To determine the sensitivity of using Material Allocations in Construction
Networks completion time.
4. To build a Model that includes Cost-Time Trade-Offs, with Material Constraints.
12
scheduling software programs use the Activities on Node (AON) format, but the
mathematical modeling languages require that schedules be in the AOA format.
1.3
Research Justification
The main contributions of this research are: (1) the analysis of a different
methodology to determine the length of a construction project and its critical path based
on mathematical modeling; (2) the identification of common problems during the
13
conventional scheduling of construction projects with material constraints; and (3) the
development of a fuzzy mathematical model for analyzing the cost-time trade off in
schedules.
The research is summarized in the following manner: definitions and explanations
used to conduct the research are included in Chapter Two, the literature review is in
Chapter Three, the problem definition and background information are in Chapter Four,
the methodology is in Chapter Five, the results are in Chapter Six, the conclusion and
recommendations are in Chapter Seven, the references and the appendices with the
mathematical models and figures are located after Chapter Seven..
14
2.1
methods that need to be implemented to achieve planned goals. The types of technology,
or methods, used differ from project to project. Practitioners usually choose to use
technologies, or methods, that have worked best for them in the past. Decisions are made
taking into account the advantages and disadvantages that each technique offers to a
specific project. In order to make efficient decisions it is wise to include all of the
participants affected in the decision process. Therefore, final decisions will include
different positions and the experience of everyone involved, which will make the
decisions more realistic. In addition, participants will understand and incorporate the
decision process into their work (Halpin and Woodhead, 1998).
Construction management deals with the optimization of the utilization of
resources. Resources are usually grouped into four categories: manpower, materials,
machines, and money. Real projects always differ from planned projects. This variability
15
could make the cost of a project increase.
Materials are important to the construction process because without them projects
could not be completed at all. In the June 2006 issue of the Building Construction Cost
Data, it states that construction costs are increasing due to rising oil prices. Using
mathematical models that incorporate material restrictions could help determine adequate
cost-time trade-offs in construction schedules (Perdomo-Rivera, 2004).
2.1.1
Construction Projects
Construction projects are built under several constraints including a specific
budget, limited resources, and a specific time frame. Construction projects have definite
deadlines, unique contracts and specifications, and the final product is unique due to
different conditions on every project. Project Managers are the people in charge of
planning, controlling, and directing specific projects.
2.1.2
Activity Networks
Activity networks are graphical representations of the relationships between the
activities that result from a project break down or Work Breakdown Structure (WBS).
This breakdown of complex, or general, activities into more specific ones depends upon
the nature, classification, and location of a project. Using WBS makes it easier to
manage, plan, and schedule a project and it improves quality because more specialized
crews might be perform specific activities.
The way activities are organized in networks follows a logical path, where some
activities depend upon the completion of other activities (logical precedence) before they
16
can be performed. Activities may be presented in Activity-on-Arrow (AOA) networks
which are known as the Arrow Diagramming Method (ADM). In Activity-on-Arrow
networks the activities are represented by arrows and the events are represented by circles
called nodes. The Activity-on-Node (AON) method is also known as the Precedence
Diagramming Method (PDM). In Activity-on-Node networks the activities are
represented in the nodes by boxes; and arrows only show relationships among the
network activities not length of time. The Work Break Structure (WBS) of a project may
be presented in a table and codified to be compatible with project management software
programs such as Primavera Project Planner (P3), as is shown in Table 2.1, which is a
Project Network Example. There are two types of activity relationships represented in
Table 2.1 and they are Start to Start (SS) and Finish to Finish (FF) ( Mubarak, 2005).
17
Table 2.1
A1000
NTP
Original
Dur.
(days)
0
A1010
Mobilization
A1000
A1020
12
A1010
A1030
Excavation
A1020
SS
A1040
Foundation
15
A1030
SS
A1030
FF
Activity
ID
Activity Name
A1050
A1060
A1070
SOG
A1080
1st-Floor Columns
A1090
10
Predecessors
Relationship
Types
Figure 2.1 shows an example of AOA network notation. The arrow heads indicate
the direction of the flow through the network and they do not represent time.
Figure 2.1
18
Figure 2.2 shows a project network using AOA representation where the activity
identifications are listed over the arrows, the durations in days are written under the
arrows, the sequence directions are represented by arrow heads, and activities are joined
by nodes. The arrows that are drawn using dashed lines represent dummy activities which
do not have any duration. Dummies represent logical constraints only.
Figure 2.2
The mathematical modeling software used for the research was the Optimization
Programming Language (OPL studio), which is a commercial optimization computer
software. The minimum completion times for each activity are entered for each activity
and their relationships are shown in the mathematical model. For example, the first node
is (X10) and it has a duration of zero. The difference between the first node and its
successor nodes represents activity times. The entire construction network and its
interrelationships are represented in this form and the difference between the last activity
19
node and the first activity node is the construction project duration that is represented by
the lower case letter z. In OPL, the process would resemble the following:
minimize z;
subject to
{z==X70-X10;
X10==0;
X20-X10>=3;
X30-X20>=4;
X40-X30>=7;
X50-X30>=13;
X50-X40>=0;
X60-X50 >= 2;
X70-X60 >= 3};
Network Scheduling
Scheduling is defined as the delineation or specification of the sequences, times,
and relationships of different activities on a project. Schedules are created based on work
break down structures, the sequential and logical organization of tasks, critical path
analysis, and time estimations. The more accurate the information is, the better the
schedule (Knutson and Bitz, 1991, p. 71).
2.1.3
20
was implemented on government projects in order to determine the best-trade off
between cost and time. Critical path methods are deterministic techniques that do not
consider variations in estimated activity times (Morder et al, 1983, p. 17; Daellenbach
and Flood, 2002, p.70). Currently, CPM schedules are used on large construction projects
and they are required in most public contracts. The information from CPM schedules is
used by Decision Makers (DMs) or people involved in the decision process to plan and
control projects and to train new DMs (Revelle et al, 1993). Critical path method
scheduling techniques are used to analyze all of the possible paths to finish a construction
project considering all interrelationships and it focuses on determining the longest
sequential path in a network.
2.2
Mathematical Models
Models are used to represent how something will function under certain
conditions. Models are simulations of the real world. Researchers have been using
mathematical models since the early 1900s to try to explain non-artificial events in a
physical sense. It is important to keep models as simple as possible for better
understanding (Revelle et al, 1993).
The goals of construction projects are to finish on time, under budget, and with
good quality, which means finishing no later than the time stipulated in the contract.
When an optimal schedule is determined, companies benefit from the following aspects:
receiving monetary compensation; enhancing their reputation; and improving resource
efficiency and availability. For instance, equipment would be available for use on other
projects or a combination of all of these benefits.
21
In this research, mathematical models were used to optimize various scheduling
objectives such as minimizing costs or minimizing completion times. The research
variables were represented by linear functions using integer values and mixed integer in a
feasible domain and by using mathematical techniques such as Linear Programming.
2.1.2
Linear Programming
Linear Programming (LP) techniques have been used for optimization problems
(1)
min .z 2 = x n x0
(2)
In the objective function formulation (1) and (2), z1 could represent the
maximization of profits and z 2 the minimization of completion time.
Subject to (constraints):
k
ij
x j yi
22
xij 0
Equation (3) is defined and expanded later in the constraints section.
Decision Variables
For any problem there are different unknowns, or elements that are key to finding
a solution for a model and Decision Variables (DV) are used to represent these
unknowns. Decision variables should represent the choices to be made by decision
makers, and they are usually defined by lower case letters with or without subscripts for a
better interpretation. For example, when an activity i will be completed and also how
many days it will take to finish a project is represented by:
(4)
Therefore, project finish time is represented by a summation of the times of each activity,
n
x .
i
When formulating a model, the first step is to identify what the problem is and then to
define the DV. All the variables used in this research investigation were considered to be
positive, i.e. x i 0.
Objective function
An Objective Function (OF) represents a Decision Makers (DM) goal as a
function of a combination of the DVs. Objective functions are usually defined by the
letter z and they determine the purpose of the analysis, which normally is the
maximization or minimization of the objective function (Winston, 1994, p.50).
max z = p i x i
i
(i = 1,2,..., n)
(5)
23
Expanding equation (1) results in max .z = p1 x1 + p 2 x 2 +,...,+ p n x n
Where,
Constraints
When optimizing a function the variables, limitations, and boundaries have to be
taken into account; otherwise the optimization would not be realistic. Therefore,
constraints are the restrictions for the decision variables that are used when generating an
optimal solution. Defining constraints is fundamental to solving problems in a more
realistic way. A group of linear inequalities represents the restrictions for a project and
these restrictions are based on the availability of personnel, materials, equipment, money,
time, or any other decision variables defined for a project.
Constraints are represented in the following manner:
k
ij
x j y i (i = 1,2,..., k ; j = 1,2,..., m)
(3)
24
may lead to infeasible (impracticable) solutions.
The following example shows how all these elements interact.
Table 2.2
*Type of Fiber
Reinforced and
Aggregate Materials
(FRAM)
1
2
Cost
$/C.Y.
Availability
(C.Y.)
70
350
20
410
25
Table 2.3
80
Demand
Requirements
(C.Y.)
360
70
400
Type of concrete
Profit
$/C.Y.
Finally, in Table 2.4, it can be seen that both concrete A and B need more than
fifty percent of the respective FRAM.
Table 2.4
Concrete Type
A
70%
60%
The objective is to determine how many Cubic Yards (C.Y.) of FRAM 1 and 2 are
needed in order to produce concrete A and concrete B.
Therefore,
x11 + x12 C.Y. of concrete A
(1)
26
x11 + x21 C.Y. of FRAM1
x21 + x22 C.Y. of FRAM2
max .z = p i x ij
j
(2)
Objective Function
(3)
b x
ij
y j ( i = 1,2; j = A, B)
(4)
27
x11
0.6 0.4 x11 0.6 x12 0
x11 + x12
x22
0.7 0.7 x21 + 0.3 x22 0
x21 + x22
xij 0
Sign restriction
The final solution from the OPL program calculations is the following:
Materials (Optimal combination)
x11 = 230; C.Y. of FRAM 1 to produce concrete A
x12 = 130; C.Y. of FRAM 2 to produce concrete A
x21 = 120; C.Y. of FRAM 1 to produce concrete B
x22 = 280; C.Y. of FRAM 2 to produce concrete B
Concrete Production
x11 + x12 C.Y. of concrete A (360 C.Y.)
x21 + x22 C.Y. of concrete B (400 C.Y.)
2.3
28
such as time or budgets. Networks may be optimized using MP, however, some research
indicates that there are drawbacks to using models due to the time required to get optimal
solutions when the complexity of the project and size increases. In some cases, it is better
to use heuristic approaches where solutions would be feasible (practical). Mathematical
Programming generates optimal solutions in construction networks by analyzing the
activities that are on the critical path, which during the process may change in a realistic
dynamic fashion (Williams, 1999).
2.4
results in a quantitative style. This presentation of the results can be done by using
membership functions which are defined in this research.
2.4.1
formulation of Linear Programming (LP) models. In the case of FLPM, the DM has the
opportunity to define a set of possible values and create a curve of satisfaction level or
membership function. For instance, using LP one objective at a time can be optimized
such as completion time or cost. By using goal programming two objectives may be
addressed giving priority to only one of the two objectives. Using FLPM, two objectives
may be addressed at the same time. In addition, the use of membership functions gives a
more realistic approach.
29
Membership Functions (MFs) are curves that define how one value (decision)
belongs to a set of values previously determined or to an interval of possible answers.
Frequently, MFs are linear or trapezoidal functions because the correlation between the
objectives is much more straightforward although not more precise. Therefore, DMs may
make use of the MF to evaluate their decisions and figure out how much a decision
satisfies or belongs to a previously defined interval of possible values.
Making a decision is a process that involves vagueness, ambiguity, or fuzziness.
In the construction industry, Decision Makers have to deal with project changes,
unavailable information when it is needed, the work experience of project managers, and
project managers subjective reasoning. However, defining a successful decision is not
always possible because a decision may be successful in a specific activity or project and
at the same time a similar decision may be completely wrong in similar or identical
projects. Moreover, sometimes several decisions have to be compared and/or associated
and they are presented in different terms. For instance, when one wants to minimize the
completion time of a project, more resources and equipment have to be involved in the
process meaning an increase in costs. In this case, time and cost are compared. For these
issues, defining a set of possible outcomes and a respective grade of satisfaction level for
each one is a useful tool for DM in decision processes.
In other words, a membership function is a curve representing the possible grades
of association that are between 1 (which represents a maximum association), and 0
(which represents non association). These grades of association give a correlation
between the DMs outcomes (answers) and the stated satisfaction level.
30
31
extra funds or resources. The optimal values for these objectives may be found using
mathematical models.
Table 2.5
Objectives
Bounds
z1 ($)
z2 (days)
Maximum
390
38
Minimum
30
Membership Functions are drawn based on the information in Table 2.5, which
includes the interval for the project objectives, and an agreement between the DM and the
personnel involved in a decision process. Having a flexible, instead of a crispy (unique),
interval of possible solutions allows the use of weighted language in these models. For
instance, as shown in Table 2.5, if an objective is to minimize the cost to finish a project
and the fixed budget is $390, then an agreement between the DM and the owner will
define the boundaries of the MF (see Figure 2.3). In addition, the membership function is
generated using graphical correlations between the interval defined by the minimum and
maximum bounds and the graded approval of the respective outcomes. Therefore, any
cost within this interval can be evaluated using a percentage grade of satisfaction
(weighted language).
In Fuzzy Mathematical Models (FMMs), the objective functions ( zi ) could have a
respective grade of membership curve. These membership curves correlate the possible
32
objective function outcomes with the DMs aspiration level, as shown in Figure 2.3. This
figure shows that for the minimum crashing cost ( z1 ), spending 1/3 of the maximum
possible cost, the DM will get a high satisfaction level. In Figure 2.4, the minimum
completion time shows a coincidence in the high satisfaction level found by completing
the project in 37 % more time than the ideal completion time. Any project duration less
than 75% of the total duration is ideal for the DM aspiration level, and proportionally
diminishes up to the due date for the project where the aspiration level will not be
acceptable.
u1
1
67%
z1
0
130
Figure 2.3
390
($)
33
u2
1
67%
z2
0
30
Figure 2.4
2.5
33
38
390
(days)
Material Restrictions
The delay, or unpredicted shortage of any material, especially in the critical path,
negatively affects the whole project. Therefore, having a tool that would let the DM
consider alternative paths is extremely useful during decision processes.
34
35
3.1
CPM
Critical Path Method Schedules are an important technique that has been used
since the 1950s, and the construction industry benefits from their use in the planning and
controlling of projects, when communicating plans, and when training new managers.
Newer versions of CPM scheduling software make CPM techniques easier for
practitioners to use them and this has increased efficiency in some construction projects.
Schedules that neglect material constraints could mislead planners and affect the
control of projects (Yates, 1993). Real life projects present a wide range of variables that
are difficult to control due to the fact that resources are limited in construction projects;
therefore, float calculated using CPM techniques will lose its significance and new
critical sequences will be created (Kim and de la Garza, 2005; Wiest, 1964). Critical Path
Method schedules show the critical path, or paths, of a project but resource profiles
present fluctuations that are not desirable for the efficiency of projects.
Project managers, or decision makers, now use commercial project management
software, such as Primavera Project Planner (P3) and Microsoft Project, which are based
on heuristic methods to plan and control schedules. In the article Project-Network
Analysis Using Fuzzy Sets Theory, the authors discuss the use of different applications
in project planning and control (Liberatore, Pollack-Johnson, and Smith, 2001). They
found that the construction industry mainly uses critical path analysis applications and
that Primavera Project Planner is used most frequently. The research showed that two out
36
of four companies use Primavera and one out of four uses Microsoft Project for
scheduling. The authors had a sample of forty-two construction companies that responded
to their survey out of two hundred forty companies. Uncertainties and methods to forecast
activity durations were two of several issues that practitioners are inquiring about
nowadays. Labor productivity improved by 6% when resources were considered in CPM
and an additional 4-6% improvement was obtained when using computerized systems as
stated in the PhD. dissertation, A Framework for a Decision Support Model for Supply
Chain Management in the Construction Industry, (Perdomo-Rivera, 2004).
37
processed is stored creating historical databases that help support future DM (Yates,
1993). This research gives practitioners an opportunity to consider historical information
and have a more realistic approach to decision-making processes.
A different approach has been followed by other authors, who stressed that
considering project constraints is not enough. The constraints have to be analyzed and
prioritized depending on their repercussions on the whole project. In their article Key
Constraints Analysis with Integrated Production Scheduler (IPS), Chua and Shen (2005)
categorize and analyze the most critical constraints and the causes of delays in
construction projects. Moreover, the examination presents a methodology that modifies
traditional Critical Path Method (CPM) scheduling. They incorporate neglected
constraints such as availability of resources and information in the CPM analysis (i.e.
paperwork documentation and design approval or changes). The methodology is
implemented by using the Integrated Production Scheduler (IPS), which is a scheduling
tool. This methodology provides an opportunity to study neglected information and offers
more realistic output.
3.2
been successful for large networks. Networks with more than 100 activities could not be
handled due to computer hardware limitations. Critical path method techniques, with
discrete information instead of continuous membership functions have proven to be more
efficient and they provide not optimal, but useable solutions. (Molder J., 1993). However,
some optimization techniques present the opportunity to analyze more than one objective
38
at a time and this permits a more realistic approach. On the other hand, other authors
analyze the uncertainties in a project by using fuzzy sets.
The principal objectives in a network project are to minimize completion time and
cost. During the literature review only a few articles that concentrated on construction
networks using goal programming were located. In the article An Application of Fuzzy
Goal Programming to a Multiobjective Project Network Problem an analysis of the
optimization of a project network facing two constraints simultaneously is presented.
Unlike other optimizing methods, the optimization is made using nontraditional
mathematical modeling, called Fuzzy Goal Programming (FGP). Therefore, having two
constraints and the fact that the two objectives are presented in different terms such as
time and cost, make the optimization project more difficult. Subjective judgments made
by the Decision Maker (DM), and the fuzziness of the objectives, are the basis of the
Fuzzy Set Theory (FST). Fuzzy Goal Programming is used to achieve an optimal solution
while considering two objectives at the same time and using membership functions
(Arikan and Gungor, 2001).
In their article Project-Network Analysis Using Fuzzy Set Theory, Lorterapong
and Moshelhi (1996) present a practical network scheduling method for construction
projects based on Fuzzy Set Theory (FST), which allows the consideration of
uncertainties coming from diverse project settings. The authors emphasize that this
method is more pragmatic than using stochastic or probabilistic techniques and does not
show sophisticated computational calculations. However, the authors state that when a
project consists of large networks a common spreadsheet package will facilitate the
39
process. Consequently, taking into consideration uncertainties to decide possible
completion times for each activity in a network makes this contribution a useful
optimization tool for decision makers and researchers.
The fuzzy set theory method is an important contribution to decision processes
because it allows consideration of uncertainties coming from a diverse project setting,
which is valuable because every project presents a large percentage of uncertainties. If
these uncertainties are known in advance, decisions could be made with more precision.
Delays are always present during construction projects; therefore, trying to avoid dealing
with potential uncertainties in the planning stage of a project could lead to problems
during the construction stage. Even if the delays do not occur, it would be better to be
safe. In the article Fuzzy Logic Approach for Activity Delay Analysis and Schedule
Updating, the authors, Ordoez-Oliveros and Fayek, formulate a new tool that gives
DMs the opportunity to have an updated schedule and to evaluate the consequences of
delays in order to make adequate decisions when they are required during a project
(Ordoez-Oliveros and Fayek, 2005).
3.3
Microsoft Project are the most frequently used commercial software in the construction
industry in the United States, users or practitioners have to analyze and interpret the
information generated by the software, which are merely tools for displaying schedules
(Kelleher, 2004). A critical analysis needs to be done when using any kind of commercial
software because the heuristic scheduling procedure, or hypothesis followed, in these
40
commercial models is neither clear, nor available for practitioners in many cases. In the
article Evaluation of the Resource-Constrained Critical Path Method Algorithms, the
authors, Kim and de la Garza, present a case where some activity relationships and lags
related to nonworking days are inconsistent (Kim and de la Garza, 2005).
Other research has been conducted that considers more than Critical Path Method
(CPM) scheduling aspects including: a Progress Monitoring System with GIS (PMSGIS), which is an investigation that focuses on the use of Computer-Aided Drafting
(CAD) programs (AutoCAD), Primavera Project Management software (P3) and
Geographical Information Systems (GIS) program (ArcViewGIS). They generate 3D
drawings and show synchronized P3 schedules that permit a faster and better
conceptualization of projects. They may be useful in scheduling, planning, controlling,
and decision making processes. ArcViewGIS knowledge is needed to operate, edit, and
update models (Arditi and Poku, 2006).
3.4
Projects
Most schedules need to strive for optimization and earlier completion times,
which sometimes results in quality being neglected. Excessive pressure to finish a project
may be disadvantageous; it may generate poor quality results and rework and this could
lead to an increase in the project duration. The responsibility is on managers who should
understand what activities are critical and know how much pressure to apply to increase
worker efficiency (Nepal et al., 2006). In addition, some actions such as motivation and
training might be taken to counteract the effects of time pressure on quality. Increasing
41
the number of hours worked per day, or the number of days per week, does not always
mean increased productivity. A study by Proctor and Gamble that was published as a
Business Roundtable (BRT) report (BRT 1974, and BRT 1980) analyzed the reduction of
productivity when working overtime 50-60 hours per week as supposed to 40 hours per
week. Productivity at construction sites decreased during the first week by about 15% as
a result of requiring overtime, and then during the next two weeks productivity returned
to about 95% out of the original 100% productivity, after the first three weeks
productivity decreased again at a constant rate for the next six weeks to 65% out of
100%, and finally after the ninth week the productivity variations stabilized at 65%
productivity (Randolph, 1990, p.7-8)
3.5
42
Background
Construction projects are made up of many different activities. These activities
are related to each other in networked systems and grouped into special categories that
require different levels of expertise and resources for their completion. The main
objective of a Project Manager (PM) is to determine the most efficient way to coordinate
construction project activities to achieve the best possible quality and to minimize time
and costs. Even though some construction companies use commercial resource allocation
software, they might have material shortages in real-life situations, and that is where
having a mathematical model with material constraints is useful to PMs.
The building construction industry has been experiencing a loss of benefits due to
several problems such as: materials not available where and when they are needed, a lack
of information about where materials are located, fragmented communication between
different project stages, and the lack of experience of the Decision Makers (DM). These
problems may cause increasing expenses and necessitate extra time to finish a project;
furthermore, established schedules and budget goals may not be met. However, having a
way of tracking and controlling real resource information, helps DMs make better
decisions.
Using Fuzzy Goal Programming (FGP), which is a mathematical tool for solving
optimization problems, Project Managers (PM) try to achieve the two main project
objectives which are to minimize project cost and time, and consider the ambiguities in
the decision process derived from the PM (Arikan and Gungor, 2001). Fuzzy set theories
43
and objectives may be converted into constraints in a fuzzy mathematical model; and the
optimal solutions generated address both objectives at the same time. In Fuzzy Set
Theory (FST), membership functions are used for presenting both objectives in similar
terms. As can be seen in Figure 4.1, objectives have a grade of association to the
membership function or gradual transition from 0 (non satisfaction level), to 1 (maximum
satisfaction level), which allows new constraints to be added to models.
Degree of Membership
Degree of Membership
1.0
1.0
(a) 0
t1
t2
t3
t(time)
(b) 0
t4
t5
t6
t7
t(time)
Degree of Membership
Degree of Membership
1.0
1.0
(c) 0
t8
t9
t(time)
(d) 0
t8
t9
t(time)
44
traditional mathematical model there is a non-grade of membership and the possible
answer either belongs to, or not to, a certain satisfaction level. There is not the possibility
of analyzing subjective decisions or including fuzziness into the model. These are called
crisp states (times and intervals). As can be seen in Figure 4.2 an answer will belong, or
not belong, in a defined interval or specific point. (Lorterapong and Moselhi, 1996).
Degree of Membership
Degree of Membership
1.0
1.0
(a) 0
t (time)
t1
(b) 0
t2
t3
t (time)
Degree of Membership
Degree of Membership
1.0
1.0
(c) 0
t4
t (time)
(d) 0
t4
Figure 4.2
Crisp Times
(a) Exactly t1 ; (b) between t2 and t3 ; (c) after t4 ; (d) before t4 .
Moselhi, 1996).
t (time)
(Lorterapong and
45
4.2
follows a unique procedure to achieve project goals. Even though the same, or a similar,
project is going to be built again in a different place the planning, controlling, and
execution of each project is different. Therefore, having similar characteristics (designs,
project duration, and equipment), standard conditions, and resources for two projects
does not imply the same procedure (techniques) and results for the projects because
technical conditions, delays, and unexpected situations arise during construction. Here is
where the experience of the Decision Maker and his/her subjective judgment helps
achieve the goals or objectives as effectively as possible by replanning and rescheduling
projects when it is necessary. However, even though Project Managers (PMs) may have
experience in construction projects, imprecision and uncertainty in their decision making
practices exist in the scheduling of real-world construction processes. In other words, the
efficiency of a construction project depends on many variables, conditions, and
uncertainties that need to be forecast by the best approximation to a realistic construction
network. Nevertheless, the problem is to find the most favorable schedule that minimizes
project completion time and costs considering material constraints. A decision makers
goal is to make decisions that will result in the best utilization or allocation of resources,
the highest possible profit for the company, and using the minimum possible time.
On the other hand, traditional techniques used in the construction industry for
minimizing project completion time and cost, such as Critical Path Method (CPM)
scheduling and crashing procedures, assume that resources are always available. Linear
46
Programming Methods (LPMs) have been used to minimize these objectives, but one at a
time. According to Deporter and Ellis (1990) in his article Optimization of project
networks with goal programming and fuzzy linear programming Goal Programming
(GP), which is a more powerful mathematical optimization tool, can be used to process
multiple objectives. Furthermore, it also allows the development of fuzzy linear
programming models and membership functions.
47
was written and presented in a civil engineering seminar class. Next, a research plan was
developed and implemented in order to address the research topic. Definitions were
researched for inclusion in the thesis to facilitate an understanding of the research topics.
In addition, the article An Application of Fuzzy Goal Programming to a Multiobjective
Project Network Problem by Arikan and Gungor (2001) that uses optimization models
in a construction network was used as the basis for conducting further research on the
topic of fuzzy programming in construction.
Phase Two: The second phase of the research included a literature review of previous
48
The Journal of the Operations Research Society of America (JSTOR-INFORMS).
The Ohio-link web site was used to locate a wide range of books on the subject of
operations research, construction management, fuzzy sets, and other mathematical
related books
Phase Three: The third phase consisted of extracting data relevant to the research
topic and analyzing and learning the software programs used for the research. In order to
gain an understanding of the research approach two courses were taken: (1) ISE 689
Scheduling and Sequencing, where the mathematical model language and the
optimization software to be used in the research were discussed; (2) ISE 541 Introduction
to Operations Research, where the basic principles of optimization and the advantages of
optimization software were taught.
Phase Four: In this phase the literature review was written and the research
research project as a case study to demonstrate the applicability of using OPL computer
models to verify network schedules.
In order to select a case study project, different projects were analyzed based on
their type, size, complexity, and origin of the information. Several construction networks
were reviewed in order to find an appropriate network that could be used to validate the
capabilities of the mathematical model. The network needed to take into account the
availability of cost information. A construction network with real normal costs and
49
crashing cost information was required for the research. Although communication with
some local companies was established, finding a real network with cost information was
impossible due to company policies about not releasing cost information. Therefore, a
construction network with normal and crash cost from the book Construction Project
Scheduling and Control was selected. The network selected for the case study originally
consisted of a list of sixty activities with their respective identification numbers (ID),
immediate predecessor activities (IPA), cost information, and some general information,
and it was customized to suit the requirements of the research project (Mubarak, 2005).
Phase Six: In this phase the computer software programs to be used for the
research were selected and they included Primavera Project Management (P5), which is a
commercial project management software program and the commercial mathematical
model software known as Optimization Program Language (OPL). Information from the
case study construction project was input into Primavera. Although commercial
scheduling software such as Primavera Project Management may be used to analyze and
schedule project network information, the logic diagram generated is in an Activity on
Node (AON) format as is shown in Figure 5.1 (Primavera representation), but AON
graphical representations cannot be used in mathematical modeling. In order to use the
mathematical language, the logic for the diagram has to be drawn as an Activity on
Arrow (AOA) network by hand (see Appendix B). The representation of immediately
preceding activities in the network system creates a large amount of crossing dummy
lines.
50
Figure 5.1
Phase Seven: In this phase the CPM schedule was drawn and calculations were
performed in order to determine the critical paths of the project using P5. First some
activities were added or modified and the logic of these activities and their logical
dependencies were checked. Some relationships were simplified by eliminating lags that
were in conflict in the network in this phase in order to eliminate some dummies that
were considered unnecessary. Then, the Activity on Arrow (AOA) logic diagram was
drawn.
In Critical Path Method scheduling techniques there is a forward pass and a
backward pass, and both of these were used to find the Early Start (ES), the Early Finish
(EF), the Late Start (LS), the Late Finish (LF), the Free Float (FF), and the Total Float
(TF). The first CPM schedule was done using standard conditions which means taking
into consideration the originally estimated activity time.
Phase Eight: In this phase a mathematical model for the optimal completion time
was written based on the AOA logic diagram for the network. Running this model with
the optimization software OPL generated the same solution as the CPM scheduling
technique solution proving that the network was well modeled. The technological
51
constraints at this phase were time and precedence relationships.
Phase Nine: In this phase, Primavera Project Management (P5) was updated to
take into account some activities that were represented in a different way. The change
was done because the logic diagram representation was different than Activity on Node
(AON). This difference is that Primavera uses AON logic diagrams and a combination of
lags such as Start to Start (SS) and Finish to Finish (FF), while in AOA logic diagrams to
represent these combinations, one of the activities has to be split into two as shown in
Figure 5.2, ( Activity 1040). In AON networks for SS relationships successor activities
are dependent on the first portion of predecessor activities. As shown in Table 5.1 the
same activity can have SS and FF relationships at the same time.
Table 5.1
Activity
ID
Activity
Name
Original
Dur. (days)
A1030
A1040
Excavation
Foundation
7
15
A1050
Fill,
Compact,
and Treat
Soil
Predecessors
Relationship
A1020
A1030
A1030
Types
SS
SS
FF
A1040
FS
LAG
Dur.
(days)
The three activities listed in Table 5.1 have two types of representations: (1) the Activity
on Arrow (AOA) representation as shown in Figure 5.2, and (2) the Activity on Node
(AON) representation as shown in Figure 5.3.
In Figure 5.2 the Activity on Arrow (AOA) example is known as the Arrow
Diagramming Method (ADM). Figure 5.2 (a) illustrates the typical AOA configuration
52
for network scheduling calculations. Figure 5.2 (b) illustrates the calculations for the
three activities listed in Table 5.1. In this representation activity A1040 had to be split
into activities A1040 (a) and A1040 (b) in order to show that activity A1030 is a
predecessor of A1040 with a FF relationship type and lag of two days.
FF(ES,EF)TF
ACT. ID
DUR.
(LS,LF)
(a)
0(0,13)0
A1040(a)
Foundation
13
(0,13)
30
6(0,7)6
A1030
Excavation
7
(6,13)
40
0(13,15)0
A1040(b)
Placing
Concrete
2
(13,15)
0(15,18)0
A1050
Fill,
80
90
Comp.
and
treat soil
3
(15,18)
10
(b)
Figure 5.2
Activity on Arrow (AOA)
(a) Activity on Arrow configuration; (b) AOA network representation
Figure 5.3 shows an Activity Precedence Diagram (APD). Figure 5.3 (a) displays a
typical AON configuration of that shows the types of relationships that are used in
network scheduling calculations. Figure 5.3 (b) illustrates the calculations for the three
activities listed in Table 5.1. Activity on Node schedules are the most common
scheduling techniques being used now and most of the commercial software generates
53
this type of graphical representation. As shown in Figure 5.3 in AON there is no need to
split activity A1040 in order to represent that its predecessor activity A1030 has a FF
relationship with 2 days of lag.
SS LAG
ES,EF
ACT. i ID
DUR.
ES,EF
ACT. j ID
DUR.
FS LAG
LS,LF
LS,LF
FF LAG
SF LAG
(a)
0,15
0,7
A1030
7
A1040
15
6,13
0,15
15,18
A1050
3
15,18
(b)
Figure 5.3
Activity on Node (AON)
(a) AON configuration; (b) AON network representation
Phase Ten: In this phase membership functions for time as an objective were
created based on the information from the completion time generated by the methods
54
used in Phases Eight and Nine. The normal completion time for the case study project
was 142 days, but the objective was to reduce extra costs involved such as indirect costs;
therefore, a decision was made to find the optimal amount of time that the project could
be crashed. Assuming that the project needed to be crashed by 32 days, the following
membership functions were generated. At this point, the objective was to finish the
project 32 days ahead of the schedule, which is a satisfaction level of 1, and this is the
maximum satisfaction level. On the other hand, if the project is not crashed at all, the
satisfaction level will be 0, which means that the satisfaction level for this objective is not
satisfied or fulfilled; any value between these ranges has a weighted value of satisfaction
as shown in Figure 5.4.
if z A 110,
1
uA ( % )
if x = 1 1 4 ( d a y s ) ,
y = 8 7 .5 % ( S a tis fa c tio n L e v e l)
y %
zA
0
Figure 5.4
110
142
(d a ys)
55
Phase Eleven: In this phase a cost analysis was calculated using the mathematical
model. Primavera does not perform crashing network procedures; therefore a
mathematical model was used. Heuristic methods can be used for crashing schedules but
they are cumbersome and time consuming. An analysis was performed in order to find
the minimum possible cost for the construction network by using the mathematical
model. After the model was run, the cost-time trade-off curve was plotted by running the
model with the crashing information and reducing the maximum number of days the
project could be reduced one by one. This information is presented in the results section
on Table 6.5 and Figure 6.5.
Phase Twelve: In this phase the membership function for cost as an objective was
generated by using the information from Phase Eleven (cost analysis), and the Decision
Maker criteria. This membership function is shown in Figure 5.5 and it is one of the
parameters of the fuzzy mathematical model.
if z B = 0,
1
56
uB ( % )
1
y%
zB
0
30600
(cost)
Figure 5.5
Completion Time (Membership Function)
Phase Thirteen: In this phase the analysis focused on combining all of the results
from the previous phases (minimum completion time, minimum cost, and the
membership functions) to create a fuzzy mathematical model that generated optimal
solutions and that considered time and cost. In the fuzzy mathematical model the
previous objectives (time and cost) were considered as constraints. The model optimized
each objective individually by using membership functions and maximized the
satisfaction level between them.
Phase Fourteen: In this phase trade-off calculations were performed using
different variations of material allocations and constraints and the OPL computer
software to determine the optimal duration of a sample project.
Phase Fifteen: In this phase material restrictions were added to the fuzzy
mathematical model and several cases were tested to validate how the project network
57
was affected each time.
Phase Sixteen: In this phase the results of the research were summarized and
58
General Results
When using manual CPM calculations, and only considering time and precedence
relationships as constraints, a total of 142 days was obtained for the project completion
time. The parameters and results for the manual procedure are shown in Figure C.3 and
Figure C.4.
The original presentation of the project network was full of dummy activities due to
the way relationship were presented in the original network (see Figure C.5).
The mathematical model was run a second time and the same number of days were
generated although the free float was used in a different manner. The individual
completion time for many activities was the earliest possible. Finally, Primavera Project
Management (P5) was used to generate the schedule and some parameters were revised
in order to obtain the same conditions. The calendar in Primavera P5 was set up with
every day as a working day, 8 hours per day. The starting date for the project was set as
February 8, 2006 and the completion time was December 21, 2006, which was 142
working days. The schedule completion times are shown in Table 6.1. The completion
time was the same number of days as was calculated manually and using the
mathematical model.
59
Table 6.1
Completion Time
(Working days)
Working Days/Month
August
September
October
November
December
Total
6.2
30
30
31
30
21
142
network in order to facilitate the process. The analysis was accomplished by creating a
20-activity sample model with its respective technological constraints. Precedence
relationships, durations, and maximum crashing time for the model are listed in Table 6.2
and shown in Figure 6.1and the cost information used in the mathematical models is
presented in Figure 6.2.
60
Table 6.2
Activities
Nodes
Predec.
Mobiliz.
A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
P
Q
R
Punch list
10 - 20
30 - 20
40 - 20
50 - 30
60 - 30
70 - 40
80 - 50
90 - 60
100 - 60
110 - 70
120 - 80
140 - 90
130 - 100
110 - 100
140 - 120
150 - 130
150 - 110
160 - 140
160 - 150
170 - 160
Mob.
Mob.
A
A
B
C
D
D
E
F
G
H
H
J
L
M-I
K-N
O-P
Q-R
Duration (days)
Normal Max.Crash
1
0
6
2
3
1
5
3
4
2
8
3
7
0
2
0
9
3
4
2
4
2
3
1
5
0
3
1
2
0
8
3
7
2
5
3
9
3
1
0
61
Figure 6.1
62
Figure 6.2
63
The sample model was drawn using Microsoft Visio, a diagramming software,
and it was analyzed using five different random allocation possibilities. The material x1
was assumed to be needed in every activity and the required amounts for each activity are
presented in Table 6.3.
Table 6.3
Demand of Materials x1
Material
Activities
Nodes
Mobiliz.
A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
P
Q
R
Punch
list
10 - 20
30 - 20
40 - 20
50 - 30
60 - 30
70 - 40
80 - 50
90 - 60
100 - 60
110 - 70
120 - 80
140 - 90
130 - 100
110 - 100
140 - 120
150 - 130
150 - 110
160 - 140
160 - 150
0
3
4
3
4
2
5
4
4
3
4
5
3
2
6
5
3
6
2
170 - 160
x1
The objective of the analysis was to verify the results when considering both the
critical path and when not considering the critical path in the allocation of material x1 .
Mathematical models for the minimum completion time and minimum crashing cost were
generated and modeled on the computer using OPL. Figure 6.3 shows a flowchart based
64
on the heuristic procedure followed in the initial planning stage. The methodology,
(strategies) used for material allocation for the planned and five alternative case studies
are presented in this section. Considering the early start allocation results generated by
the planned case, the amount of material required by day 20 was 29 units of material x1 ,
as shown in Table 6.4. In the five cases analyzed, only 20 units of material x1 were
available and the rest was going to be delivered after day 20, which means that by day 21
material x1 would be available.
Table 6.4
Nodes
30 - 20
40 - 20
50 - 30
60 - 30
70 - 40
80 - 50
90 - 60
100 - 60
units
Units of
Materials
x1
3
4
3
4
2
5
4
4
29
The analysis presented had an ideal case where there was no limitation in terms of
materials. Five cases were analyzed that had material restrictions (see Figure 6.3)
65
Figure 6.3
Analysis Flowchart
Planned: The planned network was drawn using Microsoft Visio and the CPM
calculations were made manually, and the results are shown in Figure 6.4. A
mathematical model, with a minimum completion time as the objective, was generated
and it obtained 43 days for the completion time. In this planned situation, materials were
unlimited; therefore, the critical constraint was time.
66
Figure 6.4
67
The results generated with the mathematical model and the CPM calculations,
using a preset material allocation are shown in Figure 6.4. A time-cost trade-off curve
was drawn based on the information previously assumed, such as the maximum crashing
time and cost for each activity. The time-cost trade-off curve was generated by using a
crashing cost mathematical model. The curve provided essential information, such as the
maximum number of days that the project can be crashed and its crashing cost. The
calculations generated using the mathematical model for the original case (planned) with
a duration of 43 days are presented in Table 6.5 and Figure 6.5.
Table 6.5
Cost
0
1000
2000
3000
4000
5000
6000
7000
8000
11000
14000
17000
25000
33800
68
Cost-time Tradeoffs
Case Planned
35000
30000
Cost($)
25000
20000
15000
10000
5000
0
30
32
34
36
38
40
42
44
Time(days)
Figure 6.5
6.3
optimal solution generated by the mathematical model, and an analysis of the float. This
was done by primarily considering the critical path or the activities with the least amount
of float. As can be seen in Figure 6.6, activities were moved within the available float,
such as activity F, represented by nodes X80 and X50, which had a float of 13 days. This
float allowed activity F to start 8 days later than planned, which did not delay the total
completion time of the project. Activity G, represented by nodes X90 and X60, had 22
days of float.
69
Figure 6.6
70
However, in this case, a new node had to be inserted into the network (X65) and
linked to the existing node X60 using a dummy, as can be seen on Figure 6.7, in order to
maintain the correct logical precedence. The event (X65-X60) does not require time to be
performed, the dummy event was added because activity G and H are successors of
activity D, and activity G is the only one that needed to be delayed. The new nodes for
activity G are X90 and X65.
Figure 6.7
The material allocation for this case required 20 units by day 20 (see Table 6.6), which
was the amount available. When the mathematical model was run for this case, the
completion time of 43 days was not affected.
When considering the critical path, and the activities with the least amount of
float, the completion time did not increase; therefore, the cost for completing the project
on time was the original cost. In addition, the total number of days the project could be
crashed was 6 with a cost of $16,400. Case One is viable alternative.
71
Table 6.6
Activities
Nodes
Mob
A
B
C
D
E
H
20-10
30 - 20
40 - 20
50 - 30
60 - 30
70 - 40
100 - 60
Units of
Material
x1
0
3
4
3
4
2
4
Accumulated
Units Allocated
0
3
7
10
14
16
20
Case Two: In case two, the allocation focused on the critical path and the path
that had the least float as is shown in Figure 6.8. Activity C, represented by nodes X50
and X30, had a float time of 13 days, and it had to be started exactly at its latest start
time, which was after day 20.
72
Figure 6.8
73
In addition, as can be seen in Figure 6.9, activity C is not the only successor of activity
A, so a new node (X35) had to be inserted in the network using a dummy relationship.
Node X35 was linked to node X30 with the dummy, which meant that this event does not
have any duration. A similar procedure was followed for activity G. As presented in Case
One, activity G had to start after day 20 (X65>=20), as is shown in Figure 6.7.
Figure 6.9
The new nodes for activity G are X90 and X65.The completion time for case II increased
by one day, so that the completion time was now 44 days and the cost to finish on time
(43 days) was $ 1,400. The amount of units of material allocated was 20, as is shown in
Table 6.7.
The critical path was considered in this case, but activity C, which had zero total
float, was not. Activity C, starting after day 20, created a new critical path and a new
completion time of 44 days. The cost increased 4.6% up to $1,400.
74
Table 6.7
Activities
Nodes
Mob
A
B
C
D
E
H
I
20-10
30 - 20
40 - 20
50 - 30
60 - 30
70 - 40
100 - 60
110 - 70
Units of
Material x1
Accumulated
Units Allocated
0
3
7
7
11
13
17
20
3
4
0
4
2
4
3
The total number of days the project could be crashed was 6 with a cost of $31,460.
Case Three: In case three, although the allocation was focused on the critical
path, the next path chosen was the one that had higher activity float time. As can be seen
in Figure 6.10, activity A and B are successors of Mobilization. Activity A was allocated,
but activity G was delayed. In order to delay only activity B a node X25 was created.
Node X25 was linked to the existing node X20.The total amount of material x1 required
per activity was tabulated and the cumulative amount allocated is shown Table 6.8.
Table 6.8
Activities
Nodes
Mob
A
B
C
D
E
F
G
H
I
20-10
30-20
40-20
50-30
60-30
70-40
80-50
90-60
100-60
110-70
Units of
Material
x1
3
0
3
4
0
5
0
4
0
Accumulated
Units Allocated
0
3
3
6
10
10
15
15
19
19
75
Figure 6.10
76
Then, activity B was represented by X40 and X25, in order to delay the start time of B,
X25 was started after day 20 (X25>=20), as shown in Figure 6.11. Activity G,
represented by X65 and X90, followed the same pattern (X65>=20). After that activity J,
represented by node X80 and X120, was also started after day 20.
Figure 6.11
When the critical path was considered, and the activities with higher activity float
were the next activities to be considered, the completion time increased by 20 % from 43
to 53 days, and the cost to finish on time without crashing (day 43) increased by
approximately 99% up to $29,800 (Assuming that $ 30,000 was the maximum amount
approved for the project). The total number of days the project could be crashed after the
normal completion time (43 days) was 1 day. The cost for completing the project on day
42 was $34,800.This happened because attention was not on the most critical activities.
Case Four: In case four, the allocation strategy was to finish activities without
considering critical floats as is shown in Figure 6.12. Activity H and G were started later
than planned. Activity G had less impact because of its large amount of float. However,
activity H, represented by nodes X100 and X60, was a critical activity; therefore,
77
delaying this activity caused the whole project to be delayed by 9 days. Activity E,
represented by nodes X70 and X40, had a float of 10 days and an early start of day 4. It
was started after day 20, which means that it was delayed by 4 days; but because activity
H had a longer delay time, 9 days, this was more critical. The amount of materials
x1 allocated in this case was 19 units.
Table 6.9
Activities
Mob
A
B
C
D
E
F
G
H
I
Nodes
20-10
30 - 20
40 - 20
50 - 30
60 - 30
70 - 40
80 - 50
90 - 60
100 - 60
110 - 70
x1
0
3
0
3
4
0
5
0
4
0
Accumulated
Units
Allocated
0
3
3
6
10
10
15
15
19
19
78
Figure 6.12
79
The completion time in this case was 52 days and the cost to finish on time was $ 21,800.
The total number of days the project could be crashed after the normal completion time
was 0 days and the cost to finish on day 43 was $21,800.
Case Five: In case five, the allocation strategy was to consider the critical path and
Activities
Mob
A
B
C
D
E
F
G
H
I
Nodes
20-10
30 - 20
40 - 20
50 - 30
60 - 30
70 - 40
80 - 50
90 - 60
100 - 60
110 - 70
Units of
Material
0
3
0
3
4
0
0
4
4
0
Accumulated
Units
0
3
3
6
10
10
10
14
18
18
80
Activities not Allocated
X50>=20
80
50
120
30
X90>=20
0(1,7)0
A
0(0,1)0
MOB.
1
(0,1)
90
0(7,11)0
D
4
(7,11)
6
(1,7)
10
20
60
Timeline by day 20
0(20,25)0
L
5
(20,25)
130
0(25,33)0
O
8
(25.33)
Units Material xi
Allocated
TF(ES,EF)FF
Act.
Dur.
(LS,LF)
X25>=20
Critical Allocated
i
j
Critical Non Allocated
i
25
0(1,4)12
B
150
Xj>=time (days)
TF=LF-EF
170
X100>=20
100
xi
0(42,43)0
PUNCHLIST
1
(42,43)
0(33,42)0
R
9
(33,42)
0(11,20)0
H
9
(11,20)
Conventions
160
140
3
(13,16)
40
0(4,12)10
E
8
(16,22)
70
7(12,16)10
I
4
(22,26)
110
i
j
i
Figure 6.13
Dummy activity
81
The information from the planned network and the five cases was tabulated and it
is presented in Table 6.11, where the path followed and the amount of materials allocated
in the first 20 days was shown. The amount of materials needed was 29 and the available
was 20 units.
Table 6.11
Mob-A-B-C-D-E-F-H
Mob-A-B-C-D-E-H
Mob-A-B-D-E-H-I
Mob-A-C-D-F-H
Mob-A-B-C-D-F
Mob-A-C-D-G-H
29
20
20
19
19
18
Table 6.12 presented the constraints generated by following the different paths
during the allocation process.
Table 6.12
CASE
Generated Restrictions
PLANNED
ONE
TWO
THREE
FOUR
FIVE
Planned
I>=20, G>=20,L>=20,M>=20,F>20
C>=20. G>=20,L>=20,M>=20,P>=20
B>=20, G>=20, J>=20,L>=20,M>=20
B>=20, G>=20, H>=20,E>=20,J>=20
B>=20, K>=20, F>=20,L>=20,M>=20
Completion
Time
43
43
44
53
52
53
Crashing
Cost($)
43 days
0
0
1400
29800
21800
29800
For each case a cost-time trade-off mathematical model was developed and the
results are presented in Table 6.13 and Figure 6.14. The results generated by the models
were tabulated up to the maximum number of days the project could be crashed.
82
Therefore, the blank spaces on the table mean that it was not possible to crash the project
for that time.
As can be seen on Table 6.13, having the ideal situation (planned) where no
material constraint existed, the project could be crashed 13 days and with a maximum of
$33,800. The importance of considering the critical path can be seen from Case One and
Case Two, where the maximum crashing time was reduced approximately 50%. In Case
Two activity I was finished one day early, but activity P was not able to be started until
activity M was completed. Therefore, it is best to conduct some other activity instead of
activity I in order to minimize the completion time. If the total float of an activity is not
taken into account it will result in having more critical paths and extending the
completion time. For Cases Three and Five, the Free Float (FF) for activity B was 12
days and this activity was not allocated and it meant starting activity B after day 20 and
completing the project 10 days later than the normal duration.
The previous analysis demonstrates that it is important for the success of projects
that Decision Makers be knowledgeable about all of the project details. The appreciation
of the different aspect of projects, such as critical activities, materials constraints, cost
time trade-offs, etc, are more valuable than having the high powered scheduling tools.
Unless the tools are used properly according to the aspects listed above, the results could
lead to more problems than solutions.
83
Table 6.13
Time
43
42
41
40
39
38
37
36
35
34
33
32
31
30
Planned
0
1000
2000
3000
4000
5000
6000
7000
8000
11000
14000
17000
25000
33800
Time-Cost Trade-Off
Crashing Cost
Case
Case
Case
Case
One
Two
Three
Four
0
1400
29800
21800
1000
3800
34800
*
2800
7820
*
*
4600
11840
*
*
7600
17860
*
*
12000
24660
*
*
16400
31460
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
* means that the network cannot be crashed
Case
Five
29800
34800
*
*
*
*
*
*
*
*
*
*
*
*
The result presented on Table 6.13, can be better visualized on Figure 6.14.
84
Time-Cost Tradeoffs
40000
36000
32000
28000
Planned
24000
Case One
Case Two
20000
Case Three
16000
Case Four
12000
Case Five
8000
4000
0
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
6.4
Sensitivity Analysis
In order to check how sensitive the model was in terms of cost to variations in the
amount of days that the critical path could be crashed, a sensitivity analysis was
conducted. The sensitivity analysis consisted of analyzing how different crash time
durations (see Table 6.14) would affect the completion time and cost in six time
scenarios as follows: in time-scenario I normal crash times were used; in time-scenario
II the crash time was assumed to be zero in the critical path; in time-scenario III the crash
time was assumed to be high, which meant greater than 3 days and less than 5; in time-
85
scenario IV the crash time was assumed to be low, which meant less than 3 days and
greater than zero days, in time-scenario V the crash time was assumed to be high in the
first stage of the network and low in the final stage; in time-scenario VI the crash time
was assumed to be low in the first part of the network and high at the end. The results
are presented in Table 6.16 and plotted in Figure 6.15.
The analysis of the critical path was done by creating six Time-Scenarios (TS)
and analyzing the effect of these TS on the five cases formerly defined. As can be seen
on Table 6.14, the general specifications followed in the Time-Scenarios were presented.
Table 6.14
TS-I
TS-II
TS-III
TS-IV
TS-V
TS-VI
(0 Crash(Time) 2)
In Table 6.15, the maximum number of crashing days for the six Time-Scenarios in the
original critical path is presented. The analysis was done just by changing the maximum
crash duration of the original critical path in each of the five cases analyzed. The original
critical activities are highlighted on the Table 6.15 with a darker color.
86
Table 6.15
Activities
Nodes
Mobilization
A
B
C
D
E
F
G
H
I
J
K
L
M
N
O
P
Q
R
Punch list
10 - 20
30 - 20
40 - 20
50 - 30
60 - 30
70 - 40
80 - 50
90 - 60
100 - 60
110 - 70
120 - 80
140 - 90
130 - 100
110 - 100
140 - 120
150 - 130
150 - 110
160 - 140
160 - 150
170 - 160
COST
CODE
C00
C10
C20
C30
C40
C50
C60
C70
C80
C90
C100
C110
C120
C130
C140
C150
C160
C170
C180
C190
TS-I
0
2
1
3
2
3
0
0
3
2
2
1
0
1
0
3
2
3
3
0
TS-II
0
0
1
3
0
3
0
0
0
2
2
1
0
1
0
0
2
3
0
0
TSIII
0
4
1
3
3
3
0
0
5
2
2
1
3
1
0
5
2
3
5
0
TSIV
0
2
1
3
2
3
0
0
2
2
2
1
2
1
0
1
2
3
2
0
TS-V
0
4
1
3
3
3
0
0
5
2
2
1
2
1
0
1
2
3
2
0
TS-V
0
2
1
3
2
3
0
0
2
2
2
1
3
1
0
5
2
3
5
0
In the first sensitivity analysis, the critical path was evaluated for the Case
Planned by using the six crash time scenarios described in Table 6.14. The results are
presented in Table 6.16, where the blank cells mean that crashing was not possible in that
specific time, and Figure 6.15. In the sensitivity analysis for the Case Planned, a check of
the model is verifying that the TS-I results generated should be exactly the same values
as the cost time trade-off for the planned case Table 6.13. The values obtained in the Case
Planned and using TS-I were used in the sensitivity analysis for the other cases as a point
of reference.
87
Table 6.16
Completion
Day
43
42
41
40
39
38
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
TS-I
0
1000
2000
3000
4000
5000
6000
7000
8000
11000
14000
17000
25000
33800
*
*
*
*
*
*
*
*
TS-II
0
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
TS-III
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10800
12600
15600
18600
21600
24600
27600
32600
37600
45620
61140
77160
TS-IV
0
1000
2000
3000
4000
5000
8000
11000
18000
25000
33000
41000
*
*
*
*
*
*
*
*
*
*
TS-V
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10800
13800
16800
24600
34200
44800
56800
71200
*
*
*
*
*
*
93960
*
*
* Means that the network cannot be crashed
TS-VI
0
1000
2000
3000
4000
5000
6000
7000
8500
10300
13300
16300
19300
22300
25300
34500
43900
57520
*
*
*
*
*
88
SENSITIVITY ANALYSIS PLANNED
70000
60000
50000
S-I
S-II
40000
S-III
S-IV
30000
S-V
S-VI
20000
10000
0
22
27
32
37
42
47
Figure 6.15
From Table 6.16, it can be seen that in the Ideal (planned) case, TS-II cannot be
crashed at all, because the TS-II maximum crashing time in the original critical path is
zero.
All the Time Scenarios had the same constant increasing cost for the first five
crashed days. Time-Scenario IV is the most critical, allowing the lowest network
crashing time with the highest cost. TS-IV was defined as low crashing, meaning that the
maximum crashing time for this scenario was 2 days, as shown in Table 6.15.
Sensitivity analysis for Case One
89
In the second sensitivity analysis, the critical path was evaluated for Case One by
using the six crash time scenarios described in Table 6.14. The original Time-Cost curve
for the planned case was used as a reference point. The results are presented in Table
6.17 and Figure 6.16. The results are presented up to their maximum possible crashing
time for each Time-Scenario. Again in order to check the model, the results for TS-I from
sensitivity analysis Case One (see Table 6.17) must be the same as the values found in
the cost time trade-off for Case One (see Table 6.13).
90
Table 6.17
Completion
Day
43
42
41
40
39
38
37
36
35
34
TIME SCENARIOS
Planned
TS-I
TS-II
TS-III
TS-IV
TS-V
0
0
0
0
0
0
1000
1000
*
1000
1000
1000
2000
2800
*
2800
2800
2800
3000
4600
*
4600
4600
4600
4000
7600
*
7600
7600
7600
5000
12000
*
12000
12000
12000
6000
16400
*
16400
17000
17000
7000
*
*
23200
24400
24400
8000
*
*
30000
*
*
11000
*
*
37400
*
*
* Means that the network cannot be crashed
TS-VI
0
1000
2000
3800
5601
10001
14401
21201
28001
34801
TS-I
25000
TS-II
TS-III
20000
TS-IV
15000
TS-V
TS-VI
10000
Planned
5000
0
35
37
39
41
43
45
Figure 6.16
As can be seen in Table 6.17, TS-III, TS-IV, and TS-V had the same constant
increasing cost for the first 6 crashed days. When put side by side, the curve from the
planned case and the curves from this case show how the cost increases at a higher rate
for all the Time-Scenarios (see Table 6.13).
91
In the third sensitivity analysis, the critical path was evaluated for Case Two by
using the six crash time scenarios described in Table 6.14. The results are presented in
Table 6.18 and Figure 6.17. In this case, it can be seen that even though for TS-II the
crash time for the original critical path was zero, there is a cost for finishing on day 43.
This cost exists because the original critical path changed; now the longest path has a
duration of 44 days. The results generated for Case Two while evaluating TS-I must be
the same values as the results obtained for the cost time trade off for Case Two (see Table
6.17).
Table 6.18
Completion
Day
43
42
41
40
39
38
37
36
Planned
0
1000
2000
3000
4000
5000
6000
7000
TS-I
1400
3800
7820
11840
17860
24660
31460
*
TIME SCENARIOS
TS-II
TS-III
TS-IV
1400
1400
1400
*
3800
3800
*
7820
7820
*
11840
11840
*
16360
15860
*
21960
21960
*
28760
28760
*
36660
35560
TS-V
1400
3800
7820
11840
15860
21960
28760
35560
TS-VI
1400
3800
7820
11840
16360
21960
28760
35560
92
SENSITIVITY ANALYSIS CASE TWO
40000
35000
TS-I
30000
TS-II
25000
TS-III
TS-IV
20000
TS-V
15000
TS-VI
planned
10000
5000
0
36
37
38
39
40
41
42
43
44
Figure 6.17
As can be seen in Table 6.18, the cost of crashing Case Two was similar in all the
different time-scenarios; even TS-II had a one day crashing cost because a new critical
path developed. Therefore, constraining the original path would not affect the completion
time.
Sensitivity Analysis for Case Three
In the fourth sensitivity analysis, the critical path was evaluated for case three by
using the six crash time scenarios described in Table 6.14. The first column (Planned) on
Table 6.19 represents the cost for the ideal case and it was presented up to maximum
crashing time period in Case Three, which was 4 days for TS-VI. Blank cells in the tables
presented mean that it was not possible to crash the project up to that date or number of
days, such as TS-II where the project could not be completed before 43 days. The results
are presented in Table 6.19 and Figure 6.18 .
93
Table 6.19
Completion
Day
Planned
0
1000
2000
3000
4000
43
42
41
40
39
TIME SCENARIOS
TS-II
TS-III
TS-IV
*
25800
26800
*
30800
*
*
35800
*
*
40800
*
*
*
*
TS-V
26800
31800
*
*
*
TS-VI
21801
25801
30801
35800
40800
35000
TS-I
30000
TS-II
TS-III
25000
TS-IV
20000
TS-V
TS-VI
15000
Planned
10000
5000
0
39
40
41
42
43
44
Figure 6.18
In this case, TS- I and Case TS-III are the most critical for two reasons: (1) the
maximum crashing time period is reduced to two days, and (2) the crashing cost is the
highest for these two Time-Scenarios. In general, this sensitivity analysis presented a
high crashing cost for all the TS compared with the planned crashing cost information.
TS-II was not possible to crash and TS-VI was the most favorable among the six TS.
94
Sensitivity Analysis for Case Four
In the fifth sensitivity analysis, the critical path was evaluated for case four by
using the six crash time scenarios described in Table 6.14. The original Time-Cost curve
for the planned case was kept as a benchmark. The results are presented in Table 6.20 and
Figure 6.19.
Table 6.20
Completion
Day
43
42
41
40
39
38
TIME SCENARIOS
Planned
TS-I
TS-II
TS-III
TS-IV
0
21800
*
20600
*
1000
*
*
24200
*
2000
*
*
27800
*
3000
*
*
33800
*
4000
*
*
39800
*
5000
*
*
45800
*
* Means that the network cannot be crashed
TS-V
30800
42800
*
*
*
*
TS-VI
20600
24200
27800
33800
46300
40000
35000
TS-I
30000
TS-III
TS-IV
25000
TS-V
20000
TS-VI
15000
Planned
10000
5000
0
38
39
40
41
42
43
Figure 6.19
44
95
In this analysis TS-V was the most critical, because the maximum crashing time
was one day down to 42 days, and the cost was the highest among the TS. TS-II and TSIV could not be crashed at all. TS-III and TS-VI had similar behavior when crashed up to
3 days, but after that TS-VI became more expensive and less flexible to be crashed.
Sensitivity Analysis for Case Five
In the sixth sensitivity analysis, the critical path was evaluated for case five by
using the six crash time scenarios described in Table 6.14. The maximum number of days
that the project could be crashed was 3 days up to day 40 in this case. Therefore, the
Case Panned cost information was presented up to day 40 to be compared with the results
from the TS in this case. The results are presented in Table 6.21 and Figure 6.20.
Table 6.21
Completion
Day
TIME SCENARIOS
Planned
TS-I
TS-II
TS-III
TS-IV
0
29800
*
25800
31800
43
1000
34800
*
30800
*
42
2000
*
*
35800
*
41
3000
*
*
40800
*
40
* Means that the project cannot be crashed
TS-V
31800
*
*
*
TS-VI
25800
30800
35800
40800
96
SENSITIVITY ANALYSIS CASE FIVE
45000
40000
35000
TS-I
30000
TS-II
TS-III
25000
TS-IV
20000
TS-V
TS-VI
15000
Planned
10000
5000
0
40
40.5
41
41.5
42
42.5
43
43.5
Figure 6.20
This results presented have a more constrained crashing time period, as shown in
Table 6.21 the maximum number of days that the project could be crashed was 3 days for
TS-III and TS-VI. In the case of TS-IV and TS-V the project could not be crashed and in
order to finish in the normal completion time, which is 43 days, the cost was high
($31,800). This is because TS-IV and TS-V have low crashing times at the end of the
critical path, in contrast with TS-III and TS-VI that have high crashing times.
In general even though the TS for each case presented low amounts of crashing
costs in most cases, it is due to the project already being constrained by the initial
allocations. For instance, sensitivity analysis three and five reduce the amount of
maximum crashing days to 1 day. However, with TS-III and TS-VI that amount increased
97
from 1 to 4 days. The allocation of materials is critical as presented in the five cases
analyzed, but when the critical path had a high float or the maximum amount of crashing
days is high the effect is reduced.
Time-Scenario III resulted in favorable results in the analysis, because it allowed
high crashing times. A similar behavior was seen in TS-VI because at the end of this
scenario the maximum amount of crashing days was high because the first half of the
project was already allocated.
98
6.5
Using the information from the completion time and cost models, membership
functions were generated. The normal completion time was 43 days and the objective was
to finish 6 days early by day 37. Therefore, the time membership function was
represented as:
if zT 37,
1
uT ( % )
1
y %
zT
0
Figure 6.21
37
43
(d a ys)
The planned cost for crashing the project 6 days was $6,000. Therefore, the
membership function was:
99
if zC = 0,
1
In Figure 6.22, a graphical illustration for the Cost Membership Function was
presented. The dependent variable zC represents the cost for the respective completion
time.
uC ( % )
1
y%
zC
0
Figure 6.22
6000
(Cost)
In Table 6.22, the results generated using single objective models for completion
time and cost are presented. Although 100% satisfaction was reached in this model, just
one of the objectives was considered. For instance, the completion time is 37 days, which
means 100% satisfaction, but the cost of crashing the project is $6,000, which was
defined in the membership function with zero satisfaction. In these mathematical models,
100
fz represents the final satisfaction level, fz1 the satisfaction level for the minimum
completion time, and fz 2 the satisfaction level for the minimum crashing cost.
Table 6.22
Cost
100%
100%
z1 (days)
37
43
fz1 (%)
100
z 2 ($)
6000
100
Parameter
fz
fz 2 (%)
In Table 6.23, the results generated from the fuzzy mathematical model where
completion time, minimum cost, and material constraints were considered for the Case
Planned are presented. The result from the fuzzy mathematical model generated the same
completion date ( z1 = 40 days ) for all the TS with 0.5 satisfaction level ( fz = 50% ),
and minimum cost ( z2 = $3,000 ) with 0.5 satisfaction level ( fz1 = 50% ).
TIME SCENARIOS
TS-III
TS-IV
TS-I
TS-II
50%
0%
50%
z1 (days)
40
43
fz1 (%)
50
fz
z 2 ($)
fz 2 (%)
TS-V
TS-VI
50%
50%
50%
40
40
40
40
50
50
50
50
3000
3000
3000
3000
3000
50
100
50
50
50
50
In Figure 6.23, the results from the fuzzy mathematical models are presented,
along with the results from the single objective mathematical models for completion time
101
and cost, which were basis for defining the boundaries for the membership functions in
the fuzzy mathematical models.
Case Planned
7000
TS-VI
6000
Planned
Complet.
TS-I
5000
Cost ($)
TS-II
4000
TS-I
TS-III
TS-V
TS-IV
3000
TS-V
TS-IV
TS-VI
TS-III
2000
Complet.
Planned
Cost
1000
Cost
TS-II
0
36
37
38
39
40
41
42
43
44
Time (days)
Figure 6.23
The results from the fuzzy mathematical models for the Case Planned were constant for
all of the case scenarios. However the path followed to reach the solution were different
as shown in Table 6.24, where three Time Scenarios (TS) followed path 1, two TS
followed path 2 and one was not possible to be crashed. These were the ideal solutions
with no material constraints.
Table 6.24
Path 1
Path 2
C10 (2)
C10 ( 3)
C150 (1)
I - IV - VI
III - VI
102
Fuzzy Mathematical Model for Case One
The cost for crashing the project 6 days in Case One was $16,400. Therefore, the
membership function was:
if z C = 0,
1
uC ( % )
1
y%
zC
0
Figure 6.24
16,400
(Cost)
The single objective model that was used to define the membership function
103
values generated the results that are listed in
Table 6.25.
Table 6.25
Cost
100%
100%
z1 (days)
37
43
fz1 (%)
100
16400
100
Parameter
fz
z 2 ($)
fz 2 (%)
In
Table 6.26, the results generated from the fuzzy mathematical model for Case
One are grouped.
Table 6.26
Parameter
fz
TS-I
TIME SCENARIOS
TS-III
TS-IV
TS-V
TS-II
TS-VI
53.7%
0%
53.7%
50%
50%
53.7%
39
43
39
40
40
39
fz1 (%)
66.7
66.7
50
50
66.7
z 2 ($)
7600
7600
7000
7000
7600
fz 2 (%)
53.7
100
53.7
57.3
57.3
53.7
z1 (days)
In
Figure 6.25, the results from the single objective models and the fuzzy
mathematical models are presented.
104
Figure 6.25
18000
16000
Completion Time
14000
Cost
Cost ($)
12000
TS-I
10000
TS-II
8000
TS-III
TS-IV
6000
TS-V
4000
TS-VI
2000
0
36
37
38
39
40
41
42
43
44
Time (days)
The optimal solutions for Case One are reaching a satisfaction level of 49.7 and
50%, and these solutions consider the material constraints.
The cost for crashing the project 6 days in Case Two was $31,460. Therefore, the
membership function was:
if z C = 0,
1
Figure 6.24 shows a graphical illustration for the Cost Membership Function. The
105
dependent variable zC represents the cost for the respective completion time.
uC ( % )
1
y%
zC
0
31,460
Figure 6.26
(Cost)
Time-Scenarios were analyzed and the results from the minimum completion time
model and the minimum cost time model were used as reference points.
Table 6.27
Cost
100%
100%
z1 (days)
37
43
fz1 (%)
100
31460
100
Parameter
fz
z 2 ($)
fz 2 (%)
106
In Table 6.28, the results generated from the fuzzy mathematical model for Case Two
are shown with the satisfaction level for each objective and for the general objective.
Table 6.28 Fuzzy Mathematical Model Results (Case Two)
Parameter
fz
TS-I
TIME SCENARIOS
TS-III
TS-IV
TS-V
TS-II
TS-VI
50%
0%
50%
49.70%
49.70%
50%
z1 (days)
40
43
40
40
40
40
fz1 (%)
50
50
50
50
50
11840
12360
11840
15840
15840
11840
62.4
60.7
62.4
49.7
49.7
62.4
z 2 ($)
fz 2 (%)
Case Two
35000
30000
TS-I
Cost ($)
TS-II
TS-III
25000
TS-IV
TS-V
20000
TS-VI
Completion Time
Cost
15000
10000
37
38
39
40
41
42
43
44
Time (days)
Figure 6.27
Using fuzzy mathematical models, it was shown that although the models may have
the same satisfaction levels for the three cases analyzed (Planned, Case One, and Case
Two), the costs are quite different (see Table 6.29). The fuzzy mathematical models had
107
the same interval of time for the three cases (crashed the project 6 days). In order to find
the same satisfaction level in these cases, they have to be crashed 3 days (Cases planned,
and Two) and 4 days (Case Two), as shown in the following table.
Table 6.29
Case
Case
Parameter Planned One
Two
fz
50%
50%
50%
z1 (days)
40
39
40
fz1 (%)
50%
67%
50%
z 2 ($)
3000
7600
11840
fz 2 (%)
50%
53%
62%
108
109
Fuzzy mathematical models allow the inclusion of time and costs in the analysis
process. In addition, material restrictions can be included as constraints that could
determine more realistic solutions. The more constraints that are included in an analysis
the more the analysis supports the decisions generated using fuzzy mathematical models.
Sensitivity analysis techniques were used to analyze different paths in the networks in
order to determine material constraints and their effects on the cost of projects. Using
allocation analysis for the case study showed how important it is to monitor the critical
path during an analysis. Assuming that the maximum possible cost is $30,000, and 43
days is the completion time, the following resulted from the analysis. When the critical
paths was not considered, the project completion time was increased by 20% from 43 to
52 days (Case Four), and the cost in this case increased approximately 73% up to
$21,800. This increased time and cost was due to the activities that were not critical
becoming critical.
Primavera Project Management (P5) does not directly perform cost-time trade-off
calculations. For crash cost comparison purposes, the baseline of a project must be first
created with regular cost, and then generate the crash cost on resources, activities, etc. to
compare the baseline costs with the current schedule. It then encompasses a process in
which schedules are revised again and again. Finally, this information is exported to
Excel to create the cost-time trade-off graphics. These processes to create the cost-time
trade-off graphics make the Primavera Project Software not practical in terms of this
aspect. Mathematical models can be used to directly perform cost-time trade-off
analysis.
110
From the results, it was determined that it is not just whether there is a shortage of a
material but the way materials are allocated that affects projects in different ways. It is
important to analyze schedules to check for the activities that have minimum float,
instead of just allocating materials to activities that are immediately ready to start.
Mathematical Model Development Issues
Creating the Activity on Arrow (AOA) network was a lengthy process due to the
difficulty in representing some network relationships, such as Start to Start (SS) with lags
and Finish to Finish (FF) with lags. These relationships created many dummy
connections that made the network confusing to read and understand, as shown in Figure
C.5.
Attention should be focused on the technological constraints of the model, on logic
flow, and critical activities because the results from the fuzzy mathematical models
would the optimal for the conditions created in the model.
Even though schedules can be planned and developed for construction projects,
there is always a possibility for unexpected material shortages for construction projects.
Therefore, taking this constraint into account helps project managers to evaluate
situations and make better decisions. Furthermore, traditional mathematical models may
generate optimal solutions, but even though they may be single objective-oriented or
multiobjective, they do not consider fuzziness as in the case when using fuzzy
mathematical methods. There are computers that can process complex fuzzy
mathematical models and that are able to generate more realistic approaches to finding
the optimal trade-off between cost and time for network projects in construction.
111
Decision makers have to deal with uncertainties from parameters such as time,
cost, aspiration levels, and material availability. These uncertainties should be addressed
to improve efficiency and labor productivity for construction projects. Membership
functions are used in fuzzy mathematical methods in order to incorporate the DMs
uncertainties and to serve as a tool in decision making processes. Membership functions
serve as tools that can be used by researchers and practitioners in decision making
processes that incorporate material constraints when planning different construction
projects. The more constraints that are initially considered in construction projects, the
better the results that will be generated although a project will never reflect reality since
common sense from the DM is also required in decision making. The model considered
in this investigation was based on linear programming and many real-life environmental
relationships are nonlinear.
The optimizations generated when using mathematical models could create a gap
between researchers and practitioners in engineering, due to the high level of complexity
required to process these models. In addition, real network projects have more complex
precedence relationships that involve start to start relationships, finish to finish
relationships, and dummy activities. Therefore, a user-friendly interface needs be
developed in order to encourage the use of mathematical models by practitioners. The
model should be user-friendly in terms of input requirements and output interpretation
(Daellenbach and Flood, 2002, p. 178).
Mathematical models allow decision makers to have an opportunity to analyze the
possible effects of material delays before making final decisions. When a project is going
112
to be delayed, or stopped, because a material is delayed the decision to be made has to be
analyzed in terms of incurring extra cost and finishing on time or waiting for the planned
material and delaying the whole project with possibly secondary effects on costs.
This model would be a valuable tool for analyzing possible material shortages and
for determining the critical activities to be analyzed in networks. An analysis could be
done to compare other possibilities such as expediting materials, finding a different
supplier, or using a more expensive material that could be delivered right away may be,
in the long term, more favorable for projects than crashing the schedule.
7.1
models in construction project scheduling. This lack of advance research in the use of
mathematical models in project scheduling leaves the commercial software as the only
alternative for analyzing construction project schedules.
It is important to first create a case study where all the advantages of the
mathematical model can be tested, better understood and finally validated with a real
construction project network.
Future research using mathematical models could be conducted to create and
analyze different project environments. These environments could consider more
variables and constraints, such as considering more than one material to help validate a
more robust model.
This research considered the fuzziness of the aspiration levels from the DM, and
the personnel involved in the decision making process, for the project in general. Future
113
research could also address the fuzziness in different stages of the model, such as fuzzy
individual activities.
Another avenue of research would be to consider how to reschedule a network
with a fixed amount of materials, or more rigorous artificial intelligence models could be
used such as Genetic Algorithms (GA).
Real time tracking of materials and equipment would be useful in order to control
inventory on projects. Assuming there is a shortage of materials, when the materials are
actually located somewhere at the job site has a double negative effect. Simulations to
show these effects could be performed using mathematical modeling techniques.
Mathematical models should be simple, but at the same time they need to be quite
detailed and also consider critical constraints to be realistic. No matter how efficient a
mathematical model is, they are prepared by people, which means the model could be
incorrect; therefore, using several different modeling techniques would allow for
verification of the results of the models.
114
REFERENCES:
1.
2.
3.
Ayyub, B. M., and Haldar, A. (1984). Project scheduling using fuzzy set
concepts. Journal of Construction Engineering and Management, ASCE, 110(2),
189-204.
4.
Chua, D. K. H., and Shen, L. J., (2005). Key constraints analysis with integrated
production scheduler Journal of Construction Engineering and Management,
ASCE, 131(7), 753-764.
5.
6.
Deporter, E. L., and Ellis, K. P., (1990). Optimization of project networks with
goal programming and fuzzy linear programming. Computer and Industrial
Engineering, 19(1-4), 500-504.
7.
Halpin, D., and Woodhead R. (1998). Construction Management, 2nd ed., John
Wiley & Sons, Inc., New York, N.Y.
8.
9.
115
Optimization Programming Language (OPL) module of ILOG OPL Development
Studio (IDE), Version 4.2, (2006). ILOG, S.A., Mountain View, California, CA.
10.
11.
Kim, K., and de la Garza, J. M. (2005). Critical path method with multiple
calendars. Journal of Construction Engineering and Management, ASCE,
131(3), 330-342.
12.
13.
14.
15.
16.
17.
Moder, J. J., Phillips, C. R., and Davis, E. (1983). Project management with
116
CPM, PERT, and precedence Diagramming. 3rd ed., New York: Van Nostrand
Reinhold Company.
18.
19.
Nepal, M. P., Park, M., and Son B. (2006). Effects of Schedule Pressure on
Construction Performance. Journal of Construction Engineering and
Management, ASCE, 132(2), 182-188.
20.
21.
22.
23.
24.
Primavera Project Management (P5), release 5.0, (2005), Primavera Systems, Inc.
Bala Cynwyd, Philadelphia, PA.
25.
117
Pennsylvania State University, Pennsylvania, P.A.
26.
27.
28.
Wiest, J. D. (1964). Some properties of schedules for large projects with limited
resources. Operation Research, 12 (3), 395-418.
29.
30.
31.
32.
Zheng, D. X., and Ng, T., and Kumaraswamy, M. M. (2004). Applying a genetic
algorithm-based multiobjective approach for time-cost optimization. Journal of
Construction Engineering and Management, 130(2), 168-176.
33.
118
Appendix A.
Abbreviations
AOA
activity on arrow
AON
activity on node
CAD
computer-aided design
CPM
DM
decision maker
FMM
FST
GP
goal programming
IP
integer programming
LP
linear programming
OPL
PDM
PERT
PM
project manager
119
Appendix B.
Definitions
Activity. Each of the jobs or tasks which can be related to each other or have different
evaluate how well a project execution has been done (Mubarak, 2005, p.49).
Schedule Crashing. Process of completing a project in less time than planned by using
the possible paths to achieve a defined goal. Any delay in this path will delay
the whole project (Knutson anf Bitz, 1991, p, 66).
Delay. Activities that are started later than planned. Time that these activities are
120
Free Float. Amount of time that an activity can be postponed without delaying any of
the following activities-successors ( Halpin and Woodhead, 1998, p. 111;
Winston, 1994, p. 422; Schexnayder and Mayo, 2004, p. 111).
Heuristic Methods. Techniques that do not used mathematical rigor, instead they
depend on the rules of thumb of DMs. They work well in many cases and give
good solutions but not optimal solutions (Zheng et al., 2004).
Total float. The total mount of time that an activity in a project network can be
objective at the same time. These mathematical models have trade-off between
the different objectives(Revelle et al, 1997).
Mathematical Programming. In this context the word programming is meaning
121
Predecessors. In a network project activities that have to be done before a new activity
can start (Winston, 1994, p. 413; Mubarak, 2005, p. 20).
Project network. Graphical representation of the different relationships among the
activities in a project.
Successors. Activities that can be started after an activity is completed (Mubarak,
2005, p. 20).
122
Time-Cost
(Tradeoffs)
Timeduration
(days)
142
141
140
139
138
137
136
135
134
133
132
131
130
129
128
127
126
125
124
123
122
121
120
119
118
117
116
115
114
113
112
111
110
109
108
107
106
105
104
103
102
101
100
123
Figure C.1
124
Figure C.2
125
Figure C.3
126
Figure C.4
127
Figure C.5
128
x21+x22 >=400;
x11+x21 <=350;
//FRAM1 constraint
x12+x22 <=410;
//FRAM2 constraint
129
0.4*x11-0.6*x12 >=0; //minimum FRAM1 contain in CONCRETE A
-0.7*x21+0.3*x22 >=0; //minimum FRAM2 contain in CONCRETE B
};
130
131
Network Relationships for the 60 activity network
X10==0;
X20-X10>=3;
X290-X280>=2;
X30-X20>=4;
X40-X30>=3;
X300-X270>=5;
X50-X40>=10;
X310-X280>=5;
X50-X30>=7;
X320-X250>=8;
X60-X50>=2;
X320-X310>=10;
X70-X60>=3;
X330-X300>=10;
X80-X70>=3;
X330-X240>=6;
X90-X80>=4;
X340-X260>=8;
X350-X340>=0;
X100-X90>=8;
X350-X170>=6;
X110-X100>=0;
X360-X340>=0;
X110-X90>=12;
X360-X80>=0;
X120-X100>=0;
X370-X360>=2;
X120-X30>=8;
X380-X340>=4;
X130-X110>=0;
X390-X380>=0;
X130-X70>=7;
X390-X370>=0;
X140-X130>=1;
X395-X340>=8;
X150-X140>=3;
X400-X220>=6;
X160-X140>=6;
X400-X395>=0;
X160-X150>=0;
X170-X160>=1;
X405-X150>=0;
X180-X170>=2;
X405-X395>=0;
X190-X170>=3;
X410-X395>=0;
X200-X170>=12;
X410-X240>=2;
X200-X180>=2;
X420-X350>=5;
X210-X190>=7;
X420-X400>=5;
X210-X200>=0;
X430-X380>=4;
X220-X210>=8;
X430-X270>=2;
X230-X220>=5;
X440-X405>=2;
X240-X230>=1;
X450-X410>=4;
X250-X240>=3;
X460-X370>=0;
X260-X240>=7;
X460-X440>=0;
X270-X260>=0;
X470-X440>=0;
X270-X110>=0;
X470-X450>=0;
X280-X260>=0;
X480-X290>=0;
X280-X200>=0;
X480-X450>=4;
X490-X430>=6;
X540-X530>=0;
X500-X480>=6;
X550-X510>=10;
X510-X500>=0;
X550-X320>=0;
X510-X470>=11;
X550-X330>=0;
132
X530-X390>=10;
X520-X490>=0;
X520-X420>=6;
X520-X500>=0;
X530-X490>=0;
X540-X510>=0;
X550-X540>=4;
X550-X460>=2;
X550-X530>=12;
X550-X520>=1;
X560-X550>=2;
X550-X120>=15;
X570-X560>=1; };
133
subject to
{
fz<=(30600-z1)/30600;
fz<=(142-z2)/32;
z2==X570-X10;
z1==1000*C10+4000*C20B+3020*C30+8000*C40B+500*C50+300*C60+1000*C70
+800*C80+2600*C90+1400*C100+1600*C110B+300*C120+1400*C130+500*C140+6
00*C150+800*C160+1800*C170+1600*C180+1000*C190+200*C200+500*C210+400
*C220+600*C230+500*C250+2000*C260+200*C270+800*C280+900*C290+400*C30
0+400*C310+500*C320B+600*C340+800*C350+1000*C360+600*C370+600*C380+4
00*C390+500*C400B+600*C420+400*C430+600*C450+800*C460+2000*C470+200*
C500+400*C510;
Maximum Activities Crashing Times for the 60 activity network
C10<=1;
C260<=2;
C20A<=0;
C270<=1;
C20B<=6;
C280<=3;
C30<=2;
C290<=5;
C40A<=0;
C300<=2;
C40B<=4;
C310<=2;
C40C<=0;
C320A<=0;
C320B<=3;
C50<=1;
C330<=0;
C60<=1;
C340<=4;
C70<=1;
C350<=2;
C80<=2;
C360<=4;
C90<=4;
C370<=5;
C100<=6;
C380<=2;
C110A<=0;
C390<=2;
C110B<=2;
C395<=0;
C110C<=0;
134
C120<=1;
C130<=2;
C140<=2;
C150<=2;
C160<=2;
C170<=5;
C180<=7;
C145<=0;
C190<=2;
C135<=0;
C200<=1;
C220<=2;
C230<=2;
C240<=0;
C250<=1;
C400A<=0;
C400B<=3;
C405<=0;
C410<=0;
C420<=4;
C430<=2;
C440<=0;
C450<=3;
C460<=2;
C470<=4;
C480<=0;
C490<=0;
C500<=1;
C510<=1;
C520<=0;
135
X210-X190 + C160>=7;
X210-X200>=0;
X220-X210 + C190>=8;
X230-X220 + C230>=5;
X240-X230 + C240>=1;
X250-X240 + C260>=3;
X260-X240 + C280>=12;
X270-X260>=0;
X270-X110>=0;
X280-X260>=0;
X280-X200>=0;
X290-X280 + C200>=4;
X300-X270 + C150>=5;
X310-X280 + C220>=5;
X320-X250 + C460>=8;
X320-X310 + C370>=10;
X330-X300 + C290>=10;
X330-X240 + C265>=6;
X550-X530 + C360>=12;
X550-X520 + C490>=1;
X560-X550 + C510>=2;
Final solution for fuzzy model:
fz = 0.59375;
z2 = 123;
z1 = 11800;
C10 = 1;
C20B = 0;
C30 = 0;
C40B = 0;
C50 = 1;
C60 = 1;
C70 = 1;
C80 = 0;
C90 = 0;
C100 = 0;
C110B = 0;
C120 = 0;
C130 = 0;
C140 = 0;
C150 = 0;
C160 = 0;
C170 = 0;
X480-X290>=0;
X480-X450 + C400B>=4;
X490-X430 + C350>=6;
X500-X480 + C430>=1;
X510-X500>=0;
X510-X470 + C420>=11;
X530-X390 + C340>=10;
X520-X490>=0;
X520-X420 + C440>=6;
X520-X500>=0;
X530-X490>=0;
X540-X510>=0;
X540-X530>=0;
X550-X510 + C450>=10;
X550-X320>=0;
X550-X330>=0;
X550-X540 + C500>=4;
X550-X460>=2;
X550-X120 + C470>=15;
X570-X560 + C520>=1;
};
C300 = 2;
C310 = 0;
C320B = 0;
C340 = 0;
C350 = 0;
C360 = 0;
C370 = 0;
C380 = 2;
C390 = 0;
C400B = 0;
C420 = 4;
C430 = 0;
C450 = 1;
C460 = 0;
C470 = 0;
C500 = 0;
C510 = 1;
X570 = 123;
X10 = 0;
C20A = 0;
C40A = 0;
C40C = 0;
136
C180 = 0;
C190 = 0;
C200 = 0;
C210 = 0;
C220 = 0;
C230 = 2;
C250 = 0;
C260 = 0;
X120 = 106;
X100 = 38;
X80 = 25;
X90 = 28;
X110 = 42;
X130 = 42;
X140 = 43;
X150 = 46;
X160 = 49;
X170 = 50;
X180 = 52;
X190 = 53;
X200 = 68;
X210 = 68;
X220 = 76;
X230 = 79;
X240 = 80;
X250 = 113;
X260 = 89;
X270 = 89;
C270 = 0;
C280 = 3;
C290 = 0;
X280 = 89;
X290 = 109;
X300 = 111;
X310 = 111;
X320 = 121;
X330 = 121;
C265 = 0;
X340 = 95;
X350 = 109;
X360 = 95;
X370 = 97;
C110A = 0;
C110C = 0;
C145 = 0;
C135 = 0;
C240 = 0;
C320A = 0;
C330 = 0;
C395 = 0;
C400A = 0;
C405 = 0;
C410 = 0;
C440 = 0;
C480 = 0;
C490 = 0;
C520 = 0;
X20 = 2;
X30 = 6;
X40 = 9;
X50 = 19;
X60 = 21;
X70 = 23;
X380 = 99;
X390 = 99;
X395 = 101;
X405 = 101;
X400 = 109;
X410 = 101;
X420 = 114;
X430 = 103;
X440 = 103;
X450 = 105;
X460 = 119;
X470 = 105;
X480 = 109;
X490 = 109;
X500 = 110;
X510 = 112;
X530 = 109;
X520 = 120;
X540 = 117;
X550 = 121;
X560 = 122;
137
The variables defined and the objective will be the same for all the completion
time models for the five cases. Therefore, the only information presented from the
models was the network relationships for each case and the final solutions.
138
Final solution
z = 43;
X170 = 43;
X10 = 0;
X20 = 1;
X30 = 7;
X40 = 4;
X50 = 12;
X60 = 11;
X70 = 106;
X80 = 19;
X90 = 22;
X100 = 20;
X110 = 26;
X120 = 23;
X130 = 25;
X140 = 25;
X150 = 33;
X160 = 42;
X100-X60>=9;
X100>=20; //*ACTIVITY L AND M*
X110-X70>4;
X110-X100>=3;
X110>=20; //*ACTIVITY P*
X120-X80>=4;
X130-X100>=5;
X140-X90>=3;
X140-X120>=2;
X150-X130>=8;
X150-X110>=7;
X160-X150>=9;
X160-X140>=5;
X170-X160>=1;
};
X70 = 18;
X80 = 32;
X90 = 22;
X100 = 20;
X110 = 23;
X120 = 36;
X130 = 26;
X140 = 38;
X150 = 34;
X160 = 43;
139
Network Relationships for Case Three
X10==0;
X20-X10>=1;
X25-X20>=0;
X25>=20; //*Activity B*
X30-X20>=6;
X40-X25>=3;
X50-X30>=5;
X60-X30>=4;
X65-X60>=0;
X65>=20; //*Activity G*
X70-X40>=8;
X80-X50>=7;
X80>=20; //*Activity J*
X90-X65>=2;
X100-X60>=9;
X100>=20;//*Activity L and M*
X110-X70>4;
X110-X100>=3;
X120-X80>=4;
X130-X100>=5;
X140-X90>=3;
X140-X120>=2;
X150-X130>=8;
X150-X110>=7;
X160-X150>=9;
X160-X140>=5;
X170-X160>=1;
};
X65 = 20;
X70 = 12;
X80 = 19;
X90 = 22;
X100 = 20;
X110 = 26;
X120 = 35;
X130 = 25;
X140 = 37;
X150 = 33;
X160 = 42;
X100-X60>=9;
X110-X70>4;
X110-X100>=3;
X120-X80>=4;
X130-X100>=5;
X140-X90>=3;
X140-X120>=2;
X150-X130>=8;
140
X70-X40>=8;
X80-X50>=7;
X80>=20;//*Activity J*
X90-X60>=2;
X60>=20;//*Activity G AND H*
X150-X110>=7;
X160-X150>=9;
X160-X140>=5;
X170-X160>=1;};
X70 = 28;
X80 = 20;
X90 = 43;
X100 = 29;
X110 = 35;
X120 = 44;
X130 = 34;
X140 = 46;
X150 = 42;
X160 = 51;
X100-X60>=9;
X100>=20;//*Activity L and M*
X110-X70>4;
X110-X100>=3;
X120-X80>=4;
X130-X100>=5;
X140-X90>=3;
X140-X120>=2;
X150-X130>=8;
X150-X110>=7;
X160-X150>=9;
X160-X140>=5;
X170-X160>=1;};
X80 = 27;
X90 = 30;
X100 = 20;
X110 = 36;
X120 = 31;
X130 = 35;
141
X40 = 23;
X50 = 20;
X60 = 11;
X70 = 31;
X140 = 33;
X150 = 43;
X160 = 52;
142
143
Maximum Activities Crashing Times
C00<=0;
C10<=2;
C20<=1;
C30<=3;
C40<=2;
C50<=3;
C60<=0;
C70<=0;
C80<=3;
C90<=2;
C100<=2;
C110<=1;
C120<=0;
C130<=1;
C140<=0;
C150<=3;
C160<=2;
C170<=3;
C180<=3;
C190<=0;
The variables defined, the maximum crashing times, and the objective for the crashing
cost was the same for all of the five model cases. Therefore, the only information
presented from the models was the network relationships and the final solutions.
C00<=0;
C10<=2;
C20<=1;
C30<=3;
C40<=2;
C50<=3;
C60<=0;
C70<=0;
C80<=3;
C90<=2;
C100<=2;
C110<=1;
C120<=0;
C130<=1;
C140<=0;
C150<=3;
C160<=2;
C170<=3;
C180<=3;
C190<=0;
144
Final solution for crashing the network its maximum possible number of days (13):
Planned case
X110-X70+C90>4;
z = 33800;
X110-X100+C130>=3;
X10==0;
X120-X80+C100>=4;
X20-X10+C00>=1;
X130-X100+C120>=5;
X30-X20+C10>=6;
X140-X90+C110>=3;
X40-X20+C20>=3;
X140-X120+C140>=2;
X50-X30+C30>=5;
X150-X130+C150>=8;
X60-X30+C40>=4;
X150-X110+C160>=7;
X70-X40+C50>=8;
X160-X150+C180>=9;
X80-X50+C60>=7;
X160-X140+C170>=5;
X90-X60+C70>=2;
X170-X160+C190>=1;
X100-X60+C80>=9;
X170-X10<=30 ;};
For Case One the path that was the followed was: Mobilization-A-B-C-D-E-H
Network Relationships
Case One
X10==0;
X20-X10+C00>=1;
X30-X20+C10>=6;
X40-X20+C20>=3;
X50-X30+C30>=5;
X50>=20; //*Activity F*
X65-X60>=0;
X65>=20;//*G*
X60-X30+C40>=4;
X70-X40+C50>=8;
X70>=20; //*Activity I*
X80-X50+C60>=7;
X90-X65+C70>=2;
X100-X60+C80>=9;
X110-X70+C90>4;
X110-X100+C130>=3;
X120-X80+C100>=4;
X130-X100+C120>=5;
X140-X90+C110>=3;
X140-X120+C140>=2;
X150-X130+C150>=8;
X150-X110+C160>=7;
X160-X150+C180>=9;
X160-X140+C170>=5;
X170-X160+C190>=1;
X170-X10<=35 ;};
Final solution for crashing the network its maximum possible number of days (8):
z = 31200;
C190 = 0;
C10 = 2;
C00 = 0;
C20 = 0;
X20 = 1;
C30 = 0;
X30 = 5;
C40 = 0;
X40 = 4;
C50 = 0;
X50 = 20;
145
C60 = 0;
C70 = 0;
C80 = 3;
C90 = 2;
C100 = 2;
C110 = 0;
C120 = 0;
C130 = 0;
C140 = 0;
C150 = 0;
C160 = 2;
C170 = 2;
C180 = 3;
X10 = 0;
X65 = 20;
X60 = 9;
X70 = 20;
X80 = 27;
X90 = 22;
X100 = 15;
X110 = 23;
X120 = 29;
X130 = 20;
X140 = 31;
X150 = 28;
X160 = 34;
X170 = 35;
For Case Two the path that was the followed was: Mobilization-A-B-D-E-H-I
Network Relationships
Case Two
X10==0;
X20-X10+C00>=1;
X30-X20+C10>=6;
X35-X30>=0;
X35>=20; //*ACTIVITY C*
X40-X20+C20>=3;
X50-X35+C30>=5;
X60-X30+C40>=4;
X65-X60>=0;
X65>=20; //*ACTIVITY G*
X70-X40+C50>=8;
X80-X50+C60>=7;
X90-X65+C70>=2;
X100-X60+C80>=9;
Final solution for crashing the network its maximum possible number of days (7):
z = 30260;
C10 = 2;
C20 = 0;
C30 = 3;
C40 = 0;
C50 = 0;
C00 = 0;
X10 = 0;
X20 = 1;
X30 = 5;
X35 = 20;
X40 = 4;
146
C60 = 0;
C70 = 0;
C80 = 3;
C90 = 0;
C100 = 2;
C110 = 0;
C120 = 0;
C130 = 0;
C140 = 0;
C150 = 2;
C160 = 0;
C170 = 3;
C180 = 0;
C190 = 0;
X50 = 22;
X60 = 9;
X65 = 20;
X70 = 14;
X80 = 29;
X90 = 22;
X100 = 15;
X110 = 19;
X120 = 31;
X130 = 20;
X140 = 33;
X150 = 26;
X160 = 35;
X170 = 36;
For Case Three the path that was the followed was: Mobilization-A-C-D-F-H
Network Relationships
Case Three
X10==0;
X20-X10+C00>=1;
X25-X20>=0;
X25>=20; //*Activity B*
X30-X20+C10>=6;
X40-X25+C20>=3;
X50-X30+C30>=5;
X60-X30+C40>=4;
X65-X60>=0;
X65>=20; //*Activity G*
X70-X40+C50>=8;
X80-X50+C60>=7;
X80>=20; //*Activity J*
X90-X65+C70>=2;
X100-X60+C80>=9;
X110-X70+C90>4;X110X100+C130>=3;
X120-X80+C100>=4;
X130-X100+C120>=5;
X140-X90+C110>=3;
X140-X120+C140>=2;
X150-X130+C150>=8;
X150-X110+C160>=7;
X160-X150+C180>=9;
X160-X140+C170>=5;
X170-X160+C190>=1;
X170-X10<=42;};
Final solution for crashing the network its maximum possible number of days (2):
z = 34800;
C10 = 0;
C20 = 1;
C30 = 0;
C40 = 0;
C50 = 3;
X10 = 0;
X20 = 1;
X25 = 20;
X30 = 7;
X40 = 22;
147
C60 = 0;
C70 = 0;
C80 = 0;
C90 = 2;
C100 = 0;
C110 = 0;
C120 = 0;
C130 = 0;
C140 = 0;
C150 = 0;
C160 = 2;
C170 = 0;
C180 = 3;
C190 = 0;
C00 = 0;
X50 = 13;
X60 = 11;
X65 = 20;
X70 = 27;
X80 = 20;
X90 = 22;
X100 = 20;
X110 = 30;
X120 = 34;
X130 = 25;
X140 = 36;
X150 = 35;
X160 = 41;
X170 = 42;
For Case Four the path that was the followed was: Mobilization-A-B-C-D-E-F
Network Relationships
Case Four
X10==0;
X20-X10+C00>=1;
X30-X20+C10>=6;
X40-X20+C20>=3;
X40>=20; //*Activity E*
X50-X30+C30>=5;
X60-X30+C40>=4;
X60>=20; //*Activities G AND H*
X70-X40+C50>=8;
X80-X50+C60>=7;
X80>=20; //*Activity J*
X90-X60+C70>=2;
X100-X60+C80>=9;
X110-X70+C90>4;
X110-X100+C130>=3;
X120-X80+C100>=4;
X130-X100+C120>=5;
X140-X90+C110>=3;
X140-X120+C140>=2;
X150-X130+C150>=8;
X150-X110+C160>=7;
X160-X150+C180>=9;
X160-X140+C170>=5;
X170-X160+C190>=1;
X170-X10<=43 ;};
Final solution for crashing the network its maximum possible number of days (1):
z = 21800;
C10 = 0;
C00 = 0;
C20 = 0;
X10 = 0;
C30 = 0;
X20 = 1;
C40 = 0;
X30 = 7;
148
C50 = 0;
C60 = 0;
C70 = 0;
C80 = 3;
C90 = 2;
C100 = 0;
C110 = 0;
C120 = 0;
C130 = 0;
C140 = 0;
C150 = 3;
C160 = 2;
C170 = 0;
C180 = 3;
C190 = 0;
X40 = 20;
X50 = 13;
X60 = 20;
X70 = 28;
X80 = 20;
X90 = 22;
X100 = 26;
X110 = 31;
X120 = 35;
X130 = 31;
X140 = 37;
X150 = 36;
X160 = 42;
X170 = 43;
For Case Five the path that was the followed was: Mobilization-A-C-D-G-H
X90>=20;//*Activity K*
X100-X60+C80>=9;
X100>=20;//*ACTIVITY L AND M*
X110-X70+C90>4;
X110-X100+C130>=3;
X120-X80+C100>=4;
X130-X100+C120>=5;
X140-X90+C110>=3;
X140-X120+C140>=2;
X150-X130+C150>=8;
X150-X110+C160>=7;
X160-X150+C180>=9;
X160-X140+C170>=5;
X170-X160+C190>=1;
X170-X10<=42;};
Final solution for crashing the network its maximum possible number of days (1):
Final solution with objective 34800: C190 = 0;
z = 34800;
C00 = 0;
C10 = 0;
X10 = 0;
C20 = 1;
X20 = 1;
C30 = 0;
X30 = 7;
C40 = 0;
X40 = 22;
C50 = 3;
X25 = 20;
149
C60 = 0;
C70 = 0;
C80 = 0;
C90 = 2;
C100 = 0;
C110 = 0;
C120 = 0;
C130 = 0;
C140 = 0;
C150 = 0;
C160 = 2;
C170 = 0;
C180 = 3;
X50 = 20;
X60 = 11;
X70 = 27;
X80 = 27;
X90 = 33;
X100 = 20;
X110 = 30;
X120 = 34;
X130 = 25;
X140 = 36;
X150 = 35;
X160 = 41;
X170 = 42;
C110<=1;
C120<=0;//Orig. Critical Path
C130<=1;
C140<=0;
C150<=3;//Orig. Critical Path
C160<=2;
C170<=3;
C180<=3;//Orig. Critical Path
C190<=0;//Orig. Critical Path
150
C80<=0;//Orig. Critical Path
C90<=2;
C110<=1;
C120<=2;//Orig. Critical Path
C130<=1;
C140<=0;
C150<=1;//Orig. Critical Path
C160<=2;
C170<=3;
C180<=2;//Orig. Critical Path
C190<=0;//Orig. Critical Path
151
C00<=0;//Orig. Critical Path
C10<=2;//Orig. Critical Path
C20<=1;
C30<=3;
C40<=2;//Orig. Critical Path
C50<=3;
C60<=0;
C70<=0;
C80<=2;//Orig. Critical Path
C90<=2;
C100<=2;
C110<=1;
C120<=3;//Orig. Critical Path
C130<=1;
C140<=0;
C150<=5;//Orig. Critical Path
C160<=2;
C170<=3;
C180<=5;//Orig. Critical Path
C190<=0;//Orig. Critical Path
152
After defining the variables every case was modeled as shown here:
153
maximize fz;
subject to
{ fz<=((43 -z1)/6); Membership Function for the Completion Time
fz<=((6000-z2)/6000); Membership Function for the Crashing Cost
z1==X170-X10;
z2==1000*C10+ 4000*C20+ 3020*C30+ 8000*C40+ 5000*C50+ 500*C60+ 300*C70+
1000*C80+ 2600*C90+ 1400*C100+ 4600*C110+ 7000*C120+500*C130+500*C140+
1000*C150+ 800*C160+ 3800*C170+ 3000*C180+ 500*C190;
Network Relationships Case Planned.
X10==0;
X110-X100+C130>=3;
X20-X10+C00>=1;
X120-X80+C100>=4;
X30-X20+C10>=6;
X130-X100+C120>=5;
X40-X20+C20>=3;
X140-X90+C110>=3;
X50-X30+C30>=5;
X140-X120+C140>=2;
X60-X30+C40>=4;
X150-X130+C150>=8;
X70-X40+C50>=8;
X150-X110+C160>=7;
X80-X50+C60>=7;
X160-X150+C180>=9;
X90-X60+C70>=2;
X160-X140+C170>=5;
X100-X60+C80>=9;
X170-X160+C190>=1;
X110-X70+C90>4;
X170-X10<=30 ;};
Then, the model was analyzed for the six Time Scenarios keeping the same variables
and network relationships:
1. Maximum Activities Crashing Times (TS- I for Critical Path)
C00<=0;//Orig. Critical Path
C100<=2;
C10<=2;//Orig. Critical Path
C110<=1;
C20<=1;
C120<=0;//Orig. Critical Path
C30<=3;
C130<=1;
C40<=2;//Orig. Critical Path
C140<=0;
C50<=3;
C150<=3;//Orig. Critical Path
C60<=0;
C160<=2;
C70<=0;
C170<=3;
C80<=3;//Orig. Critical Path
C180<=3;//Orig. Critical Path
C90<=2;
C190<=0;//Orig. Critical Path
154
Final Solution (TS- I for Critical Path)
C160 = 0;
fz = 0.5;
C170 = 0;
z1 = 40;
C180 = 1;
z2 = 3000;
X170 = 40;
C190 = 0;
X10 = 0;
C00 = 0;
C10 = 2;
X20 = 1;
C20 = 0;
X30 = 5;
C30 = 0;
X40 = 4;
C40 = 0;
X50 = 21;
C50 = 0;
X60 = 9;
C60 = 0;
X70 = 20;
C70 = 0;
X80 = 28;
C80 = 0;
X90 = 11;
C90 = 0;
X100 = 18;
C100 = 0;
X110 = 23;
C110 = 0;
X120 = 32;
C120 = 0;
X130 = 23;
C130 = 0;
X140 = 34;
C140 = 0;
X150 = 30;
C150 = 1;
X160 = 39;
2. Maximum Activities Crashing Times (TS-II for Critical Path)
C00<=0;//Orig. Critical Path
C100<=2;
C10<=0;//Orig. Critical Path
C110<=1;
C20<=1;
C120<=0;//Orig. Critical Path
C30<=3;
C130<=1;
C40<=0;//Orig. Critical Path
C140<=0;
C50<=3;
C150<=0;//Orig. Critical Path
C60<=0;
C160<=2;
C70<=0;
C170<=3;
C80<=0;//Orig. Critical Path
C180<=0;//Orig. Critical Path
C90<=2;
C190<=0;//Orig. Critical Path
Final Solution (TS- II for Critical Path)
fz = 0;
C170 = 0;
z1 = 43;
C180 = 0;
z2 = 0;
X170 = 43;
C190 = 0;
X10 = 0;
C00 = 0;
C10 = 0;
X20 = 1;
C20 = 0;
X30 = 7;
C30 = 0;
X40 = 4;
C40 = 0;
X50 = 12;
155
C50 = 0;
C60 = 0;
C70 = 0;
C80 = 0;
C90 = 0;
C100 = 0;
C110 = 0;
C120 = 0;
C130 = 0;
C140 = 0;
C150 = 0;
C160 = 0;
X60 = 11;
X70 = 21;
X80 = 19;
X90 = 13;
X100 = 20;
X110 = 26;
X120 = 35;
X130 = 25;
X140 = 37;
X150 = 33;
X160 = 42;
156
C140 = 0;
C150 = 1;
X150 = 30;
X160 = 39;
C110<=1;
C120<=2;//Orig. Critical Path
C130<=1;
C140<=0;
C150<=1;//Orig. Critical Path
C160<=2;
C170<=3;
C180<=2;//Orig. Critical Path
C190<=0;//Orig. Critical Path
C110<=1;
C120<=2;//Orig. Critical Path
157
C30<=3;
C40<=3;//Orig. Critical Path
C50<=3;
C60<=0;
C70<=0;
C80<=5;//Orig. Critical Path
C90<=2;
C100<=2;
C130<=1;
C140<=0;
C150<=1;//Orig. Critical Path
C160<=2;
C170<=3;
C180<=2;//Orig. Critical Path
C190<=0;//Orig. Critical Path
158
Final Solution (TS- VI for Critical Path)
C160 = 0;
fz = 0.5;
C170 = 0;
z1 = 40;
C180 = 1;
z2 = 3000;
X170 = 40;
C190 = 0;
X10 = 0;
C00 = 0;
C10 = 2;
X20 = 1;
C20 = 0;
X30 = 5;
C30 = 0;
X40 = 4;
C40 = 0;
X50 = 21;
C50 = 0;
X60 = 9;
C60 = 0;
X70 = 18;
C70 = 0;
X80 = 28;
C80 = 1;
X90 = 11;
C90 = 0;
X100 = 17;
C100 = 0;
X110 = 23;
C110 = 0;
X120 = 32;
C120 = 0;
X130 = 23;
C130 = 0;
X140 = 34;
C140 = 0;
X150 = 30;
C150 = 0;
X160 = 39;
maximize fz;
subject to
{
fz<=((43 -z1)/6); Membership Function for the Completion Time
fz<=((16401-z2)/16401); Membership Function for the Crashing Cost
z1==X170-X10;
z2==1000*C10+ 4000*C20+ 3020*C30+ 8000*C40+ 5000*C50+ 500*C60+ 300*C70+
1000*C80+ 2600*C90+ 1400*C100+ 4600*C110+ 7000*C120+500*C130+500*C140+
1000*C150+ 800*C160+ 3800*C170+ 3000*C180+ 500*C190;
159
Network Relationships Case One.
X10==0;
X20-X10+C00>=1;
X30-X20+C10>=6;
X40-X20+C20>=3;
X50-X30+C30>=5;
X50>=20; //*ACTIVITY F*
X65-X60>=0;
X65>=20; //*ACTIVITY G*
X60-X30+C40>=4;
X70-X40+C50>=8;
X70>=20; //*ACTIVITY I*
X80-X50+C60>=7;
X90-X65+C70>=2;
X100-X60+C80>=9;
X100>=20; //*ACTIVITY L AND M*
X110-X70+C90>4;
X110-X100+C130>=3;
X120-X80+C100>=4;
X130-X100+C120>=5;
X140-X90+C110>=3;
X140-X120+C140>=2;
X150-X130+C150>=8;
X150-X110+C160>=7;
X160-X150+C180>=9;
X160-X140+C170>=5;
X170-X160+C190>=1 ;};
Then, the model was analyzed for the six Time Scenarios keeping the same variables
and network relationships:
160
C70<=0;
C80<=3;//Orig. Critical Path
C90<=2;
C170<=3;
C180<=3;//Orig. Critical Path
C190<=0;//Orig. Critical Path
161
Final Solution (TS- II for Critical Path)
fz = 0;
C170 = 0;
z1 = 43;
C180 = 0;
z2 = 0;
X170 = 43;
C190 = 0;
X10 = 0;
C00 = 0;
C10 = 0;
X20 = 1;
C20 = 0;
X30 = 7;
C30 = 0;
X40 = 4;
C40 = 0;
X50 = 20;
C50 = 0;
X65 = 20;
C60 = 0;
X60 = 11;
C70 = 0;
X70 = 20;
C80 = 0;
X80 = 27;
C90 = 0;
X90 = 22;
C100 = 0;
X100 = 20;
C110 = 0;
X110 = 25;
C120 = 0;
X120 = 35;
C130 = 0;
X130 = 25;
C140 = 0;
X140 = 37;
C150 = 0;
X150 = 33;
C160 = 0;
X160 = 42;
8. Maximum Activities Crashing Times (TS-III for Critical Path)
C00<=0;//Orig. Critical Path
C100<=2;
C10<=4;//Orig. Critical Path
C110<=1;
C20<=1;
C120<=3;//Orig. Critical Path
C30<=3;
C130<=1;
C40<=3;//Orig. Critical Path
C140<=0;
C50<=3;
C150<=5;//Orig. Critical Path
C60<=0;
C160<=2;
C70<=0;
C170<=3;
C80<=5;//Orig. Critical Path
C180<=5;//Orig. Critical Path
C90<=2;
C190<=0;//Orig. Critical Path
Final Solution (TS- III for Critical Path)
fz = 0.53661;
C170 = 0;
z1 = 39;
C180 = 1;
z2 = 7600;
X170 = 39;
C190 = 0;
X10 = 0;
C00 = 0;
C10 = 0;
X20 = 1;
C20 = 0;
X30 = 7;
C30 = 0;
X40 = 4;
162
C40 = 0;
C50 = 0;
C60 = 0;
C70 = 0;
C80 = 0;
C90 = 0;
C100 = 0;
C110 = 0;
C120 = 0;
C130 = 0;
C140 = 0;
C150 = 3;
C160 = 2;
X50 = 20;
X65 = 20;
X60 = 11;
X70 = 20;
X80 = 27;
X90 = 22;
X100 = 20;
X110 = 25;
X120 = 31;
X130 = 25;
X140 = 33;
X150 = 30;
X160 = 38;
C110<=1;
C120<=2;//Orig. Critical Path
C130<=1;
C140<=0;
C150<=1;//Orig. Critical Path
C160<=2;
C170<=3;
C180<=2;//Orig. Critical Path
C190<=0;//Orig. Critical Path
163
C110 = 0;
C120 = 0;
C130 = 0;
C140 = 0;
C150 = 1;
C160 = 0;
X110 = 25;
X120 = 32;
X130 = 25;
X140 = 34;
X150 = 32;
X160 = 39;
C110<=1;
C120<=2;//Orig. Critical Path
C130<=1;
C140<=0;
C150<=1;//Orig. Critical Path
C160<=2;
C170<=3;
C180<=2;//Orig. Critical Path
C190<=0;//Orig. Critical Path
164
11. Maximum Activities Crashing Times (TS-VI for Critical Path)
C00<=0;//Orig. Critical Path
C100<=2;
C10<=2;//Orig. Critical Path
C110<=1;
C20<=1;
C120<=3;//Orig. Critical Path
C30<=3;
C130<=1;
C40<=2;//Orig. Critical Path
C140<=0;
C50<=3;
C150<=5;//Orig. Critical Path
C60<=0;
C160<=2;
C70<=0;
C170<=3;
C80<=2;//Orig. Critical Path
C180<=5;//Orig. Critical Path
C90<=2;
C190<=0;//Orig. Critical Path
Final Solution (TS- VI for Critical Path)
fz = 0.53661;
C170 = 0;
z1 = 39;
C180 = 1;
z2 = 7600;
X170 = 39;
C190 = 0;
X10 = 0;
C00 = 0;
C10 = 0;
X20 = 1;
C20 = 0;
X30 = 7;
C30 = 0;
X40 = 4;
C40 = 0;
X50 = 20;
C50 = 0;
X65 = 20;
C60 = 0;
X60 = 11;
C70 = 0;
X70 = 20;
C80 = 0;
X80 = 27;
C90 = 0;
X90 = 22;
C100 = 0;
X100 = 20;
C110 = 0;
X110 = 25;
C120 = 0;
X120 = 31;
C130 = 0;
X130 = 25;
C140 = 0;
X140 = 33;
C150 = 3;
X150 = 30;
C160 = 2;
X160 = 38;
165
maximize fz;
subject to
{
fz<=((43 -z1)/6); Membership Function for the Completion Time
fz<=((31460-z2)/31460); Membership Function for the Crashing Cost
z1==X170-X10;
z2==1000*C10+ 4000*C20+ 3020*C30+ 8000*C40+ 5000*C50+ 500*C60+ 300*C70+
1000*C80+ 2600*C90+ 1400*C100+ 4600*C110+ 7000*C120+500*C130+500*C140+
1000*C150+ 800*C160+ 3800*C170+ 3000*C180+ 500*C190;
X40-X20+C20>=3;
X50-X35+C30>=5;
X60-X30+C40>=4;
X65-X60>=0;
X65>=20;//*ACTIVITY G*
X70-X40+C50>=8;
X80-X50+C60>=7;
X90-X65+C70>=2;
X100-X60+C80>=9;
X100>=20;//*ACTIVITY L AND M*
X110-X70+C90>4;
X110-X100+C130>=3;
X110>=20;//*ACTIVITY P*
X120-X80+C100>=4;
X130-X100+C120>=5;
X140-X90+C110>=3;
X140-X120+C140>=2;
166
X150-X130+C150>=8;
X150-X110+C160>=7;
X160-X150+C180>=9;
X160-X140+C170>=5;
X170-X160+C190>=1;};
Then, the model was analyzed for the six Time Scenarios keeping the same variables
and network relationships:
1. Maximum Activities Crashing Times (TS- I for Critical Path)
C00<=0;//Orig. Critical Path
C100<=2;
C10<=2;//Orig. Critical Path
C110<=1;
C20<=1;
C120<=0;//Orig. Critical Path
C30<=3;
C130<=1;
C40<=2;//Orig. Critical Path
C140<=0;
C50<=3;
C150<=3;//Orig. Critical Path
C60<=0;
C160<=2;
C70<=0;
C170<=3;
C80<=3;//Orig. Critical Path
C180<=3;//Orig. Critical Path
C90<=2;
C190<=0;//Orig. Critical Path
167
12. Maximum Activities Crashing Times (TS-II for Critical Path)
C00<=0;//Orig. Critical Path
C100<=2;
C10<=0;//Orig. Critical Path
C110<=1;
C20<=1;
C120<=0;//Orig. Critical Path
C30<=3;
C130<=1;
C40<=0;//Orig. Critical Path
C140<=0;
C50<=3;
C150<=0;//Orig. Critical Path
C60<=0;
C160<=2;
C70<=0;
C170<=3;
C80<=0;//Orig. Critical Path
C180<=0;//Orig. Critical Path
C90<=2;
C190<=0;//Orig. Critical Path
Final Solution (TS- II for Critical Path)
C170 = 0;
fz = 0;
C180 = 0;
z1 = 43;
C190 = 0;
z2 = 12360;
X170 = 43;
C00 = 0;
X10 = 0;
X20 = 1;
C10 = 0;
X30 = 7;
C20 = 0;
X35 = 20;
C30 = 3;
X40 = 4;
C40 = 0;
X50 = 22;
C50 = 0;
X60 = 11;
C60 = 0;
X65 = 20;
C70 = 0;
X70 = 21;
C80 = 0;
X80 = 29;
C90 = 0;
X90 = 22;
C100 = 2;
X100 = 20;
C110 = 0;
X110 = 26;
C120 = 0;
X120 = 35;
C130 = 1;
X130 = 25;
C140 = 0;
X140 = 37;
C150 = 0;
X150 = 33;
C160 = 0;
X160 = 42;
13. Maximum Activities Crashing Times (TS-III for Critical Path)
C00<=0;//Orig. Critical Path
C100<=2;
C10<=4;//Orig. Critical Path
C110<=1;
C20<=1;
C120<=3;//Orig. Critical Path
C30<=3;
C130<=1;
C40<=3;//Orig. Critical Path
C140<=0;
C50<=3;
C150<=5;//Orig. Critical Path
C60<=0;
C160<=2;
168
C70<=0;
C80<=5;//Orig. Critical Path
C90<=2;
C170<=3;
C180<=5;//Orig. Critical Path
C190<=0;//Orig. Critical Path
C110<=1;
C120<=2;//Orig. Critical Path
C130<=1;
C140<=0;
C150<=1;//Orig. Critical Path
C160<=2;
C170<=3;
C180<=2;//Orig. Critical Path
C190<=0;//Orig. Critical Path
169
Final Solution (TS- IV for Critical Path)
C170 = 0;
fz = 0.4965;
C180 = 2;
z1 = 40;
C190 = 0;
z2 = 15840;
X170 = 40;
C00 = 0;
X10 = 0;
X20 = 1;
C10 = 0;
X30 = 7;
C20 = 0;
X35 = 20;
C30 = 2;
X40 = 4;
C40 = 0;
X50 = 23;
C50 = 0;
X60 = 11;
C60 = 0;
X65 = 20;
C70 = 0;
X70 = 18;
C80 = 0;
X80 = 30;
C90 = 0;
X90 = 22;
C100 = 2;
X100 = 20;
C110 = 0;
X110 = 23;
C120 = 0;
X120 = 32;
C130 = 0;
X130 = 25;
C140 = 0;
X140 = 34;
C150 = 1;
X150 = 32;
C160 = 0;
X160 = 39;
15. Maximum Activities Crashing Times (TS-V for Critical Path)
C00<=0;//Orig. Critical Path
C10<=4;//Orig. Critical Path
C20<=1;
C110<=1;
C30<=3;
C120<=2;//Orig. Critical Path
C40<=3;//Orig. Critical Path
C130<=1;
C50<=3;
C140<=0;
C60<=0;
C150<=1;//Orig. Critical Path
C70<=0;
C160<=2;
C80<=5;//Orig. Critical Path
C170<=3;
C90<=2;
C180<=2;//Orig. Critical Path
C100<=2;
C190<=0;//Orig. Critical Path
Final Solution (TS- V for Critical Path)
C170 = 0;
fz = 0.4965;
C180 = 2;
z1 = 40;
C190 = 0;
z2 = 15840;
X170 = 40;
C00 = 0;
X10 = 0;
X20 = 1;
C10 = 0;
X30 = 7;
C20 = 0;
X35 = 20;
170
C30 = 2;
C40 = 0;
C50 = 0;
C60 = 0;
C70 = 0;
C80 = 0;
C90 = 0;
C100 = 2;
C110 = 0;
C120 = 0;
C130 = 0;
C140 = 0;
C150 = 1;
C160 = 0;
X40 = 4;
X50 = 23;
X60 = 11;
X65 = 20;
X70 = 18;
X80 = 30;
X90 = 22;
X100 = 20;
X110 = 23;
X120 = 32;
X130 = 25;
X140 = 34;
X150 = 32;
X160 = 39;
171
C100 = 2;
C110 = 0;
C120 = 0;
C130 = 0;
C140 = 0;
C150 = 3;
C160 = 0;
X100 = 20;
X110 = 23;
X120 = 32;
X130 = 25;
X140 = 34;
X150 = 30;
X160 = 39;