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

net/publication/331989177

OPTIMIZATION OF RESOURCE ALLOCATION AND LEVELING USING GENETIC

ALGORITHMS

Article · August 2011

CITATIONS READS

2 70

1 author:

Ameer Ahmed
University Of Kufa
5 PUBLICATIONS 2 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Knowledge exchange, View project

All content following this page was uploaded by Ameer Ahmed on 26 March 2019.

The user has requested enhancement of the downloaded file.

 Number 4 Volume 17 August 2011 Journal of Engineering

OPTIMIZATION OF RESOURCE ALLOCATION AND

LEVELING USING GENETIC ALGORITHMS
Assist. Prof.: Dr. Sawsan Rasheed Mohammed Ameer Ahmed Abdul Ameer
University of Baghdad University of Baghdad
College of Engineering College of Engineering
Department of Civil Engineering Department of Civil Engineering

ABSTRACT
Resource allocation and leveling are of the top challenges in project management, due to the
complexity of projects. This research aims to develop an optimization model for resource
smoothing, so that.
The proposed model is formulated using C++ program for resource smoothing. The project
management software MS-Projects is adopted hereto perform resource leveling to facilitate
achieving the optimal solution.
The proposed model utilizes a system that depends on Genetic Algorithms (GAs) procedure built
in C++ program to find the optimum solution.
This research reach concludes that it is possible to smooth resources using Genetic Algorithms
program and compares then with MS-Project when the GA results are better than MS-Project.
Three case studies have been applied in this research and the application results come identical
with research objectives, to form the conclusion.
Then comes the recommendations regarding adopting and using the research results in
construction planning and project management. Further suggestions related to the research subject
are proposed for future works.

‫اﻟﺨﻼﺻﺔ‬
.‫ان ﺗﻌﻴﻴﻦ وﺗﺴﻮﻳﺔ اﻟﻤﻮارد هﻲ ﻣﻦ اآﺜﺮ اﻟﺘﺤﺪﻳﺎت ﻓﻲ ادارة اﻟﻤﺸﺮوع ﺑﺴﺒﺐ ﺗﻌﻘﻴﺪ اﻟﻤﺸﺎرﻳﻊ‬
‫ ﻟﺘﺴﻮﻳﺔ اﻟﻤﻮارد‬C++ ‫ ﻟﻘﺪ ﺗﻢ ﺑﻨﺎء اﻟﻨﻤﻮذج اﻟﻤﻘﺘﺮح ﺑﺎﺳﺘﺨﺪام ﺑﺮﻧﺎﻣﺞ‬.‫ﻳﻬﺪف هﺬا اﻟﺒﺤﺚ اﻟﻰ ﺑﻨﺎء ﻧﻤﻮذج اﻣﺜﻞ ﻟﺘﺴﻮﻳﺔ اﻟﻤﻮارد‬
‫ ﻳﻌﺘﻤﺪ اﻟﻨﻤﻮذج‬.‫( ﻻﺟﺮاء ﺗﺴﻮﻳﺔ اﻟﻤﻮارد ﻟﺘﺴﻬﻴﻞ اﻟﻮﺻﻮل اﻟﻰ اﻟﺤﻞ اﻻﻣﺜﻞ‬MS-Project) ‫وآﻤﺎ ﺗﻢ اﻋﺘﻤﺎد ﺑﺮﻧﺎﻣﺞ ﻻدارة اﻟﻤﺸﺎرﻳﻊ‬
.‫( ﻟﻠﺒﺤﺚ ﻋﻦ اﻟﺤﻞ اﻻﻣﺜﻞ‬C++) ‫اﻟﻤﻘﺘﺮح ﻋﻠﻰ اﺳﻠﻮب اﻟﺨﻮارﻣﻴﺎت اﻟﺠﻴﻨﻴﺔ ﺣﻴﺚ ﺗﻢ ﺑﻨﺎؤﻩ ﺑﺎﺳﺘﺨﺪام ﺑﺮﻧﺎﻣﺞ‬
‫ وﻟﻘﺪ وﺟﺪ ان‬MS-Project ‫ﺗﻮﺻﻞ اﻟﺒﺤﺚ اﻟﻰ اﻣﻜﺎﻧﻴﺔ ﺗﺴﻮﻳﺔ اﻟﻤﻮارد ﺑﺎﺳﺘﺨﺪام اﻟﺨﻮارزﻣﻴﺎت اﻟﺠﻴﻨﻴﺔ وﻣﻘﺎرﻧﺘﻬﺎ ﻣﻊ ﺑﺮﻧﺎﻣﺞ‬
‫ وﻃﺒﻘﺖ ﻧﺘﺎﺋﺞ اﻟﺒﺤﺚ ﻋﻠﻰ ﺛﻼث ﺣﺎﻻت درﺳﺖ وآﺎﻧﺖ ﻧﺘﺎﺋﺞ اﻟﺘﻄﺒﻴﻖ‬MS-Project‫ﻧﺘﺎﺋﺞ اﻟﺨﻮارزﻣﻴﺎت اﻟﺠﻴﻨﻴﺔ اﻓﻀﻞ ﻣﻦ ﻧﺘﺎﺋﺞ‬
‫ وﺗﻢ اﻟﺘﻮﺻﻞ اﻟﻰ ﻋﺪد ﻣﻦ اﻻﺳﺘﻨﺘﺎﺟﺎت آﻤﺎ وﺿﻌﺖ ﻋﺪد ﻣﻦ اﻟﺘﻮﺻﻴﺎت واﻟﻤﻘﺘﺮﺣﺎت ﺑﺸﺎن ﺗﺒﻨﻲ‬.‫ﻣﻄﺎﺑﻘﺔ ﻟﻤﺎ هﺪف اﻟﻴﻪ اﻟﺒﺤﺚ‬
.‫ﻧﺘﺎﺋﺞ اﻟﺒﺤﺚ ﻓﻲ ﺣﻘﻞ اﻟﺘﺨﻄﻴﻂ وادارة اﻟﻤﺸﺎرﻳﻊ آﻤﺎ وﺿﻌﺖ ﻋﺪد ﻣﻦ اﻟﻤﻘﺘﺮﺣﺎت ﻟﺒﺤﻮث ﻻﺣﻘﺔ ذات ﺻﻠﺔ ﺑﻤﻮﺿﻮع اﻟﺒﺤﺚ‬

929
 Dr. Sawsan Rasheed Mohammed OPTIMIZATION OF RESOURCE
Ameer Ahmed Abdul Ameer ALLOCATION AND LEVELING USING
GENETIC ALGORITHMS

schedule. The schedule development process

must be iterated (along with processes that
provide inputs, especially duration estimating
INTRODUCTION and cost estimating) prior to the
determination of the project schedule.
Resource allocation and leveling have been [Project Management Institute 1996]
dealt with as two distinct subproblems solved The basic objectives of the scheduling
mainly using heuristic procedures that process are as follows:[Callahan 1992]
cannot guarantee optimum solutions. In this a. To find out how long the total project
research, improvements are proposed to duration is. Timely completion of the
resource smoothing and leveling heuristics, project is particularly important when
and the Genetic Algorithms (GAs) technique fails to complete within the time
is used to search for near-optimum solution. required by contract that carries a
In dealing with project resources, two main financial penalty or liquidated
types of techniques have been used: resource damages;
allocation and resource leveling. Resource b. evaluating the early and late times at
allocation (sometimes referred to as which activities start and finish.
Constrained-Resource Scheduling) attempts c. Identifying the group of critical
to schedule the project tasks so that a limited activities so that special care is taken
number of resources can be efficiently to make sure they are not delayed;
utilized while keeping the unavoidable d. since the construction environment is
extension of the project to a minimum. always exposed to constraints and
Resource leveling (often referred to as changes, it is important to be able to
resource smoothing), on the other hand, evaluate the implications of changes
attempts to reduce the sharp variations in start and/or finish times of
among the peaks and valleys in the resource activities on the overall project
demand histogram while maintaining the duration;
original project duration [Mselhi 1993]. e. the follow up execution of the project;
Genetic Algorithms (GAs) are search f. to monitor the usage of Resources;
procedures that combine an artificial survival g. to Expect the stoppage in the
of the fittest strategy with genetic operators execution and reasons behind;
abstracted from nature [Michell 1998]. GAs h. to prepare the financial requirements
are optimization search procedures inspired and cost control of work; and
by the biological system improved fitness to control the sub-contractors work and
through evolution. GAs employ a random yet their interference.
directed search for locating the globally
optimal solution. [Goldberq 1989] SCHEDULING TECHNIQUES
There are many types of schedules which
SCHEDULING DEFINITION & can be used for many construction project.
OBJECTIVES The choice of which type of planning to be
Scheduling deals with time order in which used depends on the characteristics of each
project activities are to take place, and also project.
the manpower, material machinery and Several tools and techniques assist in the
money (the 4m’s) required at every stage of planning development process.
production that should be shown in the - Gantt Chart (Bar Chart)
scheduling. [Senupta 1995] - Network Analysis
Schedule development means determining - Program Evaluation and Review
the start and finish dates for project activities. Technique PERT
If the start and finish dates are not realistic, - Line of Balance
the project is unlikely to be finished as

930
 Number 4 Volume 17 August 2011 Journal of Engineering
OPTIMIZATION SCHEDULING Mx = ∑ [1*Resource Demandi)*1/2Resource
METHODS (MODELS) Demand1]
Modeling can be defined as the process of Mx = 1/2 ∑ (Resource Demandi)2 (1)
producing a model; model is a representation Where n is the working day number of the
of the construction and working of some projects finish date. Equation – to be a
system of interest. [Maria 1997] minimum becomes:
Modeling is one of the most powerful tools Mx = ∑ (Resource Demandi)2 (2)
that have ever been employed in various
research disciplines. Often it’s the typical GENETIC ALGORITHMS DESCRIPTION
way and some times the only way to conduct Genetic Algorithms are search algorithms
experiments on a wide range of systems of based on the mechanics of natural selection
various types. and natural genetics. They combine survival
- Linear Programming Model of the fittest among string structures with a
- Simulation Model structured yet randomized information
- Monte Carlo Simulation exchange to form a search algorithm with
- Genetic Algorithms Techniques some of the innovative flair of human search.
In every generation, a new set of artificial
RESOURCE SMOOTHING creatures (strings) is created using bits and
Resource smoothing attempts to reduce pieces of the fittest of the old; an occasional
peak requirements and smooth out period-to- new part is tried for good measure. While
period fluctuations in resource assignment randomized, genetic algorithms are no simple
without changing project duration. The random walk. They efficiently exploit
objective of resource leveling procedures is historical information to speculate on new
to schedule project activities so that the search points with expected improved
project duration does not exceed a specific performance.
limit and the variation in the projects demand The central theme of research on genetic
for a resource from one time period to algorithms has been robustness, the balance
another is held to a minimum. [Harris 1990] between efficiency and efficacy necessary for
The project managers' objective to hire the survival in many different environments. The
minimum number of resources, to reduce implications of robustness for artificial
resource fluctuation and to ensure better systems are manifold. If artificial systems
utilization of resources. Typical situations can be made more robust, costly redesigns
include full utilization of a rented piece of can be reduced or eliminated. If the higher
equipment that needs to be returned early, levels of adaptation can be achieved, existing
also reducing the number of skilled workers systems can perform their functions longer
who need to be hired for the job. Project and better. Designers of artificial systems –
managers have desired resource profiles that both Software and hardware, whether
they try to get their resource profiles to Engineering Systems, Computer systems, or
match. Business systems – can only marvel at the
A well-known heuristic algorithm is the robustness, the efficiency, and the flexibility
minimum moment algorithm that assumes of biological systems. Features for self-
limited project duration and unlimited repair, self-guidance, and reproduction are
resources [Hiyassat 2001]. The objective in the rule in Biological systems, whereas they
this algorithm is to minimize daily barely exist in most sophisticated artificial
fluctuations in resource use while keeping the systems.
total project duration unchanged. As a proxy Genetic Algorithms are now finding more
to this objective the Algorithm minimizes the widespread application in Business,
moment of the resource histogram around the Scientific, and Engineering circles. The
horizontal axis. [Harris 1978] reasons behind the growing numbers of
The moment MX is calculated by summing applications are clear. These algorithms are
the daily moments as follows: [Haider 1999] computationally simple yet powerful in their
improvement. Furthermore, they are not

931
 Dr. Sawsan Rasheed Mohammed OPTIMIZATION OF RESOURCE
Ameer Ahmed Abdul Ameer ALLOCATION AND LEVELING USING
GENETIC ALGORITHMS

fundamentally limited by restrictive Usually, this solutions replace unfit

assumptions about the search space solutions), until criterion is met (e.g. one
(assumptions concerning continuity, solution becomes satisfactory). At the end of
existence of derivatives, unimodality, and the process, the member of the population
other matters).[Baker 1985] with the best performance becomes the
optimum solution [Tarek 1999]. Figure (1)
THE WORK OF GAS illustrates the basic cycle of genetic
GAs work with a family of solutions, algorithm operations.
known as the "current population" from
which the "next generation" of solutions is STEPS OF GENETIC ALGORITHM
obtained, better solutions from one SYSTEM
generation to the next are progressively The following points illustrate the principle
obtained [Harmanani 2001]. GAs procedure work of genetic algorithm:
begins by generating an initial collection
(referred to as population) of random Encoding the Application of GA
solutions that are encoded in the form of GAs require a representation scheme to
strings called Chromosomes. encode feasible solutions to the Optimization
Each individual Chromosome represents problem. Each Chromosome represents one
one solution that is better or worse than member, i.e., one solution, which is better or
others in the population. The fitness of each worse than other members in a population.
solution is determined by evaluating its A Chromosome represents a sequence of
performance with respect to an objective genes that require Optimization.
function. To stimulated the natural survival There are two basic chromosome formats
of the fittest process, best Chromosomes in GAs,
(potential solutions) exchange information to a. Binary Coding.
produce offspring that are evaluated and can b. Ordinary Coding.
replace less fir members in the population.

New Population Initial

Mutation Selection

Crossove

Figure 1 Basic Cycle of GA Operations [Tarek 1999]

932
 Number 4 Volume 17 August 2011 Journal of Engineering

1 2 3 4 5 6 P

3 10 22 14 5 9 4

Legend:
X Gene position

Y Gene Value

Figure 2 Chromosome Structure [Lue 1999]

Generating an Initial Population of two parent Chromosomes is randomly

Chromosomes selected in a manner such that its probability
The process of genetic algorithms starts of being selected is proportional to its
with a randomly created first generation of relative merit.
population. Every individual in a generation
represents one solution and consists of one Crossover
Chromosome with a number of genes. Crossover is the process of combing the
[Haider 1999] chromosomes of two potentially good
solutions to form two offsprings [Kevin
Deciding the Chromosome Evolution 2003]. Crossover is performed by randomly
Criterion (Objective Function) selecting two members from the Population
The performance of the strings is often and exchanging their chromosomal
called fitness, and the fitness of each string information.
(Chromosome) in the population is evaluated Figure (3) illustrates that two
with respect to an Objective Function. Chromosomes (parents 1 and 2) are randomly
selected and broken at a random point (at gen
Selection 5), and after the exchange at genetic material
A pair of parent Chromosome from the two new Chromosomes (Offspring 1 and 2)
current population is selected. Each of the are generated.

933
 Dr. Sawsan Rasheed Mohammed OPTIMIZATION OF RESOURCE
Ameer Ahmed Abdul Ameer ALLOCATION AND LEVELING USING
GENETIC ALGORITHMS

A1 A2 A3 A4 A5 A6 A7 A8 A9 A10
Parent 1

B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 Parent 2

A1 A2 A3 A4 A5 B6 B7 B8 B9 B10 Offspring 1

B1 B2 B3 B4 B5 A6 A7 A8 A9 A10 Offspring 2

Figure 3 Crossover [Kevin 2003]

Mutation PARAMETERS OF GAS

Mutation is a rare process that resembles To implement GAs, its Parameters
the process of a sudden generation of an odd must be set having: [Michell 1998]
offspring that turns out to be a genius. This a. Population Size: is the number of
can be done by randomly selecting one population to be generated randomly.
Chromosome from the population and then Population size is an important factor
randomly changing some of its information. that affects the solution and
The benefit of the mutation process is that it processing time it takes. Large
can break any stagnation in the evolutionary Population size (in the order of
process and avoid sub optimal solutions. hundreds) increases the likelihood of
Figure (4) displays some forms of mutation obtaining a Global Optimum solution,
[Goldberg 1989]. In this figure characters however, it substantially increases
have mutation operation. processing time.
b. Chromosome Length: A solution is
represented as a linear string called a
1100110101 Chromosome whose length varies
100111101 with each application.
c. Crossover Probability (Pc): A rate to
ABDACE perform Crossover between any pairs
GBDANE of two selected Chromosomes
(strings). Pc is real number from 0 to
4 5 1 8 11 1 7 3 1 and the mostly used of Crossover
4 20 1 8 2 1 7 3 probability are from 0.3 to 0.95.
[Tarek 2000]
Figure 4 Some Examples of Mutation in d. Mutation Probability (Pm): A rate to
GA [Goldberg 1989] randomly alter one or more genes of a
selected string (Pm) is real number

934
 Number 4 Volume 17 August 2011 Journal of Engineering
e. from 0 to 0.1 and the mostly used The aim of this project is to get knowledge
values of mutation probability are the optimum resources to get the minimum
from 0.001 to 0.1. [Tarek 2000] ∑R2 for each project and the early start of
Stopping Criteria: There are two types of each activity which only has TF.
stopping criteria included in GAs. The first This work does not happen by isolated the
one is in state of meaning the value of activity from the other one because the
objective function, and the second is the activities is interfusion, therefore ∑R2 is
number of Offspring generations. The two cumulative.
types above can be used as a termination Any programmer sees this idea thought to
criterion for the GA process. [Tarek 1999] move the first activity and calculate ∑R2 and
then the second activity, this way is correct
BENEFITS OF GAS IN CONSTRUCTION and it is long. But in the event of using
GAs are particularly suited for Genetic Algorithm we will adapt this concept
Optimization problems in Construction and make the movement which depend on
Scheduling because: correct programming grammars used in our
a. They do not experience combinatorial program which is called cumulative test
explosion: [Que 2002] GAs perform procedure and relation movement.
well on problems whose complexity The Activities Features are as Follows:
increases exponentially with the a. Each activity has duration.
number of input Parameters. Such b. Each activity has early start that is
problems are called NP-Complete. important.
The Resource Leveling problem c. Each activity has total float that is
could be defined as a (NP-Complete) more important. Then we will make
problem. That is computational time the permutation and calculate the
which grows exponentially as the size fitness function (∑R2).
of the problem increases. [Son 1999] d. It’s noted there are critical activities
b. They are Robust: GA has the feature without movement and uncompatible
of robustness because of its ability to with the permutation because of the
evaluate many possible solutions total float is equal zero.
simultaneously and use the Every time in our project we are keep
Chromosome fitness to direct the generated the offsprings to get the best
search. cumulative resources of each day of project
duration. The visual C++ program illustrates
COMPUTER IMPLEMENTATION the optimum solution.
GAs procedure can be implemented on a
Visual C++ Program. APPLICABILITY OF GENETIC
C++ is more faster than any Programming ALGORITHMS TO RESOURCE
languages to reach the Optimal solution LEVELING
because it is deeply treat with computer and The minimum resource moment algorithm
upon of this feature it is object oriented. was improved using both Mx and My
Always C++ depends on specifying the resource moments. The moment Mx
object to get a specific class (or specific (Moment of the resource histogram about the
character) then create other target (object) x-axis) represents the resource fluctuation
from that specific character which have the and the moment My (Moment of the resource
same characters of that class. Each class histogram about y-axis) represents the
should have some functions, each of it do resource utilization. The minimum value of
special work depend on that variables. [De these two combined moments serves as a
Jon 1980] good indicator of efficiently utilized
The objective of our program of C++ is resources where fluctuations from period to
getting the minimum of MX, are equal ∑R2 another are avoided. The random activity
whereas R is the resource of each activity. priorities and the combined moments
The target of that is getting specific project.

935
 Dr. Sawsan Rasheed Mohammed OPTIMIZATION OF RESOURCE
Ameer Ahmed Abdul Ameer ALLOCATION AND LEVELING USING
GENETIC ALGORITHMS

approach form the basis of the optimization resource fluctuations and resource utilization
process. [Tarek 1999] moments. This process is executed under
The detailed GA Procedure is outlined in unlimited resources Figure (7) illustrates
Figure (5) process chart of resource smoothing.
The optimization parameters associated
PROPOSED RESOURCE SMOOTHING with this process are outlined as follows:
SYSTEM a. Objective function which will
The proposed model comprises two main minimize the fluctuation and/or
sections: processing project data in MS and utilization moments;
optimizing resource leveling using genetic b. optimization variables which consist
algorithms program. of activity TF values, which range
Figure (6) illustrates process chart accepts integers between zero and the
diagram of the proposed model. maximum TF value obtained from
Microsoft project before applying
MINIMIZING RESOURCE MOMENTS resource smoothing;
The minimizing resource moments c. optimization constraints which state
objective represents the optimization of the that the project duration should be
resource smoothing procedure. Which equal to the project deadline.
involves the process of optimally minimizing

936
 Number 4 Volume 17 August 2011 Journal of Engineering

d. START

1
User input of evaluation criteria (wts)
for project duration & moments),
population size, & No. of offspring

2
Initialize Scheduling Software

3
Save Initial Project Duration and moment

4
Generate population of random chromosome;
each chromosome has a value for the priority
of each project task

5
For each chromosome in the population:
ƒ Input its data to the software
ƒ Get project Duration & moments; &
ƒ Calculate the Chromosome's fitness

6
Evaluate the relative merit of each END
END Chromosome in the population and
determine the worst Chromosome
yes 12
Yes
14 7
Pick 2 Chromosomes at random with Is this the last
Is this the last No probability of being picked proportional
No Offspring
Offspring to their relative merits

8 11
13
Perform Crossover or mutation to
generate an Offspring Replace the worst
Discard the 9 chromosome with the
Offspring Offspring
Determine the fitness of the Offspring
chromosome, as in step 5

10
No Yes

Is Offspring better than

the worst Chromosome
in the population?

Figure 5 Genetic Algorithm Procedure [Tarek 1999]

937
Dr. Sawsan Rasheed Mohammed OPTIMIZATION OF RESOURCE
Ameer Ahmed Abdul Ameer ALLOCATION AND LEVELING USING
GENETIC ALGORITHMS

Start MS project

Input resources information

manually
ƒ Labour sheet
ƒ Equipment sheet
ƒ Materials sheet

Performing Time Schedule sheet

Performing Resources allocation Histogram

of limited time

Performing Resources smoothing Histogram

Setting up genetic algorithms program

(visual stiduo C++ but in time schedule
sheet and resources information sheet

Performing activities ES, Project Duration

and resources smoothing sheet

Input activities information get from MS project

manually in GA program

Performing resources smoothing of limited time

Compare result with Mx to the two histogram

Figure 6 Process Chart Diagram of the Proposed Model [researcher]

938
 Number 4 Volume 17 August 2011 Journal of Engineering

Time schedule sheet

from Ms project

Bar chart scheduling Resource planning

computation screen screen
- Project duration
- Activity duration
- Activity resource
- Activity ES
- Activity TF

Setting up optimization program

C++

Selecting objective function Defining optimization options

- Minimum fluctuation moment - Population size
- Minimum utilization moment or - Stopping conditions
- Minimum total moment

Defining constraints
- Project duration dead line
- Critical activities

Running the optimization process

Evaluating each possible solution with

respect to fluctuation and/or utilization
moment and constraints

Selecting the best possible solution

Performing activities with their ES, TF, and

Cumulative resource on unit time

Figure 7 Optimization Process Chart of the

Resource smoothing [researcher]

939
 Dr. Sawsan Rasheed Mohammed OPTIMIZATION OF RESOURCE
Ameer Ahmed Abdul Ameer ALLOCATION AND LEVELING USING
GENETIC ALGORITHMS

OPENING AND USING C++

when C++ screen appears follows these steps:
1. File
2. Open
3. Write the file name contains the code
4. Press F7
5. Press Yes
6. Press Ctrl+F5
After the above steps GA input screen which we will input the number of activity, duration of
project, for each activity input we will input:
a. Duration
b. Early start (ES)
c. Total Float (TF)
d. Resources (R)
e. Is it critical or not
f. When Click Enter the GAs is operate and give us Resource of each day and the minimum
of (Mx)
These steps illustrate in Figure (8) that input the data of one activity

Figure 8 GA Input Screen

940
 Number 4 Volume 17 August 2011 Journal of Engineering
MODEL APPLICATION

Case Study1 of 7 Activity (Description and Data)

A summary of the case study data is demonstrated in Table (1)

Table 1 Case Study Data of 7 activity

Activity Duration Labour (R) Predecessors

A 2 2 -
B 6 4 A
C 3 3 A
D 1 1 B
E 6 3 B
F 3 3 C,D
G 2 2 E,F

From MS Project we get figure (9) and (10) network and a bar chart representation
show a hypothetical case study of 7 activities respectively.
in the form of activity on node (AON)

B E

A D G

C F

Figure 9 AON Network of 7 Activity

Figure 10 Bar Chart of 7 Activity Project

941
 Dr. Sawsan Rasheed Mohammed OPTIMIZATION OF RESOURCE
Ameer Ahmed Abdul Ameer ALLOCATION AND LEVELING USING
GENETIC ALGORITHMS

Figure (11) illustrate the Resource Histogram before Smoothing where Mx = 353

Figure 11 Resource Histogram Before Smoothing

Figure (12) illustrate resource Histogram after Smoothing in MS Project where Mx = 353

Figure 12 Resource Histogram After Smoothing

Table (2) show the information needed to GA Resource Smoothing System which get it from
MS Project before Smoothing and Table (3) show the information after Smoothing.

942
 Number 4 Volume 17 August 2011 Journal of Engineering

Table 2 Activity Information Before Smoothing

Act Dur E.S T.F. R Prec. Critical or not

A 2 0 0 2 - Cri
B 6 2 0 4 A Cri
C 3 2 6 3 A Non-Cri
D 1 8 2 1 B Non-Cri
E 6 8 0 3 B Cri
F 3 9 2 3 C,D Non-Cri
G 2 14 0 2 E,F Cri

Table 3 Activity Information After Smoothing in MS-Project

Act Dur E.S T.F. R Prec. Critical or not

A 2 0 0 2 - Cri
B 6 2 0 4 A Cri
C 3 2 6 3 A Non-Cri
D 1 10 0 1 B+2day[FS] Cri
E 6 8 0 3 B Cri
F 3 11 0 3 C,D Cri
G 2 14 0 2 E,F Cri

It is noted that MS-Project after smoothing

make the activity D,F is critical and the Total Performing Resource Smoothing
Float = 0 while the moment of resources This application demonstrates the proposed
(Mx) is not changed and equal (Mx = 353). model ability to optimize the process of
resource smoothing to compare the results
Performing Optimization Process with MS-project program and manual results.
This application is perform to compare the The optimization parameters associated are
proposed model results with manual solution as follows:
of the case study. Optimization options are - Objective function: minimizes
set as follow: fluctuation moment (Mx).
- Population size: After initial - Change TF, ES to non-critical
experimentation with different activities.
population size (chromosomes) = No. - Constraints:
of duration of any project. Project duration = deadline duration
- Crossover rate = No. of total float of (16)
each non-critical activity. ES = early start to critical activity.
- Stopping condition: as GA gives The results of resource smoothing in GA are
results that show no further illustrated in table (4)
improvements.

943
 Dr. Sawsan Rasheed Mohammed OPTIMIZATION OF RESOURCE
Ameer Ahmed Abdul Ameer ALLOCATION AND LEVELING USING
GENETIC ALGORITHMS

Table 4 Activity Information of Resource Smoothing in GA

Act Dur E.S T.F. R Prec. Critical or not

A 2 0 0 2 - Cri
B 6 2 0 4 A Cri
C 3 8 0 3 A,B Cri
D 1 8 2 1 B Non-Cri
E 6 8 0 3 B Cri
F 3 11 0 3 C,D Cri
G 2 14 0 2 E,F Cri

And the bar chart after smoothing in GA as shown in Figure (13)

Figure 13 Bar Chart of 7 Activity After Smoothing in

GA

We conclude from the above that the MS because its small project where the Mx from
project made the activity D,F is critical where GA equal Mx from manual solution.
Mx = 353 but GA is made the activity C,F is It’s noted No. of peak and No. of changes
critical and activity D is still non-critical = 2 and highest peak = 7 from Ms results
where Mx = 341 and it is less than the while No. of changes and No. of peak = 1
calculated Mx from MS project and equal mx from GA results
from manual solution and the Histogram

944
 Number 4 Volume 17 August 2011 Journal of Engineering
Figure (14) show resource smoothing histogram from GA results where Mx = 341

Figure 14 Resource Histogram After

Smoothing From GA Results

Figure (15) show the GA output screen from C++

Figure 15 GA Output Screen

945
 Dr. Sawsan Rasheed Mohammed OPTIMIZATION OF RESOURCE
Ameer Ahmed Abdul Ameer ALLOCATION AND LEVELING USING
GENETIC ALGORITHMS

CONCLUSIONS help with it development, the fact that

Through the research work, there are success in finding a good solution is usually
groups of conclusions that can be achieved by optimize project plan during
summarized by the following points: planning stage.
1. Through the literature review, the 2. For the future work, it is suggested to
researcher found that : transfer the information automatically (to
a- There are no researches or studies perform Resource Smoothing).
accomplished in Iraq deal with the It is suggested to adopt the research
subject of Resource Leveling in findings in the application field throughout
spite of the importance of this part using the model in the planning and
of Management. managing of construction project in Iraqi
b- GA can be utilized in the other construction industry.
areas of Project Management such
as Productivity, Value Engineering, REFERENCES
Scheduling…etc. Baker, J.E. (1985). Adaptive Selection
2. Through the development of Methods for Genetic Algorithms.
Resource Smoothing Model using Proceedings of an International Conference
GA, the following points are on Genetic Algorithms and Their
concluded: Applications, 101-111.
a- GA have demonstrated to be a
promising tool for use in the initial Callahan, M. T., Enbush. D.G.Q. and
stages of Construction Projects. Rowings, J. E., 1992, "Construction Project
b- The scheduling model has the Scheduling", McGraw-Hill, Inc.
capability to obtain optimum
solution for performing resource De Jon, K.A. (1980b). “A Genetic-Based
smoothing. Genetic algorithms Global Function Optimization Technique”
program searches for the optimum Pittsburgh: University of Pittsburgh.
set of activities methods of Department of Computer Science.
constructions that minimize the
moment of resources (MX). The GOLDberq, D. E., 1989, "Genetic
purpose model can perform Algorithms in search optimization and
resource leveling is better than the Machine Learning", Addision Wesley
Microsoft project program (MS- Publishing Co.
project). We also find that the
calculated moment of resources Haider, A., Naoum, S., Howes, R and Tah, J.,
(MX) of proposed model (C++) is 1999, "Genetic Algorithms Application and
less than the results of (MS-project) Testing for Equipment Selection", ASCE
and given better smoothing. Journal of Construction Engineering and
Resource smoothing can be applied Management, vol. 125, No.1.
with the proposed model to obtain
results identical (improved further) Harmanani, H., Zouein, P., Hajar, A., 2000,
to that obtained from adopting "An Evolutionary Site Layout" , ASCE,
Microsoft project software. Journal of Construction Engineering and
Management, vol. 128, No.6.
RECOMMENDATIONS
1. Despite the good performance of GA in this Harris, R. B., 1990, "Packing Method for
work and in many situations in civil Resource Leveling (PACK)", Journal of
engineering, it suffer from a number of Construction Engineering and Management,
shortcomings, notably, the lack of theory to vol. 116, No. 2.

946
 Number 4 Volume 17 August 2011 Journal of Engineering
Engineering and Management ASCE, vol.
Harris, R. B., 1978, "Precedence and Arrow 125, No. 1.
Networking Techniques Press, London,
England. Tarek, H, 1999 "Optimization of
Construction Time-Cost Trade off Analysis
Hiyassat, A. M., 2001, "Applying Modified Using Genetic Algorithms", Canadian
Minimum Moment Method to Multiple Journal of Civil Engineer, Vol. 26.
Resource Leveling", ASCE, Journal of
Construction Engineering and Management, Tarek, H. 1999 "Optimization of Resource
vol. 127, No. 3. Allocation and Leveling Using Genetic
Algorithms", ASCE, Journal of Construction
Kevin, P. and Tarek, H., 2003, "Genetic Engineering and Management, Vol. 125, No.
Optimization for Dynamic project Control", 3.
ASCE journal of Construction Engineering
and Management, vol. 129, No.4. Tarek, S. A., 2000, "conjugation and
Controlled Mutations New Operatos in
Leu, S. and Yang, C. H.,1999, "GA-Based Genetic Algorithms", Ph. D. Thesis,
Multicriteria in Optimal Model for Computer Science and Information System,
Construction Scheduling", ASCE, Journal of University of Technology.
Construction Engineering and Management,
vol. 125, No. 6.

Maria, A., 1997 "Introduction to Modeling

and Simulation" Proceedings of the 1997
Winter Simulation Conference, Atlanta,
Georgia, USA.

Michell,M.,1998,"An Introduction to Genetic

Algorithms", MIT press, London, England.

Moselhi, A., and Lorterapong, P. (1993).

“Least Impact Algorithm for Resource
Allocation.” ” Can. J. Civ. Engrg., CSCE,
20(2).

Project Management Institute, 1996, " A

guide to the Project Management Body of
Knowledge" USA.

Que, B. C., 2002, "Incorporating

Practicability into Genetic Algorithms-Based
Time-Cost Optimization", Journal of
Construction Engineering and Management.

Senupta, B. and Guha, H.,1995,

"Construction Management and Planning",
Tata Mc Gra-Hill.

Son, J. and Skibuiewski, M. J., 1999, "Multi

Heuristic Approach for Resource Leveling
Problem in Construction Engineering and
Management", Journal of Construction

947

View publication stats