Sei sulla pagina 1di 12

GENETIC ALGORITHM FOR OPTIMIZING CONSTRUCTION PROJECT

MANAGEMENT PROCESSES IN PHILIPPINES

Michael V. Almeida¹

¹ Philippine Institute of Civil Engineers Inc., Manila, Philippines

Abstract : Construction project management, one of the fastest growing in the country, experienced a boom, generating jobs and
enhancing growth of construction industry. The success of Construction Project Management is measured by the efficiency and
effectiveness of the decisions made in managing construction project. During the management process all objectives, functions
and constraints have to be optimized. In this research paper the use of genetic algorithm as optimization tool is suggested for
making best decisions in construction project management. The use of this optimization technique for construction project
management in managing construction projects in Philippines is proposed. The objective of this paper is to present the
optimization models in developing decision making frameworks for managing construction projects.

Key words : optimization; genetic algorithm; construction project management


constraints which normally refer to scope, quality, time, cost,
1 INTRODUCTION human resource, communication, risks and procurement, and
Construction Project Management is the planning, control and directly links them to the project life cycle. These constraints
coordination of a project from conception to completion are used for measuring the success of construction project
(including commissioning) on behalf of a client. It is management. Therefore, in order to improve the efficiency
concerned with the identification of the client’s objectives in and effectiveness of managing construction projects in the
terms of utility, function, quality, time and cost, and the Philippines all these constraints should be considered. The
establishment of relationships between resources. The objective of this research paper is to introduce the application
integration, monitoring and control of the contributions to the of optimization technique in developing decision making
project and their output, and evaluation and selection of frameworks for managing construction project constraints.
alternatives in pursuit of the client’s satisfaction with the One of the most important decisions to be made in
project outcome are fundamental aspects of Project construction project management before executing any project
Management (Walker 1984). According to the Code of is during pre-qualification process. The proper selection
Practice for Project Management for Construction and process increases chances of project delivery within time, cost
Development (Chartered Institute of Building 2003), Project and quality. In this research paper the use of Genetic
Management as an emergent professional discipline which Algorithm is suggested in developing decision making model
separates the management functions of a project from the in determining best criteria for contractor’s selection during
design and execution functions is the overall planning, pre-qualification process. Six criteria are evaluated for the
coordination and control of a project from inception to primary objective of making decision faster and more efficient.
completion aimed at meeting a client’s requirements in order The six criteria are randomly selected to generate possible
to produce a functionally and financially viable project that solutions for making optimal decisions. The results show that
will be completed on time within authorized cost and to the there is chance to make possible the objectives, rationalized
required quality standards as defined by. The Project and optimized the decisions during pre-qualification process in
Management Institute (PMI 2013) defines Project selecting contractors for construction projects.
Management as the application of knowledge, skills, tools, and
techniques to project activities to meet project requirements. It Several researchers (Holt et al, 1994; Russell et al, 1992; Ng,
is accomplished through the appropriate application and 1992) have identified different criteria in use for contractor
integration of the 47 logically grouped project management selection. In a recent study, Hatush and Skitmore (1996a)
processes, which are categorized into five Process Groups; found that all clients use what are implicitly the same type of
Initiating, Planning, Executing, Monitoring and Controlling, criteria, but vary in the way they quantify the criteria, with
and Closing. The common point of all the Project Management most having to resort to a very subjective assessment based on
definitions stresses the achievements of competing project information provided by the contractors. Also many

1
techniques are proposed and applied as a solution (Hatush and The objective of this research paper is to introduce the
Skitmore, 1998; Cheng and Heng, 2004; Plebankiewicz, 2009; application of Genetic Algorithm on the contractor pre-
Jaskowski et al., 2010). Because of its wide application in qualification process to deal with the problem of inconsistent
construction project management Analytical Hierarchy Process comparisons of judgment provided by decision makers during
AHP is, as decision making method, widely used for multiple prequalification of contractors. The paper will briefly review
criteria decision-making (MCDM) in construction project the concepts and application of GA’s implementation steps,
management. (Saaty, 1990; Kamal et al., 2001; Chun-Chang and demonstrate GA application on the contractor selection
Lin et al., 2008; Jaskowski et al., 2010). Some areas of problem. It is hoped that this will encourage its application in
construction project management where AHP method is used construction project management in Philippines.
are contractor selection (Kamal et al., 2001; Jaskowski et al.,
2010; Abudayyeh et al., 2007), technology selection
(Skibniewski and Chao, 1992), equipment
selection(Shapiraand Goldenberg, 2005), analysis of causes of
disputes in the construction industry (Cakmak and Cakmak,
2013). AHP based contractor selection procedure for highway
infrastructure projects in serbia (Petronijević et al., 2015).

Despite of the wide acceptance of AHP in construction project


management as decision making tool in contractor selection
during prequalification process has been criticized on the
ground of providing consistent comparisons. This problem of
decision makers often occurs when criteria and alternatives
being compared are of large numbers. In Saaty’s AHP,
consistency is validated by the Consistency Ratio (CR) which
indicates the probability that the matrix ratings are randomly
generated. If CR is more than 0.10, the judgment matrix is Fig. 2 Some combinations of quality, cost, time and
inconsistent and should be revised (Saaty, 1980). environmental impacts

Recent study identified that most of Artificial Intelligence


Algorithm applications used in supporting decision making
process in construction management are focused on resources
allocation, evaluation design alternatives and optimizing
decisions considering time, cost and quality constraints as
shown in figures below. Where: [1]Genetic Algorithm(G.A.),
[2] Enumerative Branch-And-Cut (EBAC),[3] Fuzzy Logic
(FL), [4]Ant Colony Optimization (ACO), [5] Artificial Neural
Network (ANN), [6]Memetic Algorithm (MA) [7]Particle
Swarm Optimization (PSO) [8]Simulated Annealing
Algorithm (SAA) (Elhassan et al., )

Fig. 1 Time Problem Fig.3 Time and Cost Tade-off Problem

2
2 METHODOLOGY
In this research paper the priority weights of each criteria
commonly used in contractors prequalification process are
obtained using Analytic Hierarchy Process prioritization
methods and by using Genetic Algorithm in solving the
inconsistency of the pair-wise comparison matrix and
optimization problem for deriving priorities of the given
criteria. Genetic Algorithm (GA) is a search-based
optimization technique based on the principles of Genetics and
Natural Selection were developed by John Holland and his
students and colleagues at the University of Michigan, most
notably David E. Goldberg and has since been tried on various
optimization problems with a high degree of success It is
frequently used to find optimal or near-optimal solutions to
difficult problems which otherwise would take a lifetime to
solve. It is frequently used to solve optimization problems, in
research, and in machine learning. We start with an initial
population (which may be generated at random or seeded by
other heuristics), select parents from this population for mating.
Apply crossover and mutation operators on the parents to
generate new off-springs. And finally these off-springs replace
the existing individuals in the population and the process
repeats.

In this paper we start with rank selection for selection, one


point crossover for crossover and uniform mutation for
mutation. To ensure that best population always survives,
elitism is applied as an additional selection strategy figure 4
shows the flow chart for determining priorities, normalizing
decision matrix and genetic algorithm procedure (Aly and Abd
El-hameed, 2013).

The proposed parameters of GA set in this research paper are


often performed only on binary encoding. Crossover rate of
60%, mutation rate of 0.5%-1% and population size of 20-30.

A. Population Initialization
Population is a subset of solutions in the current generation. It
can also be defined as a set of chromosomes. In this research
paper the six criteria use by decision makers during
contractors pre-qualification process represents six element
position or set of genes of a chromosome. The six criteria as
follows:

C1 (Financial Capability) – involves contractor’s sound


financial position and profitability, here is considered
minimum average annual construction turnover within the last Fig.4 Flow chart for normalized decision matrix and genetic
five years; algorithm (Aly and El-hameed, 2013)

C2 (Past Performance) – past client’s levels of satisfaction C3 (Past Experience) – minimum value of contracts which are
with the quality of previous works and maintenance services similar to the proposed works and which were successfully
during defects liability period by the contractors in the past completed within the last five years;
five years;

3
C4 (Resources) – availability of competent personnel, owned B. Fitness Function Calculation
major plants and equipments for construction; In GA each problem to be solved requires the definition of a
unique fitness function describing the characteristics of an
C5 (Current Workload) – construction activities which are appropriate solution. The purpose of the fitness function is to
underway, on-going and nearing completion; evaluate the suitability of candidate solution with respect to
the overall goal. Given a particular chromosome, the fitness
C6 (Safety Performance) – safety performance/ accident rate function returns a numerical value corresponding to the
in the past five years; chromosome’s quality using Saaty’s judgment consistency
ratio of smaller or equal to 10% (CR) of CI = (ƛmax - n)/ (n -
These six criteria are compared in pairs. It means that one 1), n is the matrix size with the appropriate value in table 3.
unfamiliar with the methodology of AHP can compare two
elements from the same level according to verbal description Table 3. Random index values
scale. Fundamental scale used to compare the elements
consists of verbal judgments ranging from equal to extreme n 1 2 3 4 5 6 7 8 9 10
(equal, moderately more, strongly more, very strongly more, RI 0 0 0.58 0.90 1.12 1.24 1.32 1.41 1.45 1.49
extremely more) (Petronijevic et al., 2015). Corresponding to
the verbal judgments are the numerical values (1, 3, 5, 7, 9) The candidate solution will set as parent which will produce
and intermediate values (2, 6, 8). (Saaty, 1990) Saaty’s scale is offspring, store the offspring chromosome and be included in
given in table 1 . the selection process.

Table 1. Pair-wise comparison scale C. Encoding

Numerical The size of the candidate solution string is six (6) which
Verbal Judgments of Preferences composed of the criteria used in selecting the best contractor.
Rating
Each bit of string represents the rank of the corresponding
9 Extremely preferred
criterion (string indices). Then this string indices are translated
8 Very strongly to extremely to set of binary numbers (binary encoding) for each population
7 Very strongly preferred or chromosomes in initial population as shown in table 4.
6 Strongly to very strongly
5 Strongly preferred Table 4. Binary codes
4 Moderately to strongly
Criteria Weights
3 Moderately preferred Criteria Ranking
(Candidate Solution Binary Codes
(String Indices)
2 Equally to moderately String)
C1 (w1) 1 001
1 Equally preferred
C2 (w2) 2 010
C3 (w3) 3 011
In genetic algorithm we can start with either a set of random C4 (w4) 4 100
chromosomes, or ones that represent already approximate C5(w5) 5 101
solutions to the given problem. The comparison results using C6(w6) 6 110
the Analytical Hierarchy Process (AHP) for the six criteria
C1, C2, C3, C4, C5 and C6 represent a possible solution in D. Crossover
determining the best criteria for contractor’s selection during
pre-qualification process. The initial population of candidate During crossover new solution strings are created by
solution is generated randomly as shown in table 2 below . combining aspects of the selected candidate solutions like how
reproduction works in nature. By combining the traits from the
Table 2 Primary questionnaire design two candidate solution ‘fitter’ offspring are created which
inherit the best traits from each of it’s parents. In this
Factor Weighting Scores research A one point random crossover point was selected and
Fac
More preferred than Eq Less preferred than tails of its two parents are swapped to get new off-springs as
tor
ual
C1 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9 illustrated below.
C2 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9
C3 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9
C4 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9
C5 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9
C6 9 8 7 6 5 4 3 2 1 2 3 4 5 6 7 8 9

4
Parent a the comparison matrix to form a new matrix and the average
C2 C3 C4 C5 C6 C1 of each row was calculated to determine the priority vector.
Parent b After all pair wise comparisons are made consistency ratio was
C5 C6 C1 C2 C3 C4 calculated by multiplying the weight column by the Level-1
matrix in table 5 obtain a new matrix. The sum of each row
was calculated and the sum column was divided by the weight
column to find the average of the column (ƛ max). The
consistency ratio was calculated by dividing the consistency
Child a index by corresponding (RI) given in table 3. The calculated
C2 C3 C4 C2 C3 C4 consistency ratio does not exceed 10%, the judgement matrix
Child b is acceptable and consistent. Therefore the normalized matrix
C5 C6 C1 C5 C6 C1 is a candidate solution on our selection process.

E. Mutation Table 6. Normalized matrix for six criteria

Priority
Mutation is the part of the GA which is related to the Solution C1 C2 C3 C4 C5 C6 Vector
“exploration” of the search space. It has been observed that C1 0.510 0.686 0.412 0.381 0.384 0.455 0.471
mutation is essential to the convergence of the GA while C2 0.085 0.114 0.353 0.286 0.308 0.091 0.206
crossover is not. In this paper a little bit randomness into each C3 0,073 0.019 0.059 0.095 0.077 0.091 0.069
populations’ genetics were added in order to avoid every 0.128 0.038 0.059 0.095 0.077 0.182 0.096
C4
combination of solutions that can be created would be in the 0.102 0.029 0.059 0.095 0.077 0.091 0.075
C5
initial population. 0.102 0.114 0.059 0.048 0.077 0.091 0.082
C6
Mutation of Child a Ʃ= 1.00
C2 C3 C2 C4 C3 C4 ƛmax= 6.551, CI= 0.11, RI=1.24, CR= 0.089 < 0.10 OK.
Mutation of Child b Table 7 shows the size of the candidate solution string and
C5 C6 C5 C1 C6 C1 respective priority weights of each criteria.. Each bit of string
are ranked accordingly (string indices). Then this string
indices are translated to set of binary numbers (binary
3 RESULTS AND DISCUSSION encoding) which represent the candidate solution or
chromosomes in initial population.
These six criteria were evaluated with respect to the primary
objective, to select the best and capable contractor for the Table 7. Ranking translation to binary codes
project. Scores were evaluated from the interviews and survey
questions distributed to experts in the field of construction Equivalent
Criteria Weights
management and area of contractor’s pre-qualification. Table 5 Criteria Ranking Binary Codes
(Candidate
shows a comparison matrix of order six (6) where the scores (String Indices) (Candidate
Solution String)
for each criteria C1, C2, C3, C4, C5 and C6 are pairwise Solution Code)
compared against the objective (selection of competent C1 (0.471) 1 001
contractor) for importance. C2 (0.206) 2 010
C3 (0.069) 6 110
Table 5. Pair-wise comparison matrix scores for six criteria C4 (0.096) 3 011
C5(0.075) 5 101
Solution C1 C2 C3 C4 C5 C6 C6(0.082) 4 100
C1 1 6 7 4 5 5
C2 1/6 1 6 3 4 1 Implementing binary genetic algorithm in Java manages all
C3 1/7 1/6 1 1 1 1 individuals of a population, manages an individuals, manages
C4 1/4 1/3 1 1 1 2 the crossover and mutation, and allows set a candidate solution
C5 1/5 1/4 1 1 1 1 and calculate an individual’s fitness, and output as shown in
C6 1/5 1 1 1/2 1 1 Appendix 1, 2, 3, 4 ,5 and 6 respectively. Figure 5 shows the
synthesis scores of each criteria. Obviously, C1 (Financial
Table 6 shows the normalized relative weight, priority Capability) is the best and C3 (Past Experience) being the least
vector and consistency ratio. The weight of each element was among each criteria. The result shows that the initial
calculated by dividing each score by the sum of its column in population used in the decision making model is consistent.

5
Objectives and Constraints
OPTIMUM SOLUTION
Integration Scope Time Cost Quality
Communication Risks HumanResources
60 Procuremen t
50
40
Construction Project Phases
FITNESS

C1
30 C2 Initiating Planning Executing
C4 C5 C6
20 C3 Controlling & Monitoring
Closing
10
0
1 2 3 4 5 6
Type of Optimization Problem
GENERATION Project Scheduling Personnel Assignment
Alternative Selection

Fig. 5 Optimum solution


Data Availability
4 CONCLUSIONS

The paper has presented GA as a decision-making tool in Optimization Algorithms


determining the order of each criteria used to select the best
alternative. GA allows options for owner and construction Genetic Algorithm (GA ) Fuzzy Logic (FL)
managers in the selection of the best contractor for high-rise Ant Colony Optimization (ACO) Memetic
building project in Philippines. This selection method avoids Algorithm (MA) Enumerative Branch-And-
many risks which may result to problems if the project was Cut (EBAC) Artificial Neural Network
awarded to less capable contractor. (ANN) Particle Swarm Optimization (PSO)
Simulated Annealing Algorithm (SAA) Back
Managing complex projects involves complex decision Propagation Algorithm (BPA) Wavelet (W)
making abilities. Project failures not only result to poor Case-Based Reasoning (CBR)
selection of contractors but who made the selection process.
The method can also be used in selecting who will be the best
capable construction project manager.
Data Collection
Figure 6 shows the flowchart for developing decision making Questionnaire and Interviews Historical
framework in construction project management to optimize Records Case Studies Automated Process
competing project objectives and constraints. The marked
portion and arrows show the decision making model using
optimization algorithm in determining the best criteria for
selecting competent contractor during pre-qualification Software
process. The flowchart also illustrates combinations of project Visual Basic M.S. Excel JAVA C++
objectives and constraints for future research studies in MATLAB
construction project management.

ACKNOWLEDGMENTS
Data Analysis and Development of Decisions
The author would like to thank De La Salle University Optimization Model
Graduate School, Design Coorinates Inc., colleagues, staffs
and family for providing the guidance and support to conduct Fig. 6 Flow chart for developing decision making
this research paper. framework in construction project management

6
APPENDICES

Appendix 1. Population.Java

7
Appendix 2. Individual.Java

8
Appendix 3. Mutation and Crossover.Java

9
Appendix 4. Fitness Calculation.Java

10
Appendix 5. Genetic Algorithm.Java

Appendix 6. Output.Java

11
REFERENCES Shapira, A. and Goldenberg, M. “AHP-Based Equipment Selection Model for
Construction Projects.”Journal of Construction Engineering and
Abudayyeh, O., Zidan, S., Yehia, S., and Randolph, D. “Hybrid Management (2005): 131(12), 1263-1273.
Prequalification-Based, Innovative Contracting Model Using AHP.” Skibniewski, M. and Chao, L. “ Evaluation of Advanced Construction
Journal of Management in Engineering (2007); 23(2), 88-96. Technology with AHP Method.” Journal of Construction Engineering
Adeli, H., and Karim, A. (2001). “Construction Scheduling, Cost and Management 118(3); (1992), 557-593.
Optimization, and Management”. A New Model Based on Ng, S.T.T., (1992), Decision support system for contractor prequalification,
Neurocomputing and Object Technologies. University of Salford, Department of Surveying, UK.
Al-Subhi Al-Harbi, K.M., Application of AHP in project Management. Petronijevic, M., Nikolic, A., Mikic, M., Ivanisevic, N. (2015 September).
International Journal of Project Management 19 (2001) 19-27. AHP Based Contractor Selection Procedure for Highway Infrastructure
Aly, M., and Abd El-hameed, H. (2013). Integrating AHP and Genetic Projects in Serbia. Croatian Association for Construction Management.
Algorithm Model Adopted for Personal Selection. International Journal pp. 206-214. Retrieved November 03, 2015 from Miljan Mikic.
of Engineering Trends and Technology – Volume 6 Number 5-Dec 2013 http://www.researchgate.net/publication/282732904
Balubaid, M., Alamoudi, R. (2015). Application of the Analytical Hierarchy Plebankiewicz, E., 2009, Contractor prequalification model using fuzzy sets,
Process (AHP) to Multi-Criteria Analysis for Contractor Selection. Journal of Civil Engineering and Management 15 (4), p. 377-385.
American Journal of Industrial and Business Management, 2015, 5, Satty, T.L., “The Analytic Hierarchy Process”, McGrawHill, New York, 1980.
pp.581-589. Thomas L. Saaty. How to make a decision: The Analytic Hierarchy Process.
Cakmak, P. and Cakmak, E. (2013) An Analysis of Causes of Disputes in the European Journal of Operational Research 48;(1990) 9-26.
Construction Industry Using Analytic Hierarchy Process (AHP). AEI Sivanandam, S.N and Deepa, S.N., Introduction to Genetic Algorithm
2013: pp. 94-102. Zgigniew, Mtr., Genetic Algorithms + Data Structures = Evolution Programs.
Cheng, W.L.E., Heng L., 2004, Contractor selection Using the analytic 3rd, Revised and Extended Edition
network process, Construction Management and Economics 22, p. 1021-
1032.
Chun-Chang Lin, Wei-Chih Wang, Wen-Der Yu. Improving AHP for
construction with an adaptive AHP approach (A3). Automation in
Construction 17; (2008) 180-187.
Eddie, W.L., Cheng, Li, H. (2004, December). Contractor selection using the
analytic network process. Construction Management and Economics 22:
pp.1021-1032
Elhassan, Zawawi and Ghazali. Decision making framework for optimizing
construction management objectives: A review.
Hatush, Z., Skitmore, R.M., (1996a), Common criteria for contractor selection,
Construction Management and Economics (in press).
Hatush, Zedan and Skitmore, Martin R. (1997) Assessment and evaluation of
contractor data against client goals using pert approach. Construction
Management and Economics 15(4):pp. 327-340.
Hatush, Z., Skitmore, M., 1998, Contractor selection using multicriteria utility
theory: an additive model, Building and Environment 33, p. 105-115
Haupt, R.L., Haupt, E.S., Practical Genetic Algorithms
Huang, X. (2011, March). An Analysis of the Selection of Project Contractor
in the Construction Management Process. International Journal of
Business and Management. Vol.6, No.3, pp. 184-189.
Holt, G.D., Olomolaiye, P.O., Harris, F.C., 1994, Factors influencing UK
construction clients’ choice of contractor, Building and Environment,
1994 29 (2) p. 241-248.
Jaskowski, P., Biruk S., Bucon R., 2010, Assessing contractor selection criteria
weights with fuzzy AHP method application in group decision
environment, Automation in Construction 19, p. 120-126.
Kamal, M., Subhi, A., Harbi, A. Application of the AHP in project
management. International Journal of Project Management 19 (2001).
pp.19-27.
Russell, J.S. (1992). Decision models for analysis and evaluation of
construction contractors. Construction management and economics, 10,
185-202.
Saaty, T.L. (2008). Decision making with the analytic hierarchy process. Int. J.
Services, Vol.1, No.1, pp.83-98.
Shiau, Y.C., Tsai, T.P., Huang, M.L. Use questionnaire and AHP techniques to
develop subcontractor selection system.pp.1-6.

12

Potrebbero piacerti anche