Professional Documents
Culture Documents
(1)
C. The normalize criterion matrix
Because there are different dimension and magnitude
among the judgment criterions, we cant directly contrast
them. To ensure the reliability of the result, we should
normalize the original value.
Assume that the value interval of the k-th criterion is
k1 2
] [ , e
k
e
,
k1
e is the smallest value of the k-th criterion in all
methods, and
k2
e is the largest value of k-th criterion in all
methods, then we can use the following formula to transfer
the original value into dimensionless value
ij
f (0,1) e
.
2
ij
1
/
/
f
ij k
k ij
e e when j is positive criterion
e e when j is nagative criterion
(2)
Where 1, 2, , ; 1, 2, , . i m j n = =
The normalized matrix F as follows:
11 12 1
21 22 2
ij
1 2
) =(f
n
n
m n
m m mn
f f f
f f f
f f f
F
Where 1, 2, , ; 1, 2, , . i m j n = =
D. Identification weight of the criterions
The general methods used to calculate the criterion weight
are Delphi method, Brain storming, Analytic Hierarchy
Process (AHP) and so on. Here, the information entropy is
employed to determine the criterion weight. Shannon
proposed the conception of information entropy for the
discrete stochastic variables. In information theory,
information is the measurement of system order degree, while
information entropy is the measurement of system disorder
degree. Information entropy is defined as:
( ) ( ) ( )
1
j
x ln
1
m
ij ij
i
H p x p x
k =
= (3)
Where
ij
x is the possible state of random independent
events.
( )
ij
p x is the probability of a particular state and
( )
0
ij
p x > ,
1
) 1 (
ij
m
i
p x
=
=
.The value of
( )
ij
p x can be
determined by following:
( )
1
ij
m ij
ij
i
f
p x
f
=
=
(4)
1
( ) ( ln ( ) )
m
i
ij ij ij
H x p x p x
=
=
(5)
1
1
1 ( ln (
ln
) )
m
ij
i
ij ij
v p x p x
n =
= +
(6)
1
m
j ij ij
i
w v v
=
=
(7)
1 2
[ , , , ]
T
n
W w w w = is the weight distribute vector , and
1
1
n
j
j
w
=
=
.
E. Grey relational grade matrix
Assume that
ij
c
is the relationship coefficient of the i-th
method of the j-th criterion.
ij
min min 1 max max 1
1 max max 1
ij ij
j i j i
ij ij
j i
f f
f f
G
p
c
+
=
+
=
(8)
where p is the identification coefficient, its interval is [0,1],
commonly , it is 0.5. So we can obtain the gray relational
grade matrix:
11 12 1
21 22 2
ij
1 2
n
n
m m mn
G
c c c
c c c
c c c
c
=
=
Where 1, 2, , ; 1, 2, , . i m j n = =
F. Determine the decision-making vector
The model of the grey relation synthesis decision- making
is on basis of follows:
R=G W (9)
T
1 2 m
,r , ,r R=[r ] is the integrative judgment result vector of
the m evaluated objects; W is weight vector.
1
n
i ij j
j
r w c
=
=
(10)
If the value of relation degree
i
r is larger, it proves that the
i-th method is closest to the optimal method, and we can get
the priority order of the methods.
1035
IV. REAL-WORLD CASE ANALYSIS
In this section, we apply the grey relation analysis model to
a real-world infrastructure project.
Nanjing Metro Co.Ltd. wanted to alternate and innovate
project delivery system in second section of metro shield
tunnel project. The alternative set of delivery systems is
design-bid-build (DBB), design-build (DB), engineer-
procureconstruct (EPC) and construction management at
risk method (CM at-Risk).
We invited ten experts including academics, contractors,
owners and engineers to set up a selection committee. Firstly,
they were invited to investigate the construction site and
discussed together. Secondly, we provided the preliminary
design document of the project and the organizational
structure and construction experience documents to the
experts. Thirdly, the questionnaires documents about the
criteria for selection of project delivery systems were
presented to the experts individually, and ask them to choose
the questions. Finally, we collected the questionnaires and
dealt with the data.
The process of decision-making is as following:
A. Step 1 Calculate the value of criterion matrix
The ten invited experts respectively appraise the
alternative methods under each attribute, using the comment
set V. In this paper, comment set V = {very poor (VP), poor
(P), fair (F), good (G), very good (VG)} is chosen, value of
comment set V are represented by 1,2,3,4,5. Then we
calculate the mean value of each criterion for every alternate
delivery method, and obtain the value of criterions matrix as
Table II showing:
TABLE II
VALUES OF THE CRITERIA MATRIX FOR EACH DELIVERY SYSTEM
Criteria
Method
C S Q Com SC E FG RM U Size
DBB 2.0 2.7 3.8 2.0 3.4 1.5 1.6 3.8 4.0 1.5
DB 1.9 1.8 1.9 3.4 1.5 3.1 2.6 4.0 4.9 3.8
CM-at risk 2.9 1.0 3.1 3.7 1.8 1.8 3.8 3.4 2.6 4.6
EPC 2.5 4.9 1.4 2.5 4.6 4.2 3.7 4.6 1.3 4.2
B. Step 2 Normalization of criterion matrix
Normalizing the criteria matrix in table II according to (2),
the result is as the following Table III.
TABLE III
NORMALIZATION OF CRITERION
Criteria
Method
C S Q Com SC E FG RM U Size
Optimal criterion set 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
DBB 0.6897 0.5510 1.0000 0.5405 0.7391 0.3571 0.4211 0.8261 0.8163 0.3261
DB 0.6552 0.3673 0.5000 0.9189 0.3261 0.7381 0.6842 0.8696 1.0000 0.8261
CM-at risk 1.0000 0.2041 0.8158 1.0000 0.3913 0.4286 1.0000 0.7391 0.5306 1.0000
EPC 0.8621 1.0000 0.3684 0.6757 1.0000 1.0000 0.9737 1.0000 0.2653 0.9130
C. Step3 Determine the weight of the criterion
The calculation of criteria weights is according to (4)-(7)
and the values of table II. The value of the weight vector is as
follows:
1 2 10
[ , , , ]
T
W w w w =
=[0.078,0.083,0.143,0.062,0.118,0.098,0.089,0.114,0.1
33,0.082]
D. Step 4 Calculate the relational grade coefficient
The relational coefficient matrix is calculated by (8), the
results are showed as following:
0.57 0.45 1.00 0.53 0.67 0.35 0.43 0.79 0.80 0.34
0.48 0.30 0.49 0.89 0.23 0.61 0.63 0.83 1.00 0.57
1.00 0.26 0.62 1.00 0.38 0.50 1.00 0.68 0.54 1.00
0.71 1.00 0.32 0.78 1.00 1.00 0.87 1.00 0.21 0.89
G
=
E. Step 5 Calculate the comprehensive judgment result
It is known that the weight W is given above, according to
(10), the comprehensive judgment result vector R is
R=G W=(0.563,0.798, 0.432,0.678)
According to the result of R, we can get the order of the
four methods:
DB> EPC> DBB >CM-at risk
So the candidates of the project delivery systems: DBB,
DB, CM-at risk and EPC, DB is the most appropriate method.
This result can be proposed to the owners, and help them to
make decision.
V. CONCLUSION
This study proposed a decision-making model for the
selection of project delivery system which is based on
information entropy and grey relational grade analysis. The
application of the model in a real-world practice was
provided. The proposed methodology is more comprehensive
compared with the previous work, especially in the
uncertainty environment. But there are some work should be
considered, such as how to deal with the imprecise and
subjective information given by the experts; how to
determine the weight of the experts`; finding a set of
importance factors influencing the selection of a delivery
system is a complex task to further research. Furthermore it is
1036
the essential and most important work to design and innovate
the project delivery system.
REFERENCES
[1] Adetokunbo A. Oyetunji, Stuart D. Anderson, Relative effectiveness
of project delivery and contract strategies, Journal of Construction
Engineering and Management, 2006, Vol132,no.1, pp. 3-12.
[2] Gerald Yakowenko, Megaproject procurement: breaking from
tradition, Public Roads, 2004, Vol.18, no.1,pp. 48-53.
[3] Contractual Arrangements (Report A-7), Business Roundtable, New
York, NY. 1982.
[4] Molenaar K R., Songer A D. Model for public sector design-build
project selection, Journal of construction engineering and
management, 1998,Vol.124, no.6,pp.467-479.
[5] Koncharl M, Sanvido V. Comparison of U.S. project delivery system,
Journal of Construction Engineering and Management, 1998, Vol124,
no.6 , pp.435-444.
[6] Chan A P, Ho D C, Tam C M. Design and build project success factors:
Multivariate analysis, Journal of Construction Engineering and
Management, 2001,Vol.127,no.2,pp.93-100.
[7] Florence Yean Yng Ling, Swee Lean Chan, Edwin Chong, Predicting
performance of design-build and design-bid-build projects, Journal of
construction engineering and management, 2004, Vol13.no.1, pp.
75-83.
[8] Ibrahim M M, Khaled A, Decision support system for selecting the
proper project delivery method using analytical hierarchy process
(AHP), International Journal of Project Management, 2005, Vol.
23.pp.564-572.
[9] Mohammed I, Khalil A, Selecting the appropriate project delivery
method using AHP , International Journal of Project Management,
2002 ,Vol.20,pp.469-474.
[10] Gordon. C. M. Choosing appropriate construction contracting
method, Journal of Construction Engineering and Management, 1994,
Vol.120, no.1, pp. 196-210.
[11] Alhazmi T, McCaffer R. Project procurement system selection
model, Journal Construction Engineering and Management, 2000,
Vol.126,no.3,pp. 176-184.
[12] Spink C M, Choosing the right delivery system, Proceedings of the
1997 ASCE Construction Conference, pp. 63-71.
[13] J. L. Deng, Control problems of grey system, System and Control
Letters, 1982, vol. 5, pp. 288294.
[14] J. L. Deng, The primary methods of grey system theory, Huazhong
University of Science and Technology Press,Wuhan, 2005.
[15] Ziyang Zhen, Zhou Gu, and Yuanyuan Liu, A novel fuzzy entropy
image segmentation approach based on grey relational analysis,
Proceedings of 2007 IEEE International Conference on Grey Systems
and Intelligent Services, November 18-20, 2007, Nanjing, China,
pp.1019-1022.
[16] Bao Zhen-qiang, Wang Peng, Yang Fang, Zhu Cong-wei, Guo Lei,
Grey relation degree analysis for the facilitys location of logistics
distribution network, 2008 International Symposiums on Information
Processing, 23 - 25, May 2008, Moscow- Russia, pp: 615-619.
[17] Gordon CM, Choosing appropriate construction contracting method,
Journal of Construction Engineering and Management,
1994,Vol.120,no.1, pp.196-210.
[18] Fereshteh Mafakheri, Liming Dai,Dominik Slezak,and Fuzhan Nasiri,
Project delivery system selection under uncertainty: multi-criteria
multi-level decision aid model , Journal of Construction Engineering
and Management, Vol.23, no. 4, pp. 200-206.
[19] J. L. Deng. Grey theory, Huazhong University of Science and
Technology Press,Wuhan, 2002.
[20] Chan WK and Tong TKL, Multi-criteria material selections and
end-of-life product strategy: Grey relational analysis approach,
Materials & Design, 2007, Vol. 28, no. 5, pp.1539-1546.
[21] Sallehuddin, Roselina Shamsuddin, Siti Mariyam Hj. Hashim, Siti
Zaiton Mohd . Application of grey relational analysis for multivariate
time series,Intelligent Systems Design and Applications,ISDA,2008
8th International Conference on , 26-28 Nov. 2008 , Kaohsuing,
Taiwan, pp.432-437.
1037