Académique Documents
Professionnel Documents
Culture Documents
www.elsevier.com/locate/autcon
Abstract
When a number of bridges are to be maintained yearly, contract engineers need to package certain bridges based upon the specified conditions.
Traditional contract packaging is performed subjectively and manually. The purpose of this paper is to develop an objective and automatic
decision-making process for contract packaging. Contract Packaging Technique (CPT), which is derived from k-prototypes, can determine
appropriate contract packages by considering the inherent conditions of the infrastructure to be maintained. It can simultaneously handle userspecified constraints and mixed data types such as bridge locations, bridge types, bridge construction materials and maintenance costs. Validation
was performed in this study to illustrate that the CPT can effectively divide a huge bridge maintenance plan into several appropriate contract units.
2007 Elsevier B.V. All rights reserved.
Keywords: Bridge maintenance; k-prototypes; Clustering; Contract packages
1. Introduction
Many bridges in Taiwan were designed and built 30 to 40 years
ago. They have endured public overuse and the negative
influences of the environment over their lifetimes. Today, a
majority of them are deteriorating and need to be maintained,
rehabilitated or replaced. In recent years, there have been
numerous studies in the field of bridge maintenance. Kong [1]
developed a computer program to analyse the life-cycle
performance of deteriorating structures based on system reliability. Fwa [2] proposed a genetic-algorithm-based procedure for
solving the multi-objective network-level pavement-maintenance
programming problems. Zayed [3] applied the dynamic programming (DP), integer programming (IP) and greedy heuristic (GH)
approaches to optimise a two-year budget allocation for
repainting the steel bridges in Indiana. Bandara [4] focused on
Corresponding author. Department of Construction Engineering, National
Taiwan University of Science and Technology, 43 Kee-Lung Road, Section 4,
Taipei, Taiwan 10672, ROC. Tel.: +886 2 27333141x7511; fax: +886 2 2737
6606.
E-mail addresses: yingmei.cheng@msa.hinet.net (Y.-M. Cheng),
leuss@mail.ntust.edu.tw (S.-S. Leu).
0926-5805/$ - see front matter 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.autcon.2007.12.001
the development of a subjective pavement evaluation methodology for priority ranking and future condition prediction. Labi [5]
developed a model for estimating seal coating effectiveness. Chen
[6] studied 14 SPS-3 test sites in Texas to determine the
effectiveness of preventive maintenance treatments. Zayed [7]
proposed an economical model to provide a rational framework
for the alternative evaluation of the paint maintenance of steel
bridges. Morcous [8] proposed an approach that combines the use
of genetic algorithms and Markov-chain models for programming
maintenance alternatives and then applied this approach to
concrete bridge decks. Metni [9] depicted the dynamics of an
unmanned aerial vehicle for monitoring of structures and
maintenance of bridges. Lorenc [10] deeply described the Robotic
Bridge Maintenance System (RBMS) developed by the Construction Automation and Robotic Laboratory (CARL). Nevertheless, only a few studies have been conducted to package
contract units for the maintenance service. For example, Tsai [11]
used constrained fuzzy c-mean clustering algorithm for determining bridge let projects.
The traditional bridge management process is summarised
in Fig. 1 [11]. The process consists of the following four steps:
(1) bridge condition survey, (2) bridge tasks determination, (3)
683
2. Literature review
Table 1
The major decision factors in engineering contract packaging
Sewerage system
Expressway
1. Engineering
characteristic
2. Time limit for a project
3. Construction resource
4. Budget
1. Requirement to open
to traffic partially
2. Small firms cultivated
3. Construction interface
4. Capacity of construction
firms
5. Work interface
6. Construction
resource
684
685
P W ; Q
Minimise
k
X
Pl r Plc CF
l1
In Eq. (1),
Pl r
n
X
wi;l
i1
Plc g
n
X
i1
p
X
2
xi; j ql; j
j1
wi;l
m
X
d xi; j ; ql; j
jp1
686
p
X
j1
j1
m
X
DPc gxvo
p
X
2
2
xv;j qo;j xva
xv;j qa;j
m
X
d xv;j ; qo;j gxva
d xv;j ; qa;j
jp1
jp1
p
X
j1
DPc g
m
X
jp1
p
X
2
2
xv;j qo;j xvs
xv; j qs;j
j1
m
X
d xv;j ; qo;j g
d xv;j ; qs;j
jp1
687
Table 3
Purity validation (entropy measure)
Cluster number
k=2
Group attribute
k=3
B
k=4
B
Material
Type
0.2958
0.4804
0.2762
0.4746
0.2933
0.5645
0.2774
0.4680
0.2921
0.5497
0.2590
0.2958
s
X
pi log1=pi :
11
i1
k
P
Table 2
The r-value of different constraints
Constraint
n N 250
n N 300
n N 330
n b 340
n b 400
0.8018
0.8100
0.8000
0.8000
0.8018
Constraint
250 b n b 340
250 b n b 400
300 b n b 340
300 b n b 400
330 b n b 340
330 b n b 4000
0.7973
0.8018
0.7973
0.8018
0.7943
0.7943
688
group (the object number is zero). This condition can purge the
contract package, promote similarity in each contract package
and, thus, improve the contractor's work efficiency. Obviously,
Table 4(b) has more zero appearances than 4 (a). Hence, CPT is
better than the traditional packaging approach.
According to the abovementioned analysis, if engineers
choose k = 3 to create contract packages, Table 5(a) shows the
entropy value of considering different conditions to group the
382 bridges. Besides considering only the bridge locations and
all the attributes of the bridge data set, it adds more conditions,
including C: no bridge type, D: no bridge material and E: no
bridge location. Table 5(b) shows the total entropy value. The
total entropy was calculated as follows:
a materialentropy b typeentropy ; a b 1
11
12
689
Table 4
The number of bridges in each cluster and their components
(a) Grouped by only bridge location; did not consider cost constraint
(b) Grouped by bridge location, cost, bridge material and type; consider cost constraint
Table 5
Sensitivity analyses for k = 3
(a) Purity validation (entropy measure) for k = 3
Grouping attribute
Material(entropy)
Type(entropy)
0.2933
0.5645
0.2774
0.4680
0.2960
0.5797
0.2774
0.4630
0.1419
0.4424
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0.5645
0.5374
0.5103
0.4831
0.4560
0.4289
0.4018
0.3747
0.3475
0.3204
0.2933
0.4680
0.4489
0.4299
0.4108
0.3918
0.3727
0.3536
0.3346
0.3155
0.2965
0.2774
0.5797
0.5514
0.5230
0.4946
0.4663
0.4379
0.4095
0.3811
0.3528
0.3244
0.2960
0.4630
0.4445
0.4259
0.4073
0.3888
0.3702
0.3517
0.3331
0.3145
0.2960
0.2774
0.4424
0.4124
0.3823
0.3522
0.3222
0.2921
0.2621
0.2320
0.2020
0.1719
0.1419
Table 6
Objective values under different cluster numbers and RV
k Average cost (1)
Objective Coefficient
value
of variation
2 380.98
3 329.19
4 290.99
12.24%
7.79%
11.78%
Objective Coefficient
value
of variation
Objective Coefficient
value
of variation
349.46
329.08
284.67
349.02
326.09
287.99
0.00%
6.96%
12.07%
0.00%
6.39%
9.58%
690
References
[1] J.S. Kong, D.M. Frangopol, Life-cycle reliability-based maintenance cost
optimisation of deteriorating structures with emphasis on bridges, ASCE
Journal of Structural Engineering 129 (6) (2003) 818828.
[2] T.F. Fwa, W.T. Chan, K.Z. Hoque, Multiobjective optimisation for
pavement maintenance programming, ASCE Journal of Transportation
Engineering 126 (5) (2000) 367374.
[3] T.M. Zayed, Budget allocation for steel bridge paint maintenance, ASCE
Journal of Performance of Constructed Facilities 18 (1) (2004) 3646.
[4] N. Bandara, M. Gunaratne, Current and future pavement maintenance
prioritisation based on rapid visual condition evaluation, ASCE Journal of
Transportation Engineering 127 (2) (2001) 116123.
[5] S. Labi, K.C. Sinha, Effectiveness of highway pavement seal coating treatments, ASCE Journal of Transportation Engineering 130 (1) (2004) 1423.
[6] D.H. Chen, D.F. Lin, H.L. Luo, Effectiveness of preventative maintenance
treatments using 14 SPS-3 sites in Texas, ASCE Journal of Performance of
Constructed Facilities 17(3) (2003) 136143.
[7] T.M. Zayed, L.M. Chang, J.D. Fricker, Life-cycle cost-based maintenance
plan for steel bridge protection systems, ASCE Journal of Performance of
Constructed Facilities 16(2) (2002) 5562.
[8] G. Morcous, Z. Lounis, Maintenance optimization of infrastructure
networks using gengtic algorithms, Elsevier Automation in Construction
14 (2005) 129142.
[9] Najib Metni, Tarek Hamel, A HAV for bridge inspection: visual servoing
control law with orientation limits, Automation in Construction 17 (2007) 310.
[10] Steven J. Lorenc, Brian E. Handlon, Leonhard E. Bernold, Development of
a robotic bridge maintenance system, Elsevier Automation in Construction
9 (2000) 251258.
[11] Y. Tsai, C.T. Yang, Constrained Fuzzy c-mean clustering algorithm for
determining bridge let projects, ASCE Journal of Computing in Civil
Engineering (2004) 215224.
[12] K.N. Michael, A note on constrained k-means algorithms, Elsevier Pattern
Recognition 33 (2000) 515519.
[13] P.S. Bradley, K.P. Bennett, A. Demiriz, Constrained k-means clustering,
MSRTR-200065, Microsoft Research, 2000.
[14] J.C. Gower, A general coefficient of similarity and some of its properties,
International Biometric Society Biometrics, 27 (1971) 857874.
[15] K.C. Gowda, E. Diday, Symbolic clustering using a new dissimilarity
measure, Elsevier Pattern Recognition, 24 (1991) 567578.
[16] A.K. Jain, R.C. Dubes, Algorithms for Clustering Data, Prentice Hall, New
Jersey, 1988.
[17] H. Ralambondraint, A conceptual version of the k-means algorithm,
Elsevier Pattern Recognition Letters, 16 (1995) 11471157.
[18] M. Lebowitz, Experiments with incremental concept formation, Kluwer
Academic Publishers Machine Learning, 2 (1987) 103138.
[19] Z. Huang, Extensions to the k-means algorithm for clustering large datasets
with categorical values, Data Mining and Knowledge Discovery, vol. 2,
Springer, Netherlands, 1998, pp. 283304.
[20] K.A. Reshaid, N. Kartam, Design-build prequalification and tendering
approach for public projects, Elsevier International Journal of Project
Management, 23 (2005) 309320.
[21] C.K. Chau, W.L. Sing, T.M. Leung, An analysis on the HVAC maintenance
contractors selection process, Elsevier Building and Environment, 38 (2003)
583591.
[22] C.H. Wong, Contractor performance prediction model for the United
Kingdom construction contractor: study of logistic regression approach,
ASCE Journal of Construction Engineering and Management, 130(5) (2004)
691698.
[23] A.M. Elazouni, D-SUB: decision support system for subcontracting construction works, ASCE Journal of Construction Engineering and Management, 126(3) (2000) 191200.
[24] M.G. Daz, B. Arruada, A. Fernndez, Causes of subcontracting: evidence from panel data on construction firms, Journal of Economic
Behavior and Organisation, 42 (2000) 167187.
[25] V. Albino, A.C. Garavelli, A neural network application to subcontractor
rating in construction firms, Elsevier International Journal of Project
Management, 16(1) (1998) 914.
[26] I.C. Yen, The Study of the tendering policy on major projects using Taipei
transit tending experiences as example, Master Thesis, Department of
Construction Engineering, National Taiwan University of Science and
Technology, 2001.
[27] N.F. Wu, Factors of contract division in Kaohsiung sewerage system,
Master Thesis, Department of Construction Engineering, National
Kaohsiung First University of Science and Technology, 2003.
[28] P.S. Wang, A study on tendering strategies for national expressway
projects in Taiwan, Master Thesis, Department of Engineering, National
Chiao Tung University, 2005.
[29] P. Berkhin, Survey of Clustering Data Mining Techniques, Accrue
Software, Inc., 2002
[30] J. Han, M. Kamber, Data Mining: Concepts and Techniques, Morgan
Kaufmann Publishers, 2001.
[31] R. Ng, J. Han, Efficient and effective clustering methods for spatial
data mining, Proc. of the 20th Conf. on VLDB, Santiago, Chile, 1994,
pp. 144155.
[32] J. Hartigan, Clustering Algorithms, John Wiley & Sons, New York, 1975.
[33] J. Hartigan, M. Wong, Algorithm AS136: a k-means clustering algorithm,
Publishing Applied Statistics, 28 (1979) 100108.
[34] A. Jain, R. Dubes, Algorithms for Clustering Data, Prentice-Hall,
Englewood Cliffs, NJ, 1988.
[35] L. Kaufman, P. Rousseeuw, Finding Groups in Data: An Introduction to
Cluster Analysis, John Wiley and Sons, New York, 1990.
[36] M. Ester, H.-P. Kriegel, J. Sander, X. Xu, A density-based algorithm for
discovering clusters in large spatial databases with noise, 2nd International Conf. on Knowledge Discovery and Data Mining (KDD-96), 1996,
pp. 226231.
[37] R. Agrawal, J. Gehrke, D. Gunopulos, P. Raghavan, Automatic subspace
clustering of high dimensional data for data mining applications, Proc.
ACM-SIGMOD Int. Conf. Management of Data (SIGMOD'98), Seattle,
WA, 1998, pp. 94105.
[38] W. Wang, J. Yang, R. Muntz, STING: a statistical information grid
approach to spatial data mining, Proc. of the 23th Conf. on VLDB, Athens,
Greece, 1997, pp. 186195.
[39] S. Goil, H. Nagesh, A. Choudhary, 1999 MAFIA: Efficient and scalable
subspace clustering for very large data sets. Technical Report No.CPDCTR-9906-010, Center for Parallel and Distributed Computing, Northwestern University Technological Institute, Evanston.
[40] A.K.H. Tung, R. Ng, L. Lakshmanan, J. Han, Constraint-based clustering
in large databases, Proc. 8th Intl Conf. on Database Theory (ICDT'01),
London, UK, 2001, pp. 405419.
[41] Y. Qian, K. Zhang, W. Lai, Constraint-based graph clustering through node
sequencing and partitioning, 8th Pacific-Asia Conference, PAKDD,
Sydney, Australia, 2004, pp. 4151.
[42] P.M. Murphy, D.W. Aha, UCI Repository of Machine Learning Databases,
1992, www.ics.uci.edu/~mlearn/MLRepository.html.
[43] J.R. Quinlan, C4.5: Programs for Machine Learning, Morgan Kaufmann
Publishers, SanMateo, Calif., 1993
[44] M.H. Dunham, Data Mining Introductory and Advanced Topics, Prentice
Hall, 2003.