Upload
jonathan-jingshen-shi
View
215
Download
2
Embed Size (px)
Citation preview
Object-oriented resource-based planning method (ORPM) forconstruction
Jonathan Jingshen Shi, Zhongming Deng
Department of Building and Construction, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong
Received 20 February 1998; received in revised form 20 August 1998; accepted 5 January 1999
Abstract
This research intends to develop a construction scheduling and planning method, named Object-oriented Resource-based
Planning Method (ORPM), for meeting the di�erent requirements at various planning stages. Object-oriented representation isadopted for modelling construction activities. Each object has attribute values to detail the required conditions to construct theactivity, such as logical dependency, and resource demands. The planner can de®ne a set of constraints as the active planningconstraints based on the planning stage and availability of project information. The planning process checks the active
constraints against actual availability of resources for each activity. An activity cannot be scheduled for construction until all ofthe active constraints are satis®ed. At pre-construction stage, less constraints such as technological dependency and resourcecapacity can be selected into the active planning constraint set; and more constraints can be selected at the construction
planning stage when more detailed project information is available. The planning mechanism is demonstrated for variousplanning purposes through an example project. # 2000 Elsevier Science Ltd and IPMA. All rights reserved.
Keywords: Planning; Scheduling; Critical path method (CPM); Program evaluation and review technique (PERT); Construction resource plan-
ning
1. Background
Construction is one of the largest industries in manycountries. However, the industry has been experiencingsuch problems as cost overrun, delayed completion,disputes and even litigation. Scheduling and planningare identi®ed among the top potential areas needingimprovements.[1]
For its simplicity, the bar chart is still the predomi-nant scheduling method in the industry, especially forshort-term planning.[2] The main drawback of a barchart is that it cannot detail the logical interrelation-ships among activities. If an activity is behind sche-dule, it may not be possible to ascertain the e�ect ofsuch delay on the completion of the project.
Network models can overcome the drawbacks of abar chart by detailing activity interrelationships. CPM(Critical Path Method), to a less extent PERT (Pro-gram Evaluation and Review Technique), provide the
most widely used network models. Many successful ap-plications have been reported.[3±5] In the mean time,many drawbacks have also been identi®ed.[6±12] Someof these criticisms are:
. CPM considers only logical constraints during plan-ning, which is not the real world of a constructionprocess. In practice, the logical constraint is onlyone of many conditions to determine whether an ac-tivity can be scheduled for construction or not.
. CPM treats the construction of an activity as a non-stop process, i.e., FT (®nish time)=ST (starttime)+D (activity's duration). In practice, an activi-ty's construction may be interrupted if its requiredresources or other conditions cannot be met.
. CPM lacks the ability to model the projects as astepwise, dynamic decision process.
Research has been conducted by addressing thesedrawbacks in order to enhance CPM-based techniques.
International Journal of Project Management 18 (2000) 179±188
0263-7863/00/$20.00 # 2000 Elsevier Science Ltd and IPMA. All rights reserved.
PII: S0263-7863(99 )00013 -7
www.elsevier.com/locate/ijproman
Enhanced CPM based scheduling methods were pre-sented, incorporating resource capacity into the sche-duling process.[11,13] Project resource constraint issueshave been studied extensively.[11,14±16,17±19] Re-de®-nition and re-calculation of ¯oats were discussed at thetime of incorporating resources into scheduling.[20±23]
On the other hand, new planning and schedulingmethods have also been studied. Russell and Wong[24]presented a planning method by integrating CPM andlinear scheduling methods. Pultar[25] developed a pro-gress-based scheduling technique based on progresscharts and bar charts to overcome the problem of frag-mentation of activities during the application of theconventional CPM to construction projects. Espedal etal.[8] discussed TOPP project planning concepts, whichintends to ®nd a schedule that gives the highestexpected life cycle pro®tability given a set of restric-tions. Jaafari[26,27] proposed the Time And PriorityAllocation Scheduling (TAPAS) method. TAPAStreats each activity as `critical' and of equal priorityunless there are reasons to the contrary, such asresource or economic constraints. It can generate theproject plan without a network and be integrated withrisk management and earned value techniques.
Construction process simulation has been progres-sing very fast in the past two decades with theadvancement of computer technology sinceCYCLONE was introduced by Halpin[34,35] in 1977.It is an e�ective planning tool which allows the user toinvestigate the detailed operations of a constructionprocess with a dynamic and stochastic nature. Becauseof the di�culties and required knowledge in usingsimulation, research e�orts are still needed to makesimulation an applicable planning tool, especially atproject level.[29]
Knowledge-based systems have been studied since1980's with the intention of enhancing constructionplanning[30,28,31] in some or all of the followingaspects:[32] representation of the project, generation ofactivities, determination of dependencies between ac-tivities, resource allocation and scheduling of activities.However, no applicable system has been reported yet.
In summary, the current planning methods in theconstruction industry are bar charts and CPM-basedtechniques despite their criticisms and drawbacks. Thisresearch presents a new methodÐObject-orientedResource-based Planning Method (ORPM). ORPMprovides an object-oriented representation for model-ling construction activities. A set of constraints (e.g.,technological dependencies and resource demands) canbe selected by the user as the criteria for determiningwhether an activity can be scheduled for constructionor not. The planning process checks the satisfaction ofthe selected constraints. An activity cannot be sched-uled for construction until all of its constraints are sat-is®ed. With the ¯exible customer-tailored set of
constraints, ORPM is ¯exible for di�erent require-ments at various stages of a construction project. Itshould be mentioned that ORPM is in the process ofbeing implemented into a computer system. The paperaddresses the planning method.
2. Construction plans and resource constraint patterns
Both preconstruction and construction plans arewidely used in the industry. The preconstruction planof a project considers timing of works, work schedulesand resource requirements before the project starts.The construction plan, similar to the short-term plan,incorporates the progress information in constructionprocess to update the original plan in order to guidethe works on site.
At pre-construction planning stage, the detailed pro-ject information, such as equipment, materials andlabour, is usually unknown except for certain brief®gures like maximum available units of resources. Thepre-construction plan includes target schedule, mile-stones, and resource pro®le. The common resourceconstraints are: 1) The total demand should not exceedthe maximum available units, and 2) the demandshould be levelled.
At the construction stage, more detailed project in-formation becomes available. The construction plan-ning incorporates the available information andconstruction progress information to update the orig-inal schedule. Resource constraints can be detailedwith real time, for instance, one crane in week 2, andthree cranes in week 5.
A uni®ed resource constraint pattern can be de®nedfor both planning stages, mathematically,
Ri�t�Rri�t� 1RtRT �1�where i is the ith resource; t is current period; T is thetotal duration; ri(t ) is the maximum available unit ofresource i at time t; and Ri(t ) is the demand forresource i at time t.
ri(t ) is a constant value at pre-construction stageand is represented by the maximum available unit. It isthe function of time at construction planning stageand can be given by the actual availability pro®lealong project duration.
3. Object-oriented representation of activities
A construction project can be represented as aninterrelated collection of activities through work break-down structure.[33] An activity may represent di�erentlevel of information at di�erent planning stages. Forinstance, it may indicate a work package at pre-con-
J.J. Shi, Z. Deng / International Journal of Project Management 18 (2000) 179±188180
struction stage. Therefore, activities can be further bro-ken down into lower levels. This research would liketo focus on the planning method and activities treatedas the basic planning elements. The hierarchical rep-resentation of activities will be left to future researchto enhance the ORPM method.
Although an activity is always di�erent from theothers, it has common features as well: certain con-straints must be met before construction can start; anactivity consumes both time and resources. A generalobject-oriented data structure can be developed to rep-resent an activity as in Eq. (2).
Activity fD,T,P,R,C,A,W g �2�
where,
. D=Activity duration.
. T=Technological dependence.
. P=Priority of the activity with level (1, 2, 3, . . . ). Itde®nes the priority of an activity competing for lim-ited resources.
. R=Resource demands. The availability of requiredresources and an activity's priority for resourcesdetermine whether an activity's demands can be sat-is®ed or not. Availability information of resources isprovided in a project database at project level.
. C=Contract or special planning requirements (cer-tain activities have to start/complete on speci®cdates). Contrasting the scheduled dates to therequired dates of an activity can determine the satis-faction of these requirements.
. A = Activity characteristics (e.g., some activitieshave to be continuously constructed). In a CPMplan, all activities are presumed to be constructedwithout interruption. Interrupting non-essential ac-tivities and giving priorities to essential ones canreduce project duration and/or reduce peak resourcerequirements.
. W = Required weather condition. This informationcan be used for monitoring and controlling the con-struction progress, especially for delay analysis.
Among these parameters, T, P, R, C, and A func-tion as constraints determining whether an activity canbe scheduled for construction. To evaluate resourceconstraints, the availability information of resources isneeded.
Technological dependence can be represented as inCPM techniques with either FSЮnish to start orSSÐstart to start, i.e.
T � fFS,SSg �3�
Furthermore, resources can be broken down intolower level objects consisting of equipment (E ), labour(L ), and material (M ) as in Eq. (4).
R � fE,L,M g �4�The objects can be supplied with information as to
their attribute values, which can be updated with con-struction progress. Updated information drives theplan in subsequent update. An example activity objectis illustrated in Fig. 1.
4. ORPM for construction planning
If project information cannot be detailed, forinstance at the pre-construction stage, the attributes ofan activity's object cannot be fully de®ned, or the pro-ject database cannot provide detailed resource infor-mation. Then, not all of the previously discussedconstraints can be applied for scheduling. Forexample, a concreting activity needs three trucks, butit may not be clear how many trucks will be availableat the early planning stage. In order to provide a ¯ex-ible planning method which can generate a relevantplan corresponding to the available project infor-mation, a set of constraints Pp is de®ned as the activeplanning constraints in ORPM. Pp is selected from theentire constraints {T,P,R,C,A } by the planner basedon the availability of project information, i.e.,
Pp 2 fT,P,R,C,Ag �5�Less items can be selected at the pre-construction
stage when there is no detailed project information;otherwise more items can be selected for constructionplanning.
The scheduling process is period by period andadvances by one time period at one time. The unit oftime period can be an hour, day, week, or month as
Fig. 1. Example activity object.
J.J. Shi, Z. Deng / International Journal of Project Management 18 (2000) 179±188 181
long as it is consistent in a project. Day is the mostcommon time period in construction planning, and isused thereafter.
Starting from day 1, all activities are analysed identi-cally. Each of them can be at one of the three states:(a) completed; (b) on-going; and (c) not started. Both(b) and (c) are uncompleted activities, which arescanned against the selected active constraints. Activi-ties satisfying all active constraints constitute a feasibleset of activities for construction de®ned as{A1,A2, . . . ,An }, from which activities are selected forconstruction based on their priorities and are allocatedwith needed resources until no resources are availableor no activity can be selected. If the feasible set isempty, i.e., no activity can be scheduled at day i, theschedule period will be increased by 1, i.e., day iCdayi+1. The feasible activity selection process is similar as
in TAPAS.[27] Interested readers can refer to it formore detailed information. The project duration isobtained if all activities have been completed. The pro-cess is illustrated in Fig. 2.
Resource allocation is not addressed in this research.Instead, conventional heuristic approaches areemployed to select activities for construction from thefeasible set of activities {A1,A2, . . . ,An } as follows, butnot limited to those:
. Activities which have started and require continuitymust be selected;
. Activities should be selected based on the order ofpriorities;
. Resource utilisation should be as smooth as poss-ible, mathematically,
minfjRi�t� ÿ Ri�tÿ 1�jg �6�
. The utilisation of resources should be maximised,mathematically,
minfjRi�t� ÿ ri�t�jg �7�
ORPM is di�erent from the CPM approach inwhich technological dependency is the only planningconstraint, resource demands are attached to a sche-dule, and an activity's construction is a no-stop pro-cess. Obviously, ORPM has advantages in all of theseaspects over critical path method (CPM).
4.1. Pre-construction planning using ORPM
Because project information cannot be detailed atthis stage, the common active constraints are {T,R,A },and resource constraints can be re-de®ned as:
Ri�t�Rri 1RtRT �8�where Ri(t ) is the total demand for resource i at timeperiod t; and ri is the maximum available unit ofresource i.
If only technological dependencies are active, theresulting schedule is comparable to the one obtainedfrom CPM computations. Accordingly a resource pro-®le can be obtained by detailing resource requirementsalong the project duration. If the schedule is treated asa target schedule, the required resources must be avail-able on schedule in order to complete the project onschedule; otherwise, the project may not be able tocomplete as scheduled. The target schedule can beupdated as necessary similarly to a CPM target sche-dule.
Moreover, according to the purchase or renting pol-icy of resources (e.g., lag time between sending pur-Fig. 2. ORPM planning process.
J.J. Shi, Z. Deng / International Journal of Project Management 18 (2000) 179±188182
chase orders and receiving materials on site), a detailedpurchase or renting schedule can be generated (e.g.,when and how much a material should be ordered).This information will be stored in the project databaseassociated with the target schedule.
4.2. Construction planning using ORPM
At the construction planning stage, project infor-mation becomes more detailed with construction pro-gress. Another set of constraints can be selected from{T,P,R,C,A } as the active planning constraints for thispurpose, i.e.,
Pr 2 fT,P,R,C,Ag �9�The planning mechanism is the same as pre-con-
struction planning with the di�erence as illustrated inFig. 3.
The project status and resource information areupdated with construction progress. The project data-base can detail the available time and units of equip-ment, the quantity and delivery date of materials alongproject duration.
ORPM can easily incorporate construction changesinto construction planning, such as variations in con-struction productivity (remaining duration), changedresource demands, and even changed logical relation-
ships. All of these changes can be substantiated bymodifying relevant attribute values of activity objects.
5. Example project
To illustrate the ORPM planning method, a simpleproject is planned with the network shown in Fig. 4and activity data in Table 1.
With CPM as the planning method without resourceconstraints, the project schedule is summarised inTable 2.
5.1. Planning project using ORPM with technologicalconstraints only
Working day: Day =1.Only activity A can start.Day=2.Activity A is on-going with 4 days remaining. No
other activities can start. Activity A is to be completedat Day 5.
Advance time, Day=6.Activity A is completed. Logically eligible activities
for construction are Activities B, E and H. Becauselogic is the only active constraint, start all of them,and their ®nishes are 15, 19 and 17 respectively.
Advance time, Day=16.Activity B is completed. Activities E and H are on-
going. Logically eligible activity for construction is C.Start activity C on Day 16 and will complete it on day30.
Repeat above process. The planning results can besummarised in Table 3 for the entire project.
The project duration is 60 days, which equals theone obtained from CPM computation. The scheduledstart and ®nish of each activity are the same as theearly schedules in CPM (see Table 2).
Fig. 3. Real-time planning mechanism.
Fig. 4. Project network.
J.J. Shi, Z. Deng / International Journal of Project Management 18 (2000) 179±188 183
5.2. Add resource capacities into active constraints
If both technological constraints and resource ca-pacities are selected as the active planning constraints,the planning method and process are the same asabove but with di�erent results detailed as:
Day=1.
Only activity A can start, and A will be completedat Day 5.
Advance time, Day=6.
Activity A is completed. Activities feasible to startare B, E and H. If all of them start, total resourcedemands will exceed their availability. Therefore, onlysome of them can be selected to start. Many methodshave been studied to assist in the selection.[11,17,36]The rule adopted here is to maximize the utilisation ofresources based on the order of their priorities. Thethree resources are weighted with di�erent priorities:
R1 (high), R2 (medium) and R3 (low). Five optionsare available for selection and are summarised inTable 4.
Based on the optimisation objective, activities E andH (Option 5) are selected to start. The entire planningprocess is summarised in Table 5. The project durationis 110 days.
5.3. Allow interruptions for on-going activities
On a construction site, it is very common to inter-rupt some non-essential activities and to give prioritiesto other activities. ORPM provides the ¯exibility toadd one more constraint into the active planning con-straint set, i.e., to allow interruptions for on-going ac-tivities.
For the planning example discussed above (seeTable 5), there is only one on-going activity J betweenDay 42 and Day 51 with resource demands 0, 5, and 3for resources R1, R2 and R3 respectively. The man-
Table 2
CPM schedule
Activity Duration Early start Early ®nish Late start Late ®nish Total ¯oat Free ¯oat Critical
A 5 1 5 1 5 0 0 �
B 10 6 15 12 21 6 0
C 15 16 30 22 36 6 0
D 18 31 48 37 54 6 6
E 14 6 19 6 19 0 0 �
F 16 20 35 20 35 0 0 �
G 19 36 54 36 54 0 0 �
H 12 6 17 9 20 3 0
I 14 18 31 21 34 3 0
J 20 32 51 35 54 3 3
K 6 55 60 55 60 0 0 �
Table 3
Planning results using ORPM
Activities to Completion dates of
Day start scheduled activities
1 A 5
6 B,E,H 15, 19 and 17
16 Ea,Ha,C 19, 17 and 30
18 Ca,Ea,I 30, 19 and 31
20 Ca,Ia,F 30, 31 and 35
31 Ia,Fa,D 31, 35 and 48
32 Fa,Da,J 35, 48 and 51
36 Da,Ja,G 48, 51 and 54
49 Ja,Ga 51 and 54
52 Ga 54
55 K 60
a On-going activities.
Table 1
Activity information
Resource requirementa
Activity Duration R1 R2 R3 Technological
dependencies
A 5 0 5 6 ±
B 10 1 3 0 A
C 15 0 5 2 B
D 18 1 4 3 C
E 14 0 3 2 A
F 16 1 4 4 E
G 19 0 5 6 F
H 12 1 4 3 A
I 14 1 5 2 H
J 20 0 5 3 I
K 6 0 5 6 D, G, J
a Maximum resource capacity: R1R 1; R2R 9; R3R 6.
J.J. Shi, Z. Deng / International Journal of Project Management 18 (2000) 179±188184
Table 4
Select activities for start at Day 6
Resource requirement
Option Selected activities R1 R2 R3 Remarks
1 B 1 3 0
2 E 0 3 2
3 H 1 4 3
4 B,E 1 6 2
5 E,H 1 7 5 Selected option
6 B,E,H 2 10 5 Not feasible
Table 5
Summary schedule results with resource constraints
Resource requirement
Day Eligible activities Activities for construction Completion time of scheduled activity R1 R2 R3
1 A A 5 0 5 6
6 B;E;H E and H 19 and 17 1 7 5
18 Ea;B;I Ea and I 19 and 31 1 8 4
20 Ia;B;F Ia 31 1 5 2
32 B;F;J B and J 41 and 51 1 8 3
42 Ja;C;F Ja 51 0 5 3
52 C;F C and F 66 and 67 1 9 6
67 Fa;D Fa 67 1 4 4
68 D;G D 85 1 4 3
86 G G 104 0 5 6
105 K K 110 0 5 6
a On-going activities.
Table 6
Summary schedule results with interruptions
Resource requirement
Day Eligible activities Scheduling decision Scheduled activity completion time R1 R2 R3
1 A A 5 0 5 6
6 B;E;H E and H 19 and 17 1 7 5
18 Ea;B;I Ea and I 19 and 31 1 8 4
20 Ia;B;F Ia 31 1 5 2
32 B;F;J1 B and J1 41 and 41 1 8 3
42 C;F C and F 56 and 57 1 9 6
57 Fa;D;J2 Fa 57 1 4 4
58 D;G;J2 D;J2 75 and 75 1 9 6
76 G G 94 0 5 6
95 K K 100 0 5 6
a On-going activities; J1 (duration 2) and J2 (duration 18) are the parts of activity J construct before and after the interruption.
J.J. Shi, Z. Deng / International Journal of Project Management 18 (2000) 179±188 185
agement feels that other activities, such as activities Cand F are more important and should be completed asearly as possible. In the mean time, activity J can beinterrupted. Therefore, activity J is adjourned sinceDay 42. Instead activities C and F are selected for con-struction. The planning results are summarised inTable 6.
The resulting project duration is 100 days, a 10-dayreduction comparing to non-interruption schedule inTable 5. Moreover, resources are more e�ciently uti-lised and critical activities are scheduled earlier.
5.4. Construction planning
As the project progresses, more detailed informationbecomes available. It is assumed that the project hasbeen on the way for 23 days. The project status is thatonly three activities, namely A, E and H are completedat Day 23. The reason is due to the overestimate ofproductivity at the preconstruction stage. The actualduration for activities A, E and H is 6, 17 and 14 daysrespectively. Activity durations are adjusted accordingto the actual productivity, and are listed in Table 7.
Technological relationships between activities anddaily resource requirements of activities are assumedunchanged for this sample project. The updatedresource availability information is detailed in Table 8.
With the updated information, the project can be re-scheduled by following the same process as above, andthe results are summarized in Table 9.
Comparing the results to the ones in Table 6, theproject duration is increased from 100 days to 116days, i.e., an overrun of 16 days.
6. Conclusion
This paper has presented an object-orientedresource-based planning method (ORPM), whichincludes an object-oriented representation for construc-tion activities and a planning mechanism for variousplanning stages. Based on the availability of project in-formation, a set of constraints can be selected as theactive planning constraints. If only technologicaldependency is selected, for instance, ORPM-basedschedule is the same as the early schedule from CPMcomputation. On the other hand, if detailed project in-formation is available at the construction stage, a cor-responding practical construction plan can be
Table 7
Revised activity data
Activity Original duration Revised duration
A 5 6
B 10 12
C 15 18
D 18 20
E 14 17
F 16 18
G 19 22
H 12 14
I 14 16
J 20 24
K 6 8
Table 8
Actual resource availability
Period R1 R2 R3
1-40 1 7 5
41- 1 10 7
Table 9
Real time schedule results
Resource requirement
T Eligible activities Activities for construction Completion time of scheduled activities R1 R2 R3
1 A A 6 0 5 6
7 B;E;H E and H 23 and 20 1 7 5
21 Ea;B;I Ea and B 23 and 32 1 6 2
24 Ba;F;I Ba 32 1 3 0
33 C;F;I I 48 1 5 2
41 Ia;C;F Ia and C 48 and 58 1 10 4
49 Ca;F;J Ca and F 58 and 66 1 9 6
59 Fa;J;D Fa and J 66 and 82 1 9 7
67 Ja;D;G Ja and D 82 and 86 1 9 6
83 Da;G Da 86 1 4 3
87 G G 108 0 5 6
109 K K 116 0 5 6
a On-going activities.
J.J. Shi, Z. Deng / International Journal of Project Management 18 (2000) 179±188186
generated using ORPM with the updated information.Interruptions on activities are allowed in the planningprocess. The example shows that interrupting a non-essential activity can reduce the project duration by 10days and improve the utilisation of resources.
This planning methodology can be incorporatedwith monitoring, control and delay analysis functions.The research endeavour is currently being undertaken.In the meantime, the methodology is in the process ofbeing implemented into a computer system.
References
[1] Koehn E, Caplan S. Work improvement data for small and
medium size contractors. J Constr Engrg and Mgmt, ASCE
1987;113(2):327±39.
[2] Mawdesley M, Askew W, O'Reilly M. Planning and controlling
construction projects. Longman, 1997.
[3] Cohenca D, Laufer A, Ledbetter WB. Factors a�ecting con-
struction planning e�orts. J Constr Engrg and Mgmt, ASCE
1989;115(1):70±87.
[4] McCough EH. Scheduling: e�ective methods and techniques. J
Constr Div, ASCE 1982;108(1):75±84.
[5] Paulson BC. Project planning and scheduling: uni®ed approach.
J Constr Div, ASCE 1973;99(1):45±58.
[6] Badiru AB. Project management tools for engineering manage-
ment professionals. Georgia, USA: Norcross, 1991.
[7] Bent JA, Thuman A. Project Management for engineering and
construction. New York: Fairmont Press, 1989.
[8] Espedal R, Hetland PW, Jordander I. TOPP: a new project
planning concept. Int J Project Management 1992;10(2):102±5.
[9] Jaafari A. Criticism of C.P.M. for project planning analysis. J
Constr Engrg and Mgmt, ASCE 1984;110(2):222±33.
[10] Lewis JP. Project planning, scheduling and control a hand-on
guide to bringing projects in on time and on budget. Chicago:
Probus Publishing Company, 1991.
[11] Matthews M. Resource scheduling: incorporating capacity into
schedule construction. Project Management Journal
1994;25(2):44±54.
[12] Thrush KB, Diekmann JE, Wilson TA. Project control in design
engineering. Cost Engineering 1987;29(3):14±9.
[13] Nkasu MM. COMSARS: a computer-sequencing approach to
multiresourceconstrained scheduling. Int J of Project
Management 1994;12(3):183±92.
[14] Demenlemeester E, Herroelen W. A branch-and-bound pro-
cedure for the multiple resource-constrained project schedule
problem. Management Science 1992;38:1803±18.
[15] Gemmill DD, Tsai YW. Using a simulated annealing algorithm
to schedule activities of resource constrained project. Project
management Journal 1997;28(4):8±20.
[16] Lorterapong P. A fuzzy heuristic method for resource-con-
strained project scheduling. Project Management Journal
1994;00(0):12±8.
[17] Moselhi O, Lorterapong P. Near optimal solution for resource-
constrained scheduling problems. Construction Management
and Economics 1993;11:293±303.
[18] Ozdamar L, Ulusoy G. A survey on the resource-constrained
project scheduling problem. IIE Transactions 1995;27:574±86.
[19] Patterson J. A comparison of exact procedures for solving the
multiple constrained resource project scheduling problem.
Management Science 1984;30:854±67.
[20] Bowers JA. Criticality in resource constrained networks.
Journal of the Operational Research Society 1995;46(1):80±91.
[21] Gong D, Rowings Jr. JE. Calculation of safe ¯oat use in risk-
analysis-oriented network scheduling. Int J Project Management
1995;13(3):187±94.
[22] Gong D. Optimization of ¯oat use in risk analysis-based net-
work scheduling. Int J Project Management 1997;15(3):187±92.
[23] Raz T, Marshall B. E�ect of resource constraints on ¯oat calcu-
lations in project networks. Int J of Project Management
1996;14(4):241±8.
[24] Russell AD, Wong WCM. New generation of planning struc-
tures. Journal of Construction Engineering and Management
1993;119(2):196±214.
[25] Pultar M. Progress-based construction scheduling. J Constr
Engrg and Mgmt, ASCE 1990;116(4):671±88.
[26] Jaafari A. Real time planning and total risk management. Paper
presented to presentation to the ASCE Third Congress on
Computing in Civil Engineering, June 17±19, 1996, Anaheim,
CA, USA.
[27] Jaafari A. Time and priority allocation scheduling technique for
projects. Int J of Project Management 1996;14(5):289±99.
[28] Echeverry D, Ibbs CW, Kim S. Sequencing knowledge for con-
struction scheduling. J Constr Engrg and Mgmt, ASCE
1992;117(1):118±30.
[29] Shi J, AbouRizk S. Resource-based modeling for construction
simulation. J Constr Engrg and Mgmt, ASCE 1997;123(1):26±
33.
[30] De La Garza J, Ibbs CW. Knowledge elicitation study in con-
struction scheduling domain. J Computing in Civ Engrg, ASCE
1988;4(2):135±54.
[31] Levitt RE, Kartam NA, Kuntz JC. Arti®cial intelligence tech-
niques for generation of construction plans. J Constr Engrg and
Mgmt, ASCE 1988;114(3):329±43.
[32] Shaked O, Warszanwski A. Knowledge-based system for con-
struction planning of high-rise buildings. J Constr Engrg and
Mgmt, ASCE 1995;121(2):172±82.
[33] Halpin DW, Riggs LS. Planning and analysis of construction
operations. John Wiley, 1992.
[34] Halpin DW. CYCLONEÐa method for modeling job site pro-
cesses. J Constr Div, ASCE 1977;103(3):489±99.
[35] Halpin DW. MicroCYCLONE User's manual. Division of
Construction Engineering and Management. Purdue University,
West Lafayette, Indiana, 1990.
[36] Morse LC, McIntosh JO, Whitehouse GE. Using combinations
of heuristics to schedule activities of constrained multiple
resource projects. Project Management Journal 1996;00:34±40.
Dr. J. Shi is an Associate Professor in the Department of Building
and Construction at City University of Hong Kong. He obtained his
Ph.D. from the University of Alberta in 1995. He specializes in con-
struction simulation, construction planning, neural networks and
general computer applications in civil engineering. He has published
more than 40 technical papers in recent years, 20 of which in presti-
gious journals. Currently, Dr. Shi is a Principle Investigator for ®ve
major research projects funded by various government organisations
and individual private companies.
J.J. Shi, Z. Deng / International Journal of Project Management 18 (2000) 179±188 187
Dr Michael Z. M. Deng has been an assistant professor at the
Department of Building and Construction, City University of Hong
Kong since late 1994. He received a BEng from Hehai University,
China in 1983 and a PhD in Construction Management from the
University of New South Wales, Australia in 1995 Before starting
his teaching career, he was an engineer of the Yangtze Valley Plan-
ning Commission, China and was involved in the design of the
Three Gorges Project from 1983 to 1986. His main research interests
are: construction/project management, computer applications in con-
struction, lean construction, system dynamics.
J.J. Shi, Z. Deng / International Journal of Project Management 18 (2000) 179±188188