ETtt)~etpqotctm] "Epea)vct / Oper at i onal Research. An Int ernat i onal J oumal . Vol. 6, No 3 (2006), pp.

255-269
Multiobjective Optimization
in Linear Repetitive Project Scheduling 1
Pa n d e l i s G. I p s i l a n d i s
Professor, Department of Project Management, Technological Educational Institute,
41110 Larissa, Greece.
Abstract
The Critical Path Method (CPM) and the Repetitive Scheduling Method (RSM) are the most
often used tools for the planning, scheduling and control Linear Repetitive Projects (LRPs).
CPM focuses mostly on project' s duration and critical activities, while RSM focuses on
resource continuity. In this paper we present a linear programming approach to address the
multi objective nature of decisions construction managers face in scheduling LRPs. The Multi
Objective Linear Programming model (MOLP-LRP) is a parametric model that can optimize a
schedule in terms of duration, work-breaks, unit completion time and respective costs, while at
the same time the LP range sensitivity analysis can provide useful information regarding cost
tradeoffs between delay, work-break and unit delivery costs. MOLPS-LRP can generate
alternative schedules based on the relative magnitude and importance of different cost
elements. In this sense it provides managers with the capability to consider alternative
schedules besides those defined by minimum duration (CPM) or minimum resource work-
breaks (RSM). Demonstrative results and analysis are provided through a well known in the
literature case study example.
Keywords: Linear Projects, Linear Programming, Scheduling
1. I ntroducti on
Li near Repet i t i ve construction Projects (LRPs) consist of a set of activities that are
repeat ed sequentially at different locations or proj ect units. The activities fol l ow a
logical and technological driven sequence descri bed by time or distance constraints
for the entire life span of the proj ect [Kallantzis and Lambropoul os (2004)]. Because
LRP resources are used in a sequential manner, effect i ve resource management is
ver y important bot h in terms o f project cost and duration. Al so, because of the
division of the proj ect to individual units meet i ng intermediate unit del i very times is
another important issue in LRPs.
The Critical Path Met hod (CPM) is most oft en used for the planning, scheduling and
control of such projects. Nonetheless, net work analysis techniques are duration
1 This paper is part o f a research proj ect of the ARCHIMEDES programme o f the Operational
Programme f or Education and lnitial Vocational Training in Greece under the 3 r" Common Support
Framework and is 75% co-funded by the European Social Fund (ESF).
256 Operat i onal Research. An Int ernat i onal Jour nal / Vol. 6, No. 3 / Sept ember - December 2006
oriented and cannot sufficiently address resource management issues. Other more
suitable techniques that focus on resource usage, such as the Repetitive Scheduling
Method (RSM) have been developed for scheduling and controlling repetitive
projects.
However the compl ex nature of repetitive projects calls for decisions that cannot be
based solely on time or resource management but require a more holistic approach. In
this paper we explore the multi objective nature of decision maki ng in repetitive
construction projects and present a linear programmi ng parametric model formulation
for supporting the decisions of construction managers. The rest of the paper is
organized as follows: The following section refers to definitions, literature review and
a classification of linear / repetitive projects. It also includes a critical presentation of
the RSM. In section 3 we propose a Linear Programming formulation for modelling
LRPs, which is designed in a parametric way such as to allow single or multiple
objective optimization regarding time or cost elements of the project. Section 4
follows with demonstrative results based on the application of the LP model to a well
known literature example under different sets of assumptions and parameters. Finally,
section 5 contains the conclusions and proposed directions of future research.
2. L i near- Repetitive Projects
During the last decades various terms have been used in the literature, to describe
mainly construction projects that can be divided in a number of units consisting of the
same project activities, which after their completion in one unit are repeated in the
next one [Mattila and Abraham (1998)]. The most popular among t hem is Linear and
Repetitive Projects (LRPs), where the term linear refers to construction projects the
activities of whi ch are repeated continuously in a horizontal flow, such as in the case
of segments of highways, bridges, tunnels, railways, pipelines, sewers, etc. while the
term repetitive describes construction projects where activities are repeated in discrete
repeating units as in the case high-rise and multistory buildings, multi-housing
projects, etc. many times in a vertical direction.
However the not i on of LRP can be extended beyond the construction industry since
other types of projects can be classified as LRPs as they have similar characteristics.
Typical examples are reengineering projects and projects involving the
implementation of new or upgrading of comput i ng systems where the same tasks are
taking place sequentially to a series of branches or units of large organizations at he
same or different geographical locations.
Different met hods have been proposed for planning, scheduling and controlling the
construction process of these types of projects. The devel opment of Repetitive
Scheduling Met hod (RSM) by Harris and Ioannou (1998) and Yang (2002a) whi ch
can be i mpl ement for both project categories, is the milestone for classifying these
projects into a unique category as LRPs.
P. Ipsilandis / Multiobjective Optimization in Linear Repetitive Project Scheduling 257
2.1 Scheduling methodologies for LRPs
The most frequently used method for design, planning, scheduling and control of
construction projects is the Critical Path Method (CPM) [Mattila and Abraham,
(1998)]. Nevertheless, since the seventies researchers have been challenging its
applicability in an attempt to prove its inadequacies, while at the same time propose
other approaches and methods more suitable for the construction process [Peer
(1974); Carr and Mayer (1974); Dressler 1974; O' Brien (1975); Ashley (1980);
Selinger (1980); Birrell (1980); Johnston (1981); Stradal and Cacha (1982); Russell
and Caselton (1988); Reda (1990); Harmelink and Rowings (1998); Harris and
Ioannou (1998)]. The main reasons that instigated the development of new
approaches were the insufficiency of network analysis to describe the repetitive
nature of the construction process and to allocate and planning the work in the various
construction sites, its weakness to provide uninterrupted utilization (work continuity)
of resources (work teams) and minimization of the idle time, and also the weakness to
model the learning effect from the work repetition. Moreover, the arbitrary estimation
of activities' duration instead of the estimation of production rate, the large number of
activities that are required in the network for large and complicate projects and the
lack of information contained in the diagram are also considered significant
weaknesses of the CPM [Birrell (1980], since emphasis is given to optimizing project
completion time and not into the optimization and utilization of the resources and the
cost control, which is the main concern with great importance for construction
contractors.
The alternative approaches that have been developed for LRPs are based mainly on
graphical methods on X - Y diagrams, where one axis represents time and the other
work progress. These methods can be organized, according to Mattila and Abraham.
(1998), in three basic categories: i) those that are based on the Line-of-Balance (LOB)
technique, applied mainly to discrete projects, ii) those that are based on Linear
Scheduling Method (LSM), more appropriate for continuous projects and iii) those
that combine the previous techniques with others such as dynamic programming,
stochastic programming, linear programming, simulation etc.
While, all of these techniques achieved to overcome the inadequacies of PERT/CPM,
they failed to obtain wide acceptance and practical use. The main reasons are because,
they have been more complicated than they should have been while there is not an
acceptable algorithm (process) for identifying the project' s critical path and
subsequently the project completion time [Harris and Ioannou (1998); Harmelink and
Rowings (1998); Kallantzis and Lambropoulos, (2003)]. The lack of a critical path
identification algorithm can be attributed to the fact that in all these methods the
underlying assumptions is that all activities must considered as critical, for better
control of a project [Peer (1974)]. Significant efforts however have been made the last
few years towards the development of an acceptable and accurate algorithm to
258 Operational Research. An International Journal / Vol.6, No.3 / September - December 2006
determine the critical path in LRPs [Harmelink and Rowi ngs (1998); Harmelink
(2001); Mattila and Park (2003); Kallantzis and Lambropoulos (2003); Kallantzis and
Lambropoulos (2004)].
The Repetitive Scheduling met hod (RSM) was introduced by Harris and Ioannou
(1998), and was further developed with work that followed [Yang and Ioannou
(2001); Yang (2002a); Yang (2002b); Ioannou and Harris (2003); Ioannou and Yang
(2003); Ioannou and Yang (2004); Yang and Ioannou (2004)]. It is primarily a
graphical representation of the project on an X-Y diagram and its objective is to
integrate existing methods into a generalized one that ensure continuous resource
utilization. RSM can be used for the scheduling of both discrete and continuous
projects. For discrete project the repeated units (work progress) are usually drawn in
the Y axis and the elapsed project time on the X axis is the time, while for continuous
projects the time is drawn on the Y axis and the repeated units in x axis.
RSM uses a pull-system approach, where the finish-time of the predecessor activity is
pulled forward in order to meet the start date of the successor in order to achieve
work continuity and uninterrupted resource utilization, in contrast with the CPM
push-system, where the start of every activity is pushed in time to maintain the
precedence relationships with its predecessors. The objective in RSM is not the
minimization of the project completion time but achieving work continuity which
leads to mi ni mi zi ng the overall project cost. In construction projects, the
minimization of the cost may be more desirable than the reduction of the project
duration [Yang (2002b)].
The RSM algorithm has been programmed into comput er implementation, the
Repetitive Project Planner (RP2) [Yang (2002a)]. The algorithm involves two stages:
The first stage is similar to the forward pass computations of CPM and results in the
computation mi ni mum project duration. In the second stage each continuity
relationship ' pul l s' the predecessors to eliminate the time gap with the successor to
ensure work continuity.
Although RSM optimizes for continuity its analysis features are limited. It can only
allow trade offs between time gaps and project duration on a trial and error basis.
Cost considerations and other control variables are not taken directly into
consideration in the computation, but only as back-end calculations.
P. Ipsilandis / Multiobjective Optimization in Linear Repetitive Project Scheduling 259
3. Multi Objective LP Scheduling for LRPs
3.1 Decision variables in scheduling LRPs
Scheduling of LRPs is in practice more complicated and relevant decisions could
involve more control variables than just duration (CPM) or resource continuity
(RSM). The following list includes criteria that are important to construction
managers in their decision maki ng regarding scheduling of LRPs
i. Durat i on: Project duration is a key variable to any project since it affects the
projects financial performance (cost overruns, penalties) and reputation of the
project organization
ii. Res ource delay: Violating the work continuity of the same task between
successive project units introduces work-gaps that increase the cost of the project
because of idle resources.
iii. Uni t compl et i on time: Completion of the work on a project unit affects the
project deliverables and it could have significant financial implications since the
project' s cash receipts depend on completing intermediate deliverables. In other
cases, the completed units of the project can become operational before the
completion of the entire project thus resulting in earlier cash inflows.
iv. Sl ack time: Reducing activity slack time introduces higher risk to the project,
regarding unit compl et i on time and overall project duration.
V. Numbe r of uni t s the project is divided into: Each unit introduces a certain fixed
cost for maintaining a construction site whi ch increases the cost of the project but
at the same time splits the project into smaller deliverables which could improve
the project cash flow.
Furthermore scheduling decisions are rarely based only on any single variable.
Alternative project schedules, comparisons and cost tradeoffs are often needed to
arrive at an acceptable or opt i mum project schedule. In this aspect the scheduling
probl em can be seen as a multi objective probl em that can be addressed by linear
programmi ng techniques.
The following linear programmi ng formulation of LRP scheduling creates a model
whi ch has a parametric structure that can address some of the criteria (individually or
combined) raised above.
3.2 Formal description of the MOLPS - L RP model
Consider an LRP that consists of a set of m tasks represented by the nodes of a graph
and a set of P dependency relationships that can be represented as the arcs of the
graph. These dependencies can be of any type of the known of project dependencies
(SS, FS, SF, FF, with or without time-lag between the dependent activities). The
project is divided into n separate units in a linear way where without loss of
generality the following general assumptions hol d in the most part:
260 Operational Research. An International Journal / Vol.6, No.3 / September - December 2006
i. Al l t asks are p e r f o r me d in all uni t s
ii. A t ask c a nnot be pe r f or me d i n any pr oj ect uni t bef or e t he s a me t ask is c ompl e t e d
i n t he pr e vi ous uni t
iii. The set o f de pe nde nc i e s r emai n t he s a me i n all units.
Yang, (2002a) l i st s a set o f pr act i cal c onc e r ns i n s c he dul i ng r epet i t i ve pr oj ect s t hat
are speci al e xc e pt i ons o f t he gener al a s s umpt i ons ma de a bove and can be eas i l y
ha ndl e d in t he LP f or mul a t i on t hat f ol l ows .
Bas ed on t he f or ma l def i ni t i on gi ven above, t he LRP p r o b l e m c a n be f or mul a t e d as a
l i near p r o g r a mmi n g mode l as f ol l ows:
Model vari abl es and paramet ers
Let i= 1, 2, . . . , m de not e t he pr oj ect t asks andj =l , 2, . . . , n t he pr oj e c t uni t s.
Def i ne: d o t he dur at i on o f t ask i i n u n i t j (al t ernat i vel y it can be f or mul at ed as
the amount o f the correspondi ng wor k wi/ di vi ded by the product i on
rat e piQ
so , f i , t he start and f i ni sh t i me r es pect i vel y o f t ask i i n u n i t j
Pi t he set o f pr e de c e s s or act i vi t i es t o t ask i
E t he set o f all act i vi t i es wi t hout successor s
WBi
UCj
cj
f i
Cons t rai nt def i ni t i ons
t he t ot al t i me o f wor k- br e a ks becaus e o f di s cont i nui t i es o f t he t as k i
be t we e n s ucces s i ve uni t s
t he c ompl e t i on t i me o f pr oj e c t uni t j
t he t i me uni t cost ( penal t y or f i nanci al ) f or c o mp l e t i o n del ay o f uni t j .
t he t i me uni t cost f or wor k- br e a k o f r es our ces us e d i n t ask i.
Tas k dur at i on const r ai nt s f j = s o + d o
Pr oj ect l i near i t y const r ai nt s so+l >_ f j
(t ask in unit j + l f ol l ows the same
t ask in uni t j . Except i ons to this
rule can be handl ed accordi ngl y)
Tas k de pe nde nc i e s so > f kj
(the exact f or m o f the const rai nt
depends on the dependency type.)
Uni t c o mp l e t i o n t i me:
(UC, defines the pr oj ect duration)
Res our ce del ay f or t ask i:
V i--1,2 ..... m and j =l , 2 ..... n (1)
V i=1, 2 ..... m and j =l , 2 ..... n-1 (2)
V i=1, 2 ..... m, j =l , 2 . . . . . n and k~Pi (3)
UCy >)~j V k ~ E (4)
n-1
WBi = Z (s~+, - L ), v i = 1 ..... m (5)
j=l
The f ol l owi ng s et o f const r ai nt s des cr i bes t he oper at i on o f act i vi t i es i n an LRP:
P. Ipsilandis / Multiobjective Optimization in Linear Repetitive Project Scheduling 261
m
Total resource delay: WB = ~ WBi (6)
i 1
Gl obal objecti ve f unct i on
n m
Mi n i mi z e Z c i . ( UCj - Mi n UC j ) + ~ f .WB i (7)
j-1 i=l
wher e MinUCj is the CPM completion time for uni t j .
Dependi ng on the values of the parameters cj a n d f the above defined global objective
funct i on can be used accordi ngl y for optimizing various project objectives as listed in
table 1.
Table 1. Parameter settins f or various objective functions
Objective function Minimize Paramet er values
Project Duration
Total work-break time
Total Cost of
resource work-
breaks
Unit Completion Time
Cost of Delays in
Completion
Tradeoffs between
Cost of completion
delays and resource
work-breaks
UCn On=l, rest of cj and f (7.1)
equal 0
WB All fiequal 1, all cj (7.2)
equal to 0
All cj equal to 0 (7.3)
m
Z .WB
i=l
m Allfequal O, all @ (7.4)
ZUCi equal to 1
i=1
m
2 c i ' ( g c i -- Mi n UC i )
i=l
n m
Zci.(UCj-MinUCj)+ ~" f . WBi
j=l i=l
All f equal to 0 (7.5)
(7.6)
4. Appl i cati on Resul ts of MOL PS - L RP
4.1 A case st udy exampl e o f L R P
The LP model that was defi ned in the previous section is applied to a case study of a
small linear discrete proj ect that was initially used by the RSM authors [Harris &
Ioannou (1998)], with some modifications to task duration times. The proj ect consists
of six repetitive units, each having six discrete activities repeated at each unit. All
task dependencies are finish-to-start and each activity is performed by a specific crew.
Figure 1 shows the precedence net work for one unit, the duration of each task at each
of the six project units and other project specific constraints.
262 Operational Research. An International Journal / Vol.6, No.3 / September - December 2006
Du io " Lag2 /
4-4-6-6-4-4
Duration
2-2-2-2-X-2
* Duration of activity at each of the6 project units
Activity C is not executed at the~ h unit
J Duration
. ~ ~ 10-10-10-10-10-10
Duration
6-6-6-6-6-6
Due to technological constraint activity C cannot start until two days after activity A is
finished (Lag = 2)
There is a 5 days delay in activity B between units3 and 4
Figure 1: Precedence network in Unit 1
4.2 Minimizing work-break t i me
The CPM schedule for the project produces a minimum duration of 48 days. Given
the 48 duration as a constraint (UC6=48), the LP model was run setting work-break
minimization (7.2) as the objective function. The resulting schedule is shown in
figure 2. The minimum project duration of 48 days can be satisfied with a minimum
o 26 days of work break in parts of activities B and C. Further reduction of work-
break time at activities cannot be achieved without extending the project' s duration
beyond 48 days.
Figure 2. Linear Schedule - Minimization of Resource Delays with CPM duration
P. Ipsilandis / Multiobjective Optimization in Linear Repetitive Project Scheduling 263
If the CPM duration constraint is relaxed the work-breaks can be further reduced to a
minimum of 5 days but this will result in extending the project' s duration to 62 days
as shown in figure 5. The work break of task C is eliminated, while that of task B is
reduced to 5 days as set by the task' s technological constraints. The finishing time of
all units is also pulled to 14 days later than in the previous schedule.
Figure 3. Minimization of Resource Delays with relaxation of CPM duration
4.3 Minimizing unit completion time
Unit completion time is also an important criterion in scheduling repetitive projects.
Completing units as early as possible could be critical because of seasonal constraints
and could also affect the financial standing of the projects in cases where payments
depend on the delivery of completed units. In linear projects like highway
construction the early delivery of certain units could mark the project' s income
generation, thus any delay in unit completion could change negatively the net present
value of the project. In this case the objective is set to minimization of completion
time of all or certain units, even if that means that additional work-brakes must be
inserted at tasks. Figure 4 shows the resulting schedule when minimization of
completion time for all units (7.4) is defined as the objective of the MOLPS / LRP.
Even though the overall duration of the project cannot be further reduced to less than
48 days, intermediate units are delivered at earlier times. (The numbers in parenthesis
show the gains in delivery time for each unit as compared to the RSM schedule in
figure 4). On the other hand, these savings are achieved by sacrificing work
continuity. The total work-break time in all activities is increased to 64 days, or 38
days more than minimum work break time that can be achieved for the same project
duration (figure 4).
264 Operational Research. An International Journal / Vol.6, No.3 / September - December 2006
Figure 4. Linear Schedule - Minimization of Unit Completion Time
4.4 Cost trade-offs between work-brake and unit duration
In real life projects the cost of work-breaks at different activities varies according to
the type and scarceness of the resources consumed. In addition, the lateness delivery
cost which affects the overall cost of the project in different ways based on the overall
schedule of project income receipts and payments could be associated with the
delivery of the entire project and/or with deliveries of compl et ed project units. The
cost based objective functions listed in the previous sections [(7.3), (7,4), (7,6)] can
be used for providing optimal schedules in this case.
Trade-off analysis between completion time and work-break time is possible when
associated cost data exist. However, even in the case that no exact cost estimates
exist, the tools of LP sensitivity analysis can be used to provide opt i mum trade of f
ranges of the ratio between delay and work-break unit cost. Initially the cost of delays
is set equal to the unit cost of the work-break. Since it is only the relationship between
the two cost parameters and not their absolute values that affect the optimization
process both values are initially set equal to one and using LP range analysis the
optimality ranges and the changes in the opt i mum solution can be derived. The
examples that follow demonstrate this type of analysis:
Trade off between project duration delay cost and work-break cost
The first one is a tradeoff analysis between delays in project completion and task
work-breaks. Delays are measured as deviations from the earliest finish date of the
P. Ipsilandis / Multiobjective Optimization in Linear Repetitive Project Scheduling 265
project as it is set by the CPM. It is also assumed that intermediate delays in
completing each unit do not impose any cost to the project, and that the cost of work-
breaks is the same for all tasks.
In this case the objective trade-off function (7.6) of the LP model can be written as:
Minimize UC, + WB
The range analysis on the objective function cost parameters produced the results
shown in figure 5. When the work-break unit cost is below 50% of the cost paid for
project delays the optimum scheduling produces a project duration of 48 days
(minimum possible) with a maximum work-break time of 26 days. When the work-
break unit cost is between 50% and 100% of the project delay cost, it is more
economical to let the project slip by 5 days in order to reduce work-breaks to 16 days.
And finally when the work-break cost exceeds the project delay cost, the optimum
schedule is the one that introduces the minimum of 5 days of work-breaks, which
results in extending the project duration by 16 days.
Figure 5. Tradeoffs between project completion and work-breaks.
Trade off between unit duration delay cost and work-break cost
In the second example, it is assumed that a penalty cost is associated with delays in
delivery time of individual project. Delays could be measured as deviations from the
earliest finish dates of the project units or from an agreed time. This choice will not
affect at all the range analysis that follow, since the cost coefficients of the objective
function remain unchanged. For simplicity purposes here we assume that the same
penalty c applies to delays in any unit. Also as in the previous case, the cost of work-
breaks is assumed to be the same for all tasks.
266 Operational Research. An International Journal / Vol.6, No.3 / September - December 2006
In this case the objective function (7.6) of the LP model is set to:
Minimize Zc. ( UCj - Dy) + ZT. WBi =c + mo i - c Dj
j=l i=1 ~j =l '= j=l
The range analysis results are shown in figure 6. Minimizing the work brake time in
the task is necessary only when the associated work-break cost is 6 times higher than
the cost paid for unit delays 4.
A significant break point is when work-break cost is 3 times higher than cost of
delays. Under this level total delays in the units are kept below 17 days in total (about
3 days per unit) while above this level, they range from 51 to 114 days (about 8,5 to
19 days). Figure 5 gives a graphical representation of the results.
Figure 6. Tradeoffs between unit completion delays and work-breaks.
5. Conclusions
Scheduling of linear repetitive construction projects is not a single dimension decision
process. Factors such as the duration of the project the delivery of the individual
project units on time, the continuation of resource usage, and the slack time are
factors that construction managers must take into consideration in deciding an
optimum scheduling for the project. A scheduling decision must take into
consideration more than a single factor and most of the times tradeoffs are required
between unit completion times, project duration and work-breaks. Strictly scheduling
methods like CPM and RSM can provide mainly answers regarding time dimension
but cannot address in an integrated way, cost considerations that are raised since
many cost factors such different cost of work-breaks at each task, penalty costs
related to delays in project duration, financial costs arising from inadequate cash flow
P. Ipsilandis / Multiobjective Optimization in Linear Repetitive Project Scheduling 267
because of lateness in delivery of partial units. The MOLPS-LRP model can address
these issues and provide answers to relative questions. Furthermore the MOLPS-LRP
can be used to determine the optimum number of segments in a linear project when
segments are not defined as physical units (i.e. floors, apartments etc.) given that an
increase in the number of units affects positively the duration of the project in a way
of diminishing retums but also increases the cost of work-breaks and the total
employment of the resources.
References
A Guide to the Project Management Body of Knowledge, 3 rd Edition, (2004). PMI
Standard Committee, PMI Institute.
Ashley, D. B. (1980), "Simulation of repetitive-unit construction" Journal of
Construction Division, Vol. 106, No. CO2, pp. 185-194.
Birrell, G. S. (1980), "Construction planning-beyond the critical path" Journal of
Construction Division, Vol. 106, No. CO3, pp. 389-407.
Car l R.I. and Meyer, W.L. (1974), "Planning construction of repetitive building
units" Journal of Construction Division, Vol. 100, No. CO3, pp. 403-412.
Dawood, N. (1998), "Estimating project and activity duration: a risk management
approach using network analysis" Construction Management and Economics, 16,
pp. 41-48.
Dressier, J. (1974), "Construction scheduling of linear construction sites" Journal of
Construction Division, Vol. 100, No. CO4, pp. 571-587.
Harmelink, D.J. and Rowings, J.E. (1998), "Linear Scheduling Model: development
of controlling activity path" Journal of Construction Engineering and
Management, Vol. 124, No. 4, pp. 263-268.
Harmelink, D.J. (2001), "Linear Scheduling Model: float characteristics" Journal of
Construction Engineering and Management, Vol. 127, No. 4, pp. 255-260.
Harris, R.B. and Ioannou, P.G. (1998), "Scheduling project with repeating activities"
Joumal of Construction Engineering and Management, Vol. 124, No. 4, pp. 269-
278.
Ioannou, P.G. and Harris, R.B. (2003), "Discussion of algorithm for determining
controlling path considering resource continuity" Journal of Computing in Civil
Engineering, Vol. 17 ,No. 1, pp. 68-70.
Ioannou, P.G. and Yang, I.T. (2003), "Discussion of algorithm for determining
controlling path considering resource continuity" Joumal of Computing in Civil
Engineering, Vol. 17 ,No. 1, pp. 70-72.
Ioannou, P.G. and Yang, I.T. (2004), "Discussion of comparison of Linear
Scheduling Model and Repetitive Scheduling Method" Joumal of Construction
Engineering and Management, Vol. 130, No. 3, pp. 461-463.
Johnston, D.W. (1981), "Linear Scheduling Method for highway construction"
Joumal of Construction Division, Vol. 107, No. CO2, pp. 247-261.
268 Operational Research. An International Journal / Vol.6, No.3 / September - December 2006
Kallantzis, A. and Lambropoulos, S. (2003), "Correspondence of activity
relationships and critical path between time-location diagrams and Critical Path
Method" Proceedings of the 16th National Conference of E.E.E.E., Larissa, pp.
67-76.
Kallantzis, A. and Lambropoulos, S. (2004), "Critical path determination by
incorporation minimum and maximum time and distance constraints into linear
scheduling" Engineering, Construction and Architectural Management, Voh 11,
No. 3, pp. 211-222.
Mattila, K.G. and Abraham, D.M. (1998), "Linear scheduling: past efforts and future
directions", Engineering, Construction and Architectural Management, Vol. 5, No.
3, pp. 294-303.
Mattila, K.G. and Park, A. (2003), "Comparison of Linear Scheduling Method and
Repetitive Scheduling Method", Journal of Construction Engineering and
Management, Vol. 129, No. 1, pp. 56-64.
Mulholland, B. and Christian, J. (1999), "Risk assessment in construction schedules"
Journal of Construction Engineering and Management, Voh 125, No. 1, pp. 8-15.
Nasir, D., McCabe, B., Hartono, L. (2003), "Evaluating risk in construction -
schedule model (ERIC-S-): Construction schedule risk model" Journal of
Construction Engineering and Management, Vol. 129, No. 5, pp. 518-527.
O' Brien, J.J. (1975), "VPM scheduling for high-rise buildings" Journal of
Construction Division, Vol. 101, No. CO4, pp. 895-905.
Palisade Corporation (2004), "User Guide for @RISK for Project Risk analysis and
simulation add-in for Microsoft Project" Version 4.1
Peer, S. (1974), "Network analysis and construction planning", Journal of
Construction Division, Vol. 100, No. CO3, pp. 203-210.
Reda, R.B. (1990), "RPM: Repetitive Project Modeling" Journal of Construction
Engineering and Management, Vol. 116, No. 2, pp. 316-330.
Russell, A.D. and Ceselton,W.F. (1988), "Extensions to linear schedule optimization"
Journal of Construction Engineering and Management, Vol. 114, No. 4, pp. 36-52.
Russell, A.D. and Wong, W.C.M. (1993), "New generation of planning structures"
Journal of Construction Engineering and Management, Vol. 119, No. 2, pp. 196-
214.
Selinger, S. (1980), "Construction planning for linear projects" Journal of
Construction Division, Vol. 106, No. CO2, pp. 195-205.
Stradal, O. and Cacha, J. (1982), "Time space scheduling method" Journal of
Construction Division, Voh 108, No. CO3, pp. 445-457.
Vorster, M.C., Beliveau, Y.J. and Bafna, T. (1992), "'Linear scheduling and
visualization" Transportation Research Record, 1351, 32-39.
Yamin, R.A. and Harmelink, D.J. (2001), "Comparison of Linear Scheduling Model
(LSM) and Critical Path Method (CPM)" Journal of Construction Engineering and
Management, Vol. 127, No. 5, pp. 374-381.
P. Ipsilandis / Multiobjective Optimization in Linear Repetitive Project Scheduling 269
Yang, I.T. and Ioannou, P.G. (2001), "Resource-driven scheduling for repetitive
projects: a pul l -syst em approach", Proceedings of the 9th International Group for
Lean Construction Conference, Singapore, 6-8August, pp. 365-377.
Yang, I.T. and Ioannou, P.G. (2004), "Schedul i ng with focus on practical concerns in
repetitive projects" Construction Management and Economics, Vol. 22, pp. 619-
630.
Yang, I.T. (2002a), "Repetitive Project Planner: resource-driven scheduling for
repetitive construction projects" PhD. Dissertation, Uni versi t y of Michigan, Ann
Arbor, MI, USA.
Yang, I.T. (2002b) "Stochastic analysis on project duration under the requirement o f
continuous resource utilization", Proceedings of the 10th International Group for
Lean Construction Conference, Gramado, Brazil, August , pp. 527-540.
Reproducedwith permission of thecopyright owner. Further reproductionprohibited without permission.