Automation in Construction 21 (2012) 99–113

Contents lists available at ScienceDirect

Automation in Construction
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / a u t c o n

Simulation and analytical techniques for construction resource planning

and scheduling
Shih-Ming Chen a, F.H. (Bud). Griffis b, Po-Han Chen c,⁎, Luh-Maan Chang c
a
Dept. of Civil Engineering and Engineering Mechanics, Columbia Univ., New York, NY 10027, United States
b
Center for Construction Management Technology, Dept. of Civil Engr'g (Construction), Polytechnic Univ., Brooklyn, NY 11201, United States
c
Dept. of Civil Engineering, National Taiwan Univ., Taiwan

a r t i c l e i n f o a b s t r a c t

Article history: To date, few construction methods or models in the literature have discussed about helping the project
Accepted 29 May 2011 managers decide the near-optimum distributions of manpower, material, equipment and space according to
Available online 6 July 2011 their project objectives and project constraints. Thus, the traditional scheduling methods or models often
result in a “seat-of-the-pants” style of management, rather than decision making based on an analysis of real
Keywords:
data. This paper presents an intelligent scheduling system (ISS) that can help the project managers to find the
Computerized scheduling
Resource management
near-optimum schedule plan according to their project objectives and project constraints. ISS uses simulation
Artificial intelligence techniques to distribute resources and assign different levels of priorities to different activities in each
Optimization simulation cycle to find the near-optimal solution. ISS considers and integrates most of the important
Simulation construction factors (schedule, cost, space, manpower, equipment and material) simultaneously in a unified
environment, which makes the resulting schedule that will be closer to optimal. Furthermore, ISS allows for
what-if analyses of possible scenarios, and schedule adjustments based on unforeseen conditions (change
orders, late material delivery, etc.). Finally, two sample applications and one real-world construction project
are utilized to illustrate and compare the effectiveness of ISS with two widely used software packages,
Primavera Project Planner and Microsoft Project.
© 2011 Elsevier B.V. All rights reserved.

1. Introduction
Construction projects are becoming progressively larger and more
complex in terms of physical size and cost, hence the risks and
potential for losses require better control. Project management has
evolved mainly because of the need to control costs and schedule. To
date, the researchers have proposed a number of varying techniques
for reducing the project costs and project duration. The crucial
questions that they tried to answer are:
• What is the near-optimum resource (i.e., manpower, material and
equipment) distribution for each activity for different combinations
of project objectives (e.g., minimum project cost and minimum
project duration) and project constraints (e.g., project due date and
daily resource constraints)?
⁎ Corresponding author.
E-mail addresses: sc489@caa.columbia.edu (S.-M. Chen), griffis@poly.edu
(F.H.(B.) Griffis), pohanchen@ntu.edu.tw (P.-H. Chen), luhchang@ntu.edu.tw
(L.-M. Chang).
• What is the near-optimum space distribution that can satisfy the
requirements of resources with the minimum prohibition for the
project objectives?
• What is the probability to reach the project objectives? Thus, the
contractors can decide if they should buy the insurances for their
projects and how much that they should buy accordingly. More
importantly, is there a more reliable way to do risk analyses?
• What are the impacts if schedule adjustments are required based
on unforeseen conditions (e.g., change orders and late material
delivery)?
This paper presents an intelligent scheduling system (ISS) to
produce different answers to the above questions. ISS was developed
from combining computer simulation with analytical techniques. This
approach can use the advantages and avoid the disadvantages of both
simulation and mathematical modeling techniques. ISS was devel-
oped under the environment of Simphony (NSERC/Alberta Construc-
tion Industry, University of Alberta), a general-purpose simulation
language, because it provides flexibility in the form of user-written
simulation code. A resource library that incorporates atomic models is
included in ISS. The project managers can combine related atomic
models to build a bigger and more complicated schedule model. ISS
considers and integrates the most important construction factors

0926-5805/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.autcon.2011.05.018
100 S.-M. Chen et al. / Automation in Construction 21 (2012) 99–113

(schedule, cost, space, manpower, equipment and material) simulta-
neously during the process of finding the near-optimal schedule,
which makes the resulting schedule that will be close to optimal. The
major objectives of ISS are listed as follows:
• Applies evolutionary techniques to decide the near-optimum distri-
butions of manpower, equipment, material and space according to
project objectives and project constraints.
• Considers and integrates most of the important construction factors
(schedule, cost, manpower, equipment, material and space)
simultaneously in a unified environment, which makes the resulting
schedule that will be close to optimal.
• Improves risk analyses on project duration, project cost, project
revenue, and net present.
The outputs of ISS include: (1) resource distribution; (2) dynamic
space allocation; (3) production rates of driving resources; (4) utilization
rates of resources and space; (5) activity duration; (6) project schedule
and cost (including project revenue and project net present value);
(7) materials used for each activity; and (8) stochastic risk analyses of
project duration, project cost, project revenue and project net present
value. Furthermore, it allows for what-if analyses of possible scenarios,
and schedule adjustments based on unforeseen conditions (change
orders, late material delivery, etc.).

Fig. 1. ISS flow chart.

 S.-M. Chen et al. / Automation in Construction 21 (2012) 99–113 101

heuristic rules and select the schedule with minimum duration. This
procedure, however, has little diversity since the number of effective
rules to enumerate is small and it is not expected that less effective
rules will change much when effective rules are not improving the
schedule. Therefore, without introducing new rules or changing the
mechanics of heuristic procedures, a simple approach of forcing
random activity priorities is presented to improve the goodness of the
schedule. Consequently, ISS uses simulation techniques to distribute
resources and assign random activity priorities to different activities
Fig. 2. Project network for sample application.
in each simulation cycle to find the near-optimum solution.
Although the aforementioned dynamic site layout planning models
made their different contributions on solving space distribution
2. Major objectives of ISS problems, these models all use the chronological procedure to find the
local optimum solution for each of the identified stages of the project
2.1. Applies evolutionary techniques to decide the near-optimum duration. Such chronological procedure will induce the space distribu-
distributions of manpower, equipment, material and space according to tion of later stage greatly affected by the space distributed of early stage,
project objectives and project constraints which does not guarantee a global optimum solution. Furthermore, the
chronological procedure may cause the early located resources occupy
Traditional resource optimization techniques mainly include the only space that the future resource can use, and cause infeasible
mathematical method and heuristic method. Mathematical models, solutions.
however, suffer from being complex in their formulation and may be Besides regular resources such as manpower, equipment and
trapped in local optimum [12,7]. Heuristic models are problem material, space is also tracked as a resource and distributed according
dependent, with varying effectiveness on different cases. Thus their to the same algorithm as distributing the other resources in ISS, which
rules of thumb cannot be equally applied to all construction cases. can avoid the aforementioned drawback. ISS allows designers to assign
Furthermore, there was no way to know a priori the best heuristic the required space area and region for each regular resource in each
models to use for a given case. Optimal solutions from the heuristic activity, and search for the near-optimum dynamic space distribution
models are not guaranteed. Their inconsistent solutions have according to the random resource distributions with the minimum
contributed to large discrepancies among the resource scheduling prohibition for the project objectives and project constraints. ISS allows
features of commercial project management software [6,8]. the project managers to assign the project objectives (e.g., minimum
During rapid development of computer technology, simulation project cost and minimum project duration) and project constraints
modeling and analysis of construction processes have gained impor- (e.g., maximum project duration and maximum resources) with daily
tance in recent years in light of an increase in the complexity of resource constraints?
construction processes [21].
Heuristic, numerical and genetic algorithms have been applied to
solve static layout planning problems in several research projects (e.g., 2.2. Considers and integrates most of the important construction factors
[29,19,5,26,27,30,9,14,31,3,24,16,10,11]). These models statically allo- (schedule, cost, manpower, equipment, material and space) simultaneously
cate space to objects over the entire project duration and include in a unified environment, which makes the resulting schedule that will be
permanent and temporary facilities needed throughout construction. close to optimal
They ignore the possible reuse of site space to accommodate different
resources at different times, relocating resources, and varying space Schedule, cost, manpower, equipment, material and space are the
needs of resources over time. major construction factors and are intricately related. Most researches on
Generating layouts that change over time as construction pro-
gresses is termed dynamic layout planning. The problem of dynamic
layout planning has been recognized recently [22,27]. This problem is
Table 1
characterized by spatial project data that varies with time as work Sample application I databases.
progresses and materials get used. Alternative means exist to solve
dynamic layout problems. In the area of production facilities, Activity Duration Predecessors Daily resource requirements
(days)
Rosenblatt [20] and Montreuil and Venkatadri [15] solved different R1 R2 R3 R4 R5 R6
formulations of the dynamic facility layout problem subject to A 6 – 5 2 2 2 7 4
nonoverlap constraints between facilities and bounds on the facilities' B 3 – 3 5 2 3 9 6
shape and area. Optimum algorithms for solving this problem are NP- C 4 A 2 4 4 2 3 1
D 6 – 5 4 3 5 5 4
complete, and exact solutions can be computed only for small or
E 7 A, B 3 5 2 3 8 0
greatly restricted problems. In construction, Tommelein [26], Cheng F 5 C 4 1 4 9 2 5
[2], Thabet [25], Tommelein and Zouein [28], Riley [18], Lin and Haas G 2 D 4 1 4 3 9 8
[13], Osman et al. [17], and Elbeltagi et al. [4] explored alternative H 2 A, B 5 5 4 0 9 1
means to solve the dynamic layout problem, by using either I 2 G, H 3 2 4 3 4 2
J 6 F 1 5 4 6 7 3
interactive selection or computer-based positioning of resources.
K 1 C, E 3 3 2 4 5 1
ISS uses stochastic approach with dynamic feature to model the L 2 E, G, H 3 2 2 8 3 4
uncertainty of activity duration in PERT. In ISS, the true properties of the M 4 I, K 2 2 2 2 4 8
resource productivity, instead of activity duration, distribution (includ- N 2 F, L 1 4 4 3 4 1
O 3 L 5 5 4 6 2 3
ing distribution shape, mean and variance) are used to calculate activity
P 5 J, M, N 3 2 3 4 7 8
duration, and the concept of activity criticality rather than path Q 8 O 4 5 4 2 3 4
criticality, introduced by Ahuja et al. [1], is used to overcome most of R 2 D, O 5 3 3 3 7 8
the PERT drawbacks. S 6 P, R 2 4 6 2 3 4
Since it is not possible to select an optimum heuristic rule for a T 2 Q 1 6 2 7 5 2
Daily resource limits 7 10 10 16 18 13
given project network, one common procedure is to try a series of
102 S.-M. Chen et al. / Automation in Construction 21 (2012) 99–113

the resource distribution didn't consider the space restrictions of
resources, which may cause infeasible results. ISS provides a mechanism
(refer to next section) to experiment with the resource distribution
model so that the dynamic behaviors of the construction process can be
observed by tracing the movement and interactions of simulation
entities in the model. Allowing project managers to assign the project
objectives and project constraints with daily resource constraints, ISS
considers and integrates all of these construction factors (schedule, cost,
manpower, equipment, material and space) simultaneously in a unified
environment, which makes the resulting schedule that will be close to
optimal.
2.3. Improves risk analyses on project duration, project cost, project revenue,
and net present
To perform risk analysis, the past researchers proposed a lot of
probabilistic models such as program evaluation and review technique
(PERT) and Monte Carlo simulation. In these models, the duration of
each activity is entered or evaluated independently of the durations of
the other activities in the network, and activity cost is calculated
accordingly. These approaches do not consider the correlations that
exist in the activity durations, activity cost, resources and spaces in a
schedule network.
The author argues that the risk analysis on project duration should
be based on the stochastic distribution of resource production rate,
which leads to the activity duration and project duration with
consideration of daily resource constraints. The risk analysis on project
cost should be based on the stochastic distribution of resource unit cost,
which leads to the activity cost and project cost with consideration of
previous activity duration. In ISS, stochastic risk analysis is conducted
by considering most of the important factors including required spaces,
unit costs, and production rates for resources (i.e., manpower, material,
and equipment). All of these factors could be assigned different
probability distributions (constant, uniform, triangular, normal, beta,
or exponential) for different activities at the same time. Moreover, as
mentioned before, the required space and resources are generated from
running a simulation process, and thus possess a stochastic nature. With
this function, project managers could perform close-to-reality risk

Table 2
Activities performed and total resources used each day.

Activity resource Daily activities Resource quantity

Day A B C D E F G H I J K L M N O P Q R S T R1 R2 R3 R4 R5 R6

1 X 5 2 2 2 7 4
2 X 5 2 2 2 7 4
3 X 5 2 2 2 7 4
4 X 5 2 2 2 7 4
5 X 5 2 2 2 7 4
6 X 5 2 2 2 7 4
7 X X 5 9 6 5 12 7
8 X X 5 9 6 5 12 7
9 X X 5 9 6 5 12 7
10 X X 7 8 7 7 8 5
11 X 5 4 3 5 5 4
12 X 5 4 3 5 5 4
13 X 5 4 3 5 5 4
14 X 5 4 3 5 5 4
15 X 5 4 3 5 5 4
16 X X 7 6 6 12 10 5
17 X X 7 6 6 12 10 5
18 X X 7 6 6 12 10 5
19 X X 7 6 6 12 10 5
20 X X 7 6 6 12 10 5
21 X X 7 6 6 6 17 8
22 X X 7 6 6 6 17 8
23 X X 6 10 8 6 16 4
24 X X 6 10 8 6 16 4
25 X X X 7 10 10 13 16 6
26 X X 4 7 8 9 11 5
27 X X 4 7 6 14 10 7
28 X X 4 7 6 14 10 7
29 X X 7 7 6 8 6 11
30 X X 7 7 6 8 6 11
31 X X 7 7 6 8 6 11
32 X X 3 6 6 5 8 9
33 X X 6 7 7 6 11 9
34 X 5 3 3 3 7 8
35 X X 7 7 7 6 10 12
36 X X 7 7 7 6 10 12
37 X X 7 7 7 6 10 12
38 X X 7 7 7 6 10 12
39 X X 7 7 7 6 10 12
40 X X 6 9 10 4 6 8
41 X X 6 9 10 4 6 8
42 X X 6 9 10 4 6 8
43 X X 3 10 8 9 8 6
44 X X 3 10 8 9 8 6
45 X 2 4 6 2 3 4
 S.-M. Chen et al. / Automation in Construction 21 (2012) 99–113 103

Table 3
Activity definitions (predecessors, required resources, quantities and resource relationship constraints).

Activity Predecessors Work quantity for Required resources Resource Working region
manpower or equipment (manpower or equipment) relationship constraint

A – 1000 R1, R2, R3, R4, R5, R6 R1:R2 = 5:2 A1

B – 2000 R1, R2, R3, R4, R5, R6 R3:R4:R5 = 2:3:9 A1
C A 3000 R1, R2, R3, R4, R5, R6 R2:R5 = 4:3 A2
D – 4000 R1, R2, R3, R4, R5, R6 R2:R4 = 4:5 A2
E A, B 5000 R1, R2, R3, R4, R5 R2:R5 = 5:8 A3
F C 1000 R1, R2, R3, R4, R5, R6 R1:R4 = 4:9 A4
G D 2000 R1, R2, R3, R4, R5, R6 R5:R6 = 9:8 A4
H A, B 3000 R1, R2, R3, R5, R6 R2:R5 = 5:9 A3
I G, H 4000 R1, R2, R3, R4, R5, R6 R3:R4 = 4:3 A6
J F 5000 R1, R2, R3, R4, R5, R6 R4:R5 = 6:7 A6
K C, E 1000 R1, R2, R3, R4, R5, R6 R4:R5 = 4:5 A5
L E, G, H 2000 R1, R2, R3, R4, R5, R6 R1:R4 = 3:8 A5
M I, K 3000 R1, R2, R3, R4, R5, R6 R5:R6 = 4:8 A8
N F, L 4000 R1, R2, R3, R4, R5, R6 R3:R4 = 4:3 A7
O L 5000 R1, R2, R3, R4, R5, R6 R2:R4 = 5:6 A7
P J, M, N 1000 R1, R2, R3, R4, R5, R6 R5:R6 = 7:8 A9
Q O 2000 R1, R2, R3, R4, R5, R6 R1:R2 = 4:5 A8
R D, O 3000 R1, R2, R3, R4, R5, R6 R5:R6 = 7:8 A9
S P, R 4000 R1, R2, R3, R4, R5, R6 R2:R3 = 4:6 A10
T Q 5000 R1, R2, R3, R4, R5, R6 R2:R4 = 6:7 A10

analyses on project duration, project cost, project revenue, and net
present values by considering most of the important factors mentioned
above.
In conclusion, ISS incorporates the effect of correlation in network
schedules to support schedule risk management.
3. The other tools and concepts utilized for ISS development
3.1. Captures quantity takeoffs from a 3D CAD model automatically
Simulation tools require a large amount of input data. Some of the
input data can only come from the user's knowledge; the remainder
can typically be retrieved from other company systems including
historical databases, accounting systems, equipment databases and
CAD systems.
ISS allows project managers to capture quantity takeoffs from a 3D
CAD model automatically through database linkage system (built by the
author and will be introduced in a companion paper) and Dynamic Data
Exchange (DDE) technique, or assign them manually. After quantity
takeoffs are captured automatically or assigned manually, and resource
production rate distributions, resource unit costs, yearly interest rate,
overhead, contingency, profit, tax, and mark up are assigned, ISS could
get the activity duration, project schedule, project cost, project net
present values, and project revenue automatically after running the
simulation process.
3.2. Develops production rates of driving resources
Driving resources are defined as those resources applied to an
activity that controls the duration of an activity. The production rate of
the driving resource determines the duration of the activity. ISS
affords a manual interface to allow project managers to assign a
possible maximum production rate for each resource in each activity.
Production rates of driving resources in each activity will be identified
and posted after running the simulation process. This information can
help the project managers to adjust the resource distribution.
3.3. Develops the utilization rate of each resource/space allocation
The purpose of developing utilization rates for each resource
(manpower, equipment, material and space) allocated to a whole
project in ISS is to afford project managers the necessary databases to
adjust the required resource amounts and space distributions. For
instance, if some resources have low utilization rates, it may mean that
the same project duration could be achieved with fewer resources.
Similarly, if some divided space areas have low utilization rates and
some of the other divided space areas have high utilization rates, project
managers should consider the possibility of adjusting these divided
areas to improve the project duration.
3.4. What-if analyses of possible scenarios
ISS allows project managers to set up project constraints for any
variable (e.g., maximum project duration b365 days, etc.), project
goals (e.g., minimum project cost, minimum project duration, etc.),
maximum available resources for the whole project. Once any of these
settings is changed, ISS will re-allocate the resource distribution for
every activity. Thus activity duration, project schedule, net present
values, and all of the other values will change accordingly. With ISS's
what-if analysis function, project managers could correctly and clearly
see the impacts caused by the changes that they plan to make.
3.5. Schedule adjustments based on unforeseen conditions
ISS allows project managers to do the schedule adjustments based on
their requirements and practical situations. ISS will change the schedule
based on the changes made by the project managers. For instance,
owing to a change order from the owner, the first twenty activities'
durations were forced to change. ISS allows project managers to
manually set up these twenty activity durations according to the latest
information, and re-run the whole scheduling process. Then all of the
optimum resources/space distribution, duration, cost for all of the other
activities, and project duration/cost will change according to the latest
information.
4. Structure of ISS
4.1. ISS flow chart
The flow chart of ISS consists of: (1) activity network building;
(2) simulation execution; and (3) risk analysis, as shown in Fig. 1. The
16-step algorithm is presented in the following paragraphs.
104 S.-M. Chen et al. / Automation in Construction 21 (2012) 99–113

Activity network building Table 5

Material unit cost, material type required, and material productivity in each activity.

• Step 1: the available manpower/equipment/material/space types in Activity Work quantity Material unit cost/material type required/material
the whole project are identified. Then, the activity network is built for material (Q) productivity (Q/unit)
and the resource types required in each activity are identified. M1 M2 M3 M4 M5 M6
• Step 2: the work quantity takeoffs for manpower/equipment/material
20 30 40 15 25 35
are imported from 3D CAD model or assigned manually in each activity. ($/unit) ($/unit) ($/unit) ($/unit) ($/unit) ($/unit)
And, the maximum available resource/space quantity and mean unit
A 5000 800 – – – 250 –
cost in the whole project are identified. In this simulation model all of B 4000 – 800 – 500 – –
the resources, except material resources which are assumed without C 3000 400 – 250 – – –
daily resource constraints, are assigned integer variables with uniform D 2000 – – 400 – 250 –
distribution from one to their maximum available resource amount in E 1000 200 – – 100 – –
F 5000 – – 800 – 500 –
each activity. The mean resource productivity and resource relation-
G 4000 – – – 800 – 500
ship constraints (e.g. one truck needs one driver) in each activity are H 3000 250 400 – – – –
assigned. The project objectives (e.g., minimum project duration, I 2000 400 – 250 – – –
minimum project cost, or minimum project cost without exceeding J 1000 – 200 – – 100 –
assigned project due date) and project constraints (e.g., project due K 5000 – – 500 – 800 –
L 4000 – – – 500 – 800
date and daily resource constraints) are defined. M 3000 – 250 – – 400 –
N 2000 400 – – 250 – –
O 1000 – – 200 – – 100
P 5000 500 – 800 – – –
Q 4000 – 800 – 500 – –
Table 4 R 3000 400 – 250 – – –
Possible maximum resource productivity, unit cost, daily resource constraints and S 2000 – – – 400 – 250
required working area per unit resource. T 1000 100 – – – 200 –

Item Activity Resource (manpower or equipment)

R1 R2 R3 R4 R5 R6 Simulation execution
Max A Triang Triang Triang Triang Triang Triang
productivity (4,5,6) (3,4,5) (2,3,4) (3,4,5) (4,5,6) (5,6,7) • Step 3: the number of simulation iterations is set.
(Q/unit-hr) B Triang Triang Triang Triang Triang Triang • Step 4: in each activity, activity priority (the right to take precedence
(3,4,5) (2,3,4) (4,5,6) (3,4,5) (2,3,4) (4,5,6)
in obtaining required resources) is an assigned integer variable with a
C Triang Triang Triang Triang Triang Triang
(2,3,4) (3,4,5) (4,5,6) (2,3,4) (3,4,5) (4,5,6) uniform distribution from one to the total activity quantity in the
D Triang Triang Triang Triang Triang Triang whole project.
(4,5,6) (3,4,5) (2,3,4) (3,4,5) (4,5,6) (5,6,7) • Step 5: manpower/equipment/space is distributed according to
E Triang Triang Triang Triang Triang - activity priority assigned in step 4.
(3,4,5) (2,3,4) (4,5,6) (3,4,5) (2,3,4)
• Step 6: activity duration, activity schedule and project duration are
F Triang Triang Triang Triang Triang Triang
(2,3,4) (3,4,5) (4,5,6) (2,3,4) (3,4,5) (4,5,6) calculated according to the resources distributed in step 5.
G Triang Triang Triang Triang Triang Triang • Step 7: after the resources are assigned to each activity and the
(4,5,6) (3,4,5) (2,3,4) (3,4,5) (4,5,6) (5,6,7) activity schedule is calculated, the resources used in every hour/day
H Triang Triang Triang - Triang Triang
are summarized. If a resource is assigned a certain value that
(3,4,5) (2,3,4) (4,5,6) (2,3,4) (4,5,6)
I Triang Triang Triang Triang Triang Triang violates the assigned resources' relationship constraints, or the total
(2,3,4) (3,4,5) (4,5,6) (2,3,4) (3,4,5) (4,5,6) resource amount used in any hour/day exceeds the hourly/daily
J Triang Triang Triang Triang Triang Triang resource limit, this resource will be re-assigned another value so
(4,5,6) (3,4,5) (2,3,4) (3,4,5) (4,5,6) (5,6,7) that the above constraints are satisfied.
K Triang Triang Triang Triang Triang Triang
• Step 8: material quantity required in each activity is calculated.
(3,4,5) (2,3,4) (4,5,6) (3,4,5) (2,3,4) (4,5,6)
L Triang Triang Triang Triang Triang Triang • Step 9: activity cost and project cost are calculated.
(2,3,4) (3,4,5) (4,5,6) (2,3,4) (3,4,5) (4,5,6) • Step 10: the simulation model produces outputs on a number of
M Triang Triang Triang Triang Triang Triang parameters that can be used towards optimization of the project. Such
(4,5,6) (3,4,5) (2,3,4) (3,4,5) (4,5,6) (5,6,7)
project parameters are: activity and project durations, activity and
N Triang Triang Triang Triang Triang Triang
(3,4,5) (2,3,4) (4,5,6) (3,4,5) (2,3,4) (4,5,6)
project costs, project net present value, project revenue, material
O Triang Triang Triang Triang Triang Triang quantity and optimum crew in each activity, actual resource
(2,3,4) (3,4,5) (4,5,6) (2,3,4) (3,4,5) (4,5,6) productivity in each activity, risk analysis of project duration, project
P Triang Triang Triang Triang Triang Triang cost, and project net present value at the project's start date. The
(4,5,6) (3,4,5) (2,3,4) (3,4,5) (4,5,6) (5,6,7)
output data is saved in each simulation iteration.
Q Triang Triang Triang Triang Triang Triang
(3,4,5) (2,3,4) (4,5,6) (3,4,5) (2,3,4) (4,5,6) • Step 11: at the end of every run, the simulation model will compare
R Triang Triang Triang Triang Triang Triang the outputs produced in step 10 with the optimum computer
(2,3,4) (3,4,5) (4,5,6) (2,3,4) (3,4,5) (4,5,6) outputs in memory (from the previous simulation run). If the new
S Triang Triang Triang Triang Triang Triang
outputs are better than the previous optimum ones, the model will
(4,5,6) (3,4,5) (2,3,4) (3,4,5) (4,5,6) (5,6,7)
T Triang Triang Triang Triang Triang Triang
replace the old optimum outputs with the new outputs and update
(3,4,5) (2,3,4) (4,5,6) (3,4,5) (2,3,4) (4,5,6) the underlying databases (e.g., the assigned resources and space
Unit cost All Normal Normal Normal Normal Normal Normal
($/unit-hr) (30,3) (40,4) (50,5) (30,3) (40,4) (50,5)
Daily resource All 7 10 10 16 18 13 Table 6
constraints Total working area of each working region.
Req'd working All 2 4 6 8 10 12
Area name A1 A2 A3 A4 A5 A6 A7 A8 A9 A10
area per unit
resource Total area 450 400 400 400 500 400 500 350 350 400
 S.-M. Chen et al. / Automation in Construction 21 (2012) 99–113 105

4.2.1. Stage 1: calculate production rate of driving resource, activity

duration, activity cost, activity revenue, and space requirements of activity
Before ISS can begin the resource/space allocation, the activity
duration, activity cost, activity revenue, and the space requirements of
each activity must be calculated first. They are calculated as follows:
Fig. 3. Project duration, project cost, project revenue and project net present value for
the optimum resource distribution. DN ðiÞ = min fGði; jÞNði; jÞg
 
Q ði; jÞ
t ðiÞ = max
Gði; jÞNði; jÞ
for each activity). Alternatively, if the simulation produces
successively worse results, the simulation model will maintain m n
C ðiÞ = ∑ ½U ð jÞGði; jÞt ðiÞ + ∑ ½Gði; kÞU ðkÞ
the optimum results in memory. Successive simulation alterations j=1 k=1
(to specified number of simulation runs) will therefore produce
optimal results. Optimality is dependent on the project objectives RðiÞ = C ðiÞð1 + overhead + contingency + profit + tax + marku pÞ
defined in step 2.
• Step 12: the algorithm repeats step 4 to 12, until the total number of Q ði; kÞ
Gði; kÞ =
simulation runs reaches the number of simulation iterations set in Nði; kÞ
m n
step 3.
AðiÞ = ∑ ½aði; jÞGði; jÞ + ∑ ½aði; kÞGði; kÞ
j=1 k=1
Risk analysis
where
• Step 13: the number of extra simulation iterations for risk analysis is
set.
DN(i) production rate of driving resource* for activity i
• Step 14: the resource productivity and resource unit cost are set as
t(i) activity duration of activity i
probability density function.
Q(i, j) quantities being acted on by resource* j for activity i
• Step 15: the resource quantity is assigned as constant as the
Q(i,k) quantities being acted on by material k for activity i
optimum result obtained from simulation execution stage.
G(i, j) number of resource* j for activity i
• Step 16: the probability distribution/cumulative distribution of
G(i, k) number of material k for activity i
project duration/project cost/project revenue/project net present
N(i, j) production rate of resource* j for activity i
value are produced.
N(i, k) quantities being produced by unit material k for activity i
C(i) activity cost for activity i
R(i) activity revenue for activity i
4.2. ISS algorithm U(j) unit cost of resource* j
U(k) unit cost of material k
Activities are the basic elements that ISS directly deals with during A(i, j) total space requirement for activity i
a simulation experiment. The dynamic behavior of a construction a(i, j) space requirement of unit resource* j for activity i
process is portrayed by detailing the changes in the state of activities. a(i, k) space requirement of unit material k for activity i
The major simulation algorithm of ISS involves two stages: (1) calculate resource* manpower or equipment
production rate of driving resource, activity duration, activity cost,
activity revenue, and space requirements of an activity; and (2) select Note that the activity requirements of manpower, equipment,
activity for construction. material and space will be calculated in this stage; materials are

Table 7
Part one of resource distribution, resource productivity, activity cost, activity start time, activity finish time and activity duration.

Activity title A B C D E F G H I J

R1 Req'd no 5 6 5 6 6 4 6 5 7 4
Productivity 1.6 2 4.2 5 2 5 2.67 2.4 3.43 4.5
R2 Req'd no 2 3 8 8 10 7 4 10 8 6
Productivity 4 4 2.63 3.75 1.2 2.86 4 1.2 3 3
R3 Req'd no 10 4 7 10 4 7 8 4 8 6
Productivity 0.8 3 3 3 3 2.86 2 3 3 3
R4 Req'd no 10 6 6 10 15 9 14 – 6 12
Productivity 0.8 2 3.5 3 0.8 2.22 1.14 – 4 1.5
R5 Req'd no 2 18 6 6 16 14 9 18 12 14
Productivity 4 0.67 3.5 5 0.75 1.43 1.78 0.67 2 1.29
R6 Req'd no 6 11 5 13 – 7 8 6 11 7
Productivity 1.33 1.09 4.2 2.31 – 2.86 2 2 2.18 2.57
M1 Req'd no 6 – 8 – 5 – – 12 5 –
M2 Req'd no – 5 – – – – – 8 – 5
M3 Req'd no – – 12 5 – 6 – – 8 –
M4 Req'd no – 8 – – 10 – 5 – – –
M5 Req'd no 20 – – 8 – 10 – – – 10
M6 Req'd no – – – – – – 8 – – –
Cost ($) 176,875 325,270 213,487 292,400 779,417 97,000 240,355 442,965 357,087 536,511
Start time (h) 0.000 125.000 425.000 291.667 567.857 1359.524 1234.524 984.524 1584.524 1751.190
Finish time (hr) 125.000 291.667 567.857 425.000 984.524 1409.524 1359.524 1234.524 1751.190 2028.968
Duration (h) 125.000 166.667 142.857 133.333 416.667 50.000 125.000 250.000 166.667 277.778
106 S.-M. Chen et al. / Automation in Construction 21 (2012) 99–113

Table 8
Part two of resource distribution, resource productivity, activity cost, activity start time, activity finish time and activity duration.

Activity title K L M N O P Q R S T

R1 Req'd no 7 6 4 6 7 4 4 7 3 4
Productivity 3.43 2.5 5 4 2.57 2.25 3 3.43 5 5
R2 Req'd no 8 9 10 6 10 8 5 6 4 6
Productivity 3 1.67 2 4 1.8 1.13 2.4 4 3.75 3.33
R3 Req'd no 8 5 9 8 6 3 8 10 6 8
Productivity 3 3 2.22 3 3 3 1.5 2.4 2.5 2.5
R4 Req'd no 12 16 8 6 12 14 6 10 10 7
Productivity 2 0.94 2.5 4 1.5 0.64 2 2.4 1.5 2.86
R5 Req'd no 15 4 4 5 11 7 6 7 16 9
Productivity 1.6 3.75 5 4.8 1.64 1.29 2 3.43 0.94 2.22
R6 Req'd no 13 4 8 9 13 8 2 8 4 13
Productivity 1.85 3.75 2.5 2.67 1.38 1.13 6 3 3.75 1.54
M1 Req'd no – – – 5 – 10 – 8 – 10
M2 Req'd no – – 12 – – – 5 – – –
M3 Req'd no 10 – – – 5 6 – 12 – –
M4 Req'd no – 8 – 8 – – 8 – 5 –
M5 Req'd no – – 8 – – – – – – 5
M6 Req'd no 6 5 – – 10 – – – 8 –
Cost ($) 106,389 217,628 266,048 275,220 656,106 188,227 206,937 241,880 451,022 495,325
Start time (h) 1409.524 1451.190 2473.413 2306.746 2028.968 2915.079 2623.413 2790.079 3276.190 3026.190
Finish time (h) 1451.190 1584.524 2623.413 2473.413 2306.746 3026.190 2790.079 2915.079 3542.857 3276.190
Duration (h) 41.667 133.333 150.000 166.667 277.778 111.111 166.667 125.000 266.667 250.000

assumed without daily resource constraints and material delivery delay
is not considered during schedule design phase; Q(i, j) and Q(i, k) are
either imported from 3D CAD models or assigned by the project
managers; G(i, j) are assigned randomly from uniform distribution and
limited to integer values; N(i, j), N(i, k), U(j), U(k), a(i, j) and a(i, k) come
from the assigned stochastic distribution or constant value by the
project managers.
4.2.2. Stage 2: select activity
All activities are equally checked against their logical, resource and
space requirements. If an activity meets all conditions for start, it is
selected into the feasibility activity set, and the early start time of an
activity is equal to the maximum of the earliest available time of required
resources, required spaces, and dependent activities. The activity with
the minimum early start time is then selected for construction.

Fig. 4. Project duration, project cost and project net present value with stochastic distribution in resource productivity and unit cost.

Fig. 5. Probability distribution of project duration.

 S.-M. Chen et al. / Automation in Construction 21 (2012) 99–113 107

Fig. 6. Cumulative distribution of project duration.

5. Case studies and discussions
5.1. Sample application I
In this sample application, the authors set the daily resource
requirements and activity durations as constants and used the
databases in a case study from Talbot and Patterson [23], with twenty
activities and six resources with daily availability constraints (Fig. 2
and Table 1). To compare the effectiveness of ISS versus Microsoft
Project software (Microsoft Project) and Primavera Software (Prima-
vera), the authors input the same databases, including daily resource
requirements and activity durations, into all three software programs
for analysis, and all project activities were assigned the same priority
level (i.e., activity priority is an assigned integer variable with a
uniform distribution from one to the total activity quantity in the
whole project) in both Microsoft Project and Primavera. The project
objective of this sample application is assigned to be the minimum
project duration.
Without considering the given resource constraints, the total
project duration, determined by simple critical path method (CPM)
analysis, is 32 days. When the resource-leveling feature (leveling is
used in the software's terminology for both allocation and leveling) of
Microsoft Project was set to “Automatic,” total project duration was
extended to 49 days, avoiding resource over-allocations. This solution
was obtained using the software's “standard” set of heuristic rules,
which maintains logical relationships and applies the “minimum total
slack” rule to resolve conflicts. The same results were also obtained
using the “minimum total slack” rule on Primavera as a high-end
system. Several other heuristic rules were also tried on Primavera,
without improving the schedule. A project duration of 49 days is,
therefore, the best result that can be obtained from widely used
commercial softwares.

Fig. 7. Probability distribution of project cost.

108 S.-M. Chen et al. / Automation in Construction 21 (2012) 99–113

Fig. 8. Cumulative distribution of project cost.

By using ISS to run the same databases one thousand times, the
author got a project schedule with a 45-day project duration (Table 2).
5.1.1. Discussion
Most commercial scheduling software systems allow project
managers to specify priority levels for activities. Microsoft Project
implements this in a direct manner by allowing project managers to
select among eight priority levels (“Highest,” “High,” etc., to “Lowest”),
and assign them in a simple spreadsheet form. The software also
provides a second set of heuristic rules for resource allocation in which
activity priority takes precedence over its “standard” set of heuristic
rules. It is possible, therefore, to introduce some bias into some activities
and consequently monitor the impact on the schedule. From the case
study reported by Hegazy and El-Zamzamy [6,8], the examined
software, including Microsoft Project and Primavera, lacks efficient
near-optimum resource-scheduling algorithms.
Since it is not possible to readily identify, from a given network, which
activities to assign higher priorities than others to improve the schedule,
ISS uses the simulation techniques to simulate all of the possible priority
levels of each activity to get the shortest project duration.
From the aforementioned results, the 45-day project duration
resulted from ISS saves around 8.2% project time compared with the
time obtained from Microsoft Project and Primavera.
5.2. Sample application II
The sample application presented below (partly adapted from
[23]) is to test the abilities of the aforementioned framework, and to
assist readers in understanding ISS.
The sample application is based on twenty activities and several
different resources including manpower, equipment, material and space
(Fig. 2, Tables 3–6). The quantities of manpower or equipment of each

Fig. 9. Probability distribution of project revenue.

 S.-M. Chen et al. / Automation in Construction 21 (2012) 99–113 109

Fig. 10. Cumulative distribution of project revenue.

activity were specified as variables, and the relationships' constraints of
resource amounts, work quantity in each activity and working regions
were specified as shown in Table 3. Table 4 lists possible maximum
productivities and unit costs (also defined using stochastic distribu-
tions) of manpower or equipment, as well as daily resource constraints
and required working area per unit resource. Table 5 lists material unit
cost, material required, and material productivity in each activity.
Table 6 lists total working area of each working region.
Also, the following financial conditions were assumed: 15% annual
interest rate, 10% overhead, 25% contingency, 15% profit, 8.25% sales
tax, and 17% markup.
The goals of this sample application were defined to be the arrival at
optimum crews for each project activity such that the project cash flow
and resulting net present value was the lowest, subject to a constraint on
the project duration being less than 5000 working hours.
Information is also to be posted on the optimized project databases
regarding: resource productivity for each activity, activity start time,
activity finish time, activity duration, activity cost, project start time,
project finish time, project duration, project cost, project net present
value, project revenue, resources' utilization percentages in the whole
project, project duration's probability distribution and cumulative
distribution, project cost's probability distribution and cumulative
distribution, the probability distribution and cumulative distribution
of project net present value, and cash flows of each activity and the
whole project.
The mean values of the unit costs and resource productivities
served as the simulation's start point (based on Tables 4 and 5). By
using ISS to run the above databases one thousand times, the
optimum solution was achieved at the end of the 796th simulation
run, as shown in Fig. 3 and tabulated in Tables 7 and 8.
The results obtained from the previous step were then assigned to
the resource amounts in each activity. The unit costs and productivities
of resources were defined as Tables 4 and 5. An additional 500
simulation runs produced the project databases shown in Figs. 4–13,
and Tables 9 and 10.
5.2.1. Discussion
Different from sample application I, the daily resource require-
ments and durations of activities, instead of being set as constants,
were obtained as simulation results in this sample application.

Fig. 11. Probability distribution of project net present value at project start time.
110 S.-M. Chen et al. / Automation in Construction 21 (2012) 99–113

Fig. 12. Cumulative distribution of project net present value at project start time.

Fig. 13. Cash flow forecast for the whole project.

Moreover, project duration, project cost, project revenue, minimum
project net present value, resource productivity, activity cost, activity
start time, activity finish time, probability distribution of project
duration/project cost/project net present value, cumulative distribu-
tion of project duration/project cost/project net present value,
resource utilization percentage, space utilization percentage, data-
bases of each activity, and cash flow forecasts of each activity and the
whole project were also obtained as part of the simulation results.
The aforementioned results can help the project managers to
manage their projects effectively. For example, from the resource

Table 9
Resource utilization percentage in the whole project.

Resource R1 R2 R3 R4 R5 R6

Mean resource utilization percentage 74.84% 71.42% 65.42% 59.06% 60.46% 55.92%
Maximum resource utilization percentage 100% 100% 100% 100% 100% 100%
 S.-M. Chen et al. / Automation in Construction 21 (2012) 99–113 111

Table 10
Space utilization percentage in the whole project.

Area name A1 A2 A3 A4 A5 A6 A7 A8 A9 A10

Mean space utilization percentage 6.23 7.21 16.22 4.62 3.38 12.24 9.82 6.45 6.48 13.25
Maximum space utilization percentage 90.67 100 89 93.5 99.20 98.50 90.40 86.29 98.29 95.50

The goals of the activity scheduling were defined to be the arrival

Fig. 14. Project network for case study (numbers indicate specific activity).
utilization percentage of each resource, the maximum utilization
percentages of all resources are 100%. This means that the project
duration may be extended if any of these resources' daily resource
limits are reduced. And, from the cumulative distribution of project
cost, with a 90% probability, the project cost will not exceed 7,250,000
dollars.
5.3. Case study
This case study presented below was conducted at Columbia
University in New York City. The construction site is a brotherhood
synagogue located at 28 Gramercy Park South, New York, New York
10003.
The construction schedule network, activity description, activity
duration, required resources for each activity, and daily resource
constraints are shown in Fig. 14, Tables 11 and 12. By using Primavera
to schedule these activities and allocate these resources with daily
resource constraints, the total project duration is projected to be
316 days.
Table 11
Activity description for each activity.
Activity ID Activity description Predecessors
at optimum crews for each project activity, so that the project
duration would be the shortest possible within the daily resource
constraints.
To test the ability of ISS to calculate the optimum resource
distribution and scheduling activities in this case study, the author
used the same schedule network, required resource amount and
activity duration for each activity shown in Fig. 14, Tables 11, and 12.
Fig. 15 shows how the brotherhood synagogue project was created
from the ISS template.
By using ISS to run these data one thousand times, the authors
obtained a project schedule with 292-day project duration.
5.3.1. Discussion
Although ISS produced a project schedule that is around 7.6% of
project time reduced compared with the time produced by using
Primavera, this case study does not completely illustrate ISS's
effectiveness in distributing resources, because the activity's duration
and required resources, instead to be obtained from ISS, are set to be
the same for both ISS and Primavera for using the same baseline
purpose.
6. Limitation and future research
Instead of using activity duration directly, the resource produc-
tivity is used in this research to decide activity duration, activity cost,
project duration, project cost, risk analysis on project duration and
project cost, and so on. However, the actual data of resource
productivity is not collected from job sites, and the probability
density function of best fit for modeling resource productivity
distribution is not studied and used in sample application II. The
above limitation of applying ISS can be an interesting issue for future
research.
7. Conclusions

32 Construct stairs 259 ISS uses simulation techniques to assign different levels of
38 Construct scaffolding 259
62 Construct girders for 4th floor 38
priorities to different activities in every simulation cycle to find
74 Walls, windows 202, 201, 209 284 near-optimum distributions of manpower, material, equipment and
80 Doors 200, 201, 202, 204 284 space according to their project objectives and project constraints.
86 Construct joists for 4th floor 62 This helps ISS to avoid the resulting schedule to be trapped in local
113 Complete 4th floor 86
optima and get rid of application limits.
259 Start N/A
265 Start demo 259 The major contributions of this model are listed as follows:
271 Fabricate millwork 259 (1) applies evolutionary techniques to decide the near-optimum
277 Demo for sprinkler and AC 259 distributions of manpower, equipment, material and space according
284 Finish 2nd floor 265 to project objectives and project constraints; (2) considers and
290 Sprinkler 277
integrates most of the important construction factors (schedule, cost,
296 Install duct 277
302 Start electrical 201, 202, 205 284 manpower, equipment, material and space) simultaneously in a
308 Install millwork 74, 271 unified environment, which makes the resulting schedule that will be
314 Ceiling 201, 202, 205 302 close to optimal; (3) improves risk analyses on project duration,
320 Complete electrical for 2nd floor 302
project cost, project revenue, and net present. And, the other tools and
327 Install lighting 314, 320
333 Take down scaffolding 113 concepts utilized in this model are listed as follows: (1) captures
339 Finish floor 80, 308, 327 quantity takeoffs from a 3D CAD model automatically; (2) develops
345 Ceiling (lights) 32, 290, 333 production rates of driving resources; (3) develops the utilization rate
352 Finish trim 339 of each resource/space allocation; (4) what-if analyses of possible
359 Finish up 296, 345, 352
scenarios; (5) schedule adjustments based on unforeseen conditions.
112 S.-M. Chen et al. / Automation in Construction 21 (2012) 99–113

Table 12
Activity duration, required resources for each activity, and daily resource limits.

ID Activity Daily resource requirements

duration
LF L EOL EOM BL HH FEL PRB C E CF R CEF GEV SWF SW SMW SMA LA
(days)

32 40 – 1 – – – – – – 2 3 1 1 1 1 – – – – –
38 20 – 4 – – – – – – 3 3 4 – – – – – – – –
62 60 – 1 – – – – – – 4 2 1 – – – – – – – –
74 15 – 1 – – – – – – 4 3 1 – – – – – – – –
80 20 – 2 – – – – – – 2 2 2 – – – – – – – –
86 90 – 2 – – – – – – 1 2 2 – – – – – – – –
113 30 – 3 – – – – – – 1 2 5 – – – – – – – –
259 0 – – – – – – – – – – – – – – – – – – –
265 4 1 2 1 1 1 1 1 1 5 1 2 – – – – – – – –
271 60 – 3 – – – – – – 2 2 3 – – – – – – – –
277 20 – 2 – – – – – – 6 1 5 – – – – – – – –
284 6 – 5 – – – – – – 1 4 4 – – – – – – – –
290 30 – 2 – – – – – – 4 4 2 – – – 2 8 – – –
296 30 – 4 – – – – – – 5 3 4 – – – – – 2 1 –
302 40 – 3 – – – – – – 4 1 3 – – – – – – – –
308 10 – 4 – – – – – – 1 3 2 – – – – – – – –
314 25 – 3 – – – – – – 3 4 3 – – – – – – – 1
320 20 – 2 – – – – – – 4 1 4 – – – – – – – –
327 10 – 3 – – – – – – 5 1 3 – – – – – – – –
333 20 – 2 – – – – – – 3 4 3 – – – – – – – –
339 10 – 4 – – – – – – 1 2 3 – – – – – – – –
345 30 – 3 – – – – – – 5 1 4 – – – – – – – –
352 5 – 2 – – – – – – 1 3 4 – – – – – – – –
359 1 – 4 – – – – – – 1 3 3 – – – – – – – –
Daily Resource 1 6 1 1 1 1 1 1 9 6 8 1 1 1 2 8 2 1 1
Limit

Where LF = Labor Foreman; L = Laborers; EOL = Equip. Oper. [light]; EOM = Equip. Oper. [medium]; BL = Backhoe Loader; HH = Hyd. Hammer; FEL = F.E. Loader; PRB = Pavt.
Rem. Bucket; C = Carpenters; E = Electricians; CF = Carpenter Foreman; R = Rodman; CEF = Cement Finisher; GEV = Gas Engine Vibrator; SWF = Skilled Worker Foremen; SW =
Skilled Workers; SMW = Sheet Metal Workers; SMA = Sheet Metal Apprentice; LA = Lathers.

In the sample application I presented herein, ISS obtained 45-day
project duration which saves around 8.2% of project time compared
with the result obtained by Microsoft Project and Primavera.
In the sample application II presented herein, most ISS outputs
were obtained as simulation results, which can help the project
managers to manage their projects effectively.
In the case study presented herein, the activity's duration and
required resources, instead to be obtained from ISS, are set to be the
same for both ISS and Primavera for using the same baseline
purpose. The 292-day project duration obtained from ISS saves
around 7.6% of project time compared with the result obtained by
Primavera.

Fig. 15. Brotherhood synagogue project modeled in ISS.

 S.-M. Chen et al. / Automation in Construction 21 (2012) 99–113
Supplementary materials related to this article can be found online
at doi:10.1016/j.autcon.2011.05.018.
References
[1] H. Ahuja, S.P. Dozzi, S.M. AbouRizk, Project management techniques in planning
and controlling construction projects, 2nd Ed. Wiley, N.Y., 1995
[2] M.-Y. Cheng. “Automated site layout of temporary facilities using geographic
information systems (GIS),” PhD dissertation, Civil Engineering Dept., University
of Texas, Austin, Tex. 1992.
[3] S. Cheung, T.K. Tong, C. Tam, Site pre-cast yard layout arrangement through
genetic algorithms, Autom. Constr. 11 (2002) 35–46.
[4] E. Elbeltagi, T. Hegazy, A. Eldosouky, Dynamic layout of construction temporary
facilities considering safety, J. Constr. Eng. Manage. 130 (4) (2004) 534–541.
[5] A. Hamiani, “CONSITE: a knowledge-based expert system framework for
construction site layout,” PhD dissertation, Civil Engineering Dept., University of
Texas, Austin, Tex. 1987.
[6] T. Hegazy, H. El-Zamzamy, Project management software that meet the challenge,
Cost Engineering, AACE International, Vol. 4, No. 5, 1998, pp. 25–33.
[7] T. Hegazy, Computer-based Construction Project Management, Prentice-Hall,
Upper Saddle River, N.J., 2002.
[8] T. Hegazy, H. El-Zamzamy, Project management software that meets the
challenge, Cost Eng. 40 (5) (1998) 25–32.
[9] T. Hegazy, E. Elbeltagi, EvoSite: evolutionary-based model for site layout planning,
J. Comput. Civ. Eng. 13 (3) (1999) 198–206.
[10] A. Khalafallah, K. El-Rayes, Trade-off between safety and cost in planning
construction site layouts, J. Constr. Eng. Manage. 131 (11) (2005) 1186–1195.
[11] A. Khalafallah, K. El-Rayes, Minimizing construction related hazards in airport
expansion projects, J. Constr. Eng. Manage. 132 (6) (2006) 562–572.
[12] H. Li, P. Love, Using improved genetic algorithms to facilitate time-cost
optimization, J. Constr. Eng. Manage. 123 (3) (1997) 233–237.
[13] K.-L. Lin, C.T. Haas, An interactive planning environment for critical operations, J.
Constr. Engr. Manage. ASCE 122 (3) (1996) 212–222.
[14] M.J. Mawdesley, S.H. Al-jibouri, H. Yang, Genetic algorithms for construction site
layout in project planning, J. Constr. Eng. Manage. 128 (5) (2002) 418–426.
[15] B. Montreuil, U. Venkatadri, Strategic interpolative design of dynamic manufacturing
systems layouts, Manage. Sci. 37 (6) (1991) 682–693.
113
[16] H. Osman, M.E. Georgy, M.E. Ibrahim, A hybrid CAD based construction site layout
planning system using genetic algorithms, Autom. Constr. 12 (2003) 749–764.
[17] H. Osman, M.E. Georgy, M.E. Ibrahim, An automated system for dynamic
construction site layout planning, Proc., 10th Int. Colloquium on Structural and
Geotechnical Engineering, Ain Shams University, Cairo, Egypt, 2003.
[18] D.R. Riley, “Modeling the space behavior of construction activities,” PhD
dissertation, Dept. of Architectural Engrg., Penn State University, University
Park, Pa. 1994.
[19] W.E. Rodriguez-Ramos, “Quantitative techniques for construction site layout
planning,” PhD dissertation, University of Florida, Gainesville, Fla. 1982.
[20] M.J. Rosenblatt, The dynamics of plant layout, Manage. Sci. 32 (1) (1986) 76–86.
[21] A. Sawhney, O. Abudayyeh, T. Chaitavatputtiporn, Modeling and Analysis of a
Concrete Production Plant using Colored Petri Nets, Journal of Computing in Civil
Engineering, American Society of Civil Engineers (ASCE), 1999 To Appear in July
1999 Issue.
[22] D.M. Smith, An investigation of the space constraint problem in construction
planning, Major Paper, MS, Virginia Polytechnic Institute and State University,
Blacksburg, Va, 1987.
[23] F. Talbot, J. Patterson, Optimal methods for scheduling projects under resource
constrains, Proj. Manage. Q. (Dec. 1979) 26–33.
[24] C.M. Tam, T.K. Tong, W.K.W. Chan, Genetic algorithms for optimizing supply
locations around tower crane, J. Constr. Eng. Manage. 127 (4) (2001) 315–321.
[25] W.Y. Thabet, “A space-constrained resource-constrained scheduling system for
multi-story buildings,” PhD dissertation, Civil Engineering Dept., Virginia
Polytechnic Institute and State University, Blacksburg, Va. 1992.
[26] I.D. Tommelein. “SightPlan—an expert system that models and augments human
decision-making for designing construction site lay-outs,” PhD dissertation, Dept.
of Civil Engineering, Stanford University, Stanford, Calif. 1989.
[27] I.D. Tommelein, Site layout: where should it go? Proc., Constr. Congr., 91, ASCE,
New York, 1991, pp. 632–637.
[28] I.D. Tommelein, P.P. Zouein, Interactive dynamic layout planning, J. Constr. Eng.
Mgmt. ASCE 119 (2) (1993) 266–287.
[29] A. Warszawski, S. Peer, Optimizing the location of facilities on a building site,
Oper. Res. Q. 24 (1) (1973) 35–44.
[30] I.C. Yeh, Construction-site layout using annealed neural network, J. Comp. Civ.
Eng. ASCE 9 (3) (1995) 201–208.
[31] P.P. Zouein, H. Harmanani, A. Hajar, Genetic algorithms for solving site layout
problem with unequal-size and constrained facilities, J. Comput. Civ. Eng. 16 (2)
(2002) 143–151.