You are on page 1of 6

2011 Third International Conference on Computational Intelligence, Modelling & Simulation

Comparison of Two Simulation Software Systems for Modeling a Construction


Process
Amin Nikakhtar

Kuan Yew Wong

Faculty of Mechanical Engineering


Universiti Teknologi Malaysia
Skudai, Malaysia
Aminnikakhtar1987@gmail.com

Faculty of Mechanical Engineering


Universiti Teknologi Malaysia
Skudai, Malaysia
wongky@fkm.utm.my

Mohammad Hossein Zarei

Ashkan Memari

Faculty of Mechanical Engineering


Universiti Teknologi Malaysia
Skudai, Malaysia
Zarei_2002@yahoo.com

Faculty of Mechanical Engineering


Universiti Teknologi Malaysia
Skudai, Malaysia
ashkan.eng@gmail.com
Examples of simulation software used in simulating
construction processes are SDESA, MicroCYCLONE,
STROBOSCOPE, etc. In addition to these packages, there
are generic simulation packages with better graphical
environments. ARENA 13 and WITNESS 2004
Manufacturing Edition are two simulation tools with
powerful animation features.
This paper is aimed at comparing the features of two
simulation software, WITNESS 2004 Manufacturing
Edition and ARENA 13. To do this, a construction process
is simulated using ARENA 13 and WITNESS 2004
Manufacturing Edition. Also, the reports generated by both
of the simulation software are compared in order to
provide a valid comparison.
The paper is presented as follows. First, the difficulties
in construction industry which bring about the application
of simulation in construction are discussed. Second,
different simulation software and languages used for
simulating a construction process are reviewed. Third, two
simulation models are constructed for a case study,
concrete pouring of beams and slabs, using ARENA 13
and WITNESS 2004 Manufacturing Edition. Fourth, these
models are verified and validated. Finally, the reports are
compared in order to highlight different features of each of
the software.

Abstract Construction process complexities lead to many


difficulties imposed on construction planners and designers
who are facing with different issues such as developing new
methods or solving problems during a construction process.
In order to solve these problems in construction industry,
simulation can be an acceptable solution. In this regard,
graphical methods have become a useful tool for process
simulation. To do construction process simulation with a
suitable graphical display, many simulation software are
developed such as PROMODEL, SDESA, and etc. This
paper aims at comparing two simulation tools (software),
ARENA 13 and WITNESS 2004 Manufacturing Edition, for
making a construction process simulation model. It shows
that both software produce almost the same results and
outputs for a given construction process. Also, the paper
shows different kinds of features and reports provided by
ARENA 13 and WITNESS 2004 Manufacturing Edition.
Keywords-construction process simulation;simulation
software;ARENA 13;WITNESS 2004 Manufactring Edition

I.

INTRODUCTION

The Designing of a construction process is becoming


more complicated in construction industry. This
complexity leads to many difficulties imposed on
construction planners and designers who are faced with
different issues such as developing new methods or
solving problems during a construction process. Because
of the existence of complex relationships in construction
processes, describing a construction process is difficult [1].
In order to solve these problems in construction industry,
simulation can be an acceptable solution which is applied
for designing the construction operations and analyzing
their behavior [2, 3].
With the advancement of graphical computer
techniques, graphical methods have attracted attention to
themselves for using them as useful tools for process
simulation [4]. There are various simulation software that
provide environments to implement any new principle.
978-0-7695-4562-2/11 $26.00 2011 IEEE
DOI 10.1109/CIMSim.2011.42

II.

COMPUTER SIMULATION OF CONSTRUCTION


PROCESSES

In daily construction practices, many decisions related


to the construction processes are made by construction
designers. In some cases, decisions are made with
unforeseen outcomes. The reason for this is that
visualizing all the processes of construction operations is
difficult [1].
In addition, in real-life, it is preferable to test a
construction process to examine its performance prior to
implementation. From an economic perspective, it is not
suitable to test the process physically since the physical
200

testing of a process is time consuming and very expensive


[5].
In order to cope with the above mentioned issues and
problems, simulation can be a useful tool. It is a
convenient technique used to model actual construction
operations. It provides an environment in which decision
makers can better analyze and understand the problems by
doing experiments in a low-cost environment [6].
III.

Automation as a new generation simulation environment.


It uses the SIMAN processor and simulation language. It
can be used for both commercial and academic purposes
[13]. The fundamental elements of modeling in ARENA
13 are called modules that are built around SIMAN. In
order to build a simulation model with ARENA 13, a
modeler first selects the appropriate modules from a
template panel and places them on the simulation
environment. After that, these modules are connected to
each other graphically in order to show the flow of
transactions between them. Finally, the required and
considered codes are written in each module [14].
On the other hand, WITNESS 2004 Manufacturing
Edition is simulation software that is used for designing
systems and demonstrating their operation in reality.
WITNESS 2004 Manufacturing Edition can help people
by predicting and solving problems related to production
bottlenecks, overly-idle resources, etc [15]. In WITNESS
2004 Manufacturing Edition, models are built based on
fundamental parts called elements or building blocks.
Anything used for constructing a model in WITNESS
2004 Manufacturing Edition is an element. In order to
build a simulation model first the required elements are
defined then, displayed and detailed. In contrast to
ARENA 13 that modules are connected to each other
graphically, in WITNESS 2004 Manufacturing Edition,
the connections can be defined as codes in each element.
After defining the connections, finally, the required codes
are written in each element.

CONSTRUCTION PROCESS SIMULATION SYSTEMS

Using simulation in construction processes has been


broadly done in various situations. For instance,
introducing lean construction concept into construction
industry caused simulation to be applied in cases
pertaining to work flow variability [7]. Also, Tommelin
[8] evaluated the applicability of a production concept in a
pipe-spool installation process with the aid of simulation.
Today there are various simulation software developed
for simulating different construction processes such as
SDESA, PROMODEL, MicroCYCLONE, Extend,
STROBOSCOPE, etc. In the next section a brief review of
different features of the mentioned software is done.
IV.

COMPUTER SIMULATION SOFTWARE

Simulation tools provide a realistic approach for doing


analysis since they considered the randomness in activity
duration and the influence of resource availability as a
constraint to construction work flow [9]. Typical
simulation software used in construction process
simulation are SDESA, MicroCYCLONE, and
STROBOSCOPE. SDESA is a simple platform based on
activity modeling that makes construction simulation as
easy as applying the current construction planning
approaches such as critical path method (CPM).
MicroCYCLONE is a simulation program developed
specifically for modeling cyclic processes. It is based on
CYCLONE modeling elements. CYCLONE, which stands
for cyclic operation network, is the best known simulation
language designed for simulating construction operations
and activities [10]. Finally, STROBOSCOPE is a
simulation language applied for construction activity
simulation. It can also be used in domains other than
construction industry [11]. It aims at modeling
uncertainties related to activity duration, quantity, and
resource assignment [12].
In addition to the above simulation software, there are
computer simulation packages which have more attractive
graphical environments. ARENA 13 and WITNESS 2004
Manufacturing Edition are two examples of this type. In
the next section, a brief review of different features of
these two software packages is done and after that the rest
of the article is about doing a simulation study for a
construction operation using theses two software.
V.

VI.

CASE STUDY: A CONCRETE POURING OPERATION OF


BEAMS AND SLABS

The case study is part of a concrete building


construction located in the city of Mashhad, Iran. It
consists of two floors and each floor, according to design
specifications, needs 420 cubic meters of concrete to be
completed. The concrete operations are divided into two
main operations. The first one is concrete pouring of slabs
and beams and the second is concrete pouring of walls and
columns. In this paper, the authors aim at focusing on
concrete pouring operation of beams and slabs which
needs 301 cubic meters of concrete. The operation applies
mobile pumping for concrete pouring. Concrete trucks that
contain seven cubic meter of concrete are used for hauling
concrete to the construction site.
VII. PROCESS MAPPING AND DATA COLLECTION
In the first step toward building the simulation model
of a construction process, activities of the process and their
sequences, labors and resources used, and flow of the work
were determined based on the actual behavior.
Interactions between resources, activities, and the flow of
information in any construction process can be depicted by
process mapping. Figure 1 illustrates the schematic
process map of concrete pouring operations using
operation process chart (OPC) symbols. More details of
the process are considered in the simulated model.

OVERVIEW: ARENA 13 VS. WITNESS 2004


MANUFACTURING EDITION

ARENA 13 is simulation software developed by


System Modeling Company and acquired by Rockwell

201

model reflects the actual behavior of the process [17]. In


the next two sections, the verification and validation are
done for both of the models constructed by ARENA 13
and WITNESS 2004 Manufacturing Edition.

A simulation model of a construction process needs


random durations for each activity. Therefore, after
developing the process map of the construction process,
data related to the duration of each activity were gathered.
Having collected the data sets, a probability distribution
should be fitted to each data set in order to reflect the
randomness of the process. For doing data collection, the
stop watch method was used in order to gather activity
duration data. In this method, first of all the activities were
recorded by a camera and after that the duration of each
activity was recorded using a chronometer. It should be
noted that recording should be done in a way that do not
affect the performance of workforces. For fitting a
probability distribution to a sample data, various computer
packages can be used. EasyFit, which was used in this
study, is one commercial package that fits a wide variety
of distributions to sample observations. Using EasyFit,
many continuous distribution functions (such as
Exponential, Beta, Gamma, Uniform, etc.) were tested
against the collected data, and the most promising one
according to the goodness-of-fit tests is selected. Actually,
the goodness-of-fit tests (the chi-square, Anderson Darling
and the KolmogorovSmirnov tests) were used to validate
the assumed distribution functions.

Figure 2. Simulated model of concrete pouring process by WITNESS


2004

Next
member

Construction
Site Entrance

Concrete
hauling

Spreading

Yes
Vibrating
Is pump
busy?
(Queue)

NO

Slump
Testing

Concrete
pumping

Concrete
truck
finished?

NO

Next
member

Yes

Figure 3. Simulated model of concrete pouring process by ARENA 13

Finishing

Waiting

X.
VIII. MODEL DEVELOPMENT

Having defined the best probability distributions, it is


time to construct the simulation model of the considered
construction process. The process map, distributions
parameters, and actual behaviors were used to accurately
model the conventional concrete pouring operations via
ARENA 13 and WITNESS 2004 Manufacturing Edition.
Figures 2 and 3 illustrate the models constructed by
WITNESS 2004 Manufacturing Edition and ARENA 13,
respectively.
IX.

MODEL VERIFICATION

Verification, more specifically, includes inspection of


the logic of the model, performing simulation test runs,
tracing the entities in sample path trajectories, and
evaluating the consistency of statistics of the model [19].
In order to verify the models built in ARENA 13 and
WITNESS 2004 Manufacturing Edition, transactions were
checked to see if they go where they are supposed to go
and if they do what they are supposed to do under every
condition.
For instance, performance verification of labor 2 in
both models is explained in detail through the precise
tracing of the activities behavior in the model. To do so,
the labor 2s performance results are compared with the
total throughput in a random run of the model. For
ARENA 13, results of a test run in Table I demonstrate
that labor 2 has been busy in 46.44% of the total process
duration which is 186.11 minutes. It means that the time
that labor 2 spends on pumping a cubic meter of concrete
in the entire concrete pouring process is equal to 86.43
minutes (46.44%*186.11). On the other hand, according to
collected data and fitted distribution for pumping, labor 2

Figure 1. General process mapping of concrete operations

VERIFICATION AND VALIDATION OF THE MODEL

A process of modeling will be successful when the


simulated model accurately reflects the present workflow
process. Therefore, before any experimentation is done, it
is essential to verify and validate the model [16, 17, and
18]. Model verification ensures that the model behaves as
expected and it does not have any logical error. On the
other hand, model validation ensures that the simulated

202

estimate of m number of runs; = level of confidence; =


allowable percentage of error; and t
,
/ = critical
value of the two-tailed t-distribution at a level of
significance, given m-1 degrees of freedom.
The mean and standard deviation estimate are
determined for an initial m number of runs of five. Then at
a level of confidence of 95% and allowable percentage of
error of 5%, t , . is equal to 2.776. Equation (1)
represents that the number of simulation runs to achieve
the desired level of accuracy is 4 replicates or greater.
Table III shows the calculation of X m and S m . It
should be noted that the data gathered for estimating the
mean and standard deviation are the cycle time of the
construction process achieved by running the model.
After determining the number of simulation runs,
validation was done for the simulation model constructed
by both ARENA 13 and WITNESS 2004 Manufacturing
Edition. As mentioned before, the actual cycle time was
compared with the results of the five simulation runs.
Table IV shows the data for validation. It shows that the
averages of five simulation runs are 197.82 and 199 for
ARENA 13 and WITNESS 2004 Manufacturing Edition
model, respectively. As can be seen in Table IV, the
percentage variations in simulation prediction are 3.88 and
3.40 for ARENA 13 and WITNESS 2004 Manufacturing
Edition respectively, which is considered acceptable.

averagely pumps one cubic meter of concrete in 0.93


minutes. Therefore labor 2 pumps the concrete 92.93 times
(86.37/0.93) in the examined test run. The difference
between the above total seized number and the number
produced by ARENA 13 is 2.12% which is considered
acceptable for the model. For WITNESS 2004
Manufacturing Edition, the procedure is the same as for
ARENA 13. As can be seen from the results of WITNESS
2004 Manufacturing Edition in Table II, labor 2 has been
busy in 44.68% of the total process duration which is
190.51 minutes. Therefore, the time labor 2 spends on
pumping is 85.119 minutes (44.68*190.51). Given that the
average time labor 2 spends on pumping one cubic meter
of concrete is 0.93 minutes, labor 2 pumps the concrete
91.52 times. The calculated number of jobs is equal to the
number calculated by WITNESS 2004 Manufacturing
Edition, shown in Table II. Similarly, all the transactions,
modules, linkages and resources were carefully examined
and subsequently, necessary modifications were done to
solve the probable problems and verify the model.
TABLE I.

RESOURCE RESULTS OF THE MODEL FOR LABOR 2 IN


ARENA 13

Usage
Instantaneous Utilization
Scheduled Utilization
Number Scheduled
Total Number Seized
Number Busy

TABLE II.

Labor2
Average
0.464
0.464
1.000
91.000
0.464

Minimum
0
1.000
0

Maximum
1.000
1.000
1.000

TABLE III.

CALCULATION OF MEAN AND STANDARD DEVIATION


ESTIMATE FOR DETERMINING THE NUMBER OF SIMULATION RUNS

RESOURCE RESULTS OF THE MODEL FOR LABOR 2 IN


WITNESS 2004 MANUFACTURING EDITION
Name
%Busy
%Idle
Quantity
No. Of Jobs Started
No. Of Jobs Ended
No. Of Jobs Now
Avg Job Time

XI.

Simulation Cycle Time (Minutes)

Labor2
44.68
55.32
1
91
91
0
1.03

Replication

204.51

200.76

196.09

201.61

186.11

197.816
7.209

MODEL VALIDATION

Once the examiner is convinced with the verification


stage, validation activities can get under way. Model
validation is a critical step to construct a credible
simulation model. The standard approach of validation is
to compare the collected actual data to the simulated
model outputs. One of the appropriate factors to show how
actual process and simulated process are alike is cycle time
[19]. In order to do validation, first the number of
simulation runs to produce the desired level of accuracy
should be determined. To do so, Ahmed [20] proposed the
following formula, given the m initial replications:
,

TABLE IV.

(1)

Where N m = number of simulation runs to achieve


the desired level of accuracy; X m = the mean estimate of
an initial m number of runs; S m = the standard deviation

203

Replication

Simulated
Cycle Time
(Minutes)ARENA 13

1
2
3
4
5
Average
Variation (%)

204.51
200.76
196.09
201.61
186.11
197.82
3.88

VALIDATION DATA
Simulated Cycle
Time (Minutes)WITNESS 2004
Manufacturing
Edition
198.27
206.57
190.51
189.78
208.84
198.79
3.40

Actual
Cycle Time
(Minutes)

205.8

A significant feature of ARENA 13 is the ability to


categorize the activities into different types such as valueadded, non value-added, waiting, transfer, and others. This
is very useful in determining the different time spent on
different types of processes. This can help a manager to
improve the system by minimizing non value-adding
processes. However, this is not provided by WITNESS
2004 Manufacturing Edition. Tables VIII and IX show the
process reports of the case study provided by WITNESS
2004 Manufacturing Edition and ARENA 13, respectively.

XII. REPORTS ANALYSIS


After verifying and validating the simulation model, it
is time to get the results produced by the software.
ARENA 13 and WITNESS 2004 Manufacturing Edition
provide simulation reports that can be helpful for
evaluating the simulated system. These reports include
report of the part (entity) performance, labor performance,
machine performance, etc. This section aims at presenting
some typical reports generated by ARENA 13 and
WITNESS 2004 Manufacturing Edition in order to show
different features of the software. As an instance, the
queue (buffer) report generated by ARENA 13 is shown in
Table V. Also, WITNESS 2004 Manufacturing Edition
generates this report as shown in Table VI. As can be seen
in both tables, the results produced by ARENA 13 and
WITNESS 2004 Manufacturing edition are very near to
each other. As an instance, in ARENA 13, the average
amount of time that an entity, one cubic meter of concrete,
waits in the queue of Finishing is 0.31 minutes while in
WITNESS 2004 Manufacturing Edition this time is 0.39
minutes. This minor difference shows that these two
simulation packages generate the same output for a
verified and valid model.
An attractive feature of the reports generated by
WITNESS 2004 Manufacturing Edition is the charts
generated by each report. For example, figure 4 represents
a labor report produced by WITNESS 2004 Manufacturing
Edition in which the bar chart of the report table is also
presented. However, ARENA 13 does not have this feature
for generating the chart of each report. The labor report
produced by ARENA 13 is shown in Table VII in order to
compare it with the labor report of WITNESS 2004
Manufacturing Edition.

Figure 4.

Labor report generated by WITNESS 2004


Manufacturing Edition

TABLE VII.
TABLE V.

QUEUE REPORT GENERATED BY ARENA 13

Queue Detail Summary


Time
Name
Waiting Time (Minute)
Spreading
2.53
Vibrating
0.08
Finishing
0.31
Other
Name
Number Waiting
Spreading
1.12
Vibrating
0.04
Finishing
0.14

Name
Labor1
Labor2
Labor3
Labor4
Labor5
Labor6
Labor7

Inst
Util
0.18
0.46
0.38
0.38
0.36
0.36
0.43

TABLE VIII.
TABLE VI.

QUEUE REPORT GENERATED BY WITNESS 2004


MANUFACTURING EDITION

Name
Total In
Total out
Now In
Max
Min
Avg Size
Avg Time (Minute)

Vibrating
Queue
91
91
0
1
0
0.05
0.12

Finishing
Queue
91
91
0
2
0
0.17
0.39

Spreading
Queue
91
91
0
7
0
1.07
2.45

204

LABOR REPORT GENERATED BY ARENA 13


Resource Detail Summary
Usage
Num
Num
Num
Busy
Sched
Seized
0.18
1.00
13.00
0.46
1.00
91.00
0.38
1.00
91.00
0.38
1.00
91.00
0.36
1.00
91.00
0.36
1.00
91.00
0.43
1.00
91.00

Sched
Util
0.18
0.46
0.38
0.38
0.36
0.36
0.43

PROCESSES REPORT GENERATED BY WITNESS 2004


MANUFACTURING EDITION

Name

%Idle

%Busy

No. Of
Operations

Pumping

55.32

44.68

91

Spreading

64.35

35.65

91

Vibrating

66.41

33.59

91

Finishing

58.64

41.36

91

Ready to Pump

82.52

17.48

13

TABLE IX.

PROCESSES REPORT GENERATED BY ARENA 13

[5]

Process Detail Summary


Time per Entity
Value Added
Time
(Minute)
1.28
0.97
1.04
1.04
2.86
2.86
3.38
0.85
0.88
0.80
Accumulated Time
Value Added
Wait Time
Time
(Minute)
(Minute)
88.56
27.77
94.98
0.00
37.22
0.00
77.66
229.92
72.90
7.54
Others
Number In
Number Out
91.00
91.00
91.00
91.00
13.00
13.00
91.00
91.00
91.00
91.00
Total Time
(Minute)

Finishing
Pumping
Ready to Pump
Spreading
Vibrating

Finishing
Pumping
Ready to Pump
Spreading
Vibrating

Finishing
Pumping
Ready to Pump
Spreading
Vibrating

[6]

Wait Time
(Minute)
0.31
0.00
0.00
2.53
0.08

[7]

[8]

[9]

[10]

[11]

[12]
[13]
[14]

XIII. CONCLUSION

[15]

Simulation software provides a realistic approach for


analysis by means of graphical methods and visualization.
In this study, two simulation packages, ARENA 13 and
WITNESS 2004 Manufacturing Edition were used for
developing a construction process model in order to
compare them with each other. The results generated by
both packages are fairly valid and realistic. It was shown
that the outputs produced by ARENA 13 and WITNESS
2004 Manufacturing Edition are almost the same. Also,
different features of the software were compared with each
other. WITNESS 2004 Manufacturing Edition can present
bar charts for each report while ARENA 13 does not have
this feature. On another aspect, ARENA 13 is able to
classify the activities under one or more specific groups
but WITNESS 2004 Manufacturing Edition does not offer
this feature.

[16]

[17]

[18]

[19]
[20]

REFERENCES
[1]

[2]

[3]

[4]

S. Al-Masalha, A Common taxonomy for modeling construction


operations, PhD Dissertation, Virginia Polytechnic Institute and
State University, USA, 2004.
A. OpdenBosch, Design/Construction Processes Simulation in
real-time object-oriented Environments. Department of Civil
Engineering. Atlanta, GA, Georgia Institute of Technology, USA,
1994.
K. Naji, The development of virtual environment for simulating
equipment-based construction operation in real-Time objectoriented systems, Department of Civil Engineering, USA,
University of Florida, 1997.
P. Ghoddousi, A. Nikakhtar, A. Abbasian Hoseini, Improvement
of Concrete Pouring Process by Integrating Lean
Principles,Computer Simulation and Sensitivity Analysis, Proc.
the 2011 International Conference on Technological
Advancements in Civil Engineering, chennai, India., 2011, IEEE
press

205

X. Mao, X. Zhang, Construction Process Reengineering by


Integrating Lean Principles and Computer Simulation
Techniques, Journal of construction engineering and management
ASCE, vol. 371, pp. 134-135, May 2008
S. Wang, D. W. Halpin, Simulation experiment for improving
construction processes, Proc. the 2004 Winter Simulation
Conference, West Lafayette, USA.
A. Sawhney, H. Bashford, K. Walsh, and A. R Mulky, Agentbased Modeling and Simulation in Construction, Proc. the 2003
Winter Simulation Conference, Piscataway, New Jersey.
I. Tommelein, Models of lean construction processes: example of
pipe-spool materials management, Proc. Construction Congress
V, Minneapolis, pp. 405-13, October 1997.
S. AbouRizk, D. Hajjar, A framework for applying simulation in
the construction industry, Canadian journal of civil engineering,
vol. 25, pp. 604617, 1998.
A. Gonzalez-Quevedo, S. AbouRizk, D. Iseleyf, D. W. Halpin,
Comparison of two simulation methodologies in construction,
Journal of Construction Engineering and Management, vol. 119,
pp. 573-589, 1993
T. Alves, I. Tommelein, G. Ballard, Simulation as a tool for
production system design in construction, Proc. IGLC-14,
Santiago, Chile, 2006.
I. Tommelein, Discerete-event simulation of lean construction
processes, Pro. IGLC-5, Gold Coast, 1997.
WWW. Arenasimulation.com/arena_Home.aspx
T. Altiok and B. Melamed, Simulation Modeling and Analysis
with ARENA. Academic Press, NJ: Rutgers University, 2007
P. L. Market, M. H. Mayer, WITNESS simulation software a
flexible suite of simulation tools, Proc. the 1997 Winter
Simulation Conference, USA.
A. Al-Sudairi, J. Diekmann, A. Songer, and H. Brown,
Simulation of construction processes: traditional practices versus
lean principles, Proc. 7th Annual Conference of International
Group for Lean Construction, Berkeley, CA, 1999.
A. Al-Sudairi, Evaluating the effect of construction process
characteristics to the applicability of lean principles, Construction
Innovation, vol. 7, pp. 99-121, 2007.
M. Hassan, and S. Gruber, Simulation of concrete paving
operations on Interstate-74, Journal of Construction Engineering
and Management, vol. 134 (1), pp. 2-9, 2008.
T. Altiok, and B. Melamed, Simulation modeling and analysis with
ARENA.Elsevier, p 66, 2007.
K. Ahmed, Modeling Drivers' Acceleration and Land Changing
Behavior, PhD thesis, ITS Program, Massachussettes Institute of
Technology, Cambridge, MA, February 1999.

You might also like