Harrell−Ghosh−Bowden: I. Study Chapters 1.

Introduction to © The McGraw−Hill

Simulation Using Simulation Companies, 2004
ProModel, Second Edition

C H A P T E R

1 INTRODUCTION
TO SIMULATION

“Man is a tool using animal. . . . Without tools he is nothing, with tools he

is all.”
—Thomas Carlyle

1.1 Introduction
On March 19, 1999, the following story appeared in The Wall Street Journal:
Captain Chet Rivers knew that his 747-400 was loaded to the limit. The giant plane,
weighing almost 450,000 pounds by itself, was carrying a full load of passengers and
baggage, plus 400,000 pounds of fuel for the long flight from San Francisco to
Australia. As he revved his four engines for takeoff, Capt. Rivers noticed that San
Francisco’s famous fog was creeping in, obscuring the hills to the north and west of
the airport.
At full throttle the plane began to roll ponderously down the runway, slowly at first
but building up to flight speed well within normal limits. Capt. Rivers pulled the throt-
tle back and the airplane took to the air, heading northwest across the San Francisco
peninsula towards the ocean. It looked like the start of another routine flight. Suddenly
the plane began to shudder violently. Several loud explosions shook the craft and
smoke and flames, easily visible in the midnight sky, illuminated the right wing.
Although the plane was shaking so violently that it was hard to read the instruments,
Capt. Rivers was able to tell that the right inboard engine was malfunctioning, back-
firing violently. He immediately shut down the engine, stopping the explosions and
shaking.
However this introduced a new problem. With two engines on the left wing at full
power and only one on the right, the plane was pushed into a right turn, bringing it
directly towards San Bruno Mountain, located a few miles northwest of the airport.
Capt. Rivers instinctively turned his control wheel to the left to bring the plane back
on course. That action extended the ailerons—control surfaces on the trailing edges
of the wings—to tilt the plane back to the left. However, it also extended the

3
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

4 Part I Study Chapters

spoilers—panels on the tops of the wings—increasing drag and lowering lift. With
the nose still pointed up, the heavy jet began to slow. As the plane neared stall speed,
the control stick began to shake to warn the pilot to bring the nose down to gain air
speed. Capt. Rivers immediately did so, removing that danger, but now San Bruno
Mountain was directly ahead. Capt. Rivers was unable to see the mountain due to the
thick fog that had rolled in, but the plane’s ground proximity sensor sounded an au-
tomatic warning, calling “terrain, terrain, pull up, pull up.” Rivers frantically pulled
back on the stick to clear the peak, but with the spoilers up and the plane still in a
skidding right turn, it was too late. The plane and its full load of 100 tons of fuel
crashed with a sickening explosion into the hillside just above a densely populated
housing area.
“Hey Chet, that could ruin your whole day,” said Capt. Rivers’s supervisor, who
was sitting beside him watching the whole thing. “Let’s rewind the tape and see what
you did wrong.” “Sure Mel,” replied Chet as the two men stood up and stepped out-
side the 747 cockpit simulator. “I think I know my mistake already. I should have used
my rudder, not my wheel, to bring the plane back on course. Say, I need a breather
after that experience. I’m just glad that this wasn’t the real thing.”
The incident above was never reported in the nation’s newspapers, even though
it would have been one of the most tragic disasters in aviation history, because it
never really happened. It took place in a cockpit simulator, a device which uses com-
puter technology to predict and recreate an airplane’s behavior with gut-wrenching
realism.

The relief you undoubtedly felt to discover that this disastrous incident was
just a simulation gives you a sense of the impact that simulation can have in avert-
ing real-world catastrophes. This story illustrates just one of the many ways sim-
ulation is being used to help minimize the risk of making costly and sometimes
fatal mistakes in real life. Simulation technology is finding its way into an in-
creasing number of applications ranging from training for aircraft pilots to the
testing of new product prototypes. The one thing that these applications have in
common is that they all provide a virtual environment that helps prepare for real-
life situations, resulting in significant savings in time, money, and even lives.
One area where simulation is finding increased application is in manufactur-
ing and service system design and improvement. Its unique ability to accurately
predict the performance of complex systems makes it ideally suited for systems
planning. Just as a flight simulator reduces the risk of making costly errors in ac-
tual flight, system simulation reduces the risk of having systems that operate inef-
ficiently or that fail to meet minimum performance requirements. While this may
not be life-threatening to an individual, it certainly places a company (not to men-
tion careers) in jeopardy.
In this chapter we introduce the topic of simulation and answer the following
questions:

• What is simulation?
• Why is simulation used?
• How is simulation performed?
• When and where should simulation be used?
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

Chapter 1 Introduction to Simulation 5

• What are the qualifications for doing simulation?

• How is simulation economically justified?
The purpose of this chapter is to create an awareness of how simulation is
used to visualize, analyze, and improve the performance of manufacturing and
service systems.

1.2 What Is Simulation?

The Oxford American Dictionary (1980) defines simulation as a way “to repro-
duce the conditions of a situation, as by means of a model, for study or testing or
training, etc.” For our purposes, we are interested in reproducing the operational
behavior of dynamic systems. The model that we will be using is a computer
model. Simulation in this context can be defined as the imitation of a dynamic sys-
tem using a computer model in order to evaluate and improve system perfor-
mance. According to Schriber (1987), simulation is “the modeling of a process or
system in such a way that the model mimics the response of the actual system to
events that take place over time.” By studying the behavior of the model, we can
gain insights about the behavior of the actual system.

Simulation is the imitation of a dynamic system

using a computer model in order to evaluate and
improve system performance.

In practice, simulation is usually performed using commercial simulation

software like ProModel that has modeling constructs specifically designed for
capturing the dynamic behavior of systems. Performance statistics are gathered
during the simulation and automatically summarized for analysis. Modern simu-
lation software provides a realistic, graphical animation of the system being mod-
eled (see Figure 1.1). During the simulation, the user can interactively adjust the
animation speed and change model parameter values to do “what if” analysis
on the fly. State-of-the-art simulation technology even provides optimization
capability—not that simulation itself optimizes, but scenarios that satisfy defined
feasibility constraints can be automatically run and analyzed using special goal-
seeking algorithms.
This book focuses primarily on discrete-event simulation, which models the
effects of the events in a system as they occur over time. Discrete-event simula-
tion employs statistical methods for generating random behavior and estimating
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

6 Part I Study Chapters

FIGURE 1.1
Simulation provides animation capability.

model performance. These methods are sometimes referred to as Monte Carlo

methods because of their similarity to the probabilistic outcomes found in games
of chance, and because Monte Carlo, a tourist resort in Monaco, was such a
popular center for gambling.

1.3 Why Simulate?

Rather than leave design decisions to chance, simulation provides a way to validate
whether or not the best decisions are being made. Simulation avoids the expensive,
time-consuming, and disruptive nature of traditional trial-and-error techniques.

Trial-and-error approaches are

expensive, time consuming, and disruptive.

With the emphasis today on time-based competition, traditional trial-and-error

methods of decision making are no longer adequate. Regarding the shortcoming of
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

Chapter 1 Introduction to Simulation 7

trial-and-error approaches in designing manufacturing systems, Solberg (1988)

notes,
The ability to apply trial-and-error learning to tune the performance of manufacturing
systems becomes almost useless in an environment in which changes occur faster than
the lessons can be learned. There is now a greater need for formal predictive method-
ology based on understanding of cause and effect.
The power of simulation lies in the fact that it provides a method of analysis
that is not only formal and predictive, but is capable of accurately predicting the
performance of even the most complex systems. Deming (1989) states, “Manage-
ment of a system is action based on prediction. Rational prediction requires sys-
tematic learning and comparisons of predictions of short-term and long-term
results from possible alternative courses of action.” The key to sound manage-
ment decisions lies in the ability to accurately predict the outcomes of alternative
courses of action. Simulation provides precisely that kind of foresight. By simu-
lating alternative production schedules, operating policies, staffing levels, job
priorities, decision rules, and the like, a manager can more accurately predict out-
comes and therefore make more informed and effective management decisions.
With the importance in today’s competitive market of “getting it right the first
time,” the lesson is becoming clear: if at first you don’t succeed, you probably
should have simulated it.
By using a computer to model a system before it is built or to test operating
policies before they are actually implemented, many of the pitfalls that are often
encountered in the start-up of a new system or the modification of an existing sys-
tem can be avoided. Improvements that traditionally took months and even years
of fine-tuning to achieve can be attained in a matter of days or even hours. Be-
cause simulation runs in compressed time, weeks of system operation can be sim-
ulated in only a few minutes or even seconds. The characteristics of simulation
that make it such a powerful planning and decision-making tool can be summa-
rized as follows:
• Captures system interdependencies.
• Accounts for variability in the system.
• Is versatile enough to model any system.
• Shows behavior over time.
• Is less costly, time consuming, and disruptive than experimenting on the
actual system.
• Provides information on multiple performance measures.
• Is visually appealing and engages people’s interest.
• Provides results that are easy to understand and communicate.
• Runs in compressed, real, or even delayed time.
• Forces attention to detail in a design.
Because simulation accounts for interdependencies and variation, it provides
insights into the complex dynamics of a system that cannot be obtained using
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

8 Part I Study Chapters

other analysis techniques. Simulation gives systems planners unlimited freedom

to try out different ideas for improvement, risk free—with virtually no cost, no
waste of time, and no disruption to the current system. Furthermore, the results are
both visual and quantitative with performance statistics automatically reported on
all measures of interest.
Even if no problems are found when analyzing the output of simulation, the
exercise of developing a model is, in itself, beneficial in that it forces one to think
through the operational details of the process. Simulation can work with inaccu-
rate information, but it can’t work with incomplete information. Often solutions
present themselves as the model is built—before any simulation run is made. It is
a human tendency to ignore the operational details of a design or plan until the
implementation phase, when it is too late for decisions to have a significant
impact. As the philosopher Alfred North Whitehead observed, “We think in gen-
eralities; we live detail” (Audon 1964). System planners often gloss over the
details of how a system will operate and then get tripped up during implementa-
tion by all of the loose ends. The expression “the devil is in the details” has definite
application to systems planning. Simulation forces decisions on critical details so
they are not left to chance or to the last minute, when it may be too late.
Simulation promotes a try-it-and-see attitude that stimulates innovation and
encourages thinking “outside the box.” It helps one get into the system with sticks
and beat the bushes to flush out problems and find solutions. It also puts an end to
fruitless debates over what solution will work best and by how much. Simulation
takes the emotion out of the decision-making process by providing objective
evidence that is difficult to refute.

1.4 Doing Simulation

Simulation is nearly always performed as part of a larger process of system design
or process improvement. A design problem presents itself or a need for improve-
ment exists. Alternative solutions are generated and evaluated, and the best
solution is selected and implemented. Simulation comes into play during the eval-
uation phase. First, a model is developed for an alternative solution. As the model
is run, it is put into operation for the period of interest. Performance statistics (uti-
lization, processing time, and so on) are gathered and reported at the end of the
run. Usually several replications (independent runs) of the simulation are made.
Averages and variances across the replications are calculated to provide statistical
estimates of model performance. Through an iterative process of modeling, simu-
lation, and analysis, alternative configurations and operating policies can be tested
to determine which solution works the best.
Simulation is essentially an experimentation tool in which a computer model
of a new or existing system is created for the purpose of conducting experiments.
The model acts as a surrogate for the actual or real-world system. Knowledge
gained from experimenting on the model can be transferred to the real system (see
Figure 1.2). When we speak of doing simulation, we are talking about “the
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

Chapter 1 Introduction to Simulation 9

FIGURE 1.2 System

Simulation provides a
virtual method for
doing system
experimentation.

Concept

Model

process of designing a model of a real system and conducting experiments with

this model” (Shannon 1998). Conducting experiments on a model reduces the
time, cost, and disruption of experimenting on the actual system. In this respect,
simulation can be thought of as a virtual prototyping tool for demonstrating proof
of concept.
The procedure for doing simulation follows the scientific method of (1) for-
mulating a hypothesis, (2) setting up an experiment, (3) testing the hypothesis
through experimentation, and (4) drawing conclusions about the validity of the
hypothesis. In simulation, we formulate a hypothesis about what design or
operating policies work best. We then set up an experiment in the form of a
simulation model to test the hypothesis. With the model, we conduct multiple
replications of the experiment or simulation. Finally, we analyze the simulation
results and draw conclusions about our hypothesis. If our hypothesis was cor-
rect, we can confidently move ahead in making the design or operational
changes (assuming time and other implementation constraints are satisfied).
As shown in Figure 1.3, this process is repeated until we are satisfied with the
results.
By now it should be obvious that simulation itself is not a solution tool but
rather an evaluation tool. It describes how a defined system will behave; it does
not prescribe how it should be designed. Simulation doesn’t compensate for one’s
ignorance of how a system is supposed to operate. Neither does it excuse one from
being careful and responsible in the handling of input data and the interpretation
of output results. Rather than being perceived as a substitute for thinking, simula-
tion should be viewed as an extension of the mind that enables one to understand
the complex dynamics of a system.
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

10 Part I Study Chapters

FIGURE 1.3
The process of Start
simulation
experimentation.

Formulate a
hypothesis

Develop a
simulation model

No

Run simulation
experiment

Hypothesis
correct?

Yes

End

1.5 Use of Simulation

Simulation began to be used in commercial applications in the 1960s. Initial
models were usually programmed in FORTRAN and often consisted of thousands
of lines of code. Not only was model building an arduous task, but extensive de-
bugging was required before models ran correctly. Models frequently took a year
or more to build and debug so that, unfortunately, useful results were not obtained
until after a decision and monetary commitment had already been made. Lengthy
simulations were run in batch mode on expensive mainframe computers where
CPU time was at a premium. Long development cycles prohibited major changes
from being made once a model was built.
Only in the last couple of decades has simulation gained popularity as a
decision-making tool in manufacturing and service industries. For many compa-
nies, simulation has become a standard practice when a new facility is being
planned or a process change is being evaluated. It is fast becoming to systems
planners what spreadsheet software has become to financial planners.
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

Chapter 1 Introduction to Simulation 11

The surge in popularity of computer simulation can be attributed to the

following:

• Increased awareness and understanding of simulation technology.

• Increased availability, capability, and ease of use of simulation software.
• Increased computer memory and processing speeds, especially of PCs.
• Declining computer hardware and software costs.

Simulation is no longer considered a method of “last resort,” nor is it a tech-

nique reserved only for simulation “experts.” The availability of easy-to-use sim-
ulation software and the ubiquity of powerful desktop computers have made
simulation not only more accessible, but also more appealing to planners and
managers who tend to avoid any kind of solution that appears too complicated. A
solution tool is not of much use if it is more complicated than the problem that it
is intended to solve. With simple data entry tables and automatic output reporting
and graphing, simulation is becoming much easier to use and the reluctance to use
it is disappearing.
The primary use of simulation continues to be in the area of manufacturing.
Manufacturing systems, which include warehousing and distribution systems,
tend to have clearly defined relationships and formalized procedures that are well
suited to simulation modeling. They are also the systems that stand to benefit the
most from such an analysis tool since capital investments are so high and changes
are so disruptive. Recent trends to standardize and systematize other business
processes such as order processing, invoicing, and customer support are boosting
the application of simulation in these areas as well. It has been observed that
80 percent of all business processes are repetitive and can benefit from the same
analysis techniques used to improve manufacturing systems (Harrington 1991).
With this being the case, the use of simulation in designing and improving busi-
ness processes of every kind will likely continue to grow.
While the primary use of simulation is in decision support, it is by no means
limited to applications requiring a decision. An increasing use of simulation is in
the area of communication and visualization. Modern simulation software in-
corporates visual animation that stimulates interest in the model and effectively
communicates complex system dynamics. A proposal for a new system design can
be sold much easier if it can actually be shown how it will operate.
On a smaller scale, simulation is being used to provide interactive, computer-
based training in which a management trainee is given the opportunity to practice
decision-making skills by interacting with the model during the simulation. It is
also being used in real-time control applications where the model interacts with
the real system to monitor progress and provide master control. The power of
simulation to capture system dynamics both visually and functionally opens up
numerous opportunities for its use in an integrated environment.
Since the primary use of simulation is in decision support, most of our dis-
cussion will focus on the use of simulation to make system design and operational
decisions. As a decision support tool, simulation has been used to help plan and
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

12 Part I Study Chapters

make improvements in many areas of both manufacturing and service industries.

Typical applications of simulation include
• Work-flow planning. • Throughput analysis.
• Capacity planning. • Productivity improvement.
• Cycle time reduction. • Layout analysis.
• Staff and resource planning. • Line balancing.
• Work prioritization. • Batch size optimization.
• Bottleneck analysis. • Production scheduling.
• Quality improvement. • Resource scheduling.
• Cost reduction. • Maintenance scheduling.
• Inventory reduction. • Control system design.

1.6 When Simulation Is Appropriate

Not all system problems that could be solved with the aid of simulation should be
solved using simulation. It is important to select the right tool for the task. For
some problems, simulation may be overkill—like using a shotgun to kill a fly.
Simulation has certain limitations of which one should be aware before making a
decision to apply it to a given situation. It is not a panacea for all system-related
problems and should be used only if the shoe fits. As a general guideline,
simulation is appropriate if the following criteria hold true:
• An operational (logical or quantitative) decision is being made.
• The process being analyzed is well defined and repetitive.
• Activities and events are interdependent and variable.
• The cost impact of the decision is greater than the cost of doing the
simulation.
• The cost to experiment on the actual system is greater than the cost of
simulation.
Decisions should be of an operational nature. Perhaps the most significant
limitation of simulation is its restriction to the operational issues associated with
systems planning in which a logical or quantitative solution is being sought. It is
not very useful in solving qualitative problems such as those involving technical
or sociological issues. For example, it can’t tell you how to improve machine re-
liability or how to motivate workers to do a better job (although it can assess the
impact that a given level of reliability or personal performance can have on over-
all system performance). Qualitative issues such as these are better addressed
using other engineering and behavioral science techniques.
Processes should be well defined and repetitive. Simulation is useful only if
the process being modeled is well structured and repetitive. If the process doesn’t
follow a logical sequence and adhere to defined rules, it may be difficult to model.
Simulation applies only if you can describe how the process operates. This does
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

Chapter 1 Introduction to Simulation 13

not mean that there can be no uncertainty in the system. If random behavior can
be described using probability expressions and distributions, they can be simu-
lated. It is only when it isn’t even possible to make reasonable assumptions of
how a system operates (because either no information is available or behavior is
totally erratic) that simulation (or any other analysis tool for that matter) becomes
useless. Likewise, one-time projects or processes that are never repeated the same
way twice are poor candidates for simulation. If the scenario you are modeling is
likely never going to happen again, it is of little benefit to do a simulation.
Activities and events should be interdependent and variable. A system may
have lots of activities, but if they never interfere with each other or are determin-
istic (that is, they have no variation), then using simulation is probably un-
necessary. It isn’t the number of activities that makes a system difficult to analyze.
It is the number of interdependent, random activities. The effect of simple interde-
pendencies is easy to predict if there is no variability in the activities. Determining
the flow rate for a system consisting of 10 processing activities is very straightfor-
ward if all activity times are constant and activities are never interrupted. Likewise,
random activities that operate independently of each other are usually easy to ana-
lyze. For example, 10 machines operating in isolation from each other can be ex-
pected to produce at a rate that is based on the average cycle time of each machine
less any anticipated downtime. It is the combination of interdependencies and ran-
dom behavior that really produces the unpredictable results. Simpler analytical
methods such as mathematical calculations and spreadsheet software become less
adequate as the number of activities that are both interdependent and random in-
creases. For this reason, simulation is primarily suited to systems involving both
interdependencies and variability.
The cost impact of the decision should be greater than the cost of doing the
simulation. Sometimes the impact of the decision itself is so insignificant that it
doesn’t warrant the time and effort to conduct a simulation. Suppose, for example,
you are trying to decide whether a worker should repair rejects as they occur or
wait until four or five accumulate before making repairs. If you are certain that the
next downstream activity is relatively insensitive to whether repairs are done
sooner rather than later, the decision becomes inconsequential and simulation is a
wasted effort.
The cost to experiment on the actual system should be greater than the cost of
simulation. While simulation avoids the time delay and cost associated with ex-
perimenting on the real system, in some situations it may actually be quicker and
more economical to experiment on the real system. For example, the decision in a
customer mailing process of whether to seal envelopes before or after they are
addressed can easily be made by simply trying each method and comparing the
results. The rule of thumb here is that if a question can be answered through direct
experimentation quickly, inexpensively, and with minimal impact to the current
operation, then don’t use simulation. Experimenting on the actual system also
eliminates some of the drawbacks associated with simulation, such as proving
model validity.
There may be other situations where simulation is appropriate independent of
the criteria just listed (see Banks and Gibson 1997). This is certainly true in the
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

14 Part I Study Chapters

case of models built purely for visualization purposes. If you are trying to sell a
system design or simply communicate how a system works, a realistic animation
created using simulation can be very useful, even though nonbeneficial from an
analysis point of view.

1.7 Qualifications for Doing Simulation

Many individuals are reluctant to use simulation because they feel unqualified.
Certainly some training is required to use simulation, but it doesn’t mean that only
statisticians or operations research specialists can learn how to use it. Decision
support tools are always more effective when they involve the decision maker, es-
pecially when the decision maker is also the domain expert or person who is most
familiar with the design and operation of the system. The process owner or man-
ager, for example, is usually intimately familiar with the intricacies and idiosyn-
crasies of the system and is in the best position to know what elements to include
in the model and be able to recommend alternative design solutions. When per-
forming a simulation, often improvements suggest themselves in the very activity
of building the model that the decision maker might never discover if someone else
is doing the modeling. This reinforces the argument that the decision maker should
be heavily involved in, if not actually conducting, the simulation project.
To make simulation more accessible to non–simulation experts, products
have been developed that can be used at a basic level with very little training.
Unfortunately, there is always a potential danger that a tool will be used in a way
that exceeds one’s skill level. While simulation continues to become more user-
friendly, this does not absolve the user from acquiring the needed skills to make
intelligent use of it. Many aspects of simulation will continue to require some
training. Hoover and Perry (1989) note, “The subtleties and nuances of model val-
idation and output analysis have not yet been reduced to such a level of rote that
they can be completely embodied in simulation software.”
Modelers should be aware of their own inabilities in dealing with the statisti-
cal issues associated with simulation. Such awareness, however, should not pre-
vent one from using simulation within the realm of one’s expertise. There are both
a basic as well as an advanced level at which simulation can be beneficially used.
Rough-cut modeling to gain fundamental insights, for example, can be achieved
with only a rudimentary understanding of simulation. One need not have exten-
sive simulation training to go after the low-hanging fruit. Simulation follows the
80–20 rule, where 80 percent of the benefit can be obtained from knowing only
20 percent of the science involved (just make sure you know the right 20 percent).
It isn’t until more precise analysis is required that additional statistical training
and knowledge of experimental design are needed.
To reap the greatest benefits from simulation, a certain degree of knowledge
and skill in the following areas is useful:
• Project management.
• Communication.
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

Chapter 1 Introduction to Simulation 15

• Systems engineering.
• Statistical analysis and design of experiments.
• Modeling principles and concepts.
• Basic programming and computer skills.
• Training on one or more simulation products.
• Familiarity with the system being investigated.
Experience has shown that some people learn simulation more rapidly and
become more adept at it than others. People who are good abstract thinkers yet
also pay close attention to detail seem to be the best suited for doing simulation.
Such individuals are able to see the forest while still keeping an eye on the trees
(these are people who tend to be good at putting together 1,000-piece puzzles).
They are able to quickly scope a project, gather the pertinent data, and get a use-
ful model up and running without lots of starts and stops. A good modeler is some-
what of a sleuth, eager yet methodical and discriminating in piecing together all
of the evidence that will help put the model pieces together.
If short on time, talent, resources, or interest, the decision maker need not
despair. Plenty of consultants who are professionally trained and experienced can
provide simulation services. A competitive bid will help get the best price, but one
should be sure that the individual assigned to the project has good credentials.
If the use of simulation is only occasional, relying on a consultant may be the
preferred approach.

1.8 Economic Justification of Simulation

Cost is always an important issue when considering the use of any software tool,
and simulation is no exception. Simulation should not be used if the cost exceeds
the expected benefits. This means that both the costs and the benefits should be
carefully assessed. The use of simulation is often prematurely dismissed due to
the failure to recognize the potential benefits and savings it can produce. Much of
the reluctance in using simulation stems from the mistaken notion that simulation
is costly and very time consuming. This perception is shortsighted and ignores the
fact that in the long run simulation usually saves much more time and cost than it
consumes. It is true that the initial investment, including training and start-up
costs, may be between $10,000 and $30,000 (simulation products themselves
generally range between $1,000 and $20,000). However, this cost is often recov-
ered after the first one or two projects. The ongoing expense of using simulation
for individual projects is estimated to be between 1 and 3 percent of the total
project cost (Glenney and Mackulak 1985). With respect to the time commitment
involved in doing simulation, much of the effort that goes into building the model
is in arriving at a clear definition of how the system operates, which needs to be
done anyway. With the advanced modeling tools that are now available, the actual
model development and running of simulations take only a small fraction (often
less than 5 percent) of the overall system design time.
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

16 Part I Study Chapters

Savings from simulation are realized by identifying and eliminating problems

and inefficiencies that would have gone unnoticed until system implementation.
Cost is also reduced by eliminating overdesign and removing excessive safety
factors that are added when performance projections are uncertain. By identifying
and eliminating unnecessary capital investments, and discovering and correcting
operating inefficiencies, it is not uncommon for companies to report hundreds of
thousands of dollars in savings on a single project through the use of simulation.
The return on investment (ROI) for simulation often exceeds 1,000 percent, with
payback periods frequently being only a few months or the time it takes to com-
plete a simulation project.
One of the difficulties in developing an economic justification for simulation
is the fact that it is usually not known in advance how much savings will be real-
ized until it is actually used. Most applications in which simulation has been used
have resulted in savings that, had the savings been known in advance, would have
looked very good in an ROI or payback analysis.
One way to assess in advance the economic benefit of simulation is to assess
the risk of making poor design and operational decisions. One need only ask what
the potential cost would be if a misjudgment in systems planning were to occur.
Suppose, for example, that a decision is made to add another machine to solve a
capacity problem in a production or service system. What are the cost and proba-
bility associated with this being the wrong decision? If the cost associated with a
wrong decision is $100,000 and the decision maker is only 70 percent confident
that the decision is correct, then there is a 30 percent chance of incurring a cost of
$100,000. This results in a probable cost of $30,000 (.3 × $100,000). Using this
approach, many decision makers recognize that they can’t afford not to use simu-
lation because the risk associated with making the wrong decision is too high.
Tying the benefits of simulation to management and organizational goals also
provides justification for its use. For example, a company committed to continu-
ous improvement or, more specifically, to lead time or cost reduction can be
sold on simulation if it can be shown to be historically effective in these areas.
Simulation has gained the reputation as a best practice for helping companies
achieve organizational goals. Companies that profess to be serious about perfor-
mance improvement will invest in simulation if they believe it can help them
achieve their goals.
The real savings from simulation come from allowing designers to make
mistakes and work out design errors on the model rather than on the actual sys-
tem. The concept of reducing costs through working out problems in the design
phase rather than after a system has been implemented is best illustrated by the
rule of tens. This principle states that the cost to correct a problem increases by a
factor of 10 for every design stage through which it passes without being detected
(see Figure 1.4).
Simulation helps avoid many of the downstream costs associated with poor
decisions that are made up front. Figure 1.5 illustrates how the cumulative cost
resulting from systems designed using simulation can compare with the cost of
designing and operating systems without the use of simulation. Note that while
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

Chapter 1 Introduction to Simulation 17

FIGURE 1.4 Concept Design Installation Operation

Cost of making
changes at subsequent
stages of system
development.

Cost

System stage

FIGURE 1.5
Comparison of
Cost without
cumulative system
simulation
costs with and without
System costs

simulation.
Cost with
simulation

Design Implementation Operation

phase phase phase

the short-term cost may be slightly higher due to the added labor and software
costs associated with simulation, the long-term costs associated with capital
investments and system operation are considerably lower due to better efficiencies
realized through simulation. Dismissing the use of simulation on the basis of
sticker price is myopic and shows a lack of understanding of the long-term sav-
ings that come from having well-designed, efficiently operating systems.
Many examples can be cited to show how simulation has been used to avoid
costly errors in the start-up of a new system. Simulation prevented an unnecessary
expenditure when a Fortune 500 company was designing a facility for producing
and storing subassemblies and needed to determine the number of containers re-
quired for holding the subassemblies. It was initially felt that 3,000 containers
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

18 Part I Study Chapters

were needed until a simulation study showed that throughput did not improve sig-
nificantly when the number of containers was increased from 2,250 to 3,000. By
purchasing 2,250 containers instead of 3,000, a savings of $528,375 was expected
in the first year, with annual savings thereafter of over $200,000 due to the savings
in floor space and storage resulting from having 750 fewer containers (Law and
McComas 1988).
Even if dramatic savings are not realized each time a model is built, simula-
tion at least inspires confidence that a particular system design is capable of meet-
ing required performance objectives and thus minimizes the risk often associated
with new start-ups. The economic benefit associated with instilling confidence
was evidenced when an entrepreneur, who was attempting to secure bank financ-
ing to start a blanket factory, used a simulation model to show the feasibility of the
proposed factory. Based on the processing times and equipment lists supplied by
industry experts, the model showed that the output projections in the business
plan were well within the capability of the proposed facility. Although unfamiliar
with the blanket business, bank officials felt more secure in agreeing to support
the venture (Bateman et al. 1997).
Often simulation can help improve productivity by exposing ways of making
better use of existing assets. By looking at a system holistically, long-standing
problems such as bottlenecks, redundancies, and inefficiencies that previously
went unnoticed start to become more apparent and can be eliminated. “The trick is
to find waste, or muda,” advises Shingo; “after all, the most damaging kind of waste
is the waste we don’t recognize” (Shingo 1992). Consider the following actual
examples where simulation helped uncover and eliminate wasteful practices:

• GE Nuclear Energy was seeking ways to improve productivity without

investing large amounts of capital. Through the use of simulation, the
company was able to increase the output of highly specialized reactor
parts by 80 percent. The cycle time required for production of each part
was reduced by an average of 50 percent. These results were obtained by
running a series of models, each one solving production problems
highlighted by the previous model (Bateman et al. 1997).
• A large manufacturing company with stamping plants located throughout
the world produced stamped aluminum and brass parts on order according
to customer specifications. Each plant had from 20 to 50 stamping presses
that were utilized anywhere from 20 to 85 percent. A simulation study
was conducted to experiment with possible ways of increasing capacity
utilization. As a result of the study, machine utilization improved from an
average of 37 to 60 percent (Hancock, Dissen, and Merten 1977).
• A diagnostic radiology department in a community hospital was modeled
to evaluate patient and staff scheduling, and to assist in expansion
planning over the next five years. Analysis using the simulation model
enabled improvements to be discovered in operating procedures that
precluded the necessity for any major expansions in department size
(Perry and Baum 1976).
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

Chapter 1 Introduction to Simulation 19

In each of these examples, significant productivity improvements were real-

ized without the need for making major investments. The improvements came
through finding ways to operate more efficiently and utilize existing resources
more effectively. These capacity improvement opportunities were brought to light
through the use of simulation.

1.9 Sources of Information on Simulation

Simulation is a rapidly growing technology. While the basic science and theory
remain the same, new and better software is continually being developed to make
simulation more powerful and easier to use. It will require ongoing education for
those using simulation to stay abreast of these new developments. There are many
sources of information to which one can turn to learn the latest developments in
simulation technology. Some of the sources that are available include

• Conferences and workshops sponsored by vendors and professional

societies (such as, Winter Simulation Conference and the IIE
Conference).
• Professional magazines and journals (IIE Solutions, International Journal
of Modeling and Simulation, and the like).
• Websites of vendors and professional societies (www.promodel.com,
www.scs.org, and so on).
• Demonstrations and tutorials provided by vendors (like those on the
ProModel CD).
• Textbooks (like this one).

1.10 How to Use This Book

This book is divided into three parts. Part I contains chapters describing the sci-
ence and practice of simulation. The emphasis is deliberately oriented more to-
ward the practice than the science. Simulation is a powerful decision support tool
that has a broad range of applications. While a fundamental understanding of how
simulation works is presented, the aim has been to focus more on how to use sim-
ulation to solve real-world problems.
Part II contains ProModel lab exercises that help develop simulation skills.
ProModel is a simulation package designed specifically for ease of use, yet it pro-
vides the flexibility to model any discrete event or continuous flow process. It is
similar to other simulation products in that it provides a set of basic modeling
constructs and a language for defining the logical decisions that are made in a sys-
tem. Basic modeling objects in ProModel include entities (the objects being
processed), locations (the places where processing occurs), resources (the agents
used to process the entities), and paths (the course of travel for entities and
resources in moving between locations such as aisles or conveyors). Logical
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

20 Part I Study Chapters

behavior such as the way entities arrive and their routings can be defined with lit-
tle, if any, programming using the data entry tables that are provided. ProModel is
used by thousands of professionals in manufacturing and service-related indus-
tries and is taught in hundreds of institutions of higher learning.
Part III contains case study assignments that can be used for student projects to
apply the theory they have learned from Part I and to try out the skills they have ac-
quired from doing the lab exercises (Part II). It is recommended that students be as-
signed at least one simulation project during the course. Preferably this is a project
performed for a nearby company or institution so it will be meaningful. If such a
project cannot be found, or as an additional practice exercise, the case studies pro-
vided should be useful. Student projects should be selected early in the course so
that data gathering can get started and the project completed within the allotted
time. The chapters in Part I are sequenced to parallel an actual simulation project.

1.11 Summary
Businesses today face the challenge of quickly designing and implementing com-
plex production and service systems that are capable of meeting growing de-
mands for quality, delivery, affordability, and service. With recent advances in
computing and software technology, simulation tools are now available to help
meet this challenge. Simulation is a powerful technology that is being used with
increasing frequency to improve system performance by providing a way to make
better design and management decisions. When used properly, simulation can re-
duce the risks associated with starting up a new operation or making improve-
ments to existing operations.
Because simulation accounts for interdependencies and variability, it pro-
vides insights that cannot be obtained any other way. Where important system
decisions are being made of an operational nature, simulation is an invaluable
decision-making tool. Its usefulness increases as variability and interdependency
increase and the importance of the decision becomes greater.
Lastly, simulation actually makes designing systems fun! Not only can a de-
signer try out new design concepts to see what works best, but the visualization
makes it take on a realism that is like watching an actual system in operation.
Through simulation, decision makers can play what-if games with a new system
or modified process before it actually gets implemented. This engaging process
stimulates creative thinking and results in good design decisions.

1.12 Review Questions

1. Define simulation.
2. What reasons are there for the increased popularity of computer
simulation?
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

Chapter 1 Introduction to Simulation 21

3. What are two specific questions that simulation might help answer in a
bank? In a manufacturing facility? In a dental office?
4. What are three advantages that simulation has over alternative
approaches to systems design?
5. Does simulation itself optimize a system design? Explain.
6. How does simulation follow the scientific method?
7. A restaurant gets extremely busy during lunch (11:00 A.M. to 2:00 P.M.)
and is trying to decide whether it should increase the number of
waitresses from two to three. What considerations would you look at to
determine whether simulation should be used to make this decision?
8. How would you develop an economic justification for using simulation?
9. Is a simulation exercise wasted if it exposes no problems in a system
design? Explain.
10. A simulation run was made showing that a modeled factory could
produce 130 parts per hour. What information would you want to know
about the simulation study before placing any confidence in the results?
11. A PC board manufacturer has high work-in-process (WIP) inventories,
yet machines and equipment seem underutilized. How could simulation
help solve this problem?
12. How important is a statistical background for doing simulation?
13. How can a programming background be useful in doing simulation?
14. Why are good project management and communication skills important
in simulation?
15. Why should the process owner be heavily involved in a simulation
project?
16. For which of the following problems would simulation likely be useful?
a. Increasing the throughput of a production line.
b. Increasing the pace of a worker on an assembly line.
c. Decreasing the time that patrons at an amusement park spend
waiting in line.
d. Determining the percentage defective from a particular machine.
e. Determining where to place inspection points in a process.
f. Finding the most efficient way to fill out an order form.

References
Audon, Wyston Hugh, and L. Kronenberger. The Faber Book of Aphorisms. London: Faber
and Faber, 1964.
Banks, J., and R. Gibson. “10 Rules for Determining When Simulation Is Not Appropri-
ate.” IIE Solutions, September 1997, pp. 30–32.
Bateman, Robert E.; Royce O. Bowden; Thomas J. Gogg; Charles R. Harrell; and Jack
R. A. Mott. System Improvement Using Simulation. Utah: PROMODEL Corp., 1997.
Harrell−Ghosh−Bowden: I. Study Chapters 1. Introduction to © The McGraw−Hill
Simulation Using Simulation Companies, 2004
ProModel, Second Edition

22 Part I Study Chapters

Deming, W. E. Foundation for Management of Quality in the Western World. Paper read at
a meeting of the Institute of Management Sciences, Osaka, Japan, 24 July 1989.
Glenney, Neil E., and Gerald T. Mackulak. “Modeling & Simulation Provide Key to CIM
Implementation Philosophy.” Industrial Engineering, May 1985.
Hancock, Walton; R. Dissen; and A. Merten. “An Example of Simulation to Improve Plant
Productivity.” AIIE Transactions, March 1977, pp. 2–10.
Harrell, Charles R., and Donald Hicks. “Simulation Software Component Architecture
for Simulation-Based Enterprise Applications.” In Proceedings of the 1998 Winter
Simulation Conference, ed. D. J. Medeiros, E. F. Watson, J. S. Carson, and
M. S. Manivannan, pp. 1717–21. Institute of Electrical and Electronics Engineers,
Piscataway, New Jersey.
Harrington, H. James. Business Process Improvement. New York: McGraw-Hill, 1991.
Hoover, Stewart V., and Ronald F. Perry. Simulation: A Problem-Solving Approach.
Reading, MA: Addison-Wesley, 1989.
Law, A. M., and M. G. McComas. “How Simulation Pays Off.” Manufacturing Engineer-
ing, February 1988, pp. 37–39.
Mott, Jack, and Kerim Tumay. “Developing a Strategy for Justifying Simulation.” Indus-
trial Engineering, July 1992, pp. 38–42.
Oxford American Dictionary. New York: Oxford University Press, 1980. [compiled by]
Eugene Enrich et al.
Perry, R. F., and R. F. Baum. “Resource Allocation and Scheduling for a Radiology
Department.” In Cost Control in Hospitals. Ann Arbor, MI: Health Administration
Press, 1976.
Rohrer, Matt, and Jerry Banks. “Required Skills of a Simulation Analyst.” IIE Solutions,
May 1998, pp. 7–23.
Schriber, T. J. “The Nature and Role of Simulation in the Design of Manufacturing
Systems.” Simulation in CIM and Artificial Intelligence Techniques, ed. J. Retti and
K. E. Wichmann. S.D., CA.: Society for Computer Simulation, 1987, pp. 5–8.
Shannon, Robert E. “Introduction to the Art and Science of Simulation.” In Proceedings
of the 1998 Winter Simulation Conference, ed. D. J. Medeiros, E. F. Watson,
J. S. Carson, and M. S. Manivannan, pp. 7–14. Piscataway, NJ: Institute of Electrical
and Electronics Engineers.
Shingo, Shigeo. The Shingo Production Management System—Improving Process Func-
tions. Trans. Andrew P. Dillon. Cambridge, MA: Productivity Press, 1992.
Solberg, James. “Design and Analysis of Integrated Manufacturing Systems.” In W. Dale
Compton. Washington, D.C.: National Academy Press, 1988, p. 4.
The Wall Street Journal, March 19, 1999. “United 747’s Near Miss Sparks a Widespread
Review of Pilot Skills,” p. A1.