View the PDF.
Acrobat Reader is needed to view the PDF.

The vast amount of software available for discrete-event simulation can be overwhelming for the newcomer to the field. The selection process is especially complicated when you consider the 1996 IIE Solutions Simulation Buyer's Guide which featured 57 entries. Of these, 25 indicated that they were useful in process control. Thirty-two indicated shop scheduling as an area of application, and 38 of the entries provide animation. With this many options available, a careful software selection decision can take as long as six months.

Some of the areas that you may need to consider in software selection include: input, processing, output, environment, vendor, and cost.

Prior to discussing these features, however, we must give a few warnings:

Warning 1: Know which features are appropriate for your situation.

You may not need a particular feature. For example, a package may include a feature for modeling cranes. IF your company does not use cranes, this feature would not be necessary.

Warning 2: Don't judge on the basis of "Yes" or "No."

Assume that the software being considered has the ability to model conveyors. This doesn't mean that it can accurately model your conveyors. A simplification of your conveyors may be the limits of the software being considered. If you need more detail, simplification can cause problems downstream. For example, for your use, both the length and width of the conveyor may be relevant; however, if the software under consideration is concerned only with the length of the conveyor, it won't work for you.

Warning 3: You may not need certain features.

Assume that one of the features of the software under consideration is that an interface with C code is possible. Because C code can be difficult to work with, you might prefer a simulation software package that avoids C code. Powerful constructs within this type of software can be very useful.

• Point and click Capability
  Simulation software users expect and desire this familiar feature for their software applications. And, most simulation software has migrated to this environment.

• CAD Translation
  If a CAD drawing of the static background exists, it should be converted into the drawing system used by the simulation. A CAD translation feature can save a great deal of time when laying out a large or complex conveyor model.

• Importing a file
  Often, you will need to be able to import a data file for use in the simulation. For example, if you have a large database that is generated and updated electronically, having to re-key such information would be a waste of time, and would likely cause errors.

• Exporting a file
  The output file may be a spreadsheet for drawing business graphics beyond or different than those generated by the simulation software.

• Syntax
  The syntax should be easily understood, consistent and unambiguous. There are many counter examples to this. For instance, in several of the popular simulation languages, a QUEUE block is not necessary to have a waiting time. The QUEUE block just initiates data collection. This is just one of many ambiguous notions that long term users have come to accept.

• Interactive run controller (IRC)
  Some call this feature debugger. Even the best of simulation analysts makes mistakes or commits logical errors when building a model. The IRC may assist in finding and correcting those errors in the following ways:
  1. The simulation can be monitored as it progresses. This can be accomplished by advancing the simulation until a desired time has elapsed, then displaying model information at that time. Another possibility is to advance the simulation until a particular condition is in effect, and then display information.
  2. Attention can be focused on a particular area of the simulation, or a particular entity. For example, every time an entity enters a specified area, the simulation will gather information, or every time a specified entity becomes active, the simulation will pause.
  3. Values of selected model components can be observed. When the simulation has paused, the current value or status of variables, attributes, queues, resources, and counters can be observed.
  4. The simulation can be temporarily suspended to view information or to reassign values or redirect entities

• Interface to another language
  This is the ability to change into another language for additional detail. We questioned this particular capability above. However, in some instances, its use can be of great important

• Input data analysis capability
  This feature provides the ability to determine whether input data can be described by a statistical or mathematical distribution. Many software vendors have included the ability of third party software to provide this capability.

• Powerful constructs
  For some situations, powerful capabilities may be needed. For example, modeling bridge cranes with interference requires powerful constructs.

• Speed When there are many entities in the system, the software speed should not degrade to the point of slow motion.

• Runtime flexibility
  An example of this feature is batching, that is, the ability to stack a series of runs, and the necessary data collection in an orderly fashion, without intervention. Another example is scenario generation. In this case, with some prodding, the simulation software should be able to automatically generate alternate possibilities for simulation. An example of scenario generator allows input data to vary over a range (for example, the cycle time of a machine can range from 4 to 8 seconds).

• Random variate generator
  There are about 12 statistical distributions that are commonly used in simulation. Most simulation software has the ability to generate random variates using these 12 distributions

• Reset
  For stead state analysis, it is important to have the ability to reset the statistics that have been collected. This reset is accomplished without clearing the entities that are currently in the system.

• Independent replications
  Multiple replications using different sets of random numbers should be possible. Otherwise, the same results would occur repeatedly.

• Attributes and global variables
  Attributes are local values available to the entity possessing that attribute and global variables are available to all entities. The question is how many of each of theses is available. Real simulation models can use large numbers of attributes and global variables

• Programming
  This feature is also custom logic representation. The ability to accurately mimic custom logic to any desired degree of detail usually requires some type of internal programming capability or underlying language. It is an absolute necessity for modeling complex problems or systems in order to build high fidelity models (that, like a good audio system, can reproduce the sound of the original system). Consider whether the software allows you to represent a problem to any desired level of detail or complexity, or if it forces you to choose canned representations or simplifications.

• Portability
  This feature enables the software to be run on various classes of computer without any changes in the software

• Standardized reports
  Examples of standardized output measures are the average number in queue, average time in queue, and throughput. The software can produce these and other values automatically, or upon request.

• Customized reports
  These are tailored presentations such as those that would be shown to a manager. The format can be set by the simulation analyst.

• Business graphics
  The software can have the ability to generate high quality bar charts, pie charts, and histograms, that can be used in presentations and reports.

• Database maintenance
  Once possibility is to collect stacks of paper output representing the replications from each scenario. However, this could become a rather large stack of paper. At the other extreme is a database that contains all of these outputs in an organized fashion.

• Collect mathematical expressions desired
  This feature allows the specification of measures of interest to the modeler. For example, suppose that each Type 1 output completed is to be multiplied by a royalty of $1.53 and each Type 2 output is to be multiplied by a royalty of $1.86. These royalties are to be reported by shift and week, as well as their sum.

• Custom performance measures
  Does the software allow the analyst to define and create new or custom measures of performance for a model? Or, is the analyst constrained to a predefined set of performance measures?

• Write to a file
  Does the software allow data, events, or system variable to be written to a file whenever desired? This feature allows the analyst to later import the file into a spreadsheet or database program for further customized analysis or manipulation.

• Ease of use
  Power of the software is probably much more important than ease of use.

• Ease of learning
  This feature is important to the casual user, not so important to the frequent or continuous user.

• Quality of documentation
  Documentation needs to be on-line and easy to understand.

• Animation capability
  Not all animations are created equally. Consider the ease of development, the equality of the picture, the smoothness of movement, and the portability for remote viewing.

• Run only version
  Does the vendor offer an "execute only" version, or some other convenient way of creating a stand-alone demonstration/animation of the model or specific scenarios? This capability can be quite important in showing results, without the need to have the licensed software system present at every viewing

• Stability
  How long has the vendor been in business? Is simulation software their primary business, or just a sideline?

• History
  Does the vendor have a history of providing annual updates to the software? Are they committed to continuous improvement of the product?

• Track record
  Are the vendor's new releases on time and error free? Do they maintain upgrade paths from older versions (for both models and animation) so that previous models can be kept current and enhanced? Have they ever released a new version that wasn't compatible with previous model versions? S

• upport
  Is the vendor committed to supporting the product and offering adequate telephone technical support and access to a BBS system to download patches or new releases? What do other users have to say about the quality and responsiveness of the vendor's technical support staff? What computer platforms and operating systems are supported?

Conclusion
Which simulation modeling tool is the right one? Based on the discussion above, and some additional pointers, the following considerations are made:

• Accuracy and detail are critical;
• Powerful capabilities can increase simulation analyst productivity;
• Get the fastest speed that you can afford, because the greater expense is the analysts' time;
• Fancy ads and demos do not always accurately demonstrate the software's capabilities;
• Beware of checklists–go beyond the checkmark; and
• Implementation and capability are important.

• Have simulation vendors solve a small version of your problem;
• Seek references that can attest to the software's capabilities and limitations;
• Seek the opinion of consultants that use several products;
• Seek the opinions of companies with similar applications;
• Attend user group meetings and annual symposia;
• visit the vendors at the Winter Simulation Conference or other related conferences; and
• Take a class on the software.

Further Reading

Akbay, K.S. "Using Simulation Optimization to find the Best solution," IIE Solutions, May 1996.

Banks, J., "Selecting Simulation Software," Proceedings of the 1991 Winter Simulation Conference", Phoenix, Arizona, December 8-11, 1991.

Banks, J., "Selecting Simulation Software," Conference Proceedings, Vol. 1,", AUTOFACT '95, Chicago, Illinois, November 13-16, 1995.

Banks, J., and R. Gibson, "Getting Started in Simulation Modeling," IIE Solutions, November 1996.

Banks, J., and R. Gibson, "Simulation Modeling: Some Programming Required," IIE Solutions, February 1997.

Banks, J. and V. Norman, "Justifying Simulation in Today's Manufacturing Environment," IIE Solutions, November 1995.

Brewer, S.K., "Simulation: Why Aren't We Where We Should Be?" IIE Solutions, January 1995.

Mabrouk, K.M., "Create Your Own Low-Risk Manufacturing Environment," IIE Solutions, January 1996.

Mazziotti, B.W., "Get More Mileage from Flexible Simulations," IIE Solutions, May 1996.

Porcaro, D., "Simulation Modeling and DOE," IIE Solutions, September 1996.

Routh, G., "Industrial Engineers Play Critical Role in Boeing Celebration," IIE Solutions, July 1995.

Schelasin, R.E.A., and J.L. Mauer, "Creating Flexible Simulation Models," IIE Solutions, May 1995.

Thompson, M., "Simulation-Based Scheduling: Meeting the Semiconductor Wafer Fabrication Challenge," IIE Solutions May 1996.

Randall Gibson is president of Automation Associates Inc., an independent engineering firm based in Solana Beach, California, that specializes in systems design, analysis, and simulation modeling. He is a senior member of IIE.