OR/MS Today, October 1997

Simulation Goes Mainstream

By James J. Swain

Popular methodology is becoming increasingly integrated within the decision-making of many organization's significant operations

Early this summer we were shown how the Martian lander commands were tested on a simulated lander on a simulated Martian surface, prior to being sent to the lander itself. Before the Mir cosmonauts began their repair job in space, they simulated their activities to familiarize themselves with the restricted space in which they would have to work and to practice the necessary repair activities. It is increasingly the case that if an application or a mission is critical, simulation will be used to try it out, troubleshoot it or just check that it will proceed as planned. Simulation has gone mainstream and applications abound &endash; it may well be among the most pervasive analysis tools currently in use.

Simulation is not relegated to special projects, either. Karl G. Kempf, the chief scientific officer of the Intel Corporation, reports that Intel attempts to use simulation through all stages of its electronic fabrication process, from the earliest design stages, into early production and ramp up, throughout the high volume phase, and finally through the commodity phase and product phase-out. The philosophy is that the models used to study the process are maintained as the production facility itself is upgraded, and that system study and facility "tuning" are ongoing processes. As he notes, with the typical facility costing over $2 billion, and with annual income streams more than double that amount, the potential for significant improvements through simulation are always there.

Nor is simulation limited to the confines of a single corporation. Motorola, Lockheed and Raytheon have developed a joint simulation project for the production of their Iridium satellites. They call this virtual manufacturing, and it has allowed them to reduce expenditures and to coordinate an extremely complex manufacturing problem involving 66 satellites whose precision components are manufactured in one place, assembled in another, and eventually will be "delivered" to their operational site by rocket.

In addition to helping decision makers quantify their designs and study the implication of their operations, simulation can be used to study more conceptual what-if questions. For instance, Sterman et al. [1997] uses simulation to study the effect of implementing TQM in a firm (Analog Devices, Inc.) and the paradoxical results they achieved: quality and productivity improved but the firm's performance suffered. Simulation was used to model the internal workings of the firm and the external dynamics of their market in an effort to determine the operation of Analog under its TQM program and to contrast it with the situation in which TQM had not been implemented. Their model suggested that although TQM played a role in their decline, the problem would ultimately have been more severe without TQM. And this is not an isolated case: simulation can also be used for the basis of forming broad public policy. The United Network for Organ Sharing (UNOS) and Pritsker Corporation studied the effect of various allocation strategies for liver transplantation. The Board of UNOS ultimately recommended changes in the allocation method based upon this work.

Increasingly, simulation is integrated within the decision-making of any significant operation, and simulation software is growing to reflect that integration. For instance, simulation software is able to accept a variety of inputs, from spreadsheets, databases and even CAD drawings, while also able to export results to these applications. Animation is now integral to almost all products, and there are general purpose animators (such as PROOF) which can perform animation for any simulation which can write to a text file. Increasingly, tools for both input and output analysis are provided within the software or are available as optional modules or products. Simulation optimization modules are becoming increasingly popular.

The proliferation of personal computers and simulation software has greatly expanded the accessibility of simulation, so that potentially any desktop machine could be used to write and to perform simulation studies. The interconnection of computers through the World Wide Web and the development of Web-based simulation capabilities may increase our simulation abilities even further. Demonstrations and completed applications may now be placed on the Web and accessed anywhere in the world.


Survey
This survey is the fourth biennial survey of simulation software for discrete event systems simulation and related products [Swain, 1995]. As in previous years, the information was provided by the vendors based upon a survey of important product features, price and platform. We have concentrated on products that run on desktop computers with an emphasis on discrete event simulation because they are the most suitable for usage in our field. Simulation products whose primary capability is continuous simulation (systems of differential equations) or training (e.g., aircraft and vehicle simulators) are omitted here.

There are 46 products listed in this survey, and the range and variety of these products continues to grow, reflecting the growth and increasing sophistication of the user base. The information presented has been chosen to provide a general gauge of the product's functionality, special features and price. A separate listing gives contact information for all of the vendors in the survey.

The interaction between vendors and their customers has developed as products have matured, and most of the vendors maintain contact with their users through product mailings, regular newsletters and annual conferences. These user group conferences showcase the application and usefulness of their products, nurture close contact with their users, and allow the users to learn from each other.

In addition to users groups, there are a number of organizations and conferences devoted to application and methodology of simulation. Articles on simulation methodology and applications appear in the INFORMS publications Management Science, Operations Research and Interfaces.

The INFORMS College on Simulation is a small but active group dedicated to simulation. The College sponsors simulation sessions at the national INFORMS meetings and sponsors regular awards for both the best simulation publication and lifetime achievement for service to the area of simulation. For further information about the College on Simulation, its activities and pointers to sources of information about simulation, visit their Website at www.isye.gatech.edu/informs-sim.

The College is also a co-sponsor of the annual Winter Simulation Conference to be held Dec. 7-10 at the Renaissance Waverly Hotel in Atlanta. This year's conference has the theme, "The shortest distance from research to application," and the program includes special sessions on simulation success stories. The program includes introductory and advanced tutorials, computer, manufacturing and military applications, and presentations and exhibits by many of the vendors in this survey, as well as methodology and analysis sessions. The full program and registration information is available from the permanent Winter Simulation Website, www.wintersim.org. Other sponsors of the conference can be linked from this Website.


Trends in Simulation Software
From the very beginning the trend in simulation products has been from generality to increasing specificity and ease of use. Programming in procedural languages gave way to subroutine libraries and eventually to general purpose simulation languages. In time these general purpose languages acquired improved interfaces and menuing systems, so that the actual programming was less apparent. As the user base of the software increased, it became feasible to tailor the interface to specific application, such as manufacturing, and services such as healthcare and telecommunications. In some cases, this meant much more than simply altering the model-building interface; as simulation products become more focused, it is also possible to imbed more detailed and specific information into the model. For instance, COMNET includes predefined constructs for particular transmission cables and protocols, while MEDMODEL provides logic and animation icons for typical elements of a healthcare service facility.

One group of products can be characterized in the integration of the model-building functions in their interface. Integrated products such as AweSim, based upon the SLAM simulation system, WITNESS, PROMODEL, TAYLOR II and AUTOMOD, for example, emphasize ease of use and comprehensiveness.

Other products have evolved using a hierarchical structure. Both ARENA and SLX, for instance, have multiple layers of generality. The lowest level elements can be built into modules that either represent commonly encountered composite elements (e.g., in a manufacturing setting, the combination of input conveyor, buffer, server and perhaps inspection station) or specialized modules for a specific application area (e.g., a work unit plus animation). Such a structure appears to promise many of the advantages of an applications-oriented simulator, while the lower levels of the hierarchy permit the generality associated with a general purpose simulation language. Some of the programs have it both ways, through a hierarchical design which at the top allows the user to interact with the program as if it is dedicated to their particular application, while at the same time providing a mechanism for accessing the language at its most basic, providing the full range of flexibility available to that product.

In parallel to other computer software, simulation software is increasingly able to share information with other applications. Many of the products here can accept information from a variety of spreadsheets or database formats, and can output data to them as well. Most of the products also provide access to procedural languages which can be linked to the simulation code (whether C code or Visual Basic), which can access specialized information sources or perform specialized computations, such as scheduling and decision-making within the simulated system.

Simulation packages are increasingly integrated through the vendor-provided add-on software for such tasks as capacity planning, scheduling, input modeling or the run control (specification of runs, collection of run data, and analysis of the results). For those products that don't include input modeling, ExpertFit fits observed data to a wide variety of distributions and can then specify the syntax for realization of the fitted distributions for almost all of the simulation packages.

Perhaps the most interesting development for simulation is the interaction between the World Wide Web and simulation. The Web has, of course, been a source of information for simulationists for years. In recent years the development of JAVA-based simulations promises that simulations can be shared anywhere in the world. Fishwick provides an overview of this field (and pointers to examples) via the Web page www.cise.ufl.edu/~fishwick/websim.html. The Web may also make it possible for simulations to be run interactively between dispersed sites, both to enlarge the domain of the simulations (cooperating companies can run their simulations together in order to simulate joint operations) or to run interactively in a "gaming" mode to see how a given configuration or strategy will compete in the uncertain marketplace.


Applications
The growth of simulation applications has not slowed in the last two years. Improvements in simulation interfaces, and the development of application-oriented simulators, will increase the accessibility of simulation. Animated simulation output provides credibility and an immediacy for model developers which is harder to achieve with sample statistics.

Transportation is a growing area of simulation use. Examples of simulation use can be cited for all modes of shipment, and most major carriers &endash; whether air, rail, ship or truck &endash; now employ simulation to study and streamline their operations. Singapore has used simulation to study its automated ship-unloading operation in conjunction with the scheduling algorithms since this had the greatest impact on its capacity and its ability to process ships within the harbor. Carriers such as UPS must process large numbers of packages in a short time, in an operation involving a large number of arriving and departing aircraft, loading and unloading crews, and assorted materials handling equipment. They have used simulation to understand and to improve their operations.

The military employs simulation to model many aspects of its operations within an operational area and throughout the world. Logistics, communications and the operation of individual systems and munitions (alone and in joint operations) are among the key areas of military simulation. The military has been a leading proponent of animation and interconnected simulations, and is already interested in combining these into "virtual" simulations.

Simulation use is also heavy throughout the commercial world. As service systems become automated, simulation is used to tune equipment configurations and predict performance. Sensitivity studies are also made of these systems to examine system robustness and likely failure modes. Service industries use simulation to test scheduling algorithms and to ensure that performance can be maintained under a wide variety of operating conditions.


Teaching and Training
Simulation has long been used for training, especially where simulation-based training provides a more cost effective method of providing experience. Aircraft simulators run at a fraction of the cost of actual flying and without the risk of catastrophe. Moreover, the experience can be tailored to particular training scenarios or flight missions.

Simulation use in the classroom can be expected to increase as well. Educational use of simulation can be beneficial for several reasons. First, there is a growing appreciation that many student learning styles involve generalization from specific experiences, experience which can be provided by simulation-based experiments and demonstrations. Second, simulation can be a useful means of integrating material from several sources, as when a manufacturing simulation is used to illustrate the impact of product design on ease of manufacturing, product reliability and financial success of a simulated firm. Thus, simulation can be used to improve the effectiveness of many different courses throughout the curriculum.

1. Sterman, J. D., N. P. Repenning, and F. Kofman, "Unanticipated Side Effects of Successful Quality Programs: Exploring a Paradox of Organizational Improvement," Management Science, Vol. 43, No. 4, pp. 503-521, 1997.

2. Swain, J. J. (1995), "Tools for Process Understanding and Improvement: Simulation Software Survey," OR/MS Today, Vol. 22, No. 4, pp. 64-79.

E-mail to the Editorial Department of OR/MS Today: orms@lionhrtpub.com


OR/MS Today copyright © 1997, 1998 by the Institute for Operations Research and the Management Sciences. All rights reserved.


Lionheart Publishing, Inc.
2555 Cumberland Parkway, Suite 299, Atlanta, GA 30339 USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com


Web Site © Copyright 1997, 1998 by Lionheart Publishing, Inc. All rights reserved.