Modeling Concepts

  • George S. Fishman
Chapter
Part of the Springer Series in Operations Research book series (ORFE)

Abstract

Every simulation study consists of at least 10 steps:
  1. 1.

    Translating the qualitative description of a system into a formal abstraction that explicitly accounts for all logical and mathematical relationships

     
  2. 2.

    Identifying all parameters of the abstraction that require numerical values as input

     
  3. 3.

    Identifying all measures of performance whose values require estimation

     
  4. 4.

    Estimating the values of all unknown input parameters from available data, expert opinion, etc.

     
  5. 5.

    Designing the set of sampling experiments, at least one for each distinct set of values for the vector of input parameters

     
  6. 6.

    Converting the logical abstraction to executable code in a simulation programming language

     
  7. 7.

    Incorporating into the code statements for computing time averages of performance measures during program execution

     
  8. 8.

    Performing the set of sampling experiments by executing the code repeatedly, at least once for each vector of input-parameter values

     
  9. 9.

    For each measure of performance, evaluating how well its time average approximates its unknown long-run average

     
  10. 10.

    Comparing corresponding sample time averages for each performance measure across experiments

     
While this numbering allows for orderly progress from beginning to end, actual practice may turn out to blur the distinction between steps, rearrange their order, or in some cases, neglect or pay too little attention to them. For example, the availability of point-click-drag-drop simulation software entices a simulationist to merge steps 1 and 6, possibly making her or him oblivious to a wider range of modeling concepts than the selected language implementation may accommodate. Frequently, the choice of long-run time averages to estimate in step 3 depends solely on what the programming language automatically makes available in converting the model to code in step 6. Too often, step 9 is ignored, making any conclusions based on comparisons across experiments difficult to justify statistically.

Keywords

Elementary Event Outflow Rate Input Buffer Compound Event Completion Event 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. Cheng, R.C., and J.D. Lamb (1999). Making efficient simulation experiments interactively with a desktop simulation package, University of Kent at Canterbury, United Kingdom.Google Scholar
  2. Daley, D. and L. Servi (1999). Personal communications.Google Scholar
  3. Fishman, G.S. (1973). Concepts and Methods in Discrete Event Digital Simulation, Wiley, New York.Google Scholar
  4. Gordon, G. (1969). System Simulation, Prentice-Hall, Englewood Cliffs, N.J.Google Scholar
  5. Pritsker, A.A.B., J.J. O’Reilly, and D.K. LaVal (1997). Simulation with Visual SLAM and AweSim, Wiley, New York.Google Scholar

Copyright information

© Springer Science+Business Media New York 2001

Authors and Affiliations

  • George S. Fishman
    • 1
  1. 1.Department of Operations ResearchUniversity of North Carolina at Chapel HillChapel HillUSA

Personalised recommendations