This chapter contains a set of sample simulations, by Gerard F. Cummings and other authors, based on a wide spectrum of applications.
In this collection of models, there are examples from traffic control, inventory control, distribution, assembly, quality control, Flexible Manufacturing Systems, PBX design, network modeling and many others. Each example is listed here, followed by a sample report, and a set of suggested interactions with the simulation. These examples are intended primarily to present the "artistic" side of a simulation project: how simulation techniques can be used to capture the essential behavior of a real world system. It is usually the case that any given process can be modeled in many different ways, and studying the methods used by experienced professionals is an efficient way to build ones own simulation skills.
It is desirable to develop a"feel" for which factors will have significant effects. You will find that many suggested changes to a simulated system, are, in fact, ineffective with respect to many overall system metrics, such as speed or cost. Since realizable designs to be suggested by the simulation study should not be based on ineffective changes, it is important to develop a "feel" for the effectiveness of changes before the final designs are simulated. You should not waste time studying changes which dont matter.
In simulation studies it is important to keep track of what real world behavior and/or data can be taken as given and what is assumed to be true. Unlike most other disciplines, the experimental nature of simulation allows you to use"sensitivity analysis" to examine the effects of error in your assumptions. It may be permissible to include very inaccurate data, if its accuracy has no effect on the results. On the other hand, if the results are highly sensitive to missing data, it may be ill-advised to continue the simulation study unless the proper data can be obtained. Such relationships can be quantified by preliminary simulations.
It is the burden of the simulation analyst to prove that the reported results are not due to random variation. Many simulation studies in the past have reported so-called "effects" which were actually indistinguishable from random noise. GPSS World makes it easy to perform a first level statistical analysis. The GPSS World ANOVA command is discussed in Lesson 13 of the last chapter, and in Chapter 6 of the GPSS World Reference Manual.
The objective of this chapter is to present the application of simulation techniques to real world problems. For this reason, the examples in this chapter have been simplified in two ways.
lFirst, the run times have been made much shorter than you would use in a real simulation study. Longer runs would lead to less variable results.
lSecond, in this chapter, we have not considered the use of statistical methods, which you would apply in any actual application.
The sample programs are in the GPSS World Samples Folder. The examples are named by the corresponding GPSS World Program File names. They are:
1. TURNSTIL.GPS Simulation of a turnstile at a football stadium.
2. TELEPHON.GPS Simulation of a simple telephone system.
3. PERIODIC.GPS Simulation of inventory with periodic review.
4. TVREPAIR.GPS Simulation of a television repair shop.
5. QCONTROL.GPS Simulation of a quality control system.
6. ORDERPNT.GPS Simulation of an order point inventory system.
7. MANUFACT.GPS Simulation of an electronics manufacturing system.
8. TEXTILE.GPS Simulation of a textile factory.
9. OILDEPOT.GPS Simulation of an oil storage depot.
10. ASSEMBLY.GPS Simulation of a pump assembly process.
11. ROBOTFMS.GPS Simulation of a robot operated FMS.
12. BICYCLE.GPS Simulation of a bicycle factory.
13. STOCKCTL.GPS Simulation of a warehouse and branch inventories.
14. LOCKSIMN.GPS Simulation of a canal and lock.
15. FOUNDRY.GPS Foundry simulation.
16. TAPEPREP.GPS Simulation of NC tape preparation.
17. TRAFFIC.GPS Simulation of traffic at a T-junction.
18. POWDER.GPS Simulation of customer brand loyalty.
19. QTHEORY.GPS Simulation of a solvable queuing network.
20. SUPERMRK.GPS Simulation of a supermarket.
21. SHIPPORT.GPS Simulation of a port.
22. EXCHANGE.GPS Simulation of a PBX.
23. FMSMODEL.GPS Simulation of a flexible manufacturing system.
24. ETHERNET.GPS Simulation of a 10 Mbps Ethernet Network.
25. PREDATOR.GPS Simulation of Predator and Prey Populations.
We recommend that you study each example, then read it into your GPSS World Simulation Environment, to interact with it. You may then interactively explore a wide variety of"What if?" scenarios, starting with the suggested interactions. Feel free to use and explore your own methods.
At this point, you might also want to refer to the chart included in the "How This Manual Describes Actions" section at the beginning of Chapter 1 of this manual. This chart will tell you how to interpret the color changes in GPSS World icons.
Some GPSS statements are introduced without explanation. You will need to refer to Chapters 6, 7, and 8 in the GPSS World Reference Manual for details.
The first five examples include detailed discussions of the programming of each GPSS model. You may want to study these first, and then proceed to the application(s) that match your own most closely.
You can use your mouse to select the Samples folder in the GPSS World folder by double clicking on it. Then a double click on the sample you want to run will automatically start a GPSS World Session and open the selected model or you can
CLICK ON Start / Programs / GPSS World...
and when the Main Window opens follow the steps in any of the samples that follow.
One additional note: While running GPSS World, files that are written are placed in the directory where their parent file has been saved. In other words, a newly created simulation will be saved in the location where the model file that produced it exists. Reports are written to the location where the saved simulation exists. If the simulation has not been saved then the reports are written to the location of the simulation's parent, the model file.
Now, let's go on and try running some sample models.
[Table of Contents]