Simulation

General

In simulations, processes or systems are imitated by models. Different parameters can be set and changed as incoming variables, and it is illustrated how the system would behave under the given parameters.

Goal

The aim of this method is to illustrate complex processes of reality in simplified models in order to generate insights into the real processes.

Execution

Law and Kelton [1] have proposed a model for conducting a simulation study:

1. Preliminary Study: definition of the problem
2. Conceptual modelling and data collection: collection of information on the system structure and processes and collection of representative input data
3. Validation: checking the validity of the conceptual model by e.g., structured model checks If the model is not valid, the system jumps back to step 2.
4. Implementation: implementation of the model in the selected development environment. Subsequently, the implementation is checked for technical correctness.
5. Test Runs: The simulation system is tested with real data, e.g., from existing, comparable systems or plausibly generated.
6. Validation of the Implemented Model: application and simulation experts also check the correctness. There is no general validation approach, a suitable approach must be selected (approaches to be found in [4]). In any case, interviews with experts, comprehensive documentation and a sensitivity analysis are useful. If the validation is negative, we return to step 2.
7. Experiment Design: Determine the length of the simulation run, the startup phase and the number of independent replications for each system configuration. In practice, statistical design of experiments is the most commonly used method.
8. Perform Production Runs: The production runs should be carefully recorded in order to be able to perform the desired analyses afterwards.
9. Output analysis: determine absolute performance indicators of systems, compare different configurations, optimize the vector of input factors as targets.
10. Documentation, Presentation of Results: preparation of documentation as a parallel task throughout the study All assumptions, the developed simulation system as well as the results must be well documented.

Core Literature

• [1] Law, A. M., Kelton, W. (2007). Simulation Modeling and Analysis. Boston : McGraw Hill, 4th edition.
• [2] Liebl, F. (1995). München : Oldenbourg.
• [3] Mehl, H. (1994). Methoden verteilter Simulation. Braunschweig : Vieweg.

Further Literature

• [4] Rabe, M., Spieckermann, S., Wenzel, S. (2008).Verifikation und Validierung für die Simulation in Produktion und Logistik: Vorgehensmodelle und Techniken (VDI-Buch) , Springer Verlag.