Improving Simulation in Supply Chain Research

“Improving the rigor of discrete-event simulation in logistics and supply chain research” is a 2009 paper by Ila Manuj, JT. Mentzer and MR. Bowers.

The paper consists of three distinct sections:

1. Introduction of the SMDP (Simulation Model Development Process)

1. Evaluation of the current literature in comparison to the SMDP

1. Explanation of the SMDP using an example

Simulation Model Development Process

The SMDP is discribed as a normative model to create a rigor and trustworthy simulation model. The promise seems simple, eight steps ought to be followed to create a model which can be described sufficiently.

1. Problem formulation
2. Choice of dependent and independent variables
3. Validation of conceptual model
4. Data collection
5. Verification of computer model
6. Model validation
7. Performing simulations
8. Analysis

Following those steps can lead to a systematical description of the simulation and the underlying model. The result is a more complete model by using validating techniques like expert input and comparison with real world data.

Literature

The most important part for writing papers and theses is to not only generate this information but actually write it down.

The second part of the paper deals with this question. Are current simulation studies implicitly or explicitly following the above mentioned steps?

The answer is no. Especially steps 3 and 5 through 7 are not documented very well. Of course this does not mean that for example the simulation models have not been validated in some form, but very often there seems to be only very little documentation of the methods used to enable other scientists to reexamine the models.

Prove of concept

In the third part of the paper Manuj et al. show that it is actually possible to follow the steps by documenting a simple simulation model very exhaustively.

Summary

Overall I very much like the papers approach.

It provides a checklist for improving the model generation and simulation process.

The prove of concept already includes some risks though it is especially interesting read if you want to use this as a (very) basic model for SCRM.

The paper tackles a problem which is intrinsic to many articles in many papers, they are written to provide you with a glimse of the results. Very often the methodical foundations and side results are just ignored and not documented. Progress in science very much depends on the possibility to build on those models.