A simulation study of simplification strategies in the development of optimization models
A major decision in the development of optimization models concerns the degree of accuracy to be used in the model. In practica, this simplification decision is widely met in an ad hoc fashion. Recently, several ways of Computing (worst-case) bounds for simplifying linear programming (LP) models have been developed. On the other hand, the need for empirical studies of simplification effects has been expressed. This paper discusses LP-models for product-mix planning in a job-shop in which manufacturing options exist. In this decision Situation some simplification (compared to the data used in nonsimultaneous planning) is often necessary to obtain a computa-tionally manageable LP model. Several simplification strategies are defined and applied in the Simulation study. Properties of the solutions obtained after applying the different simplification strategies are compared to those of the "real" Optimum. Significant differences of the objective function values allow ranking of the simplification strategies and giving recom-mendations for practical model development.