Demand and cost forecast error sensitivity analyses in aggregate production planning by possibilistic linear programming models

Abstract

A production management system contains many imprecise natures. The conventional deterministic and/or stochastic model in a computer integrated production management system (CIPMS) may not capture the imprecise natures well. This study examines how the imprecise natures in the CIPMS affect the planning results. Possibilistic linear programming models are also proposed for the aggregate production planning problem with imprecise natures. The proposed model can adequately describe the imprecise natures in a production system and, in doing so, the CIPMS can adapt to a variety of non-crisp properties in an actual system. For comparison, the classic aggregate production planning problem given by Holt, Modigliani, and Simon (HMS) is solved using the proposed possibilistic model and the crisp model of Hanssmann and Hess (HH). Perturbing the cost coefficients and the demand allows one to simulate the imprecise natures of a real world and evaluate the effect of the imprecise natures to production plans by both the possibilistic and the crisp HH approaches. Experimental results indicate that the possibilistic model does provide better plans that can tolerate a higher spectrum of imprecise properties than those obtained by the crisp HH model.