Issues of Hierarchical Planning in Multi-Stage Production Systems.

摘要

The multi-stage production system is viewed as a production process in which component parts have to be obtained by manufacturing or by purchasing then assembled into subassemblies, assemblies, and finally into the finished good. At the aggregate level a new formulation for the aggregate planning model is given, in order to bring computational feasibility to situations in which older formulations went beyond the capabilities of current linear programming codes. The resulting aggregate model is a large scale system that lends itself to solution by column generation. A dynamic programming algorithm for the generation of columns is developed. Next, at the disaggregation level, the problem of computing optimal lot sizes in multi-stage systems is addressed and solved. Since exact solution procedures are found to be either very expensive or computationally infeasible, a heuristic approach is adopted and results are reported. For more complex situations, in which parts are common to several end products or where there is independent demand for parts, even the heuristics become infeasible; therefore it is suggested that myopic lot sizing policies be used.