Go Back to Homepage
Charles R. Sox
Auburn University
Industrial Engineering
June 1995
We provide some computational results for a set of new production planning heuristics that generate near-optimal solutions for the stochastic, multi-item production planning problem with or without setups. The advantage over stochastic dynamic programming is that this approach can quickly generate solutions for realistic problem instances. The heuristics rely upon a Lagrangian decomposition of the formulation which is used in a subgradient optimization algorithm. A procedure for generating feasible solutions from the Lagrangian dual solutions is also presented. We report here the results of some computational experiments using this approach which indicate that it is robust over a wide range of realistic problem instances and has the potential for use as a real-time decision support tool.
Key Words: production planning, stochastic programming, subgradient optimization
This paper is based upon work supported by the National Science Foundation under Grant No. DMI-9409344.
Go Back to Homepage