CONTINUOUS-TIME REPRESENTATION IN BATCH SEMICONTINUOUS PROCESS SCHEDULING - RANDOMIZED HEURISTICS APPROACH/

This paper address the short-term scheduling problem for multipurpose/ multiproduct batch and semicontinuous processing systems. The nonunifo rm time discretization model (NUDM) of Reklaitis and Mockus (1995b), u nder which binary variables are used to represent occurrence of start and stop events for the various recipe tasks, is extended to accommoda te sequence dependent changeovers, non-dedicated storage, and semicont inuous tasks. Since the short term scheduling problem of large batch/s emicontinuous plants may be expensive to solve because of its high com binatorial complexity, we adapt and test a bayesian approach to discre te optimization, namely the randomized heuristics technique of Mockus el al (1994). Under this approach, instead of solving the original pro blem with its large number of binary variables, one solves low a dimen sionality heuristics calibration problem which has embedded in it a he uristic solver to the discrete optimization problem. Although any of a number of heuristics suitable for a specific class of problems can be employed within this framework, three different heuristics are tested in this work: simulated annealing, a general polynomial scheme, and a specialized heuristics tailored to the batch scheduling problem struc ture. The proposed framework readily lends itself to parallelization. Computational comparisons are also reported to solutions obtained via an existing uniform discretization based MILP formulation. It is shown that for test problems proposed formulation and bayesian solution app roach consistently outperforms the UDM formulation solved via conventi onal branch and bound based solution techniques. The results suggest t hat the NUDM/bayesian approach shows considerable promise for the solu tion of a class of large and realistic batch scheduling problems.