AN INTERACTIVE SCHEDULING SYSTEM FOR THE OPERATION OF MULTI-PRODUCT PLANTS (FLOWSHOPS, CONSTRAINTS, GRAPHICS)

WILLIAM WIEDE, Purdue University

Abstract

An interactive solution methodology is developed for the scheduling of products in processing networks typically encountered in the chemical processing industry (CPI). These multi-product plants can frequently be represented as multi-stage serial processes or as single stage parallel processes. A number of recursive algorithms are presented for the construction of production schedules in serial unit networks operating under any of five intermediate storage policies: unlimited storage (UIS), finite storage (FIS), no storage (NIS), no wait (NW) and mixed storage (MIS). New recursive algorithms are developed for obtaining schedules for a two unit FIS serial processing network and for the multi-unit FIS and MIS serial processing networks. A rudimentary parallel unit scheduling procedure is also given for processing products on multiple units having identical and uniform production rates. For a selected product sequence, completion times are computed for each product on each unit in a recursive fashion based on a number of simplifying assumptions. This set of product completion times can then be used to evaluate any selected performance criteria and production constraints. The best sequence with respect to the selected performance criteria; makespan, maximum tardiness and total set-up time, and with respect to any specialized production constraints can then be constructed from heuristic algorithms and a branch and bound optimization procedure. An interactive scheduling package, INTSPAK, is developed for the scheduling of products in multi-product plants. This scheduling package will allow schedules to be constructed for serial and parallel unit processing networks either interactively by the user or with the help of the heuristic and optimal scheduling algorithms. The interactive features of INTSPAK provide flexible data input/output problem definition facilities, accommodate many scheduling problem structures for serial and parallel unit networks, improved numerical and visual display of intermediate and final generated sequences, allows user step by step control of the scheduling algorithms, and provides a sensitivity analysis of generated schedules. A number of actual industrial processing situations are investigated by using the scheduling algorithms and program capabilities available in INTSPAK.

Degree

Ph.D.

Subject Area

Operations research

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS