Optimal operation of quality controlled product storage

The mixed integer linear programming model has been developed for the produ ction scheduling of polybutene (PB) process of Daelim Industrial Co. (DIC) in Korea. The PB process is composed of one reactor, 13 storage tanks and t wo package types. It produces seven reactor products. Most of the tanks are practically dedicated to a specific product but in general, any tanks can be shared by multiple products. Also, two major products use multiple tanks . The reactor is operated on block mode. The products are liquid and they s hould be stored in tanks before packaging. The unique characteristics of th is scheduling problem exists on the fact that the product in tank should be locked for 3-15 days in order to check the quality specifications after ru n-down from the reactor. The primary objective of scheduling is reducing th e number of quality checking process because it causes great loss in produc tion time and tank utilization. The model is composed of about 1500-2000 in teger variables. The branching priorities are adjusted according to the imp ortance of variables based on current manual scheduling practice and it gre atly enhanced the convergence of mixed integer model. The scheduling system was successfully implemented in real plant and replaced the manual work of scheduler. (C) 2000 Elsevier Science Ltd. All rights reserved.