OPERATION OF PIPELESS BATCH PLANTS - I - MILP SCHEDULES

Pipeless batch plants are an emerging technology for multipurpose chem ical batch production. Optimization formulations for the design, layou t and scheduling of pipeless batch plants have been developed. These m ixed integer linear programming (MILP) formulations make certain assum ptions about the operation of this new type of batch plant. For exampl e, the detailed movement of vessels is left up to a control algorithm, and the travel times are assumed to be long enough to allow conflicts to be resolved. In this paper we examine the implementation of plant schedules generated by the MILP. A discrete event simulator (DES) has been coupled to the output of the MILP to simulate plant operation whi le allowing for variations in its operation such as vessel travel time s and task durations. First, the paper describes how the MILP schedule is implemented in the DES, given that DES are event driven rather tha n time driven. Second, the performance of two different plant layouts, with the same equipment, is compared. The DES is able to differentiat e the performance of these layouts whereas the differences were invisi ble to the MILP due to the assumptions involved in the formulation. Th ird, small pertubations to processing and travel times are introduced in the plant. The overall plant performance does not degrade significa ntly indicating that the tightly coupled MILP schedules are not hypers ensitive to the variations. (C) 1998 Elsevier Science Ltd. All rights reserved.