Scheduling of batch chemical plants with resource constraints
Abstract
A wide variety of processing industries face the problem of determining the order in which different products are to be processed so as to optimize an appropriate performance criterion and satisfy a set of production constraints. Given the product demand pattern and due dates, the product recipes including storage policy and resource availability patterns, the scheduling program generates a detailed schedule based on the performance criteria specified, while trying to satisfy all the constraints. The general network multiproduct batch plant under due date restrictions and with resource constraints including the availability of intermediate storage between some or all of the stages is examined. Parallel units are allowed to have different sizes and to have constrained connectivity. The storage vessels are allowed to hold more than one batch of an intermediate, thus allowing aggregation of batches in storage. The product recipe allows for product-dependent setup times, material transfer times (in and out of units) clean-up times and resource constraints. Also, after processing of a batch, a storage policy may be imposed depending on the stability of the product and availability of storage vessels. The scheduling problem is decomposed into two subproblems: sequence determination and completion time calculation. In view of the highly combinatorial nature of this problem (it is undoubtedly NP-complete), the solution of the problem relies on heuristic methods both to select the product sequence and to determine the resource-constrained completion time. Two heuristic approaches are used for the completion time; these methods allow for merging and splitting of batches between stages and units. Thus, in this work the concept of product batch size exists only at the stage and unit level and the product demands are described in amounts rather than batches. A simplified and efficient resource satisfaction strategy is used for the resource constraints. The sequence selection subproblem involves repeated calculation of the performance criteria in order to select the preferred sequence. A number of sequencing heuristic rules are used for generation of the product sequences. All the results are presented graphically. The effectiveness of the composite method is demonstrated through extensive test cases.
Degree
Ph.D.
Advisors
Reklaitis, Purdue University.
Subject Area
Chemical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.