An interval-based framework for the scheduling of resource-constrained batch chemical processes
Abstract
With the shift from producing high volume chemicals to more specialized production of high value, low volume specialty chemicals, batch processing has become an important mode of operation. Additional complexities in plant operation arise in noncontinuous processes which are not significantly present in continuous processes. The scheduling of production tasks subject to resource constraints is in particular an important problem. Accepted methods for solving scheduling problems rely on either heuristics or mathematical formulations. In this work, a framework with many aspects of both approaches is produced. The essence of the method is in the treatment of a set of continuous time intervals and the relationships between the intervals. After processing intervals for each recipe step and resolving resource conflicts, a single interval is left for each recipe step during which the step may float freely and not cause resource constraint violations as long as the generated precedence constraints are followed. Exactly specifying a start time for each step is then accomplished using a linear program formulation, which is also structured in order to discover areas in the schedule where storage must be used and where the use of storage may result in a more efficient schedule. In contrast to traditional modelling techniques, time is only discretized when real process events occur. The framework is able to completely consider the entire search space without using the conventional fine, uniform discretization. Examples solved with the prototype implementation of the method indicate promise for such a framework to be used in a practical environment. Finally, a mathematical formulation is given for the framework in the form of a mixed-integer non-linear program, which is subsequently linearized in an efficient manner.
Degree
Ph.D.
Advisors
Reklaitis, Purdue University.
Subject Area
Chemical engineering|Operations research
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.