GOALI: Multiscale Decomposition Techniques for the Integration of Optimal Planning and Scheduling of Batch and Continuous Multiproduct Process Systems

Carnegie Mellon University, Pittsburgh PA

Investigators

Abstract

This GOALI grant provides funding for the development of multiscale decomposition techniques that will allow the temporal and spatial integration of planning and scheduling problems that arise in the process industry. The proposed work will initially address the case of simultaneous planning and scheduling of multiproduct continuous plants. In the second phase, it will address a real-world problem provided by The Dow Chemical Company that deals with planning and scheduling of geographically distributed manufacturing systems consisting of batch reactors that must satisfy demands for a set of markets. For the temporal integration we intend to develop a bilevel iterative scheme that involves the solution of aggregate MILP planning and detailed MILP scheduling models. To reduce the gap in the bounds predicted by each of these problems, a novel set of logic and mixed-integer cuts will be investigated. For the spatial integration we will investigate a multiscale decomposition approach through Lagrangean duality that will incorporate the bilevel temporal scheme. If successful, the proposed research has the potential of making significant contributions in the areas of planning and scheduling, which are basic building blocks for achieving the goal of Enterprise-wide Optimization in the process industry. A major result that is expected is a new methodology for integrating planning and scheduling in order to achieve consistency and coordination in long and short term production targets, which can translate into major economic savings. While the proposed GOALI project will primarily involve collaboration with Dow Chemical, it will also involve potential collaboration and dissemination of the research results with the petroleum, chemical and engineering/software companies that are members of the Center for Advanced Process Decision-making (CAPD) at Carnegie Mellon. The findings of this research will be documented as case studies that will be made available through the internet, potentially benefiting universities and companies.