Issues in solving large scale planning, design and scheduling problems in batch chemical plants

Abstract The solution of the planning/ design/ scheduling problem is limited by existing solution algorithms and computing resources. A combined planning/scheduling problem over an extended horizon of several months yields a formulation with thousands of variables. A natural approach presented earlier is to decompose the large problem into smaller ones, the solution of the decomposed problems constituting the solution to the original problem. The decomposition is however, a nontrivial task because of the global decision making process. Infeasibilities may arise at the lower stage since we have only considered the design/planning stage at a macroscopic level, therefore, a diagnostic formulation δRF is presented here. The power of the approach in identifying the infeasibilities and its use to solve the design problem is discussed in this paper. Illustrative examples for batch plant design are presented.