Superstructure Approach to Batch Process Scheduling by S-graph Representation

Scheduling plays a key role in batch process operation; it has a major effect on the process' performance. Available methods for determining the optimal schedule are primarily based on either MILP/MINLP formulation in conjunction with mathematical programming (Floudas and Lin, 2004; Vaklieva-Bancheva and Kirilova, 2010) or graph representation in conjunction with combinatorial algorithms (Sanmarti et al., 2002).The current work comprises three major contributions. First, an algorithm has been crafted to generate a superstructure for a scheduling problem. The problem is defined in the form of an S-graph representing the recipe. The superstructure contains exclusively every step potentially performed by any of the functional or operating facilities or equipment units capable of completing at least one task to be scheduled. These steps involve executions of tasks and changeovers from one task to another. Second, an MILP formulation is elaborated on the basis of the superstructure, which guarantees the optimal solution of the scheduling problem. Third, a relaxation of the MILP model is incorporated into the S-graph algorithms to support the selection of subproblems and decision variables in the branch- and-bound procedure. © 2011 Elsevier B.V.