Improvement of the computational performance of STN/RTN formulations for short-term scheduling of multipurpose batch plants

GUILLERMO A. DURAND; J. ALBERTO BANDONI

Lyon (Francia)

Congreso; 18th European Symposium on Computer Aided Process Engineering; 2008

The short-term scheduling of multipurpose batch plants is a difficult problem for which several formulations exist in the literature. These formulations generate MILP models using binary variables to indicate the starting/activation/finishing or not of the tasks involved in the plant operation. However, in these formulations the process recipe is modeled thru equations using only continuous variables or mixing continuous variables with binary variables. When the scheduling problem resulting from such formulations is solved by a branch-and-bound strategy, the binary variables’ relaxation does not generally find the correct values (0-1) for these variables straightforwardly. In this work we present a new scheme of modeling the process recipe in short-term scheduling of multipurpose batch plants, using only the binary variables present in each formulation, that improve the computational performance of the solution by helping the branch-and-bound strategy to find the correct values of binary variables in a faster way. The technique, originally developed for the short-term scheduling of mono-purpose mixed continuous/batch plants, increases its potential of improvement when the process recipe is carried out in multipurpose equipment units that can perform several tasks.