Logic cuts for processing networks with fixed charges

Abstract: "We show how some simple logical constraints can substantially accelerate the solution of mixed integer linear programming (MILP) models for the design of chemical processing networks. These constraints are easily generated in a preprocessing stage and can be applied either symbolically during a branch-and-bound search or as constraints in the MILP model. Furthermore, they represent a new class of cuts, 'logic cuts,' that generalize traditional cutting planes. A logic cut can cut off feasible points but does not change the optimal solution. We establish some elementary properties of logic cuts and use them to show that our logical constraints for processing networks exhaust all possible logic cuts for these problems. Preliminary computational results are presented, using OSL."