In energy-intensive industries, the traditional strategy is to schedule the production first. From the production schedule, the demand for energy is predicted and an optimization of the energy-supply cost is performed. The academic approach is to combine all the production and energy-cost related constraints into a single monolithic problem. In contrast, in this work a different approach is proposed. The energy-cost optimization problem is solved using Multiparametric Programming (mp), separately from the scheduling problem. Based on the solution from the mp-MILP (Mixed Integer Linear Programming) problem, several production scheduling problems with sensitivity information from the mp-MILP solution embedded can be solved in parallel in order to find the system-optimal solution. The approach is tested on realistic data instances of a stainless-steel process and obtains the optimal solution. However, the computational performance is strongly limited to very small instances due to limitations of the mp-MILP solvers.
QC 20170224