Parametric programming technique for global optimization of wastewater treatment systems

This paper presents a parametric programming technique for the optimal design of industrial wastewater treatment networks (WTN) featuring multiple contaminants. Inspired in scientific notation and powers of ten, the proposed approach avoids the non-convex bilinear terms through a piecewise decomposition scheme that combines the generation of artificial flowrate variables with a multi-parameterization of the outlet concentration variables. The general non-linear problem (NLP) formulation is replaced by a mixed-integer linear programming (MILP) model that is able to generate near optimal solutions, fast. The performance of the new approach is compared to that of global optimization solver BARON through the solution a few test cases