OPTIMAL DESIGN OF WASTEWATER TREATMENT SYSTEMS THROUGH A MILP-BASED INITIALIZATION PROCEDURE

This work presents a new approach for the optimal design of distributed wastewater treatment networks (WTN) featuring multiple contaminants. It consists of a two-stage solution procedure. The first stage makes use of a sequence of mixed-integer linear programs (MILP) to generate a striking number of structurally different suboptimal networks. Each will correspond to a starting point for the solution of the general nonlinear program (NLP), in the second stage. The initialization strategy assumes that the wastewater streams go through the treatment units in series from the moment they enter the system up to the discharge point. On each calculation stage, binary variables are used to select the best unit for that position, while ensuring that the remaining treatment units are powerful enough to take the contaminants concentration down to the discharge limits. The approach is illustrated through the solution of several test cases taken from the literature and compared to that of our previous algorithm. Best solution returned and total computational efforts are the critical indicators for model performance evaluation.