A multi-objective optimization framework to support integrated stormwater management

This thesis presents a multi-objective optimization framework to support the application of models to integrated stormwater management processes. The framework includes three main stages, (i) multi-objective calibration of the hydrologic model, (ii) multi-objective optimization of stormwater best management practices (BMPs), and (iii) evaluation of selected BMP designs using additional calibration solutions. The benefits of the multi-objective optimization framework are illustrated by using two case studies. Results from the multi-objective calibration showed that calibration trade-offs may exist. Also, selection of a calibration solution to be applied as the evaluation tool is not a straightforward process, particularly when there is more than one objective that conflict among each other. Furthermore, the design of detention ponds at the watershed scale, using an approach that combines watershed-wide performance criteria, and standard design methods, was successfully implemented using the multi-objective optimization algorithm. Finally, it was shown that the evaluation of selected detention pond designs using alternative calibration solutions may render benefits in terms of minimizing unexpected system performance due to model uncertainty. Both the calibration and the design optimization are based on the evolutionary multi-objective optimization algorithm called Non-dominated Sorting Algorithm NSGA-II (Deb et al., 2002), and the Storm Water Management Model (SWMM).