Generating Alternative Watershed-Scale BMP Designs with Evolutionary Algorithms

The first part of a two-step decision-making framework for watershed-scale stormwater runoff control (Kaini et al., 2008, this meeting) involves identification of the most cost-effective combination of structural BMPs that meet target peak flow and sediment reduction criteria. This paper presents the second half of that framework: simultaneous generation of near-optimal alternative design strategies using a Euclidean distance metric. Structural BMPs included in this model include detention ponds, infiltration ponds, field borders, grade stabilization structures, and grassed waterways. Alternative designs are identified by coupling Soil and Water Assessment Tool (SWAT) and a Species Conserving Genetic Algorithm (SCGA). In addition, we demonstrate SCGA's flexibility and efficiency at generating alternative designs as well as varying numbers of alternatives. The model is demonstrated on Silver Creek watershed, a sub-watershed of the larger Lower Kaskaskia watershed in southern Illinois.