Generating Alternatives Using Evolutionary Algorithms for Water Resources and Environmental Management Problems

Contemporary heuristic search procedures [e.g., evolutionary algorithms (EAs)] continue to offer increased capabilities for systematic search for a range of water resources and environmental management problems. These problems are often riddled, however, with numerous unquantifiable issues that are important when making decisions, but escape being incorporated in the system model. The mathematically optimal solution to such an incompletely defined model may be found unrealistic or altogether incorrect for the real problem. Optimization procedures could still be made useful if they can be utilized effectively to generate, in addition to the optimal solution, a small number of different alternatives that are near optimal. Alternatives with maximal differences in the decision variable values are expected to perform differently with respect to the unmodeled issues, providing valuable choices when making decisions. Although successful alternative generation procedures have been reported for mathematical programming-based search procedures, they are yet to be explored fully for EAs. This paper describes an extensive investigation of a new EA-based alternatives generation procedure, the evolutionary algorithm to generate alternatives (EAGA). A previously published regional wastewater treatment optimization study is used as a basis for establishing and demonstrating the capabilities of EAGA, and the set of results from the previous study is used as a benchmark for comparing the performance of EAGA. Comparisons of results indicate that EAGA is effective in generating good alternative solutions that perform differently with respect to several unmodeled issues. EAGA is sufficiently flexible to be applied to a wide range of water resources and environmental management problems. Further, EAGA can be applied to any problem that is set up to be solved using an evolutionary algorithm.