A Randomization Process for Modeling Constructed Wetlands with an Optimization Example

摘要

Constructed wetlands are becoming increasingly popular in managing stormwater and other wastewaters. Studies examining the relationship between performance and design parameters suffer from two flaws: 1) small sample size and 2) non-random samples. These constraints limit the applicability of existing performance predicting equations. Therefore, the goals of this research were to 1) develop a methodology for creating randomly generated wetland designs and 2) use these designs to develop a set of equations predicting peak flow reduction.Using various software packages, 2,000 constructed wetland designs were randomly generated. A flood wave was routed through each design to assess the peak flow leaving each design. Channel length was the best statistical predictor of peak flow reduction accounting for >70% of the total variation within the dataset. This equation represents the most general equation to date and can be used to better design constructed wetland systems.