Physical and numerical modeling of performance of detention dams

Abstract

In this study, the trap efficiency (TE) of detention dams was investigated using a laboratory study. To express the mathematical relationship between dependent and independent parameters, nonlinear regression and soft computing techniques including multivariate adaptive regression splines (MARS), gene expression programming (GEP) and group method of data handling (GMDH) were utilized. To find out the most effective parameters on TE, analyzing the Gamma Test (GT) and reviewing the structures of GMDH, MARS and GEP were performed. The accuracy of empirical formula derived based on the nonlinear regression with (R2=0.65 and RMSE=15.17) was inappropriate. Evaluating the performances of GMDH, GP and MARS with regarding calculation of standard error indices such as (R2 and RMSE) indicated that all of them with minimum value of R2=0.95 and maximum value of RMSE=5.79 have suitable accuracy for modeling and predicting the TE. However, the MARS model is a bit more accurate than the others. Results of GT and reviewing the structure of prepared models (GMDH, MARS and GP) declared that ratio of volume of flood to volume of sediments (VF/VS), ratio of mean diameter of sediment size to the volume of flood (D50/VF3) and specific gravity (Gs=γs/γw) are the most effective parameters on TE.