Simulation of Scour Depth Around Twin and Three Piers Using Group Method of Data Handling

Estimation and computation of scouring around structures such as piers has a significant importance. In this study, scour depth in the vicinity of twin and three piers was simulated using Group Method of Data Handling (GMDH). First, effective parameters on scour depth were identified and then four different GMDH models were defined. To verify the simulation results, some experimental measurements were applied and 70% of these data were utilized to train the GMDH models, whereas 30% of the data were employed to test the models. Subsequently, the best GMDH model and the most influencing input parameters were introduced by conducting a sensitivity analysis. The sensitivity analysis showed that the GMDH models estimated the scour depth with acceptable accuracy. For instance, the correlation coefficient (R), scatter index (SI), and variance accounted for (VAF) for the best GMDH model were respectively calculated to be 0.949, 0.212, and 90.129. In addition, the Froude number was detected as the most important input variable to estimate the scour depth through GMDH model. Moreover, the mean discrepancy ratio (DRave) for the superior GMDH model was computed to be 1.228. For different GMDH models, four equations were presented and lastly a computer code was provided to simulate scour depth by means of the GMDH model.