Estimation of Pipeline Scour due to Waves by GMDH

Abstract

In the present study, the group method of data handling (GMDH) network is applied to predict scour depth below pipelines exposed to waves. The GMDH network is trained using a back-propagation (BP) algorithm. The pipeline scour is modeled as a function of three-dimensionless parameters, including the Keulegan-Carpenter number, the ratio of the initial gap to pipe diameter, and the Shields parameter. The performances of the GMDH network are compared with the adaptive neuro-fuzzy inference system (ANFIS) model, the model tree (MT), and empirical equation. The results indicated that the GMDH network produced a more accurate prediction of scour depth compared with other models.