Estimation of Scour Depth in Pipelines Laid on the Seabed under Live Bed Condition using a Hybrid 〖M5〗^'-GLUE Model

Estimation of scour depth in pipelines laid on the seabed is an important subject in pipeline design process. Due to the variability of scouring nature, a hybrid stochastic scouring decision tree model (M5ʹ-GLUE) was developed using GLUE approach and it was compared with other deterministic scouring decision tree model under live bed condition. The results obtained in both train and test steps showed that the mean value of determination coefficient of hybrid model in estimation of scouring was 0.72. Therefore, the M5ʹ-GLUE model has good accuracy in the estimation of pipeline scouring. Furthermore, having employed two other index Root Mean Square Error and Agreement Coefficient, it was revealed that the accuracy of scouring estimation was improved by 18 percent using hybrid model in comparison to deterministic tree model. Less improvement was occurred under live bed condition with small (e/D). Generally, it can be concluded that the developed hybrid model in comparison to the deterministic tree model has good accuracy for estimation of scour depth in pipes laid on the seabed.