Various arrangements of pile groups are widely being used as supports of marine structures. As piles are located on erodible beds of the sea, scouring is a threat to such structures and the scour depth amounts should be considered well in their designs. Though most of these supports are constructed in form of groups of piles, majority of studies were concentrated on predictions of scouring around single piles whereas the arrangement of the piles and the spaces between them in arrangements as well as their geometry, sediment and wave characteristics should also be studied. Despite the importance of the scour hole depths, the existing prediction formulas are not capable of accurate estimations around pile groups with different arrangements. Hence, developing a robust model for the estimation of scour depth seems necessary. One of the most common approaches as an alternative to empirical ones is the soft computing methods. Artificial Neural Network as the most famous data-mining method has been successfully applied in scour studies. But there are still needs of more assessments in their applications on pile group case studies. In addition, Support Vector Machines as one of the recently applied soft computing models in scouring has scarcely been studied so far. In this study, series of large scale scouring experiments were done for various arrangements of pile groups with different pile and arrangement characteristics exposed to waves of shallow water and equilibrium scour depth around them were measured in wave basin of Ujigwa Open Laboratory of Kyoto University. Finally, by applying the provided experimental data, the applicability of data mining models were assessed in predictions of pile group scour properties. Results indicate that, data mining approaches can provide more reliable predictions of scouring properties due to waves compared to current available empirical formulae.
Ghazanfari Hashemi, Samaneh; Hiraishi, Tetsuya; and Mansoori, Amir Reza, "Study Of Wave-Induced Scour Depth Around Group Of Piles Using Support Vector Machines" (2014). CUNY Academic Works.