Modeling Scour around a Spur Dike in Bends Using Fuzzy Logic and Artificial Neural Network

Spur dike is a structure made of stone, sand, rock, soil or concrete, which is usually built with an angle relative to the bank to divert the flow from the banks and concentrate it towards the centerline of the river in order to prevent bank scouring. One of the main problems regarding this structure is its stability due to possibility of the scouring around the nose of structure. Therefore modeling the amount of the scouring around the structure according to the flow conditions is important and essential. In this research the laboratory data of scouring around a spur dike for different flow conditions in a 180° bend were applied for modeling this phenomenon using Fuzzy Logic model (FLM) and Artificial Neural Network (ANN). The scour was modeled as a function of the length and the installation angle of spur dike in bend, and the upstream Froude number. The results showed that the ANN and FLM models were able to predict the amount of scouring, accurately. A regression equation was also developed for describing the amount of scouring around the spur dike using the corresponding measured values empolyed for producing and calibrating the pattern of the ANN and FLM models. The results obtained from ANN, FLM and regresion models were then compared together using another series of existing data, which had not been applied for developing those models.