Prediction of cavitation damage using SVM model based on air–water two-phase flow over dam spillway

Cavitation is one of the primary causes of breakdown and failure on chute spillways, causing surface damage and structural destruction. In this research, a three-dimensional two-phase flow over an ogee spillway was modeled using the FLOW-3D model for the Gelevard-Neka spillway and validated with the available field data. After analyzing the hydrodynamic parameters of flow, a method was presented to predict the intensity and location of cavitation damage on the spillway surface based on the support vector machine (SVM) model. The hydraulic parameters, including flow velocity, pressure, and cavitation index, were introduced to the SVM model, and the cavitation damage level, from no damage to major damage, was predicted along the spillway structure. The validation flow results agreed well with the field data, and the normalized root-mean-square error value of 0.0196 was obtained. In the prediction of cavitation damage using the SVM model, the MAE, R, and RMSE for the training stage were, respectively, 0.32, 0.882, and 0.127, and for the testing stage were 0.024, 0.857, and 0.133. The results show reasonable performance of the SVM model in the prediction of cavitation damage. According to the results, the spillway is susceptible to cavitation damage with the most significant damage anticipated to occur in the distance range of 70–190 m from the spillway origin. Based on the importance of the aerators in protecting the spillway from cavitation damage, it is recommended to investigate the various effects of aerators on mitigating cavitation damage.