Methodologies for cavitation prediction on high-head spillways: Review and propose a hybrid approach

Abstract

Cavitation on high-head spillways can cause rapid concrete damage, posing serious threats to structural safety during flood releases. This review synthesizes methodologies for predicting cavitation inception, intensity, and damage risk across laboratory, field, numerical, and machine learning approaches. Both physical and numerical methods have been widely applied; however, they present limitations such as high cost, time consumption, and applicability restricted to individual case studies. The literature indicates that for spillways with similar configurations, the locations most at risk of cavitation damage are typically the same. Building on this insight, this study introduces a novel procedure for predicting cavitation risk by integrating advanced techniques including machine learning and 3D CFD modeling to develop a rapid prediction tool for quantifying the cavitation index. Keywords: High-head spillway, procedure of prediction cavitation, 3D CFD, ML method