System reliability of a gravity dam-foundation system using Bayesian networks

Abstract Traditionally, the reliability of a gravity dam-foundation system is evaluated using several typical dam sections. However, the performance of different dam sections may vary greatly even under the same environment and loading conditions due to the different dam structures and geological conditions. Therefore, an improved method was proposed in this study to assess the system reliability of a gravity dam-foundation system more reasonably. Based on the synchronous load increment in the whole dam and the response surface method (RSM), the critical failure modes were determined, and the RSM agent models for the strength failure and instability failure performance functions were constructed. Through the Monte Carlo simulation and the Bayesian progressive deduction, the method of system reliability was proposed. The application results of a gravity dam showed that the reliability in different dam sections was different for the same loading and environment conditions, and the system safety assessment was unreasonable using several typical dam sections. The proposed method overcame the defects of excessively searching the failure path, distorting the failure probability assessment, and incompletely describing the characteristics of the system using the traditional methods. It provides a new method for the system reliability evaluation of large-scale and complex structural systems.