Abstract: The response surface method is in wide use for calculating the reliability of gravity dams against sliding, but traditionally it is short of parameter screening analysis and its previous studies lack calculations of reliability index using an intelligent algorithm. Aiming at the problems, this paper develops a weighted dynamic response surface reliability analysis method using the cat swarm optimization that can overcome the shortcoming of massive repeated sampling in the Monte Carlo method. First, considering the dam foundation seepage-stress coupling, we choose key rock mass parameters through quantitative global sensitivity analysis. Then, the implicit performance function is approximated by a weighted response surface, and dynamic updating along with reliability index calculations is achieved using the cat swarm optimization. Finally, we analyze the sensitivity of dam foundation parameters and its engineering reliability in a case study. The results show that the shear strength parameters of dam foundation have a significant influence on dam stability and the reliability index is 2.71, which agrees well with the calculation results using the standard method. Our new response surface method is effective and reliable and would help further research.

Key words: gravity dam, reliability, cat swarm optimization, EFAST method, seepage-stress coupling