Application Of Importance Sampling Method On Caisson Breakwaters Reliability Analysis

Sliding, the main failure mode in caisson breakwaters stability analyses, is the instability phenomenon frequently occurred in Actual Engineering. The traditional reliability design method of the caisson breakwaters do not allow to takes place any sliding. If breakwaters have some sliding, we define it failure. In practice, caisson breakwaters appear some sliding do not affect its function. In recent years, the relevant researchers, in Japan, proposed the designing conception of allowing caisson to appear slipping, and controlled the slippage within the range of allowable design. The allowable slippage is the total sliding distance in design reference period. Because of the complexity and randomicity of marine environment, the calculation of total sliding distance generally use Monte Carlo Simulation Methods. The failure probability, the probability of caisson slippage oversteps allowable value, have to simulate more than 10000 times by Monte Carlo Simulation Methods, which efficiency is so low and affect its application in Project. This paper improve basic Monte Carlo Method, apply kernel method of importance sampling to simulate the reliability of the caisson breakwaters, estimate the probability of failure in the complete life cycle of breakwaters, and improve the efficiency of simulation. Regarding the total sliding distance in the complete life cycle of breakwaters as the failure criteria, use wave spectrum to simulate the wave in front of breakwaters, and apply mass-spring-dashpot mode to calculate total sliding distance.The main content included in the paper is as follows:1. A random wave of year storm has been generated by the wave spectrum. Using the improved time history of wave force and the dynamic analysis model of caisson breakwaters to calculate the cumulate sliding distance in year storm.2. The reliability of the caisson breakwaters is simulated by kernel method of importance sampling, improved Monte Carlo Method, which can greatly enhances the efficiency of simulation.3. Changing the parameters of kernel sampling density to contrast the efficiency of sampling. Contrast the efficiency difference between basic Monte Carlo Method and kernel method of importance sampling.4. The effect of parameters on the probability of failure and ESD is analyzed by changing the parameters in dynamic model.