A new framework for system reliability-based design from quantile value perspective

Abstract

This study develops a new framework for system reliability-based design (SRBD) from a quantile value perspective. The design parameter of interest is treated as a variable and separated from each limit state function, enabling the failure threshold of the design parameter for each failure mode to be obtained. The so-called system failure threshold is defined as the maximum or minimum individual failure thresholds and a single run of MCS is employed to obtain the cumulative distribution function (CDF) of the system failure threshold. An SRBD aiming at any failure probability is performed by finding an upper or lower quantile value on the determined CDF. Four illustrative examples including a simple beam, a semi-gravity retaining wall, a cantilever retaining wall, and a soil-nailed wall are used to demonstrate the implementation procedures. The results are validated according to direct Monte Carlo simulations. The proposed approach is applicable when the design parameter is monotonically related to the original performance functions concerning multiple failure modes. The proposed approach can play a complementary role to the existing SRBD methods due to its simplicity and efficiency. Complex optimization or iteration procedures involved in the conventional SRBD methods are not required.