Efficient Reliability-Based Design Optimization Framework Based on an Improved Inverse Quantile-Based First-Order Second-Moment Method

Reliability-based design optimization (RBDO), which aims to achieve the optimal design with minimal cost while simultaneously meeting the required reliability level, generally necessitates an extensive number of evaluations of the limit state functions of geotechnical structures. As a result, RBDO is often computationally prohibitive, especially when simulation-based methods are used for reliability analysis. To address these challenges, this paper proposes an efficient RBDO framework by transforming the optimization of multiple design parameters and multiple target constraints into a series of estimations of a single design parameter with a single target constraint. This transformation can be accomplished using an inverse reliability analysis method. An improved inverse quantile-based first-order second-moment algorithm is proposed to perform the inverse reliability analysis. The inverse quantile-based first-order second-moment algorithm is both efficient and accessible to the physical normal space of random variables. Its procedure for searching the design parameter shares certain features with the widely accepted load and resistance factor design method, making it more easily understandable to geotechnical engineers. Four geotechnical design examples are presented to demonstrate the performance of the proposed method. The results indicate that the proposed method can accurately compute multiple design parameters while minimizing construction costs. It offers a practical pathway for geotechnical engineers without a strong background in probabilistic theory to transition from semi-probabilistic to fully probabilistic design.