Determination of the Load-Resistance Factors for Reliability-Based Codes by Optimization

Abstract

In current years, the limit state design is the general trend for bridge design specifications which suggests load-resistance factors for structural members of a bridge with dominant ultimate limit states to equally satisfy the target reliability index. To verify that existing load-resistance factors equally satisfy the target reliability index, the reliability assessment is executed for AASHTO LRFD and KHBDC-LSD by calculating the required strength which can be generated from load-resistance factors. Prior to the reliability assessment, the normalization is introduced in this study, which nominal loads and nominal resistance are expressed as the ratio to the total nominal load. The generalized reliability assessment is possible by applying the normalization without the length and the cross sectional details of a bridge related to absolute values of nominal loads. According to the reliability assessment results, both specifications do not guarantee equally the target reliability level. To resolve this problem, it is necessary to establish calibration formulation which can generate load-resistance factors to satisfy evenly the target reliability index. By applying this calibration process, revised load-resistance factors are proposed for both specifications. In this study, the exact values of load-resistance factors are calculated by applying the inverse reliability analysis given statistical properties of variables. From the results, ultimate limit state regions determined newly. In addition, optimization process within each ultimate limit state region is carried out for proposed load-resistance factors. Load-resistance factors are decided relatively by optimization following the variance of resistance and those of loads. According to the reliability assessment with proposed load-resistance factors, both specifications satisfy the target reliability level equally. This study is thoroughly based on statistical properties of load-resistance variables, which means there is a need of the strict research about statistical properties of variables considered in the bridge design preveniently.