ABSTRACT
The failure of a structural system is usually governed by multiple failure criteria, all of which are to be taken into consideration for the reliability estimation. If all the uncertain parameters are defined as random variables, then the reliability of a structural system can be estimated accurately by using the existing techniques. But when the knowledge about some of the variables is limited to lower and upper bounds, the entire range of these bounds should be explored while estimating the bounds on the reliability. The computational cost involved in estimating these bounds increases tremendously because a single reliability analysis, which is a computationally expensive procedure, is performed multiple times for each configuration of the interval variables. To reduce the computational cost involved, high-quality function approximations for each of the limit states and the joint failure surface are considered in this paper. The proposed technique is demonstrated with a numerical example.