| New aircraft systems are likely to be constructed of novel materials, operate in extreme environments -- such as hypersonic, and be of limited production runs. As a result, experienced-based methods for determining critical failure locations and failure modes may be lacking, and large complex models may be too time consuming to analyze every location. A recently proposed hierarchical system reliability-based methodology uses FORM and a formal pair-wise error metric to filter out noncritical locations in a model. In this paper, this technique is expanded to efficiently identify critical failure locations in models with spatially varying parameters -- modeled as Gaussian random fields -- such as distributed loads, material properties, or geometry. The Karhunen-Loeve Transform, an eigen-decomposition of the random field, is used to reduce the number of random variables in the problem and reduce computational cost. Once the critical locations have been determined, a system reliability estimate for the structure is obtained. A numerical example is presented to demonstrate the methodology and the effect of the random fields on the critical locations. It was found that decreasing correlation lengths of the random field can result in more critical failure locations. |
