This conference paper was submitted for presentation at the NAFEMS World Congress 2025, held in Salzburg, Austria from May 19–22, 2025.

Abstract

Metamodels of Failure Probabilities (Fragility Surfaces) make it possible to explore the transition region between failure and non-failure, not only in dependence of for example the required lifetime, but also in dependence of other important parameters like maximum temperature of a thermal load cycle, etc. This approach gives new insights on Wöhler- or Weibull-like curves, surfaces. For example, the Wöhler-like curves, surfaces are the isolines of the Fragility Surfaces. The Fragility Surfaces have many possible cross domain applications, for example in Prognostics and Health Management based on digital twins. We will focus on several of these cross domain applications. Qualitative and quantitative failure analysis is important especially for the reliability of components or systems. The information about the transition region from non-failure to failure is a critical information. This information should be passed along the value chain while keeping IP non-disclosed. Metamodels provide the possibility to exchange this information without exchanging other more IP related information about the simulations. Usually, a huge number of simulations is necessary to study the transition region between failure and non-failure, if we take tolerances into account. Fragility Surfaces opens a door to a quantitative approach for a more detailed discussion and examination of the transition region between failure and non-failure enabling many cross-domain applications with a strongly reduced number of necessary simulations. We made new workflows that can be used for creating Fragility Surfaces. Especially the methodology of nested workflows is very useful. In the inner loop we run robustness analysis with stochastic parameters (tolerances for example) and in the outer loop we have the control and requirement parameters. The control parameters are typically the most important parameters (maximum temperature, etc.) and a requirement parameter can be a minimum requested lifetime. With that developed methodology we are now able to apply some insights, techniques from robust design optimization in a very flexible way giving possibilities to a lot of new applications especially using Fragility Surfaces. We can use this methodology to understand the transition region between failures and non-failures in an early phase of the development based on quantitative failure analysis. The new workflows finally enable the robustness analysis of designs using fragility surfaces, for example helping to ensure that optimized designs will fulfil their required reliability to a high degree of certainty.