A Reliability Model Validation Method for Mitigating the Effects of Measurement Uncertainty
Abstract With the increasing role of numerical modeling in engineering design and development processes, improved techniques are needed for validating computational results against experimental measurements. Most existing validation methods suffer from two main limitations: (i) they are often highly sensitive to the experimental measurement uncertainty, and (ii) extending these methods for reliability model validation requires large quantities of failure data that may be very time-consuming or costly to obtain. In order to overcome the aforementioned limitations, this study proposes an indirect reliability model validation method. First, a new procedure for computing a validation metric is developed based on Richardson extrapolation (RE) to reduce the sensitivity of the metric to the experimental measurement uncertainty. Second, a new validation metric is defined based on the limit state function (LSF) approximation to extend numerical model validation to reliability model validation. The proposed method is illustrated by validating a reliability estimation model for a cantilever beam under a vertical load.