Efficient Computational Methods for Probabilistic Damage Tolerance Design

Abstract

Aircraft engine safety is influenced by material anomalies that can form during manufacturing melt and machining operations. These anomalies may occur at any location within a component and have led to catastrophic engine failure events and loss of life. A probabilistic damage tolerance (PDT) approach is used to assess the fracture risk associated with these events. It involves probabilistic treatment of material anomaly size, frequency, and location as well as other variables that are related to applied loads and material properties and the effectiveness of non-destructive inspections. During the design process, PDT is applied to large finite element (FE) component models to assess the suitability of components for use in commercial aircraft engines. The design process is directly impacted by the computation time of the PDT assessment. In this paper, approaches are presented to improve the computational efficiency of PDT assessments. A strategy for the optimal discretization of a FE component model into regions called zones is described to address the uncertainty in the location of material anomalies. This strategy reduces computation time by a factor of ten thousand or more depending on the size of the FE model. Anomaly frequency affects the size of the largest anomaly that may be found in a component. A conditional probability method is combined with an extreme value-based formulation to adjust the anomaly size cumulative distribution function considering the expected number of anomalies in each zone. This approach significantly reduces the number of simulations that are needed to satisfy computational accuracy requirements. These strategies are applied to the efficient certification of aircraft engines according to United States Federal Aviation Administration (FAA) and European Aviation Safety Agency (EASA) requirements. Examples are presented to illustrate the approaches.