Probabilistic Fatigue Methodology for Aircraft Landing Gear

Abstract

Aircraft landing gear comprise of safety-critical structural components that are exposed to large cyclic loads over long in-service design lives. To prevent the occurrence of fatigue crack initiation, fatigue analysis methods are employed to identify the `safe-life’ at which the component must be retired, to guarantee the structural integrity of the component. However, the engineering parameters relating to the fatigue design of landing gear components, including material properties and loading, demonstrate significant variability. This variability is currently mitigated using design conservatism.

Design conservatism can ultimately lead to over-weight components and as a result, probabilistic design approaches have been proposed to better represent design parameter variability within fatigue analysis processes. Unfortunately, a large number of inhibiting factors, or `blockers’, currently prevent the wider-scale implementation of probabilistic fatigue design approaches.

The aim of this research is to develop a probabilistic fatigue methodology that overcomes the blockers to a probabilistic design approach. It is hypothesised that careful selection of a Monte Carlo Simulation based methodology, definition of systematic processes and frameworks, along with the exploitation of recent advances in surrogate modelling and `big-data’ sources, can help to combat the blockers to probabilistic design.

Following application of the developed probabilistic fatigue methodology to landing gear component case studies, it was demonstrated that probabilistic methodologies can support the fatigue design of landing gear components through identifying the conservatism in existing practices, along with highlighting areas of component over-design. From implementing the methodology, it was observed that the proposed methodology could overcome the blockers to probabilistic design concerning computational expense, required assumptions, availability of data, accuracy of data characterisation and the large amount of required knowledge to implement such approaches. The remaining blockers to probabilistic design approaches therefore concern the development of reliability targets and the engineering mindset change required to implement such approaches.

Date of Award	23 Jun 2020
Original language	English
Awarding Institution	University of Bristol
Supervisor	Julian D Booker (Supervisor) & Jonathan E Cooper (Supervisor)