Eventos Anais de eventos
MECSOL 2022
8th International Symposium on Solid Mechanics
Aircraft Structural Reliability/Risk Estimate with Limited Data Using Possibility Theory
Submission Author:
Mateus Cunha Cavecci Zequi de Oliveira , SP
Co-Authors:
Mateus Cunha Cavecci Zequi de Oliveira, Moacyr Machado Cardoso Júnior, Mariano Arbelo
Presenter: Mateus Cunha Cavecci Zequi de Oliveira
doi://10.26678/ABCM.MECSOL2022.MSL22-0052
Abstract
One of the most important field of aircraft structural reliability is related to fatigue failures. Although several analysis, evaluations and predictions are made during the design phase, only Full-Scale tests or fleet occurrences can bring to light some failures modes on specific details. Test data (and, in some cases, also fleet data) can usually be a limited set, due to the inherent complexity, costs and time that a complete Full-Scale test demands (or due to a single isolated fleet leader occurrence). This works proposes an alternative approach to estimate structural reliability when limited data is available. While there is a known (possible) range, but not a properly known distribution for parameters as material properties, geometrical variation due to manufacturing tolerances, loading variation, etc, traditional analysis typical proceed by picking one value inside this range. It may be the mean value, or, in some cases, the most conservative one. As predictions and conclusions made by a single observation point may carry a high amount of uncertainty, it will be useful to somehow quantify the uncertainty related to the assumed parameters, by exploring the results throughout its range. Probabilistic calculation depends on well characterized distributions for input parameters, which typically is not often available, so, alternatively, the Possibility Theory was used to explore the possible range of parameters that are usually estimated. Possibility Theory is a special branch of Evidence Theory that deals with bodies of evidence. It is devoted to the handling of incomplete information. R/RStudio software was used and, more specifically, the HYRISK library to perform the required calculation. Independent Random Sampling (IRS) algorithm was used to perform uncertainty propagation. A typical scenario from literature was modelled and evaluated considering the proposed approach using possibility distribution for some specific variables. Uncertainty propagation was performed and besides typical mean results, lower and upper bounds were determined (Possibility/Necessity curves), as well as the quantification of modeled uncertainty effect. The example application consists of the analysis of a trunnion collar failure from a Main Landing Gear during a Full-Scale Fatigue Test, and risk prediction for fleet. The results are assessed and compared with typical calculation of reference literature. The proposed approach helps to better quantify and understand uncertainty in the prediction of aircraft structural reliability/risk when only limited data is available. This gives a better and most realistic picture of reliability/risk ranges and each factor contributions, allowing for more appropriated and sound decisions.
Keywords
Risk, Structural Reliability, Aircraft structures, fatigue, Probability of failure, Possibility Theory, uncertainty
