Probabilistic Fail-Safe Size Optimization of Aerospace Structures Under Several Sources of Uncertainty

Abstract

his work presents a research on the probabilistic fail-safe size optimization of aerospace structures. The goal is to design minimum weight structures taking into account possible damage scenarios, as well as several sources of uncertainty. The first type of uncertainty refers to the one present in structural parameters, which can be characterized as aleatory, epistemic or hybrid uncertainty. The second type of uncertainty pertains to the ignorance of what partial collapse will occur in an accidental failure event. The last type of uncertainty is related to debris characterization in the event of an engine failure, due to the randomness in the parameters defining the debris, such as the number of impacts or the location and size of holes in the fuselage. Several methodologies have been developed to deal with the first type of uncertainty: fail-safe Reliability-Based Design Optimization (fail-safe RBDO) using the Sequential Optimization and Reliability Assessment method (SORA), fail-safe Evidence-Based Design Optimization (fail-safe EBDO) using the decoupled EBDO approach, and fail-safe Hybrid Reliability-Based Design Optimization (fail-safe HRBDO) using a fast-convergence decoupled strategy that was developed by the author to deal with random and evidence variables simultaneously. Concerning the second type of uncertainty, two methodologies are proposed in this research to address the probability of occurrence of each damage scenario: the Probability-Damage approach for Fail-Safe Design Optimization (PDFSO) and the Reliability-Index based strategy for the Probability-Damage Approach in Fail-Safe Design Optimization ( -PDFSO) where the latter also considers aleatory uncertainty in random structural parameters. Several application examples have been carried out, including a curved stiffened panel of an aircraft fuselage and the rear section of an aircraft fuselage. The last contribution of this research is the development of a framework (DamageCreator) to automatically generate a large enough set of possible damage scenarios from an aircraft mesh, due to an uncontained engine or propeller blade failure event. The debris parameters, such as number of impacts, impact area, spread angles, hole location, debris orientation, size, and velocity, can be considered as random or deterministic. The tool is applied to a cylindrical barrel structure and to a fuselagewing assembly corresponding to a narrow-body aircraft. The programming codes of the proposed methodologies were fully implemented by the author using Matlab and Python environments, as well as Abaqus and Nastran as finite element solvers.