Dissertation/Thesis Abstract

Survival analysis of aging aircraft
by Benavides, Samuel, Ph.D., The George Washington University, 2010, 199; 3397234
Abstract (Summary)

This study pushes systems engineering of aging aircraft beyond the boundaries of empirical and deterministic modeling by making a sharp break with the traditional laboratory-derived corrosion prediction algorithms that have shrouded real-world failures of aircraft structure. At the heart of this problem is the aeronautical industry's inability to be forthcoming in an accurate model that predicts corrosion failures in aircraft in spite of advances in corrosion algorithms or improvements in simulation and modeling. The struggle to develop accurate corrosion probabilistic models stems from a multitude of real-world interacting variables that synergistically influence corrosion in convoluted and complex ways. This dissertation, in essence, offers a statistical framework for the analysis of structural airframe corrosion failure by utilizing real-world data while considering the effects of interacting corrosion variables.

This study injects realism into corrosion failures of aging aircraft systems by accomplishing four major goals related to the conceptual and methodological framework of corrosion modeling. First, this work connects corrosion modeling from the traditional, laboratory derived algorithms to corrosion failures in actual operating aircraft. This work augments physics-based modeling by examining the many confounding and interacting variables, such as environmental, geographical and operational, that impact failure of airframe structure. Examined through the lens of censored failure data from aircraft flying in a maritime environment, this study enhances the understanding between the triad of the theoretical, laboratory and real-world corrosion. Secondly, this study explores the importation and successful application of an advanced biomedical statistical tool—survival analysis—to model censored corrosion failure data. This well-grounded statistical methodology is inverted from a methodology that analyzes survival to one that examines failures.

Third, this work demonstrates the development of a probabilistic corrosion failure model using survival analysis methods and techniques. Using a parsimonious approach, the coefficients of a Cox proportional hazards model were derived from a set of environmental, geographical and operational predictor variables. To determine if the variables satisfied the proportional hazard assumption, numerous statistical tests were performed—such as the equivalence tests of the log rank, Wilcoxon, Peto-Peto and Fleming-Harrington—and graphical plots generated such as observed-versus-expected plots and log(-log) survival curves.

Finally, in a paradigm enhancement to current design methodologies, this dissertation place sets survival analysis modeling in the context of an emerging holistic structural integrity philosophy. While traditional aircraft design and life prediction methodologies consider only the cyclic fatigue domain without consideration to the environmental or unique operating spectrum that aircraft may fly in, a holistic approach considers the cradle-to-grave driving forces in the life of a component, such as corrosion assisted crack nucleation in a material. This dissertation, which uses real-world failure data obtained from structural aircraft components, is poised to narrow the cradle-to-grave loop and provide holistic feedback in the understanding of aircraft structural system failures.

Indexing (document details)
Advisor: Mazzuchi, Thomas A.
Commitee: Allario, Frank, Hoeppner, David W., Murphree, Edward L., Sarkani, Shahram
School: The George Washington University
Department: Systems Engineering
School Location: United States -- District of Columbia
Source: DAI-B 71/04, Dissertation Abstracts International
Subjects: Aerospace engineering, Systems science, Materials science
Keywords: Aging aircraft, Corrosion modeling, Structural integrity, Survival analysis
Publication Number: 3397234
ISBN: 9781109689457
Copyright © 2019 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy