Computational efficient robustness analysis of aircraft component distortion

Abstract

In this thesis, a methodology is developed that allows for both reliable and computational efficient robustness analysis of aircraft component distortion. This research applies to dis- tortion caused by reductive manufacturing processes that are used to obtain monolithic components from pre-stressed stock material. With the developed methodology, orienta- tions of any component within rolled plate stock material can be found where distortion is most robust. Part distortion is defined as a deviation in shape of an aircraft component from original intent as a result of the component’s reductive manufacturing process. As extreme precision is required in aircraft component assembly, the distortion phenomenon is highly undesired. The developed methodology in this thesis contributes to AIRBUS’ objectives to minimize part distortion related issues. In this thesis, the fundamentals of part distortion are studied. It is found that aircraft com- ponents distort as a result of residual stresses that are present in stock material from which the components are manufactured. As residual stress in rolled plate is subjected to substan- tial variation, part distortion is stochastic in nature. Positions of the components in rolled plate are searched for where distortion is most robust. The robustness of distortion refers to the insensitivity of distortion to uncertainty in residual stress. A mathematical stochastic representation of residual stress in rolled plate is developed showing high coherence with experimental measurement data provided by AIRBUS. For elementary geometries, the rela- tionship between distortion robustness and residual stress is derived analytically. The developed method for predicting distortion robustness is more than one hundred times more efficient in terms of computation cost compared to state-of-the-art methods and al- lows for reliable robustness predictions. In the developed method, three-dimensional po- sitioning of a component in rolled plate can be simulated where state-of-the-art distortion modeling tools usually stick to one dimension. The developed methodology is put to the test in a case study concerning an aircraft stiffener component. The case study emphasizes the significance of robustness predictions; distor- tion dispersion is found to be relatively large compared to the distortion magnitude and significant correlation is found between the component’s orientation in rolled plate and the level of robustness. Positions of components in rolled plate can be found where distortion is extremely robust. Moreover, a relationship is found between the component’s degree of symmetry and the level of robustness.