Robust design through identification of main components from multivariate random loads

ifasd2024 Tracking Number 44

Important progresses in terms of methods and computer technique made possible to numerically analyse the dynamic response of the aircraft and loads with high level of accuracy. Also, modern testing techniques can capture highly dynamic pressure fluctuations and accelerations for full wetter surfaces. All of this leads to important amounts of data that, in case of turbulence or buffeting analyses, produce complex and fluctuating loading conditions that need to be considered in structural design adequately. Aircraft structural design traditionally relies on a down-selection of critical loading conditions, using adequate monitoring stations in order to derive nodal load cases. This process is especially complex for stochastic phenomena such as buffeting, where unsteady patterns need to be robustly captured with a large number of monitoring stations. Consequently, many unneeded similar load cases are derived for airframe sizing due to the high correlation of some results. This problem is addressed in the current paper. Accepting that one cannot predict the unsteady patterns and that these need to be captured, a finer monitoring grid is applied. But instead of deriving directly nodal load cases associated to each monitoring station, a principal component analysis (also called proper orthogonal decomposition) is applied to the results either at full aircraft or at component level, extracting the main loading components to be applied for airframe sizing. Therefore, the accuracy provided through the huge input data is kept and the number of sizing cases is drastically reduced for structural design, still covering the most critical load cases. This method is applied to a full aircraft configuration, using high accuracy buffeting loading and response data. The main loading directions are identified and structural sizing cases are derived. A comparison in terms of quality, robustness and efficiency is presented in order to underline the advantage of applying this method for structural design loads compared to the traditional process. To conclude, measurement and analysis of very complex multivariate random loads has become possible, and the presented method allows to adequately take benefit of this unprecedented level of quality for structural design.