An Automated Robust Process for Physics Based Aerodynamic Prediction 2000-01-5565
By Combining the Response Surface Methodology with a classical Design of Experiments formulation, a robust method was developed to facilitate the aerodynamic analysis of conceptual designs. These aerodynamic predictions, presented in a parametric form, can then be furnished to a sizing and synthesis code for further evaluation of the concept at the system level. The computational basis of this methodology is a set of numerical codes that work in unison to both optimize the geometry for minimal drag and evaluate key aerodynamic parameters such as lift, friction, wave and induced drag coefficients. Code fidelity and sensitivity to a wide variety of input parameters such as aircraft geometry, panel layout, number of panels used, flow theory used within the numerical code, etc. was investigated. The numerical results were compared with experimental data for different configurations, and the code input parameters required for the best correlation were grouped according to aircraft type. These input parameters could later be used to analyze new designs that fall within an identified aircraft type. The resulting numerical predictions should then have the smallest possible error. An estimate of the error at discrete points was provided by comparing numerical predictions of each group with the experimental data. The error data was later fitted with a continuous Chebyshev polynomial.