Combining Computational Fluid Dynamics Analysis Data with Experimentally Measured Data

ABSTRACT

This article presents an approach to reducing the time and cost of experimentation in large wind tunnels, such as the 10 × 10 ft. supersonic wind tunnel at NASA Glenn Research Center, by combining computer simulations of test models from Reynolds averaged Navier-Stokes analysis with small sets of wind tunnel data. To demonstrate the viability of the approach, the impact of microramp flow control on the shock wave boundary layer interaction using paired sets of data from both computational fluid dynamics (CFD) analysis and experimental measurements was compared. By combining the CFD results consisting of 15 central composite face-centered (CCF) simulations with a smaller subset of four/five experimental wind tunnel cases, augmented, interlocking combined data sets were generated from which models were developed that allow the prediction of wind tunnel results. No statistically significant differences were found to exist between the predictions from models generated using the augmented interlocking data sets and the models generated using the complete set of 15 wind tunnel cases based on a paired t-test. From an engineering perspective, the same optimal microramp configuration was obtained using models derived from the combined data set as obtained with the complete set of experimental wind tunnel data.

KEYWORDS:

• CFD
• D-optimal
• data scaling
• DOE
• interlocking DOE
• multiple regression
• scale-up