Multi-Fidelity Aerodynamic Modeling in a Mission-Based Conceptual UAV Design Process

This paper presents and evaluates a multi-fidelity approach for the calculation of lifting surface aerodynamics in a conceptual mission-based design process for civil unmanned aerial vehicles (UAVs). In general, aerodynamic computations significantly influence the final aircraft design. Thus, deviating accuracy of aerodynamic modeling methods may evolve towards substantially different designs and mission performances. The effect aggravates with regard to iterative design loops. Design space exploration in optimization frameworks at conceptual stage requires efficient aerodynamic calculations. Hitherto, the aerodynamic calculations were carried out exclusively using an adapted three-dimensional lifting line theory denoted in the following as the medium-fidelity model. For a faster design space exploration, however, an even simpler aerodynamic model is of interest. Therefore, an analytical aerodynamic model was additionally integrated as the low-fidelity model. Eventually, to benefit from the efficiency and accuracy of both, the analytical model and the lifting line method were fused in a multi-fidelity approach based on hierarchical kriging. All aerodynamic modeling methods are assessed on three levels. First, this paper investigates the effect of accumulating aerodynamic discrepancies on the overall design level. The final configuration is found to be little sensitive to the aerodynamic modeling method applied to the wing. The masses of the configurations deviate within the range of +/- 0.3%. Additionally, deviations between low-, medium-, and multi-fidelity models are evaluated directly on their outputs regarding the aerodynamic coefficients of each final UAV configuration. The multi-fidelity model demonstrates a higher overall accuracy than the low-fidelity method and shows a high level of agreement with the medium-fidelity results. Finally, the study focuses on the simulated aircraft performance level. For this purpose, trimmed configuration polars were generated and mission simulations were carried out. Again, the multi-fidelity method has a higher overall accuracy than the low-fidelity method and demonstrates good agreement with the medium-fidelity results. However, the outcome indicates that adaptive sampling strategies need to be applied in order to enhance the robustness of the multi-fidelity method.     «

This paper presents and evaluates a multi-fidelity approach for the calculation of lifting surface aerodynamics in a conceptual mission-based design process for civil unmanned aerial vehicles (UAVs). In general, aerodynamic computations significantly influence the final aircraft design. Thus, deviating accuracy of aerodynamic modeling methods may evolve towards substantially different designs and mission performances. The effect aggravates with regard to iterative design loops. Design space expl...     »