Uncertainty Quantification Approach for Rotorcraft Simulations

The goal of this work is to quantify the uncertainty and sensitivity of freestream velocity and wind direction on wing download and rotor thrust predictions for the Joint Vertical Experiment (JVX) tiltrotor configuration in hover. Even light winds can have a significant impact on hover performance. Accordingly, an accurate representation of hover performance with uncertainties due to variability in atmospheric wind conditions needs to be understood. To support this effort, mid-fidelity simulations with a Reduced Order Aerodynamic Model in CREATETM-AV Helios is used to generate training and testing data for constructing the surrogate models. Uncertainty propagation is facilitated using a surrogate-based approach which integrates stochastic expansions based non-intrusive polynomial chaos method in the Dakotaenvironment. The first test case considers wind velocity and directions treated as epistemic uncertain variables. Post uncertainty analysis, parameter sensitivities are established using Sobol indices to rank the relative contribution of input parameters to the total uncertainty in download and thrust. Sensitivity analysis showed that the interaction of wind velocity and direction has the largest influence on download predictions. The second case includes collective as an uncertain input and additionally carries out a sensitivity analysis. Computed Sobol indices identified collective as the major contributor of uncertainty. Ultimately, uncertainty quantification procedure laid out in this work can facilitate informed design decisions based on quantifiable data that is formed using validated computational approaches integrated with established data science principles with statistical metrics.