Analysis of HVAB Rotor Hover Performance with Uncertainty in CFD Meshing Strategies

Over the past few decades, advances in computing hardware capabilities have enabled increasingly higher fidelity Computational Fluid Dynamics (CFD) simulations to be performed for a variety of aerospace applications. One area of particular importance to the rotorcraft community is the prediction of hover performance. Although the computational cost of hover simulations is tractable – especially for the steady-state calculations considered in this work – ensuring their accuracy remains a challenge. There are many sources of uncertainty that could affect the results of hover simulation predictions, but this work aims to focus solely on the importance of discretization error due to the use of finite-resolution meshes in steady-state NASA OVERFLOW simulations. Specifically, several common choices for best-practice gridding parameters are considered to be uncertain, and the effects of this uncertainty are quantified using a combination of automated parametric grid generation and surrogate modeling techniques. The Hover Validation and Acoustic Baseline (HVAB) rotor was chosen as the geometry due to its frequent usage in previous AIAA Hover Prediction Workshop studies. Surrogate modeling techniques include the Non-Intrusive Polynomial Chaos, Gaussian Process Regression, and two Deep Neural Network approaches due to their common usage for CFD and other Uncertainty Quantification (UQ) applications. Among the six uncertain gridding parameters considered, the first spacing off the wall and leading edge spacing were found to have the largest impact on the uncertainty in HVAB hover performance. If future studies aim to reduce discretization error in hover simulation predictions, these two parameters should be a primary focus.