Combined Feature-Driven Richardson-Based Adaptive Mesh Refinement for Unsteady Vortical Flows

摘要

An adaptive mesh refinement strategy that couples feature detection with local error estimation is presented. The strategy first selects vortical regions for refinement using feature detection, and then terminates refinement when an acceptable error level has been reached. The feature detection scheme uses a local normalization of the Q-criterion, which allows it to properly identify regions of swirling flow without requiring case-specific tuning. The error estimator relies upon a Richardson extrapolation-like procedure to compute local solution error by comparing the solution on different grid levels. Validation of the proposed approach is carried out using a theoretical advecting vortex and two practical cases, namely, tip vortices from a NACA 0015 wing and the wake structure of a quarter-scale V22 rotor in hover.