Wall model based on neural networks for LES of turbulent flows over periodic hills

Abstract

A data-driven wall model for turbulent flows over periodic hills is developed using a feedforward neural network and wall-resolved large-eddy simulation data. The developed wall model employs wall-normal distance, near-wall velocities, and pressure gradients as input features, and the wall shear stresses as output labels, respectively. In the $a$ $p\phantom{\rule{0}{0ex}}r\phantom{\rule{0}{0ex}}i\phantom{\rule{0}{0ex}}o\phantom{\rule{0}{0ex}}r\phantom{\rule{0}{0ex}}i$ test, the accuracy of the trained wall model is examined using periodic hill cases at different Reynolds numbers and with different hill geometries. In the $a$ $p\phantom{\rule{0}{0ex}}o\phantom{\rule{0}{0ex}}s\phantom{\rule{0}{0ex}}t\phantom{\rule{0}{0ex}}e\phantom{\rule{0}{0ex}}r\phantom{\rule{0}{0ex}}i\phantom{\rule{0}{0ex}}o\phantom{\rule{0}{0ex}}r\phantom{\rule{0}{0ex}}i$ test, the trained wall model is applied to the flow over periodic hills and turbulent channel flows.