Gravity currents and wall behavior modeling at high Reynolds numbers

Abstract

Gravity currents are buoyancy-driven flows governed by horizontal density gradients, originating from both natural and anthropogenic sources. They play a critical role in a variety of environmental and geophysical processes, and their interaction with human-made structures can be highly significant. These flows are often studied numerically using advanced techniques such as Large Eddy Simulation (LES), which are capable of capturing the complex physics involved. However, the high computational cost associated with LES makes the study of realistic cases prohibitively expensive. To address this challenge, the present study investigates the use of coarse-grid simulations, both with and without wall-model implementations, to evaluate the potential for reducing computational costs while maintaining reasonable accuracy. Gravity currents were analyzed using the lock-exchange configuration at a Reynolds number of 136,000, based on the bulk velocity and the domain height. The analyses indicate that the coarse-grid cases are able to qualitatively reproduce the main characteristics of the current. In one case, based on a wall modification of the eddy viscosity, the front evolution, during the self-similar phase, exhibits an error of 0.25% relative to a wall-resolved reference case. Generally, cases with an eddy viscosity wall models perform better during the self-similar phase and in representing the head of the current, whereas cases without eddy viscosity modification perform better in capturing the integral quantities of a gravity current. Overall, the use of coarser grids reduces computational costs by approximately two order of magnitude while preserving the main characteristics of the gravity current.