A Preliminary Study Of Large Eddy Simulation Of Subaqueous Gravity Currents

As an important method of modeling turbulent flows, Large Eddy Simulations(LES) can obtain more informations than Reynold Averaged methods and need lesscomputational sources than Direct Numerical Simulation. In this study, some previouslaboratory experiments of lock-exchange gravity currents have been simulated byLES. It is discussed the differences between2D,3D LES and RANS modeling,2DLES improved with the vortex dissipation model, kinetic energy spectra and coherentstructures of gravity currents, the influence of heterogeneity of grids and theapplication of three subgrid-scale models have been discussed. The followingconclusions have been reached:(1)3D LES may obtain more accurately the largeeddies of gravity flows than the2D LES and the corresponding energy spectra satisfythe-5/3power law in the inertial sub-range;(2)3D LES well reproduced the coherentstructures from two different turbulent mechanisms of gravity currents, the inviscidfree-shear and viscous wall boundary layer, including Kelvin-Helmholtz billows, lowand high speed streaks and horizontal rollers etc.;(3)the energy spectra of3D LESand3D RANS are very close in the range of Reynold averaged scales and separated inthe range of fluctuation scales and the-5/3power law is not observed in the spectra ofRANS method, which have been proved by theories and experiments;(4) the impactof heterogeneity grids on the3D LES is very small and the3D LES may be appliedon heterogeneity grids for gravity currents;(5) the results of three subgrid-scalemodels, namely standard Smagorinsky model, dynamic Smagorinsky model andLiu-Meneveau-Katz similarity subgrid-scale model, have been compared. It is foundthat dynamic Smagorinsky model functions best and similarity subgrid-scale modelworst. But the hydrid model of latter two obtains close results to those by dynamicSmagorinsky model with less computational sources. These may provide importantreference for subsequent LES of gravity currents with sediments in complexgeometries...