Detached-eddy Simulation Of Wing-tip Vortex

DES (Detached Eddy Simulation), which was proposed in1997, is ahybrid Reynolds Averaged Navier-Stokes (RANS)/Large Eddy Simulation(LES) model. It is designed to automatically turn to a RANS model inboundary layer and a LES model in massively separation regions. Theaccuracy of DES predictions for massively separation flow is far superiorto that of RANS, at the same time the cost of DES is less than that ofLES.On the other hand, it is considered to be difficult for RANS to predicta wing-tip vortex flow accurately because of the complicated pressuregradient and velocity gradient in all three dimensions.In this paper, the basic idea and mathematical expression is displayedat first. Several problems of DES and different versions of DES designedto correct these problems is then introduced.Before the simulation of a wing-tip vortex, several typical cases ofmassively separation flows are simulated by both DES and RANS, includinga circular cylinder, back-facing step and a NACA0012airfoil in deep stall.Results show that the three dimensional flow structure and the unsteadycharacteristic of massively separation flows can be captured by DES, butnot by RANS, and the quantitative results of DES is also superior to thoseof RANS, especially in NACA0012case.The wing-tip flow of a plate plane is then simulated by both RANS andDES in three grids of different quality. The results show that althoughthe lift/drag coefficient got by different grid is just the same, thevorticity and the pressure in vortex are far more different. On the otherhand, in the downstream field closed to the plane, pressure and vorticitysimulated by DES and RANS are very close, but these vortex characteristicdissipate very quickly alone streamwise in RANS case.The near-field characteristic of wing-tip vortex of NACA0015airfoilin angle of attack (AOA) of8and10degree are simulated by both DES andRANS. The results of these cases are compared to experiments[9][10], whichshow that DES can capture the complicated three-dimensional structure inthe vortex, and quantitatively, the streamwise vorticity as well as the cross-flow velocity agree the experiment results quite well. But theresults got by RANS, which used the same grid as that of DES, failed tocapture the correct structure in vortex, and RANS under-predicts thevorticity in tip vortex by40%. All the results show that DES can simulatethe flow characteristic in tip vortex effectively, where RANS can not.