Numerical Study Of A Transitional Ventilation Flow With Coanda Effect

Coanda effect and transition flow are a kind of flow phenomena that widely exist in nature and industry.Transition flow refers to the process of fluid transition from laminar flow to turbulent flow.At present,there is no mature accurate theory about the mechanism of transition.Coanda effect refers to the tendency of fluid to flow along the protruding solid surface instead of deviating from the original direction of motion.The mechanism of the Coanda effect is that the fluid will deflect to the nearby solid surface due to the entrainment movement of the environmental fluid between the solid surface and the jet,and then attach to it.The forces induced by the Coanda effect can be used to generate lift or other forms of motion.The research on these two phenomena is of great value,and at present,only a few scholars have conducted in-depth research on them.In this paper,six kinds of low-Reynolds k-ε turbulence models and three kinds of large eddy simulation subgrid models are used to simulate the plane jet with the presence of the Coanda effect and transition flow.The flow is affected by the high anisotropy of transition,attachment,separation and other factors.The following conclusions are drawn:(1)The boundary layer predicted by all six kinds of low-Reynolds k-ε turbulence models is thinner than that in the experiment,because the value of ε near the wall is seriously too low when the flow separation occurs or approaches the separation.AB,LB,YS,AKN and CHC models predicted the location of the wall jet separation too far forward,while LS model predicted the location of the wall jet separation too far back.When the Reynolds number is low(Re=1000),LB and YS models can not predict the Coanda effect well,but when the Reynolds number increases to 1750 and 2500,the performance of LB and YS models becomes better.The CHC model is just the opposite.The results of LB model’s prediction of turbulent kinetic energy and dissipation rate are completely unreasonable,which leads to its prediction of wrong velocity distribution.LS model has a good prediction of velocity distribution,because the additional terms D and E in the equations of k and εof LS model can make up for the lack of isotropic Boussinesq hypothesis in some aspects,while AB model is very suitable for predicting the Coanda effect.(2)At x/L=0.07 and 0.27,the boundary layer thickness predicted by all three subgrid models is thinner than the experimental data,while at x/L=0.47,more accurate results can be predicted.DSM model performs best in predicting the time average velocity,and it is also very good at predicting the location of the large eddy center in the recirculation region.DSM model and WALE model are more suitable for the simulation of Coanda effect,while DKE model is not suitable for the simulation of Coanda effect at low Reynolds number.The three subgrid models can only accurately predict the distribution of root mean square velocity in the boundary layer at x/L=0.27 and 0.47.In other cases,it is difficult for the three models to obtain good results consistent with the experimental data.