Hello,

I would like to know what is importance the importance of a right or false predicted transition point concerning lift and drag of an airfoil. In case of heat transfer at wall it is clear that the heat fluxes strongly depend on the position of the transition point. But how is strong is the influence on airfoil flow ?

Tim

Hi Tim, The transition point inside a boundary layer is crucial for the overall performance of an aerofoil under extreme situations such as landing or take-off configurations (i.e. large incidence angles). If you have a turbulent boundary layer, you know you can reach higher angles of attack without stalling than when employing a laminar aerofoil. Hence, under take-off conditions, you would prefer a turbulent b/l if it wasn't because of the drag induced due to the turbulent flow. Therefore, you look for a compromise between a laminar and turbulent b/l flow . There is also evidence that a transitional flow aerofoil can withstand stronger shock waves than a fully TURBULENT aerofoil provided the position of the shock is located a short distance dowstream of the transition point, that is, where the turbulent boundary layer has just becomed "mature". There is a lot of experiments which have also highlighted the influence in trailing edge separation, i.e. a turbulent profile withstand stronger adverse pressure grad. and hence, is such cases, you want a turbulent b/l in that region and not in regions where laminar flow can maintain and drag not be as large. Hope it helps. Javier

Javier,

thanks for your explanation. Regarding CFD I have the problem to predict the transition point correctly. I would guess that my boundary layer will be turbulent from the leading edge if I use the standard k-epsilon turbulence model due to high TKE production at the stagnation point. So what I'm not sure of is what error this deficiency induces in predicting drag and lift. Is it as high as I would be interested in heat transfer ?

Tim

You can simply get rid of the over-production at the leading edge by introducing a couple of lines in your code. See "On the k-3 Stagnation Point Anomaly" P.A.Durbin, Int.J.Heat and Fluid Flow, 17:89-90, 1995. It is a physically correct limiter that avoids the over-production of TKE at the stagnation point. I've used myself and makes a considerable difference. After that, if you want to pursue a good modelling, I suggest you go for a point-transition criterion, i.e. the production of TKE is limited only in regions downstream of the previously specified transition point. It certainly makes a difference in skin friction predictions. If you want to go further, you can attempt for intermittency modelling and, if you want to go for the max. try and couple CFD with stability techniques... Regards, Javier

(1). Assuming that you are solving only the turbulent boundary layer using the low Reynolds number version of two-equation k-epsilon model, then I think, it is possible to show that one can compute the laminar flow region as well as the turbulent flow region. (2). I don't know whether the results are right or wrong. But it is a research area. (3). When you extend the model outside the boundary layer to cover the whole flow field, then it has been shown that high eddy viscosity region will appear in the pressure side, in the normally inviscid region. Whether this is correct or not, I don't know. But based on the commercial codes I used in the past, and the current commercial code I am using, the eddy viscosity field distribution is not consistent with the conventional viscous + inviscid region definition. That is much higher eddy viscosity region is observed in the normally inviscid region. (4). The consequence of this is the excessive diffusion of the boundary layer and the total pressure loss. When I was doing the low Re modeling back in 70's and 80's, I did not have this problem. So, this is related to the proper handling of the model. (5). Now you see that there are already two problems, one is non-realistic inviscid flow field, and the other is the capturing of the transition in the boundary through the low Re model. (6). I would say that the displacement effect due to the transition of boundary layer on the pressure distribution will be small. (7). But since the viscous drag is normally an order of magnitude smaller, and is a direct function of the wall shear stress, the drag predicted with/ or without transition will be quite different, not to mention the effect due to the improper handling of the inviscid region linked directly to the use of k-epsilon model. (8). In addition to these problems, I also have found that excessive loss can also be produced by the use of numerical algorithms in the code. By selecting different numerical method options, you get different total pressure losses. (9). The heat transfer will be even more sensitive than the skin friction because it is a local quantity. In most cases, one is not interested in the overall heat loss, but he is more concern about the local temperature. In this case, it has been shown that the local heat transfer can be off several times, if the turbulence model is not calibrated for that particular flow. (10). So, the conclusion I have is: as long as the boundary layer remains attached, the transition and the boundary layer displacement effect on the pressure lift and drag would be small. On the other hand, when the local temperature or the heat transfer is of interest, it is still very hard to predict the accurate values without fine tuning of the turbulence model. (a factor of two in the drag or total pressure loss prediction is fairly common in the use of CFD codes.)