Does the Pressure or Pressure Coefficients on the roof of a bluff body depend on the inlet values of k and e, when we are doing a turbulent flow analysis using a k-e model ? How sensitive are the results to the inlet boundary conditions, like u, v, p, k and e ? What about the dependence of other turbulence models like k-w. k-e RNG and K-e nonlinear models and LES on the inlet value of k and e ?

Thanks to all of you..

Calighan

(1). The pressure distribution depends on the continuity equation. (2). The viscous effect will change the flow field distribution (continuity) through two mechanisms. (3). One is the displacement effect (wall boundary layer), and the other is the flow separation. (4). So the answer to the question is : for the attached flow, at high Reynolds number, the effect is secondary (displacement effect). (5). But if the flow separation is involved, the effect can be large, such as the problem of separated diffuser flow with non-uniform inlet conditions. (6). So, it depends on whether the flow is attached or separated.

I agree with what John says. Let me add a few things about the models though. Most classical k-epsilon models will be insensitive to the inlet k and epsilon values, even if you have separation (an effect that, as John said, could be affected). The reason for this is that around bluff bodies the production of turbulent energy, k, will probably be so high in the stagnation region that any turbulence in the free-stream will negligeble. However, note that this high production of k predicted by these models is unphysical. RNG is less affected by this problem and thus is more likely to be affected.

With a k-w model you often see that they are too sensitive to the inlet w value. Also these models suffer from the problem with high stagnation region k production. The problem is less pronounced though and in many cases you will not notice it.

A well tuned non-linear model is more likely to have the correct sensitivity to free-stream k and epsilon/w. Same goes for LES, but there you also have a lot of additional problems with how you prescribe your inlet.

Whether or not this sensitivity to k and epsilon/w is important for your results very much depends on your case. It transition important? Is separation important? If so, is it separation from a corner or from a curved wall? Is boundary layer blockage important?

As far as RANS modeling for high-Re (fully turbulent) flows like yours, inlet turbulence values do not seem to much affect the results, especially the mean flow quantities, as long as reasonably realistic values are used, although the sensitivity to inlet turbulence will vary from model to model. At least, this is what "models" do.

Chance is that inlet "velocity" BCs will affect pressure distribution far more than other boundary conditions. For your application (building aerodynamics), you may need to specify a velocity profile consistant with the atomospheric boundary layer your building is immersed in. If you want, you can also specify k and epsilon profiles at inlet boundaries using correlations for typical equilibrium boundary layer flows.

Now when it comes to LES, I think that, at least, the pressure fluctuation (not mean pressure) on the roof ought to depend upon inlet turbulence a lot.

Guys,

The answers above have really impressed me - very good and right to the point. One small addition: if your bluff body is close to the floor (like in car aerodynamics), the crucial effect is the flow split below and above the body. This, in turn, depends on the growth of the boundary layer (in the wind tunnel?), which also depends on inlet turbulence (typically out of equilibrium). If you get the inlet turbulence wrong, the thickness of the boundary layer at the front of the body is wrong, which effects the flow split and may produce silly lift coefficient.

Air-foil of an aircraft generates lift which is already explained in this forum. The top surface of the air-foil is partially convex and the bottom surface is almost flat. Now lets come to analysis of the shape of a sports car. If the bottom surface shape along with bluff body design is optimized then silly lift coefficient can be reduced. An engineer has to be smart enough to design the shape of the car so that drag is minimum (along with enough space to sit comfortably in front and rear seat) at the same time silly lift co-efficient is also minimum (to make the car safe to control and drive at high speed). This is what design optimization is all about.