[fluid23]

I need some intelligent insight into a problem I face.

Consider a wall bounded (duct) flow that has a porous medium somewhere in the duct modeled using a momentum sink. I know the turbulence intensity and length scale upstream of the media. Inside of the porous region, the turbulence transport equations are not solved since the porous structure is going to invalidate the turbulence model (SST in this case, but that is irrelevant). The normal procedure is to specify turbulence parameters to the porous region outflow... which is great if you have that data available.

In the absence of that data, assumptions are necessary. I think it is reasonable to assume that turbulence length scales would be limited to to the average pore size of the media. However, I am not clear on the conservative approach to turbulence intensity.

If the goal is to investigate downstream separation, is assuming no reduction in turbulence intensity conservative, non-conservative? Is it silly to even think in these terms... i.e. wrong is wrong?

[FMDenaro]

I think that an issue is the fact that at the level of the characteristic lenght scale of the pore there is no turbulence but the flow is laminar. You should consider the ration to the Taylor microscale to see that.

[fluid23]

Interesting, I had neglected to think about that. I will have to investigate a little further. My pore structure if fairly coarse, but it certainly warrants investigation.

What about outflow intensity though? Regardless of whether its laminar inside the medium or not, I still have to specify outflow parameters in the software. Is assuming upstream intensity conservative or a bad idea? Just trying to understand the ramifications of that assumption.

[FMDenaro]

Originally Posted by fluid23

Interesting, I had neglected to think about that. I will have to investigate a little further. My pore structure if fairly coarse, but it certainly warrants investigation.

What about outflow intensity though? Regardless of whether its laminar inside the medium or not, I still have to specify outflow parameters in the software. Is assuming upstream intensity conservative or a bad idea? Just trying to understand the ramifications of that assumption.

I have no direct experience in porous media, maybe the assumption of a flow rate equal to the inflow is enough appropriate but I suggest ot check further

[LuckyTran]

If we are talking about pore sizes much smaller than turbulence length scales, then it shares a lot in common with flow through a perforated plate (with many small bores). The small pores/perforations wipes out most of the history of the turbulence upstream. The flow doesn't re-laminarize, but it is like tripping a tubulent boundary layer and causing it to restart. The porous media / perforated plate needs to be very thick/long in the streamwise direction for the flow to fully re-laminarize. In either case, I agree that the turbulent length scales should be adjusted and limited to a fraction of the pore size.

The LIH porous media approach used in most CFD transports the turbulence through the porous media. Some approaches force the turbulent viscosity to 0 in the porous region, but still choose to transport the the k & epsilon/omega thru the porous region. Both of these are similar to your conserving upstream intensity approach. The ramification here is that the porous media doesn't have any interaction with the turbulence and hardly any with the flow (except for adding a bit extra pressure drop). Although I understand the reason for doing things this way, there are clearly major shortcomings to this approach. It phenomenologically does not agree with our practical observations.