Dear All,

Currently I'm involved in simulating the flow through several grids using CFX. Since I had some troubles in the original (complex) problem, I started to research the effects of a (simple) grid placed in a two-dimensional duct with constant cross section. One boundary a inlet, opposite boundary mass flow boundary, halfway inbetween the grid covering the complete cross section.

I thought of modelling the grid using the porous medium option in CFX. But a porous medium must be modelled with at least three cells width for interpolation reasons. For simulating a grid I then need to refine the local mesh too much to be attractive (or even possible??).

And since I'm not interested in simulating the flow IN the grid (porous medium) but on the effect of it on the SURROUNDING flow, I thought of modelling the resistence of the grid using the fortran routine USRBF to define some bodyforces. I now defined the grid as a USER3D patch and skipped the definition of porosity and just set the resistence parameters. This reduced the width of the modelled grid from three cells to one cell, which comes closer (but not close enough) to reality.

From the CFX manual I found: B = Bf - (Rc + Rf|v|)v For me Bf = 0.

From the 'Chemical Engineer's Handbook' I found: Rc = 0 Rf = 3.25E+4 for a 50% porous grid.

So for a grid of 1 mm depth and an inlet velocity of 1.11 m/s I would expect a pressure drop of: dP = Rf*v*v*dx = 3.25E+4*1.11*1.11*0.001 = ~40 Pa.

Since the modelled grid has a width of 1 cell, I need to rescale the body forces in USRBF. Duct length 1.5 m No. cells in length direction: 200 Actual width of grid: 0.001m So the scale factor would be: 0.001/(1.5/200) = 0.1333

In contrast to my expectations CFX returns (after convergence) with a pressure drop of ~0.006 Pa.

Question: Can someone tell me what's wrong in my models, or in my assumptions???

Thanks in advance, Hans.

Dear Hans: I am modeling roughly the same problem as you do -a flow through the thin porous screen. The physics behind this flow is simple. On the one hand screen is effectively contraction of the flow area. Thus, by mass conservation, the velocity inside the screen must be is higher than in the region before the screen. On the other hand, screen can be thought of as a distributed wall causing resistance to the flow. When porosity of the flow is low (as in your case) the increase in velocity due to flow contraction is counterreacted by the friction resistance. In the model you described, you modeled friction resistance but you did not model flow contration. In my point of view, you DO HAVE to model the flow contraction, i.e., refine the grid towards the screen and place at least 4 control volumes along the screen width. That is what I do.

Regards, Andrey.