code_saturne User's Forum

Hello,
 
We would like to model a perforated plate, and I see that CS 2.0-rc1 has a 'head losses' option in the GUI.
My main concern is that the perforated plate in my model would be positioned only a few centimeters from the tank inlet, and the air would enter the tank parallel to the plate it has to pass (see attached .jpg).
 
Do you think it is reasonable to use a head loss coefficient to account for the plate?
If so, would you have other recommendations regarding mesh quality and solver settings?
 
We are interested in having a better idea of the pressure distribution downstream the plate. Ideally, it should be even, but we suspect some variation at high flow rate.
We can measure the pressure at the wall, in what's labeled 'area of interest' on the sketch.
 
Thanks in advance,

Jean-Baptiste

Hello,

I am not a big fan of using head loss zones too close to areas of interest, but several studies that I am aware of or participated in tested/compared models with meshing of drilled plates (or zooms on parts thereof) and using head losses, and results were pretty consistent. In all cases, though, flow was predominantly normal to the plate, and not tangential as in your case.

If the plate is thin (i.e. represented by very few cell thicknesses), in any case you should use the "improved pressure interpolation in stratified flow" option (under physical properties, gravtity/hydrostatic pressure in the GUI, or IPHYDR in usini1.f90).

Also note that when using head losses, no specific modelization of the effects of the obstacle causing the head loss on turbulence is done (as a drilled plate or a mixing device could have similar head loss coefficients but very different effects on turbulence, it would require a database of sorts).

I believe using head losses could be a reasonable "initial approach" to your question, as meshing would be simpler (and you could use a coarser mesh, with faster runs), so you could do parametric tests on both inlet velocity and head loss coefficient for comparative data, but it would be safer to do at least one or two flow rate configurations with the real geometry to confirm the head loss approach's validity.

Best regards,

  Yvan

Hello Yvan,
Thanks for the input. Insightful, as usual!
Best regards,
Jean-Baptiste