[ATIKADAR]

Hi, I am simulating laminar water flow through a straight rectangular porous channel. Two side walls are symmetric and the top and bottom are no-slip walls. I know the isotropic porosity and permeability and I am using Forchheimer-Brinkman extended Darcy model (please see the attached figure). I have calculated the isotropic loss coefficient using the attached equation for the superficial velocity formulation. I did the same simulation in Fluent using the same parameters. Surprisingly, in the case of CFX, I am getting much lower pressure compared to the Fluent results. For example, at vel. 0.4 m/s, I got DelP=1023Pa from Fluent whereas CFX gave only 835Pa.

For actual conjugate heat transfer, I need to use CFX. That's why I am trying to calibrate my CFX model. Can someone please tell me why CFX is giving a much lower pressure difference than Fluent? Is there anything wrong with velocity formulation or coefficient calculation?

TIA

[Opaque]

Are you using the porous domain, or fluid domain + subdomain loss source ?

If using porous media, are you using a superficial velocity formulation loss, or true velocity formulation?

[ATIKADAR]

Hi. Thanks for your reply. I am using the porous domain. I have checked both superficial and true velocity using the same loss coefficients. I found true velocity formulation gives high DelP than superficial velocity formulation. For true velocity formulation for porous domain, should I need to make any changes in loss coefficient calculation than my previous calculation (see the attached image in the first post)?

Thanks.

[Opaque]

The loss coefficients are not unique. Say I have the expression

Drag = C * V

which V? That is the where both formulation differentiate. You must have a very clear understanding which V is used in a given formulation, and V is expected to be used from the documentation.

V_superficial = V_true * Porosity

if I recall correctly. From there, you can figure out which one you should use.

One thing to keep in mind, both solutions (superficial and true) must give the exact same results for your case if the input is consistent. The choice is to make it easier to set up, not to change the results.