[NewToAnsys]

I thought it would stabilize the solution at the fluid-porous interface, because that's where it begins to oscillate and diverge.
But that was just a thought, I would be grateful for suggestions!

So far I thought of changing the under-relaxation parameters (which one?!), smaller time steps, better mesh quality at the interface.

Thank you in advance for your suggestions!

[Opaque]

How is the flow around the interface?

Normal to the interface, or parallel to the interface?

In CFX, timestep is the "under-relaxation" mechanism. More information is needed to determine if you need under-relaxation or not.

Under-relaxing can solve any divergence problem to the point of not making progress during the iterations--> never converges but it does not diverges.

The motto in a CFD algorithm is to force/enable/enhance/drive towards convergence, not to prevent divergence.

[NewToAnsys]

Flow is normal to the porous domain and flowing into it. It is also a multiphase set up : Fluid A is present in all domains at initialisation, and Fluid B enters through the inlet at a constant mass flow rate. Both fluids are continuous, using VOF, mixture model. I'm using adaptive time steps (min time step is 10^-6 s). Mesh has a max skewness of 0.95.

The discontinuity seems to occur when Fluid B reaches the fluid-porous interface.

Is there an ideal mesh quality requirement for multiphase simulations? This setup for single phase, steady state converges fine.

Thank you!

[ghorrocks]

Regarding mesh quality - The quality requirement is different for different simulations so it is not possible to give general answers. But every simulation will run better with better mesh quality. So time spent improving mesh quality is never wasted, even if the quality was pretty good to start with.

[NewToAnsys]

Can someone explain the effect of this setting? (more than what is mentioned in the documentation)

[Opaque]

Not sure you want to apply a global parameter to solve a localized issue at a domain interface.

However, if you are willing to try another expert parameter, you could also try the following

porous cs linearization option = 1 or 2

Its default value is no-linearization as far as I know, and 1 and 2 are different strategies to improve convergence at porous domain interfaces.

Hope the above helps,

[Opaque]

Not sure you want to apply a global parameter to solve a localized issue at a domain interface.

However, if you are willing to try another expert parameter, you could also try the following

porous cs linearization option = 1 or 2

Its default value is no-linearization as far as I know, and 1 and 2 are different strategies to improve convergence at porous domain interfaces.

Hope the above helps,

[Goenitz]

maarsalan_1@yahoo.com

The attached file has Foam type, Its porosity phi, df fibre diameter, dp pore diameter, Cf inertial coefficient, Kp permeability and more.

My question is that for a CFX input we need
Volume porosity which is phi,
Permeability is given Kp,
Resistance loss coefficient (need to be calculated as inertial loss coefficient is unit-less)
and interracial area density by formula = 6(1-phi)/particle diameter

My questions is:
1. how to calculate resistance loss coefficient (1/m) form inertial loss coefficient (unit-less).
2. The data is enough to calculate interfacial area density? as diameter of pore is given but not of solid converted into sphere..

[Goenitz]

Originally Posted by Opaque

Not sure you want to apply a global parameter to solve a localized issue at a domain interface.

However, if you are willing to try another expert parameter, you could also try the following

porous cs linearization option = 1 or 2

Its default value is no-linearization as far as I know, and 1 and 2 are different strategies to improve convergence at porous domain interfaces.

Hope the above helps,

When I was using ANSYS Fluent, there were problems so I switched to CFX also bcz my advisor knows CFX.

Anyway
1 Try using refine mesh near interface.
2 If there is no turbulence e.g Re is low then don't use any Turbulence model (Shear Stress Transport is good though) in solving scheme as it causes velocity-pressure coupling (idk what that is but my advisor told me).
3. Try structure mesh instead of unstructured. There can be comformality issue.