[arash93]

Hi,
I am working on multi-phase flow in porous media using ANSYS-CFX. In this kind of problems, porous domain might change in thickness because of external pressure. My main problem is I can not model this king of compaction in my transient simulation.

1- Do I need an external solver for this problem?
2- Compaction can be modeled as a function of pressure and other parameters. Is it possible to get a real-time observation of this parameters and use this compaction models to update domain thickness during run?

Does any body have Ideas about this issue?
Thanks

[ghorrocks]

First of all - do not post the same post many times. I have deleted all the duplicate posts. Do not waste my time like that again please.

Can you explain how external pressure changes the thickness? Why can't this be modelled with a CEL expression and moving mesh? That would be easiest way to do it.

[arash93]

Sorry for my mistake.

I am working on simulation of Vacuum Assisted Resin Transfer Molding.
In this simulation Resin impregnates the porous media.
This injection starts when vacuum applies to mold and cause fluid to move forward in porous media.
Because of flexible vacuum bag on top of the mold, atmospheric pressure put force on it and
cause LOCAL compaction to porous media and LOCAL change in domain thickness.

P_compaction=P_atm-P_Resin %%(P_Resin change LOCALLY)


in pictures you can see:
a- fibrous porous media
b- Resin volume fraction(my simulation of flow propagation in porous media without changing in thickness)
c- Compaction of porous media due to atmospheric pressure




As I said before, there are models that predict thickness change using P_compaction.
But it's a transient process and thickness is changing locally during flow propagation, specially in flow front position.
In this situation, monitoring P_compaction during simulation for thickness prediction seems to be inevitable.

I have used CEL just for Boundary and initial condition as a function of time.

Now the question is:
1-how can I identify thickness of mold as a variable and update it during run?
2- Is it accurate to monitor P_compaction during process?

Thanks for your help

[ghorrocks]

Are you sure compaction is the correct mechanism? Compaction implies compression, but I bet the resin and fibres are incompressible. What is actually happening is the air is being pushed out and the resin is flowing somewhere else. Would not a pressure boundary on the top face be more realistic?

[arash93]

I'm sure that compaction is the right mechanism. Resin is in compressible but fiber bed in some models is assumed as non linear spring in thickness direction. The most important parameter in flow propagation in Darcy's law is permeability and it's highly in relation with fiber volume fraction of preform. Fiber volume fraction changes by thickness variation. All of these parameters are relatively functions of P_compaction.
Because of this effect many works are focused of compaction modeling and considering it's effect on permeability and flow propagation behavior. In addition, thickness variation cause local injection between elements. In fact changing in thickness between neighbor elements cause a local network of flow propagation witch is calculated by mass conservation equation. (it is just because of transient element volume variation)
I think if it would be possible to monitor pressure in some nodes using monitor point, thickness and permeability could be determined by mathematical interpolation models. But next step seems more difficult. Geometry or mesh thickness should be updated at current time step. It's quite a problem!!!
Another solution is coupling problem with an external structural solver to determine thickness. I have done some FSI simulations on Wing and blood vessels with Ansys-Fluent and Ansys-CFX, but it's different. I don't think in could be simulated as FSI problem.

[ghorrocks]

Can you model it as a free surface? Then you could apply the atmospheric pressure to the air phase, and the resin phase would flow and the interface would move. The thickness will be automatically calculated. For the fibres you might be able to implement this as a source term (rather than just the default porous model) so you can make the interface track the resin surface.

[arash93]

I have used free surface option. It works and verify experimental data perfectly.
I am not familiar with source term. Could you please explain more or introduce me a reference to follow the topic.
Thank you

[ghorrocks]

Source terms are explained in the CFX documentation.