Learning Forum

[alexandern.zhao]

Hi everyone,

I am currently working on a biomechanical simulation involving a flexible hose that undergoes large rotational motion and significant flexible deformation. My primary objective is to simulate fluid seepage (leakage) through micro-pores distributed along the wall of this hose, while accurately capturing the two-way Fluid-Structure Interaction (FSI) between the internal flow and the deforming solid structure. I am using ANSYS System Coupling to handle the bidirectional coupling (Fluent + Mechanical).

My Initial Approach & The Bottleneck:
To model the porous seepage, I initially extended the fluid domain by adding a thin porous zone adjacent to the inner wall and applied a "Porous Jump" boundary condition at the interface. However, this geometric modification has proven to be highly problematic for the FSI process. Because the hose deforms and rotates, the additional porous region requires complex mesh motion and, more critically, introduces a massive amount of additional mapping/interpolation at the FSI interface. This significantly increases the computational cost and severely deteriorates the convergence stability of the coupling iterations.

The Core Question:
I would like to eliminate the extended domain and the "Porous Jump" condition altogether. Is it possible to implement this seepage effect directly on the original, unchanged wall boundary (i.e., keeping the default no-slip wall topology) using a User-Defined Function (UDF)?

Specifically, I am looking for a UDF that can impose a local mass flux or velocity normal to the wall (simulating the seepage) directly on the wall faces, without changing the boundary type from "Wall" in the Fluent GUI. The boundary should simultaneously:

1. Act as a standard wall for the tangential flow (no-slip).

2. Allow a specified normal outflow (seepage) driven by the local pressure gradient.

3. Maintain the exact same geometric face zones so that the FSI load transfer (pressure and viscous forces to Mechanical) remains straightforward and does not require additional mapping interfaces.

Specific Inquiries:

1. Which UDF macro is most suitable for this purpose? Alternatively, would it be better to apply a source term (DEFINE_SOURCE) to the adjacent wall-adjacent cell layer?

2. If I impose a mass flux via UDF on a standard wall, will the pressure force calculated on that face and transmitted to System Coupling automatically account for the momentum loss due to the outflow, or do I need to manually adjust the force vectors in the UDF?

3. Are there any inherent limitations in System Coupling regarding permeable walls defined via UDF (e.g., does the mesh displacement mapping still function correctly when the wall is permeable)?

Thank you in advance for your time and insights.

Best regards,
Alex

[Rob]

CFD is generally best when the flow gradients are sensible so porous jump or media won't tend to work well for permeable walls. The pressure loss over the PJ is high relative to along the pipe so you'll have numerical issues.

The "easy" fix is to use a sink/source (DEFINE_SOURCE) on the near wall cells. But you then need to pair the wall & wall-shadow facets as they're not linked automatically other than for energy as it's built in. That's likely to interact badly with system coupling as the wall pressure may be a little out of sink with the wider domain: it'll not matter in a fixed domain. The flux/value option in species might work, but once you've paired the wall & shadow facets you'll need a DEFINE_PROFILE UDF. 

I'm not aware of any specific limitations for system coupling, but the extra UDFs might make it somewhat unstable. 

Finally, staff are not permitted to go into too much detail other than what's considered public domain and/or in the documentation so assistance on this will be limited. Your starting point is to get the non-permeable model running. 

[alexandern.zhao]

Thank you for your valuable suggestions. I have already managed to get the non‑permeable model running successfully. However, I must admit that my knowledge of Fluent is still quite limited, and I am not very familiar with the concept of wall‑shadow facets and the associated pairing methods. I will take some time to study this further and explore other possible approaches. Your advice is greatly appreciated.