Biphasic Contact & Mesh Convergence

[wajihaahmad]

Hello all members.
I am reaching out to kindly request your assistance with

Question 1:
Does the tied biphasic contact allow the fluid to automatically flow between the corresponding two biphasic bodies, or do we have to apply fluid pressure = 0 (free draining) on nodes of both contacting surfaces?

Question 2:
Since FEBio Studio utilizes NETGEN for meshing STEP files, how can one perform a Mesh Convergence Analysis on CAD geometry in FEBio Studio?

Dear Dr. maas and Dr. ateshian , I have shared my file via email, and I would be most grateful if you could kindly provide your valuable suggestions and feedback.
I’m feeling quite anxious that there may be an issue with our model. I know this is a significant request, and I completely understand if it's not feasible, but could you possibly review the attached model and check if there’s anything I may have overlooked or done incorrectly? I sincerely appreciate your time and help with this, and I apologize for any inconvenience this may cause.

Best,
Wajiha.

[ateshian]

Hi Wajiha,

Question 1:
Does the tied biphasic contact allow the fluid to automatically flow between the corresponding two biphasic bodies, or do we have to apply fluid pressure = 0 (free draining) on nodes of both contacting surfaces?

Yes, the tied-biphasic interface automatically enforces continuity of normal fluid flow across the interface. Do not set the fluid pressure = 0 on that interface, as that would defeat the enforcement of this continuity.

Question 2:
Since FEBio Studio utilizes NETGEN for meshing STEP files, how can one perform a Mesh Convergence Analysis on CAD geometry in FEBio Studio?

Netgen is a very powerful meshing software and we haven't fully figured out how to best integrate all its capabilitiest into FEBioStudio yet, so the current implementation of Netgen meshing is sub-par compared to what the software can do. For example, Netgen has the ability to perform a "uniform" mesh refinement, which means that the mesh is refined by different amounts based on local mesh refinement of the original mesh. That would be a standard way to perform mesh convergence analyses with Netgen. We are still in the process of properly implementing the full capabilities of Netgen but we are somewhat hampered by the rapid changes occurring in associated libraries and we are typically cautious with linking FEBioStudio with the latest libraries of each software package on which FEBio Studio depends. But this is an ongoing process and we hope to keep moving forward with this task.

Dear Dr. maas and Dr. ateshian , I have shared my file via email, and I would be most grateful if you could kindly provide your valuable suggestions and feedback.
I’m feeling quite anxious that there may be an issue with our model. I know this is a significant request, and I completely understand if it's not feasible, but could you possibly review the attached model and check if there’s anything I may have overlooked or done incorrectly? I sincerely appreciate your time and help with this, and I apologize for any inconvenience this may cause.

Please be patient with us, as this is the start of the semester for those of us who work in university environments and things become very hectic across the board for all our developers around this time of year. We will try to address each request in a timely manner.

Best,

Gerard

[wajihaahmad]

Hi Professor Gerard,
Thank you so much for your detailed response.

Thank you for your understanding and patience.
I completely understand the demands of the start of the semester in university environments, and I appreciate your efforts despite the hectic schedule.

Best,
Wajiha

[ateshian]

Hi Wajiha,

I took a look at the model you emailed and, based on my initial review I would recommend the following: When performing a biphasic analysis you should be cognizant of the fact that the early time response of the material (immediately upon loading) is nearly isochoric, because the interstitial fluid does not have enough time to escape high pressure regions, and the fluid and solid components are modeled as intrinsically incompressible. Since TET4 elements are known to exhibit mesh locking for isochoric responses, it is better to use other element types (e.g., TET10, HEX8 or HEX20). For the irregular geometry that you are using, this may become somewhat challenging, but it may be important, especially if you focus on the early time response to loading.

You have a number of "prescribed fluid pressure" boundary conditions on internal surfaces of your mesh. This may be challenging to explain physically: How does one prescribe a fluid pressure at the interior of a domain? You may need to rethink whether these are the correct boundary conditions you want to apply.

As a general rule, when you tie two dissimilar biphasic domains across an interface (such as the interfaces between the SSCTs and median nerve, or SSCTs and flexor tendons), you need to create biased meshes on both sides of the interface to properly capture any fluid flow and pressure gradients that develop across those interfaces. Currently your model has a coarse mesh which makes it unlikely to capture such gradients.

These are my remarks based on my initial review of the model you shared with me via email.

Best,

Gerard

[wajihaahmad]

Hi Dr. Gerard,
I am truly grateful for your thoughtful review and the time you have dedicated to assisting me.

[wajihaahmad]

Hi Dr.Gerard.
I modified the model according to your kind review.
While running the model, I encountered an error: Fatal error in factorization of stiffness matrix. Aborting run.
Could I get some assistance with it? Thanks alot

Grateful!

[ateshian]

Hi Wajiha,

Your problem now has nearly 5.5 million degrees of freedom in it (5490249 equations), so you're probably running out of memory, that's why you are getting that error.

I would suggest rethinking how to create your model and asking yourself if it is absolutely necessary to include so many elements in the Y-direction (e.g., you can try to simplify your model by making it more of a plane strain analysis in 2D, with only one layer of elements along Y). You may also need to consider using a different method for generating your mesh (e.g., using quadrilateral elements in the plane of the cross-section, then extruding that cross-section into a single layer of elements along Y). This will help reduce the number of elements in your model, and also deal with boundary layer meshing more easily. Please remember that FEBioStudio is not designed as a CAD software nor as a powerful meshing software, so you may have to turn to other options for creating a quadrilateral mesh in 2D, e.g., using Coreform Cubit or other similar software. Then you can use FEBioStudio to create the boundary layer mesh needed for your analysis, e.g., using one of several options available in the software as outlined in this tutorial.

Best,

Gerard

[ateshian]

Hi Wajiha,
See the two attached documents.
Best.
Gerard
Wajiha.pdf
Circle-Square-BigCircle-Facebinder.step

[wajihaahmad]

Hi Dr.Gerard.
I am grateful beyond words for your time and efforts.
Thank you so much.
I will follow the same procedure for my analysis.

Thank you so much.

[wajihaahmad]

Hello once again, Dr. Gerard. I hope this message finds you well.
I have followed the approach you suggested, but unfortunately, I encountered an issue at the final step during the merge process. The error message I received states: "Cannot merge selection."
Moreover, if I prescribe Contacts like "Tied Biphasic", and "Tied Elastic" between the bodies, do still i have to merge them before?

I would be grateful for any further guidance or suggestions you might have to resolve this issue.
Thank you for your time and assistance. I look forward to your advice


Best,
Wajiha

[maas]

Hi Wajiha,

I tried Dr. Ateshian's approach and was able to repeat the procedure successfully. I suspect you must have missed a step. One thing I think is important to clarify is that in the step that asks you to use the "Detach elements" tool, make sure you use the tool from the main toolbar (right next to the Merge Objects tool). Don't use the "Detach Elements" tool from the Mesh panel. The difference between these two tools is that the former detaches elements into a separate object, whereas the latter simply detaches the elements within the same object. After that step, where the document refers to "part" it should actually refer to "object".

Regarding your question, if you use tied interfaces, you indeed don't have to merge the objects. However, since the meshes are conforming, I recommend merging the objects and avoid the need for a tied interface.

Best,

Steve

[wajihaahmad]

Hi Dr. Maas.
I am truly grateful to you for your time and efforts.

Thank you so much for clarifications. I really appreciate your recommendations.

One question I want to ask that:
Q: What's the difference between the Effective Fluid Pressure and the Fluid Pressure.

* I searched in the Forum and came to this: https://forums.febio.org/forum/febio...fluid-pressure. But unfortunately I didn't understand it for Biphasic Analysis
* I read on website that: In a biphasic material however, since c=0 in p~=p−RθΦc , the effective and actual fluid pressures are the same, p=p~, but in my output they have different values in

I am sincerely grateful to you!
Thank you.

Best,
Wajiha

[ateshian]

Hi Wajiha,

In a biphasic analysis the "effective fluid pressure" is the nodal fluid pressure, whereas the "fluid pressure" is the integration point pressure. If the mesh is coarse you will see significant differences between these values, but if the mesh is refined those differences should decrease with decreasing mesh size.

Best,

Gerard

[wajihaahmad]

Hi Dr. Gerard,

Thank you so much for your time.
Thanks a bunch, I get it.

Best,
Wajiha.