Volumetric Locking workaround by leaving nodes free

[mR1ch]

I wanted to ask if this is ok.

I am working with a biphasic model which has a quasi static formulation. In my setup, I have a BC on all the nodes on the outer surface of my model. Trying to run like that does not work very well even if I make permeability of the tissue very large. I was thinking of having a system where I displace only some of the nodes leaving other free. So for example odd numbered nodes move for short period of time before moving the even nodes. I would switch back and forth nudging the nodesets until the load is fully applied.


An example of my approach is shown below. If I keep the time steps very small and switch often is this an ok approach when the model wont run due to locking?
image.png

[maas]

Hi,

What kind of problems are you running into when you apply the BC on all the nodes at once?

Regarding your approach, it's definitely a bit unusual. What are you plotting in the images above? You can definitely see some banding patterns appearing, so this approach might lead to a mesh-dependent solutions. If you pursue this approach, I would definitely recommend doing some verification tests to make sure that you are indeed getting the correct answer.

Best,

Steve

[mR1ch]

I attached a version that moves all the nodes at once. I believe the problem locks.

[maas]

Hi,

I'm not sure if locking is the issue here. There are elements inverting from the very first iteration. The elements that are inverting are extremely thin: they have an aspect ratio of 2000. I recommend remeshing so that the elements near the edge have a better aspect ratio.

Best,

Steve

[mR1ch]

I am not sure how to resolve this issue... On the one hand I need small element sizes to satisfy the biphasic characteristic time constant, on the other febio does not handle these thin elements very well. Do I just need to find a sweet spot in terms of the mesh that has think elements with decent aspect ratios?

Also can you look at aspect ratio for your model through FEbioStudio? The mesh inspector does not have it as an option.

[maas]

I determined the aspect ratio of measuring edge lengths with the Measure tool (2-point distance).

You are right, biphasic analyses often require a sufficiently refined mesh near inlet and outlet boundaries, but there is of course a limit: elements that have a large aspect ratio can cause problems as I suspect is the case here. I don't know how easy it is to mesh this geometry, but if possible, you could consider doing a mesh convergence study, where you start from a relatively course mesh, and then create several iterations of refinements to see what mesh resolution you need.

Best,

Steve