Adding stiffness to a surface

[tsims]

Hi everyone,

I am working on simulating an object that has a thin membrane around an internal compartment, and I was wondering if there would be a way to add stiffness properties or some sort of force to nodes on a surface?
I have tried meshing a surface around the object, but getting it as thin as it needs to be results in very dense meshes that take a long time to simulate, so I was wondering if there might be a way to use the surface of the mesh to implement these parameters?

Thanks!

[ateshian]

Hi,

You can use shell elements to simulate very thin membranes. This would require you to create a surface mesh only, and it doesn't have to be too fine.

Best,

Gerard

[tsims]

Hi Gerard,

Thank you so much for your reply! We implemented this solution and are starting to get results, which has been awesome!
We have ran into an issue though with the surface thickness. We are able to complete a simulation using a surface thickness of 1mm, 1.2mm and 1.25mm successfully, but if we try and set the surface to be 2mm, which is our target parameters, the model will not initialize, and we get the following error:

************************************************** ***********************
* ERROR *
* *
* Negative jacobian detected during domain initialization *
* Domain: tissue *
* *
* *
************************************************** ***********************
************************************************** ***********************
* ERROR *
* *
* Model initialization failed *
* *
************************************************** ***********************

Do you have any ideas what may be causing this and how we can overcome it?

[ateshian]

If the shell elements are attached to solid elements, FEBio assumes that the solid element loses a height equal to the shell thickness. If the solid element is not thick enough, subtracting the shell thickness will create solid elements with negative Jacobians. That's probably what's happening in your model.

Best,

Gerard

[Fluvio]

Gerard,

Is it possible for the shell elements to be 'too thick' then? And if so, what is happening in that case?

Robert and I have been trying to use 'shell elements' as an efficient way of modeling skin. It works for some models, but not all.
In this particular model that Robert mentions, we increased the thickness of the elements from 1 - 2.5 mm. When using a thickness of 2.75 mm, the simulation cannot initialize due to negative jacobians!
Also, at 2.50 mm, we only get through 51% of the sim... which is the furthest we have gotten (with this method)

Given the behavior/interaction between shell elements and the underlying solid tets, would you consider this to be a good method for simulating skin at all? perhaps skin is too thick for this method?

Thank you!

[ateshian]

Hi Fluvio,

Is it possible for the shell elements to be 'too thick' then? And if so, what is happening in that case?

Yes, it is possible for the shell elements to be too thick. In that case it is easier to use the extrude tool on FEBioStudio to extrude penta (from triangles) or hex elements (from quadrilaterals) with the desired finite thickness, instead of using shell elements. This extruded layer then represents the skin in your model. In general the extrusion tool should work well, but if the underlying surface has sharp creases, the extrusion may produce inverted elements. So there is no foolproof system of creating a skin layer of a desired thickness, the underlying geometry determines the feasibility.

Best,

Gerard

[tff]

Hi everyone,

Sorry in advance for my simple questions but I have got confused with the concept of shell elements that I would appreciate if someone can help me with it.
In my model, I have a thin membrane tissue between two other geometries and that membrane tissue is in contact with them based on the defined contacts. I want to use shell elements instead of volumetric elements in my model for the mentioned tissue. However, I have several questions about the shell elements as listed below:

- Previously, I had modeled this tissue with a separated geometry meshed with tetrahedrons, and I was using contact definitions to connect this tissue to its adjacent surfaces. For using the shell elements, would it be fine if I use the surface mesh of the same tissue's geometry to create the shell element from it (for example the 3D geometry of the tissue as STL file)? or should I use the contact interfaces of adjacent bony geometries for defining the surface of the shell element and then increase the thickness of the shell element to fill the gap in between the two bony geometries?

- What are the general steps that one should follow in order to create shell elements with a specific thickness in FEBio?

- When I specify a thickness to a shell element, would the shell geometry be visible, and can one actually see the thickness of the generated geometry (similar to the result of the extrude tool)?

- Can these shell elements be used to study the stress/strain in the membrane tissue? I understood that using shell element has a big advance in reducing the computational time, what about the limitations of using such elements in the FE model?

Best

[ateshian]

Hi,

would it be fine if I use the surface mesh of the same tissue's geometry to create the shell element from it

Yes, that's fine. There is a tool in Edit Mesh called Create Shells from Faces that can do what you want conveniently. It lets you specify the shell thickness.

When I specify a thickness to a shell element, would the shell geometry be visible, and can one actually see the thickness of the generated geometry (similar to the result of the extrude tool)?

No, shell elements are represented as surfaces and FEBioStudio does not show you an extruded version of those elements.

Can these shell elements be used to study the stress/strain in the membrane tissue?

Yes, but to display strains in shells make sure to select the plot variables "shell strain" and (optionally) "shell relative volume".

I understood that using shell element has a big advance in reducing the computational time, what about the limitations of using such elements in the FE model?

There is no gain in computational efficiency when using shells. The two primary reasons for using shells are: (1) It may be easier to represent thin structures as surfaces, especially when segmenting image data to create a mesh. (2) A standalone mesh (not attached to a solid) is so thin (e.g., sheet metal) that a finite element analysis using solid elements would produce mesh locking (i.e., the material may behave as though it is much stiffer than in reality), in which case one should use EAS shells (see the User's Manual). EAS shells are actually computationally more expensive than solid elements, but they represent the only way to avoid mesh locking.

Best,

Gerard