Creating a 2D shell element and Extruding the surface.

[Aarash]

Hi All

I'm modeling a segment of a residual limb confined in a rigid cast. To do that I'm trying to extract the outer surface of the limb as a separate part(Cast) and assign rigid properties to it. my problems are:

#1 when I extract the surface and assign shell thickness to the elements, there are negative Jacobians in the part that I'm not able to fix by inverting the elements or by refining the mesh. what does a negative Jacobean mean in a 2D shell element? how do i fix it?

#2 I would like to extrude the shell elements to make a 3D part, but I haven't been able to use this feature in Preview. how do i go about this?

I've attached a Preview file of the mentioned shell elements. I appreciate it if anyone can help me out with this.

Thanks in advance

Arash

[maas]

Hi Arash,

Usually, negative jacobians for shells usually means the thickness is too large. However, reducing the thickness of this model does not eliminate all negative jacobians, so I'll need to take a closer look.

It also seems that the extrude feature is not working for shells. I'll try to work on that.

As a work-around (depending on what exactly you want to do), you might be able to just use a rigid contact interface. Just create a rigid interface and assign the outer surface of the limb to the rigid interface. That way you won't need any shells or create a separate rigid part. Would that work?

Cheers,

Steve

[Aarash]

Hi Steve

Thanks for you reply. I will try your suggestion with rigid interface and see how that works. But, ideally I want to have sliding contact between the skin and the cast to investigate the effects of friction between the surfaces and/or to introduce a cushioning layer in between.

Thanks again, I let you know here if I made any progress.

Arash

[Aarash]

Hi Steve

I finally found out where those Shell elements negative Jacobian stem from. Its due to any changes that I make to the element's geometry(e.g scaling up/down) or orientation (e.g rotating about the X axis).

however, even with no negative Jacobians, I still cant extrude the elements and make a 3D part.

I didn't have any luck with creating sliding contact with Shell elements either. It is possible to use surface element for sliding contacts, isn't it?

Cheers

Arash

[maas]

Hi Arash,

What problem did you run into when creating sliding contact with shells? Can you attach the FEBio input file?

Thanks,

Steve

[Aarash]

Hi steve

No matter what parameters I choose for the sliding contact, I still get penetration of skin into the Cast (Rigid part). Although the penetrations are very small (~1 e-5 mm) and can be neglected in most cases; I haven't been able to run a full analysis with large/prolong prescribed displacements.

Attached is an example of the model I am working on. I have tried with alternative master and slave surfaces and a range of values for Penalty ,Augmentation, Gap and search tolerances.

Thanks in advance for your help.

Arash

[Aarash]

Hi Steve

I was wondering if you had a chance to look at my model? I am still struggling to get the sliding contact to work with applied force .

I really appreciate any suggestion you may have for me.

Cheers
Arash

[maas]

Hi Arash,

Sorry for the delayed response. I took a look at the model and I suggest making a simplification. I'm not sure why you have both a force applied to the cast as well as contact with the spring to induce effectively a prescribed displacement. I recommend applying the displacement directly to the cast and eliminating the contact with the spring. If you want to run the model to a certain reaction force on the cast you can simply run it displacement driven and make sure that you get enough displacement to reach the desired reaction force. This is what I did in the attached model. Please take a look and let me know if you have any questions.

Cheers,

Steve

[Aarash]

Hi Steve

Thanks for getting back to me. Because I was was trying to run a force driven model with a single step, I introduced the spring to bring the two surfaces in contact so I can apply a load on the Cast (which is a rigid body).
I am going to try the displacement driven model. If I was to calculate the amount of displacement required to reach a certain reaction force; that would vary for every model, right?!

Can I also please ask you to have look at the sliding contact on this attached model. The model wont converge with relatively high Loads(~2200 N) or displacements(~57 mm).

p.s. I couldn't import your FeBio file to Preview. It gave an fatal error on the very last line of the file, which is an empty line!!!

Many thanks
Arash

[maas]

Hi Arash,

Indeed, the amount of displacement would vary from model to model, but you would run the model until you reach the desired reaction force and then terminate it.

In the upcoming FEBio 2.3 release, we have introduced an alternative mechanism for running force driven contact problems: you would set up a two-step problem. In the first step you apply a prescribed displacement until contact is established, and then in the second step you switch to force control. Once this new version is released you could try that approach. In this case you won't need to know the exact displacement and you will always hit your target load.

Which version of PreView are you using? I am able to import it in PreView 1.16.3. If you are using the latest version on the web (1.16.2) then I guess you'll need the newer version which we'll try to upload shortly. I do hope you were able to run that model in FEBio2?

Regarding the last model you attached (180215) I think the normals on the shell are in the wrong direction. You will need to invert the elements. You can do this in PreView from the MeshPanel: select the shell elements and click the Invert button. Then export the model again and try again.

Let me know if I can help with anything else.

Best,

Steve

[Aarash]

Hi Steve

Thanks for clarification. In my case, I am trying to simulate the reaction forces of a gait cycle which means, the forces vary with time. As a result, I cant just aim to reach a maximum reaction force and terminate the run at that point.

I was using the 1.15 version of Preview, but now I upgraded to 1.16.2 and managed to import your Febio file. Yes, I ran your file in FeBio 2.1 and managed to see the changes you made by reading the code.

Because I wasn't able to complete a run due to contact errors, I experimented with the sell surface normals, and in my case I didn't find changing the normals to improve the convergence.

Cheers,

Arash

[hipop1901]

Hi Aarash,

May I ask what software did you use to model this segment of limb?

Bo