Tutorial 9: Axisymmetric analysis of biphasic indentation

[SucaBlyat]

Hi,

While working through the tutorial, I have run into a problem with tutorial 9.
I have even redone the entire model, step by step but I cannot seem to find the source fo the error.

Right now the errors are:

"Problem is diverging. Stiffness matrix will now be reformed"
"NAN Detected. "
------- failed to converge at time : 0.0166667
Max. nr of retries reached.

Thereś a number of things that kind of bother me, hope someone can help clarify them for me:
1) When I select elements (for mesh editing), clicking "Invert" does not do a lot (in fact, only seems to add a kind of shadow to the elements?)
2) When welding the two meshes, choosing the given threshold does not really seem to do anything, so to exclude uncertainties I just went with the "Weld" option given during "Attach"-ing the meshes.
3) What is the difference between absolute and relative moving/rotating? Is it with respect to the original and current frame of reference/ coordinate system, respectively?
4) When adding the Rigid Constraint>Prescribed force for the indentor, should I leave prescribed force in z-direction as 0, and then only edit the value in the load curve?
5) In the analysis step, is it correct that the second point is (1 000 , 200) rather then (10 000 , 200)?

Thanks a lot

PS: I have attached my files for reference

[ateshian]

Hi,

I have even redone the entire model, step by step but I cannot seem to find the source fo the error.

You need to add the fixed displacement boundary condition that prevents motion of the bottom surface of the tissue layer. Also, for the indenter load curve, you either specify the value of -0.002 in the load curve itself or you specify that value in the rigid force constraint, but not both places. Whichever has the value -0.002, the other should have the value of 1. FEBio multiplies the prescribed BC's with the load curve value and uses that as the prescribed value.

1) When I select elements (for mesh editing), clicking "Invert" does not do a lot (in fact, only seems to add a kind of shadow to the elements?)

This tutorial does not ask to invert elements. However, if you do find the need to invert an element you can check the result of the inversion using the Mesh Inspector tool: Inverted elements will have their Volume or Jacobian switch sign.

2) When welding the two meshes, choosing the given threshold does not really seem to do anything, so to exclude uncertainties I just went with the "Weld" option given during "Attach"-ing the meshes.

Tha'ts fine. This option did not exist in PreView when the tutorial was originally written but it performs the same function.

3) What is the difference between absolute and relative moving/rotating? Is it with respect to the original and current frame of reference/ coordinate system, respectively?

The absolute position of an object is determined by a rotation matrix about the origin and a vector from the origin to the centroid of the object. Rotations are always performed about the global coordinate system. So specifying absolute rotation/translation means that you directly modify this rotation matrix and translation vector. The "relative" option means that you perform a rotation/translation relative to the current orientation/position of the object. Again, all rotations and translations are performed in the same, global coordinate system.

4) When adding the Rigid Constraint>Prescribed force for the indentor, should I leave prescribed force in z-direction as 0, and then only edit the value in the load curve?

Use 1 instead of 0 as explained above.

5) In the analysis step, is it correct that the second point is (1 000 , 200) rather then (10 000 , 200)?

In this tutorial this model runs to a final time of 1000, so (1000,200) is the correct value for the second point.

Best,

Gerard

[SucaBlyat]

Thanks so much!

1) When I select elements (for mesh editing), clicking "Invert" does not do a lot (in fact, only seems to add a kind of shadow to the elements?)
This tutorial does not ask to invert elements. However, if you do find the need to invert an element you can check the result of the inversion using the Mesh Inspector tool: Inverted elements will have their Volume or Jacobian switch sign.

I was actually referring to 2.4.1 in the tutorial, this was probably not evident. This does not seem to work, is this correct?

The absolute position of an object is determined by a rotation matrix about the origin and a vector from the origin to the centroid of the object. Rotations are always performed about the global coordinate system. So specifying absolute rotation/translation means that you directly modify this rotation matrix and translation vector. The "relative" option means that you perform a rotation/translation relative to the current orientation/position of the object. Again, all rotations and translations are performed in the same, global coordinate system.

It seems that PreView accepts both inputs at the same time (absolute as well as relative), does it not get mixed up then?

In this tutorial this model runs to a final time of 1000, so (1000,200) is the correct value for the second point.

This is not explicitly stated in the tutorial, does this mean that time span = 0-1000, in 10 000 time steps?

Please find attached a screenshot of the final result. I noticed that the elements at the interface of the indenter and the tissue layer run into each other. Is there a solution to this?

Best

[ateshian]

Hi,

I was actually referring to 2.4.1 in the tutorial, this was probably not evident. This does not seem to work, is this correct?

I followed the tutorial instructions just now and the invert selection feature seems to work fine. Please remember to select the elements then go to the Edit menu and select Invert selection.

It seems that PreView accepts both inputs at the same time (absolute as well as relative), does it not get mixed up then?

The code does not get confused since it has a hard-coded order in which to perform these operations when specified simultaneously. I did not write this part of the code but I would assume logically that relative transformations are performed on top of absolute transformations. It should not be difficult to test this assumption by trial and error (which I did...).

This is not explicitly stated in the tutorial, does this mean that time span = 0-1000, in 10 000 time steps?

You are correct that the time step of 0.1 (which is the default in PreView) was not explicitly stated in the tutorial. (Most other default values are also not stated explicitly, but we can update the tutorial in future releases to mention the time step.)

Please find attached a screenshot of the final result. I noticed that the elements at the interface of the indenter and the tissue layer run into each other. Is there a solution to this?

Yes, you can (a) increase the penalty or (b) decrease the tolerance for Lagrangian augmentation, or (c) switch to a gap tolerance and set a tight tolerance. Please keep in mind that some overlap will always exist at a perfectly sharp point, but (d) you can also improve the response by refining the mesh further. All of these options will require a longer solution time.

Best,

Gerard