Large deformation on Shell Elements ?

82 views
Skip to first unread message

SoMdt

unread,
Dec 10, 2024, 11:26:15 AM12/10/24
to ProjectChrono
Hello,

I am implementing a simple test of a cylinder with fixed extremities under pressure (length of 42 mm, radius of 2.5 mm and thickness of 0.6 mm). Material has a Young's modulus of 0.6 MPa and nu=0.45.
I tested ShellBST, ShellReissner and ShellANCF4234 in chrono and compared the results with Abaqus.

With a pressure of 0.001 MPa, the results are the same as Abaqus. However, when increasing the pressure to 0.03 MPa, the deformation of the cylinder is roughly twice lower in chrono than in Abaqus (results are in the same magnitude for all 3 shell elements).

To completed the analysis, in Abaqus, when setting the NLGEOM parameter to NO, the deformation goes down to the range of chrono deformation.

My question then : are there shelle elements able to consider large deformations ? Did I forget a hidden parameter somewhere ?

Thanks a lot for you help,
Solenne

Mike Taylor

unread,
Dec 12, 2024, 5:51:13 PM12/12/24
to ProjectChrono
Solenne,

I'm most familiar with the ANCF elements.  The ANCF formulation supports both geometric and material non-linearities.  Currently most of the ANCF elements in Chrono are formulated with only a linear viscoelastic material, but they do inherently support moderately large deformations (i.e. large angle bending, but not plastic like flows.)

Here are a few thoughts on things to try:

If you've only tried to run a static analysis, you might want to try running a dynamic analysis with a slowly applied load with a relatively large amount of damping to dynamically settle to the "static" solution?  

If you are running a static analysis, are you running the linear static or nonlinear static analysis version?  If you are using the nonlinear static analysis, what argument are you giving that function?

If you haven't already examined refining your mesh (using more elements), you might want to give that a try as well.

Finally, the ANCF Shell 3423 element has a relatively low order (linear) interpolation function.  You could also try an element with a richer/higher order interpolation function like the "ChElementShellANCF_3833"element.

Best Regards,

Mike Taylor

SoMdt

unread,
Jan 8, 2025, 4:31:55 AM1/8/25
to ProjectChrono

Dear Mike,

 

Thanks you very much for your answer, here are our latest advancement on the topic.

To answer in the order :

  • I tried both non linear static and dynamic tests and obtained the same results
  • I specifically launched a simulation with a very fine mesh + very small timestep, but this did not improve the results.
  • Eventually, we implemented ANCF_3833 elements, but the result is the same as 3423 and Reissner.
  • I also tested using SetStiff(true) for the pressure load but the results are the same.

 

I am thinking we might have missed something in the implementation.

 

To complete my message, here are all the results for the maximum displacement in mm, both in Abaqus and in Chrono for two different values of pressure (p=0.001MPa -> small deformation, p=0.03MPa -> large deformation):

Abaqus                                       p=0.001 : d= 1.51e-2               p=0.03 : d= 7.81e-1
Chrono - Shell Reissner          p=0.001 : d= 1.46e-2               p=0.03 : d= 4.18e-1
Chrono - ShelANCF_3423      p=0.001 : d= 144e-2                p=0.03 : d= 3.73e-1
Chrono - ShellANCF_3833     p=0.001 : d= 1.42e-2               p=0.03 : d= 3.73e-1

 

I join you all the simulation files (cpp + mesh + nodesets for boundary conditions). If you have time to look into it, I would be immensely grateful: we are currently completely blocked in our project due to this issue.

Best regards,

Solenne

nset_left_Cylinder_quad4.inc
nset_right_Cylinder_quad8.inc
nset_left_Cylinder_quad8.inc
nset_right_Cylinder_quad4.inc
Cylinder.cpp
cylinderQuad.vtk
cylinderTri.vtk

Mike Taylor

unread,
Jan 8, 2025, 8:58:46 PM1/8/25
to ProjectChrono
Solenne,

Best case, it will likely be a couple of days before I can dive deeper into the files you sent.

Looking at your latest message, I was wondering what kind of strains you are seeing in each of these models with ABAQUS. Is your enclosed volume changing significantly, similar to inflating a balloon with the large pressure and not significantly with the lower pressure?  If so, that would mean that the surface area of the elements significantly changed with the second pressure, which might narrow down where I need to start looking.

Could you share a few images from ABAQUS of the undeformed vs. deformed shape (true scale, not exaggerated) for each of the two pressures for additional visual context.  Maximum principal strain contours from ABAQUS might be helpful as well.

Best Regards,
Mike

SoMdt

unread,
Jan 9, 2025, 3:02:10 AM1/9/25
to ProjectChrono
Mike,

Here are some snapshots from Abaqus. In the second case, the surface area indeed changes significantly like an inflating baloon. I also added an image overlaying Abaqus simulation result (in green) and Chrono simulation result (in white) on the large deformation case.

Moreover, I tried calling element->SetDimensions() before each iteration to update the element's dimensions but it did not change anything.
Thanks a lot for the time you can make to dive into this.
Solenne


Cylinder_SmallDefs_MaxPrincStrain.pngCylinder_SmallDefs_Disp.png
Cylinder_largeDefs_MaxPrincStrain.pngCylinder_SmallDefs_Disp.pngChronoVSAbauqs.png

SoMdt

unread,
Jan 15, 2025, 3:53:49 PM1/15/25
to ProjectChrono
Hello Mike,
I was wondering if you had the chance to look into our issue since last week ?
Thanks a lot,
All the best,
Solenne
Reply all
Reply to author
Forward
0 new messages