Hello all,
I am reaching out to this community to seek help with a ballistic simulation of layered composites. I am working with Sapphire (*MAT002) and PC (*MAT024) and a steel (*MAT015) bullet so there is a large difference in stiffness between the two-layered materials. My issue is that the shear stress at the contact region is orders of magnitudes higher in sapphire compared to polycarbonate as shown in the attached picture. Right now the contact is defined as tied surface to surface but I tried merged nodes as well with no significant change in output. I also checked a finer element size down to 0.05 mm. I am grateful for any help or hint in which direction I should put my focus to.
Thank you all Fabian
Hi @fabian
First off, pretty cool simulation setup! Were you not expecting this higher shear stress in the Sapphire region? Can you elaborate more on why this may be an issue?
Also, is the shear stress increase only happening at the contact region of the Sapphire? If so, this may simply be a bi-product of the contact between the bullet and the Sapphire. What is the contact that you are using and how is the contact implemented? (part to part, part to seg set, seg set to seg set ect.) Also make sure you are using realistic friction values for the contact. If you are worried about the contact and only currently have 3D mesh, you should create a 2D, nodally connected, overlay for both the bullet and Sapphire contact surface and assign it as *Mat_Null. This is a bit of a trick to help LS-Dyna with contacts.
Lastly, how are you retrieving your YZ stress? ELOUT?
Thank you for your comments. The magnitude of the stress is not surprising to me, rather than the fact that the stress does not seem to be transferred into the polycarbonate so it appears there is no traction between the two layers. This issue appears at the entire contact region between sapphire and polycarbonate the magnitude, of course, is different depending on the relative distance to the bullet impact.
I used the tied surface to surface contact definition with segments in this particular analysis. However, I have also merged the contact nodes with no additional BC but the results are pretty much the same as the once shown here. The plot is created in the post-processor history plot of two elements in the contact region.
My question in this case mainly concerns the contact between the 1st and 2nd layer of my composite and not the bullet impact contact definition.
Thank you again for your help and I hope I answered your question
Gotcha. Do you know that the shear stress does in fact transfer to the polycarbonate experimentally? It may just be due to the differences in material stiffnesses.
Nodally connecting the two layers is generally a fine way to model a composite, but you are interested in the interaction between the layers so that may not be your best option. Tied contacts can be finicking as some of them cannot transfer forces / moments across the contact. And since you are interested in the two layers sliding across each other, you may want to try a sliding option. So try this:
*Contact_Automatic_Surface_to_Surface_Tiebreak
and set the Option parameter to 4 which allows sliding along the two surfaces.
You should be able to use seg sets or part ids and make sure that you define some friction coefficients that are reasonable. Let me know if this is helpful.
Thank you for the suggestion I will set up an analysis with this contact definition.
I can not be certain how the transfer of shear stress experimentally will be but I assume it is greater than what we see in this analysis. On this topic, I was thinking about another possibility there is some rotation of the element due to the deflection plate. Is there a way to plot the XY-stress in the local rather than the global coordinate system? I was not able to toggle the local element system in post-processing?
Thank you again for your help
I’m not 100% positive how to export stress in the local coordinate system of the element, but I’m pretty sure it depends on the material model and it needs to be orthotropic. Here’s a previous thread on this topic which may or may not be useful.
https://feassistant.com/forums/ls-dyna/local-coordinate-system-reconstruction-from-global-results/