LS Dyna

LS Dyna

Coupled structural-thermal modeling of Taylor impacts using Eulerian mechanics

    • magliarj
      Subscriber

      I am simulating a Taylor impact for AA6061-T6 plugs using Eulerian, Lagrangian and ALE models. I have noticed inconsistent findings for the thermal contours between the Eulerian (ELFORM=12) predictions and the Lagrangian (ELFORM=1) and Eulerian (ELFORM=5) predictions. The former exhibits peak temperatures at the periphery of the mushroom while the latter predict peak temperatures of similar magnitude towards the center of the specimen at the impact surface. Additionally, we noted that the von Mises stresses are moderately comparable between modeling approaches and the plastic strains/deformed profiles are near-identical. 

       

      I have attempted to change the thermal solver from nonlinear to linear and implemented various thermal solvers other than the default but none of these combinations had a significant impact on the findings. Modifying the thermal timestep and inputs in the *CONTROL_ALE command also had no impact on the results. A brief presentation is attached to this post, which illustrates the above points, along with the current input deck for our Eulerian simulation. The analyses are performed using a double precision, SMP version of R10.1. Based on our current attempts I am unsure of there is an issue/shortcoming with our modeling approach or if this is potentially associated with the solver. Any suggestions or insights would be greatly appreciated.

    • Aniket
      Ansys Employee
      Ansys staff can not download images/attachments on the student portal, so if you want to reach a larger audience to get answers from, please insert the images inline.n-AniketnHow to access Ansys help linksnGuidelines for Posting on Ansys Learning Forumn
    • magliarj
      Subscriber
    • ido
      Ansys Employee
      Hello magliarj,nIf you zoom in closely you will see that LAG and ALE5 (ALE with ELFORM=5) has sharper gradients then Eulerian (ALE12 or ALE11). This may be due to 2 factors: (a) the mesh is finer for the first 2 cases ==> more elms to resolve/capture sharper gradients. ALE12 has less elms to resolve the same region so the gradient may be diffused; (b) ALE12 resolve the mat interface within half the ALE elm width due to volume fraction representation of the ALE mat. The LAG and ALE5 cases has precise mat interface. To compare better, you will need much finer meshes for all 3 cases until they converge.nHope this helps,nIan Do, PhDn
    • magliarj
      Subscriber
      Hello Dr Do,nnThank you for your reply, in response to your suggestions:nn(a) An identical mesh was implemented for the plug in all 3 cases. For the Eulerian case a comparable airmesh was added as an extension. I created an additional 'fine' mesh for the Eulerian case where the number of elements was increased by a factor of 10. The results are shown in this reply.n(b) The temperature-time history for the center of the plug was plotted for the coarse (original) and fine (revised) Eulerian simulations and compared to the Lagrangian results. The Eulerian results generally converge with each other but do not match the Lagrangian results. Additionally, note that peak temperature at the end of the impact, shown in the attached image, was in a consistent location for the Eulerian simulations but with a peak value of 760 K for the fine mesh (in comparison to 550 K in the original, coarse mesh).nnThank you again for your reply, any further suggestions would be greatly appreciated. If you need further information please let me know.nnnnn
Viewing 4 reply threads
  • You must be logged in to reply to this topic.