General Mechanical

General Mechanical

Using initial stress after first loadstep

    • Jurjen
      Subscriber

      I am trying to simulate the stress according to a method from literature.
      However, I am struggling with transferring stresses in between load steps. Since I am also resetting the deformation after each step, this normally also resets the stress. Therefore I am using APDL INISTATE command to store and restore stresses (see attachment). Unfortunately, I am not allowed to store stress in the INISTATE parameter after the first load step limiting my simulation to two steps. Therefore my question was if there is another way to store and restore stresses in the geometry? Or is there a way for using the INISTATE even after the first load step?


      I also tried using restarting analysis but without success.


      The method should basically do these steps:



      1. Geometry is imported with no stress and/or displacement. The inlet is fixed in both longitudinal and rotational direction (to prevent rigid body rotation and translation).

      2. Set number of iterations and end-pressure From the set parameters: the pressure (working on the inside of the vessel) in each step can be calculated, since this pressure is building up from zero to your pressure of interest with an S-curve (according to the number of iterations of choice).

      3. Simulation of loadstep 1

      4. After each step the displacements has to be reset to zero (thus resetting the geometry to the initial geometry) and calculated stresses (at step N) have to be kept as initial stress for the next step (n+1). Therewith estimating the stress in the measured geometry. Furthermore the pressure for the N+1 step is incremented according to the pressure curve in 2.

      5. After performing all steps, the geometry is reset another time. Ending up with the stress calculated in the last step and the original geometry.





       

    • peteroznewman
      Subscriber

      Can you share the literature that describes this method?  I don't understand the benefit of following this process. It seems to create a non-physical result. If you solve for the stress normally, that is the real stress and there is a deformed shape. You can make plots or probe the stress on the original geometry by setting the Result display scale to Undeformed.

    • Jurjen
      Subscriber

      Here you can find the paper: https://www.ncbi.nlm.nih.gov/pubmed/16822515 


      Common for a specific case, estimating stresses in geometries from medical images.

    • peteroznewman
      Subscriber

      Thanks, now I understand.



      "the method can predict an unloaded configuration if the loaded geometry and the load applied are known"



      I am not a skilled APDL programmer, but someone who is wrote the code in the attached file here that executes multiple restarts in a loop to get to the full load, and uses ekill to delete elmements from the model and keep going. Maybe you can use this code as a framework to implement your INISTATE into the loop and take out the ekill.

    • Jurjen
      Subscriber

      I tried to rewrite the code according to the framework. Unfortunatelly I still receive the error:


       *** ERROR ***                           CP =      34.195   TIME= 14:19:48
       INISTATE data cannot be edited after the first load step


       


      Is it somehow possible to disable and or force the program to continue? Or are there other ways to store stresses?

    • sk_cheah
      Subscriber

      Perhaps this example from SimuTech provides the framework you are after.

      Kind regards,
      Jason

    • Jurjen
      Subscriber

      Thanks, sk_cheah, I continued combining both your import and exporting stresses as well as with the loop provided by peteroznewman.


      Unfortunately it still does not fully function. My best guess is that it involves the resetting of the displacements. Since both the looping and restoring stress seem to work every step (I actually don't exactly know how to check this?).


      At the moment I am resetting my displacements with looping over all nodes like this:

      *DO,ii,1,maxnodenr
         nx = UX(ii)
         ny = UY(ii)
         nz = UZ(ii)
         D,ii,UX,-nx
         D,ii,UY,-ny
         D,ii,UZ,-nz
      *ENDDO

      Although with this method I do not observe any residual displacement higher than 1e-6 m thus neglectable. displacements The results of the whole procedure end up in a checkerboard pattern. I am still uncertain why, but I would like to exclude the effect of the displacements.
      See movie here: https://drive.google.com/open?id=1wpavRCUBBik7qn0AgE6nW48idQgsvEOu


      Therefore my question, is there a more elegant APDL to reset displacements or use the original geometry for this calculation?


      Thanks in advance and sorry for the question!


       

    • pedrocarneiro
      Subscriber
      Do you successfully implement your code? Can you provide an example of how to implement the stress after first load step as initial stress in the second and at the same time delete the displacements.n
Viewing 7 reply threads
  • You must be logged in to reply to this topic.