Regarding usage of /AMS card
Dear All ,
I am performing a UNDEX simulation using the LAW51 material card . In my simulation, I have water, TNT and cylinder structure.
The overall mesh is 25 lakh. The simulation has TYPE 18 contact to estabilish Fluid structure interaction .
When I am solving the simulation the time step is coming 3.8116E-06 which shows the remaining time REMAINING TIME= 12430098.62 s, which is more.
I am interested in introducing the /AMS card to speed up computational time.
I am added /AMS in stater file and /DT/AMS 0.67 3.8116E-5 in engine file .But I am getting negative rho error every time .
If anybody helps me to use /AMS to speed up the computatioal time will be helpful . I also tried /DT/NODA/CST but facing same issue .
I am attaching .rad files from 2023 version for reference .
Thanking you in advance .
Regards ,
SACHIn
Answers
-
Hi sachin,
The /AMS and /DT/NODA/CST commands are not compatible with ALE solutions, so they can not be used to speed up your model.
Instead you can identify the element or node that causes the low time step and try to re-mesh, because element size directly affect the time step. Also, the /EOS can have a huge effect on time-step, so you should double check if anything can change to help your solution (some time lower values of bulk modulus can be used to speed-up the simulation with small accuracy errors).
Finally, you can try to speed-up your method with the use of Multi-Domain Technique. You can find more info on how to set-up and a Multi-Domain Technique in the following KB:
Polyvios
0 -
Dear Polyvios Romanidis ,
Thank you very much for your detailed reply. I understood that the Multi-Domain Technique can be useful for my case.
Regarding /AMS the following radioss highlights it's a limitation for ALE/ Eulerian mesh ?
Regards ,
SACHIN
0 -
sachin awasare said:
Dear Polyvios Romanidis ,
Thank you very much for your detailed reply. I understood that the Multi-Domain Technique can be useful for my case.
Regarding /AMS the following radioss highlights it's a limitation for ALE/ Eulerian mesh ?
Regards ,
SACHIN
Hi Sachin,
In the screenshot you shared inside the brackets you can see the description that AMS can be used for ALE applications but only apply to lagrangian parts.
Furthermore, the model you have shared is an underwater explosion and hence a dynamic situation. AMS is only proposed for quasi-static models, because it changes the mass matrix and hence can not produce accurate results for dynamic phenomena.
So, I would not recommend this method for your model.
Polyvios
0 -
@Polyvios Romanidis Thank you for your feedback. But in case where Eulerian element governs the time step. How to speed the simulation?
Sachin
0 -
sachin awasare said:
@Polyvios Romanidis Thank you for your feedback. But in case where Eulerian element governs the time step. How to speed the simulation?
Sachin
Hi sachin,
If your simulation time step is controlled by Eulerian elements you can not use /AMS or /DT/NODA/CST to speed up your solution. Actually there are not specific methods to implement this speed up.
The thing you can do is to change your model. The time step is related to element size and the speed of sound (based on /EOS). You can try change these parameters and see the effect they have in your simulation.
Generally, the change in these parameters can bring a difference in the results so you have to be careful and check the solution created. I would recommend to first change the element size by making your elements bigger and then reduce the bulk modulus.
Finally, keep in mind that you try to simulate an underwater explosion case so a low time-step is expected.
Polyvios
0