Difficulty finding correct time step
Altair EmployeeI am running a quarter model of an indentation test shown below to take advantage of the symmetry. I used to run the simulation using an imposed displacement, and it ran fine with a time step of 0.1 and took about 30mins to finish. I now want to change from imposed displacement to a concentrated force on the rigid indenter. I have applied my cload on the mnode of the rigid body, but now if I try to use a .1 time step it gives me the mass energy exceeded error. So I have been trying to lower my time step so it wont give me this error, and currently I am using 5.0E-4 as my time step. The issue with this now is that my simulation went from taking 30 minutes to now its saying potentially 2 days. I will attach all the relevant files you may need, but I am wondering if i have something set-up incorrectly, or if anyone has any advice on how to decrease sim time without getting the mass energy exceeded error.
Answers
-
Hi Brandon,
Which unit system are you following in this model?. Is it kg, mm, ms?
The run time specified also seems high. I'm presuming that the unit system is kg, mm, ms and so, the total run time provided is 5000 ms, that is 5 seconds which is very high for explicit analysis. Please recheck the same.
0 -
Yes you are correct about the units. Well my thought process was that the indentation machine takes about 2-3 seconds to apply the load fully so I was trying to replicate that. What would be a recommended range for explicit?
0 -
Hi Brandon,
Normally explicit analysis will range in 100-300 milliseconds, but this depends on the application. For crash analysis or bird strike analysis, it will last for hardly 50 milliseconds.
0 -
Okay, any suggestions on how I should go about this then? My only goal at the moment is to just see the displacement after the load is fully applied.
edit: I've changed the cload so that it is fully applied at 250 milliseconds, and changed the entire sim time from 5000 to 250, and i am using a time step of 2.5 E-5 at the moment. It is now saying it will take approximately 7 hours which is much more manageable. Will update if it works correctly for anyone who needs info in future.
0 -
Hi Brandon,
Please review the results with this run time.
Ensure the mass error is less than 5% (.05), due to the imposed time step, which is always a good practice.
0 -
The simulation has finished successfully, however, the output files show a mass error of around 38%. Will lowering the time step by a magnitude to say 2.5e-5 lower the mass error?
0 -
Hi Brandon,
Yes, lowering the Tmin in /DT/NODA/CST will reduce the mass error. Mass error of 38 percent is very high. Reduce the Tmin value and try running.
0 -
I changed the Tmin to 1.0e-4, and it is now saying it will take approximately 24hours. If this seems normal, i'll let it run and see what the results are. Other than decreasing the time step, do you see anything else I could change about my model that would decrease the simulation time?
I could probably get rid of the outer elements on the cube in order to decrease the amount, but those elements are fairly large so not sure how much that'll help. So, just wondering if you know of anything from previous experience or what not to improve my simulation time.
0 -
Hi Brandon,
Please review the engine out (_0001.out) file while running so that you can review the time step and can find what is causing the time step to drop.
It can be due to an element, a node or an interface.
•For an element, check the related material (especially its Young modulus and density in case of an elastic-plastic material; and its viscosity in case of a visco-elastic material). There must not be an error in the units system that this data is given in. Check the size of the element, since elemental length is proportional to time step.
•For a node, check the characteristics of connected elements. If the node is on the master side or the slave side of an interface, this interface must be verified.
•For an interface, the gap of the interface must be verified if some failure happens on the master or the slave side of the interface.
These are the possible workarounds for time step drop.
0