Two step impact analysis using RESTART
Hello
I am using RADIOSS to simulate a crash test which I have previously done. The trajectory of the impactor before impact is complex and requires several linkages and mechanisms to set up.
The area of interest is the post-impact behaviour (stress/strain) on the target component, as well as the trajectory of the impactor.
When I run the impactor by itself, to see the launch trajectory, I get a reasonable run time of about an hour.
When I run the impactor with a simplified initial velocity + the target with all the contact and failure cards defined, I get a reasonable run of about 3 hours.
When I run the full model: impactor with the initial trajectory + the target with all the contact and failure cards defined, I get a run around 12 hours to reach the impact, and another 12-24 hours to get results from post impact trajectory.
I believe this is partly due the finer mesh required for the target model. I was hoping there is a way I can split the analysis in two, using the results from a the first as an initial condition for the second.
Consider the following scenario:
Model A: Simple model with only what is required to set off the trajectory of the impactor. The target is not modeled. Run from T=0 to T=T_impact.
The run is stopped at the moment where the impact would have occurred.
Model B: The target, with finer mesh and contact and fail definition is introduced, and the analysis is 'RESTART' picking up from where Model A left off.
In other words the target (with finer mesh) only appears just before impact onwards. This should theoretically cut down the majority of the analysis time. Is there a way to set up a RESTART file with introduction of new elements / properties / fail cards, and linking it to a 'Stopped' run?
I hope the diagrams attached illustrates what I'd like to do. Please let me know if there is a way to do this. Thank you.
Best Regards,
MN
<?xml version="1.0" encoding="UTF-8"?>
<?xml version="1.0" encoding="UTF-8"?>
Answers
-
Hi,
Put a rigid body on finer mesh and turn it off just before the impact with a sensor. Rigid bodies can be activated or deactivated with a sensor or with the options /RBODY/ON and /RBODY/OFF in the Engine files.
To activate two rigid bodies with the master node 1 and 2, add the following option in Rootname_0001.rad:
/RBODY/ON
1 2
All elements included in both rigid body are deactivated.
To deactivate one rigid body whose master node ID is 3, add:
/RBODY/OFF
3
All elements are reactivated, strains and stresses are those from previous deactivation time.Contact interfaces can be turned on/off with Tstart and Tstop parameters. Use additional engine files (runname_0002.rad) to change simulation settings during the run.
Refer to example RD-E: 1200 Jumping Bicycle for more details.
0 -
Hello
Thank you for your response. If I understand correctly this is only applicable if the target is a rigid body?
What if the target is a regular element with typical properties (P1_SHELL, M2_PLAS_JOHNS_ZERIL, FAIL_JOHNSON). Would non-rigid body elements be able to turn off and on?
Best Regards
0 -
You have misunderstood- deformable (regular) elements are made non-deformable with a rigid body. Once rigid body is turned off, regular elements will become deformable again.
Refer to attached document for more information.
0 -
Hello,
Ah I see what you mean. I will give it a try. Thank you.
0 -
Hi MN,
In another way you can define this problem setup.
Define the initial time of the trajectory with implicit simulation and before impact make it explicit dynamic.
Hope it will reduce your computation time using implicit card.
Thanking You.
0 -
Hello Pritam,
That sounds like a possible solution. Have you tried this approach before? I have only used RADIOSS for explicit, since we have been using other solvers to do implicit analysis.
Would this approach require you to make the changes manually in the *0001.rad file or can it all be done within hypermesh? Thank you for your suggestion.
MN
0 -
Hi MN,
You can do it by imposing some control cards or by changing the engine file. Although i Have not used in my case, you cay do it.
You can find some help in reference guide regarding Implicit solvers and control cards. See the path in attached image.
Also you can find some example file also.
See this example: RD-E: 1602 IMPLICIT Solver
<?xml version="1.0" encoding="UTF-8"?>
Hope it helps.
Thanking You.
0 -
Thank you Pritam, I will look into this.
0
Altair Employee