Rigid Part Contact ...
 
Notifications
Clear all

Rigid Part Contact Not Behaving as Expected  

 

pjay
 pjay
(@pjay)
Junior Member
Joined: 1 year ago
Posts: 26
July 8, 2020 4:08 pm  

Hi Everyone,

I have a simple simulation of a *MAT_RIGID shell part that is contacting a rigid plate (*RIGID_WALL_PLANER) under the influence of gravity, essentially a drop test. The simulation performs as expected until the contact occurs at which point the shell part rebounds off of the plate at a very high speed and acceleration. If I change the material to *MAT_ELASTIC this behaviour is not seen and the contact continues as expected however run times increase dramatically. This simulation is a subset of a larger simulation where run times will be very long if I use *MAT_ELASTIC. Does anyone have any ideas why this might be happening?

In order to obtain the contact between a rigid part and a *RIGID_WALL I have to set RWPNAL = 1.0 in the *CONTROL_CONTACT card but everything else was left as the default. My only other thought on how to proceed is to create a physical plate with *MAT_RIGID and hope I don't see this behaviour again but I have a strong suspicion that I will.

Cheers!


Quote
Topic Tags
Negative Volume
(@negativevolume)
Senior Member Admin
Joined: 2 years ago
Posts: 326
July 8, 2020 5:21 pm  

Hi @pjay,

Direct contacts between rigid parts should be avoided in LS-DYNA. You say that the simulation takes much longer with *Mat_Elastic for the impacting plate. This will be highly dependent on things like material coefficients, mesh size, and ELFORM. Is it possible to remedy the plate with larger elements, use ELFORM=2, and define the *Mat_elastic material coefficients as steel, or even stiffer if it needs to be for you simulation?


ReplyQuote
pjay
 pjay
(@pjay)
Junior Member
Joined: 1 year ago
Posts: 26
July 8, 2020 5:33 pm  

@negativevolume Good to know! If I was not clear I meant changing the falling debris to *MAT_ELASTIC and keeping the plate as *RIGID_WALL_PLANAR. The reason I am using *RIGID_WALL_PLANAR is to contain SPH particles and the falling debris in my simulation which is the only way I know how to keep parts from falling indefinitely, is there another way / better way to go about doing this?

I am currently using ELFORM=2 for the shell section of the falling debris with a relatively coarse mesh compared to the rest of the simulation but I will investigate making the falling debris even coarser. As for material properties for the falling debris, they have to be quite specific as I am modelling some physical experiments (HDPE material).


ReplyQuote
Negative Volume
(@negativevolume)
Senior Member Admin
Joined: 2 years ago
Posts: 326
September 26, 2020 7:23 pm  

@pjay Did you ever figure this out? I am going back and looking for any topics that I let slip by so I wanted to check in. 


ReplyQuote
pjay
 pjay
(@pjay)
Junior Member
Joined: 1 year ago
Posts: 26
September 27, 2020 4:28 pm  

@negativevolume, I never did figure out how to fix this issue other than changing the part to *MAT_ELASTIC. The only other options that I looked into were reducing the time step significantly, which helped somewhat (less rebound but still unrealistic) and using DEFORMABLE_TO_RIGID_AUTOMATIC to change the material to elastic before impacting the shell.


ReplyQuote
Negative Volume
(@negativevolume)
Senior Member Admin
Joined: 2 years ago
Posts: 326
September 27, 2020 4:40 pm  

@pjay Yeah I was just thinking that your long simulation times were probably a product of small debris shell elements, and to try mass scaling to increase your timestep by using *Control_timestep and inputting your new target timestep in DT2MS. 


ReplyQuote
pjay
 pjay
(@pjay)
Junior Member
Joined: 1 year ago
Posts: 26
September 27, 2020 5:46 pm  

@negativevolume, Yes the long simulation times seem to be a product of a long simulation time (i.e. 5.15s) and a small time step (~1e-6s) from the shell elements. I will look into mass scaling but I am not to sure exactly how it works. I know that the contact time step between the RIGIDWALL and debris shell elements is already pretty small (~1e-5s) so I am not sure how much I can increase the time step.


ReplyQuote
Share: