Learning Forum

[harolis.susanade]

Hi everyone,

I'm trying to tune a simulation in LS-DYNA to match experimental data as closely as possible (green curve in the image). The case I'm working on is the impact of a solid wooden ball against a solid metal cylinder. I have both the force vs. time curve from the experiment and the ball’s initial and rebound velocities. In the experiment, the ball hits at around 18 m/s and rebounds at about 9 m/s, which gives a coefficient of restitution (COR) of approximately 0.45–0.47.

In the simulation, however, the rebound velocity is almost equal to the incoming one, around -17 m/s, so the COR is practically 1. This clearly doesn’t reflect the real behavior. Also, while the force peak looks somewhat similar, the unloading phase is much faster and cleaner in the model, which suggests the energy loss isn’t being captured correctly.

I’ve tried tweaking the SFSA and SFSB parameters in the automatic contact definition (CONTACT_AUTOMATIC_SURFACE_TO_SURFACE). That helped a bit, but even with low values like 0.05, I can’t get close to the ~9 m/s rebound velocity I’m aiming for. I also experimented with DC, VC, and VDC, hoping to introduce damping without adding actual friction, but they didn’t seem to have much effect. I tried modifying SLSFAC under CONTROL_CONTACT too, but the results barely changed.

My professor suggests damping is key to reproducing the energy loss seen in the experiment, but I haven’t found an effective way to implement it in LS-DYNA. I’m still new to the software, so I might be missing something obvious.

Has anyone worked on a similar case or has tips on how to properly introduce damping to lower the rebound velocity without messing up the rest of the model? I’d really appreciate any insights.

I've attached images of the experimental vs simulation force curve and the rigid body X-velocity graph.

Thanks in advance for any help!

 

[Erik Kostson]

 

 

 

 

Hi

Good to check all contact related param.

Some ideas – perhaps other members have more.

Think also about how energy can be reduced in the system (via deformations and plasticity possibly).

Not sure if it is good idea to model any part as rigid (you have a rigid velocity plot) since you want to capture energy loss due to deformations which is important.

Also do not use linear materials – use nonlinear plasticity for steel (harder to know for wood, but mat143 can be used).

A last point to think, the impact happens earlier in the simulation so something to think about perhaps.

The velocity in the meas. must be tricky as it is a sphere impacting a cylinder (ball could go anywhere – so hard to measure rebound vel. along one dir. correctly)



All the best

Erik