Damping in ElasticSolve Neo-Hooke model available?

[bengt]

Hi, dear Elmer friends,
does anybody know if / how material damping can be activated within the nonlinear ElasticSolve using Neo-Hookean material??
I desperately need some damping in a transient model comprising pseudo-contact (displacement soft limiter) in order to reduce the oscillations.
I've already tried out the according keywords within material and / or boundary condition section, however without noticing any different behaviour...
If, in this case, damping is not yet available as a standard feature, would it be possible to add or "activate" it in the sources?

I'd greatly appreciate any suggestions!!
Kind regards
bengt

[mika]

Hi,

I guess you have already tried to give "Damping = Real ..." in the material section.

- Mika

[bengt]

Hi Mika,

thanks for your reply!
- yes, I've given "Damping = Real ..." in the material section, and additionally in all boundary conditions. From the sources I guessed that damping should indeed be available, however, using 0.7, at first sight I didn't see any significantly different behaviour. Using 0.9, during a very long run (> 400 timesteps, timestep size = 5e-7) over the weekend, the oscillations decreased very slowly, so I guess that there must be some damping effect... however, I don't know where the damping should be given in the sif (ok, material section, obviously). Moreover, I wasn't able to figure out which value should be optimally applied (currently trying 0.5) in order to achieve the strongest effect (closer to one, I guess..).
Would it be possible for you to briefly clarify if damping is principally available with this model and if a higher value such as >= 0.9 could indeed be needed here?
Thanks so much for your time!
Regards
bengt

[mika]

Hi,

I must say that I'm not an expert on material damping, but I'm giving some thoughts anyhow. The only damping mechanism which should be available in this connection is a version of proportional damping where the fully discrete equations are modified by adding the first-order time derivative term with a coefficient matrix C = b * M (M is the mass matrix and b is the parameter to be tuned). One might expect that the effect of this type of damping increases for oscillatory motions that are associated with low frequencies, while high-frequency modes may be nearly unaffected under this damping. Might it be that the characteristic time scales of your problem are so short that this type of damping is unable to produce a noticeable effect? I know that in the context of linear theory the high-frequency modes can also be damped by taking C = a * K + b * M where K is the stiffness matrix, but I don't know how the similar effect should be produced in the context of nonlinear theory.

- Mika

[bengt]

Hi Mika,

thanks so much for your detailed explanations, I think the timescale indeed could be the problem... Using the sif-reload feature, I'm currently trying to gradually increase the timestep size throughout the course of the simulation, let's see what happens. Just one more question in order to be sure about the correct implementation within the sif: is the "damping = real ..." entry given only within the material section or do I have to give it within the boundary conditions, too (guess it should only be needed in material?? however, in the models manual I found the damping keyword also in the bc context - was a bit confusing..)?
Do you have any suggestions regarding a common useful choice of the damping value?
Again, thanks a lot for your time!!
Regards
bengt

[raback]

Hi

It is the Material section.

I don't think that damping values should be limited to [0,1]. Looking at dimensions I would think that if you want to have damping of the same order as inertia then having b=1/<t>, where <t> is the timescale of motion, would make sense.

-Peter

[bengt]

Hi Peter,

thanks a lot for your explanations!! I'll try out and keep you all informed about results.
Regards,
bengt

[bengt]

Hi,
"b=1/<t>" is working perfectly!!!
Thanks again and best regards
bengt