Learning Forum

[Tonmoy]

Hi everyone,

We are trying to use the Rayleigh Damping model for an explicit simulation case in LS-Dyna. However, we are facing some issues with the Stiffness Proportional damping which can be implemented with the *DAMPING_PART_STIFFNESS. For the implicit case, the implementation is pretty straightforward - setting the calculated beta coefficient as negative solves the system correctly.

However, for explicit analysis, setting the beta as negative can solve the system for high frequency motions but gives erroneous results for lower frequencies. The applied motion of the explicit case was assumed to be damped for frequencies ranging from 1 to 10Hz. We found through some simplified analysis that for this case, the Rayleigh damping result through implicit and explicit analysis is different. Hence, we are confused if there is "actually" a way to implement the Rayleigh damping model in LS-Dyna. Otherwise we might need to solve the problem implicitly, which will be a nightmare considering the contact surfaces and assemblies we have modeled.

I would really appreciate it if anyone could help us with this.

Thank you.

- Tonmoy

[Jim Day]

Because no stiffness matrix is assembled in explicit analysis, *DAMPING_PART_STIFFNESS, whether COEF is entered as positive or negative, is not equivalent to traditional Rayleigh stiffness damping which, as you noted, can be obtained via implicit analysis. Rather, stiffness damping in explicit is implemented at the element level and is generally used with COEF positive and not greater than 0.10 as a means to reduce high frequency noise in the results.

[Jim Day]

A damping alternative is a frequency-independent damping option which targets a range of frequencies and a set of parts (*DAMPING_FREQUENCY_RANGE). Notes on this keyword may be obtained by Googling DAMPING_FREQUENCY_RANGE.

[Tonmoy]

Hi Dr. Day,

Good morning. Thank you for your response.

We were also exploring the frequency range damping (with the deform option, as we had rigid body modes) that you mentioned. We found that for our case, a frequency ratio of 10 (Fhigh/Flow = 10) is more appropriate. We also wanted to implement a damping coefficient of 0.05. However, with some simple analysis we found that the response was accurate for 1-2% damping when using the keyword *DAMPING_FREQUENCY_RANGE_DEFORM.

Now, considering all of it, what can be a good approach to apply 5% or 10% damping for our model? We were trying to do some nonlinear analysis by imposing earthquake motion at the base of the system. Do we need to play with the stiffness of the system as the manual mentions?

[Jim Day]

Beyond the remarks in the Manual concerning *DAMPING_FREQUENCY_RANGE(_DEFORM) and the notes I asked you to Google, there's nothing more I can add concerning use of *DAMPING_FREQUENCY_RANGE. If you don't see remarks on the "Iterative Method" in the Manual, please download a newer version of the Manual. As I read it, the Iterative Method makes stiffness or frequency adjustments essentially unnecessary.

[Tonmoy]

Thanks you Dr. Day for you kind observations. I think our last resort in explicit case will be to opt for mass damping, but we will continue to play with the other damping keywords available in the software to see if any of them will work best for the task in hand.

Thank you again for you time. Have a good weekend ahead!

- Tonmoy.