Methods to Approximate/Guess the damping of a structure w/o any experimental data

[ResearchGuy53]

MAIN QUESTION:
Is there a procedure/method for approximating the damping properties (Rayleigh coefficients for example) of a structure when you have no data on the damping [and you cannot run experiments to determine the damping properties]?




I was wondering if there is a way I could quickly 'determine' the damping properties of a skid landing gear, without actually going out and running experimental tests.

Basically I'm running a hard-landing FEA simulation of a rotorcraft skid landing gear. To get meaningful results, the input data needs to make sense. Inertia and mass properties are known, and the stiffness properties are obtained implicitly from defining the material and geometry. So, I'm wondering what the common practice is for damping, when you have no actual data. I have to 2 sloppy ideas so far.


IDEA 1
Can I just enter no data on damping (effectively setting damping coefficients to zero, aside from plasticity and friction which would indirectly damp out energy)? This seems like a sloppy solution. Damping would still be included by the plastic limit of the material as well as friction at the skid-ground interface. However, I feel it's common practice to include other types of damping, and this seems incomplete. If the material stays within its plastic limit, that leaves friction as the ONLY damping mechanism. Hence, the damping shown in the model cannot be very accurate.


IDEA 2
Should I model damping by including Rayleigh (proportional) damping? Is there some sort of standard procedure to estimate the alpha, beta coefficients [again, without having any experimental data]? In my experience, Rayleigh damping is a non-physical model used to "force" your results to match known data and thus the (rayleigh) damping coefficients are obtained backwards. In other words, without any physical data, I do not think a Rayleigh damping model is any use to me. I could be wrong though.

Do you guys have any better ideas? Basically, if I add some sort of damping coefficient (Rayleigh or something similar) to my model, it would either need to be physically accurate OR justifiable (cited and obtained from an accepted procedure, methodology)??


Thanks

 

[BUGGAR]

When I have so many unsure aspects of a problem, I try to bracket the solution with both the highest likely solution and the lowest. Then make a judgement. A non-answer to be sure but perhaps adequate for a "non-soluble" problem.

 

[GregLocock]

Bolted steel structures typically have at least 3% critical damping, usually more. Or you could find a similar structure and hit it with a stick. You should b able to guess at how long it rings for, and the frequency, then create an SDOF model and make it do the same.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[BUGGAR]

That "stick hitting" trick sounds interesting. My wife tunes her guitar with an electronic gadget that detects frequencies - could that facilitate reading the structure vibrations?

Bob

 

[petb]

Maybe? I'm not really familiar with tuning instruments. You could tape your smartphone to the structure and hit it with a a stick. Almost all of them have acceleration sensors for three dimensions. You should be able to at least capture the first eigenfrequency. I'm not sure about the sampling frequency of said accelerometers, but IT should be sufficuent to capture any mode of interest.

/petb

 

[glass99]

For building structures seismic analysis the usual assumption is 5% of critical damping, though this includes significant energy loss for plastic deformation. For my glass structures I typically assume 1% damping.

The thing that makes damping hard to estimate is that it comes from weird places. Joint friction, aerodynamic loss, local yielding, contacts, vibration transmitted to other elements etc. Its not really a material property, at least in the application of a building structure. If you had a vibrating simply supported steel beam in a vacuum with ball bearings for supports, I assume it would vibrate forever.

petb: I have done the iPhone vibration monitoring thing before. iPhone accelerometers at least in the older models have a sampling frequency of 50Hz, which is good enough to capture a vibration of approx 12Hz if you assume 4 datapoints per oscillation.

 

[ResearchGuy53]

You guys are on the same exact page as me. The exception is that I don't have a real structure to "hit", but other than that I'm doing the exact same thing as GregLocock suggested.

Basically, I'm trying to do the same thing that GregLocock is talking about, except I replace the real-life structure (skid gear) with an FEA skid gear. The problem is that I'm wondering if this successfully approximates REAL damping. I'll summarize my process below:

1) Set up my simple mathematical model/sim. of a rotorcraft landing, which consists of a series of rigid bodies attached by joints where the stiffness and DAMPING needs to be defined. Here the damping is still unknown.

2) Set up an FEA model of a skid gear [skid attached to a rotor modeled as a single node with defined fuselage mass/inertia], impacting the ground. So there are now 2 simulations: the rigid multi-body sim. (my main model) above, and an FEA sim.

3) For the FEA model (#2), observe the dynamic behavior as if it was an SDOF system (fuselage attached to a spring). And use the time history plots to approximate stiffness and damping.


So, as you can see, I am doing a very similar process to what you're talking about except I do not have a "real life" structure. This is where the problem emerges. Is there a a way I could model the FEA skid gear to successfully predict the damping?

The short answer I believe is "no" because w/o an real structure/data, we cannot get the real damping. But I'm wondering, if I set nonconservative, energy dissipating phenomena in my model such as friction and a plastic limit, then run a sim. and get the time history response. Using the equivalent SDOF model combined with this new time history plot, could that give me a justifiable damping coefficient? In short, could I create a FEA sim that gives me approximately accurate damping, or at least a justifiable damping coefficient?