PSD interpretation 1

[Mica188]

Hi,
I'm newbie in vibration field. How can you interpete PSD data? In other words, I have data of a PSD (g^2/Hz vs Hz)curve for a truck, how can I know or calculate the objet that I want to put in is safe (not damage by the truck's vibration)?

Thanks,

Mica

 

[CESSNA1]

MICA188: This is difficult to explain I suggest you do some research or take courses (

http://www.tti4edu.com).

The Power Spectral Density (PSD)Curve you have shows a plot of Acceleration Spectral Density (G^2(rms)/Hz) verses frequency (Hz). The vertical lines that you see are the end views of sine waves. The rms amplitude of the sine wave is read in G^2(rms)/Hz. To find information about the sine wave take the square root of the value of the wave at one frequency. That gives the rms amplitude of the sine wave at that frequency. Then multiply by 1.414 to get the peak value. To find the G(rms) value of the entire spectrum,find the area under the curve. This is log-log plot so you just cannot do it as if it was linear. You need equations dervived for log-log plots. The best way to evaluate the equipment is to put it on a shaker table driven by the PSD spectrum you have. Rmemeber the spectrum you have is for one road surface and other road surfaces may be used. You can also do accelerated testing on the shaker such as HASS or others

Hope this helps
Regards

Dave

 

[Mica188]

So, if I do a shaker table test to my equipment. I will obtain an Acceleration Spectral Density curve then I compare to my Acceleration Spectral Density curve of the truck. With these two curves, I can compare which frequency to be avoid or truck transportation is to be avoid. Is it that what you mean?

Thanks again,

Mica

 

[IRstuff]

But you need to know what your device can handle to start with. Otherwise, I'd suggest packing it all in a box of styrofoam peanuts.

TTFN

 

[IRstuff]

Your perspective is odd. Usually, you design your device to survive the transport. You do not design the transport to protect the device.

Also, you've not indicated what external environment the truck is seeing. If all it's doing is primary roads, there's no much that would affect most devices. On the other hand, a military truck going over railroad tie test track is wholely different matter.

TTFN

 

[Mica188]

Yes, but I just want to know what is the worst scenario. And also, which type of transportation is the worst for my equipment. I have several Acceleration Spectral Density transportation curves. I have to know the limitation of the equipment while transported with different transport.

Thanks,

Mica

 

[CESSNA1]

MICA188: Do a shaker test using the PSD you have as a spectrum to drive the shaker. A Highly Accelerated Stress Screen (HASS) test will proabably be more beneficial.

Regards
Dave

 

[Mica188]

Thanks for all the feedbacks. For the interpretation of the PSD curve, for example, at 10 Hz I read 0.0012 g^2/Hz. If my equipment has the same response, this means is bad?

Mica

 

[Twoballcane]

Check this site out for some intro to random vibe...
“

http://www.electronics-cooling.com/html/2005_august_techbrief.html”

Tobalcane
"If you avoid failure, you also avoid success."

 

[SomeYahoo]

Mica,
The PSD is a tool for looking at time/distance based data in the frequency domain. In your case (if I am interpreting correctly), you have a road profile. The road profile was likely created by measuring acceleration/force on a wheel as it rolled down the road.

That said, the PSDs you have are the INPUT into your system. The reccomendation by CESSNA is spot on; put your equipment on the shaker table and give the PSD to the operator. They will convert this frequency data back into time domain and give your equipment a ride.

These are the basics... you still need to look at acceleration factors (how much time), what vehicle it will be on (remember you have road profiles, there is a suspension and tire between you and the road), and other factors.

Hope I didn't muddle that up too much...

 

[Mica188]

After a shaker table test, I will have a PSD profile of my equipment. With these data I will be able to compare with my PSD truck's curve. What should my equipment's curve look like to be not interaction with the truck's vibration? Should my equipment PSD curve below the PSD truck's curve?

Also, does the shaker table test will be able to provide the naturel frequency of my equipment?

Thanks,

Mica

 

[SomeYahoo]

After the shaker test if your equipment is functional in most cases it is a pass. If you choose to put accelerometers on your equipment you can compare input and output to create transfer functions (AKA attenuation/amplification/transmissibility curves). Your natural frequency will show up as a peak in the transfer function curve.

It is likely your equipment will have some transmissibility greater than one, and some less than one. All depends on stiffness, accel location, etc.

In essence, you will have to see what conclusions you can draw based on the data after the test is complete.

 

[IRstuff]

You need to more clearly explain what your equipment is supposed to be doing on the truck. You made it sound like the truck was pure for transportation of the equipment. If so, then the PSD of your equipment is completely irrelevant. The only thing that matters is whether your equipment can survive the shock and vibration of the truck.

I suggest that you get a copy of MIL-STD-810 and do some reading or get another engineer to help you.

TTFN

 

[Twoballcane]

“After a shaker table test, I will have a PSD profile of my equipment. With these data I will be able to compare with my PSD truck's curve. What should my equipment's curve look like to be not interaction with the truck's vibration? Should my equipment PSD curve below the PSD truck's curve?

Also, does the shaker table test will be able to provide the naturel frequency of my equipment?”

When you put your unit under test (uut) on the shaker, you have to put an accelerometer (response) on the part that you have the most concern. Now two things can happen. If your control plot (truck curve) matches your response curve (uut) you can not get any better than that. This means that the uut is basically moving with the truck. If your uut survives with out incident, you’re golden. If you see that the response plot has a spike anywhere along the control plot, you may have an issue. One, this is the natural frequency of spot you have the response accelerometer on, and second depending the calculated Q (transmisibility) you can calculate Gout. Once you do that, you have to go back into your design and calculate the stress and strain due to the Gout load. If your stress due to the Gout is less by a factor of safety your golden, if not, you may have to redesign.




Tobalcane
"If you avoid failure, you also avoid success."

 

[ripo]

In my understanding you are correct in selecting, as first what is going to be the worst "realistic" profile for your equipment. As second step, would be to know the transfer functions of your equipment against vibration (sine survey 10hz to 500Hz) in order to correlate the response of your equipment to the "realistic" random profile.
As third step, would be to know if the equipment has been designed for support shock or not. If yes, it is possible to compute the responses multiplied by the "realistic" random profile, and compute the resultant grms, and see if it is less than the shock to support.
If the equipment has not been designed to support shock, the way will be to select the appropriate transportation package.
MIL-STD-810 gives quite a lot about how to proceed.

 

[Mica188]

Does exist an other way to find the nature frequecy of my uut? What I mean is, if I have the naturel frequency of my uut, for example my uut has a naturel frequency of 20 Hz, so I can go to the truck's and train's PSD curves and compare at that frequency what are the values. Then if my uut is below or above 10% of that value my uut is safe to transport in these vehicles. I think this method is simple but, for the first order of anaylsis, is it correct to think like this? After, if my uut naturel frequency interact with my truck's PSD curve then I can do a shaker table test of my uut for the truck's profile. Is that right?

Thanks,

Mica

 

[IRstuff]

The two ways are FEA or test. Testing generally has better verisimilitude, since FEA is only as good as the model you input.

Also, generally, you should be running your UUT against the truck profile.

TTFN

 

[Mica188]

Thanks for the info. You said testing, what do you mean by that? Is it the same method that you put your UUT in a shaker table? FEA = Finite Element Analysis ?


Thanks again,

Mica

 

[IRstuff]

yes
yes

TTFN

 

[Mica188]

Thanks,

Mica