How does one accurately measure the effective dynamic load of equipment trasmitted to the floor? 1

[yihshawn]

How does one accurately measure the effective dynamic load in frequency domain exerted by heavy equipment (huge vacuum pump, multistage centrifugal pump, and etc.) transmitted to arbitrary floor without reallocating the equipment?

One method I can think of is to work out the FRF of the floor at a point close to equipment foundation with a force hammer when the equipment is in rest and capture the floor response when the machine is operating. The dynamic load can be then be determined by the relationship of input and output. According to theory it should work for cases with single-toned harmonic excitation, however the equipment is expected to emit multi-toned or non-harmonic excitation.

Is there any other established method to deal with this kind of problem?

Is there a sweet spot on the equipment where its acceleration can give us a good approximation of dynamic force generated?


P/S:
A lot of discussions in the forum leave the responsibilities to manufacturer.

http://www.eng-tips.com/viewthread.cfm?qid=335609

http://www.eng-tips.com/viewthread.cfm?qid=406382

Many textbooks or design guide in foundation engineering derive the dynamic load by empirical formulae concerning these parameters:
- bearing load rating
- maximum allowable vibration and bearing stiffness
- mechanical unbalance
Have no idea how reliable these formulae are.

 

[scheah]

I too am not aware of any established method to deal with this type of problem.

If your equipment is on isolators, you could measured the motion above and below the isolators, the difference multiplied by the stiffness gets you vertical transmitted force. The catch is you know what the stiffness is at that particular pre-load and the isolator is linear enough to make that assumption.

Depending on your floor, measuring transfer function of the floor could be difficult as the acceleration levels could be very low (seismic accelerometers?).


Best regards,
Sze Kwan (Jason) Cheah

 

[IRstuff]

FRF, by definition, is spectral, so multi-frequency shouldn't be a problem, should it?

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[BrianE22]

You mention "accurately measure" but then you list equations that would only be accurate if you knew the masses, damping and spring rates. With heave, pitch and roll I would think it would be difficult to know the effective masses and inertias. For me, "accurate" means installing load cells on the floor under the mounts and actually measuring the forces.

 

[yihshawn]

@ sk_cheah
You are right that the amplitude of the response in this circumstance is generally low (5um/s - 50um/s). Can this be compensated by moving the floor response sensor closer to the forced hammer?
From my experience a 1000g/mV sensor are likely to pick up the signal in such scenario.
It is true that with isolators beneath equipment the subsystems are well separated to be dealt as classical lumped mass model. The dynamic load in vertical direction would be k(x2-x1) + c(v2-v1). Even so, the dynamics load can be unevenly distributed on isolators which result in rocking motion, does it mean that we need to probe on N isolators + 1 floor response point following your suggestion?


@ IRstuff
May I ask what are the implications to the FRF when non-harmonic load (or even non-periodic) is exerted?


@ BrianE22
>>You mention "accurately measure" but then you list equations that would only be accurate if you knew the masses, damping and spring rates. <<
This is the main reason why I am looking to measure the force rather than deducing them from empirical formulae.
These formulae tend to overestimate the load and also lose out the spectral information, which is important to foundation and floor design.
Installing load cells on the floor under the mounts involve reallocating/lifting the equipment which is not ideal.

 

[GregLocock]

sk_cheah has identified the method I'd suggest. To make it a bit more interesting we also measure the dynamic stiffness of the isolators.

If the foundation is stiff then its displacement is zero so k*(x2-x1) merely resolves to k*x2. x2 is just the double integral of acceleration so if you are working in the frequency domain just whack an accelerometer on there and you are done.

Cheers

Greg Locock


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

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

 

[yihshawn]

Equipment that rest on column-supported floor can trigger floor response at its fundamental mode (typically 9-15Hz).

 

[scheah]

You are right that the amplitude of the response in this circumstance is generally low (5um/s - 50um/s). Can this be compensated by moving the floor response sensor closer to the forced hammer?

I'd be very nervous in swinging an impulse sledge hammer real close to the accelerometer . This is not my first choice as it is not in line with the actual load path.

It is true that with isolators beneath equipment the subsystems are well separated to be dealt as classical lumped mass model. The dynamic load in vertical direction would be k(x2-x1) + c(v2-v1). Even so, the dynamics load can be unevenly distributed on isolators which result in rocking motion, does it mean that we need to probe on N isolators + 1 floor response point following your suggestion?

Ideally, you measure just above the isolators and below the isolators all at once to get the relative motion (2*N measurements). This should capture the entire load path.


Good luck,
Sze Kwan (Jason) Cheah

 

[BrianE22]

I'm guessing the mounting lugs aren't stressed too highly but if you can find an area of bending then maybe you can strain gauge them.

 

[SwinnyGG]

I supply large/heavy equipment that vibrates and gets mounted to floors of various types.

It is absolutely my responsibility to know the loads applied by my equipment to whatever it is attached to.

In short, I would call the manufacturer first. If they give you numbers, remember that they will be quite conservative, in other words your actual service load into your floor is likely lower than what the manufacturer says it is.

 

[yihshawn]

sk_cheah hopefully I will be able to setup an experiment to validate on that (not so soon)

BrainE22 strain-gauging is more complex in operation but worth a try too on a scale down experiment, thank you for the suggestion.

jgKRI do you get the dynamic load in frequency domain? What are your method of measurement? Manufacturers (most of it) from my place can only supply things such as acceleration data(frequency domain) at bearing housing as per machine health standard, or bearing rating (which is not frequency dependent and very conservative), or balancing grade (which is not complete if hydraulic unbalance or other dynamic source is concerned).

 

[SwinnyGG]

For some machine types, yes. The typical spec on a spec sheet would be an RMS loading into any mounting provisions, but the full spectral picture is available if the customer has a specific concern or unique floor mounting.

We measure by testing equipment with load cells, accelerometers, and strain gauges in various combinations.

 

[yihshawn]

jgKRI can you enlighten me under what circumstances strain gages can be used? Did you try to measure load of equipment without isolation system?


I have a feeling that this topic opened up a large can of worms. It belongs to ill-conditioned inverse problem which recent research can be decomposed into several aspects:
1. minimize the sensor or sampling (optimized sensor placement and compressive sensing, etc)
2. optimization techniques in solving underdetermined matrix and noise reduction
3. determine force location (wave propagation)

Now I have a large collection of literature about this topic and if anyone is interested to read them do let me know.

 

[SwinnyGG]

jgKRI can you enlighten me under what circumstances strain gages can be used? Did you try to measure load of equipment without isolation system?

If our machine includes its own isolation, the specified forces on the spec sheet will be at the isolator/floor interface.

If the machine does NOT include isolation, the specified forces on the spec sheet will be at the machine mounting/floor or machine mounting/isolator interface.

So, in short, we measure both, although we do not necessarily release both- we tell the customer what they need to know to design their floor or building, and no more.

With regard to strain gauges: we have several proprietary tools we have designed which use strain gauges in combination to measure loads in configurations that would be difficult by other means.

Unfortunately I can't provide any more detail than that.

I have a feeling that this topic opened up a large can of worms.

Yup.

 

[Panther140]

I'm not sure how this machine is mounted to the floor, but maybe my idea can help or lead you to something that can help.

Strain gauge rosette on a structural component of the machine that is coupled post-dampening to the floor. Digitally record the gauge's voltage readings with whatever daq system you have (hopefully it can read and log at a few kilohertz). Process the data from each axis through a program of your choice that can do an FFT and then give you a PSD. Apply Young's modulus.

"Formal education is a weapon, whose effect depends on who holds it in his hands and at whom it is aimed." ~ Joseph Stalin