This theoretical and also a very practical question may sound naive but I could not find an answer in my mind without performing several bump tests on various components. Here it goes.

When performing a bump test and descovering a confirmed resonance peak(s) how do we know it's (their) association with a particular component of rotating equipment? Is it the pedestal, rotor, or a blade resonating?

Let's rephrase this question. Will bumping of a rotor, instead of a pedestal or bearing housing, produce different response allowing to narrow down the resonant culprit and why?

We all have seen more then once that as a result of a bump test of a pedestal (while using seismic sensors on the pedestal), bracing of the strucure would be suggested. Why? May be this is the rotor going into a flexural mode shape, not the structure?

Will appreciate you opinion.
Dave

This message has been edited. Last edited by: Dave_man,

Hello Dave

You must "see" the mode of the resonance, what parts moves, where are the nodes (points of no movement).
You have to look for points of max respons, and for points of zero respons. This wil also be the points of max vib ans min vib in operation.
You must do some OSD in your mind.

Once you found the most critical place (max deflection, direction of mouvement) you have to immobilise this point bij connecting to a fix point.
And the miracle happens ( in most cases).

Dave,
Your question is a good question, as you are trying to really understand what is going on.
In general one should expect that whacking one part of the system will excite the whole system.
The exception would be modes where parts of the system are essentially "decoupled" from one another, ie part of the system vibrates but the other part doesn't.
Of course to excite any particular mode you have to strike it at a point where there is some motion in that mode. If you strike it at a node you won't get any energy into that mode.
So, if you have modes where the whole system participates, and you whack it at other than a node, you'll get it going.
To make a thorough study of a system you should impact at different locations, and measure response at different locations.

One more puzzle piece: To identify a rotor critical, well in a standing still rotor, you can normally identify it from being fairly sharp, i.e. less damping than the structure, plus a dip, a minimum just very near. A rotor, being symmetrical could maybe expected to have just a single frequency vertically and horsiontally. But since the rotor rests at two very stiff points in the bearings in the vertical direction and the shaft is relatively "loose" looking horisontally, there are in fact two frequencies, very near each other. When you measure in one direction, you see the peak in that, and the ringing in the other will suck energy an cause a dip at the other. Shift the sensor 90 and hit 90 and see how the former dip is now a peak and the former peak is now a dip. Vertical resonance is normally less damped, hence sharper than the horisontal.

Arne – as you mentioned recently, the bearing stiffness can be different at rest than rotating. Under what conditions might we expect that a bump test would show the rotor critical? (mabye more likely in a/f bearings vs sleeve?) .

Also if we are looking for the translational (vs rotational) modes, would you bump in the center of the stator rather than on the bearing housing? If it is a large horizontal sleeve bearing motor, the outer frame is fairly flimsy and won’t transmit much to the bearings... any suggestions for that case?

If you are looking for the first banana bending mode and have luck that the nodes are smack on where the rotor has its support in space, then the bump test standing still to real speed will only differ a slight bit due to the polar moment of inertia. Most critical speed softwares can do two versions of the calculation, with and without rotation impact. A motor is almost the same running/rotating. A slim multi stage pump can have a great difference. Like my favorite charging pump from United Pumps, 1780 to past 4245 from standing still to running. If the bearings are way off from the nodes, the tuning effect can be substancial. This can be used to sve a machine. If the critical is on the running speed, placing the bearing on a soft pad can tune nicely. I learned that from prof Gunther. Saved me several times.

A further case, a vertical large motor with sleeve bearings: The upper bearing was at first stiff - steel to steel. The tuning effect was just some 5 percent downwards and the crit was very near to speed. But as the level was a bit high I took on to trim balance it. When I had taken the unbalance down nicely, the bearing had no radial force anymore so it started shaking violently with 0,44xspeed going slowly further downwards as it wore and after some half a minute the bearing wiped. After licking the wounds and new bearing installed we added some unbalance and it ran perfectly again but on the up-slope of the crit. So the vendor entered the scene and said place these plastic blocks to hold the bearing housing and you can balance to nothing if you like. Damping was enough to keep bearing stable. But even nicer, the crit went up more than 4 Hz and gave a nice margin to the running speed. So a changed bearing stiffness can do good results.

A/f bearings are basically much stifffer, right, but difference vertical to horisontal can be substancial. For the frequencis of interest, I would say up to 120 Hz and a bit more, the stators are often soft and have one or even two important resonances, but down in the speed range, it is rarely a stator problem, but I have had such a case for a large 10 MVA refiner motor. In another case, the bearing was sitting on the endbell structure with too little connections to the stator body. So the resonance and crit was way too near running speed. But adding bracing inside gave the needed stiffness to get both the structure up above 35 hz (running 1500 RPM) and the crit tuned to 42 Hz, nicely in between speed and 2x.

I would not consider a flimsy stator structure as a reason not to bumnp both at coupling and on bearings and on rotor core end and stator core end and as a support on stator midspan. Although, if the problem is strong and machine large and has a fat price, it might be ok to do a good modal test. That can either have arowing sensor or rowing excitation. I have done both but like the rowing sensor better. The drawback is that choosing the excitation point is tricky if you like to see all modes nicely. But better excite three rounds in at least two 90 degree off points and one at 45 to the machines axis and have patience with a rowing sensor all three rounds. Often the setup time is the big part. When you finally are hitting along with the hammer, you soon get through even some hundred ponts reasonably fast.