In our previous issue of Vibration Matters, we concluded our look at system dynamics with the topic   System Dynamics: Multi-degree of freedom (MDOF) systems.  In today's issue, we are starting our new topic Resonance, where we will look at how to identify it and what type of vibration tests you can use to troubleshoot and ultimately correct the problem.

Unusual Failures

If you have machinery failures or structural failures that seem to be as a result of fatiguing and you do not have any other explanations, then you should consider resonance as a possible cause. Structures should last a very long time, and fatigue failures should only occur after MANY years of service. Examples of failure modes include:

• Broken welds
• Cracked and leaking pipes
• Premature machine failure
• Broken or cracked shafts
• Foundation cracks

In the following example, high vibration due to resonance caused the hand rail so slowly wear away.

(Images courtesy of Simon Hurricks, Genesis Power, NZ)

Tell-Tale Signs in the Spectrum

There are four tell-tale signs:

1. Unusually high peaks in the spectrum.
• Peaks are amplified

2. Directional vibration
• High vibration levels in one direction/axis but not in another.

3. “Haystacks” and “humps”
• Areas in the spectrum where the noise floor, and any peaks in the vicinity, seem to have been raised.

4. Peaks that change amplitude when machine speed changes.
Because resonances amplify vibration, peaks in the spectrum that may normally be quite low (at the blade pass frequency, or gear mesh frequency, etc.) can increase dramatically. When you see a high peak, remember to consider the possibility that it is being amplified by a resonance.

If you have a variable speed machine, you may observe that when the machine is operating at its highest speed, the blade pass frequency may coincide with a natural frequency, so the machine resonates, and the blade pass frequency peaks will increase in amplitude. However, you may also observe that when it is operating at a lower speed, the blade pass frequency would be below the natural frequency, so the amplitude would drop to “normal” levels, and the machine will not be in resonance.

Also, when you study the structure more closely, you will notice that the mass, stiffness and damping can be very different vertically, horizontally and axially. So, there may be a resonance in the horizontal direction (i.e. it is vibrating from side-to-side), but not so vertically (there will be natural frequencies, but they may not be excited). In the case of vertical pumps, however, resonances can quite often exist radially, but not at the same frequency vertically (axially along the machine).

“Humps” and “Hay Stacks”

As we have just seen, any vibration that coincides with a natural frequency will be amplified. And, as we have just seen, it does not have to be a perfect match. In the spectra below, you can see the “hump” in the data from 30 to 65 Hz. It looks like the 3X peak is being amplified.

The data above came from the vertical pump pictured below. You can see the photo of the installed machine (where the data was taken), as well as an old machine that had previously failed:

In our next issue, we will look at the various methods to test for resonant conditions.