Practical Solutions to Machinery and Maintenance Vibration Problems
Chapter 3, Detuning and Proving Resonance
Section 21, Vibration Due to Foot-Related Resonance (FRR)
Most vibration specialists are well aware of the usual machine and structural parts that are most subject to resonance, such as spans of piping, sections of steel bases, beams, columns, pedestals, portions of ductwork (and very rarely, the rotor itself).
However, there was always another source for resonance which was generally unknown because it was not easily detected with the usual methods, such as determination of modal shapes, bump tests, shaker tests and so on. This type of resonance was most usually affected by the machine's feet. Therefore, Update created the expression "foot-related resonance" with its abbreviation "FRR."
Foot-related resonance doesn't seem to be well-known by most vibration specialists. There is usually some knowledge of the possible damaging effects of machinery "soft feet." However, as Update gained more and more experience on this subject, it realized that they are really two separate subjects, each distinctly different from the other. One is the well-known "soft feet," that is, when a machine's frame or feet deflect slightly when the hold-down bolts are tightened or loosened.
Whether it is just the foot that is deflecting or the machine's frame, the "soft feet" phenomenon does not necessarily result in increased vibration. When vibration does increase, it is usually due to resulting changes in shaft-to-shaft alignment, bearing-to-bearing alignment, air gap concentricity and so on. Most of the time, however, moderate levels of soft feet do not generally cause increases in vibration amplitudes.
While the machine is running, look for a definite and appreciable drop in vibration amplitude somewhere around 30 percent or more, when a certain bolt or combination of bolts are loosened. For machines with high amplitudes, the decrease in vibration may drop several hundred percent. For example, an amplitude of 0.6 in/sec (approximately 15.0 mm/sec) could suddenly decrease to 0.05 in/sec (approximately 1.3 mm/sec) when one or two bolts are loosened. It used to be presumed that the change was due to soft feet deflection. However, the drastic change often occurred when the loosened bolt caused no deflection change! Most often the vibration suddenly increased the instant that the finger-tightened bolthead touched the foot.
By various means, including "before and after" phase changes, it was determined that the phenomenon was due to resonance rather than foot or frame deflection. For the usual resonant parts, several methods are taught on how to determine which specific part is resonant. However with FRR, almost always there is no specific length of "spring system" that is found to be resonant. All the symptoms of resonance are present, but finding the actual resonant part by the usual means is almost impossible. It has been determined that it is the resonance of the machine's total structure.
To understand this different resonance situation, imagine that a force is applied axially at the top 12:00 o'clock position of, for example, a motor's frame. The resulting deflection will depend on the rigidity in that direction. Its resonant frequency will be related to that deflection. Applying the same amount of force at or near the same point, but in the horizontal direction, will most likely result in an entirely different amount of deflection. Obviously, its resonant frequency will be different. This can be repeated throughout the motor frame, not only point-by-point, but also by forces applied at more than one point at a time, such as horizontally in one direction at the outboard bearing and, at the same time, horizontally in the opposite direction at the inboard bearing. The resulting deflection due to twist, will have its own resonant frequency.
It's easy to see that rather than just individual part resonances, there are also several resonances of the total structure. It is suspected that these resonant frequencies are changed appreciably when the motor frame is bolted down as compared to when it is totally free. In like manner, certain frame resonances are changed appreciably when only three bolts are tight and the other one loose (motor supported by a triangle rather than by a rectangle), or when two bolts are loose at diagonally opposite corners.
It should now be easier to visualize some of the possible frame-related resonant frequency changes that occur not due to foot/frame deflections, but instead due to the various changes in the frame rigidities when certain bolts are loosened and the resulting changes in resonance frequency.
FRR cannot only increase vibration amplitudes, but can make vibration analysis considerably more difficult. For example, resonance to a specific frequency usually occurs in only one direction for a specific machine situation. The very large amplitude in one direction and not in the other, might cause the analyst to conclude that a machine is misaligned rather than out of balance. Or, resonance may change the phases in one direction from what they would have been without resonance, but in the other non-resonant direction, phases are not altered. If the analyst doesn't truly understand or consider resonance as part of the problem, then analysis can be extremely difficult.
Even evaluation of amplitudes at various harmonic frequencies gets distorted. For example, the 1 x rpm vibration can be magnified so much that, to keep the plot on scale, the non-resonated harmonics look very small in comparison. If the 1 x rpm vibration is not resonated by FRR, its amplitude and those of the accompanying harmonics will be much easier to compare and therefore to properly analyze.
FRR can resonate any frequency to which it is tuned. Most usually, it resonates the 1 x rpm frequency. Sometimes it resonates a harmonic frequency, such as 2 x rpm or rpm x number of impeller vanes. Resonating the amplitude at electrical hum frequency is also relatively common.
A real life situation will dramatically show how FRR can distort analysis. Fig. 1 shows the original spectrum obtained at the point of maximum vibration. Note that the amplitude at 1798 rpm is about 8.5 mils, converting to approximately 0.85 in/sec (approximately 21.5 mm/sec). The harmonics are extremely small in comparison. Based only on the spectrum in Fig. 1, balancing the rotor seems to be the most probable solution. Yet, if the analyst proceeds with balancing, what will be the result?
To find out, see Fig. 2 for the spectrum obtained at the same point after only one bolt was loosened. Amazing isn't it? The amplitude dropped from over 8 mils to less than 0.5 mils! Imagine what would have happened if FRR was not discovered! Either the rotor would have been removed and sent to the balancing shop, and the shop would have found it in balance; or, each field balancing trial weight would have unbalanced the already balanced rotor. As attempts to balance failed, it would have been tempting to try other solutions, such as better alignment and so on. This too would only have made it worse.
About 80 percent of the time, the FRR results from machine feet that
require re-shimming. Usually this involves carefully measuring the spaces
under each corner of the foot and can require three to four different
shim heights. For about 20 percent of the remaining situations, no variation
of shims seems to work. Instead, either the bolt(s) has to remain loose
or the specific resonant section must be found and detuned. This usually
involves resonance at the angle between two connecting parts, rather
than resonance at a span of one part. As this is not easy to correct,
in some extreme situations, the loosened bolt that cured the problem
remained loose. At some plants, safety requirements mean that all bolts
must be tightened. If so, the only solution is to truly locate the resonant
section, and detune it. Sometime, the FRR is cured by locating a resonant
part, such as an attached pipe or portion of a steel base, and detuning
This textbook contains only part of the information in our Practical Vibration Analysis seminar.