Practical Solutions to Machinery and Maintenance Vibration Problems
Chapter 14, Mental Approach
Section 2, Approaching a Machinery Vibration Problem
Why is it that some people find vibration analysis relatively easy, and others find it so much more difficult? Experience reveals that it doesn't directly relate to intelligence, but instead to the way the problem is approached. Intelligence cannot be taught, but an intelligent approach can be. This section will concentrate on those steps used to get at the source of the trouble in the easiest manner and the shortest amount of time.
Meeting Before Going to the Machine Site
As a result of the urgency of most vibration situations, the analyst will often be coerced into rushing directly to the machine, taking a few readings and almost immediately be faced with the anxious questions, "What's wrong with it?" or "What did you find out?" Therefore, try hard to first have a quick meeting with those who would know something about the machine's history. Answers to the following questions will often enable a considerably faster and more accurate solution to the problem.
A. How long has this machine been in operation? (Ask for only a rough approximation.)
B. What specific problems has the vibration created, such as excessive bearing wear, seal wear, bad product surface or cracking? If cracking, where?
C. How bad was the vibration when the machine was first installed?
D. If good, when did it first start vibrating, and what physical changes occurred at that time?
E. If vibration was bad from the beginning, why didn't the manufacturer correct it? What did the manufacturer claim was not the reason for vibrating? Did the field service representative understand resonance or perform foot-related resonance tests? Were phase readings taken for more complete analysis? Is the rotor narrow or overhung, suggesting it was erroneously balanced in only a single plane? What was the alignment method?
When at the machine, don't let the vibration scare you or make you
uptight. Some severely vibrating machines will seem almost impossible
to correct. But with step-by-step analysis, the trouble can be found
and often requires a very ordinary correction, like alignment, balancing
or bracing the right part to eliminate resonance. Even the more rare
vibration troubles, such as that occurring at gearmesh frequency, will
usually be related to the large orbits caused by unbalance, coupling
misalignment, or magnification of resonance. Here are the steps taken,
and the questions asked, for a typical vibration analysis problem. If
there is a larger than usual amplitude at vanepass frequency, inquire
about flow rates, possible over-design, etc.
1. Holding a coin in your hand or the stem of a vibration pickup, "hand feel" the vibration at the usual points for measuring vertical, horizontal and axial vibrations at bearing caps or housings. Mentally record which seem to be the roughest, considering both amplitude and frequency. Continue by feeling the attached typical items, such as pipes, sections of base, pedestals, valve stems, valves, ribbing, ductwork and so on. Walk around the floor area, reaching about three yards or meters beyond the machine's base to determine whether certain areas are producing a lot more vibration than others. Especially note whether the floor vibration is considerably greater away from the machine than right at the machine (waves in floor that may indicate resonance).
2. If there seems to be considerably more vibration on a portion of an attached part than at the worst bearing, then be prepared to check that section for a resonant condition.
3. Repeat the above, this time noting amplitudes as indicated on the vibration instrument. It isn't necessary at this stage to go to each point, but measure at those that felt the worst.
4. Measure the 1 x rpm vibration at all bearings in the horizontal, vertical and axial directions.
Enter the 1 x rpm vibration amplitudes and phases on the "Running Speed Amplitude and Phase Data" form.
5. Obtain a spectrum of amplitude vs frequency starting with a wide
frequency range, such as 0 to 2 kHz (in order to include the "higher
frequencies"). Then obtain another spectrum for a lower frequency
range that includes at least the first 5 or 6 harmonics of
6. Most importantly -- after analyzing the frequency of the worst vibration, determine what is not causing that vibration. For example, if the worst vibration is at rpm x number of vanes but is very good at 1 x rpm, determine that it is not caused by unbalance, bent shaft or eccentric armature.
7. Step-by-step, eliminate all the usual vibration sources for all the vibration frequencies that shows acceptable amplitudes.
8. Concentrating now on the "bad" vibration amplitude, list all the possible sources that could create vibration at that specific frequency. For example, if the worst vibration is at a frequency of 2 x rpm, list such items as loose base, loose bearings, misaligned coupling or pulleys, etc.
9. Concentrating only on what is left, analyze the phase readings to determine the shaking modes in the vertical, horizontal and axial directions, rather than just the amplitudes. Analyze other symptoms such as beats.
10. Determine if resonance is part of the problem and if so, how or why.
11. Review again the machine's history, and mark those items that could have caused vibration at the trouble frequency. For example, if the motor was rebuilt and balanced, not only could the balancing be in error but, more probably, the realignment may have gone wrong. If something got excessively hot, something could have warped. If bearings were pressed into place, the fits may not have been proper, etc. Unfortunately, it is also not uncommon to get dirt, burrs, or debris in a bearing bore during replacement.
12. From the analysis, now determine the cause. Although almost all machines will have only one major vibration source, occasionally expect two or three problems, such as a combination of unbalance and misalignment. Or, the vibration may be magnified by a resonance, such as a section of a weldment.
13. When your conclusion is almost positive, ask yourself: “What are the reasonable chances for this to really happen?”
14. The last but very important step -- when you think you have the answer, ask yourself:
"What else could cause the same symptoms?"
"What are the real probabilities of this going wrong?"
Sometimes, this will help you search just a bit longer or with a little more perception. Finally, you can never be 100 percent sure. Using these procedures and the basic teachings from Update's course, accuracy of diagnoses should be within 90 to over 95 percent.
This textbook contains only part of the information in our Practical Solutions seminar.