A couple of months ago, I was invited to speak on the subject of pump reliability to a group of sixty engineers during a Vibration Institute session. All of these vibration experts, with excellent knowledge of rotating machinery, vibrations and alignment, were stunned when, at the end of the talk, I challenged them to tell me why a dial indicator alignment (or even laser alignment) works better than the good old "straight edge" method.
While it would be intuitively silly to question (!?) the importance of driving the alignment to "as best as is humanly possible," few attending this meeting were able to substantiate this belief by specific, hard, field-proven facts. Will a dial-indicator-aligned pump last 10 times longer than a straight-edge-aligned one? Will its vibrations be 10 times lower? 5 times lower?
In my field troubleshooting experience, a "bad pump" is not the one that was aligned by a straight-edge method instead of a laser, but the one that was not aligned at all! When I look at the pump-to-motor shafts of such "bad actors," I do not see 0.020-in off, I usually see a full inch off! These are the units that produce 0.80-in/sec vibration values. After these are pointed out and the problem is corrected, even a straight-edge-aligned pump immediately becomes a "good" pump, one that is no longer the cause of trouble and talk at the plant.
So I challenged the sixty vibration experts to give me personal knowledge of two pumps, side by side, one aligned by a straight edge and the other by a dial indicator (or laser), with specific and appreciable difference in vibrations. Was the straight-edge-aligned pump a source of talk as a bad actor at the plant, and the other (the properly laser-aligned one) a pride of the plant, a no-problem unit? No one in the audience could answer that.
Ironically, I could not answer that either. I could present all sorts of theoretical reasons and emotionally charged convictions that a "good alignment is great!", but I had no field data to support such emotions.
A few months later, haunted by my own sense of incompleteness, I did my own study of a pump in the field, a relatively small single stage, end-suction ANSI pump, operated at 3600-rpm. During field work, I first aligned the pump using a good-old straight-edge method, much to the horror of the theoreticians and delight of the local maintenance folks. Though I am not the best field mechanic, I can still align a pump within 0.020-in relatively quickly, which is what a typical straight-edge alignment does. After we started the pump and measured the vibrations on the bearing housing, they were recorded at 0.08-in/sec. Not bad after ten minutes of work.
Next, I stopped the pump, realigned it using the reverse dial indicator method, and aligned it under 0.002-in after about an hour. This included attaching indicators, shimming the motor, and other "please-help-me" techniques. After the pump was restarted, the vibrations still measured 0.08-in/sec.
Is it possible to predict the tangible impact on pump reliability that laser alignment has over straight edge alignment?
Photo courtesy of Ludeca, Inc.
So the bottom line here is this: does better alignment really improve reliability? According to our limited vibration comparison test, the answer is no. However, we did not monitor pump reliability over an extended period of time. Even though the vibrations did not change, would a better-aligned pump last longer, if only we were willing to wait, monitor, and see? Probably so, and the following explanation may clarify why.
What are the elements of the pump that would be an immediate suspect for lowered life due to sloppy alignment? Coupling, seals, and bearings. Many couplings these days, however, have allowable misalignment substantially beyond the noted 0.020-in, so this is not an issue as long as at least some attention is given to alignment.
Let's consider bearings next. For a typical small end suction pump (an MTX frame, for example), the internal bearing load is approximately 1400-lb (ref. [1]), and L10 life calculates for its 5309 size bearing (Cbrg = 16,400-lb) for 20,052 hours at 3600-rpm, as reproduced by a formula from reference [1]:
L10 = 2.5 x (16,400/1400)3 x 106 / (60 x 3600) = 20,052 hrs ~ 2.3 years
For a typical 20-lb rotor of such a pump, a 0.020-in TIR produces additional load of:
F = mR/g x (π x RPM/30)2 = 20 x (0.020/2) x (3.14 x 3600/30)2 = 400 lbs (ref. [2])
Thus, the new bearing life is:
L10 = 2.5 x [16,400 / (1400 + 400)] x 106 / (60 x 3600) = 9276 hrs ~ 1.1 yrs
And there you have it: bearing life reduces in half when a "quick and simple" straight edge method is used. Now that we have developed some tangible numbers to actually justify those earlier emotions, this phenomenon seems even more interesting because it is not immediately apparent from the overall vibrations standpoint.
But a challenge still remains. Though all of these calculated numbers work well in theory, what would real, field-substantiated data tell us? So far, unfortunately, that answer still remains a mystery. As we all know, the field has its own rules, despite what any calculations made in air-conditioned offices try to convince us otherwise.
By the way, what about seals? Let's leave them for you to consider. What do you think? Can you present a defendable number? Let us know.
As always, the best answer wins a free admission to our next Pump School. Send your feedback to me at my website (see below). Alignment experts, what's wrong with this picture? Let's see some facts!
References
[1] Nelik, L., "Bearing Life extension and reliability Features of Modern ANSI Pumps," 2nd international Conference on Improving Reliability in Petroleum Refineries and Chemical and natural Gas Plants, H. Bloch, Conference Chairman, 1993.
[2] Nelik, L., and Jackson, C., "Effects of Mechanical Unbalance on Vibrations, Forces and reliability of a Single-Stage centrifugal Pump," 12th Texas A&M Pump Symposium, Houston, 1995.
Pumps and Systems, February 2007
