Non-Repeatability, a Little Movement can Cause a Lot of Headaches!

A recent Fixturlaser GO Basic Training class performed a field alignment on a 100 HP, 3 Phase, 5600 RPM electric motor coupled to a crude oil pump in a refinery.  The class decided it would be permissible to use the 3600 RPM setting in GO Basic active tolerance table, which is +/- 0.5 mils/1” angular and +/- 2.0 mils offset misalignment, both in the vertical and horizontal planes.  

During the alignment process, we were getting non-repeatable results in both the vertical and horizontal planes.  The angular and offset coupling values would change just enough to throw us out of tolerance.

After correcting and re-measuring several times, were we getting a little frustrated.  Something HAD to be moving.

It occurred to me that during the horizontal adjustment phase, the motor shaft was being “held in place” by one of the mechanics to steady it.  This is simply the wrong procedure.  Even on a heavy electric motor, it is difficult to prevent shaft movement by hand.  On some larger 3 phase electric motors, the shaft will roll very freely in the normal direction of rotation, but there will be some resistance in the opposite direction due to “residual magnetism” in the windings and stator.

I tried to steady the shaft myself on the next alignment and had the same results.  I did notice although, that no matter how steady I held the motor shaft during the horizontal adjustment, the shaft was indeed rolling slowly causing the inclinometers to change.  Almost unable to perceive, it moved just 1/10th of a degree every couple of seconds. During the process of moving the motor horizontally, this movement would add up to a few degrees of rotation, enough to skew the readings and create inaccuracy. So if the shaft was rolling during the horizontal correction process it was probably “drifting” during the measuring process affecting our repeatability. 

Aligning to a close alignment tolerance will be extremely difficult if you have movement in the sensors.

My solution was one that everyone should be doing.  We built a “steady rest” from a 2×4 and a steel cylinder to rest the sensors on when rotated to a horizontal plane.  With the sensors gently resting on the steady rest, we were able to take the measurements, make the live horizontal adjustments, and follow a bolt tightening sequence without any shaft roll.

The pump came right into alignment on the very next check.  This goes to show that shaft movement (sensor movement) during an alignment can cause unexpected and unrepeatable results.

Tags: ,,,,

6 Comments
  1. Stephen, I couldn’t agree more. Monitoring angular positions before taking any measurement is the best way to avoid errors. The second step is to control those errors, if they occur. I’ve tried everything from steady rests to duct tape.

    Like anything else, the more accurate the inputs are, the more accurate the output will be.

    Great post!

  2. I think you hit the nail on the head with the ‘it was probably “drifting” during the measuring process affecting our repeatability’ comment. Trying to hit a moving target is tough.

  3. Stephen Beeson, Mechanical Instructor at the Northwest Renewable Energy Institute

    Very common problem on uncoupled and flex-coupling alignments, on a 100-Hp motor no less!

    I always explain this phenomenon to the classes I teach during the instructional portion of our wind turbine technician training. The difference is that these students perform alignments on a significantly smaller, 1/3-Hp motor. The problem is typically picked up by a indicator on the DU, what I call a “backlash” detection error, (the little icon triangle with the exclamation point inside).

    The Vibralign GO manual states this icon will appear for one of 2 reasons:
    (1) Initially (prior to the readings), the sensors are MORE than 3 degrees apart in the radial direction (I.e., not parallel)
    (2) The sensors move more than 0.7 degrees apart BETWEEN readings.

    There is a way to avoid the “backlash” problem.
    A technician should carefully monitor the inclinometer readings to ensure neither of the conditions I just mentioned develops during both the sensor “reading” process or the alignment correction procedure itself. I typically have the students use a fixed object similar to your “steady rest” mechanism. 1-2 solid aluminum risers or a dial indicator base stand firmly magnetized in place normally does the trick.

    There are some other potential root causes of this problem, however, that the “steady rest” mechanism will only superficially address.

  4. Andrew Martin

    Great post James. Often the smallest things are the ones that make the biggest difference.

  5. Andrew Martin

    Great post…thanks James!

  6. Jim Robinson

    Great post about an important topic.

Leave a comment