And it just may be a lot more than you would think.
While visiting a customer site to do an alignment check using the EVO laser alignment tool on a 1750 HP Electric Motor / Boiler Feed Water Pump / 3574 rpm, measurements were taken several times (x3) to make sure the values were repeatable. All three sets of measurements repeated (angular and offset values less than 0.001” variance).
The results indicated a vertical angle of +0.9mils/1″ (almost 2x 3600 RPM tolerance) with a vertical offset of +18.5 mils (9x 3600 RPM tolerance).
The mechanics I was working with had done a rim and face alignment and said their numbers were repeating as well. But their measurement results were “less than a thou rim and face” as in “dead nuts”. We decided to remeasure, dials and laser. And guess what, both methods of alignment measurement were repeating.
So, what gives? During discussions it was said that there was a “little sag” to be concerned with. So after rechecking set-ups and re-measuring, it was determined that was the only variable. We decided to take off the dial bracket as it was set up and test it on a piece of pipe in the shop.
At the top set the dial to zero.
Then rotate 180 degrees to the bottom. Watch the dial as it travels to ensure an accurate reading and to note the direction the dial needle moves + or -.
WOW! -37 thousands! -0.037” (-37 mils)
Let’s do the math: Total Indicator Reading (TIR) of 37 mills divided by 2 gives the actual offset (or error in this case). That equals 18.5 mills. Now reference the initial measurement with the EVO. The vertical offset was measured 18.5 mils.
Conclusions: If you are using dials, always check bar sag even if the set-up is over a short distance. When comparing dial values to lasers, if the only value that is different is the vertical offset, you should suspect dial/bar sag as the reason.
We weren’t done with the alignment so we returned to the boiler feed water pump and used the EVO laser alignment tool. Made one shim correction and easily made the 3600 rpm specification/tolerance.