How to Tell Angular Misalignment From Parallel Offset

Why this matters

A single reading at the coupling cannot tell you whether the shafts are offset, angled, or both, and the three cases need three completely different corrections at the feet. A tech who measures one plane, sees eight mils, and pulls eight mils of shim out of every foot will get the indicator to read zero and leave the angle exactly where it was. The machine comes back. This is the measurement discipline that separates the two errors, and the arithmetic that turns them into a shim number for each foot.

Lock it out, then keep the rotor controlled while you turn it

Open and lock the motor disconnect and prove the terminals dead with a live-dead-live check on a known live source before and after (29 CFR 1910.333(b)(2); NFPA 70E-2021, 120.5). Work the mechanical side under 29 CFR 1910.147: confirm zero rotation, and isolate, lock, relieve and drain any pressurized or hot line into the machine before a foot bolt is touched, because a hot line releases into the machine when a foot lets go. Casings that ran above about 140 F burn on contact, so read surface temperature with a non-contact infrared thermometer and let the machine cool rather than checking it by hand.

This procedure requires you to turn the shafts by hand, which is a controlled exception to the blocked-rotor rule and needs its own handling. Turn with a bar on a designated barring point or on the shaft itself, never by gripping the coupling hub or reaching between the hubs, keep the disconnect locked for the whole session, and re-block the rotor any time a hand goes into the coupling gap to move a bracket or read a feeler. Mount and dismount indicator brackets with the rotor blocked, not mid-sweep.

Step 1: Understand why one plane is not enough

Two shafts have a relationship that varies along their length. At any single axial station, you can measure how far the movable shaft's centerline sits from the stationary shaft's centerline extended. That one number is a position, and a position by itself has no slope in it.

Offset and angle are position and slope. You get a slope by measuring position at two stations and dividing the change by the distance between them. That is the entire method. Every tool below is a different way of getting two stations.

If you skip this step and measure only at the coupling, you learn the position at one place and nothing about how it is changing, which is the same as knowing where you are and not which way you are pointing.

Step 2: Pick the tool, knowing what each one actually gives you

Straightedge and feeler gauge. Reads the offset between the two hub outside diameters at one station, and only if both hubs are the same diameter and both run true to their shafts. Useful for finding a gross error before you set up properly. It resolves angle poorly and it inherits any hub runout.

Rim and face with a dial indicator. The rim indicator reads offset at its measurement plane. The face indicator reads angle directly, because the face is a second station displaced from the rim by the radius of the sweep. It is fast and it is the method most affected by axial float.

Reverse dial. Two rim readings, one indicator on each shaft reading the other hub, at two axial planes. Gives both position and slope from two clean rim measurements, and it is immune to axial rotor float because nothing is read off a face.

Laser systems. Same geometry, done electronically, with the two stations built into the heads. They do not remove the need to understand the two-station principle, they just do the arithmetic.

Skipping the tool decision costs you on sleeve-bearing machines specifically: a rotor that floats axially during the sweep corrupts every face reading, and the resulting angle is fiction. On those machines either push the rotor to the same axial position for every reading, or use reverse dial.

Step 3: Measure and subtract bracket sag before anything else

An indicator bracket spanning a coupling droops under the indicator's weight, and the droop reverses as the bracket rotates from top to bottom. That reversal lands directly in your vertical readings and it is not small: on a long bracket it can exceed the misalignment you are trying to measure.

Measure it, do not estimate it. Clamp the bracket and indicator on a stiff piece of pipe at the same span you will use on the machine, zero the indicator at the twelve o'clock position, roll the pipe 180 degrees, and read the six o'clock value. That reading is your sag, and it is subtracted from every vertical rim reading taken with that bracket at that span. Change the span and you re-measure.

Skip it and every vertical number you take is wrong by the same amount in the same direction, which is the most dangerous kind of error because the readings stay self-consistent.

Step 4: Establish the sign convention on this setup, not from memory

Which way the needle moves for a shaft that is high depends on which way the indicator plunger faces and which shaft carries the bracket. Do not carry a convention from the last job.

Establish it once, on this setup: with the indicator at the twelve o'clock position and zeroed, lift the movable machine a known amount with a feeler or a shim under a foot, moving the machine with a bar and keeping fingers out from between the foot and the base, and watch the direction and size of the needle movement. Write down "movable shaft up equals needle plus" or the reverse, on the sheet, before you take a single real reading.

Also note the factor of two. A rim indicator sweeping from twelve to six crosses the shaft centerline, so it sees the offset twice: the total indicator reading at six o'clock, with the indicator zeroed at twelve, is twice the actual centerline offset. Half your rim TIR is the offset. Face readings carry no such factor; a face TIR divided by the diameter of the circle the indicator swept is the slope directly.

Step 5: Take four readings and make them close

Read at twelve, three, six and nine o'clock, rotating both shafts together so the indicator stays on the same spot of the target hub. Rotating one shaft alone measures hub runout, not alignment.

The validity check is arithmetic: the twelve plus six sum must equal the three plus nine sum, within a mil or two. They are two independent measurements of the same thing, the diameter of the sweep. If they do not close, stop and find the mechanical cause - a loose bracket clamp, an indicator tip not perpendicular to the surface, a rotor floating axially, a bearing with enough clearance to let the shaft lift during the sweep - rather than averaging your way past it. A set of readings that does not close will still produce a confident-looking shim calculation.

Step 6: Convert two stations into a slope and a position

With vertical offsets known at two axial stations, the angle is the difference in offset divided by the distance between the stations, and the position is whichever station you choose as your reference.

Watch the units when you compare to a tolerance. A slope of one mil per inch is a slope, and a coupling manufacturer may publish angular tolerance either as a slope in mils per inch or as a gap difference measured across a stated coupling diameter. Those two are only comparable when you know that diameter, because a gap difference is the slope multiplied by the diameter it was quoted across. Convert one into the other before comparing, and state which you are using.

Worked example: the reading sheet on a 6-inch station spacing

Pump is hard-piped and stationary. Motor is the movable machine. Reverse dial, two measurement planes 6.0 inches apart, bracket sag already measured and subtracted. Let x be inches measured from the near station toward the motor.

Angle first: 8 minus 2 is 6 mils over 6.0 inches, so the slope is 1.0 mil per inch. That is 0.001 inch per inch, which as an angle is 0.001 radian, or 0.057 degrees. Position at station 1 is 2 mils high.

The motor centerline follows y equals 2 plus 1.0 times x, in mils. Project it out to the feet:

• Front foot at x equals 10.0: y equals 2 plus 10.0, which is 12 mils high.
• Back foot at x equals 26.0: y equals 2 plus 26.0, which is 28 mils high.

So the correction is to remove 12 mils of shim from the front feet and 28 mils from the back feet. Check it: dropping the front 12 and the back 28 changes the slope by 28 minus 12 over 26.0 minus 10.0, which is 16 over 16, exactly 1.0 mil per inch of reduction, cancelling the measured slope, and the front foot lands at 12 minus 12 equals zero. The line is flat and on zero at every station.

Now the same reading treated as pure offset, which is what a one-plane measurement produces. Suppose the tech reads only station 2, sees 8 mils high, and pulls 8 mils from all four feet. Removing the same amount at every foot drops the machine without changing its slope, so the new line is y equals 1.0 times x minus 6. At station 2 that reads exactly zero and the indicator confirms it. At station 1 it reads minus 6 mils, the motor now low by 6, and the angle is still 1.0 mil per inch, completely untouched.

That is the failure mode, and it is worth stating plainly because it does not look like a failure: the indicator read zero at the plane the tech chose, the job was signed off, and the joint is now carrying six mils of offset and the entire original angle. The bearings see the same reaction force they saw before. The tell in the field is a set that reads good at the coupling and keeps failing bearings, and the way to catch it is that the correct correction is almost never the same number at the front and back feet.

What changes the answer. If the motor turns out to be the piped machine and the pump the movable one, every projection is done from the other end and the front and back foot numbers swap character. If the horizontal move you calculate exceeds the clearance between the hold-down bolt and its hole, you are bolt-bound and the answer is to relieve the hole or move the other machine, not to force it. And if the set is a hot machine, the numbers above are cold numbers and need a deliberate cold offset, which a sibling article covers.

How to verify you got this right

Re-read at both stations after the shims change, not just at the one you corrected to, and confirm two things: the position is inside tolerance and the slope between stations is inside tolerance, separately. Then re-run the twelve-plus-six against three-plus-nine check, because a bracket that shifted while you were under the machine will produce a clean-looking pair of numbers that are both wrong. Last, write both stations and the foot distances on the record. The next tech cannot repeat your projection without the distances, and a coupling reading with no station geometry attached is not a measurement anyone can reuse.

References

• 29 CFR 1910.147 for mechanical isolation, stored energy and rotor control; 29 CFR 1910.333(b)(2) with NFPA 70E-2021, 120.5, for the motor circuit.
• Coupling manufacturer documentation for the installation alignment tolerance at operating speed, and for whether their angular tolerance is published as a slope or as a gap difference across a stated diameter.
• See related: "Why a Flexible Coupling Does Not Fix Misalignment"; "How to Set Up a Shaft Alignment and Know It Held"; "Shimming, Leveling, and Alignment".