What Balance Actually Means on a Rotating Part

Why this matters

"It needs balancing" gets used as a verdict on anything that shakes, and half the time it is wrong. Balance is a specific, measurable property of how mass sits around a shaft centerline, and it is not the same thing as how much the machine moves. A rotor can be inside its balance tolerance and still shake the building, and a rotor can be badly out of balance and barely move if it is bolted to something stiff enough. If you cannot say what quantity a balance job actually corrects, you cannot tell a customer why the balancing service came back with a clean report on a wheel that still shakes, and you will end up paying for a second one.

Before you put a hand or an indicator on the wheel

Open the disconnect, apply your own lock and tag, and then confirm the rotor has actually stopped rather than slowed. A fan wheel on good bearings coasts for minutes after the contactor drops, and that coasting mass is exactly the stored mechanical energy 29 CFR 1910.147 exists to control, including its verification-of-isolation step before anyone reaches in. Do not wear gloves or loose sleeves while reaching near a shaft or coupling even when it is stopped, because a glove caught by a rotor someone else energizes is a far worse outcome than the scrape it was preventing. If you need to open the motor terminal box or a drive enclosure as part of the same call, that is electrical work under 29 CFR 1910.333(b)(2) rather than 1910.147, which excludes electric utilization installations at 1910.147(a)(1)(ii)(C); prove the conductors dead using the live-dead-live sequence in NFPA 70E-2021, 120.5.

Unbalance is a product, not a weight

The quantity a balance job corrects is unbalance: a mass multiplied by the radius it sits at, pointing in a specific angular direction. It is written in ounce-inches or gram-millimeters, and both halves of that unit carry equal weight in the arithmetic.

That means 4 grams sitting 250 mm out from the centerline (1,000 g-mm) and 8 grams sitting 125 mm out (also 1,000 g-mm) are the identical fault. A tech who hears "you need 4 grams" and adds 4 grams at whatever radius the bolt hole happens to be has corrected the wrong amount. The correction weight and the correction radius travel together or neither number means anything.

Unbalance also has an angle. A rotor is not "3 grams out", it is 3 grams out at a particular clock position, and adding the right mass 90 degrees from where it belongs makes the rotor worse than leaving it alone. This is why a genuine balance job needs a phase reference (a reflective mark and a tachometer, or an accelerometer plus a trigger), not just an amplitude reading. Amplitude alone tells you something is wrong; it does not tell you where to put the weight.

What the balance specification deliberately does not cover

This is the part that costs shops money, so read it as a list of things a clean balance report is not claiming:

• It does not say the machine will not shake. A balance tolerance is a permissible mass-radius product on the rotor. How much the machine actually moves also depends on how stiff the bearings, base and structure are, and whether a natural frequency of that structure sits near running speed. See the companion articles on resonance and on how a base changes the machine standing on it.
• It does not cover the shaft, the coupling, or the sheave. A wheel balanced on an arbor in a shop is balanced by itself. Put it back on a bent shaft, a coupling half with a keyway that is not filled correctly, or a sheave with its own eccentricity, and the assembly is out of balance again while the wheel's report stays true.
• It does not survive a change of operating condition. Dust build-up on one side of a wheel, a chunk of ice thrown off, one blade eroded thinner than the others, a set screw backed out, or a corroded section that finally sheds all change mass distribution. A wheel handling dirty air is a balance job with an expiry date.
• It is not a statement about misalignment. Those are separate faults with separate mechanics. Cross-check with the article on telling imbalance from misalignment before you order either correction.
• It does not mean zero force. Every tolerance is a permissible residual, and residual unbalance at speed is a real rotating load on the bearings. The worked example below puts a number on it.

Static, couple and dynamic: how many planes the fault lives in

Where the extra mass sits along the length of the rotor decides how many correction planes you need, and this is a geometry question, not a severity question.

Static unbalance is one heavy spot, with the rotor's mass centerline displaced parallel to the shaft centerline. Rest the rotor on knife edges and it rolls the heavy spot to the bottom every time. One correction plane fixes it.

Couple unbalance is two equal heavy spots on opposite sides of the rotor at opposite ends. The mass centerline is tilted rather than displaced. It rolls to no preferred position on knife edges, so a static check calls it good, and it still produces a rocking moment at speed that hammers both bearings in opposite directions. It needs two correction planes, because correcting at one end alone just converts it into a different two-plane fault.

Dynamic unbalance is a combination of both, which is what almost every real rotor has.

The practical fork: a narrow rotor, meaning one whose width is small compared with its diameter (a single-width fan wheel, a thin pulley, a grinding wheel), carries mostly static unbalance and single-plane correction usually resolves it. A rotor whose length is a meaningful fraction of its diameter, or one with two wheels on a shaft, needs two planes. That rule of thumb is about the ratio of the rotor's axial length to its diameter and about running speed, and the crossover is not a hard line, so where a wheel sits near the boundary the manufacturer's own service data decides, not the rule of thumb.

Working a tolerance all the way through

Take a fan wheel with a rotor mass of 40 kg turning at 1,750 rpm, and suppose the manufacturer calls out balance quality grade G6.3, a common grade for general fans and pump impellers under ISO 21940-11 (the standard that replaced ISO 1940-1). The grade number is the permissible specific unbalance multiplied by angular velocity, expressed in mm/s, so it converts to a per-kilogram allowance with:

permissible specific unbalance (g-mm per kg) = 9549 x G / speed in rpm

For this wheel: 9549 x 6.3 / 1750 = 34.4 g-mm per kg.

Multiply by the 40 kg rotor mass: 1,376 g-mm total permissible residual unbalance.

Now translate that into something you can act on. The wheel has two correction planes and its center of gravity sits close to midway between them, so the total splits roughly evenly, about 688 g-mm per plane. (That even split holds for a rotor whose mass is centered between the planes; on an overhung wheel the allocation shifts toward the plane nearer the mass and the balancing house calculates it rather than halving it.) Correction weights go on a rim at a radius of 250 mm, so:

688 g-mm / 250 mm = 2.8 grams per plane.

Under 3 grams on a wheel that weighs about 88 pounds. That is the number that surprises people, and it is why field correction with whatever washers are in the truck rarely lands inside a published grade.

Now change one thing. Suppose the only place you can physically attach a weight is an inner hub ring at 125 mm rather than the rim at 250 mm. The permissible unbalance has not changed, so the mass doubles: 688 / 125 = 5.5 grams per plane, and the resolution of your correction has halved too, because each gram you add or remove now moves the rotor half as far. Correcting at a small radius makes both the target and the error bigger.

Finally, the part that connects balance to bearing life. That 1,376 g-mm residual is about 1.9 ounce-inches. For a rigid rotor treated as a single point mass in one plane, the rotating force it produces is roughly 1.78 x unbalance in ounce-inches x (rpm/1000) squared, in pounds:

1.78 x 1.9 x (1.75 x 1.75) = about 10 pounds of force, rotating with the shaft, 1,750 times a minute, forever.

That is a rotor inside its tolerance. The bearings are carrying a 10 lb rotating load on top of the actual work the machine does, by design and by agreement. A rotor five times out of tolerance is carrying 50 lb, which is why "it is only a few grams" is a bad instinct. The force law behind that multiplication, and what happens when you change speed, is covered in the companion article on how a small imbalance becomes a large force.

What to do when the numbers say the wheel is fine

If a balance report comes back inside grade and the machine still shakes, do not order a second balance. The chain of custody for the fault has moved off the rotor, and there are only a few places left it can be: the shaft and its runout, the mounting of the wheel to the shaft, the coupling or sheave, an alignment condition, the bearings, or the structure the whole assembly stands on amplifying a force that was always there. Cost the structural check first, because it is the cheapest to rule in or out with an impact test and it is the one most often skipped.

One field note on correction method. Adding mass is nearly always the right field call. Removing mass by grinding is a shop operation, not a truck one, and grinding a coated, galvanized or primed wheel releases metal-bearing dust and fume at the point of the wheel, so it needs local exhaust plus respiratory protection under a written 29 CFR 1910.134 program, with hexavalent chromium exposure assessed under 29 CFR 1910.1026 wherever a chromate primer is suspected. A glove and safety glasses do not address an airborne route.

How to verify you got this right

Ask yourself three questions before you accept any balance number, whether it came from a service or from your own trial-weight work.

1. Does every mass in the conversation have a radius attached? If a report, a work order or a tech's verbal handoff names grams without naming the radius they go at, the number is incomplete and cannot be executed.
2. Does the correction plane count match the rotor's geometry? A single-plane correction on a rotor with meaningful axial length can leave a couple unbalance that the single-plane reading never saw. If a one-plane correction reduces the shake at one bearing and raises it at the other, that is the tell.
3. Was the assembly measured, or the part? A wheel balanced off the machine and a wheel balanced in place on its own shaft and bearings are different tests. The in-place result is the one that matches what the customer experiences.

References

• ISO 21940-11 (mechanical vibration, rotor balancing, balance tolerances; replaced ISO 1940-1) for balance quality grades and the permissible-residual formula used above
• 29 CFR 1910.147, control of hazardous energy, for the stored-rotational-energy isolation and verification requirement before reaching into a rotor
• 29 CFR 1910.333(b)(2) and NFPA 70E-2021, 120.5 for the electrical fork when the same call includes terminal-box or drive work
• 29 CFR 1910.134 and 29 CFR 1910.1026 for respiratory protection and hexavalent chromium exposure where a coated wheel is abraded
• See related: How a Small Imbalance Becomes a Large Force; How to Tell Imbalance From Misalignment Mechanically; How a Base and a Foundation Change the Machine on Them