AXILINEWhy balancing does not help: 7 cases where vibration stays
You balance a fan, the software says "within tolerance," and the machine hums just as before. Or the vibration drops and comes back within a shift. That happens not because the balancing was done badly, but because the vibration was never coming from unbalance. Below are seven cases where weights are useless or harmful, and what to do about each one.
The first test: 1x against overall vibration
Balancing works with a single quantity: the running-speed component, 1x — the part of the signal at exactly the rotation frequency, once per turn. The instrument picks it out using the tachometer mark. Weights do not touch anything else in the spectrum.
That gives a simple rule. You measure overall vibration in mm/s RMS in the 10–1000 Hz band and check the 1x amplitude alongside it. If 1x accounts for most of the overall figure, balancing makes sense. If overall is several times higher, look for a different cause.
Do the arithmetic. A fan reads 9 mm/s overall and 2 mm/s at 1x. Even a perfect balancing job removes only those 2 mm/s, and the machine stays at around 8.8 mm/s. You spend a shift and get no result.
The second most important sign is phase repeatability (the angle showing where in the turn the vibration reaches its peak). Run the machine twice at the same speed. The 1x phase should match within a few degrees. If it wanders by tens of degrees, the system is nonlinear, and the correction-weight calculation will be unstable.
| What you see in the spectrum | Likely cause | Will balancing help |
|---|---|---|
| 1x dominates, everything else is small, phase repeats | Unbalance | Yes |
| Strong 2x alongside 1x, noticeable axial vibration | Shaft misalignment, soft foot | No, needs shaft alignment |
| A comb of 2x, 3x and higher, unstable phase | Loose mounting | No |
| A steady component at 0.42–0.48x | Oil whirl in a sleeve bearing | No |
| A component below running speed (subsynchronous), 50–80% of 1x, drifting in frequency | Rotating stall | No |
| A broad rise in the background ("pedestal") in the 1–10 kHz band with no distinct lines | Cavitation, an early-stage bearing defect | No |
| A 100 Hz line on a 50 Hz supply, sidebands around 1x | An electromagnetic cause | No |
A caveat the rule does not work well without: 1x does not rise only from unbalance. A bent shaft, shaft misalignment, pulley eccentricity and a soft foot also produce 1x. So one number is not enough. You need phase at both bearings, the axial direction, and how the reading behaves when the speed changes.
Case 1. Resonance in the structure, frame or pipework
Resonance amplifies vibration when the running speed lands on a natural frequency of the structure: the frame, the support, the housing, the platform, the pipework. The unbalance itself can be small while the vibration is huge, because the system multiplies any excitation at 1x.
Here is what happens if you balance in resonance. The influence coefficient — the machine's response to the trial weight, which the whole calculation is built on — drifts from run to run, the phase does not repeat, and the software gives different answers to similar data. Sometimes the vibration drops and comes back an hour later, because the temperature, the speed or the belt tension shifted.
- During start-up or coast-down, 1x amplitude shows a sharp, narrow peak over a small speed range, then drops off
- The 1x phase shifts by around 180° through the peak
- The frame, housing or pipework vibrate several times harder than the bearing housing
- The vibration changes when you simply lean a bar or a jack against the frame
- You have already balanced it once: the vibration dropped, then returned to the previous level
- Step 1
Find the natural frequencies
A bump test on the stopped machine gives repeatable candidate peaks. One response spectrum is not proof by itself: confirm it with a start-up, a coast-down, or measurements at several points on the structure.
- Step 2
Check whether the operating speed falls in the peak zone
Compare the rotation frequency in Hz (divide RPM by 60) with the natural frequencies you found. It is not only the running speed itself that is dangerous — 2x under shaft misalignment, or the blade-pass frequency, can be just as much of a risk.
- Step 3
Change the system, not the rotor mass
Stiffen the support with a brace or a rib, add or remove mass from the structure, fit a support roller on the belt span, brace the pipework with an extra support. On a machine with a VFD, sometimes shifting the operating speed by 5–10% is enough.
You can balance a machine that has a nearby resonance, but only at a steady speed outside the peak zone. And the result you get applies to that speed alone. You cannot promise quiet running across the whole range in that case.
Case 2. Oil whirl in sleeve bearings
On sleeve bearings, the oil wedge drags the shaft around, and the shaft centre starts moving in a circle — precessing — at roughly 0.42–0.48 of running speed. This is oil whirl: self-excitation of the shaft-and-oil-film system, not unbalance.
In the spectrum you see a steady, well-defined component just below half of running speed. Weights have no effect on it whatsoever, because it is not tied to the angular position of any mass. You can bring 1x down almost to zero and overall vibration will not move.
The causes of whirl are mechanical and operational: a lightly loaded bearing, excess clearance, oil viscosity that is too low, low temperature or pressure in the lubrication system.
There is one especially dangerous moment: when the whirl frequency during start-up coincides with the rotor's first critical speed. The whirl locks onto the resonance, its frequency pins to the critical speed and stops rising with speed. This is oil whip. Amplitudes grow fast, and the machine has to be shut down.
- Measure the bearing clearance and compare it against the drawing
- Check the oil's viscosity, temperature and pressure — the cause is often there
- Assess the load on the bearing; a lightly loaded bearing loses stability
- Check for whirl across different operating conditions: it appears and disappears along with the load
- If it keeps recurring, discuss switching to a more stable bearing-shell design with the manufacturer
Case 3. Pulley and belt drive: geometry and tension come first
Balancing a pulley that runs out means losing a day. A belt drive produces its own frequencies, and they do not match the pulleys' running speeds.
Belt frequency is always below running speed (subsynchronous). It is worked out as f_belt = π · D_pulley · f_pulley / L_belt, where D_pulley is the pulley diameter and L_belt is the belt length. Belt defects (a crack, uneven wear, a splice) show up at this frequency and its harmonics, most often the second harmonic, 2x belt. The belt span also has its own natural frequency and can resonate on its own.
There is a second trap. Worn grooves, run-out in the fit, an eccentric pulley and an over-tightened belt all produce vibration at 1x — exactly where you would mistake it for unbalance. You fit weights, the software shows "within tolerance," and a week later the vibration is back: the belt has worn the groove further.
A simple trick helps separate these causes. Remove the belt and turn the rotor over from the drive, or run it down freely. If the vibration pattern changes noticeably, the drive is the culprit.
- Check the pulleys' radial and face run-out with a dial indicator before balancing
- Inspect the grooves: uneven wear, signs of slippage, shiny bands
- Check the pulleys are parallel and the shafts aligned; pulley misalignment produces both 1x and faster belt wear
- Set the tension to the manufacturer's instructions, not by feel
- Replace belts as a matched set, not one at a time
- Only after that, measure again and decide whether balancing is needed
Case 4. A wheel that changes while it runs
Balancing assumes the mass distribution on the rotor stays constant. The moment the mass starts changing, the result lasts only until the end of the shift.
Build-up. On an extractor fan, a mulcher, a screw conveyor or a fan handling damp dust, material settles unevenly on the impeller. Vibration climbs over hours. You balance it in the morning, and by evening it is all back.
Erosion. An abrasive flow or cavitation strips metal from the blades unevenly. The process is slower than build-up, but it never stops, and balancing turns into a routine chore rather than a fix.
Cracks. This is the nastiest case. A crack in a blade, at the root or along a weld changes the rotor's stiffness, and vibration climbs gradually. If you mask it with weights, the machine keeps running with a growing crack until the wheel fails.
The mirror-image situation looks like this: a fan starts humming right after cleaning. Here balancing works fast and predictably — uneven cleaning has created a real unbalance, and 1x will dominate the spectrum.
Before balancing any wheel that works with abrasive material, dust or a damp environment, inspect it. Look for cracks at blade roots and along welds, uneven edge wear, flaking build-up, signs of rubbing against the housing. A cracked wheel must not be balanced. A wheel with build-up is not worth balancing until it has been fully cleaned, and it has to be cleaned evenly — partial cleaning creates unbalance of its own.
Case 5. The method does not fit the rotor or the machine
Sometimes the cause really is unbalance, but balancing still fails to deliver. Either the rotor needs more than one correction plane — the cross-section where the correction weights go — or the machine no longer holds its geometry, and any calculation based on influence coefficients comes out unreliable.
One plane where two are needed
Couple unbalance is two equal masses in different planes, 180° apart. The centre of mass sits on the axis, the rotor is statically balanced, but in rotation it produces a pure moment. One weight will not remove it: you bring down the vibration at one bearing and raise it at the other. A working rule of thumb by rotor shape: at L/D (the ratio of rotor length to diameter) below 0.5, a short disc-shaped rotor is often balanced in one plane; above 0.5, you need two. The sign of the mistake is obvious: after a single-plane correction, the second bearing gets worse.
Loose mounting
Looseness produces a comb of harmonics — 1x, 2x, 3x and on up — plus half- and sub-harmonics, a marked direction to the vibration, and an unstable phase. It is the unstable phase that kills balancing: the influence coefficient does not repeat. Check the torque on the foot bolts and bearing-housing bolts, cracks in the feet, the condition of the grout and foundation, wear in the fits.
Soft foot
One of the machine's feet does not sit flush on the foundation: a gap, a tilt, a distorted frame, dirt under the foot. Tightening that bolt twists the casing, and you get an induced misalignment and a rise in 1x and 2x. It is found by loosening each foot's bolts in turn while watching vibration or the gap, and corrected with calibrated shims. This has to be done before shaft alignment and before balancing.
Worn bearings
A worn-out bearing with excess clearance does not hold the shaft in one position. The response to the trial weight becomes nonlinear, the phase drifts, and the vibration comes back after correction. On top of that, the defect itself produces frequencies that are not multiples of running speed, plus high-frequency noise. Replace the bearing, and balance afterwards — the influence coefficients will change anyway.
Shaft misalignment
Strong 2x alongside 1x, plus noticeable axial vibration, is the classic sign of shaft misalignment. Parallel misalignment gives more radial 2x; angular misalignment gives pronounced axial 1x and 2x. Balancing does not cure it. It needs shaft alignment that accounts for thermal growth: the cold offsets are set so the shafts come into line once the machine reaches operating temperature.
Case 6. This is not unbalance at all
There is a class of problems where balancing does not "work poorly" — it simply does not apply. Here the question is not the quality of the weights, but the physics of the source.
Cavitation in a pump
A broad pedestal in the spectrum from roughly 1 to 10 kHz with no distinct lines, sometimes modulated by the blade-pass frequency, BPF = z · f_run, where z is the number of blades. This is an operating condition, not a rotor defect. The fix lies in the hydraulics: inlet head, valve position, fluid temperature and level, air in the system. Cavitation also erodes the impeller, so unbalance can turn up later as a consequence.
Aerodynamics and flow separation
Rotating stall produces a subsynchronous component at 50–80% of 1x, unstable in frequency. A rise in blade-pass frequency points to an uneven clearance or an interaction between the flow and the structure. Here you change the operating condition, not the mass: the damper, the guide vanes, the clearance to the housing, the condition of the inlet cone (the narrowing section ahead of the wheel).
An electrical cause
A 100 Hz component on a 50 Hz supply (2×LF, twice the mains frequency) points to air-gap eccentricity or a loose stator. Sidebands around 1x spaced at 2 · s · LF (s is slip, LF is the mains frequency) give away broken rotor bars. The on-site check is simple: cut the power and watch the vibration during coast-down. The electromagnetic component disappears instantly; the mechanical one decays along with the speed.
Reciprocating machines and crankshafts
A reciprocating compressor or an internal combustion engine has a spectrum that is inherently full of running-speed harmonics. These are reciprocating forces, normal for that type of machine. Levels that are normal for an engine will easily land in the "bad" zones C or D on tables built for rotating machinery. A crankshaft is not balanced on site from vibration readings: there are no accessible correction planes, and the measured 1x does not reflect residual unbalance. That kind of shaft gets balanced on a balancing machine, removed from the engine.
Order of action: measure first, decide after
All these cases come down to one idea: the decision gets made after the measurement. Below is a sequence that covers most situations in a single visit.
- 01
Inspection and fasteners
Walk round the machine. Torque on the foot and bearing-housing bolts, cracks, the condition of the frame and grout, clearances, the belt and pulleys, how clean the wheel is, signs of oil or rubbing. Some cases get resolved right here, without any instruments.
- 02
Measure overall vibration
mm/s RMS in the 10–1000 Hz band, at the bearing housings, in three directions, sensor on a rigid mount: a stud or a magnet on a clean, flat spot. Record the point and direction so the next measurement is comparable with this one.
- 03
1x, phase and spectrum
Compare 1x with overall. Read the spectrum: harmonics, half-orders (0.5x, 1.5x), components below running speed, a high-frequency "pedestal," mains-frequency lines. Compare the phase at both bearings — that is what separates unbalance from shaft misalignment and from couple unbalance.
- 04
Check for resonance
A controlled start-up and coast-down with 1x amplitude and phase recorded. A bump test on the stopped machine if you suspect the frame, the housing, the platform or the pipework.
- 05
Decision
Unbalance — we balance. Shaft misalignment — we align the shafts. Looseness — we tighten and repair. Resonance — we change the structure's stiffness or the speed. A bearing — we replace it. Hydraulics or aerodynamics — we work on the operating condition.
- 06
Balancing and verification
If it is unbalance, we balance in the machine's own bearings at operating speed. The trial weight has to change the 1x amplitude by 20–30% or the phase by 20–30°, or the influence coefficient will come out unreliable. After that comes a check run and a repeat measurement of overall vibration, with a before-and-after report.
The numbers and grades in this text are working guidance. Check which part and edition of the standard applies to your specific machine: overall-vibration condition is assessed against ISO 20816 (formerly ISO 10816), balance quality against ISO 21940-11 (formerly ISO 1940-1), flexible rotors against ISO 21940-12, and reciprocating machines against their own separate parts. For contractual acceptance, record the standard's part and edition, the measurement points, the frequency band, the operating condition and the bearing type.
Sources: ISO 21940-11:2016 · ISO 20816-1:2016 · ISO 281:2007 · ISO 21940-12:2016 · ISO 13373-3:2015 · ISO 13373-5:2020 · Balanset-1A operation manual · Balanset-1A manufacturer specification
What AXILINE brings, and what you get
We are engineers who design and manufacture the Balanset instruments and use them to balance on site ourselves. So the first thing we do at your site is not weld on a weight — it is measure. If 1x turns out to be a small part of the overall reading, or the phase does not repeat between runs, you will hear that straight away.
The Balanset-1A takes in two channels at once and displays overall vibration, 1x amplitude and phase, and the FFT spectrum, and it saves the signals and spectra to an archive for the report. The kit includes two accelerometers, a laser phase-and-speed sensor, a two-channel USB module with preamplifiers and an ADC, and Windows software.
When the cause really is unbalance, we balance in one or two planes, split the weight across fixed positions (blades or holes), and if there is nowhere to weld, we calculate a drilling correction instead. We work out the tolerance against the G grades, add weights (trim balancing) if we do not land within tolerance on the first pass, and save the influence coefficients for repeat visits.
We keep the outlook on results honest. A typical fan at around 12 mm/s, where the vibration genuinely comes from unbalance, usually comes down to 1.5–2 mm/s. If unbalance only accounts for a small share of the overall vibration, no one — us included — will get you that number, and we will say so in the report, along with where to look next.
You can also buy the instrument: the Balanset-1A is used for on-site balancing in a machine's own bearings, and as a measuring system for soft-bearing balancing machines. It comes with consultation support, including going through your own spectra and readings with you.
Frequently asked questions
I balanced a fan, the vibration dropped by 20% and stopped there. What does that mean?
Most likely, unbalance was only part of the vibration. Compare overall vibration with 1x after balancing: if 1x is now small but overall has barely changed, you removed your share and something else is producing the rest. Look at 2x and axial vibration (shaft misalignment), a comb of harmonics with an unstable phase (looseness), subsynchronous components, and high-frequency noise (bearings, hydraulics).
How can I tell quickly whether it is resonance, without special equipment?
Change the speed. On a machine with a VFD, sweep the range and watch the 1x amplitude: a sharp, narrow peak with a phase shift of around 180° points to resonance. If the speed is fixed, take the vibration during coast-down instead. There is also a rough tell: the frame, housing or pipework vibrate several times harder than the bearing housing.
Vibration rose right after cleaning the impeller. Should I balance it or look for something else?
This is usually exactly the case where balancing fits: uneven cleaning creates a real unbalance, and 1x will dominate. First make sure the build-up was removed completely and evenly, and that the blades are sound. There is no point balancing a wheel with build-up still left on it — the mass will keep changing while it runs.
Can you balance a rotor on sleeve bearings?
Yes, if the vibration genuinely comes from unbalance and 1x dominates. Check for a steady component at 0.42–0.48 of running speed. If it is there, that is oil whirl, and weights will not help. Work on the bearing clearance, the oil's viscosity, temperature and pressure, and the load on the bearing instead.
Why can I not just use one plane and save a trial run?
If the rotor has couple unbalance, one plane will not remove it: you improve one bearing and make the other worse. A rule of thumb by shape: L/D below 0.5 often allows single-plane work, L/D above 0.5 needs two. One plane means two runs (the baseline and the trial); two planes need three. That is a difference of one run, but the difference in the result is fundamental.
What do you do if it turns out on site that balancing will not help?
We measure overall vibration, 1x with phase, the spectrum, and how the reading behaves when the speed changes, and then we tell you plainly and give you a finding based on the measurements, with pointers on where to look next. We do not start balancing in that situation: it would not change the overall vibration and would only use up your downtime.
Related content
Safety during on-site balancing: start-up, lockout, weights, and the throw zone
On-site balancing is work on rotating equipment, and safety here rests on three decisions. Your own person in charge of start-up and shutdown commands the machine, and the technician never touches the controls under any circumstances. While people are in the zone, the drive is de-energized and locked out with a lock and tag - not just switched to 'stop'; trial and correction weights are mounted with a bolt and lock nut or a full weld, never with glue, tape, or a magnet. Before every run, everyone clears the plane of rotation, and any abnormal sign during run-up means an immediate stop, not 'let's finish the measurement first.'
Rotor unbalance: what it is and why it is dangerous
Rotor unbalance is a mismatch between the rotor's principal central axis of inertia and the axis of rotation set by the bearings. In plain terms: the mass is distributed asymmetrically, and rotation produces an uncompensated centrifugal force that sweeps around the support once every turn. The force equals the unbalance multiplied by the square of the angular speed, so doubling the speed means quadrupling the force. Unbalance is dangerous not because it breaks the machine outright, but because of the cyclic loading: bearing life falls by a large factor, fatigue cracks grow in welds and the frame, fasteners work loose, and machining accuracy drifts.
On-site balancing of centrifugal fans: single-inlet and double-inlet
Yes, we balance centrifugal fans right on site, both single-inlet and double-inlet wheels. The rotor turns in its own bearing housings, and there's no need to remove the wheel or take the volute apart. Three conditions apply. The machine has to hold a stable speed at one damper setting. The correction plane — the spot on the wheel where balancing weights go — has to be reachable: an inspection hatch in the volute, a removable inlet cone, or an open inlet pocket, and on a double-inlet wheel both sides have to open. And most of the vibration has to come from the 1x component (vibration exactly at rotating speed — the signature of unbalance), not from bearings, a worn pulley, shaft misalignment, or pedestal resonance. We work out access and regime from your photos at the request stage, we measure the 1x share ourselves in the first half hour on site, and we tell you plainly if weights won't help here.
Describe your equipment and the problem
We'll answer your questions, clarify the details, and let you know what's needed for an estimate and a visit.