# Balancing compressors, blowers and gas blowers at the point of operation

> A compressor machine is almost always high-speed, often sits behind a step-up gearbox, and almost never gives access to the rotor without stripping it down. So the answer here doesn't start with weights, it starts with a measurement: we separate the once-per-turn component — vibration at the rotation speed, which unbalance creates — from oil whirl, blade-pass frequency and the gearbox's tooth-mesh frequencies. Part of the compressor fleet we balance right on site, in the machine's own bearings. Part we send honestly to a workshop, and we say so before the visit, not after.

**In short:** Yes, but not every piece of compressor equipment. On site we balance whatever gives access to a correction plane: open impellers on blowers, air-blower units and gas blowers, overhung impellers on single-stage centrifugal compressors, pulleys, flywheels, half-couplings, compressor shafts and drive-motor rotors. Screw-compressor rotors, multistage rotors in a closed casing, and high-speed turbocompressor rotors don't get balanced on site — those need a workshop and a balancing rig. Before we reach for weights, we always prove with a measurement that unbalance is genuinely what's producing the vibration.

Source: https://axiline.pt/en/equipment/balancing-compressors-blowers-gas-blowers-at/  
Publisher: AXILINE · Vila Nova de Gaia, Portugal · +351 931 831 229 · axilinegeral@gmail.com

## Symptoms: when it's time to measure the compressor, not put up with it

A compressor machine rarely dies suddenly. First the level rises at the bearing supports, then oil appears on the seal housing, then the frame's gusset plates crack and the coupling's flexible insert tears. If the measurement is taken in time, you have weeks for proper planning instead of one night for an emergency swap.

Below are the symptoms where a measurement is already worthwhile. Not all of them mean unbalance, and that's exactly what we sort out on site.

- [x] Vibration at the bearing supports has risen to one and a half times its usual level or more, even if you're technically still inside the allowable zone.
- [x] Humming and runout appeared right after cleaning, flushing or replacing the blower impeller.
- [x] Vibration changes together with discharge pressure or valve position: the flow is part of the picture too.
- [x] The same bearing support has failed a second and a third time.
- [x] The compressor unit is rocking the frame, and vibration is travelling down the pipework into the workshop floor above.
- [x] The screw block's sound has changed, a whine has appeared, and the coastdown is different.
- [x] A belt has torn, a coupling's flexible insert has broken up, or the foundation bolts have worked loose.
- [x] Oil mist on the casing, darkened oil, marks on the shaft from the seal working.

> Note down which condition gives the higher vibration. An observation like 'shakes harder at full pressure' or 'only shakes on a cold machine' cuts the on-site diagnostics down considerably, because it rules out whole groups of causes straight away.

## What compressor equipment we balance

The compressor fleet isn't uniform. Within a single installation there are rotors you can reach in twenty minutes, and rotors you can't reach without fully stripping the casing. We group equipment by access to a correction plane, not by what the machine is called.

### Blowers, air-blower units, gas blowers

The most productive group for on-site balancing. Gas-blower rotors and blower impellers pick up unbalance from product build-up, abrasive-dust erosion, uneven blade wear and weld repairs. The impeller is usually overhung and reachable through a hatch or with the cowling off. A belt drive adds subsynchronous components of its own, and those need separating out before balancing.

### Centrifugal compressor impellers

We balance a single-stage machine with an open or semi-open overhung impeller on site if the hatch gives access to the disc or hub. A multistage rotor in a closed casing doesn't get balanced on site: you have neither access to the planes nor the right to touch the impellers without the manufacturer's (OEM) protocol. A rotor like that is balanced assembled, on a rig.

### Compressor rotors and compressor shafts

Drive and intermediate shafts, half-couplings, pulleys, flywheels, the drive motor's rotor, the compressor unit's cooling fan. These are the most accessible correction planes on the unit, and quite often the source of a high once-per-turn component sits here rather than in the compressor's impeller.

### Screw-compressor rotors

A separate case, and almost always a workshop job. The male and female screw rotors run in a casing with clearances of a few tenths of a millimetre, there's no access to them, and metal must not be removed from the profile. High vibration on a screw block more often means worn bearings or enlarged clearances, not unbalance. On site we run diagnostics and balance whatever is accessible from outside.

### Turbocompressors and rotors behind a step-up gearbox

Speed on the high-speed shaft runs into the tens of thousands, the rotor is usually flexible and works above the first critical speed (the speed at which the rotor goes into resonance). Rotors like these are balanced on high-speed balancing rigs in a specialist workshop. On site we measure, separate out the gearbox frequencies, and balance the low-speed end, the fan and the pulley.

### Vacuum-unit rotors

Centrifugal vacuum pumps and vacuum blowers with an external impeller are balanced on site the same way as gas blowers. Liquid-ring and rotary-vane machines with the rotor inside the casing can't be balanced on site: the correction planes aren't accessible, and the vibration there more often comes from wear, cavitation and liquid in the working chamber.

> A reciprocating compressor doesn't fit this logic. There, high vibration at the rotation frequency and its harmonics is produced by the reciprocating masses and the gas forces. Balancing the rotor doesn't remove that vibration; on site we can only work with the flywheel, the pulley and the cooling fan.

## What we check before balancing on compressor units specifically

We fit two accelerometers to the bearing supports and set the optical tachometer to read a reflective marker on the shaft we're about to balance. This matters: on a unit with a step-up gearbox, two shafts turn at different speeds, and a marker on the wrong shaft gives a meaningless once-per-turn component.

Next we compare overall vibration with the once-per-turn level. If the overall level is several times higher than 1x, weights will only remove a small part of the problem, and we'll say so before we start any trial runs. The logic behind this comparison is covered in detail in the article on overall vibration, the once-per-turn component and phase.

On machines with sleeve bearings we pay particularly close attention to the subsynchronous part of the spectrum — frequencies below running speed. Oil whirl produces a peak at roughly 0.4 to 0.5 of the rotation frequency, and that's instability in the oil film, not unbalance. The causes lie in clearance, viscosity, oil temperature and pressure, and an underloaded bearing. Balancing a machine like that is pointless and sometimes dangerous, because it masks real deterioration of the bearing.

We check for misalignment. On compressor units it turns up after every repair, after grouting a foundation, and after thermal growth of the casing on a warmed-up machine. The signs are the usual ones: a noticeable second harmonic, strong axial vibration, a phase shift (the angle showing where the vibration peak sits in the turn) of around 180 degrees across the coupling. It's fixed with shaft alignment and clearing up soft foot, not with weights.

We check the running condition and temperature separately. A compressor is hot on the discharge side, the shaft develops thermal bow, and a cold machine gives different figures from a warmed-up one. We always measure at a warmed-up, steady-state condition and record it in the report, otherwise the before-and-after readings end up comparing two different machines.

- Fixings and stiffness: foundation bolts, soft foot, the condition of the anti-vibration mounts, cracks in the frame.
- The fit of the impeller, pulley and half-coupling on the shaft: a worn-loose fit gives an unstable 1x from run to run.
- The belt drive: tension, pulley runout, belt condition and belt-related frequencies.
- The step-up gearbox and gearing: a tooth-mesh frequency with sidebands points to wear, not unbalance.
- Rolling-element bearings: peaks at high frequencies not related to running speed by a whole number, and a rising crest factor (the ratio of the peak level in the signal to its average level).
- Resonance of the rotor, the frame or the discharge pipework: a sharp, narrow peak and a shifting phase as speed changes.
- The process condition: flow, pressure, valve position, recirculation, the condition of the inlet filter.

> Here, doing things in order matters more than speed. Balancing a unit with loose feet, or with oil whirl in a bearing, eats up a shift and gives no result, because you're fighting the wrong cause.

Sources: [ISO 13373-3:2015](https://www.iso.org/standard/40840.html) · [ISO 281:2007](https://www.iso.org/standard/38102.html)

## Frequencies on a compressor machine: what's what

| What the measurement shows | What it usually means | What we do |
| --- | --- | --- |
| A peak at the rotation frequency dominates, phase is stable from run to run | Rotor unbalance | On-site balancing, in the machine's own bearings |
| A subsynchronous peak at roughly 0.4 to 0.5 of the rotation frequency on a machine with sleeve bearings | Oil whirl: instability in the oil film, clearance, viscosity, an underloaded bearing | We don't balance. We work through clearance, oil temperature and pressure together with your maintenance team |
| A noticeable second harmonic plus strong axial vibration across the coupling | Misalignment or soft foot | Shaft alignment and the supports get corrected, we repeat the measurement afterwards |
| A peak at the blade count multiplied by speed, and its harmonics | Blade-pass frequency and gas pulsation, sometimes acoustic resonance in the header | Weights won't help. We look at the process condition, the impeller-to-casing clearance, the pipework and the pipe supports |
| A peak at the gearbox's tooth-mesh frequency with sidebands | Wear or a defect in the step-up gearbox | Gearbox diagnostics. We handle balancing separately, shaft by shaft |
| Peaks at high frequencies not related to running speed by a whole number, crest factor rising | A defect in the rolling-element bearings | Bearing replacement, balancing after the repair |
| Amplitude rises sharply within a narrow speed range, phase shifts | Resonance of the rotor, the frame or the pipework | We change the operating condition or the stiffness. In resonance, a balancing result won't hold |

> A compressor unit usually has more than one shaft turning, so we tie every line in the spectrum to a specific rotation frequency. Without the recorded speed and a marker on the right shaft, this table doesn't work.

Sources: [ISO 13373-3:2015](https://www.iso.org/standard/40840.html) · [ISO 13373-5:2020](https://www.iso.org/standard/62202.html)

## How the work goes on site

1. **Inspection, access and the limits of intervention** — We check hatches, cowlings, the fitted balancing points and the manufacturer's requirements. We agree straight away on what can be touched: only the fitted threaded holes and balancing screws, or whether removing metal is allowed. If the manufacturer forbids interference with the impeller, we work with the planes that are accessible and say so up front.
2. **Baseline measurement at the warmed-up condition** — Two accelerometers on the bearing supports, mounted on a magnet on a cleaned, flat spot, radial-horizontal direction, the same for every reading. The optical tachometer reads the reflective marker. We record overall vibration, 1x, phase, speed, spectrum and time waveform at every point.
3. **Separating out the causes** — We compare 1x with the overall level, check the subsynchronous range, the second harmonic, the blade-pass and tooth-mesh frequencies, and the bearings' high-frequency zone. If unbalance isn't confirmed, we stop here and hand over a finding on the cause instead of running pointless trial weights.
4. **Trial mass** — We fit a trial weight in the first correction plane, then, for a two-plane job, in the second. We treat a run as usable if the 1x amplitude changes by at least twenty percent, or the phase by at least twenty degrees. If there's no response, we increase the mass and repeat, rather than work out an influence coefficient (the machine's response to a trial weight) from noise.
5. **Correction and fitting the weights** — The software gives the mass and angle, or a fixed-position number against the disc's fitted holes. On blower impellers we fit weights into threaded holes or remove metal from the hub rim by a drilling calculation. On pulleys, flywheels and half-couplings we use balancing bolts. In a hot gas path we don't use bonded or tack-welded weights: a weight coming loose isn't acceptable.
6. **Check run and trim** — We run the machine at the same condition and the same temperature. If the level has come down but the target hasn't been reached, the software works out an add-on weight for the ones already fitted. We save the influence coefficients so the next trim balance can be done without repeating the trial runs.
7. **Report and recommendations** — We record the before-and-after figures for each support, the positions and masses of the weights, the running condition, and the measurement points and directions. We list separately what balancing doesn't cure: clearances, bearing condition, pipework, fixings.

> One correction plane needs a baseline run and one trial run; two planes need a baseline and two trial runs, plus a check run. On a compressor machine, every run means a shutdown, depressurising and opening a hatch, so the real schedule is set by access, not by the measurements.

Sources: [Balanset-1A operation manual](https://vibromera.eu/balanset-1a-operation-manual/)

## One plane or two: how it's decided on compressor rotors

The rotor's geometry and speed set the number of planes. The rule of thumb is simple: work out the length-to-diameter ratio in the zone where the weight goes. A short, disc-shaped impeller behaves as a disc, and one mass is usually enough. An elongated rotor almost certainly has a couple unbalance, and one mass won't remove the vibration at both supports at once.

The higher the speed, the more often two planes are needed. Compressor equipment is exactly the high-speed kind, so in practice we choose a two-plane scheme wherever two planes are accessible. A detailed look at the rule and its exceptions is in the article on choosing the number of correction planes.

- One plane: the overhung impeller of a single-stage blower, a pulley, a flywheel, a half-coupling, a cooling fan, the short impeller of a vacuum unit.
- Two planes: a long gas-blower rotor, a rotor with impellers at both ends of the shaft, the drive motor's rotor, a compressor shaft with two fitting locations.
- Two planes regardless of geometry: if the level at the second support stays high after a one-plane correction.
- Workshop only: a flexible rotor working above the first critical speed. A rotor like that needs modal balancing, not two planes on site.

> Sometimes only one plane is physically accessible on site, even though the rotor calls for two. We do a one-plane correction and state plainly in the report that a residual couple unbalance remains, along with the level we managed to reach. That's more honest than promising a complete result.

Sources: [ISO 21940-11:2016](https://www.iso.org/standard/54074.html) · [ISO 21940-12:2016](https://www.iso.org/standard/50429.html)

## What gets in the way, and when it won't work on site

We prefer to say this before the visit. Below are the cases where on-site balancing is either physically impossible or won't solve your problem.

Screw rotors head this list. The rotor profile must not be drilled or have weight added, the casing gives no access, and the clearances between the rotors rule out any improvisation. If a screw block vibrates, we measure and say where the cause lies: bearings, clearances, the fit, the drive. Screw-rotor balancing is done in a workshop, on a rig, with the block fully stripped and the bearings replaced.

Multistage centrifugal-compressor rotors and turbocompressor rotors are workshop jobs too. The first are enclosed by the casing and have no accessible planes; the second run at speeds that call for a high-speed rig and a vacuum chamber. On site we're useful for diagnostics and for balancing the peripheral components.

The manufacturer's requirements constrain us for real, not just on paper. On many impellers only the fitted balancing points are allowed, welding is forbidden because of the material or an explosive atmosphere, and interfering with the impeller's geometry voids your warranty. We work within these limits and ask for written agreement if metal removal is needed.

- No access to the correction plane: an enclosed casing, no hatch, the rotor assembled with its seals in place.
- The rotor is already deformed: a bent shaft, a blade torn off or badly eroded, a wrecked hub. Repair or replacement comes first.
- The machine runs in a resonance zone: the phase jumps from run to run, the result doesn't repeat.
- Speed can't be held stable, or the machine can't be brought to its working condition for a check run.
- The cause of the vibration isn't unbalance: oil whirl, blade pulsation, gearbox wear, cavitation, or liquid in a vacuum unit's working chamber.
- Fouling keeps happening: if the build-up comes back within a month, balancing only buys time, and the source of the fouling needs solving.
- An explosive zone or a hot environment where welding, adhesives and safe access during operation are all off the table.

> When on-site balancing isn't the right fit, we don't disappear. You get a finding from the measurement, an explanation of the cause, and a clear next step: a workshop, bearing replacement, shaft alignment, or work on the pipework. A rundown of typical cases where balancing doesn't help is in a separate article.

Sources: [ISO 21940-12:2016](https://www.iso.org/standard/50429.html)

## What you get in hand

The result of the work isn't just a quieter machine, it's a document you can put in the repair file and show to management.

We keep the condition assessment separate from the balancing result. The software reports 'within tolerance' when the residual once-per-turn component is below the target value you set. The unit's overall condition is assessed against overall vibration and the zones of the applicable part of ISO 20816, while the quality of the rotor balance is assessed against residual unbalance and an ISO 21940 balance quality grade. These are three different tolerances, and we don't mix them up. We fix the exact part and edition of the standard for your type of machine, including power, speed and bearing type.

- A report with before-and-after figures for each bearing support: overall vibration and the once-per-turn component, phase, speed.
- The measurement conditions: points, directions, frequency band, running condition and machine temperature.
- The masses and exact locations of the weights fitted, or the amount of metal removed.
- Spectra and time waveform at each point, for comparison at the next service.
- This rotor's influence coefficients, saved: the next trim balance will do without repeated trial runs.
- A separate list of what balancing doesn't cure, prioritised by urgency.
- A recommended baseline level for ongoing vibration monitoring on your route.

Sources: [ISO 20816-1:2016](https://www.iso.org/standard/63180.html) · [ISO 21940-11:2016](https://www.iso.org/standard/54074.html) · [Balanset-1A manufacturer specification](https://vibromera.eu/product/balanset-1/)

## Price and how to book

vibration diagnostics with a report costs 300 EUR per unit, balancing adds from 250 EUR, and the minimum invoice per visit is 500 EUR. The final figure depends on the number of rotors, the number of correction planes, how far the site is, and how long it takes to get access to the plane. The calculator on the website gives you an exact figure for your situation, and after a short conversation we confirm it with a number.

We're based in Vila Nova de Gaia, near Porto, and we travel all over the country. The people who come out are the engineers who design and manufacture the Balanset instruments, and who balance with them themselves. The instrument is two-channel: two accelerometers on the supports, an optical tachometer reading a reflective marker, one- and two-plane calculation by the influence-coefficient method, fixed positions and a drilling calculation, spectrum and reports.

- [x] The type and model of machine: blower, gas blower, air-blower unit, centrifugal or screw compressor, vacuum unit.
- [x] Working speed, and whether there's a step-up gearbox or a belt drive.
- [x] Bearing type: rolling-element or sleeve.
- [x] Current vibration figures, if you have them, and the points they were taken at.
- [x] Photos of the machine showing the supports, hatch and drive, plus a photo of the impeller if access has already been opened up.
- [x] What gets in the way: a hot environment, an explosive zone, manufacturer restrictions, or not being able to stop the machine.
- [x] The site address and the window in which the machine can be stopped.

> If the photos and data show that on-site balancing won't help, we'll say so in writing before the visit. A pointless visit doesn't work out for either side. We've covered the checklist for getting equipment ready for a site visit in a separate article — it saves more on site than any discount would.

## Frequently asked questions

**Can a screw compressor be balanced on site?**

Practically never. Screw rotors run in a casing with clearances of a few tenths of a millimetre, there's no access to them, and the rotor profile must not be touched. Screw-rotor balancing is done in a workshop, on a rig, with the block stripped down. If your screw compressor is vibrating, we come out to measure: more often the cause is the bearings, the clearances, the pulley fit or drive misalignment, in which case balancing isn't needed at all.

**The spectrum shows a large peak at the blade-pass frequency. Will balancing help?**

No. Blade-pass frequency is the number of blades multiplied by speed. A peak there means gas pulsation, uneven flow feed, too small a clearance between the impeller and the casing, or acoustic resonance in the discharge header. Balancing only reduces the once-per-turn component, and blade-pass frequency isn't part of that. Here it's the process condition, the clearances, the pipework and the pipe supports that matter.

**We have sleeve bearings. Is there any point measuring vibration on the casing?**

There's a point to it, but with caveats. The casing of a machine like this transmits less than the casing of a machine on rolling-element bearings, so the figures look more modest for the same rotor condition. We still get the once-per-turn component and phase, and that's enough for balancing. What matters more: on sleeve-bearing machines we always check the subsynchronous range. A peak at roughly 0.4 to 0.5 of running speed is oil whirl, and balancing doesn't remove it.

**The manufacturer forbids interfering with the impeller. What then?**

We work within the limits it leaves us. Usually that's the fitted threaded holes, balancing screws or rings, sometimes an allowed zone for removing metal. If there are no fitted points and drilling and welding are both forbidden, we won't balance the impeller on site. What's left then is the accessible planes on the shaft, pulley, half-coupling and fan, or workshop balancing of the rotor to the manufacturer's protocol.

**How many runs are needed, and does the compressor station have to be shut down?**

One correction plane needs a baseline run and one trial run; two planes need a baseline and two trial runs, plus a check run. Between runs the machine has to be stopped, depressurised and opened up to reach the plane so the weight can be moved. So the real schedule is set not by the measurement but by your shutdown and hatch-opening procedure. We agree the window in advance and work within it.

**What figures do you get vibration down to?**

We set the target for the once-per-turn component before the work starts and write it into the report. We assess the unit's overall condition against overall vibration and the zones of the applicable part of ISO 20816, and the quality of the rotor balance against residual unbalance and an ISO 21940 balance quality grade. Promising a specific figure in advance would be dishonest: on site it's limited by bearing condition, fixings, access to the planes and closeness to resonance. We tell you what's realistically achievable after the baseline measurement.