AXILINEBalancing turbines, turboexpanders and high-speed spindles at the point of operation
A turbine or spindle doesn't shake from unbalance alone. On top of the usual task here you get critical speeds — the speeds at which the rotor hits resonance — thermal bow, sleeve bearings, and a manufacturer's procedure that states outright where a weight may go. So we don't start with weights, we start with a measurement. After the first run and a coastdown — a reading taken as the rotor freely slows down — we tell you honestly whether your case can be solved on site or the rotor will have to come out.
Symptoms: when it's worth calling
On high-speed rotors, the margin between 'gone up a bit' and a failure is shorter than on a slow-running fan: the energy from unbalance rises with the square of the speed. Look at the change, not the absolute figure. A spindle that always gave 0.6 mm/s and is now giving 2.0 mm/s needs attention.
- Vibration has risen to one and a half times its level or more, at the same speed and load.
- Humming shows up in a narrow speed range and disappears above it: looks like resonance.
- The level rises as the machine warms up and doesn't come back down: possible thermal bow in the shaft.
- The spindle has started leaving waviness or chatter marks on the workpiece at the same settings as before.
- Things got worse after cleaning the flow path, replacing blades, an arbor or a half-coupling.
- The fitted monitoring system shows rising amplitude at running speed while the other channels stay stable.
A jump in level over a single run usually means a lost or bent blade, a weight that has shifted, or a fit that has worked loose. Don't balance it: a correction would mask the defect.
What equipment in this family we take on
Below are the subgroups people call about most often, and where the unbalance comes from in each.
Steam-turbine rotors
Unbalance from erosion and deposits on the blades, a bent blade, worn seals, residual shaft bow. Weights only go into the built-in planes: balancing grooves and rings, half-coupling bolts. On site we take on small drive turbines with a rigid rotor, under an agreed programme. Large multistage rotors are a job for a specialist facility.
Gas-turbine rotors and turboexpanders
Fouling and coking of the flow path, erosion, blade replacement, uneven heating. Speeds are high, the rotor is often flexible, and the supports run on sleeve bearings. We start with a measurement: 1x amplitude and phase at the supports, the spectrum (how the vibration breaks down by frequency), a coastdown. We balance on site if the rotor behaves as rigid and a correction plane is allowed — somewhere the construction lets us fit a weight.
Hydro-turbine runners
Speeds are low, the masses are large, and the vibration is more often hydraulic: blade-pass frequency, a vortex rope, cavitation. On top of that come shaft-line misalignment and clearances in the guide bearing. Balancing the runner helps in a minority of cases. We separate the causes first, and only fit a weight on the rim by an agreed fixing method.
Turbocharger rotors
Turbocharger rotors run at tens and hundreds of thousands of rpm: they're balanced component by component and on a high-speed rig — it isn't done in the machine. We measure the rotors of industrial centrifugal compressors on sleeve bearings and work to the owner's programme.
High-speed spindles and machine-tool spindles
The most productive subgroup for a site visit. Unbalance from the arbor and tool, the chuck, the faceplate, dents and carbon build-up on the mounting taper. We balance them in their own bearings at running speed, fitting the weight into the fitted balancing rings, screws or flange holes.
Test-rig rotors and special-purpose process equipment
Run-up and drive shafts on test rigs, run-in rig rotors, the rotor heads of dispersers and mixers, faceplates and tooling. These are usually rigid rotors at moderate speeds, and our format suits them best of all: two or three runs, one or two planes, a report with numbers.
Rigid rotor or flexible: the main fork in the road
We balance using the influence-coefficient method: we fit a trial weight of known mass and see how the amplitude and phase change. The method needs a stable system — the same response to the same mass from run to run. A rigid rotor turns below the first critical speed, its bow shape doesn't change, and a correction found once holds.
Above the first critical speed the rotor bends, the bow shape depends on speed, and every mode shape needs its own combination of weights. A weight that removes vibration at running speed can raise the amplitude when passing through a critical speed. Rotors like this are balanced by a modal scheme: several planes, several speeds, readings on run-up and coastdown. A coastdown tells you which kind of rotor you have.
| Rotor | Sign | What's possible on site |
|---|---|---|
| Rigid, running speed comfortably below the first critical | 1x amplitude and phase are stable, the response repeats | A full cycle in one or two planes within a single visit |
| Near-critical condition | An amplitude peak close to running speed, phase drifting | Only at a stable condition outside the resonance zone |
| Flexible, above the first critical | Bow shape depends on speed | A modal programme: our format doesn't fit — we provide a measurement and a finding |
| On sleeve bearings | The casing is damped by the oil film | Readings at the supports plus the fitted channels, decided against the owner's programme |
| Small turbocharger rotor | Tens and hundreds of thousands of rpm | Not possible on site — a run-up rig is needed |
Where weights can be fitted is set by the construction: grooves, rings, washers and half-coupling bolts, set screws. Welding a weight onto a disc or a blade is not allowed. For rotors with flexible behaviour there's a separate part of the standard covering procedures and tolerances; check the applicable part and edition for your machine.
Sources: ISO 21940-12:2016 · ISO 21940-11:2016
What we check before we touch any weights
Unbalance isn't the only thing that raises the once-per-turn component: misalignment, a worn-loose fit, thermal bow, rubbing in a seal and an unstable oil film all raise it too. That's why the first run is always diagnostic.
Sleeve bearings damp the signal: the casing shows less than the shaft is actually doing. So we fit the accelerometers on the bearing housings close to the split line, and cross-check against the fitted shaft-displacement channels, where they exist.
- 1x against the overall level (all the vibration at all frequencies combined): if 1x is only a small share of it, balancing will remove little.
- The spectrum: does 1x dominate, or is there 2x together with axial vibration, a comb of harmonics, subsynchronous peaks?
- How stable the 1x phase is from run to run: a drifting phase is a sign of resonance, looseness or bow.
- Coastdown: where the critical speeds sit, and how much margin there is from running speed.
- Thermal state: a reading on the warmed-up machine, and tracking the drift as it warms up.
- Fixings and supports: foot-bolt torque, soft foot (a foot that doesn't sit flush on the frame), the stiffness of the frame and foundation.
- Shaft alignment, the coupling's condition, and the fit of the half-coupling, arbor or chuck on the shaft.
- On a spindle, additionally: the mounting taper, the clamping, and the balance of the tool arbor.
Misalignment is a cause of vibration; shaft alignment is the action that fixes it. On a train made up of a turbine, a gearbox and a generator, the order is strict: fixings, shaft alignment, a repeat measurement — and only then balancing.
Sources: ISO 13373-3:2015 · ISO 281:2007
How the work goes on site
- 01
Agreeing the programme before the visit
We gather the rotor type, speed, bearing type, the support layout, the allowed correction planes and the access conditions. This is also where we say whether we're taking on balancing or only measuring for now.
- 02
Measurement points
We clean off spots on the bearing-housing casings, and fit two accelerometers on a magnet or a stud, in the direction of the largest vibration. We stick a reflective marker on the shaft for the optical tachometer.
- 03
Diagnostic run and coastdown
The Balanset-1A records the overall level, 1x with phase, speed, spectrum and time waveform, then a coastdown recording amplitude and phase. This is where it's decided whether we go on.
- 04
Trial weight
We fit a weighed mass in the allowed plane at a known radius. A run is usable if the 1x amplitude changes by 20–30%, or the phase by 20–30°. If the response is weak, we increase the mass.
- 05
Calculating and fitting the correction
The software gives the mass and angle for each plane, or, in fixed-position mode, the hole number directly. On half-couplings that means washers and balancing screws; on spindles, the fitted rings.
- 06
Check run
A measurement at the same condition and the same points. If the level has dropped but the target hasn't been reached, we add small masses to the ones already fitted.
- 07
Report and influence coefficients
We record the before-and-after figures under identical conditions, and save the influence coefficients. The next top-up balance will go ahead without trial runs.
Sources: Balanset-1A operation manual
One plane, two, or more
The rule of thumb by shape: at a length-to-diameter ratio below roughly 0.5, the rotor behaves as a disc, and one plane is often enough. An elongated rotor needs two, otherwise a couple unbalance remains — a pair of 'heavy spots' at opposite ends that rocks the rotor.
In this family, almost everything is long: a turbine shaft, a turboexpander rotor, a spindle, a test-rig drive shaft. One plane shows up on disc-shaped items: an overhung impeller, a faceplate, a rotor head.
- Two accessible planes on an elongated rotor: two trial runs instead of one.
- A flexible rotor needs its own set of weights for each mode shape: more than two planes is a different task altogether.
- If the level at the second support stays high after a one-plane correction, we move to two.
- Running speed close to a critical speed: only a stable condition outside the resonance zone will do.
Every trial run on a turbine or a large spindle costs more time than on a fan: it needs warm-up, the condition to settle, and a coastdown.
What gets in the way, and when it won't work on site
- A flexible rotor above the first critical: several planes, several speeds and a controlled run-up are needed.
- Small turbocharger rotors: only component-level balancing on a high-speed rig will do.
- There's no allowed correction plane, or the casing blocks access. If the casing is opened up anyway, the rotor is balanced outside the machine.
- Speed can't be held stable: the influence coefficients drift, and the result won't reproduce.
- Thermal bow: the 1x amplitude and phase drift at a constant speed.
- Rubbing in a seal, a cracked shaft, a rotor fit that has worked loose. That's a repair, not weights.
- A bearing defect or an unstable oil film: balancing doesn't cure a bearing.
- Hydraulic and aerodynamic causes: blade-pass frequency, a vortex rope, cavitation, flow separation.
- The manufacturer's procedure forbids intervention outside its own service, or outside a planned shutdown.
A measurement is worthwhile here too: you get the numbers, the spectrum, a coastdown chart and a finding to take to the manufacturer or into a repair.
What you get at the end
The result isn't 'it's quieter now', it's numbers taken under identical conditions before and after.
- A measurement before the work: the overall level in mm/s RMS (the root-mean-square value — the standard measure of vibration level), 1x with phase at each support, speed, running condition.
- Spectrum and time waveform, so you can see exactly what was producing the level.
- A coastdown chart: where the critical speeds sit relative to running speed.
- The masses, angular positions and radius of the weights fitted, by plane.
- A measurement after the work, at the same points and the same condition, with residual 1x against the target value.
- The influence coefficients, saved, and a finding: what was fixed and what we recommend.
'Within tolerance' in the software means one thing: residual 1x is below the target value you set. It isn't a condition assessment of the machine against overall vibration, and it isn't a confirmed balance quality grade in g·mm/kg. For acceptance, fix the applicable part of the standard, the measurement points and the running condition.
Sources: ISO 20816-1:2016 · ISO 21940-11:2016
Price and how to book
We're the engineers who design and manufacture the Balanset instruments, and we balance with them ourselves on site visits. We're based in Vila Nova de Gaia, near Porto, and we travel throughout Portugal.
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 and planes, the time spent on runs, and how far the site is. The calculator gives you the exact figure. So we can give you a useful answer from the first email, tell us:
- The type and purpose of the machine, working speed, power.
- Bearing type, and whether there's fitted monitoring.
- The support layout: between bearings or overhung, and the number of stages.
- Which correction planes the documentation allows and which are accessible without stripping the machine down.
- Current vibration figures, and from when the level started rising.
- Whether speed can be controlled, plus photos of the supports, the coupling and the area where weights would be fitted.
For more on getting the site ready for a visit, choosing between on-site work and a workshop rig, trim balancing (a quick touch-up using saved coefficients), and vibration coming back afterwards, see the separate articles in this section.
Sources: Balanset-1A manufacturer specification
Frequently asked questions
Can a turbine rotor be balanced without removing it?
Sometimes, yes. Three conditions are needed: the rotor behaves as rigid at running speed, a correction plane the documentation allows and that's accessible exists, and the condition repeats over several runs. If even one of these is missing, on-site work won't give a result.
Why can't you balance as normal close to a critical speed?
There, the rotation frequency coincides with the structure's natural frequency: amplitude rises sharply, and the 1x phase flips by around 180°. The response to a trial weight becomes unstable. That's why we record a coastdown and balance with a margin away from the critical speed.
The rotor is on sleeve bearings. Will an accelerometer on the casing show anything useful?
Yes, but with a caveat: the oil film damps the signal, and the casing transmits less than the shaft is actually doing. For balancing that's usually enough: what we need is the change in 1x amplitude and phase between runs. We assess the machine's condition taking the fitted shaft-displacement channels into account.
The spindle got noisy after the arbor was changed. Do you balance the spindle or the arbor?
The arbor and the whole tooling set-up first. The set-up changes from one operation to the next, and a spindle balanced for one tool falls apart the moment the tooling changes. Check the mounting taper and the clamping: a dent or carbon build-up produces runout that no weight will remove.
Vibration changes as the machine warms up. Is that unbalance?
Probably not. A drift in 1x amplitude and phase at constant speed as the temperature rises is a sign of thermal bow in the shaft, or rubbing in a seal. Balancing would lock in one thermal state and make another one worse. So we take a series of readings as the machine warms up first.
What happens if balancing won't help?
We won't start it. If the level isn't coming from unbalance, or the rotor needs specialised conditions, you get the readings, the spectrum, a coastdown chart and a finding with a recommendation. vibration diagnostics with a report costs 300 EUR per unit, balancing adds from 250 EUR, and the minimum invoice per visit is 500 EUR.
Related content
On-site balancing of high-speed spindles and tool assemblies, where they operate
Yes, but what we balance isn't the spindle unit itself — it's what turns on it: the tool assembly — the tool holder, collet, or chuck together with the cutting tool — and the spindle's standard balancing rings, if the factory built them in. We mount two accelerometers (vibration sensors) on the spindle head housing, radially at the front and rear bearing, and aim the laser phase sensor at a reflective mark on the spindle or on the tool holder's flange. We run the full cycle with an overall-level assessment up to roughly 12,000 rpm: the instrument computes vibration velocity RMS (the root-mean-square vibration level) across the 5–200 Hz band, and 200 Hz works out to exactly 12,000 rpm. Above that limit, we have two honest options. First: balance the rigid assembly at a reduced, stable rotation speed, then verify the result at operating speed by the running-speed component (vibration at the rotation frequency, denoted 1x) and the spectrum. Second: send the tool holder to a specialised machine built for tool assemblies. We never intervene in the spindle housing, the bearings, or their preload: that's the manufacturer's service work, not the balancer's.
Balancing machine-tool spindles and tooling at the point of operation
Yes, we balance machine-tool rotors on site, but not the spindle assembly itself. We balance whatever turns on the spindle and comes off it: a grinding wheel assembled with its flanges and arbor, a chuck, a faceplate, a tool arbor, a knife shaft, a cutter block, pulleys and the drive motor's rotor. We fit two accelerometers to the headstock casing, at the front and rear supports; the optical tachometer is aimed at a reflective marker — on the spindle, not the motor. There are three conditions: there's access to a correction plane (somewhere a mass can physically be fitted or removed), working speed is up to roughly 12,000 rpm, and the spindle bearings are in good shape. If a bearing in the assembly is worn out or the taper runs out, balancing won't give a result, and we'll say so before any work starts.
Critical speed and flexible rotors: why a correction made at one speed doesn't work at another
Critical speed is the rotational speed at which the rotation frequency coincides with a natural frequency of the rotor shaft's own bending vibration: the shaft bows outward, vibration in the 1x running-speed component spikes, and the phase swings by roughly 180°. Below the first critical speed, the rotor counts as rigid, and a correction in one or two planes works at any speed. Above it, the rotor behaves flexibly: the deflection shape depends on speed, so weights found at one speed can increase vibration at another. For rotors like this, modal balancing and multi-speed balancing apply, not the usual two-plane scheme.
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.