# Balancing axial fans: impellers, variable-pitch blades, cooling fans > An axial fan shakes harder than its weight would suggest: a light, large-diameter impeller can rock the casing, ductwork, and roof even with a fairly small unbalance. Axial machines also have a trap of their own: a good share of what looks like "unbalance" turns out to be scatter in blade-setting angle, which weights won't fix. We measure, tell the two apart, and balance the impeller right in its own supports, without taking it off the shaft. We're based in Vila Nova de Gaia, near Porto, and travel across all of Portugal. **In short:** Yes, we balance axial fans on site: duct and roof units, wall-mounted axials, cooling and smoke-extraction fans, and wheels with variable-pitch blades. The impeller stays on the shaft, we mount sensors on the motor's bearing end shields, and we need access to the hub through the inlet collector or a service hatch. Two conditions are mandatory: stable operating speed, and the once-per-revolution component, 1x — vibration at rotor speed, the only kind weights can remove — dominating the spectrum. And one caveat specific to axial machines: if the vibration is coming from scatter in blade-setting angle, we'll see it in the measurements and say so right away. In that case, the angles get set first, and balancing follows as a second operation. Source: https://axiline.pt/en/equipment/balancing-axial-fans/ Publisher: AXILINE · Vila Nova de Gaia, Portugal · +351 931 831 229 · axilinegeral@gmail.com ## Symptoms: how an unbalanced axial fan behaves What gives an axial machine away is that the vibration rarely stays in one spot. The impeller is light, the casing shell is thin, and the supports are compliant. So the whole structure shakes: a duct-mounted unit hums along the entire ductwork, a roof fan passes vibration into the curb and the floor slab, and on an air-cooled heat exchanger the diffuser and frame shake. The complaint often comes not from the maintenance crew but from people in the room a floor below. - [x] Vibration and noise increased after replacing one or more blades - [x] The impeller was taken apart for cleaning, the blades went back in the wrong positions, and the machine started shaking - [x] A clicking or a periodic whistle is audible: a blade is catching the ring for part of each revolution - [x] Vibration changes noticeably when the blades are repitched or the system's flow rate changes - [x] Vibration is felt by hand on the motor casing, and its bearings have started running hot - [x] A factory balancing weight has shifted or gone missing from the hub - [x] Cracks have appeared at the blade roots on a plastic impeller, and the material has gone cloudy - [x] The roof fan can be heard inside the building, even though its datasheet says it's quiet > An axial wheel is sensitive to small things. At a one-metre diameter, just five grams of mass at a blade tip already produces noticeable vibration. So a lost factory weight, dirt stuck to one blade, or a drip of paint left over from a repair is already reason enough to measure — not to guess. Sources: [ISO 13373-5:2020](https://www.iso.org/standard/62202.html) ## Impeller construction, and where the unbalance comes from An axial fan impeller is a hub and blades, most often removable ones: cast aluminium, reinforced plastic, or, less often, welded steel. Every blade has its own mass, and a difference of tens of grams within one set is normal for manufacturing. As long as the blades sit where the factory placed them after balancing, the set is in balance. Any swap, replacement, or lost part breaks that arrangement. ### Scatter in blade mass Replacing one blade with a new one from a different batch shifts the whole wheel's centre of mass. A new blade can easily differ from its neighbours by 20–50 grams, and it's working at a radius of half a metre or more. That's hundreds of gram-millimetres of unbalance on a light rotor. ### Motor inside the airstream On duct and roof units, the motor sits inside the shell on struts, and the main impeller is mounted overhung on its shaft. The motor's own cooling fan turns on that same shaft. Both sources produce vibration at the same frequency, and the instrument sees their sum. Only a two-plane calculation and careful sensor placement separate the two. ### Erosion and buildup Dust wears down aluminium leading edges and sticks to plastic ones. Mass leaves and arrives unevenly around the circumference. The general mechanics of fouling are described in our article on balancing fans; for axial machines, add one thing: the same dirt sits at a large radius here, so it produces more unbalance than it would on a centrifugal wheel. ### Factory balancing weights Factory weights sit on the hub or the rim: clips, bolt-on plates, or segments in a slot. Over time they corrode, work loose, and go missing. A lost weight hands the impeller back the entire unbalance the factory once compensated for — and it does it in one sudden step. ### Pitch-adjustment mechanism On machines with variable-pitch blades, there's also a pitch-adjustment mechanism. Play in it lets a blade rock in its mount: the angle drifts, and the vibration's phase (its tie to the rotor's angle of rotation) wanders from run to run. It's too early to balance a wheel like that. The play has to be taken up and the angles fixed first. ## Blade-angle scatter: it looks like unbalance, but it isn't A blade sitting at a different angle produces a different aerodynamic force. That force rotates with the wheel and shakes the machine once per revolution, at the same 1x frequency as a mass unbalance. A single measurement at operating speed can't tell them apart. We tell them apart by how the machine behaves, and that's the main reason an axial fan can't be balanced using the same playbook as a centrifugal one. - We measure the angles directly. On the stopped machine, we check every blade's setting angle with a template or a protractor, referenced off the hub face. A scatter of more than one or two degrees already produces a noticeable aerodynamic imbalance. - We change the regime. A mass unbalance doesn't depend on flow rate: partly close the damper, and the 1x amplitude barely changes. An aerodynamic component shifts noticeably as flow rate and angle of attack change. - We check phase repeatability. A fixed mass gives a stable 1x phase from run to run. A variable-pitch blade with play in its mount gives a phase that wanders by tens of degrees. - We try a mass correction. If a correctly calculated weight only brings the vibration down to a certain level, and what remains depends on the regime, that means the weight is fighting the aerodynamics. That's a dead end: change the regime, and the vibration comes back. > The order of work follows directly from this: set all the blades to the same angle first, using a template or the mechanism's built-in scale, and only then balance. Doing it the other way round means doing the work twice, because the correction has to be recalculated from scratch after the angles are adjusted. ## What we check before the first weight We've described the general method for telling unbalance apart from other causes in our article on identifying the cause of vibration. Here, we'll cover only what we check specifically on an axial machine, on stopped, locked-out equipment. - [x] The setting angle of every blade and its scatter, plus play and locking of the pitch-adjustment mechanism - [x] The blades' fit in the hub: fastener torque, and signs of fretting (wear marks from micro-movement in a joint) and work-hardening on the mating surfaces - [x] The radial blade-to-ring clearance all the way around, and signs of rubbing on the blade tips and the casing shell - [x] Cracks at the blade roots, chips and degradation on plastic blades, and erosion on aluminium leading edges - [x] Factory balancing weights: whether they're all in place, tight, and free of corrosion underneath - [x] The motor bearings, via the spectrum: on an in-stream motor the impeller is carried overhung, so the motor's bearings are the rotor's supports - [x] The motor's mounting on its struts and the weld seams of the struts in the shell: a cracked strut changes both stiffness and clearance - [x] How much the supporting structure gives: the duct-insert's vibration isolators, the roof curb, the cooler unit's frame. During coast-down we check for resonance (vibration amplified when speed coincides with the structure's natural frequency) near operating speed > A separate rule for machines running inside a ring shroud. If the blade-to-ring clearance is locally under the datasheet figure, we find out why first: sagging vibration isolators, a motor that's shifted on its struts, or a deformed shell. Fitting a trial weight on a wheel that's already nearly catching the ring is not an option: a changed deflection could end in rubbing at operating speed. Sources: [ISO 13373-3:2015](https://www.iso.org/standard/40840.html) ## How balancing an axial fan goes 1. **Reviewing the machine from photos** — You send us photos of the impeller from both sides, the diameter, blade count and material, the speed, and the configuration: duct, roof, wall-mounted, or air-cooled heat exchanger. We tell you where the correction plane will be (the spot on the rotor for correction weights), what we'll use to fit them, and what access is needed. 2. **Angles, fits, clearances** — We lock out the start and go through the checklist above. If the blade-angle scatter is large, we set them to a template before any runs. This removes the aerodynamic part of the vibration and leaves only the mass part for balancing. 3. **Initial measurement** — We mount two accelerometers on the motor's bearing end shields, near the front and rear bearing, radial direction, kept the same for every run. We stick the reflective mark on the hub or the nose cone, and aim the laser phase sensor through the inlet collector or a hatch. We record speed, overall vibration (the total level of vibration from all causes at once), 1x with phase, and the spectrum — vibration broken down by frequency. 4. **Blades as fixed positions** — We number the blades in the direction of rotation and enter them into the software as fixed positions. From then on the instrument gives a blade number and a mass instead of an angle, splitting the mass between two neighbouring positions if needed. Getting the reference direction wrong is ruled out. 5. **Small mass, secure fitting** — For a light impeller the trial weight is small, often 5–15 grams. We fit it with a bolt onto the hub or a dedicated mounting point, never onto the middle of a blade. We count a run as valid where the 1x amplitude has changed by 20–30 percent or the phase by 20–30 degrees. 6. **Weights or blade reshuffling** — The software gives a mass and a position. We fit a permanent weight at a dedicated spot on the hub, or, if the unbalance is large, redistribute mass by matching blades and take up the remainder with a small weight. A trim run clears the remaining vibration without new trial runs. 7. **Post-measurement and report** — The verification measurement happens at the same regime and speed. The report includes the before-and-after figures at each point, blade angles, weight masses and positions, spectra, and your machine's saved influence coefficients. Sources: [Balanset-1A operation manual](https://vibromera.eu/balanset-1a-operation-manual/) ## One plane or two, and where they sit An axial impeller is a classic disc: the hub width is much smaller than the diameter, and the length-to-diameter ratio, L/D, is well under 0.5. So one correction plane is enough in most cases. It sits on the hub: the back face, the rim under the factory weight positions, or the blade-mounting ring. One trial run, one weight fitting, a short job. Two planes are needed less often, but the cases are typical. A long, drum-type hub. A two-stage machine with two impellers on one shaft. And the most common case: an in-stream motor whose own cooling fan, on the rear end of the shaft, adds its own contribution to the vibration. Then that fan or the rotor's rear face becomes the second plane, and the instrument calculates both corrections in one series of runs, using the influence coefficient method — from the machine's measured response to a trial weight. > The L/D rule and the cost of a second plane are covered in our article on single- and two-plane balancing. The practical sign for an axial machine is simple: if vibration at the motor's rear bearing stays high after single-plane correction, there's a couple component in the system (a pair of forces rocking the rotor from both ends), and we move to two planes. Sources: [ISO 21940-11:2016](https://www.iso.org/standard/54074.html) ## Fitting weights without welding, and matching blades by mass We never use welding on axial impellers, full stop. Cast aluminium warps under heat, the structure changes near the weld, a thin blade distorts, and blade profile is exactly what gives the machine its aerodynamics. Plastic can't be welded at all. So the entire fitting toolkit here is mechanical, and the first thing we look for is what the manufacturer built in: clips on the hub rim, segment weights in a slot, threaded holes for plates. A dedicated spot with a factory mass limit always beats an improvised one. Where there's no dedicated spot, bolted fitting works: a stack of washers or a plate on a bolt through a hole in the hub or the blade-mounting ring, with torque checked and the thread locked. Removing metal by drilling is allowed only in a solid part of the hub, and only where the manufacturer permits it; the Balanset-1A software calculates the drilling diameter and depth. What we never do: glue weights on, or hang them off the blades. Centrifugal force at the blade tip tears a glued weight off, the loose mass flies down the duct, and any bump on the blade profile adds noise and spoils the airflow. A magnetic weight is allowed only as a trial weight on a steel hub, and is removed before the machine is handed back. On a wheel with removable blades, it's often more effective not to add mass at all, but to redistribute what's already there. The blades come off and are weighed one by one, to the nearest gram, together with their fasteners. The set is then arranged by position so the total mass vector is minimised: the heaviest blade opposite the next-heaviest, and so on around the wheel. This arrangement removes most of the unbalance, and the remainder is taken up with a small weight on the hub through ordinary balancing. - Nothing extra appears in the airstream: no bumps, no fasteners, the aerodynamics stays exactly as the factory built it - Nothing can work loose and fly off: the result doesn't depend on the tightness or corrosion of any added weights - Blade-mass difference is the main source of unbalance in an assembled wheel, so reshuffling addresses the cause, not just the symptom - Replacing a damaged blade stops being a lottery: the new one is matched by mass to the one it replaced, or the whole set's arrangement gets recalculated > When reassembling, we check the tightening torque and the setting angle of every blade. Otherwise, even with the mass order right, it's easy to introduce angle scatter, and the vibration switches character from mass-based to aerodynamic. Correction-weight mounting methods in general are covered in a separate article of ours on fitting weights. ## When on-site balancing won't work, or isn't needed An honest list of limits saves you a site visit and saves our reputation. Here are the cases where we'd stop and suggest something else. - The vibration comes from angle scatter, and there's no adjustment available: the blades are cast integrally with the hub and are deformed. Weights can't fix aerodynamics — the wheel gets repaired or replaced - Play in the pitch-adjustment mechanism: the phase is unstable and the influence coefficient is unreliable. Repair the mechanism first, then balance - A blade is rubbing on the ring. We restore the clearance first: realign the motor on its struts and check the vibration isolators and the shell's geometry - A crack at a blade root or in the hub. It can't be balanced: the vibration would drop, the one external warning sign would vanish, and a blade tearing off at operating speed punches through the casing shell - The impeller's plastic has delaminated or degraded from heat and UV exposure. Adding mass to material like that is dangerous — the wheel gets replaced whole - Resonance of the roof curb, the duct insert, or the frame: a peak shows up near operating speed during coast-down, and the phase is unstable. Structural stiffness or a speed change comes first — the signs of resonance are covered in a separate article - Overall vibration several times higher than 1x: motor bearings, electrical causes, looseness. We'll run diagnostics and tell you what to fix, but we won't prescribe weights - Speed won't hold steady: the drive is controlled by temperature and constantly drifts. We only balance if process staff can lock the frequency for the duration of the work > If vibration doesn't drop after correctly performed balancing, or comes back quickly, the cause usually isn't the weights. Seven such scenarios are covered in our article on why balancing doesn't help. ## What you get, and how to book a visit The result isn't just a quiet machine. You get a report: initial and residual vibration at each measurement point in mm/s RMS (root mean square — how every vibration meter measures it), the 1x component shown separately, the measured blade-setting angles and their scatter, and the mass and position of every weight fitted. If we did a mass-matching pass, the report includes the blade arrangement layout. Plus a zone assessment under the applicable part of ISO 20816, with a note on the edition. Your machine's influence coefficients are kept on file: the next balancing job on the same fan skips the trial runs. Balancing is done by engineers who design and manufacture the Balanset instruments and work with them out in the field themselves. Vibration diagnostics with a report is 300 EUR per unit, balancing adds from 250 EUR, the minimum invoice per visit is 500 EUR, and the calculator on the site gives you the exact figure for your machine. Several fans on one site are more economical to close out in one visit. - [x] Photos of the impeller from both sides, plus a photo of the motor and its mounting - [x] Wheel diameter, blade count, and material: aluminium, plastic, or steel - [x] Operating speed and the type of control: damper, VFD, or blade pitch - [x] Configuration: duct, roof, wall-mounted, air-cooled heat exchanger, or smoke extraction - [x] How access opens to the hub: an inlet collector, a hatch, or a removable duct section - [x] Whether there are factory balancing weights and dedicated mounting spots for them - [x] Your shutdown window, and who's responsible for lockout > We're based in Vila Nova de Gaia, near Porto, and travel across all of Portugal. Send us answers to this list, and we'll tell you plainly: whether your machine can be balanced on site, whether the blade angles need setting first, or whether the wheel needs repair before any weights go on. Sources: [ISO 20816-1:2016](https://www.iso.org/standard/63180.html) · [Balanset-1A manufacturer specification](https://vibromera.eu/product/balanset-1/) ## Frequently asked questions **Can a weight be glued or welded directly onto a blade?** No. Welding on cast aluminium warps the blade and changes the metal's structure near the weld, and it's simply not possible on plastic. A glued weight out at the impeller's radius is held on by not much more than hope: centrifugal force tears it off, and the loose mass flies down the duct. On top of that, any bump on the blade profile adds noise and spoils the airflow. We fit weights only on the hub and dedicated mounting spots, with mechanical fasteners and locked threads. On the blade itself, we never drill, weld, or glue. **We replaced one blade and the fan started shaking. What do we do?** This is the most common scenario on axial machines. A new blade from a different batch differs in mass by tens of grams, and the set was balanced by the factory with the old lineup. There are three options, in increasing order of effort: add a compensating weight on the hub calculated by the instrument, match the new blade by mass to the one it replaced, or weigh the whole set and rearrange the blades for a minimum total mass vector. The first option is usually enough, but with a large mass difference, rearranging gives a cleaner result. **We have a fan with variable-pitch blades. Is balancing possible?** It's possible, with a mandatory order of operations. First we check the pitch-adjustment mechanism for play: a blade that rocks in its mount gives an unstable phase, and the influence coefficient for a machine like that is unreliable. Then we set all the blades to the same angle using a template or the mechanism's built-in scale, because angle scatter produces an aerodynamic force at the 1x frequency that weights can't remove. Only after that do we balance the residual mass part. Do it the other way round, and the balancing has to be repeated after the angles are adjusted. **The impeller is plastic. Do you balance it?** Yes, with two conditions. First: before the job, we assess the material's condition. Plastic that's gone cloudy and cracked from heat or UV exposure may not hold an added mass — a wheel like that gets replaced, not balanced. Second: fitting is mechanical only, and only on the hub or dedicated spots, most often a stack of washers on a bolt through an existing hole. Masses on a light plastic wheel are small, just a few grams, so we choose the trial weight carefully too, starting from the minimum. **The motor sits inside the duct, and the bearings aren't reachable. How do you mount the sensors?** The sensors need a rigid contact with the motor's bearing end shields, or with a structure directly connected to them, such as the struts right at the flange. Access is usually available through a service hatch, a removable duct section, or the inlet collector, or, on roof units, after the cap comes off. If the insert is sealed and can't be opened up, the honest answer is: we won't get a reliable balance, because you can't measure vibration through the casing across soft vibration isolators. In that case we discuss a partial teardown or pulling the insert out. **How much does it cost, and how long does it take?** Vibration diagnostics with a report is 300 EUR per unit, balancing adds from 250 EUR, the minimum invoice per visit is 500 EUR, and the calculator on the site gives you the exact figure for your machine. On time: an accessible axial fan with a single correction plane takes a few hours — checking the mechanics and angles, an initial measurement, one trial run, fitting the weights, and verification. It runs longer if the blade angles need setting, the set has to come off and be weighed for mass matching, or a duct section has to be opened up for access. Several machines on one site are more economical to plan as a single visit.