AXILINEOn-site balancing of an induced-draft fan: the hot gas path, ash, and thermal regime
Your induced-draft fan shakes a little more every week, and after you cleaned the wheel it got worse, not better. A familiar story: ash builds up unevenly, abrasive dust wears down the edges, and thermal bow throws off the picture on a cold machine. We balance induced-draft fans and boiler-room and furnace draft fans right in their own bearing housings, measuring on the warmed-up machine and fitting weights after it cools down. We're based in Vila Nova de Gaia, near Porto, and travel across all of Portugal.
The symptoms that bring induced-draft fan owners to us
On an induced-draft fan, unbalance runs on its own schedule. The wheel works in a stream of flue gas loaded with ash and soot, so the rotor's mass shifts not over years but over weeks: buildup in one place, wear in another, a chunk breaking off somewhere else. The vibration doesn't just rise — it behaves in a characteristic way, and its behaviour tells us a lot before we even measure.
One trait specific to draft fans: vibration on the warmed-up machine differs from vibration on the cold one. If the fan shakes at start-up in the morning and settles down after two hours of running, or the other way round, that's thermal bow of the shaft and discs, and it isn't fixed the way ordinary unbalance is. This effect is covered in more detail in our article on rubbing and thermal bow in a rotor.
- Vibration climbs gradually over weeks after the last cleaning or repair, even though the boiler's operating regime hasn't changed
- The amplitude jumped within an hour: looks like a chunk of ash buildup or a wear plate broke off
- Vibration is one thing on the cold machine and another on the warmed-up one, and the level drifts for the first few hours after start-up
- Shaking got worse after cleaning the wheel, not better
- The volute and the duct hum, the service platform shakes, and the bearing-housing bolts keep working loose
- The bearing housings run hotter than usual, and the grease turns black before the service interval is up
- Vibration changes noticeably when the guide vanes are repositioned, even though the speed stays the same
- The heavy wheel's coast-down got shorter or sounds different: possible rubbing inside the volute
The trend matters more than the absolute figure: overall vibration at the bearing housings rising to one and a half times the usual level is already a reason to measure. Check the zone boundaries against the applicable part and edition of ISO 20816, for draft fans taking the support type into account.
Sources: ISO 20816-1:2016 · ISO 13373-5:2020
How an induced-draft fan is built, and where its unbalance comes from
The typical layout: the running gear, with a shaft on two bearing housings, sits on a frame outside the gas stream, and the impeller is mounted overhung into the volute. The drive is direct, through a coupling; on smaller machines, by belt. The wheel is heavy: a back plate, backward-curved blades, often a shroud too, and welded-on wear plates at the leading edges. The volute may carry a lining on the inside, and on the outside it's a shell with a hatch. The machine runs under negative pressure, so air gets drawn in through gaps at the hatch and the shaft seal, rather than dust blowing out.
This layout is convenient for balancing: the bearing housings are accessible, sit outside the hot gas path, and are usually only moderately warm. That's exactly where we mount the sensors, and we aim the laser phase sensor at a reflective mark on the shaft or coupling half. There's no need to go into the duct to measure.
Ash and soot
Deposits build up unevenly: thicker where local flow speed is lower, on the back side of the blades and in the corners by the discs. A few millimetres of buildup on one sector already produces unbalance that isn't visible from outside. The buildup has a life of its own: it compacts, sinters, and breaks off in chunks.
Abrasive wear
Fly ash acts like sandblasting. The blade leading edges and their wear plates thin out fastest. Wear isn't even around the circumference, so the wheel loses mass unevenly and the unbalance drifts. A wear plate tearing off produces a jump in amplitude: the level changes in a step, not gradually.
Thermal bow
Hot gas heats the wheel and shaft unevenly, especially at start-up and when the boiler load changes. The shaft bows slightly, the discs warp a little, and an apparent unbalance shows up that disappears once the machine cools down. That's exactly why a measurement on a cold induced-draft fan isn't reliable.
Old weights
On a wheel that's been balanced more than once, welded plates from past years pile up. Some of them were compensating for buildup that's no longer there. After a thorough cleaning, those old weights become a source of unbalance in their own right. We record them in the report: mass, position, and weld condition.
Why we measure the hot machine, not the cold one
This is the main methodological difference between an induced-draft fan and an ordinary fan. The rotor's geometry depends on temperature: as the machine comes up to its operating regime, the shaft and discs heat through, the thermal bow redistributes, and clearances and fit interference change. The amplitude and phase of the once-per-revolution component, 1x — vibration exactly at rotating speed, the component unbalance produces — differ between the cold and the warmed-up machine: sometimes the amplitude differs several times over, and the phase by tens of degrees.
You have to balance the state the machine actually lives in. So we take the initial measurement, the trial runs, and the verification measurement on the warmed-up induced-draft fan, in its settled thermal regime: after it's been running under load long enough for the 1x amplitude and phase to stop drifting. That's usually one to two hours after start-up, longer on bigger machines. We don't guess at stability — we see it on the instrument: several measurements in a row have to repeat in phase within a few degrees of each other.
Balance an induced-draft fan off cold readings, and you'll end up with a machine that's quiet on coast-down and shakes while it's running. The reverse is also true: don't be alarmed by higher vibration in the first half hour after start-up — judge it by the settled regime.
We keep the guide-vane position and the boiler's operating regime the same on every run. Repositioning the vanes changes the flow rate and the aerodynamic forces on the wheel, and measurements at different vane positions aren't comparable.
What we check before the first weight
We inspect the induced-draft fan wheel from the inside, through the hatch, on a stopped and cooled-down machine. This isn't a formality: on a dirty or cracked wheel, balancing either won't hold or will mask a defect that must not be masked.
- Cleanliness of the wheel: sector-by-sector ash buildup, on both sides of the blades, in the corners by the back plate and shroud. A dirty wheel gets cleaned completely first, then balanced
- Blades: thinning of the leading edges, through-wear spots, cracks at the root and along weld seams
- Wear plates: lifted edges, cracked welds, signs of a recent tear-off
- Old balancing weights: how many there are, where they sit, and whether the welds still hold
- The wheel's fit on the shaft and the condition of the key connection, coupling tightness, and, on a belt drive, pulley runout
- Fasteners on the bearing housings and frame, foundation condition, cracks in the gusset plates
- Signs of the wheel rubbing against the volute, and the condition of the lining: a piece of lining breaking off changes the clearances
- The shaft and hatch seals: heavy air ingress throws off the regime and lets ash dry out the bearing
- The guide vanes: play in the vane linkage, and whether the actuator is locked in its operating position
- Speed stability: on machines with a VFD, the setpoint is frozen for the duration of balancing
After that comes the measurement on the warmed-up machine: we compare the overall vibration with the 1x component, look at the spectrum (vibration broken down by frequency), and check phase repeatability. Balancing only reduces 1x. If bearing frequencies or the twice-per-revolution component, 2x — a typical sign of shaft misalignment — dominate the spectrum, we'll say so before fitting any weights, not after. The method for telling the causes apart is described in our article on how to identify the cause of vibration.
Sources: ISO 13373-3:2015
How balancing an induced-draft fan goes
- Request
Reviewing the machine from photos and data
You send us the machine type, speed, power, and drive type, wheel diameter, gas temperature, a photo of the wheel through the hatch, and photos of the bearing housings. We tell you whether it can be balanced on site, how many correction planes to expect, and how big a shutdown window to plan for, cooldown included.
- Run 0
Initial measurement on the hot machine
The induced-draft fan runs under load in its settled thermal regime. Two accelerometers sit on the bearing housings, and the laser phase sensor watches the shaft mark. We record speed, overall vibration, the 1x amplitude and phase at both housings, the spectrum, and phase repeatability. This is where we decide whether it's unbalance or something else.
- Shutdown
Cooldown and clearance to enter the duct
The machine is stopped, the drive locked out, and the guide vanes closed and fixed in place. The casing is ventilated down to a safe temperature and air composition. We check there's no draft pulling through the open hatch: neighbouring units sharing the same flue can set up a flow that both spins the wheel and draws gas into the work area. We lock the wheel mechanically so this draft can't turn it. Only after that do we work inside.
- Marking out
Blades and disc as fixed positions
We number the blades in the direction of rotation, measure the weight-mounting radius on the back plate, and enter the positions into the software. From then on the instrument gives us a position number and a mass instead of an angle. No protractor needed inside a tight volute.
- Trial
Trial weight and calibration run
We fix a weighed trial weight with a clamp or a bolt, as securely as a permanent one. The machine is brought back up to the same thermal regime, and only then do we take the reading. A valid trial run changes the 1x amplitude by 20–30% or the phase by 20–30 degrees — the criterion is covered in our article on trial weights. This is how the instrument gets the influence coefficients of your specific system — the machine's measured response to a known weight, the basis of the whole calculation.
- Weights
Fitting the correction after cooldown
The software gives a mass and a position for each plane. We stop the machine again, wait for it to cool, ventilate, and lock out. We clean the mounting spot down to bare metal, removing ash and scale, and weld the plates to the back plate. We never weld onto hot metal: it's both a safety rule and a recipe for a bad weld.
- Verification
Post-measurement and report
The verification run happens at the same speed, the same guide-vane position, and the same thermal regime as the initial one. If needed, we do trim balancing: a short fine-tuning of the weights already fitted, without new trial runs. Then comes the report, with before-and-after figures.
Plan the time window honestly: on an induced-draft fan, every cycle takes longer than on an ordinary fan. The heavy wheel coasts down slowly, the casing cools slowly, and warming back up before each measurement takes time too. It usually works out to two or three stop-start cycles, and the cooldown in them eats up more time than the actual work with the weights. The requirements for working inside the duct and for lockout are collected in our article on safety in on-site balancing.
Sources: Balanset-1A operation manual
One plane or two, and where they sit on an induced-draft fan
An induced-draft fan wheel is wide: the width-to-diameter ratio on most machines is above 0.5, and the wheel itself is mounted overhung, beyond both bearing housings. A weight in either plane affects both housings at once, and single-plane correction often leaves couple unbalance behind: one housing settles down, the other doesn't. So by default we plan for a two-plane scheme and confirm it with measurement.
Physically, the correction planes here are as follows. First: the outer face of the back plate, near the hub, where the metal is thick and welding is safe. Second: the shroud area near the rim, wherever a hand can reach through the hatch. On narrow, single-inlet wheels at moderate speed, one plane at the back plate is sometimes enough — you can see this from the residual at the second housing after the first correction. On double-inlet wheels, two planes are mandatory. The general rule for choosing is covered in our article on one and two correction planes; here, access matters more: if the hatch only opens onto one side of the wheel, we flag it at the request stage.
Induced-draft fans run at moderate speed, usually 500 to 1500 rpm, and the rotor behaves as rigid. Machines running close to their first critical speed (the speed at which a shaft starts to bend noticeably) need a different approach, under the applicable part of ISO 21940 for flexible rotors — on draft fans, this is rare.
Sources: ISO 21940-11:2016 · ISO 21940-12:2016
What we use to fit weights on an induced-draft fan wheel, and where
For an induced-draft fan there's one standard method: welding on steel plates. We fit bolted weights only where the manufacturer built in dedicated holes, and we use magnetic weights strictly as trial weights, removing them before handing the machine back: on hot metal a magnet loses its holding force, so it can never serve as a permanent weight.
- We clean the mounting spot down to bare metal: a weld over ash and scale won't hold
- We weld the plate all the way around its perimeter. An open-ended seam gets undercut by ash in the abrasive stream, and the weight eventually tears off
- We weld to the back plate, in the thick zone near the hub, or to the shroud. We never weld onto the blades, in the abrasive-flow zone, or onto the wear plates
- We cut the plates from the same steel grade as the disc. For heat-resistant wheels, we agree the material and electrode separately
- We keep the plate shape low and streamlined, so it doesn't collect buildup or create a local wear spot
- We weld only after the wheel and casing have cooled: that's both a requirement for clearance into the duct and a condition for a sound weld
- We don't just chisel off old weights of unknown origin: measurement comes first, the decision after. Sometimes it's more correct to treat them as part of the rotor
- We rarely remove mass by drilling on induced-draft fans: the disc is already thinned by erosion, and weakening it further isn't desirable
Our article on fitting correction weights covers the mounting methods and their limits in more detail. For an induced-draft fan, remember the essentials: a full perimeter weld, bare metal, a cooled-down machine.
When on-site balancing won't help
Some induced-draft fan requests we stop at the photo stage or after the first measurement. Here's an honest list of cases where weights won't solve the problem.
- The wheel is caked in ash. Balancing the buildup is pointless: that mass isn't fixed in place, and the result disappears the moment a piece breaks off. Full cleaning first, then measurement. Partial cleaning is worse than none at all
- Cracks in the blades, welds, or discs. Balancing would lower the vibration and hide the one external sign of the defect. A blade tearing off at operating speed destroys the volute. Repair comes first
- Critical thinning of the edges or lifted wear plates. The wheel keeps losing mass, and the unbalance comes back within weeks. Repair by weld build-up or replace the wheel, with balancing as the final step
- Residual shaft bow that doesn't go away on the cooled-down machine. That's no longer thermal bow, it's deformation. The shaft needs straightening or replacing
- Play in the wheel's fit on the shaft, a worn-out key. Any play means repair — the influence coefficients aren't reliable with it present
- Overall vibration several times higher than 1x: bearings, looseness, shaft misalignment. Shaft misalignment is corrected by shaft alignment — for a hot machine, accounting for the thermal movement of the supports
- Unstable regime: speed that drifts, constant repositioning of the guide vanes, a boiler that can't hold its load. Measurements don't repeat, and calibration isn't possible
- No cooldown window. If the machine can't be stopped for longer than an hour, fitting weights simply isn't physically possible. Then the work gets planned around the boiler's next shutdown
- Resonance of the frame or platform: vibration depends sharply on speed and isn't fixed by weights. The structure's stiffness needs addressing first
If the wheel is coming off for repair anyway, shop balancing on a machine may be the more sensible route. Our article comparing on-site and shop balancing covers both paths.
What you get, and how to book a visit
The result isn't just a quiet duct. You get a report: initial and residual vibration at each bearing housing in mm/s RMS (root mean square — the standard measure of vibration level), the 1x component shown separately, the speed, the guide-vane position, and the thermal regime at the time of measurement. Plus the mass and position of every weight fitted, before-and-after spectra, a zone assessment under the applicable part of ISO 20816, and a tolerance calculation by G accuracy class where the rotor mass is known.
There's a specific benefit for an induced-draft fan: we keep your machine's influence coefficients on file. Unbalance on a draft fan comes back by the nature of how it works, after the next cleaning or the next time buildup falls away. A repeat balancing job run from saved coefficients skips the trial runs entirely — one less cooldown cycle, and noticeably shorter.
Balancing is carried out by the engineers who design and manufacture the Balanset instruments and use them out in the field themselves. We work with the Balanset-1A: two accelerometers on the bearing housings, a laser phase sensor off a reflective mark, single- and two-plane calculation using the influence coefficient method, a fixed-position mode for the blades, and an archive with reports. You can buy the same instrument and balance your own fleet yourself.
Prices: vibration diagnostics with a report is 300 EUR per unit, balancing adds from 250 EUR, the minimum invoice per visit is 500 EUR. The final figure depends on the number of rotors, the number of planes, and how far away the site is — the calculator on the site gives you the exact number. A practical tip for a boiler room: it pays to balance the induced-draft fan and the boiler's forced-draft fan in one visit, since the second machine costs substantially less than a separate trip.
- Machine type and make, speed, power, and drive type
- Wheel diameter and width, gas temperature, and the fuel used by the boiler or furnace
- Photos of the wheel through the hatch from both sides, plus photos of the bearing housings and coupling
- When the wheel was last cleaned, and whether a cleaning is planned before we arrive
- Your shutdown window, cooldown included, and who's responsible for lockout and clearance to enter the duct
Send us a request with answers to this list, and we'll tell you plainly: whether your induced-draft fan can be balanced on site, whether the wheel needs repair first, and how much time to allow for the shutdowns. We're based in Vila Nova de Gaia, near Porto, and travel across all of Portugal.
Sources: ISO 20816-1:2016 · Balanset-1A manufacturer specification
Frequently asked questions
Why does the measurement have to be on the hot machine? Wouldn't it be easier on a stopped one?
Because a running induced-draft fan and a cooled-down one are effectively two different rotors. Thermal bow of the shaft and discs changes the amplitude and phase of the once-per-revolution component, sometimes several times over. Balancing off cold readings would give you a machine that spins quietly at low load and shakes at operating regime. That's why we take the initial, trial, and verification measurements in the settled thermal regime, and we see stability on the instrument through phase repeatability. And nobody goes inside the hot casing to do it: the sensors sit outside, on the bearing housings.
How long do we wait for cooldown before fitting the weights?
It depends on the machine's size, the gas-path temperature, and how well the casing ventilates. You can only enter the duct after it's been vented and cooled to a safe temperature, and weights can only be welded onto cooled-down metal: a weld over a hot, ash-covered surface won't hold. In practice, cooldown takes longer than fitting the weights themselves — on large induced-draft fans, that's a matter of hours. That's exactly why we agree the shutdown window in advance, and why repeat balancing jobs run from saved influence coefficients are so valuable: they remove one whole cooldown cycle.
We cleaned the wheel and vibration went up. Was the cleaning pointless?
Not pointless at all. Ash buildup is distributed at random and sometimes partly balances the wheel out, while also hiding weights from past balancing jobs that were fitted on an already-dirty rotor. Once the buildup comes off, that old compensation turns into a new unbalance. This is a normal, predictable effect: cleaning gives the wheel back its true, fixed mass, and that's exactly the mass that can and should be balanced. The right order is: full cleaning, inspection for cracks and wear, then balancing. Partial cleaning of just one sector is the worst option — it almost always increases the unbalance.
Can an induced-draft fan be balanced without a long boiler shutdown?
Not entirely without stops: the measurements are taken on the running machine, but the trial weight and the correction weights only go on once the fan is stopped, locked out, and cooled down. It usually takes two or three stop-start cycles, and the lion's share of the time goes into cooling down and warming back up to the settled regime. A realistic scenario: work during an agreed low-load window or at a scheduled shutdown. If you have saved influence coefficients from a previous balancing job on this machine, you need fewer cycles.
What about safety when working inside the duct?
The procedure is strict, and we don't cut corners on it. The drive is disconnected and locked out, the guide vanes are closed and fixed, and the casing is ventilated down to a safe temperature and air composition. We separately check there's no draft: neighbouring units on a shared flue can pull air through the open hatch, and that flow can turn the wheel. So we lock the wheel mechanically before anyone enters the casing. Welding inside happens only on cooled metal and under the permit-to-work procedure adopted at your site. From your side, we need someone responsible for lockout and the work permit.
How much does balancing an induced-draft fan cost?
Vibration diagnostics with a report is 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 correction planes, how far away the site is, and whether diagnostics is needed before balancing. The calculator on the site gives you the exact figure for your machine. Keep in mind the particular nature of induced-draft fans: because of the cooldown cycles, the work takes longer than on an ordinary fan, so it pays to combine it with a scheduled shutdown and balance neighbouring machines, such as the same boiler's forced-draft fan, in one visit.
Related content
On-Site Balancing of Centrifuge Drums and Baskets
We balance the drum or basket on site, in the machine's own bearing housings. But first we answer a different question: is this imbalance at all. On a filtering or sedimentation centrifuge, the dominant vibration more often comes from unevenly settled product than from the rotor. We tell them apart by phase repeatability — the angle by which the vibration is tied to the rotor. We run several starts in a row on a clean, dry basket and check whether the phase of the 1x running-speed component — vibration at the rotation frequency — stays put. If it does, it's mechanical imbalance and weights will help. If it drifts from run to run, weights would only mask the problem for a single cycle. We fit masses at the standard spots on the rim or the reinforcing ring. We don't drill the perforated shell, we don't weld on stainless steel, and we don't put foreign parts in the product zone.
On-site balancing of centrifuges and separators: drums, baskets, separator rotors
Yes, we balance centrifuges and separators at the site where they operate, in the machine's own supports, but under two conditions. First: measurements have to confirm that the vibration is dominated by the 1x running-speed component — vibration at the rotor's rotational frequency, a sign of imbalance — rather than by shaft misalignment, bearings, a loosened fit, or resonance. Second: the manufacturer has to permit fitting correction masses on that rotor. On high-speed disc-stack separators and on sealed drums, intervention is often prohibited or requires written approval. In that case we carry out vibration diagnostics only and hand you numbers you can take to the manufacturer or a service centre.
Multi-bearing rotor balancing: a long shaft on three or more bearings
A multi-bearing rotor is a distinct problem, not an extended two-plane one. For a rigid rotor on two bearings, you're solving a system of two equations with two unknowns, and the influence coefficient matrix is 2×2. Three bearings give a 3×3 system with nine coefficients and a minimum of four runs; four bearings give sixteen coefficients and five runs. You have to measure at every bearing and solve the system as a whole: if you work through the planes one at a time, the vibration will keep running from bearing to bearing, and the process won't converge.
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.