# On-site balancing of rolling mill rolls, calender rolls, and print cylinders > A roll lives by its own rules. The working surface cannot be touched, so mass correction goes to the end faces, to the standard balancing plugs, or to the drive flange. The roll heats up internally and changes shape, works in a pair under nip load, and the result shows up not only on the instrument but on the product itself. We arrive with a two-channel analyzer, mount sensors on the chocks and bearing housings, and start by answering the main question: is this even imbalance at all. **In short:** Yes, we balance rolling mill rolls, calender rolls, print cylinders, and paper machine rolls on site, in their own bearings. The condition is strict: the roll must have permitted correction planes — places where mass may be added or removed — outside the working surface. These are the end-face balancing groove, standard threaded plugs on a bolt circle, an end-face bore, the drive flange, or the half-coupling. We never weld, drill, or centre-punch the barrel or its covering. We balance in two planes, on a warmed-up roll, under running conditions. If there are no standard correction points, if the barrel runs out, or the covering has gone wavy, on-site work won't do it: the roll is removed, ground, and balanced on a machine, supported on its journals. Source: https://axiline.pt/en/equipment/on-site-balancing-rolling-mill-rolls/ Publisher: AXILINE · Vila Nova de Gaia, Portugal · +351 931 831 229 · axilinegeral@gmail.com ## Symptoms: the roll writes its diagnosis on the product For this group of machines, vibration rarely shows up as noise. It shows up as scrap. A band across the web at the pitch of the roll's rotation, uneven sheet thickness, doubling in print, waviness in the winding. The operator sees the defect before the vibration meter shows an alarm level. So the first thing we ask is what pitch the defect repeats at. If the period matches the circumference of a specific roll, the list of suspects narrows to that roll. Then we measure and see what's behind that pitch: imbalance, barrel runout, or corrugation of the covering. - [x] A periodic defect on the product at a pitch equal to the circumference of one of the rolls in the pair. - [x] Uneven strip or sheet thickness that disappears when speed is reduced. - [x] Vibration appeared after a roll change or regrinding, where there was none before. - [x] The level at the chocks increased after the covering was recast or a journal was repaired. - [x] Closing the nip makes vibration change in a sudden jump. - [x] A thermal roll behaves differently cold and warmed up; the numbers don't repeat from run to run. > Balancing only reduces the 1x running-speed component — vibration at the roll's rotation frequency, exactly once per revolution; that's exactly what imbalance produces. This is especially noticeable on rolls: here the overall level is built up by the drive, the gear train, and the nip — the line where the two rolls press together. The relationship between overall vibration, 1x, and phase is covered in a separate article; on site we simply measure the 1x share and tell you how much of your level the weights will remove. ## Design: where imbalance on a roll comes from A roll rarely leaves the factory out of balance. Operation and repair are what make it that way. ### Rolling mill rolls and chocks Rolls sit in chocks, and the chocks sit in the housing windows with clearance. Imbalance comes from uneven barrel wear, a neck built up by weld repair, or a misaligned drive spindle. Rotation speed is low, so the running-speed component falls below the usual 10–1000 Hz band. ### Calender rolls A pair typically combines a hard steel or cast-iron roll with a roll carrying a filled covering. The soft one wears unevenly, absorbs moisture, and changes its mass and mass distribution after regrinding. Hence the typical complaint: it ran smoothly before regrinding, and vibration appeared after. ### Thermal rolls: heating and cooling Peripheral channels or a central channel carrying oil, steam, or water. Uneven filling, condensate, a blocked channel, and temperature variation along the barrel length all produce an imbalance that changes as the roll warms up. On a cold roll it may not show up at all. ### Print cylinders and sleeves Plate and offset cylinders, interchangeable sleeves, inking unit rollers. Mass is added by the sleeve, the mounted plate, tape under the plate, and built-up ink. The amounts are small, speeds are higher than on a calender, and the requirements for smooth running are tighter. ### Paper machine rolls Press, dryer, suction, and reel rolls. Length becomes a factor: the length-to-diameter ratio exceeds ten, and operating speed approaches the first critical speed — the rotation rate at which the roll enters resonance. Water inside the hollow shell acts as a variable mass. ### Drive side The drive gear, spindle, half-coupling, universal joint. The imbalance you hear on the roll often actually lives here. And that's the good news too: unlike the barrel, you can mount weight on the drive flange. ## Where correction planes are actually allowed on rolls like these This is the central question of the whole job, and it's settled before any trial runs. The working surface is untouchable. Welding changes hardness and draws a crack under the covering, drilling ruins the barrel profile and the crown — the factory-built convexity of the barrel toward the middle — and centre-punching leaves a mark that will show up on the product within one revolution. Mass goes only where the manufacturer built it in. That's why we ask for the roll drawing, or at least photos of the end faces, before the visit. If there's an annular groove for slide blocks or a bolt circle with threaded plugs, the job is predictable. If there isn't, the conversation turns to the drive flange, the half-coupling, and the neck outside the bearing zone — and sometimes it ends in an honest no. | Location on the roll | What's allowed | What we never do | | --- | --- | --- | | End-face balancing groove, T-slot, ring | Standard slide blocks and weights around the slot | We don't widen the slot or weld on it | | End-face flange, standard threaded plugs | Plugs of a different mass, washers, calculation by position number | We don't drill new holes without the manufacturer's approval | | Drive flange, half-coupling, gear hub | Bolt-on weights, clamp collar, metal removal on non-working shoulders | We don't touch the gear rim or the fit surfaces | | Neck outside the bearing and seal zone | A split ring or clamp collar with screws, if the design allows | We don't weld on the neck or touch the bearing fit | | End-face bores and internal cavities | Only factory-engineered solutions with reliable mass locking | We don't block heating channels or leave weight unsecured | | Barrel, working surface | Nothing | No welding, drilling, weld overlay, or centre-punching | | Covering: rubber, polyurethane, composite, chrome, ceramic | Nothing | We don't drill, cut, or spot-grind | > Not every roll has standard balancing points. If there aren't any and there's nowhere safe to add mass, on-site balancing isn't performed. We'll say so from the drawing and photos, before the visit, not after three trial runs. ## Heating, cooling, and thermal bow A cold thermal roll and a thermal roll at operating temperature are two different rotors. The barrel expands, the bow changes, and the heat-transfer fluid inside redistributes. Balancing a cold machine is pointless: a correction found in the morning on an unwarmed roll will stop working by lunchtime. We bring the machine up to operating speed and temperature and wait until the numbers stop drifting. The criterion is simple: the 1x amplitude and phase — the angular position of the vibration relative to the mark on the roll — repeat from run to run. If the phase keeps rotating around at a constant speed, you're looking at thermal bow or rubbing, and weights won't help here. Uneven warm-up along the barrel length is just as deceptive: the level tracks temperature, not speed, and the fix is adjusting the heat-transfer fluid supply. - We balance only on a warmed-up roll, at its operating temperature and speed. - We check the heat-transfer fluid supply: a blocked channel creates imbalance, and compensating for it with a weight is the wrong fix. - We drain condensate from the cavities and channels, otherwise the readings drift from run to run. - We record the barrel temperature so the verification measurement runs under the same conditions. - If the roll operates across several temperature regimes, we fit the correction to the main one and flag the spread on the others. > Thermal bow and rubbing are covered in a separate article. On site, what matters isn't the mechanism but the criterion: until the phase settles, it's too early to fit a trial weight. ## The roll pair and the nip: what we measure under load, and what we measure without it A roll almost never works alone. It's pressed against the adjacent roll, and the nip changes everything: system stiffness, bearing load, the path vibration travels through. The level with the rolls apart and the level under the nip are two different numbers, and you cannot compare them to each other. The sequence is this. First we measure and balance with the nip open, or at minimum nip force: there the running-speed component is seen cleanly, and the influence coefficients — the roll's response to a trial weight, from which the software calculates the correction — come out stable. Then we bring the nip up to operating force and take a verification measurement at the same points. Both states go into the report. We separate the two rolls of the pair by rotation frequency, since their diameters differ, and the frequency of the peak immediately shows whose imbalance it is. - [x] Measurement with the nip open — the basis for calculating the weights. - [x] Measurement under the operating nip — what the machine actually lives with. - [x] Separating the two rolls of the pair by rotation frequency in the spectrum (the breakdown of vibration by frequency), so we don't balance someone else's imbalance. - [x] Checking the covering for corrugation: periodic waviness produces vibration that doesn't match 1x and can't be removed with weights. - [x] Assessing the drive's contribution: gear-mesh frequency and its harmonics have nothing to do with imbalance. - [x] A jump in level when nip force changes points more often to structural resonance than to mass. ## What we check before trial runs On a roll, we reach for the dial indicator before the weights. Balancing works with mass distribution, not shape, and on a precision roll that distinction is the most costly one to get wrong. - [x] Barrel runout measured with a dial indicator at several cross-sections along the length, on slow rotation, plus neck runout separately. - [x] Barrel profile and crown: wear, taper, traces of the last regrinding. - [x] Condition of the covering: waviness, corrugation, delamination, moisture uptake, hardness variation along the length. - [x] Bearing units and chocks: chock clearance in the housing window, wear on the liner plates, bearing condition. - [x] Fasteners on the supports and posts, soft foot (a foot not sitting flush against the frame), condition of the calender or stand housing. - [x] Drive alignment. Shaft misalignment calls for shaft alignment, not balancing, and it's especially easy to mask at low speeds. - [x] Cavities and channels: water, condensate, blocked holes. - [x] Spectrum at operating conditions: 1x share, drive gear-mesh frequency, harmonics, high-frequency bearing peaks. - [x] Rotation speed. At 60–200 rpm the running-speed component sits below 3.5 Hz, so we extend the lower end of the measurement range. Sources: [ISO 13373-3:2015](https://www.iso.org/standard/40840.html) · [ISO 281:2007](https://www.iso.org/standard/38102.html) ## How the work proceeds on site 1. **Baseline measurement** — We mount two accelerometers on the bearing supports of the roll being balanced — on the chocks or bearing housings, close to the bearing, on cleaned mounting spots. Direction is horizontal-radial, the same from measurement to measurement. We attach a reflective mark to the roll end face, the neck, or the drive flange. We record overall level, 1x, phase, speed, and spectrum at operating temperature. 2. **Decision: balancing or repair** — We calculate the running-speed component's share, cross-check the defect period on the product against the roll's rotation, and check runout and clearances. If something other than imbalance dominates, we stop here and hand over the diagnosis. 3. **Agreeing on correction planes** — We determine from the drawing and from the actual roll where mass can be placed: an end-face groove, standard plugs, an end-face bore, the drive flange, the half-coupling. We record the mounting method and radius in writing, before the first trial weight. 4. **Trial weight, first plane** — We fit a weighed mass at a known radius and run the machine under the same conditions. We consider the response valid at 20–30% or more in amplitude or angle. A weak response is a reason to increase the mass, not to calculate a correction from noise. 5. **Trial weight, second plane** — We move the weight to the second end face and repeat the run. The software obtains influence coefficients for both planes and outputs the mass and angle for each. 6. **Correction and verification run** — We fit the weights at the agreed points. For standard plugs and grooves we work in fixed-position mode: instead of an angle and a protractor, you get a position number and a mass. We remove the trial mass, run a verification start, and if needed add a small mass in trim mode. 7. **Measurement under nip and report** — We bring the nip up to operating force and repeat the measurement at the same points, in the same direction, at the same temperature. We record the before-and-after numbers, masses, radii, and position numbers. We save the influence coefficients: the trim after the next regrinding will go through without trial runs. > We work with the Balanset-1A instrument: two vibration channels, a laser phase sensor keyed to the reflective mark, a laptop with the software, fixed-position mode, calculation of metal removal by drilling, tolerance by G grades (balance quality grades), archiving and reports. If the roll shop already has a balancing machine with an outdated measuring section, the Balanset-1A OEM version integrates into it without the case. Sources: [Balanset-1A operation manual](https://vibromera.eu/balanset-1a-operation-manual/) ## When you need the shop, regrinding, and a balancing machine Declining on-site balancing is also a result of the visit, and on rolls it happens more often than on fans. The reason is always one of two: either geometry, or nowhere to put the mass. - The barrel runs out, there's taper, or the crown has been lost. Regrinding comes first, then balancing on the restored geometry. - The covering has gone wavy, is delaminating, or has absorbed moisture. After recasting and grinding, balancing is mandatory: the mass of the covering has changed. - Corrugation has appeared on the covering. This isn't imbalance — it's a trace of vibration in the nip, and it's removed by grinding plus eliminating the source. - There are no standard balancing points, and the working surface can't be touched. The roll is removed and balanced on a machine, supported on its journals. - The roll is long and flexible, and operating speed is close to the first critical speed. On-site results don't hold up well; a spin-test rig is needed. - Acceptance to a G balance quality grade with residual imbalance calculated in g·mm/kg is required. On site we give an estimate; a guaranteed grade is confirmed on the machine. - Bearings are worn out, or the chock is loose in the housing window. Balancing will mask the level and won't restore service life. > The comparison between on-site and shop work is covered in a separate article. The rule for rolls is simple: if the roll has to come off for regrinding anyway, it makes more sense to balance it there, and leave the verification measurement and trim for the site visit. Sources: [ISO 21940-12:2016](https://www.iso.org/standard/50429.html) · [ISO 21940-11:2016](https://www.iso.org/standard/54074.html) ## What you get, the price, and how to order The main result is numbers you can present to other departments and to the roll manufacturer. Measurements before and after at the same points, under the same conditions, at the same temperature, and at the same nip force. Without these qualifications, comparison on rolls doesn't work. Vibration diagnostics with a report costs 300 EUR per unit, balancing adds from 250 EUR, the minimum invoice per visit is 500 EUR, and the calculator on the site gives an exact figure. We are the engineers who design and manufacture Balanset instruments, and we do the on-site balancing ourselves. Base in Vila Nova de Gaia near Porto, service visits across Portugal. - Roll type: rolling mill, calender, print cylinder, press, or dryer. - Barrel diameter, length, mass, and the machine's operating speed or roll rotation speed. - A drawing of the end faces or photos: grooves, bolt circles, standard plugs, drive flange. - The covering, its condition, and the date of the last regrinding. - Heating or cooling mode and the barrel's operating temperature. - The defect on the product and its repeat pitch, if there is one. - Constraints: what can't be touched, the shutdown window, site access requirements. - [x] A report with overall vibration and the running-speed component for each support, before and after the work, with the nip open and under the nip. - [x] The masses fitted, radii, angles, or standard position and plug numbers. - [x] Spectrum and condition assessment: bearings, chocks, fasteners, drive, signs of shaft misalignment and resonance. - [x] A geometry assessment: barrel and neck runout, condition of the covering, notes for the next regrinding. - [x] An assessment against the applicable part of ISO 20816 as a reference and, if needed, a residual imbalance calculation for the selected G grade. - [x] Saved influence coefficients for that specific roll. > Preparation cuts the visit time almost in half: access to both bearing supports, cleaned mounting spots for the sensors, a visible spot for the phase sensor mark, and the ability to warm up the machine and stop it as many times as needed. There's a checklist in the article on preparing equipment for balancing. 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/) ## Frequently asked questions **Can a correction weight be welded or drilled directly onto a calender roll?** No. The barrel and the covering are untouchable: welding changes hardness and draws a crack, drilling ruins the profile and the crown, and any mark on the working surface will show up on the product within one revolution. We fit mass only in the end-face groove, in standard threaded plugs, in an end-face bore, on the drive flange, or on the half-coupling. If none of these exist, we don't do on-site balancing, and we say so before the visit. **Should a thermal roll be balanced cold or warmed up?** Warmed up, at its operating temperature and speed. A cold roll and a hot roll are different rotors: the barrel expands, the bow changes, the heat-transfer fluid redistributes. A correction found on a cold roll will stop working by the end of the shift. We wait until the 1x amplitude and phase start repeating from run to run, and only then fit a trial weight. **A band is appearing on the web at the roll's rotation pitch. Is this definitely imbalance?** Not necessarily, but the pitch is half the diagnosis. Barrel runout, local covering wear, and corrugation all produce the same period, and they're fixed by grinding, not by a weight. We measure the running-speed component's share, check the barrel with a dial indicator, and look at the spectrum. If 1x dominates and the geometry is sound, balancing will remove the band. **Should the nip be closed or open during balancing?** We run the main cycle with the nip open or at minimum nip force: there the running-speed component is seen cleanly and the influence coefficients come out stable. We take the verification measurement under operating force, since that's what the machine actually runs with. Both states go into the report, and they can't be compared to each other: the system stiffness is different. **The roll is back from regrinding. Does it need to be balanced again?** As a rule, yes. Removing metal or covering material changes the mass and its distribution, and asymmetric removal adds imbalance. The good news is that for a roll we've already worked on, we keep the influence coefficients on file, so the trim goes through without trial runs: measurement, calculation, weight fitting, verification run. **The roll turns at 80 rpm. Will the instrument detect imbalance at that speed?** Yes, but with different settings. 80 rpm is 1.33 Hz, below the usual 10–1000 Hz band used to assess the machine's general condition. We extend the lower end of the range and look at more than just vibration velocity. At speeds like this, vibration is often caused not by imbalance but by chock clearance, a worn bearing, or barrel runout.