# On-Site Pump Impeller Balancing: Closed, Semi-Open and Open Impellers > The pump shakes, the mechanical seal weeps, the front bearing runs hot, and the impeller is the first suspect. Sometimes rightly so: over a year, cavitation eats away at the inlet edges, deposits settle in one blade channel, or a new impeller arrives with residual imbalance. But on a pump, flow rather than mass often accounts for half the vibration level, and that's something we can check on site within the hour. We arrive with a two-channel analyser and first separate the running-speed component — vibration at the rotational frequency, which is what imbalance produces — from the blade-pass frequency (the rate at which blades pass a fixed point) and cavitation noise. If the numbers support it, we balance the rotor right in its own bearing supports. **In short:** Yes, we balance pump impellers on site, in their own bearing supports, without removing the impeller from the shaft. Three conditions apply: the running-speed component accounts for most of the vibration level, and its phase repeats from start to start; the impeller has been cleaned of deposits; and there is at least one correction plane — a place where mass can be added or removed — that is actually reachable. If erosion has eaten the blades down to uneven thickness, the disc has cracked, or the clearance on a semi-open impeller has gone beyond the manufacturer's tolerance, balancing will only postpone replacement. In that case we say so directly, before the visit. Source: https://axiline.pt/en/equipment/on-site-pump-impeller-balancing/ Publisher: AXILINE · Vila Nova de Gaia, Portugal · +351 931 831 229 · axilinegeral@gmail.com ## Symptoms that point specifically to impeller imbalance Impeller imbalance behaves differently from hydraulic effects. It doesn't respond to the discharge valve, doesn't disappear when the level in the suction tank rises, and holds steady at the same level for months, until the impeller changes again. The phase of the running-speed component keeps repeating from start to start: the heavy spot sits at one angular position and doesn't move. - [x] Vibration level rose gradually over months, then dropped in a single start after flushing the casing and impeller. That's buildup, and it will come back. - [x] The pump started shaking harder right after a new impeller was fitted, or after an impeller of a different size was installed. - [x] Vibration increased after the impeller was trimmed down in diameter for a different head. - [x] The pump caught reverse rotation on shutdown, and the noise changed afterward. Check the impeller nut and the fit. - [x] On an overhung pump, the front bearing — the one nearer the impeller — vibrates noticeably more than the rear one. - [x] Axial vibration at the thrust bearing has increased along with the radial vibration. - [x] The pump has been pumping sand, sludge or slurry for a season, and the blade inlet edges are visibly worn. - [x] Inspection shows pitting and a rough, eaten-away surface on the inlet edges or the shroud. - [x] Vibration level doesn't change when you operate the discharge valve. > The last point is the most valuable one. Vibration that doesn't depend on flow and pressure is mechanical, and that's exactly what we reduce with correction weights. Anything that responds to the operating point is a separate matter — covered in our article on hydraulic causes of pump vibration. ## How the impeller is built, and where its imbalance comes from The impeller is a casting or a welded assembly with blades, fitted onto a taper or a key and secured with a nut. Imbalance gets into it in four ways. First: casting non-uniformity and residual imbalance left over from manufacturing or repair. Second: wear, which almost never proceeds symmetrically around the circle. Third: product buildup in the blade channels and on the discs. Fourth: assembly — that is, an eccentric fit on the shaft, misalignment from an under-tightened nut, or a trimmed or lost key. The impeller type determines the main thing: where you'll be able to add or remove mass. That's the key question for a visit, and it's often visible straight from a photo of the assembly. ### Closed impeller The blades are enclosed by the back and shroud discs, and the blade channels are sealed off. You can't get inside, and that's exactly where deposits and cavitation pits sit. The correction plane stays on the outside: the outer face of the back disc, the impeller nut, or the coupling half. The back disc usually has balance holes that reduce axial thrust. These are not used for correction, and drilling near them is avoided. ### Semi-open impeller Only the back disc is present; the blades are open on the inlet side and run with a small clearance to the cover or wear plate. The blade tips must not be touched: this clearance sets the head and efficiency, and metal removed there can't be put back. We take mass from the outer face of the back disc, or move the correction to the coupling half. ### Open impeller The blades sit on the hub with no discs at all: sludge pumps, sewage pumps, and contaminated media. Access to the blades is best here, but each blade's mass is small, and removing metal noticeably changes the hydraulics. Here, cleaning and replacement usually win out, with balancing serving as an interim measure between repairs. ### Between-bearings and double-suction impellers A rotor with the impeller mounted between two bearing supports responds differently from an overhung one: the reaction forces from the same unbalanced mass are shared between the supports. There are two correction planes, and both only become accessible once the casing cover is removed, so on-site work is done through the coupling half and the free end of the shaft. > Note the number of blades separately. We don't need it to calculate the correction weights — we need it to compute the blade-pass frequency, so we don't mistake its peak for imbalance. ## What we check before we bring out the weights We mount the sensors on the bearing supports: one on the coupling-side bearing, the other on the impeller-side bearing, both radial-horizontal, magnet-mounted on a cleaned spot. On an overhung pump we add an axial reading at the front bearing: axial vibration there indicates wear on the wear-ring clearances and the condition of the balance holes. We stick the reflective tape onto the coupling half or a dry section of the shaft. Under the packing gland it gets washed off by leakage, and the laser phase sensor loses the signal. Next comes a short list of things specific to the impeller and its fit. Balancing over the top of these issues is pointless — the result won't hold through the end of the shift. - [x] Whether the impeller and casing have been flushed. A dirty impeller can't be balanced: you'd be fixing weights around the buildup, and when a piece breaks off, vibration will end up higher than the starting level. - [x] Impeller nut: tightening torque, locking, thread direction. A loose nut produces an unstable running-speed component, and the influence coefficient — the machine's response to a trial weight, the basis of the whole calculation — can't be obtained from it. - [x] Taper or key fit: shaft runout at the hub and at the free end, checked with a dial indicator. - [x] Clearance between the blade tips of a semi-open impeller and the wear plate. Beyond the manufacturer's tolerance means adjustment or a new impeller comes first. - [x] Balance holes on a closed impeller. Clogged holes raise the axial thrust and load up the thrust bearing. - [x] Axial play of the rotor and the condition of the front wear ring. - [x] Packing gland: packing not over-tightened, with a drip-rate leak present. Over-tightened packing runs hot and bends the shaft, and the phase drifts between starts. - [x] Mechanical seal: no dry running while turning the shaft, and no welding nearby. - [x] Piping strain on the nozzles and soft foot. Shims and shaft alignment between pump and motor come first, balancing after. - [x] Resonance. On a pump with a variable-speed drive, we check the whole working speed range, not just one point. - [x] Bearing housing temperature and the allowable limit for a magnet-mounted accelerometer. Sources: [ISO 13373-3:2015](https://www.iso.org/standard/40840.html) · [ISO 281:2007](https://www.iso.org/standard/38102.html) ## Hydraulics that look like impeller imbalance Half the calls about impeller imbalance end with a "flow" diagnosis, and that's checked without any disassembly. We ask you to hold one operating point across all starts and log the numbers: flow rate, suction and discharge pressure, valve position, level in the suction tank, media temperature. Then, if the process allows, we change the operating point and see what the vibration did. We calculate the blade-pass frequency right away: number of blades multiplied by rotational speed and divided by 60. A peak at this frequency is never removed by correction weights. A detailed breakdown of cavitation, recirculation and the operating point is in our separate article on hydraulic causes of pump vibration. | What we do at the operating point | If the cause is flow | If the cause is impeller mass | | --- | --- | --- | | We throttle the discharge valve | Level and sound change noticeably, crackling weakens or intensifies | Amplitude and phase of the running-speed component stay the same | | We raise the level in the suction tank or clean the inlet strainer | High-frequency noise and crackling go away | Nothing changes | | We change speed on the variable-speed drive | The peak moves along with the blade-pass frequency, unstable peaks appear below running speed | Amplitude grows roughly as the square of the speed, phase holds steady | | We do three consecutive starts at the same operating point | Amplitude and phase jump around | Repeatable in both amplitude and phase | | We look at the vibration-acceleration spectrum up to the kilohertz range | A continuous noise floor with no distinct peaks | A single clean peak dominating at running speed | > Without a fixed operating point, trial runs simply can't be compared to each other. The instrument sees a vibration change from a weight and a change from flow rate the same way, and the influence coefficient comes out false. ## How the on-site work proceeds 1. **Photos and data before the visit** — You tell us the impeller type, number of blades, speed and pump configuration, and send photos of the coupling assembly and front bearing. The photo shows whether a correction plane exists at all, which is half the answer. At the same time we clarify any constraints: no-welding rules, sanitary design, hazardous area. 2. **Inspection and cleaning, if the casing opens** — If the pump has already been opened, or the cover comes off quickly, we look at the blade inlet edges, disc thickness, and deposits in the channels. The question is often settled right here: the impeller needs flushing or replacing, not balancing. 3. **Measurement at a fixed operating point** — Two accelerometers on the bearing supports, laser phase sensor on the mark. We take overall vibration (total level across all frequencies), the running-speed component with phase, rotational speed, the spectrum — vibration broken down by frequency — and the time waveform. This is where we decide whether correction weights make sense. 4. **Mechanics and impeller fit** — We check the impeller nut, runout, clearances, packing gland, fasteners and shaft misalignment. We fix what we can on site, log the rest in the report, and take a repeat reading. 5. **Trial weight** — We place a trial weight of known mass at a known radius in the accessible correction plane. A valid trial run changes amplitude or phase by at least 20-30%. Less means the weight is too small, and we increase it. 6. **Correction** — The software gives mass and angle for each plane. On the impeller face, we compute metal removal by drilling at a pre-measured radius. On the coupling half, we work in fixed-position mode: instead of an angle, you get a position number and a washer mass. 7. **Confirmation run, assembly mark and report** — We start up at the same operating point and compare. If the level dropped but the tolerance hasn't been reached, we add a small trim correction to the weights already fitted. We mark the relative position of impeller and shaft so the next assembly doesn't undo the result, and we save this machine's influence coefficients. ## One plane or two, and where the correction planes are on a pump The impeller itself is almost always short: the length in the correction zone is less than half the diameter, and as a standalone part it balances in a single plane. But you're not balancing the impeller — you're balancing the assembled rotor, and from there the pump's configuration decides. On an overhung pump, the impeller sits beyond the bearing assembly. The unbalanced mass on the overhang produces a large radial reaction at the near bearing and an opposite-sign one at the far bearing. If you correct in only one plane, offset along the shaft from the location of the imbalance, part of the unbalance remains as a moment: vibration drops at one bearing while staying the same or growing at the other. That's why on overhung pumps we more often work in two planes, using the impeller or its nut and the coupling half. A between-bearings impeller and a multistage rotor produce an elongated geometry, where two planes are mandatory. Flexible rotors are a separate case: a result obtained at one speed doesn't carry over to another. That's covered in our article on critical speed and flexible rotors. - The outer face of the impeller's back disc. Closest to the plane of imbalance, with a large radius, so the required mass is small. Accessible only with the casing cover removed. - The impeller nut and the washers under it. The radius is small, so the same correction needs noticeably more mass. Suitable for small residual values. - The coupling-half bolts. The most accessible plane on a pump: we build up mass with washers, work in fixed positions, and don't weld or drill anything. - The free end of the shaft, a pulley, or a motor cooling fan on belt-driven units. - Factory-provided balance holes and washers, where the design includes them. Balance holes are not the same as the axial-relief holes. > To be honest about relocating the correction: if the imbalance sits in the impeller but only the coupling half is accessible, vibration will still drop as long as the rotor behaves as rigid. Accuracy falls off as the distance between impeller and coupling half grows, and as speed rises. On long multistage rotors this workaround often doesn't work at all, and we'll tell you that up front, from the numbers, not after the visit. Sources: [ISO 21940-11:2016](https://www.iso.org/standard/54074.html) · [ISO 21940-12:2016](https://www.iso.org/standard/50429.html) ## How we attach the correction mass on this rotor On a pump impeller, a welded weight is disallowed more often than allowed, so the main correction method here is metal removal. The software calculates not just the mass but also the drilling parameters at a given radius, so you get a diameter, depth and position — not just abstract grams. ### Metal removal by drilling The main method on an impeller. We drill the outer face of the back disc at a radius measured and recorded beforehand. We keep to the required remaining wall thickness, deburr, and restore the coating. We don't drill near the hub, near the balance holes, or in the taper-fit zone. ### Washers under the coupling-half bolts The fastest and fully reversible method. We build up mass with washers under the factory bolts; the instrument gives a position number instead of an angle, and we tighten the bolts to the specified torque. No welding, no drilling, no marks on the wetted part. ### Washers under the impeller nut Suitable for a small residual correction. The radius is small, so a large mass is needed, and at high speeds this method quickly runs into its limit. We always restore the nut's locking feature. ### Welding Only with your team's sign-off, and only outside the product-contact zone. We don't weld on stainless or lined impellers, on cast iron, near the mechanical seal, or in sanitary designs. In a hazardous area, any spark-producing work goes through a permit. - We don't glue or clamp weights in the wetted passage: the fluid and pressure would strip them off within a shift. - We don't place mass in a blade channel. You'd lose flow and create a new erosion site. - We don't drill the shroud of a closed impeller: it's thinner than the back disc and is often already eaten away by cavitation. - We don't touch the blade tips of a semi-open impeller: the clearance to the wear plate can't be restored. - We don't leave bare metal on an impeller used for drinking water or food production. Any intervention in the wetted part is agreed with your process engineer before the visit. > The correction location and radius go into the report along with the mass. Six months from now, or after the next teardown, you need to know not just "how much" but also "where," otherwise the next assembly starts from zero. We've covered the general rules for attaching correction weights in a separate article. ## When on-site impeller balancing won't work There are cases where correction weights either won't help or will do harm. We prefer to name them before the visit, not after. - Cavitation erosion has gone all the way through, or has eaten into the shroud. Drilling and loading such an impeller is dangerous, and balancing won't restore head or efficiency. Replacement is needed. - A blade is chipped or broken off, or there's a crack in the disc or hub. Replacement, no alternative. - Abrasive wear has eaten the walls down to uneven thickness around the circle. We can lower the vibration level, but it's a delay: the impeller's geometry is already compromised, and we'll note in the report that remaining life is limited. - The clearance on a semi-open impeller has gone beyond the manufacturer's tolerance. Adjustment or a new impeller comes first, then a measurement. - The fit on the shaft is damaged, the key is stripped, the nut won't hold. The running-speed component wanders from start to start, and no influence coefficient can be obtained. - A bent shaft. Imbalance here is a consequence, not the cause. - A blade channel is solidly plugged with fibre or crust. Cleaning comes first, and often balancing turns out unnecessary afterward. - The impeller is entirely inaccessible: a lined casing, a sealless chemical pump with a magnetic coupling, a submersible pump in a flooded housing. If there's no coupling half either, the only option left is disassembly and rotor balancing on a stand. - A sanitary or potable-water design with a ban on any intervention in the wetted part. We work only outside it, and if there's no access there either, the impeller gets swapped for one matched by mass. - The operating point won't hold steady: the pump cycles on and off, flow and pressure drift. Trial runs can't be compared to each other. > On-site balancing is strong precisely because the rotor is balanced in its own bearings, together with the shaft, coupling half and the frame's actual stiffness. An impeller removed and balanced on an arbor picks up new imbalance from fit eccentricity when reassembled. So we prefer to work on site, but not at any cost: if the right answer is a new impeller, you'll hear that from us. ## What you get, and how to order The result of the work is numbers and weight locations, not just "it's quieter now." Everything below goes into the report, and you can compare it against a fresh reading six months later. - Overall vibration and the running-speed component with phase, before and after, for each bearing support, in mm/s RMS (root-mean-square vibration velocity), plus an axial reading on an overhung pump. - Spectra before and after: you can see what went away and what remains at the blade-pass frequency and in the high-frequency range. - The operating point at the time of measurement: speed, flow, pressure, valve position. - Masses, radii and angular positions, or fixed-position numbers. For drilling: diameter, depth and location on the face. - Residual unbalance and an assessment against the selected balance quality grade G — the balance tolerance — where one is specified for the machine. - A machine-condition assessment by zones from A to D (A - good, D - unacceptable), stating the standard part and edition applied, the measurement points and operating conditions. - A mark of the relative position of impeller and shaft, so the next assembly doesn't undo the result. - This pump's saved influence coefficients: a repeat balancing job will go ahead without trial runs. - A list of what balancing did not resolve: seals, bearings, operating point, piping, impeller wear. - vibration diagnostics with a report cost 300 EUR per unit, balancing adds from 250 EUR. The minimum invoice for a visit is 500 EUR, because travel, sensor setup and diagnostics take time regardless of the number of machines. The calculator on the website gives an exact figure for your case, and several pumps in one visit work out cheaper: the instrument and sensors are already on site. - We're the engineers who design and manufacture Balanset instruments, and we do the on-site balancing ourselves. We're based in Vila Nova de Gaia, near Porto, and travel throughout Portugal. - [x] Impeller type: closed, semi-open, open; number of blades. - [x] Pump configuration: overhung, between-bearings, multistage, close-coupled. - [x] Speed, drive power, whether there's a variable-speed drive or belt drive. - [x] Media and temperature, whether it's abrasive and prone to buildup. - [x] What changed and when: after a repair, after cleaning, after an impeller change, or gradually. - [x] Photos of the coupling assembly, the front bearing, and the impeller itself if accessible. - [x] Constraints: no-welding rules, sanitary design, hazardous area, allowable shutdown duration. > If the photos and numbers show the impeller needs replacing rather than weights, we'll say so before the visit and suggest what to do instead of 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 operation manual](https://vibromera.eu/balanset-1a-operation-manual/) · [Balanset-1A manufacturer specification](https://vibromera.eu/product/balanset-1/) ## Frequently asked questions **Can an impeller be balanced without removing it from the shaft?** Yes, and it's the usual approach. We balance the rotor in its own bearing supports over two to three starts at running speed, placing the mass wherever it's accessible: the coupling-half bolts, the impeller nut, or the outer face of the back disc with the cover off. An impeller removed and balanced on an arbor picks up new imbalance from fit eccentricity when reassembled, so in-place balancing often gives a more accurate result. **We have a semi-open impeller. Can metal be removed from the blade tips?** No. The clearance between the blade tips and the wear plate sets the head and efficiency, and metal removed there can't be put back. We take mass from the outer face of the back disc, and if the cover can't be removed, we move the correction to the coupling half and tell you in advance what accuracy to expect. **The impeller is closed and there's no access to it. Will a weight on the coupling half help?** As long as the rotor behaves as rigid, yes: vibration at the bearings will drop. Accuracy falls off as the distance between impeller and coupling half grows and as speed rises, and on long multistage rotors this workaround often doesn't work at all. We decide from the reading and tell you plainly if the result won't hold. **We flushed the pump and vibration dropped. Is balancing still needed?** A measurement first. A drop in level right after flushing means buildup was causing the imbalance. If the running-speed component is within tolerance after cleaning, no weights are needed - a flushing schedule is. If a high running-speed component with a repeatable phase remains, then real imbalance was sitting underneath the buildup: wear, the fit, or the impeller's geometry. **Can a weight be welded onto a stainless-steel impeller in a drinking-water circuit?** We don't do that. Welding in the wetted part changes the metal's structure, leaves a zone that starts corroding and fouling, and it's not acceptable in a sanitary design. We work by drilling outside the product-contact zone, or by correcting at the coupling half, and we agree the method with your process engineer before the visit. **The pump handles abrasive media. How long will the balancing last?** Until the next noticeable wear. As the blades wear away, the mass distribution keeps shifting, and the correction weights are chasing a moving target. On such machines balancing gets built into the schedule together with wall-thickness and clearance checks, and the impeller gets replaced by wear, not by vibration. Saved influence coefficients make every following visit shorter and cheaper.