AXILINEHydraulic Causes of Pump Vibration: Cavitation, Vane-Pass Frequency, and Operating Point
The pump has started humming, the instrument reads 6.8 mm/s at the bearing housing, and someone suggests pulling the impeller and balancing it. For a pump, that call is wrong more often than for any other machine. The impeller works inside a flow, and the flow generates forces of its own: they shake the rotor, the casing, and the piping, but they have nothing to do with mass distribution. Below we go through how to tell hydraulics apart from imbalance without taking the pump apart, and what to check first.
A Pump Doesn't Shake from Imbalance Alone
Hydraulic forces in a pump fall into three groups. First: a radial force in the volute casing that pulls the shaft sideways and depends on flow rate. Second: periodic pressure impulses from vanes passing the diffuser tongue or volute cutwater, which act at the vane-pass frequency and its harmonics. Third: random forces from the collapse of vapour cavities and from vortex shedding, which produce broadband noise with no distinct lines.
The tricky part is that some of these forces repeat exactly once per revolution. An uneven gap around the circumference, a clogged interblade channel, an eccentric impeller fit, a worn wear ring — all of these add to the running-speed component, 1x, vibration at exactly the rotation frequency. You see a high 1x, install weights, the software honestly reports "within tolerance," and a day later the level is back, because the cause is in the flow, not in the mass.
Hence the rule for working with a pump: before you touch a trial weight, you need to know not just the speed but the operating condition — flow rate and pressure. Without those numbers, the diagnosis is a guess.
- Flow-driven radial force: minimal near the best-efficiency point, grows with deviation in either direction, loads the bearings and mechanical seal to one side.
- Vane-pass frequency f_v = z · n / 60, where z is the number of impeller vanes and n is the speed in rpm. Present on any healthy pump.
- Cavitation and flow separation: a broadband rise across roughly 1 to 10 kHz, sometimes modulated by the vane-pass frequency.
- Low-flow recirculation: unstable subsynchronous components in the 0.5–0.9× running-speed range.
- Pressure pulsations: shake the piping first and foremost, and come back onto the pump casing through the flanges.
- Air in the flow: readings jump around, and head and motor current drift along with them.
Hydraulics and imbalance coexist just fine. A year of cavitation eats away unevenly at the vane inlet edges, and you end up with genuine imbalance as a consequence. The order of work doesn't change because of this: first sort out the flow, then take a fresh measurement, and only then decide about weights.
Cavitation: How to Recognize It Without Disassembly
Cavitation begins where the absolute pressure at the impeller inlet drops to the fluid's vapour pressure. Bubbles boil up in the flow, get carried into the high-pressure region, and collapse at the vane surfaces. Every collapse is a micro-impact on the metal.
Your ear works faster than an instrument here. A cavitating pump sounds as if it's pumping gravel along with the water: an uneven crackle that changes with the valve position. A steady hum on one note is not cavitation.
In the spectrum — the breakdown of vibration by frequency — cavitation doesn't look like a peak but like a pedestal: a broadband rise across roughly 1 to 10 kHz with no discrete lines, often modulated by the vane-pass frequency. To see it, raise the spectrum's upper limit, Fmax, into the kilohertz range and look at acceleration rather than velocity: in mm/s the high-frequency part is barely visible. We've written separately about choosing Fmax and frequency resolution.
Confirm the diagnosis with the operating condition, not a guess. Throttle the discharge valve: flow drops, the required NPSH drops, and the crackle usually weakens. Raise the level in the suction tank, or let the fluid cool: cavitation recedes because the margin to vapour pressure grows. Clean a clogged suction strainer: the suction head is restored, the crackle disappears. Imbalance doesn't react to any of this at all.
Inspection settles the question for good. Cavitation erosion leaves a characteristic pitted, rough surface on the vane inlet edges and on the shroud, not the smooth wear pattern of abrasion. If those marks are there, the machine has been cavitating for a long time, and the impeller is almost certainly already asymmetric in mass.
- Sound: an uneven crackle, "gravel in the flow," changes together with the valve.
- Acceleration spectrum: a broad rise across 1–10 kHz with no distinct lines.
- Suction pressure gauge: absolute pressure close to the vapour pressure at the operating temperature.
- NPSH margin: actual suction head minus the required value from the manufacturer's curve at the actual flow rate. As a practical guideline, a margin of 0.5–1 m or 10–30% is often used; check the pump documentation for the exact requirement.
- Hot fluid: vapour pressure rises with temperature, and a pump that ran fine in winter starts cavitating in summer.
- The suction filter or strainer is partly clogged, or the suction valve is throttled.
- Vane inlet edges: pitted surface, cavities, local thinning.
Balancing doesn't cure cavitation, and the reason is simple. The cavities collapse randomly in time and location, so this force isn't tied to the rotor's angular position. A correction weight creates a force strictly once per revolution, and there's nothing for it to counter. You'll reduce 1x while the crackle and the high-frequency noise stay exactly the same, and the impeller keeps deteriorating.
Operating Off the Design Point: Radial Force and Recirculation
A pump is designed around one point on its performance curve. At the best-efficiency point, flow enters the vane at the design angle, and the radial force in the volute casing is at a minimum. Move away from that point, and the picture changes in both directions.
At low flow, when the valve has been throttled or the system has scaled up, recirculation sets in: part of the flow turns back on itself, at both the impeller inlet and outlet. Strong vortices appear and shed irregularly. In the spectrum this shows up as unstable subsynchronous components at 0.5–0.9× running speed plus a raised broadband floor that looks a lot like cavitation. Recirculation damage also resembles cavitation damage, except it sits on the other side of the vane.
At the same time the radial force grows. The shaft deflects, the mechanical seal works at an angle, the bearing takes a constant load to one side. From there the chain is predictable: seal leakage, bearing wear, growing clearances, and a rising 1x that you'll mistake for imbalance.
A fully open valve isn't any safer. The required NPSH rises with flow rate, and the pump slides into cavitation simply because it's pumping more than its design flow, while the motor can exceed its rated current at the same time.
Signs of Low Flow
Valve throttled, flow noticeably below design, motor current below rated, casing running hot. Subsynchronous components at 0.5–0.9x in the spectrum, unstable in frequency and amplitude, plus a raised noise floor. The level changes noticeably when you move the valve.
Signs of High Flow
Valve fully open, discharge pressure low, motor current above rated. The crackle gets louder as you open the valve further. Check the required NPSH from the curve at the actual flow rate — it has risen along with the flow.
Minimum Stable Flow
The manufacturer specifies this separately. As a working guideline, noticeable recirculation problems begin below roughly half the best-efficiency-point flow, but the actual threshold is set by the specific pump. A bypass line solves the problem better than a permanently throttled valve.
If the Operating Point Can't Be Restored
Sometimes the pump simply wasn't selected to match the system. In that case the operating condition isn't corrected with the valve: the impeller's outer diameter is trimmed, the impeller is replaced, or a variable-frequency drive is installed. Keep one thing in mind: after trimming, the impeller needs to be rebalanced, because the metal removal almost never comes out symmetrical.
Vane-Pass Frequency: Calculate It Before You Reach for Weights
Every healthy pump has a vane-pass frequency, and on its own it's not a defect. It takes ten seconds to calculate: f_v = z · n / 60. A five-vane impeller at 2950 rpm gives a running speed of 49.2 Hz and a vane-pass frequency of 5 · 49.2 = 246 Hz.
You have to calculate it, otherwise you'll confuse the vane-pass frequency with a harmonic of running speed. On a six-vane impeller it lands exactly on 6x, on a four-vane one on 4x, and on a two-vane screw-type impeller it sits right on 2x — exactly where you'd be looking for misalignment. The same spectral peak gets two opposite diagnoses, and which one is correct depends on the vane count.
The vane-pass amplitude rises with both geometry and operating condition. The main culprit is the radial gap between the impeller's outer diameter and the diffuser vane or volute cutwater. The smaller the gap, the sharper the pressure impulse as each vane passes, and the higher the peak. The gap changes on its own: the impeller shifts along the shaft, the cutwater wears from the flow, the casing wears.
The second factor is the ratio between the number of impeller vanes and the number of diffuser vanes. Unfavourable combinations, especially equal counts, produce strong pulsations at the vane-pass frequency and its second harmonic. This can't be fixed on site: it calls for changing the impeller to a different vane count or trimming the cutwater, and that has to be done in consultation with the manufacturer.
| What You See in the Spectrum | What It Means | What to Do |
|---|---|---|
| Peak at z·1x is moderate and stable, not rising in trend | Normal for this pump | Record it as the baseline level and leave it alone |
| Peak at z·1x has grown several times over, 2·z·1x and 3·z·1x have appeared | The gap to the diffuser vane or cutwater has shrunk, the cutwater is worn, or the impeller has shifted along the shaft | Measure the clearances and the impeller's axial position at the next teardown |
| Peak at z·1x grows with deviation of flow from the design value | Operating condition, not geometry | Restore the operating point, check the valves and the bypass |
| z·1x plus a broadband rise and modulation | Cavitation or recirculation layered on top of the normal vane-pass frequency | Deal with suction head and flow, not with mass |
| Sharp rise in both 1x and the vane-pass frequency at once | A broken or cracked vane is possible | Shut down and inspect. Never balance an impeller with a crack |
The vane-pass frequency is never removed with weights. A correction mass acts at running speed, while the pressure pulsation lives at z·1x, and there's no link between the two. All that balancing such a pump will get you is a smaller share of 1x in the overall level (total vibration across all frequencies), while the vane-pass peak stays exactly where it was.
Pressure Pulsations and Piping Resonance
Some complaints about "pump vibration" aren't really about the pump. The impeller generates pressure pulsations at the vane-pass frequency, which travel into the piping and excite it. The piping responds with its own natural frequencies: spans between supports vibrate like beams, and the fluid column has acoustic natural frequencies that depend on the length of the runs and the speed of sound in the fluid.
The check is simple and nearly free. Take a sensor and walk the line with it: measure at the bearing housings, at the flanges, at the midpoints of spans, at supports, and on the header. If the level at the midpoint of a span is several times higher than at the pump bearing housing, you've found a poorly secured pipe, not a rotor defect.
Resonance is confirmed by changing the speed. If the pump is on a variable-frequency drive, sweep through the range and watch the amplitude: resonance produces a sharp, narrow peak within a small speed band, with vibration dropping on either side of it. From there you either move off that speed or modify the structure, because the cause isn't in the rotor.
Vertical pumps with a long column are a separate story. Their first bending natural frequency often sits close to running speed, and a machine like that amplifies any residual imbalance. We have a separate article on the signs of resonance and ways to shift it.
- Measurements not only at the supports but on the pipe itself: flanges, midpoints of spans, bends, header.
- Spans between supports: sag, broken or missing clamps, torn-off hangers.
- Supports that have effectively become rigid restraints and transmit thermal-expansion forces onto the pump flanges.
- Expansion joints and flexible connectors: intact, not over-tightened, not acting like a rigid link.
- Check valve: banging on closure, water hammer on shutdown.
- Suction line: an elbow right before the inlet nozzle, a short straight run, an air pocket at a high point.
- Pumps running in parallel: pulsations from a neighbouring unit arrive via the shared header.
Air, Clogging, and Uneven Wear
These four causes produce the most confusing picture, because part of their contribution lands right on the running-speed component. This is exactly where a balancing job that solves nothing gets ordered most often.
Air in the Flow
Air drawn in through the shaft seal, a vortex from insufficient submergence of the suction nozzle, an air pocket at a high point in the suction line. Vibration is unstable, and head and motor current drift along with it, while the sound is dull and "hollow," not as sharp as with cavitation. Check the tank level, submergence depth, the tightness of the flanges and packing, and the operation of the air release valve.
Clogged Impeller Channel
Fibre, rags, or plastic lodge in one interblade channel. You get two effects at once: genuine imbalance and hydraulic asymmetry. The tell-tale sign: vibration jumped rather than climbed gradually, and head dropped at the same time. Clear the impeller and re-measure. Balancing a clogged impeller is pointless — you'll have to remove the weights again after clearing it.
Uneven Wear and Erosion
Abrasive slurry and cavitation remove metal from the vanes unevenly, and 1x rises slowly, over months. Balancing does work here, but it's maintenance, not a fix: as long as the source of abrasive or the cavitation is still there, you'll be back at this pump. Plan for an impeller replacement or a protective coating.
Wear-Ring Clearance Wear
Worn wear rings increase internal recirculation: head drops while flow through the clearance rises. At the same valve position, the pump slides down its performance curve, and you end up with an operating condition you never chose. Measure head and flow, not just vibration: a drop in head at unchanged speed points to worn clearances.
Inspecting the impeller is mandatory before balancing. A crack at the vane root, local thinning from erosion, a delaminated coating — with any of these defects the impeller needs to be repaired or replaced, not balanced. A weight on a cracked impeller only masks the growing defect until it fails.
How to Tell Hydraulics from Imbalance in One Hour
One technique helps separate these causes: change the operating condition, not the rotor. Imbalance is a property of the rotor: it depends only on speed and knows nothing about the valve. Hydraulics is a property of the flow, and it responds to any change in flow rate, pressure, and level.
Here's the sequence. Take a measurement at the operating point: overall level and 1x at each bearing housing, phase (the angular tie between vibration and the rotor's revolution), spectrum, speed. Then throttle the discharge valve in steps, and at each step record vibration together with pressure and flow, waiting for the readings to settle. Three to four steps is enough.
Read the result like this. If 1x barely changed while broadband noise and subsynchronous components rose or fell, the cause is hydraulic. If 1x stays at the same level regardless of valve position and grows as the square of speed, the imbalance hypothesis is alive, and balancing will work. The logic behind this sign-based separation is laid out in the ISO guides on vibration diagnostics; check the applicable part and edition for your machine.
| Sign | Hydraulics | Imbalance |
|---|---|---|
| Dependence on valve position and flow rate | Level changes noticeably | Level barely changes |
| Dependence on suction head, level, and fluid temperature | Changes, up to the crackle disappearing entirely | Doesn't change |
| Frequency content | Broadband rise, subsynchronous components, a peak at z·1x | One dominant peak at 1x |
| 1x phase from run to run | Drifts together with the operating condition | Repeats within a few degrees |
| Sound | Crackling, hissing, "gravel in the flow" | A steady hum on one note |
| Response to a trial weight | Amplitude and phase change unpredictably, the influence coefficient doesn't repeat | A repeatable change of at least 20–30% in amplitude or 20–30° in phase |
| Share of 1x in the overall level | Usually less than half | Typically more than 70% |
The trial run itself doubles as a linearity test. If, after installing a known weight, the 1x amplitude and phase change predictably and repeat from run to run, the system is linear and balancing will proceed normally. If the response is different every time, you're not fighting mass, and further runs will just eat up the shift. We've written separately about the 20–30% trial-weight criterion and about influence coefficients.
Sources: ISO 13373-3:2015 · Balanset-1A operation manual
What to Record Alongside Vibration
Pump vibration without process parameters is a number without meaning. A month later you're looking at a rise from 2.1 to 5.8 mm/s and can't tell whether the impeller wore down or an operator throttled the valve because a second pump got started on the same header in the next shop. Context gets written down at the moment of measurement — it can't be reconstructed afterward.
The minimum set fits in one logbook line and takes five minutes to record, if gauges and a flow meter are already installed. If they aren't, that's the first thing worth installing: diagnosing a pump without suction and discharge pressure turns into guesswork.
- Actual speed from a tachometer, not the figure off the nameplate. Especially on a variable-frequency drive.
- Suction and discharge pressure, noting the elevation where the gauges are mounted.
- Flow rate. If there's no flow meter, record at least the valve position and the number of pumps running in parallel.
- Motor current and power: they show where you are on the performance curve.
- Temperature of the pumped fluid and of the bearing housings.
- Level in the suction tank and the submergence depth of the suction nozzle.
- Condition of the suction filter or strainer, and the date it was last cleaned.
- Overall level and 1x with phase at each bearing housing, in mm/s RMS (root-mean-square) over the 10–1000 Hz band, plus a separate acceleration spectrum with Fmax in kilohertz for the high-frequency part.
- What you heard and saw: crackling, hissing, a banging check valve, a shaking pipe, a seepage stain at the seal.
Calculate the NPSH margin right on the spot: convert the absolute suction pressure into metres of head of the pumped fluid, subtract the vapour pressure at the operating temperature, and compare against the required NPSH from the manufacturer's curve at the actual flow rate. If there's no margin, there's nothing further to look for, and weights certainly won't be needed. Assess the overall level per the applicable part of ISO 20816: this series has a dedicated part for rotodynamic pumps, so check which part and edition apply to your machine.
Sources: ISO 20816-1:2016
Correction Sequence and Where Balancing Fits In
Send us the pump type, speed, impeller vane count, suction and discharge pressure, flow rate, and vibration level with the measurement point noted. With this data we can usually tell, from the first conversation, whether it looks like imbalance or hydraulics. If it's hydraulics, you don't need a balancing visit, and we'll tell you so. Consulting support on the method and the instrument is included with the purchase.
When In-Situ Balancing Is Genuinely Needed
The impeller was trimmed on its outer diameter for a new operating point. The impeller was replaced or repaired by welding. Wear and erosion have progressed for years under normal operation, and 1x has risen gradually. The motor rotor was rewound. In all these cases the cause is in the mass, and the issue is closed without sending the impeller to a machine shop. Set the tolerance deliberately here: residual imbalance by G class per ISO 21940-11 and the overall level per the applicable part of ISO 20816 are two different things.
How We Help
We're engineers who design and manufacture Balanset instruments and use them to do field balancing ourselves. We come out to the pump, measure vibration at the bearing housings and along the piping, separate out the overall level from 1x, calculate the vane-pass frequency from the vane count, and look at the spectrum and phase. If the picture is hydraulic, we say so directly and explain what to check in the operating condition, instead of selling you a balancing job.
If You'd Rather Work It Out Yourself
Balanset-1A is two accelerometers, a laser phase sensor using a reflective tape mark, a two-channel USB module, and Windows software. The instrument gives you overall vibration and 1x with phase, speed, the FFT spectrum, and the time waveform; it calculates trial and correction weights in one and two planes, splits the weight across fixed positions, calculates drilled-hole corrections, stores influence coefficients for trim balancing, and keeps an archive for reports. With one instrument you both make the diagnosis and carry out the correction when it's actually needed.
- 01
Restore the Operating Point
Open the valve to the design flow rate, or set up a bypass. Check against the manufacturer's curve and the minimum stable flow. This is the cheapest step and the most common answer.
- 02
Check the Suction Side
Suction head, tank level, submergence depth, filter, air pockets, tightness of the packing and flanges. Calculate the actual NPSH margin at the actual flow and temperature.
- 03
Inspect the Impeller and Clearances
Cavitation erosion at the inlet edges, cracks at the vane roots, clogged channels, wear-ring wear, the radial gap to the diffuser vane or cutwater, the impeller's axial position on the shaft.
- 04
Address the Piping
Supports and clamps along the spans, expansion joints, the absence of imposed forces at the flanges, the check valve. Check for span resonance by changing the speed.
- 05
Close Out the Mechanics
Soft foot, foundation bolt tightness, alignment of the pump and motor shafts accounting for thermal growth, bearing condition. Misalignment produces its own 2x and axial vibration, and balancing won't remove it.
- 06
Re-Measure, and Only Now Decide About Weights
If, after all the steps, 1x still accounts for most of the overall level, the imbalance is real. In-situ balancing, in the pump's own bearings, will remove it in two to three runs without disassembly.
Sources: ISO 21940-11:2016 · Balanset-1A manufacturer specification
Frequently asked questions
Can a cavitating pump be balanced?
Technically yes, and the software will even show "within tolerance," because the residual 1x will drop below the target value. There's no practical point to it. The crackle, the 1–10 kHz broadband noise, and the impeller erosion will remain, and within a few months the uneven metal removal will create a new imbalance, and you'll be back here again. First remove the cavitation: suction head, flow, temperature, the inlet filter. Then re-measure and decide about weights.
How do you tell whether a spectral peak is the vane-pass frequency or a harmonic of running speed?
Calculate f_v = z · n / 60 using the actual vane count and actual speed. If the vane count is a small whole number, the vane-pass frequency lands exactly on a multiple of running speed, and the spectrum alone won't tell them apart. In that case, work with the operating condition: move the valve and see what changes. The vane-pass frequency and the broadband floor respond to flow rate, while a harmonic from misalignment or looseness stays put.
The pump got noisier after the valve was throttled. Is that dangerous?
Yes, and the danger isn't in the noise. At low flow, recirculation sets in, the radial force in the volute casing grows, the shaft deflects, the mechanical seal works at an angle, and the bearing takes a constant load to one side. A few months later you're paying for a seal and a bearing. The right fix is a bypass line or a speed change, not a permanently throttled valve.
Vibration on the pipe is higher than on the pump casing. What should be done?
Look for the cause in the piping, not in the rotor. Check the supports and clamps along the spans, the hangers, the expansion joints, and the thermal-expansion forces at the flanges. Span resonance is confirmed by changing the speed: a sharp, narrow peak within a small band and a drop in level on either side of it. Rotor balancing won't remove this vibration, because the source is the pressure pulsation at the vane-pass frequency, and the amplifier is the pipe itself.
What NPSH margin counts as sufficient?
The one specified by the pump manufacturer for the actual flow rate. As a practical guideline, a margin of 0.5–1 m or 10–30% above the required NPSH is often used, but that's exactly a guideline, not a standard. Calculate the margin at the fluid's operating temperature: vapour pressure rises with temperature, and a pump that ran fine in winter starts cavitating in summer at the same tank level.
Is a balancing instrument enough to sort out hydraulics?
For making the diagnosis, a two-channel system with a phase sensor is enough: you get the overall level and 1x at each bearing housing, phase, speed, and the spectrum. That's enough to calculate the 1x share, find the vane-pass peak, and see the broadband rise from cavitation, provided you raise Fmax into the kilohertz range and look at acceleration. What the instrument won't give you is pressure, flow, and tank level. You take those readings with gauges and a flow meter and record them alongside the vibration — otherwise you won't be able to separate hydraulics from imbalance.
Related content
Electrical Causes of Motor Vibration: How to Separate Them from Mechanics
Electrical vibration gives itself away through a frequency tied to the mains supply, not to the shaft: a line at twice the supply frequency (100 Hz on a 50 Hz supply) and sidebands around 1x (running speed) spaced at the pole-pass frequency F_p = s · f_line · P, where s is slip and P is the number of poles. This is checked with a single shutdown: cut the power and keep recording vibration through the coast-down, while the shaft slows under its own inertia. The electromagnetic component cuts off within a fraction of a second; the mechanical component decays smoothly along with the speed. Balancing doesn't remove an electrical cause at all, because the source of the force isn't the rotor's mass, and a weight on the rotor has no effect on that source.
How to read a vibration spectrum: 1x, 2x and harmonics
Convert the frequency of every peak into a multiple of running speed. If 1x dominates and overall vibration sits close to 1x, that is unbalance, and balancing makes sense. A clear 2x plus a rise in axial vibration usually points to shaft misalignment, which needs shaft alignment, not balancing. A comb of 1x, 2x, 3x with half-order harmonics and a raised noise floor means mechanical looseness, cracks or rubbing. Peaks higher up the frequency range that are not multiples of running speed are usually bearings. The spectrum gives you a hypothesis; the diagnosis is confirmed with the time waveform, phase, repeatability and history.
On-site pump balancing: impellers, rotors, and shafts of pump units
Yes, we balance pump rotors on site, in their own bearing supports, with no dismantling and no sending the impeller to a machine shop. Three conditions apply. First: most of the vibration has to come from the 1x running-speed component — vibration at the rotor's rotational frequency, the main sign of imbalance. Second: at least one correction plane has to be accessible, meaning a spot where a weight can be fitted — usually the coupling bolts or the impeller nut. Third: the pump has to be able to hold a stable mode with a constant flow rate. If the level is being driven by cavitation, operation away from the best-efficiency point, shaft misalignment, or worn wear rings, we'll show that with a measurement and say plainly: weights won't help.
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.