Cavitation is what happens when the pressure at a pump’s suction falls low enough for the liquid to boil, forming vapour bubbles that collapse violently as they reach the higher pressure inside the impeller. It is not a sudden failure. It announces itself in a predictable order, and the first two symptoms arrive long before anything is physically damaged. The plants that catch it there replace a calculation. The plants that miss it replace an impeller.
TL;DR
The first symptom of cavitation is audible, a rattling or gravel-like sound at the casing. Then discharge pressure and flow start fluctuating. Then the pump measurably drops below its performance curve. Only at the end do you see pitting on the impeller. Either of the first two symptoms is the point to check available NPSH against required NPSH, before the damage becomes physical. The published NPSHR on a pump curve is not the point where cavitation begins. It is the point where it has already cost you 3% of head.
What is actually happening inside a cavitating pump?
As liquid flows through a centrifugal pump’s impeller, static pressure drops, most sharply at the inlet to the vane passage. KSB attributes the extent of that drop to rotational speed, fluid density and viscosity, the impeller’s inlet geometry, the operating point, and the velocity profile of the approach flow. If the pressure upstream of the impeller falls to the liquid’s vapour pressure, the liquid boils.
The bubbles do not stay bubbles. They travel into the higher-pressure region of the impeller and collapse, and each collapse is a small implosion against the metal. Repeat that a few thousand times a second and you get material loss on the vane leading edge, which is the pitting that engineers eventually photograph and send to us.
The damage is not confined to the impeller. KSB’s own planning literature notes that prolonged operation past the head breakdown point damages impeller, bearings and shaft seal. That matters commercially, because a plant that treats cavitation as an impeller problem will replace the impeller and keep the fault.
What are the signs of cavitation, in the order you will meet them?
This sequence is the most useful thing in this article, because it is ordered by how early you can act, not by how obvious the symptom is.
| Stage | What you observe | What it costs to act now |
|---|---|---|
| 1 | Rattling or gravel-like sound at the casing | A calculation and possibly a valve or pipe change |
| 2 | Fluctuating discharge pressure and flow | A calculation and possibly a valve or pipe change |
| 3 | Measurable drop below the published performance curve | Impeller trim, speed change or reselection |
| 4 | Visible impeller pitting on inspection | Replacement parts and the downtime to fit them |
Stage one sounds like gravel being pumped. That is not a metaphor for the sound, it is the standard description, and once you have heard it on a running pump you will never mistake it. Stage two is the one most often blamed on something else. Fluctuating discharge pressure gets attributed to the control valve, the level in the sump or the instrument.
Either of the first two is the point to check available NPSH against required NPSH. Not stage three, and certainly not stage four. By stage four you are no longer diagnosing, you are costing.
One caution on stage one. Rattling and vibration can also mean bearing wear or misalignment, and a drop in flow can also mean impeller wear or internal leakage. Cavitation is one of several explanations for the early symptoms, which is why the next two sections matter. On solids-bearing duties a non-clog pump adds a further explanation, since partial blockage produces very similar readings.
Why is NPSHR on the curve not the point where cavitation starts?
Because of how NPSHR is defined. KSB lists four possible cavitation criteria for specifying NPSHR: incipient cavitation, a certain extent of the cavitation zone on the vanes, the start of head drop, and a cavitation-induced head drop of 3%. It then states plainly that the first three are less common, that proving incipient cavitation requires demanding and expensive testing, and that it is commonly agreed NPSHR equals NPSH3.
Read that again in operational terms. The NPSHR curve your supplier printed is the suction head at which cavitation has already reduced the pump’s head by 3%. It is a reproducible test endpoint, not a cavitation-free boundary. Cavitation begins at a considerably higher NPSH than the number on the sheet.
KSB puts the consequence explicitly: if NPSHA is higher than the incipient value, cavitation will not develop and the impeller runs without bubble formation, and the lower NPSHA falls, the longer the bubble trail becomes. So the honest reading of the curve is that satisfying NPSHA greater than NPSHR keeps you out of head breakdown. It does not promise you a pump with no bubbles in it.
That is why a margin above NPSHR is engineering, not padding. It is also why “NPSHA exceeds NPSHR, therefore the pump is fine” is one of the most common wrong conclusions in a design review.
Two more things the curve does not tell you. NPSHR is measured on clean cold water, so it does not transfer directly to another fluid, particularly one carrying dissolved or undissolved gas. And KSB notes that measured NPSH3 for hydrocarbons and hot water is lower than for cold water, so a cold-water acceptance test is not the same test as your service.
What causes cavitation in a centrifugal pump?
Most of what we see traces back to selection rather than operation. Oversized selection is common. Buyers add margin to be safe, which lands the duty point away from the best efficiency point, raising energy cost and mechanical wear and introducing cavitation risk at low flow.
There is a mechanism behind that, and KSB describes it. Every impeller has a shock-free flow rate at which the approach flow enters the vanes cleanly. Below it, cavitation develops on the vane’s suction side. Above it, on the pressure side. Running away from the best efficiency point in either direction moves you off shock-free entry, and the required NPSH rises accordingly. An oversized pump throttled back to the flow you actually need is running exactly where it should not be.
The rest is suction-side installation, and it is where the cheapest fixes live:
-
Suction lift and pipe run. Locate the pump as close to the liquid source as possible, and minimise lift. Every metre of static lift comes straight off NPSHA.
-
Pipe sizing and fittings. Suction and delivery pipes must follow the manufacturer’s recommended sizes. NPSHA is reduced by entry losses and by pressure drops across valves and fittings, so an undersized suction line, a strainer nobody cleans or a sticking reflux valve is a live cavitation cause.
-
Air pockets. Use eccentric reducers at the suction and avoid short bends close to the flange. A concentric reducer on a horizontal suction line traps air at the top and feeds it to the impeller.
-
Poor suction geometry by design. Basement-level pump installations in steel plants routinely have poor suction conditions, which is why we specify low NPSH designs such as end suction process pumps there rather than arguing with the building.
How do you confirm it is cavitation and not something else?
Run the shut-off head check first, because it takes minutes and needs no instruments beyond the discharge gauge. Most performance charts show a shut-off head, the maximum pressure the pump generates with the delivery valve closed. Close the delivery valve and read the pressure. If it matches the shut-off head on the chart, the pump itself is generating what it should, and the problem is in the piping or the system rather than in the pump’s hydraulics.
That single test separates two very different jobs. A pump making its shut-off head with fluctuating flow points you at the suction side, the strainer, the sump level or the valve. A pump that cannot reach shut-off head points you at wear rings, impeller condition or internal leakage.
Then run the differential. Rising power consumption with falling output usually means hydraulic inefficiency or internal wear rather than cavitation. Overheating bearings or motor casing usually means lubrication, overload or a cooling problem. Visible leakage at the seal is its own fault. Cavitation is specifically the combination of noise at the casing plus fluctuating discharge, with the shut-off head intact.
Finally, check the duty point against the curve rather than against memory. Duty points should always sit on or below the pump’s head-flow curve for the impeller diameter actually fitted, and it is worth confirming which diameter that is. Head, flow and power all change with impeller diameter, which is why the chart carries a separate curve for each one.
How do you fix it?
There are only two categories of fix, and knowing which one you are doing keeps the conversation honest.
| Approach | What it changes | Typical measures |
|---|---|---|
| Raise NPSHA | The system delivers more suction head | Raise the source level or lower the pump, shorten and upsize the suction line, remove or clean restrictions, fix eccentric reducers and short bends, lower the liquid temperature where the process allows |
| Reduce NPSHR | The pump needs less suction head | Move the duty point back toward best efficiency point, reduce speed, select a larger impeller eye or a low NPSH design, use a vertical turbine or submersible arrangement where suction geometry cannot be fixed |
We prefer to frame the correction around total cost of ownership rather than blame. Show the customer the energy delta between the actual operating point and the best efficiency point, then present right-sizing, impeller trimming or adding a variable frequency drive as optimisation rather than as a correction of somebody’s mistake. It is a more accurate description of the situation and it gets accepted more often. Where a drive is added, the motor has to be checked for converter duty at the same time.
Where suction conditions cannot be improved, the answer is usually a different pump arrangement rather than a bigger one. A vertical turbine pump puts the first stage below the liquid level, which removes the suction lift problem instead of managing it.
What we ask before we requote a pump
New India Electricals Ltd has supplied pumps and pump systems to demanding industrial applications since 1969, including centrifugal pumps for steel rolling mills across India, boiler feed and cooling water duties in power and process plants, multistage and fire fighting sets, and water distribution and pump station projects in Malawi and Zambia. Six things let us give you a real answer rather than a catalogue number.
-
The actual duty point, flow and head, measured rather than designed. Not the rating of the pump being replaced.
-
The full suction arrangement. Static lift or flooded suction, pipe size and length, number and type of fittings, strainer condition, and the sump or tank level range.
-
The fluid. Temperature, specific gravity, viscosity, solids content and whether it carries dissolved or entrained gas. All four change NPSHR.
-
Site altitude, because atmospheric pressure at the source is part of NPSHA and a Copperbelt or highland site starts with less of it than a coastal one.
-
What you are hearing and seeing, using the four-stage sequence above, plus the shut-off head reading.
-
The nameplate and curve of the existing pump, photographed, with the impeller diameter if it is recorded. We hold more than 4,000 motors in ready stock, so a pumpset rebuild rarely waits on the drive end.
Send those and we will tell you whether you have a selection problem, a piping problem or a wear problem. Contact our team to discuss an application.
Frequently asked questions
- What is cavitation in a pump?
- Cavitation is the formation and violent collapse of vapour bubbles inside a pump, caused by the liquid boiling when suction pressure falls to its vapour pressure. The bubbles form at the impeller inlet where static pressure is lowest, then collapse as they reach higher pressure inside the impeller. Each collapse is a small implosion against the metal, and repeated collapse removes material from the vane leading edge.
- What are the first signs of pump cavitation?
- The first symptom is audible, a rattling or gravel-like sound at the casing. The second is fluctuating discharge pressure and flow. The third is a measurable drop below the published performance curve, and the fourth is visible impeller pitting on inspection. Either of the first two is the point to check available NPSH against required NPSH, before the damage becomes physical.
- What is the difference between NPSHA and NPSHR?
- NPSHA, net positive suction head available, is what the system delivers at the pump suction, and it depends on source pressure, static lift, pipe and fitting losses and the liquid’s vapour pressure. NPSHR, required, is what the pump needs, and it is measured by the manufacturer. Continuous operation requires NPSHA to exceed NPSHR, with a margin.
- Does NPSHA greater than NPSHR mean no cavitation?
- No, and this is widely misunderstood. KSB notes that the industry commonly agrees NPSHR equals NPSH3, the suction head at which cavitation has already reduced head by 3%. Cavitation actually begins at a higher NPSH than that. Meeting NPSHA greater than NPSHR keeps you out of head breakdown, but a margin above NPSHR is what keeps bubble formation down.
- What causes cavitation in a centrifugal pump?
- Most commonly, selection. An oversized pump runs away from its best efficiency point, and required NPSH rises either side of the shock-free flow rate at which liquid enters the vanes cleanly. The rest is suction-side installation: excessive static lift, undersized or long suction pipe, dirty strainers, concentric reducers trapping air, and short bends close to the suction flange.
- How do you stop a pump cavitating?
- Either raise NPSHA or reduce NPSHR. Raising NPSHA means lowering the pump or raising the source, shortening and upsizing the suction line, clearing restrictions and correcting the fittings. Reducing NPSHR means moving the duty point back toward best efficiency point, reducing speed, or selecting a low NPSH design. Where suction geometry cannot be fixed, a vertical turbine or submersible arrangement removes the lift entirely.
- Can I use the NPSHR curve for a fluid other than cold water?
- Not directly. NPSHR is established by testing with clean cold water, and converting it to another fluid, another speed or another pump size is itself a matter of testing. Fluids carrying dissolved or undissolved gas behave differently. KSB also notes that measured NPSH3 for hydrocarbons and hot water is lower than for cold water, so a cold-water acceptance test does not represent those services.
- How can I tell cavitation from ordinary pump wear?
- Run the shut-off head check. Close the delivery valve and compare the pressure against the shut-off head on the pump chart. If it matches, the pump’s hydraulics are intact and the fault lies in the piping or system, which points toward cavitation or a suction problem. If the pump cannot reach the shut-off head, suspect wear rings, impeller condition or internal leakage instead.
