A pump that is cavitating is not failing because the material is wrong. It is failing because the static pressure at one point inside it dropped below the pressure at which the liquid turns to vapour, and the bubbles that formed there are collapsing against a solid surface a fraction of a second later. The damage is the symptom. The cause is almost always upstream of the pump.
This article separates the three mechanisms behind cavitation in centrifugal pump service, gives three measurements that identify which one is running, and works one NPSH calculation from start to finish. The last section is specific to moulded polypropylene impellers, because the repair decision there is different from the one for metal.
Key takeaways
- Cavitation starts where static pressure falls below vapour pressure — normally at the impeller eye, not at the suction flange.
- Three mechanisms produce it: vapour cavitation, gas entrainment and internal recirculation. Each leaves damage in a different place.
- A 25% NPSH margin protects against a 3% head drop, not against cavitation, which is already well developed at that point.
- At low flow the NPSH at which cavitation begins rises while NPSH3 stays flat, so a pump can cavitate on a margin that looks adequate.
- Three measurements identify the mechanism: the NPSH margin, head at shut-off against head at duty, and the damage pattern after a strip-down.
- Most fixes are on the system side and do not touch the pump.
- A moulded polypropylene impeller cannot be weld-repaired. It is replaced.
What Cavitation in a Centrifugal Pump Actually Is
Cavitation occurs when the fluid’s static pressure at a given flow rate falls below its vapour pressure at that temperature (DOE). Nothing has to be hot in absolute terms. It only has to be hot relative to the pressure at that point inside the machine.
The Pressure Drop Happens at the Impeller Eye
In a centrifugal pump the fluid accelerates into the impeller, and that acceleration is what drops the static pressure (DOE). The lowest pressure in the machine therefore sits at the eye, which is where the first bubbles appear. Alaska DEC states it directly: bubbles begin to form in the eye of the impeller.
This matters for diagnosis. A gauge on the suction flange reads upstream of the eye and always reads higher than the pressure where cavitation starts, so a suction gauge that looks healthy does not rule cavitation out.
Collapse, Not Boiling
As the fluid travels along the vane the local pressure recovers, the vapour condenses, and the bubble collapses. DOE describes the collapse as violent enough to throw tiny destructive jets of liquid onto the impeller surfaces next to it. That is the removal mechanism.
Boiling is heat-driven and happens throughout the liquid. Cavitation is pressure-driven, happens at one location, and reverses a moment later. That is why a pump can cavitate on cold water, and why raising the temperature is one of the fastest ways to start it.
Three Mechanisms, Not One
The usual habit in cavitation in centrifugal pump literature is to sort it into suction and discharge types. That split mixes three different physical events, and the mixing causes two of them to be misdiagnosed. Sorting by what produces the bubble is more useful, because each mechanism leaves damage in a different place and needs a different fix.
Vapour Cavitation: The Liquid Boils Where It Should Not
The liquid itself changes phase. Static pressure falls below vapour pressure at the eye, bubbles form, and they collapse as the pressure recovers along the vane. The trigger is usually upstream: a suction lift that is too high, a suction line that is long or narrow, an elbow at the inlet, a partly closed valve, or a liquid hotter than the system was designed for.
DOE adds a point that catches people out. Cavitation usually occurs at high flow rates, when a pump runs far to the right of its performance curve. Running a pump past its design flow raises the velocity into the eye, and velocity is what drops the pressure.
Gas Entrainment: Air That Was Never Vapour
Here the gas was already there. Air is drawn in at the suction — a vortex in the sump, a leaking flange, liquid falling into a tank from above — or dissolved gas comes out of solution as the pressure drops.
It needs its own heading because it breaks the usual rule of thumb. The TAMU work reports that the release and expansion of dissolved gases can cause performance breakdown at suction pressures well above the ones corresponding to the established liquid NPSHR. A pump can lose head at an NPSH the vendor’s curve calls fine, because the gas is not vapour and the curve was not drawn for it.
Internal Recirculation: Cavitation at Low Flow
The third mechanism has no vapour and no air. At low flow the flow pattern inside the impeller separates, and the resulting vortices produce the same kind of collapse and the same kind of damage. DOE defines recirculation as a condition that occurs during periods of low flow, usually below the minimum flow requirement of a pump, and states that it causes cavitation-like damage usually to the pressure side of an impeller vane.
The location is the giveaway. Vapour cavitation attacks the suction side, where the pressure is lowest. Recirculation attacks the pressure side, because the vortex sits at the trailing edge of the passage rather than at the eye.
Several published guides call a version of this “discharge cavitation” and attribute it to high discharge pressure. The recirculation is real; the causal claim is not. The flow pattern is set by how far the duty point has moved from the best efficiency point, not by absolute discharge pressure.
| Mechanism | Where the bubbles come from | Where the damage lands | Typical trigger |
|---|---|---|---|
| Vapour cavitation | The liquid flashes where static pressure drops below vapour pressure | Impeller eye and the suction side of the vane | Hot liquid, long or narrow suction line, high suction lift |
| Gas entrainment | Air or dissolved gas already in the liquid | Impeller eye, spread across several vanes | Vortex in the sump, leaking suction joint, falling liquid |
| Internal recirculation | Low flow lets the flow separate inside the passages | Pressure side of the vane, inlet or discharge edge | Pump running well below its minimum flow |
What the Damage Looks Like on the Impeller
Damage position is the most reliable evidence available after the fact, because each mechanism attacks a different surface. A centrifugal pump impeller that has cavitated usually carries a pattern specific enough to name the cause.
Where Each Mechanism Leaves Its Mark
Vapour cavitation pits the suction side of the vane and the shroud near the eye, where the pressure first drops. Gas entrainment spreads across several vanes at once, because the air arrives in the inlet flow rather than forming at one point. Internal recirculation concentrates on the pressure side and, in the discharge case, at the vane tips.
TAMU describes two patterns in detail: fine pitting on the blade suction side, and uniformly spread pitting with many nearly identical minicraters. The second points to a different mechanism again — pressure pulsation rather than steady low pressure — which is why a strip-down is worth more than a guess.
Symptoms You Can See Without Opening the Pump
DOE gives the audible signature: crackling and popping noises that often sound like marbles passing through the pump. Alaska DEC describes the same pinging or crackling noise. The pump cavitation symptoms most often reported — falling flow, a valve opened further than it used to be, vibration, a swinging suction gauge — tell you that something has changed, not which mechanism is responsible.
Three Measurements That Confirm It
Three measurements separate the mechanisms, and they follow the order a cavitation in centrifugal pump investigation normally takes: two need only a gauge and a flow reading, the third needs the pump open.
Measurement One: The NPSH Margin
Compute NPSHA at the actual duty flow and compare it with the NPSHR on the vendor’s curve at the same flow. DOE’s design convention is that NPSHA should be 25% higher than NPSHR for all expected flow rates. Below that margin, vapour cavitation is the leading explanation and the arithmetic in the next section applies.
TAMU adds a caution that changes how the number should be read. The 3% head drop that defines NPSHR is not the point at which cavitation begins. It is a point at which cavitation is already well developed, to the extent that it has produced a measurable drop in head, efficiency, capacity and power. A 25% margin over NPSHR therefore buys a margin against performance loss, not against cavitation damage — a distinction the design rule does not make.
Measurement Two: Head at Shut-Off Against Head at Duty
Close the discharge valve briefly and read the head, then open to duty and read it again. Compare both with the commissioning figures. A shut-off head that has fallen, or a duty head that falls month by month at the same flow, points to impeller wear rather than to a suction problem. Worn wear rings raise the NPSH required, which can start a cavitation problem on a duty that used to be fine.
Measurement Three: The Damage Pattern Itself
After a strip-down, the location of the pits decides the answer, using the map in the previous section. Mixed patterns mean more than one mechanism is running, which is common on a pump that has been throttled for a long time.
The NPSH Arithmetic, Worked Once
NPSH is the total head of the fluid at the centre line of the impeller above the fluid’s vapour pressure, and it measures the margin against vaporisation at the pump inlet (TAMU). In head terms, NPSHA is the atmospheric head at the liquid surface, plus or minus the static height of the liquid above the impeller, minus the friction loss in the suction line, minus the vapour pressure head.
Two of those terms belong to the installation rather than the pump. The static height is a drawing dimension. The friction loss is a pipe calculation that changes with flow. Neither appears on a pump curve, and those are the two people get wrong.
A Worked Example at 20 m³/h
Water at 40 °C, pump set 2.5 m above the liquid surface, suction line friction loss 1.2 m at 20 m³/h. Atmospheric head is 10.3 m. Water’s vapour pressure at 40 °C is 7.38 kPa from the NIST Antoine fit, which is 0.75 m of head.
NPSHA = 10.3 − 2.5 − 1.2 − 0.75 = 5.85 m.
The pump curve gives NPSHR 4.5 m at 20 m³/h. The 25% rule asks for at least 5.63 m. The installation passes by 0.22 m — about the height of a coffee cup, on a number assembled from a drawing, a pipe table and a temperature reading.
What the Margin Number Means
Now move the flow. NPSHR rises with flow, so at 24 m³/h the required value on the same curve is 5.6 m and the rule asks for 7.0 m, against an NPSHA that has also fallen because friction loss rises with the square of flow. The margin that passed at duty fails at 120% of duty. This is the mechanism behind DOE’s warning that the difference between NPSHA and NPSHR can become dangerously small when a pump runs far to the right of its design point.
Then move the flow the other way. TAMU’s data show NPSH3% holding almost flat from half flow to full flow, while the NPSH at which cavitation first appears rises steeply below the best cavitation point, and that rise is associated with the onset of suction recirculation. A pump at 60% of design flow can be cavitating on a margin that looks comfortable on the NPSH3 curve.
| Term | Symbol | Value | Unit |
|---|---|---|---|
| Atmospheric head at the liquid surface | Ha | 10.3 | m |
| Static suction head (2.5 m lift, taken negative) | Hz | −2.5 | m |
| Suction line friction loss | Hf | 1.2 | m |
| Vapour pressure head at 40 °C | Hvp | 0.75 | m |
| NPSHA = Ha + Hz − Hf − Hvp | 5.85 | m |
| Flow relative to duty | NPSHR | Margin the rule asks for | What actually happens |
|---|---|---|---|
| 60% of duty | Close to the rated value | 25% above that value | The inception NPSH rises before NPSH3 does, so recirculation damage can start first |
| 100% of duty | Rated value | 25% above NPSHR | Design point, and normally the only point checked |
| 120% of duty | Higher | 25% above the higher value | NPSHA has also fallen, because friction rises with the square of flow |
Fixing the System, Then the Pump
Almost every fix is on the system side, and the system side is cheaper than a new pump. The order is deliberate: exhaust the plumbing before changing hardware.
System-Side Fixes
Raise the liquid level, lower the pump, shorten and enlarge the suction line, remove the elbow at the inlet, and reduce the temperature if the process allows. Each changes one of the four terms. None changes the pump.
The suction line is where most of the loss usually sits. Sizing it for the same velocity as the discharge is common and wrong: suction velocity should be lower, and every fitting between the tank and the flange spends NPSH the pump cannot get back. On a polypropylene installation the same rule applies to the industrial water pump loop as to any other.
Pump-Side Fixes, and When the Answer Is a Different Pump
Speed and impeller diameter both change NPSHR. Reducing speed lowers the required NPSH. Trimming the impeller is the one that catches people out: DOE notes that in some pumps trimming increases the required NPSH, and instructs that the effect be evaluated from the manufacturer’s data over the full operating range rather than assumed. The polypropylene pump selection process treats the NPSH terms as system inputs for exactly this reason.
Change the pump when the arithmetic cannot be closed by plumbing: the source cannot be raised, the line cannot be enlarged, the temperature cannot be reduced, and the margin stays negative at the real duty flow. Running below minimum flow and accepting recirculation is not a workaround. It is a scheduled impeller replacement.
| Measure | What changes | Cost | When it fits |
|---|---|---|---|
| Raise the liquid level or lower the pump | Increases Hz | Civil work | The sump can be deepened or the pump relocated |
| Enlarge and shorten the suction line | Reduces Hf | Pipework | The line is long, narrow, or has an elbow at the inlet |
| Remove the elbow at the inlet | Reduces Hf and inlet swirl | Pipework | The straight inlet run is short |
| Cool the liquid | Reduces Hvp | Process change | The real temperature is above the design value |
| Reduce pump speed | Lowers NPSHR | Drive change | A variable speed drive or pulley change is available |
| Trim the impeller | Often raises NPSHR | Impeller work | The duty head is above what the system needs |
Why a Polypropylene Impeller Cannot Be Repaired
Everything above applies to any centrifugal pump. The repair decision is where cavitation in centrifugal pump service diverges for polypropylene, and the divergence is not in the polymer’s favour on the damage side. It is in the economics.
What the Collapse Does to a Moulded Vane
In a metal impeller the damage is pitting, a spongy surface, and in severe cases deep gouges where material has been torn away. A centrifugal pump impeller moulded in polypropylene sees the same energy act on a much softer surface, and the result is a honeycombed, chalky area with no structural value left.
The failure mode after that is mechanical. Material loss changes the balance of the rotor, and DOE describes the consequence for any impeller: enough loss makes the impeller unbalanced, creating alternating bearing loads that accelerate bearing wear. On a polymer impeller the loss happens faster, so the bearing symptom arrives sooner. The centrifugal pump parts article covers that chain in more detail.
Why Welding Does Not Apply
A metal impeller can sometimes be repaired by weld build-up and re-machining, or dressed back if the loss is shallow. Neither route exists for a moulded PP part. There is no weld repair that restores a moulded vane, and a patch would change the balance of the rotor rather than restore it. The part is replaced.
That is the practical difference, and it is worth stating without decoration. On a metal pump, cavitation is sometimes a repair bill. On a polypropylene pump, it is always a replacement.
The Honest Limit of PP in Cavitating Service
DOE’s guidance where cavitation is to some extent unavoidable is to specify high-tensile-strength materials for the impeller, because tougher materials withstand higher energy cavitation. Polypropylene is not a high-tensile-strength material. Any claim that a PP impeller is more cavitation-resistant than a metal one should be treated with suspicion, including when it comes from a supplier of PP pumps.
The correct answer for polypropylene is not a tougher material. It is not to cavitate: hold the NPSH margin above the design rule at every flow the pump will see, keep the temperature inside the rating, and keep the pump off the low-flow end where recirculation does the damage. Those are system decisions, and they are why this article spends most of its length on the suction side rather than on the impeller.
| Metal impeller | Moulded polypropylene impeller | |
|---|---|---|
| Damage form | Pitting, spongy surface, deep gouges in severe cases | Honeycombed and chalky, with no structural strength left |
| Can it be repaired? | Sometimes, by weld build-up and re-machining, or dressing back | No. No weld repair, and a patch disturbs rotor balance |
| Consequence | Repair or replace | Replace |
| Best defence | A higher-tensile alloy | Hold the NPSH margin, and keep the temperature down |
Where the duty is aggressive chemistry at moderate head and 60 °C or below, a solid PP centrifugal pump is a sound choice, provided the suction side is designed so the impeller never sees a cavitating condition in the first place.
Frequently Asked Questions
What does cavitation in a centrifugal pump sound like?
Like marbles or gravel passing through the pump, with a crackling or popping quality. DOE and Alaska DEC describe it the same way. Treat the noise as a late indicator: by the time it is audible, material has been coming off for a while.
Can a pump cavitate at low flow?
Yes, through internal recirculation rather than through low suction pressure. DOE defines recirculation as a flow condition below the pump’s minimum flow requirement that causes cavitation-like damage, usually to the pressure side of the vane. The NPSH3 curve does not show it, which is why low-flow cavitation is missed.
What is the difference between NPSHA and NPSHR?
NPSHA is available and belongs to the system; NPSHR is required and belongs to the pump. DOE states the split as NPSHA being a function of the system and the flow rate, and NPSHR a function of the pump and the flow rate. Plumbing changes one; only the pump or its speed changes the other.
Does a bigger pump reduce cavitation?
Usually the opposite. A larger pump throttled back to a small duty runs far from its best efficiency point, which raises the recirculation risk, and an oversized pump pushed to high flow runs where NPSHR is highest. Sizing to the duty is the safer route.
Can cavitation damage be repaired?
On a metal impeller, sometimes, by weld build-up and re-machining. On a moulded polypropylene impeller, no: there is no weld repair that restores a moulded vane, and a patch changes the rotor balance, so the impeller is replaced.
What NPSH margin should I specify?
DOE’s design convention is NPSHA at least 25% above NPSHR for all expected flow rates, not only the design point. Read that as a margin against a 3% head drop, and check it at the lowest and highest flows the pump will actually see.
Sources
- U.S. Department of Energy, Improving Pumping System Performance: A Sourcebook for Industry, 2nd ed., DOE/GO-102006-2078. Section 6, Centrifugal Pumps, and Appendix A, Glossary of Basic Pumping System Terms.
- Bradshaw, S., Liebner, T. and Cowan, D., Pump Cavitation — Various NPSHR Criteria, NPSHA Margins, and Impeller Life Expectancy. Turbomachinery Laboratory, Texas A&M University. oaktrust.library.tamu.edu
- Alaska Department of Environmental Conservation, Introduction to Pumping Systems, Chapter 6. dec.alaska.gov
- National Institute of Standards and Technology, NIST Chemistry WebBook, Water, Antoine Equation Parameters. webbook.nist.gov
