Troubleshooting Pump Cavitation With Adequate NPSH

A pump can cavitate with ample inlet NPSH because NPSH addresses pressure at the impeller inlet, while vapor bubbles can also form through internal recirculation at low flow. Here, the pump runs at half its design flow, the stated NPSHr is 4–5 m, and the calculated NPSHa is about 11.8–12 m before the remaining local corrections. That combination makes low-flow recirculation the leading diagnostic path, while suction loss, air ingestion, speed mismatch, and curve selection still require checks.

Symptom Pattern and Initial Diagnosis

The reported operating pattern combines several useful observations:

• Water temperature is 45 °C.
• The suction pressure tap is after the strainer and before the pump.
• The suction gauge reads 0.2 bar gauge.
• The pump operates at half its design flow for process reasons.
• The stated NPSHr at that operating point is 4–5 m.
• The pump sounds as though rocks are passing through it and has substantial vibration.
• The discharge pressure fluctuates between 28 and 30 bar but is not strongly unstable.
• The impeller has pitting on both sides of the vanes, with worse damage on the pressure side.
• Rotation is correct, the suction valve is fully open, and the straight suction run exceeds ten pipe diameters.

The term here means vapor cavitation only when local static pressure falls below the liquid vapor pressure and the resulting bubbles collapse in a higher-pressure region. Noise, vibration, and pitting fit that mechanism, but they do not establish where the low-pressure region occurs. With a large nominal NPSH margin and operation far below design flow, investigate the internal velocity field before blaming bulk suction pressure.

Suction Pressure and NPSHa Interpretation

NPSHa must use absolute pressure. A gauge-pressure reading is relative to local atmospheric pressure, so a reading of 0.2 barg is not an absolute pressure of 0.2 bar. Using the stated atmospheric head of 10.2 m and the stated conversion of 0.2 bar to approximately 2 m of water, the pressure head at the tap is about 12.2 m absolute.

At a pressure tap close to the pump inlet, the applicable form is:

NPSHa = absolute pressure head at tap + velocity head − vapor-pressure head − losses from tap to impeller inlet

Using the supplied vapor-pressure estimate of 0.18 m gives approximately:

NPSHa ≈ 10.2 m + 2.0 m − 0.18 m + velocity head − local downstream losses

This is about 12.0 m before the velocity and short downstream-loss corrections. Do not add the tank static head again when using the local suction reading; that pressure measurement already reflects upstream static head and losses through the tank connection, piping, and strainer.

The alternative system calculation is:

NPSHa = 10.2 m atmospheric head + 10 m static head − 8.2 m tank vacuum head − 0.18 m vapor-pressure head − suction head loss

NPSHa = 11.82 m − suction head loss

The two methods must reconcile when they use the same datum, velocity terms, and loss boundaries. A disagreement points to an incorrect gauge reference, wrong elevation basis, blocked pressure connection, inaccurate instrument, or double-counted static pressure. With NPSHr of 4–5 m, the nominal margin is roughly 7–8 m before applicable corrections. NPSHr is a pump-test performance criterion, not a promise that no bubbles exist anywhere inside the casing.

Low-Flow Internal Recirculation Mechanism

A centrifugal pump operating well below its intended flow range develops unfavorable incidence at the impeller eye and discharge. Flow no longer enters each vane passage at the intended angle. Part of the liquid can reverse direction, form rotating vortices, and circulate through impeller clearances or between vane passages.

Those local velocities can reduce static pressure below vapor pressure even while the average pressure at the suction gauge remains high. Bubble collapse then occurs as the recirculating liquid moves into a higher-pressure part of the impeller. Bulk NPSHa therefore may remain comfortably above published NPSHr while internal recirculation produces noise, broadband vibration, and material loss.

Operation at half design flow is the strongest discriminator in this case. A fully open suction valve does not correct low flow because the system resistance, downstream controls, pump speed, and impeller geometry determine the operating point. The 28–30 bar discharge-pressure variation is only a two-bar range and does not exclude cavitation; recirculation can remain damaging without dramatic gauge instability.

Pitting on both vane sides, with greater damage on the pressure side, supports an internal hydraulic disturbance but is not a standalone cavitation classification. Record the axial, radial, inlet-to-outlet, and pressure-side-to-suction-side location of the damage. Compare that map with the manufacturer’s recirculation and allowable-operating-range guidance for the actual impeller.

Symptoms-versus-Causes Decision Table

Diagnostic and Corrective Procedure

1. Validate the pressure basis. Confirm that 0.2 bar is gauge pressure, record local atmospheric pressure or absolute suction pressure, and verify the pressure tap is clear. Use one NPSHa method at a time and retain velocity head and loss boundaries consistently.
2. Establish the operating point. Record actual flow, suction pressure, discharge pressure, water temperature, speed, and installed impeller identification. Calculate developed head from suction and discharge conditions instead of treating the 28–30 bar discharge reading alone as pump head.
3. Select the correct curve. Plot the measured flow and head on the manufacturer’s curve for the installed speed and impeller. Read NPSHr at that actual point and locate the permitted operating range and minimum continuous-flow requirement.
4. Check the low-flow condition. The measured flow is half design flow. Under an approved test condition, raise flow toward the manufacturer’s acceptable range while trending noise, vibration, suction pressure, discharge pressure, and motor current. A rapid reduction in noise and vibration identifies recirculation more strongly than a static NPSH calculation does.
5. Measure suction restrictions. Take pressures on both sides of the strainer at the same flow and calculate its differential head. The downstream gauge already includes upstream strainer loss, but a rising differential identifies fouling and explains loss of margin.
6. Check air-entry paths. Inspect flange joints, valve stems, instrument connections, vessel vortices, and suction-pipe high points. Vary suction-vessel level within the approved operating range and observe whether noise or vibration changes.
7. Verify rotational speed and pump configuration. Rotation is already reported correct. Confirm actual speed against the curve and nameplate data. A pump intended for 1800 rpm but operated at 3000 rpm requires direct manufacturer review; also check whether the design is offered for 3600 rpm before accepting the impeller geometry at that speed.
8. Review impeller inlet velocity. Obtain the impeller-eye diameter and allowable tip-speed guidance from the manufacturer. If only a wear-ring diameter is available, treat any geometric estimate as a screening calculation, not an acceptance criterion. The cited 85 ft/s value is a vertical-pump rule of thumb and cannot replace product-specific limits.
9. Correct the operating arrangement. Move the duty point into the manufacturer’s allowable range through an approved process-flow change, pump selection change, speed change, impeller change, or recirculation path. Do not improvise a minimum-flow value from design flow alone.

Numbered Verification Checks

1. Check 1: NPSHa reconciliation. Expect the local-pressure method and tank-to-pump method to agree after applying the same datum, velocity terms, and pipe losses. The preliminary result is approximately 11.8–12 m, compared with the stated 4–5 m NPSHr.
2. Check 2: Curve agreement. Expect measured flow and developed head to land on the curve for the installed speed and impeller within normal measurement uncertainty. A different curve invalidates the NPSHr comparison.
3. Check 3: Flow-response test. Expect cavitation-like noise and broadband vibration to decrease when flow moves out of the low-flow recirculation region. Little or no change directs attention back to air ingestion, suction restriction, speed, or mechanical damage.
4. Check 4: Strainer condition. Expect a stable, acceptably low differential pressure at the tested flow. A differential that increases with runtime points to accumulating restriction.
5. Check 5: Pressure and current stability. Expect repeatable suction pressure, discharge pressure, flow, and motor current after the operating-point correction. Erratically low current combined with erratic flow or head indicates intermittent loss of hydraulic loading.
6. Check 6: Damage arrest. Establish a documented impeller condition after repair or replacement and inspect it after a defined plant maintenance interval. Expect no new pitting in the previously attacked regions.

Recurring Diagnostic Pitfalls

The first wrong practice is comparing a gauge-pressure value directly with vapor pressure or NPSHr. Convert gauge pressure to absolute pressure before calculating NPSHa.

The second is subtracting all suction-pipe loss from a pressure reading taken after that piping. Upstream losses are already represented in the local reading. Subtract only losses between the pressure tap and the NPSH reference location, with the correct velocity and elevation terms.

The third is treating published NPSHr as a cavitation-free boundary. It identifies a tested performance condition for a particular pump configuration and operating point. Internal recirculation, localized inlet vortices, or gas ingestion can still produce damaging bubbles above that value.

The fourth is accepting “half design flow” without checking the permitted operating range. Design flow, best-efficiency flow, and minimum continuous flow are different terms. Read each value from the curve or operating-limit documentation for the installed pump.

The fifth is diagnosing from sound alone. A rock-like sound and high vibration justify an immediate hydraulic investigation, but looseness, rubbing, bearing defects, solids, and gas entrainment can overlap. Use the controlled flow test, pressure trends, vibration spectrum, and damage map together.

Frequently Asked Questions