Cooling-water pump cavitation after a suction strainer is added
A 37 kW end-suction centrifugal pump, 2,950 rpm, rated 120 m³/h at 65 m head, lifts cooling water from an open concrete sump. In summer the water reaches 42 °C (vapour pressure 8.2 kPa abs, density 991 kg/m³); site atmospheric pressure is 101.3 kPa. The pump centreline sits 3.5 m above the lowest sump level. The suction line is 12 m of 150 mm ID pipe with two 90° elbows and a fully open gate valve (total K = 6.5, f = 0.018), plus a Y-strainer added last month whose clean pressure drop is rated 1.5 m at 120 m³/h. Published NPSHr is 4.2 m at rated flow. Since the strainer was fitted, operators report a rattling noise 'like gravel', discharge pressure swinging ±0.4 bar, and they have throttled the control valve to 95 m³/h to quieten it. The suction gauge reads −0.60 bar. The maintenance supervisor proposes trimming the impeller.
1. The noise could be cavitation, air entrainment, or suction recirculation. Use the evidence to discriminate, and state what inspection would confirm it.
Read the worked answer
Evidence for cavitation.
- Onset is tied to a change that increased suction loss (the strainer), not to a change in sump level or seal condition.
- The noise is described as gravel-like rattling, and it is accompanied by discharge pressure fluctuation of ±0.4 bar - vapour pockets collapsing and re-forming on the impeller.
- Throttling from 120 to 95 m³/h reduces the noise. NPSHr falls roughly with Q² while NPSHa rises as suction loss drops, so a flow-dependent symptom is the classic cavitation signature.
- The suction gauge is consistent with NPSHa ≈ 3.4 m, below NPSHr.
Why not the alternatives. Air entrainment from a vortex or a leaking flange would give erratic, surging gauge behaviour, air in a sight glass and noise that does not improve when the flow is reduced; it would also worsen at low sump level regardless of strainer condition. Suction recirculation shows up at low flow, not at 120 m³/h, and would not have started with the strainer.
Confirming inspection. Open the pump and examine the impeller suction side and blade leading edges: cavitation leaves a spongy, dark, pitted-eroded surface with material loss near the inlet, whereas air entrainment leaves no erosion pattern. Measure strainer differential pressure, check the sump for a vortex with the level at minimum, verify gauge calibration, and check for any suction-flange air leak by the bubble test. A vibration spectrum with broadband high-frequency energy also supports cavitation.
2. Recommend corrective actions, ranked. Evaluate the supervisor's proposal to trim the impeller.
Read the worked answer
Immediate (days). Clean or back-flush the strainer and fit a differential-pressure gauge with an alarm; restore the sump to its highest normal level and check for vortexing; keep the pump derated to about 95 m³/h with a written restriction. These are stop-gaps, not fixes - the clean-strainer case still gives only 4.64 m, so any dirt immediately re-creates the problem.
Permanent, in order of effectiveness.
- Remove the suction-side loss: replace the Y-strainer with a large low-loss basket strainer (Δp ≤ 0.3 m) or move the strainer to the discharge side where the pump can tolerate it.
- Upsize the suction line to 200 mm (v ≈ 1.06 m/s, hv ≈ 0.057 m): pipe plus fittings fall to roughly 0.7 m and NPSHa rises to about 5.9 m - a genuine margin, not a marginal one.
- If piping cannot change, lower the pump or raise the sump level by 1 m, and/or reduce summer water temperature.
- As a final option, fit a low-NPSH impeller or an inducer, and confirm NPSHr with the supplier's certified curve.
On trimming the impeller. Trimming does reduce NPSHr (roughly with Q², about 19 % for a 10 % diameter cut), but it also cuts head and flow, and the system needs 65 m at 120 m³/h. It treats the symptom, costs efficiency, and the cavitation returns the first time the strainer fouls. Reject it as the primary fix; only consider trimming as part of a deliberate re-rate of the duty point.
Verification. Recalculate NPSHa for worst-case conditions (maximum water temperature, minimum sump level, partially dirty strainer) and confirm a margin of at least 0.6 m above the certified NPSHr, then performance-test the pump at that condition.