Cooling-Water Pump Cavitation After 18 Months
A cooling-water pump draws 120 m³/h from an open sump through 8 m of DN150 (150 mm ID) suction pipe with friction factor 0.02 and fittings equivalent to K = 6.5 velocity heads. The pump centreline sits 0.8 m above the sump water level. Water is at 30 °C (ρ = 996 kg/m³, vapour pressure 4.25 kPa); site barometric pressure is 101.3 kPa. A clean basket strainer adds 0.5 m of head loss. The pump curve gives NPSHr = 4.5 m at 120 m³/h, rising with the square of flow. After 18 months in service the operators report a rattling noise at the pump, a fluctuating discharge pressure and slight vibration, always worst in the afternoon when demand is highest.
1. Compute the NPSH available at the design flow with a clean strainer and state whether the installation has an adequate margin.
Read the worked answer
Q = 120/3600 = 0.03333 m³/s; A = π(0.150)²/4 = 0.01767 m²; v = 1.886 m/s; v²/2g = 0.181 m
Friction: hf = f(L/D)(v²/2g) = 0.02 × (8/0.15) × 0.181 = 0.193 m Fittings: 6.5 × 0.181 = 1.179 m Strainer (clean): 0.500 m Total suction loss = 1.87 m
Pressure head available: (p_b − p_v)/(ρg) = (101.3 − 4.25) × 10³/(996 × 9.81) = 9.93 m Static: −0.80 m (pump centreline above sump level)
NPSHA = 9.93 − 0.80 − 1.87 = 7.26 m Margin over NPSHr = 7.26 − 4.50 = 2.76 m (about 60% of NPSHr, well above the usual 1.0–1.5 m or 10% rule).
So the as-designed installation is sound. The problem is therefore not the layout - it is something that changed in service, which is the useful conclusion to carry into troubleshooting.
2. A differential pressure gauge across the strainer now reads 35 kPa. Test the cavitation hypothesis quantitatively, including the afternoon peak.
Read the worked answer
Strainer loss now: 35 000/(996 × 9.81) = 3.58 m, i.e. 3.08 m more than assumed.
At 120 m³/h: NPSHA = 9.93 − 0.80 − 1.87 − 3.08 = 4.18 m < NPSHr 4.5 m → cavitation.
At the afternoon peak, take flow 15% above design (v² rises by 1.15² = 1.32): Suction piping loss = 1.37 × 1.32 = 1.81 m; strainer = 3.58 × 1.32 = 4.73 m NPSHA = 9.93 − 0.80 − 1.81 − 4.73 = 2.59 m NPSHr = 4.50 × 1.32 = 5.95 m Margin = −3.36 m
The hypothesis fits every symptom: rattling (bubble collapse), fluctuating discharge pressure, vibration, and a peak-demand bias because NPSHr climbs faster with flow than NPSHA can follow. Diagnose a fouled suction strainer compounded by a high-flow operating point, not a worn impeller or a bearing fault.
3. Recommend remedial actions, ranked, and give the NPSHA that results from the most durable one.
Read the worked answer
Clean or replace the strainer element and fit a dP transmitter with an alarm at 20 kPa. This restores the 7.26 m margin at essentially zero capital cost and prevents recurrence - the real failure here is the absence of condition monitoring, not the strainer.
Raise the sump level or lower the pump. Every metre of static head is a metre of NPSHA; often the cheapest permanent fix if civil work is easy.
Increase the suction line to DN200. At DN200, A = 0.03142 m², v = 1.06 m/s, v²/2g = 0.0574 m; friction plus fittings = (0.02 × 8/0.20 + 6.5) × 0.0574 = 0.42 m. With the strainer still fouled at 3.58 m: NPSHA = 9.93 − 0.80 − 0.42 − 3.58 = 5.13 m, which survives fouling and peak flow better than anything else on the list.
Cap maximum flow (trim impeller or throttle to a set maximum). Since NPSHr rises as Q² and NPSHA falls as Q², limiting flow protects the margin - but do not open the discharge valve further hoping to 'flush' the problem; that moves the duty point the wrong way.
Sequence: clean now to restore production, then instrument the strainer and schedule the suction-line upgrade at the next outage.