Cooling-Water Pump Cavitation After a Debottleneck Request
A cooling-water pump takes suction from an open concrete sump whose normal water level is 2.5 m below the pump centreline. Duty is 240 m³/h at 45 m head; the pump curve shows NPSHr of 5.0 m at duty and 6.2 m at 280 m³/h. The DN200 suction line has a calculated loss of 1.8 m at 240 m³/h. Site conditions: water at 35 °C (vapour pressure 5.6 kPa, density 994 kg/m³), barometric pressure 101.3 kPa. Production wants to debottleneck by running the pump at 280 m³/h, and during peak draw-off the sump level can fall a further 0.8 m below normal. Operators already report a rattling, gravelly noise from the pump and periodic discharge-pressure fluctuation, worst in the afternoon. A spare pump of identical model is in the store. You must decide what can be run this week and what permanent modification to fund, using the numbers rather than the spare pump as the answer.
1. Calculate the NPSH available at 240 m³/h and at the proposed 280 m³/h with the low sump level, and state the operating decision for this week.
Read the worked answer
NPSHa = (p_atm − p_v)/(ρg) − static suction lift − suction-line loss.
Pressure head available: (101.3 − 5.6) × 1000 / (994 × 9.81) = 95 700 / 9751 = 9.81 m.
At 240 m³/h, normal level: 9.81 − 2.5 − 1.8 = 5.51 m against NPSHr 5.0 m. Margin 0.5 m - thin, not comfortable, and consistent with the noise reported.
At 280 m³/h with level down 0.8 m: losses scale with flow squared, 1.8 × (280/240)² = 1.8 × 1.36 = 2.45 m; static lift becomes 3.3 m.
NPSHa = 9.81 − 3.3 − 2.45 = 4.06 m against NPSHr 6.2 m. A deficit of over 2 m.
Decision this week: the pump runs at no more than 240 m³/h, and only with the sump level held at or above normal by level control - not by operator attention. The 280 m³/h debottleneck is refused until the suction system is modified. Do not fit the identical spare as a fix: it will cavitate in exactly the same way, and installing it only removes the evidence.
2. Quantify the permanent options and recommend the one you would fund, showing the arithmetic.
Read the worked answer
Target: NPSHa ≥ NPSHr at 280 m³/h (6.2 m) plus a working margin of about 0.8 m, i.e. ≥ 7.0 m. The deficit to close at 280 m³/h with low level is 7.0 − 4.06 ≈ 2.94 m.
- Raise the sump operating level by 3.0 m (weir/level control): NPSHa = 9.81 − 0.3 − 2.45 = 7.06 m. Works, but needs civil work and may not be possible within the pit.
- Upsize suction to DN250: loss falls roughly as (200/250)^4.8 ≈ 0.29 of 2.45 m ≈ 0.55 m (recalculate properly for the final layout): NPSHa = 9.81 − 3.3 − 0.55 = 5.96 m - still short of 7.0 m on its own.
- Lower the pump centreline by 1.5 m: NPSHa = 9.81 − 1.8 − 2.45 = 5.56 m - better but not sufficient alone.
- Combined: lower pump 1.5 m plus DN250 suction = 9.81 − 1.8 − 0.55 = 7.46 m. Comfortable.
Recommendation: fund the combined suction modification (DN250 line with a short, rising-to-falling, air-tight route plus pump lowering), and specify a vortex-breaker and a minimum-submergence check at the new low level. A suction strainer, if fitted, must be sized on clean pressure drop and monitored - a fouling strainer silently removes the margin you just bought.
3. Explain the afternoon pattern of noise and pressure fluctuation, and say what condition assessment you would require on the running pump before the modification.
Read the worked answer
The afternoon worsening is a vapour-pressure effect. Water temperature rises through the day; at 40 °C the vapour pressure is about 7.4 kPa rather than 5.6 kPa:
- Available pressure head = (101.3 − 7.4) × 1000 / (991 × 9.81) ≈ 9.65 m - about 0.16 m less than at 35 °C
- NPSHa at 240 m³/h then falls to about 5.35 m, right at the NPSHr of 5.0 m
Combine that with the 0.8 m level swing when draw-off peaks and the margin disappears entirely. The rattling/gravelly noise with discharge-pressure fluctuation is the classic signature of cavitation: vapour bubbles collapsing on the impeller inlet, not a bearing fault (bearings give tonal, speed-related noise, not broadband crackle). At low sump level, air-entraining vortices add the same symptoms without true cavitation, which is why submergence and vortexing must be checked.
Before modifying anything I would require: impeller inlet and shroud inspection for erosion pitting at the next stop, check of wear-ring clearances, seal condition, a vibration spectrum (broadband high-frequency energy plus 1× and vane-pass components), suction pressure logged with flow and temperature, and a check that the suction pipe has no high point or air pocket. If the impeller is already eroded, the pump will not recover its curve after the suction fix, so the modification and the impeller overhaul should be planned together.