Suction-Side Cavitation on a Cooling-Water Pump Asked to Deliver More Flow
A utility operates a cooling-water transfer pump drawing from an open, vented sump at essentially atmospheric pressure. At the original design flow of 120 m³/h the pump requires 35 m of head and has a published NPSH required of 4.5 m. The sump water surface sits 2.5 m below the pump centerline, and the suction line (strainer, two elbows, about 6 m of DN150 pipe) loses 1.8 m at 120 m³/h. Water is at 35 °C: density 994 kg/m³, vapor pressure 5.6 kPa. Barometric pressure is 101.3 kPa. A process change means the plant now wants 140 m³/h. At that flow the suction losses rise to 3.2 m and the pump's NPSH required rises to 5.8 m. Operators report rattling noise from the casing, vibration, and a flow indication that swings between 115 and 135 m³/h. A variable-speed drive is installed but currently runs at fixed speed, and the discharge valve is 40% throttled.
1. Calculate the available NPSH at 120 m³/h and at 140 m³/h, and state what the result means for the requested uprate.
Read the worked answer
NPSH available is the total absolute head at the pump suction flange above vapor pressure:
NPSHa = (p_atm − p_v)/(ρg) + z − h_f
Velocity head in the sump is negligible, so it is omitted.
Pressure term: (101.3 − 5.6) kPa = 95.7 kPa = 95,700 N/m² ρg = 994 × 9.81 = 9,751 N/m³ 95,700 / 9,751 = 9.81 m
At 120 m³/h: NPSHa = 9.81 − 2.5 − 1.8 = 5.51 m NPSHr = 4.5 m → margin = 1.01 m
At 140 m³/h: NPSHa = 9.81 − 2.5 − 3.2 = 4.11 m NPSHr = 5.8 m → margin = −1.69 m
The pump is already operating on a thin 1 m margin at design flow (many specifications require 1.0–1.5 m, so this is borderline), and at 140 m³/h it is more than 1.5 m short. The requested uprate cannot be met with the existing suction arrangement; the machine is cavitating.
2. Interpret the reported symptoms, evaluate the 40% throttled discharge valve as a remedy, and give the corrective options in order of practicality.
Read the worked answer
Symptoms: rattling from the casing, vibration, and an unstable flow indication that will not settle are the classic signature of cavitation and pump instability - vapor bubbles collapsing on the impeller, reducing head and disturbing the flow. The indication swings because the pump is operating on an unstable part of its curve as cavitation develops.
Throttling does not fix this. A discharge throttle raises system head and reduces flow, which slightly reduces NPSHr, but it addresses the symptom from the wrong side of the machine, wastes energy, and gives no reliable margin. Throttling cannot be counted on for a controlled uprate and should be opened while diagnosing.
Corrective options:
- Raise the sump level. Required submergence above pump centerline: z ≥ (NPSHr + 1.0 m margin) − 9.81 + 3.2 = 5.8 + 1.0 − 9.81 + 3.2 = +0.19 m. That means raising the water surface roughly 2.7 m above its current elevation - often the cheapest fix if the sump can be modified.
- Reduce suction losses: larger suction pipe, long-radius elbows, low-loss strainer, shorter run. Every 0.1 m of loss recovered is 0.1 m of NPSHa.
- Lower water temperature (reduces vapor pressure) if the process allows.
- Reduce speed/flow: NPSHr falls roughly with Q² under affinity laws, so a VFD can restore margin - but only if 140 m³/h is not truly required.
- Replace the pump with a low-NPSHr impeller or a different suction-specific design. Verify after any change with a suction pressure gauge and vibration measurement, not by ear alone.
3. The plant decides to run at 80 m³/h instead of 140 m³/h. Using a static head of 12 m and the given system data, compare the shaft power for throttling versus variable-speed operation and estimate the annual energy difference at 6,000 h/yr.
Read the worked answer
System friction at 120 m³/h = 35 − 12 = 23 m, so k = 23/120² = 0.00160 h²/m⁵. At 80 m³/h: H_sys = 12 + 0.00160 × 80² = 12 + 10.2 = 22.2 m. Hydraulic power = ρgQH = 9,751 × (80/3600) × 22.2 = 4.81 kW. At a wire-to-water efficiency of 0.72 for the VFD case: P_shaft ≈ 4.81/0.72 = 6.7 kW.
For throttling, the pump stays on its own curve. Take the curve as H = 42 − 0.000486 Q² (which passes through 35 m at 120 m³/h). At 80 m³/h the pump develops 42 − 3.1 = 38.9 m, and part-load efficiency drops to about 0.60: Hydraulic power = 9,751 × (80/3600) × 38.9 = 8.43 kW P_shaft ≈ 8.43/0.60 = 14.0 kW
Difference ≈ 14.0 − 6.7 = 7.3 kW, a saving of roughly 50%. Annual energy saved = 7.3 kW × 6,000 h = ~44,000 kWh/yr; at USD 0.10/kWh that is about USD 4,400/yr. Caveat: the affinity-law estimate assumes the VFD is programmed to follow the system curve and that the motor remains efficient at reduced speed. Also confirm that reduced flow does not violate the pump's minimum-flow limit, which is a separate reliability risk.