Raw Water Pump Cavitating: Fix Suction or Fit a VFD?
A 45 kW, 1450 rpm centrifugal pump at a raw-water pumping station delivers 180 m³/h against a total head of 45 m, of which about 25 m is static lift to the delivery reservoir and 20 m is friction and fitting loss at rated flow. The pump's NPSHr at this duty is 6.5 m. In summer the sump water level is 3.5 m below the pump centreline; the 200 mm suction line loses 1.2 m at rated flow, the water is at 30 °C (vapour pressure 4.24 kPa abs, density 996 kg/m³) and the site atmospheric pressure is 95 kPa abs. Operators report a rattling noise from the casing, falling discharge pressure and rising vibration since the sump level dropped. The station runs 18 hours a day at a flat energy charge of ₹6.20/kWh. The utility proposes to trim flow to 140 m³/h by throttling the discharge valve; a vendor instead offers a VFD for ₹4.2 lakh plus ₹0.4 lakh installation.
1. Calculate NPSHa and decide whether the pump is cavitating. What physical change would give at least 1 m of NPSH margin?
Read the worked answer
NPSHa = (p_atm − p_v)/(ρg) + z − h_f = (95,000 − 4,240)/(996 × 9.81) + (−3.5) − 1.2 = 90,760/9,771 − 4.7 = 9.29 − 4.70 = 4.6 m.
Against NPSHr = 6.5 m this is a shortfall of about 1.9 m, so the pump is cavitating. That matches the symptoms - casing rattle, falling discharge head and vibration are classic cavitation, not bearing or alignment faults. Note that the margin worsens further if the water gets warmer or if the suction strainer chokes with silt, which is likely as the sump level falls.
Options ranked by effect per rupee
- Lower the pump (or raise the sump level). Dropping the pump 2 m, or restoring sump level by desilting/raising the weir, adds 2 m of NPSHa → 6.6 m, giving a 0.1 m margin; combine with option 2 to reach a safe 0.5–1.0 m margin.
- Enlarge the suction line to 250 mm. Velocity falls from 1.59 m/s to 1.02 m/s, and friction from 1.2 m to roughly 0.5 m, recovering about 0.7 m.
- Clean/relocate the strainer - recovers unrecorded losses and is nearly free; do this first on site.
- Reduce flow - lowers both h_f and NPSHr, but only as a stop-gap.
A 0.5 m NPSH margin is the practical minimum; I would not run with 4.6 m against a 6.5 m requirement even briefly.
2. Compare throttling to 140 m³/h with the VFD option in shaft kW, using the static head correctly.
Read the worked answer
Throttling. The pump moves up its own curve; head rises to roughly 55 m while efficiency falls to about 0.65, flow is 0.0389 m³/s. P = ρgQH/η = 996 × 9.81 × 0.0389 × 55/0.65 ≈ 32.2 kW, against 30.5 kW at the original duty (180 m³/h, 45 m, η ≈ 0.72). Throttling increases energy use and saves nothing.
Naive affinity-law answer (wrong here). Speed ratio 140/180 = 0.778 would give head 45 × 0.778² = 27.2 m and about 14.8 kW - tempting, but it ignores the 25 m of static lift.
Corrected VFD answer. The system curve is H = 25 + 20(Q/180)². With a pump curve of the form H = 55 − 3.09×10⁻⁴Q², solving at Q = 140 gives speed n ≈ 0.886 (about 1285 rpm) and head H = 25 + 20(140/180)² = 37.1 m. With efficiency about 0.70: P = 996 × 9.81 × 0.0389 × 37.1/0.70 ≈ 20.1 kW, plus about 3% VFD loss ≈ 20.7 kW.
Comparison. Throttled 32.2 kW; VFD 20.7 kW; saving ≈ 11.5 kW. The static head has already cost about 3.5 kW of the saving the affinity laws appear to promise, and that is the figure to carry into any payback calculation.
3. Compute the annual saving and payback and give your recommendation, listing the caveats you would record.
Read the worked answer
Energy and money. 11.5 kW × 18 h/day × 365 days = 75,555 kWh/year; at ₹6.20/kWh this is about ₹4.68 lakh/year. Capex is ₹4.2 + 0.4 = ₹4.6 lakh, so the simple payback is roughly 12 months - attractive, and it is the VFD that is attractive, not throttling.
Recommendation. Approve the VFD for flow control, but fix the suction first. The VFD partly masks cavitation (NPSHr falls roughly with n², to about 5.1 m at 0.886 speed, still below the 4.6 m available), so it is not a cure; run the station at full speed until the suction work is done.
Caveats to record
- The saving is sensitive to static head. If the delivery main's static component rises (higher reservoir level, longer route), the pump can only trim speed slightly and the saving falls towards zero.
- If 140 m³/h is a permanent duty, compare a trimmed impeller - cheaper than a VFD and slightly more efficient - and reserve the VFD for genuinely variable demand.
- Verify savings with the VFD's own kWh meter over a month rather than trusting the calculation.
- Check minimum speed and minimum flow for the pump, motor insulation and bearing currents on the VFD, harmonic injection into the station's supply, and ventilation of the VFD panel.
- Set the operating band so the pump never runs far left of its curve, where recirculation destroys the impeller.