Cavitation Margin at a Water Treatment Works
A 132 kW centrifugal pump at a water treatment works lifts clarified water at 62 °C from an open sump to a deaerator. The pump centreline is 3.6 m below the minimum sump level. Suction piping is 6 m of DN250 steel with two long-radius bends and a wide-open gate valve; at the design duty of 210 m³/h the suction loss is 1.9 m and the manufacturer's NPSHr is 8.9 m. Site atmospheric pressure is 100.2 kPa, water density at 62 °C is 982 kg/m³, and vapour pressure is 21.5 kPa. Since a recent demand increase pushed the pump to 105% of design flow, operators report crackling noise and a 3.1 kHz vibration peak that disappears below 85% flow. The suction pressure gauge shows a 12 kPa fluctuation at the same frequency as the noise. The project manager proposes a restriction orifice in the discharge to protect the pump and asks you to sign off continued operation at 105% until the next shutdown in seven weeks.
1. Calculate NPSHa at design flow and at 105% flow. Assume suction loss scales with Q² and NPSHr rises to about 9.4 m at 105%. What margin exists, and what does that imply for the PM's request?
Read the worked answer
Use NPSHa = (p_atm − p_v)/(ρg) + z − h_f.
The pressure term is 100.2 − 21.5 = 78.7 kPa, or 78,700 Pa. Divide by ρg = 982 × 9.81 = 9,634 Pa/m: 78,700 / 9,634 = 8.17 m.
At design flow: NPSHa = 8.17 + 3.6 − 1.9 = 9.87 m. Against NPSHr 8.9 m, the margin is 0.97 m. That is just below the common 1.0 m guideline and leaves almost nothing for strainer fouling, level dips, or future flow increases.
At 105% flow, suction loss is 1.9 × 1.05² = 2.09 m. NPSHa = 8.17 + 3.6 − 2.09 = 9.68 m. If NPSHr rises to 9.4 m, the margin is only 0.28 m. Even if NPSHr stayed at 8.9 m, the margin would be 0.78 m.
The calculation does not support an unrestricted seven-week sign-off. The pump is already close to its NPSH limit at design flow, and the higher duty consumes what little margin remains. A sign-off would need an operating limit, monitoring, and a credible plan to restore margin, not just a discharge orifice.
2. How much weight should you place on the 3.1 kHz vibration peak and the 12 kPa suction-pressure fluctuation? What additional evidence would you require before calling this cavitation?
Read the worked answer
The 3.1 kHz peak and the suction-pressure fluctuation are consistent with cavitation, but they are not conclusive. Cavitation normally produces broadband high-frequency noise and pressure pulsation, and it can excite a structural resonance that appears as a discrete peak. A vane-pass frequency, recirculation, air entrainment, or a loose component can also produce a narrow peak.
Useful checks:
- Does the frequency track pump speed or flow, or is it fixed at 3.1 kHz?
- Does the noise disappear abruptly below 85% flow, or fade gradually?
- Is there a sudden drop in discharge pressure or flow at the same condition?
- What is the strainer differential pressure, sump level, and vortex pattern?
- Does a fast-logged suction pressure show sharp, high-frequency spikes rather than a slow fluctuation?
- Does impeller inspection show pitting or erosion?
I would treat the symptoms as a prima facie indication of inadequate NPSH margin, not as a confirmed diagnosis. The calculation already shows a thin margin at 105% flow. The inspection and pressure log should confirm the mechanism before spending capital on a fix.
3. Evaluate the proposed discharge restriction orifice and set out your response to the request to sign off seven weeks at 105% flow.
Read the worked answer
A discharge restriction orifice would reduce flow, but it wastes energy and treats the symptom rather than the suction limit. It may also shift the pump to a different part of its curve and increase internal recirculation if the flow is pushed too low. It is acceptable only as a temporary, controlled throttle with a clear operating limit.
Better options, in order:
- Raise the sump level if the operating envelope allows it; every 0.5 m adds 0.5 m NPSHa.
- Clean or replace the suction strainer and verify its pressure drop.
- Replace the wide-open gate valve with a low-loss isolation valve if the valve is a significant loss.
- Reduce flow to 95% or less using a VFD or controlled throttling, and confirm NPSHr at that flow with the vendor.
- If margin is still inadequate, modify the suction line or install a low-NPSH impeller/inducer.
I would not sign off seven weeks at 105% flow on the current evidence. I would issue a written operating limit, require daily vibration and suction-pressure checks, confirm spare seals and bearings, and define a shutdown trigger. If commercial pressure requires continued operation, I would require a formal risk assessment, a shorter review interval, and a documented acceptance by the accountable manager.