NPSH Shortfall on a Hot Water Transfer Pump
A centrifugal pump takes suction from an open tank of 80 C water (vapour pressure 47.4 kPa abs, density 971 kg/m3) whose surface is 3 ft below the pump centreline. At the design flow of 400 gpm the suction line friction loss is 4.5 ft and the curve shows NPSHR of 14 ft. The site is at sea level (101.3 kPa). Since commissioning the pump has been noisy - a rattling like gravel - with the discharge gauge needle flickering, flow about 8 percent below design, and no change in motor current. The suction line is 4 in with two elbows and a fully open gate valve; the discharge is throttled to 60 percent open. Maintenance has ordered a new impeller, believing the old one is worn. The pump was recently moved to a new skid and the suction piping reworked.
1. Calculate the available NPSH and decide whether the worn-impeller theory is supported.
Read the worked answer
NPSHA = Ha - Hv - Hs - Hf (all in feet of the pumped liquid)
- Ha = 101.3 kPa / (971 x 9.81) = 10.6 m = 34.9 ft (or 14.7 psi x 2.31 = 33.9 ft using standard head; the difference is density - use the tabulated value)
- Hv = 47.4 kPa / (971 x 9.81) = 4.98 m = 16.3 ft
- Hs = 3 ft static lift
- Hf = 4.5 ft suction friction at 400 gpm
NPSHA = 34.9 - 16.3 - 3.0 - 4.5 = 11.1 ft (about 10.1 ft on the 33.9 ft basis)
Against NPSHR of 14 ft, the pump is 3 to 4 ft short before any margin, and a sane design wants 2 ft of margin on top - so it needs roughly 6 ft more NPSHA.
The impeller theory is not supported. A worn impeller would show a steady loss of head and flow with smooth running, not gravel noise, gauge flutter, and unchanged current. The cause is hydraulic: the pump is cavitating at design flow, and a new impeller will cavitate exactly as the old one did. Note that 80 C water is the dominant term - 16.3 ft of the 34.9 ft available is spent just on vapour pressure.
2. Rank the practical remedies and quantify what each one buys.
Read the worked answer
Options, with the NPSHA gain each delivers:
- Raise the tank level / lower the pump - 1 ft of level = 1 ft of NPSHA. Need about 6 ft to reach 17 ft (14 + 3 margin). Usually the most reliable fix if the tank or the foundation can move.
- Cool the water - 60 C water has a vapour pressure of about 19.9 kPa, i.e. 2.06 m = 6.8 ft of head instead of 16.3 ft: gains about 9.5 ft. The biggest single win, but it is a process change, not a maintenance one.
- Enlarge the suction line - 4 in to 6 in at the same flow cuts velocity, and friction goes roughly as 1/d^5, so 4.5 ft becomes about 0.6 ft: gains about 3.9 ft. Not enough on its own, but it removes the cheap-to-fix portion and also kills the air-pocket risk.
- Pressurize the tank - 3 psi of nitrogen blanket = 6.9 ft of head.
- Reduce flow - lower flow drops both NPSHR and friction. A VFD is attractive because NPSHR falls roughly with speed squared; slowing to 90 percent speed cuts NPSHR disproportionately.
Recommendation: start with (1) plus (3) - relocation and suction re-sizing get to 17+ ft without touching the process. Verify by measuring actual suction pressure at the flange rather than trusting the calculation.
3. Distinguish cavitation from air ingress and internal recirculation using the evidence, and state what you check first.
Read the worked answer
Cavitation - calculated NPSHA below NPSHR (confirmed above); noise like gravel that worsens as flow increases; discharge gauge flutter; flow short of design; damage on the suction (concave) side of the vanes near the inlet, with a frosted or sponge-textured appearance.
Air ingress - the pump will not hold prime after shutdown; the noise may be irregular and change when the suction valve is throttled; look for a leaking suction flange gasket, valve packing, a high point or air pocket in the reworked suction line, or a vortex in the tank if submergence is low. This matters here: the suction piping was reworked when the pump moved, so a new high spot or a badly made flange is entirely plausible.
Internal recirculation - noise at low flow, with damage on the discharge side and the vane trailing edges rather than the inlet. With the discharge throttled to 60 percent open, low-flow recirculation is a real secondary risk and can be adding to the damage even after the NPSH problem is fixed.
Check first, cheapest to most expensive: tank level against the design value; the suction line for air pockets, gasket leaks and correct eccentric-reducer orientation (flat side up); then measure actual flow and suction pressure at the flange to verify the NPSH calculation; then inspect the impeller - the pitting pattern alone tells you which mechanism has been at work.