Feedwater Pump Cavitation After Deaerator Pressure Increase
The deaerator storage tank water level is 4.5 m above the centreline of the 120 m³/h boiler feedwater pump. At design flow, suction-line friction loss is 1.2 m; the pump's NPSHr is 4.0 m. The deaerator has been raised from 0.20 bar(g) to 0.35 bar(g) to improve oxygen removal; water temperature is now about 108°C. Since the change, the pump has become noisy, flow has fallen from 120 to 95 m³/h, and discharge pressure fluctuates. The suction strainer is clean and the pump is not air-bound. The operator wants to raise deaerator pressure further to 0.50 bar(g) 'to keep the water from flashing' and to throttle the discharge to stabilize flow. You are asked to calculate the available NPSH and recommend a safe course of action.
1. Calculate the available NPSH at design flow and determine whether cavitation is expected. Explain why the deaerator pressure does not simply add to NPSHa.
Read the worked answer
Use the NPSH equation for a pump taking suction from a tank:
NPSHa = (p_surface - p_vapour)/(rho g) + z - h_f
where z is the static height of the liquid surface above the pump centreline and h_f is the suction-line friction loss. At the deaerator surface, the water is at saturation temperature for the deaerator pressure. Therefore p_surface equals p_vapour, and the pressure term cancels. The available NPSH is essentially the static head minus the suction friction:
NPSHa = 4.5 m - 1.2 m = 3.3 m
The pump requires 4.0 m NPSHr, so the margin is 3.3 - 4.0 = -0.7 m. Cavitation is expected at design flow. At 95 m³/h the friction loss is lower, perhaps 0.75 m, giving about 3.75 m NPSHa, which is still below 4.0 m. That is consistent with the noise and fluctuating discharge pressure.
The deaerator pressure does not improve NPSHa because raising the pressure also raises the water temperature and vapour pressure. At equilibrium, the surface pressure and vapour pressure rise together, so the net pressure head remains near zero. The static height is what provides the subcooling at the pump suction. If the water at the pump is hotter, it is closer to flashing, not further from it. A temporary increase in pressure before the water heats up can create apparent subcooling, but that is not a stable operating condition.
2. Evaluate the operator's two proposals: raising deaerator pressure to 0.50 bar(g) and throttling the discharge to stabilize flow.
Read the worked answer
Raising deaerator pressure to 0.50 bar(g) is the wrong direction. The deaerator will heat the water to a higher saturation temperature, and the vapour pressure at the pump suction will rise. Once equilibrium is reached, the pressure term still cancels, and the available NPSH remains approximately the static head minus friction. In addition, the higher temperature reduces the subcooling margin at the pump inlet and can make cavitation worse. It may improve oxygen removal, but that benefit is lost if the pump cavitates and the boiler loses feedwater. The correct approach is to lower the deaerator pressure if the original oxygen removal was acceptable, or to provide more subcooling or more suction head.
Throttling the discharge shifts the pump to a lower flow point. At lower flow, suction friction falls and NPSHr may decrease somewhat, so the noise may seem to improve. However, this masks the root cause and creates new problems: the pump may operate too far left on its curve, recirculation can overheat the casing, and the boiler may not receive enough feedwater during a demand increase. It also leaves the system vulnerable because the NPSH margin is still negative or barely positive. Throttling is not a safe permanent fix; it may be used only as a brief, monitored measure while the real correction is made.
3. Recommend two practical corrective measures and state the constraints you would place on temporary operation.
Read the worked answer
The first measure is to reduce suction-line friction. Replace the suction piping with a larger diameter, shorten the run, remove unnecessary elbows and fittings, and use full-port isolation valves. The existing 1.2 m loss is significant relative to a 3.3 m NPSHa. Reducing it to 0.5 m would raise NPSHa to about 4.0 m, which is only equal to NPSHr, so I would want a margin of at least 0.5 to 1.0 m above NPSHr. The suction strainer should be checked and kept clean, but it must not be removed as a substitute for proper piping.
The second measure is to increase the static head or provide a low-NPSH booster pump. Raising the deaerator storage tank or lowering the feedwater pump gains NPSHa directly, but structural changes are costly. A booster pump with a low NPSHr installed ahead of the main pump is often more practical; it raises the suction pressure at the main pump while the booster itself is selected for the available NPSHa. Another option is to subcool the feedwater slightly with a small heat exchanger or a bypass arrangement, which lowers the vapour pressure at the pump suction, but this adds complexity and must not compromise deaeration.
For temporary operation, I would lower the deaerator pressure back to 0.20 bar(g) if oxygen removal remains acceptable, reduce feedwater flow to a point where NPSHa exceeds NPSHr by at least 0.5 m, monitor pump noise, vibration, and discharge pressure, and keep the boiler load stable. I would not accept throttling as a fix. If the pump continues to cavitate, I would shut down the boiler on a controlled basis rather than risk losing feedwater and damaging the pump or boiler.