Feed Pump Cavitation and Poor Deaeration After a Load Expansion

The key point: in a deaerator the water surface is at saturation, so the pressure on the surface is exactly the vapor pressure of the water. Those two terms cancel, and NPSHa reduces to the static height of liquid above the pump suction.

Compare in psi and convert:

• Surface pressure: 5 psig = 19.7 psia
• Static: 8 ft of liquid at specific gravity 0.962 = 8 × 0.962 ÷ 2.31 = 3.33 psi
• Vapor pressure at 228°F = 19.7 psia
• NPSHa = 19.7 + 3.33 − 19.7 = 3.33 psi = 8.0 ft of liquid

Available = 8.0 ft, required = 8.0 ft. Zero margin at normal level - the pump is running at its NPSHr limit with no allowance for level variation, suction strainer fouling, or dissolved-gas release.

At the low-level alarm (2 ft lower, so 6 ft of static):

• Static = 6 × 0.962 ÷ 2.31 = 2.50 psi
• NPSHa = 19.7 + 2.50 − 19.7 = 2.50 psi = 6.0 ft, i.e. 25% below NPSHr.

That is precisely when the operators hear the gravelly roar. Any further rise in flow pushes the pump further into cavitation because NPSHr rises with flow while NPSHa falls with the increased suction-line losses.

Raising the deaerator to 15 psig does not help: the surface pressure rises to 29.7 psia, but the saturation temperature rises to about 250°F and the vapor pressure rises to 29.7 psia with it. The two terms still cancel, and NPSHa is still just the static height - now with hotter water that is marginally easier to flash and a specific gravity of about 0.94, which if anything reduces the head in feet. The deaerator pressure is not a lever on NPSH; the height is.

Deaeration is a two-part job: heat the water to saturation so that gas solubility falls to near zero, then give the released gases a path out (steam scrubbing plus a vent). Chemical scavengers are only the polish for the last few ppb - they cannot keep up with tens of ppb, and they add cost and solids.

With the spray valve wide open and the vent clear, the likely mechanism is a hydraulically overloaded deaerator: 30% more flow through a tray section sized for the original load. Symptoms fit: water backs up over the trays, the steam can no longer heat it to saturation and strip it, and the non-condensable load (air in with the make-up and condensate) exceeds what the fixed vent orifice can pass. The result is a deaerator that is running wet and cold at the top.

Verification, in order:

1. Calibrate the deaerator pressure gauge and the temperature element. If the water is 5°F below the saturation temperature for the measured pressure, it is not deaerating, period.
2. Check the vent - an "unobstructed" vent may be an undersized orifice. Feel the vent plume: a healthy deaerator vents a visible wisp continuously.
3. Survey air ingress on the condensate side: condensate pump seals, receiver vents, and any condensate header running under vacuum.
4. Inspect the tray stack and spray nozzles at the next outage for fouling, missing trays and plugged nozzles. A partially plugged nozzle pattern gives poor heating and high DO at any load.

Scavenger will pull the number down on paper for a few days and then the residual climbs again, while the real problem - a deaerator operating outside its hydraulic range - remains.

Ranked by effect per dollar:

1. Raise the deaerator (free to cheap, largest NPSH effect). Every foot of elevation above the pump adds a foot of NPSHa directly. Setting the deaerator on a 4 ft stand gets NPSHa to 12 ft at normal level and 10 ft at the alarm - comfortable against 8 ft NPSHr. If structural steel makes this impossible, go to item 2.
2. Operate the level control at a higher setpoint and shrink the alarm band (near zero cost). Whatever elevation you have, keep the deaerator at the top of its range. This buys margin today but does not create it.
3. Lower the suction losses (moderate cost). Increase suction pipe size one step, shorten the run, remove unnecessary elbows, and replace any restricted valve with a full-port gate or butterfly. A 2 psi reduction in suction loss is 4.6 ft of NPSHa at 0.96 SG - often larger than the elevation change.
4. Address the hydraulic overload (design fix). Re-rate the deaerator's tray section, enlarge the vent orifice for the new load, or split the make-up so the unit sees less cold water. This is the only fix that corrects the underlying deaeration shortfall as well as the pump.
5. Reduce flow per pump (operating fix). Run both pumps at part flow rather than one at its limit; NPSHr falls roughly with the square of flow.

First action: raise the level setpoint today and start the deaerator re-rate engineering. The level change is free and immediately removes the cavitation at normal operation; the re-rate is what makes the 7 ppb target achievable and keeps the pumps out of cavitation at the low-level alarm, which is where they currently destroy themselves.