Cavitating Charge Pump on a Knockout Drum
A centrifugal pump takes light hydrocarbon (SG 0.62) from a knockout drum at 41 psia and delivers it to a fractionator at 95 °F. The liquid vapor pressure is 38 psia at that temperature. The drum's normal liquid level is 6 ft above the pump centerline, and the suction line loses 5 ft of head at the rated flow of 400 gpm. The pump's NPSHr is 12 ft at 400 gpm. For the last month the pump has been noisy, the discharge pressure gauge fluctuates, and the flow meter reads erratically whenever operators open the discharge valve to raise rate. The pump has been rebuilt twice; the impeller shows pitting on the suction side of the vanes. Instruments have been recalibrated and are accurate.
1. Calculate the available NPSH at rated flow and state the margin.
Read the worked answer
NPSHa = (P_vessel − P_vapor) × 2.31 / SG + static head − suction friction
Pressure term: (41 − 38) × 2.31 / 0.62 = 3 × 3.726 = 11.2 ft
Static: +6.0 ft
Friction at 400 gpm: −5.0 ft
NPSHa = 11.2 + 6.0 − 5.0 = 12.2 ft
NPSHr = 12 ft → margin = 0.2 ft
A margin of about 3 ft is the usual minimum for a hydrocarbon near its bubble point. This pump has essentially none at rated flow, so it is at the edge of cavitation at design conditions and will cavitate at any higher rate or any loss of drum level. The repeated rebuilds and the suction-side vane pitting are classic cavitation-erosion signatures, not a mechanical defect in the pump.
2. Estimate what happens when operators open the discharge valve and flow rises 20%.
Read the worked answer
Both sides of the margin move the wrong way.
NPSHa falls. Friction scales roughly as Q^1.8:
h_f = 5 × (1.2)^1.8 = 5 × 1.39 = 6.9 ft
NPSHa = 11.2 + 6.0 − 6.9 = 10.3 ft
The static and pressure terms are fixed; only friction grew.
NPSHr rises. Near runout the required NPSH climbs roughly as Q² and the curve steepens:
NPSHr ≈ 12 × 1.2² = 17.3 ft (call it 15–17 ft to be conservative)
Margin at 480 gpm ≈ 10.3 − 17 = −7 ft. Deep cavitation. The pump will lose head and flow, the discharge gauge will hunt, and the flow meter will read erratically - exactly the reported symptoms. This also explains why the symptoms appear only when the valve is opened: everything is fine until the flow moves off the rated point.
Do not chase this with a bigger motor or another rebuild. The system NPSH is the problem.
3. Recommend fixes and quantify what each one buys.
Read the worked answer
Target about 15 ft of NPSHa at rated flow (12 ft NPSHr + 3 ft margin). Current is 12.2 ft, so a 2.8 ft deficit.
Options, in order of cost:
- Raise drum level 3 ft. Gains 3.0 ft directly. Cheapest fix, but only if the drum has the level span and the drum pressure is unaffected.
- Limit maximum flow. A restricting orifice in the discharge, or a VFD with a maximum-speed limit, keeps the pump off the steep part of its NPSHr curve. Since the problem appears only when the valve is opened, this is the most direct and inexpensive control.
- Enlarge the suction line. Line loss scales roughly as D⁻⁴·⁸; going from 6 in to 8 in cuts 5 ft to about 1.3 ft - a gain of 3.7 ft.
- Cool the feed. Each 1 psi of vapor-pressure reduction is worth 2.31/0.62 = 3.7 ft. The 2.8 ft deficit is recovered by roughly 0.75 psi, i.e. about 1 °F of cooling. Even a small trim cooler is a large NPSH gain, but it is capital and it adds a heat exchanger to maintain.
- Lower the pump (if the layout is still open) - 1 ft of elevation, 1 ft of NPSHa.
What I would do: raise the level 3 ft plus install a discharge restriction so the pump cannot be run past 420 gpm, immediately. Put the suction-line enlargement in the next turnaround scope. Also add a low-level alarm at the drum and a note in the operating procedure that opening the discharge valve on this pump makes cavitation worse, not better - and stop rebuilding the pump.