Pump Cavitation After an Exchanger Bypass
Toluene at 90 C is transferred from a blanketed surge drum at 0.15 barg to a distillation feed drum by a centrifugal pump rated 100 m3/h at 55 m head. The drum liquid level is 4.0 m above the pump centerline. The suction line is 6-inch with two elbows and a fully open gate valve; at the current flow the calculated friction loss is 3.6 m. The pump curve shows NPSHr of 3.8 m at 100 m3/h and 7.9 m at 140 m3/h. Since a downstream exchanger was bypassed last week, the pump has been running at 138 m3/h, the discharge control valve is wide open, and the flow indicator swings by 8% around that value. The pump sounds like it is moving gravel, and the discharge pressure gauge oscillates by 0.4 bar. The suction gauge reads 0.9 barg and is steady. Toluene at 90 C has a vapor pressure of 54 kPa abs and a density of 785 kg/m3. The existing suction line cannot be changed before the turnaround.
1. Calculate the NPSHa at the observed flow and compare it with the NPSHr. State the margin criterion and your conclusion.
Read the worked answer
NPSHa
Available NPSH is the absolute pressure at the pump suction flange, converted to head, plus the static head, minus the suction-line friction:
NPSHa = (P_drum - P_vap)/(rho x g) + z - h_f
P_drum = 0.15 barg = 116.3 kPa abs P_vap at 90 C = 54 kPa abs Delta P = 62.3 kPa rho = 785 kg/m3, g = 9.81 m/s2
(62.3 x 1000)/(785 x 9.81) = 8.09 m
Static head z = 4.0 m Suction friction at 138 m3/h = 3.6 m
NPSHa = 8.09 + 4.0 - 3.6 = 8.49 m
Margin
At 138 m3/h the pump curve gives NPSHr = 7.9 m. Margin = 8.49 - 7.9 = 0.59 m.
The usual criterion is at least 1.0 m, or 10% of NPSHr, whichever is greater. Here 0.59 m is below both. The pump is cavitating at the impeller eye.
At the design flow of 100 m3/h the suction friction would fall to about 3.6 x (100/138) squared = 1.9 m, and NPSHr is 3.8 m. NPSHa = 8.09 + 4.0 - 1.9 = 10.2 m, a margin of 6.4 m. The pump is fine at design; the problem is the runout flow, not the pump or the drum height.
2. Explain why the suction pressure gauge can look healthy while the pump cavitates, and why the flow indication is oscillating.
Read the worked answer
Why the gauge looks healthy
The gauge is on the suction flange, upstream of the impeller. At that point the fluid is still subcooled: 0.9 barg is about 191 kPa abs against a vapor pressure of 54 kPa, so the gauge shows a comfortable margin. Cavitation does not happen at the flange; it happens at the impeller eye, where the local velocity is highest and the static pressure is lowest. The eye pressure is lower than the flange pressure by the velocity head and the inlet losses, and that is where the pressure crosses the vapor pressure.
A gauge reading is also a time average. It cannot show a pressure that collapses for a few milliseconds per impeller passage.
Why the flow oscillates
Once vapor bubbles form at the eye, they collapse in the higher-pressure region of the impeller. That produces:
- a loss of head and flow while the vapor pocket occupies the passage;
- a rumbling or gravel noise, which is the collapse of the bubbles;
- a recovery when the flow drops enough to raise the eye pressure, followed by a return to the high-flow condition.
The result is the 8% flow swing and the 0.4 bar discharge pressure oscillation. The suction gauge stays steady because the cavitation is downstream of it.
The gravel noise, the flow swing and the discharge pressure oscillation together are the classic cavitation signature; a healthy suction gauge does not contradict it. Do not use the suction gauge alone as an operating criterion.
3. What would you change, in order of preference, and what would you not do?
Read the worked answer
In order of preference
- Throttle the discharge to bring the flow back to about 100 m3/h. This is the immediate fix. At 100 m3/h the suction friction drops to about 1.9 m and the NPSHr drops to 3.8 m, giving a margin of about 6.4 m. It costs capacity, but the exchanger bypass is the reason for the high flow.
- Raise the drum pressure or level, if the drum is rated for it. The blanket is at 0.15 barg. Raising it to 0.5 barg adds about 4.5 m of NPSHa, but check the drum design pressure and the blanket gas supply first; this also changes the drum's vent and relief requirements.
- Reinstate the bypassed exchanger or add system resistance so the pump stops running out to the end of the curve.
- At the turnaround, replace the suction line with a larger diameter or install a low-NPSH impeller or inducer. The 6-inch line with two elbows and a gate valve is marginal at 138 m3/h.
What not to do
- Do not open any suction valve further. There is nothing to gain, and if a valve is partly closed it is probably the discharge, not the suction.
- Do not add a discharge-to-suction recirculation line to keep the flow up. It heats the suction liquid and reduces the vapor pressure margin.
- Do not run at 138 m3/h and simply monitor. Cavitation destroys the impeller, the wear rings and the mechanical seal, and the damage can happen in days.
- Do not raise the drum pressure without checking the drum relief and design rating.