As soon as I looked at the knockout drum, a sinking feeling took over. The plant was mystified by level trips from a compressor interstage knockout drum on a compressor system and complained the liquid pump on the drum “never seemed to work right.” Going out with the unit engineer quickly led to identifying one problem with the system. The liquid drum was so close to the ground the clearance gave barely enough room for the isolation valve and an elbow to squeeze in under the vessel. The liquid height in the vessel was about 2 feet above the pump inlet nozzle.
The system originally was designed for a steady-state operation where the interstage pressure could force the liquids out with no pump. To save money, the interstage drums were placed a minimum height above grade. The bottom flanges of the vessels were barely 18 inches above the concrete slab.
Intermittent loads in the compressor system caused fluctuating drum pressures. Some times the pressure in the drum dropped below the pressure of the liquid collection system for the interstage liquids. Therefore, the plant decided to add a pump to ensure liquid could get out of the system. However, the net positive suction head (NPSH) available for the pump was extremely low — less than 18 inches.
The plant was most familiar and comfortable with centrifugal pumps and, so, installed one to remove the liquids. The pump suffered continual problems and often stopped working. In fact, on some occasions no liquid at all seemed to get to the pump. The pump had lost prime.
The changing system loads on the compressor were one reason for this. When the compressor gas load decreased, the interstage pressure dropped. Liquid that was in the drum started to vaporize when the pressure fell. In such cases, the feed to the pump contained vapor bubbles, not just liquid.
So, the pump had to contend with two issues. First, very little head was available to get liquid into the pump. Second, vapor intermittently formed in the liquid.
The final analysis was that the pump must meet three requirements:
• operate at low suction head available;
• tolerate some vapor in the feed; and
• be robust enough to survive (with reasonable reliability) some amount of damage due to vapor bubble collapse.
This led us to investigate four types of positive-displacement (PD) pumps:
• regenerative turbine; and
For PD pumps, you must evaluate the net positive inlet pressure (NPIP) required rather than NPSH used with centrifugal pumps.
We quickly vetoed reciprocating pumps for two reasons. First, acceleration effects on the NPIP would be large. Second, vapor in the pump would cause problems with pistons and valves.
Regenerative turbine, vane and gear pumps had a huge advantage for this service. They all could provide smooth flow characteristics. They all could have flow control added by using variable frequency motors to regulate capacity.
We opted for a gear pump. One key reason for the choice was that a physically robust unit could be found for a reasonable price. (For details on such pumps, see: “Get Up to Speed on Gear Pumps.”)
Gear pumps also are self-priming. Indeed, they can pull all the way down into a vacuum if required. They can tolerate bubbles in the feed. Nevertheless, these benefits come with some costs. Figure 1 shows the typical shape of the curve for capacity versus suction pressure for a gear pump. As vaporization starts in the feed (suction pressure drops below NPIP), the capacity drops. So, coping with varying system pressure requires changing operating speed to maintain the same capacity. A flow controller based on pump speed may need some tuning to handle unstable feed conditions.
In addition, pumping vapor remains a severe service. The most difficult operation is when the vapor bubbles in the feed collapse completely. The cavitation shock wave can create severe damage. Even though a gear pump is more damage resistant than a centrifugal unit, damage will occur. Fortunately, the much slower speed of the gear pump decreases the damage.
No pump can completely avoid problems at low suction heads. However, if you can’t change the process, select a pump that can reduce problems to a tolerable level.
ANDREW SLOLEY is a Chemical Processing contributing editor. You can email him at ASloley@putman.net.