Self-priming

A plain centrifugal impeller cannot shift air. Air weighs almost nothing against water, so an impeller spinning in it develops almost no pressure difference, and the liquid never comes up the pipe. The pump sits there making noise.

A self-priming design gets round that by holding liquid in the casing when the pump stops. On start, the impeller throws that liquid outwards and drags air out of the suction line with it. The air and liquid mixture passes into a separation chamber, where the air rises and leaves through the discharge and the liquid drops back to the impeller to be used again. Every pass takes a little more air out of the suction line. When the last of it has gone, atmospheric pressure pushes medium up the pipe and the pump runs as a normal centrifugal.

Two things follow from that cycle. The discharge has to be open, because the air the pump strips out needs somewhere to go, and a closed valve or a non-return valve that will not lift with air behind it will stop a prime dead. And priming is not instant. The time it takes grows with the volume of air in the suction line, so a long or oversized suction pipe is paid for at every start.

Self-priming does not mean the pump reaches any depth. Nothing pulls liquid up a suction line. Atmospheric pressure pushes it, and atmospheric pressure is finite.

That gives a ceiling of about 10 m of water column in theory, and roughly 6–7 m in practice once friction in the suction pipe and fittings and the vapour pressure of the medium are counted, which is the same limit we set out in the notes on abstracting from a river or lagoon. A warm medium or a volatile one takes more off it again, because its vapour pressure is higher and the margin to flashing is smaller. Past that margin the medium turns to vapour at the impeller inlet and what you have is cavitation, not priming.

So the honest answer on a deep source is to move the pump rather than specify a better self-primer. If the water sits below what atmospheric pressure will lift, a submersible or a borehole set removes the suction question instead of stretching it. We would tell you that rather than quote you a bigger surface pump.

Initial priming happens once. Somebody fills the casing through a plug at commissioning, or the pump is run with the suction flooded until it is wet. Re-priming is what the pump has to manage on its own every time it starts after that, and it is the one that decides whether the installation works unattended.

A self-priming pump re-primes because it retained liquid in the casing between runs. Anything that empties the casing takes the capability away. A suction line that siphons back through a failed foot valve, a drain plug left loose after a strip-down, or a long shutdown in warm weather will all do it. A pump that primed perfectly at commissioning and will not start on a Monday morning has usually lost its prime liquid rather than its impeller.

Air ingress is the other cause and the harder one to trace. A suction line under vacuum draws air in through joints that would hold pressure quite happily, so a leak that never shows as a drip on the floor will still stop the pump priming. Gland packing, the suction gasket and any threaded fitting on the vacuum side are where to look first.

Self-priming is usually discussed as a centrifugal feature, because the centrifugal pump is the one that cannot manage without help. Self-priming centrifugals are the right answer for water and near-water at high flow where the pump has to sit above the source.

Positive displacement pumps arrive at the same capability from the other direction. A pump that seals a fixed volume on every stroke shifts air much as it shifts liquid, so it clears its own suction line as a consequence of how it works rather than as a feature bolted on. That is why the capability crops up right across our reciprocating pump range.

The Penn Valley Double Disc Pump is the one we reach for on sludge, and it is the concrete example. Penn Valley states that the pump will dry prime at 2.1 m (7 ft), re-prime at 5.5 m (18 ft) and generate up to 25 inHg. The dry prime figure is the interesting one. It means the pump pulls the line down with nothing in the casing at all, which a self-priming centrifugal cannot do, because it has to start with liquid in it.

It manages that because two discs do both the pumping and the valving, seating positively on every revolution, and because the fluid chamber is sealed with flexible trunnions. Penn Valley's own description is that the trunnions eliminate packing, mechanical seals and flushing water, and that the low friction design lets the pump run dry without damage. The two go together. A pump that primes from dry is only worth having if the dry running it does while priming is harmless, and on a sewage works the suction line is empty far more often than anyone plans for.

Name the trade-off before you specify one. The double disc is a reciprocating pump with an operating ceiling of 4.1 bar, so a long high-head rising main is not its duty, and it delivers a pulse that a pulse-sensitive dewatering feed wants dampeners for. It earns its place on the medium, not on the head.

A self-priming centrifugal is a bigger, heavier and more expensive machine than the equivalent flooded-suction pump, because the priming and separation chambers are extra casing volume that does no pumping. The recirculation that clears the line costs efficiency, and the priming time costs you at every start on a pump that cycles.

So the verdict is simple. If you can flood the suction, flood it. Drop the pump below the liquid level, or put a submersible in the chamber, and the priming question never arises. Specify self-priming when the site will not allow that: a tanker or trailer set, a portable unit that moves between sumps, a pump that has to sit on a bund above a lagoon, or a wet well whose level swings below the pump on a normal day.

Send us the flow, the total dynamic head and the medium, and add the vertical distance from the lowest liquid level to the pump centreline along with the length and bore of the suction pipe. Those last three are what decide whether self-priming is the right answer or whether the pump simply needs to go lower. If it needs to go lower, we will say so.