What happens if a submersible pump runs dry?

A mechanical seal needs liquid to lubricate it. Almost every part of what a dry run does to a submersible pump follows from that one sentence. The two seal faces run against each other separated by a fluid film that John Crane puts at about one thirtieth the width of a human hair, and that film is what holds them apart. For the seal to function, the faces need a process fluid or a gas for cooling and lubrication. Take the liquid away and they run on each other, generate heat with nothing to carry it off, and wear.

On a submersible pump a failing seal shows up in one place. Between the wet end and the motor sits an oil chamber with a seal at each end of it. When the lower seal goes, pumped medium works its way into the oil. Drain it and you get a milky emulsion instead of clean oil, and that is the tell that a seal has failed. A station that monitors its pumps sees it earlier than that, because a moisture probe in the oil chamber raises a seal-failure alarm while the upper seal is still holding.

The useful part of that arrangement is that it buys you a stripdown rather than a scrap. Catch it at the seal-failure alarm and you are changing seals and oil. Ignore it and the medium eventually reaches the windings, and then you are rewinding a motor or buying a pump. On a duty-and-standby set the cheap version of this is to act on the alarm and run the standby while the pump comes out.

On a sewage works three causes account for most of what fails, and dry running is the third of them.

First is rag and solids build-up. Wipes and fibre bind up the wet end, and the pump loses its duty long before it loses a component. Second is grit, which usually means the grit removal on the works is not doing its job. Grit abrades, wear follows, and the pump fails early. Third is dry running, and unlike the first two it is nearly always something the installation did rather than something the medium did.

We put it third because it is the least common of the three, not because it is the mildest. Rag and grit degrade a pump over months and give you warning. A dry run can take a seal out in a single unattended night shift, and the pump gives no warning at all unless something on the panel is watching for it.

Rag and grit are answerable at the specification stage. Put a solids-handling pump in and the first cause largely goes away. Dry running is the one no choice of pump saves you from, because the fault is in the system the pump sits in.

The same fault carries three different price tags, and which one you get is decided by the pump type long before anything goes wrong.

On a progressing cavity pump it is catastrophic. The rotor turns inside an elastomer stator and relies on the medium to lubricate that interface. Run it dry and it burns through the stator and damages the rotor as well, so you are replacing both. That is why progressing cavity pumps are the type we are most insistent about protecting.

On a rotary lobe pump the rotors never touch each other, so a dry run does not destroy the wet end. What goes is the mechanical seal, starved of the liquid it needs to lubricate it. That is a stripdown rather than a write-off, and it holds for rotary lobe pumps in hygienic and heavy-duty form alike.

On a centrifugal pump, which covers most submersible pumps, the usual result is again a failed mechanical seal. The impeller touches nothing, so it survives. The submersible case adds the motor, since the motor is normally cooled by the medium around it and a dry run takes that cooling away at the same moment it takes the seal's.

Then there is the pump with no mechanical seal to lose. A double disc pump can run dry, because it has no mechanical seals in it. That matters more on sludge than anywhere else. Rag, grit and dry running are the three things that kill pumps on a sewage works, and a double disc shrugs off all three, which is why it keeps coming up on duties that chew through conventional pumps.

Two things fail on a submersible pump when the level drops, and they fail on different clocks.

The seal loses its lubrication when the wet end stops passing liquid. The motor loses its cooling when the casing is no longer surrounded by medium. Those usually happen together, but not always. A pump sitting high in a wet well can still be passing liquid up the column while the motor housing is in air, and a pump snoring on a falling level is passing air and liquid in turn while the motor is still submerged. The failure you get depends on which one you have.

There is a variant that changes the answer. Some submersible and dry-installed units carry a cooling jacket, so a coolant circuit takes the heat off the motor instead of the surrounding medium. Those units tolerate being unsubmerged in a way a jacketless pump does not. It is a build option, not a property of the type, so it is a question about the specific unit rather than about submersible pumps in general.

How long a given pump survives either condition is the manufacturer's figure, for that model and that build, and it belongs to them. Read it off the installation and operating manual for the unit you have, with the model and serial number from the nameplate in hand. A number picked up from a forum for a different pump is worth nothing, and acting on one is how a repairable seal failure becomes a rewind.

Dry running is a symptom. The fault is nearly always upstream of the pump, and the pump gets blamed for it.

The two we see most often are a suction tank that is too small and a suction head that is too small. Either one means the pump is capable of emptying its own supply, so the level falls to the point where the wet end is passing air and the pump has no way of knowing it should stop. That is a system fault presenting as a pump fault, which is the pattern behind most of what lands on our desk. A pump specified in isolation from the pipework, the tank and the level control will find whichever of the three is weakest.

Oversizing gets there by a different road and ends in the same place. Margin stacks: the customer states a flow, the consultant adds some for future duty, the contractor adds some, the installer adds a bit more. Nobody in that chain is careless and the pump that comes out of it is far too big. Size for 10 m³/h when the tank will only give you 5 m³/h and the pump empties the tank and pumps air. NPSHa has to exceed NPSHr at the duty point on the curve, and when it does not you cavitate. Our pump selection guide sets out the order those questions actually get asked in.

Then there is the one with no engineering in it at all. Somebody shuts a valve, forgets it is shut, and runs the pump. Same outcome, same failed seal, no design fault involved, and it happens on well-run sites. If you are working back from a dry run and the suction design checks out, check the valves before you conclude anything about the pump.

A pump that has run dry once will run dry again unless something in the system changed. Take the causes in the order that clears the cheapest and most likely first.

Fit dry-run protection, then go and fix the reason it would ever operate.

The usual arrangements are a low-level float or electrode set that inhibits the start, a motor power or current monitor that spots the drop in absorbed power when the pump loses its prime, thermal sensors in the windings, and a moisture probe in the oil chamber that flags a seal failure before the motor is at risk. On a wastewater station a level transducer with a float as the independent backstop is the standard belt-and-braces arrangement, because a transducer in a wet well is one rag away from reading nonsense.

Protection costs you something, though, and it is worth saying what. Every one of those devices works by stopping the pump, and a pump that keeps stopping on a station with no standby is a station that overflows. You have converted a mechanical failure into an operational one. Nearly always the right trade, but it is a trade, and it is the argument for duty and standby on any station where a stop means a spill.

The honest verdict is the unglamorous one. If your pump is tripping on dry run, the protection is working and the installation is wrong. Send us the flow, the head, the medium and the tank volume and we will tell you which of the two needs changing, even when the answer is that the pump you already have is fine.