A mechanical seal on a boiler feed water pump fails mainly from heat: when the seal chamber is not cooled enough, the feedwater stops lubricating the faces, flashes to steam across them, and takes the O-rings past their temperature limit. Vibration and misalignment, cavitation and low flow, dirty feedwater, and start-up temperature swings are the other causes.
Boiler feed pumps move water at 100 to 200 °C and 20 to 48 bar, and a single pusher seal survives that duty only because a support system, usually an API Plan 23 cooling loop, keeps the seal chamber much cooler than the pump. When a seal keeps failing, check that loop before ordering another seal.
What Causes Boiler Feed Pump Seal Failure
A boiler feed pump seal fails when its chamber runs too hot for the faces and O-rings, or when the pump around it vibrates, cavitates, carries dirt to the faces, or changes temperature faster than the seal can follow.
Inadequate Seal Chamber Cooling
A Plan 23 loop fails the seal when it stops carrying heat away: the chamber warms toward pumping temperature, the water film between the faces thins and flashes, and the faces wear fast.
The pumping ring that drives the loop makes only a small fraction of the head the impeller makes, so every restriction in the circuit cuts the flow. On the cooling-water side, hard water fouls the coil, and fouling is more common when the water moves slower than 0.9 m/s (3 ft/s).
Air left in an unvented loop collects around the seal faces and runs them dry. Venting too much does the opposite damage: it draws hot process water past the throat bushing into the loop and raises the temperature at the faces and O-rings. A worn throat bushing, or one left in from the last seal, lets the same hot water mix into the chamber.
The cooler itself can be too restrictive or too small. In a published case study of a refinery boiler feed pump on Plan 23, the double-wrapped, 36 ft cooling coil let less than 0.5 GPM through the loop, and the cooler was too small for the heat load.
Flashing at the Seal Faces
Flashing turns the water film between the faces into steam, so the faces lose the liquid that cools, lubricates and separates them, and they wear fast.
The closer seal chamber pressure sits to the water’s vapor pressure, the earlier the water vaporizes as it crosses the faces. Cooling the chamber widens the gap between the two.
API 682, the petrochemical seal standard, asks for a seal chamber pressure at least 3.5 bar (50 psi) above vapor pressure, or 1.3 times it. It does not as a rule govern power-plant feed pumps, so use the figure as a yardstick.
Vibration and Misalignment
Vibration and misalignment move the shaft at the seal faces, so the faces wear faster and the dynamic O-ring rolls until it is cut in a spiral or jams the sliding face so it can no longer follow its mate.
On a feed pump the usual sources are damaged bearings, pipe loads on the nozzles, and an alignment made cold. A multistage feed pump grows as it heats, so the final alignment check is made with pump and driver at operating temperature.
API 610, the petroleum-industry pump standard, caps total shaft deflection at the primary seal faces at 0.002 in (0.05 mm) under the most severe dynamic conditions. That is a limit on the running shaft; the static runout you measure with a dial indicator before installing a seal is a different reading.
Cavitation and Low-Flow Operation
Running a feed pump well away from its best efficiency point, on either side, brings cavitation, vibration and recirculation, and raises the radial load on the seal, so it wears and fails sooner.
On a feed pump, low flow comes from the system: a feedwater control valve throttling the pump back on its curve, or a minimum-flow bypass line too small for the job. The allowable operating range and the minimum flow are set by the pump manufacturer, so check the operating point against their curve.
Cavitation can also come from a transient lasting seconds. After a steam turbine trip or a sudden load cut, deaerator pressure decays while the suction line is still full of hot water. On a hot restart, the pump and suction pipe are hotter than the water arriving from the deaerator, and steam can flash at the suction.
Contaminated Feedwater
Dirty feedwater wears the seal faces by abrasion when hard particles reach the gap between them. On feed pumps the particles are iron oxide and pipe scale, and they arrive in bursts at start-up.
Dissolved minerals that drop out of the water wear the faces the same way, and where they build up on the atmospheric side they can jam the O-ring so the face hangs up.
Thermal Transients at Start-Up and Shutdown
A seal is damaged at start-up and shutdown when the pump’s temperature changes faster than its parts can follow.
- Thermal shock: hot feedwater into a cold pump, or the reverse, can fracture the seal faces. This hits the spare pump that sat at ambient temperature until the running pump tripped.
- Axial shaft growth: during warm-up the shaft and casing heat at different rates, so the shaft grows through the seal chamber and moves the seal off its setting. On between-bearings multistage pumps the drive-end seal usually sees the most growth, because the thrust bearing typically sits at the non-drive end. A high-temperature cartridge seal can typically take up to 1 mm either way from where it was set cold.
- Thermal stratification: in an idle or poorly warmed pump, hot water rises and cooler water settles, so the casing and rotor expand unevenly. Starting the pump in that state can cause rubbing, vibration and seizure.
O-Rings Above Their Temperature Limit
An O-ring run above its limit in hot water degrades and loses its resilience, so the seal leaks past it, or the O-ring sticks and holds the sliding face back.
The Fluid Sealing Association does not recommend FKM O-rings in water above 70 °C (160 °F), yet FKM is the base elastomer in API 682 seals and common in chemical-duty seals. EPDM is the usual elastomer for hot water and steam.
Warning Signs Before a Boiler Feed Pump Seal Fails
The signs that come before a boiler feed pump seal fails are leakage that trends upward or turns erratic, a Plan 23 loop that runs hotter, and water turning up in the bearing oil.
Judge leakage against the pump’s own baseline. On boiler feedwater, 1 to 2 drops per minute is normal and some services run normally at 10, so a drip by itself does not mean the seal is failing. A dry gland is no proof of a tight seal either: on hot water the leakage often evaporates before it can be seen, and steam escaping at the gland can be invisible.
Read the trend at steady running conditions, because leakage during start-up or load swings can reach 100 times its steady value. The patterns that matter:
- Leakage rising steadily: usually gradual face damage such as chipping, scratches or taper. At steady conditions the seal can stay in service until it reaches an agreed limit.
- Leakage swinging by orders of magnitude: on a critical pump such as a feed pump, pull and inspect the seal now or at the next opportunity.
- Leakage jumping with a change in vibration: a rise of less than ten times may settle; more than ten times usually means damage that will not reverse.
- Loop temperature rising: leakage leaving the Plan 23 loop is replaced by hot process water, which heats the loop. Thermocouples at the cooler inlet and outlet show it; compare their trend with collected leakage.
- Water in the bearing oil: spray from a failing seal gets through labyrinth seals into bearing housings and motors.
How to Find the Cause From the Failed Seal
Trace the failure by reading the failed parts against the Plan 23 loop temperatures and the pump’s operating history: the damage shows how the seal failed, and the loop data and history show why.
While the pump is still running, record the flush inlet and outlet temperatures at the cooler and the gland temperature. A water flush is normally sized for a rise under 15 °F across the seal. With the pump stopped, a static pressure test separates an O-ring leak from a leak across the faces.
Then read the parts. Face damage and O-ring condition point to these causes:
- Carbon face worn far down, nose gone: the film was lost to heat or flashing. Check loop flow and cooler capacity first.
- Cracked or fractured face: thermal shock. Check what happened at the last start, trip or standby changeover.
- Hard, flattened or cracked O-rings: overheating. Check the elastomer against chamber temperature.
- Rolled or spiral-cut dynamic O-ring: shaft deflection or vibration.
- Scored or grooved faces: particles in the feedwater.
- Mineral build-up on the atmospheric side: scaling that hung the face up.
Finally, line the failure up with the history. Leakage that rose after a change in operating conditions can be the faces wearing in to the new conditions. A new seal that leaked 100 times more than expected at start-up was most likely damaged or set wrong at installation. Seals failing every few months on the same pump point to the system: a feed control valve throttling the pump back on its curve, an undersized bypass, or pipe loads on the nozzles.
How to Stop Repeat Seal Failures
Stop the repeats by fixing the cause the failed seal and the loop temperatures point to before the next seal goes in, because a new seal in the same conditions fails the same way.
In the case study, a tighter, insulated throat bushing, a better pumping ring, and a shell-and-tube cooler with more capacity and less restriction to flow took mean time between repairs from 6–12 months to more than 48 months.
For the Plan 23 loop itself:
- Cooler: shell-and-tube rather than plate, and cleaned at every seal change.
- Tubing: short runs, as few valves as possible (full-bore where needed), long-radius bends, and a steady rise to a high-point vent.
- Throat bushing: renewed with every seal change.
- Venting: both sides of the cooler vented before start and periodically in service, but not so often that hot process water is pulled into the loop.
- Filter: a magnetic filter in the loop to catch rust and pipe scale.
Warm the standby pump before it takes load. Without figures from the manufacturer, start it once the casing is within 16 °C (30 °F) of pumping temperature and has soaked there for 2 hours.
Where the evidence points to assembly, the fix is the setting and handling at installation. Where it points to the seal itself, such as FKM O-rings in hot water, the replacement is a selection question for boiler feed pumps.