How a Mechanical Trap Fails and What You See

Why this matters

The two mechanical trap families use buoyancy the opposite way round from each other, and that one geometric difference decides which failure you get told about. A float rides up on condensate to open its valve, so a float that fills with water sinks and holds the valve shut: the equipment floods, the heat stops, and somebody calls you the same day. An inverted bucket rides up on steam to close its valve, so a bucket that loses its water seal cannot rise and leaves the valve open: steam blows through continuously, nothing gets cold, nobody calls, and the loss runs for as long as it takes someone to notice a water treatment order. One of those failures costs a service visit. The other costs a percentage of everything, indefinitely.

Before you test or pull one

Never identify a trap's condition by hand. The body is at line temperature and the outlet is live.

To test in service: use an ultrasonic probe with its stinger held against the trap body through the instrument's own handle, and a contact temperature probe on a cleaned spot with your hand and the lead clear of the pipe. If the trap has a test valve, operate it only where that valve discharges to a rated drain point, stand beside it and out of the plane of the outlet, and wear face protection under 29 CFR 1910.133 with sleeves down.

To remove one: isolate the inlet and the outlet, lock and tag both under 29 CFR 1910.147, vent the section to atmosphere through a drain you operate from the side, confirm zero gauge pressure on a gauge you can read, and let the body cool to a temperature you have measured before a joint is broken. Condensate held between two closed valves is above its atmospheric boiling point and will flash out of the joint you open. When you shake a removed float to check for water inside, do it after the body has cooled and over a container, not over your own boots.

The two geometries, and the failure each one defaults to

Float and thermostatic. A hollow ball floats on rising condensate and lifts a lever off a seat. Its default failure is a punctured or waterlogged float, which sinks and holds the valve closed. What you see is flooded equipment, no heat, an inlet at saturation temperature and an outlet that is cold, and no discharge at all on an ultrasonic probe. The overwhelming cause of a punctured float is water hammer, which is why this family is the wrong choice anywhere hammer is a live risk. Two other failures belong to the same trap and run the other way: the separate thermostatic air vent element can fail open and blow steam through the vent, and a seat that has been wire-drawn by dirt leaks through. Neither of those is the buoyancy failure, and neither is as common as the float.

Inverted bucket. An open-bottomed bucket sits mouth-down in a water seal. Steam entering underneath makes it buoyant, it rises, and the lever closes the valve. Condensate displacing that steam takes the buoyancy away, the bucket sinks, and the valve opens. Its default failure is loss of the water seal, usually called loss of prime, after which the bucket cannot float at all and the valve stays open. What you see is a continuous rush on an ultrasonic probe with no cycle, a return line running hot, and no complaint from anybody. The counterpart failure runs the other way: the small vent hole in the top of the bucket exists to bleed air out, and when scale plugs it the trapped air keeps the bucket buoyant and the valve stays shut, which produces the flooded-equipment picture instead.

So the families are inverted in their default direction and each one can also fail the other way through a secondary mechanism. That is worth holding precisely, because it is what lets you predict the complaint from the nameplate.

Why a prime gets lost, and at which end of the operating range

A bucket's prime is water, and water at line pressure is above its atmospheric boiling point. Drop the pressure quickly and part of that seal flashes to vapour and leaves. The mechanism therefore lives at the fast-depressurisation end of the operating range: a header shut down abruptly, a pressure reducing station that swings, a load that goes from full to nothing in seconds, or steam arriving with superheat that boils the seal off. At the other end, a header held at steady pressure and cycled gently, a bucket holds its prime indefinitely, which is why the same trap model gives twenty years of service in one plant and fails every season in another.

The standard remedy follows directly from the mechanism: a check valve at the trap inlet, so that when upstream pressure falls the seal cannot be pushed backwards out of the body. It is a small fitting and it addresses the actual cause rather than the symptom.

The case: one loud failure and five quiet ones

Two unrelated reports arrived in the same month at a plant with a 100 psig yard header and a low-pressure building distribution.

The loud one. A service call for no heat in the north bay. The unit heater there drains through a float and thermostatic trap. An ultrasonic probe on the body showed no discharge at all. A contact probe on cleaned spots read the inlet at saturation for the measured pressure and the outlet cold, which is the flooded signature. After isolating, locking out, venting and letting the body cool, the trap came out and the float rattled with water in it. Two hours of labour and a part, and the customer was satisfied that afternoon.

The quiet one. Separately, the facility manager mentioned that water treatment chemical orders had roughly doubled over the past year. Nobody had reported a fault, because nothing had got cold. The make-up water log, which the plant had been keeping without looking at, showed a step change about fourteen months earlier that matched a weekend when the yard header had been dropped fast for a repair.

A survey of the twenty inverted bucket drip traps on that yard header found five discharging continuously, with the same rush and no cycle on every one. All five had lost prime, and the shared cause was the shared event: this header is shut down abruptly every weekend, and each shutdown is a chance to flash the seal away.

Putting a bound on the quiet failure

You cannot measure a blowing trap's loss with the ultrasonic probe that found it, because that instrument reports relative acoustic energy with no absolute basis. It distinguishes flow from no flow and it cannot tell you pounds per hour.

What you can do is bound it. Napier's relation gives steam flow through an orifice as roughly 24.24 times the absolute upstream pressure in psia times the orifice area in square inches, in pounds per hour. Read the conditions it was derived under before you use it: it assumes dry saturated steam, a sharp-edged orifice, a discharge coefficient of one, and critical or choked flow, which for steam requires the downstream absolute pressure to be below roughly 0.58 of the upstream absolute pressure.

Check the condition first. Upstream is 100 psig, or 114.7 psia; the return is near atmospheric, 14.7 psia; and 14.7 divided by 114.7 is about 0.13, comfortably choked. Now take a one-eighth inch seat, whose area is about 0.0123 square inches: 24.24 times 114.7 times 0.0123 is about 34 pounds per hour through one trap. Five traps is about 170 pounds per hour, against a plant producing 3,000 pounds per hour, or about 5.7 percent of production.

Report that as a bound, not as a measurement, and never as a plus-or-minus. Every one of the assumptions runs in the same direction: a worn seat is not sharp-edged, the real discharge coefficient is below one, and what leaves a failed trap is a two-phase mixture of flash and water rather than dry steam, all of which put the true figure under the bound. What the number is good for is exactly one decision, which is whether five silent traps are worth a survey programme. Under 6 percent of production, running for fourteen months and counting, against a survey that takes a technician a day, answers that without needing a better number.

Set the two failures side by side and the asymmetry is the whole lesson. The flooded coil cost two hours and was fixed the day it happened, because its failure direction generates a complaint. The five open traps cost a share of every pound the boiler made for over a year, because their failure direction generates silence. Nothing about the second failure is harder to find than the first. It just has to be looked for.

What would change the picture

A pressurised return changes what a blowing trap does. Where back pressure is high enough that flow is no longer choked, Napier's relation does not apply at all and the loss is lower than the bound it gives. Check the ratio before you use the formula, not after.

A float trap on a hammer-free low-pressure heating main rarely punctures a float and will more often fail through a dirt-cut seat instead, which reverses its usual direction to a leak-through. If you are seeing leak-through on float traps rather than flooding, look at the strainers.

An inverted bucket on a steady, gently cycled header is not the trap in this case study. The prime failure is caused by the shutdown practice, so a plant that ramps its header down over minutes rather than seconds will not reproduce it however many bucket traps it has.

Superheat. Steam arriving with superheat will boil a bucket's prime away during normal running rather than at shutdown, and no inlet check valve fixes that. That is a family selection problem, not a maintenance problem.

How to verify a mechanical trap without guessing

Take three readings in this order and let them agree before you call it.

Acoustic. Hold the probe against the body. A healthy float trap gives a continuous modulated flow that varies with load. A healthy bucket gives discrete discharges with quiet between them. A continuous unvarying rush from either is a fail-open, and total silence from either is a fail-closed.

Temperature across the trap. Contact probe on cleaned spots at inlet and outlet, hand and lead clear. Use the same probe for both readings: its systematic offset is common to both and largely cancels in the difference, leaving that percentage of the difference rather than of either reading, while its random spread does not cancel and gives a difference about 1.4 times as spread as a single reading. An inlet well below saturation on a mechanical trap is not that trap subcooling on purpose, because these families do not subcool; it is condensate backing up ahead of it.

Downstream line temperature during a known-idle period. A drip trap on a warmed main should make very little at steady state. A return line that stays hot at a drip point when the main has been up for hours is passing steam, whatever the trap sounds like.

References

• Napier's relation for steam flow through an orifice under critical flow, as published in trade steam engineering references, with its dry-saturated-steam and sharp-edged-orifice conditions
• Steam trap manufacturer published data for seat sizes, capacities at stated differentials, and inlet check valve recommendations for prime retention
• 29 CFR 1910.147 for isolation of pressure and thermal stored energy before removing a trap; 29 CFR 1910.133 for eye and face protection when operating a test valve
• See related: What a Steam Trap Is Actually Deciding; The Trap Families and What Each One Is Good At; Why Condensate Is the Expensive Half of a Steam System