The Trap Families and What Each One Is Good At
Why this matters
Trap selection gets made on the shelf: whatever the truck carries, in the size that matches the pipe. The pressure rating is checked, the family is not, and the family is the part that decides whether the equipment works. Selection is driven by the shape of the condensate load, not by the pressure, and it comes down to three questions you can ask standing in front of the equipment. Five of the six columns in the comparison table below are derivable from those questions plus the test each family runs, so you can rebuild most of it from memory on a job with no literature. The last column is not derivable at all: tolerance of hammer and superheat is a fact about how the body is built, and anything with a thin bellows or a thin float is what dies first.
Before you fit or 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 beside rather than in front of the outlet, confirm zero gauge pressure on a readable gauge, and let the body cool to a temperature you have measured on a probe before a joint is broken. Condensate held between two closed valves sits above its atmospheric boiling point and flashes out of the first opening you make.
When you read a return-main pressure to size a trap, read it from beside the gauge, not in front of any plug, cap or gauge fitting, with face protection under 29 CFR 1910.133. Where a trap is being fitted outdoors in a freezing climate, mount it so the body drains when the system is off and lag or trace it as the design requires, because a trap that holds water in a dead system splits its body and that failure is discovered as a live leak on the first cold start.
The three questions
Does the load surge, or is it steady? A drip leg on a running main makes condensate at a steady rate. A modulating coil, a batch vessel and any equipment that starts from cold make almost all of their condensate in the first minutes and then very little. A surging load needs a family that can pass a large slug and then modulate down without dumping steam.
Is there air to clear? Air and other non-condensable gases arrive whenever equipment cools and breathes, and they do not condense, so they collect at the coldest surface and blanket it. A trap that runs a density test cannot pass air by itself: a float sinks in air the same way it sinks in steam, and an inverted bucket cannot float on it. That is why some families ship with a thermostatic element attached and some need a separate air vent.
Can the equipment tolerate condensate backing up? A trap running a temperature test refuses to open until the condensate has cooled below saturation, so it necessarily holds water back. On a radiator or a traced line that is harmless or actively useful. On a temperature-controlled heat exchanger or an outside-air coil in winter it is the mechanism that floods tubes and splits them.
The families
| Family | Test it runs | Discharge | Passes air | Backs condensate up | Needs differential | Tolerates hammer and superheat |
|---|---|---|---|---|---|---|
| Float and thermostatic | Density, plus a separate thermostatic vent | Continuous, modulating | Yes, through its vent | No, drains at saturation | Very little | Poor, the float is the weak part |
| Inverted bucket | Density | Intermittent | Slowly, through a small bucket vent hole | No | Some, and it must keep a water prime | Good |
| Balanced-pressure thermostatic | Temperature, tracking the saturation curve | Intermittent | Yes, it is essentially an air vent | Yes, by design | Very little | Poor, the bellows is the weak part |
| Bimetallic thermostatic | Temperature, on a fixed element curve | Intermittent | Yes | Yes, and usually by more than a bellows | Some | Good |
| Thermodynamic disc | Flash velocity at a seat | Rapid cycling by design | Poorly, it can air-bind on start-up | No | Yes, and it has a back-pressure limit | Good |
Read the columns against the three questions and the picks fall out without memorising anything. Surging load with air and no tolerance for backup: only one row satisfies all three. Steady drip on a high-pressure header where backup is harmless and dirt is the real enemy: the rugged density rows win. Radiator with a large air load and full tolerance for subcooling: the temperature rows.
A sixth device gets sold hard and deserves naming: the fixed-orifice or venturi trap, which has no moving part and no test at all. It is a hole sized for one condensate rate. On a genuinely constant load it works; at any load below the one it was sized for it passes live steam continuously, and a load that varies is most loads.
The artifact: a selection card for one plant
A single facility with six distinct steam loads, sized at the lowest differential each trap will actually see rather than at the design differential.
| Load | Conditions | Family chosen | Why that one |
|---|---|---|---|
| Boiler-room drip legs | 100 psig header, old system carrying scale, return near atmospheric | Inverted bucket | Steady load, minimal air after warm-up, backup harmless, and the bucket is the family that shrugs off dirt and hammer |
| Yard main drip legs | 100 psig, outdoors, freezing climate | Thermodynamic disc | High differential is available, freezing and superheat are tolerated, and a mount that drains when the system is off protects the body |
| Indoor heating main drip legs | 5 psig main, 2 psig return | Float and thermostatic | Only 3 psi of differential exists, which the density test survives and the flash-velocity test does not |
| Modulating outside-air preheat coil | Deep turndown, must not flood, freeze exposure | Float and thermostatic | Surging load, air on every start, and zero tolerance for backup; fitted with a vacuum breaker and a gravity drain to a vented point below the coil |
| Two-pipe radiators | Low pressure, gravity return, large air load on every cycle | Balanced-pressure thermostatic | Enormous air duty, backup entirely acceptable in a radiator, and the element opens a consistent margin below saturation as the system pressure moves |
| Outdoor process tracing | Freezing climate, subcooling desired | Bimetallic thermostatic | Backup is wanted here because holding condensate in the tracer recovers sensible heat into the line, and the element survives freezing and hammer |
Six loads, five distinct families, and float and thermostatic is the only one that wins twice, on the two loads whose defining feature is that they surge and must not flood.
Notice the third and fourth rows both landed on float and thermostatic for different reasons: the third because the differential is small, the fourth because the load surges and the equipment cannot flood. Notice also that the first three rows are all drip legs and all three landed on different families, purely on differential, dirt and exposure. Drip service is not a family.
Sizing: two checks, and both have to pass
Check one is capacity at the minimum differential the trap will ever see, not at the header pressure. On a modulating coil that differential collapses as the valve throttles, and the stall condition where it reaches zero is owned by a separate article. On a drip leg it is the header pressure minus the full return-side back pressure including any lift.
Check two is the start-up load, which on a system that warms a long main from cold is several times the running load and arrives in minutes.
Both, not either. Where they disagree badly, do not resolve it by fitting a trap sized for the surge: a trap sized several times over for running discharges poorly at running load and throws away far more when it eventually fails open. Resolve it with drip leg volume to buffer the surge and, on a long main, a warm-up drain that is opened for start and closed after.
Read a published capacity with its basis attached, the same way you would read an instrument specification. Capacity quoted for cold water is not capacity for hot condensate, because hot condensate flashes as it crosses the seat and the resulting vapour occupies a large share of the passage. The two bases differ enough to change a size. Check which one the table you are holding uses, and check the differential the column was derived at, because a capacity is meaningless without it. The manufacturer's table at your actual differential is the only authority on this; there is no generic number.
Apply the manufacturer's stated safety factor for the application rather than a habit. Where none is stated, ask for one rather than inventing it.
What would change a selection
A return that gets pressurised later. A plant that adds a flash recovery vessel raises the return pressure everywhere behind it. Disc traps that were comfortably inside their back-pressure limit can drop outside it, and the symptom is rapid cycling and a jump in make-up.
A pressure reduction at the load. Reducing header pressure at a terminal cuts the differential at that trap, which can push a selection from an intermittent family to a modulating one without anything visibly changing.
A change of duty on the same equipment. A vessel converted from continuous to batch operation turns a steady load into a surging one, and the trap that suited it before will now underdrain on every start.
Freezing exposure added by a re-route. Moving a trap outside changes the family shortlist immediately, and it changes it toward the families whose failure mode leaves the body drained.
Verifying a selection after it is in
Give it one heating season and check three things, none of which need the trap opened.
Watch a cold start. The correct family for an air-heavy load clears its air within minutes and the equipment comes up evenly. A density-only trap with no vent on that load leaves a cold zone that persists until something sweeps it, and that persistent cold zone is a selection error, not a fault.
Listen at running load. Match what you hear to the discharge column: a modulating family should be quiet with a steady flow, an intermittent family should give discrete discharges with silence between, and a disc should click at a regular slow rate rather than a rapid one. Hold the ultrasonic probe against the body through the instrument's own handle, never with a hand on the trap.
Check the equipment, not the trap. On the loads where backup is unacceptable, take a temperature across the equipment at part load, in the air stream where there is one and otherwise with a contact probe on a cleaned spot with your hand and the lead clear of the surface. Output that sags at part load and recovers at full load is condensate sitting in the bottom of the equipment, and it points at either the family or the differential rather than at a broken trap.
References
- Steam trap manufacturer published capacity tables at stated differential pressures, with the basis (cold water or hot condensate) named, which is the only authoritative source for a capacity or a safety factor
- 29 CFR 1910.147 for isolation of pressure and thermal stored energy; 29 CFR 1910.133 for eye and face protection
- See related: What a Steam Trap Is Actually Deciding; Why Pressure Sets Temperature and What That Decides; How a Mechanical Trap Fails and What You See