Views: 0 Author: J-VALVES Publish Time: 2026-11-05 Origin: Site
A steam trap is not a pipe fitting; it is a self-contained automatic valve with three simultaneous jobs - discharge condensate as it forms, vent air and non-condensable gases, and hold live steam. Most trap problems in the field are selection problems made at the desk: the wrong family for the duty, a capacity chart read at inlet pressure instead of true differential pressure, a warm-up load that was never calculated, or a stall condition that was never checked on a modulating application. This guide walks through the selection logic in the order a specifier should follow: duty, family, load, differential pressure, installation, and the testing regime that keeps the population healthy. A trap that fails open burns fuel continuously; a trap that fails closed waterlogs the equipment and feeds water hammer - the failure mode you accept is part of the selection.
The ANSI/FCI definition used across the industry describes a steam trap as a self-contained valve that drains condensate automatically while remaining steam-tight. In practice the trap must perform three duties at once, and every selection decision comes down to which of those duties the application makes hardest. First, discharge condensate the moment it forms: condensate that pools on a heating surface blankets it with a low-conductivity water film and collapses heat transfer; in a steam main it becomes the slug that produces the water hammer described in check valve water hammer and non-slam selection. Second, vent air: steam admitted to a cold system arrives behind a mixture of air and CO2 that collects on heating surfaces. Armstrong's drainage guidance notes that under certain conditions as little as 0.5 percent air by volume can cut heat transfer efficiency by up to 50 percent, which is why system air venting - by the trap or by dedicated thermostatic air vents - is a selection criterion, not an accessory afterthought. Third, hold steam: a trap that passes live steam is a continuous fuel leak, and Section 10 below quantifies what one blown-through trap costs per year.
No single trap family does all three jobs equally well across all duties. A thermodynamic disc trap holds steam superbly and shrugs off superheat and water hammer, but vents air sluggishly. A float trap discharges continuously and vents air well, but cannot tolerate water hammer or freezing. The job of the specifier is to rank the three duties for the specific application, then pick the family whose weaknesses matter least.
Traps separate steam from condensate by one of three physical properties, and that property determines the family's personality. Mechanical traps (float, inverted bucket) act on density: liquid lifts a float or sinks a bucket, and the mechanism opens a valve regardless of temperature. They discharge at condensate temperature, continuously or in batches, and their capacity is essentially independent of sub-cooling. Thermostatic traps (balanced pressure bellows, bimetallic) act on temperature: they stay open until condensate has cooled a few degrees below saturation, then close as steam approaches. Their defining feature is sub-cooling - a deliberate disadvantage on a heat exchanger that needs condensate removed hot, but a deliberate advantage on tracing, where the sensible heat in sub-cooled condensate is useful and the closed trap holds hot condensate in the line instead of flashing it. Thermodynamic (disc) traps act on velocity: flash steam generated across the seat accelerates over the disc, drops the pressure above it, and snaps the disc shut; when the pressure above decays, the disc pops open and blasts condensate. They are intermittent, small, and rugged.
Two mechanical nuances matter at specification time. An inverted bucket cycle depends on a small vent in the bucket, so it vents air slowly unless a separate thermostatic air vent is fitted through the cover. A float trap needs a separate thermostatic element for air venting - which is why the standard process product is the combined float-and-thermostatic (F&T) trap. And because a float trap is a gravity device, it passes condensate at any pressure differential including zero, provided there is gravity head above it - the property that makes it the default choice for modulating applications, as Section 7 explains.
The table below compresses the vendor literature into the properties that actually drive selection. Figures are typical published envelopes across major manufacturers (TLV, Spirax Sarco, Armstrong, Watson McDaniel); a specific product may differ, so always confirm against the manufacturer's own pressure-temperature chart.
Property | Thermodynamic disc | Float & thermostatic (F&T) | Inverted bucket | Balanced-pressure thermostatic | Bimetallic |
|---|---|---|---|---|---|
Discharge pattern | Intermittent blast | Continuous modulating | Intermittent | Continuous at load | Intermittent / continuous |
Typical min operating pressure | ~0.25-0.3 bar g | Vacuum up | Needs working differential | Low | Low |
Typical max pressure | ~40 bar g (forged versions higher) | ~32 bar g | ~65 bar g | ~30-40 bar g | 100+ bar g |
Air venting | Poor | Excellent | Poor unless air-vented | Excellent at start-up | Good |
Sub-cooling of condensate | None (discharges hot) | None | None | Several deg C - useful on tracing | Adjustable, up to tens of deg C |
Superheat tolerance | Excellent (to ~250 deg C) | Poor | Good | Poor | Excellent |
Water hammer resistance | Good | Poor | Excellent | Poor | Excellent |
Freeze resistance | Good | Poor | Poor | Good | Good |
Dirt tolerance | Poor | Moderate | Excellent | Moderate | Good |
Dominant failure mode | Fails open | Fails closed | Fails closed (loses prime) | Fails open | Varies with setting |
Relative size and cost | Smallest / lowest | Largest / highest | Large / medium | Small / low | Small / medium |
Read the "dominant failure mode" row carefully when the trap protects critical equipment: a failed-closed F&T floods a heat exchanger and stops production, while a failed-open disc trap only burns fuel. Some plants deliberately specify inverted bucket traps on lines where waterlogging is intolerable, accepting higher steam-loss risk instead.
Choose the trap by the duty it drains, never by the pipe size it connects to. The application matrix below is the industry-standard starting point, combining Spirax Sarco and TLV selection guidance.
Application | Preferred trap | Why |
|---|---|---|
Steam main drip legs, superheated mains, outdoor tracing on windy sites | Thermodynamic disc | Rugged, tolerates superheat and water hammer, wide pressure range, small; needs ~0.25 bar g minimum and dislikes high back pressure |
High-pressure or dirty service, lines subject to water hammer | Inverted bucket | Handles dirt and hammer, positive shut-off; fit an air vent and expect periodic priming checks |
Modulating process heat exchangers, air heating coils, shell-and-tube heaters | Float & thermostatic | Continuous discharge the moment condensate forms, excellent air venting, works down to zero differential; check stall (Section 7) |
Modulating duty where stall occurs | Pump-trap or F&T plus condensate pump | Steam pressure alone cannot lift stalled condensate out; mechanical pumping replaces the lost differential |
Steam tracing, non-critical | Balanced-pressure thermostatic | Sub-cooling keeps hot condensate in the tracer, saves flash losses; fast air venting at start-up |
Freeze-critical outdoor tracing | Bimetallic thermostatic | Robust, discharges without holding condensate; needs a cooling leg, typically 1-3 m |
Rotating dryers and cylinders, jacketed vessels with batch loads | F&T (with steam-lock release where needed) or inverted bucket | High cold-start loads and air accumulation; size with 3-4x safety factor |
System air venting at start-up and shutdown | Dedicated thermostatic air vents | Air vents open at steam temperature regardless of condensate; a trap is not a substitute at the top of vessels and headers |
Three selection errors dominate field surveys. Putting a disc trap on a modulating heat exchanger: when the control valve throttles, the differential collapses and the disc locks closed - waterlogging follows. Putting an F&T trap on an outdoor main that freezes: the float chamber is full of water and splits. Putting a thermostatic trap on a heat exchanger that must run hot: the trap deliberately holds sub-cooled condensate, which floods the surface and throttles the duty. Each error is invisible on the purchase order and expensive on the plant.
The sizing load is the larger of two numbers, and the one most often skipped is the warm-up load. The running load of a steam main is the steady heat lost through the insulation: for an insulated DN100 main at 10 bar g this is typically of order 0.1 kW per metre, so a 100 m main condenses roughly 15-25 kg/h at steady state. The warm-up load is entirely different: at start-up, every kilogram of steel between ambient and saturation temperature must be heated by condensing steam. Work the arithmetic for DN100 pipe (114.3 mm OD, about 12 kg/m of steel) warmed from 20 to 184 deg C (10 bar g saturation): 12 kg/m x 0.49 kJ/kg K x 164 K is roughly 1,000 kJ per metre. If the main is to be warmed in 15 minutes, that heat must arrive at about 1.1 kW per metre, which at a latent heat of about 2,000 kJ/kg condenses roughly 2 kg/h per metre - so a 100 m main produces about 200 kg/h of condensate during the warm-up, ten times the running load, concentrated in the first minutes.
Warm-up rate is a policy decision, not a property of the pipe: many operators deliberately limit warm-up to control thermal shock and manage the condensate. Whatever the policy, the trap at the bottom of each drip leg must be sized to clear the resulting load, because a trap sized only for the running load cannot clear start-up condensate - it accumulates, and the first slug of water hammer arrives before the main is even hot. Process equipment behaves the same way: a jacketed vessel or air heater going cold to hot condenses several times its steady-state load while the metal mass comes up to temperature. Manufacturer sizing routines therefore always start from the maximum of running and start-up load, and published guidance notes that cold-start load can run 2-3 times the running figure even before the warm-up rate is tightened.
Trap capacity is read from the manufacturer's chart at the true differential pressure, not the supply pressure. Differential pressure equals inlet pressure minus back pressure, and back pressure has three components: the condensate return main pressure, static lift after the trap, and line resistance. The classic field arithmetic: a trap on a 7 bar g supply discharging into a 2 bar g return with a 10 m vertical lift has a differential of about 7 - 2 - 1 = 4 bar, not 7 bar. Sizing at 7 bar on the capacity chart undersizes the trap badly, because chart capacity roughly halves between those two differentials. Allow about 1 bar of differential per 10 m of static lift, and remember that a lift immediately after the trap on a modulating application is exactly the geometry that produces stall (Section 7).
On top of the load, apply a safety factor. No international standard mandates the multiplier - it is manufacturer and engineering practice, with published values across major OEMs spanning roughly 1.5 to 5.0. TLV's published minimums for its own products run from 1.5 for float traps to 3-5 for bimetallic types; Watson McDaniel's selector uses 2.0 as the baseline for insulated main drip legs at 50-150 psig and 3.0 for uninsulated mains in cold ambients; Nicholson's guidance applies 4.0 to heavy rotating equipment such as large jacketed vessels. Typical published practice:
Application | Typical safety factor |
|---|---|
Insulated steam main drip legs, running load | 2x |
Uninsulated or cold-ambient mains (warm-up dominates) | 3x |
Heat exchangers under modulating control | 2-3x |
Jacketed vessels, batch and process heating | 2-3x |
Steam tracing | 2x |
Unit heaters and air heating coils | 2-3x |
Rotating cylinders, dryers, heavy rotating equipment | 3-4x |
Instrumented, well-traced drip legs (documented load) | 1.5x |
Do not inflate the factor as insurance. A capacity ratio (selected capacity divided by required capacity) of about 1.15-1.50 is the engineering sweet spot: below 1.00 the trap is undersized and waterlogs the duty; above about 1.5 on intermittent types (disc, inverted bucket), gross oversizing shortens open cycles, accelerates seat wear, and wastes steam on every blast. Also verify two independent limits: the trap body's maximum operating pressure (PMO) must cover the system design pressure and temperature - a Class 600-and-above selection typically means a forged steel body - and the available differential must exceed the trap's minimum operating differential, typically 0.25-0.5 bar for thermostatic types and around 0.25-0.3 bar for a disc trap. A disc trap with 0.3 bar of back pressure against a low inlet pressure simply locks closed; this cold-loop failure never appears in a sizing spreadsheet unless the minimum differential is checked explicitly. Note the trap itself is sized on capacity, not Cv; the Kv arithmetic of the control valve upstream of the heater is a separate calculation covered in the valve Cv/Kv sizing guide.
Stall is the single most misunderstood failure in steam trap selection. Consider an air heater supplied through a temperature control valve, running at full load with 4 bar g in the steam space (saturation temperature 152 deg C), and returning condensate to a 1 bar g main with a 5 m lift behind the trap - a total back pressure of 1.5 bar g (saturation 127 deg C). As the control valve throttles to meet falling demand, the steam space pressure and temperature fall. Using the standard air-heater approximation, the load fraction at which the steam temperature reaches the back-pressure saturation point is (127 - 20) / (152 - 20) = 0.81: whenever the heater runs below roughly 81 percent of full load, the pressure behind the condensate is no longer sufficient to push it up the lift against the 1.5 bar g return, and condensate stalls in the heater. The trap is healthy, correctly sized, and behaving exactly as designed - the installation has simply run out of differential.
Stall is therefore a system check, not a trap check: compute the steam space pressure at minimum load and compare it with return pressure plus lift. If the condensate stalls, no trap family fixes it; the cures are structural. A pump-trap (a float trap with an integral condensate pumping section using steam or air pressure) or a separate electric condensate pump replaces the lost differential mechanically; alternatively, drain by gravity to a receiver below the equipment and pump from there. Specify the stall check on every modulating application with any lift or pressurized return - it is a half-page calculation that prevents chronic waterlogging, cycling temperatures, and premature tube or coil failure. The pressure-side logic is the same discipline applied to steam pressure reducing valve station design: the downstream pressure at minimum load, not the nameplate pressure, decides whether the station works.
Condensate leaving a trap at high pressure contains far more energy than the return main can hold as liquid, and it flashes. Condensate at 7 bar g flashing to atmospheric releases roughly 13 percent of its mass as flash steam, and that vapour occupies hundreds of times the volume of the water. Two consequences belong in the specification. First, size the condensate return line for the flash steam volume, not the water volume: an undersized return line builds back pressure, the back pressure cuts trap capacity, and the equipment floods - a chain of cause that looks, on the plant floor, exactly like a failed trap. Second, never put a lift immediately after a trap on a modulating application (Section 7), and remember that every metre of lift is back pressure paid for out of the trap's working differential.
Flash steam is not only a nuisance; it is recoverable energy. A flash vessel on a high-pressure condensate header can feed low-pressure steam to a deaerator or tracing system, and the sizing logic mirrors the trap logic: separate the phases by density, size the vessel for the vapour volume at the flash pressure, and vent incondensables. Where flash steam is not recovered, at least route the trap discharge into a properly sized, self-draining return line falling in the direction of flow, so the trap's full differential is spent on moving condensate, not on pushing a flooded line.
Most "trap failures" traced in surveys are installation faults, and four of them recur on every plant. First, the strainer: pipe scale, weld slag and rust score trap seats and are the leading cause of failed-open disc and bucket traps. Fit a Y-type strainer with blowdown valve immediately upstream of every trap station - a Y-type strainer with the screen pocket mounted horizontally, not pointing down, so it does not fill with condensate and defeat itself. Second, the drip leg: on horizontal mains the pocket should be the same diameter as the main up to DN100 and at least half the main diameter above that, with a minimum depth of about 1.5 pipe diameters; the trap connects to the side of the pocket above the bottom so dirt settles below the take-off. Spacing follows the load: every 30-50 m on horizontal runs, at every low point, before every rise, before every control valve and isolation valve, and at the end of every main, with the main falling in the direction of flow at about 1:100. Third, group trapping: never drain multiple steam spaces through one trap. If their pressures ever differ, the higher-pressure space blows steam through the trap and blocks the lower-pressure one, which waterlogs - one trap per steam space is the rule, with a common manifold acceptable only downstream of the individual traps. Fourth, two float-trap specific faults: steam locking (trapped steam pockets above the float hold it open or closed depending on geometry - specify a steam-lock release on rotating dryers and similar duties) and air binding on vessels vented only through small connections - specify balanced air venting paths for large vessels that fill with air at shutdown.
Body material follows the piping class rather than the trap function: cast iron dominates low-pressure PN16 drip duty, carbon steel (A216 WCB or A105 forged) covers Class 150-600 saturated steam, and forged steel is expected at Class 600 and above or on superheated mains; stainless 316 bodies are specified for corrosive condensate and sealed maintenance-free designs. Internals are hardened stainless for seat wear. As with every pressure component, the trap station's isolation and check valves are selected on the same basis as any other valve in the line, and the station's water hammer exposure should be assessed with the same physics as the non-slam check valve selection guide.
Traps fail in two directions and the economics differ by an order of magnitude. A failed-closed trap waterlogs the equipment: heat transfer collapses, product quality or output suffers, and slugging water hammer threatens the pipework - but the failure is visible as a cold trap and lost production. A failed-open trap is silent: it passes live steam into the return all day, and nothing downstream complains. Quantify one case. A disc or bucket trap with a 6 mm seat blowing through at 10 bar g upstream is a choked steam orifice; critical-flow arithmetic with a realistic discharge coefficient puts the live steam loss in the order of 35-45 kg/h. Take 40 kg/h over an 8,000-hour year: 320 tonnes of steam - roughly 13,000 USD per year at a delivered steam cost of 40 USD per tonne, from one DN15 component. Multiply by a population and the stakes are clear.
The population stakes are well documented: the US Department of Energy notes that in plants left unmaintained for 3-5 years, 15-30 percent of steam traps may have failed, with the failed fraction typically skewed toward open failures. A 200-trap plant at a 20 percent failure rate and a similar loss profile carries a five-figure annual steam bill for failed traps alone, before counting waterlogged equipment and the return-line back pressure that blowing traps impose on every healthy trap downstream. The corrective economics are equally clear: a trap costs a small fraction of one year's blow-through loss, so survey intervals shorter than two years usually pay for themselves on the failed-open population alone.
Trap testing uses three independent channels - temperature, ultrasound and sight/flow observation - and the misdiagnosis risk lies in trusting one channel alone. Temperature alone misleads in both directions: a trap blowing steam through is hot, so "hot" does not mean healthy; a trap upstream of a hot flooded exchanger reads warm while failed closed. Ultrasound hears velocity, so it discriminates well between the intermittent blast cycle of a working disc trap and the continuous high-velocity hiss of blow-through, but it needs an operator who knows what each pattern sounds like, and it over-reads on a leaky isolation valve upstream. Visual observation at the discharge of an open-ended trap is definitive but rarely available on closed return systems.
Three specific misdiagnoses recur. First, misreading blow-through on an intermittent trap: a disc trap in normal operation blasts periodically at full velocity; between blasts it is silent. An operator unfamiliar with the cycle reports the blast as a failure - or worse, reports the silence between blasts as a failed-closed trap. Second, blaming the trap for symptoms caused elsewhere: a heat exchanger that will not hold temperature may have an undersized control valve, fouled tubes, a stalled condensate condition (Section 7) or air in the steam space, none of which any trap replacement will fix. Third, testing every trap identically regardless of duty: a tracing trap discharging sub-cooled condensate is cold by design, and flagging it as failed-closed generates wasted work. A competent survey programme therefore assigns each trap an expected signature - discharge pattern, surface temperature band, ultrasonic pattern - at selection time, and tests against that signature. Population surveys at least annually, with critical-duty traps on a shorter cycle, are the published norm.
The table below maps the symptoms an operator sees to the root cause and the field verification before any trap is replaced. Work top to bottom: replacing a healthy trap because of an upstream fault is the most expensive form of maintenance.
Symptom | Likely root cause | Field verification | Corrective action |
|---|---|---|---|
Continuous ultrasonic hiss at trap; return line hot at discharge | Failed open (blow-through) - worn seat, dirt on seat, lost bucket prime | Upstream/downstream temperature differential; listen across isolation valves to exclude seat leakage | Replace internals; clean strainer; check upstream strainer condition |
Trap cold; equipment flooded; temperature cycling | Failed closed - debris, mechanical damage, frozen float chamber | Confirm condensate at trap inlet; open strainer blowdown | Clear or replace; add freeze protection if frozen |
Heat exchanger output falls at reduced load; trap found healthy | Stall - differential collapses below back pressure at part load | Compute steam space pressure at minimum load vs return pressure + lift (Section 7) | Pump-trap or condensate pump; remove lift after trap |
Disc trap cycles too often or blows continuously on light load | Low differential or cold/wet disc; oversizing shortens cycles | Check true differential and back pressure; insulate cap in exposed sites | Insulating cover; correct sizing at actual differential |
Inverted bucket loses prime intermittently | Pressure fluctuation or steam blow-through between spaces sharing the trap | Confirm one trap per steam space; check for group trapping | Separate traps per space; fit lock-release or check valve where dictated |
Heater slow at start-up; air venting audible problem | Air binding - trap or system not venting non-condensables | CO2/air test at vents; check trap air-handling rating | Dedicated thermostatic air vents; F&T or balanced-pressure trap |
Tracing trap cold but tracer line warm and functional | Normal sub-cooling behaviour of thermostatic trap | Confirm expected signature for the trap family | None - update survey records, do not replace |
Write the selection into the purchase document, not just the size. The checklist below turns the preceding sections into specification language a supplier can price and a receiver can verify.
# | Specification item |
|---|---|
1 | State the duty explicitly: equipment drained, running load and start-up load in kg/h, warm-up policy, modulating or constant pressure |
2 | State true differential pressure: inlet pressure at minimum load minus return main pressure minus static lift, not the supply header nameplate |
3 | State the trap family per the application matrix, and the stall check result for modulating duties (pump-trap requirement, if any) |
4 | State the safety factor applied and the resulting required capacity, so the capacity ratio can be audited (target 1.15-1.50) |
5 | Require manufacturer capacity charts at the stated differential, and a PMO rating covering design pressure and temperature |
6 | State body and internals materials per piping class (WCB/A105, forged above Class 600, 316 for corrosive condensate), connections per ASME B16.5/B16.11 or EN 1092-1 |
7 | Require classification and testing traceability per ISO 6704 classification, ISO 6948 production and performance characterisation tests, and the ANSI/FCI definition of trap function; capacity data per the manufacturer's published test method |
8 | Specify upstream Y-type strainer with blowdown valve, screen pocket horizontal, and a drip leg detail drawing for main drain points |
9 | Specify one trap per steam space; prohibit group trapping unless each space is proven to share identical pressure at all loads |
10 | Specify steam-lock release for rotating dryer duties, dedicated thermostatic air vents for vessel air venting, and the expected operating signature per trap for the survey programme |
Why does a steam trap blow through, and what does it cost?
A trap fails open when dirt or wear stops its seat from closing, or an inverted bucket loses prime, and it passes live steam into the condensate return continuously - a failure that is silent because nothing downstream complains. The loss is large: a DN15 trap with a 6 mm seat blowing through at 10 bar g is a choked steam orifice passing roughly 35-45 kg/h of live steam. Over an 8,000-hour year that is about 320 tonnes of steam, roughly 13,000 USD at 40 USD per tonne - from one small component. Published survey data shows 15-30 percent of trap populations fail after 3-5 years without testing, so the population economics dominate any single repair.
What safety factor should I apply when sizing a steam trap?
No standard mandates the multiplier; published manufacturer practice spans roughly 1.5 to 5.0. Typical values: 2x for insulated main drip legs at running load, 3x for uninsulated or cold-ambient mains where warm-up dominates, 2-3x for modulating heat exchangers and air coils, 3-4x for rotating dryers and heavy rotating equipment, 1.5x only for instrumented, well-traced drip legs with a documented load. Aim for a capacity ratio of 1.15-1.50: below 1.0 the trap waterlogs the duty, while gross oversizing above 1.5 on intermittent types shortens open cycles, wears seats and wastes steam on every blast.
Why does my heat exchanger flood at low load even though the trap tests healthy?
Because the condensate has stalled, and the trap is behaving exactly as designed. On a modulating application, throttling the control valve lowers the steam space pressure; when it falls to the return pressure plus any static lift, no differential remains to push condensate out. Example: an air heater at 152 deg C steam temperature (4 bar g) with a 1.5 bar g back pressure stalls whenever load drops below (127-20)/(152-20) = 81 percent of full load. No trap family fixes this - install a pump-trap or condensate pump, or drain by gravity to a receiver below the equipment.
Can I drain several pieces of equipment through one steam trap?
No. If two steam spaces sharing one trap ever operate at different pressures, the higher-pressure space blows steam through the trap and blocks the lower-pressure space, which waterlogs. Start-up and shutdown timing differences alone are enough to create the condition. The rule is one trap per steam space; a common manifold is acceptable only downstream of the individual traps. Where a shared trap already exists and one unit runs cold while another blows steam, this is the first thing to check before replacing anything.
How often should steam traps be tested, and using what method?
Published practice surveys the whole trap population at least annually, with critical-duty traps on a shorter cycle, because 15-30 percent of an unmaintained population may be failed after 3-5 years. Testing combines three channels - surface temperature, ultrasound and visual/flow observation - and each trap should be tested against the expected signature recorded at selection time (discharge pattern, temperature band, ultrasonic pattern). Temperature alone misleads: a blow-through trap reads hot, and a thermostatic tracing trap is cold by design. The most expensive error is not a missed failure but replacing a healthy trap whose symptom actually comes from stall, air binding or an upstream fault.
