How to Size a Trap From the Load Rather Than the Pipe
Why this matters
The single most common steam trap selection in the field is "whatever size the pipe is." It is fast, it is free, and it is wrong in both directions at once: on drip service it typically lands an order of magnitude oversized, and on a modulated heat exchanger it can land undersized at exactly the condition that matters. A trap is a capacity, and a capacity is a load and a pressure difference. The connection size falls out at the end; it is never an input.
What follows is a selection sheet, filled in one field at a time, with the reasoning for each field and what a wrong entry does downstream.
Before any of it: condensate is not a benign fluid. At 15 psig it is roughly 250 F, and the moment a line is opened it flashes to steam and throws hot water with it. No trap, strainer, drain leg, or return line in this procedure is opened while it is live. Isolate at both ends and lock and tag both isolations under the written energy control procedure required by 29 CFR 1910.147 for stored pressure energy. Then drain through the drip leg to a safe discharge point, vent to atmosphere, and confirm zero on a gauge you have just watched read something else. Zero pressure is not zero energy: let the metal fall below 120 F, verified with a non-contact reading taken at arm's length rather than assumed from the clock, before a wrench goes on it. Everything measured below is measured from existing gauges, a tape measure, and a nameplate.
Field one: the running condensate load
Condensate load is the equipment's heat duty divided by the latent heat of the steam at the pressure the equipment operates at.
Both terms carry conditions. The duty is the design duty from the equipment's own rating sheet, at the conditions the equipment actually runs at, not a nameplate number from a different entering temperature. The latent heat comes from the steam tables at the operating pressure, and it moves: about 952 Btu per pound at 10 psig, about 945 at 15 psig, about 912 at 50 psig, about 881 at 100 psig. Using the wrong pressure's latent heat is a modest error, a few percent, and it is not where sizing goes wrong. It is worth doing right anyway because it costs nothing.
If the load is not a heat exchanger but a length of pipe, the load is the pipe's heat loss, which depends on size, insulation, and ambient, and the insulation manufacturer's heat loss tables are the authority for it rather than any rule of thumb.
Field two: the load that governs
Running load is not the largest load. Starting from cold, every pound of metal in the equipment and its piping has to be raised from ambient to saturation temperature, and every Btu of that comes out of condensing steam. That startup load is larger than the running load and it lasts until the metal is hot.
Two ways to handle it, and you pick one, not both:
- Where the equipment manufacturer publishes a startup or warm-up condensate load, that is a load. Use it and apply a safety factor of about 1.5.
- Where they do not, use the running load and apply a safety factor of 2 to 3, which is absorbing the startup surge and the uncertainty in the duty together.
Do not multiply a published startup load by a startup factor. That is the double-count that produces the wildly oversized traps a sibling article covers, and the shop that did it will defend the number because each step looked conservative on its own.
There is a third path that changes the answer entirely: if the startup can be slowed, either with a warm-up bypass or with a control that ramps the valve open over several minutes, the startup load stops governing and the trap is selected on running load. Slowing the startup is nearly always the better engineering answer, because it also removes the water hammer risk that a fast cold start creates.
Field three: the actual differential pressure
Here is where sizing genuinely fails, and it fails quietly because nobody writes the number down.
A trap moves condensate because the pressure at its inlet is higher than the pressure at its outlet. The differential is:
inlet pressure, minus return main pressure, minus static lift, minus return line friction.
The term everyone forgets is static lift. If the trap discharges upward into a return main above it, every foot of that rise is back pressure the trap has to push against. Water column converts at roughly 2.31 feet per psi, so an 8 foot lift is about 3.5 psi of back pressure that no gauge in the room displays.
Return main pressure is read from a gauge on the return main, not assumed to be zero. A return system with other equipment discharging into it, or with a receiver under pressure, sits well above atmospheric. Where no gauge exists on the return, one is added with the return isolated, drained, vented, confirmed at zero, and locked and tagged under 1910.147, because a return main is full of water hot enough to flash the instant it is opened.
For the capacity look-up itself, the relationship is that flow through a fixed orifice on a liquid rises with the square root of the differential, so cutting the differential to a quarter cuts capacity in half. That relationship is derived for incompressible flow through an orifice, and real trap condensate is at saturation and flashes as it crosses the seat, which chokes the flow and makes the trap deviate from it. Use the square root as a sanity check on your own arithmetic. Use the trap manufacturer's published capacity curve as the selection authority. Those two sentences settle which of the three possible sources owns the answer.
Field four: the differential at minimum load, and the gate before capacity
The differential is not one number. On any equipment with a modulating control valve, the pressure inside the equipment falls as the load falls, because a lower load needs a lower saturation temperature. Back pressure does not fall with it. So the differential shrinks as the load shrinks, and at some load it goes to zero and then negative.
This gate comes before any capacity calculation: confirm the differential stays positive across the whole load range the equipment will see. If it does not, the answer is not a larger trap, because no trap of any capacity passes condensate against a negative differential. The answer is a pump-trap, or a vented receiver with a condensate pump, or a control arrangement that keeps pressure up. A sibling article covers what happens to a coil when this gate is ignored, and it is the most expensive failure in this whole subject.
Field five: the trap type, which is a separate axis from capacity
Capacity and type are two independent selections and the type follows the service, not the size.
| Service | Type usually selected | Why |
|---|---|---|
| Modulated heat exchanger or coil, load and pressure both varying | Float and thermostatic | Drains continuously at any load, and the thermostatic element vents air on startup |
| Drip leg on a high pressure main | Inverted bucket | Tolerates high differential and dirt, handles small continuous loads |
| Outdoor drip, high differential, freeze exposure | Thermodynamic disc | Small, drains itself, no water to freeze in the body |
| Tracing, unit heaters, small loads where subcooling is acceptable | Thermostatic | Simple, holds condensate back until it subcools, which is a fault on a coil and a feature on tracing |
The last row carries the trap in its own caveat, which is how these should be read: a type that holds condensate until it cools is fine on a tracing line and is exactly what you must not put on a heat exchanger, because held condensate is surface removed from service.
The sheet, filled in
Shell and tube converter heating building water. No manufacturer startup load available. Gravity return, vented receiver below the trap.
| Field | Entry | Where it came from |
|---|---|---|
| Duty | 800,000 Btu per hour | 40 gpm from 140 F to 180 F, at 500 times gpm times temperature rise; the 500 is 8.33 pounds per gallon times 60 minutes times a specific heat of 1.0, valid for water and not for a glycol loop |
| Steam pressure at the shell | 15 psig | Downstream gauge at the reducing station, confirmed at the converter |
| Latent heat | about 945 Btu per pound | Steam tables at 15 psig |
| Running load | about 847 lb per hour | 800,000 divided by 945 |
| Governing load basis | running load, factor 2.5 | No manufacturer startup figure, so one factor absorbs startup and uncertainty; the 1.5 path is not used |
| Required capacity | about 2,120 lb per hour | 847 times 2.5 |
| Return main pressure | 0 psig | Gauge on the vented receiver |
| Static lift | none | Receiver is below the trap |
| Return friction | about 1 psi | Return line length and fittings |
| Differential at full load | 14 psid | 15 minus 0 minus 0 minus 1 |
| Load at minimum turndown | about 254 lb per hour | 30 percent of running load, from the building's own load profile |
| Shell pressure at that load | 2 psig | Converter capacity curve, read at 30 percent |
| Differential at minimum | 1 psid | 2 minus 0 minus 0 minus 1, positive, so the gate passes |
| Selection point | 2,120 lb per hour at 14 psid | Manufacturer capacity curve |
| Type | float and thermostatic | Modulated service, varying load and pressure |
| Trap connection | whatever the selected model carries | Output of the selection, not an input |
Now check the second point, because a selection that satisfies full load can still fail at minimum load. Using the square root relationship as the sanity check: the square root of 1 over 14 is about 0.27, so a trap passing about 2,120 pounds per hour at 14 psid passes roughly 570 pounds per hour at 1 psid. Required at that condition is 254 pounds per hour. It clears, with margin, so the full load point governs and the selection stands. Confirm that against the manufacturer's own curve at 1 psid rather than accepting the square root, for the flashing reason given above.
Note what did not appear anywhere in that sheet. The converter's drain connection is 1.5 in. The trap the chart lands on is smaller than that, and that is correct: the drop leg from the converter stays full 1.5 in size all the way down to the trap so condensate can fall freely, and only the trap itself is the smaller number. Full-size drop leg, chart-size trap. Reversing that, running a small drop leg into a large trap, starves the trap and produces the symptoms of undersizing on a correctly sized trap.
Confirming the selection in service
Three checks, all from a distance, none requiring anything to be opened.
Watch a full cycle at low load, not at design. Low load is where selections fail and it is the condition nobody commissions at. On a float and thermostatic trap draining properly the discharge is continuous or nearly so; on any trap, a discharge that has become a rapid open-and-shut at low load is telling you the trap is far larger than the load, which a sibling article covers.
Read the equipment's own temperature at low load. A converter that holds its leaving water temperature at design and drifts down at 30 percent load, with steam pressure available, is waterlogged. That is either an undersized trap at the low-load differential, or the differential gate failing, and Field four tells you which.
Read the condensate line temperature on high-emissivity tape, or with a contact probe on metal that is already bare. Bare steel reads low on an infrared thermometer, sometimes by tens of degrees, and here the bias runs toward missing a blowing trap rather than inventing one, which is the direction that costs you the finding. A line running at or near supply temperature continuously is a trap passing live steam. A line running at condensate temperature with a rise on each discharge is normal. Do not touch the line to judge this; it sits at condensate temperature at best and supply temperature at worst.
Write the sheet into the job record with the differential and the load, not just the trap model. The next tech who replaces this trap will otherwise buy the connection size, and the whole selection dies with the paperwork.
References
- Steam trap manufacturer capacity curves, which are the governing authority for the flow a specific model passes at a specific differential, and which supersede the square root relationship used above as a check
- Equipment manufacturer rating sheets, for the design duty and for any published startup or warm-up condensate load
- Saturated steam tables from any engineering handbook, for latent heat at the operating pressure
- 29 CFR 1910.147, the OSHA general industry energy control standard, for isolating stored pressure energy before opening a trap, drain leg, or return line
- See related: Why an Oversized Trap Fails Differently Than an Undersized One; What a Steam Coil Does and Why It Freezes