How to Put a Number on What a Trap Population Is Wasting

Why this matters

The number a trap survey produces is the number the owner decides on, and it is almost always built the same wrong way: add up the published capacity of every failed-open trap, call that the loss, and divide by nothing. That figure has no denominator, so it cannot be checked or argued with, and it is routinely two to three times the truth. Once an owner has spent against it and measured the result, the next survey there has to overcome the last one.

The fix is not a better trap-side estimate. It is that the plant already holds an independent measurement of the same loss, and the two numbers cap each other. This procedure runs the denominator first, because a numerator with no denominator is not a number, and runs the independent check before anything is reported, because that check is what turns an estimate into a bound you can defend.

The hazards this procedure creates

The survey walk itself, with its asbestos, confined space, noise and burn exposures, belongs to the sibling procedure on surveying a trap population and is not repeated here. Two hazards belong to the steps below specifically.

Step 4 sends you to look at a condensate receiver vent. A vented receiver discharges saturated steam and hot droplets, and the plume is invisible for the first stretch out of the pipe. Observe it from beside and upwind at ground level. Never from beneath it, never from a ladder in line with it, and never with a hand or a face in the plume to judge how much is coming out.

Any step that needs a gauge fitted where one does not exist is a separate scheduled job: isolate supply and return, lock and tag under 29 CFR 1910.147 in general industry or the construction energy control requirements of 29 CFR Part 1926 on a site, drain to a point piped away from any walkway, confirm zero on a gauge you have just watched read something else, and let the metal fall below 120 F verified non-contact at arm's length before a wrench goes on it.

Step 1: Decide which loss you are claiming, before you measure anything

Skipping this makes the rest incoherent, because there are two different losses here and they are two different sizes.

• Mass leaving the plant. Water that has to be replaced, treated and heated from cold. It shows up in the make-up meter.
• Latent heat delivered nowhere. Steam whose heat of vaporisation went into a return line instead of into a building or a process. It does not necessarily leave the plant at all.

A failed-open trap discharging into a return that terminates at a vented receiver sends live steam into a pipe where it partly condenses. Whatever still arrives above 1,150 Btu/lb leaves as flash out the vent, so the mass is retained only to the extent the return line condensed it on the way. That is what step 4 measures. The latent heat is entirely wasted. The latent number is the larger; the mass number is the one with an instrument behind it. Calling either "the loss" without saying which is where most survey reports go wrong.

Three steam-table values carry the arithmetic, read at your own pressures rather than remembered. Latent heat is about 881 Btu/lb at 100 psig and about 970 at atmospheric, which is why a pound flashed at a vented receiver costs more heat than a pound made at header pressure. Liquid water carries about 1 Btu per pound per degree F here.

Step 2: Establish the denominator, from a meter rather than from a nameplate

The denominator is steam generated over the same window the survey covers, and the boiler nameplate is not it. A boiler rated at 10,000 lb/hr running at part load all year overstates generation by whatever its load factor is, and using the nameplate deflates every percentage you are about to compute.

Take the totalised feedwater meter over a full year and divide by the hours. Feedwater is not generation: it is generation plus blowdown. Subtract blowdown as its measured or estimated fraction of feedwater, and say which of those two it is, because an estimated blowdown fraction is a systematic term that does not average away. With no feedwater meter, fuel records over a season plus the boiler's steady-state efficiency give a serviceable denominator carrying the efficiency assumption's error too.

Step 3: Build the numerator as a bound, not as a value

For each failed-open trap, take the manufacturer's published capacity for that body and orifice at that trap's measured differential, and sum them. The per-trap ranking measure and the choked-flow condition it rests on are owned by the survey sibling and are not re-derived here.

That sum is an upper bound and is written as a bound, never as a value with a plus-or-minus attached. A worn seat is not an open orifice: it is partly obstructed by scale and by whatever wore it, and nobody has measured the actual open area. Assuming full published capacity is one assumption applied once per failed trap, so it biases the total in one direction and stays at full strength no matter how many traps are in the sum. That is what makes it a bound rather than a spread.

Step 4: Cap the bound with a whole-plant measurement that knows nothing about traps

Make-up water measures mass leaving the plant from every cause at once. Subtract blowdown, the deliberate loss quantified in step 2, and any steam the plant consumes rather than returning: humidification, direct injection, a process that uses the steam itself. What is left is every unintended route together, meaning atmospheric leaks, vents, deaerator venting and receiver flash.

Then convert your trap-side bound into the flash it would have to produce, and compare. Live steam entering a vented return heats the returning condensate, and a vented receiver cannot hold water above about 212 F at atmospheric pressure, so every Btu beyond what it takes to reach 212 F leaves as flash out the vent. If the trap-side bound implies far more flash than the make-up meter can account for, that bound is wrong, in exactly the direction its own systematic assumption predicts.

Step 5: Report the smaller bound, and name the assumption behind the larger one

Report both, say which governs, and say why. "At most X percent of generation, capped by the make-up balance rather than by the trap capacities" is a sentence an owner can take to anyone for a second opinion. "Y percent" on its own is not.

Do not average the two. They are not two estimates of one quantity with independent errors; they are two ceilings, and the lower ceiling is the answer.

Worked example: 84 traps, two bounds, and the one that governs

A single-boiler plant on a 100 psig header. All steam goes to heating coils and exchangers and every condensate line returns to one vented receiver. Nothing is injected or deliberately consumed, which is what makes the mass check clean here.

The denominator, first. Totalised feedwater over the year, divided by hours, averages 6,316 lb/hr. Blowdown is metered at 5 percent of feedwater, which is 316 lb/hr. Generation is therefore 6,000 lb/hr, and that is the number every percentage below is taken against.

The numerator as a bound. The survey covers 84 traps and returns 11 failed open. Summing each one's published capacity at its own measured differential gives 1,050 lb/hr. Against 6,000 lb/hr of generation that is 17.5 percent, and 17.5 percent is where most reports stop.

The independent measurement. The make-up meter averages 620 lb/hr over the same year. Feedwater minus make-up gives 5,696 lb/hr of returned condensate, a 90 percent return rate against feedwater or 95 against generation, two bases 5 percent apart, so print the one you used. Either way it is healthy, and that is itself a clue that this plant is not hemorrhaging mass. Of the 620 lb/hr, 316 lb/hr is the blowdown already accounted for, leaving 304 lb/hr as the ceiling on mass leaving by every unintended route combined.

Now convert the trap-side bound into flash and see whether it fits. Do not run this on latent heat alone. A pound of live steam arrives at the receiver carrying its full 1,190 Btu, and 1,190 is above the 1,150 Btu/lb of saturated vapour at 212 F, so it cannot exist there as liquid at all. Run the whole receiver as one balance instead. The return carries 5,696 lb/hr of condensate arriving at roughly 200 F, which is 168 Btu/lb; bringing all of it to 212 F, or 180 Btu/lb, absorbs 5,696 times 12, which is 68,400 Btu/hr. Each pound of live steam falls from 1,190 to the 180 Btu/lb of 212 F liquid, releasing 1,010 Btu, so the trap side puts 1,060,500 Btu/hr in. Less the return's 68,400, the remaining 992,100 Btu/hr has nowhere to go but flash at about 970 Btu/lb, which is 1,023 lb/hr out of the vent.

That number cannot be true. The make-up meter says at most 304 lb/hr leaves the plant by all unintended routes together. A steady 1,023 lb/hr of flash is more than three times that entire budget, before allowing anything for atmospheric leaks or deaerator venting, and a receiver shedding it would be heavily and continuously plumed. Confirm that with your own eyes from beside and upwind of the vent.

So run the check backwards and let the plant set the ceiling. Allow the whole 304 lb/hr to be trap-driven flash, which is generous because some of it is not. Flashing 304 lb/hr absorbs 304 times 970, or 294,900 Btu/hr, and bringing the return to 212 F absorbs the same 68,400 Btu/hr as before, for 363,300 Btu/hr into the return. At the 1,010 Btu/lb each pound of live steam releases on its way down to 212 F liquid, that is about 360 lb/hr through the traps, which is 6.0 percent of the 6,000 lb/hr denominator. Running it on latent heat alone returns 412 lb/hr and 6.9 percent, wrong in the flattering direction: a procedure that exists to stop a survey overstating a loss cannot afford to overstate its own.

Read the two bounds against each other. The trap side says at most 17.5 percent, the plant side at most 6.0 percent, and the plant side governs because it is measured rather than assumed. The gap between them is a factor of about 2.9, which is the direction and roughly the magnitude the published-capacity assumption predicts once eleven worn seats have each been credited with a clean full orifice.

What gets reported. Trap losses on this plant are at most about 7 percent of steam generated, capped by the make-up balance. The trap-side sum of 17.5 percent is stated alongside it as what the published capacities would allow, noted as unsupported by the plant's own water balance. Both numbers go in the report, in that order, with the denominator printed next to them.

The failure mode if step 4 is skipped. The 17.5 percent goes in the proposal, the traps get replaced, and the measured improvement comes in at roughly 6 percent. Every trap replaced was genuinely failed and every hour of that work was justified, and the shop still spends the next visit explaining why the result landed at about a third of the promise. The work was right and the number was not, and it is the number the owner remembers.

What would change the answer. If this plant vented its condensate to drain, or if the traps discharged to atmosphere, the two checks would collapse into one and the make-up meter would be a direct measurement rather than a cap. The bounds diverge here because the condensate is returned rather than dumped while the receiver itself is vented, so the two checks measure different things. Be careful with the word closed: a closed return system is not a closed volume.

How to verify the number before it goes in front of an owner

• Print the denominator next to every percentage. Generation and feedwater sit 5 percent apart on this plant, and a percentage whose base is not on the same line is the commonest way a correct figure gets repeated wrongly.
• Confirm the survey window and the meter window are the same window, and re-check the deliberately-consumed steam term. A humidifier discharging to atmosphere puts mass through the make-up meter that has nothing to do with traps, and leaving it in inflates the cap until it stops capping.
• Re-read the direction words in your own summary sentence. The bound that governs is the smaller one, and a sentence saying the survey "showed" 17.5 percent when the water balance caps it at 6.0 percent is the defect that survives every arithmetic check in this procedure.

References

• ASME Steam Tables, or an equivalent saturated-steam table, for latent heat and liquid enthalpy at the pressures you actually measure
• Trap manufacturer capacity data for the body and orifice at the measured differential, which owns the trap-side bound
• 29 CFR 1910.147 in general industry, and the energy control requirements of 29 CFR Part 1926 on a construction site, for any gauge connection opened as part of this work
• See related: How to Survey a Trap Population and Rank What to Fix; Why Condensate Is the Expensive Half of a Steam System; What a Flash Tank Is Recovering