Why Pressure Sets Temperature and What That Decides

Why this matters

Somebody is going to ask you to make a steam coil run cooler, and the only lever that does it is the pressure inside the coil. That is not a design preference, it is the physics of a saturated fluid, and it means a modulating steam valve is a pressure controller wearing a temperature controller's label. It also means that below a certain load the coil's own pressure falls under the pressure in the return line, the condensate stops leaving, and the coil floods with the valve still calling for heat. That failure has split more outside-air coils than every frozen pipe in the building put together, and it is fully predictable from a steam table and two pressures.

Safety before you go near a modulating coil

A flooded steam coil holds hot condensate at a pressure you cannot infer from the supply gauge, because the coil may be at partial vacuum on one side of the valve and at line pressure on the other. Isolate the supply valve and the return, lock and tag both under 29 CFR 1910.147, then break the vacuum and drain from a valve you stand beside rather than in front of, with face protection under 29 CFR 1910.133 and the discharge routed to a floor drain or a rated container, not onto the floor at your feet. Let it cool to a temperature you can read on a probe before any joint is opened.

If the coil is on an outside-air stream and the space is below freezing, treat a flooded coil as a live burst risk: shut the outside air damper and stop the fan before you begin, because a fan pulling freezing air over stagnant tubes will split them while you work.

The relationship, and the part everyone misses

Where liquid water and its own vapour share a space, one pressure means exactly one temperature. Around 0 psig that pair is 212 F. Around 15 psig it is roughly 250 F, and around 100 psig roughly 338 F. Read the exact pairs from a steam table; the point here is the shape of the curve, not the values.

The shape is what people miss. The curve is steep at the bottom and flat at the top. Going from 0 to 5 psig moves saturation temperature by roughly 15 F, about 3 F per psi. Going from 100 to 105 psig moves it by roughly 3 F, closer to 0.7 F per psi. Same 5 psi of pressure change, more than four times the temperature effect at the low end.

Three decisions fall straight out of that, and none of them are optional:

• A coil fed at 5 psig cannot deliver 250 F, ever. Valve size, coil size and fan speed are irrelevant. The ceiling is the saturation temperature at the pressure available, and the only way to raise it is to raise the pressure.
• Low-pressure steam control is twitchy in temperature terms and high-pressure steam control is sluggish, because the same increment of valve travel buys several times more temperature change down at the bottom of the curve. A loop tuned on a 100 psig header and then reused on a 5 psig header will hunt.
• Distribute high, use low. A pound of steam at 100 psig occupies roughly a seventh the volume of a pound at 0 psig, so the main is much smaller for the same heat. A pressure reducing valve at the load then buys back the lower temperature you actually wanted. This is the standard arrangement and it exists entirely because of the two ends of this one curve.

The gate that decides whether a modulating coil works

State it plainly, because everything below is this one rule run twice: a steam trap moves condensate on differential pressure only. It has no pump in it. If the pressure on the coil side falls to or below the total back pressure on the return side, condensate stops leaving the coil regardless of what the trap is doing. That condition is called stall.

Now watch what happens to coil pressure as the valve throttles. At constant airflow, a coil's output tracks the difference between its saturation temperature and the mean air temperature passing over it, near enough for this purpose. Cut the required output and the valve throttles, coil pressure falls, saturation temperature falls with it, and the driving difference shrinks to match. On a deep turndown the coil pressure goes below atmospheric and the coil sits in a partial vacuum, which is normal and not itself a fault.

The gate is whether the coil's falling pressure crosses the return-side back pressure before it reaches the bottom of the required load range. Back pressure is the sum of everything downstream of the trap: static lift, friction, and any pressure held in the return main.

Case one: gravity drain to a vented receiver

A face-and-bypass air handler coil, entering air 40 F, leaving air 100 F at design, fed through a modulating valve from a 10 psig header. Saturation temperature at 10 psig is about 240 F. Mean air temperature at design is 70 F, so the design driving difference is 240 minus 70, or 170 F.

The trap discharges downward into a vented receiver on the floor below, with the receiver open to atmosphere and a vacuum breaker fitted on the coil. Back pressure at the trap is effectively 0 psig, and saturation at 0 psig is 212 F.

Find the load at which the coil pressure reaches atmospheric. Let f be the fraction of design output. Leaving air is 40 plus 60f, so mean air is 40 plus 30f, and the available difference at 0 psig is 212 minus that, or 172 minus 30f. Set that equal to the required difference, 170f:

170f = 172 - 30f, so 200f = 172, so f = 0.86.

The coil goes sub-atmospheric below about 86 percent of design load, which on a heating coil is nearly all of the operating year. That is fine here, and only because of the vacuum breaker and the downward drain. The vacuum breaker admits atmospheric air as the coil pressure falls, which stops a vacuum from holding condensate up in the tubes, and the condensate then falls out of the coil under its own static head into a receiver that is not fighting it. The price you pay is real and worth naming: the air the breaker admits is a non-condensable gas that will blanket tube surface until the next high-fire cycle sweeps it out, so this coil will always be slightly lazy on the first minutes of a call.

Case two: same coil, overhead return

Now change one thing. The receiver is on the roof, the trap discharges up into a return main that sits at 5 psig, and nobody fitted a vacuum breaker because the coil drains uphill anyway. Saturation at 5 psig is about 227 F, so the available difference is 227 minus mean air, or 187 minus 30f. Run the same rule:

170f = 187 - 30f, so 200f = 187, so f = 0.935.

This coil stalls below about 94 percent of design load. It drains for the few hours a year the building sits at design condition and floods for every other hour. What the operator sees is a coil that works on the coldest morning of the year and swings wildly on a mild afternoon, which reads backwards to anyone expecting things to fail when they are working hardest. The mechanism says the failure lives at the low-load end, and the symptom has to match: at high load this coil is fine, and at low load it hunts, because at low load and only at low load the coil pressure is under the return pressure.

Then it floods, output collapses toward zero, the controller drives the valve wide open, full pressure blows the accumulated slug out through the trap, output spikes past setpoint, and the cycle repeats every few minutes. Put that coil on outside air in January and the flooded tubes freeze and split between two of those cycles.

The fix is not a bigger trap, and it is not a bigger valve. Neither one can raise the pressure inside the coil. The fix is either to remove the back pressure (drain by gravity to a vented point below the coil, add a vacuum breaker), or to add something that actually makes differential pressure - a pump-trap or a condensate pump under the coil. A third answer, and often the cheapest, is to stop modulating: two-position control holds full header pressure in the coil whenever it is on and never stalls, at the cost of a wider temperature swing in the air stream.

What would change the call

Raise the header to 25 psig and re-run the same arithmetic and the picture improves, because you gain margin above the return pressure across more of the turndown. Lower the entering air temperature and the required difference rises at every load, which helps. Cut the airflow and the mean air temperature climbs, which hurts. Add a lift after the trap and you have added back pressure equal to the static head of that lift, which on a two-storey rise is not a rounding error. The rule does not change, only the numbers you feed it.

Verifying the diagnosis on site

Do not trust the header gauge to tell you what the coil sees. Read a gauge on the coil side of the control valve, between the valve and the coil, plus a second gauge or a reliable reading of the return main. If a gauge has to be added, that is not a live-line job: isolate, lock and tag under 29 CFR 1910.147, vent to atmosphere and confirm zero gauge pressure before opening the fitting. Watch both as the load falls.

Before you call a 2 psi margin real, give the reading its basis and its character. Most dial gauges are specified as a percentage of full scale, which means a 1 percent of full scale gauge on a 0 to 60 psi dial is worth about 0.6 psi anywhere on the face, whether it is reading 50 psig or 3 psig. That is a fixed systematic offset for that instrument and it does not average away. If you are comparing coil pressure and return pressure using two different gauges, those offsets are independent, so they combine in quadrature and two similar terms give a combined figure about 1.4 times one of them - roughly 0.85 psi here, not 1.2. A margin smaller than that is not a margin.

The confirming observation costs nothing: on a coil you suspect is stalling, watch the discharge side of the trap through a sight glass or listen at it with an ultrasonic probe held against the body, never with a hand on the pipe. A stalled coil's trap is silent and cold on the outlet while the valve is calling, and then passes a violent slug when the valve goes wide open. A healthy coil's trap cycles at a rate that rises and falls smoothly with load.

References

• ASME Steam Tables (IAPWS industrial formulation) for the saturation pressure and temperature pairs used above
• 29 CFR 1910.147, control of hazardous energy, for isolating pressure and thermal energy before opening a coil; 29 CFR 1910.133 for eye and face protection
• Steam trap and pump-trap manufacturer sizing literature for capacity at a stated differential pressure, which is the only place the actual capacity of a given trap lives
• See related: What Makes Steam Different From Every Other Working Fluid; How to Read a Steam Table Without Memorising It; What a Steam Trap Is Actually Deciding