What a Phase Change Does to a Temperature Reading

Why this matters

A tech puts a probe on a line, gets exactly the temperature the design drawing calls for, and writes "operating normally" on a system delivering three quarters of its rated output. Nobody made an arithmetic error and the probe was accurate. The reading was simply blind, because in the region where a fluid is boiling or condensing, temperature stops responding to how much heat is moving.

This is not a measurement problem you can solve with a better instrument. It is a property of the fluid. Knowing exactly what that reading has stopped telling you, and what it still tells you with unusual precision, is the difference between diagnosing a steam coil, a DX evaporator, or a condenser and guessing at one.

Before you put a probe anywhere near this

The systems where phase change matters most are the ones that injure fastest, and each route needs its own control.

• Steam and condensate. Saturated steam at atmospheric pressure is already at 212 F and causes a full-thickness burn on contact, and condensate in a return line sits at saturation too, so a "return" line is not a cool line. Never crack a union, a gauge port, a strainer cap, or a trap on a live steam or condensate path. Isolate at the supply valve, let the section cool, drain to an open point, and confirm a gauge on that section reads zero before you loosen anything.
• Searching for a steam leak. A leak at a small orifice can be nearly invisible and it cuts. Never sweep a hand or an arm through the suspected area. Use a length of board held ahead of you, or an ultrasonic listening instrument, and approach from upwind.
• Flashing. A condensate line that is not hot at the surface can still flash to steam the moment you drop its pressure by opening it. Treat every line downstream of a trap as live until it is isolated and vented.
• Refrigerant circuits. Do not open a sealed circuit to install a probe. Liquid refrigerant freezes skin on contact, so wear splash goggles and gloves rated for cryogenic contact when connecting at a service port, and crack the fitting away from your body. Vapor is heavier than air and pools in pits, crawlspaces and equipment rooms where it displaces oxygen, so ventilate a low or enclosed space before working in it.
• Insulation on steam and older hydronic lines is presumed to contain asbestos until sampled, and lifting it puts fibers in your breathing zone. That control is covered in the profile-reading card and carries the same duty under 29 CFR 1910.1001 in general industry and 29 CFR 1926.1101 for maintenance and renovation work that disturbs it.

What actually happens to the reading

Add heat to a liquid below its boiling point and its temperature climbs, roughly in proportion to the heat added divided by its mass and its specific heat. Keep adding heat once it reaches saturation and the temperature stops climbing. The energy goes into breaking the molecules apart instead, and the reading sits still until the last of the liquid is gone.

The amounts involved are not small corrections. Water takes about 180 Btu per pound to go from freezing to boiling, and about 970 Btu per pound to cross from liquid to vapor at atmospheric pressure. Vaporizing takes roughly 5.4 times the heat that warming it through that entire liquid range does, and none of it shows on a thermometer. Melting ice takes about 144 Btu per pound, which is why a ton of refrigeration is defined as melting a short ton of ice in a day: 2,000 lb times 144 Btu per lb is 288,000 Btu, and over 24 hours that is the familiar 12,000 Btu/h.

Latent heat is why heat exchangers that boil or condense are compact, and it is why a probe on them goes quiet.

The negative space: what a two-phase reading cannot tell you

This is the part worth carrying in your head, because each item on it has cost somebody a return visit.

It cannot tell you how much heat is flowing. A coil at 20 percent of duty and the same coil at full duty read the same saturation temperature if pressure is the same. Rate of transfer does not appear in the number.

It cannot tell you how far along the change has gone. Quality, the fraction that has turned to vapor, is invisible. A point that is 10 percent vapor and a point that is 90 percent vapor read identically, which is precisely why a probe cannot find the boundary inside a coil.

It cannot tell you whether anything is flowing at all. A dead-headed circuit that has reached the same pressure reads the same as a working one. Stagnant saturated fluid and moving saturated fluid are indistinguishable to a thermometer.

It cannot tell you whether the surface is fouled. Fouling reduces duty. Duty does not appear in the reading. On a condensing or boiling side, fouling shows up as a change in the pressure the system settles at, not as a temperature that is wrong relative to that pressure.

It cannot tell you which direction the process is going. Boiling and condensing at the same pressure are at the same temperature. On a reversing system, the reading alone will not tell you which mode a given exchanger is in.

And it cannot distinguish a trap passing condensate from a trap blowing live steam. Both put 212 F fluid into the return at atmospheric pressure. This one item has probably wasted more fuel across the trades than any other single fault.

The one thing it tells you with unusual precision

Inside the two-phase region, temperature and pressure are locked together for a pure substance. That makes the reading a precise pressure indicator rather than a heat indicator, and the trades use it constantly in that direction: a suction pressure converted to saturation temperature, a condensing pressure converted to condensing temperature, a steam pressure read off the saturation table.

The whole superheat and subcooling method exists to work around the blindness. You measure temperature where the fluid is no longer two-phase (superheated vapor after the boiling is finished, subcooled liquid after the condensing is finished) and compare it against the saturation temperature that the measured pressure implies. The difference is a real number that moves with the system's condition, because outside the two-phase region temperature responds to heat again.

Buying the information back

When the temperature reading has gone quiet, three independent measurements recover what it lost.

• Mass flow of the phase-changing stream. Collected condensate over a timed interval, or refrigerant mass flow from compressor data, multiplied by the latent heat, gives duty directly.
• An energy balance on the other side. Air-side or water-side sensible math measures the same heat from the side that has not gone blind.
• The pressure the system settles at. A degraded exchanger forces its boiling or condensing pressure to move so the driving temperature difference can grow to compensate. Approach temperature is the tool for reading that, and it is owned by a sibling card.

Worked case: a coil at 76 percent with every temperature at design

A steam heating coil in a commercial air handler, fed at near atmospheric pressure, condensate returned through a trap. Complaint: the space does not hold on cold mornings.

Everything reads correct. Steam supply line surface 212 F. Coil header 212 F. Condensate line at the trap 212 F. The design drawing calls for saturated steam at atmospheric pressure, which is 212 F. A tech reading only these three points writes "steam present at the coil, temperatures at design."

The air side disagrees. Airflow measured by traverse at 4,000 cfm, entering air 60 F, leaving air 95 F.

Sensible duty: 1.08 x 4,000 x (95 - 60) = 4,320 x 35 = 151,200 Btu/h.

The coil is scheduled at 200,000 Btu/h with 60 F entering air, which would put leaving air at 200,000 / 4,320 = 46.3 F of rise, so 106 F. Delivered duty is 151,200 / 200,000, about 76 percent of schedule, with every temperature in the steam path sitting exactly where the drawing says.

Why nothing in the steam path moved. If the trap is failing to clear condensate fast enough, water backs up and floods the lower passes. Flooded surface still sits at saturation, so it still reads 212 F. What it does not do is condense steam, and condensing is where the heat transfer lives. Duty falls and the thermometer never flinches.

The mirror failure produces the same silence. A trap failed open blows live steam straight into the return, which also reads 212 F, and the energy leaves the building rather than the air stream. At 970 Btu per pound, a trap passing 60 lb/h of live steam wastes 58,200 Btu/h. State the base you are comparing against: that is 29 percent of the 200,000 scheduled, but 38.5 percent of the 151,200 the coil is actually delivering, and the second figure is the one that describes the waste against real output. No temperature anywhere reports it.

Verifying with a second, independent measurement

The coil needs 200,000 / 970, about 206 lb/h of steam, to make its scheduled duty. Condensate was diverted to a drum at the receiver and weighed over a timed 10 minute interval: 26 lb, which is 156 lb/h.

156 / 206 is 76 percent. That is the same figure the air side gave, arrived at from the steam side by a completely different route, which is what makes it a verification rather than a repetition. Two independent measurements landing on the same number also rules out the alternative fault: a trap blowing through would have shown condensate rate at or above design while duty stayed low, and this one is short on both.

The trap was replaced with the correct type and capacity for the application. Re-measured with airflow confirmed still at 4,000 cfm: condensate 34 lb over 10 minutes, so 204 lb/h, and leaving air 105 F.

Air side: 4,320 x (105 - 60) = 194,400 Btu/h, about 97 percent of schedule. Steam side: 204 x 970 = 197,880 Btu/h. The two agree within about 2 percent, which is the normal disagreement between an air traverse and a weighed condensate sample rather than a residual fault.

Where the constant-temperature assumption itself breaks

The whole card above rests on saturation temperature being fixed by pressure. Three real conditions break that, and each one produces a diagnosis that is wrong in a specific direction.

Blends with temperature glide. A pure substance boils at one temperature for a given pressure. A zeotropic mixture does not: it starts boiling at one temperature and finishes at a higher one at the same pressure, and the spread between them is the glide. On a system charged with a glide blend, "the" saturation temperature is not one number, and superheat and subcooling have to be figured against the correct end of the glide, using the fluid's own pressure-temperature data rather than a single-value chart. Reading the wrong end inflates or deflates superheat by the whole glide.

Pressure drop along the exchanger. Saturation temperature follows pressure, and pressure falls as the fluid travels. So the saturation temperature at the outlet of a long evaporator is genuinely lower than at its inlet, and comparing a temperature measured at one end against a pressure measured at the other builds that drop into your result. Take both readings at the same point, or account for the drop deliberately.

Non-condensables. Air or another gas that will not condense at operating conditions adds its own partial pressure to the total. The gauge then reads a pressure higher than the fluid's true saturation pressure, and converting that reading to a temperature gives a saturation temperature that the fluid is not at. It looks exactly like a fouled or undersized condenser on every number except the measured surface temperature, which sits below what the pressure implies. That gap between measured surface and converted pressure is the tell.

References

• ASHRAE Handbook, Fundamentals volume, chapters on thermodynamic properties and on the properties of refrigerants including glide data
• Steam table data for saturation temperature, latent heat of vaporization, and condensate properties at operating pressure
• 29 CFR 1910.1001 and 29 CFR 1926.1101 for asbestos-containing thermal system insulation on steam and hydronic lines
• Manufacturer pressure-temperature data for the specific refrigerant in the system, used in place of any generic chart on a blend
• See related: Why Approach Temperature Matters; How Boiling and Condensing Differ From Heating and Cooling