What a Delta-T Is Actually Telling You

Why this matters

Delta-T is the most quoted number in the trades and the most misread. It gets treated as a health score, as though a number inside a published band means the equipment is fine and a number outside it means the equipment is broken. It is neither. It is a ratio of two things, and a ratio can sit perfectly in range while both of its terms are badly wrong. Shops lose whole afternoons, and sometimes whole seasons of a customer's comfort, to a reading that was in range on a system with two cancelling faults.

The identity behind the number

Every delta-T you take across a component comes from the same relationship:

heat rate = mass flow x specific heat x temperature difference

Rearranged so the measured value is on the left:

temperature difference = heat rate / (mass flow x specific heat)

For a fixed fluid, specific heat is close to constant over the range you work in, so the sentence to carry is: delta-T is load divided by flow. Two field forms of the same identity, both for standard conditions:

• Water: heat rate in Btu per hr = 500 x GPM x delta-T
• Sensible air: heat rate in Btu per hr = 1.08 x CFM x delta-T

Those constants are not magic. The water one is 60 minutes per hour x 8.33 lb per gallon x 1.0 Btu per lb per F. The air one is 60 x 0.075 lb per ft3 x 0.24 Btu per lb per F. Knowing where they come from is what lets you correct them when the fluid or the air is not standard, which the last section covers.

A ratio hides both of its terms

Because delta-T is load over flow, four different situations produce the identical reading:

Rows two and three are the ones that cost money. A system delivering half its rated output at half its rated flow shows a textbook delta-T and a customer who is still cold. A delta-T inside the published band is not evidence of capacity. It is evidence that load and flow are in the designed proportion to each other, which is a different claim.

This is also why delta-T alone can never separate a heat-transfer problem from a flow problem. That separation needs a second, independent measurement, and the sibling card on that split covers which one to reach for.

What a published target silently carries

A nameplate temperature-rise range or a published design delta-T is a conditional statement, and the conditions almost never travel with the number when techs quote it to each other. Every published target carries at least three:

• A design flow. The target was derived at a specific airflow or a specific water flow. At any other flow the correct target is different, and it moves inversely: half the flow doubles the correct delta-T at the same load.
• Design entering conditions. Entering fluid temperature, and on the air side entering wet bulb as well, change how much heat the component can move before the target means anything.
• A fluid. A glycol mix is not water and does not carry the same heat per unit of volume per degree.

Quoting a delta-T target without its flow is like quoting a torque figure without saying which fastener. The correct habit is to state them together every time: "the target rise at the design airflow on this nameplate," never just "the target rise."

Two faults that cancelled

A gas furnace, complaint of long run times and a house that never quite catches up. The nameplate gives a temperature-rise range and a rated output.

Before any of this is measured: work on a fired appliance is combustion work. If anyone smells gas at any point, everyone leaves the building immediately, nothing is switched on or off, no lights and no phone are used inside, and the call is made from outside. The limit control is never jumpered or bypassed to keep the appliance running for a reading, because that control is what stops a heat exchanger from being run past its temperature rating. Combustion analysis and gas-side verification belong to the sibling cards on combustion; this article stays on the heat side.

Design numbers. Rated output 60,000 Btu per hr. Design airflow, at the widely used residential convention of roughly 400 CFM per nominal ton with a common working band of 350 to 450, comes out at 1,200 CFM for this unit. The design rise follows from the identity:

60,000 / (1.08 x 1,200) = 60,000 / 1,296 = about 46 F

That sits inside the nameplate band, as it should, because the band was derived the same way.

Measured. The tech takes a settled supply and return pair with one instrument, with the supply probe placed out of line of sight of the heat exchanger so it is not reading radiant heat instead of air. Measured rise: 48.6 F.

Against a design rise of 46 F, that is 48.6 / 46 = about 5 percent high. Inside any reasonable band. A tech who stops here writes "temperature rise normal" and starts looking at the thermostat, the ductwork or the customer's expectations.

What the two independent checks found. An airflow measurement came back at 800 CFM, which is 800 / 1,200 = about 33 percent below design. A clocked input check came back at roughly 70 percent of rated, so an actual output near 42,000 Btu per hr, about 30 percent low.

Run the identity on the real numbers:

42,000 / (1.08 x 800) = 42,000 / 864 = 48.6 F

The measured rise was correct. Both terms were about 30 percent low, so their ratio landed within 5 percent of design. The furnace was delivering 42,000 Btu per hr into a house sized for 60,000, which is exactly the long-run-time complaint, and the single number the tech trusted was the one number in the system that looked healthy.

Why it presents this way rather than as a limit trip. Low airflow alone would have driven the rise to 60,000 / 864 = 69.4 F, above any normal nameplate band, and the limit would have started cycling the appliance, which is a loud, obvious fault. The underfire quietly removed that symptom. Two faults, each of which would have been findable on its own, hid each other because delta-T reports only their ratio.

The failure mode to name on the ticket. "Rise in range" is not a finding. It is one equation with two unknowns. The finding is "rise in range at measured airflow of 800 CFM and measured input at about 70 percent of rating," and that sentence is diagnosable by the next person.

What would have changed the conclusion. If the airflow measurement had come back near 1,200 CFM, the 48.6 F rise would have implied an output near 63,000 Btu per hr, slightly above rating, and the complaint would have pointed at load or distribution rather than the appliance. Same measured delta-T, opposite conclusion, decided entirely by the second measurement.

When delta-T is legitimately the answer

Nothing above says the number is useless. It is the right tool in three situations, and it is worth knowing which one you are in:

• When you know the flow and want the capacity. Measure the difference, multiply by the flow and the constant, and you have a heat rate. This is the most defensible capacity measurement available on most residential equipment.
• When you know the load and want the flow. Rearranged the other way, a known heat input plus a measured difference gives an implied flow, which is often easier and more accurate than trying to measure airflow directly in a bad duct.
• When you are comparing the same system to itself over time. A rise that has moved from 46 F to 62 F on the same equipment at the same setting is real information even without knowing either term, because one of them changed. Trend readings are worth logging for exactly this reason.

Notice that the first two both require you to already know one term. That is the honest scope of the number.

What moves delta-T without either term changing

Three conditions change the arithmetic itself, and each one has caught techs who moved between markets or between fluids:

• Altitude. The 1.08 constant is built on sea-level air density. Density falls with elevation, and at roughly 5,000 ft the constant drops to around 0.90. Using 1.08 there overstates capacity by roughly the same proportion you ignored. Correct the constant for site elevation, or state in the notes that it is uncorrected.
• Glycol. A propylene or ethylene glycol mix carries less heat per unit volume per degree than water, and the shortfall grows with concentration and falls with temperature. The correct figure comes from the fluid manufacturer's property data at your concentration and your operating temperature; do not carry the water constant into a glycol loop and do not assume a single correction factor covers every mix.
• A wet coil. The 1.08 form is sensible-only. Any coil pulling condensate is also doing latent work that a dry-bulb difference cannot see, so the sensible form understates total capacity. Use the enthalpy form on a wet coil; the sibling card on latent heat carries that arithmetic.

Two more make the reading itself invalid rather than the arithmetic: a probe in a stratified stream, and a bypass between the two probes. Both belong to the measurement procedure rather than to the interpretation, and the sibling how-to covers them.

Verifying you read it right

Before the number goes on a ticket, three questions, each of which has to have a written answer:

1. What flow was this taken at, and how do I know? If the answer is "the fan table" or "the design drawing," label it as assumed rather than measured. An assumed flow makes every capacity figure derived from it assumed too.
2. Did I check the second term independently? One equation, two unknowns. If you only measured the difference, you have not measured capacity, and no band on any nameplate changes that.
3. Do my two derived figures agree with each other? If you know the load and back out the flow, does that flow look physically plausible for the equipment and the duct or piping in front of you? An implied airflow well above what the fan can produce at any tap means one of the inputs is wrong, and it is usually not the temperature.

References

• ASHRAE Handbook Fundamentals, sensible and total heat rate relationships and air density corrections
• Manufacturer nameplate temperature-rise range and rated output for the specific appliance
• Fluid manufacturer property data for glycol concentration, specific heat and density
• See related: How to Read a Temperature Difference Across a Component; How to Tell a Heat-Transfer Problem From a Flow Problem; What Latent Heat Explains That Sensible Heat Cannot