Input, output and why the two numbers differ
Every gas furnace carries two capacity numbers and they are not interchangeable. Input is the rate at which the burner consumes fuel: an 80,000 BTU/h furnace burns 80,000 BTU of gas every hour on high fire. Output is the rate at which heat leaves the heat exchanger and enters the supply air. The rest goes up the flue as hot combustion products, plus a small amount of jacket loss.
The bridge between them is AFUE, the annual fuel utilisation efficiency measured under the US Department of Energy test procedure and printed on the yellow EnergyGuide label. AFUE is a seasonal average that includes cycling and off-cycle losses, so it is slightly lower than steady-state combustion efficiency, but it is the number the industry sizes with. An 80% AFUE furnace with 80,000 BTU/h input delivers about 64,000 BTU/h. A 96% condensing furnace of the same input delivers 76,800 BTU/h.
That is why replacing an old 80,000 BTU/h furnace with a new 80,000 BTU/h furnace usually oversizes the house. The input matched, but the output went up by 20% because the new machine condenses its flue gas and recovers the latent heat of the water vapour. Size on output against load, then work backwards to the input you must buy.
The third correction is what happens between the furnace and the rooms. Ducts running through a vented attic or a crawlspace lose heat by conduction and by leakage, and that heat never reaches the occupied space. Building America field work and the duct-loss tables in ACCA Manual J both treat this as a percentage adder on the equipment side, which is what the duct distribution loss field does here. Set it to zero when all the ductwork is inside the thermal envelope.
The sizing calculation and where the design load comes from
The design heat loss is the whole-house transmission and infiltration load at the outdoor 99% design temperature, meaning the temperature that only 1% of winter hours fall below. It is computed as Q = UA × (Tin − Tdesign), where UA is the sum over every envelope assembly of its area divided by its R-value, plus the 1.08 × CFM term for infiltration. Get that number from the Manual J load calculator, from a rating firm's report, or from a blower-door-informed model. Do not get it from a per-square-foot rule: the same 2,000 sq ft house can have a 25,000 BTU/h load or an 80,000 BTU/h load depending on vintage and climate.
Once you have Q, the arithmetic is short. Multiply by one plus the duct loss to get the output the furnace itself must produce. Divide by AFUE to get input. Divide again by the altitude derate factor if the site sits above 2,000 ft, because a naturally aspirated burner at altitude draws less oxygen per cubic foot of air and the appliance's usable input falls. The customary US rule reduces input 4% for each 1,000 ft above 2,000 ft, so a 6,000 ft site keeps 84% of nameplate input.
Then compare the nearest catalogue size against the load. ACCA Manual S is the standard that governs equipment selection once Manual J has produced the load, and for furnaces it limits output to no more than 140% of the design heating load. Single-stage furnaces are made in coarse steps, so the 140% window is what makes the selection possible at all, but a selection at 105% is better than one at 138%.
Worked example: a 60,000 BTU/h load with attic ducts
A Manual J report gives a 2,400 sq ft house a design heating load of 60,000 BTU/h at the 99% design temperature. The trunk and branches run through a vented attic, so you allow 10% distribution loss. You are quoting a 95% AFUE condensing furnace, and the site is at sea level.
- Output needed at the furnace. 60,000 × (1 + 0.10) = 66,000 BTU/h.
- Divide by AFUE. 66,000 ÷ 0.95 = 69,473.7 BTU/h of input.
- Altitude derate. Sea level, so kalt = 1.00 and the required input stays 69,473.7 BTU/h.
- Nearest catalogue size. The next standard input at or above that is 70,000 BTU/h.
- Delivered output. 70,000 × 0.95 = 66,500 BTU/h.
- Oversize check. 66,500 ÷ 66,000 = 1.008, so the furnace delivers 100.8% of what the house and its ducts need. Comfortably inside Manual S.
Now change one thing: keep the same house but quote an 80% AFUE furnace. Required input becomes 66,000 ÷ 0.80 = 82,500 BTU/h, the next size is 88,000, and delivered output is 88,000 × 0.80 = 70,400 BTU/h, or 107% of the load. Both selections are fine. The 80% furnace burns 82,500 BTU/h of gas to deliver 66,000, while the 95% burns 69,474 to deliver the same, a difference of about 13,000 BTU/h of fuel every hour the burner is on.
Reading the oversize ratio
The oversize ratio compares delivered output against the load including duct allowance. A ratio between 1.00 and 1.40 satisfies Manual S for a furnace. Where inside that window you land changes how the system behaves.
Near 1.00 the furnace runs long cycles on design days and effectively continuously during a cold snap, which is what you want: steady supply temperatures, even room-to-room distribution, and a blower that keeps mixing the house. The risk is recovery time. If the thermostat is set back 8 °F overnight, a furnace at 105% has almost no spare capacity to bring the house back up quickly, so pick a smaller setback or accept a slower recovery.
Near 1.40 the furnace satisfies the thermostat quickly and then stops. Short cycles mean the heat exchanger spends a larger fraction of its running time warming up, the supply air arrives hot and then stops, and rooms at the end of long duct runs never get their share. In a house with marginal duct design this is where comfort complaints come from.
Above 1.40 the calculator flags the selection. The usual fixes are a two-stage or modulating furnace, whose low fire may sit at 40% to 70% of nameplate and therefore lands much closer to the real load for most of the season, or a smaller single-stage model if one exists. Do not solve an oversize problem by reducing the blower speed: airflow has its own requirements, and starving a furnace of air raises the temperature rise across the heat exchanger and can trip the limit switch. Check the temperature rise against the rating plate and set airflow from the room CFM calculator instead.
The load-per-square-foot output is a cross-check on the heat loss itself, not on the furnace. If the figure looks implausible for the vintage and climate, the load calculation is where to look, and the envelope inputs behind it can be sanity-checked with the R-value to U-value converter.
Required input for a given output, by AFUE
| Output needed (BTU/h) | 80% AFUE | 90% AFUE | 95% AFUE | 98% AFUE |
|---|---|---|---|---|
| 30,000 | 37,500 | 33,333 | 31,579 | 30,612 |
| 40,000 | 50,000 | 44,444 | 42,105 | 40,816 |
| 50,000 | 62,500 | 55,556 | 52,632 | 51,020 |
| 60,000 | 75,000 | 66,667 | 63,158 | 61,224 |
| 80,000 | 100,000 | 88,889 | 84,211 | 81,633 |
| 100,000 | 125,000 | 111,111 | 105,263 | 102,041 |
The output column is the design heat loss already multiplied by one plus the duct loss allowance. Above 2,000 ft, divide the input again by the altitude derate factor.
Assumptions and limits of this calculation
- It sizes on the heating load only. If the same air handler carries a cooling coil, the blower must also move the cooling airflow, which is often the binding constraint. Size the cooling side separately with the air conditioner BTU calculator or a full load calculation.
- AFUE is a seasonal average, not the efficiency at any instant. Steady-state efficiency on a long call is higher; efficiency on a short cycle is lower. Using AFUE for sizing is conventional and conservative enough, but it is not a thermodynamic statement about the machine at design conditions.
- The duct loss allowance is a lump percentage. Real duct loss depends on surface area, insulation, leakage rate and the temperature of the buffer space. A tested duct system with measured leakage deserves a measured number rather than a default.
- The altitude rule is the customary US one. The appliance rating plate, the manufacturer's high-altitude kit instructions and the local gas code govern the actual derate, and many current furnaces are certified to 10,000 ft with a conversion kit.
- Catalogue sizes vary by manufacturer. The list here covers the common residential steps. Check the actual model family before you commit to a selection.
- Nothing here evaluates the venting. Changing input rating changes vent sizing, and a condensing furnace needs a different vent material and a condensate drain entirely. That is a separate calculation under the National Fuel Gas Code.
Where furnace sizing sits in the design sequence
The correct order is load, then equipment, then distribution: Manual J gives the load, Manual S selects the equipment, Manual D designs the duct system, and Manual T places the registers. Skipping straight to equipment is the single most common failure in residential HVAC, and it is why so much installed equipment is one to two sizes too large.
Once the furnace is chosen, its blower table sets what happens next. The rated external static pressure at the design airflow is the starting point for the available static pressure calculator, which subtracts the filter, coil and accessory losses and turns what is left into a friction rate. That friction rate then drives every duct size through the duct size calculator. A furnace correctly sized to the load and then connected to undersized ducts will still deliver poor comfort, because the blower cannot move its rated airflow against the pressure the ducts impose.
If you are comparing a furnace with a heat pump, note that the two are sized on different criteria. A furnace is selected against the heating load with a 140% ceiling. A heat pump is normally selected against the cooling load, with supplemental heat covering whatever the heating load exceeds, because a heat pump sized to the heating load in a cold climate would be badly oversized for cooling and would dehumidify poorly all summer.