Manual J S and D: How to Tell If Sizing Was Skipped

The line on the bid is nine words long: 3.5-ton 16 SEER2 system, sized per ACCA Manual J. Nothing is attached to it. The house is 1,900 square feet, which, on the old 500-square-feet-per-ton habit, rounds to 3.8 tons — and there is your 3.5. The word Manual J is doing the work of a calculation that, on this evidence, may never have happened.

That is the problem this piece is about. Not whether the tonnage is right — nobody can know that from the bid alone — but whether the sentence claiming it was calculated can be checked, and what a checkable version looks like. Three ACCA documents are named on nearly every HVAC quote in the country. Each one produces a specific piece of paper with specific numbers on it. If the paper is not there, the manual was not used, whatever the line says.

The code sections, the EPA form and the ACCA edition pages quoted here were checked on 30 August 2026. Standards get revised; check the editions your jurisdiction cites.

Three manuals, three pieces of paper, one chain

ACCA, the Air Conditioning Contractors of America, publishes the three, and each is an ANSI-recognised standard rather than a trade guideline:

The order matters. Manual S cannot be run without a Manual J load to compare capacity against, and Manual D cannot be run without knowing which blower is going in. So a bid that offers Manual D "on request" but has no Manual J is offering the last link of a chain with the first two missing.

One thing the table does not say, and should: Manual J does not produce a tonnage. It produces a Btu/h figure, and two of them — sensible and latent for cooling, plus a separate heat loss for heating. The tonnage appears only after Manual S has picked a machine. If the only number you are shown is "3 tons", you are being shown the output of a selection step that had nothing documented feeding it.

What the code actually requires, and its two exceptions

The model residential code has said this for several cycles. 2021 IRC Section M1401.3, Equipment and appliance sizing, reads:

Heating and cooling equipment and appliances shall be sized in accordance with ACCA Manual S or other approved sizing methodologies based on building loads calculated in accordance with ACCA Manual J or other approved heating and cooling calculation methodologies.

The 2021 edition then adds two exceptions under which sizing "shall not be limited to the capacities determined in accordance with ACCA Manual S": first, where the equipment uses multistage or variable refrigerant flow technology and the calculated loads fall within the manufacturer's published capacity range; second, where no published capacity satisfies both the total and sensible gains and the next larger standard size is specified. Read those carefully, because a contractor will sometimes cite the first one as if it waives the calculation. It does not. Both exceptions still begin from "the loads calculated in accordance with the approved ... methodology". They relax the selection limit, not the requirement to calculate.

The energy code says the same thing from the other side. 2021 IECC R403.7, Equipment sizing and efficiency rating, requires equipment "sized in accordance with ACCA Manual S based on building loads calculated in accordance with ACCA Manual J or other approved heating and cooling calculation methodologies" — wording that is unchanged from the 2015 edition. Some states go further. Florida's energy code (the Sixth Edition, 2017, at R403.7) required that the manufacturer and model number of the outdoor and indoor units be submitted along with sensible and total capacities at design conditions, and stated flatly that the code "does not allow designer safety factors, provisions for future expansion or other factors that affect equipment sizing". The current Eighth Edition should be read for the present wording, but the direction has been consistent.

So: the requirement exists in the model code. What varies enormously is enforcement. Many inspectors ask for the report only when a permit shows equipment going up in size, or a house that has been remodelled. And if the job is done without a permit at all — the discount this site has already priced — nobody asks. Which means on a replacement job, the person most likely to ever look at the Manual J is you.

The nine inputs that have to be on a real report

You do not need to own Manual J to check a Manual J. The federal ENERGY STAR programme publishes a form designed to be filled from one, and it is the best public checklist of what a load calculation must contain. EPA's National HVAC Design Report, Rev. 14 (revised 15 January 2025, Form 5900-426) is for new homes, but Section 3 — Room-by-Room Heating & Cooling Loads — is nothing more than the list of inputs any Manual J run has to have. Lay it beside the contractor's printout:

Insulation R-values for walls, ceiling and floor, and where the ducts run (attic, crawlspace, conditioned space) sit on the same pages. Every one of these is a question the contractor had to answer by walking the house or asking you. If nobody measured a window, counted an orientation or asked about the attic insulation, the software was fed defaults, and defaults in every Manual J package lean towards more load.

Two tells. A one-page report with a single tonnage on it is a summary, not a calculation; ask for the room-by-room pages behind it. And a report dated the same afternoon as a fifteen-minute visit that never involved a tape measure has defaults in it. Neither is evidence.

Manual S: where the percentage comes from

The EPA form's Section 4 is a Manual S selection laid out as fill-in-the-blanks, and it shows what the manual is really about: capacity at your design conditions, not nominal tonnage. Item 4.11 asks for "total capacity at design conditions, from OEM expanded performance data". A "3-ton" condenser is rated at 36,000 Btu/h under AHRI test conditions of 95 °F outdoors and 80 °F/67 °F indoors; at a 100 °F design day it delivers less, and the manufacturer's expanded tables say how much less. Manual S reads that table. A rule of thumb does not know it exists.

Then the form divides: item 4.13, cooling sizing % = total capacity ÷ maximum total heat gain, and checks it against the limits the form reproduces from Manual S (the form's footnotes cite Manual S Second Edition; the 2023 Third Edition is now the ANSI-approved version and ACCA says it widened the tolerances for variable-capacity heat pumps, so confirm current figures with ACCA):

For a furnace the form gives 100–140 % of heat loss when used for heating only, and up to 400 % allowed when paired with cooling — because in a mild climate the blower needed for the air conditioner can force a furnace larger than the heat loss wants.

Put the opening bid through it. Suppose the contractor's own Manual J, once produced, shows a total gain of 26,000 Btu/h. The 3.5-ton unit at design conditions delivers perhaps 39,000. That is 150 %: outside recommended for anything, outside allowed for single-speed. If the compressor is variable-speed it scrapes in. If not, the bid has to explain the exception it is relying on — and IRC M1401.3's second exception only lets you round up to the next standard size, which from 26,000 is a 2.5-ton, not a 3.5.

The AHRI reference number on the bid is the bridge to this step. EPA's form asks for it at item 4.5 because the expanded performance data belongs to a specific indoor-and-outdoor pairing, and a Manual S run on a different coil from the one being installed is a Manual S for somebody else's system.

Manual D: the one nobody attaches

A replacement bid that leaves the existing ducts alone will almost never carry a Manual D, and there is a defensible reason: the ducts are not being built. But the equipment is changing, and the new blower has to push its design airflow — item 5.2 on the EPA form, cooling and heating CFM — through those same ducts at a total external static pressure the manufacturer allows, item 5.4, in inches of water column. A 3-ton system wants roughly 1,050–1,200 CFM. Ducts sized for a 2-ton furnace in 1994 may not pass it without noise and a starved coil.

You will not get a duct design on a like-for-like swap. What you can reasonably ask for is a measured static pressure on the existing system, written on the bid, and the manufacturer's maximum for the new air handler beside it. That single comparison is what Manual D would have protected, and it costs the contractor a manometer and five minutes. If the number is over the limit, the bid needs either a duct line item or an explanation. Line 9 of the fourteen a complete HVAC bid must contain is that line.

The line the rule of thumb leaves behind

Energy Vanguard's Allison Bailes has written that a rule of thumb "like 400 to 600 square feet per ton will work for very few houses", and a later sample of forty of his firm's Manual J runs came out with a majority above 1,000 square feet per ton and a low of 624. That is one consultancy's field data, not a standard, but it points at the trace the rule leaves. A house sized by area produces round tonnages that track the floor plan: 1,500 sq ft becomes 3 tons, 2,000 becomes 4. A house sized by Manual J produces odd numbers — 26,400 Btu/h, 31,800 — that land between nominal sizes and force a Manual S decision about which way to round.

So when three bids for the same house all say 3.5 tons and none has a load attached, the agreement is not corroboration. It is three people using the same shortcut. And when a fourth says 2.5 tons with twelve pages behind it, the twelve pages are the reason to take that bid seriously, not the reason to doubt it.

Before signing: what the bid must show for each manual

None of this asks a contractor to do anything the model code has not already required of them since well before the 2021 edition. It asks them to show it. The one who can produce the pages in a day had them all along.