Manual S Oversizing Limit for Air Conditioners Under Manual S-2023

Manual S-2023 caps a single-speed air conditioner at 115% of the Manual J total cooling load, or 120% when that load is 24,000 Btu/h or less. Two-speed equipment may go to 125%, and variable-capacity equipment sized by the simplified procedure to 130%. Those limits come from Tables N2.3.1 and N2.3.4 as reproduced in ACCA's Addendum c public-review draft, which amends Manual S-2023 as already revised by 2024 Addenda a and b. Section M1401.3 of the 2024 IRC references ANSI/ACCA 3 Manual S-2023. The capacity you compare is the one in the manufacturer's expanded performance data at your design conditions.

The Manual J vs S vs D breakdown covers the order of the three manuals. On the Pro and Team plans, the Manual S equipment selection tool still uses the first-edition window, so run the 2023 percentage yourself.

What does the Manual S oversizing limit cap?

Total cooling capacity from expanded data at design conditions, divided by the Manual J total load. Manual S-2023 limits that size factor to 1.15 for single-speed equipment (1.20 at 24,000 Btu/h or less), 1.25 for two-speed and 1.30 for simplified variable-capacity sizing. Total capacity must reach 0.90 of load and latent 1.00, and sensible 0.90 except in simplified variable-capacity sizing. Single-speed and two-speed equipment in houses with a Manual J sensible heat ratio of 0.95 or higher is checked under a separate dry sizing condition.

ACCA announced ANSI approval of the 2023 edition on September 11, 2023, "in time to be referenced by the International Code Council's (ICC) 2024 edition of the International Residential Code (IRC)." The 2024 IRC referenced standards on UpCodes list ANSI/ACCA 3 Manual S-2023 for M1401.3.

None of these limits is marked as a proposed change, and neither table sets a latent maximum.

The dry sizing condition (Table N2.3.2) takes over for single-speed and two-speed equipment when the Manual J sensible heat ratio is 0.95 or higher. A single-speed unit then passes when total capacity divided by the total load plus 6,000 Btu/h is 1.00 or less, with the same 0.90, 0.90 and 1.00 minimums.

The 15% wording goes back to section 3-4 of the earlier Manual S.

ACCA Manual S (pre-2023) § 3-4

"Cooling equipment shall be sized so that the total cooling capacity does not exceed the total cooling load by more than 15 percent." - ACCA Manual S, quoted in viridiant.org

Check which edition your adopted code references before you apply a number.

Which capacity number goes into the check?

The total capacity from the OEM expanded performance table at your Manual J outdoor design temperature, the indoor entering wet bulb (63°F for 75°F and 50% RH) and the airflow you'll run. The AHRI certificate is tested at 95°F outdoor and 80°F/67°F indoor, which is not your design point.

"OEM expanded performance data should be used to select properly sized equipment." - Wes Davis, ACCA, ICC Building Safety Journal

The same article adds that cooling equipment size "must be based on the same temperature and humidity conditions that were used to calculate the Manual J loads."

Read three things off the table. When your design temperature falls between two outdoor columns, interpolate. A hotter column understates capacity and flatters the oversizing check, and a cooler one flatters the minimum check. Take the 63°F EWB row and the CFM row that matches your blower setting. Davis's 3-ton unit delivers the required capacity at 1,050 CFM. A different CFM row changes total and sensible, so the CFM you read is the CFM you commit to in Manual D.

How do you check a real AC selection against the limit?

Pick the size factor for your equipment type and load, multiply the Manual J total load by it for the ceiling, read total and sensible capacity at design, subtract to get latent, then run three comparisons. In ACCA's published example a 30,000 Btu/h load gets a unit delivering 31,510 total, 105% of load, which passes under the first edition and Manual S-2023 alike.

Davis's house is designed at 95°F outdoor, 75°F indoor, 50% RH and 63°F wet bulb. Manual J returns 30,000 Btu/h total, 22,000 sensible and 8,000 latent.

1. Pick the condition and limit. The load sensible heat ratio is 22,000 / 30,000 = 0.73, under 0.95, so the standard condition applies. The load is over 24,000 Btu/h, so a single-speed unit gets 1.15, a ceiling of 34,500 Btu/h. The floor is 30,000 x 0.90 = 27,000 (Davis used the first-edition 100%).
2. Try the obvious size. The 2.5-ton model is nominally 30,000 Btu/h, but at 95°F outdoor and 63°F EWB its expanded data falls short of the first-edition 100% total floor Davis applies. Under Manual S-2023, recheck it against 27,000 total, 19,800 sensible and the full 8,000 latent.
3. Read the next size at design. The 3-ton model delivers 31,510 Btu/h total and 23,000 sensible at those conditions.
4. Check total. 31,510 / 30,000 = 1.050, so 105.0%. That's under 115% and over the floor.
5. Check sensible. 23,000 / 22,000 = 1.045, so 104.5%, over the 0.90 sensible minimum.
6. Check latent. 31,510 - 23,000 = 8,510 Btu/h, and 8,510 / 8,000 = 1.064, so 106.4% of the latent load, over the 1.00 minimum.

On nameplate the 3-ton unit is 120% of load. At design it is 105%, the number Manual S uses.

What if no unit fits under the limit?

First check the next smaller unit against the Manual S-2023 minimums, 90% of total and sensible load and 100% of latent, since a unit below the load can still comply. If neither size fits, the 2024 IRC section M1401.3 lets sizing exceed Manual S when published capacities cannot satisfy both total and sensible gains and the next larger standard size is specified.

The Shums Coda Manual S breakdown, written for Xcel Energy's Building Codes Support Program in Colorado, walks through a 3,600 sq ft Denver house. A unit at about 136% to 137% of load was accepted there because the next smaller unit fell short of the full load. Under Manual S-2023 that dry house goes to the dry sizing condition, worked out below.

1. Start from the loads. Denver, 90°F outdoor design, 75°F indoor. Equipment sensible load 15,756 Btu/h, latent load 0, so the load sensible heat ratio is 1.00.
2. Read expanded data. The Carrier 24ABB324 unit at 800 CFM, 63°F EWB and the 95°F column delivers 21,600 total and 16,080 sensible, so 5,520 latent.
3. Apply the latent adjustment. With no latent load, half of the excess latent moves to sensible, 5,520 / 2 = 2,760, and 16,080 + 2,760 = 18,840 sensible.
4. Check total. 15,756 x 1.15 = 18,119 Btu/h ceiling. 21,600 / 15,756 = 1.371, so 137.1%, about 3,480 Btu/h over.
5. Record the decision. The worksheet calls the unit slightly oversized, "however the next smaller unit does not have the capacity to meet the load. In this case this air conditioner is acceptable."

The worksheet applies the 15% limit and doesn't cite the M1401.3 exception.

Manual S-2023 changes the arithmetic twice. The load is under 24,000 Btu/h, so the standard-condition ceiling would be 15,756 x 1.20 = 18,907 Btu/h, and at 137% the unit still breaks it. But a load sensible heat ratio of 1.00 sends single-speed equipment to the dry sizing condition, where 21,600 / (15,756 + 6,000) = 21,600 / 21,756 = 0.99, under the 1.00 limit. Total (1.37) and sensible (16,080 / 15,756 = 1.02) clear 0.90, and with a zero latent load the latent check passes by default. On the worksheet's own reading the unit passes without an exception.

That pass is thin. The PDF computes the ceiling from 15,832 (18,206 Btu/h) while its report page shows 15,756. The worksheet took the 95°F column as "within 5 degrees" of a 90°F design, but the 85°F column reads 22,720 total at 800 CFM and 63°F EWB. Our linear estimate, not the worksheet's, puts 90°F near 22,160, about 141% of 15,756, and 22,160 / 21,756 = 1.02 misses the dry limit. That worksheet corrects airflow for altitude, 800 CFM / 0.832 = 962 CFM, so its 995 CFM is about 828 standard CFM, just above the 800 CFM row.

The half-latent step comes from section 3-10 (step 4) of the earlier Manual S, which lets you "add half of the excess latent capacity to the sensible capacity" when sensible falls short, per the Viridiant guide. Raw sensible already covered the load, so here it changed nothing.

ACCA's proposal RM 3-15 in ICC's 2015 Group A changes asked to delete the two M1401.3 exceptions. The 2024 IRC text on UpCodes still carries both.

Check local adoption before you lean on the third row. It is a Virginia amendment, in the 2015 and 2021 Virginia code, absent from the model IRC.

Does the limit apply to sensible or total load?

Total. In Manual S-2023 the cap divides total capacity by total load. Sensible capacity has only a 0.90 floor and latent a 1.00 floor. The 2014 table added a 1.50 preferred latent maximum; the 2023 table has none.

Ed Janowiak of ACCA cites 15, 20 and 30 percent, the same numbers as the 2014 table. The consolidated text in the Addendum c draft shows 25 for two-speed.

ACCA HVAC Blog § Hey Ed

"The correct airflow for an air conditioner will be one that meets the sensible, meets the latent, and doesn't exceed the total by 15, 20, or 30 percent, depending on the compressor technology." - Ed Janowiak, ACCA, hvac-blog.acca.org

A July 2026 SlashGear explainer near the top of search results says rated capacity "can't be more than 115% of the calculated sensible cooling load." Apply that ceiling to the design-condition capacity in Davis's example and the 3-ton unit fails at 31,510 / 22,000 = 143%, for a unit ACCA's own reviewer accepts at 105% of total load.

Checking the Manual S window before you quote

Run the whole-house load, pick the sizing condition and size factor for your compressor type, and quote only from units inside it. BuildSolver's Pro and Team tool picks a nominal size inside the first-edition 100% to 115% window from a derate model, as a sales-phase estimate. Check the 2023 limits, the dry condition and a specific unit's size factor by hand.

The Manual S equipment selection tool takes the total and sensible loads you give it in the chat, typed in or carried over from a cooling load run there. It picks the smallest standard size whose estimated capacity at your design temperature lands between 100% and 115% of total load, or 125% for a heat pump or dual-fuel system in IECC zone 4 and colder, the first-edition heat pump rule. Against Manual S-2023 that window is off in both directions. The 100% floor is stricter than 0.90 and the 115% cap stricter than the 1.20 allowed at 24,000 Btu/h or less, while 125% for a single-speed heat pump is looser than both 1.15 and 1.20. It has no two-speed, variable-capacity or dry-condition limits. If no size fits, it takes the smallest one up to 5 tons that meets the load and warns which limit it breaks.

When you enter a specific unit's AHRI-rated capacity, the tool checks that it meets 100% of the total, sensible and latent loads, taking sensible from a typical ratio unless you enter it, but doesn't test its total capacity against a cap. Enter the AHRI rating, measured at 95°F outdoor and 67°F indoor wet bulb. The tool scales it to your outdoor design temperature with a model built at 67°F wet bulb, and at 95°F that factor is 1.000. Above sea level it also cuts sensible capacity by the air density ratio at your site altitude, when the chat looked up your design conditions by ZIP or city. A house held at 75°F and 50% RH puts 63°F wet bulb on the coil, where the coil delivers less total capacity than at 67°F, so the tool can pass a unit that falls short on the 63°F row. For the size factor and the 90% minimums, divide the expanded-data total and sensible at your design temperature, 63°F EWB and your CFM by the Manual J loads, as the hand check above does. The tool also holds sensible capacity to 115% of the sensible load. When it picks a size, it estimates sensible from a fixed typical sensible heat ratio, and it marks a unit you enter as over that line only if you also enter its sensible capacity at those rating conditions, which you read from expanded data. That dehumidification flag isn't in the 2014 or 2023 table. Pull the submittal's expanded table before you order.

The heat pump sizing guide covers the heating side and backup heat. Many AHJs require the manufacturer's data and output from ACCA-approved software for a permit. BuildSolver isn't ACCA-approved software, and the permit path is in Manual J for permit. Start with the whole-house load in the free Manual J calculator, then run your candidate unit through the check above.

Sources

• ACCA, Proposed Addendum c to the Normative Sections of ANSI/ACCA Manual S-2023 with 2024 Addenda a and b (public-review draft, 2025), Tables N2.3.1, N2.3.2 and N2.3.4: higherlogicdownload.s3.amazonaws.com
• ACCA, ANSI Approves ACCA Manual S - 2023, September 11, 2023: acca.org
• 2024 IRC Chapter 44 and section M1401.3, GSA Residential Code 2024 on UpCodes: up.codes
• ACCA Manual S sections 3-4 and 3-10 and the 2014 size-limit table, as reproduced in Viridiant, Heating/Cooling Design, Equipment, and Installation (2021 VRC/VECC Code Guide): viridiant.org
• Wes Davis, ACCA, Reviewing HVAC Designs for Compliance with ACCA Manual S, ICC Building Safety Journal, Jan-Feb 2009: media.iccsafe.org
• Ed Janowiak, ACCA, Hey Ed, What Is the Correct Airflow for an AC: hvac-blog.acca.org
• Shums Coda, Manual S Breakdown, Building Codes Support Program: shumscoda.com
• ICC, 2015 Group A Proposed Changes, IRC Mechanical (M1401.3 exceptions text): iccsafe.org
• ARI Standard 210/240, standard rating conditions table: law.resource.org
• SlashGear, What Is the 115% Rule for Air Conditioners, July 18, 2026: slashgear.com

For estimation purposes only. Not a substitute for a licensed engineer, and not ACCA-approved for permit submission.