An explainer, not a calculator. BuildSolver runs Manual S equipment selection on the Pro and Team plans; nothing on this page picks equipment for your job. Final selection belongs to a licensed mechanical engineer or a certified contractor working from the manufacturer's expanded performance data.

Maintained by the BuildSolver engineering team · Updated September 21, 2026

Method · ACCA Manual S

Manual S calculator and the ACCA equipment selection method

There is no form on this page. The oversize limits and the capacity checks run in the Pro and Team tool. This page shows the procedure so you can audit any Manual S you are handed.

Below are the sizing limits by equipment class for both editions of Manual S, and four jobs run through the BuildSolver engine. The five steps run in this order.

Write down total cooling, sensible cooling, latent cooling and heating, together with the outdoor design temperature and the indoor conditions those loads were figured at.
Find the candidate outdoor unit and indoor coil in the manufacturer's tables and read capacity at your outdoor design temperature, your entering wet bulb and the airflow you plan to run.
Total capacity at design has to cover the load and stay under the limit for that equipment class. Under the first edition that is 115 percent for an air conditioner, 125 percent for a heat pump in a heating dominant climate, and 140 percent of the heating load for a furnace or a boiler. Manual S-2023 restates the ceilings by compressor type.
Sensible capacity at design has to cover the sensible load and latent capacity has to cover the latent load. A unit that passes on total can still miss one of them.
Read heating capacity at the heating design dry bulb, find the outdoor temperature where capacity crosses the load, and size supplemental heat from that balance point.

What a Manual S run returns

Ask for equipment selection in chat on a Pro or Team account and the answer comes back in six parts. Knowing the shape of it tells you what to demand from any Manual S, including one a competitor hands your customer.

Recommended size: Nominal tons and BTU/h for cooling equipment, or input and output BTU/h with the AFUE for a furnace or a boiler.
Acceptable range: The floor and ceiling in BTU/h for this load, the commercial sizes that fall inside it, and the wording of the rule that produced them.
Capacity check at design: Total, sensible and latent capacity at your outdoor design temperature against the matching load, each with a pass or fail, plus the derate factors used.
SHR check: Load sensible heat ratio against equipment sensible heat ratio, with a plain reading of what the gap means for humidity.
Heat pump detail: Heating capacity at the winter design temperature and the balance point, the outdoor temperature below which the unit can no longer carry the house alone. The free heat pump sizing page already returns the shortfall and the kilowatts of resistance heat behind it, for a unit whose 47°F rating you type in yourself. Picking that unit for you, and the balance point, are what the paid run adds.
Warnings and citations: Every assumption the run had to make, and the standard behind each limit with its version and the date it was read.

See what Pro includes

Manual S runs on Pro and Team · whole-house loads stay free

The Manual S sizing limits

Manual S sets a floor and a ceiling for every equipment class. The floor is the load itself. Without a ceiling a contractor can round up two sizes and call it safe. The rows below are the first-edition set from Table 1 of the ACCA review article by Wes Davis, and they are what the BuildSolver tool enforces today, while Manual S-2023 has moved several of them and renumbered the text. Section numbers in this list locate a rule in the first edition only; the 2023 edition carries these limits at N2.3.1 through N2.3.4, and its factors have their own section further down.

Air conditioners: load ≤ total capacity at design ≤ 1.15 × total cooling loadTable 1 gives 115 percent of total cooling load, pointing at section 3-4 of the first edition. The worked example in the same article reads the window as 100 to 115 percent of the cooling load.
Heat pumps, cooling dominant: total capacity at design ≤ 1.15 × total cooling loadFootnote 1 of Table 1 reads, heat pumps in a cooling dominant climate are allowed to be 115% of the cooling level. First-edition section 4-4.
Heat pumps, heating dominant: total capacity at design ≤ 1.25 × total cooling loadFootnote 2 of Table 1 reads, heat pumps in a heating dominant climate are allowed to be 125% of the cooling level. BuildSolver treats IECC zone 4 and colder as heating dominant, which is a BuildSolver assumption and not wording from the table.
Furnaces and boilers: output capacity ≤ 1.40 × total heating loadTable 1 gives 100 to 140 percent of total heating load for both, at first-edition section 2-2. Output capacity, not input, so the AFUE comes off first.
Electric supplemental heat: sized from the heat pump balance pointTable 1 gives no percentage for electric backup, only the words based on equipment balance point, at first-edition section 4-8. Emergency heat is left to local codes in section 4-9.
Dual fuel backup: furnace output ≤ 1.40 × total heating loadTable 1 puts the gas side of a dual-fuel system back on the 100 to 140 percent heating rule, at first-edition section 6-8.
Conditions the check runs at: same design conditions as the Manual J runThe asterisk under Table 1 reads, the size of the cooling equipment must be based on the same temperature and humidity conditions that were used to calculate the Manual J loads.

Four jobs run through the checks

Every figure below is the literal output of the BuildSolver Manual S engine for the loads and design temperatures shown. All four clear the total-capacity line, and three of them then fail somewhere else. That is the ordinary result, and the reason the procedure has more than one step.

Four Manual S selections with the acceptable window, the recommended size and the capacity delivered at design conditions
Job	Manual J load, BTU/h	Acceptable window, BTU/h	Pick	At design conditions
Houston, TXIECC zone 2A, hot and humidSplit air conditioner	35,500 total, 25,000 sensible, 10,500 latent	35,500 to 40,825	3 ton, 36,000	36,000 total, 28,080 sensible, 7,920 latent. Total and sensible clear, latent short.
Phoenix, AZIECC zone 2B, hot and drySplit air conditioner	32,000 total, 27,500 sensible, 4,500 latent	32,000 to 36,800	3 ton, 36,000	33,358 total, 25,087 sensible, 8,271 latent. Total clears, sensible 9 percent short.
Atlanta, GAIECC zone 3A, mixed and humidAir-source heat pump	30,000 total cooling, 22,000 sensible, 32,000 heating	30,000 to 34,500	2.5 ton, 30,000	30,477 total, 23,983 sensible, 6,494 latent. Total and sensible clear, latent short, heating finished by backup.
Chicago, ILIECC zone 5A, coldGas furnace, 80 percent AFUE	52,000 heating at a -3°F design day	52,000 to 72,800	80,000 input, 64,000 output	64,000 output, inside the window

Houston, TX

Three tons is the only commercial size that lands inside the window. Houston sits at the 95°F rating point, so the derate curve is 1.00 and the unit delivers its full 36,000 BTU/h against a 35,500 BTU/h load. Sensible clears with room to spare, and latent falls 2,580 BTU/h short. The generic unit moves 7,920 BTU/h of moisture against a 10,500 BTU/h latent load, and tonnage is not the lever that closes it: the window ends at 40,825 BTU/h, so 3.5 tons is already out of bounds on this load, and a bigger box would raise sensible capacity faster than latent anyway. On the Gulf coast that gap is the difference between holding the indoor design humidity the Manual J run assumed and a callback about a clammy living room. Bring the sensible heat ratio down with a different coil and airflow match, add standalone dehumidification, and confirm both against a real manufacturer table.

Phoenix, AZ

Here the desert does the damage. At a 108°F design day the model puts total capacity at 92.7 percent of the rating and sensible capacity at 89.3 percent, so a 3-ton unit that reads 36,000 on the label delivers 33,358 BTU/h total and only 25,087 BTU/h of sensible. The load is 86 percent sensible, so the sensible side is what governs, and 25,087 against 27,500 comes up 9 percent short. Buying a 3.5-ton unit would break the 36,800 BTU/h ceiling on total capacity, so the fix is a larger coil and more airflow on the same 3 ton condenser, and the corrected sensible capacity comes off the manufacturer's expanded performance data.

Atlanta, GA

Zone 3A is cooling dominant, so the ceiling stays at 115 percent and the window runs 30,000 to 34,500 BTU/h. Cooling governs the pick, and 2.5 tons is the only commercial size inside that window. Heating is then whatever that same unit happens to deliver: 23,400 BTU/h at the 23°F winter design against a 32,000 BTU/h heating load, which puts the balance point, where capacity crosses the load line, at 31°F. Below it the remaining 8,600 BTU/h comes from backup, about 3 kW of resistance heat. Running the winter on the heat pump alone would take roughly a 48,000 BTU/h unit at 47°F, and the cooling ceiling does not allow one here, so this house either accepts the strips or moves to dual fuel. Latent capacity is 6,494 BTU/h against an 8,000 BTU/h latent load, the same humid-climate squeeze Houston showed, in a milder form.

Chicago, IL

Fuel-fired equipment has no design-condition derate to apply, so the whole job is arithmetic on the 140 percent rule. Watch which side of the burner each number describes. The window of 52,000 to 72,800 BTU/h is output, and furnaces are sold by input, so every candidate has to be converted before it can be compared. An 80,000 BTU/h input furnace at 80 percent AFUE puts out 64,000, which is 123 percent of the load and lands in the window with room on both sides. Jump to the next bin, 100,000 input, and output goes to 80,000, over the ceiling. This is the case where the rule alone decides the answer.

Design temperatures are the 1 percent cooling and 99 percent heating dry bulbs from the BuildSolver metro dataset, derived from NREL TMY3 station files and cross-checked against the city examples in ACCA Manual J 8th Edition Chapter 7. Houston is station KIAH at 95°F cooling, Phoenix is KPHX at 108°F, Atlanta is KATL at 92°F cooling and 23°F heating, Chicago is KORD at -3°F heating. The free lookup on the design-temperature page takes a ZIP and answers with the worst case of that ZIP's climate zone, which runs more extreme than these station values.

Every job above except the Chicago furnace is sized on the cooling side, and those picks do not land on one number. The two air conditioners come out at nominal 3 tons and the Atlanta heat pump at 2.5. A heat pump picked that way is not also picked for winter, which is why the Atlanta row ends with backup heat rather than a bigger unit. Pull the manufacturer's expanded performance table for each coil before you quote, or run the selection in chat on a Pro account.

Standards and sources

ACCA Manual S, Residential Equipment Selection - the procedure itself. The current edition is ANSI/ACCA 3 Manual S-2023, the third; the second is ANSI/ACCA 3 Manual S-2014. ACCA describes the newest edition as carrying new expanded size tolerances for variable capacity heat pumps.
ACCA Addendum c public-review draft, March 6, 2025 - the source of the Manual S-2023 size factors quoted here. It consolidates the 2023 text with the 2024 addenda a and b, and its Tables N2.3.1, N2.3.2 and N2.3.4 carry the single-speed, two-speed, variable-capacity and dry-condition limits.
Reviewing HVAC Designs for Compliance with ACCA Manual S, Wes Davis, ICC eNews 2009 - Table 1 of this article, written by ACCA's manager of technical services and published by the International Code Council, is where every percentage on this page comes from, with the Manual S section number behind each row.
ACCA Manual J 8th Edition - the load calculation that feeds Manual S. The same article quotes the code requirement as heating and cooling equipment shall be 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.
AHRI Certified Directory - confirms that a given outdoor unit and indoor coil were certified as a pair and gives the efficiency ratings. The ACCA article is blunt about the limit of that data, saying AHRI directories should only be used to compare equipment efficiency ratings and that manufacturer expanded performance data should be used to select properly sized equipment. That sentence is the reason this page sends every capacity check to the manufacturer tables.
NREL/TP-5500-56354, Cutler et al. 2013 - the biquadratic capacity model BuildSolver uses to estimate cooling capacity away from the rating point. This is a BuildSolver modeling choice, not an ACCA requirement, and it stands in for the manufacturer's table only until you open the real one.
NEEP cold climate air-source heat pump database - the source BuildSolver averaged to build its heat pump heating derate curves. The averaging is a BuildSolver assumption, so a specific inverter unit can hold capacity far better than the curve suggests.

What this page does not do

Equipment selection runs in chat on Pro and Team, so the checks below are here to audit a selection, not to produce one.
No page and no chat answer picks a model number for you. Manual S narrows the field to a size and a set of checks; the specific outdoor unit and coil come from a manufacturer's catalog.
The capacity numbers in the examples come from a generic derate curve, not from any one manufacturer's expanded performance table. Two manufacturers' expanded performance tables at 95°F can put the same nominal 3 ton pair at different total capacities, which is why the check belongs in the catalog.
The BuildSolver tool applies the first-edition sizing window. Manual S-2023 is the current edition and the one the 2024 IRC references, so where a plan review works from the newer text, run the 2023 size factor yourself against the same capacity numbers.
Where equipment data is not supplied, the engine assumes a sensible heat ratio of 0.78 at the rating point. BuildSolver treats the working range for residential equipment as roughly 0.70 to 0.85, an assumption rather than a published figure, and it is wide enough to flip a latent check.
Altitude is applied to sensible capacity only, and only when the job's elevation is known. Combustion derate for fuel-fired equipment at altitude is not modeled at all.
Manual S is not the load calculation and does not correct a bad one. If the Manual J inputs were guessed, the equipment selection inherits the guess.
Preliminary sales-phase output. Not ACCA-approved software and not accepted for permit submission, where the engineer of record uses approved software and the manufacturer's data.

Where Manual S selections go wrong

Ceiling used as a target. The 115 percent line is the point past which a selection stops being acceptable. A pick that lands at 114 percent of the load runs shorter cycles than one at 102 percent, and in a humid climate the runtime you lose is the dehumidification you needed.
Capacity read off the nameplate. Nominal tonnage is a catalog label tied to a rating point, and the Phoenix job above shows what happens to it on a 108°F afternoon. Manual S asks for capacity at the conditions the load was figured at, which is a different number in every climate.
Stopping once total capacity fits. Total, sensible and latent are three separate checks. Houston passes total and sensible and fails latent. Phoenix passes total and falls short on sensible. A selection that only clears the total line has answered one third of the question.
Sizing a heat pump on the 47°F rating. That rating describes a mild autumn day. What matters in January is capacity at the heating design temperature, the balance point below it, and how many kilowatts of backup sit behind that. Skip that and the strips carry the house at resistance-heat prices.
Selecting from the certification directory. The directory confirms that a condenser and a coil were certified together and lets you compare efficiency ratings. ACCA's own guidance says the selection itself belongs to the manufacturer's expanded performance data, because directory ratings describe one test point rather than your design day.

What changed in Manual S-2023

Manual S-2023 rewrote the size factors by compressor type. A size factor is capacity at design divided by the Manual J load, and it is the term the standard uses, so this section states every limit in that form: 1.15 rather than 115 percent. Single-speed equipment keeps the 1.15 ceiling, and four things arrive that the first edition had no answer for, a 1.20 ceiling on loads of 24,000 BTU/h and under, a higher ceiling for staged and inverter compressors, a 0.90 floor in place of the old 1.00, and a separate track for very dry climates. The factors are read from ACCA's Addendum c public-review draft of March 6, 2025, which consolidates the 2023 text with the 2024 addenda.

Single-speed, over 24,000: 0.90 ≤ total size factor ≤ 1.15Table N2.3.1. The ceiling is unchanged from the first edition, while the floor drops from the load itself to 0.90 of it.
Single-speed, 24,000 and under: 0.90 ≤ total size factor ≤ 1.20Table N2.3.1 sets the ceiling at 1.20 for loads of 24,000 BTU/h and under, 0.05 above the single-speed limit, because the commercial size ladder is coarse down there.
Two-speed: 0.90 ≤ total size factor ≤ 1.25Table N2.3.1 sets the two-speed ceiling at 1.25. A second stage buys runtime at part load, so the limit sits 0.10 above single-speed.
Variable capacity, simplified: 0.90 ≤ total size factor ≤ 1.30Section N2.3.4 reads, the total cooling size factor shall not exceed 1.30, and shall not be less than 0.90, with the latent size factor equal to or greater than 1.00.
Sensible and latent floors: sensible ≥ 0.90 × load; latent ≥ 1.00 × loadSection N2.3.1 states that sensible capacity divided by the sensible load must be at least 0.90 and latent divided by latent must be 1.00 or greater. The 2023 table sets no latent ceiling.
Dry sizing condition: total capacity ÷ (load + 6,000) ≤ 1.00Section N2.3.2 replaces the standard ceiling once the Manual J sensible heat ratio reaches 0.95. That is the dry-climate case, a house with almost no moisture to remove, where the ordinary limit rules out every size on the shelf.

The BuildSolver tool has not moved to those factors. It still applies the first-edition window, a 100 percent floor with a 115 percent ceiling for air conditioners and 125 percent for heat pumps in heating dominant climates. For a single-speed job over 24,000 BTU/h the two editions agree at the top, so the tool answer stands. For staged equipment, inverter equipment, or a load at 24,000 BTU/h and under, take the capacity numbers the tool reports and divide them by the load yourself, then compare against the factor in the row above.

Our write-up on the 2023 oversizing limit runs the arithmetic through two real selections, including a Denver house that only passes under the dry sizing condition. ACCA's Addendum c draft carries the tables themselves.

Why the manufacturer's table beats any generic model

The ACCA review article works a 30,000 BTU/h cooling load with 22,000 BTU/h sensible and 8,000 BTU/h latent. At the 95°F design day the 2.5 ton unit delivers 29,000 BTU/h at a 67°F entering wet bulb and 26,770 BTU/h at 63°F, short of the 30,000 BTU/h load either way. The 3 ton pair clears it with 31,510 BTU/h total under the 34,500 BTU/h ceiling and 23,000 BTU/h sensible over the sensible load.

Hand the same load to the BuildSolver engine and it clears a 2.5-ton unit, because its derate curve is an industry average rather than one manufacturer's coil. In the Atlanta job above that curve puts a 2.5-ton heat pump at 30,477 BTU/h against a 30,000 BTU/h load. The ACCA numbers in the paragraph above show how wide that can run, with a real 2.5 ton table falling under a 30,000 BTU/h load that the generic curve clears. The gap is the honest scope line for sales-phase sizing, and it is why the engine attaches a warning to every defaulted capacity it uses.

ACCA is direct about which data to trust, in the sentence quoted in the sources list above. Use the directory to prove the condenser and the coil were certified together, then open the catalog and read the number at your design day. Pull the article and read Table 1, then confirm the limit against the current Manual S-2023 wording before you rely on it for inverter equipment.

Questions contractors ask about Manual S

Why is there no Manual S calculator on this page?

Equipment selection is one of the paid features. It runs on the Pro and Team plans, where the chat can call the Manual S tool directly, and putting the same oversize limits and capacity checks behind a public form would move a paid tool into the free tier. The whole-house Manual J load that Manual S starts from is free in chat on every plan, with no signup for three questions a day.

What does Manual S do that Manual J does not?

Manual J answers how much heating and cooling the building needs at design conditions. Manual S answers whether a particular piece of equipment delivers that much at those same conditions, and whether it splits the delivery between temperature and moisture the way the house needs. Run the Manual J first. Manual S has nothing to check until the load exists.

How much larger than the load can the equipment be?

Two answers, depending on the edition your reviewer works from. Table 1 of the first edition prints percentages: air conditioners at 115 percent of the total cooling load, heat pumps at 115 percent in a cooling dominant climate and 125 percent in a heating dominant one, furnaces and boilers at 100 to 140 percent of the heating load, with the load itself as the floor. Manual S-2023 prints size factors instead, holding single-speed equipment at 1.15, lifting it to 1.20 where total cooling comes in at 24,000 BTU/h or below, allowing 1.25 for two-speed and 1.30 for variable capacity under the simplified procedure, and dropping the floor to 0.90. BuildSolver applies the first-edition window.

Which edition does the BuildSolver tool use?

The first edition, with its window running from 100 percent of the load up to the ceilings listed in the sizing-limit rows above. That holds equipment to the full load where Manual S-2023 accepts a 0.90 floor, and matches the newer ceiling for single-speed gear. Where the difference matters is staged and inverter equipment and loads at or under 24,000 BTU/h, so on those jobs take the capacity numbers the tool reports and apply the 2023 size factor to them yourself.

Why is nominal tonnage not the same as capacity?

Nominal tonnage is a size class tied to a standard rating point. Capacity falls as outdoor air gets hotter for cooling and as it gets colder for heating, and the sensible and latent halves move at different rates. In the Phoenix example a nominal 36,000 BTU/h unit delivers 33,358 BTU/h total at a 108°F design day, and its sensible capacity falls further than its total.

Should I select from the AHRI directory or from the manufacturer's tables?

Use the directory to confirm the outdoor unit and the indoor coil were certified as a pair and to compare efficiency ratings. Select from the manufacturer's expanded performance data. The ACCA article states the reason directly, that directory test conditions represent a very small geographic area, so they rarely match the design day you are sizing for.

How does Manual S size the backup heat on a heat pump?

Table 1 gives no percentage for electric supplemental heat. It says the sizing is based on the equipment balance point, below which the heat pump can no longer carry the house on its own, and it sends emergency heat to local codes. Dual-fuel backup goes back on the furnace rule of 100 to 140 percent of the heating load.

Can I submit BuildSolver output for a permit?

No. This is sales-phase preliminary sizing meant for the first conversation with a customer. Where a jurisdiction requires ACCA-approved software, the engineer of record runs the selection there with the manufacturer's data. What BuildSolver gives you is a preliminary size with every limit it applied written out.

What does the Manual S step cost?

It is part of the Pro plan and the Team plan, along with room-by-room Manual J, full Manual D and the branded PDF. Current prices are on the pricing page. Design temperatures by ZIP, whole-house cooling and heating loads, dew point, psychrometrics and a simplified duct size stay free for anonymous visitors as well as free accounts. That duct number covers a single run; a full Manual D is on the Pro and Team plans.

Used on the same job

Where the free line sits

Describe a house in chat and the assistant runs the whole-house Manual J cooling and heating loads, pulls design conditions from the ZIP, and hands back the numbers Manual S needs as input. That part costs nothing, for anonymous visitors at three questions a day and free accounts at ten a day.

Equipment selection itself, room-by-room Manual J, full Manual D and the branded customer PDF sit on the Pro and Team plans. Ask for a Manual S on a free account and the assistant says which plan it belongs to, points at the pricing page, and offers the rough tonnage the free load already supports instead of running the selection. The product page for the Manual S tool walks one Houston job through the worksheet the Pro run produces.

Run the load in chat