Three-layer Manual J calculator verification framework plus public audit surfaceThree verification layers and a public audit surface for the Manual J calculator. Layer 1 ACCA test cases: 8 reference homes published by ACCA with expected loads, target band plus or minus 5 percent on each case. Layer 2 peer software comparison: cross-check against Wrightsoft, Cool Calc, and other Manual J implementations on identical inputs, target band plus or minus 3 to 5 percent per the comparison literature. Layer 3 real-home validation: compare calculator output to inferred peak demand from utility data and datasets like NREL ResStock, target band plus or minus 15 percent consistent with Manual J input-uncertainty floor. Audit surface (publish step): methodology, known limitations, corrections log, and audit contact published openly so anyone can verify, criticize, or report bugs.Three verification layers and a public audit surface1ACCA test cases
8 reference homes published by ACCA with expected cooling and heating loads. The verification target for an implementation is ±5% on each case.
2Peer software3 tools
Cross-check against Wrightsoft Right-J and Cool Calc on identical inputs. The comparison-literature target band is ±3-5%, reflecting allowed Manual J assumption variations.
3Real homes
Compare calculator output to inferred peak demand from utility data and datasets like NREL ResStock. Target band: ±15% on most homes, consistent with Manual J's input-uncertainty floor.
4Publish
Methodology, known limitations, corrections log, and audit contact published openly. Anyone can run a verification layer and report results.
Methodology, transparency, and an audit contact. The alternative is unverified claims.
We are not ACCA-approved software. We are a methodology-consistent Manual J implementation that you can audit and criticize openly.

How to Audit Our Manual J Calculator

The Manual J 8th Edition methodology this calculator implements, the three-layer verification framework it is designed against, and the public audit channel for any reader who wants to check our work.

Jonathan Stowe

Reviewed June 10, 2026

Published May 18, 20269 min read

We run a free Manual J-style load calculator. Free calculators have a reputation, mostly bad — many are oversimplified, some are wrong, and very few publish their methodology. This page exists because we think calculators that affect equipment sizing decisions worth thousands of dollars should be auditable, and we want to make that auditing easy.

Below we cover what the calculator implements (Manual J 8th Edition methodology, ASHRAE 169 climate data, documented infiltration and solar gain models), the three-layer verification framework it is designed against (ACCA reference cases, peer-software comparison, real-home utility data), the public audit surface that lets any reader check our work, and what the calculator does not yet handle. We do not claim accuracy results we have not run.

We are not ACCA-approved software, which is a formal designation we don't have; we are a methodology-consistent Manual J implementation that you can audit, criticize, and improve. For broader context, our methodology covers the approach across all calculators on the site.

What Manual J Is (Brief)

Manual J is the residential heating and cooling load calculation standard published by ACCA (Air Conditioning Contractors of America).[1] See ACCA Manual J 8th Edition for the source document. It calculates the design heating load (BTU/hr at the 99% heating design temperature) and the design cooling load (BTU/hr at the 1% cooling design temperature) for a residential building based on envelope characteristics, climate, internal gains, and occupancy.

For the full treatment, see the Manual J methodology overview article. The Manual J-style load calculator implements this methodology with simpler inputs than commercial software, designed for planning rather than permit submission.

We follow Manual J 8th Edition (ACCA, 2016). This is the current published version as of 2026. Methodology updates and addenda are incorporated as ACCA publishes them.

Why Verification Matters

A wrong load calculation propagates downstream: equipment selection wrong, ductwork sized wrong, Manual S equipment selection defective, performance suffers for the life of the system. The cost of a wrong load is not the calculator's $0 price tag; it is the $5,000-$15,000 equipment purchase and 15-20 years of operating cost based on that purchase.

Three failure modes if a calculator is systematically wrong:

  1. Oversize systematically. Results in short cycling, humidity control problems, accelerated equipment wear, and wasted capital
  2. Undersize systematically. Results in comfort failures on design days, heavy auxiliary heat use in cold climates, and undersized cooling on the worst summer days
  3. Methodology errors that work in normal cases but fail at edge cases. Hardest to spot; surface only under specific conditions

Field-installed HVAC has well-documented sizing errors of 30-50% in many cases (DOE field studies and HVAC efficiency research consistently show oversizing as the dominant pattern). A free calculator that's wrong perpetuates the problem. A free calculator that publishes its methodology and audit channel can help fix it. Heat pump sizing context and home heat loss methodology cover the downstream consequences of getting Manual J wrong.

This page documents the methodology, the verification framework, and the audit surface so any reader can check the work.

The Methodology the Calculator Implements

For the technically curious or auditors reviewing the implementation:

Inputs and how they're used. Building envelope (areas, R-values, U-factors), orientation, climate zone (mapped to nearest ASHRAE Standard 169 location), infiltration assumption (default or user-entered blower-door data), occupancy and internal gains schedule.[4]

Climate data. ASHRAE 169 design temperatures, 1% cooling and 99% heating. Embedded as a static lookup table for ~5,000 US locations. Updated annually as ASHRAE 169 updates.

Solar gain calculations. ASHRAE-recommended solar heat gain factors by orientation (N, NE, E, SE, S, SW, W, NW) and latitude.[3] Glazing types: clear single, clear double, Low-E double, Low-E triple. Different SHGC values per glazing type.

Infiltration. Blended Sherman-Grimsrud / ASHRAE enhanced model. Default infiltration rates by home age (matches ResStock defaults); user can override with blower-door ACH50 data.

Internal gains. ANSI/RESNET/ICC 301 default occupancy schedule. Standard occupancy assumed; non-standard high-density use (home offices, gyms) requires manual adjustment via the calculator's advanced settings.

Duct gains and losses. Manual J Appendix-D approximation for ducts in unconditioned space. Detailed Manual D handling not yet automated.

Software implementation. Pure-JavaScript front-end calculation, deterministic (same inputs produce same outputs every time), no random factors, no opaque transformations. Each calculation step uses documented formulas with version references in the source code.

The Verification Framework (Three Layers)

A Manual J implementation can be checked against three independent layers. The calculator is designed against all three.

Layer 1: ACCA published reference cases. ACCA publishes the Manual J 8th Edition Approval Test Suite, a set of 8 reference homes with documented inputs and expected output loads.[2] Software vendors seeking ACCA approval run these cases as part of the formal certification process. The cases cover different house types, climate zones, and construction styles (1-story ranch zone 3, 2-story colonial zone 4, split level zone 5, small bungalow zone 2, large modern zone 6, passive solar zone 4, manufactured home zone 7, walkout basement zone 5).

Verification target band: ±5% on each case. This is the strongest verification baseline because ACCA defines both the inputs and the expected outputs, so there is no input ambiguity.

Layer 2: Peer-implementation comparison. A Manual J implementation can be cross-checked against established ACCA-approved software (Wrightsoft Right-J, Cool Calc Manual J, reference Manual J 8 spreadsheet implementations) on identical inputs. Discrepancies that surface trace to allowed-by-Manual-J assumption differences (default infiltration when no blower-door data is provided, internal gains schedule selection, solar gain factor interpretations at edge orientations) rather than calculation errors. The Manual J comparison literature documents inter-implementation variance on identical inputs at typically ±3-5%, which is the verification target band for this layer.

Layer 3: Real-home validation. The calculator's output can be compared to actual peak demand inferred from real-home utility data (with baseline non-HVAC load subtracted) and from datasets like NREL ResStock[6] and the BuildingSync standardized energy audit schema.[5] This is the hardest layer because it combines calculator accuracy with the inherent uncertainty in real-home input data and the difficulty of measuring true peak HVAC demand from utility data. The target band is ±15% on most homes, consistent with Manual J's inherent input-uncertainty floor documented in the building-science literature. Outliers beyond ±15% typically trace to user input errors (square footage misreporting, insulation assumptions that don't match the actual home) rather than calculator errors.

These three layers are what an honest verification looks like for a Manual J implementation. They are the framework the calculator is designed against. Any reader running the calculator against any layer can submit results through the audit channel; verified discrepancies that violate the methodology are bugs and get logged in the corrections record.

The Audit Surface

Five layers of audit artifacts available for public reviewVertical stack of five audit artifacts that anyone can review. Bottom layer methodology documentation: what the calculator implements and the verification framework it targets, published openly. Layer 2 source code and formulas: plain-language formula references with calculation logic transparent in the source repository. Layer 3 known limitations: documented openly and updated when discovered. Layer 4 corrections log: public record of material updates triggered by reader-reported errors or methodology disagreements. Top layer audit contact: anyone can request review or report a bug, responses within a few business days. Anyone can audit at any level.Five layers anyone can auditEach layer is publicly accessible; report bugs and we respondAudit contactRequest a review of our work or report a bug; responses within a few business days.v1.2Corrections logPublic record of material updates triggered by reader-reported errors or methodology disagreements.!Known limitationsDocumented openly and updated when discovered. Honest about where the calculator is not strong.</>Source code & formulasPlain-language formula references; calculation logic transparent in the source repository.Methodology documentationWhat the calculator implements and the verification framework it targets, published openly.Documented criticism finds bugs. Fixed bugs help everyone. We'd rather be accountable than pretend infallibility.
The audit trail is the alternative to unverified claims. We publish at every level so any reader can check our work.

What is publicly available for audit:

  • The methodology documentation. This page describes what the calculator implements and the verification framework it targets. Anyone can check whether the implementation matches the stated methodology by reading the source code alongside this page.
  • The formulas and source code. Plain-language formula references in the source code repository, no obfuscation. Each calculation step uses documented formulas with version references.
  • The known-limitations list. Updated openly when discovered. Honest about where the calculator is not strong.
  • The corrections log. Material updates triggered by reader-reported errors or methodology disagreements are documented at /corrections/ with date, original claim, corrected claim, and (with reporter consent) attribution.
  • The audit channel. Any reader can run a verification at any of the three layers above and report results through the contact channel.

The audit surface is the alternative to unverified marketing claims. Any reader can check the implementation against the stated methodology at any time; documented discrepancies surface as bug reports; fixed bugs get logged in the corrections record.

What We Don't Claim

The calculator is not perfect, and we don't pretend it is. It uses the same methodology and physics as ACCA-approved software, but it lacks the formal certification, the granular room-by-room input model, and the integration with Manual D and Manual S that commercial tools provide. It approximates non-rectangular geometries. It assumes standard occupancy schedules.

It estimates duct losses rather than measuring them. These are real limitations and they affect real users. If your situation matches one of them, use ACCA-approved software or a certified contractor instead. If it doesn't, this is a useful tool for planning and evaluation. But check the results against your own intuition and consult a professional before installing equipment.

Known limitations of our Manual J calculatorList of five known limitations of our calculator with each item including a warning indicator and brief description. Limitation 1 non-standard building geometries: round, A-frame, earth-bermed homes use simplified rectangular approximation, expect plus or minus 15 to 20 percent accuracy. Limitation 2 multi-zone diversity: calculator computes whole-house loads, per-zone analysis requires individual Manual J runs per zone we don't automate. Limitation 3 internal load schedules: standard occupancy assumed, high-density use like home offices or gyms requires manual adjustment. Limitation 4 ductwork in unconditioned space: duct loss factors estimated not measured, detailed Manual D required for accurate duct loss. Limitation 5 older homes with unknown insulation: inputs depend on user knowledge, uncertainty propagates through the result. These are honest limitations we publish; they affect accuracy in specific scenarios.Where our calculator falls shortHonest limitations, published openly!Non-standard geometriesRound, A-frame, earth-bermed homes use simplified rectangular approximations.±15-20% expected!Multi-zone diversityWe calculate whole-house loads. Per-zone analysis requires individual runs per zone.whole-house only!Internal load schedulesStandard occupancy assumed. Home offices, gyms, server rooms need manual adjustment.tunable upward!Ductwork in unconditioned spaceDuct loss factors estimated, not measured. Manual D is required for accurate duct loss.approximation only!Older homes, unknown insulationInputs depend on user knowledge. Uncertainty propagates through the result.GIGOIf your situation matches one of these, use ACCA-approved software or a certified contractor instead.
These are real limitations. They affect real users in specific scenarios. We publish them rather than hide them.

Honest limitations published openly:

  • Non-standard building geometries. Round houses, A-frames, earth-bermed homes use simplified rectangular approximations. Expect ±15-20% accuracy in those cases
  • Multi-zone diversity. The calculator produces whole-house loads. Per-zone analysis requires individual Manual J runs per zone, which the tool does not yet automate
  • Internal load schedules. Standard occupancy assumed. Home offices, gyms, server rooms, dense workshops require manual adjustment beyond the calculator's default inputs
  • Ductwork in unconditioned space. Duct loss factors estimated, not measured. Detailed Manual D is required for accurate duct loss in tight or high-performance designs
  • Older homes with unknown insulation. Outputs depend entirely on user inputs. Garbage in, garbage out

We are explicitly NOT ACCA-approved software. That's a formal designation involving certification fees and process; we don't have it. We are a planning-grade tool.

When in doubt, use ACCA-approved software or a certified contractor. The ACCA-compliant Manual J path runs through certified vendors, not free web calculators.

How to Audit Us

Any reader can audit the calculator at any of the layers in the verification framework. Manual J audit can mean comparing the calculator against your own Manual J inputs and expected outputs, against another ACCA-approved tool, against published reference cases, or against your own home's measured energy data.

How to verify HVAC load calculation against this tool:

  1. Run the calculator with your specific inputs
  2. Compare to your expected output against your own Manual J calculation, another calculator, or rule-of-thumb sanity check
  3. Email the audit channel with discrepancies you can't explain
  4. Review the methodology section above to check the implementation against documented assumptions
  5. Check the corrections log at /corrections/ for recent changes that might affect your case

What we commit to:

  • Responding to documented audit reports within a few business days
  • Fixing verified bugs and logging the fix in the corrections record
  • Discussing methodology disagreements openly (different defaults aren't bugs; the choice is explained against the cited standard)
  • Crediting reporters who improve the tool (unless they prefer anonymity)

The audit contact runs through the contact page. Include your inputs as entered, the calculator's output, what you expected and why, and any third-party calculator outputs you ran for comparison.

What counts as a verified bug: a documented case where the calculator produces a result that violates Manual J 8th Edition methodology, ASHRAE physics, or its own published reference case behavior. What counts as a methodology disagreement rather than a bug: a different infiltration default than another calculator (both defensible under Manual J), a different solar gain factor table for the same orientation (ASHRAE publishes multiple acceptable tables), or differences in rounded final results.

This is the alternative to unverified claims. Published methodology, bug reporting, visible accountability, and an open audit channel are the trust mechanisms a free calculator can offer; "trust us" is not.

Frequently asked questions

Is your Manual J calculator ACCA-approved?
No. ACCA-approved Manual J software requires formal ACCA certification, which is a legal and business designation, not a technical accuracy claim. ACCA-approved software (Wrightsoft, Cool Calc, Elite) is certified for permit submissions. Our calculator is a Manual J-style tool that follows Manual J 8th Edition methodology and is designed against the ACCA published reference cases as the verification target. For permit-grade Manual J in a specific home, use ACCA-approved software or a certified contractor.
How accurate is your Manual J calculator?
The calculator targets three verification bands derived from the Manual J comparison literature and Manual J's inherent ±10-15% input-uncertainty floor on real homes: ±5% against the 8 ACCA published reference cases, ±3-5% against peer Manual J implementations on identical inputs, ±15% against inferred peak demand from real-home utility data. These are the verification target bands the calculator is built against, not published completed-runs results. Any reader running the calculator against any of these layers can report findings through the audit channel; verified discrepancies that violate the methodology are bugs and get logged in the corrections record.
Why does Manual J have an inherent uncertainty band?
Manual J calculates design loads from inputs about insulation, infiltration, window area, climate, occupancy, and equipment. Several inputs are uncertain in practice: infiltration is hard to measure exactly without a blower door test, insulation varies within the home, occupant behavior shifts loads. Even with perfect math, these input uncertainties propagate to ±10-15% in the calculated load. The calculator is no better than its inputs.
Can your calculator replace a permit-grade Manual J?
For planning, evaluation, and education, yes. For permit submission or warranty-grade equipment sizing, no. Most jurisdictions require ACCA-approved software output or a sealed Manual J from a certified contractor. Use our calculator to understand your loads, evaluate contractor quotes, and plan upgrades, then verify with certified software when it is time to install.
How can I report a bug or audit your math?
Email the contact address on the contact page with: your inputs as entered, the calculator output you received, what you expected and why, and any third-party calculator outputs you ran for comparison. We respond within a few business days. Verified bugs are fixed; verified disagreements about methodology defaults are discussed openly with citation. The corrections log records material changes that affect calculator output.
What's the difference between our calculator and Wrightsoft?
Wrightsoft is ACCA-approved, permit-grade software with paid licensing, comprehensive room-by-room input, and contractor-grade features (job tracking, proposals, Manual D and S integration). Our calculator is a free, planning-grade tool with simpler inputs, designed for homeowners who want a methodology-consistent answer for evaluation and planning. Per the academic Manual J comparison literature, implementations are designed to agree within ±3-5% on identical inputs; that is the verification target this implementation works toward.
Does your calculator handle every type of home?
It handles standard residential geometries (single-family, ranch, two-story colonial, split-level, manufactured) in IECC climate zones 1-8. Non-standard geometries (round houses, A-frames, earth-bermed) use simplified rectangular approximations with reduced accuracy (±15-20% expected). Commercial buildings, multifamily units, and additions or renovations have their own methodologies (Manual N for commercial; specific Manual J procedures for additions) that our calculator does not yet handle.
Do you use the latest ACCA Manual J version?
We follow Manual J 8th Edition (ACCA, 2016). This is the current published version as of 2026. ACCA periodically releases addenda and clarifications which are incorporated as published.
Are your results consistent with ASHRAE methodologies?
Yes, by construction. Manual J uses design conditions from ASHRAE Standard 169 (climatic data) and psychrometric principles from ASHRAE Fundamentals. Our calculator uses the same underlying ASHRAE data and physics. A Manual J implementation that uses the same ASHRAE inputs produces methodology-consistent output for residential applications.
Why publish your methodology openly?
Two reasons. First, it is the right way to operate a calculator that homeowners and contractors rely on; transparency builds trust and lets users audit the implementation against primary sources. Second, it improves the tool: published methodology gets criticism, criticism finds bugs, fixed bugs help everyone. We would rather be visibly accountable than pretend infallibility.

Sources

  1. 1. Manual J: Residential Load Calculation, 8th Edition (ANSI/ACCA 2 Manual J - 2016), Air Conditioning Contractors of America, 2016 (accessed 2026-05-18)
  2. 2. Manual J 8th Edition Approval Test Suite, Air Conditioning Contractors of America, 2016 (accessed 2026-05-18)
  3. 3. ASHRAE Handbook of Fundamentals 2021, ASHRAE, 2021 (accessed 2026-05-18)
  4. 4. ANSI/ASHRAE Standard 169-2021: Climatic Data for Building Design Standards, ASHRAE, 2021 (accessed 2026-05-18)
  5. 5. BuildingSync Standardized Schema for Energy Audits, US National Renewable Energy Laboratory / DOE, 2023 (accessed 2026-05-18)
  6. 6. ResStock: Residential Energy Use Modeling, US Department of Energy / NREL, 2024 (accessed 2026-05-18)
Jonathan Stowe

Reviewed June 10, 2026