§ Methodology · Tool No. 01
How the calculator
actually works.
Every number on the results page comes from this document. If you disagree with a constant or a data source, this is where to argue.
§ I.
Energy load model
The calculator uses the degree-day method — the simplest defensible way to estimate annual HVAC consumption without running a full Manual J or hourly simulation:
annual_heating_btu = HEATING_FACTOR × sqft × HDD_base65 annual_cooling_btu = COOLING_FACTOR × sqft × CDD_base65
Where HEATING_FACTOR = 8.8 and COOLING_FACTOR = 3.0 (BTU per sqft per degree-day). These are calibrated against US Residential Energy Consumption Survey (RECS 2020) averages. Expected accuracy: ±30–40% without bill calibration; ±15–20% with it.
For homes substantially better or worse insulated than the RECS average, the bill-calibration step is the fix — it scales our estimate to match your actual annual spend.
§ II.
HSPF2 climate derate
Manufacturer HSPF2 ratings are derived from AHRI 210/240 Region IV (~8,300 HDD). They systematically overstate delivered heating efficiency in colder climates because heat pump COP drops at low outdoor temperatures. We apply a multiplicative derate by the first digit of your IECC climate zone:
| Zone | Derate | Effect |
|---|---|---|
| 1–3 (hot) | 1.00 | No derate |
| 4 | 0.95 | -5% |
| 5 | 0.90 | -10% |
| 6 | 0.85 | -15% |
| 7 | 0.80 | -20% |
| 8 (arctic) | 0.75 | -25% |
The NEEP Cold-Climate Heat Pump list publishes per-model COP at 47°F / 17°F / 5°F / -15°F. We're planning to replace the zone-digit derate with per-model curves in a future release.
§ III.
Data sources
- Climate: NOAA NCEI Climate Normals 1991–2020 (station-level HDD/CDD), joined to ZIPs via US Census ZCTA-to-county crosswalk and the nearest-station assignment.
- IECC zones: DOE Building America county-level climate-zone map (philngo/2014 dataset).
- Electricity rates: EIA Form 826 residential state averages, refreshed monthly. ~2 month publication lag.
- Heating oil & propane: EIA weekly retail petroleum prices (national average). Heating season only; fallback to prior reading out of season.
- Natural gas: EIA monthly state-level residential prices (currently using national-average fallback where the feed returns empty).
§ IV.
What we don't model
- Time-of-use rates. EIA state averages mask 2–3× variation within a state. If your utility has tiered or peak/off-peak pricing, the bill-calibration field corrects for this at the aggregate level.
- Latent / dehumidification load. In humid cooling-dominant climates (FL, TX, LA) real cooling consumption runs 20–40% higher than our model due to latent load.
- Duct losses. The calculator assumes ductless or efficient ducting. Leaky ducts in an unconditioned attic/crawlspace can add 20–30% to existing-system cost — bill calibration partially corrects.
- Setpoint schedules. Assumed constant setpoint. If you set back aggressively, consumption will be lower than modeled.
- Solar gain and thermal mass. Modeled only through the load factors above.
§ V.
Confidence by input
Our model's relative accuracy (existing vs new) is much better than its absolute accuracy. The savings-delta number is usually within ±15% of reality because the same load factors apply to both sides. The absolute cost estimates are where bill calibration matters most.
For highest accuracy:
- Enter your actual last-12-month HVAC bill
- Enter your specific unit's SEER2 / HSPF2 (not defaults)
- Use per-zone mode if different rooms have different systems
§ VI.
Incentives note (2026)
The federal §25C Energy Efficient Home Improvement Credit was repealed effective December 31, 2025 (One Big Beautiful Bill Act, late 2025). Heat pumps installed in 2026 do not receive the $2,000 federal credit — any calculator still showing it is out of date.
State and utility rebates are now the entire incentive stack. Mass Save, NYSERDA NY Clean Heat, Efficiency Maine, BPA Northwest, and California's TECH Clean remain substantial.