Three quantities you need first
1. Annual electricity consumption
Pull your most recent 12-month bill history from your utility's online portal (every major US utility now offers this; it is often a CSV download). Sum the kWh column. This is your annual consumption. The national average is about 10,800 kWh per residential customer per year, with wide variation: California averages closer to 6,500 kWh due to mild climate; Louisiana and Texas push 14,000 kWh due to air conditioning load.
If you have made or plan major load changes — installing a heat pump, electrifying a hot water heater, buying an electric vehicle — adjust upward. An EV driven 12,000 miles per year typically adds 3,000–4,000 kWh of household consumption.
2. Peak sun hours at your location
A peak sun hour is one hour at which solar irradiance averages 1,000 W/m². That is the standard test condition used to rate a solar panel's nameplate output. If your location averages 5.5 peak sun hours per day, a 1 kW (nameplate) system will produce approximately 5.5 kWh on an average day — before applying real-world losses.
State-level averages range from about 3.0 in Alaska to 6.5 in Arizona and New Mexico. NREL's National Solar Radiation Database is the canonical source; our calculator pulls population-weighted state averages from it.
3. Performance ratio
A real installed system never produces the nameplate-times-sun-hours figure. The performance ratio (PR) captures everything between the DC nameplate and the AC kWh that lands on your bill:
- Inverter conversion losses (typically 95–98% efficient).
- DC and AC wiring losses (1–3%).
- Soiling (dust, leaves, pollen — 2–5% in most climates, higher in agricultural areas).
- Module temperature derating (silicon loses about 0.4% per °C above 25 °C; on a 60 °C rooftop, that's a 14% loss before any other factor).
- Light-induced degradation and panel mismatch (2–3%).
The industry default PR is 0.80, codified in NREL's PVWatts engine and used throughout this site. Premium installations with cool climates and oversized inverters can achieve 0.83–0.86; hot, dusty, or partially shaded sites may fall to 0.70.
The sizing equation
system_size_kW = annual_kWh / (peak_sun × 365 × performance_ratio)
Worked example: a household in Texas consuming 14,000 kWh/year at 5.3 peak sun hours and PR 0.80 needs:
14,000 / (5.3 × 365 × 0.80) = 14,000 / 1,547.6 = 9.05 kW DC
That is approximately 23 × 400 W panels — a typical 9 kW residential installation in the Texas market.
Should you size to 100% offset?
Not always. There are good reasons to undersize, and several places where oversizing is actively destructive:
- Net metering rules. Many states only allow you to "size to load" — the utility will not interconnect systems projected to produce more than your historical consumption. Even where allowed, surplus annual production typically receives no compensation at year-end "true-up".
- Roof orientation forces oversize. A north-facing roof in the Northeast may need a 20% larger system to deliver the same annual production as a south-facing roof of the same square footage.
- Future load growth. If you plan to add an EV or heat pump within 5 years, size the system today for the future load — adding panels later is disproportionately expensive on a per-watt basis.
- Diminishing returns on the last kilowatts. The first 4 kW pay back the fastest; an additional 2 kW that produces marginal surplus during midday low-value hours may not be economical, especially under net billing.
Why panel wattage isn't the right unit
Manufacturers compete intensely on per-panel wattage — last decade's premium 300 W panels are this decade's mid-range 400 W panels. But what matters is the system's total nameplate capacity and its area-adjusted efficiency. A 9 kW system made of 22 × 410 W panels at 20.5% efficiency and a 9 kW system made of 25 × 360 W panels at 18.5% efficiency produce essentially identical energy. The difference shows up in roof area required, not in output.
The relevant trade-off:
- High-efficiency panels (Maxeon, REC Alpha, Q.PEAK DUO): premium price, less roof area used. Right answer when roof space is the binding constraint.
- Standard-efficiency panels (Jinko, Trina, LONGi): lower cost per watt, more area used. Right answer when you have plenty of unshaded south-facing roof.
Site-specific accuracy: use PVWatts
The four-input model on our home page produces a reasonable estimate, but it cannot model your specific roof's geometry. For a free site-specific calculation that does, use NREL's PVWatts calculator. PVWatts takes:
- An exact address (down to building level via Google Maps).
- Roof tilt and azimuth (the orientation of the slope, measured in degrees from south).
- Module type (standard, premium, thin-film).
- Loss factors that you can tune individually rather than rolled into a single PR.
For a homeowner doing due diligence on an installer quote, a 5-minute PVWatts run is the single most valuable cross-check available. If the installer's projected first-year production differs from PVWatts by more than 10%, ask why.
Last reviewed May 2025. PVWatts and NSRDB references verified against the current NREL documentation.