What solar output actually depends on
A solar array's energy production is set by three things and nothing else: how much DC capacity you installed, how much sunlight lands on the plane of the modules, and what fraction of that sunlight survives the trip to your meter. Everything a designer argues about — tilt, azimuth, module brand, inverter topology, string length — reaches the answer through one of those three terms.
The middle term is the one most people get wrong. Peak sun hours is not the number of hours the sun is up. It is the daily solar energy arriving on one square metre of the array, expressed as the number of hours of full-strength sun (1,000 W/m²) that would deliver the same energy. Phoenix gets roughly twice the peak sun hours of Seattle, but Seattle still has fifteen hours of daylight in June. Peak sun hours already fold in latitude, cloud cover, tilt, azimuth and season, which is why a single number can carry the whole solar resource.
The third term is the performance ratio. A module rated 400 W is rated at 25 °C cell temperature under 1,000 W/m² of light. On a roof in August the cells run near 60 °C and the module delivers perhaps 340 W in the same irradiance. Add dust, a chimney shadow at 4 p.m., resistance in 90 feet of conductor, and the inverter's own conversion loss, and the array reliably delivers 75–85% of the ideal figure over a year. That ratio is the number a commissioning engineer checks first, because it is the only production metric that is independent of the weather.
Use this alongside the solar system size calculator when you are working the other direction — from a target kWh consumption back to the array size needed.
The formula, term by term
The production equation is deliberately simple: E = P_dc × PSH × PR. Multiply the array's kilowatts by the peak sun hours to get the ideal daily kilowatt-hours, then multiply by the performance ratio to strip out everything that goes wrong between the glass and the revenue meter.
P_dc is the sum of the module nameplate ratings, in kilowatts. Twenty 400 W modules is 8.0 kW DC. Note that this is the DC rating; inverters are rated in AC kilowatts and are routinely undersized relative to the array, so an 8.0 kW DC system may carry a 6.6 kW AC inverter. Production is driven by the DC number, with the inverter's clipping losses rolled into your loss stack.
PSH comes from an irradiance dataset, not from a weather app. NREL's National Solar Radiation Database and the PVWatts tool built on it will give you the plane-of-array irradiation for a specific tilt and azimuth anywhere in the United States. Use the annual average for an annual estimate and the monthly value if you are checking a single month's bill.
PR is a product, not a sum. Each loss acts on whatever energy survived the previous one, so six independent 5% losses do not cost you 30% — they cost you 1 − 0.95⁶ = 26.5%. Authors who add their loss percentages consistently understate production, and the error grows with the number of loss terms. The PVWatts default loss stack sums to about 14% of non-thermal system losses, which combined with inverter efficiency and a typical annual temperature penalty lands most rooftop systems between 0.75 and 0.85.
Two loss terms deserve individual attention. Temperature loss is the annual energy penalty from cells running above 25 °C; it depends on the module's temperature coefficient (typically −0.29 to −0.40 %/°C for modern silicon) and on how freely air moves behind the array. A ground mount or an elevated rack loses less than modules flush-mounted three inches above dark shingles. Shading loss is the term that varies most between sites and is the only one you can usually engineer away.
Worked example: twenty 400 W modules at 4.5 peak sun hours
Take a suburban roof with twenty 400 W modules, an annual-average 4.5 peak sun hours, and the default loss stack: 3% shading, 2% soiling, 8% temperature, 3% wiring, 4% inverter, 6% mismatch and availability.
- Array rating. 20 × 400 W = 8,000 W = 8.0 kW DC.
- Ideal daily energy. 8.0 kW × 4.5 h = 36.0 kWh before any losses.
- Performance ratio. Multiply the survival factors: 0.97 × 0.98 × 0.92 × 0.97 × 0.96 × 0.94. Working left to right: 0.97 × 0.98 = 0.9506; × 0.92 = 0.874552; × 0.97 = 0.848315; × 0.96 = 0.814382; × 0.94 = 0.765519, or 76.6%.
- Daily production. 36.0 × 0.765519 = 27.56 kWh per day.
- Annual production. 27.56 × 365 = 10,059 kWh per year.
- Monthly average. 10,059 ÷ 12 = 838 kWh per month.
- Specific yield. 10,059 ÷ 8.0 = 1,257 kWh per kW installed.
Sanity-check the last figure. Specific yield strips out system size entirely, so it is directly comparable with any other array anywhere. A well-built fixed-tilt system in the American Southwest reaches roughly 1,600–1,800 kWh/kW; the Northeast and Pacific Northwest run closer to 1,050–1,300. A figure of 1,257 kWh/kW is exactly what you would expect from a 4.5-PSH site with an ordinary loss stack, so the arithmetic hangs together.
Now suppose the homeowner's monitoring shows 8,400 kWh in the first full year. That is a specific yield of 1,050 kWh/kW and an implied performance ratio of 0.639. The array is not underrated — something is taking a 16% bite. Restack the losses with a shade report in hand and you will usually find it.
How to read the result
Read the specific yield first, not the kWh. Total kWh tells you how big the system is; specific yield tells you how well it works. Compare your specific yield with other systems at a similar latitude, and if it is more than about 10% below them, you have a problem to find rather than an estimate to revise.
A performance ratio between 0.75 and 0.85 is normal for a fixed rooftop array. Ground-mounted systems with good rear ventilation and no shading push toward 0.85–0.88. Anything above 0.90 over a full year means you have omitted a real loss, most often temperature. Anything below 0.70 means shading, soiling in a dusty climate, an inverter operating outside its efficient range, or a string that is offline and nobody noticed.
The daily figure is an annual average and no single day will match it. In the continental United States a fixed south-facing array typically produces two to three times as much in June as in December. Budget your battery and your net-metering expectations on the monthly numbers, not the annual average — the net metering savings calculator handles that seasonal split explicitly.
Production falls a little every year. Crystalline silicon modules degrade at roughly 0.4–0.7% per year under most manufacturers' warranties, which typically guarantee about 80–92% of nameplate output at year 25 depending on product. This calculator gives you year one; apply degradation yourself when you model a 25-year cash flow, or use the solar LCOE calculator, which discounts a degrading energy stream properly.
Typical annual-average peak sun hours and expected specific yield
| Region | Annual-average PSH (h/day) | Ideal yield (kWh/kW/yr) | Expected yield at PR 0.78 |
|---|---|---|---|
| Desert Southwest (Phoenix, Las Vegas) | 6.0 | 2,190 | 1,708 |
| Inland California, west Texas | 5.5 | 2,008 | 1,566 |
| Colorado Front Range, New Mexico | 5.3 | 1,935 | 1,509 |
| Florida, Gulf Coast | 4.8 | 1,752 | 1,367 |
| Mid-Atlantic, Carolinas | 4.5 | 1,643 | 1,281 |
| Midwest (Chicago, Minneapolis) | 4.2 | 1,533 | 1,196 |
| New England, upstate New York | 4.0 | 1,460 | 1,139 |
| Pacific Northwest (Seattle, Portland) | 3.5 | 1,278 | 997 |
Regional peak sun hour bands are approximate planning figures. Pull the exact plane-of-array value for your address, tilt and azimuth from the NREL National Solar Radiation Database before you size anything you are going to build.
Mistakes that make a production estimate wrong
- Adding losses instead of multiplying them. Six 5% losses cost 26.5%, not 30%. The error always runs in the same direction, so an added stack systematically underestimates production.
- Using hours of daylight as peak sun hours. Daylight hours are roughly double peak sun hours in summer and can be triple in winter. If your PSH figure is above 7, you have almost certainly made this substitution.
- Applying a flat-roof or horizontal irradiance value to a tilted array. Plane-of-array irradiation on a 30° south-facing surface is meaningfully higher than horizontal irradiation at mid-latitudes. Ask the dataset for the tilt and azimuth you are actually building.
- Leaving temperature loss at zero. It is the single largest loss on most rooftops and the easiest to forget because it does not appear on any invoice.
- Counting the inverter's AC rating as the array rating. A DC-to-AC ratio of 1.15–1.30 is standard practice; using the inverter's kW as P_dc understates production by that same margin.
- Ignoring the second roof plane. Modules on a west-facing plane receive different irradiation from the south-facing ones. Run each plane separately and add the results rather than averaging the orientation.
- Assuming a shade report's percentage is a summer figure. Shade reports quote annual solar access. Winter shading is far worse because the sun is lower, so a 5% annual figure may be 15% in December.
Where this method sits among the alternatives
This calculator implements the simplest defensible model: one irradiation number, one multiplicative loss stack. It is the model behind every rule-of-thumb estimate a salesperson quotes and it is accurate to within a few percent of far more elaborate tools when the loss assumptions are honest.
NREL PVWatts is the next step up and remains the reference implementation in the United States. It reads a typical meteorological year hour by hour, computes cell temperature from ambient temperature and wind speed, applies the module temperature coefficient, models the inverter efficiency curve at part load, and reports month-by-month output. Where this calculator asks you for a single temperature loss, PVWatts derives it. Use PVWatts for anything that goes into a contract or an incentive application.
SAM (System Advisor Model), also from NREL, goes further still, modelling specific modules and inverters from the CEC database, string-level shading, bifacial gain and detailed financial structures. It is the right tool for commercial and utility-scale work.
For an existing array, the honest check is not a model at all: it is the specific yield from your own monitoring compared with a neighbouring system of similar orientation. Modelled production carries an irreducible uncertainty of several percent from the weather year alone, so a single month that comes in 10% low means very little and a full year that comes in 15% low means a great deal.
Once you have a credible annual kWh figure, feed it into the solar payback period calculator for the economics, or into the home battery backup sizing calculator if you are pairing the array with storage. If you are sizing an off-grid system instead, start from consumption with the off-grid daily load audit.
Key terms
- Peak sun hours (PSH)
- Daily solar energy on the array plane divided by 1,000 W/m². A site receiving 4.5 kWh/m² per day has 4.5 peak sun hours, whatever the actual length of the day.
- Standard test conditions (STC)
- 1,000 W/m² irradiance, 25 °C cell temperature and air mass 1.5. The nameplate wattage on every module label is measured here, and no roof in service ever matches it during production hours.
- Performance ratio (PR)
- Actual energy delivered divided by the energy the array would deliver if every module produced its nameplate rating in the measured irradiance. It is dimensionless and site-independent, which makes it the standard commissioning metric.
- Specific yield
- Annual kWh per kW of installed DC capacity. Two systems of different sizes at the same site should report nearly the same specific yield; a gap points to shading or a fault.
- DC-to-AC ratio
- Array DC rating divided by inverter AC rating, commonly 1.15–1.30. Deliberate oversizing captures more energy in the shoulder hours at the cost of clipping a few peak hours.