Sizing a solar array is an energy balance, not a roof measurement
You size a grid-tied solar system by matching annual energy, not by filling the roof. Your utility bill tells you how many kilowatt-hours you consume in a year; the site tells you how many kilowatt-hours each installed kilowatt will generate in a year; dividing the first by the second gives the kilowatts you need. Everything else — module count, inverter rating, roof area — follows from that quotient.
The number that does the work is the denominator, the annual specific yield in kWh per kW. It is the product of 365 days, the site's peak sun hours, and the system's performance ratio. In Phoenix that product is near 1,700; in Seattle it is near 1,000. Two households with identical 12,000 kWh bills therefore need arrays that differ by 70% in size. Any sizing rule quoted in kilowatts per thousand kWh of consumption is silently assuming one particular climate.
Sizing to annual energy works because net metering and net billing tariffs settle over a year: you bank surplus in the summer and draw it back in the winter. If your tariff does not let you carry credit across months, or credits exports well below retail, annual matching is the wrong objective and you should size closer to your daytime load instead. The net metering savings calculator shows the difference that makes to the bill.
The formula and what each input is really doing
Annual consumption should come from twelve consecutive months of bills. Utilities print the kWh for each billing period, and many provide a rolling twelve-month total or an interval-data download. One summer month in a house with air conditioning can be double the annual average; one shoulder month can be half. Sizing from either produces a system that is wrong by a factor you will live with for twenty-five years.
Load growth belongs in the numerator, not in a fudge factor. An electric vehicle driven 12,000 miles a year at 3.5 miles per kWh adds roughly 3,400 kWh — around a third of a typical household's consumption. A cold-climate heat pump replacing gas heat can add more. Add the kWh you can name; do not pad the target offset instead, because that hides what you are actually buying.
Target offset is a policy choice as much as an engineering one. One hundred percent is the usual objective under full retail net metering. Under net billing tariffs where exports earn a fraction of retail, the economically optimal offset is lower — often well under 100% — because every exported kilowatt-hour is worth less than the one you consume on site. Many utilities also cap interconnected system size relative to historical consumption, so check the interconnection rules before entering a number above 100.
Peak sun hours is the daily plane-of-array irradiation in kWh/m². Get it from NREL's PVWatts for your address, at the tilt and azimuth you will actually build. A west-facing array at 30° receives materially less annual irradiation than a south-facing one at the same tilt, and that difference belongs here rather than in the performance ratio.
Performance ratio is everything between the module glass and the meter: shading, soiling, cell temperature, conductor resistance, inverter conversion, module mismatch, and downtime. Rooftop systems land between 0.75 and 0.85. Entering an optimistic ratio undersizes the array, and the shortfall shows up as a bill you did not expect. If you want to build the ratio up loss by loss, the solar panel output calculator does exactly that.
The last step is integer arithmetic. You cannot buy 20.5 modules, so the count rounds up and the installed size lands slightly above the requirement. That is why the achieved offset in the results is almost always a little higher than the target you asked for.
Worked example: a 1,000 kWh per month home at 5.0 peak sun hours
A household averages 1,000 kWh a month, wants to cover all of it, and lives at a site with 5.0 annual-average peak sun hours. The installer's loss stack gives a performance ratio of 0.80, and the proposed module is 400 W with a 1.20 DC-to-AC ratio.
- Annual consumption. 1,000 × 12 = 12,000 kWh.
- Energy the array must generate. 12,000 × 100% = 12,000 kWh.
- Annual yield per installed kW. 365 × 5.0 × 0.80 = 1,460 kWh/kW.
- Array size required. 12,000 ÷ 1,460 = 8.219 kW DC.
- Module count. 8.219 × 1,000 ÷ 400 = 20.55, rounded up to 21 modules.
- Installed size. 21 × 400 W = 8.40 kW DC.
- Inverter rating. 8.40 ÷ 1.20 = 7.0 kW AC.
- Annual production. 8.40 × 1,460 = 12,264 kWh.
- Achieved offset. 12,264 ÷ 12,000 = 102.2%.
Rounding up added 0.18 kW and 264 kWh a year — a 2.2% overshoot, which is the normal consequence of integer module counts. Rounding down to 20 modules would give 8.0 kW, 11,680 kWh and a 97.3% offset. Either is defensible; which one you choose usually comes down to roof space and to whether your utility's interconnection cap makes the overshoot a problem.
Now move the same household to a 3.5 peak sun hour site and nothing else changes. Yield per kW drops to 365 × 3.5 × 0.80 = 1,022 kWh/kW, the requirement rises to 12,000 ÷ 1,022 = 11.74 kW, and the module count goes from 21 to 30. That is the whole argument for using a real irradiation dataset rather than a national average.
How to read the size the calculator gives you
Check the roof before you fall in love with the number. A 21-module array at roughly 21 ft² per module needs about 440 ft² of unshaded module area, plus fire-access setbacks. Run it through the solar panel count and roof area calculator before you commit; on many roofs the binding constraint is area, not consumption, and you end up sizing to the space and accepting a partial offset.
An achieved offset a few points above target is normal and harmless. An achieved offset 30% or more above target usually means the consumption figure is too low — a common error when someone enters a winter month for a house with electric heat, or forgets that the bill they are reading covers a shared meter.
The inverter rating is a starting point, not a specification. Inverter selection also depends on string voltage limits at the coldest expected temperature, on whether you need module-level electronics for shading or rapid shutdown compliance, and on the discrete ratings the manufacturer actually sells. A 7.0 kW calculated rating typically means buying a 7.6 kW unit.
Watch what a partial offset does to the bill, not to the kWh. Fixed monthly charges, minimum bills and non-bypassable charges do not scale with consumption, so a 90% energy offset does not produce a 90% bill reduction. On most residential tariffs the bill reduction is smaller than the energy offset.
Array size needed per 1,000 kWh of annual consumption
| Peak sun hours (h/day) | Annual yield (kWh/kW) | kW per 1,000 kWh/yr | kW for a 12,000 kWh home |
|---|---|---|---|
| 3.0 | 876 | 1.142 | 13.70 |
| 3.5 | 1,022 | 0.979 | 11.74 |
| 4.0 | 1,168 | 0.856 | 10.27 |
| 4.5 | 1,314 | 0.761 | 9.13 |
| 5.0 | 1,460 | 0.685 | 8.22 |
| 5.5 | 1,606 | 0.623 | 7.47 |
| 6.0 | 1,752 | 0.571 | 6.85 |
| 6.5 | 1,898 | 0.527 | 6.32 |
Yield per kW is 365 × PSH × 0.80. Scale the last two columns by (0.80 ÷ your PR) if you use a different performance ratio.
Sizing mistakes that cost real money
- Sizing from one month's bill. Summer and winter consumption commonly differ by a factor of two or more. Always use a twelve-month total.
- Reading the dollar amount instead of the kWh. Tiered and time-of-use rates mean the dollar figure is not proportional to energy. Size from kilowatt-hours.
- Assuming 100% offset is optimal under every tariff. Where exports credit at an avoided-cost rate well below retail, the last kilowatt of array earns much less than the first, and the best offset can be well under 100%.
- Forgetting the interconnection cap. Many utilities limit interconnected DC capacity relative to historical consumption or to the service rating. Oversize past it and the application is rejected after you have paid for the equipment.
- Adding an EV as a percentage rather than as kWh. Compute the actual annual kWh from your mileage and the vehicle's efficiency; a guessed percentage is usually low.
- Using an optimistic performance ratio. Every point of PR you overstate is a point of array you fail to buy. If shading is real, model it — do not assume it away.
- Sizing to the roof and calling it an offset. If the roof limits you to 6 kW, say so and quote the honest offset. A partial offset is a perfectly good outcome; a misdescribed one is not.
When to size a different way
Under net billing tariffs, size to daytime load. California's NEM 3.0 and similar successor tariffs credit exports at hourly avoided-cost values that are a small fraction of the retail rate for most of the year. The economically correct array under such a tariff generates roughly what the house consumes while the sun is up, and the remaining value comes from adding storage rather than adding modules. Size the battery separately with the home battery backup sizing calculator.
For commercial buildings, size to the demand charge as well as the energy charge. Where demand charges are a large share of the bill, the array's value depends on whether its output coincides with the facility's peak. West-facing orientation and storage often beat a larger south-facing array.
For off-grid systems, do not use this method at all. An off-grid array is sized to the worst month, not the annual average, and to the battery's recharge requirement rather than to a bill. Start with the off-grid daily load audit and then the battery bank amp-hour calculator.
Once the size is settled, the two questions that follow are how much it produces and whether it pays. This calculator's production figure is a first-order estimate; run the address through PVWatts for the version you would put in a contract, and take the economics to the solar payback period calculator.