Why annual totals mislead
The standard way to describe a solar array is the percentage of your annual consumption it covers. It is a useful sizing figure and a poor savings figure, because production and consumption are on different annual curves. A fixed array in the northern hemisphere makes roughly twice as much in June as in December. Household electricity use, meanwhile, either peaks in summer with air conditioning, peaks in winter with electric heat, or stays fairly flat, and only one of those three lines up with the sun.
This page keeps the twelve months separate. In each month it works out how much of the production is consumed as it is generated, which is worth full retail because it is electricity you did not buy, and how much is surplus, which is worth whatever your utility gives you for exports. Then it adds them up. The gap between the percentage of kilowatt-hours covered and the percentage of dollars removed is what the annual-total method hides.
Net metering, net billing and nothing
What a utility does with your exports has changed a lot and is still changing, and it is the single largest determinant of what a system saves after the hardware is on the roof.
| Arrangement | What an exported kWh is worth | Effect on sizing |
|---|---|---|
| Full retail net metering with kWh banking | The same as one you consume, if you can spend the credit later | Sizing to annual consumption makes sense |
| Net billing with an export rate | The export rate, often a third of retail or less | Sizing to daytime consumption makes more sense |
| No export compensation | Nothing | Only self-consumed production has value; oversizing is waste |
Two practical warnings. What your neighbour receives is frequently not what a new interconnection is offered, because most changes to these rules have grandfathered existing customers. And banking credits in kilowatt-hours is only as good as your ability to use them: they usually expire at an annual true-up, and where that true-up falls in the calendar decides whether a summer surplus survives long enough to cover a winter deficit.
The floor under the bill
Fixed charges do not respond to solar. The customer charge, service charge, meter charge and any flat riders are billed whether the meter runs forwards, backwards or not at all, and in many territories there is a minimum monthly bill that sits above the fixed charge alone. A household paying $12 a month in fixed charges cannot get below $144 a year no matter how much it generates, and on a small bill that floor is a large fraction of the total.
This is why the percentage of the bill removed is almost always lower than the percentage of consumption covered, and why the gap is widest for low users. It is also why interconnection agreements sometimes add a charge specific to solar customers, which is worth asking about before signing anything, since it comes straight off the savings this page calculates.
What would change these numbers most
Time-of-use pricing, if you are on it, changes everything and is not modelled here. Under time-of-use rates a kilowatt-hour has a different value depending on the hour, and solar production is concentrated in the middle of the day. In territories where the expensive hours are in the late afternoon and evening, after production has fallen away, a flat-rate model like this one overstates the savings; where the peak sits in the early afternoon, it understates them. Batteries change it again, by moving production into the expensive hours, which is a separate calculation with its own economics.
The other assumption worth testing is the specific yield. It is the one input on this page you cannot measure yourself before installing, and it depends on latitude, tilt, azimuth, shading and how honestly the losses were counted. Proposals commonly assume less shading than a roof turns out to have. Rerun the model with the yield 10 or 15 percent lower and see how much of the conclusion survives; if the answer changes your decision, the yield figure needs verifying before anything else does.
Questions people ask
How is this different from the solar payback calculator?
They answer sequential questions. This page asks what a system does to your electricity bill in a typical year, modelling twelve months separately so seasonal mismatch, banked credits and fixed charges show up. It knows nothing about what the system cost. The payback calculator takes an annual saving and an installed cost and runs twenty-five years of rate escalation, panel degradation, maintenance and inverter replacement to find a break-even year. A sensible order is to work out the annual saving here, then take that figure over to the payback page along with your quote.
Should I size the array to cover 100 percent of my usage?
Only under full retail net metering with credits you can actually spend. On that arrangement an exported kilowatt-hour and a consumed one are worth the same, so matching annual consumption is a reasonable target. On an export tariff they are not the same at all: self-consumed energy is worth full retail and exported energy might be worth a third of that, so each panel beyond what you can use during the day earns much less than the ones before it. Run the model at your intended size and again at 70 percent of it, and look at the average value per kilowatt-hour produced. If that figure falls sharply as the array grows, the marginal panels are being sold cheap.
Why does the bill not fall as much as production covers?
Three reasons, all visible in the results. Fixed charges are unaffected by solar, so a portion of the bill is untouchable. Seasonal mismatch means some production arrives in months when you cannot use it, and unless it banks at full retail and gets spent later, it is worth less than the electricity it displaced in a different month. And exports, on most current tariffs, are compensated below retail. Together these routinely turn a 100 percent kilowatt-hour offset into a 70 or 80 percent bill reduction, which is not a failure of the system, only of the way the offset percentage is usually quoted.
Where do I find my specific yield?
From a modelling estimate for your specific roof, not from a rule of thumb. Public irradiance tools will produce an annual kWh figure from your location, array tilt, orientation and an assumed loss factor; divide that by the system size in kW to get the yield to enter here. An installer proposal contains the same thing implicitly, and dividing its annual production estimate by its system size is a good way to see what it actually assumed. The number varies across the United States by something like 70 percent between the cloudiest and sunniest regions, and shading can take a large bite out of it that generic estimates miss entirely.