Solar, Explained
No quotes, no installer referrals, no advertising. Rules and rates vary by state and utility.

How many solar panels a house needs: the arithmetic

Annual consumption divided by site yield gives system size; divided by module wattage gives the panel count. Then roof, string and inverter limits decide.

The number of panels on a roof is not chosen. It falls out of four quantities: how much energy the building uses in a year, how much energy one kilowatt of panels produces at that site in a year, how many watts each panel is rated for, and how much of the resulting array the roof, the inverter and the string topology will actually accept. Each of those is knowable before anything is ordered. The order matters, because the last two can shrink the answer the first two produce.

Annual consumption comes from twelve months of bills, not one

A single winter bill describes a winter. A single summer bill describes a summer. The figure the calculation needs is the sum of a full twelve consecutive months of consumption, in kilowatt-hours, taken from the utility’s own history rather than from memory. Most utilities expose this as a downloadable interval or monthly history; the twelve-month total is the only number that should enter the arithmetic.

If the building is new, or the history is short, the same figure can be assembled from a year of usage at the previous address, or from a year of a comparable building, but it should be labelled as an estimate and revisited after twelve months of real data.

Site yield: production per kilowatt, not panel count

A panel’s wattage is measured at standard test conditions: 1000 W/m² irradiance, 25 °C cell temperature, a defined air mass. Those conditions do not occur on a roof. What a roof delivers over a year is described by its annual production per installed kilowatt, in kWh per kW per year, and that number varies with latitude, weather, tilt, azimuth and shading.

A common shorthand is peak-sun-hours: the number of hours per day at 1000 W/m² that would deliver the same daily energy. A site with 4.5 peak-sun-hours averages 4.5 kWh per kW per day, or roughly 1,640 kWh per kW per year before losses. Real systems lose energy to temperature, soiling, wiring, inverter conversion and mismatch, so the delivered figure is lower than the irradiance figure. Design tools model those losses explicitly; the National Renewable Energy Laboratory’s PVWatts calculator is the standard free implementation and takes the site, the array geometry and the loss assumptions as inputs.

From annual energy to system size to panel count

The arithmetic is two divisions.

System size in kW = annual consumption in kWh ÷ annual yield in kWh per kW.

Panel count = system size in kW × 1000 ÷ module wattage.

Worked example, with round hypothetical numbers: a building uses 10,000 kWh per year. The site yields 1,400 kWh per kW per year after losses. The system size is 10,000 ÷ 1,400 ≈ 7.1 kW. At 400 W per module, the count is 7,100 ÷ 400 ≈ 18 modules. None of those figures describes a real building; they exist to show the shape of the calculation.

Two adjustments are commonly made at this point. If the goal is to offset only the portion of consumption the utility will credit, the target energy is smaller than the annual total. If future loads are planned, such as an electric vehicle or a heat pump, the annual figure should include their expected consumption, because adding panels later is constrained by the same roof and string limits that apply now.

The roof decides whether the arithmetic survives

A count of modules has to fit somewhere. Usable area is not the same as roof area: setbacks required by fire code, chimneys, vents, skylights and shading all remove space. A modern 400 W module is around 1.7–1.8 m², so 18 modules need roughly 31 m² of usable, unshaded roof in the chosen orientation.

Orientation and tilt change the yield figure rather than the count directly. A north-facing roof in the northern hemisphere, or a steeply pitched east or west roof, produces less per kW, so more kW are needed for the same energy, and the count rises until the roof runs out. Shading does the same, and it is not linear: a shadow across one cell string can reduce a module’s output far more than the shaded fraction suggests.

String and inverter limits cap the count independently

Panels are wired in series into strings, and a string’s voltage is the sum of its modules’ voltages. A silicon cell produces roughly half a volt, and a 60-cell module around 30 V at operating conditions, but the voltage that matters for equipment ratings is the open-circuit voltage at the coldest expected temperature, because voltage rises as temperature falls. A string sized for a 25 °C day can exceed the inverter’s maximum input voltage on a cold clear morning. The inverter’s maximum input voltage, its MPPT window and its maximum input current are hard limits, and the string sizing calculation has to satisfy all of them at the temperature extremes for the site.

Inverter capacity is a second cap. Inverters are commonly rated below the array’s nameplate wattage, a deliberate ratio that trades a small amount of clipping at peak irradiance for better production at other times. The ratio is a design choice, but it limits how much array the inverter will accept.

Where the count changes after the design is fixed

SituationEffect on panel countWhen this does not apply
Shaded roof sectionMore panels elsewhere, or microinverters/DC optimisers to limit the lossThe shading is seasonal and falls outside the main production hours
Steep east or west roofMore kW for the same annual energyThe roof is near-flat and tilt can be corrected with mounting
Planned EV or heat pumpHigher annual consumption, so more panels nowThe load is small relative to the annual total
Inverter at its input limitCount capped regardless of roof spaceThe inverter is specified after the array, not before
Cold climateString length limited by cold-morning open-circuit voltageThe site’s record low is mild

Why the panel count on a proposal may not match the arithmetic

A proposal’s count can differ from the calculation for legitimate reasons: the designer used a different yield assumption, included or excluded a future load, or hit a string or inverter limit. It can also differ because the design was fitted to the roof rather than to the annual energy. The useful check is to ask for the annual consumption figure, the assumed yield per kW, and the resulting system size, and to confirm that the three are consistent. A count without those three numbers cannot be checked.

What the reader has to verify locally

Net metering, export compensation, interconnection limits, permitting and any incentive are set by state and by utility, and they change. A system sized to offset annual consumption may be credited differently depending on the local rule, and some utilities cap the size relative to historical consumption. The reader has to check the current rule with their utility and their state energy office before fixing the count, and should treat any published rule as provisional.

The DC side of an array carries dangerous voltage whenever light falls on it and cannot be switched off at the panel. String wiring, combiner work, roof work and the mains connection are licensed trades in most jurisdictions, and the design calculations that set string length and inverter limits are part of that licensed work.

Common questions

How many solar panels do I need for a 2,000 sq ft house?

Floor area does not determine panel count; annual consumption does. Two houses of the same size can differ by a factor of two in kWh used. Start from twelve months of bills, divide by the site's annual yield per kW, then divide by module wattage.

How many kWh does one solar panel produce per year?

It depends on the site and the panel's wattage. A 400 W module at a site yielding 1,400 kWh per kW per year produces about 560 kWh per year after losses. A sunnier site or a different tilt changes that figure.

Can I add panels later if I need more?

Sometimes, but the roof space, the string voltage limits and the inverter's input capacity all constrain later additions. It is usually cheaper in design effort to size for planned loads such as an EV or heat pump at the outset.

Why does my installer's panel count differ from my own calculation?

The likely differences are the yield assumption, whether future loads were included, and whether the design was capped by roof area or inverter limits. Ask for the annual consumption, the assumed yield per kW and the resulting system size, and check that they are consistent.

Does a bigger panel count always mean more energy?

No. Panels in shade, on a poor orientation, or beyond the inverter's input limit contribute less than their nameplate suggests. The count is only meaningful alongside the yield assumption and the equipment limits.

Read next