What array sizing is really asking
Array sizing is a division problem with three honest numbers and one dishonest one. The honest numbers are how much energy you use, how much sunlight lands on your roof, and how big a panel is. The dishonest one is nameplate wattage, because no array ever produces its nameplate output for a full day.
A 400 W module is rated at standard test conditions: 1,000 W/m² of irradiance, a 25 °C cell temperature and a defined spectrum. Those conditions occur on a clear spring day for a few minutes. Real output is lower for the whole of the rest of the year, and the gap has a name — system losses — that covers inverter conversion, cell heating, soiling, module mismatch, wiring resistance and shading.
Peak sun hours resolve the first half of that. Peak sun hours are the daily energy landing on the plane of the array, in kilowatt-hours per square metre. Because modules are rated at exactly 1,000 W/m², a site with 4.5 kWh/m²/day behaves as though the sun switched on at full rated intensity for 4.5 hours and then switched off. That makes the arithmetic simple: one kilowatt of array in a 4.5 sun-hour location harvests 4.5 kWh of DC energy per day before losses.
The derate factor resolves the second half. Multiply by the fraction of that DC energy that reaches your meter. NREL's PVWatts model uses a default total system loss of about 14% alongside a 96% inverter, which lands near 82%; a planning value of 75-80% is the conservative choice, and it is what this calculator defaults to.
Divide your daily energy by the product of those two, and you have the array in kilowatts. Everything after that is rounding to whole panels.
Each term and where to get it
Daily energy. Take a twelve-month total from your utility bills and divide by 365, or take a monthly figure and multiply by 12/365. A single month misleads badly in either direction: a July bill in a hot climate and a January bill in a cold one describe two different households. If your consumption is about to change — an electric vehicle, a heat pump, a new workshop — add the expected load now rather than resizing later, because inverter capacity and roof space both constrain expansion.
Peak sun hours. This is site-specific and cannot be guessed from latitude alone, because tilt, azimuth and local climate all move it. Get it from NREL's PVWatts, which reports monthly and annual plane-of-array irradiation for any address once you enter your tilt and azimuth. A south-facing array at latitude tilt maximises the annual figure in the northern hemisphere; a west-facing array yields less energy but more of it in the late-afternoon peak, which matters under time-of-use rates.
System efficiency. The largest single component is temperature. Silicon modules lose roughly 0.3-0.4% of output per degree Celsius above 25 °C, and a rooftop module in summer runs far hotter than the ambient air. Inverter conversion costs a few percent, soiling a few more in dusty or pollen-heavy regions, and DC wiring and mismatch a little each. Shading is not a percentage but a design problem: even partial shade on one module can cut a whole string, which is why module-level power electronics exist.
Panel wattage and area. These two together give the module efficiency, which the calculator reports as a sanity check. Divide watts by area in square metres, then by 1,000 W/m². Current residential silicon modules land roughly between 19% and 23%; a figure well outside that means one of the two datasheet numbers has been entered wrongly.
Rounding is always upward. You cannot buy 23.4 panels. The installed size therefore exceeds the calculated requirement, and the calculator reports both so the headroom is visible.
Worked example: a 1,000 kWh per month home
A household averages 1,000 kWh a month. PVWatts reports 4.5 peak sun hours for their roof at its tilt and azimuth. They plan on 78% system efficiency, 400 W modules of 21.5 ft² each, and want to offset 100% of their use.
- Convert to daily energy. 1,000 kWh/month × 12 ÷ 365 = 32.877 kWh/day.
- Energy to offset. 32.877 × 100% = 32.877 kWh/day.
- Yield per kilowatt of array. 4.5 sun hours × 0.78 = 3.51 kWh per kW per day.
- Array size. 32.877 ÷ 3.51 = 9.367 kW DC.
- Panel count. 9,367 W ÷ 400 W = 23.42, rounded up to 24 panels.
- Installed size. 24 × 400 ÷ 1,000 = 9.6 kW DC.
- Annual production. 9.6 × 4.5 × 365 × 0.78 = 9.6 × 1,281.15 = 12,299 kWh/year, against the 12,000 kWh the household consumes.
- Module area. 24 × 21.5 = 516 ft² of panel, before walkways, setbacks and obstructions.
- Module efficiency check. 21.5 ft² ÷ 10.764 = 1.997 m²; 400 W ÷ (1.997 × 1,000) = 20.0%, which is normal for a current residential module.
Now test the sensitivity. Move the same house to a 5.5 sun-hour location and the array falls to 32.877 ÷ (5.5 × 0.78) = 7.66 kW, or 20 panels — four fewer for the identical consumption. Drop the efficiency assumption from 78% to 70% instead and the array rises to 32.877 ÷ 3.15 = 10.44 kW, or 27 panels. Those two inputs deserve more care than any other on the page.
How to read the result
Check the array against your roof before anything else. The module area figure is bare panel area. Real installations need fire-service access pathways, setbacks from ridges and edges, and clearance around vents and skylights, and those requirements come from the International Fire Code as your jurisdiction has adopted it. Plan on needing meaningfully more roof plane than the module area alone, and lay the array out on a satellite image before you commit to a panel count.
Annual production is an annual average, not a monthly promise. A system that covers 100% of consumption over a year will overproduce in June and underproduce in December, sometimes by a factor of three between the two. Under net metering that is fine, because the summer surplus banks against the winter deficit. Off-grid it is not fine at all: an off-grid array must be sized on the worst month, not the average, which typically means 1.5 to 2 times the grid-tied figure plus a backup generator.
The DC-to-AC ratio decides the inverter, not this calculation. Arrays are routinely paired with an inverter smaller than the nameplate DC, commonly at a ratio between 1.1 and 1.3, because the array rarely reaches nameplate and a smaller inverter runs closer to its efficient region. Clipping a few peak hours a year costs less than the larger inverter would.
Match the array to the storage if there is any. A battery bank has to be refilled within the available sun hours as well as sized for the load, so run your storage requirement through the battery bank sizing calculator and confirm the array can deliver that energy plus the daytime load on a short winter day. String voltages and conductor sizes then follow: check the DC run with the voltage drop calculator and the wire size and ampacity calculator.
Array size by monthly consumption
| Monthly use | Array required | Panels | Installed size | Annual production |
|---|---|---|---|---|
| 500 kWh | 4.68 kW | 12 | 4.8 kW | 6,150 kWh |
| 750 kWh | 7.03 kW | 18 | 7.2 kW | 9,224 kWh |
| 1,000 kWh | 9.37 kW | 24 | 9.6 kW | 12,299 kWh |
| 1,250 kWh | 11.71 kW | 30 | 12.0 kW | 15,374 kWh |
| 1,500 kWh | 14.05 kW | 36 | 14.4 kW | 18,449 kWh |
| 2,000 kWh | 18.73 kW | 47 | 18.8 kW | 24,086 kWh |
Every row is scaled to one site. Change the sun hours and the whole column moves: at 6.0 sun hours each array is 4.5/6.0 = 75% of the size shown, and at 3.5 sun hours it is 4.5/3.5 = 129% of it.
Where this estimate goes wrong
- Sizing from one month's bill. Consumption swings seasonally by more than most people expect. Use a twelve-month total, which every utility prints on the bill or exposes in an online account.
- Assuming an unshaded roof. A chimney, a vent stack or a neighbouring tree costs far more than its shadow area suggests, because a shaded cell drags down every module in series with it. Site the array around obstructions, or specify optimisers or microinverters.
- Using an optimistic derate. Anything above 85% assumes a cool, clean, perfectly matched, lightly wired system. Modules run hot on a roof, and the temperature coefficient on the datasheet tells you exactly what that costs.
- Forgetting that panels degrade. Manufacturers warrant a declining output over 25 years, commonly guaranteeing somewhere above 80% of nameplate at the end of the term. Read the specific warranty schedule for the module you are buying; the array you size today produces less in year twenty than in year one.
- Sizing an off-grid array on annual average sun hours. Off-grid systems must carry the worst month. Take December's figure from PVWatts rather than the annual mean, or budget for a generator to cover the difference.
- Ignoring the interconnection limit. Many utilities cap the exported system size at historic consumption or at a fraction of the service rating, and some restrict it further on constrained feeders. Check the rules before you design past 100% offset.
Code and standards that shape the design
NEC Article 690 governs photovoltaic systems: maximum system voltage calculated at the record low temperature, rapid shutdown at the array boundary, DC and AC disconnects, grounding and labelling. Article 705 covers interconnection with the utility supply, including the busbar rule that limits how much back-fed current a panelboard may accept. UL 1741 lists the inverter, and IEEE 1547 defines how it must behave when the grid misbehaves.
On the structural side, the array is a wind and snow load on the roof, and ASCE 7 is the governing standard for both. Racking manufacturers publish span tables derived from it for a given wind speed and exposure category. The rating method behind the panel wattage is IEC 61215 at standard test conditions, which is why the nameplate is optimistic relative to real operating temperature.
Grid-tied, hybrid and off-grid sizing differ
Grid-tied. The grid is an infinite battery, so annual energy balance is the right criterion and this calculation is the whole answer. The economics turn on the net-metering rules: full retail credit makes a 100% offset the natural target, while a low export rate pushes the optimum toward covering only daytime consumption.
Hybrid with storage. Now two constraints apply at once. The array must generate the annual energy, and it must also recharge the battery within the sun hours available on a poor day while still serving the daytime load. In practice the second constraint usually governs in winter, so size the array from this calculator, then check the charge rate against the bank size you have chosen and increase the array if the bank cannot return to full.
Off-grid. The annual average is irrelevant; the worst month governs. Take December sun hours for a northern site, size the array from those, and accept substantial summer surplus — it is cheaper than the alternative, which is a much larger bank. Most off-grid designs still keep a generator for multi-day storms, and sizing that set is a separate calculation because a generator is limited by power rather than energy.
Whichever case applies, the fastest way to shrink the array is to shrink the load first. A single old chest freezer or an electric water heater on a poor schedule can be worth several panels; the appliance energy cost calculator ranks your loads by annual kWh so you know which one to address before you buy modules to feed it.