PV array size from daily load

The answer arrives in kilowatts-peak, and the suffix is doing real work. A kWp is the output a module family produces under Standard Test Conditions — irradiance of 1,000 watts per square metre on a cell held at 25 °C — which is a laboratory reference point, not a promise about a roof. An array almost never delivers its nameplate in service; nameplate is simply the common currency in which arrays are bought, quoted and compared, and it is the currency the rest of this cluster trades in.

Peak sun hours compress a whole day of varying sunlight into an equivalent number of hours at full laboratory strength. A site quoted at 5.5 PSH receives, over a typical day, the same energy it would receive from five and a half hours of STC-grade sun and darkness the rest of the time. That compression is what makes the sizing a single division — but the figure is an ANNUAL AVERAGE, and averaging is where it bites: a mid-latitude site can see winter days at a fraction of its summer figure, so an array sized on the average will overshoot the bright months and fall short in the dark ones.

The performance ratio concedes that a real system is not the laboratory. Cell temperature above 25 °C, inverter conversion, wiring resistance, soiling, mismatch between modules — each takes its share, and the performance ratio is the fraction of the ideally available energy that survives all of them together. It sits in the denominator, so a lower ratio grows the array: the machine is deliberately oversized to cover its own losses.

The structure of the division is worth reading before any values go in. Load sits alone in the numerator, so the required array scales directly with demand; sun and performance sit together underneath, so halving either doubles the array. The pack also declares the reverse workflow: fix an array size — a roof or a budget already decided it — and the same relation, run backwards by the engine, returns the daily load that array can support.

A roof is a constraint, not a requirement. Starting from the roof answers the question “how much solar could I fit?”, which is almost never the question being asked. The question is “how much of my consumption can the sun carry?”, and that starts at the meter: a year of bills, divided into a daily rhythm, tells you what the array must produce on an average day before a single panel is chosen.

Working from consumption also exposes the cheapest improvement first. A household that trims its daily load before sizing buys a smaller array, a smaller inverter and — if storage comes later — a smaller battery. Every saved daily kilowatt-hour is bought once; every generated one is bought as hardware that must also be installed, maintained and eventually replaced.

The load number deserves more scepticism than either of the other two inputs, because it is the one the household controls and the one that changes. An electric vehicle, a heat pump, a new workshop — each rewrites the daily load, and an array sized to last year’s bills quietly becomes undersized without anything failing.

Peak sun hours are not daylight hours, and the difference is the most common sizing error in either direction. Daylight counts every pale morning hour at full value; peak sun hours weigh each hour by how much energy it actually delivers. A site’s figure comes from irradiance data — solar atlases, meteorological services, installer databases — and using a generic value from a different climate moves the array size by more than any other single mistake.

The performance ratio is guessed wrong in the optimistic direction almost exclusively. Heat derates panels precisely when the sun is strongest; dust accumulates fastest in the seasons it rains least; inverters convert at slightly less than their brochure figure all year. None of these is a defect — they are what “real installation” means — and the ratio exists so they are counted once, deliberately, instead of discovered later as disappointment.

Neither number requires precision to be useful. What both require is provenance: a peak sun figure traceable to the site’s own climate, and a performance ratio chosen from the installation’s actual conditions rather than from hope. The calculator’s guards refuse values outside the physically plausible range, but they cannot know your climate — that judgement stays with you.

The calculator below is the same certified engine the calculator page runs — fetched, verified and mounted mid-lesson. It arrives pre-filled with the pack’s own worked example: a modest household load on a sunny site with realistic losses. Change the daily load to your own and watch the array size follow; then try the peak sun hours of a cloudier climate and see how sharply geography prices solar.

Read the result as capacity, not certainty: the array that covers the stated load on an average day. Every figure shown is computed by the verified engine as you type — nothing on this page stores an answer.