An array is sized against the household, not against the roof

Roof area is a constraint, not a target

The usual design conversation starts with how many panels will fit. It is an understandable place to begin, since the roof is the physical limit, but it answers a question about geometry when the question that decides the outcome is about consumption. The roof tells you the maximum. It says nothing about the optimum.

A household draws electricity in a pattern that is largely fixed by how it lives: when people are home, what appliances run, whether anything heavy is on the circuit. An array produces in a pattern fixed by the sun and the weather. The value of the system comes from where those two patterns overlap, and area does not appear anywhere in that sentence.

This matters because the two patterns overlap poorly by default. Generation peaks around the middle of the day. Domestic demand in a house that is empty from morning to evening peaks at either end of it, which means much of what the roof makes has nowhere to go except the grid.

Annual totals match too easily and mislead badly

The tempting calculation is to divide annual consumption by expected annual generation per panel and size the array to cover the year. It produces a satisfying number and it conceals the whole problem, because electricity generated in June cannot be spent in December and a unit exported is not a unit avoided.

Import and export are priced separately in nearly every arrangement, and a unit you use as it is generated is worth what you would otherwise have paid to buy it, while a unit you export is worth whatever the export arrangement pays. The gap between those two figures is the entire economics of sizing, and it varies enormously by country and by contract.

Where export is paid at something close to the import price, an annual matching approach is roughly defensible. Where export pays a token amount, or nothing, sizing to annual consumption produces an array whose upper half is largely donated. Check which situation applies before anybody counts panels.

Each additional panel is worth less than the one before

Think about what happens as an array grows against a fixed household demand. The first panels produce less than the house is drawing at almost every daylight hour, so essentially everything they make is consumed on site. As capacity increases, midday production begins to exceed the baseline load, and the surplus goes out to the grid.

The fraction consumed on site therefore falls steadily as the array grows, and it falls fastest once midday generation passes the household’s daytime draw. Doubling capacity does not double the saving on imports. Depending on the load pattern, it may add rather little to it at all.

The panels themselves get slightly cheaper per unit of capacity in a larger order, and the fixed costs of scaffolding, labour and paperwork are spread further, so the cost curve moves the other way. Somewhere those two curves cross. Finding that crossing is the actual design problem, and it needs a consumption profile rather than a tape measure.

Loads that move the answer

Anything that can absorb midday electricity changes the calculation fundamentally, because it converts export into self-consumption. A car charged at home during daylight is the clearest example, since it represents a large and flexible load that can often be scheduled. A heat pump shifts a share of heating demand onto the electricity meter, though it does so mostly in the season with the least sun.

A hot water diverter absorbs surplus into an immersion heater, which is a modest and reliable sink for a few units a day in a house with a cylinder. A battery does the same thing with far more flexibility and far more cost. Both raise the ceiling on a sensible array size.

Timing matters as much as quantity. A dishwasher that runs at two in the afternoon rather than eleven at night converts imported units into generated ones at no cost whatsoever, and households that make that shift habitually get more from a small array than households that don’t get from a large one.

When filling the roof is right anyway

There are honest arguments for a larger array than the current household needs. The fixed costs of an installation are considerable and are incurred once, so adding capacity while the scaffolding is up is far cheaper than returning later. Future loads are plausible: a car, a heat pump, a different household in the same building.

Where export is reasonably paid, the marginal panel keeps earning something even when its output leaves the property, and the case for filling the roof strengthens accordingly. Where a connection limit or an export limitation constrains what can leave, it weakens.

And there is a straightforward case for restraint that nobody makes often enough. If a household is out all day, has no car to charge, no cylinder to heat, no intention of buying a battery, and an export arrangement that pays close to nothing, then the last several panels are producing electricity that is worth almost nothing to anybody. A smaller array on the sunniest part of the roof will do better per unit spent, and the money not spent stays in the account.