An east and west array can beat a south-facing one on the bill

Annual total is not the objective

The default assumption in siting an array is that more generation is better, so the equator-facing slope wins: south in the northern hemisphere, north in the southern. On pure annual kilowatt-hours that is correct, and it is not the only thing worth maximising.

What a household actually banks is the value of the electricity, and under any tariff where export pays less than import, a unit consumed at home is worth more than a unit sent out. That turns the question from how much the roof produces into how much of what it produces the house can absorb as it arrives.

Those two questions have different answers, and occasionally opposite ones. A configuration that generates less in total can be worth more in the year, and the mechanism is entirely in the shape of the curve.

What the two curves look like

An equator-facing array produces a tall, narrow curve centred on solar noon. It climbs through the morning, peaks hard in the middle of the day and falls away through the afternoon. Most of its output arrives in a few hours around midday.

A split east and west array produces something much flatter. The east-facing half starts early and peaks in mid-morning; the west-facing half peaks in mid-afternoon and runs later into the evening. Added together they give a broad plateau with a shallow dip at noon rather than a spike.

The annual total from the split configuration is lower, because neither half ever points squarely at the sun at the moment the sun is strongest. How much lower depends on latitude and on roof pitch, and it is a meaningful reduction rather than a rounding error. But it is a reduction in a quantity that was never quite the right thing to measure.

Why the flatter curve is worth more

Household demand is not shaped like a midday spike. It has a morning peak when people get up, a trough while the house is empty or quiet, and a substantial evening peak when cooking, lighting and heating coincide. The equator-facing curve delivers its maximum precisely into the trough.

The split array delivers less at noon, which was going to be exported anyway, and more at breakfast and more in the late afternoon, which are hours the house can actually use. Self-consumption rises, and every unit moved from the export column to the self-consumption column is worth the difference between the two rates.

There is a second benefit that shows up on the equipment rather than the bill. A flatter curve means the inverter runs closer to a steady load for longer instead of hitting its ceiling for two hours and idling either side, which allows an inverter smaller than the array to be used with less clipping.

When the trade does not pay

It only pays if the household is there to consume the shoulders. A house empty from eight until six gains nothing from an east-west split; both configurations export the bulk of what they make, and in that case the higher annual total simply wins.

It also only pays where the gap between import and export rates is wide. Under net metering at parity, exported units are worth the same as consumed ones, self-consumption is irrelevant to the arithmetic, and maximising annual generation is straightforwardly correct. The whole argument for east-west depends on a pricing structure that penalises export, and that structure is a policy choice which differs by country and changes over time.

And a battery changes the calculation again, by absorbing the midday spike and releasing it in the evening. With storage, the shape of the generation curve matters much less, because the battery is doing the reshaping. A household considering both should not pay twice for the same fix.

What usually decides it in practice

For most people the choice is not free. A roof has the slopes it has, and if the equator-facing slope is large and unshaded, filling it is the obvious move. The east-west question arises mainly on houses whose ridge runs the wrong way, and there the useful conclusion is that this is not the disadvantage it is often presented as.

Where the ridge runs east to west and both slopes are usable, there is a genuine decision, and it deserves the four numbers rather than a rule. Compare import and export rates, look at when the house is occupied, and consider whether an inverter shared across both slopes is being asked to handle two peaks or one.

The broader point survives the specifics. An array should be designed against the household’s consumption profile and its tariff, not against a table of annual yields. The best roof is the one whose output the house can actually use.