The winter wall: why 80% energy self-sufficiency is affordable and 90% is not

Image above: AI impression, not a construction drawing.

What does it honestly cost for a small residential community to supply its own energy? Not as a feeling, as a number. In the lab I simulate a neighbourhood of eight households, anchored in Breda, the Netherlands, on the real 2022 weather year from the Dutch national weather service and official consumption profiles. The model now also covers the largest energy item of a Dutch household: heat. Heat pumps, hot water, a heating curve that follows the outdoor temperature.

That is not a detail. Without heat, the self-sufficiency figures were quietly optimistic. With heat, the base neighbourhood drops from 57.9% to 45.5% self-sufficiency, because the electric winter demand lands exactly where the solar output disappears.

So the question becomes: what does it cost to close that gap? I let an optimisation model search for the cheapest route to every self-sufficiency level, with perfect knowledge of the whole year. That gives a hard lower bound: it does not get cheaper than this.

Up to roughly 70% self-sufficiency the route is almost boring. Solar up to the roof ceiling (80 kWp for the whole neighbourhood), a modest shared battery, done. The marginal cost of an extra percentage point is close to zero there.

Above that, everything tips over:

• 80% self-sufficiency costs about 6,100 euros per year for the whole neighbourhood.
• 90% costs 117,000 euros per year. The model wants a battery of over 4,000 kWh for that, and the shadow price (what one extra percentage point costs) jumps to 34 euros per kWh of avoided import.

That is not a gradual slope, that is a wall. The roof ceiling binds first: beyond 80 kWp there is simply no room for another panel, so every additional percentage point has to come entirely from storage. And storage that is only truly needed a few times a year is absurdly expensive per avoided kilowatt-hour.

The cause is seasonal mismatch. In the Netherlands, solar panels produce the most in the months when an all-electric neighbourhood needs the least, and close to nothing in the weeks when the heat pumps run at full power. A battery is a tool for day and night, not for July and January. Trying to bridge winter with daily batteries means buying thousands of kilowatt-hours of storage that sit idle eleven months a year.

The policy translation is concrete: all-electric neighbourhoods without seasonal storage should aim for 70 to 80% self-sufficiency, not 90% or more. That is a defensible design number that no solar calculator produces, because those calculators work per home and per year, not per neighbourhood and per hour.

Honesty about the edges of this result:

• The optimisation model runs with perfect foresight of the entire year. The curve is therefore a lower bound; a real control strategy never quite reaches it.
• Weather year 2022, a single year. Sensitivity to a cold or dark year is still open.
• The heat parameters are market indications in the range of the Dutch building standard, not yet calibrated on measured homes.
• Net metering is switched off, because it ends on 1 January 2027, exactly the regime in which this question becomes urgent.

Every run behind these numbers closes the energy balance every hour to machine precision; a result that breaks the balance is a bug, not a breakthrough.

And then the real question: is the wall final? No. There is one component that breaks it, and the effect is larger than I expected. That is part 2 of this diptych.