Sizing solar and battery systems

Sizing is where solar and battery projects are won or lost, and it happens before any hardware is chosen. A correctly sized system is an asset that pays for itself by displacing expensive imported power. An oversized system is a machine for generating low-value export, bought at full price. The difference between the two is not visible in the quote. It is visible in your half-hourly consumption data, which is why sizing without that data is guesswork with a capital budget.

The core economics are fixed by one asymmetry: electricity you generate and use on site displaces the full delivered import price, while electricity you export earns only the export rate. Self-consumption is therefore the engine of every solar business case. The right system size is the one that keeps self-consumption high across the year, including the weekends, holidays and seasonal troughs that an annual average hides.

Why oversizing happens

The incentives lean that way. The party doing the sizing usually profits from the system size, roof area presents itself as the natural constraint, and annual consumption figures make generous sizing look justified. None of this requires bad faith. It just requires that nobody in the transaction is paid to say smaller. The result is a market with a persistent bias towards systems larger than the load behind them.

One site we modelled came into the conversation convinced it exported around 17% of its solar generation. That figure had already shaped the thinking on the roof: a battery pitched to capture the surplus, an export case built around it. The half-hourly data told a different story. Actual export was under 1%, at most 0.7% of everything the array produced, because a large, continuous base load was absorbing almost all of it as it was generated. The site was self-consuming roughly 96% of its own solar without any help from storage. Nothing about the roof or the array had changed. What changed was which system made sense once the assumption was replaced by the meter. A battery sized to manage a 17% export surplus that barely existed would have been an expensive way to solve a problem the site did not have.

Batteries: size for the job, not the brochure

A battery has no single purpose; it has candidate jobs, including shifting consumption out of expensive periods, increasing solar self-consumption, managing demand peaks, and providing resilience. Each job implies a different capacity, power rating and configuration. Sizing a battery before defining its job produces an asset that does several things poorly. The job definition comes from the same place sizing does: the half-hourly data and the tariff structure it interacts with.

Tariff structure decides which job is worth paying a battery to do. On a flat import rate, a battery’s main opportunity is limited to shifting demand away from peak charges and levies, because there is no price spread within the day to exploit. Move the same site onto a day-ahead tariff, where the import price tracks the wholesale market half hour by half hour, and the battery gains a second job: buying cheap and discharging expensive, on top of the network charge avoidance it was already doing. We modelled exactly that combination for a client portfolio moved onto a day-ahead structure, where the tariff switch alone saved 12.0% against the previous flat-rate contract, and adding battery storage against the day-ahead price on top of that took the combined saving to a 3.3-year payback. The tariff did not just change the savings figure. It changed what the battery was for. Sizing a battery without knowing which tariff it will sit under is sizing it for the wrong job by default.

The modelling standard to insist on

Before committing capital, require a model that uses at least a full year of your metered half-hourly consumption, realistic generation for your location and orientation, your actual tariff structure on both import and export, and honest treatment of degradation and downtime. Then require the model to be free of any interest in the outcome. A model built by the party selling the system fails that last test by construction, however good its engineering.

To have a proposed system independently modelled against your real consumption, request a benchmark at /benchmark or book a review at /book.