When commercial solar battery storage earns its place

Where the return on a solar project actually comes from

If you are signing this off, the figure that decides the investment is not panel wattage or roof area. It is the proportion of what the array generates that gets used inside your own building. A unit consumed on site cancels a unit you would otherwise have bought at your full commercial rate, levies and standing charges included. A unit that leaves the meter earns your export tariff, a fraction of the price you pay to import. That difference is the whole game. Two buildings with identical 200 kWp arrays and identical generation can sit years apart on payback, purely because one uses what it makes and the other sells it cheaply.

That gap is the entire reason storage comes up. A battery generates nothing. It moves energy in time, from the middle of the day when the roof produces more than the building can absorb to the hours when the building is buying from the grid. Same roof, same array, same annual generation. What changes is where the units end up.

So the test is narrow, and it cuts both ways. Storage earns its place when there is a real volume of solar spilling to export and a real volume of demand outside generating hours to take it. Miss either half and the capital is better spent on more roof, on the plant driving your load, or on nothing at all.

Reading your load shape against the generation curve

A UK array does most of its work through the middle of the day, and far more of it between April and September. Your building has its own shape: when the plant starts, when the second shift comes on, what runs overnight, what the meter does on a Sunday. Lay one curve over the other and the answer is usually visible before anyone quotes a price.

Four or five patterns in the metered demand tell you a store will find work to do.

• A flat overnight baseload: refrigeration, compressed air, servers, or handling equipment on charge. The strongest single indicator, because it draws every night of the year, not seasonally.
• An evening peak: kitchens, laundry, function space, floodlighting, or a late shift starting as generation falls away.
• Weekend demand in a building whose roof generates all weekend regardless. That export is the most predictable volume for a store to recover, because it repeats every week whatever the trading pattern.
• A long summer export shoulder, where the meter runs backwards for hours a day across whole weeks.
• A pre-dawn winter ramp, though a store will rarely hold enough charge in December to serve it.

The quantity that decides the sizing is not peak kW. It is kilowatt-hours in the wrong place: the annual volume you would otherwise export cheaply, and the annual volume you import when the roof cannot cover it. A battery can never be worth more than the smaller of those two, and that single sentence eliminates a good number of proposals.

This needs a full twelve months of interval data, not a fortnight. Your supplier will release them on request. A year matters because a store sized on a June week sits idle in January, and one sized on a January week is full by mid-morning all summer.

Size it from the spill, not from the nameplate

The most expensive mistake in commercial solar battery storage is buying capacity by rule of thumb. Matching kWh to kWp, or taking an hour of peak array output, produces a number that sounds proportionate and has nothing to do with your building. Capacity that never fills earns nothing and costs the same as capacity that cycles every day.

Take a 250 kWp warehouse array. At UK yields of 900 to 1,050 kWh per kWp it generates roughly 225,000 to 262,500 kWh a year. At 65% self-consumption, something like 79,000 to 92,000 kWh leaves the site annually. Averaged across 365 days that reads as 215 to 250 kWh a day, which is where the tempting sizing figure comes from. The average is a fiction. Almost all of that export falls between May and August, so a clear June Saturday produces a surplus many times larger than a December day, when the array generates little and the building absorbs it all.

The real question is not how much spills, but how much of it the building can take back before the next morning. If demand between dusk and sunrise is only 40 kWh, a 200 kWh store discharges a fifth of itself and the rest of the capital sits still. Sizing runs off recoverable overnight demand first, then gets sense-checked against the summer surplus, never the other way round.

Ask for two modelled numbers before approving any capacity: full-equivalent cycles per year, and kWh discharged per year. If the cycle count is low against the store's rated capacity, the store is larger than the job it has been given, and the modeller should justify the sizing.

The losses that do not appear on the front page

Every unit that passes through a battery comes out smaller. Inversion, cell losses, thermal management and the store's own standby draw each take a slice. The number to insist on, in writing, is round-trip efficiency measured at the AC terminals, because DC-to-DC figures flatter and are not what your meter sees. Apply whatever AC figure the manufacturer declares to every stored unit in the model.

Then apply that loss to the correct price. A stored unit is not worth your full import rate, because it was saleable anyway. It is worth the difference between your import price and your export tariff, multiplied by round-trip efficiency. Run that arithmetic early and a fair number of storage cases deflate.

Degradation is the second omission. Cells lose capacity with cycles and with age, and warranties are written in years and cycles together, with the lower of the two ending cover. Panel performance warranties commonly run to 25 years. Battery warranties do not, so a model that runs both to year 25 with no replacement or augmentation line in it is not a model.

None of this makes storage a bad purchase. It makes an unmodelled one a bad purchase, which is a different thing entirely.

When the right answer is no battery

For a large share of UK commercial buildings, a battery lengthens payback and should not be bought. Where the building runs weekday daytime hours and goes dark in the early evening, generation and demand already overlap for most of the year: self-consumption sits near the top of the 55 to 75% band and little spills out. Storage there buys a small increment of saving with a large increment of capital. Solar alone typically returns inside the usual 5 to 8 year range, and a battery it does not need only pushes that out.

If the modelling says the array alone is the better investment, the proposal says the array alone and shows the workings, so the conclusion can be challenged. No battery gets recommended here on the basis that storage is the direction of travel.

One further case is worth naming. Storage is sometimes proposed as the fix for an export constraint on the grid connection, and the connection answer can cap the sensible size of the scheme. The units a constraint curtails are units you would have sold cheaply, so a battery bought to recover them only pays where genuine after-dark demand is waiting.

Battery-ready first, battery later

For most projects the strongest position is a solar installation designed so a store can be added cleanly later. It costs little at design stage and a great deal to retrofit into a finished installation.

• An AC-coupled architecture, so commercial batteries can be added later without touching the array, its inverters or its commissioning records.
• A spare way in the switchboard, labelled and left free, with metering positions at the incomer ready to control a store.
• Reserved floor space or an external plinth, with access, ventilation and fire separation agreed with your insurer at design stage.
• Cable route and containment installed while access equipment and the installation team are on site.
• A grid connection application scoped so that adding storage later is a variation rather than a fresh agreement.
• Circuit-level monitoring from day one, recording what the building really does with the array running.

That last item is the argument. The best data for sizing a battery is twelve months of your own array operating: what it generated, what the building took, what left the meter and when. Buy the store on day one and it is sized from a model. Wait a year and it is sized from measurement, on an asset already paying back. Storage prices and the value of flexibility move too, so a store that looks marginal today can become clearly worthwhile within the array's life.

How the decision gets made on your project

The sequence is deliberately unexciting. Twelve months of interval data and recent bills, an array modelled to cover 60 to 85% of annual consumption, then two costed cases side by side: PV only, and PV with a store sized to your load shape. Each case states self-consumption, exported volume, annual saving and payback, with the assumptions visible.

A third case is usually worth pricing: PV built battery-ready, with the store deferred and the cost of adding it later stated plainly. That case is often the one that survives scrutiny. If you are holding competing quotes, reduce each to a price per kWp before comparing; the cost page at /cost/ shows the ranges. Annual Investment Allowance treatment can apply to storage capital as it does to the array, letting a profitable company write the qualifying outlay off against taxable profit in year one, up to the annual AIA limit. Treatment depends on your circumstances, so confirm it with your accountant. That improves both cases without changing which one ranks higher.

If storage does not stack up on your building, the document will say so, with the workings attached. Desk feasibility of this kind costs nothing and commits you to nothing.