What size solar system does your business actually need?

The principle that drives commercial sizing is self-consumption. A kilowatt-hour your building uses the moment it is generated offsets a full unit of imported electricity, which on an open-market business tariff is the expensive number on your bill. A kilowatt-hour you export instead earns only whatever export rate you can secure, and for commercial supplies that figure is typically a good deal lower than what you pay to import. Every unit you can shift from export to self-consumption captures that spread, and that spread is where the return lives.

So the target is not the maximum the roof can hold. It is the steady midday load, the base demand that is reliably present while the sun is up. Size the array so its peak output lands at or below that daytime load and most generation is consumed on the spot, with little spilling into low-value export. Sites that run a flat daytime load, manufacturing lines, cold storage, food retail, refrigeration, suit this best, because their demand sits right under the generation curve. A site that is busy mainly after dark self-consumes far less of the same array, and should be sized down accordingly or paired with storage.

The honest way to find the figure is to overlay two curves. The first is your site's demand through the day, read from half-hourly meter data; the second is the generation an array would produce on your roof, modelled across the seasons. The self-consumed energy is the area where they overlap. Anything generated above your load curve at that moment is export.

That is why the sizing exercise needs at least twelve months of half-hourly or AMR meter data, profiled in design software against the modelled solar curve. A full year matters because both curves move with the seasons: winter generation is a fraction of summer, and your own load may rise or fall across the year. A monthly bill cannot show this. Two sites with identical annual consumption can carry completely different arrays depending on whether the load sits in daylight or after it. Our half-hourly metering guide covers how to get that data and what it shows.

For domestic solar there is a loose convention of sizing a touch above annual use. It does not transfer to commercial. Right-sizing a system over 50 kWp is an engineering exercise that needs your interval data and your tariff structure rather than a multiplier read off a chart. Where export is limited or poorly paid, the design should bias deliberately toward self-consumption rather than carry a generation buffer.

Be wary of any guide that quotes a fixed self-consumption rate or a kWp-per-base-load figure. Illustrative sector ranges do circulate, cold storage in the nineties, manufacturing in the seventies to mid-eighties, hotels and restaurants lower, pure offices lower still, but these are averages from secondary sources, not a guarantee for your building. A business with significant overnight operation that assumes a high figure will see its savings materially overstated. The only reliable basis is your own half-hourly profile. One related caution: MCS look-up tables for self-consumption are built around domestic occupancy patterns, when a home is empty or occupied, and are not a substitute for site-specific analysis on a commercial project.

There is a hard ceiling above the load target, and it is the grid. A system above the per-phase threshold needs an ENA Engineering Recommendation G99 application accepted before it can energise, and for a system over 50 kWp the network operator typically requires witness testing. The point that matters for sizing is that the DNO can impose an export limit below your array's nameplate where local network capacity is constrained. That cap is then enforced with G100-compliant export limitation.

So a roof-filling array is not only worth less per exported unit, it can be physically capped at the meter, which reinforces the case for sizing to self-consumed load in the first place. Where the daytime base load is smaller than peak generation but total daily consumption is high, battery storage can extend self-consumption beyond the live overlap by holding midday surplus for evening or early-morning demand. We model whether storage earns its place rather than overbuilding an array that would export cheaply. Once the target capacity is set, translating it into a panel count and roof area is the next step, covered in how many solar panels.

Array size and inverter size are two separate numbers, and the relationship between them is its own sizing decision. The panels are rated in kilowatts of DC, the inverter in kilowatts of AC, and most commercial designs fit more DC panel capacity than the inverter's AC rating. That deliberate overbuild is the DC-to-AC ratio, sometimes called the inverter loading ratio, and a modest ratio above one is normal practice rather than a fault.

The reason is that a panel almost never produces its full nameplate output. Standard Test Conditions assume 1,000 watts per square metre of irradiance at a 25C cell temperature, and a UK roof rarely meets both at once. Real output sits below nameplate for most of the year, so an inverter sized to the panels' label would run well under capacity nearly all the time. Fitting more DC behind it fills the morning, evening and overcast hours when irradiance is low, and the few peak moments where DC would exceed the inverter's AC limit are clipped. A small amount of annual clipping is an accepted trade for a fuller generation curve, and on a self-consumption design that fuller curve is exactly what lands more units under your daytime load.

The ratio that suits your roof depends on orientation, pitch and how flat your daytime demand is. An east-west split benefits from a higher ratio because the two planes peak at different times and rarely clip together; a single south plane clips sooner. We set the ratio in the PV*SOL model for your specific roof rather than carry a figure across from another site, and the same logic underpins why commercial-size panels differ from the domestic kit. Where the network imposes an export cap, a higher DC ratio also keeps the inverter nearer its limit for longer, which we weigh against the G100 export limitation that enforces that cap.

The right size is set by a full year of generation rather than a peak summer day, because UK output swings hard across the seasons. A commercial array delivers the great majority of its annual yield in the summer half of the year, with winter months producing only a fraction of a clear June day. Sizing to the summer curve alone overstates what the array does in the months when your bill may actually be highest, so the model has to run the whole year against the whole-year load.

The anchors we model from are an annual specific yield in kilowatt-hours per kilowatt-peak, and a performance ratio that accounts for temperature, soiling, cabling and inverter losses, typically in the region of 0.80 to 0.85 for a well-designed commercial array. These are starting points for the model rather than promises. The generation curve itself is built from long-run irradiance data for your location, drawn from datasets such as the Met Office and the European Commission's PVGIS tool, then reconciled against the actual yield of comparable installed systems rather than left as a theoretical figure. Because winter generation is low, a site whose load peaks in winter, heated warehousing for instance, will see a smaller share of its annual demand met by solar than its annual kilowatt-hour total suggests, and the array should be sized with that in mind rather than to the headline yearly figure. This seasonal shape is also why the roof area available matters less than it looks, a point covered in how much roof space, and why half-hourly metering across a full twelve months is the input we insist on.

Yes, and it is a constraint a finance director should know about before settling on a capacity, because it can move the answer independently of the load and the grid. Most rooftop solar on commercial buildings is permitted development under the Town and Country Planning (General Permitted Development) (England) Order, which means it does not need a full planning application provided it stays within the limits in that Order. Those limits cover how far the panels project from the roof surface and proximity to the edge, and they tighten on listed buildings and in conservation areas, where the permitted development right may not apply at all.

For a larger ground-mounted array the position is different again, and above certain capacity thresholds a standalone solar scheme can require full planning permission rather than permitted development. The thresholds and the precise projection limits are set in the Order and are subject to amendment, so we treat them as orientation rather than settled fact and confirm the current position with the local planning authority for your site before fixing the design. The practical effect on sizing is that the planning envelope can cap the array below what the roof or the load would otherwise allow, particularly on a heritage-sensitive building, and that ceiling is checked alongside the structural survey rather than discovered late. We set out how the survey, design and consents fit together in design and engineering.

Often it should, because the daytime load you meter today is not the load the asset will serve for its full life. Many commercial sites are electrifying, adding workplace EV charging, swapping gas process heat for electric, or fitting heat pumps, and each of those lifts daytime demand in a way that raises the self-consumption a larger array can capture. Sizing strictly to today's flat base load can leave a building under-provisioned within a few years, exporting nothing it could have used because the array was built for the demand it had rather than the demand it was about to grow into.

The honest way to handle this is to model a credible future load alongside the current one, not to inflate the array on a hope. Where a fleet is genuinely moving to electric, the charging schedule that we would model against feeds straight into the sizing, and our EV charging work exists to make that demand land in daylight where the array can serve it. A string-based design also lets capacity be added in phases as the new load materialises, so the array can grow with the building rather than be fixed in one go, and translating the agreed capacity into a panel count is covered in how many solar panels. Whether the extra capacity earns its place is a modelling question we answer against the projected load and the relevant finance route, not a default to building big.