Commercial PV systems: how a good one is specified

What you are actually buying

For a finance director or estates manager weighing a proposal, the first thing to understand is that a commercial PV system is not a product ordered by the metre. It is six subsystems, and each fixes a constraint the others live with for the life of the asset. Specification is where the return is won or lost; good installation cannot rescue a system sized against the wrong load.

Each part decides something different.

• The array. Panel wattage and layout set capacity in kWp, and orientation sets the shape of the generation curve. An east-west duo-pitch gives a flatter, wider curve than a south-facing roof, which often suits a building occupied through the working day better than a taller midday peak.
• The mounting. Non-penetrative clip-fix rails suit most profiled metal industrial roofs and normally preserve the roof manufacturer's warranty. On single-ply membrane the choice is mechanical fixing or ballast, and ballast adds tonnes of dead load to a structure never designed for it.
• The inverters. These set the electrical architecture: how many strings, how the array is zoned, what happens when one zone shades or one unit fails. Inverter count also decides how maintainable the system is years into its life.
• Isolation, protection and cabling. AC and DC isolators, surge protection, the connection into an existing board or a new one, and the cable route from roof to plant room. Spare capacity at that board is often the awkward part.
• Metering and export control. Generation metering, an import and export meter, and, where the connection agreement caps export, the control scheme that holds output at or below the agreed figure.
• Monitoring. String-level or inverter-level data, so a dead string shows up as a number rather than as an unexplained rise in your electricity bill months later.

Battery storage is a seventh element where it earns its place. It moves the self-consumption figure rather than the generation figure, and it is decided after the PV design settles.

The order the decisions have to be made in

Roof area is the last input, not the first. A proposal that opens with a panel count drawn off a satellite image says nothing about whether those panels will pay.

The order that produces a system worth owning runs differently. Your load shape comes first: when the building actually draws power, hour by hour and season by season, read from interval data rather than inferred from an annual kWh total. The network constraint comes second, because the connection answer or an export limit can cap the sensible size before the roof does. The survey comes third, deciding how much of the area is genuinely usable. Panel layout comes fourth.

This is why two competent specifiers can look at the same warehouse roof, arrive at capacities almost twice apart, and both be right. One has read a load that collapses in mid-afternoon and at weekends; the other, refrigeration running a base load through the night. Same roof, different economics.

Season matters as much as time of day. A UK array generates a large multiple of its December output in June, so a winter-peaking building such as a heated workshop self-consumes a smaller share of a given system than a summer-peaking office.

The marginal kWp, and where it stops paying

The design target for almost every commercial rooftop array is annual generation equal to 60 to 85 per cent of annual consumption. That is not a rule of thumb dressed up as engineering; it falls out of what each additional kilowatt-peak is worth.

Electricity generated and consumed on site displaces grid electricity at your full delivered rate, network charges and non-commodity levies included. Exported units earn an export tariff instead: a fraction of the delivered rate, not contractually stable across the life of the asset, and partly curtailed rather than sold on an export-limited connection.

So the first kWp is nearly all self-consumed and pays back fastest, and each one after it self-consumes slightly less. Past roughly 85 per cent of consumption the marginal panel exports most of what it makes and blended payback lengthens. A daytime-occupied building without storage self-consumes 55 to 75 per cent of what it generates, and pushing the array past the target drops that figure quickly.

A bigger quote is therefore not a better one. An array sized well beyond what the building can use shows a larger headline saving and a worse return on capital, which is the number your board will ask about.

What a properly sized system looks like in numbers

Expect the array to remove 30 to 60 per cent of total grid electricity cost, with payback of five to eight years. Factories on a day shift and refrigerated warehousing sit at the short end, because demand is high and steady through daylight. Offices and retail units that go quiet at weekends sit at the long end.

Reading the specification behind a quote

A quote is a price. A specification is what you get for it, and the gap between the two is where disputes live. These items make a document worth signing against.

• Capacity in kWp, panel count and panel wattage stated separately. Capacity drives everything; panel count on its own is marketing.
• Modelled annual yield in kWh with the assumption shown. UK rooftop arrays produce roughly 900 to 1,050 kWh per kWp per year, and anything above that needs explaining.
• The self-consumption percentage and where it came from. This is the most manipulated number in commercial proposals, and if it is absent the savings figure beneath it means nothing.
• A roof layout drawing with array zones, walkways, edge setbacks, plant and rooflights, plus the DC string arrangement.
• The mounting method named, whether it penetrates the covering, and a statement on the effect on any existing roof warranty.
• Inverter quantity, rating and the DC to AC ratio, with the modelled clipping loss stated. Oversizing the DC side can be a deliberate design choice; it should be shown as a number, not left implicit.
• The connection point, the spare capacity at that board, and whether new switchgear is included or excluded.
• The grid application route, who submits it, who carries the application cost, and what happens to the price if an export limit is imposed.
• Access and plant. Scaffold, mobile elevating work platforms and crane lifts are real costs, often left out.
• Warranties split out: panel product warranty, panel performance warranty with its length stated, inverter warranty, and workmanship warranty naming the insurance-backed scheme behind it.
• Commissioning and handover: test certificates, the connection paperwork, the operation and maintenance manual, monitoring credentials and as-built drawings.

Items marked TBC before survey are not automatically a problem; several genuinely cannot be fixed at that stage. A document where they are absent entirely is the problem, because then the risk has not been priced. It has just been left with you.

Sites where the answer should be no

Some buildings should not have solar on them, and it is cheaper for everyone to say so before a survey is paid for.

• A roof with only a north-facing pitch. A due-north array gives up a large share of what a south-facing one would produce at the same tilt, and payback stretches beyond the point where the case holds. East-west duo-pitch is fine; north-only usually is not.
• Shading that cannot be designed around: taller neighbouring buildings, mature trees to the south, lift overruns, large plant. Zoning and inverter selection soften the loss but do not remove it, and a roof shaded through the middle of the day is a poor host.
• A covering with less life left in it than the payback period. Mounting an array on a roof that will need replacing first means paying twice to strip and refit it. Re-roof first, then install.
• Very low daytime demand. A warehouse that is lit and little else, a site that runs at night, a building that shuts for weeks at a time. Self-consumption drives the return and there is little here to consume.
• A lease with less time to run than the payback period, and no agreement with the landlord about what happens at the end of it.
• Listed buildings, and street-facing arrays in conservation areas, where Listed Building Consent or a full application is needed and the answer may reasonably be no. Permitted development rights do not reach every site.

If your site is on that list, a desk feasibility will say so before you have spent anything.

Making two competing proposals comparable

Competing proposals are rarely comparable as written, because each embeds its own assumptions in the savings column. Three normalisations fix that.

• Reduce each price to £ per kWp and compare against the going rate for that size band, which our cost page carries. A figure well under it usually means something has been excluded rather than that someone is cheaper.
• Restate both yields at the same kWh per kWp. If one has assumed 1,150 kWh per kWp, above the 900 to 1,050 a UK rooftop actually delivers, and the other 950, the savings columns are not measuring the same thing.
• Restate both at the same self-consumption percentage, then recompute the annual saving yourself at your own current unit rate. That single exercise resolves most apparent gaps between quotes.

Then look at what is missing rather than what is listed. Scaffold, switchgear, the network application fee, structural survey, asbestos survey, out-of-hours working where the building cannot take a daytime shutdown. The proposal that looked tens of thousands cheaper frequently is not, once those come back as variations.

What the paperwork has to itemise for your accountant

A single invoice line reading 'solar installation' creates work later. Solar PV is a special-rate pool asset for capital allowances, and the route to relieving the whole cost in year one is the Annual Investment Allowance, which lets a profitable company deduct qualifying capital cost from taxable profit in the year of expenditure, up to the annual AIA limit. It is not full expensing, and a proposal claiming otherwise has the tax treatment wrong.

So ask the specification to separate plant and equipment from any genuine building work done at the same time, such as a roof repair, and ask for VAT shown separately. If the system is funded rather than bought outright, allowance treatment depends on the structure, so that conversation belongs with your accountant before contracts are signed.

One further line for the board pack. Commercial premises with a working array typically show a 5 to 15 per cent uplift in value, and relocating that capability to another site costs roughly 15 to 25 per cent of the original installed cost. Neither figure belongs in the payback calculation.