Alectrona
Commercial solar system sizes · over 50 kWpCommercial solar divides into four capacity bands, and each one is a different project rather than the same project with more panels. Pick the band nearest your likely capacity and read what changes at that scale: the structural duties, the inverter strategy, the grid connection route and the metering.
This page is a selector. Commercial solar over 50 kWp is grouped here into four capacity bands, from 50 to 150 kWp up to 1 MW and above, because a 100 kWp array and a 700 kWp array are different projects rather than one project scaled up. Find the band nearest your likely capacity and read what changes there.
What separates one band from the next is engineering and regulation rather than panel count. Four things move as the array grows, and the same four are set out for every band below: the structural work the roof has to carry, the inverter strategy, the grid connection route, and how the site is metered and settled.
The bands are a way to navigate, not a menu to order from. Working out the capacity your building should actually have is a separate exercise, driven by your half-hourly consumption rather than your roof area, and the method is set out in full in our guide to half-hourly metering.
| Band | Indicative modules | Usable roof area | Indicative output | What activates |
|---|---|---|---|---|
| 50 to 150 kWp | roughly 85 to 280 modules | roughly 250 to 1,050 m² | roughly 42,000 to 145,000 kWh per year | G99 grid connection |
| 200 to 300 kWp | roughly 340 to 555 modules | roughly 1,000 to 2,100 m² | roughly 170,000 to 285,000 kWh per year | CDM 2015 and half-hourly metering |
| 500 to 750 kWp | roughly 850 to 1,390 modules | roughly 2,500 to 5,250 m² | roughly 425,000 to 712,000 kWh per year | RC62 shutdown and G100 export limit |
| 1 MW and above | roughly 1,665 to 1,850 modules per MW | roughly 5,000 to 7,000 m² rooftop, or 2 to 2.5 hectares ground-mount, per MW | roughly 850,000 to 950,000 kWh per MW per year | Grid queue and network charges |
There is no cost column, and that is deliberate. A meaningful figure for a commercial system comes from an on-site survey and a PV*SOL model of your roof and your load, not from a price per kWp. The figures above are indicative engineering rules of thumb, modelled for your site before anything is specified. Each band is set out in full below.
Even at the lower end the array is a real imposed load on the structure, and on a flat roof the ballast holding it down adds more again. The purlins, the deck and the spans are appraised, and on an older building the covering and its fixings alongside them, before the panel count is signed off. Across a run of multi-let units one design is templated over comparable roofs and the rollout phased unit by unit.
The array is split across a handful of three-phase string inverters, each carrying its own group of strings. That keeps the power-point tracking close to roofs that face different ways or shade at different times, limits a fault to one inverter’s share of the output, and makes the AC side easier to land on your existing three-phase distribution board.
Above the G98 notification ceiling of 16 A per phase, every system here needs a G99 connection application made to the network operator before the work goes ahead. The offer that comes back may grant your full export, cap it, or in a constrained area call for reinforcement, and the design is shaped around whichever arrives. Getting the application moving early is what keeps the project on programme.
A generator this size usually puts the site on half-hourly settlement metering, so the import the array offsets and any surplus it exports are measured in half-hour blocks rather than estimated against a profile. That granularity makes self-consumption auditable, and it is what an export route needs in order to pay you for surplus at all.
On a flat roof at this size the frames are ballasted or fixed to the roof’s structural loading and the local wind zone, and uplift rather than dead load is what governs the design. The work is a construction project in its own right, so the CDM 2015 duties on you as client, on the principal designer and on the principal contractor apply in full, and access, scaffold or mast-climbers become a real line rather than an afterthought.
The array feeds in through switchgear carrying the isolation, metering and G99-compliant protection the connection requires, tied into your main distribution board or the low-voltage side of the supply transformer. Where the board already sits close to its rating, or the transformer is sized tight to the present load, accommodating the array can mean an upgrade on your side of the meter.
The application stops being a formality here. Whether your incomer and transformer have the headroom to carry the array, and whether the local network can accept what you might export, is exactly what the enquiry resolves, so it runs alongside the roof survey rather than after it. Where the offer comes back limited, an agreed export cap engineered to G100 lets the array proceed instead of waiting on reinforcement.
The industry’s P272 change moved larger non-domestic supplies off profile estimation and onto actual half-hourly readings, so the way solar reshapes your demand across the trading day is measured rather than inferred. The same granularity exposes the Use of System charges on the bill, which are weighted toward winter peaks and the red-band hours of the working day.
An array here covers roughly 2,500 to 5,250 m², so the cable runs across the roof, the edge protection and the access provision are as much a part of the design as the frame. The mounting has to suit your particular roof build-up rather than a standard detail, and the insurer’s view of roof safety starts bearing on the specification as well as on how the work is run.
Past roughly 300 kWp, property insurers increasingly expect the array to meet the RC62 loss-prevention standard, which calls for the panels to be de-energised at module level so a roof full of live DC can be made safe for the fire service. In this band that tends to become a condition of cover rather than a recommendation, and it changes the inverter and DC architecture, so it belongs in the first design and not the last.
The application moves up a tier. The network operator runs studies and requires witness testing before the array may energise, and the offer can call for new switchgear, a dedicated cable run or, in a constrained area, reinforcement. Full export is rarely a given at this scale, so an export limit engineered to G100 is often what lets the connection be granted at all.
The site settles half-hourly, so distribution and transmission charges are visible against the trading day and are material at this size. Solar offsets daytime import directly, and pairing the array with storage moves load away from the charged peaks. With hundreds of strings on the roof, string-level monitoring against the modelled yield is what stops a failed section going unnoticed until a quieter bill arrives.
A megawatt needs roughly 5,000 to 7,000 m² of usable flat roof, or 2 to 2.5 hectares as a ground array. A roof reuses a structure you already have, subject to its load capacity, its age and its layout. Ground brings racking, foundations, fencing, trenching and civils, and the land itself. The two routes have genuinely different cost and programme shapes, so both are modelled before either is committed to.
At this scale the inverters, switchgear, metering and protection amount to an electrical installation with its own room and its own cabling design, and a connection at 11 kV brings high-voltage switchgear onto your side of the boundary. Storage is frequently designed in alongside, both to lift self-consumption against an export limit and to manage the charged peaks.
You are typically a Type B generator, frequently connecting at high voltage, with a formal application, network studies and mandatory witness testing before the array may energise. A larger project can need new switchgear, a cable run or a substation, and reinforcement where the network is constrained. The reformed queue weights readiness, such as planning progress and land control, so an evidenced application moves faster than a speculative one.
Transmission and distribution charges are a real line on a site this size, and the way they fall on half-hourly supplies has been moving through recent charging reviews. Where full export capacity is not available, a limited or zero-export connection means anything the site cannot use or store is curtailed, which is why the design follows what you consume rather than what the roof could theoretically generate.
None of the four axes is summarised above and then left there. The connection route is set out in full in our guides to the G99 application, the grid connection queue and G100 export limitation, with the network operator's own role explained in what a DNO is and what it decides. The structural side sits on the structural roof survey guide and the commercial mounting page, and the construction duties that apply from the first band upward on the CDM 2015 guide.
On the electrical side, the choice between string and central inverters is worked through in string versus central inverters, the platforms we specify are on the commercial inverters page, and the module-level shutdown insurers ask for above roughly 300 kWp is covered in the RC62 fire safety guide. Measurement and what it exposes are on the half-hourly metering and network charges guides, with battery storage the usual answer to a charged peak or an export cap. Once a system is running, monitoring and servicing sit under operations and maintenance, and the sequence a project follows from enquiry to switch-on is on the commercial process page.
Enough roof and enough weekday load to be worth doing properly, without yet crossing into the heavier grid and engineering regimes above.
A system that occupies a meaningful share of one industrial roof and feeds a substantial part of the site’s daytime demand.
The scale at which one large roof carries a serious generating asset, and where the insurer and the network operator shape the design between them.
No longer a roof retrofit with more inverters. A connection, a charging position and a funding route that have to be designed together.
What ties every band together is self-consumption. A unit you use on site displaces an expensive imported unit, while an exported one is paid far less, so the value of an array tracks how well its output lines up with your demand rather than how many panels the roof will hold. We confirm that match against your own half-hourly meter data before anything is sized, and where the daytime fit leaves a surplus we test whether storage or an export-limited design is the better answer.
Capacity follows your daytime load. A system sized to sit under the power your building draws through the working day is one whose output you mostly use on site, and self-consumed power is worth far more to you than exported power. That is the whole of the principle. The working behind it, the half-hourly data we ask for and the trade-off against filling the roof are set out in our guide to half-hourly metering.
The indicative arithmetic against each band gives you a feel for what a capacity means physically. As a rule of thumb, expect roughly 1.7 to 1.85 commercial modules of 540 to 600 W per kWp, so around 185 modules for a 100 kWp array; roughly 5 to 7 m² of flat roof per kWp once you allow for the spacing between rows; and a Yorkshire yield of roughly 850 to 950 kWh per kWp installed each year. These are engineering rules of thumb, useful for orientation and nothing more. Your real panel count, roof area and output come from a PV*SOL model of your specific roof, its pitch, its shading and its orientation.
Capital behaves differently across the bands, and we publish no rate for it here. The fixed elements of a project, the survey, the design, the grid application and the scaffold, spread across more capacity as the system grows, so the direction of travel is downwards, but that is a trend rather than a figure. What sits underneath that trend also changes as you go up. In the lower bands the roof dominates the figure: the mounting and the way it fixes, the access and scaffold, the cabling back to your board. By the upper bands the connection is the largest single variable, and a high-voltage connection calling for new switchgear, a cable run or reinforcement can pull against the scale economics on any given site. How a commercial figure is actually built, and what moves it, is covered in our guide to what commercial solar costs.
At the top of the range the question widens past what to buy and into whether to own the asset at all, because a developer-funded route is a genuine alternative once a project is measured in megawatts. We model ownership first and the alternatives beside it, so the comparison is made on your numbers. How the economics and returns work out is set out on the commercial finance page and in the ROI and payback guide.
Four things move as the array grows, and together they are why each band is treated as a different project rather than a bigger version of the last one.
Structure. Dead load, wind uplift and, on a flat roof, the ballast holding the array down all scale with capacity, so the structural appraisal gets heavier and strengthening or a revised mounting approach becomes more likely the further up the bands you go.
Inverter strategy. Smaller arrays run on a handful of string inverters sited near the array. Larger ones move to multiple inverters on a shared AC collection, and at the top end to a central arrangement with its own switchroom, transformer and cabling design, which changes the plant space the building has to give up.
Grid connection. The application to the network operator moves from a straightforward G99 submission to a study that can come back with a connection offer carrying a cost or a curtailment condition. On the larger bands that step, rather than the panels, is what sets the programme.
Metering and settlement. At commercial scale the site is on half-hourly metering, and the larger the export the more the settlement arrangements, any agreed export limit and the network charges bear on the design. The band tabs above set out which of these bites first at each capacity.
As engineering rules of thumb, at commercial modules of roughly 540 to 600 W and about 1.7 to 1.85 modules per kWp: 50 to 150 kWp is roughly 85 to 280 modules on roughly 250 to 1,050 m², making in the order of 42,000 to 145,000 kWh a year. 200 to 300 kWp is roughly 340 to 555 modules on roughly 1,000 to 2,100 m², making around 170,000 to 285,000 kWh. 500 to 750 kWp is roughly 850 to 1,390 modules on roughly 2,500 to 5,250 m², making around 425,000 to 712,000 kWh. 1 MW is roughly 1,665 to 1,850 modules per MW on roughly 5,000 to 7,000 m² of rooftop, or 2 to 2.5 hectares of ground, making around 850,000 to 950,000 kWh per MW. Every one of those is indicative orientation on a Yorkshire yield of roughly 850 to 950 kWh per kWp, never a fixed promise. Your real figures come from a PV*SOL model of your roof, its shading and your load, and the working is in our guide to how much roof space it takes.
Size to your consumption first. Filling the roof feels efficient, but every kilowatt-hour the array makes above what the building is using at that moment has to be exported, and export is worth a fraction of what you save by using the power yourself. A larger array does lower the cost per kWp, so there is a genuine trade-off, and the right answer is the size where you still self-consume most of what you generate. You can push that balance further by adding battery storage to move midday generation into the early evening, or by agreeing an export limit with the network operator. We model that trade-off against your real half-hourly data rather than assuming the roof should simply be full.
Not necessarily, and the larger the band the less likely it becomes. Every system over the G98 notification ceiling connects under a G99 application, and the offer the network operator returns may grant your full export, set a limit, or in a constrained area call for reinforcement. From roughly 500 kWp upward a limited offer is common. The engineered answer is export limitation designed to the G100 standard, which caps what the system pushes onto the network so the connection can still be granted while you keep the full benefit of the power used on site. Because a self-consumed unit is worth far more than an exported one, a system designed around your own load loses very little to a sensible cap. Whether you need one, and at what threshold, is for the network operator to confirm, which is why the connection enquiry runs before the design is locked.
Because of what the insurer is being asked to cover. Past roughly 300 kWp, property insurers increasingly expect a rooftop array to meet the RC62 loss-prevention standard, which calls for the panels to be de-energised at module level so a roof carrying live DC can be made safe for the fire service. Below that scale it reads as a recommendation; in the 500 to 750 kWp band and above it tends to become a condition of cover. It changes the inverter and DC architecture rather than sitting on top of it, so we design it in from the first drawing. Discovering the requirement after the layout is fixed is the kind of surprise that costs a project weeks.
It depends on the space you hold and the grid at your connection point. A rooftop array reuses a structure that already exists and sits behind your meter for direct self-consumption, but it depends on the roof's load capacity, its age and its layout. A ground array suits a site with land but not enough roof, at the cost of racking, foundations, fencing and trenching, and with a connection that turns on local network capacity and, usually, on full planning permission rather than permitted development. The two routes have genuinely different cost and programme shapes, so we model both against your site and your half-hourly demand before recommending either.
The pace is set by the grid connection, not by travel or the panels. The survey reaches you quickly from our Yorkshire base, and the design and install rarely hold things up. What governs the programme is the G99 application to your network operator, and that step scales with size: a larger array means a heavier application and sometimes network reinforcement, which the operator times. We will not invent a duration here. Your proposal carries a dated programme built around the connection your specific system needs, and the installation timeline guide walks through each stage.
Send us your half-hourly data and the site. We model the right size against your real load and come back with a designed system and a figure built for you.