Commercial solar in Sheffield, South Yorkshire and the Humber

The conventional route to a system size starts at the roof: measure the area, subtract the plant, rooflights and access zones, divide by module footprint, and a kWp figure falls out. On a Sheffield process site that is the wrong end of the problem. A stainless finishing line or a heat-treatment plant draws a load that a roof of that footprint cannot come close to covering, so the array is limited by what the building can hold rather than by what the occupier can use. The question inverts. It stops being how much of this generation can you absorb, and becomes how much of your import can we physically displace.

That is why we ask for twelve months of half-hourly consumption before a design is drawn. Three things in that file set the shape of the scheme. The overnight minimum gives the true baseload, the floor the site never drops below. The weekday daytime plateau gives the ceiling the array will work under. The ratio between them, the site’s load factor, tells us whether we are designing for a plant that runs flat around the clock or a building that wakes at seven and goes quiet at four.

Where the overnight minimum already sits above the array’s peak output, every kilowatt-hour generated is consumed on site at the price you would otherwise have paid to import it, in every half hour of every day of the year. Self-consumption stops being a modelling assumption and becomes arithmetic. That position is unusual nationally and common enough along the Don Valley to be worth testing first, because it changes the export strategy, the inverter selection and the case for storage all at once. None of it can be read off a roof plan or an aerial image: we fly the roof, model the array in PV*SOL against your metered profile and design to BS 7671, and the size follows from the two together.

Grouping sites by trade is a poor guide at this scale, because two neighbouring buildings in the same industry can behave completely differently at the meter. Grouping them by load shape is far more useful, and across the Lower Don Valley and the estates around it four shapes cover most of what we see.

The first is continuous process: melting, holding, heat treatment, anything with a furnace it is uneconomic to cool down. Demand is close to flat around the clock, the load factor is high, and an array simply removes a slice off the top of a large steady number. The second is single-shift machining and fabrication, typically starting between six and seven and finishing mid-afternoon. That draws a square block of daytime demand which fits generation well in the middle of the year and poorly at each end of it, because in December the shift both starts and finishes in the dark.

The third is two or three shift working, where the daytime block is joined by evening and night blocks the array cannot reach at all. The daytime half of that profile is still a good match; the night half is outside the conversation entirely, and pretending otherwise is how oversized schemes get sold. The fourth is the technical and research estate, where a moderate base load is punctuated by short, heavy rig and test loads. Those spikes barely move annual energy, but they do set the site’s maximum demand, and a rooftop array is not a reliable way to shave a peak that might fall at seven on a January morning.

Which of the four a building sits in tells us more than its roof area does. Larger South Yorkshire sites usually contain more than one, because a works will run a continuous furnace alongside single-shift machining, and the meter shows only the sum. Separating them out is part of the modelling.

Northern Powergrid is the distribution network operator for every town on this page, including Scunthorpe and Grimsby on the south bank of the Humber, because the Yorkshire licence area covers both banks of the estuary. Every system above 50 kWp connects under a G99 application made and accepted before the array energises. On heavy-industrial sites across this area the application is frequently assessed against a connection built for a process load far larger than anything the array will export, which is a very different starting position from a building fed off a residential primary or a rural feeder on the eastern plain. We confirm the position at your actual point of connection rather than assuming the pattern holds.

The reason is historical. The Lower Don Valley network was laid in to feed melting shops, rolling mills and forges, and that infrastructure is still in the ground. Many of the larger sites along the S9 and S60 corridors take a high-voltage supply through their own customer substation and transformer, with an agreed import capacity set for a plant that drew considerably more than the present occupier does. An array of a few hundred kilowatts behind that connection is a small thing relative to the assets it is connecting to.

Two cautions belong with that. Import capacity is not export capacity: a generous agreed supply capacity does not convert into permission to export the same figure, and the generation is assessed on its own terms. And an old connection is not automatically a healthy one, so the age of the on-site switchgear and the condition and discrimination of the protection are design work, not a detail to be discovered at commissioning.

Where a site does hold a flat baseload above the array’s peak output, the export question can be closed cheaply. A G100 export limitation scheme set to zero export costs a plant like that close to nothing, because there is no surplus to export in the first place: the process consumes the output as it is made. That turns what is often a difficult conversation elsewhere into a simple one here, and it is worth testing early, because an application seeking no export is a materially easier thing to progress than one seeking firm export capacity.

There is a second argument that only applies to sites of this kind. Where a plant already runs close to its agreed import capacity, and an expansion or an added furnace would trigger reinforcement, generation behind the meter cuts daytime import and can defer that upgrade. That is a capacity argument rather than an energy one, it is often worth more than the avoided units, and it comes out of the same half-hourly file.

An honest model shows where the match breaks down as clearly as where it holds, and on South Yorkshire process sites it breaks down in three predictable places. The first is the annual shutdown. Much of the engineering trade here still closes for a fortnight in the summer and again over Christmas, and the summer break falls directly on the highest-generating weeks of the year. A model that assumes the plant runs every weekday overstates self-consumption, and the overstatement lands exactly where the generation is largest. We put the shutdown weeks in explicitly and show the effect rather than averaging it away.

The second is the weekend. A five-day plant stands idle roughly two days in seven, and on those days the array produces into a site drawing little more than security lighting and standby. Depending on the export position that output is sold at whatever the terms give, curtailed, or moved: compressed air topping up, handling-equipment charging and batch heat treatment can sometimes be scheduled into a Saturday where the process allows it. Whether that is worth doing is a production-planning question, and we would rather put the numbers in front of you than assume the answer.

The third is the night shift, where the answer is simply that solar does not serve it. A night block changes the proportion of annual consumption an array can reach; it does not by itself decide whether there is a case. Storage gets the same treatment. Where the daytime baseload already exceeds what the roof can generate there is no surplus to store, so a battery does nothing for self-consumption there and its case has to be made on other ground, such as capacity charges or resilience. We model that separately rather than folding it into the solar figures.

At Waverley and Catcliffe, just over the Rotherham boundary, the Advanced Manufacturing Park has put a different kind of industrial building on the ground: high-bay research and production halls, university-led research centres, test cells and pilot lines, built from the early 2000s onward rather than inherited from the steel era. Sheffield Business Park off the Parkway carries a related mix of technical and single-storey commercial units, and the newer terraces around the Olympic Legacy Park at Attercliffe sit in the same family.

Electrically these buildings behave nothing like a melting shop. Connected load is large but diversity is low, because a handful of big machines dominate: overhead cranes, machine tools, induction and resistance heating, test rigs, and the compressors and process chillers supporting them. A hall can sit at a modest base for hours and then take a substantial step when a rig starts. That matters for inverter selection and for protection settings, because the inverter sees a supply that moves under it, and it matters for sizing, because annual energy and maximum demand say different things about the same building.

Occupancy structure carries more weight here than on an owner-occupied works. Much of this stock is multi-let or held on leases shorter than the operating life of an array, so who owns the system, who consumes the output and what happens at lease end has to be settled before a design is worth drawing. The same question runs through the public-sector estate, where the roof and the electricity bill often sit in different parts of the organisation. For the production halls themselves the approach is the one set out for manufacturing sites: start from the metered process load, not the floor area.

North of the Don the load story changes, and it changes twice. The Templeborough and Sheffield Road corridor between Rotherham and the Sheffield boundary is downstream metals: rolling, drawing, coating, forging and fabrication rather than melting. That work is batch rather than continuous, so demand is lumpier than on a furnace site and the daily profile has real structure to it. It makes the half-hourly file more informative and the sizing more sensitive to reading it properly.

Further north again, the Dearne Valley estates at Manvers, Wath, Cortonwood and Goldthorpe were built on reclaimed colliery land from the 1990s onward, and the occupiers are lighter: packaging, plastics, food, engineering supply chain, and service and administrative operations. Demand per square metre of roof is a fraction of what a metals plant draws and it is far more strongly diurnal. Here the surplus is genuine, midday output does exceed site demand for part of the year, and the export terms and any load-shifting materially change the answer. It is close to the opposite of the Don Valley case, on estates twenty minutes apart.

One ground-engineering point runs beneath all of it and belongs in the first conversation rather than the last. The former South Yorkshire coalfield leaves shafts, adits and shallow workings under a great deal of the land at Manvers, Cortonwood, Elsecar and around Barnsley, and the records are held by the Coal Authority. For a rooftop array that is somebody else’s concern. For a ground-mount scheme it is not, because driven piles or screw foundations into made ground over recorded workings need a coal mining risk assessment, and in a development high risk area a permit before any intrusive ground work. It is a routine step here and a slow one if it surfaces late.

This page reaches a long way east of the Don, and the reason is a road. The M18 leaves the M1 at Thurcroft, picks up the Doncaster interchange at Rossington and Armthorpe, and becomes the M180 beyond Hatfield, which runs to Scunthorpe and on as the A180 to Grimsby and Immingham. The same corridor carries the rail freight route to the Humber ports. North of it the M62 runs out through Goole to Hull. Almost every large commercial roof between Sheffield and the sea sits within a few miles of one of those two lines, which is why they are the sensible way to describe the coverage rather than a county boundary.

The county line is real and we would rather name it than let a page imply otherwise. Scunthorpe is in North Lincolnshire and Grimsby is in North East Lincolnshire. Both are unitary authorities, both sit in the ceremonial county of Lincolnshire, and neither is in Yorkshire. One thing changes at that line and one does not. The determining authority changes, and there are seven separate planning authorities across this page. The electricity distribution licence area does not: Northern Powergrid’s Yorkshire licence covers the south bank as well as the north, so the operator, the G99 route and the export-limitation options are identical from Attercliffe to Immingham. We still confirm the operator against an MPAN lookup for the specific supply before lodging anything.

What does change across that distance is the specification rather than the process. A works in the Lower Don Valley, a big-box shed at Rossington, a cold store at Pyewipe and a transit shed on the Goole dock estate are four different buildings in four different exposures, and the fourth of those is standing in salt air on open ground with nothing upwind of it. Exposure and atmospheric corrosivity are assessed for the actual site before a bill of materials is written, because they change what the array is made of rather than where the modules go. What that assessment produces, the exposure terms and the bill of materials set by the corrosivity category, is on our coastal and exposed sites page. Our drone survey travels to Grimsby and Hull exactly as it does to Tinsley.

South Yorkshire carries a deep stock of commercial solar from the feed-in tariff years, much of it fitted between 2012 and 2016 on manufacturing and metals roofs. Those arrays are long past the point where the original installer’s involvement ended, and a good number have lost their monitoring entirely, usually when the mobile data link they relied on was switched off underneath them. An array with no monitoring degrades quietly, and the owner finds out from a bill rather than an alarm.

Two failure modes are more common on this ground than on a clean site. The first is soiling. Metals processing, foundry work and grinding put particulate into the air, and on a shallow-pitch roof downwind of that activity the deposit does not simply wash off in the next shower. Soiling on a Don Valley roof is a measurable loss rather than a rounding error, and a system never cleaned in a decade will be visibly under its model. The second is the inverter environment: older string inverters were frequently hung in warm, dusty plant rooms or on south-facing external walls, and the result is thermal derating and fan failure showing as a summer output ceiling, precisely when the array should be at its best.

There is a paperwork trap too. Many of these systems still hold a feed-in tariff accreditation, and altering the installed capacity or the metering arrangement can affect it, so the tariff position, the metering and the ownership are established before any physical change. Through Solar Tech Support we take on independent, brand-agnostic operation and maintenance and the takeover of older systems whoever installed them, starting from what the array is producing against what it should be.