Vitreous Labs · Great Britain · NESO connection register
No permitted battery capacity remains in any zone
NESO's transmission and Scottish embedded registers, 19 August 2026, refreshed twice weekly. 426.7 GW of storage capacity is contracted across 1,563 projects, but permitted battery capacity remaining for future gated application windows is zero in all nineteen zones and 99.4% of the 83.2 GW gate 2 book went to projects that were already protected. So this tool does not ask where to build. It asks which positions will fail, when, and at which substation the capacity returns. Size is contracted power (MW).
Developers
Acquire
Points
Constraints
Zones
Live
Watchlist 0Clear
Layers
Gate
Consent stage
Configuration
Applicants, by contracted storage MW
Show all applicants
Ownership, from Companies House
Applicants are grouped into platforms by shared registered address, which is strong but circumstantial evidence of common control. It is graded: an address is only used to group where fewer than five hundred companies are registered at that postcode. Fifteen postcodes in this set host more than a thousand companies each and one hosts over seventy thousand, so a shared address there means nothing at all and those applicants are left ungrouped.
The three clocks
Every position runs three clocks at once: its stage in the register, its consent, and the contracted connection date NESO has agreed. Consent coverage is the share of an applicant's contracted storage that has planning approved, which is what separates a project pipeline from an option book. Overdue is capacity whose contracted date has already passed. Click a row for the detail.
Where the queue is stacking up
Marker size
Ranked, and why each sits where it does
How the ranking is built. Three components, each with a stated maximum, then a deliverability gate. Contested, up to 40, is all storage capacity seeking the point against the largest stack in the country, and answers whether the market wants to be there. Returning on evidence, up to 30, is gate 2 capacity whose contracted date has already passed. Returning on risk, up to 20, is gate 2 capacity scoring high or severe on delivery risk but not yet late. Persistence, up to 10, asks whether more than one applicant is present, so churn is not a single counterparty event. The four are then multiplied by a gate: full weight where storage is already energised at the point, four fifths where gate 2 is present and consented, half where gate 2 is present with nothing consented, a quarter where there is no gate 2 position to take at all.
The deliberate difference from Lithuania. There the returning component scored against a published expiry date, because Litgrid publishes one. NESO publishes no milestone calendar, so the component splits into evidence, a date that has already passed, and risk, an inferred consent gap. Evidence is weighted higher for that reason. Delivery risk is consent gap out of 50, days overdue out of 30 and capacity scale out of 20; the lead times behind the consent gap, 1,095 days from scoping, 730 from awaiting consents, 547 from consented and 180 from under construction, are inference and not published.
Delivery risk across the firm book
The deepest connection point stacks
Contracted storage capacity assigned to each substation the register names. Thermal ratings are not published in Great Britain any more than they are in Lithuania, so this is what is contracted at a point and not what the point can carry. Click a row to zoom.
Standalone against co-located
The register names the full plant composition on every row, so standalone and hybrid separate without inference. What it does not separate is the capacity split at a hybrid site: the megawatts are the whole site, not the battery.
Planning, from the renewable energy planning database
Shown as an independent layer rather than joined to the connection register. The two registers name projects differently and a strict match reaches only 10% of register rows, so a join would be mostly guesswork. Turn on the planning layer to see where applications sit against the queue.
What you may do with each layer
Share alike runs by feature type, not by product. Everything in the first group is published under licences that permit commercial reuse with attribution alone. The second group is derived from OpenStreetMap and carries the Open Database Licence with it.
The OpenStreetMap derived layers are offered as a separate database under the Open Database Licence, which is the honest reading of a systematic extraction of a feature type. Nothing in the analysis depends on that split: the register deduplication, the attrition score, the delivery risk model, the ownership grouping, the duration profile and the balancing mechanism cross reference are all built from NESO, Elexon, the capacity market, the planning database and Companies House, none of which carry share alike.
Where every number comes from
Duration, from the capacity market
Duration is rising
Weighted mean hours by delivery year, weighted by connection capacity. The capacity market de-rating classes state duration explicitly, which is the field neither Litgrid nor NESO publishes anywhere in the connection register.
The firm book against transmission charges
Generation TNUoS tariff for the zone each connection point sits in, 2031 forecast. North Scotland is £97.75/kW and Central London is minus £16.51/kW, a spread of £114 a kilowatt. Storage has its own charging arrangements, so read this as the locational signal rather than a bill.
120 GW is queued at connection nodes that do not exist
NESO publishes no position for a strategic connection node, because there is nothing yet to position. Each family is drawn as the region its name refers to, at a hand set centre with a radius that reflects the extent of that region. That is inference about the name and not a statement about where anything will be built, and the layer can be switched off.
Points with no published coordinate
Method and caveats
Coordinates come from two sources. NESO's own grid supply point and node lookup, matched on the connection site name the register publishes, and OpenStreetMap's substation layer, matched on the same name after normalising voltage and the word substation away. A name is only accepted where the distinctive words in it identify exactly one mapped substation and every candidate for that name sits within six kilometres, so Botley West does not become Botley Wood. That takes the located points from 334 to 508 of 937 and the located projects from 629 to 902 of 1,563. What remains unlocated is listed above. Two gaps sit behind it. The reform created strategic connection nodes, Trent Valley South, Sussex and Romney, South Anglia and East Kent among them, which do not physically exist yet; where the node names an existing parent substation, Leeds North for instance, it is drawn there and flagged. Separately, several hundred distribution level points are named in a form that no published gazetteer carries.
What the system operator pays to manage flows
Thermal constraint cost, daily settlement values summed by constraint group. This is the measured version of the constraint component the Lithuanian screen had to infer, because Litgrid publishes no cost of congestion and NESO does.
Why this is not painted onto the boundaries
NESO publishes constraint costs by constraint group, and boundary geometry by boundary name, and no machine readable mapping between the two. The network diagrams that would resolve it are PDFs whose labels do not extract. Joining them by eye would invent a precision that is not published, so the groups and the boundaries are shown side by side and the relationship is stated instead: SSE-SP and SCOTEX are both Scottish and are the two largest cost groups in every year on record, and 7,654 MW of the 10,229 MW gate 2 storage book sits in Scotland.
The firm book against that
The network geometry that is missing
The dashed lines are NESO's transmission constraint boundaries, which is the network representation the system operator itself uses to manage congestion. The quiet mesh beneath is the grid supply point regions, 362 of them, the transmission to distribution interfaces. What is not drawn is the circuits themselves. SP Energy Networks publishes 3,211,536 line assets including SP Transmission circuits, National Grid Electricity Transmission publishes its own, and UK Power Networks, Northern Powergrid and Electricity North West all publish 132 kV geometry. Every one of those endpoints returns a 403 to this environment while serving its metadata normally, which reads as a network level block rather than a permissions problem. A browser on an ordinary connection will very likely retrieve what this build could not.
Every zone is allocated to its ceiling, and two of them are not
NESO rations connections against the Clean Power 2030 targets, calculating a permitted capacity per technology per zone. Queue formation allocated all of it. The remaining column is NESO's own published figure for what is available in future gated application windows, and for battery it reads zero in all nineteen zones. For comparison, low carbon dispatchable power keeps 7,821 MW, onshore wind 5,101 MW, offshore wind 1,137 MW, solar 417 MW and long duration storage 162 MW.
The rule that makes two zones different
The technical principles state that capacity moved during rebalancing does not endure. A zone that received capacity must fall back below its original ceiling before anyone new can join. A zone that donated capacity has not reached its original ceiling, and new projects can apply there. For battery there were exactly two shifts and no substitutions: T3 donated 457.6 MW to T2, and D5 donated 127.2 MW to D6. The published remaining table reports zero for T3 and D5 as well, so either it is stated against the adjusted ceiling, in which case 585 MW is contestable now, or against the original ceiling, in which case there is no open door. The two readings differ by the whole answer to whether a new project can apply today. This is the one item in this tool unresolved on the face of the published data and it is a question for NESO.
How the reform closed the door
Zone areas are read from NESO's own map on page six of the detailed results and are approximate; the authoritative shapefiles have not been found. Individual project results are withheld as commercially sensitive, so a named project cannot be tied to its zonal outcome from the databook alone. Remaining capacity of zero is a statement about permitted capacity under the current methodology, not about physical network capability, and Ofgem and the department opened a review of entry requirements in April 2026, so the figure has a policy half life.
The balancing mechanism paid batteries £228m to discharge in a year
A battery in Scotland charges six times as often as it discharges
Where the system calls on storage, by transmission charging zone
Full power hours a year in each direction, and offer revenue per kilowatt, for the units that can be placed. This is a measured answer to where the system is short of flexible power and where it is long of energy it cannot move, which no queue analysis can give.
The operators actually trading batteries hold almost no gate 2
Matching is on exact company name after removing the legal suffix, so it is conservative: a lead party trading through a different entity from the one that applied for the connection will not match and is not counted. Read the figure as a floor.
Battery units in the balancing mechanism
What has moved since the last capture
Field dictionary
Every export and what each column means. Derived fields are marked, because they are this tool's judgement and not a published figure.
No warranty. Every source is published data supplied as is by its publisher, and NESO in particular accepts no liability for errors or omissions. Scores, risk bands, platform groupings and zone areas are inference and are labelled as such. Nothing here is a load flow study, a valuation or advice.
Version 8 · registers read 19 August 2026, balancing mechanism read 20 August 2026 · six tabs, draggable panel, three clocks dossier, delivery risk model, constraint cost series, watchlist carried in the address bar, field dictionary, print stylesheet · CSV export follows the current filter on every tab · fully self-contained file, no external requests.