How reserve capacity is procured, block by block

A capacity tender buys availability: the commitment to be ready to deliver, priced per MW per hour. It is a different market from the balancing energy the other views cover, cleared on a different timetable, and a unit can win capacity and never be activated.

The date control here reads a tender date rather than a delivery day. For this auction they are the same date, so carrying the parameter across from the other balancing views is correct rather than a coincidence worth guarding against.

A block

A four-hour window of the tender day, per reserve product and direction. Each block clears on its own, so a day carries several independent auctions rather than one.

The merit order

One block's offers, cheapest first. Choosing another block, using its table link or the block selector, redraws the curve and marks the selected row. The choice lives in the URL.

The award cut

The price the last awarded MW cleared at. Everything below it was taken; the cheapest rejected offer above it is the nearest thing to a counterfactual the publication supports.

Award share

How much of what was offered was actually taken. This is the figure the whole-day product would distort if it were pooled in - see below.

The summary reports mean offered MW per four-hour product and direction block, and the largest single block award. MW across different delivery blocks is not added into a daily capacity total. If either reserve feed is unavailable, the page names the partial tender and the statistics cover only returned blocks. Empty responses are unconfirmed publication coverage.

The selected curve's folded bid ledger lists every offer, award and price behind the chart. Its CSV includes product, direction, block and the time the bids were read.

A whole-day product competes for the same reserve across all 24 hours instead of one four-hour block. On the tenders measured it offered many times the requirement, at prices far above what the four-hour blocks cleared at, and was awarded nothing.

Pooling it into the merit order above would put never-competitive volume at the top of every block's curve and make the award-share figures meaningless. So it is reported on its own, with its own table, and only where the tender actually carried one. With no awards the table is folded and its summary says so; it opens by default when awards are present.

Great Britain's operator buys frequency response and reserve in a daily auction and publishes the result of every block: what cleared, at what price, and which units won it, by name. The view reads those published results directly each time the page loads and stores nothing, so any delivery day back to November 2023 can be opened.

The delivery day

An EFA day: it starts at 23:00 UK time the evening before and runs for a day, so it is 23 or 25 hours long when the clocks change. Block times on the page are UK local time.

Two grains

The response products (Dynamic Containment, Moderation and Regulation, each low and high frequency) clear in four-hour blocks. The reserve products (Quick, Balancing and Slow Reserve, each positive and negative) clear per half hour. The page never converts one into the other or adds MW across blocks into a daily total.

The clearing price

Published, not derived: the price per MW per hour that the block cleared at, in pounds sterling. It is never converted or compared with a Belgian price.

Not cleared

A block published with no cleared volume and no price. The page says not cleared; it does not show a price of zero, because a zero price with volume is a different fact.

Negative prices

A clearing price can be negative. It is shown as published, and the page offers no explanation for any particular one.

Awards by unit

For the selected block, who was awarded how much: by participant with a share of the block, and by unit with its technology. Names are exactly as the operator publishes them.

Which empty it is

A failed read, a day not published yet, a day the operator holds no table for, a product that did not exist yet, and a block that was published and cleared nothing are five different states, and the page says which one it is.

Product coverage differs. The response products start in November 2023, Quick Reserve in December 2024, Balancing Reserve in this dataset in October 2025 and Slow Reserve in April 2026; a product before its start is reported as not yet auctioned, not as a day on which nothing cleared.

Why no offer curve is drawn. Belgium's view ranks every offer cheapest first. Great Britain's published sell orders include parent, child and substitutable-child orders: alternatives for the same capacity, of which only one can win. Adding their quantities up would count the same MW more than once, so this view shows results and awards and does not pretend to show offered volume.

Nothing here is screened or thresholded. These are published results, not a model, and a unit's award is not evidence about its behaviour.

Nothing here is screened or thresholded. No finding is produced and no threshold is applied.

In Belgium's view nothing can be attributed to a participant, and that is measured rather than assumed: the bid identifier does not thread across tenders. Verified 0 of 25,701 bid ids recurring across 14 tender dates - so the identifier is per-tender bookkeeping and cannot be used to follow anyone between auctions. Great Britain's awards are the opposite case: they are published by named unit, and the page shows the names without drawing any conclusion from them.