Who appears in supply, who appears in accepted actions, and how those recorded actions are distributed.
September 2026, with separately defined source clocks. The archive selection was fixed on 2 October; later source revisions are not substituted.
Accepted-action figures come from Balancing Mechanism acceptance records (BOALF) only. Balancing done outside the Balancing Mechanism, such as frequency response and other contracted services, is not visible in them.
Identity scope of the accepted-action figures: they use only BOALF source rows with a non-empty string in Elexon's bmUnit field (838,282 of 838,285 rows). 3 rows have no usable value in that field; they are retained in the source record, excluded from these bmUnit-grouped figures and denominators, and not counted as zero. Any nationalGridBmUnit value is retained as source context and not substituted.
Share of the signed recorded generation sum, including net interconnector and pumped-storage values.
GB settlement dates: 1-30 September.
UTC interval: 31 August 23:00 to 30 September 23:00.
Share of observed event keys by frozen BM-unit category. One event key is counted once. Supplier BM units (2__) not otherwise classified are 2__ identifiers in the 2026-06-12 register archive that the frozen rule placed in no fuel, storage or other category. Other archived 2__ units appear under those categories, and 2__ units absent from the archive appear as Unclassified. Unclassified covers archived units of several identifier prefixes that the frozen rule placed in no category, and units absent from the archive.
Earliest observed segment: September UTC.
Selected event records may extend across the month boundary.
Supply is not an action-count denominator. These distributions cannot be subtracted to measure who contributes more. Bar lengths are scaled separately within each distribution; printed percentages name the values.
The frozen high-confidence storage classification contains 69 identifiers; 56 have selected action records. This is not a census of all GB batteries or proof of physical identity throughout September.
Storage selection uses LFR classification v0.1 on the BM Unit register archived 2026-06-12 12:16:53 UTC. Its high tier requires battery-related naming and approximately symmetric declared import/export capacities, excluding pumped storage and interconnectors. This historical record selection does not confirm physical identity or a current battery census. Units absent from that archive are not in this classification and appear as Unclassified, together with archived units the frozen rule left without a category.
Group event keys / all observed event keys
27.3%91,334 / 334,992
An event key is BMU + acceptance number. Its earliest segment observed in the selected files starts in the September UTC window. This is not a certificate of the event's global first segment.
Group start coordinates / all observed start coordinates
19.5%31,006 / 159,154
Distinct BMU + settlement date + starting settlement period, from the same selected events. Segments are not expanded into every period they cross; this is not duration, occupancy or service participation.
Group endpoint-area MWh / all selected endpoint-area MWh
2.3%112,716.5 / 4,995,888.5
Sum of (|start MW| + |end MW|) / 2 multiplied by segment hours, including retained cross-month segments. A trajectory approximation, not metered delivery or strictly within-month energy. Per-BMU areas are rounded before the share is formed.
Three separate denominators. No addition, subtraction, contribution ranking or combined score follows from placing these percentages together. Action count is not elapsed participation, and endpoint area is not physical delivery.
Each point uses its own day's event-key denominator. Date follows the earliest observed segment in UTC. A monthly share is a ratio of monthly counts, not the average of these percentages.
| UTC date | Group events | All events | Share |
|---|---|---|---|
| Sep 01 | 2,774 | 10,096 | 27.5% |
| Sep 02 | 3,053 | 9,442 | 32.3% |
| Sep 03 | 2,464 | 9,453 | 26.1% |
| Sep 04 | 4,584 | 17,484 | 26.2% |
| Sep 05 | 2,889 | 10,482 | 27.6% |
| Sep 06 | 3,096 | 11,675 | 26.5% |
| Sep 07 | 2,103 | 9,923 | 21.2% |
| Sep 08 | 2,789 | 13,119 | 21.3% |
| Sep 09 | 3,347 | 11,811 | 28.3% |
| Sep 10 | 3,063 | 11,972 | 25.6% |
| Sep 11 | 4,313 | 14,502 | 29.7% |
| Sep 12 | 3,570 | 13,543 | 26.4% |
| Sep 13 | 2,494 | 6,676 | 37.4% |
| Sep 14 | 3,225 | 11,590 | 27.8% |
| Sep 15 | 2,903 | 10,912 | 26.6% |
| Sep 16 | 2,581 | 9,705 | 26.6% |
| Sep 17 | 4,436 | 14,997 | 29.6% |
| Sep 18 | 4,074 | 14,887 | 27.4% |
| Sep 19 | 4,832 | 16,477 | 29.3% |
| Sep 20 | 4,640 | 15,070 | 30.8% |
| Sep 21 | 2,656 | 8,520 | 31.2% |
| Sep 22 | 1,611 | 5,222 | 30.9% |
| Sep 23 | 3,004 | 9,593 | 31.3% |
| Sep 24 | 2,392 | 8,628 | 27.7% |
| Sep 25 | 1,920 | 7,636 | 25.1% |
| Sep 26 | 2,713 | 10,683 | 25.4% |
| Sep 27 | 2,690 | 10,582 | 25.4% |
| Sep 28 | 2,540 | 8,396 | 30.3% |
| Sep 29 | 2,332 | 11,236 | 20.8% |
| Sep 30 | 2,246 | 10,680 | 21.0% |
The other two measures use settlement coordinates or segment duration. They are not overlaid on this UTC-date line. No dispatch cause, persistent trend or asset condition is inferred.
BM-unit identifiers with selected events
Concentration-equivalent identifier count
The second value is 1 / sum(squared event-count shares). It describes the concentration of this count distribution; it is not a count of identified units, independent resources or available replacements.
838,285 BOALF rows / 28,800 FUELHH rows were reopened from the fixed source set.
Both source sets contain a file for every September date. FUELHH contains every half-hour coordinate of those dates for every fuel type encountered. This does not certify omitted upstream categories, embedded generation or whole-event completeness.
55 retained BOALF segments extend beyond the selected UTC window. The area measure belongs to the selected action-record set; it is not a strictly clipped September total. Separate clocks are retained rather than silently aligned.
Endpoint area is calculated from the accepted MW levels at each segment's endpoints and its duration. Signed endpoints and per-unit rounding differences are retained in the internal audit record. No settlement metering, BMS or SCADA delivery is asserted.
Asset health, wear, cyber exposure, independent reserve paths, dispatch motives and system resilience. Frequency, metering, contracts and disclosures are not silently used as explanations for these charts.
The observatory's instrument changelog registers a planned outage of NESO's Wider Access API, the instruction route used by some Balancing Mechanism units, on 18 September. Units on that route take telephone instructions while it is out, so accepted-action records for that window may be thinner without any physical change. The effect is not measured here.