Before 16:52:33 BST
Normal system operation
The Great Britain power system operates close to 50 Hz while ESO secures frequency response for a 1,000 MW infeed loss.
- Frequency
- 50.00 Hz
- Infeed loss
- 0 MW
- LFDD demand
- 0 MW
Drawing No. EH–CA–005 // Engineering Case Study
An interactive engineering reconstruction of the 9 August 2019 Great Britain (UK) power outage, showing how a correctly cleared lightning fault coincided with unexpected losses at Hornsea One, Little Barford and distribution-connected generation. The resulting 1,878 MW cumulative infeed loss drove frequency to 48.8 Hz, triggered Low Frequency Demand Disconnection and required coordinated frequency recovery and demand restoration.
Select a stage or run the event.
A static summary of the same nine stages used by the interactive regional replay.
The Great Britain power system operates close to 50 Hz while ESO secures frequency response for a 1,000 MW infeed loss.
A phase-to-earth fault occurs on the Eaton Socon–Wymondley 400 kV circuit. Transmission protection clears both ends within 74 milliseconds.
Hornsea One de-loads by 737 MW, the Little Barford steam turbine trips by 244 MW and approximately 150 MW of embedded generation trips on vector-shift protection.
A further estimated 350 MW of embedded generation trips on RoCoF protection. Cumulative infeed loss reaches 1,481 MW before response arrests the first decline at 49.1 Hz.
Little Barford GT1A disconnects 210 MW while all available frequency response is being delivered.
Stage-one Low Frequency Demand Disconnection automatically removes 931 MW of demand.
Little Barford GT1B disconnects a further 187 MW, taking cumulative infeed loss to 1,878 MW.
More than 1,000 MW of response and a further 1,240 MW of ESO control-room actions stabilise the system.
ESO progressively authorises DNOs to reconnect demand. All DNOs confirm restoration complete by 17:37.
Source-based milestones with educational interpolation.
The complete event in five engineering steps.
Official milestones with explicitly labelled interpolation.
| Stage | Cumulative infeed loss (MW) | Demand disconnected (MW) |
|---|---|---|
| 1. Before 16:52:33 BST | 0 | 0 |
| 2. 16:52:33.490–.564 BST | 150 | 0 |
| 3. 16:52:33–34 BST | 1,131 | 0 |
| 4. 16:52:34–16:53:18 BST | 1,481 | 0 |
| 5. 16:53:31 BST | 1,691 | 0 |
| 6. 16:53:49.398 BST | 1,691 | 931 |
| 7. 16:53:58 BST | 1,878 | 931 |
| 8. 16:54:20–16:57:15 BST | 1,878 | 931 |
| 9. 16:58–17:37 BST | 1,878 | 0 |
| Stage | Frequency (Hz) | Condition |
|---|---|---|
| 1. Before 16:52:33 BST | 50.00 | Near nominal |
| 2. 16:52:33.490–.564 BST | 49.97 | Near nominal |
| 3. 16:52:33–34 BST | 49.55 | Low |
| 4. 16:52:34–16:53:18 BST | 49.20 | Low |
| 5. 16:53:31 BST | 49.05 | Emergency |
| 6. 16:53:49.398 BST | 48.80 | LFDD threshold |
| 7. 16:53:58 BST | 48.95 | Emergency |
| 8. 16:54:20–16:57:15 BST | 50.00 | Near nominal |
| 9. 16:58–17:37 BST | 50.00 | Near nominal |
| Stage | Frequency response shown (MW) | Control-room actions (MW) | Demand restoration (%) |
|---|---|---|---|
| 1. Before 16:52:33 BST | 0 | 0 | 0 |
| 2. 16:52:33.490–.564 BST | 0 | 0 | 0 |
| 3. 16:52:33–34 BST | 0 | 0 | 0 |
| 4. 16:52:34–16:53:18 BST | 900 | 0 | 0 |
| 5. 16:53:31 BST | 1,000 | 0 | 0 |
| 6. 16:53:49.398 BST | 1,000 | 0 | 0 |
| 7. 16:53:58 BST | 1,000 | 0 | 0 |
| 8. 16:54:20–16:57:15 BST | 1,000 | 1,240 | 15 |
| 9. 16:58–17:37 BST | 1,000 | 1,240 | 100 |
| Stage | Normal | Stressed | Critical | Offline | Restoring |
|---|---|---|---|---|---|
| 1. Before 16:52:33 BST | 5 | 0 | 0 | 0 | 0 |
| 2. 16:52:33.490–.564 BST | 3 | 2 | 0 | 0 | 0 |
| 3. 16:52:33–34 BST | 0 | 3 | 2 | 0 | 0 |
| 4. 16:52:34–16:53:18 BST | 0 | 2 | 3 | 0 | 0 |
| 5. 16:53:31 BST | 0 | 0 | 5 | 0 | 0 |
| 6. 16:53:49.398 BST | 0 | 4 | 0 | 1 | 0 |
| 7. 16:53:58 BST | 0 | 4 | 1 | 0 | 0 |
| 8. 16:54:20–16:57:15 BST | 4 | 0 | 0 | 0 | 1 |
| 9. 16:58–17:37 BST | 5 | 0 | 0 | 0 | 0 |
The transmission fault was cleared correctly. The widespread outage developed because several unexpected generation losses occurred almost simultaneously with the voltage disturbance and were followed by additional trips.
Frequency response delivered at least 650 MW within about eleven seconds and 900 MW by 16:53:04. This arrested the first decline at 49.1 Hz and supported a recovery to 49.2 Hz.
Little Barford GT1A then disconnected 210 MW while all available response was already being delivered. The system therefore had insufficient remaining headroom to absorb the new loss without emergency demand disconnection.
Official reports provide national frequency, plant and DNO information, but not a continuous administrative-region animation. The map therefore uses five broad educational regions to show stress, LFDD and restoration without claiming an official geographic sequence.
Electrical supplies were restored quickly, but some rail equipment and trains required manual restart or operational recovery. The page therefore distinguishes power-system restoration from dependent-infrastructure recovery.
What the event demonstrates about frequency security and dependent infrastructure.
A concise view of prevention, protection and recovery barriers.
Open each topic to review the principles illustrated by the event.
When connected demand exceeds generation, rotating machines initially supply the deficit from stored kinetic energy and slow down. System frequency therefore falls until response increases generation or demand is reduced.
Generators are expected to remain connected through specified short voltage disturbances. The transmission fault was cleared quickly, but the coincident plant responses produced large infeed losses.
These loss-of-mains methods detect changes in frequency behaviour or voltage phase. In 2019 they disconnected substantial distribution-connected generation during a transmission-system event.
LFDD disconnects selected demand at defined low-frequency thresholds. Removing 931 MW of demand helped arrest the 48.8 Hz decline and preserve a synchronised system.
Reconnecting too much demand at once can create a new imbalance. ESO therefore authorised DNO restoration progressively after frequency and reserve levels were secure.
Four transferable lessons for modern low-carbon power systems.
Short answers to common questions about the 9 August 2019 event.
The lightning fault initiated the sequence, but transmission protection cleared it correctly. The widespread loss of supply followed the near-simultaneous loss of major and embedded generation.
The ESO technical report records 1,878 MW of cumulative infeed loss: 737 MW at Hornsea One, 641 MW across Little Barford units and an estimated 500 MW of embedded generation.
Frequency fell to 48.8 Hz, triggering stage-one Low Frequency Demand Disconnection.
Low Frequency Demand Disconnection automatically removes selected demand when frequency reaches emergency thresholds, helping rebalance generation and demand.
DNO relays disconnected 931 MW of demand, affecting approximately 1.1 million customers.
No. The system remained synchronised. LFDD and frequency response arrested the decline and avoided a much more severe system collapse.
Some trains and rail assets shut down or required manual intervention to restart. Operational recovery therefore continued after electricity supplies had been restored.
No. The event chronology and national values follow official reports. Stage colouring is an educational illustration applied to eleven geographic regions through five broader operating-status groups.
Compare this frequency event with other EngineerHub disturbance replays.
Primary reports and the boundaries of this educational reconstruction.
National Grid ESO: Technical Report on the events of 9 August 2019.
Ofgem: Investigation into the 9 August 2019 power outage.
UK Government / E3C: GB power system disruption on 9 August 2019: Final report.
Office of Rail and Road: Report following railway power disruption on 9 August 2019.
Map geometry: Great Britain regional boundaries are adapted from the Wikimedia Commons NUTS 1 statistical regions of the United Kingdom map by Dr Greg and Nilfanion. It contains Ordnance Survey data © Crown copyright and database right 2011 and is used under CC BY-SA 3.0. Muted neighbouring coastline context is derived from GSHHG, released under the GNU LGPL. Transmission corridors and stage colouring are educational.
The map uses geographic regional boundaries but remains a teaching backbone rather than a complete transmission or DNO topology. Event times, plant losses, frequency milestones, LFDD quantity and restoration milestones follow official reports. Intermediate frequency points, regional status and corridor loading are educational interpolation.
Include the inputs and assumptions used.