A business interruption policy turns on one number: how long the asset is out. For energy assets that number is rarely published. The register now estimates it for every incident it holds, with the loss, and states the figure a source gives wherever one does.
By technologyHow long each kind of asset is out
The typical figure is the middle incident: half took longer, half less. Business interruption cover usually starts after a waiting period of 30, 60 or 90 days, so the share of incidents past each one is shown.
| Technology | Incidents | Typical time out | > 30 d | > 90 d | Typical loss |
|---|---|---|---|---|---|
| Battery storage | 133 | 7 weeks | ••%Subscribers only | ••%Subscribers only | £•••kSubscribers only |
| Solar | 222 | 12 days | ••%Subscribers only | ••%Subscribers only | £•••kSubscribers only |
| Onshore wind | 235 | 12 weeks | ••%Subscribers only | ••%Subscribers only | £•••kSubscribers only |
| Offshore wind | 28 | 9 weeks | ••%Subscribers only | ••%Subscribers only | £•••kSubscribers only |
| Grid infrastructure | 532 | under a day | ••%Subscribers only | ••%Subscribers only | £•••kSubscribers only |
| Hybrid sites | 40 | 7 weeks | ••%Subscribers only | ••%Subscribers only | £•••kSubscribers only |
Subscribers Blurred figures are in the subscriber edition. See the plans, with 14 days free.
Bars: typical days out of service. Technologies with fewer than three estimated incidents are left out.
What it costs
The typical loss, with the 75th and 90th percentiles that a limit is set against.
| Technology | Typical loss | 75th pct | 90th pct |
|---|---|---|---|
| Battery storage | £•••kSubscribers only | £•.•mSubscribers only | £••.•mSubscribers only |
| Solar | £•••kSubscribers only | £•.•mSubscribers only | £••.•mSubscribers only |
| Onshore wind | £•••kSubscribers only | £•.•mSubscribers only | £••.•mSubscribers only |
| Offshore wind | £•••kSubscribers only | £•.•mSubscribers only | £••.•mSubscribers only |
| Grid infrastructure | £•••kSubscribers only | £•.•mSubscribers only | £••.•mSubscribers only |
| Hybrid sites | £•••kSubscribers only | £•.•mSubscribers only | £••.•mSubscribers only |
Subscribers Blurred figures are in the subscriber edition. See the plans, with 14 days free.
CauseFires keep assets out longer than electrical faults
Thermal incidents, mostly fires, typically keep an asset out for 3 weeks; electrical faults for under a day.
| Cause | Typical out | > 30 d | Typical loss |
|---|---|---|---|
| Thermal / fire (492) | •• weeksSubscribers only | ••%Subscribers only | £•••kSubscribers only |
| Electrical (472) | •• weeksSubscribers only | ••%Subscribers only | £•••kSubscribers only |
| Mechanical (158) | •• weeksSubscribers only | ••%Subscribers only | £•••kSubscribers only |
| Control system (3) | •• weeksSubscribers only | ••%Subscribers only | £•••kSubscribers only |
| Gas release (4) | •• weeksSubscribers only | ••%Subscribers only | £•••kSubscribers only |
| Other (52) | •• weeksSubscribers only | ••%Subscribers only | £•••kSubscribers only |
The number of incidents is in brackets.
Subscribers Blurred figures are in the subscriber edition. See the plans, with 14 days free.
Subscriber editionWhat the full report adds
- Subscribers The loss in pounds by technology, cause, severity and stage of life, with the 75th and 90th percentiles, the figures a limit is set against.
- Subscribers The share of incidents out of service beyond 30, 60 and 90 days, the waiting periods business interruption cover uses.
- Subscribers The largest estimated losses of the last twelve months, each with its range, time out and basis.
- Subscribers The estimate on every entry: the range, the days out, the confidence and how it was worked out.
See the plans. The Analyst plan includes this report and the estimate on every entry, with 14 days free.
How far to trust itWhat the figures rest on
| Basis | Incidents | Share |
|---|---|---|
| A figure the source states | 42 | 4% |
| Estimate, high confidence | 25 | 2% |
| Estimate, medium confidence | 397 | 33% |
| Estimate, low confidence | 728 | 61% |
Each estimate is a range. The AI prices what was damaged item by item from the entry and its sources, and was tested against incidents whose real loss and downtime are known: the ranges are set wide enough to hold the real figure about three times in four. A formula checks every estimate, and where the two disagree by more than three times the confidence shown is low. How the estimates work.
What this report is not
It is a benchmark of reported incidents, not a loss ratio. The register does not know the fleet the incidents came from, so it cannot say how often an asset has an incident, only what happens when one is reported.
Most figures are modelled estimates, not settled claims. Each incident is counted at the middle of its range. The register is new and fills in its archive month by month, so the figures will move as it grows.