Forward scenarios¶
The system model rebuilds GB prices from the plants behind them. v4.0 adds two things that let it look forward, and v4.2 makes the links behave as they have and adds unit commitment to them:
- Interconnector links (
system.interconnectors: links): GB's ten interconnectors trade with France, the Netherlands, Belgium, Norway, Denmark and Ireland at those markets' actual day-ahead prices, instead of holding imports at their historic output. - NESO Future Energy Scenarios (
system.fes): a FES pathway's capacities and demand for a year from 2025 to 2035, applied to a historic weather year.
uv sync --extra system
uv run openenergy system run examples/gb-2035-holistic.yaml
name: gb-2035-holistic
system:
data: ../data/system
year: 2025 # the weather year: weather, demand shape, fuel and carbon prices
interconnectors: links # or {mode: links, hurdle_gbp_per_mwh: 5, availability: 1.0}
commitment: true # start-up costs and minimum stable output (marginal price)
fes:
pathway: Holistic Transition # also Electric Engagement, Hydrogen Evolution, Falling Behind,
# Ten Year Forecast (FES 2025) and Ten Year Outlook (2026)
year: 2035 # 2025 to 2035, not before the weather year
battery_efficiency: 0.85 # round trip; FES publishes none (stated assumption)
ldes_efficiency: 0.75 # round trip for new long-duration storage (stated assumption)
battery: {power_mw: 1, energy_mwh: 2}
Results are "this fleet and demand in 2025's weather and markets", not price forecasts: FES publishes no fuel or carbon price paths, so the weather year keeps its own.
Interconnector links¶
Each built link in NESO's Interconnector Register is a price taker on GB's single bus: it imports up to its import capacity whenever GB's price is above the neighbour's actual day-ahead price (Ember, converted at ECB rates) plus a hurdle, and exports when GB's price is below the neighbour's minus the hurdle. A link joins the system on the day it entered service, read from Elexon's flows (Nemo Link 2019, IFA2 and NS Link 2021, ElecLink 2022, Viking Link at the end of 2023, Greenlink 2025). Historic imports are dropped. Irish prices start in October 2018, so links need a year from 2019. Links work with the merit order, PyPSA (when storage is added) and unit commitment.
| Setting | Default | Meaning |
|---|---|---|
mode |
fixed |
fixed: historic imports (as before v4.0); links: price-taking links |
hurdle_gbp_per_mwh |
0 | added to import prices and taken off export prices |
availability |
flows |
flows: each link's import and export share fitted from its actual flows (below); a number from 0 to 1 applies to every link |
prices |
prices/europe_day_ahead_ember.csv next to data |
neighbours' hourly prices |
Checked against history¶
examples/interconnector_check.py solves each year four ways and compares with GB's
actual day-ahead price and the actual flows (Elexon FUELHH): historic imports (fixed);
links at full capacity (full, the v4.0 and v4.1 default); links at their fitted
availability (links, the default from v4.2); and that with unit commitment (UC, marginal
price). Prices in £/MWh, flows in TWh:
| Year | Mean: actual / fixed / full / links / UC | Spread: actual / fixed / full / links / UC |
|---|---|---|
| 2019 | 42.9 / 42.3 / 42.4 / 42.7 / 42.2 | 31.8 / 4.2 / 8.5 / 6.1 / 25.9 |
| 2020 | 35.3 / 37.1 / 36.9 / 37.5 / 36.6 | 39.6 / 3.4 / 9.4 / 5.0 / 27.5 |
| 2021 | 117.8 / 83.5 / 86.7 / 85.3 / 85.0 | 146.4 / 7.1 / 25.2 / 14.0 / 40.1 |
| 2022 | 204.8 / 157.7 / 163.1 / 160.0 / 159.9 | 155.6 / 9.5 / 42.5 / 26.7 / 61.3 |
| 2023 | 94.1 / 121.1 / 110.9 / 117.2 / 116.8 | 70.0 / 7.4 / 45.5 / 30.1 / 61.8 |
| 2024 | 72.5 / 89.2 / 81.2 / 87.0 / 86.4 | 59.9 / 5.7 / 43.2 / 26.5 / 56.6 |
| 2025 | 80.7 / 87.5 / 82.4 / 86.6 / 86.3 | 67.5 / 6.2 / 41.1 / 25.2 / 52.0 |
| Year | Imports: actual / full / links / UC | Exports: actual / full / links / UC |
|---|---|---|
| 2019 | 25.0 / 34.6 / 21.5 / 21.7 | 3.7 / 9.9 / 4.5 / 4.2 |
| 2020 | 22.8 / 34.9 / 21.4 / 21.1 | 4.5 / 10.5 / 4.4 / 4.3 |
| 2021 | 29.2 / 29.9 / 19.0 / 18.7 | 4.4 / 28.1 / 12.5 / 12.4 |
| 2022 | 16.2 / 26.1 / 16.7 / 16.4 | 20.4 / 47.2 / 20.2 / 20.3 |
| 2023 | 33.5 / 64.7 / 42.8 / 43.2 | 10.1 / 11.1 / 4.7 / 4.2 |
| 2024 | 43.9 / 69.9 / 46.4 / 46.4 | 10.6 / 18.8 / 8.0 / 7.8 |
| 2025 | 44.0 / 69.0 / 45.4 / 45.4 | 14.8 / 23.4 / 11.2 / 11.1 |
At full capacity links trade far more than happened: imports up to 93% too high (2023) and exports up to six times too high (2021). A price taker sees every hour's price gap as profit; real flows are limited by outages, losses, ramping, capacity sold ahead, trading on later markets and the neighbours' own supply. Each link's availability (below) brings imports to within 7% of actual in 2020, 2022, 2024 and 2025; they are 14% low in 2019, 35% low in 2021 and 28% high in 2023. Exports are within 3% in 2020 and 2022, 22% high in 2019, almost three times too high in 2021, and a quarter to a half too low in 2023–2025.
Matching flows takes spread away: links at full capacity were importing the neighbours' daily shape, which partly hid the model's own flat price (2025: 41.1 at full capacity, 25.2 at fitted availability). Unit commitment restores the shape on GB's own plant: the spread is 69% to 94% of actual in 2019–2020 and 2023–2025 (27% and 39% in the 2021–2022 gas crisis, when quarterly fuel prices miss the market's swings). The mean is within £2 of actual in 2019–2020, above it by £6–23 in 2023–2025 and well below it in 2021–2022 (see the system model).
Link availability¶
scripts/fit_interconnector_availability.py sets each link's import and export
availability to the energy it actually moved over what a price taker at full capacity
would have moved on actual GB and neighbour prices, 2019–2025, capped at 1 (see
Data). Fitted only on flows, never on prices. Fitted on 2019–2022 instead,
it gives 36.9 TWh of imports in 2024 against 43.9 actual (66.0 at full capacity), and
11.8 TWh of exports against 10.6. It is one number per link and direction, so it does not
respond to the size of the price gap.
Marginal or start price¶
With links and commitment, the marginal price is closer to history than the start price in 2023, 2024 and 2025 on the mean, the daily spread, the hour-to-hour change and battery value; the start price holds the night up at a running unit's cost plus its start cost. The scenarios below use the marginal price. This is a choice between two un-fitted definitions made on history. A 1 MW / 2 MWh battery (95% each way, £5/MWh wear), GBP per MW-year, last week's prices / perfect foresight:
| Year | Actual | Links | Links + UC, marginal | Links + UC, start |
|---|---|---|---|---|
| 2023 | 36,075 / 45,748 | 1,765 / 11,838 | 8,660 / 28,350 | 3,974 / 20,397 |
| 2024 | 29,148 / 39,430 | 1,648 / 10,694 | 9,538 / 26,798 | 5,108 / 17,411 |
| 2025 | 36,148 / 44,927 | 949 / 10,413 | 11,623 / 26,483 | 4,864 / 16,955 |
With perfect foresight the model now gives 59% to 68% of the actual value; planning on last week's prices, 24% to 33%.
A hurdle (hurdle_gbp_per_mwh) is still available but left at 0.
FES pathways¶
system.fes scales the weather year's system by the ratio of the FES year to the weather
year in the same pathway, so every half-hour keeps its weather and demand shape:
| Weather-year input | Scaled by FES (target year / weather year) |
|---|---|
| Wind (transmission and embedded) | Offshore Wind + Onshore Wind capacity |
| Solar | Solar PV capacity |
| Nuclear, hydro | Nuclear; Hydro capacity |
| Biomass | Biomass + CCS Biomass capacity |
| CCGT and peaking capacity (DUKES) | Gas + CCS Gas + Hydrogen capacity |
| Coal | none (set to 0) |
| Demand | ED1 annual system demand |
| Interconnector capacity | Interconnectors capacity |
| Batteries | added in full: FES GW, hours = GWh / GW |
| Long-duration storage | added: FES increase over the weather year (GW, GWh) |
Pumped storage keeps its historic output. Hydrogen and CCS gas plant are costed as gas,
because there is no open cost data for them. The weather year has to be a year the FES
edition covers: from 2024 for FES 2025, from 2025 for the Ten Year Outlook. Any scale or
capacity_mw you set overrides the FES value.
Results on 2025's weather¶
scripts/build_scenario_prices.py runs every pathway in 2030 and 2035 with default
settings (prices in £/MWh, energy in TWh, fes columns are NESO's own figures for the
same year):
| Pathway | Year | Mean | Spread | Imports | FES | Exports | FES | Curtailed | FES | Unserved GWh |
|---|---|---|---|---|---|---|---|---|---|---|
| Holistic Transition | 2030 | 86.3 | 72.0 | 36.8 | 35.7 | 24.6 | 38.9 | 8.7 | 12.4 | 68.0 |
| Holistic Transition | 2035 | 211.7 | 163.8 | 47.8 | 27.6 | 51.3 | 108.8 | 34.9 | 41.0 | 723.0 |
| Electric Engagement | 2030 | 74.3 | 58.0 | 36.3 | 37.0 | 28.4 | 40.1 | 11.6 | 12.2 | 9.7 |
| Electric Engagement | 2035 | 191.1 | 162.4 | 60.1 | 41.3 | 46.9 | 89.1 | 15.0 | 27.4 | 683.2 |
| Hydrogen Evolution | 2030 | 65.6 | 44.3 | 36.9 | 40.1 | 24.5 | 36.2 | 7.6 | 9.2 | 0 |
| Hydrogen Evolution | 2035 | 121.4 | 109.1 | 52.2 | 31.3 | 40.7 | 93.5 | 19.6 | 34.2 | 214.8 |
| Falling Behind | 2030 | 72.4 | 34.9 | 41.4 | 29.5 | 20.9 | 35.4 | 1.8 | 2.3 | 0 |
| Falling Behind | 2035 | 56.0 | 43.4 | 33.6 | 21.2 | 32.8 | 61.8 | 18.7 | 18.0 | 0 |
| Ten Year Forecast | 2030 | 66.3 | 38.6 | 37.2 | 22.9 | 24.2 | 46.8 | 6.2 | 7.4 | 0 |
| Ten Year Forecast | 2035 | 57.1 | 36.3 | 44.7 | 21.2 | 42.2 | 93.1 | 18.3 | 23.7 | 0 |
| Ten Year Outlook | 2030 | 73.8 | 32.0 | 43.2 | 42.1 | 19.6 | 34.0 | 1.6 | 7.4 | 0 |
| Ten Year Outlook | 2035 | 86.4 | 53.1 | 45.7 | 35.3 | 39.1 | 69.3 | 15.3 | 49.8 | 20.0 |
Settings since v4.2: links at each link's availability fitted from its flows, and unit commitment (marginal price). For comparison, the model's own 2025 with the same settings has a mean of £86.3/MWh and a spread of £52.0/MWh (actual £80.7 and £67.5).
Imports in 2030 are within 10% of FES's in four pathways and 40% and 62% above it in Falling Behind and the Ten Year Forecast; in 2035 they are 1.3 to 2.1 times FES's (1.5 to 3.5 times before links were given their fitted availability). Exports are 44% to 71% of FES's in every case: FES expects GB to export much more in windy hours than links that behave as they did in 2019–2025 do. Curtailment is 22% to 104% of FES's.
Six cases leave demand unserved. The price sits at the £6,000/MWh value of lost load in Holistic Transition 2035 for 174 hours on 11 days, Electric Engagement 2035 for 161 hours on 10 days, Hydrogen Evolution 2035 for 73 hours on 5 days, and on one or two days in Holistic Transition 2030, Electric Engagement 2030 and the Ten Year Outlook 2035. Without those hours the means are £56–94/MWh. The cause is link availability, not unit commitment. Holistic Transition, mean £/MWh and unserved GWh:
| Year | Full availability | Full + commitment | Fitted | Fitted + commitment |
|---|---|---|---|---|
| 2030 | 67.2, 0 | 67.6, 0 | 86.2, 68.0 | 86.3, 68.0 |
| 2035 | 77.9, 40.3 | 85.6, 50.5 | 170.8, 679.3 | 211.7, 723.0 |
The fitted availability holds on GB's tightest days in the record: on the 1% of days
with the highest GB price in 2019–2025, links imported 3% (ElecLink) to 42% (Nemo Link,
BritNed) of their capacity, below their fitted shares for every link but East West (39%
against 38%), because neighbours were often short at the same time. So derating links at stress is supported by the flows. What the model
leaves out is what FES counts on in those hours: demand response, new flexibility and
neighbours' own 2030s fleets. Read 2035 prices in these pathways as "this fleet is short
of firm capacity in 2025's weather when links behave as they have", not as a price
forecast. With every link at full capacity the same pathways shed load only in Holistic
Transition 2035 (50.5 GWh) and Electric Engagement 2035 (10.9 GWh)
(availability: 1.0).
A battery on pathway prices¶
The same 1 MW / 2 MWh battery (95% each way, £5/MWh wear) on 358 days of each modelled year, GBP per MW-year:
| Prices | Last week's prices | Perfect foresight |
|---|---|---|
| Actual 2025 | 36,148 | 44,927 |
| Model 2025, links + commitment | 11,623 | 26,483 |
| Holistic Transition 2030 / 2035 | 9,649 / 10,019 | 27,434 / 64,033 |
| Electric Engagement 2030 / 2035 | 4,853 / 14,261 | 22,640 / 70,217 |
| Hydrogen Evolution 2030 / 2035 | 6,318 / 12,138 | 18,393 / 47,247 |
| Falling Behind 2030 / 2035 | 3,892 / 4,311 | 12,106 / 16,368 |
| Ten Year Forecast 2030 / 2035 | 3,297 / 3,535 | 13,665 / 12,928 |
| Ten Year Outlook 2030 / 2035 | 2,528 / 4,259 | 9,967 / 19,929 |
Compare pathways with the model's own 2025, not with actual 2025: on modelled 2025 prices the battery's perfect-foresight value is 59% of its actual value, and the model's daily spread is 77% of the real one. Against that baseline, perfect-foresight value is 48% to 62% lower in Falling Behind 2030, the Ten Year Forecast and the Ten Year Outlook 2030, 15% to 38% lower in Electric Engagement and Hydrogen Evolution 2030, Falling Behind 2035 and the Ten Year Outlook 2035, and about the same in Holistic Transition 2030. The three 2035 pathways that rise far above it (Holistic Transition, Electric Engagement and Hydrogen Evolution, 1.8 to 2.7 times) get most of that from the lost-load days above: with perfect foresight over the full year, those days earn £49.4k of Holistic Transition's £63.1k, £54.2k of Electric Engagement's £69.2k and £30.7k of Hydrogen Evolution's £46.8k. Without lost-load days every pathway would be below the 2025 baseline: the six pathways with lost load earn £11.6k to £16.1k on their other days. A value that rests on a few days of lost load in one weather year is not a robust result. Planning on last week's prices keeps 16% to 35% of the perfect-foresight value, against 80% on actual 2025 prices and 44% on the model's own 2025.
Limits¶
- One GB bus: no constraint between Scotland and England, which is where much FES curtailment happens.
- Links take the neighbours' actual prices for the weather year, so neighbours do not respond to GB, and their 2030s fleets are not modelled. New links are not added as new neighbours; FES interconnector growth scales the existing ten.
- Capacities from the register are today's; earlier limits on Moyle, IFA and BritNed are not modelled separately, only through each link's fitted availability (see Data).
- Each link's availability is one number per direction for every hour and year; it does not grow with the price gap, and new FES capacity gets the existing links' shares.
- Ember gives one price for Norway and one for Denmark; GB's links land in southern Norway and western Denmark.
- The weather year's fuel and carbon prices are kept; storage efficiencies for FES batteries and long-duration storage are stated assumptions.
- With unit commitment (the published scenarios), storage is planned one day at a time seeing 24 hours ahead; without it, with perfect foresight over the year.
Data sources¶
FES tables: NESO, FES: Pathways to Net Zero 2025 and 2026 Ten Year Outlook, ES1 and ED1; interconnectors: NESO, Interconnector Register; both under the NESO Open Data Licence v1.0, supported by National Energy SO Open Data. Flows: Elexon BMRS FUELHH, under the BMRS data licence; contains BMRS data © Elexon Limited copyright and database right 2026. Neighbouring prices: Ember, European Wholesale Electricity Price Data, CC BY 4.0, converted with ECB reference rates (source: ECB). All accessed 13 October 2026; see Data for the rest of the system model's sources. OpenEnergy is not affiliated with or endorsed by NESO, Elexon, Ember or the ECB.