A walkthrough of the merit order dispatch and wholesale price simulation — live view · replay history
To decide what technologies get dispatched and at what price, we need to model the mechanics of the wholesale electricity market. To keep things relatively simple, we assume that all generation is sold in a single day-ahead auction, to one buyer. Reality is more complicated, with intra-day trading, balancing markets, and a variety of contract types and actors in the system, but our approach should capture the major price-setting dynamics.
Whoever is purchasing electricity wants to get it as cheaply as possible, so they buy electricity from the cheapest generators first. They continue to buy from more expensive generators until they have the required amount of electricity. One feature of the market that seems peculiar to some is that all generators get paid the price of the most expensive unit of electricity. This is called "pay-as-clear" pricing. In the example below, demand is 45GW, so the market clears at £80/MWh, and all generation gets paid that price.
Economic theory would suggest that generators should bid into the market at their short run marginal cost. In layman's terms, this is the cost of producing one additional unit of electricity (ignoring any fixed costs). For technologies like wind and solar, this cost is very close to zero, so they would typically bid in at or close to £0/MWh. For generators with a fuel cost, like gas or biomass, their short run marginal cost is heavily influenced by the price of their fuel, so this sets the price. For gas generators in our model, we pull the gas price from a market feed, and update the bid price on a daily basis accordingly.
However, the market in the UK is distorted by various subsidy schemes, which change this bidding behaviour. The most significant of these is the Renewables Obligation, which provides a subsidy for each unit of generation from eligible renewable sources.To receive this subsidy, which amounts to roughly £40-50/MWh for onshore wind and solar, the generator must sell their electricity. The generator can therefore bid in at a negative price (roughly up to the negative of the subsidy amount), which ensures they get dispatched, and still come out ahead. This was also true for the initial Contracts for Difference (CfD) scheme, but more recent CfD rounds have removed subsidy for hours where the market price is negative.
Bidding in at a negative price may seem odd, but even without the subsidy, it can make sense for some generators to do this. A nuclear plant might have very high costs for shutting down and restarting, so would rather pay to stay generating and expect to make up the loss when prices are higher. In reality, most of the negative pricing periods are driven by subsidised renewables. Within one technology type, there will therefore be a range of bid prices based on the subsidy scheme. Offshore wind generators with an RO receive 2 certificates, meaning their subsidy was worth roughly £100/MWh. For the first CfD round, the strike price was around £120, and there was no negative pricing rule, so these generators could bid in at around -£120/MWh. The strike price continued to fall, but it was only by round 4 that the strict negative pricing rule came in, removing any subsidy for hours where the market price is negative. Any wind generators from this round onwards (and any non-subsidised generators) would be expected to bid in at around £0/MWh, as they have no incentive to bid negative, but also no reason to bid above their short run marginal cost.
Note that the vast majority of the goal offshore wind capactity by 2030 should be from the later CfD rounds or merchant projects, so we would expect most of it to be bidding in at around £0/MWh. There may also be minor differences in bidding behaviour, even for generators with identical subsidy, so the real picture will be slightly messier.
To account for some of this complexity (whilst avoiding modelling every site individually), we divide each technology into its subsidy scheme, and then assign a representative bid price to each group. We divide each subsidy scheme into bands, and add some noise to the bid prices to reflect variation within the band. We get the load factor for each technology type from NESO and Elexon: for more details see the github readme.
We do this for all our technologies. We then need to sort the bands by price to get our new merit order, which creates some "overlap" between technologies, rather than all of one technology being cheaper than all of another. The resultant merit order still has noticeable steps, but has more variation in the bid prices.
It's important to remember that the clearing price is a function of both demand and supply. Two hours with the same demand can have very different prices, depending on what generation is available. The availability of renewables will vary with the weather, but gas has a fairly consistent availability, so will often set the price. It's also worth noting that the price of gas powered electricity is much more variable than the price of renewables, as the majority of the cost is the cost of fuel, which can change rapidly.
Interconnectors and storage are where things get slightly more complicated, as they can both import/export electricity depending on the price.
Interconnector flows will be governed by the difference in prices between markets. Power flows from low to high price zones. To model this, we use day-ahead prices from ENTSO-E for each neighbouring country. Remember that we're imaging that we have 2030's grid today: in reality, by 2030 our neighbours power systems will also have changed, so their wholesale prices will also be different. To simplify things, we assume that prices in our neigh Each link has a threshold representing transmission losses and friction — shorter AC links use £3/MWh and longer HVDC links use £5/MWh. Imports are handled by placing each interconnector's capacity directly into the merit order as a supply band at the foreign price plus the threshold: if the GB price would otherwise clear above that level, the link imports. Exports are computed analytically: if there is surplus generation below the foreign price minus the threshold, that surplus is exported up to the link's capacity, with the highest-value export markets taking priority.
The examples above shows a set of 5GW of interconnectors to one neighbouring country. In the left hand example, the 45GW of domestic demand clears at £3/MWh, well below the foreign price of £45/MWh. This means we won't need to import, and we also have enough generation to export to our neighbour. This pushes the final clearing price up to £5/MWh, still below our neighbour's price. The right hand side example shows a time where renewable generation is low: the interconnector sits in merit, and we therefore import to meet domestic demand. The price the interconnector will bid at is slightly higher than the foreign price, to account for losses and any other fees.
Storage is modelled in a similar way to interconnectors. We set a maximum charge price for each storage type — reflecting the price at which they expect to charge during low-price, high-renewable periods — and derive the discharge bid from that using round-trip efficiency: discharge bid = max charge price ÷ efficiency². A battery with 95% one-way efficiency (90.25% round trip) that charges at a maximum of £10/MWh will therefore bid to discharge at around £11/MWh. LDES, assumed to have a 70% one-way efficiency (49% round trip), bids to discharge at around £20/MWh for the same charge price. This means LDES is more selective about when it discharges, reflecting the greater energy cost of its lower round-trip efficiency. We've chosen the maximum charging price fairly arbitrarily: real storage operators will have a range of strategies and expectations for how price will evolve. This should be roughly acceptable for our purposes.
Modelling the price has a usefulness beyond the estimate of wholesale price itself. As you can see from above, the price is essential for understanding the flows of power between countries, and into and out from storage. Whilst our approach has some simplifcations, it can help us understand the dynamics of how the system might operate.
We do need to remember that wholesale price is only a portion of the price paid by consumers, particularly with subsidies such as the CfD which guarantee a fixed price to generators. This subsidy will be paid by consumer bills. A CfD dominated system could have average wholesale prices of around £0/MWh (compared to around £80/MWh in 2024), but with bills unchanged. The wholesale price should be thought of as a reflection of balance and supply in the system, and a signal for when to increase demand, rather than a direct signal of consumer costs.
A strict reliance on the wholesale price to control generator behaviour could also produce undesirable outcomes. For example, in hours where our neighbours have high wholesale prices, it could make economic sense for our gas generators to run solely for the purposes of export, even if there is low domestic demand. This would increase emissions and make the CP2030 goal harder to achieve.