FAQ: Fundamentals model

• How have we determined the demand forecast? E.g. in terms of future weather scenarios, electrification.
• How do you deal with price cannibalization for storage (and interconnectors)?
• How do you model scarcity pricing?
• How do you deal with negative pricing and CfD payments?
• Can the model deal with curtailment?
• Is there a volume limit to the Balancing Mechanism? If so, how do you set that?
• How did we arrive at our carbon derating factors for gas in the CM?

How have we determined the demand forecast? E.g. in terms of future weather scenarios, electrification.

We take the annual peak and minimum demand from the FES, as well as overall TWh demand, and scale to a daily demand shape based on weekday/weekend and time of year using historic trends.

There is huge uncertainty on what shape future demand will take - the success and take-up of smart appliances, smart metering with flexible tariffs, I&C schemes for demand shifting, demand (and availability) for EV charging at peak, heat pumps, home batteries, climate change weather variations etc… Rather than conjecture what the relative take up of each of these elements which could change the demand shape in an unknown way, we use historic data to inform the future.

We put demand side response in the capacity stack at various prices which effectively shifts the demand shape according to price.

How do you deal with price cannibalization for storage (and interconnectors)?

Storage charges up when power is cheap, and discharges when it's expensive.

When you have a lot of storage on the system, it might discharge into the same periods, which changes the supply stack. With the addition of storage, there is suddenly a surplus of generation, and as a result, the price crashes. What used to be an expensive period is now a cheap one (and on the charging side, a more expensive one, though to a lesser extent, as charging is typically more spread out).

This means the opportunity for storage is lost - and the storage makes no (or much less) money.

In reality, high prices would get damped or suppressed as 'peaks' spread into neighboring periods.

In our storage model, we limit the amount of the storage fleet that can dispatch into an individual half-hour period. More information on this here. This acts to soften the price cannibalization.

How do you model scarcity pricing?

Demand side response (DSR) is priced at £250/MWh, £500/MWh, and £1500/MWh. It sits near the top of the generation stack, with loss of load pricing at £6,000/MWh above it.

When there is a shortage of generation capacity (usually driven by low wind), a high price results as demand and generation intersect in this high-price region.

How do you deal with negative pricing and CfD payments?

Depending on the year the CfD has been awarded, payments stop after a certain number of consecutive hours of negative pricing.

Our CfD fleet is split into the early auctions (AR1) and later auctions. The AR1 fleet is priced at -£100/MWh - effectively, it never turns off. The later auctions are priced at £0/MWh.

You might expect, after many hours of negative pricing (above the cap of the auction round), assets turn off as they'd have to pay to generate.

We don't consider the previous half-hour's price when dispatching the subsequent half-hour. This means this is not dealt with in our model. We may therefore have more negative pricing than you might expect - as if these assets were to turn off, there would be less generation on the system (and the price would return to £0 or above).

However, there is uncertainty around how you'd turn off GW of wind or solar with a few hours notice and no major operational cost.

Can the model deal with curtailment?

Some new 'non-firm' grid connections are being offered, which have some level of curtailment. This means the transmission network operator, or the distribution network operator, could reduce (or curtail) the amount of power imported or exported through the site at any one time. It can also be referred to as 'Active Network Management'. This could be to ease pressure on the local network.

To model the impact of curtailment on battery revenues, we would need to understand the level of curtailment the site is likely to face. So, the number of hours that restrictions could be placed, and what those restrictions look like - if all exports are set to 0, or all imports, or maybe exports limited to 50% over solar peak hours in the summer months, for example.

Our dispatch model can show what the impact on revenues will be for such curtailments, under a custom run. Please contact the team so we can understand what this could look like for your site.

There is a bit more info on active network management here.

Is there a volume limit to the Balancing Mechanism? If so, how do you set that?

The volumes in the BM are set by modeling the amount of energy and system balancing the GB system operator must do to keep the grid balanced. We use our wind and demand forecast, along with the capacity of the transmission network in different areas of GB, to predict these. See here.

This determines the 'depth' of the Balancing Mechanism.

Next, we determine how much flexible generation will be on the system: the plants that can flex their output in response to BM actions, and where these will be. We allow batteries, pumped hydro, CCGTs to turn up, and these plus wind to turn down. We then estimate how much of the energy and system actions in a region (dictated by constraint boundaries) that batteries will be to provide, given this competition.

We limit the dispatch of batteries by their duration, day-ahead position, and cycling considerations.

How did we arrive at our carbon derating factors for gas in the CM?

Currently (April 2024), as part of prequalification for the Capacity Market, there is a limit of 550g of Fossil Fuel origin per kWh of electricity generated. This applies to new build, existing, and refurbishing CMUs for the delivery year commencing in 2024, as well as any subsequent delivery years. The Department of Energy Security and Net Zero (DESNZ) have indicated in their latest REMA consultation that they intend to both reduce this limit and to undertake a broader Capacity Market reform. Yet, recent government press releases have stated unabated gas will remain on the system to ensure security of supply.

There are questions about the specifics and timings of emissions limits, as well as how to incentivise low carbon technologies with desirable characteristics. One option DESNZ are considering for the CM consists of 'multiples' which is similar to derating factors for different properties (like de-rating factors in reverse).

We will adapt our Capacity Market model when more information on future auction design is released, but for now we have accounted for carbon through derating factors, using the 550g/kWh limit as a starting point. We have modeled this using a CO2 emissions allowance which drops over time. Plants with emissions below this level (550g/kWh in 2024) have a carbon derating of 100% - i.e. it receives the full CM price for its technology type. Plants with emissions above 550g/kWh have a carbon derating which means they receive less than their full CM price. For instance, if emissions for a generator are halfway between the allowance and the maximum - given by double the allowance - then it will receive a carbon derating of 50%.

We have modeled this using various different rates of decrease for the carbon allowance, as well as carbon maxima that are different multiples of this allowance.

We settle on the factors shown here. If these changes occur too slowly then emissions will not be consistent with emissions targets and Net Zero goals. If these changes occur too quickly then there are concerns around security of supply, and we see the impacts of this through extremely high wholesale prices, extremely high capacity market prices to encourage greater levels of low carbon generation, as well as loss of load. We have therefore settled on yearly allowances and maxima that we believe would make sense from a policy point of view, due to their balance between incentivising low carbon generation whilst maintaining security of supply.

How do you know that the commercial model of storage, and other generator types, is viable?

We use NREL's data on CAPEX and OPEX of storage systems to calculate IRR on battery revenues within the forecast model, using the revenues we project. NREL put 2024 battery prices CAPEX at $720/kW for a 2-hour system, which drops ~$30/kW a year to 2030, when the annual decline in price slows. Using these figures, we find the 25-year IRR stable at around 10-12% for the forecast horizon. In our central scenario we have 50GW of storage by 2050, which is probably more aggressive than other providers in the market - but with the trend of falling wholesale prices and more and more intermittent generation coming online, we find the markets are deep enough to take this much short duration storage.

Looking at the sensitivity of modeled price to grid-scale battery storage, we find daily power price spreads fall <3% with each additional 10GW of storage in the model at 2040.

We also perform a similar IRR calculation for other generation types within the model (gas, nuclear, wind, biomass, solar, etc.) to ensure the commercial model of all the generator types makes sense, including any subsidy mechanism like a CfD.