8th October 2024

The changes this quarter are mostly within the fundamentals model. Check out the changes to demand, wind buildout, and gas price - to name but a few.

1. Specific EV & heat pump models change demand shape, making it more price-sensitive

We forecast to have 30 million Electric Vehicles (EVs) on the road by 2050. We will also see many GW of heat pumps on the system, as Great Britain electrifies heat and transport.

We now isolate EVs and heat pumps and forecast their demand independently. We expect some EVs to be price sensitive (via smart charging, with a smaller capacity of Vehicle-to-Grid) and, therefore, compete with battery storage. Heat pumps will be less flexible but have a demand pattern that (in winter) peaks both in the morning and the evening as homes switch on their heating. By 2050, we expect large winter evening peaks driven by 8GW of demand from electric domestic heating.

We now capture the individual shapes of these within our demand forecast - and how they contribute to nationwide demand and, therefore, the effect on wholesale power prices.

The impact is that demand is generally flatter, but more seasonal. There is higher demand overnight driven by smart charging and lower demand at peak (compared to v3.1). This leads to higher wholesale prices overnight, which pushes daily spreads down. As heat pump demand picks up in the 2040s, winter evening peaks grow—and our demand peaks at over 110GW. This is 5GW less than in v3.1!

2. Updates to commodity prices

We update gas prices with the latest forward curves from CME for gas prices to 2029, and the latest 2024 Future Energy Scenarios for prices further along the curve. We also blend the forward curve with the long-term forecast of gas prices, which smooths a jump we previously had in the gas price in 2029.

Long term, gas prices are up 9% from the v3.1 release, and this has pushed up wholesale prices in the short term.

Carbon prices from UK ETS futures (ICE) are also updated; they have increased by 3% to 2027.

3. Updates to Capacity Market de-rating factors

Previous updates to Capacity Market de-rating factors (which take effect in 2026) and more recent changes to Capacity Market derating factors of Equivalent Firm Capacity for storage are also included. After 2028, NESO's new derating factor methodology for BESS will apply, boosting 1-hour systems from 8.4% in 2028 to 10.3% in 2029. We have revised our projections of these to 2050.

The impact is that in 2026, we see an increase in Capacity Market revenues for storage, as the de-rating factors increase before they fall again in 2027. In 2029, CM revenues jump again with the new methodology. After this, they again fall away in line with the expectation that energy-limited assets of a certain duration become less valuable to the system as loss of load events become longer.

Read more here.

4. Wind buildout adjusted for AR6 results and reduced long-term projection of offshore wind

Both onshore and offshore wind buildouts have been updated to reflect the latest AR6 results in the latest Contracts for Difference round. Our solar projections were already in line with the AR6 results so they are unchanged.

The pipeline of offshore wind sites has been updated with the latest commissioning dates of known projects, including staged commisionning for particularly large sites. Many of these dates have been delayed by a year or more compared to our previous projection. We assume the Scotwind projects will come online between 2035 and 2040, with a modest buildout of offshore wind after that.

As a result, our wind buildout has shrunk 28GW to 87GW by 2050 vs v3.1. This pushes up baseload power prices after 2040. It also pushes the buildout of CCS up, as we need more flexible thermal capacity to meet the shortfall.

We also now use a more conservative buildout of wind in our low scenario. This assumes that none of the floating wind projects are built.

5. IRR calculations of the battery energy storage fleet better reflect wholesale market cannibalization in buildout projections

The capacity expansion model for battery storage determines our buildout projection. Within this model, we must estimate battery revenues each year and how this changes as the fleet grows (or shrinks). This feeds into the Internal Rate of Return (IRR) calculation, which in turn determines how much storage can be built economically.

Further modeling of the cannibalization of daily price spreads (and therefore wholesale revenues) as the battery storage fleet grows, shows that as the capacity of installed batteries grows above 20GW, the spreads available start to shrink due to the actions of that fleet. We now apply this shrinkage depending on the fleet buildout - previously, we used an average across the time horizon.

So, the estimation of annual battery revenues now better accounts for the impact of this cannibalization. As a result, the buildout of storage has slowed. We forecast 54GW of utility-scale storage in 2050, which is a reduction of 6GW compared to v3.1.

6. Assume CCS plants get more economical

As the amount of abated gas (labeled as CCS) grows, we assume it new plants will require a higher hurdle rate. The first 5GW of CCS requires an IRR of 0% to be built, underpinned by government subsidy. For each GW of subsequent CCS, we require an IRR of +0.3% to make a build decision. By 2045, when there is xGW of CCS on the system, we have a hurdle rate of x%.

Now, we move on to some fairly minor changes in the dispatch model.

7. Batteries > 300MW follow ramp rate restrictions & ancillary service rules

Ramp rate restrictions apply to large generators on the GB grid: according to the grid code, they can ramp instantly to 300MW and then ramp at 50MW/minute. This is for physical notifications only.

Ancillary service contracts are also capped at 100MW for a single site.

Given the number of larger sites coming onto the system in the next few years, we can now account for these differences in custom runs for assets larger than 300MW.

Read more here.

8. Update to timelines of Open Balancing Platform updates impacting short-term BM revenues

The newly formed National Energy System Operator (NESO) recently gave an update to the Open Balancing Platform (OBP) timelines, which are detailed here.

"New Storage Parameters" going live within OBP has been pushed back from October 2024 to April 2025. This change will give control room engineers visibility of the energy capacity of batteries within the Balancing Mechanism, rather than assuming a 30-minute dispatch is possible. It is likely to increase dispatch volumes as the BM gets more efficient for storage.

As a result, we have pushed back the timelines for improving the efficiency of BM energy actions by 6 months within our BM model.

While we are starting to see some small growth in the number of system-flagged actions batteries are used for, particularly in North Scotland (more on this here), improvements to dispatch rates have been much slower for these constraint-driven actions than for energy actions. So, we have also pushed back the timelines for when we will see fully efficient system-flagged actions from 2027 to 2028.

In this release, most of our changes involve updates to the dispatch model, impacting BM revenues. We have only made limited changes to the fundamentals model and we have updated battery CAPEX.

1. Added Balancing Reserve

Batteries doing both merchant-focussed and ancillary-focussed strategies are now able to participate in Balancing Reserve.

This new service was launched in March 2024 and has seen participation from batteries, as well as gas peakers and CCGTs.

Revenues from Balancing Reserve are small, but it ensures flexibility is reserved for use in the Balancing Mechanism during the delivery periods.

The impact: we see higher BM revenues, as well as a new line of revenues from Balancing Reserve.

2. Updated BM dispatch rates with the latest we observe across operational battery sites

Significant progress has been made in the dispatch rates of operational battery storage in the Balancing Mechanism, with rates above 9% (on average) across April, May, and June 2024.

We have updated the locational dispatch rates at the start of the model in the latest release to reflect this.

The BM revenues to 2027 are now higher.

3. A revised indication of BM revenues in day-ahead strategies

Previously the model producing our ancillary-focused strategy had no foresight of revenues available in the Balancing Mechanism. We now give it (and the merchant-focused strategy) an indication of what BM revenues will be.

The result is higher BM revenues in the ancillary-focused strategy, as more flexibility is reserved for the BM. This also means that the ancillary-focused strategy now returns higher revenues than the merchant-focused strategy for shorter-duration systems.

4. Removed ramp rate restrictions to batteries doing frequency response

The ESO has signaled that it will remove existing ramp rate restrictions, which limit the change in power per minute to 5% of any contracted frequency response volume (in the other direction) for batteries.

The impact is that merchant revenues are now 12% higher to 2027, as batteries are not limited by how much they can charge up while performing frequency response.

5. Updated TNUoS revenues for both distributed and transmission sites

5a. Added wider generation TNUoS as standard to transmission connected BESS

We use the ESO forecast of wider generation tariffs published in April 2024 to forecast TNUoS costs for distribution connected sites greater than 100MW, or those connected directly to the transmission system.

These use projected values of Annual Load Factors, depending on the strategy of the site, which are important in determining the rates.

The impact is that transmission-connected or large sites now have a non-zero TNUoS revenue.

5b. Updated Embedded Export Tariffs to consider battery duration

Using historic data on how well battery sites of different durations hit triad periods, we apply a capture rate to Embedded Export Tariffs to get a more realistic estimate of the rates they actually receive for exporting during winter peaks.

We have also updated the forecast values of the tariffs given the latest 5-year forecast from the ESO.

The impact is that distribution-connected or large sites now have smaller TNUoS revenues.

6. Updated battery CAPEX & near-term buildout

In a change to our capacity expansion model, we have reduced the CAPEX of new-build battery sites by 30% compared to the last version. We have also reduced the amount we expect to be built each year. This is in line with recent market data.

We have also updated the build-out of BESS to 2027 in line with the latest Modo buildout report, which reduces the total capacity to 2027.

The impact is that we get a slightly higher BESS buildout by 2050, and more longer duration systems.

7. Updated commodity prices

We have updated gas prices in line with the latest curve from CME.

The impact is minor.

8. Fixed an issue in discontinuities in the battery fleet's state of charge within the fundamental model

The way the model is parallelized led to a jump in the state of charge of the storage fleet every 196 days. This had more of an impact towards the final years of the forecast, when the storage fleet gets larger - as it meant there was more 'free' energy coming onto the system at those points.

We have now removed these jumps and the impact is that the tail of average power prices sits closer to £25/MWh than £20/MWh.

9. Updated solar load factor year to align with wind

We now use 2018 for both the wind load factors and solar load factors. The two are correlated, so it's good to be consistent across both when using these load factors to forecast future years.

10. Updated degradation curve and repowering

We have updated our standard degradation curve to be more in line with cell manufacturers current estimates for new systems. In parallel, we have changed our run library re-powering default to 10,000 cycles (previously 8000 cycles).

As well as updating the commodity prices, we have made some key under-the-hood changes to our fundamentals model that we're excited to share.

1. Supply stacks - now with far more thermal variation

• We have split our half-hourly supply stacks up into smaller chunks for thermal plants, to better reflect varying short-run marginal costs due to efficiency curves and ramp rates as well as different efficiencies within the fleet
• Impact: more price variation within a day, better reflecting historic prices (checkout Backtest)

2. Fleet storage - better dealing with price cannibalization

• Previously we had occasions where the fleet storage charging was causing price increases in the power price
• Fleet storage now responds to de-rated margin, has a strike price that reflects the cycling cost of the system, and is limited by how much volume can be dispatched at once.
• Impact: lower price spreads over the forecast horizon as storage cannibalizes significant spreads

3. Capacity build out determined by economics, moving away from the NESO Future Energy Scenarios as the major input

• New build, retrofit, and retirement of unabated gas CCGT, OCGT, peakers; and CCGT CCS; and battery storage have a build-out of capacity which responds to the economics of these plants
• The capacity market model has a carbon element post-2030. Volume procured - and price - responds to the minimum margin on the system
• Impact: our capacity build-out has changed since v2.4:
  • Extended CCGT retirements: 1.7GW of high efficiency unabated CCGT online in 2050
  • Higher CCS CCGT capacity by 2050: 7.3GW vs 6GW in v2.4
  • Slightly fewer batteries by 2050: 48GW vs 50GW in v2.4.

4. Other minor changes

• Wider generation TNUoS
  • Impact: transmission system BESS now face (usually) revenue line item from TNUoS, depending on which generation zone they are in. Available via custom run.
• REGO prices are included when getting the SRMC for solar and wind
  • Impact: solar and wind SRMC is reduced by around £7/MWh in the shorter term.
• Updated commodity prices - gas & carbon & hydrogen
  • Impact: 12% reduction in price and 10% in spread in 2024 with impact reducing later
• Demand inputs aligned to 2023 FES Consumer transformation scenario
  • Impact: higher demand after 2040 leads slightly higher average power prices at the back of the curve
• Derate day ahead trading capacity to reserve more battery capacity for the BM after 2027
  • Impact: higher overall revenues post-2027 as the batteries capture more value from the Balancing Mechanism
• Get national bid and offer prices by varying demand by 8% of the average annual demand rather than 10% of half-hourly demand. This was necessary to avoid very high bid-offer spreads post-2045.
  • Impact: Minor compared to v2.4, average bid-offer spreads remain consistent.

We have delayed the operational start date of Hinkley Point C, Sizewell C, and other Pressurised Water Reactors

• On January 23rd EDF announced a delay to Hinkley Point C's expected start date, from 2027-2028 to 2029-2031.
• We expect other new nuclear reactors to now face similar delays. We have pushed back the start date of Sizewell C to 2033-2036 and delayed other new Pressurised Water reactors by two years.

Operational nuclear capacity falls between 2027 and 2037

The nuclear build-out from v2.3 to v2.4 is shown below.

The delay in Hinkley Point C has a minor impact on storage revenues

In the years affected by the delay of Hinckley Point C (2027-2029), price spreads fall on average. This is because the shortfall in generation can be met mostly by increased CCGT generation or changes in interconnector flows. Most of the time, it makes the minimum price in the day higher, but the maximum price in the day is only slightly higher, as the energy is replaced by a slightly more inefficient CCGT. More on this here.

The fall in price spread causes a maximum 3% reduction in battery energy storage revenues in 2028.

Delays to nuclear buildout in 2030s have a more significant impact

Into the 2030s, the delay in nuclear buildout leads to an increase in price spreads. This is because of the reduced size of the CCGT fleet over time, which causes an increased frequency of price spikes in the years most affected by the change.

This increase in price spread leads to a maximum 8% increase in battery revenues in 2036.

Quarterly Forecast Pro update in Jan 2024 improves ancillary, balancing and wholesale revenue forecasts

• Changes to the Modo Battery Revenue forecast reflect the release of the Enduring Auction Capability (EAC) and increasing frequency response market saturation. Frequency response revenues are now lower.
• Improvements to the forecast of Balancing Mechanism (BM) actions for batteries lead to bigger locational differences in BM revenues in the Modo Battery Forecast. Some areas are higher, and some are lower than previously.
• We've also updated commodity prices - gas & carbon to reflect the latest data. This has pushed the spreads in our Power Price Forecast down by around 20% in 2024.
• We have updated the build-out of thermal gas plants. This has made a negligible change to the Power Price Forecast.

More on this below.

📘

We've updated our methodology pages to reflect these changes - check them out!

Frequency response prices have dropped significantly with the Enduring Auction Capability release in November 2023

• Changing bidding strategies with the new platform and auction clearing algorithm have driven down prices.
• Our updated frequency response price forecast takes the new relationship that's emerged between frequency response clearing price and wholesale day-ahead price.
• We also now allow frequency response prices to be negative - as seen in the market.

As dispatch rates in the BM improve with the Open Balancing Platform, revenues depend on transmission constraints

• We use supply, demand, current dispatch rates & future transmission build-out of wires and wind, as well as modeled BESS day-ahead positions, to estimate Balancing Mechanism revenues.
• There are now significant differences in BM revenues from region to region (which are each DNO) in BM revenues, and they change over time as the electricity system develops.

Frequency response volumes of a single site are limited depending on market size

• With increasing battery buildout, more batteries will compete to provide frequency response services.
• In the dispatch model, we limit the volume that a single battery site can provide in each service, depending on how big (ie how much MW volume) each service is.

Intraday revenues for the 'Merchant + Ancillaries' grow

• As the battery fleet grows at a faster rate than ancillary markets, each site does less frequency response.
• They will have more availability for intraday trading as a result, leading to increasing intraday revenues over time.

You can read about these changes in more detail in our methodology:

• Balancing Mechanism
• Frequency response prices
• Dispatch model with frequency response saturation
• Intraday revenues

N.B. - January 25th 2024 update

After releasing v2.3 of the Forecast model on 19th January 2024, we made a hotfix after further reviewing the numbers & after getting more information from the ESO during a visit to their control room. This impacts the Balancing Mechanism revenues in the first 3 years. We have now better incorporated the precise rollout dates along with our estimated impact of the steps within the Open Balancing Programme. This has brought BM revenues down in the first few years (and closer to reality). We refreshed the v2.3 forecasts in the BESS Revenue library on the 25th January. This change affected forecasts whereby the start date was 2024 and only the first 3 years were affected.

24th October 2022

At v2.2, we have made several additions to the forecast, both in the data available and in the dispatch model:

• Scenario Databook: For each Macro Scenario, Forecast Pro users can access a Scenario Databook, using a link contained within the run library download. This contains additional information about modelling assumptions (commodity prices, demand, generation) and results (wholesale and ancillary prices, carbon intensity, hours at VoL).
• Merchant only revenues: Each data download contains a second results table showing revenues for batteries following a 'Merchant only' strategy i.e. no Frequency Response. As battery buildout grows, more and more batteries the proportion of batteries providing Frequency Response will decrease. More batteries will follow a Merchant only strategy.
• Capture rates: We have applied capture rates to our dispatch model revenues to reflect imperfect price forecasting, battery unavailability and lack of perfect acceptance in Frequency Response markets. These capture rates are based on a historical analysis of Modo's battery Leaderboard data and asset performance. We model capture rate improving with time, with improving forecasting and optimization tools. We have added a page to our documentation that discusses this in more detail.

2nd October 2023

We've made a number of improvements to the model which make revenue figures for post-2035 lower in v2.1

We have improved on the FES build-out capacities to ensure less loss of load, particularly post-2035. This has brought the post-2035 battery revenue numbers down.

These changes are in:

1. How we calculate the short-run marginal cost of assets in the generation stack.

• We have variation in the short-run marginal cost within a single generator type to reflect differences in cost (driven by efficiency, for example). Eg., the lowest efficiency of the ‘low-efficiency CCGT’ has a higher cost than the most efficient of the ‘low-efficiency CCGT’. This mimics plant-level behaviour in the generation stack, without needing to model each individual plant on the system.
• A linear fit between each point of the supply stack provides this variation. We have changed how we calculate this fit, by taking the fit from the 'start' of each generator type - rather than the 'end'. The result is, that the cheapest plant within a generator type is now slightly higher and the jump between wind and gas now better reflects reality.
• The impact is that power prices are higher as we have boosted each short-run marginal cost up a type.

2. The future load factors of wind now have a higher minimum, a lower maximum, and a lower standard deviation.

• With more wind farms being built in wider geographic locations around the UK, the turbines getting bigger, and further out to sea, ‘capture rates’ or load factors will increase. While we had this increase in wind load factor in previous versions of the forecast, we still experienced occasional very low periods of wind, as well as periods at 100% load factor.
• We changed the wind forecast so there will be fewer periods with very low wind and fewer periods with 100% load factor, while maintaining the (increasing) average annual load factors. This better reflects the anticipated behaviour.
• The impact is we get fewer periods of very high prices, particularly in the early 2040’s, when very low wind periods were driving scarcity. And, the periods of high prices that we do have are shorter in duration - a few hours instead of 6 hours or more. This makes battery revenues lower in this period.

• BM prices are set by demand elasticity around wholesale prices. In times of scarcity, offer prices are particularly high. With fewer periods of very low wind resulting in fewer periods of scarcity, less money is made by offering up availability in the Balancing Mechanism in v2.1 of the model. Thus, in the merchant-only strategy, the revenues from BM trading are now less, and more in line with BM revenues in the ancillary+merchant strategy.
  • This has the effect of bringing merchant revenues and merchant+ancillary revenues closer together, particularly post 2035.

3. Impact of AR5 has reduced offshore wind buildout between 2028 and 2035

• We have delayed the offshore wind buildout to reflect the results of the recent AR5 auction, in which no offshore wind bid in for any Contracts for Difference.
• The impact is less wind generation between 2028 and 2033, with the pipeline finally catching up in 2035. This means slightly higher power prices during these years, with slightly increased battery revenues (but impact is minor).

4. Increased levels of Demand Side Response in the model

• With increasing levels of smart charging of electric vehicles and more domestic consumers participating in flexibility turn down services particularly, we have revised up the amount of demand side response in the capacity stack across the forecast horizon to 14GW by 2035 and 25GW in 2050. We have also decreased the prices of it from £3000/MWh, £1000/MWh and £500/MWh to £1500/MWh, £500/MWh and £250/MWh (for high, mid and low price DSR).
• Demand side response acts as a competitor to battery storage, and sits near the top of the generation stack. This has meant that there are fewer periods with loss of load and subsequent very high prices - and battery revenues are lower compared to v2.0, especially post-2035.

26th September 2023

There are some big changes to the Modo Forecast

We've just done a major release, moving from v1 to v2.

This includes sweeping changes to the way you interact with the forecast, and expanding on what scenarios and site configurations are available.

• We have added the forecast to the Modo platform, rather than being in an excel file.
• We have added a host of new site configurations, including co-location with solar sites.
• As well as the possibility to run a 'Custom Run' if none of the 600+ runs are quite right.

And, there are changes to the forecast numbers themselves:

• We updated our input numbers in line with FES 2023's release as v1 was using FES 2022.

You can now find a library of runs under 'Forecasting' on the platform

You can apply filters to quickly and easily find the site configuration and setup you'd like, and hit download.

There are over 600 runs to choose from

We now run the forecast in the cloud. What that means is we're able to run a huge number of different site setups, with different fundamentals models - many more than before!

Choose between 1, 2 and 4 hour duration batteries; a maximum of 1 or 2 cycles per day, the 14 different Distribution Network regions in GB, a distribution or transmission connection, degraded or not (and 3 different re-powering options with our standard degradation profile) and from our 8 different fundamental scenarios:

• Central
• Low (delayed CCGT retirement, reduced BM dispatch, increased demand response uptake, low gas and carbon prices)
• High (increased BM dispatch, reduced demand response uptake, high gas and carbon prices)

All files show merchant, ancillary + merchant revenues, split out by month, from 2023 - 2050, with the details of the scenario and site configuration at the top of the csv file.

And co-location is now one of them!

There are currently two runs in the run library of co-located revenues.

More information about the model used to estimate revnues from a site with solar and storage is in our methodology.

Custom runs are now possible

In case you can't find the site setup you're after in our standard run library, you can create a custom run.

This opens up a handy type form to tell us exactly what you'd like, and we'll take it from there.

Any questions with how to use a Custom Run? Reach out to our sales team - [email protected]

We've also updated our inputs to align with the FES 2023 release

ESO launched their 2023 Future Energy Scenarios in July 2023 so we've taken their updates to capacity build out, gas and carbon pricing, and demand from the 'Consumer Transformation' scenario and incorporated them into our model numbers.

• There is less biomass and gas CHP, and slightly less wind in the FES 2023 numbers, which has increased scarcity on the system in the early years of the forecast
• In later years, we have more hydrogen peakers and low-priced DSR
• This has the impact of higher scarcity, and therefore higher battery revenues to 2035. After that, revenues drop relative to v1.0 as more flexibility from hydrogen peakers and DSR in later years drives down battery revenues.

For example, this table shows the percent change in generator type going from FES '22 to FES '23, in 2031:

v1.0: 30/06/2023

v1: Internal revision

v1.2: 03/07/2023

v1.3: 18/07/2023

Building on our initial excel spreadsheet, in versions 1.0 - 1.3:

1. We added ancillary service revenues to the dispatch model, so we now have both

• Merchant only revenues
• And, merchant + ancillary revenues.

2. Fixed a bug in the run hours and generation production figures in the fundamental model outputs (previously these were showing too high due to the way subtypes were being aggregated up)
3. Added some alternative fundamantal low scenarios to the databook:

• Double the DSR, half the price
• Extend CCGT retirement
• Increase gas and carbon pricing by 20%
• Lower the BM dispatch rates