Modelling electricity prices of the future

As part of my PhD research, I’ve been thinking a lot about modelling Great Britain’s electricity market over the longer term (10–40 years). In particular, I am interested in how sources like solar and wind that are intermittent but have zero marginal cost, can exist alongside on-demand generation and storage capacity. This post outlines why I believe it can work.

How Electricity Prices are Set

Firstly, I need to outline how the electricity prices are set. Firstly, I am going to assume a simpler system, similar to the gross pool system which Great Britain used to have.

Here, each generator provides offers of how much electricity they will provide at different prices. For example, suppose that various generating units offer to generate at various quantities at prices between 0 and £100/MWh. In this case, the system operator can asses the demand and determine which offers to accept, starting from the cheapest. The highest price required to meet the demand is paid for all offers accepted.

For example, as indicated by the red lines, if demand is 25,000MWh, all solar, wind, nuclear and 7,000MWh of gas will be accepted, and all will be paid £40/MWh. If demand rises to 35,000MWh, all solar, wind, nuclear, and 17,000MWh of gas will be accepted, and all will be paid £70/MWh. This means that solar and wind generation will be paid as long as the marginal unit of electricity is offered a price above £0/MWh.

A few notes:

• Generators may choose to offer supply at a negative price, to ensure it is accepted. Coal, biomass and gas (CCGT) take time to ramp up and down, so may want to ensure that their offers are accepted. In addition, some solar and wind generators may have contracts which incentivise them to generate, even at a loss.
• The system operator will not pay more than £6000/MWh for electricity. If this price is insufficient to attract sufficient supply, demand will be shed (I believe this has never happened in Great Britain).
• I mentioned above that Great Britain no longer operates a gross pool system. Instead it uses a three step process:
  - Generators nominate a quantity that they want to generate, and can lock in prices for this in the forward, day-ahead or intraday markets.
  - Generators can offer to increase or reduce supply to help the system operator balance supply and demand. This process determines an imbalance price.
  - Generators are only exposed to the imbalance price for any difference between the quantity generated and the quantity that has been locked in.
  It is unclear to me that this system changes the generation mix or expected generation revenue, so for the rest of this post I will assume a gross pool.

Generator revenue in today’s markets

In the example above, I showed how different generators might offer electricity into the market at different prices, and as demand varied the set of offers that were accepted would vary, along with the clearing price.

We can look at the last year of prices (July 2018 — June 2019), and see how many hours and at what average price generators would have achieved at different offer prices. I have used the day ahead hourly auction prices for this analysis, for which the minimum price over the year was £9.09/MWh and the maximum price was £276.51. At no half hour was all the electricity generated by solar and wind or nuclear, so it seems likely that these zero or low marginal cost sources were never the final offer accepted.

The average price of all hours was £54.45/MWh, so any generator operating the whole year would have received revenue of £476,976. In contrast, a generator that only chose to operate where the price exceeded £60/MWh would only operate for 2930 hours, but would have received an average price of £70.47.

Solar and wind farms are in a different situation, in that they aren’t operating every hour. I therefore took the day ahead wind and solar generation forecasts (taken from BMRS data), to calculate the average price weighted by generation. These showed that the average price weighted by solar was 51.75, by offshore wind was 53.00, and for onshore wind was 54.67 (for comparison, the flat average was 54.45). In other words, over this time period the average price achieved by solar and wind were not wildly different to the overall average.

Generators being built now are likely to be around for the next twenty to thirty years. My feeling is, if they knew that they’d be able to get prices like last year over that whole period, there would be no problem building more wind, solar and gas generation (making certain assumptions about gas and carbon prices). However, the reality is that there are likely to be significant changes ahead.

Likely Changes

Over the next ten years, I expect we we will a number of changes that will affect the electricity mix:

• Increased total electricity demand, as more vehicles and heating switches to electricity.
• Increased solar and wind capacity, to the extent that some of the time we will not need any gas, coal or biomass.
• Increased battery capacity, which will provide flexibility, especially within each day. This will make the fact that many power stations take time to ramp up and down, and the short-term unpredictability of solar and wind, less of a problem.
• Increased decentralised generation and demand side response which will make things more difficult in the short term, but I am confident that these will end up being helpful to ensure grid stability.

Effect of Changes on Electricity Prices

Once solar and wind provide 100% of electricity demand during some hours, we will get an increasing number of hours where the clearing price is close to zero, perhaps 20–30% of hours by 2030.

If this is not offset by extra high prices in the other hours, additional gas, nuclear and biomass plants will not be viable, and there will be shortages of power much of the time. I suppose the government could directly fund capacity, for example through capacity payments. However, my feeling is that allowing high prices is preferable: it avoids the need for the government to pick which technologies to support, and it incentivises flexible technology and demand side response (ie reducing demand when demand exceeds supply). I am not sure what the government would end up doing in such a situation, but for the rest of this post let’s assume they do allow prices to rise.

So we now have a price profile in which say 30% of hours have a price of close to zero, 60% of hours have a price of close to £70/MWh, and 10% of hours have a price of £120/MWh. Here, the average price would be £54/MWh, similar to what it was in 2018/19. A gas power station with a cost of £50 would have been used only slightly more, but would make gross profit of £166,400/MW/year, significantly more than the £71,500/MW/year it had in 2018/19. At these levels it would continue to be profitable even taking into account the need for carbon capture and storage.

Battery operators are likely to have made more money in this scenario than in 2018/19, being able to charge at close to 0 and sell at higher prices. However, this depends on the distribution of high and low prices. If the price was 0 for 30% of each day, and £120 for an hour or two of each day, batteries would prove extremely profitable. If the price was 0 for several months of the year, and then £120 for a whole month, batteries would prove significantly less valuable. The same can be said for any demand with significant flexibility, for example the production of hydrogen using electrolysis.

To the extent that solar and wind generation was uncorrelated with prices, these would achieve similar revenue as to 2018/19. However, the more likely scenario would be for prices to be lowest when wind and solar generation was highest, and the prices highest during periods of lower wind and solar generation. As a result, predicting the revenue of a solar or wind generator would depend a lot on the mix of solar, wind and batteries deployed.

This was the scenario where wind and solar provided all the electricity required 30% of the time. I can also imagine a scenario where wind and solar provided all the electricity required 80%. In this scenario, in order to make gas generators with a variable cost of £50/MWh viable (achieving the same £71k/MW/year as in 2018/19), we would need prices averaging over £90/MWh during that 20% of hours.

Finally, moving from a world where prices are around £30–70/MWh most hours, to one where 80% of the time they are £0/MWh and 20% of the time they are £90/MWh, is a lot riskier for generators. For example, if it turns out that they are only used 10% of the time, or the price when used turns out to be only £75/MWh, profit will be significantly reduced. Suitable mechanisms to mitigate or get comfortable with this risk will be necessary to attract the necessary investment.

In conclusion, I don’t see the economics of electricity markets as preventing wind, solar and flexible generation, as well as storage, in fact, I think they will actually enable them. I recognise that there will be considerable effort needed to manage grid stability and short-term imbalances. but I believe that can be done, and that there won’t be a need to significantly override our electricity markets.