An analysis of Future Energy Scenarios 2020
Published by National Grid plc

Mark Howitt
Chief Technical Officer and Co-Founder
Storelectric Limited

Introduction

This analysis has been almost entirely re-written as compared with previous years.

This FES builds on the vast improvements of the 2019 analysis, and is the best FES that National Grid have produced, incorporating many of the points on which we have provided feedback over recent years. They have adopted three scenarios consistent with the 2019 Net Zero legislation (which became law too late for incorporation into FES 2019) and, for comparison purposes, one scenario that envisages steady progression on some continuation of current policies. This excellent choice of scenarios enables us to evaluate the costs and benefits of each compliant scenario in comparison with Steady Progression, benefitting legislators, government and industry decision makers. Note that Net Zero applies to the entire economy.

The foci for Net Zero are:

• Hydrogen for 21-59% of energy needs, from methane reformation plus CCUS, heating ⅔ of homes;
• Carbon Capture, Use and Storage (CCUS) for industrial clusters;
• Continuing unabated gas combustion, its emissions balanced by BECCS (bio-energy with CCUS);
• Heating demand reduced by ¾, met by at least 8m domestic hybrid heat pumps, over 40% of which will have thermal storage;
• 75% reduction in vehicle energy demand
• Major changes in human behaviour, such as active management of heat and other energy.

In the electricity system, on which this report focuses, the scenarios require

• Over 40GW energy demand, to be met by over 132GW more renewable generation capacity –
  • Over 71% of demand is met by renewable generation;
• Up to 38GW Vehicle to Grid (V2G) electricity storage from 5.5 EVs, which needs to be reduced by a factor of 7;
• The electricity system alone drops below Net Zero by the mid-2030s.

Although the scenarios are much improved, in that they aim for Net Zero, and though there are many improvements in specific technology forecasts (e.g. overall demand, nuclear capacity), others are a combination of unaddressed issues (e.g. CCS, DSR, interconnectors) and new contributory technologies whose inclusion is good but whose treatment gives rise to new concerns (e.g. hydrogen, bio-energy).

Conclusions

The requirement for large-scale long-duration electricity storage has increased over previous years’ assessments to account for the Net Zero target, in which the power system is negative emissions by the 2030s. Total storage needed is 20-40GW as stated in FES 2020, a quantity that only large-scale storage can provide. There is no statement as to the duration of such storage, but FES 2019 stated (consistently with the National Infrastructure Plan and the Technology Innovation Needs Analysis) that most of it needs to be long-duration, defined as greater than 4 hours. This increases by a further 19-28GW by 2030 and 22-46GW by 2050 taking into account the following considerations.

Despite these excellent improvements, there are a number of points of concern, for example:

1. While FES tries hard to model the economy and energy system as a whole, Total process efficiency and costs are not fully accommodated, e.g.:

• CCS consideration does not allow for the additional capital costs, the inefficiencies (up to 30%) it imposes on the “host” plant (e.g. power station), or the risks of leaks in capturing, transporting and storing a colourless, odourless, asphyxiating gas that is heavier than air;
• Flexing renewable generation with electrolysis igores that 3-6 times as much expensive electrolysis equipment would be needed than if powered by baseload electricity;
• Flexing variable demand with hydrogen-fired power stations does not account for a 30-40% round trip efficiency (electrolysis to power station) and much more expensive round trip capital and operational costs than 70% efficient large-scale long-duration storage like adiabatic CAES.

2. There is little recognition of the effects on grid infrastructure and required investment of their chosen energy mixes; for example:

• Widespread roll-out of EVs would require prodigious reinforcement of distribution grids even if chargers are 100% smart;
• Reliance on interconnectors almost guarantees black-outs some future during times of system stress;
• Connecting large renewables (especially offshore wind) to grids directly rather than through large-scale long-duration storage –
  • doubles or triples grid reinforcement,
  • ditto whole-system reinforcement as balancing actions are remote from the farms,
  • creates a large requirement for investment in flywheels and synchronous condensers because inertia is not put into the system at the renewables’ connection points.

3. The continuing focus on statistical-average measures such as Loss of Load Expectation (and on TWh annual consumption rather than peak/trough GW usage) gives rise to inadequate treatment of resilience for predictable events, e.g.

• They only address “a 1-in-20 peak winter day” with natural gas (and consequent emissions) and don’t consider cold spells such as the kalte Dunkelflaute (cold dark doldrums) which extend such periods for up to a fortnight and simultaneously make heat pumps and batteries less efficient;
• Nearly all European countries’ energy transition plans rely on imports during largely concurrent times of system stress (low renewable generation and/or high demand, which can extend for days or even weeks), so we can’t rely on there being a surplus for us to import – yet the scenarios rely on 22-27GW interconnectors to keep the lights on;
• A focus on distributed systems relies on their self-sufficiency, whereas in reality they rely on the grid for back-up during times of system stress as it’s prohibitively expensive to provide such back-up in distributed manner, yet FES 2020 does not identify a need for the capabilities to provide such grid-based back-up;
• Renewable generation moving (due to weather patterns) from one part of the country to another requires sufficient capacity for the entire system’s needs to be carried in each part of the country, unless there is sufficient large-scale long-duration storage;
• The Lockdown proved the high costs (forecast >£1bn p.a. by the 2030s even in an 80% renewable grid) of providing real inertia even when there are no gas-fired power stations able to produce it – large-scale long-duration storage (of all types) can; Storelectric’s can 24/7.

4. CCUS (Carbon Capture, Use and Storage) doesn’t account for costs and inefficiencies imposed on the host system, e.g. power generation, or on the leakage risks in the capture transport and storage of an asphyxiating, odourless and colourless gas that is heavier than air. While it’s necessary for industry clusters that cannot decarbonise intrinsically, for the energy sector it’s too expensive, inefficient and dangerous in comparison with widespread roll-out of large-scale long-duration storage.

• Utilisation (the U in CCUS) is mostly delaying emissions (e.g. re-processing it into fuels and plastics, both of which are then used/disposed of) rather than capturing the emissions permanently.
• CCUS is less than 100% effective and its costs and inefficiencies increase exponentially towards 100%.

5. Treatment of transportation is fanciful and encourages counter-productive legislation, regulation and investments, the most egregious examples being:

• Most vehicles will have to be hydrogen / fuel cell as:
  • There are insufficient raw materials in the earth’s crust;
  • Large numbers don’t have access to personal charging points;
• Of the battery EVs, only a small fraction (maybe 10%) of their storage will be available for V2G, and the grid will have to pay for their battery degradation;
• Autonomous Vehicles will indeed decrease the number of vehicles in service, as per the scenarios, but in doing so they will actually increase mileage and hence energy consumption, not decreasing it, as they travel between rides; this will also increase the number of vehicles on the roads at any given time.

6. Consideration of grid-connected battery storage remains deficient and counter-productive, as:

• There are insufficient materials, as the point on EVs, above;
• Doubling the size and/or duration of batteries increases capital costs by 70-85%, whereas the proportion is less for other technologies;
• Their capital costs are grossly under-stated, focusing on the equipment and ignoring land, grid connection, building and ancillary costs;
• Their average life-time efficiency is less than other technologies as they use promoters’ information which tends to deliberately ignore:
  • Ancillary loads: cooling (~10%), inverters (~5%) and other, and
  • Aging: efficiencies are quoted on day 1 whereas cells require three times as much cooling by swap-out, and cell and inverter inefficiencies double or triple over their lives;
• They provide only synthetic inertia: while it helps grids recover from faults, it does not prevent faults like real inertia does – and ditto all the other related stability services that grids need.

7. The large amount of bio-energy and BECCS (Bio-Energy with CCS) is difficult due to lack of availability of feedstock, and the large environmental harm in growing, processing and distributing it. Much of it would also be unnecessary if there were more focus on cheaper and more efficient large-scale long-duration electricity storage.

8. There is no statement on duration of storage required. FES 2019 was clear that most storage requirements (other than E2V) were large-scale long-duration, defined as over 4 hours’ duration.

9. There is no statement on grid stability services, which all large-scale long-duration storage can provide (and Storelectric’s plants will be able to provide double, 24/7, whether charging, discharging or neither) even though:

• During the Lockdown, in a partial foretaste of a 2030s grid (renewables as a high proportion of demand), the National Grid paid £10-30M per day to ensure sufficient such services, and put out a forecast commensurate with spending over £1bn p.a. on them by the 2030s; and
• These services were delivered by turning up gas-fired power stations, most of which are likely to close by the 2030s – though gas+CCS and BECCS could also provide them if sufficient are built and operated.

Conclusion: these scenarios are a plan for periodic widespread blackouts.

Total Demand

Peak demand is more realistic than previous years’ FES, approaching 80GW for System Transformation and Leading the Way scenarios, and exceeding 95GW for Consumer Transformation. This looks low, due to the moderating effect of large amounts of “smart energy” systems (especially smart vehicle charging) and changes in consumer behaviour. We are not confident that such enormous smoothing is realistic, but will work with it.

However it should still be treated with some caution as there are also very substantial changes that decrease energy efficiency in the whole system, e.g.

• Hydrogen production is necessarily energy inefficient, whether by methane reformation or electrolysis;
• Hydrogen transportation through the gas grid is more energy intensive as it carries about one-third of the energy per m3 of the gas;
• Autonomous vehicles will lead to a significant increase in total mileage;
• Imports of biofuels increases energy demand – as does domestic production of feedstock, by displacing agriculture which in turn increases food imports.
• Increasing gadgetisation and “intelligent systems” all demand electricity.

Nevertheless, the remainder of this report will assume that National Grid is correct in its forecast electricity demand.

Supply Side

Generation Mix

In FES 2020, energy supply is from the following technologies (stacking baseload generation at the bottom, then dispatchable, then interconnectors and finally intermittent generation):

This looks like a serious risk: the country is depending on imports for actual demand, without even taking into account either supply margin (10-15% to be added to demand) or the lack of duration of most of the storage concerned. In the Net Zero scenarios, by 2030 this shortfall is 19-22GW; by 2050 it is 21.5-39.5GW. Even in the do-nothing Slow Progression scenario the shortfalls are 12.6GW by 2030 and 14.75GW by 2050. And in every scenario except Slow Progression, the shortfall is greater than the country’s capacity to import.
V2G is assumed to provide up to 38GW storage – which would be inadequate for the CT scenario and barely enough for the LW. But for reasons described fully elsewhere in this report, this should be reduced by a factor of 7 to 5.5GW.
Not only that, but this does not account for the duration of storage called upon. All battery storage in the country is 1 hour duration or less; a little is planned at two hours. But the evening peak is 3-5 hours, and energy is also needed overnight after windless winter evenings – not to mention the weather patterns (e.g. the kalte Dunkelflaute – see Security of Supply, below) which extend these periods to multi-day, and even to up to a fortnight.
And the shortfall is, in all cases including the do-nothing Slow Progression scenario, 19-28GW by 2030 and 22-46GW by 2050.
The above are National Grid’s unadjusted figures. However demand should be increased by the supply margin, a 10-15% additional allowance to account for the unpredictable, e.g. extra-high demand and/or plant failures such as on the cable from a wind farm. Ignoring the technologies that cannot be relied upon (intermittent and interconnectors) because there are predictable periods when they fail simultaneously (e.g. the kalte Dunkelflaute and shorter-duration similar periods, such as after sunset on a windless winter evening), then the distribution of demand and reliable supply is:

However, the true picture is much worse: these figures don’t just ignore the supply margin, but also assume that electricity generation is at nameplate capacity. The technologies should be de-rated according to these factors, which are applied elsewhere by National Grid in their planning and management of the system.

Applying these factors, the truer picture is:

De-rating factors for biomass, coal, gas, hydro, interconnectors, nuclear, storage, energy from waste, using T-1 de-ratings, section 1.3 (p6):

De-rating factors 2017 for wind, DUKES Chapter 5, paragraph 5.43: https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/729379/Ch5.pdf

De-rating factors 2017 for solar, DUKES Table 5.7 footnote (4): https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/731590/DUKES_5.7.xls

Fossil fuel: used the figures for CCGT power generation. Interconnectors are averaged, Other renewables: assumed to be equal to on-shore wind. Assumed that all technologies with CCS are 5% lower than those technologies without CCS.

Note: National Grid’s assumption is that Vehicle to Grid (V2G) is as reliable as batteries, whereas in reality vehicles are often not available.

Therefore all scenarios depend, for actual demand, not only on ALL interconnectors importing simultaneously at 100% de-rated capacity, but also on 10-30GW of intermittent generation during every single peak in demand without exception. And on storage providing energy for the entire duration of need – which the current focus on batteries and V2G does not. (Large-scale long-duration storage can, if built with sufficient duration, for which there would have to be sufficient duration-based revenue streams which don’t currently exist.)

In other words, all scenarios are planning for widespread and frequent black-outs.
Security of Supply
Security of supply means two things principally: keeping power reliably in the grid to meet variable demand, and being in charge of the source of that power. This year is the first year in which we will have insufficient generation to supply the country’s needs, relying on imports for actual demand, and also for the country’s entire supply margin during “times of system stress”

These “times of system stress” are periods of high demand and/or low renewable generation. They occur every single windless winter evening, and are extended (and occur in other seasons) by weather patterns that yield negligible generation. The largest and longest of these weather patterns is the “kalte Dunkelflaute” (cold dark doldrums) identified by the French and Germans as covering almost the entirety of Western Europe for a fortnight every couple of years; reducing the duration to a few days, and scale to a few countries, makes these weather patterns very common.

Note that as the weather gets colder, the efficiency of heat pumps diminishes, to zero for temperatures of 0 to -8oC depending on the heat pump. This in turn would increase demand for electricity and/or (in the ST and LW scenarios) hydrogen for heating. The increased demand for hydrogen would itself place more demand on electricity, if electrolysed or if electricity were used in the reformation process. Such increases in electricity demand do not appear to have been accommodated in the forecasts.

So the question is: can we rely on imports during such periods? Storelectric’s own study of our neighbouring countries’ energy transition plans says no:

The only exports will be available from Switzerland and Norway (who will primarily export to their neighbour and, in Norway’s case, Germany) and Iceland. The Norwegian interconnector is projected to cost over £5bn for 1GW, for which price Storelectric could build ~5GW storage with durations of 5-12 hours, which begs questions about interconnectors’ value for money. An Icelandic interconnector would cost much more, and Iceland only has ~1-2GW exportable energy.

FES relies (p56) on natural gas for resilience, which negates the energy transition.

System Stability

As power stations have closed, grids around the world (led by National Grid) have become increasingly aware of the grid stability services that they produce, such as phase-locked loops, fault currents, voltage and frequency control, RoCoF (Rate of Change of Frequency) reduction following faults, etc. Most of these relate to natural inertia.

As demonstrated by the UK blackouts on 9th August 2019, while synthetic inertia helps recovery from faults, it cannot prevent them; real inertia does. That is because synthetic inertia is basically an ultra-fast response time. But any response time is a delay, and any delay translates a fault into a spike on the mains; it is this spike that does the damage. Real inertia is “always-on”.

Solar and wind generation, interconnectors and batteries are all DC connected and therefore provide no real inertia. All large-scale long-duration storage (CAES, pumped hydro, even LAES though that is medium-scale) uses large rotating machines and therefore produces real inertia. Storelectric’s CAES is configured to produce twice the real inertia of a power station, and to do so 24/7 – whether charging, discharging or neither. Without such inertial plant, the grid would have to invest in flywheels, synchronous converters or similar.

System Costs

Existing Subsidies

There are many subsidies hidden in the electricity system. For example,

• Interconnectors rely on the double subsidy of cap-and-floor contracts and zero grid access charges;
• Interconnectors also provide a UK-financed subsidy to overseas generators owing to their lower grid access charges and carbon prices, and the fact that the difference between these and the UK versions are not charged on import;
• Nuclear power relies on a very highly priced cap-and-floor regime;
• The total cost of the balancing and ancillary services market has increased by ~£1bn between 2010 and 2017, which represents additional system costs for balancing intermittent renewables without sufficient large scale long duration storage;
• The £1bn+ Capacity Market seems to be a subsidy dressed up as a market, to keep fossil fuelled power stations in operation to balance intermittent renewables;
• A negative subsidy (i.e. unwarranted cost) is imposed on storage by triple charging (to import and to export, plus the cost embedded within the price of the purchased electricity) which is currently proposed to reduce to double charging, still an unwarranted commercial disadvantaging.

The balancing/ancillary markets and Capacity Market subsidies alone are already over £2bn p.a. and expected to double within 5 years and to keep on increasing.

The two compliant scenarios rely on 15.75GW nuclear power. This is proving to be one of the most expensive energy sources available. It also relies on 8.3GW CCS, analysed above. It also relies on 19.7GW interconnectors which are only viable with. In contrast, Storelectric’s CAES has a cheaper levelised cost of electricity than a gas-fired peaking plant and can therefore balance the entire system cost-effectively and (on a level playing field) without subsidy.

The electricity system can only deliver cost-effective energy to UK consumers if the playing field is levelled.

Affordability

While policy makers talk about energy cost, they mostly focus on its price. These have become divorced from each other, with cost (including both overt and covert subsidies) rising as fast as price (£/MWh wholesale) falls. Already more than half of most commercial bills is made up of non-price levies and costs; this should be under one-quarter, preferably <20%, to pay for transmission and distribution costs alone, and to penalise anti-social behaviour such as excessive consumption of fossil fuels.

Contract Length

This focus on lowest price today and in the near future is the driving force behind the current regulatory structure. As a result, another tilt in the regulatory playing field is the short term nature of all contracts. This favours investments that have a short pay-back time, and hence those that have a short operational life and/or small scale.

• The cheapest way to deliver a 2-year contract is to patch up a fully amortised plant for an additional 2 years’ life.
• Following this contract, it is repeated; only the plant is older, more polluting, more expensive to patch up and maintain, and less reliable.
• This repeats at ever increasing cost until the plant dies of old age.
• Then electricity needs to be imported or new plants built with subsides.
• The cheapest way to deliver a 15-year contract is with a new plant.
• The total cost over 15 years is less under a 15-year contract than under 7½ x 2-year contracts, and in the meantime sufficient capital investment has been put into new plant to keep the system young, without subsidy, with benefits in security of supply (both definitions), reliability and cost.

Again, the electricity system can only deliver cost-effective energy to UK consumers if the playing field is levelled.

Costly Responses

In response to these shortfalls, National Grid is taking increasingly costly measures such as creating the Capacity Market in which, according to a recent government consultation document, “Two CM auctions have now been held, for delivery in 2018/19 and 2019/20 respectively. Whilst, given the target levels that were set, the auctions procured relatively little new capacity…” for about £2bn.

Added to that, the Winter Outlook Report 2015 states that to cope with narrowing markets, National Grid “developed a set of new balancing services (NBS) to help us to manage the uncertainty and tightening margins over last winter. … Demand-Side Balancing Reserve (DSBR) and Supplemental Balancing Reserve (SBR)”. “The total costs incurred in the procurement and testing of the new balancing services was £31.2m.” This total is likely to increase in future years: “On 3 June 2015, we announced the procurement of the 2.56 GW of additional electricity reserve for the winter 2015/16”, compared with the 1.05GW purchased the previous winter.

In FES 2017, National Grid stated that there will be “a growth in balancing tools and technologies”, but admits that “What technologies will be utilised has yet to be established by the marketplace”. This must grow: “As intermittent and less flexible generation grows at transmission and distribution level, the ability to flex generation and demand is becoming increasingly important”. This supports Aurora’s analysis in 2016, whose figures have largely been borne out in practice:

Since then, further market mechanisms have also been created, adding to the costs of maintaining the system, such as Supplemental Balancing Reserve, Enhanced Frequency Response and Demand Side Top-Up. It appears that additional patches or sticking plasters are being added to a worn-out regulatory framework at ever-increasing rates, tackling the symptoms of the problem rather than its causes, the largest of which is the system-wide loss of inertial generation and load.
Worse, renewables have been allowed to bid (albeit with huge de-rating factors) for Capacity Market Contracts. Since the CM exists to ensure back-up for renewables, they can’t back themselves up, so any amount of de-rating below 100% is a logical non sequitur.

And finally the CM has been subverted: 85% of the last T-4 contracts were let as 1-year contracts, i.e. a second bite of the T-1 cherry, which destroys the market’s purpose in financing the construction of new plants.

COVID-19 Lockdown Developments

During the recent Lockdown, demand was suppressed during a sunny and windy period, to the extent that renewables generated a proportion of demand that had only been expected in the 2030s. National Grid spent £10-30m per day on measures to ensure sufficient stability in the system (which could have been delivered much more cost-effectively by 2-4GW of Storelectric’s plants), and put out a forecast that by the 2030s such measures would cost over £1bn p.a. It is notable that these measures involved turning on/up gas-fired power stations, most of which will have closed by the 2030s; one wonders where the grid would source such stability services if not from large-scale long-duration storage. For more details, see the appendix The Lockdown: A Partial Test of the 2030s Grid.