Professional Documents
Culture Documents
Draft
In a world where supply and demand for energy are completely matched there is no need for
storage. However, as variability is introduced into both supply and into demand, a means of
balancing the potential mismatches of supply and demand is required. Such mismatches can be
caused by large, predictable changes in demand due to consumer behaviour and climatic effects, or
shorter, less predictable events such as sudden cold snaps. Also, they can occur due to supply
fluctuations resulting from the effect of weather on wind or solar farms, or, in the shorter term,
when generators trip.
This Chapter will focus mainly on the growing need to balance supply and demand created by the
intermittency of wind and solar power and the ways of addressing these imbalances. Short-term
storage (a few hours or a day) has been studied before, but the longer term storage needs are the
main focus and the net-zero emissions target will remove the main sources of flexible and
dispatchable generation from the energy system, i.e., flexible fossil fuel generation.
Storage of both heat and electricity are important and are interrelated because the choices about
how to decarbonise heating in the future will have a major effect on the size of the electricity
system. We begin by identifying the factors that affect the need for energy storage before discussing
them in detail, and estimating the expected size of the UK’s energy storage system.
To address the variability of the future energy system and the different characteristics or storage
technologies, energy may be stored in a number of different ways:
Our approach seeks to find a balance between variable renewables, other low-carbon energy
sources, and energy storage by considering what options are available. A completely renewable
energy system with storage matched to the likely energy demand in 2050 is studied before
considering alternative systems which are not purely renewable but that are compatible with the
net-zero carbon target. This provides an estimate of the storage needs required to balance
variations on a net-zero carbon system.
Electricity demand, or consumption, is normally met by direct supply from the generators. Any
shortfall in supply due to either the inflexibility or intermittency of generators is met from storage. In
times of surplus power, energy is either put into the stores, traded outside the UK through
interconnectors, or curtailed. Overall grid and distribution losses – much of which are in local low
voltage systems – are of the order of 10% of demand.
Electrical energy is expected to always be available when wanted, i.e., highly reliable. Therefore, the
need for storage is created by the inability of the system always to match supply and demand – due
to both the intermittency of wind and solar generation and the inherent variability of demand.
1
Draft
Energy storage could be aligned with the different elements of supply: wind and solar and with
demand, each smoothing out their elements of supply and demand. Whilst it would sound for
renewable sources or energy of deal with the problems of variability and intermittency at source,
this solution would miss the important opportunities both for off-setting fluctuation in one source of
energy against another and smoothing the variability by difference is supply based on location.
Energy storage is never completely efficient and can be much lower than 100%, perhaps as low as
25%. Losses can occur either in the process of converting and storing energy, during storage with
time-dependent losses, or in releasing or reconverting energy to meet demand. Therefore, the
round-trip efficiency of storage technologies that includes all three effects is important for storage
system studies.
Curtailment of
Primaryexcess wind and Grid
Direct Losses
Supplysolar energy
Supply
UK Energy
Generation Demand
Storage System
Losses
In a highly renewable energy system, this mismatch between demand and supply is the Residual
power - defined as:
Residual power = Renewable Supply - Demand
2
Draft
1
National Grid (2019)
2
Barrett & Sharp (2020)
3
Draft
pumps and matches the Climate Change Committee’s3 central assumption of 2050 demand –
600 TWh.
2. Fluctuation in demand during the year. There are recognised fluctuation of daily demand of 20-
30 GW depending on the season, due to both work patterns and day-night effects. Also, demand
is lower at the week-ends and during holidays. Seasonal effects also have a major effect on
demand, being highest in winter and lower in summer. Also, variations in mean temperature
between years affect the annual level of demand.
3. District heating stores could have a significant effect on the size of electricity demand. Also,
storing solar heat in the summer for use in winter would reduce the demand for electricity
to provide heating, although – as discussed in Chapter X – this is not expected to have a
major impact on the need for electricity storage.
4. The extent to which demand can be shifted in time to better match variable supply – see
Demand Side Response below.
On the supply side
1. The amounts of wind and solar power that are assumed to be available, their variation during
the year and between years, and – since their relative contributions vary during the year – the
wind/solar mix.
2. Interconnectors, which can trade surpluses and deficits with other countries, including wind
and solar from places where the weather may not be closely correlated with the UK’s.
3. Fluctuations in renewables leading to either shortfalls or over-supply and export or curtailment.
We will consider 100% renewables as base case; we are not advocating for this system, but wish to
establish first how provide a reliable supply from renewables before considering some of the zero-
carbon alternatives. Energy deficits can be met by:
1. Supply that is dispatchable (i.e., BECCS, hydro, nuclear, etc.). This role of matching supply and
demand is currently filled by gas generation, but gas with carbon capture and storage may be
a bridging technology. However, unless capture rates can be improved to 99%, to meet the
net- zero target, gas with carbon capture will not be viable for 2050.
2. Energy storage; this will include a range of technologies because of their different
performances, economic characteristics and efficiencies. The round-trip efficiency of electricity
storage varies from over 90% for modern battery systems, 40-70% for compressed air storage
and perhaps 25-50% if ammonia or hydrogen (as discussed in Chapter 3) are used as storage
media. Storage losses are lower when the output is heat rather than electricity.
3. Additional capacity to compensate for losses in storage or to reduce the effects of periods
of low supply, curtailment, or spillage of excess energy at other times.
Storage losses increase the cost of storage and also increase the amount of primary energy that will
be needed to meet demand in systems with highly variable supply. For these reasons, studies of
highly renewable energy systems that can be considered realistic must take account of storage
efficiency.
3
Climate Change Committee (2019)
4
Draft
We now turn to the effects of the weather and remark on expected future demand (to which we
return when discussing the UK’s storage needs), before considering residual demand in high
renewable scenarios and how it might be reduced, both by imports through interconnectors and by
including dispatchable generation.
In order to estimate future residual power and the need for storage, it is necessary to model the
effects of energy efficiency on future demand and the variation of demand through the year. Future
demand estimates used in the work, described below, take into account the expected improvements
in energy efficiency.
We will return to modelling of demand when considering the UK’s future need for storage, while
meanwhile quoting the results of studies in which future demand has been estimated in a variety of
ways.
The weather
The weather affects both wind and solar energy, but to different degrees and at different times of
the year. Also, the weather affects demand through the influence of temperature and solar gain,
major drivers of winter heating demand and summer air
conditioning.
Solar irradiance fluctuations are driven by the time of day and by the season. Sunlight is available
only part of the day and requires day-night balancing. In northern climes like the UK, the load factor
4
Jaffe (2018).
4
Draft
of solar PV is four times higher in summer than winter. Also, cloud cover adds another factor of
unpredictability. The passage of a cloud at midday can cut the output of a solar farm from 100%
to 20-30%5.
Variations in both wind speed and clouds are local effects and, on relatively short timescales, can be
smoothed by distributing wind and solar farms over a large area, connecting these sources through
the Grid. Mean seasonal weather is reasonably predictable and a number of trends can be seen and,
to a degree, mitigated with an appropriate ratio of wind and solar capacity6,7,8.
Mean solar and wind power data (normalised across the years averaged for the UK are shown
below, together with the range of variability: 25th / 75th and 5th / 95th percentiles.
Figure 4 shows:
Infrequent but extreme weather is also important. Three types of extreme weather events are
expected to affect the longer term stability of a predominantly solar and wind energy system in the
UK9:
1. Winter Wind drought – Both low temperatures and low wind over an extended period
(several days) at a time when Northern Europe experiences similar conditions, but with
the potential for offsetting by energy from Southern Europe.
2. Surplus Solar – Solar energy rises above demand during the middle of the day in
summer. This extreme weather event may be important for shorter term storage.
3. Summer Wind drought – Wind power below 6% regularly for a period > 24 hours – and
occasionally 30 days – and which could not be fully compensated by Solar.
5
OECD (2019)
6
Bett and Thornton (2016)
7
Pfenninger (2017)
8
Cardenas et al. (2018)
9
Dawkins (2019)
5
These need to be taken into account in energy system design for storage. Dawkins identifies the
need for increases in longer-term power capacity and the introduction of a new category of multi-
year energy storage, requiring 4-5% of annual demand (see Table 1 below, sized for 2050 levels of
demand). Their estimates are of mean demand but the power system would have to provide the
peak residual power requirements, which could be up to 125 GW during a summer wind drought,
unless ameliorated by demand management. Future changes in weather due to climate change will
also be important, but are extremely difficult to quantify at this stage.
Table 1. Weather Extreme Stress Events applicable for longer term
10
Sinden (2007)
11
Malvaldi (2017)
Residual power in high renewable scenarios
A net-zero carbon target leads towards much higher levels of intermittent renewables. Also, it
virtually prohibits the option of using fossil-fuelled generators that currently provide flexibility and
stability, unless carbon capture rates of 99% can be achieved.
Designers of future energy systems have to consider how to design a reliable and stable electricity
system in which the major quantity of generation is provided by weather-driven solar and wind.
The questions are: How best to model such systems? How can the system be made reliable?
It is clear from the range of the weather fluctuations that all means of reducing residual power need
to be considered. These include:
Supply overcapacity
Interconnections of grids
Demand-side response
They are discussed first below, before modelling energy systems using many years of weather.
Overcapacity and Curtailment
Overcapacity is the provision of additional renewable supply in excess of annual demand. It may be
used either to provide for energy losses in storage, or to fill gaps in supply when wind or solar
output is low, with any excess energy being spilled during periods of high solar or wind output.
Studies for the US13, UK, and EU14 have demonstrated the role of overcapacity in reducing residual
power during a single year and, as a result, the need for energy storage. Reductions in residual
power through overcapacity for longer-term storage cases (year-to-year etc.) seem to be less
effective because persistent low wind speeds, or higher cloud cover cannot always be offset by over-
capacity.
Energy systems dependent solely on wind and solar can lead to periods in which weather-driven
intermittent generation either is in deficit (supplementary supply is needed), or it exceeds demand
(curtailment of supply is required). Currently in the UK, wind and solar generate less than 20% of
electricity on an annual basis, but there are frequent periods when their supply exceeds 50% of total
generation. Even with the currently more flexible energy system, curtailment of renewable energy
sometimes occurs. As the share of generation provided by variable sources rises, excess supply will
12
Dawkins (2019)
13
Shaner (2018)
14
Gils (2017)
occur, and therefore curtailment of renewables will be increasingly necessary to the extent that it is
lower cost than storing or transmitting the energy.
Modelling of highly renewable energy systems by the OECD15 (see Figure 6 below) shows that, when
the variable renewable share increases above 50%, curtailment or energy storage becomes a regular
part of the system operation (see red zone below covering almost half the year), rather than an
extreme event.
Multi-year modelling of the larger EU energy system16 suggests a 10% over-capacity could reduce the
need for year-to-year storage by 10%. In contrast, UK modelling17 demonstrates that a small amount
of overcapacity (i.e. annual supply in excess of demand) has a more powerful effect in reducing the
longer term residual power variability and hence storage needs.
The optimum level of overcapacity – that is, additional renewable supply in excess of 100% demand
– will be determined by the trade-off between the economics of renewables, flexible generation,
and storage. Given the prevalence of ‘low-wind’ periods, it is unlikely that renewable overcapacity
will provide the most economical solution to variability on its own.
Curtailment of supply inherent with high levels of overcapacity is a poor use of these high capital
cost resources, although it could be the lowest cost option in some circumstances. The alternatives
are exports of energy, energy storage, or finding buyers ready to use this extra energy at short
notice (e.g., intermittent hydrogen production by electrolysis). Exports of excess energy are
currently limited by interconnector power capacity. They could potentially make a significant
contribution to system balancing18, although potential importers might well be simultaneously
experiencing similar weather and suffering from under or over-supply, and connection costs could
be significant (see Interconnectors below).
15
OECD (2019)
16
Gils (2017)
17
Cosgrove, Roulstone & Llewellyn Smith (2020)
18
Pujianto (2013)
Interconnectors and grids
Interconnecting grids helps countries share supplies and smooth fluctuations in residual power due
either to demand or supply. At its extreme and with global grids, interconnectors could allow
continuous access to daytime solar and perhaps matching wind and solar to avoid against seasonal
shifts in renewable supply, though at an extremely high cost for interconnectors.
In some larger countries like the US, energy grids are localised, with only low power interconnectors
between regions to allow the trading of energy. However, the development of interconnections
between country grids is accelerating to meet both the growth in demand and the expansion of
renewables. The UK interconnection with Europe is planned to increase from the current 4 GW
(2019) to 12 GW over the next five years19 (about 25% of current peak demand). The connection to
Europe would need to grow to about 30 GW20 to have a significant impact in smoothing the output
of a highly renewable energy supply.
While the cost of high power capacity HV DC connections across land is $1m/GW/km21, undersea
connections are perhaps three or four times more expensive. Conversion losses (1-2% 22) and
transmission losses (0.3% per 100 km23) could be significant for the long links planned between the
UK and Norway.
China already has over a dozen high voltage interconnectors of over 1000 km connecting the east
and the west of the country. At an extreme scale, Singapore is planning a 3800 km interconnector
to join with Australia and meet a reported 20% of Singapore’s electricity needs24.
Models considering a connected electricity system across Europe 25 26 have found that
interconnection could reduce residual power and hence storage need. The distances between
Eastern and Western regions reduces the effect of local weather variation, while the connection
between Southern and Northern Europe provides a balance between areas where solar and wind
dominate. The inclusion of several time zones modifies the peaks in demand. These factors
ameliorate the intermittency of renewables, and, for a perfectly connected 100% renewable system
across Europe, would reduce residual power 33% or more, depending on the year that was
considered.
This effect of interconnectors has also been considered for the US27, where there is not currently a
single national electricity grid. Similar to Europe, the large land area and the long distances between
coasts has the potential to smooth the supply from a weather-dependant renewable energy system.
The US covers several time zones and different climatic regions – North/South and East/West.
Interconnectors would be effective in ameliorating supply and demand variations. The capital cost of
interconnectors was high – estimated at c. $400bn, though how this assessment balances with the
cost of storage is not clear, because only battery storage was considered in their study.
19
https://www.ofgem.gov.uk/electricity/transmission-networks/electricity-interconnectors
20
Gils (2017)
21
Gils (2017)
22
National Grid
23
Gils (2017)
24
https://www.theguardian.com/environment/2019/jul/14/just-a-matter-of-when-the-20bn-plan-to-power-
singapore-with-australian-solar
25
Gallo Cassarino Sharp & Barrett (2018)
26
Gils (2017),
27
Shaner (2018)
Demand side response
Demand side response impacts on storage in two ways:
1. Demand reduction, which reduces the overall need for energy.
2. Load shifting, which changes the timing of energy demand to better match the
variable output of renewable supply.
Both are likely to be major contributors in a future energy system design and to have a significant
impact on the need for storage. Also, the reduction in forecast demand growth as a result of a
programme of energy efficiency measures having the far larger potential effect. Demand growth
reduction is included in the demand forecasts used to estimate system behaviour in 2050 below.
Load shifting moves demand in time without changing total demand but often with some losses. This
is done either by smart scheduling of demands that can be delayed (e.g. overnight EV charging), or
providing economic incentive to cut demand in times of supply shortage or increase use during times
of excess supply. It is therefore often considered as an alternative to storage.
Recent analysis28 for the Committee on Climate Change shows that up to 53% of residential
electricity demand, 32% of commercial electricity demand, and 22% of industrial electricity demand
are potentially movable, mostly only for a few hours. Load shifting may be lower in cost than
storage, typically requiring capital investment only for the metering and control technologies. For
this reason, the effective potential is expected to increase as smart metering and appliances become
the norm and estimates of the potential in the UK range from 5 GW to 11 GW29 for current demand
or perhaps 10-22 GW for the larger 2050 level of demand. Some industrial demand can be managed
by delaying demand on a planned basis by up to a day by using some economic incentives. EV power
demand can be shifted by up to 8 hours by controlling the rate and the timing of charging 30 which
smooths variations in overall demand.
Load shifting is important in smoothing short-term daily peaks in demand, hence reducing the very
high range of residual power and therefore the power output requirements energy storage.
However, it is unlikely to be effective for weekly- and longer-term variations because customers
would be unable/unwilling to defer the activities for which they required power over such durations.
There are no studies showing that long-term load shifting can be effective or could make a
significant contribution to system balancing. In the context of storage needs, demand side response
should be seen as a shorter-term strategy alongside battery storage and which could reduce the
storage release power requirements from 145GW by 30 GW or more, to below 120 GW by delaying
peak demands by up to 8 hours.
There is some experience of demand suppression during national emergencies (such as a 3 day week
the in 1970s) where demand was suppressed by order of the Government; this was a serious
disruption to normal life. Also, there were political consequences, including an election and a change
of Government. Therefore, demand suppression by ‘fiat’ does not seem to be a viable basis for
future energy planning.
28
Vivid Economics (2019)
29
Torriti & Green (2019)
30
Crozier (2018)
need for storage have been made, with diverse and interesting results31 32 33 34. However, few studies
consider renewable supplies above 75%. Also, studies often included flexible fossil generation
because their carbon target was less stringent than net-zero. Some assumed a single store, hence,
not addressing the duration of energy storage. Few considered that storage is less than 100%
efficient. Taken together, these assumptions are unrealistic and would lead to incorrect estimates of
the storage needs for a net-zero emissions target.
While drawing on ideas in earlier studies, we will focus on one for the US, another for Europe, and
two for the UK that are based on multi-year simulated renewable supply, inferred from past
meteorological data, and using models of possible demand in 2050.
US & EU highly renewable energy studies
A recent US study35 of a renewable system for the whole of the country analysed hourly solar and
wind resource data for the US for the 36 year period 1980-2015, with 100% renewables It covered
the whole of the US land area, utilising the long distances between coasts and North to South to
smooth renewables supply and demand. They found that storage is always required to achieve
system reliabilities greater than 90%. Both excess renewable capacity, curtailment, and storage
increased sharply as the renewable share grew above 70%.
Storage need varies with the over-capacity level (see Figure 7 below). Limiting storage to 12 hours
requires 120% overcapacity to meet the US system reliability criteria (loss of load < 0.1 days pa).
Increasing storage to 32 days (9% of annual demand) requires only 10% overcapacity.36
Figure 7. System reliability plots for different energy mixes & areas of energy generation. [Shaner (2018)].For areas as large
as the whole US (CONUS) reliability requirements are not met without > 12 hours (0.15%) storage, even for 50% over-
capacity.
31
Blanco (2018)
32
Cebulla (2018)
33
Jacobson (2015)
34
Aurora (2016)
35
Shaner (2018)
36
Shaner (2018)
Another study37 of the large, connected European energy system for the 21 years, 1984-2004,
showed that storage, or other flexible energy needed to be 8-11% of annual demand, and there was
significant curtailment of renewables, with the amount rising sharply as variable renewable energy
volumes increases above 70%. There were high storage losses for high solar shares and large
transmission losses for high wind shares.
UK current demand storage studies
There are few UK energy studies of system that are highly reliant on solar and wind energy. Two
system-level studies38 39 look at the cost of providing energy for 2050 levels and types of demand
from very low-carbon energy supplies, including high levels of renewables. Because they are options
studies, it difficult to see the underlying assumptions about weather and about the choice of flexible
supplies or energy storage, other than short-term storage.
Detailed studies of either one year40 41 or a few years42 of weather data, only one of which included
storage efficiency, show that storage is minimised with small solar shares (15-25%). They showed
the characteristics of storage needs for current levels of electrical demand (330 TWh):
High peaks in power capacity (9-12 GW) for the short cycles of less than a week with
low storage sizes (100 GWh).
Medium term (weeks) storage with a lower power capacity (2-3 GW) but a higher volume
(1-2 TWh).
Large winter-summer storage needs (17 TWh) due to the differences in demand and
supply between winter and summer that are not fully compensated for by the solar/wind
mix.
Evidence of larger year-to-year residual energy fluctuations amounting to several 10s
TWh driven mainly by differences of supply from long term variations in mean wind speed
and demand from changes in ambient temperatures.
From these studies, an initial estimate of the storage need for a single year, is 20 TWh (c. 6% of
annual demand). More is required for periods longer than a year. Also, up to 25% of the energy
demand is supplied from storage, meaning that the round-trip efficiency of storage is
important.
UK storage for future demand
We now look ahead to 2050 when the demand for electricity will be much larger 43 – and could be in
the range 500-700 TWh – and ask:
What storage would be required if the only source of energy supply was wind and solar?
How to take account of extreme weather (by including very long periods of weather data)?
To what degree can storage needs be reduced by greater use of interconnectors and
by other, dispatchable, low-carbon sources of supply?
UCL44 have modelled the UK storage needs for 2050 using their DYNEMO model for a highly
renewable, wind-heavy system using 31 years of weather data, with both onshore and offshore
wind. Large new demands for space heating (heat pumps plus resistive heating but not EVs) are
37
Gils (2017)
38
Strbac (2018)
39
Aurora (2018)
40
Cardenas (2018)
41
Cosgrove, Roulstone & Llewellyn Smith (2020)
42
Gallo Cassarino, Sharp & Barrett (2018)
43
Climate Change Committee (2019)
44
Barrett & Sharp (2020)
modelled, together with existing demand, both modified by ambient temperature. These demands
are compared hourly with energy production from renewables.
The aim of this study was to determine the size and distribution of residual energy rather than
storage need. Neither storage efficiency, nor overcapacity, nor interconnectors were included in the
study. Longer term variations of residual energy are dominated by the effect of weather (see Figure
8 below)
Changes in temperatures affect demand by up to 7%. More significantly, solar and wind fluctuation
affect supply. Within any single year, the average monthly renewable generation varies between +60
GW and -100 GW, filling the energy stores in the spring and summer and providing back-up for the
energy system during the autumn and winter.
On average, supply exceeds demand over the 30 year period and annual residual power varies
between extremes of about +12/-10% of demand.
Figure 8. UK energy system hourly model response to renewable supply variation over 31 years, 1980-2010
45
Cosgrove, Roulstone & Llewellyn Smith (2020)
no forecast un-met demand within the following 168 hours, the excess is held in the Long duration
store, to be allocated to a later energy need.
Future electrical demand (600 TWh) is modelled in a simpler manner than Barrett. It includes hourly
profiles of current electrical demand and both electric vehicle charging (75 TWh) and a share of the
heating load (200 TWh) without modelling demand variation either with ambient temperature or
between years.
The renewable energy supply is sized to match 600 TWh of demand over 37 years (1980-2016) of
hourly weather, recognising that this could not meet demand without some combination of storage,
overcapacity and curtailment, and flexible back-up. The renewable energy supply system includes
solar share as a variable. For wind, the supply is representative of the large wind turbines that are
being planned and built, most of which are both offshore (70% of wind supply) and will be widely
dispersed.
Large year-to-year wind variations drive the longer-term (more than 10 years) patterns of residual
energy fluctuations (see Figure 9 below). Residual energy varies regularly by +/- 70 TWh, (+/- 12% of
annual demand) with higher infrequent extremes. Multi-year residual energy variations are driven
by large year-to-year changes in wind. The cumulative effect over the whole period shows an excess
of supply between 1980 and 2000 of 150 TWh and a matching deficit between 2000 and 2016, of
about 3% of demand for this time. Periods of excess supply lead to renewable curtailment. Deficits
need to be addressed either by large long-term stores, or overcapacity, or some other zero-carbon
energy source.
Figure 9. UK annual residual power and cumulative residual energy fluctuations for weather 1980-2016 for a highly
renewable system (20% solar, 80% wind) for 2050 demand (600 TWh) future fleet MERRA-2 & NUTS-2 weather data46
These multi-year weather effects can be correlated with ocean temperatures 47. In both the UK and
Europe, average wind speeds respond to North Atlantic Ocean temperatures, cycling with a period
between 10 to 20 years. These weather fluctuations are very long-term and are not currently
capable of being forecast. Also, because these effects are both widespread and last for many years,
they cannot always be reduced by interconnecting grids.
Baseline storage
Longer-term fluctuations in supply dominate the problem of ensuring supply and of sizing storage
needs for an idealised system with 100% efficient stores. Assuming 100% efficient stores in sizing
provides the baseline for the other cases, prior to introducing more realistic efficiency assumptions
and using additional means to reduce the size of storage that is required.
Modelling using a sequential energy allocation method48 shows that the baseline level of storage
needs for 20% solar share (where storage needs are close to a minimum) is large – 238 TWh (39% of
demand) as both year-to-year and longer-term weather variations are accommodated.
The power requirement of storage – up to 146 GW – is very high. This level is driven by both the
large range of power demands of the larger energy system, and by the need to cover extreme
weather events when renewable supplies are not available (see Table 1 above for ESEs). The peak
power requirements for this ESE is somewhat lower – 125 GW. The level of energy supplied to meet
demand supplied from storage reflects the size of the stores and the frequency of cycling of residual
power.
It is likely that the choice of storage technologies and perhaps flexible supplies would be different
for different storage durations – ranging a few hours with many cycles each year for short term
fluctuation of residual power, to many months for annual demands, and even longer for very multi-
year weather-driven variations.
There are several ways of reducing the amount of stored energy from the baseline of 238 TWh and
addressing the high power output demands. These may be used separately, or in concert. The
following sections explore some of the options and the modelling results are given in Figure 10
below.
Overcapacity
Overcapacity is the provision of additional renewable supply in excess of annual demand –
expressed as a percentage of average demand. It may be either to provide for energy losses in
storage, or to fill gaps in supply when wind or solar output is low, with any excess energy being
spilled during periods of high solar or wind output if it is not stored.
Energy storage is never 100% efficient and it is clear that no single storage technology will be satisfy
all the storage needs. Different storage technologies have different performances and economic
characteristics and will be required to store energy for different periods of time. Long term storage
technologies, such as hydrogen and ammonia, have low round-trip efficiencies (of perhaps 40% or
25%, respectively) reflecting the energy lost in the steps of conversion to chemical energy and the
subsequent reconversion to electricity. Other systems such as CAES can have higher efficiencies, up
to 70%, and batteries have round-trip efficiencies greater than 90%.
To illustrate the effect of inefficiency for the three-level storage system, we modelled:
Li-ion batteries for the Short store that requires rapid response with high power levels
and efficiency but has relatively small capacities and high costs,
Adiabatic Compressed Air Energy Storage (ACAES) for the Medium store that requires higher
volumes with its somewhat lower efficiency and lower capital costs for storage,
Hydrogen for the Long store that has the highest volumes and requires the lowest
capital costs, but has low round-trip efficiency and high power costs.
We find that the amount of overcapacity must be increased to balance the system. Total store size is
reduced with increasing overcapacity: 105 TWh at 20%, 63 TWh at 30%, and 55 TWh at 40%
overcapacity – all at 20% solar share49. Once there is enough overcapacity to offset storage losses,
less than 20% of the additional energy satisfying demand. Curtailment or spillage rises by more than
additional supply from overcapacity because of reduced storage losses.
200
150
Total Storage
238
100
50 98
63 53 46
0
Baseline 6% OvercapAssumed eff 30% 25% Baseload25% Baseload w
Overcap30% OvercapInterconnect
Figure 10. 2050 UK energy system - 37 years 1980-2016 – Means of reducing storage needs.
The distribution of storage needs between short and long term depends on the amount of
overcapacity, the share of baseload and the effectiveness of interconnectors. Short-term (up to day)
storage needs are 100-250 GWh and weekly storage needs 5-8 TWh with the balance (40-65 TWh)
for the long-term54. The largest amount of energy released is that for weekly storage of a total
energy supplied from storage (40-77 TWh) because of the many times that it stores and releases
energy.
All these cases require very high power outputs from each of stores, on occasions more than 140
GW. Apart from battery storage, providing this power capacity from energy storage will be very
expensive. There are several means of reducing the range of power output requirements, including:
demand management; peak lopping using lower capital costs flexible supplies such as simple cycle
chemical turbines, reciprocating engines, or load-following nuclear; altering the management of
stores to allow more than one to supply energy at a time.
Demand variability is a major driver of residual power requirements. Demand management that, for
example, delayed demand by up to 8 hours could reduce storage power output requirements for
the base case from 146 GW to 137 GW and for the 25% baseload case from 133 GW to 122 GW.
The scope for removing peaks in residual power deficits with high capacity peaking plants (68 GW
similar to the level of mean demand) is large, reducing the power demand on storage from 146 GW
to about 70 GW. Relatively small amounts of energy (0.1-1 TWh pa) are supplied during peak
lopping. As a result these plants would have very low levels of power availability (less than 0.2%) and
thereforve high marginal energy costs.
Short-term, medium-term and longer-term storage technologies have quite different requirements.
High energy-power ratios affects the choice energy storage technology – battery capacity, thermal
store size or gaseous storage volumes. Also, the round trip efficiency is important where energy is
stored and release frequently. Conversely, lower energy-power ratios raise the importance of the
means of conversion of energy to and from stores – inverters, compressors, electrolysers and
turbines.
Other storage technologies could be considered for the longer-term needs. An example is ammonia
for long-term storage rather than hydrogen, because it can be stored at low pressure. The use of
ammonia, with its lower efficiency (25%) for the Long term stores, would increase total storage size
by 32% to 71 TWh without interconnectors or by 37% to 57 TWh with interconnectors. Long-term
annual plus multi-year storage needs increase by 42% with ammonia stores, for the same level of
overcapacity (30%)55.
Similarly, the medium-term store could use non-adiabatic CAES (efficiency ~50% 56 57) or other
technologies with lower efficiency than ACAES which was modelled. This lower efficiency would
affect the storage needs for the weeks/months period.
All the estimates of storage needs are sensitive to both the weather data used and the profile of
demand. For a lower level of demand (500 TWh)58 associated with the wider use of hydrogen from
natural gas, storage needs fall to 38 TWh and for a high demand case (700 TWh)59 storage needs are
increased to 54 TWh.
The choice between the different means of providing a zero-carbon energy system will be economic,
related to the capital cost of renewables and storage, and the cost of energy from baseload
generation. Estimates are put forward very tentatively to provide a guide to the scale of storage that
the UK will require to design highly renewable energy systems for net-zero. Further studies are
required of the more complex cases to establish the interaction of the different means of reducing
storage size and power capacities in a renewables-heavy energy system.
Conclusions
A net-zero emissions target will lead to at least a doubling of the demand for electricity in 2050
even with a programme of energy efficiency measures. There are important decisions to be taken
about the main means of heating (hydrogen, or heat pumps), powering transport, and the use of
CCS and DACC that will affect the size of the energy system and hence storage needs.
If all of UK’s net-zero electricity was supplied by wind and solar – which we are not advocating – with
the solar-wind mix adjusted to follow the seasonal pattern of demand, the UK would need to be able
to store up to 39% annual demand, much of this for more than a year. This large amount of storage
could be reduced substantially by providing overcapacity, interconnectors, baseload and flexible
generation. The provision of 30% over-capacity of renewables, high power interconnectors and 25
GWe of baseload biomass or nuclear energy would cut the storage needs to 7.5% of annual demand.
In any of the scenarios that were considered, there will be a need for very large amounts – many 10s
TWh – of long-term energy storage.
Several storage technologies will be required for the different daily, weekly and longer term needs.
Short term stores must have high power capacity and must respond quickly. For long-term stores the
issues are how to provide the large quantities of storage required and also high powers, their round-
trip efficiency, and the economics of infrequent use of peaking plants. Any highly renewable energy
system will require back-up means of supply for the very infrequent and extreme weather events,
when wind power is low for several weeks and covering almost the whole of range power demands
– amounting to more than 100 GW.
The economic choice between these scenarios is uncertain for several reasons. The cost of some of
the different systems is not well known because they have not been built at scale. The trade-offs
between overcapacity, storage, and transmission losses in the UK have yet to be fully assessed. Also,
some of the storage technologies are novel and their technical and economic characteristics are not
well established.
The large costs of balancing and backing up an intermittent, highly renewable system using energy
storage could be reduced, but not eliminated, by one of the two means:
1) Scale of the renewable electricity system was reduced, by:
a. Using substantial amounts of natural gas to provide heat and/or to produce
hydrogen by steam methane reforming. In both cases, carbon capture and storage
with very high capture rates would be required. Also, the CO2 and associated
methane that is not captured would have to be offset or captured to achieve net-
zero carbon emissions. These emissions would be in competition with other
demands for carbon offsetting or capture that will be required to compensate for
the hard to reduce emissions: aviation, some industrial processing etc.
b. Demand reduction by radical energy savings, or by ‘force majeure’ to address the
effects of infrequent extreme weather events.
2) Increasing the flexibility of the energy system and restraining intermittent wind and solar to
50% share, by providing both:
a. A substantial part of the UK’s electricity (100 TWh pa), perhaps double what is
proposed by the Committee on Climate Change 60 from burning biomass and
capturing and storing the CO2 emitted (BECCS). For this to happen, the UK would
have to become a major importer of biomass, for which there will also be competing
demands in UK and internationally. BECCS have other priority demands such as
replacing fossil fuels in making synthetic fuels and plastics.
b. A much larger share of energy (200 TWh pa) from new more flexible nuclear by
constructing modern LWR reactors (including Small Modular Reactors) and providing
an economic incentive to operate these in a load-following mode.
Finally, more detailed studies are required to:
Optimise the power capacity requirements of storage for different time periods.
Determine the best approach to covering the power needs of infrequent extreme
weather events.
Establish the scale of the need for energy storage in different net-zero scenarios.
Determine their economics and how they compare with alternatives.
These studies should be periodically updated as the probable route to net-zero becomes better
defined.
21