Sir Christopher Llewellyn Smith presented an important paper on energy storage to the 2022 Oxford Energy Day, of vital relevance to the UK’s net zero ambitions. The methods and findings also have implications for, and should trigger similar analysis in, many other countries planning to rely on weather dependent forms of low carbon generation. The basic question is how best to combine the storage of surplus output from low carbon sources (nuclear or renewable) when output is high and consumer demand is low, with the need to draw down when demand is high and output low.
Drawing on material prepared for a forthcoming report to the Royal Society, and previously also presented to BEIS[1], he outlined both the complexity of the choices that lie ahead and the information, assumptions and methods necessary to resolve them. In principle the storage issue is not confined to renewables, but applies to any system with less flexible generation, eg baseload nuclear, if output is not immediately controllable to exactly match load.
A precondition for understanding the economics of storage is to appreciate the significance of both the frequency of the charge/discharge cycle and the scale involved. It’s often widely assumed by commentators that solutions are simply a matter of advancing battery technology, thus reducing the cost per kWh stored of battery capacity, together with more pumped hydro storage facilities. Batteries, plus pumped hydro and a few other technologies for short term storage at similar cost, together with measures to persuade consumers to spread their loads, are likely to be sufficient to managing short term balancing and the average within the day mismatches between consumer demand and variability in renewable supply.
But these are not not nearly adequate for a renewables based system of the nature currently envisaged for the UK. One reason is the frequency or infrequency of the charging/ energy release cycles. It is relatively easy to justify £ 75-100 per kWh of storage capacity for a battery[2], or a similar capital investment in pumped storage[3], when there is a daily or more frequent charging cycle and the fixed cost of the capital investment is therefore spread over up to 365 days a year. This is not the case for seasonal storage, or, for provision against extreme weather events. With an annual or less frequent cycles for a single charge and discharge, such a high capital cost is unaffordable.
The other factor is scale. The scale of summer/winter compared to daily imbalances is an order of magnitude greater, up to 2000 times greater than daily imbalances, although for the UK this can be substantially reduced with the right mix of wind (winter) and solar (summer) renewables.[4] The bigger issue is weather variability. This goes well beyond daily volatility and including both “wind droughts” that can last a couple of weeks and very substantial multi-year variations.
This scale necessitates much lower capital cost storage solutions, even if the “round trip” energy conversions are less energy efficient than batteries. On current knowledge this implies some form of chemical, thermal or large scale compressed air storage (ACAES). “Green” hydrogen produced by electrolysis is currently seen as the leading option, and the working hypothesis in the study for the Royal Society.
Renewables generation in the UK is currently based mainly around wind and solar, and their weather dependent nature largely defines storage requirements. Most significant are “extreme” weather events, such as an extended “wind drought” when the wind is not blowing anywhere in Northern Europe, and big variations driven by the North Atlantic Oscillation in annual renewables output – years with less sun or less wind.
This highlights the importance of meteorological data. Substantial variations in annual output, and sustained deviations from average, mean that looking at just a few years of data does not reveal the scale of the problem. Llewellyn Smith uses “middle of the road” projections for consumer load, and “real weather” meteorological data over 37 years for his assessments, though expert meteorological opinion is that even this may not be enough to capture extreme events. In itself this is an important observation and is compounded by uncertainties over the future impact of climate change on UK weather.[5]
Importance of the right mix of renewables
A first, and surprising, result is that, for the UK, and at least with projections of current load patterns, the need for storage is dominated by weather volatility not (as often asserted) by seasonality per se. In other words what matters is extreme weather or unusual weather patterns, eg two or three weeks of no wind or whole years with poor renewables output.
Typical or “average” seasonal storage requirements could actually be largely managed by having the right mix of wind and solar, with an 80/20 wind solar[6] split as one feasible option on the basis of the particular assumptions in the study.
However one of the findings of the study is that even the seasonal imbalances between summer and winter can vary substantially from year to year, reinforcing both the need for much more concentration on weather volatility, and the conclusions on the need for substantial long term storage provision.
Exactly how much storage is required will also be an economic choice based inter alia on the relative costs of storage and renewables capacity, and on conversion capital costs and efficiencies, but initial indications suggest numbers in the range 130- 200 TWh, serviced by 80-120 GW of electrolysers. The TWh of energy storage is very substantially higher than simply seasonal balancing needs, and represents over half total current annual consumption, while the electrolyser capacity in GW comfortably exceeds current peak demand.
The result demonstrates the system importance of managing the mix of weather dependent renewables. Bringing baseload nuclear power and fossil generation with carbon capture into the choice of technologies also offers further options, but at least for a system with substantial renewables content, the analysis has essentially similar characteristics with respect to storage investment choices.
Most analysis has ignored capacity required for both input to, and extraction from, storage.
In the past reliability provision has been dominated by the need to meet “needle peaks”, to have sufficient kW capacity to meet extreme conditions perhaps experienced in only one half-hour of the year, with adequate kWh of energy reserve provided by stored coal, oil or gas. In the future reliability will depend much more on use of a kWh energy reserve without use of fossil fuel.
An obvious but hitherto mostly neglected element in the calculation is the capital investment in the energy conversion capacity required for putting kWh of energy into storage and taking it out for final use. Both require significant high capital cost capacity – electrolysis for the production of hydrogen as input to store (when renewables output exceeds demand), and then use of that hydrogen to generate useful energy, as electricity or heat, when demand is high. Both these types of conversion capacity will also have to operate at relatively low load factors, and the scale of electrolysis capacity will likely need to match that of current generation capacity. The loss of energy in this “round trip” – 41% efficiency for hydrogen - is often discussed, but not the amount or cost of capacity required to charge or discharge from store. This should be sufficient capacity to match either surpluses or shortfalls in renewables output, but there will be a trade-off between the extra costs implied by “wasting” or spilling renewables output, the size of storage requirement, and spending more on electrolysis capacity. Identifying this gap in the analysis is another important result from the Llewelyn Smith analysis.
Policy Implications
There are also further general policy conclusions to be drawn from this work.
The first is that there is a clear need for whole system coordination, over and above what can be delivered by markets alone. There is no visible mechanism by which uncoordinated responses to market signals can be guaranteed to provide a credible approximation to the right seasonal balance of renewables, notably wind and solar. Nor is this deliverable through collaboration between localised, decentralised systems. Similar considerations apply to getting the right amount of conversion capacity to ensure an economic and reliable infrastructure for storage – including both the stores themselves and the conversion capacity for inputs to store, electrolysis, and outputs as eventual use of the stored energy. This coordination is essential both at the investment stage and in system operation.
A related policy conclusion is that the fundamental decisions on power sector investment and infrastructure need to be closely coordinated with decisions for the heat sector. Reliance on heat pumps for example will substantially increase the seasonal imbalances and introduce weather related volatility on the demand as well as on the supply side.
A third is that we need to re-examine and discuss the treatment of reliability and risk in power sector policy and planning. A disturbing message from the meteorological data is that we do need to consider the possibility of very occasional crises, even if perhaps only one every few decades, when there is an extreme and sustained shortfall in energy provision. Such crises are of course not confined to renewables systems. The UK experienced the 3-day week of the 1970s, induced by industrial unrest, and the petrol rationing of the 1956 Suez crisis, and could experience, for example, type fault issues with nuclear power. The issue is how we plan against extreme events. The answer is not necessarily excessive and unaffordable investment in spare capacity, but to accept that from time to time we will have to face major crises which are disruptive but can be managed without societal collapse. The recent pandemic is just one such example.
What this might imply for managing some future energy crisis might be deduced from the 3-day week[7]. A temporary shutdown of energy intensive industries, for example, might also be interpreted as the realisation that much of our “store” of energy can in fact be more cheaply held in the form of stocks of the energy-intensive products themselves. There are in reality many ways of managing energy and many other risks. What is required is identification of what the risks are and the development of good coping strategies.
Perhaps a final observation Is that we cannot assume historical weather patterns, even over many decades, are not necessarily a perfect guide to what weather we might experience as climate change has an increasing impact on weather patterns, introducing additional uncertainty and additional risks. So climate change adds further risks even to the measures that need to be taken in mitigation.
Expect more discussion of these issues as they increasingly impact on our strategic choices.
[1] The main conclusions (subsequently refined) were presented in April 2021 www.era.ac.uk/Medium-Duration-Energy-Storage and were communicated to BEIS in May 2021.
[2] A credible target given the pace of current advances in battery technology. Peter Bruce’s assessment of battery costs at the Energy Day suggested an estimate of this order.
[3] Dinorwic pumped hydro in North Wales stores about 8 GWh output of energy. It cost c £425 mn in 1990, or about £ 100 per kWh in 2022 money. Current lithium battery costs have recently fallen to this level and can be expected to fall a little further, perhaps to around £ 75 per kWh.
[4] We can define a hypothetical storage need for “flattening” a load curve as the storage needed assuming 100% conversion efficiency, 100 % load factor of generation output which exactly equals total consumption. On this basis daily load curve flattening, historically, needs a mere 80 GWh, about 10 Dinorwics. But a notional flattening of the annual curve, calculated from monthly seasonal pattern, needs about 17,000 GWh or about 2000 Dinorwics. In fact as the analysis shows, this need not depend entirely on storage and, for current seasonal patterns, could be largely managed by seasonal balancing (wind/solar mix) of renewable capacity. Current UK annual consumption is of the order of 300,000 GWh. (Seasonal balancing needs can be checked from data at statista or other data sources)
[5] The Met Office suggest this does not provide a sufficient sample of “rare” weather events, indicating the need for contingency provision.
[6] Readers may be surprised that a smaller solar percentage actually requires storage of winter output for use in summer.
[7] It is interesting that at least one commentator has suggested that Germany, given the risks of its high dependence on gas, might usefully learn from UK experience in 1973/74 measures.
