Winter Storage Costs 2 to 16 Times More Than Europe's Batteries
Twelve hours of batteries cost about 25 to 41 pence a person a day. Covering weeks of winter wind drought costs 74 pence to £4, on published estimates. All hand arithmetic, no dispatch model.
In my last grid post I argued that a northern European grid can reach about 90% wind and solar with under 12 hours of battery storage, if it has strong transmission and some dispatchable backup. A reader will ask the obvious question. What covers a cold, windless fortnight in January? This post prices that gap. My result is a range, and it is wide: winter storage costs roughly 2 to 16 times what the battery hours cost, per person per day.
I computed everything below by hand from published numbers. I did not use the Lab. My own simulation is pending (see the end).
Question
Does the storage that matters at high wind and solar shares come in hours or in weeks? And what does each cost per person per day?
All quantities are final energy (electricity delivered), not primary energy. Money is mixed: the Royal Society figures are 2021 pounds and the battery figures are 2025 dollars. I convert at £1 = $1.3. That rate is my assumption.
Assumptions
| Input | Value | Source or status |
|---|---|---|
| UK electricity demand, 2050 | 570 TWh/yr | Royal Society executive summary [2] |
| Population | 69 million | My assumption, not sourced |
| Demand per person | 22.6 kWh/day | Hand arithmetic: 570e9 kWh / 69e6 / 365 |
| Seasonal store needed (UK) | 60 to 100 TWh of hydrogen | Royal Society [1][2] |
| Seasonal store needed (Europe) | 91 to 385 TWh, 1.8% to 7.8% of yearly demand | Kittel et al. [3] |
| Battery turnkey price, 4-hour | $110/kWh global, $177/kWh Europe | BNEF 2025 survey, from a news summary [4]. I could not open the page, so treat as single-source |
| Battery size | 12 hours of average load | My choice, from the last post |
| Annuity | 5% a year, 15 years, factor 0.0963 | My assumption |
| Hydrogen storage adder | £33 to £179 per MWh of demand | Derived from Turver's replication [5] |
Per person, demand is 22.6 kWh/day, so the average load is 0.94 kW. Per person per day, please: that is the unit I use for every cost below.
Data and where it came from
The Royal Society report used 37 years of weather data. It finds up to 100 TWh of storage in 2050, mostly hydrogen in salt caverns [1]. The executive summary gives a hydrogen-benchmark system cost of £52 to £92 per MWh, with wind and solar at £30 to £45/MWh and discount rates of 5% to 10% [2]. It gives no single storage cost.
Kittel and colleagues ran 36 weather years (1982 to 2016) for Europe. Their long-duration storage need is 91 TWh (median, unlimited exchange) up to 385 TWh (max, TYNDP exchange). The worst winter was 1996/97 in every scenario. The policy-relevant case with hydrogen exchange needs 351 TWh, or 7.1% of yearly demand [3]. The page I read does not give a cost per kWh, and I did not read its supplementary information.
Bernecker and colleagues add a cost angle. Their robust optimisation of a fully decarbonised Europe finds a single worst-case regional wind-and-solar drought raises system cost by 9%. Multi-region events raise it by up to 51%, and a Europe-wide event by up to 71% [6]. Localised events need only renewables, batteries and transmission. Large events need long-term hydrogen and load shedding.
For the cost of the hydrogen chain I rely on David Turver's replication of the Royal Society model [5]. He is a critic, and his realistic-case inputs are his own. I treat his table as one end of a range, not as truth. His numbers are the only itemised ones I could read.
Method
Step 1, the size gap. Battery: 12 h × 0.94 kW = 11.3 kWh per person. Seasonal store, UK: 60 to 100 TWh / 69 million = 870 to 1,450 kWh per person. That is 77 to 128 times the battery. Using the report's own ratio, 60 to 100 TWh against 570 TWh/yr is 10.5% to 17.5% of a year's demand, or 38 to 64 days. The battery is half a day. For Europe, 1.8% to 7.8% of yearly demand is 7 to 28 days. I do not know whether the report states store size as hydrogen energy or as electricity out, so read these as storage capacity as quoted.
Step 2, the battery bill. Europe at $177/kWh: 11.3 kWh × $177 = $2,000 per person. Times 0.0963 is $193 a year, or $0.53 a day. Global at $110/kWh: $1,243 per person, $120 a year, $0.33 a day. In pounds that is £0.25 to £0.41 a day. Per MWh of demand that is £11 to £18. This excludes running costs and conversion losses. It also uses a 15-year life that I assumed. It matches the 32 cents I found in my earlier issue.
Step 3, the winter bill. Turver reports system cost per MWh delivered for several cases. Renewables alone cost £89.05/MWh at his 11% discount rate. I subtract that from each total to get a storage adder. This is my derivation, not his. For the pairs of totals he does not label, I take the lower as the "average to 2050" case. That is my reading and it could be wrong. I multiply each adder by 22.6 kWh/day.
Result
| Case (storage adder over renewables) | £/MWh | £ per person per day | Multiple of battery (£0.25 to £0.41) |
|---|---|---|---|
| Royal Society inputs, 5% rate [5] | 32.84 | 0.74 | 1.8 to 3.0 |
| Royal Society inputs, 11% rate [5] | 54.63 | 1.23 | 3.0 to 4.9 |
| Higher electrolyser and turbine capex, 11% | 91.11 | 2.06 | 5.0 to 8.2 |
| Plus lower efficiencies, 11% | 100.28 | 2.27 | 5.5 to 9.1 |
| Plus 30% capex inflation, 11% (average case) | 122.35 | 2.77 | 6.7 to 11.1 |
| Same, "today" case, 11% | 179.35 | 4.05 | 9.9 to 16.2 |
The top row against the cheapest battery gives 1.8. The bottom row against the cheapest battery gives 16. That is where "2 to 16 times" comes from. In energy terms, £0.74 to £4.05 a day is 3% to 18% of the 22.6 kWh/day if electricity cost £1 per kWh. Easier to see per MWh: the adder is £33 to £179 on top of a renewables cost of £45 to £89.
The sign is stable. Every row costs more than the battery hours, even the most generous one. The size is not stable.
Here is the verdict in three words that I try to keep apart. Physically possible: yes. The report says salt caverns can hold this, and three UK hydrogen caverns have operated since 1972 [1]. Affordable: only at the low end. Politically likely: I leave that open, because I have no evidence on permitting.
What this means for my 90% claim
My earlier position says the last 10% is the expensive part. This analysis suggests why. A 90% share leaves 10% of annual energy for dispatchable backup. The Royal Society store is 10.5% to 17.5% of annual demand. That is the same order of size. At 90%, a gas or biomethane fleet that runs a few hundred hours a year is the seasonal store, held as fuel. The cost then sits in idle power plant, not in the cavern. I have not priced that fleet. The numbers above apply to a near-100% hydrogen design, so they overstate the 90% case. How much they overstate it is the missing number.
Turver's table points the same way. Caverns are not the main cost. Electrolysers and turbines are, because they run few hours a year. Capex per kW and efficiency drive the result, not the price of the hole in the ground. He quotes £333/kW for the Royal Society electrolyser against £1,350 to £2,500/kW in his realistic case [5]. I cannot say which is right.
Sensitivity: which assumption moves the result most
The two most uncertain inputs are the discount rate and the electrolyser and turbine bundle (capex plus efficiency).
| Royal Society capex and efficiency | Turver capex, efficiency and inflation (average) | |
|---|---|---|
| 5% discount rate | £33/MWh (0.74 per person per day) | Not available in my sources |
| 11% discount rate | £55/MWh (1.23) | £122/MWh (2.77) |
Moving the rate from 5% to 11% multiplies the adder by 1.66 (hand arithmetic: 54.63 / 32.84). Moving the equipment bundle at 11% multiplies it by 2.2 (122.35 / 54.63). The equipment bundle is the larger lever, but the empty cell matters. I cannot cross the two factors without rerunning the model, so I do not claim a combined range beyond the table above.
Does the range flip a verdict? Take "cheap storage" to mean at most the battery cost, £0.41 a day. No row passes. So the verdict on cheap winter storage does not flip. What flips is the size of the bill. At the low end, storage adds about 40% to a £77/MWh system. At the high end, it more than doubles a renewables cost of £89/MWh. Change the two biggest assumptions and look again: the answer moves by a factor of 5.5 (179 / 33), which is the real finding.
What I got wrong, or cannot yet say
- The Turver case is one critic's reading of one report. Its inputs are not peer reviewed. I use it because it is itemised.
- I mix a UK study with a Europe-wide one. Kittel's store is a smaller share of demand (1.8% to 7.8%) than the UK's (10.5% to 17.5%). If the European figure holds, the costs above fall by roughly a factor of two to six. This is a rough ratio and I did not rerun any model.
- The Bernecker result, 9% to 71% extra system cost, is relative to a baseline I could not see [6]. It supports the direction, not my numbers.
- The battery price comes from a summary I could not open. Pack prices alone are $70/kWh for stationary storage in BNEF's 2025 pack survey [4b]. Turnkey Europe is much higher.
- Population is my assumption. A 5% error changes the per-person costs by 5% and leaves every multiple unchanged, because both bills use the same demand.
What would change my mind: an hourly model showing that a 90% system with a modest gas fleet needs a winter bill below £0.41 per person per day. I would also change my mind if a cavern-plus-electrolyser study with audited capex put the low case above £1.2.