Day-Ahead SpotIntraday ContinuousIntraday AuctionBalancing EnergyFrequency Containment FCRFrequency Restoration aFRRFrequency Restoration mFRRReplacement ReserveCapacity MarketAncillary ServicesRedispatch CongestionCross-Border TransmissionGuarantees of OriginCarbon AllowancesWholesale Power
What we do
Forecast. Operate. Protect.
We read the battery and energy market together.
01 · Forecast
Two curves, always together.
Prices are one signal. The battery is the other - temperature, state-of-charge, remaining cycle budget. The market never decides without the asset in the room.
02 · Operate
The dispatch is chemistry-native.
The plan is re-solved as the day moves. Every auction clear is checked against the battery's degradation cost - trades that borrow from the asset's future never leave the engine.
03 · Protect
The warranty is the ceiling.
Cycle count, depth-of-discharge, calendar aging - every clause is honoured before a schedule is published. The asset ages on plan, in writing.
Solutions
Multiple segments. One intelligence.
For battery energy storage.
We operate your battery inside its own chemistry. Every dispatch is priced against the battery itself - the schedule earns today, without spending the asset's tomorrow.
Energy in motion with intelligence. That is Storpeak. An energy management company, redefining how energy is run: renewables lead, storage carries, and intelligence sits at the centre.
01 · Forecast
Markets
Day-ahead, intraday, balancing and capacity. One live tape.
Cycle count, depth of discharge and calendar age honoured in writing.
07 · Protect
Reports
P&L per battery, per market, per cycle. Plain language, on cadence.
08 · Protect
Control
Algorithms run. Operators watch. Owners decide.
A buffer for a grid that moves.
A battery is a dynamic energy buffer. It absorbs when the grid is long, releases when the grid is short, and steadies the frequency in between. Run with intelligence, it does all of this without borrowing from the battery's own future.
Built for the asset owner.
Whether you operate a single site or a growing portfolio, Storpeak runs the battery as if it were our own. You keep the upside. The battery keeps its lifetime. The grid keeps its rhythm.
End-to-end stewardship.
With Storpeak's energy management platform, grid-scale batteries are not only traded to maximise asset value, they are dispatched inside the battery's own chemistry and kept within the warranty clause in the same motion. Taking an end-to-end approach to storage, managing forecasting, bidding, dispatching, risk and reporting all in-house, Storpeak is built to be trusted by battery owners and grid operators alike.
Shape your energy assets. Our models do the rest.
Two models, live. One reads the day, the other reads the battery.
Battery model
Ambient temperature25°C
5°45°
Depth of discharge80%
20%100%
C-rate0.5 C
0.2 C2.0 C
Cycles per day1.0
0.53.0
Energy market forecast
Solar share25%
080%
Wind share20%
060%
Volatility30%
lowhigh
Battery cycles1.0
0.52.5
Our tech.
With our energy management platform, we turn live market signals into trades that respect the battery's own chemistry. By fusing machine learning with price, weather, grid and battery data, our technology not only raises the revenue each asset earns, it protects its lifetime and gives owners a clear view of what the battery actually returns, redefining how storage is operated in a decarbonising grid.
Energy Market
Spain & Europe
Battery Knowledge
Chemistry & economics
Electricity is a strange product: you can’t store it easily at the scale of a country, and it has to be produced the instant it is used. So somewhere, someone has to match supply and demand minute by minute. In Spain and Portugal, the place where most of that matching happens is called OMIE. It runs two kinds of auctions - the day-ahead market and the intraday market - and together they set the price of almost every megawatt-hour consumed on the peninsula. For a battery, understanding how OMIE works is the first step to understanding how it earns money.
The day-ahead auction
The day-ahead market is a pay-as-cleared, uniform-price auction for the following delivery day. Generators, retailers and cross-border traders submit buy and sell curves, and OMIE matches them against each other for every delivery slot of the next day. The clearing price is the intersection of aggregated supply and demand in each slot. Until 1 October 2025, slots were hourly - a single price for each of the 24 hours. From 1 October 2025 onwards, the day-ahead market clears in quarter-hourly resolution, one of the biggest structural changes the Iberian market has seen in a decade.
A battery dispatching against the day-ahead curve no longer has 24 prices to optimise against; it has 96. Morning ramps, midday solar troughs and evening peaks all now appear at 15-minute resolution, which tightens the price spreads visible in the curve and, in principle, makes a fast-responding asset more valuable.
Intraday auctions and the continuous market
OMIE's second layer is the intraday market, which lets participants adjust positions after the day-ahead has cleared. Since 13 June 2024, Iberia participates in the three pan-European intraday auctions (IDAs) that run across the SIDC (Single Intraday Coupling) area, alongside the continuous cross-border XBID order book. The IDAs clear at fixed gate times each day; the continuous market runs between gates and closes a short time before delivery.
Intraday is where most short-notice corrections happen: a solar plant revising its output forecast, a wind farm adjusting to a weather update, or a battery repositioning itself around a price spread it did not capture in the day-ahead. Since March 2025, the intraday continuous market and its auctions trade in quarter-hourly products, and day-ahead followed in October 2025. Intraday spreads relative to day-ahead have been reported at ±€50/MWh in Spain during late 2025, confirming that arbitrage opportunities exist; liquidity remains lower than in the UK or German continuous markets, which is a practical constraint for high-frequency algorithmic strategies.
Balancing and ancillary services
Energy markets (day-ahead and intraday) are only part of the picture. After gate closure, Red Eléctrica de España (REE), as Iberian TSO, operates balancing and ancillary mechanisms - secondary and tertiary reserves, congestion management, voltage and frequency control. Batteries are increasingly eligible for these services, and the reform path Spain has taken since 2024 is pointed at opening more of these mechanisms to storage. For most standalone BESS projects today, the economic anchor is still energy arbitrage in day-ahead and intraday, with ancillary revenues layered on top as the regulatory framework opens up.
What it means for a battery
A two-hour battery in Spain now dispatches against 96 price blocks per day across three temporally overlapping venues - day-ahead, intraday auctions, and intraday continuous - each with different liquidity, different closing times and different strategic purposes. The day-ahead is where most capacity is still booked. The intraday auctions are where forecast corrections land. The continuous market is where algorithmic, short-notice trading captures residual spreads. Optimising across all three simultaneously, while respecting the battery's own state of charge and degradation cost, is the core operational problem for any Iberian BESS.
The quarter-hourly transition in 2025 is structural. It raises the value of fast, accurate forecasting and of low-latency execution, and it pushes the industry closer to the operational tempo of the ancillary markets. Expect spread structure, capture prices, and the value of flexibility to continue shifting as the market settles into the new tempo through 2026.
At 12:33 on 28 April 2025, the Iberian synchronous grid separated from continental Europe and collapsed inside a handful of seconds. Roughly 60% of Spanish demand went dark, and the lights stayed off across most of Spain and Portugal for around ten hours. It was the largest European blackout in more than two decades. Six months later, Spain published Royal Decree 997/2025 - a package of urgent measures to strengthen the electricity system and, in the same stroke, to accelerate the build-out of the storage capacity the grid was visibly short of that afternoon. For BESS developers, this is the most consequential regulatory text of the cycle.
The blackout in one paragraph
The immediate trigger was a voltage excursion the grid could not absorb; two large fluctuations led Spain to disconnect from the synchronous European grid and the Iberian system then collapsed. Independent post-incident reviews, including one by the IIT at Comillas in September 2025, pointed to insufficient synchronous generation providing dynamic voltage control, combined with limited interconnection with the rest of Europe - Spain has roughly 2–5% interconnection with neighbouring countries versus the EU's 10–15% recommendation. The blackout is not usefully described as "renewables caused it"; it is more accurately described as a system that had run ahead of its own inertia, voltage-control and flexibility services.
What RD 997/2025 actually does
The decree accelerates what Spain already had in motion and removes specific bottlenecks that had been slowing storage projects.
Environmental permitting
Permitting timelines for co-located battery storage are halved. Battery modules installed inside the boundary of a plant that already holds a favourable Environmental Impact Statement are exempt from the simplified environmental review. Projects are declared urgent and of public interest, which carries through the administrative chain to building permits.
Hybridisation
The decree prioritises hybridisation of storage with existing generation plants and simplifies authorisation procedures for such projects. For a wind or solar plant already connected to the grid, the economic case for adding a battery inside the existing connection envelope has become materially easier.
Access capacity
CNMC, the sector regulator, is tasked with publishing updated monthly access-capacity maps, short-circuit limits and related technical parameters from February 2026. In a country where access capacity has been the structural bottleneck for new projects, moving to transparent monthly refreshes is a material operational change.
Targets and funding
The decree re-anchors Spain's energy storage target at 22.5 GW by 2030, up from an earlier 20 GW in the draft NECP. This is aligned with MITECO's updated PNIEC 2023–2030 and the 76 GW solar PV target for the same horizon. In parallel, Spain has allocated roughly €840 million in combined MITECO and EU funds to storage projects, including around €699 million aimed at deploying up to 3.5 GW of new capacity.
Why it matters for developers
Three consequences are visible in the decree text and in early post-publication commentary. First, timelines compress: for qualifying co-located projects, the administrative tail that historically ran 18–24 months can be shortened meaningfully. Second, the universe of eligible sites expands, because hybridisation within an existing connection becomes a lower-friction route than greenfield development. Third, the access-capacity refresh changes how pipelines are valued - a monthly public map compresses the information asymmetry that previously favoured developers with strong TSO relationships.
None of this fixes the underlying dynamic voltage and inertia problem that caused the blackout. That is a separate programme of work for REE and the European synchronous area, and it is likely to ripple into ancillary-service design over 2026–2027. What RD 997/2025 does is remove the most obvious reason for storage projects to stall between signing and commissioning. The market reaction so far - EY's tracking has Spain at roughly 29% of the global BESS project pipeline through 2030 - suggests developers have read it the same way.
On a sunny May afternoon in Spain, the country often produces more electricity than it can use. The wholesale price then drops to zero - or goes negative, which means producers have to pay to keep generating. Hours later, when the sun sets and demand peaks, the same grid has to fire up gas plants to meet the evening load. That daily whiplash has two visible symptoms: curtailment (energy that could have been produced but was paid not to be) and negative prices (energy that was produced and was paid to stop). Both are expensive system problems. Both are also exactly the arbitrage window that batteries are being built to close.
Negative and zero-price hours are no longer rare
In 2024, Spain recorded 196 hours of negative day-ahead prices and 527 hours at zero - together, almost 10% of all traded hours. By the first week of September 2025, negative-price hours alone had already doubled the full 2024 total, passing the 500 mark for the year. The average price in those hours deepened from roughly -€0.12/MWh in 2024 to -€1.80/MWh in 2025, meaning not just more frequent negative prices but also more punishing ones.
Wind, hydro and nuclear assets remained economic on average because they capture higher prices at other hours. Solar, which by definition produces almost entirely inside the zero/negative cluster, had a much harder year. OMIE-reported solar capture prices for 2025 have been tracked at around €34/MWh against a wind capture of roughly €62/MWh - nearly a 2× gap.
Curtailment is rising and will continue to rise
Curtailment - generation that is instructed or economically forced off - ran at around 2% of Spanish PV output in 2024. Research tracking the last two years together puts average PV curtailment at roughly 2.9%, with 2.5 percentage points receiving no compensation. Forward-looking analyses from Strategic Energy Europe and others project curtailment reaching 5% by 2027–2028 if the solar pipeline is built without parallel flexibility. In May 2025, pv magazine reported that 21% of PV energy offered that month did not clear, even with bids below €5/MWh - a volume constraint driven by lack of demand, not by grid limits.
Where BESS comes in
The arbitrage window is simple to describe. A battery charges during the zero/negative cluster between roughly 11:00 and 16:00 local, and discharges into the evening peak between roughly 19:00 and 22:00. Intraday opportunities layer on top. What matters for project economics is not the average spread but the distribution - a handful of very high-spread days per month can dominate the annual capture.
Quarter-hourly clearing in day-ahead, which took effect on 1 October 2025, has made that distribution more granular. Short, sharp 15-minute swings around the shoulders of the evening ramp now show up in the price curve instead of being averaged into an hourly block. A battery with accurate price forecasting and fast execution captures more of those 15-minute swings than one optimising against the old 24-block curve.
A note on PPAs
The spread between zero-floor and no-floor PPAs has widened sharply. Solar off-takers that agreed to pay a merchant-linked tariff in 2022–2023 are increasingly exposed to curtailment and negative-price risk; developers signing new PPAs in 2025–2026 are negotiating zero-floor clauses and co-located storage much more aggressively.
Where this goes next
Two dynamics matter for the next two years. On the supply side, another 10–15 GW of PV is expected to interconnect in Spain by 2027, deepening the midday trough before storage catches up. On the demand side, electrification of heat and transport, and the first batches of contracted industrial demand response, will begin to fill the trough. The gap between those two curves defines the headroom for flexibility, and by extension for battery revenue, through the late 2020s. Every published curtailment number is, in effect, a leading indicator of how much flexibility the system will reward.
A battery project, like a power plant, has to pay for itself. How it does that - which party earns what, and who carries the risk when electricity prices move - is set by its revenue contract. Across Europe, BESS contracts cluster into three archetypes: fully merchant, floored merchant, and tolling. They differ in who bears market risk, volume risk and operational risk - and therefore in what a project looks like to a lender. Through 2024–2025 the bankability hierarchy has become clearer; the mix of deals actually signed has shifted with it.
Fully merchant
In a merchant structure, the project owner (or its optimiser) takes all risks. Revenue is whatever the asset captures in day-ahead, intraday and balancing markets, net of imbalance costs. Upside is the highest of any structure - spreads are volatile and a well-run asset in a deep-spread market can over-earn materially. Downside is correspondingly high, and because there is no floor, projects are harder to finance with bank debt.
Germany's standalone BESS market is archetypal merchant: volatile, high-upside at peak arbitrage, but exposed to grid fees, construction taxes and planning friction that compress returns. Spain is merchant-first today for a different reason - tolling liquidity is still thin and most of the early 2025–2026 cohort is being built with equity plus partial project finance against a modelled merchant stack.
Floor
A floor contract blends merchant with a minimum guaranteed revenue per MW or per MWh from an off-taker. If the market pays better than the floor, the owner captures the upside (sometimes shared with the off-taker); if it pays worse, the off-taker pays the difference. In exchange, the off-taker typically receives a share of the upside above a strike.
Floors emerged as the compromise structure that made UK projects financeable through the recent price-volatility cycle. The UK is now Europe's most mature BESS market, with typical transactions at around 70% leverage and unlevered IRRs around 12% against a weighted average cost of capital in the ~5% range, according to 2025 financing surveys. Floors do not eliminate market risk; they cap the downside at a level the lender can underwrite.
Tolling
In a tolling structure, an off-taker (typically an optimiser, utility or trading house) pays the owner a fixed fee per MW and/or per MWh for the use of the asset, and takes the dispatch decisions and the market risk. The owner becomes much closer to an infrastructure investor - revenue is a contracted annuity for a period of years, with limited exposure to spread volatility.
Tolling has been the fastest-growing revenue structure in European BESS. Standalone tolling deal counts in Europe rose from 3 in 2024 to 15 in 2025, with roughly 6 of those 15 in Germany. European buyers contracted close to 24 GWh of BESS under flexibility purchase agreements across 2025. Many banks now require at least 50% of project revenues to be secured through a bankable toll before they will lend at infrastructure terms.
Ranked by bankability
From a lender's perspective the order is simple: toll > floor > merchant. Toll gives contracted cashflows against a known counterparty; floor gives a downside-capped envelope with upside participation; merchant gives full exposure to spread. Lender LTV and debt-service coverage requirements follow the same ordering.
That said, the highest equity IRRs in Europe 2023–2025 came from disciplined merchant operation in the UK and Germany, not from tolling. Tolling trades away upside for stability. Which trade makes sense depends on the investor's cost of capital, the stage of the market, and how much of the asset's expected life is front-loaded into the merchant window before competing capacity drives spreads down.
Why early Spanish projects are merchant-first
Three reasons. The market is young, so there are few repeat optimiser counterparties willing to offer five- or ten-year tolls at prices an owner will accept. The structural spread (curtailment, negative prices, sparse evening flexibility) is unusually wide for a Western European market today, so the expected merchant capture is high. And the price-discovery work has simply not happened yet - until a pipeline of Spanish tolls clears, nobody knows what an on-market Iberian toll rate looks like. Expect the first wave of Spanish tolls and floors to land through 2026–2027, with pricing anchored to observed merchant capture in 2025 and 2026.
Not all lithium batteries are the same. The two that matter for grid-scale storage are called LFP and NMC, and they behave very differently in the real world. LFP is the chemistry now in almost every new utility-scale battery being built in Europe. NMC is the chemistry in most electric cars. Both are lithium-ion, but the cathode materials are different, and that one design choice drives very different answers on safety, lifetime and cost. The question of which one wins for grid storage was open a decade ago. Today it is effectively closed.
What the chemistries actually are
LFP uses a LiFePO₄ cathode with a graphite anode. The olivine structure of the cathode is thermally stable up to around 270°C before it decomposes. NMC (lithium nickel manganese cobalt oxide) uses a layered oxide cathode that is denser in energy but decomposes into an exothermic runaway pathway between roughly 150°C and 200°C. In shorthand: LFP gives up energy density and gains safety; NMC does the opposite.
Safety is the decisive factor
For a residential EV battery pack, the trade is marginal - both chemistries ship in production cars. For a 100 MWh container yard that sits next to a solar farm or a substation for 15 years, it is not marginal. LFP's higher decomposition temperature, its lower flammable-gas release in the event of a cell failure (reported at roughly 80% less than NMC in comparable tests), and its lack of a strong oxidiser in the cathode all translate into simpler, cheaper fire-suppression, simpler planning approvals, and lower insurance premiums. Several European fire codes - and, downstream, most utility procurement frameworks - explicitly favour LFP for large, stationary systems.
Cost has converged in LFP's favour
NMC was long the cheaper-per-kWh option, but the gap has closed. LFP cell prices dropped faster than any other chemistry through 2023–2025 as Chinese manufacturing capacity scaled on a cathode material with no cobalt, no nickel, and a far simpler supply chain. Recent market pricing from BloombergNEF and others has LFP at effectively parity or better with NMC on a cell-level $/kWh basis for stationary systems. When cycle life is factored in - manufacturer specs for modern large-format LFP run 6,000+ cycles at 0.5C, 25°C under compression, against ~3,000–4,000 typical for NMC in grid-scale applications - LFP wins on levelised cost of storage (LCOS) over a ten-year horizon by roughly 10–15% in most published models.
Cycle life and calendar life
LFP is a more cycle-tolerant chemistry. It is not immortal - calendar aging still eats into capacity and a poorly-managed LFP bank can still lose 20%+ in five years - but the headline datasheet numbers are real. The landmark 2025 ScienceDirect study on 180 Ah prismatic LFP cells, tested over 1,500 cycles and 850 days of calendar aging at 35°C and 50°C, confirms the expected pattern: low-rate, moderate-temperature operation is where LFP earns its lifetime, and where the economic case over NMC is strongest.
Energy density is a real constraint - for someone else
NMC still wins on energy density, which is why passenger EVs still ship NMC and NCA variants. On a 15 MW AC/60 MWh LFP container yard the site is one or two additional rows of containers; nobody cares. On a 75 kWh passenger-car pack at 1.8 tonnes kerb weight, every Wh/kg is fought for. Stationary storage simply does not live on the same density constraint.
Regional factors that push LFP even harder in Southern Europe
Three local factors compound the European norm. First, insurers and permitting authorities across Southern Europe have been conservative on battery fire risk after several high-profile warehouse fires over 2022–2024; the simpler LFP failure envelope makes permitting and insurance meaningfully easier. Second, most BESS projects in Iberia and Italy are co-located with solar, where the economic case is driven by levelised cost of storage across a long horizon rather than peak power density. Third, the 2–4 hour duration that dominates the Southern European pipeline is exactly the sweet spot where LFP’s cycle life advantage over NMC pays off most clearly.
The question for the next cycle is not NMC versus LFP. It is LFP versus newer chemistries - sodium-ion for grid duty (now approaching $59/kWh at pack level per recent BloombergNEF tracking), lithium manganese iron phosphate (LFMP) for higher energy density at LFP-like safety, and solid-state for specialist applications. For now, across the European utility stack, LFP is the default, and the default is the right answer.
A battery ages in two ways at once. Every time you use it, it ages a little. And even when you don’t use it, it still ages, just from sitting there. Those two effects - cycle aging and calendar aging - have different drivers, and on a grid-scale asset they don’t add up cleanly. Which one dominates depends entirely on how the battery is operated and where it sits. Getting that split wrong is the single most common mistake in BESS project-life models.
Cycle aging - the use-related wear
Cycle aging is the capacity you lose every time the battery is charged and discharged. It is driven by physical and chemical wear inside the cell: tiny amounts of lithium getting stuck on the anode when you charge too fast or too cold, a protective layer on the electrode that keeps growing with use, and slow mechanical fatigue of the electrode materials at deep discharges.
The practical takeaways are simple. Running at full power (1C) ages the cell faster than running at half-power (0.5C). Discharging to 20% remaining is harder on the cell than stopping at 50%. And charging below roughly 15°C cell temperature is not just a bit worse - it crosses into a different regime where lithium starts to plate out on the anode metal-side, and that damage does not reverse. NREL’s BLAST-Lite framework and industry calibrations agree on the direction even where they disagree on the numbers: lower power, shallower cycles and moderate temperatures buy cycle life.
Calendar aging - the passage of time
Calendar aging happens whether the battery is cycling or not. The electrolyte slowly decomposes, parasitic side reactions continue at the electrode surfaces, and the protective SEI layer keeps thickening. Two factors drive it far more than anything else: how warm the cell is, and how full it is on average (the average state-of-charge, SoC).
The Dalhousie group’s high-precision measurements on LFP/graphite cells showed that keeping the average SoC low extends lifetime measurably. A cell that sits most of its life in a 0–25% SoC window ages slower than the same cell in a 75–100% window, even at the same depth of discharge. Sitting full is harder on a lithium battery than sitting empty - the anode is at a more aggressive potential when it is fully charged.
Why two batteries age differently
Two packs with identical specs go into service at the same solar farm. One sits full most of the day waiting for the evening peak; the other is cycled shallow twice a day. At face value the second pack does more cycles and should age faster. In practice, in a hot climate, the first pack often degrades more because calendar aging at high SoC and high cell temperature dominates its trajectory. A 2025 ScienceDirect study on 180 Ah LFP cells running at 35°C and 50°C across multiple SoC conditions found capacity loss at 50°C exceeded 35°C across every test condition - cycle aging did not save the high-temperature pack.
The German utility-scale BESS thermal study is another useful anchor. In a 7 MWh frequency-regulation system, the difference between the floor (avg 23°C) and top rack (avg 32°C) reached nearly 1 percentage point of capacity per year - inside a single container. That is not cycling doing the work; that is Arrhenius.
The knee point
Both degradation modes are compounded by a non-linear end-of-life effect sometimes called the knee. Attia et al. (Nature, 2022) demonstrated that batteries often shift from a near-linear capacity fade into a rapid acceleration after reaching roughly 78–82% state-of-health, driven by a feedback loop where reduced lithium inventory increases local stress, which consumes more lithium, which accelerates further. Project lifetime models that assume the linear section extrapolates cleanly to 60% SoH understate late-life risk; lenders have started demanding knee-point stress tests explicitly.
Practical implication
A useful rule of thumb from the industry-level data: under typical Spanish ambient conditions (10–30°C cell temperature, LFP chemistry, 2–4 hour duration, 60–80% DoD, 0.5C average rate), calendar aging and cycle aging contribute roughly comparably over a 15-year horizon. Push C-rate toward 1C, DoD toward 90%, or cell temperature above 35°C, and cycle aging dominates. Idle a high-SoC pack in hot weather and calendar aging dominates. There is no single number for "battery life" - there is a response surface, and the duty cycle picks the point on it.
There’s an old rule of thumb in battery engineering: if you want the battery to last longer, don’t drain it as deeply each cycle. It is correct, but only half the picture. Where the battery sits on the charge scale - not just how far it swings - matters just as much for lifetime. On a grid-scale asset, the right question isn’t “how many cycles can this battery take?” but “which operating pattern earns the most money after paying for the wear it causes?”
What the traditional curve actually says
The textbook plot shows cycle counts climbing from a few thousand at 100% depth-of-discharge (DoD) to tens of thousands at 20%. For a modern LFP cell tested at 0.5C and 25°C, manufacturer curves typically show something like 6,000 equivalent full cycles to 80% state-of-health at 100% DoD, 15,000+ at 50%, and 30,000+ at 20%. Academic studies on large-format prismatic LFP cells (Taylor & Francis 2024) confirm the shape. Shallow cycles really do extend life - the curve is real.
Depth of discharge and SoC window - not the same thing
DoD is how much energy is drawn per cycle. The SoC window is where along the charge scale that cycle sits. A cell that cycles between 10% and 90% charge and a cell that cycles between 0% and 80% charge both have 80% DoD, but the first averages 50% state-of-charge and the second averages 40%. That difference matters.
The Dalhousie / Dahn Lab study on LFP/graphite pouch cells measured it directly with ultra-high-precision coulometry: a 0–25% SoC window beats a 75–100% window on lifetime, even at similar DoD. Lithium inventory loss and iron dissolution accelerate when the cell sits full. The familiar advice “keep DoD low to extend life” is incomplete. The better version is: keep average SoC moderate, and use only the width of the window you actually need.
Why the right window depends on the revenue stack
A battery that earns almost all its revenue from one deep evening-peak spread per day - the pattern across most continental European markets today - only needs about 80% DoD once a day. Cycling 10–90% captures close to all the available spread and significantly reduces degradation versus 0–100%. The opportunity cost of the extra 10% on each end is tiny spreads the optimiser almost never captures profitably after degradation cost.
A battery earning from ancillary services such as secondary reserve needs headroom in both directions - it can neither sit at 100% nor at 0% if it is contracted to respond symmetrically. Typical operating windows are 20–80%, often tighter. A battery doing dual-use (arbitrage plus frequency response) might operate 15–85% on arbitrage windows and tighten to 30–70% during ancillary windows.
A battery running in a high-spread, volatile market where the optimiser repeatedly sees multiple high-value windows per day - GB’s Balancing Mechanism is the standard example - may rationally run 0–100% on the days it pays, and lean on the faster degradation as a cost of doing business, because the uplift from capturing every spread swamps the extra wear.
Cost the window, don’t just choose it
The disciplined way to pick a window is to attach a cost to degradation, in €/MWh, and let the operating optimiser route around it. Ask: how much extra capacity loss does the next 5% of DoD buy, or the next 10°C of cell temperature? That cost curve turns the SoC window from an engineering choice made once into a revenue-optimisation output that shifts day by day. DNV’s 2024 Battery Scorecard is the cleanest public benchmark: operating fleets span a wide range of observed degradation, with assets that run tighter windows and better thermal management clustering at the low end and assets that don’t drifting measurably above it.
One final nuance
Manufacturer warranties typically specify a DoD band and an average temperature. Cycling outside the band does not just hurt lifetime - it voids the warranty. Before any window decision, the warranty text is the binding constraint.
If you put 100 units of electricity into a battery and take 85 back out, its round-trip efficiency is 85%. Simple on paper. On a real grid-scale system, that one number hides several underlying efficiencies, a temperature dependence, and a slow decline as the battery ages. A single percentage point up or down is worth real money over a year of dispatch - and the number on the EPC contract is rarely the number the asset actually delivers.
DC round-trip vs AC round-trip
The cleanest measure is DC round-trip efficiency: energy out of the battery terminals divided by energy into them, at the same SoC endpoints. Modern LFP cells run DC RTE around 96–98% at low C-rate and optimal temperature.
AC round-trip efficiency is the number that matters economically. It takes DC RTE and adds the losses from the power conversion system (PCS) and any transformers between the battery and the grid meter. Well-engineered modern PCS equipment runs around 97–98% one-way at partial load. Multiply these chains together and you arrive at a theoretical AC RTE in the low 90s.
Full-system AC RTE - the number the meter reports, including station auxiliaries - is typically 85–88% for grid-scale LFP systems in operation. ACCURE's 2025 Energy Storage Report, which analysed over 100 BESS sites representing 18 GWh globally, found best-in-class systems above 88% with roughly one-third of the fleet clearing that bar.
Parasitic loads are material
HVAC is the largest single parasitic line on a BESS, and it is almost never included in the datasheet RTE. A container yard in a hot climate running compressor-based cooling can spend 2–4% of throughput on HVAC alone. In a cold climate, heating is smaller but non-zero. Temperate European climates - the Iberian interior, southern France, the Po Valley - sit in the best-case window of 1–2%. Control systems, fire-suppression standby, lighting and communications add a further 0.3–0.7%. None of this shows up in the cell-level 96% DC RTE number.
Academic work - including an IEEE virtual-battery paper on HVAC impacts - has quantified the parasitic load as a function of setpoint and outside temperature. The most energy-efficient cell temperature band for LFP operation is roughly 22–25°C; below or above that band, both degradation and HVAC-driven parasitic load rise.
RTE drifts with SoH and temperature
As a cell ages, its internal resistance grows. Ohmic loss - the I²R term - scales with the square of current and with resistance, so a 10% increase in internal R at a given C-rate is a proportional efficiency hit. Over 8–10 years, DC RTE can drift 1–2 percentage points just from resistance growth, and AC RTE tracks it.
Temperature compounds. Cold cells have higher internal resistance; hot cells need more HVAC work to stabilise. Both directions erode RTE. Industry HVAC/battery studies put optimum energy efficiency in the 21.8–25.2°C range, with measurable losses either side.
The revenue impact of one point
Consider a 100 MWh system doing one full cycle per day at a €80/MWh average capture spread and operating 350 days of the year. A one-point improvement in AC RTE - from 85% to 86% - delivers roughly one additional MWh of net throughput per cycle. Across a year, that is 350 MWh × €80 = €28,000 of additional gross margin, per point, per year.
Scale that over a 15-year operating horizon, and the present value of a sustained one-point RTE improvement on a 100 MWh system is a six-figure number. On a 500 MWh portfolio it is materially larger. This is why operational telemetry matters: an asset whose HVAC runs at the wrong setpoint, or whose PCS firmware is not tuned for partial-load efficiency, is leaking real revenue every day it operates off-optimum.
What to look at
A disciplined operator tracks three numbers continuously: DC RTE at the battery terminals, AC RTE at the POI meter, and auxiliary load as a fraction of throughput. The three of them together tell you whether degradation, PCS tuning, or HVAC is the current leak. ACCURE's report also notes that only around 83% of projects surveyed met their nameplate capacity at the site acceptance test - meaning RTE drift is typically layered on top of an under-delivered baseline. Measure early, measure often.
A capacity market pays power plants and batteries simply for being ready, not for the electricity they actually produce. Think of it like paying a firefighter to stand by on shift, even on days when there are no fires. Great Britain, Ireland, Italy and most of North America already do this. Spain does not - yet. In 2026, it will, for the first time, and under a framework built jointly with the EU’s Electricity Market Reform. For storage developers this is a structural change, not a pricing tweak: the auction is the first serious attempt to pay batteries for being present on the grid, rather than only for what they happen to dispatch. Designed well, it is the instrument that lifts merchant IRRs from the low single digits into the range institutional capital actually requires.
Why Spain needs a capacity market now
The updated PNIEC 2023–2030 places 22.5 GW of storage and 76 GW of solar PV on the system by 2030. The gap between the two numbers is where the flexibility problem lives. On a bright May afternoon, Spain in 2025 is already recording multiple hours of negative day-ahead prices; on a still January evening it still leans heavily on combined-cycle gas and grid imports. A capacity market is the mechanism used in Great Britain, Ireland and most of North America to pay resources for being available at the stress hour. Spain is now following the template with Iberian-specific design, under the EU's Electricity Market Reform Regulation 2024/1747.
What the decree sets up
Royal Decree-law 7/2025, continued by the broader RD 997/2025 package published after the 28 April 2025 blackout, gives MITECO the legal basis to run competitive capacity auctions. The framework permits contracts of up to 15 years for new-build assets, including standalone and hybrid BESS. Existing assets participate on shorter terms, typically 1–3 years. The auction is technology-neutral in principle, with de-rating factors applied by technology to reflect each resource’s contribution to system adequacy at scarcity hours. The first auction is targeted for 2026; volumes and de-rating factors are being defined through CNMC consultation during 2025–2026.
The revenue math
Published revenue-stacking studies on Iberian BESS (including 2024–2025 work from the IIT at Comillas and independent consultancies) model a standalone ~50 MW / 200 MWh Spanish asset across day-ahead arbitrage, aFRR capacity and aFRR energy, with and without a five-year capacity contract. Without capacity payments the modelled IRR lands in the low single digits; a capacity annuity in the low €10k–€13k/MW/year range, layered on top, typically lifts returns by about 4 percentage points on a 20-year horizon.
That threshold is a useful anchor. If the cleared capacity price lands meaningfully above €12–13k/MW/year on 15-year terms, the 4-hour BESS pipeline that has been accumulating permits in Spain since 2023 starts to clear financial hurdles at a very different pace than it does today.
What the de-rating factors decide
Capacity markets are won or lost in the de-rating methodology. A 100 MW / 400 MWh battery is not 100 MW of firm capacity at the 18:00 stress hour unless the state-of-charge trajectory across the preceding day permits it. GB uses duration-dependent de-rating, where a 2-hour system is credited materially lower than a 4-hour one. Ireland uses a stricter availability test. Spain’s draft CNMC methodology follows the GB template with an Iberian probabilistic run on REE’s scarcity hours. 2-hour BESS will likely de-rate into the 40–60% range; 4-hour into 70–85%; 6-hour close to 95%. Those numbers drive duration sizing decisions on every project still in pre-design.
Cannibalisation - the real risk
As BESS capacity scales, arbitrage spreads narrow. Capacity payments are supposed to compensate for the component of revenue that thins as competitors arrive. The open question is whether the cleared capacity price will track the true missing-money curve or lag it. In GB the capacity market has consistently cleared below the missing-money curve implied by fundamentals, which is why GB BESS projects still lean heavily on ancillary services and merchant arbitrage. Spain is likely to follow a similar pattern in 2026–2028, with capacity clearing lower than the modelled missing-money threshold in the early auctions while the market builds out de-rating credibility. Projects modelled in 2025 against a 15-year low-€10k/MW/year assumption should stress-test at 30–40% below that number.
What developers should watch in 2026
Three documents: the CNMC final methodology on de-rating factors; the MITECO auction design, including volume cap and clearing rule; and the first list of qualified bidders. Every number on a pro-forma model is sensitive to those three. A well-designed auction will materially accelerate the Spanish BESS pipeline. A poorly-designed one will leave the sector worse off than it is today, because investors will price in regulatory risk on top of merchant risk. Regulation 2024/1747 binds Spain to have a mechanism; it does not bind anyone to clear it at an efficient price.
When a wind or solar plant is far from the main demand centres, the grid itself can become the bottleneck. The electricity is cheap to produce, but the wires cannot physically carry it to where it is needed. In those moments the system operator has to tell cheap renewables to stop generating and pay more expensive gas plants, closer to demand, to ramp up instead. That service is called redispatch - in Spain, Technical Restrictions. And in Spain, the bill for it has exploded. In 2020 it cost the system €528 million. In 2024 it cost €2,523 million. That is a 40% compound annual increase, over four years, in a single line of system cost that almost nobody outside TSO operations talks about. Post-blackout, it accelerated again. The structural case for batteries in Spain is partly the day-ahead arbitrage window. It is also, very directly, this bill.
What the bill actually pays for
Technical Restrictions is the service REE calls when the economic market dispatch - day-ahead clearing plus intraday adjustments - does not respect some physical limit on the grid. The limit is typically a thermal constraint on a congested corridor, a voltage-control shortfall in a sub-area, or a synchronous-inertia shortfall during a high-renewables hour. To rebalance, REE instructs renewables to curtail inside the constrained zone and ramps up combined-cycle gas (CCGT) or cogeneration in neighbouring zones. Both legs cost money: the renewable operator is compensated for the energy it could have sold, and the thermal operator is paid above merchant to run. The difference between what the system paid and what the pure economic dispatch would have cost is Technical Restrictions.
The trajectory
REE’s published ESIOS settlement data show a near-exponential increase in the Technical Restrictions component of the final electricity price. 2020 closed at roughly €528 million, averaging €1.4 million per day. 2021 was €851 million. 2022 reached €1,390 million, driven partly by exceptional gas prices but also by a step-change in congested hours. 2023 hit €1,940 million and 2024 closed at €2,523 million - roughly €6.9 million per day on average. In the months following the 28 April 2025 blackout, average daily redispatch spend climbed to roughly €15 million per day as REE tightened its operating envelope. In aggregate, adjustment services added €11.43/MWh to the final Spanish electricity price in 2024 - an uplift on the order of 11% over a typical wholesale clearing price.
Why the number is growing
Three forces compound. First, the renewable pipeline has grown faster than the grid-reinforcement pipeline - there is more renewable capacity asking for access to the same transmission corridors, so congestion hours rise non-linearly. Second, the retirement of legacy thermal plant and the closure of several nuclear units under the 2027–2035 phase-out reduces dispatchable capacity in exactly the zones where Technical Restrictions are most often called. That forces REE to pay a higher redispatch premium to secure the remaining thermal fleet. Third, post-blackout REE has tightened synchronous-inertia and dynamic-voltage requirements, which expands the set of hours where renewable-heavy dispatch cannot be accepted as-is.
Why BESS is the natural substitute
For most of the Technical Restrictions envelope, a battery does the same physical work as a redispatched CCGT: it injects or absorbs real power at a specific node at a specific minute. In grid-forming configurations it can provide synthetic inertia and voltage support as well. Unlike a CCGT it does not have a ramp constraint, does not require gas, does not emit CO2, and does not need to warm up. Grid-forming inverters - which MITECO’s €300/kWh bonus in the €700M scheme is explicitly designed to catalyse - close the last functional gap between batteries and synchronous machines on voltage-support services.
The economic substitution logic is direct. If a CCGT is being paid €150/MWh to run out-of-merit to lift a transmission constraint, a battery placed at the same node that can discharge into the constraint does the same job. The difference is that the battery then recharges in a low-price hour, whereas the CCGT burns gas. Over a year, the battery’s all-in cost of relieving the same constraint is materially lower. As BESS fleets scale, the merit of the Technical Restrictions stack shifts toward batteries.
What the number tells developers
The Technical Restrictions bill is the single clearest quantitative signal that Spain has a flexibility deficit, and that the TSO is currently paying for it in cash. MITECO is under explicit mandate under Royal Decree 997/2025 and the CNMC’s adequacy review to route some of that spend into new storage. For a developer siting a project, the practical question is no longer “will BESS be needed?” but “which substations sit inside REE’s most expensive Technical Restriction zones in 2024–2025?” That is a public data layer (ESIOS zone-level settlements), and it is a much better first filter than the grid-access queue alone.
A battery wears out. The phone in your pocket probably holds less charge than it did when you bought it, and grid-scale batteries behave the same way - just over 15 or 20 years instead of two or three. A 100 MWh battery installed in 2026 does not stay 100 MWh. By year 8 it is typically closer to 80 MWh; by year 15, often 65 MWh. If the revenue contract - or the capacity-market commitment - is written against nameplate energy, that degradation is a direct cashflow problem, not an engineering curiosity. Augmentation is the industry’s term for the plan to keep the battery at nameplate over its contracted life. Getting that plan right on day one is worth more than almost any other operational decision.
Why capacity fades faster than people assume
Modern LFP cells under typical Spanish ambient conditions lose roughly 2% of usable capacity per year in the first two or three years, slowing to 1.2–1.5% per year through mid-life, then re-accelerating toward end-of-life due to the knee-point effect documented by Attia et al. (Nature, 2022). NREL’s BLAST-Lite framework calibrates similar trajectories for 2–4 hour utility duty cycles. Applied to a 100 MWh system over 15 years, cumulative loss typically runs 30–40% - which means that to hold 100 MWh of usable nameplate, the physical cell inventory must grow to roughly 140–160 MWh over the life of the asset.
Three augmentation archetypes
Oversize on day one
Install 115–120% of nameplate on day one, run the asset on a narrower SoC window for the first few years, then let it drift to a full window as capacity fades. Simplest to build, but it carries the largest upfront capex and forgoes the ongoing cost-reduction curve on future cell purchases.
Augment in blocks
Install nameplate on day one, then schedule one or two augmentation events at years 5–7 and again at 10–12. Each event adds physical racks - typically 20–30% of the initial installed energy - in new cabinets or dedicated new containers. This is the mainstream approach in GB and is the one most Spanish EPCs are now building into 2026–2028 contracts.
Container swap
At a mid-life milestone - typically year 8–10 - swap out under-performing containers and replace with new, higher-density ones. More operationally disruptive, but the simplest from an accounting standpoint. Each refresh is a clean capex event with its own cycle-life warranty on modern cells.
Warranty stacking is the hidden constraint
Cell vintages drift on chemistry, firmware and physical format. An LFP cell made in 2026 is not electrically identical to an LFP cell made in 2033, and the 2033 vintage will typically have different specific energy, different internal-resistance curves, and different safety envelopes. Operating a mixed-vintage rack without partitioning creates warranty and balancing complications: the weakest vintage dictates the string voltage cutoff, and the newer cells end up under-utilised. The disciplined approach is physical segregation - new augmentation batteries sit in their own containers, on their own DC bus, and report into the EMS as a separate resource, with their own warranty envelope from their own OEM. The EMS dispatches both resources but does not mix their cells.
Capacity guarantee vs throughput guarantee
OEM warranties on grid-scale BESS now come in two families. Capacity guarantees state that the system will deliver at least X% of nameplate energy at year Y - typically 70% at year 15 for LFP. Throughput guarantees state that the system can cycle at least Y MWh over its life, typically 15,000–25,000 full-equivalent cycles. Capacity guarantees pair naturally with augmentation-based models because the OEM is on the hook for delivering the augmentation. Throughput guarantees pair naturally with overbuild models because the owner manages the trajectory and the OEM is only on the hook if the cell fails early. Bankable project-finance structures in Europe have converged toward capacity guarantees with OEM-provided augmentation as a contracted service - the lender gets a known installed-capacity trajectory without underwriting the owner’s forecast of future cell prices.
Site planning for augmentation
A project planned with augmentation must reserve four things on day one. Physical footprint: enough pad area for 25–30% additional containers at year 7 and another 15–20% at year 12. Many early European BESS projects were built against a peak-day footprint and cannot accept augmentation without procuring adjacent land. MV infrastructure headroom: transformers, MV cabling and switchgear sized for the final augmented system, not just day-one capacity. Retrofitting transformer capacity is far more expensive than specifying it up front. Interconnection: the grid-connection agreement must permit dispatch of the augmented system. In Spain this is material because CNMC’s access-capacity methodology treats AC-side capacity as the binding limit. And finally EMS / HVAC headroom: both scale with container count and need to accommodate the largest planned footprint from the outset.
Cost trajectory - the reason augmentation pays
Utility-scale lithium-ion pack prices fell from roughly $280/kWh in 2019 to $115/kWh in 2024 (BloombergNEF), and NREL projects further declines into the $80–90/kWh range through 2030. A 25% augmentation purchased in 2032 will cost significantly less than the same 25% purchased in 2026, and that cost differential is the financial case for the augment-in-blocks approach over the overbuild approach. The overbuild penalty is non-trivial: paying for 20% of capacity six years early, at a higher unit cost, on a discounted-cashflow basis, is typically worth 1–2 percentage points of project IRR.
A note on Southern European projects
Iberian and Italian projects sit at a useful point on the curve. Ambient temperatures are moderate by global standards (lower calendar aging than in the US Southwest or India). The mainstream duty cycle - 2–4 hour arbitrage plus secondary reserve - gives LFP a gentle operating profile. And the first wave of large-scale projects is only now being built; the first augmentation events will not arrive until 2031–2033, by which point cell prices will be materially lower. The design choice most Southern European developers face in 2026 is not whether to augment but how much physical and electrical headroom to reserve for the augmentation they will do later.
Ask a non-specialist what size a battery is, and they usually say how much energy it holds (kilowatt-hours). Ask an engineer, and they give you two numbers: how much energy it holds, and how fast it can deliver it. A 100 MW / 200 MWh battery is a “2-hour” system - running flat out, it takes two hours to empty. A 100 MW / 400 MWh system is a “4-hour”. That single ratio - duration - is the most consequential sizing decision on any grid-scale battery. And across Europe, it is being made against a moving target. Projects permitted in 2023 were mostly 1–2 hour. Projects entering detailed design in 2026 are mostly 2–4 hour. The reason is not fashion; it is the interaction of arbitrage spread shape, capacity-market de-rating, and a real shift in the ancillary-services stack. Duration is the one parameter that reads every piece of market design at once.
What duration means physically
Duration is the ratio between a battery’s usable energy (MWh) and its power rating (MW). A 100 MW / 100 MWh system is 1-hour; 100 MW / 200 MWh is 2-hour; 100 MW / 400 MWh is 4-hour. The power rating is set by the inverter (PCS) and the grid connection; the energy rating is set by the number of cells. Longer duration means more cells for the same PCS - more capex on batteries, the same capex on power electronics. That cost geometry shapes every duration decision.
Arbitrage favours shorter - until it doesn’t
The simplest arbitrage window is: charge at midday when solar is saturated and discharge into the evening peak. In a market with a single deep spread per day, a 1-hour battery captures a large share of that day’s value - cycling once, deeply, at the maximum spread. Adding a second hour of duration captures shoulder prices that are typically 20–30% lower than the peak; adding a third captures lower still. NREL’s Storage Futures study and the JRC’s duration analyses model the same pattern: marginal revenue per MWh of duration declines as duration lengthens, and the optimum depends on the daily spread distribution. In continental European markets with heavy midday negative pricing and a single sharp evening ramp - the Iberian and Italian pattern today - the pure-arbitrage optimum sits around 2–2.5 hours for most sites. Once multi-day spreads from weekends and weather-driven renewable lulls are included, the optimum widens toward 3 hours.
Frequency-regulation services like secondary reserve (aFRR) pay mostly for power, not energy. Capacity remuneration is typically quoted in €/MW/h regardless of whether the provider is a 1-hour or a 4-hour asset. A shorter-duration battery therefore has a lower capex-per-MW of aFRR capacity and competes more aggressively in that market. Activation (energy) remuneration adds variable revenue that favours duration - a longer battery can sustain a call for longer - but capacity is the larger line.
Capacity markets push in the opposite direction. Great Britain, Ireland, Italy’s MACSE, and now Spain all de-rate shorter-duration batteries for capacity-market participation. Indicatively, 2-hour systems de-rate into the 40–60% range, 4-hour into 70–85%, and 6-hour close to 95%. If a developer assumes a capacity-market annuity in the pro-forma, the 4-hour asset clears more of that annuity per MW of interconnection than the 2-hour.
Where the two curves cross
For each site, a developer can build two revenue curves. The “merchant + aFRR” curve typically peaks at 2–2.5 hours and declines after that. The “merchant + aFRR + capacity market” curve keeps rising and typically peaks at 3.5–4 hours, depending on the cleared capacity price. The right duration is where total stacked revenue minus duration-scaled capex is maximum. With capacity payments at an indicative €10–12k/MW/year 15-year annuity, and cell prices moving from ~$115/kWh today toward ~$90/kWh by 2030 (BloombergNEF), the crossover for most continental European sites sits in the 3–4 hour range. Below €8k/MW/year, the crossover retreats to 2.5–3 hours. Above €15k/MW/year, it pushes to 4.5–5. The lesson is that duration is a function of capacity-market clearing price, not a standalone engineering parameter.
What Great Britain tells us
The GB market is about a decade ahead of continental Europe on this trajectory. Early GB projects (2015–2020) were overwhelmingly 1-hour, optimised for frequency response. As the EFR and dynamic-containment stacks matured and capacity-market de-rating methodology settled, the mainstream shifted to 2-hour through 2022–2024. By 2025 the share of new-build 4-hour projects in GB had risen materially, with the first 6- and 8-hour projects entering planning. McKinsey’s 2025 BESS revenue analysis flagged the same directional shift across every mature European market: as ancillary services saturate and capacity markets take a larger share of the stack, durations lengthen.
Cost geometry still matters
The cost difference between 2-hour and 4-hour is not 2×. BoS (balance of system), PCS, transformer, interconnection, land, site works and permitting are largely fixed between the two sizes; only the battery cells and the HVAC / fire-suppression scale with energy. Unit-capex trajectories from Lazard’s 2024 LCOS analysis and NREL’s 2024 cost update put a 4-hour system at roughly 150–170% of the capex of the equivalent 2-hour system - not 200%. That is why 4-hour can out-economics 2-hour even when the 2-hour has higher revenue per MWh of energy.
What to lock in on day one
Three design decisions can be deferred cheaply; three cannot. The deferrable set is battery chemistry generation, cell supplier selection and augmentation schedule - all of these adjust as cells improve. The set that cannot be deferred is grid-connection capacity (which fixes the MW), site footprint (which caps total containers), and MV infrastructure sizing (which caps maximum final energy). A site designed for 2-hour today with no physical or electrical headroom cannot be re-sized to 4-hour later. Developers across Europe who expect capacity markets to pay well from 2026 are overwhelmingly specifying the civils for 4-hour even where the day-one installation is 2-hour.
Until recently, batteries on the European grid were a rounding error. That changed in 2024–2025. Large-scale batteries in Europe now total around 10 GW of installed power capacity - roughly three times the level of two years earlier - and credible sector forecasts put the number at 40 GW by 2030. The shift is no longer something the industry projects. It is something the transmission operators are already planning around.
What the numbers actually show
Ember’s European Electricity Review 2025 (covering 2024 data) records that wind and solar together produced 30% of EU electricity for the first time, narrowly edging past the 29.8% coming from the three main fossil fuels (coal, gas, oil). Coal fell to 9.2% of generation, the lowest on record. Gas briefly ticked up 8% year-on-year, mostly because 2024 was a dry year for hydro and demand rose, but the structural trajectory for fossil fuels is clearly down. Into that picture, large-scale batteries are growing rapidly: national pipelines are consistent with a 40 GW European total by 2030, up from around 10 GW at the start of 2025.
Rystad Energy’s 2026 Energy Storage report puts the same shift in global context. Cumulative global BESS capacity passed 250 GW in 2025, with 100 GW / 280 GWh added in that single year alone. Rystad’s forecast for 2026 is a further 130 GW / 350 GWh, with Europe and the US absorbing a material share. China remains the largest installer in absolute terms, but the European share has roughly doubled between 2023 and 2025.
Why 2024–2025 was the inflection point
Three things moved together. First, pack prices fell fast. BloombergNEF’s 2024 price survey put lithium-ion pack prices at around $115/kWh, down roughly 60% from $295/kWh in 2019. Chinese turnkey systems reached $150/kWh in 2025 (Rystad), and NREL projects further declines through 2030. Second, grid-scale projects started clearing financeable thresholds at volume: ~15 tolling deals signed across Europe in 2025 versus 3 in 2024, and close to 24 GWh contracted under flexibility purchase agreements across the year. Third, regulatory frameworks converged: EU Regulation 2024/1747 on the Electricity Market Reform, the EU 15-minute trading go-live on 30 September 2025, and an Iberian capacity market framework published in mid-2025 all gave projects legible revenue signals they did not have before.
The emerging duty cycle
Europe’s new batteries are not doing what GB’s 2018–2020 fleet did. That first wave was overwhelmingly 1-hour, frequency-response-focused and ancillary-dominated. The 2024–2025 wave is mostly 2–4 hour and is increasingly cycled around wholesale arbitrage. Italy provides the cleanest illustration: Terna and Ember data show battery discharge typically equivalent to roughly 3% of peak gas-hour demand in 2025, up from near-zero two years earlier - a small share of dispatch but a meaningful share of peak flexibility. California’s CAISO shows what the mature phase looks like: in April 2025, batteries provided more than 20% of CAISO’s evening generation on a typical day. Europe is tracking toward the same destination on a 4–6 year lag.
What is still scarce
The binding constraint in 2026 is not battery supply. Pack prices continue to fall and the OEM queue is long. The binding constraints are grid connection and permitting. Across most European markets, new interconnection requests sit behind multi-year queues, and the fastest-growing category of connection approval is hybrid (solar-plus-storage or wind-plus-storage) inside an existing envelope. Germany, the Netherlands, Italy and Spain have all published regulatory or legislative changes specifically to accelerate storage behind existing connections, because that is the path with least friction in the next three years.
What to watch through 2026
Three numbers will tell whether the 40 GW trajectory holds. Italy’s MACSE auctions: Europe’s first large-scale, long-duration storage-specific capacity mechanism is clearing in 2025–2026 with contract terms up to 15 years. The Iberian first auction in 2026: price discovery for a 22.5 GW target. And the German Bundesnetzagentur’s inertia market, live from January 2026, which will be the first European revenue stream specifically engineered for grid-forming inverters. If any two of those three clear at volume, the Ember / Rystad trajectory for 2026–2030 is on track. If they clear badly, European battery deployment still grows - but the financing mix becomes more equity-heavy and the tail-end forecasts soften.
In most European grids, getting a new project connected is a long wait. Developers queue for years to receive “firm” access - the right to inject or draw their full contracted power at any time. A few European regulators have now started offering a different deal: connect fast, but accept that in a small number of hours each year you will be asked to curtail. That is a non-firm grid connection, and for batteries it is emerging as one of the single biggest levers for accelerating deployment this decade.
The trade
The core bargain is simple. A firm connection costs more and takes longer because the grid operator has to reinforce the local network to accommodate worst-case conditions. A non-firm connection costs less and comes faster because the project contractually accepts curtailment in the worst-case windows. For a battery, which is physically designed to absorb and release energy flexibly, the cost of occasional curtailment is materially lower than for a wind or solar plant whose revenue loss tracks generation one-for-one.
What the Netherlands study actually modelled
A 2025–2026 paper in Energy Policy (Verhagen, Hu, Harmsen; Vol. 208, art. 114903) co-optimised firm and non-firm connection choices for BESS in the Netherlands. The headline result: a project accepting up to 15% curtailment in exchange for a 65% reduction in grid connection fees clears a higher net present value than the firm-connection counterfactual in nearly every modelled scenario. The loss from curtailment is smaller than the capex and grid-fee savings. Critically, the model accounted for the interaction between curtailment and market arbitrage: in the hours when the grid asks for curtailment (typically midday solar surplus or evening load peaks in the same local zone), wholesale spreads are often already high, and the battery compensates in the neighbouring hours.
How other European markets are moving
Germany has been the furthest along with its §14a EnWG regime, which allows flexible network tariffs for controllable loads including batteries and EV chargers. The Netherlands regulator ACM formalised non-firm connections (“flexibele aansluiting”) as a regulated product in 2024 and 2025. Belgium’s Elia and France’s Enedis both operate variants: Elia has piloted interruptible connections for industrial loads; Enedis offers “raccordement à puissance dynamique” in constrained zones. The UK’s National Grid ESO has run the Technical Limits process for years, and the ongoing reform to transmission connections (the REMA and TMO4+ processes) is explicitly expanding the space. Italy, Spain and Poland are all consulting on equivalent mechanisms in 2025–2026.
What it means for a developer
Three operational consequences follow. First, a non-firm project needs a more sophisticated optimiser. Curtailment windows are not random: they follow local congestion patterns, which are usually predictable and sometimes correlated with the same high-spread market hours. An optimiser that can pre-position state of charge ahead of likely curtailment windows recovers most of the foregone energy in neighbouring hours. Second, financing terms change. Lenders have been cautious on non-firm projects historically; the emerging precedent - including Dutch banks now underwriting non-firm BESS in 2025 - is that curtailment history over the previous 12–24 months from the local TSO is sufficient for debt sizing. Third, siting becomes more intentional: rather than filtering for the cleanest grid nodes, developers can target constrained nodes where the non-firm discount is largest and the battery is physically useful to the TSO.
The systemic logic
Non-firm connections work because they move the marginal cost of access from the system (grid reinforcement) to the project (occasional curtailment) and let batteries shoulder a cost they are physically well-placed to shoulder. Where firm connection requires 3–5 years and multi-million-euro substation works, non-firm connection can be completed in 12–24 months with minor local reinforcement. For a policy maker, it is one of the cheapest available mechanisms to accelerate deployment without raising the system cost for every other network user.
If Europe is going to triple or quadruple its battery fleet by 2030, a meaningful share of that capacity will come in through non-firm connections. The regulatory design question through 2026–2027 is how tightly to standardise the product: a patchwork of national rules will work, but a common European framework (an extension of the Network Code on Grid Connection, for example) would reduce risk premiums and accelerate the curve further.
Battery fires are rare, but when one happens it makes the news. Because there have been so few, most of what the public knows about BESS failures is anecdotal. EPRI has been quietly changing that. Its Battery Energy Storage Systems Failure Incident Database, now past 81 logged incidents and 26 fully classified root-cause investigations, is the best public dataset on where, how, and why grid-scale batteries actually fail. Two headlines from the most recent release: the failure rate per installed gigawatt-hour has dropped by 97% between 2018 and 2023, and most of the remaining failures do not happen in the cells.
The trajectory
EPRI’s failure-rate tracking, normalised against installed capacity, shows the trajectory falling from roughly 10 incidents per GW in 2018 to well under 1 per GW by 2023. The absolute count of incidents has risen modestly as the installed fleet has grown, but the intensity per unit of deployed energy has collapsed by two orders of magnitude. That is not a story about cells becoming fundamentally safer overnight - it is a story about industrialisation: better commissioning procedures, better fire-suppression design, standardised testing, and a rapid shift to LFP chemistry across grid-scale procurement.
When failures actually happen
The single most surprising finding in the EPRI dataset is temporal: 72% of failures occur during construction, commissioning, or within the first two years of operation. Mid-life failures - year 3 to year 10 - are rare. Late-life knee-point failures are typically bounded by operational policy rather than catastrophic events. The implication is that the riskiest period for a BESS is not the end of its design life but the beginning. EPC quality, commissioning rigour, and first-year operational procedures are the highest-leverage interventions for fleet safety.
Where failures come from
Of the 26 failures classified in detail, the root causes cluster not where most of the public commentary assumes. Only 11% were attributed to cell-level defects. 26% came from issues in the fire-suppression or fire-protection system, 18% from thermal management (cooling, HVAC, thermal runaway propagation), and the rest from a combination of BOS components, EMS firmware, installation errors, and site-level integration issues. The pattern is consistent across North American and European incidents in the database: modern cells, well selected and operated inside their envelope, rarely initiate failures on their own. What turns a contained cell event into a system-level incident is the response system around it.
What operators actually do differently now
Four operational changes have tracked the failure-rate collapse. First, large-format prismatic LFP has replaced most NMC in new-build grid-scale systems, raising the thermal-runaway onset temperature from 150–200°C to roughly 270°C and reducing flammable-gas release per cell. Second, NFPA 855 and UL 9540A testing are now effectively industry-standard; the tests specifically stress propagation between modules and are used to tune enclosure venting. Third, multi-stage fire detection - gas sensing before smoke detection, thermal-runaway precursor detection before gas - has moved from optional to default in European insurer requirements. Fourth, commissioning procedures have been formalised: 72 hours of monitored continuous operation under representative duty cycles before handover is now a common EPC contractual milestone.
What the database still cannot tell us
The EPRI dataset covers incidents that were reported and investigated. There is a plausible under-reporting bias for smaller thermal events that did not escalate, and a larger one for operational issues (capacity fade beyond specification, repeated parasitic-load excursions) that do not rise to “incident” status. The granular early-warning dataset - what happens before a cell hits thermal runaway - lives mostly in proprietary telemetry and has not been pooled. Several European TSOs and regulators are discussing a common reporting template in 2025–2026, which would materially improve the shared picture.
What it means for procurement
The EPRI database is the cleanest empirical answer to the question insurers and lenders ask: is this asset class getting safer? The answer is yes, decisively, and the improvements trace to specific engineering and procedural changes. For a developer, the practical takeaway is that the cheapest safety intervention available is EPC quality during commissioning - where 72% of failures originate. That is where procurement, warranty language, and commissioning acceptance procedures deserve the most attention. The cells will be fine; the question is whether everything around them is.
For a long time, the easiest way for a battery to earn money in Europe was to help the grid stay at 50 Hz. Frequency response and other ancillary services were lucrative and under-contested. That era is ending. In every European market where batteries have scaled - Great Britain, Germany, France, the Nordics - ancillary-service prices have collapsed. Batteries entering service in 2026 cannot rely on them in the way 2020–2022 projects did. The revenue stack is shifting visibly, and the design choices it forces on new projects are big.
What the numbers show
McKinsey’s 2025 analysis of European BESS revenue tracks a clear pattern: ancillary services are on a trajectory from 50–80% of the revenue stack today to below 40% by 2030, with wholesale arbitrage rising from 20–50% today to becoming the dominant line, and capacity markets moving from near-zero to 20–30%. In Great Britain, average battery revenues fell from around $300/kW/yr in 2022 to roughly $182/kW/yr in 2023 as ancillary prices corrected - the top quartile held up through wholesale arbitrage, but the fleet average collapsed. In France, the primary-reserve (FCR) market saturates at roughly 1 GW of battery participation, past which the clearing price drops to 40–50% of its 2021–2022 level. Similar dynamics in Germany, Belgium and the Nordics.
Why saturation happens fast
Ancillary services are small markets relative to wholesale energy. Primary frequency response across the European synchronous area is on the order of 3–4 GW of contracted capacity. Secondary reserve is larger but still bounded by system need. A market where 1 GW of batteries participate is already at 25–30% saturation for FCR; at 2 GW it is past it. The economic floor on ancillary price is set by the next-cheapest provider - typically a flexible generator or demand-response aggregator - and batteries are structurally cheaper than both once capex is sunk. Prices collapse toward the variable cost of a battery, which is essentially zero plus degradation cost.
What is emerging in its place
Three new revenue layers are emerging to backfill.
Wholesale arbitrage on shorter timescales. EU single day-ahead and intraday markets moved to 15-minute resolution on 30 September 2025. The finer time resolution creates more spread opportunities per day, and a fast, well-optimised battery captures a larger share of them than a slow-responding resource. This is the main reason wholesale is expected to rise from 20–50% to the dominant line by 2030.
Capacity markets and long-duration storage contracts. Italy’s MACSE, Greece’s storage CfD, the forthcoming Iberian capacity auction, and GB’s long-duration storage cap-and-floor are all attempts to pay batteries for being present rather than only for dispatch. The contract lengths (15 years) and the revenue certainty (regulated or competitively set) materially change project IRRs. McKinsey flags these as the fastest-growing share of future European BESS revenue.
System services for grid-forming inverters. Germany’s Bundesnetzagentur is launching a paid inertia market in January 2026 at an indicative €8–17k/MW/year for grid-forming-capable resources. Similar mechanisms are in consultation in the Nordics, GB, and Spain. The service is new, the price point is material, and only grid-forming inverter designs qualify - which tilts procurement toward units with the right firmware architecture specifically.
What it means for a new project
Three practical consequences. First, a 2026 pro-forma that assumes 2022-level ancillary-service revenue is wrong. The honest modelling assumption is that ancillary revenue is a residual, not a base case. Second, wholesale arbitrage increasingly sets the revenue floor, and that floor depends on forecasting accuracy, execution latency and dispatch discipline - all of which favour operators with strong optimisation stacks. Third, the capacity-market line is the closest thing left to a contracted annuity, and the cost of missing that line (either through poor auction strategy or failure to meet de-rating thresholds) is material to IRR.
The saturating-ancillary-services problem is not unique to Europe - California went through the same cycle in 2020–2023 - but Europe is going through it faster because the installed base is scaling rapidly from a small starting point. The good news is that the transition to wholesale + capacity is not a crisis for well-sited, long-duration batteries; it is a headwind for 1-hour, ancillary-only designs built to the 2020–2022 playbook. The new playbook, for 2026–2030, is 2–4 hour duration, capacity-market participation, and wholesale optimisation as the dominant skill.
Electricity is an unusual commodity. It cannot be cheaply stored in large quantities, it must be generated at the exact instant it is consumed, and if supply drifts from demand by even a fraction of a percent, grid frequency moves off 50 Hz and the system destabilises. That physical constraint is why Europe does not have one energy market but a sequence of them - each one cleaning up what the last one did not finish, closer and closer to real time. This primer walks through every product a battery can participate in, in the order they run.
The market sequence - delivery is “now”
Forward
Years → months
PPA, tolling
Day-ahead
D-1, 12:00 CET
SDAC auction
Intraday
H-1 → H-15 min
SIDC + IDAs
Balancing
Seconds → minutes
FCR/aFRR/mFRR/RR
Capacity & System
Parallel contracts
CM, inertia, VAr
Why there are so many markets
Most commodities have one market. Electricity has seven or eight, because the grid’s physics force a just-in-time match between supply and demand. The grid operator (Red Eléctrica in Spain, Terna in Italy, 50Hertz/TenneT/Amprion/TransnetBW in Germany, NESO in GB, RTE in France) runs a sequence of markets that progressively refine the plan: years ahead, then days ahead, then intraday, then balancing, then real-time system services. Each market has its own rules, its own product, and its own clearing logic. A battery can earn from almost all of them - and that layering is why revenue models feel complicated.
Forward markets - bilateral PPAs and tolling
The longest-dated products do not clear on a visible exchange. They are negotiated between two parties and held for years. A Power Purchase Agreement (PPA) is a fixed-price contract between a generator and a buyer. For batteries, the equivalent is a tolling agreement: a fixed fee paid to the battery owner in exchange for the right to dispatch the asset during the contract period. Tolling solves the financing problem for a new project by giving lenders a contracted cash flow rather than a merchant forecast.
The European flexibility tolling market grew roughly five-fold from 2024 (about three public deals) to 2025 (approximately fifteen), with aggregate contracted volume in the tens of GWh across the year. Contract tenors typically run 5 to 15 years. Utilities, traders and large industrial offtakers are the most common counterparties.
The day-ahead market - the one that matters most
The day-ahead market is the main wholesale energy market in Europe and the single largest source of revenue for most batteries. Every day at noon local time, an exchange collects bids and offers for every time block of the following day and runs a single algorithm - EUPHEMIA - that clears the continent together under the Single Day-Ahead Coupling (SDAC) rule.
The exchanges that run each zone
OMIE – Spain & Portugal (MIBEL) · EPEX Spot – DE, FR, NL, BE, AT, CH, GB, IT-North, NO · Nord Pool – Nordics, Baltics, IE, NL, PL, GB · GME – Italy · HUPX / OKTE / OTE / OPCOM – Central & Eastern Europe · EXAA – Austria
How the clearing works, step by step:
Generators offer MWh at a minimum price they will accept
Retailers, industrials and batteries bid for MWh at a maximum price they will pay
EUPHEMIA stacks the offers upward and the bids downward and finds the intersection price for each time step
Cross-border capacity is allocated simultaneously so power flows from cheap zones to expensive ones until the interconnection is full
Every accepted bid is paid the single clearing price for that interval (pay-as-cleared, marginal pricing)
Key parameters: auction closes at 12:00 CET on D-1; results publish around 13:00 CET. Resolution was 60 minutes up to 30 September 2025 and is 15 minutes from that date onward, under the EU-wide SDAC 15-minute market time unit. Price ceiling and floor inside MIBEL (Spain/Portugal) sit at +€4,000/MWh and −€500/MWh; other zones use equivalent harmonised limits. For a battery, day-ahead is the easiest market to participate in: charge during the cheapest hours, discharge during the most expensive.
Intraday markets - continuous trading plus auctions
After day-ahead clears, the intraday market opens and stays open until 15 minutes before delivery (in some zones, five minutes before). Prices change every few seconds as forecasts update: a revised wind nowcast, a plant outage, a sudden demand spike - each one moves intraday prices.
Two formats coexist under the Single Intraday Coupling (SIDC):
Continuous intraday (the pan-European XBID order book): any party can post a buy or sell order; the best bid matches the best offer instantly
Intraday auctions (IDA1, IDA2, IDA3): discrete 15-minute auctions that clear at 15:00, 22:00 and 10:00 CET respectively, providing liquidity depth in addition to continuous trading - operational on SIDC since 13 June 2024
Volume is growing fast. EPEX Spot traded 21,132.7 GWh on its intraday markets in October 2025 alone - about 14% higher than October 2024 - as the 15-minute transition pulled more continuous trading closer to delivery. For a battery, intraday is where a fast optimiser earns its money: a well-forecast asset with low execution latency captures sub-hour spreads that a static day-ahead trader cannot.
Price limits in MIBEL: continuous intraday ±€1,500/−€150/MWh; IDA auctions ±€200/−€20/MWh. Continuous trades settle pay-as-bid; IDA auctions settle pay-as-cleared.
Balancing markets - FCR, aFRR, mFRR, RR
The balancing markets keep the grid at 50 Hz in real time. They procure reserve - capacity that can be activated on demand to inject or withdraw power when the system unbalances. There are four products, layered by speed.
Product
Activation
Trigger
Procurement / platform
FCR Primary / containment
< 30 s, sustained 15 min
Local frequency deviation (automatic)
Weekly/daily auction, capacity payment only
aFRR Secondary / automatic
30 s – 5 min full delivery
TSO dispatch setpoint (4 s signal)
Daily auction, capacity + energy, EU PICASSO platform
mFRR Tertiary / manual
Within 12.5 – 15 min
TSO instruction (manual)
Daily auction, capacity + energy, EU MARI platform
RR Replacement
15 min – 1 h
TSO instruction
Used in ES (REE), IT (Terna), PT
When the grid wobbles - the activation cascade
A generator trips at t = 0. Frequency starts to fall. FCR injects within 30 seconds to arrest the fall. aFRR takes over from ~30 s, restoring frequency to 50 Hz via TSO setpoints. If the imbalance persists beyond ~5 minutes, mFRR is dispatched to free up aFRR headroom. If still unresolved after 15 minutes, RR and rebalancing redispatch complete the response.
How FCR actually works. A battery in FCR holds its state of charge near the middle of its window. When grid frequency dips below 50 Hz, the battery injects power; when it rises above 50 Hz, it absorbs. The response is fully automatic, local, and proportional to the frequency deviation - the unit responds to the frequency it reads at its own terminals, with no TSO signal. Procurement is typically weekly in Continental Europe (FCR-CE) and daily in GB.
How aFRR and mFRR work. Both are TSO-dispatched. The TSO sends a power setpoint every 4 seconds (aFRR) or dispatches the unit manually by instruction (mFRR). Units hold reserved capacity throughout the delivery window and receive a capacity payment for being available plus a separate energy payment when activated. PICASSO - the pan-European aFRR energy exchange - went live June 2022; RTE (France) joined in April 2025, and most EU member states are scheduled to connect by end-2025. MARI is the equivalent mFRR platform. ACER’s 2025 monitoring report credited the balancing platforms with over €1.6 bn of cross-border benefits in 2024.
What is happening to prices. Balancing was historically the largest revenue source for batteries, but in every European market that crossed roughly 1 GW of battery participation, prices have fallen sharply. Great Britain, Germany, France and the Nordics all show the same pattern: initial scarcity, rapid battery build-out, and a clearing price that drops toward the variable cost of a battery (near zero plus degradation). GB’s 2024–25 annual balancing bill hit £2.7 bn - up 10% year-on-year - but on a per-MWh basis balancing prices compressed because BESS imported 506 GWh and exported 537 GWh through the Balancing Mechanism.
Capacity markets - paid to be there
A capacity market is a different logic. It pays resources for being available during system stress events, not for dispatching energy in normal conditions. The TSO or regulator runs an auction that procures, say, 40 GW of “firm” capacity for a delivery year three or four years ahead. Winners receive an annual payment in €/kW-year in exchange for an obligation to respond when called.
Key design features:
Delivery year - typically 3 to 4 years ahead of the auction
De-rating factor - batteries are credited at less than nameplate. A 1 MW × 4 h battery might be credited at ~85% of 1 MW in GB; a 2-hour asset drops to ~65%; a 1-hour asset to ~35%. The exact factor depends on system tightness and duration
Contract length - 1 year for legacy assets, 5 to 15 years for new-build storage
Penalties - failure to deliver during a stress event triggers clawbacks that can equal or exceed the annual payment
European capacity markets - active or launching 2025–2026
GB - T-1 and T-4 auctions, annual, longest-running
Italy - Mercato della Capacità; MACSE storage-specific auction cleared at 15-year tenor
France - Marché de Capacité, reforming
Poland - Rynek Mocy
Ireland - CRM (all-island)
Belgium - CRM since 2021
Iberia - Spanish framework published mid-2025; first auction expected 2026
System services - voltage, inertia, black start
Beyond balancing, a grid needs services that keep the system stable as a physical machine.
Voltage control (reactive power). Inverters in modern batteries can absorb or inject reactive power on command and can be paid for it. Procurement is usually bilateral with the TSO or through a local auction. REE in Spain procures voltage services under P.O. 7.4 with payments that are re-tendered each year.
Inertia. The ability of spinning masses to resist frequency change in the first seconds of a disturbance. Synchronous machines provide it naturally; grid-forming inverter batteries can emulate it. Germany’s Bundesnetzagentur launched a paid inertia product from January 2026 at an indicative €8–17k/MW-year for qualifying resources. NESO (GB) has procured “stability” since 2020 via Stability Pathfinders; Ireland runs a DS3 system-services suite; Spain, the Nordics and France have consultations ongoing.
Black start. The ability to re-energise a dead grid without external power. A small number of new BESS contracts pay specifically for this capability, typically as a long-term bilateral with the TSO. After the 28 April 2025 Iberian blackout, Spain’s RD-law 7/2025 mandated a larger black-start portfolio as part of its anti-blackout response.
How a battery stacks revenue
A 2026-built 2- or 4-hour battery in Europe typically stacks revenue along four lines.
Typical revenue stack - merchant battery, 2026 Europe
Wholesale arbitrage (day-ahead plus intraday) forms the base layer, usually 40–60% of revenue. Balancing (FCR and aFRR together) is a residual layer, 15–30% and falling as the market saturates. The capacity market, where available, grows toward 10–25%. System services (inertia, voltage) contribute 5–10% in markets where those products have been procured. A tolling contract, if the developer chooses one, replaces some or all of the above with a fixed annuity.
The three rules that matter most in 2026
If you read only the fine print on three regulatory details, read these. First, the EU-wide move to 15-minute resolution on day-ahead and intraday, live since 30 September 2025 under the SDAC 15-minute MTU, benefits batteries more than slow resources because they can capture sub-hour spreads the old hourly market invisibly averaged out. Second, EU Regulation 2024/1747 on Electricity Market Reform pushes every member state toward capacity-market and long-term-contract frameworks as primary investment signals - that is why MACSE, the Iberian framework and GB’s long-duration cap-and-floor all appeared in the same 18-month window. Third, updates to national grid codes on grid-forming inverters (GB, Germany, Spain) decide which batteries can qualify for inertia and voltage products as they are launched - a procurement decision made in 2025–2026 determines which revenue streams a project can access through 2040.
A lithium-ion battery is not magic. It is chemistry. Lithium ions shuttle back and forth between two layered materials, packaged carefully enough that the motion repeats millions of times without failure. Once you see the pieces, most of the behaviour of a grid-scale battery makes sense - including why it ages, and what an operator can actually do about it.
What “lithium-ion” actually means
Inside a lithium-ion cell are two active materials: a cathode (the positive electrode, a metal oxide or phosphate that can host lithium inside its atomic structure) and an anode (the negative electrode, usually graphite). When the cell charges, lithium ions leave the cathode, travel through a liquid electrolyte, and slot into the anode. When it discharges, they move back. Electrons cannot pass through the electrolyte - they flow through the external circuit instead and do useful work on the way. That round trip is what we call charging and discharging a battery.
From cell to system - four nesting layers
Everything a grid-scale BESS does happens across four physical scales. Knowing which scale a problem lives at is half of diagnosing it.
Cell → Module → Rack → System
■
Cell
~1–1.9 kWh
3.2 V (LFP)
■■■
Module
~30–60 kWh
16–52 cells in series
◨
Rack
~250–400 kWh
~8 modules + BMS
▣
System / container
~5–6 MWh
20-ft unit, PCS, HVAC
The six parts of a cell
Anatomy of a prismatic LFP cell
Cathode (+)
LFP or NMC active material coated on aluminium foil. Sets voltage and capacity per kg.
Anode (−)
Graphite (or graphite-silicon) coated on copper foil. Hosts lithium when charged.
Separator
Porous polyolefin membrane. Lets ions through, blocks electrons from shorting.
Electrolyte
Lithium salt (LiPF6) in organic solvents. The medium that carries ions.
Current collectors
The aluminium/copper foils plus tabs carrying current in and out.
Casing
Aluminium or steel housing that seals everything and manages thermal & mechanical loads.
Hidden inside every working cell is a seventh layer: the Solid Electrolyte Interphase (SEI) - a thin passivation film that forms on the anode during the first charge cycles. The SEI is essential (it stops continuous electrolyte decomposition) but it also slowly thickens over time, consuming usable lithium. Its growth is the chemistry underlying calendar aging.
The chemistries you will see in 2026
Li-ion is not one chemistry but a family. In 2024 JRC data, LFP was more than 20% cheaper per kWh than NMC but with ~20–30% lower energy density. For stationary storage that trade-off is easy; for vehicles it is harder. DNV’s 2024 Battery Scorecard ranked CATL and Narada LFP cells at the top of the stationary-storage class on combined performance and safety. For grid-scale projects delivered in 2025–2026, LFP dominates; NMC remains the default for mobility; LMFP and sodium-ion are climbing the commercial curve.
Chemistry
Cathode
Energy density
Cycle life
Where it dominates 2026
LFP
Lithium iron phosphate
~390–434 Wh/L
8,000–10,000+
Stationary storage
NMC 811
Ni-Mn-Co oxide, high-Ni
~600–700 Wh/L
3,000–5,000
Passenger EVs, some stationary
LMFP
Lithium manganese iron phosphate
~450–500 Wh/L
5,000–8,000
Emerging EVs; early stationary pilots
Sodium-ion
Prussian-white or layered oxide
~250–330 Wh/L
4,000–6,000
Cold-climate & cost-driven stationary (TRL 9 since 2024)
Grid-scale projects in 2025–2026 are overwhelmingly built with large-format prismatic LFP cells: 280 Ah was mainstream, has mostly been replaced by 314 Ah in 2025, and the industry is now moving to 500 Ah+ formats. CATL’s 587 Ah cell reports a volumetric energy density around 434 Wh/L, and a datasheet cycle life above 10,000 full cycles under the manufacturer’s specific test conditions (0.5C, 25 °C, shallow DoD, cell under compression); real fleet cycles at utility-scale duty typically land well below that envelope. 20-foot containers using 587 Ah cells hit about 5 MWh without exceeding the 45-ton weight limit.
The metrics that describe a cell
A cell is described by a small number of numbers. Each one tells you something different.
Metric
What it means
Typical for LFP grid cell
Capacity (Ah)
How much charge the cell can store. A 314 Ah cell can supply 314 A for one hour.
280 – 587 Ah
Energy (kWh)
Capacity × nominal voltage. The useful work the cell can deliver.
~1.0 – 1.9 kWh/cell
Voltage (V)
Potential across the terminals. Changes with state of charge, current and temperature.
3.2 V nominal, 2.5 – 3.65 V range
SoC (%)
State of Charge - where the cell sits between empty and full.
0 – 100%
SoH (%)
State of Health - capacity remaining as a fraction of original. End-of-life usually 70–80%.
100% new → 80% at EoL
Internal resistance
Voltage drop per unit current. Lower is better. Rises with age and cold.
~0.2 – 0.5 mΩ new
C-rate
Current relative to capacity. 0.5 C on a 314 Ah cell is 157 A.
0.25 C (4 h) – 1 C (1 h)
Round-trip efficiency (AC)
Energy out ÷ energy in over a full cycle, AC-to-AC at system level.
85 – 94% new; DC cell-level 95–98%
Internal resistance deserves special attention. It is the number behind almost every aging symptom. When a new cell has low resistance, the voltage barely drops under load and very little energy is lost as heat. As a cell ages, resistance rises; the cell sags more under load, runs warmer, and loses a bit more energy every cycle. Monitoring resistance growth over time is one of the earliest warning signs of accelerated aging - often visible months before capacity fade becomes obvious.
Why batteries age - two mechanisms running at once
A battery ages in two ways at the same time. The observed degradation you see in the data is their sum. You cannot switch one off without understanding the other.
Cycling aging happens when you use the battery. Every charge-discharge cycle causes small, irreversible changes: lithium atoms getting trapped in the SEI, microscopic cracks forming in the cathode particles, a bit of electrolyte decomposing at each interface. NREL’s BLAST model, built on accelerated aging data, decomposes the effect into terms for SEI growth, electrode cracking, cycling-driven acceleration of SEI growth, and early-life “break-in” shifts in lithium inventory.
Calendar aging happens even when the battery sits still. The SEI continues to grow slowly - capacity loss follows a characteristic √t shape, proportional to the square root of storage time, with the rate constant set by an Arrhenius temperature dependence. A cell kept at 100% SoC and 40 °C loses capacity faster while idle than a cell kept at 50% SoC and 20 °C.
Real-world operation is never one or the other. It is always both, simultaneously, and the skill in operating a fleet is knowing which one is dominant at each moment.
SoH decay - illustrative, three duty cycles over 15 years
The six operational factors that shape degradation
Six operational choices determine how long a given cell actually lasts. Each one has a direction and a trade-off.
Factor
How it drives aging
Operational lever
Temperature
Arrhenius physics. Reaction rate roughly doubles for every ~10 °C above 25 °C.
Liquid cooling; keep cells 20–30 °C
Depth of Discharge
Deep cycles stress both electrodes at their extremes. Shallow cycles age the cell far less.
Dispatch policy; SoC window
C-rate
Higher current = more heat + more mechanical stress on electrodes.
Sizing (2 h vs 1 h); rate limits
Voltage window
Top-of-charge and bottom-of-discharge voltages stress chemistry disproportionately.
BMS upper/lower cutoff choice
Cycle count
Raw throughput. But cycles at mild conditions are not equivalent to cycles at harsh ones.
Revenue strategy (arbitrage vs FR)
Round-trip efficiency
Energy not returned comes out as internal heat - driving the temperature factor above.
System design (inverter, thermal)
Temperature
Temperature is the single biggest controllable driver of battery life. The chemical reactions that cause aging obey the rule of physics called Arrhenius: reaction rate roughly doubles for every 10 °C of temperature rise. In practice, 25 °C is the reference for most cell data sheets; 35 °C roughly doubles aging rate; 45 °C doubles it again. Below 10 °C, the opposite problem appears: lithium plating risk during charging rises sharply, which can cause both capacity loss and dendrite growth that threatens safety. This is why every serious grid-scale BESS now uses active liquid cooling and keeps cells inside a 20–30 °C band. A well-cooled cell can reach 6,000–10,000 full-equivalent cycles to 80% SoH; the same cell run hot at 40 °C may reach 80% SoH in 2,000–3,000.
Depth of Discharge
DoD is how deep each cycle goes. A 100% DoD cycle takes the cell from full to empty. A 60% DoD cycle uses only the middle 60% of the window. Shallower cycles age the cell much less, because the chemistry is stressed less at both extremes. Reducing DoD from 100% to 80% typically extends life by roughly 30–50%; reducing further to 60% extends it more. The trade-off is that shallower cycling means less energy throughput and less revenue per dispatch. Grid-scale LFP systems usually operate at 90–100% DoD in normal dispatch and dial back only in protective conditions.
C-rate and charge speed
The faster current flows in or out, the more heat is generated inside the cell and the more mechanical stress is placed on the electrode particles. A cell rated 1 C can sustain higher C-rate for short pulses but running continuously at elevated C-rate raises temperature, accelerates aging, and may trigger protective controls. 2- and 4-hour grid batteries typically run at 0.25–0.5 C, which is mild. 1-hour batteries run at 1 C, which is still well within cell ratings but produces materially more heat. Sub-1-hour dispatch (fast frequency response) pushes C-rate much higher in short bursts and is one of the reasons FR-focused assets age differently from arbitrage assets even at similar throughput.
Voltage window
Every cell has a voltage range inside which it is safe and healthy to operate. For LFP, that range is roughly 2.5 to 3.65 V per cell. Pushing the cell all the way to the upper or lower limits every cycle stresses the electrodes chemically. Narrowing the window (say, 3.00–3.45 V for LFP) reduces stress on both electrodes and slows both calendar and cycle aging. The trade-off is that a narrower window delivers less usable energy per cycle. Grid operators usually pick a middle-ground window that captures ~95% of theoretical energy while materially extending life.
Cycle count and usage intensity
Cycle count is the running total of full-equivalent cycles the battery has performed. Two cells that have done the same number of cycles under different conditions (temperature, DoD, C-rate) will not be at the same SoH. Cycle count alone is an incomplete picture - what matters is cycle count weighted by intensity. Modern fleet operators track effective cycles that credit mild cycles at lower weight and heavy cycles at higher weight. Typical European duty cycles: wholesale arbitrage at 1–1.5 full-equivalent cycles/day, balancing-focused assets at 0.5–1 with many shallow micro-cycles, heavy arbitrage tolling at up to 2 full-equivalent cycles/day.
Round-trip efficiency
RTE is usually thought of as a revenue metric - lower efficiency means more energy bought, less sold - but it also has a direct aging effect: the energy that does not come back out is mostly lost as heat inside the system. At the cell level (DC), lithium-ion RTE is high, typically 95–98%. Step up to the battery terminals on the AC side and 85–94% is more typical; published field studies of containerised systems show 88–92% at AC terminals for healthy new LFP plants and can fall another 8–13 percentage points once auxiliary consumption (HVAC, pumps, fans, BMS electronics) is accounted for on a 24-hour basis. A 90% RTE battery generates roughly half the internal heat of an 80% RTE battery over the same throughput; high-RTE designs age more slowly because they run cooler under the same duty cycle. RTE itself falls over time as internal resistance rises with age. A new LFP system at 92% AC RTE may be at 86% after 10 years and 80% at end of life.
Safety - what actually goes wrong at grid scale
The most cited safety concern in lithium-ion is thermal runaway: a self-sustaining exothermic reaction inside a cell that can propagate to neighbouring cells if not contained. Every modern BESS is engineered around stopping runaway early - thermal barriers, pressure relief, fire-rated enclosures and rapid gas detection - but the more important insight from the field is that runaway is rarely the root cause.
EPRI’s public BESS Failure Incident Database recorded 81 incidents worldwide through May 2024 and showed two striking facts: the failure rate per GW of installed capacity dropped by roughly 97% from 2018 to 2023, and of the failures that did occur, only about 11% traced to cell manufacturing defects - roughly 89% were driven by the balance of system, controls or operational error. Approximately 72% of documented BESS failures occurred within the first two years of operation, making commissioning quality and burn-in the dominant lever on project safety.
The BMS - the computer that keeps the chemistry honest
The Battery Management System is the layer of sensors, microcontrollers and software that sits between the cells and the rest of the world. Its job is to prevent any cell from leaving the safe operating envelope, to keep the pack balanced so no cell drifts far from the others, and to report state to the site controller. At minimum, a grid-scale BMS measures cell voltage, current and temperature thousands of times per second; performs active or passive balancing between cells; estimates SoC and SoH cell-by-cell; enforces upper and lower voltage and temperature cutoffs; and triggers protective isolation when limits are breached. The BMS is why a battery pack made of thousands of individually weak cells behaves as a single strong one.
Putting it all together
Four operational choices compound into a battery’s real-world life: cooling (keep cell temperature inside 20–30 °C), DoD (avoid deep cycles when not needed), C-rate (match dispatch strategy to cell rating) and voltage window (leave margin at the extremes). A system that gets all four right can reach 10,000 full-equivalent cycles at 80% SoH. A system that gets none of them right can be at end of life by 3,000.
The central insight is that degradation is not a fixed property of the cell. It is a function of how the cell is operated. The chemistry sets an envelope; operational choices decide where inside that envelope a specific battery actually lives. This is why identical hardware at two different sites can show markedly different aging curves, and it is why modern operating strategies care about cooling, SoC window and C-rate in about the same detail they care about market prices.
