How Flow Batteries Actually Work: A Deep Explainer (2026)
The strangest battery you have probably never seen is the size of a shipping container, sounds like an aquarium filter, and stores its energy in two giant tanks of colored liquid. When you charge it, one tank turns deep violet and the other turns yellow-green. When you discharge it, the colors slide back. There is no solid anode. There is no risk of thermal runaway. The cell that produces the electricity is no bigger than a stack of pizza boxes, but the tanks behind it can hold enough vanadium electrolyte to power a small town for twelve hours.
Architecture at a glance
This is a flow battery, and after thirty years on the sidelines it is finally moving to the center of the grid. China’s Dalian project hit 200 MW / 800 MWh in 2024. ESS Inc commissioned its first utility-scale iron-flow systems in Sacramento and across Europe in 2025. Sumitomo has shipped hundreds of MWh of vanadium systems to Hokkaido and Australia. Every serious long-duration energy storage roadmap from the US DOE, the IEA, and BloombergNEF now includes flow chemistry as a distinct category alongside lithium-ion.
The interesting question is not whether flow batteries are real (they are) or whether they will replace lithium (they will not). The interesting question is how they work — what physical principles let you decouple energy from power, why the same vanadium atom can play both half-cells, and why ten hours of storage suddenly becomes cheap when you stop trying to pack it into a solid electrode. This is a deep explainer aimed at engineers, energy analysts, and the merely curious. We will write out the half-reactions, do the sizing math, and look at where flow chemistry is winning real procurement battles in 2026.
The Storage Problem Lithium Cannot Quite Solve
To understand why flow batteries matter, start with the duration curve of a modern grid. A solar-heavy grid (California, South Australia, Spain) routinely sees four to twelve hours per day where supply exceeds demand. Wind-heavy grids (Texas, the UK North Sea, Inner Mongolia) face the opposite: multi-day lulls when output collapses. To firm renewable generation into something a dispatcher can rely on, you need storage that can absorb midday solar surplus and re-release it through the evening peak — and survive doing that several thousand times a year.
Lithium-ion is excellent at the first two hours of this problem and increasingly competitive at four. Beyond four hours, the economics turn against it. The reason is structural, not a matter of incremental cost reduction. In a lithium-ion cell, energy and power are physically welded together. The same lithium-intercalating electrode that delivers current also stores the lithium. To get more energy, you stack more cells, which also gives you more (unwanted) power, more (unwanted) inverters, more (unwanted) thermal management, and a faster march down the cell-life curve as you cycle deep.
The US Department of Energy’s Energy Storage Grand Challenge roadmap explicitly calls out the “long-duration energy storage gap” as the technologies that can deliver 10 to 100 hours of discharge at sub-$0.05/kWh levelized cost. Lithium chemistry, even in its cheapest LFP form, struggles to hit that envelope. Pacific Northwest National Laboratory (PNNL), which has been doing flow battery research since the 1980s, has made the same point for years: at multi-hour scale, the right architecture is one where energy lives in cheap tanks of liquid and power lives in a separately sized electrochemical stack. That is exactly what a flow battery is.
How a Flow Battery Works
A flow battery is an electrochemical cell where the active redox species are dissolved in liquid electrolytes that are stored in external tanks and pumped through the cell. Charging and discharging happen inside a small, fixed cell stack; the energy lives in the tanks.
The Two-Tank Principle
Every flow battery has at least two reservoirs: an anolyte (the negative-side electrolyte) and a catholyte (the positive-side electrolyte). Each is a solution of a redox-active species — vanadium ions, iron salts, zinc bromide, an organic quinone — dissolved in an aqueous (and usually acidic) supporting electrolyte. Pumps circulate each liquid in its own closed loop through one half of the electrochemical cell.
When you charge the battery, electrical energy drives the species in the anolyte to a more reduced oxidation state and the species in the catholyte to a more oxidized state. When you discharge, the reactions reverse and electrons flow through the external circuit. Because the active material lives outside the cell, you can store as much of it as you can afford to build tanks for. Need ten hours of storage instead of four? Build bigger tanks. The cell stack does not change.
Half-Cells and the Membrane
The two half-cells share one critical component: an ion exchange membrane (most commonly Nafion, a sulfonated tetrafluoroethylene from Chemours). The membrane has two jobs that are in tension. First, it must let charge-balancing ions — usually protons (H⁺) — cross from one side to the other to close the electrical circuit. Second, it must not let the redox-active species cross. If vanadium ions leak across the membrane, you get a self-discharge reaction that wastes energy and slowly contaminates both tanks.
On either side of the membrane sit the electrodes. They are not the active material themselves — they are usually carbon felt or graphite plates, chosen for chemical stability and high surface area. The redox reaction happens at the electrode surface as the electrolyte flows past it. The cell stack is essentially a sandwich: bipolar plate, electrode, membrane, electrode, bipolar plate, repeated as many times as needed to reach the desired voltage and power.
For a vanadium redox flow battery (VRFB), the four oxidation states of vanadium do the chemistry. The two relevant half-reactions, first demonstrated as a practical couple by Maria Skyllas-Kazacos and her group at UNSW Sydney in the mid-1980s, are:
- Negative electrode (anolyte): V³⁺ + e⁻ ⇌ V²⁺ (E° = −0.26 V vs SHE)
- Positive electrode (catholyte): VO₂⁺ + 2H⁺ + e⁻ ⇌ VO²⁺ + H₂O (E° = +1.00 V vs SHE)
- Cell voltage: ~1.26 V at standard conditions
What makes VRFB elegant is that both half-cells use vanadium. Even if a vanadium ion crosses the membrane (and a few always do), it is not a contaminant in the foreign sense — it is the same element. Over many cycles you get a slow capacity drift that can be rebalanced by remixing the tanks; you do not get permanent poisoning.
Decoupling Energy from Power
Here is the architectural pivot point. In a lithium cell, doubling the energy means doubling the cell count, which doubles the power. In a flow battery, doubling the energy means doubling the tank volume, which costs you tanks and electrolyte but zero extra cell stack. Conversely, if you want more power for the same energy, you build a bigger cell stack and keep the tanks. The two design knobs are independent.
This is what makes flow chemistry uniquely suited to long-duration storage. A four-hour lithium system and a twelve-hour flow system can have similar power ratings (the cell stack), but the twelve-hour flow system costs only a marginal amount more — three times the electrolyte and three times the tank steel — while the lithium system needs three times everything. The break-even point, in 2026 numbers and depending on how you do the math, lands somewhere between four and six hours of discharge duration.
The Chemistries
Not all flow batteries are vanadium. The chemistry zoo has expanded substantially since the 2010s, with serious commercial deployments now spanning vanadium, iron, and zinc-bromine, plus a credible emerging class of organic flow batteries.
Vanadium Redox (VRFB)
VRFB is the reference design. It was pioneered by Skyllas-Kazacos at UNSW from 1984 onward, commercialized first by Sumitomo Electric (Japan) and Prudent Energy (China), and now anchored by Rongke Power, which built the Dalian 200 MW / 800 MWh system that went fully online in 2024. Round-trip efficiency is typically 70–80%. Energy density sits at roughly 20–40 Wh/L of electrolyte. The cycle life is the headline number: 20,000 full cycles is achievable, with very little degradation, because the electrodes do not undergo intercalation or plating — they just shuffle electrons in and out of dissolved ions.
The downside is vanadium itself. The vanadium pentoxide (V₂O₅) market is small, dominated by steelmaking, and historically volatile. A 2018 price spike to over $30/kg vanadium briefly made VRFB economics unworkable. Prices have since stabilized in the $7–10/kg range in 2026, but a single battery can lock up tens of millions of dollars of vanadium for its life. Leasing models, where the electrolyte is rented from the manufacturer and reclaimed at end-of-life, are how Rongke and Sumitomo route around the capital lock-up.
Iron Flow (IFB)
ESS Inc, based in Wilsonville, Oregon, builds an all-iron flow battery using two iron-based electrolytes — a Fe / Fe²⁺ plating couple on the negative side and a Fe²⁺ / Fe³⁺ redox couple on the positive side. Cell voltage is about 1.21 V. Energy density is lower (~9 Wh/L) but the materials cost is a small fraction of VRFB. Iron is the fourth most abundant element in the Earth’s crust; no critical mineral risk.
ESS commissioned its first utility-scale “Energy Center” deployments in Sacramento (with SMUD) and at the Burns Paiute Tribe site in Oregon during 2023–2025, and announced European deployments through partnerships with Honeywell. The chemistry is non-toxic, non-flammable, and the spent electrolyte is essentially fertilizer-grade iron salts. The trade-off is the hydrogen side-reaction: in aqueous iron systems, hydrogen evolution at the negative electrode competes with iron plating, so the system needs active hydrogen-management plumbing.
Note that ESS Inc’s iron flow battery is different from Form Energy’s iron-air battery (sometimes confused in the press). Form Energy’s product is a non-flow primary-style metal-air system targeting 100-hour discharge for the deepest LDES applications — different architecture, different problem set, similar earth-abundant ethos.
Zinc-Bromine
Zinc-bromine flow batteries (Redflow in Australia, formerly Primus Power in the US) are technically hybrid flow batteries: zinc plates onto the negative electrode during charge, so part of the energy is stored in solid form. Cell voltage is 1.82 V and energy density runs around 65 Wh/L — substantially better than vanadium or iron — because plating is more compact than dissolved ions. The cost is mechanical complexity (managing zinc dendrite growth) and bromine handling, since elemental bromine is corrosive and toxic. Cycle life is typically a few thousand full cycles, much shorter than VRFB.
Organic Flow
The most interesting research frontier is organic flow batteries: redox couples based on synthetic quinones, viologens, or other small organic molecules. The Aziz group at Harvard, and startups like Quino Energy (US) and CMBlu (Germany), are pursuing chemistries where the active species is a molecule that can be made from earth-abundant feedstocks (often petroleum-derived hydrocarbons) at low cost. Energy densities are still modest (10–20 Wh/L) and long-term molecular stability is the open scientific question, but if it works, organic flow batteries solve both the critical-mineral problem and the cost problem in one move.
The chart makes the trade-off visceral. Flow electrolytes are typically 5–30× less energy-dense than lithium cells. That is the central penalty of flow chemistry: storage takes up much more space. The compensating advantage is that the storage medium is dirt cheap by comparison, the cell life is enormous, and the failure modes are gentle.
Scaling Math
How do you size a flow battery? The first-principles calculation is straightforward and worth doing once, because it makes the design knobs concrete.
The theoretical energy in a flow battery is:
E = n · F · c · V · E_cell
where:
– n is the number of electrons transferred per redox event (1 for VRFB)
– F is the Faraday constant, 96,485 coulombs per mole
– c is the active-species concentration in moles per liter
– V is the total electrolyte volume in liters (per half-cell; assume both sides are equal)
– E_cell is the cell voltage in volts
Let us run a realistic VRFB example. Suppose we have two 200,000-liter tanks (one anolyte, one catholyte) at 1.6 M vanadium — close to the practical solubility limit in sulfuric acid at room temperature. n = 1, E_cell = 1.26 V.
E = 1 × 96,485 × 1.6 × 200,000 × 1.26 = 3.89 × 10¹⁰ J
Convert to watt-hours: 3.89 × 10¹⁰ / 3,600 = 10.8 MWh theoretical
In practice you cannot use 100% of that, because state-of-charge limits prevent full reduction or oxidation (you would crash the cell voltage, hit side-reactions, or precipitate vanadium). A typical usable window is 80%, giving roughly 8.6 MWh of deliverable energy. Round-trip efficiency on top of that (say 75%) means about 6.5 MWh round the loop. Add a 2 MW power stack to discharge it over four hours, or a 600 kW stack to discharge it over fourteen hours — your choice, independent of the tank sizing.
This is the calculation that drove the Dalian project: at 200 MW / 800 MWh, the design point is four hours of full-power discharge, but the tank volume gives flexibility to run lower-power, longer-duration profiles when that is what the grid needs.
When Flow Beats Lithium
Pure cost-per-kWh comparisons miss the point. The right question is which technology wins for which duration and cycling profile.
The rough heuristics, drawing on the DOE’s 2024–2025 LDES benchmarking and BloombergNEF’s grid storage outlook, look like this:
- Under 2 hours: Lithium-ion (especially LFP) wins decisively. Flow batteries have too much capital tied up in cell stack relative to energy delivered.
- 2 to 4 hours: Lithium is still usually cheaper today, especially for systems cycled less than once per day.
- 4 to 12 hours: This is the contested zone. Flow becomes attractive if (a) you cycle deeply every day, (b) you have urban or indoor siting where thermal-runaway risk is a regulatory issue, or (c) you need 15+ year asset life with minimal augmentation.
- Above 12 hours: Flow batteries (or iron-air, or thermal, or pumped hydro) dominate. Lithium economics fall apart at 24 hours.
Two specific advantages keep showing up in 2026 procurements. The first is safety. Aqueous flow electrolytes do not burn. The 2024 fire codes (NFPA 855 in the US, equivalent updates in EU and APAC) have tightened siting requirements for large lithium installations to the point where co-locating multiple MWh of LFP in a downtown substation is genuinely hard. Flow chemistry sidesteps that entirely.
The second is deep daily cycling. A lithium system rated for 6,000 cycles at 80% depth-of-discharge degrades meaningfully when you cycle it twice a day for ten years. A VRFB rated for 20,000+ cycles does not. For a daily solar-shifting profile, the flow battery’s cycle-life advantage compounds dramatically over the asset’s life.
Trade-offs and Limits
Flow batteries are not a free lunch. Three honest limitations.
Energy density. A vanadium electrolyte at 25 Wh/L is roughly 14× less energy-dense than an LFP cell at 350 Wh/L and 24× less than NMC at 600 Wh/L. That means tanks. A 10 MWh VRFB needs about 400,000 liters of electrolyte plus pump skids, plumbing, and a building. This rules flow batteries out of EVs, consumer electronics, and anywhere space is tight. Stationary grid storage is where the volume penalty stops mattering.
Pumps and balance-of-plant. A flow battery is the only common battery chemistry with moving parts. Pumps, valves, seals, and pipes all need maintenance and contribute to parasitic energy losses (typically 1–3% of system output goes to running pumps). The good news: these are well-understood industrial components, not exotic materials. The bad news: any failure that takes the pumps offline takes the battery offline, regardless of state of charge.
Material costs and price volatility. VRFB depends on vanadium, a thinly-traded commodity. Zinc-bromine depends on bromine, a controlled chemical with global supply concentration. Even iron flow has plating-chemistry complications. Each chemistry has a different cost-stack risk profile, and project finance has to model them explicitly.
Round-trip efficiency. 70–80% for VRFB and IFB versus 88–92% for modern lithium. The 10–15-point gap is real and matters for arbitrage economics. If your business model depends on capturing every cent of price spread between cheap midday and expensive evening, lithium’s efficiency edge can matter more than flow’s cycle-life edge.
Where They are Deployed in 2026
The 2026 deployment map for flow batteries is unrecognizable from five years ago. A few anchor projects worth knowing:
- Dalian VRFB (China). Rongke Power’s 200 MW / 800 MWh vanadium system, commissioned in stages and fully online by 2024. The largest flow battery in the world by a substantial margin. It anchors Dalian’s peak-shaving program and is the proof point that VRFB scales to hundreds of MWh of installed capacity.
- ESS Inc Sacramento and partnerships. The Sacramento Municipal Utility District (SMUD) deployment, plus a broader rollout via the Honeywell partnership announced in 2023, put iron-flow systems on the map in the US and Europe. ESS public-company status (NYSE: GWH) makes the cost trajectory legible.
- Sumitomo Electric (Japan, Australia, US). Sumitomo has been shipping VRFB systems since the early 2000s. The Hokkaido Electric 60 MWh installation has been operating since 2015 — twenty thousand cycles of empirical data on VRFB longevity.
- CellCube (Austria). Long-standing European VRFB integrator, with deployments in distributed energy and microgrid roles across Europe and South Africa.
- Redflow (Australia). Zinc-bromine specialist with deployments in telecoms, off-grid, and behind-the-meter commercial sites; restructuring in 2025 reshaped the company but the technology base persists.
FAQ
Are flow batteries efficient?
Round-trip efficiency is typically 70–80% for vanadium and iron flow, versus 88–92% for modern lithium-ion. Lower than lithium, but high enough for most grid arbitrage applications when paired with long cycle life.
Why aren’t flow batteries everywhere?
Energy density. A flow battery occupies roughly 10–30× the volume of an equivalent-energy lithium system. That is fine for grid-scale storage on a few acres of land, but disqualifying for EVs, phones, and anywhere space is constrained. They are also more complex (pumps, plumbing, two electrolyte loops) than a sealed lithium cell.
Are vanadium electrolytes toxic?
Vanadium compounds are moderately toxic and the electrolyte is dissolved in sulfuric acid, so it is handled with industrial chemical safety practices — not consumer-grade safety. However, the electrolyte is contained in a closed, pumped loop with secondary containment, and at end-of-life the vanadium is recovered and reused (it does not degrade). Iron flow electrolytes are essentially non-toxic.
What is the round-trip efficiency over a real cycle?
For a vanadium redox flow battery: ~75% AC-to-AC after losses from pump parasitics, stack inefficiency, and inverter losses. Iron flow is similar. The number quoted by vendors is usually DC-to-DC, which can be 2–4 points higher than AC-to-AC.
How long do flow batteries last?
Calendar life is typically 20–25 years. Cycle life for VRFB and IFB exceeds 20,000 full cycles, sometimes much more, because there is no intercalation degradation. The electrolyte itself does not wear out; the membrane and stack components may need replacement once or twice across the asset’s life.
Can flow batteries be recycled?
Vanadium electrolyte is fully recoverable — the molecules do not degrade, so the electrolyte from a decommissioned battery is feedstock for the next one. Iron flow electrolytes are non-toxic enough to be repurposed as fertilizer-grade iron salts. This is structurally different from lithium recycling, where you must break down a complex solid composite.
What’s the difference between a flow battery and a fuel cell?
Both involve electrochemistry with liquid or gas reactants flowing through a cell. The difference is that a flow battery is rechargeable — the same electrolyte is reduced and oxidized cyclically — while a fuel cell consumes its reactants (typically hydrogen) one-way and must be refueled.
Do flow batteries work in cold weather?
Aqueous flow electrolytes have a viable operating temperature window of roughly 10–40 °C. Below 10 °C, vanadium electrolyte can precipitate; above 40 °C, the positive-side VO₂⁺ becomes thermally unstable. Modern systems include thermal management (heat exchangers, insulated containers) to hold the electrolyte in-window across climates.
Further Reading
- How Solid-State Batteries Actually Work (2026) — companion deep explainer on the other major next-generation battery architecture and how its trade-offs differ from flow chemistry.
References
- Skyllas-Kazacos, M. et al. New All-Vanadium Redox Flow Cell. J. Electrochem. Soc. 133, 1057 (1986). The foundational VRFB paper from UNSW Sydney.
- US Department of Energy. Energy Storage Grand Challenge Roadmap (2020, updated annually). Defines the long-duration storage gap and benchmarks targets.
- Pacific Northwest National Laboratory. Vanadium Redox Flow Batteries: Status and Outlook, various reports 2011–2024.
- Form Energy. Iron-Air Battery: Multi-Day Storage Architecture (technical white papers, 2023–2025) — for clarifying the iron-air vs iron-flow distinction.
- BloombergNEF. Long-Duration Energy Storage Cost Survey (2024, 2025 editions).
About the author. This piece is part of the IoT, Digital Twin & PLM deep-explainer series, where we unpack the engineering and economics of the systems that will run industry through the next decade. Subscribe for one carefully-built explainer per day on the technologies reshaping how the physical world is monitored, simulated, and powered.
