Three hundred and sixty megawatt-hours of heat. Thirty megawatts of discharge. Seventy-eight megawatts of charging. Twelve hours of continuous steam at full output.
Those four numbers describe the unit Rondo Energy and Rio Tinto have just begun engineering for the Yarwun alumina refinery near Gladstone, Queensland. The announcement landed on 21 October, and the technical shape is simple: a heat battery that charges on cheap electricity and discharges as process steam into a plant that already runs around the clock.
The money is stranger than the machine. ARENA, the Australian Renewable Energy Agency, lists A$74.9 million of funding against a A$125.19 million project. Only A$3.6 million is available now, for front-end engineering and design. Rio Tinto matches that A$3.6 million towards the study. Another A$38 million is staged for deployment if the study clears a stage gate review, and A$33.3 million sits in conditional contingency for cost overruns and electricity prices.
Read the three funding lines together and the announcement becomes a statement about risk.
What the battery actually replaces
Yarwun digests bauxite into alumina, the feedstock for aluminium. The digestion step needs high-pressure steam every hour of the year. Today that steam comes from three coal boilers and a gas-fired cogeneration plant, according to ARENA's project record. The site is brownfield: steam mains exist, electrical capacity is finite, and the heat load does not follow the sun.
Rondo stores electricity as heat in refractory brick at about 1,100°C. Resistive elements charge the brick. Stored heat leaves as hot air, passes through a heat recovery steam generator, and rejoins the plant as process steam. The unit slots into the existing steam network. Refinery processes run as before; only the energy source changes.
At full discharge the battery supplies 30 MW of thermal output for 12 hours. ARENA expects it to cover roughly 5% of Yarwun's steam demand and cut Scope 1 emissions by about 67,000 tonnes of CO₂-e a year.
Five percent is the honest number, and it is the one to hold onto.
A single unit at one refinery is a cost experiment. Its value is the price signal it produces for the next one.
Heat battery capacity at Yarwun
30 MW thermal discharge, 78 MW electrical charge, 12 hours at full output. ARENA project record, 2026
Forecast Scope 1 reduction
Onsite emissions avoided at Yarwun if the project proceeds to operation. ARENA, 2026
Why five percent is the right size for a first unit
Divide the A$125.19 million project cost by 360 MWh and you get about A$348 per kilowatt-hour of storage capacity. That figure covers the battery, the steam generator, electrical and control infrastructure, grid and site integration, commissioning, performance testing, and 36 months of trial operations. It is an integration price, not a cell price, and integration is what the next project will actually face.
Context matters for the denominator. Alumina refining in Australia emits about 14.9 million tonnes of CO₂-e a year, roughly 3% of national emissions, according to ARENA's record for an earlier Yarwun hydrogen program. Heat dominates that footprint. Electrifying heat is harder than electrifying road transport: the temperature is high, the load never stops, and the equipment sits inside a working plant.
So what does a bankable industrial heat project look like when the incumbent fuel is coal and the underwriter is a public agency? This one looks like a paid engineering study with a deployment option attached.
The contingency line is the honest part
Contingency is usually a footnote. Here it is 45% of the headline. ARENA has approved up to A$33.3 million that unlocks only in defined circumstances — cost overruns and electricity costs. That is an explicit statement that the two variables which decide the project are construction cost and the price of power.
The price of power is the sharper of the two. Rondo's model depends on charging during the four to six hours of surplus midday solar that Australia regularly produces, then selling steam into a plant whose alternative is coal and gas. Charging that drifts into the evening peak narrows the arbitrage. Higher network charges raise the operating cost with it.
Darren Miller, ARENA's chief executive, described the project as a test rather than a build. "Industrial heat remains one of the toughest decarbonisation challenges facing heavy industries," he said. "This project will test whether thermal energy storage can provide a practical pathway to lower emissions from industrial heat in a real-world operating environment, where reliability, performance and integration are critical."
Armando Torres, Rio Tinto's managing director for Pacific operations, was equally careful. "This study will help us determine whether thermal energy storage can provide a technically and economically viable alternative to fossil fuels for part of Yarwun's steam requirements," he said.
The 78 MW charging load is the unglamorous problem
Charging a 360 MWh battery in a few hours means a 78 MW electrical load. ARENA's project scope names electrical and control infrastructure, grid and site integration, commissioning and performance testing as separate line items. On a brownfield site, that is where schedules slip. A new connection of that size can take years in parts of the Australian grid, and the refinery's existing supply was sized for its own load rather than for a battery that charges at midday and discharges all night.
Rondo's pitch is that the battery is a drop-in replacement for a boiler. The electrical side is not drop-in. Someone has to fund the switchgear, the protection systems, and the connection agreement, and those costs sit inside the A$125.19 million rather than in the battery price.
What Rondo has already run
Yarwun is not Rondo's first unit. The company says a 100 MWh heat battery has run in daily automatic operation in California since October 2025, charged by an on-site solar array and supplying steam to a fuel production facility. A second 100 MWh unit for HEINEKEN's Portuguese brewery, backed by EDP and a €75 million European Investment Bank and Breakthrough Energy Catalyst package, is due next.
The step from 100 MWh to 360 MWh is 3.6×. The step from a fuel facility to an alumina refinery is bigger than the ratio suggests. Three coal boilers, a gas cogeneration plant, and a steam network that cannot tolerate interruption make this a systems test as much as a storage test.
As we wrote in October, Epyr raised €15 million to electrify industrial heat with firebricks. Epyr sells a heat battery too. The difference here is the counterparty: a listed miner with an operating refinery, and a public agency underwriting the integration study.
Can the unit deliver steam at a cost per tonne below the coal and gas baseline?
Does the 78 MW charging load fit the site without new grid capacity?
Do 36 months of trials produce data another refinery can underwrite?
The clock runs to 2033
ARENA dates the project from 21 October 2026 to 30 April 2033. That window includes the feasibility study, construction, commissioning, performance testing, and 36 months of trial operations. A commercial decision at Yarwun therefore lands years before a second unit can be ordered on the strength of it.
That timeline is the real constraint. Heat batteries compete against boilers that are already paid for. A five-year evidence cycle is slow for a technology that needs to be cost-competitive this decade, and it is the price of doing the test inside a live refinery instead of a pilot hall.
Who carries the integration risk
Rondo sells heat batteries as capital purchases or leases, and offers a heat-as-a-service model where the customer pays for steam delivered. Yarwun's structure is closer to a co-funded demonstration. Rio Tinto leads the project through RTA Yarwun Pty Ltd, matches the feasibility funding, and holds the final investment decision. Hatch and GHD are named as engineering partners on the ARENA record.
That split matters for what happens after 2033. A service contract would have pushed performance risk onto Rondo. A co-funded study keeps it with the refinery, which is why the contingency line exists and why the language from both sides stays conditional.
Five percent of Yarwun's steam is the size of a first unit that a miner and a public agency can both defend.
The number that decides the next project is cost per tonne of steam against three coal boilers, not tonnes of CO₂ avoided.
Yarwun's coal boilers will keep running through the study. The battery's job is to make one of them optional by the time the stage gate closes. If the cost per tonne of steam lands below the fossil alternative, the next five percent gets easier to finance. If it does not, the project ends as an expensive data set — a cost ARENA has already written into the contingency line.