Running an aluminium smelter with wind, solar and batteries

South Australia got over 70% of its energy from wind and sunshine in 2024. What if we diverted all our renewables to running a single aluminium smelter?

Powering an aluminium smelter on wind, sunshine and lithium

According to the Australian Financial Review, Alcoa and Rio Tinto have different views on whether batteries could backup the requirements of an aluminium smelter. Alcoa says no, but Rio Tinto, they report, says yes.

Rio Tinto seems to be having a bet both ways, because it is also seeking a bail-out of its Tomago smelter saying:

Top Tomago executives have repeatedly warned that spiralling electricity costs and the unavailability of consistent and affordable renewable alternatives were threatening to put it out of business.

The Tomago smelter needs a nice steady 950 megawatt (MW) power supply. South Australia (SA), for comparison, typically needs 900-2,000 MW during the winter and up to 3,500MW during summer heatwaves. But that’s for about 1.7mn people.

What if SA dedicated our (yes, I do live in SA) entire electricity infrastructure to running Tomago? We currently have 2,763 MW of wind turbines, 735 MW of utility scale solar farms, and 1,510 MWh of battery storage; not to mention our 2,173 MW of home PV panels.

Can we get a steady 950MW with that 5,671 MW of energy collection and 1,500 MWh of battery storage? [Technical note for non-geeks: note the ‘h’ on the end of ‘MWh’. A megawatt-hour (MWh) is a unit of energy. It’s the energy used when you use 1,500 MW of power for one hour. Similarly, 3,000 MWh is using 1,500 MW for 2 hours, or 1,000 MW for 3 hours, or 3,000 MW for 1 hour. ]

In 2023 SA got 71.1% of its electricity from wind+solar+batteries and in 2024 it was higher; but not by much; just 71.9%. We seem to be hitting a ceiling.

Powering Tomago

It’s not complex. Tomago runs day and night, but all the solar in the world does nothing at night; unless you buy a fleet of batteries. How much battery storage would you need and what would it cost? Keep reading; I’ll get to that.

What about all those SA wind turbines?

As you add more wind power, you get overproduction when it’s windy and sweet bugger all when it’s still. What do I mean by “sweet bugger all”?

Last week (ending on 13th of June) in South Australia was a good example.

Look at all that green! That’s wind power. You can see that for the first 2 days and nights of the week we got pretty much 100% of our electricity from wind power. Running Tomago for those days would be a breeze (pun intended).

Day 3 was a little windy. But then we had 4 days of very little wind. On some nights there was virtually nothing.

To keep the lights on, SA needed prodigious amounts of gas and some imports from Victoria.

Similarly, Tomago would have been buggered without gas and imports. The issue, of course, isn’t to just power Tomago, but to power it with no carbon dioxide emissions. There’s much more to decarbonising aluminium production than using carbon free electricity. Estimates vary, but it’s typically less than half of the total decarbonisation problem; and it’s the easy half.

Looking back at the image again, you can see the gas and imports that keep the lights on in SA. The different shades of light orange are different kinds of gas turbines. The purple at the bottom is electricity being imported from across the boarder in Victoria (probably from coal).

The above image is from OpenElectricity. Here is a replot of the data to make the critical features clearer. The top line is demand and the lower green line is the wind power output. We’ll discuss the orange area later.

Look at the green line on the night of June 10-11. The wind dropped incredibly fast at sunset and stayed at sweet bugger all for the rest of the night.

So, on that night, June 10-11, Tomago would have needed a good 12 hours of power with virtually nothing coming from wind or solar, despite thousands of MW of wind and solar capacity. 12 hours at 950 MW is 11,400 MWh of energy. We have, as I said, about 1,500 MWh of batteries, so we would have been 9,900 MWh short. That last statement is wrong; think about it. By that 3rd night of low wind, the 1,500 MWh of batteries would have been dead flat, so we’d have been the full 11,400 MWh short.

In nice round figures, big batteries cost about $500mn for 1,000 MWh. So Tomago would need a minimum of $5.7 billion worth of batteries to get through a single low wind night, assuming the batteries were fully charged by sunset.

Is $5.7 billion a big number for an aluminium smelter? Tomago produces close to 600,000 tonnes of aluminium annually and the price of aluminium is about $2,500/tonne. That’s about $1.5 billion worth of product. So no, Rio isn’t going to buy $5.7 billion worth of batteries anytime soon. Especially not with the short lifespan of a battery.

The role of “big” batteries in the South Australian grid

Batteries won’t be powering aluminium smelter in the foreseeable future, but what is their role more generally? Let’s ignore Tomago for a bit and think about powering South Australia again.

Look at the left hand axis of the above chart. 1000, 2000, 3000 … those are megawatts (MW). We have over 2,763 megawatts of wind turbines in South Australia, but on the night of the 10th of June, as we’ve said, there was hardly a single moving blade. For over 8 hours that night and into the next morning those turbines produced about 3% of their maximum output during this week. And that maximum was about 2,000 megawatts, quite a bit less than the 2,764 MW that you might think would be the maximum.

The orange shaded area is the shortfall (the demand minus the total amount of wind, solar and battery power. Recall there is about 1,500 MWh of battery storage in SA. But the accumulated shortfall by the end of just this single week is 102,200 MWh (102.2 gigawatt-hours). Meaning you’d need 100 times more than our current level of batteries. That’s about $50 billion; rather a lot for a state of 1.8mn people. And that’s just to cover this particular week. What if you had a longer run of still weather (like we had in winter 2024)?

The difficulty with adding more batteries, apart from the breathtaking cost, is two fold.

1. You’ve got to put them in the right places. You can only charge them when you have excess and when you have excess, your transmission lines would soon be saturated and not able to move all your power. Popular tweets about how few batteries you need to firm up the Australian grid have no basis at all in reality. The assumption that you can magically move electricity across the continent into and out of storage without transmission is both false and profoundly misleading.

2. Profitability. Nobody wants to spend millions, let alone billions, on batteries which are hardly ever used. But that’s exactly how it works. During the week shown above, the 1,500 MWh of batteries operated at 1.4% of their maximum output. That’s roughly $750 million dollars worth of batteries sitting round doing nothing for 98.4% of the time.

How then do big batteries make money?

Big batteries don’t make money from selling electricity at its cost plus some profit margin; they make money by price gouging. They only supply electricity when there is a shortage (or to prevent the grid falling over), and when there is a shortage, given our market system, the price is sky high. The worse the shortage, the more they can and do charge.

If you own a big battery, then the beauty of a wind+solar system in the context of our dysfunctional electricity market is that it guarantees frequent shortages (as you can see by the graph above). It guarantees that the system will be on its knees and allowing you to make windfall profits on the back of our desperation to avoid blackouts.

AEMO’s Quarterly report for Q1 2025 puts the average price of electricity from batteries at $215/MWh, compared with $84/MWh from coal and $153/MWh from gas.

Batteries only have to compete with each other. The other generators supplying during a shortage love it when batteries are price gouging. Because they get the same high price as batteries during the periods that batteries set the price.

Here’s how it works.

Prices are set every 5 minutes on our grid using a bidding process. Suppose you need 2,000 MW during the next 5 minutes in South Australia. Wind and solar farms say how much they can provide and at what cost; as do the gas plants. If the most they can come up with, due to a lack of wind or sunshine, is 1,950 MW then a bid by batteries will be accepted for the last 50 MW. If a couple of batteries bid at $215 to fill the last 50MW, they will both prevent a blackout and delight everybody else who bid, because everybody gets $215/MWh for the next 5 minutes. So everybody loves shortages. The only thing they like more than shortages is batteries filling those shortages. Wind, solar and batteries are a match made in heaven; in the context of our market bidding and payment rules which reward everybody for price gouging.

Duplication and inefficiency

South Australia has, in effect, multiple full sized electricity systems. We have enough rooftop PV to supply over 100% of our electricity when the sun is shining during summer. We have enough wind turbines to provide, when it is windy, 100% of our electricity, except on days of peak demand in summer. We have enough gas turbines to provide almost 100% of our electricity when the sun isn’t shining and the wind is still. We have enough interconnector capacity to handle modest shortfalls of wind, solar and gas and will shortly have another $4.1 billion interconnector to provide even more backup capacity.

But despite these mountains of electrical generation capacity, we are still only getting ~70% of our electricity from wind+solar+batteries. Guess what will happen when we build more wind farms? When it is windy we’ll have more excess excess, but when it isn’t we’ll be short.

When the wind isn’t blowing, doubling or tripling the number of turbines just means you have double or triple the number of stationary blades.

Our new $4.1 billion interconnector will allow us to export some of the wind excess, but mostly we’ll be throwing it out. Why? Because “big” interconnectors, like Project Energy Connect (PEC), aren’t actually very big; they are just very expensive. PEC, when complete, will only move about 800 MW.