The gist
The assets in the Australian electricity grid are worth about $105 billion (and that’s ignoring the separate grid in WA). The grid and it’s ongoing maintenance is responsible for a bigger portion of our electricity bills than the generators which feed electricity into it. That $105 billion represents about $10,000 per customer. That’s not the replacement cost, I’ve no idea what that is, but it’s a convenient accounting figure after depreciation.
Chris Bowen’s plan to decarbonise using wind, solar and batteries won’t get rid of the grid. Instead, the transmission component, the wires on big towers across the countryside, will get longer, by about 10,000 kms, and be more complicated. In addition, the most expensive part of the grid, the smaller wires connecting homes and businesses to the transmission, will also get thicker (requiring more copper), more complex and expensive. Why? In areas with extensive rooftop solar and batteries, wires will need to be thicker and the transformers replaced to handle reverse flows of electricity at current levels not anticipated.
So that $105 billion will rise. Added to that will be a considerable sum spent by individual households on batteries and rooftop solar. So for each of these customers, there will be more than $10,000 embedded in the grid they use, plus the thousands embodied in their photovoltaic (PV) and batteries; possibly another $15,000-20,000.
Chris Bowen has, so far, fooled the Australian public into thinking that cheap electricity generation makes for a cheap electricity system. One commentator has put it beautifully, “renewables are a cheap way of generating very expensive electricity”.
The alternative is to maintain our grid in its current form, which is relatively cheap and simple, and install nuclear power plants where we currently have coal plants. I won’t explore that option in this piece, but will go into more detail about the mess Chris Bowen is creating.
Let’s be clear. I’m not a climate change denier, I spent many years following the battles over climate science and soon realised that the people with a really deep and broad understanding of the physics and chemistry were in furious agreement. They, in turn, had spent years genuinely trying to set up experiments (usually this meant designing instruments to put on satellites to measure stuff accurately) to disprove their theories in precisely the way that the best scientists do. If Bowen or the Government were serious about winding back our contributions to the problem they wouldn’t be ignoring the cattle and sheep industries, our biggest contributor to the problem now and for as many decades as climate change has been reasonably well understood. The hyperlink is to an article in the Age which I wrote 15 years ago with the then Professor of Climate Change at Adelaide University, Barry Brook, and Professor Peter Singer. It’s as true now as when we wrote it. If you think adding seaweed to cattle feed can solve the ruminant methane problem, then you haven’t thought about it. Ditto the Greens; still backing our biggest climate forcing (being ruminants rather than coal).
So I desperately want deep decarbonisation, but I prefer actual solutions over poorly considered wishful thinking. That said, read on if you want a little more detail on the trainwreck that is our electricity decarbonisation “plan”.
Who the hell is running this show?
Journalists can, perhaps, be forgiven for not knowing what they don’t know about energy when writing about it. But what about the head of the Australian Energy Market Commission (AEMC), Ms Anna Collyer? Consider the following quote from a story in the Australian Financial Review (AFR):
Australian Energy Market Commission chairwoman Anna Collyer said Australia’s 3.5 million household solar systems provided an equivalent amount of generation capacity into the national energy market as four Snowy Hydros. And with better management, such consumers could be the ”heroes on the road to net zero”.
Let’s see how many ways this claim is wrong. Always keeping in mind that perhaps the journalist misquoted her.
Snowy 2.0 will, assuming it is finished, be a 2,200MW/350,000MWh hydro system.
Why are there two numbers? They are power and energy respectively. They tell you that if the reservoir is full, it can supply 2,200 MW of power for 159 hours (350,000 divided by 2,200). Whether it can supply a lower power for longer is far more complex, but the answer is usually yes, but with caveats. If you are vague about the distinction between power and energy, then please read the Appendix.
Note the word “capacity” in the excerpt. It is annoyingly common for energy people to use the word capacity to mean power, not just when talking to other experts but when talking to the general public.
They know what they mean, but it gets pretty confusing for non-experts. It isn’t obvious that Collyer knows what she means. Capacity could easily be taken to refer to the amount of stored energy. But coming from an expert, it would mean power. Let’s assume that Collyer is claiming that the combined power of our 3.5 million rooftop systems is the same as the combined power of four Snowy 2.0s. I.e., that it is something close to four times the 2,200 megawatts (MW) of Snowy 2.0. i.e., 2,200 x 4 = 8,800 MW.
So is the power of our 3.8 million rooftop photovoltaic (PV) systems equivalent to 4 Snowy 2.0s? Not at night it isn’t.
How will better management help? It won’t; not at night, and not when it’s cloudy.
But what if by better management she means by adding batteries? It’s quite a stretch to spin spending billions on extra hardware as “management”, but maybe there’s a marketing qualification in her CV that I missed.
What would it cost to add a Tesla Powerwall 3 (5/13.5 kw/kwh each) batteries to each of the 3.5 million PV systems? Origin Energy is quoting about$14k for these. What’s 3.5 million times $14k? About $49 billion. That’s a lot of management.
The latest cost blowout has Snowy 2.0 costing about$12 billion. So if we built four and they all went over budget by the same amount, that would still be cheaper (just) than batteries.
But … and it’s a huge ‘but’, four Snowy 2.0s would store 1,400,000 MWh of energy compared to the 47,000 MWh that 3.5 million Powerwalls could store.
In short, it’s hard to imagine any definition of “equivalent” or “capacity” in which Australia’s 3.5 million rooftop PV systems was equivalent to four Snowy 2.0. Not at night, not when it’s cloudy and not even with $49 billion worth of batteries.
The actual number of batteries is rather less than 3.5 million. At of the end of 2022, the number installed over the previous 7 years was just 179,000, with a combined energy storage of less than 1,920 megawatt-hours.
What is stalling investment in “backup” infrastructure?
More statements from Collyer at the same AFR event indicates that her confusion about basic concepts goes far deeper. Consider this, a direct quote:
“The stronger our CER [consumer energy resources] is, the more confident we can be about supply, and the less back-up infrastructure we need to build at utility-scale.”
The quote comes in the context of widespread concerns about the lack of investment in utility scale “back-up” infrastructure.
What is a back-up generator? It’s a thing that hospitals have. Nobody who buys one actually wants to use it and they don’t care if they never use it. But now imagine you are investing billions in some kind of utility scale power or storage infrastructure. Do you want to use it? That’s a stupid question. You want it to be running and earning money for as big a proportion of the day as possible. If your facility is merely a back-up. You’ll want to be paid handsomely by somebody to have it standing by doing nothing. What could be worse than rooftop PV limiting the time your plant is earning money during the day? Yes, that’s right the only thing worse would be rooftop PV with batteries; eating into the time your plant is earning money at night. It should be blindingly obvious why nobody wants to build back-up plant. I’ll return to this point later.
Communal grid vs communal greed
Successive governments have exploited the genuine desire of Australians to “do their bit” to fight climate change and abrogated their responsibilities to decarbonise our grid (among other things!). I’ll only discuss our electricity system in this post, but most of what I say will apply to other areas as well.
Recall the cost of that 5kw/13.5kwh Tesla Powerwall battery; about $14k. Also recall that 3.5 million of these would be about $49 billion.
How does that compare with the value of our current electricity grid?
The Australian Energy Regulator (Australian Energy Regulator (AER)) has just release its State of the Energy Market report 2023. Chapter 4 is about the eastern Australian grid; which supplies electricity to SA, VIC, QLD, and TAS. This grid services 10.8 million customers. I’ll ignore WA for now and pretend that all of the 3.5 million rooftop solar systems are in the eastern states.
The assets, the physical stuff of the poles, wires, transmission lines and towers, the transformers, switches and stuff with names that few people understand, is worth (regulated asset value) about $105 billion. Divide that by 10.8 million and you get about $10k per customer (connection). That distributes electricity to everybody 24x7. If you spent an additional $105 billion, you’d get enough of those PowerWalls for about 7.5 of those 10.8 million connections; and you’d still need the grid for those PowerWall owners to extract money from the non-PowerWall owners by selling them electricity. So now you’ll have $208 billion worth of assets on the grid.
And what happens if people do add those batteries?
Firstly, the overall power of those batteries, when fully charged, is certainly substantial, about 37.5 gigawatt (GW), assuming you can get it from where it is to where it is needed. That could indeed power the entire National Electricity Market (NEM), keeping the distribution caveat in mind, for a little under 3 hours.
But, everybody will then also have to pay to upgrade the other $105 billion dollars worth of infrastructure to cope with the additional problems they have caused by putting generators and storage where they were never intended to be – on people’s roofs. Far from being a help, all this distributed power is a liability, because it pushes up the cost of the most expensive part of the electricity system; the grid (transmission and distribution).
But wait there’s more.
What Chris Bowen and the string of Energy Ministers before him haven’t been telling people, is that while utility scale solar is cheap, putting it on a roof isn’t.
Picture a B-double delivering a load of PV panels to a field where a few techies walk around (on the ground) and install it all. Easy. Now imagine a bunch of little trucks wandering around every little street delivering a few panels here and there and having a bunch of people crawling around on roofs and up ladders doing the installation; roofing has always been dangerous. And think about the layers of sales people also being paid. It’s chalk and cheese. It’s hard to think of anyway to make PV panels more expensive. Oh yes, and the batteries are definitely not included.
Lastly, who remembers any of the times Chris Bowen has claimed renewables are the cheapest way to make electricity? Neither he, nor the Commonwealth Scientific and Industrial Research Organisation (CSIRO) in its study, nor Australian Energy Market Operator (AEMO) in their Integrated System Plan (ISP) add in the costs to consumers of buying all the kit (batteries and PV panels) that used not to be needed. They aren’t talking about costs to society of the entire system, just about the costs to the Government and business.
The GenCost report only deals with “large-scale” electricity generation, not rooftops and batteries. The report on which it bases much of it’s costing is from Aurecon. It also doesn’t deal with rooftop PV panels. But, interestingly, its 2022 report does have data on the cost of both big (200 MW is it’s hypothetical target size) and residential sized batteries.
There are two big differences between the costs and technology that an energy company might roll out and the consumer grade products that Collyer’s heros would use. The cost for a 200 MW/400 MWh battery would be about $255 million dollars. That will buy you about 16,000 of it’s target residential batteries with an aggregate 80MW/160MWh of storage. Collyer’s heros are missing not just economies of scale, but the consumer grade batteries are only warranteed for 10 years; rather than 20 for the bigger gear.
In summary, instead of Governments doing what they should be doing and replacing coal with carbon-free generators feeding into a perfectly serviceable grid, the government and Ms Collyer want people to spend not just on panels but they are also expected to shell out $14k for a 10 year battery.
Aurecon have an interesting little note in their report:
Given the volatility of the RBESS [residential battery energy storage] market and observed problems with product quality, it is reasonable to assume that many RBESS products will not reach or operate beyond their warranted period. Given the volatility of the RBESS market and observed problems with product quality, it is reasonable to assume that many RBESS products will not reach or operate beyond their warranted period.
This is a pretty blunt assessment of the consumer grade junk in this market; but they didn’t name names.
Expanding the grid
South Australia has the most wind, solar and rooftop PV per person of any Australian state. How is the grid travelling with these renewables? There is only a single entity that manages the grid in SA, called SA Power Networks (SAPN). It’s Distribution Annual Planning Report 2022/23 to 2026/7 was released this year. Remember how we speculated about having a bunch of Powerwalls … millions of them? The grid is SA will not cope with this unless those batteries are limited in how much power they can export to the grid. Recall that our grid was designed to carry power from big generators along thick wires by ever thinner wires to peoples homes and businesses; not in the other direction! Carrying power in the other direction is called “reverse flow”. SAPN is pretty blunt:
“Zone substation reverse N and N-1 constraints that cannot be cost effectively mitigated through network augmentation.”
What does this heavily jargonised sentence mean? An “N contraint” is just a grid limitation under normal operation. An “N-1 constraint” is a grid limitation after a generator failure. Obviously when you lose a generator, you will get some unusual current flows and grids are typically designed to handle the biggest generator failing. The word “handle” here doesn’t mean nothing goes wrong, but that the system stays up; usually at the expense of some customers being disconnected; usually briefly and in a controlled manner. SAPN describes the current policy of allowing people to export 5kW to the grid as “unsustainable”. It foreshadows changes to reduce the amount of power people can export just as Ms Collyer is spruiking the addition of more power (and batteries) in peoples homes. Put simply the workers and management are running in opposite directions. Images of bulls and Pamplona come to mind.
The obvious analogy
Here’s what’s happening. Imagine you are in Europe in a place with an efficient diesel bus based public transit system and lots of old narrow streets. You start to encourage car ownership because you want to shut down the stinky diesel buses. People start buying cars and you start widening streets; at enormous cost. As the date to shutdown the buses draws nearer, you realise plenty of people still won’t have cars, so try to convince some schmuck to invest in a bunch of nice new electric buses for the odd occasion when somebody can’t drive because they have broken their arm, or are pissed. You want a back-up system! But, surprise, surprise, nobody seems keen to supply it!
Rooftop solar is to our electricity supply what the private motor car is to mass transit.
The situation is even worse than that. Imagine you are in South Australia, where there is already enough wind power on the grid to occasionally supply 100% of the demand (similarly with PV). Who wants to build more wind power in such a situation? At this point, any extra wind power will simply be curtailed. Your only hope is to persuade somebody to build another interconnector to move the extra electricity to somewhere with less wind power. That’s obviously just a short term fix.
Sure, a little extra wind could go to batteries. But there will never be enough batteries to soak up much of the extra wind; they are simply too expensive and guess what? Nobody wants to supply really big batteries that will only serve as backup. Why? Because who wants to spend money building things that are never (or rarely) used?
If instead of wind, you have invested in a dispatchable energy source, you might try lobbying to get somebody to pay you to have your system on-call. The jargon for this is a “capacity market”. The now sackedEnergy Security Board (Energy Security Board (ESB)) shows the fate of anybody trying to suggest such a thing to the people who reckon you can power the country by harvesting weather.
As coal plants retire, will they be replaced by wind and solar? Easy question, no.
Why not? On still nights it doesn’t matter how much solar you have.
On still nights wind power can drop to 5-10 percent of its maximum. So if you want a reliable 1 gigawatt, you’ll need 10 or more gigawatts of wind. Which will create the same problem … a bunch of very expensive stuff making no money for whoever built it – ultimately some bank. Banks insist on their loans being repaid.
What about batteries? Consider the calculations above. Multiply 10.8 million users by the cost of a consumer scale battery and you still won’t have a system anywhere near as reliable as the current grid. You might do better with centralised bigger batteries. Remember economies of scale? Those who favour distributed consumer grade grids seem to forget this; or more likely they are simply promoting a product they sell and they very much don’t want economies of scale for anybody.
In any event, it’s a pipe dream. There aren’t enough mines to provide the materials. Do we really love big mines that much? Remember that it isn’t just 26 million Australians wanting grid batteries, there are 100 million new cars a year needing them. Which brings us to “V2G” (vehicle to grid), another terrific way to ensure that none of the current $105 billion grid will be fit for purpose and that all of it will require a redesign and rebuild.
Building batteries and training AI models
The International Energy Agency (IEA) 2023 update to its Net Zero by 2050 scenario is very bullish on battery factories. The monster caveat being that mining expansion isn’t keeping up. But there is another caveat deserving of monster status. Look at the blue hatched areas in the image below.
The IEA reckon the current battery and PV factories are only running at half the capacity they should be. You should be able to double the output, by “simply” running the factories longer.
So the IEA reckon these factories should be running 24 hours a day 6 days a week instead of 12 hours a day 6 days a week.
Who wants to do the night shifts? Who cares? They are mostly in China anyway.
Assuming you can find the people, you still need electricity to run factories. Can you run a PV panel factory with PV panels? Provided you are happy to have your very expensive factory sit idle for long periods. At a recent AFR summit on Energy and Climate, cement maker Boral described how it is now more efficient to let 5,500 workers sit and do nothing when electricity prices spike; as they now do regularly in Australia. So they just stop work. This kind of inefficiency is spun as a feature of a renewable grid. It is called demand side management.
Northvolt is a Swedish battery maker, dedicated to making batteries with the lowest possible carbon footprint. It recently announced a deal to build a battery gigafactory in Quebec.
Quebec also recently announced that it planned to refurbish and reopen a nuclear plant. You’d be right to suspect a connection between these announcements. Canada designed and built its nuclear fleet with a technology called CANDU (Canadian Deuterium). Most of Quebec’s electricity comes from hydro dams which it built decades ago when displacing indigenous people was rather easier than it is today. There is no stomach for more hydro in Quebec, so nuclear is the obvious choice. In any event, there is plenty of 24x7 reliable nuclear power next door in Ontario; along with teams experienced in refurbishing CANDUs.
Given that most (about 80%) of the world’s battery factories are in China, energy won’t be an issue; it will be coal. In China most of the heavy industry is inland, whereas most of the nuclear plants are in the more heavily populated coastal provinces. China produces about half the world’s coal, as well as being the biggest importer. Can they, and will they, continue to produce solar panels and batteries for us if we or other countries stop selling them coal to do so?
In short, reaching Net Zero by 2050 (NZ2050) goals will depend on continued use of coal for 24x7 electricity and extending night shifts in Chinese battery factories. It will also be critical that China’s current economic malaise is sorted. Do we have time to spend a decade building the kind of vertically integrated battery supply chain that China has created over the past decade? We have allowed ourselves to be wedged in an unfortunate situation; which is more important, control over supply chains or rapid action on climate change?
And then there’s the AI revolution and it’s impact on global electricity needs. The field is young and growing fast. A2021 paper estimated that it took 1.3 gigawatt-hours of electricity to train the popular AI Chat GPT-3; over a period of 15 days. That looks relatively small!
But training the AI isn’t the same as using it. Training (and retraining) happens from time to time, but the real requirement for large amounts of reliable electricity comes when millions of people, from all over the world, start firing questions at your AI. One estimate is that AI alone could be using between 85 and 134 terrawatt-hours of electricity a year by 2027. How does that compare with other industries. A typical nuclear reactor has a power rating of about 1 gigawatt. So it will produce about 8 terawatt-hours of electricity per year. So 85 is a considerable amount! It’s roughly the same amount of electricity used each year to produce 6 million tonnes of aluminium.
So it should be no surprise that big data centres are increasingly announcing nuclear deals. Standard Power in the USrecently announced plans for 24 Nuscale SMR reactors for its data centres in Ohio and Pennsylvania. Microsoft also hasnuclear plans for its data centres and AI.
A 2018 Massachusetts Institute of Technology* (MIT) study* found that the nuclear part of building a nuclear power plant was about 20% of the cost. Fully half of the cost was in “civil works to prepare the site, including excavations and foundations, the ultimate heat sink (cooling towers or river cooling), other equipment, and the installation of plant components”. Multiple companies have taken heed. Most recently Last Energy has been signing to build reactors in Poland, the UK and Estonia. Each reactor is just 20 MW with all the components being made in factories and shipped in shipping containers. Site preparation will be trivial. Thorcon is a US company working in Indonesia with similar aims to build reactors that don’t need custom site preparation. They aim to build standardised reactors in ship yards. The IEA report mentioned above, has increased its 2021 target for nuclear as a result of the many signs of change in the industry. In 2021 it called for a doubling of nuclear power by 2050, keeping in mind that many older plants will be closing before then. It’s 2023 update adds another 100 or so big reactors (or equivalent) to the scenario. Not forgetting that the Chinese just keep churning out big reactors at a steady pace.
In contrast, Australia’s faith in renewables (without nuclear) is looking like an irrational outlier. We are looking like we haven’t learned anything from our experiences with renewables; our suspension of the market in 2022, the statewide SA blackout in 2016 and subsequent constant manual interventions in that SA grid along with the need for thousands of kilometres of transmission lines. Has nobody noticed that we seem to have replaced the traditional assumption that the grid should be reliable with the spin that we can “demand management” our way out of problems by just getting large companies to “simply” down tools (see Boral example above) when the grid is stressed? Is it acceptable to have the head of our major energy regulator being clearly ignorant of basic engineering concepts like power, energy and storage? Again the images of bulls and Pamplona flood in. We are looking increasingly like we are at Pamplona and running towards the bulls.
Appendix: Primer for the head of the AEMC and anybody else not clear on the basics of power, energy and storage
There are daily examples in media of all kinds where journalists report on energy issues without understanding the difference between power and energy. It’s not really that complicated. Your electric jug will have a power rating on it somewhere which will give, obviously, it’s power. Mine is 2,000 watts; which is pretty common. When I was growing up, jugs were smaller and typically were about 1,000 watts. Boil an amount of water in 1,000 and 2,000 watt jugs and the second will boil it twice as fast.
That’s what more power does.
But both jugs will use exactly the same amount of energy to do the job. Run a 2,000 watt jug for 1 hour and it will use 2,000 watt-hours of electricity. The term watt-hour is a unit of energy. What could be simpler, run a power source for a given length of time and the energy used is just the power multiplied by the time. Writing it with a hyphen between them makes it ever so obvious what is happening. 100 kilowatt-hours of electricity is the quantity of energy used by running:
a hundred 1,000 watt jugs for an hour or
one huge 100,000 watt urn for an hour or
one 1,000 watt jug for 100 hours.
What could be easier? This really should be done and dusted in pre-school. I’m assuming, of course, that most people get that “kilo” means a thousand, as in a kilometre is 1,000 metres. A megawatt (MW) is a thousand kilowatts, meaning a million watts, and a gigawatt is a thousand megawatts, ie., a billion which is 1,000,000,000.
So there really isn’t any excuse for a journalist reporting on a 100 megawatt battery project by only giving the power rating (100 megawatts). They should also tell you the energy. It’s usually enough to say how long the battery can run at full power.