The Australian Energy Market Operator (AEMO) has drawn up a detailed Integrated System Plan (ISP) for the eastern Australian grid (the NEM) out to 2050. Guardian journalist Graham Readfearn referred to it recently in a put down piece on nuclear power and battery requirements and costs.
He rebutted one estimate of nuclear costs with another, a CSIRO estimate. He was happy to detail the shortcomings of the first estimate while seemingly not considering that his preferred CSIRO estimate might also have issues. As it happens, the CSIRO estimate has been robustly criticised.
Readfearn is one of the best environment journalists we have, but why would I accept his technology costings simply because they came from the CSIRO? For the past 17 years, CSIRO has backed the most environmentally destructive diet on the planet, the high red meat and fish CSIRO Total Wellbeing Diet. CSIRO happily work for Australia's most climate damaging industry; the cattle industry. The CSIRO has thousands of scientists, and many are astonishingly brilliant, but many are just hired guns, and some are both! You need to look at the assumptions behind the numbers rather than rely on an argument from authority. The critique I linked to of the CSIRO numbers is by Ben Heard, I think it is damning. But I also accept that nuclear power is overly slow and expensive in the West.
But nuclear costs aren't a function of the technology, but of the regulatory structures and they can be changed. It takes 70,000-100,000 tonnes of stuff to make a gigawatt (GW) of solar photovoltaic panels (PV), but it only takes about 600 tonnes of stuff to make a 1.4 GW nuclear reactor pressure vessel. The rest of a nuclear plant should be, but often isn't, just normal thermal engineering. So there is no material reason nuclear can't be very cheap and in a future post, I'll consider the matter in detail. But this piece isn't about costs.
Costs come and go, and people argue interminably about them, but the physical reality is far more stable and worth understanding before you think about the money.
However, before moving to a discussion of the ISP and its problems, I can't resist pointing out that if we'd started building nuclear reactors when I first remember people telling me they are too slow to build, they'd be running by now.
The ISP is not a 100% renewable NEM, nor is it zero carbon
Let's consider the Integrated System Plan (ISP), the result of a bunch of modelling by our energy regulator, the Australian Energy Market Operator (AEMO) in consultation ... mostly with people wanting to make money from the system.
First, the ISP is actually not a plan; it's a set of four scenarios. The emphasis in the report is mostly on the "Step Change" scenario, so I'll stick to that in what follows.
Second, the ISP is also not a plan in the sense most people think of. It's not a list of things to do and a budget to do them.
Here's my favourite ISP quote and it convey's beautifully the nature of the "plan". The quote comes from a list of four big risks that worry the ISP authors:
the risk that storage of more than 8 hours duration takes longer than expected to materialise
Perhaps you need to remember Star Trek and transporter rooms to burst out laughing. I thought instantly about the crew waiting for people being "beamed up" to materialise. But there is no transporter; AEMO's only strategy is to convince people to become "players" in their market and to further convince these players that they can make a bucket of money if they do what the "plan" suggests. Nonetheless, it's convenient to treat this "plan" like it's a plan and its targets like there is a team and a budget to achieve them.
The next thing to stress is that the ISP isn't a zero-carbon electricity plan.
The ISP actually calls for an increase in the amount of gas capacity in the NEM; from about 7 gigawatts to about 10 gigawatts while using the generators a little less so as to reduce the amount of gas used from 530 TJ/day now to about 400 TJ/day (a TJ is a terajoule ... about 277,777 kwh) in 2050. The ISP explicitly says that the emissions from this gas burning can be offset or the gas can be replaced with hydrogen (I'll discuss this later). The ISP describes the gas plants as "peaking" plants, but the numbers imply that they'll be running flat out about 46 percent of the time.
The ISP increases the number of gas generators, but uses them less. How will people be persuaded to build generators that don't get used much? Governments are typically okay with doing things like this, but private industry? Not so much.
The peak for renewable generation on the NEM so far was on 15th November 2021 at 61.8 percent. The ISP envisages 9 times more utility-scale renewables and 5 times more rooftop solar by 2050, so clearly there will be times when we have far too much electricity. What will happen to it? When all the storage is full, about 20 percent of what is generated will be dumped; "curtailed" is the jargon for it. Persuading people to build wind or solar farms and not get paid for 20 percent of what they produce might be a little challenging. Batteries on the NEM currently make money by buying when electricity is cheap and price-gouging when it isn't in order to control the grid frequency. How will that work when there are vast amounts of storage on the NEM? I'm guessing we won't find out because car makers will consistently outbid utility suppliers for the requisite minerals for really big batteries. The crew of AEMO Star Trek will be waiting a long time for those batteries to materialise.
And by 2050, we will have ...
All up, by 2050, the ISP envisages 210 gigawatts (GW) of wind and solar + 61 or 46 GW of storage. Interestingly, the AEMO infographic says 61 GW of storage (in all forms) and the actual document says 46 GW (p.10). All that, plus an extra 10,000 km of transmission lines, will supply 320 terawatt-hours of electricity annually ... on top of the amount generated and used behind the meter, by owner-generators. I haven't quantified the grid rebuild and redesign which will be required. The 10,000 km of extra transmission lines may not be as simple to roll out as it is to plan. Farmers are generally not keen on big structures and their access roads across their land.
If instead, you wanted to supply 320 TWh with nuclear power, then instead of 271 GW of renewables and batteries, you'd need just 42 GW of nuclear plants plus another 4 or so to allow for refuelling outages and peaks. Let's round up to 50 since it obviously isn't cool to be concerned with throwing electricity out or the efficient use of equipment. So the renewable system needs around 5.4 times more GWs of stuff than the nuclear option; so nuclear could be 5-times more expensive per GW and still come out cheaper. Much cheaper. And if you fix the nuclear regulatory mess, then "much cheaper" could be "very much cheaper".
Why? Because a nuclear powered grid wouldn't need a total redesign plus robust rebuild with thicker wires; it wouldn't need electricity flowing in two directions, complicating fault handling. There would be no inverters for hackers to hack and no need for frequency stabilisation markets. There would also be no need for thousands of kilometres of extra transmission lines.
So any cost comparisons need to be mindful that you need considerably less nuclear power to replace vast fields of solar panels and wind farms cobbled together with thousands of kilometers of additional transmission lines.
There's the cost of solar and the cost of rooftop solar
Cost comparisons also need to keep in mind that the low costs of solar we are constantly reminded of are only ever about utility-scale solar farms; not rooftop solar. In the ISP, some 69 of the wind and solar GW are rooftop solar, making them very expensive; in more ways than one.
The latest IPCC report (p.6-23) notes that the panel cost is only about 30% of the installed cost of rooftop solar. For a homeowner, focused on their own bills, rather than the overall system, all you need to do to convince them to buy rooftop solar is that the amount saved on bills over the life of the system exceeds the cost of the system. The impact on total system cost caused by owners installing significant levels of rooftop solar is severe.
People acting in their own best interests won't necessarily make decisions that are optimal for the overall system. Given the ratio of installation cost to panel cost, then it's obvious that the total cost of 69 GW of solar farms would be lower than 69 GW of rooftop solar. It's an obvious economy-of-scale thing. But there's more to it than that. Suppose you want to build a huge energy storage plant. The technology is irrelevant, what matters is that you want a return on investment, or at the very least, to break even. But rooftop solar will attract household batteries; and every GW of household battery storage will be in direct competition with large bulk storage projects. The analogy with transport is obvious. Any significant penetration with private cars makes public transit less financially viable. In the end, public transit becomes a state service which nobody expects to be profitable. There's plenty more to say about this, but I want to go back and deal with that substantial gas component in the ISP.
Replacing that 400 TJ/day of gas with hydrogen
The ISP has a small sentence about the possibility of replacing the gas with hydrogen. Presumably green hydrogen. What if you were to generate that hydrogen with solar farms? It's not hard to calculate (see appendix); you'd need about 20 GW of PV ... which is about 1.4 to 2 million tonnes of panels on about 43,000 hectares of land. The Chevron Wheatstone gas project in WA translocated the wildlife from its project site. Over a period of about 7 years, they moved 30,000 animals. The site area? About 1,000 hectares. How much of this is required for green hydrogen solar farms will obviously depend on the location, but I'm betting that no solar farm will go ever to the trouble that Chevron went to for its gas plant.
But hydrogen to replace the gas is only mentioned as a possibility by the ISP and it would be in addition to 58 TWh of hydrogen already implied in the Step Change Scenario (p.31). If you were producing this 58 TWh with solar then you'd need 28-42 GW of PV. I'm assuming that AEMO factored this in somewhere in the modelling, but I couldn't find it.
Creating new kinds and levels of inequality
Just as motor cars reduce the profitability of public transport, household energy systems reduce the financial feasibility of big utilities (private or public). Household systems reduce the efficiency of the entire electricity system. Not just a little bit, but a lot. The AEMO Renewable Integration study contains a table showing the problems being caused at every level of the grid by rooftop solar. But in the ISP there are no problems, just opportunities. The simplest opportunity to explain is overheating (aka thermal overload). You can't have clusters of 6kw solar systems feeding into cabling designed for much lower loads and expect it to cope. So over the next 20 years, a vast rebuild will be required. An engineer's problem is a market operator's opportunity. The words "opportunity" appears 64 times in the ISP, and often in the plural! Why is that not surprising.
By 2050, the ISP envisages 65 percent of detached houses will have humungous PV systems on their roofs and batteries in their garages, along with their EVs. Electricity used to be a public service supplied on a cost-recovery basis. It is fast becoming a wealth and virtue signalling market opportunity for those keen on being "the smartest guy in the room". The Enron culture is winning. The ISP assumes all those households will be happy to contribute to the grid (via automated appropriation and control), but what if they decide they'd rather be independent and live off-grid? Living off-grid has always been a dream of those who see a virtue in self-sufficiency; courtesy of tens of thousands of dollars of high tech equipment supplied by vast multinational juggernauts pulling in mineral resources from all over the planet.
A few people disconnecting will be of little concern, but if it becomes a trend, then we will see exactly what we see in the transport sector; with public transport being more used by the less well-heeled members of society who have no other choice. The inefficiencies in such a system will mirror those of our car-centred culture, and could even scuttle the ISP's distributed energy vision.
The ISP may not be a plan, but it is a recipe for yet another mode of inequality; the big battery haves and have not. Big batteries will be the logical successor to the SUV. They might start as a status symbol and end up as the norm. Will "vehicle to the grid" stop this from happening? You'd only believe that if you believed that one SUV in the drive stopped people from getting a second.
Appendix: Hydrogen from PV panels
When you burn a tonne of hydrogen you get about 39 megawatt-hours of heat (not electricity). You can get pure hydrogen by splitting water, but it takes about 60 megawatt-hours of electricity to get a tonne of hydrogen this way currently and about 50 megawatt-hours is the best anybody will ever do. Whenever you turn heat into electricity, you lose energy in the process; a lot of energy. So you don't get 39 megawatt-hours (of electricity) from your tonne if you use it for that purpose; you might get 15-25 megawatt-hours. So if you use electricity to make hydrogen which you then burn to make electricity, then you put in about 60 megawatt-hours and get back 15-25 megawatt-hours. It's an incredibly inefficient way to store electricity. On the other hand, making hydrogen which is then used for its heat rather than to make electricity, only involves one very lossy conversion instead of two.
How many GW of PV do we need to generate 400 TJ/day (TJ is a terajoule... 1e12 ... meaning 1 with 12 zeros after it) worth of hydrogen; the amount the ISP suggests? The energy density of hydrogen is 140 MJ/kg (MJ is megajoule). So 400e12 / 140e6 / 1000 = 2857 tonnes and it takes 50 MWh to make each tonne. A GW of PV would produce between about 4.8 and 7.2 GWh of depending on whether the capacity factor was 20% or 30% (and in Germany, it would be about 10%!) electricity per day so we need between 20 and 30 GW of PV for the job. For example, with 30% capacity factor we'd be getting 1e9 x 24 x 0.3 = 7.2GWh per day per gigawatt of PV, so we'd need 2857 x 50e6 / 7.2e9 = 19.84 GW of PV.