All kinds of people think that Peter Dutton’s plan for nuclear power in Australia is a massive distraction.
Including me.
Plenty of climate action campaigners think that the Liberal party’s born again net-zero nuclear rollout plan is highly suspect, given their record while in Government and on-going support for coal from those backing Matt Canavan. On the other hand, Christopher Pyne, long time Liberal Party number cruncher, reckons the battle over climate denialism has been won by those backing the science. I'm hoping he's right. Still, the rotting carcase of intra-party feuds always has a whiff about it.
But how much time and energy is being wasted by journalists and the moribund anti-nuclear movement revitalising long dead “debates” about water or waste? Nuclear has been operating at scale for decades in multiple countries. It works, get over it.
This faux debate about nuclear is a massive deflection from the real issue: the massive First of a Kind (first-of-a-kind (FOAK)) open-cut renewable pit mine Australia is digging for itself.
It’s still the grid stupid!
Just over two years ago I wrote a piece called “It’s the Grid Stupid!”. It outlined the mass of new work that was (and still is) needed to solve all the problems with designing (let alone building) a new kind of grid; a grid that didn’t and doesn’t exist anywhere on the planet.
Back in 2017, well before my little blog piece, Australia’s Chief Scientist, Alan Finkel, did a review of the National Electricity Market. I remember him talking during interviews about modern grids as being the largest and most complex machines ever built by humans. Others have used the same language. And yet we are trying to dump it in favour of something about which only one thing is agreed; it will be vastly more complex.
Nobody else is doing this. They are decarbonising their grids without changing them too much.
The US already gets 20% of its electricity from nuclear, the EU gets about 25%, so both can expand nuclear and bolt on a bunch of renewables and batteries without changing the basics. They will still have grids stabilised by large synchronous generators, meaning big lumps of spinning metal whose physical inertia magically solves a bunch of really hard problems as if by magic. Dumping this is like replacing a beautifully counter-weight balanced 100 tonne bell that a child can ring with a collection of internet hackable microchips and power electronics; and doing it fast.
States in the US, like California, and countries, like Germany, may be talking the 100% renewable talk but both have the ready option of bludging on nearby jurisdictions that can rescue them if they get stuck; much as South Australia is currently rescued daily by Victorian coal. Germany can also just burn more forest to keep synchronous generators in the mix, or rely on connections to France; or just revert to coal.
Australia is unique in going it alone in trying to design and build this new style of grid; despite the urgency and the long history of even small first of a kind projects running late. The hubris of a manufacturing minnow of a country like ours trying to do this is astonishing.
In “It’s the Grid Stupid!”, I focused on Australian Energy Market Operator (AEMO) engineering reports (circa 2020) on all the problems which needed to be solved and the new software that needed to be written to solve problems, some of which had never been solved before. I also drew attention to the deep physics and old computer code underpinning the current grid modelling software used to check what changes in the grid would do to stability.
Shortly after that 2020 Renewable Integration Study: Stage 1 (RIS) report, AEMO joined the G-PST (Global Power System Transformation Consortium), a bunch of the world best and brightest regulatory labs, United States National Renewable Energy Laboratory (NREL) from the US, VTT from Finland, Imperial College London, Commonwealth Scientific and Industrial Research Organisation (CSIRO), EPRI, IEEE, etc.
G-PST is pretty much everybody who is anybody in the world of people who don’t actually make anything.
Where is Siemens? Where is Hitachi? Where are the Chinese companies who build so much of everything? Where is Huawei (about 30 percent of the world’s inverters currently)? Where are the Indian giants like Sungrow (23% of inverters)? None of the manufacturers listed in the chart below is part of G-PST.
Some of the research involves working out how to assess the stability of clusters of inverters. Inverters are increasingly complex pieces of electronics and software sitting connecting the grid with solar panels and wind turbines. The major fear is a failure which cascades because of interactions within the cluster.
In June 2024 in a CSIRO report on the state of progress on G-PST, the authors write:
“Prior work by AEMO, UNSW and Solar Analytics has shown that distributed photo voltaics (DPV) inverters can respond en masse during major power system disturbances. Synchronised generation reduction across large swathes of the photovoltaic (PV) fleet following disturbances presents a material risk to power system security.”
This isn’t just a theoretical possibility. A 2021 AEMO report presented data on the percentage of inverters spontaneously disconnecting from the grid in response to some disturbance. It anonymised the inverter brands. Why? In any event, the worst was Brand X where between 20% and 70% of inverters disconnected, depending on the event.
So … four years after the serious practical problem of cascading inverter problems was discussed in an AEMO report, it is still being discussed as a problem to be worked on … not a problem which had been solved. This problem of cascading inverter failures actually goes back much further … to Germany in 2005. German inverter specifications required inverters to disconnect when the electrical frequency hit 50.2 hz. Which they did. Except that masses of disconnections pushed the frequency higher causing a cascade of failures. How could one of the strongest engineering nations on the planet get it so wrong? Plenty of things look blindingly obvious in hindsight!
Twenty years later, engineers are still discussing these kinds of problems as serious and urgent; and unsolved. In contrast, it was just 21 years from the first demonstration of nuclear fission to having a nuclear powered submarine circumnavigate the world underwater. People's understanding of which problems are hard and which are easy doesn't always align with reality.
Who remembers the Optus outage in 2023? A cascading disconnection of routers took millions of people offline. Imagine this in our electricity grid; it’s not a pleasant thought.
A major impediment in tackling this problem is that the good folk at G-PST don’t know how inverters work.
They can certainly buy them and pull them apart and look at how they are made. They will have deep knowledge of the principles. But they didn’t write the code and they don’t have access to it; meaning they can’t predict precisely how any inverter will operate under given conditions without testing each and every model under those conditions. Which means they can’t model their operation.
So the techies at G-PST are working blind.
So one preliminary task the G-PST people have set themselves is to treat inverters as black boxes and design tools for testing how they work well enough to be modelled accurately. Meaning that they test (every) inverter under a set of test conditions and use that to build predictive models. Be assured that's even more complex than it sounds.
In comparison, AEMO (another member of G-PST) doesn’t rely on such methods when it models the impacts of new additions to the grid. AEMO has negotiated deals to get inverter software that they can use without reading it or otherwise exposing the proprietary details. For techos who love the details, they run the code in virtual computers firewalled from anything other than carefully controlled access to the inverter behaviour. So why doesn’t G-PST do the same thing? They want to develop simpler analytic approaches to modelling inverter behaviour that are faster than exhaustive physics simulations used by AEMO connections software.
There are good arguments for both approaches; the bottom line is that this is dazzlingly complex.
There are, of course, national and international standards which are intended to govern the operation of inverters, but the degree of adherence to standards has a patchy history. Back in 2019, some of the standards had compliance levels of about 10%; meaning 90% failed to work as required. Fast forward to 2023, and things have certainly improved. But inverter makers sell globally and need to introduce an array of settings to handle national differences. So now the biggest problem isn’t clear inverter violations of standards, but the failure to have the correct settings made at installation. In the 2023 survey cited above, 55% of inverters with visible settings had them incorrectly set in some way. Some inverters have invisible settings so the audit was unable to confirm what settings were applied.
As the audit summarised:
“In the field, compliance with technical settings is poor; a wide range of data sources consistently indicate that less than half of systems installed are set correctly to the required standard. This has been observed consistently in measures of compliance to specific settings within the 2015 standard, and also in compliance to the 2020 standard. This suggests significant deficiencies in governance frameworks for monitoring and enforcing compliance with technical settings in the field. ”
And even if your inverters are working fine, will the next software version of inverter code change something critical? That problem too is discussed in the G-PST.
“Compliance with standards, regulations and other related performance standards needs serious consideration given the potential for on the fly updates that could change wither knowingly or unknowingly the response of individual inverters, potentially no longer conforming with standards and regulations”
Are inverters really that important? CSIRO and everybody else at G-PST seems to think so. Of the 9 key topic areas worked on in 2023/4, three featured inverters: advanced inverter applications, analytical methods of determining stable inverter operation, the role of inverter based resources in system restoration.
How hard can it be to turn something on?
What is “system restoration”? It’s the challenge of restarting the grid should it ever fail …i.e., how to turn it back on again.
Believe it or not, we are building a grid without understanding how to restart it if it ever falls over. How hard can that be? Isn’t there a switch on the wall?
Here’s the research brief:
“ … to develop a research plan for creating new procedures for black starting and restoring a power system with high or 100% inverter-based resources (IBRs) penetrations. This topic is not only about black starting a single IBR plant, but rather the careful sequencing of restoring the entire system while simultaneously ensuring stability, security, proper protection, etc. This research topic focuses on developing 1) grid code specifications for IBRs to be able to black start, and 2) methods and procedures and analysis techniques to black start and restore a power system with various penetrations of IBRs, up to 100%. The challenges include increased complexity of system restoration due to 1) imperfect information of instantaneous capability of variable renewables, 2) protection concerns due to lower and different fault current behaviours of IBRs, and 3) changed load due to potentially high penetrations of different types of distributed energy resources (DERs).”
Note in particular that they, correctly, assume imperfect information about various renewables, different [fault current] behaviours of wind, solar and batteries (all are IBR … Inverter based resources). What's "fault current"? It's the electrical current when something is broken. Imagine sticking a screw driver in a power socket (but don't do it!), there is a surge of electrical current. It's the surge that instantly triggers your safety switch and saves your life. Outside your house, you get something superficially similar when a tree falls on a power line. Paradoxically, you need the surge to trigger the safety mechanisms which cut off the current. Wind and solar devices do things very differently and break the entire conceptual framework of fault discovery, isolation and protection.
Did you notice the ellipsis "..." at the beginning of the quote? It means I've deleted some preceding text.
The missing text tells you that it isn't CSIRO who will be undertaking this project. They have contracted it out.
There are about 5,500 scientists at CSIRO but they contracted this work out to a specialist engineering company; Aurecon. Aurecon is an Australian based, but global, group of companies. It has more employees than CSIRO.
The implication? The problem of restarting a 100% renewable grid is too hard for Australia’s top scientific research organisation; they can do quantum computing, but can’t turn the lights back on if they turn off. Clearly, if you think restarting a new kind of grid is easy, you are wrong.
Like I said in the beginning. The arguments over nuclear are a terrible distraction. We know nuclear works … everywhere. What we need to think much harder about is our new hackable, complex, fragile, ceiling wax and string, FOAK renewable master plan.
Of hackers and the grid
The grids of the 1950s had no software in them. Typically, electrical engineers aren't well versed in software. It's one thing to write a little code, SCADA interfaces and the like, but quite another to understand the intricacies of security systems. It's neither desirable nor possible to return to software free grids and it's inevitable that grids will evolve to have more rather than less software. But every time we allow a grid device to have an internet address, we open a container load of worms. Hackers gaining access to baby monitors is a little scary, but it doesn't compare to a denial of service attack on the grid via inverters. I was amazed to see in the most recent (2020) inverter standards (AS/NZS 4777.2:2020 ... amended in 2024) that they haven't progressed past what the IT security geeks call "security through obscurity".
"Changes to the settings shall require the use of a tool and special instruction not provided to unauthorized personnel."
That's a sentence which takes me back to when train doors were secured with a key that was a square bolt; how quaint. Modern internet security is premised on requiring systems to withstand attacks from people who have access to perfect information about the system they are attacking; including the source code. Assuming you can make something secure by hiding the instructions shows an astonishing level of naivety. The CSIRO have assigned 1 junior and 1 senior person to spend 3 years producing various reports into what the risks are how to mitigate them. Clearly, they haven't got a clue.
I should note that DCCEEW has been working on a cybersecurity framework for some years. And AEMO is also now involved. But it seems that the CSIRO G-PST people missed the memo and are a few steps behind.
Of seashells and science
The great, if a little nutty and nasty, Isaac Newton put his finger on a fundamental feature of (many) scientists as people:
“I do not know what I may appear to the world, but to myself I seem to have been only like a boy playing on the seashore, and diverting myself in now and then finding a smoother pebble or a prettier shell than ordinary, whilst the great ocean of truth lay all undiscovered before me.”
If you read the various June 2024 CSIRO G-PST reports (1,2,3,4,6,7,8,9), you may notice they are infused with “the thrill of the chase”; the excitement of being the first to solve complex problems and understand interesting phenomena. Note that I left out report 6, this was done by Aurecon, the engineers, these people tend to “simply” follow their brief. They get stuff done.
Scientists often can’t resist unlocking Pandora’s box when they find something new and interesting. It's in their DNA. For example, when discussing grid forming inverters, the authors note that these devices mostly tend to try hard to simulate current synchronous generators. Lift that restriction and you may be able to design totally different and, they suggest, possibly better. Alternative, you may unleash chaos and end up with hundreds of thousands of alternatives. What could be stupider than Australian inverters which we think are the best and South Korean inverters which work quite differently and which they think are the best … and Japanese inverters, and so on. What could be simpler than everybody simply emulating the physics of our current standardised synchronous machines?
With something as big and complex as the grid, the number of tangential and interesting topics to explore will be infinite and it will require considerable discipline to control what is usually called “scope creep” in projects; resisting the Newtonian diversion from the smoother pebble and prettier shell. The reports all list risks in their various tasks, but unless I missed it, nobody lists scope creep among their risks. A significant risk over the next 30 years is too many reports by too many scientists exploring too many brilliantly stupid ideas. I guess somebody has to soak up all that naive US venture capital just desperate to make a killing.
Comments