Stop talking about Blackouts
Fear and loathing in the NEM.
Another week, another article from the experts1 at The Australian warning how the electricity system is at risk of blackouts. Whether it’s vague fears about blackouts in response to the anticipated closure of Eraring in 2025. Or breathless warnings of blackouts due to Labor’s newly passed Safeguard Mechanism. Or just pandering over the achievements of the outgoing Liberal NSW government.2 Sometimes the articles don’t explicitly say blackout, but use the more euphemistic terms secure or reliable (or if you’re lucky, both!).
So, is the system actually at risk of blackouts? What actually is a blackout? Are blackouts occurring? And what does the future hold? This week we’re pointing a flashlight into the darkness.
The National Electricity Rules
Let’s start with the sexy stuff — the 1,800 pages of The National Electricity Rules (link is here you nerds). This is the key governing document defining how the electricity system is to be operated. You would be shocked to learn that it’s very long. And very dry. And it gets rewritten with increasing frequency.
Inside of this long document are two very important terms — reliability, and security. Those two links are to some excellent pages written by the Australian Energy Market Commission with a lot of information and context, but let me provide glib summaries for you.
Reliability is the property of the system relating to there being enough supply to meet demand. When you go home tonight and turn on your oven, air-conditioner and 60” plasma to watch MAFS, there will be enough power in the system for all of that work seamlessly. Indeed there will be enough power for all of east coast Australia to tune in to unmissable TV tonight. If the system is able to supply our desire to use electricity, it is considered reliable. On the other hand, an unreliable system cannot guarantee that everyone turning on their TVs to watch the MAFS finale won’t trigger a blackout.3
Security is the property that relates to the system’s ability to resist disturbances. If you’re sitting at home matching MAFS and a power station somewhere breaks down randomly,4 a secure system will mean your guilty viewing pleasure won’t be interrupted. The electricity system is complex and there are lots and lots of things that could happen to disturb the system unexpectedly; but a secure system is able to ride through all but the worst of these disturbances.
So, yes, a secure and reliable electricity system is critical for a modern first world country to provide its citizens with uninterrupted access to Vanderpump Rules.
So, what’s a blackout?
Blackouts
Let’s talk about the causes of blackouts. This sounds facetious, but it’s important. The technical term for measuring the size and impact of a blackout is Unserved Energy. The Australian Energy Market Commission has a whole document discussing and defining this term.
Broadly speaking, there are three categories of unserved energy:
A storm could wreak havoc with the power lines, a council worker could put a backhoe through a cable5 or a transformer could fail. These accidents happen frequently but usually only at a localised distribution network level affecting hundreds or thousands of customers. Generally, these events impact regional energy users more frequently that metropolitan users too.
On a really hot day, there might literally not be enough power in the system to meet the demand,6 or there are a significant number of power stations offline, or in some cases both. These events happen infrequently, but can effect hundreds of thousands or possibly millions of customers.
Something very significant and unexpected7 happens, disturbing the system plunging hundreds of thousands or millions of customers into darkness. Two recent-ish examples are the 2021 explosion of Callide C which blacked out a decent chunk of Queensland and the storm in 2016 which collapsed transmission towers and plunged the entire of South Australia into darkness.
Point one could be rephrased as network blackouts, point two reliability blackouts, and point three security blackouts. The AEMC has helpfully prepared an infographic of the relative frequencies of these events.
Because this whole unserved energy thing seems pretty important, there is a defined reliability standard, which sets guidelines over how much unserved energy is acceptable in the power system — the current standard is for 99.998% reliability, per region, per financial year.8
Importantly, the reliability standard doesn’t include blackouts attributed to accidents on the transmission and distribution networks. It also doesn’t include security blackouts — where loads are disconnected to protect the broader system from blacking out (more on this below). The reliability standard only relates to times when there is not enough supply available in the system to meet demand.
So, how does a reliability or security event actually cause blackouts? The answer lies in electrical protection systems.
In the case where there is not physically enough supply to meet demand, AEMO (the market operator) will instruct transmission networks to conduct involuntary load shedding. Transmission networks will then instruct distribution networks downstream to shut off specific substations, blacking out thousands or tens of thousands of customers. There are procedures determining the order and number of these substations to be shut off to prevent favouritism;9 shockingly these are not public documents. These outages will move between suburbs until enough supply is made available and the underlying demand has dropped away.
The core premise behind the involuntary load shedding is much like frostbite — by sacrificing the outer extremities,10 the body can be protected. By blacking out a few random suburbs, the rest of us can continue to watch Love Island in air-conditioned comfort.
In the case of security events, the protective mechanism is Under Frequency Load Shedding.11 If a major security event disrupts the system, the grid frequency will rapidly decline. Automated relays at substations will kick in, removing large numbers of customers and rapidly reducing demand, hopefully fast enough to prevent the entire region from blacking out due to frequency collapse. This is precisely what happened after the Callide C explosion in May 2021 — roughly 400,000 customers were kicked off the system in order to prevent the entire state of Queensland (or even further south) blacking out.
The outcome of involuntary load shedding and UFLS or even a bad storm or network equipment failure might look the same (entire suburbs blacking out), but the causes and mechanisms are very different.
So why not just build the system so we get no blackouts? Well, to have an infinitely reliable system, you need infinite redundancy. Building out the system to prevent any unserved energy would be astronomically expensive — there is a long tail of risk and building out the system to manage every conceivable event is wildly inefficient. It’s much more cost effective for all of us to live with a small, but non-zero risk.
The follow up is that the work done to minimise unserved energy due to reliability or security events would do little to mitigate the blackouts that affect you and I most of the time. If 95% of customer outages are due to accidents on the distribution network, the cost of gold plating the system seems a little pointless.
Forecasting Blackouts
If you worry about whether you’ll still have unimpeded access to MAFS in 2030, or if you love to write up fearful op-eds about the future of electricity, you’ll be happy to learn there is a lot of time and effort allocated to forecasting blackouts.
There are four tiers of forecasts:
The most immediate operational forecasts are called "Pre-dispatch PASA" and "Short Term PASA"12 These forecast any potential shortfalls in supply (i.e. reliability issues) on a 7-day horizon. There are three levels to this — Lack of Reserve 1, which is not great, LOR 2 is cutting it fine, and LOR 3 is literal blackout conditions and load shedding will be required.13
Beyond the PD PASA and ST PASA systems, there is also the “Medium Term PASA”, which forecasts on a three year horizon. Because of the longer forecast horizon, the MT PASA doesn’t categorise LOR events, but uses a metric called the Loss of Load Probability which is a forecast of when load shedding might be required.
The final forecast produced by AEMO is the “Electricity Statement of Opportunities”, or ESOO. The ESOO exists on a ten year horizon to provide long term planning, especially for governments and generation investors.
Have reliability blackouts been occurring?
No (well, not recently). There will always be small network blackouts, especially for those out in the sticks.
BUT, there have been no identified unserved energy events since the hot summer of 2018-19. And there have been no breaches to the reliability standard in nearly 15 years. In layman terms, the system has met the planning standard for reliability. But don’t take our word for it, the AEMC Reliability Panel has produced this nice chart.14
So when the experts at The Australian are discussing blackouts, we hope they would never conflate accidental localised distribution network outages with reliability problems (USE).
… And will there be blackouts in the future?
No (But also not no, bear with us).
There will always be blackouts in the power system — trees will keep falling on lines and random catastrophic security events will continue to happen (because that’s how tail risk works).
But with regards to blackouts due to reliability events (i.e. the things that conservative commentators spend lots of column inches on), let’s look at the forecasts. The latest ESOO was published in mid-2022, and AEMO released an update in February 2023. Let’s walk through the key findings.
There are two horizontal dotted lines on the chart - the Reliability Standard (99.998% reliability) and the stricter Interim Reliability Measure (99.9994% reliability).
No regions are currently forecast to breach the reliability standard until 2027 at the earliest. That’s the key takeaway.
No regions are currently forecast to breach the stricter interim reliability measure until 2025 (which has specific implications too mundane to describe here).
The chart has two versions — the dotted lines are the forecasts from 8 months previous in the 2022 ESOO, and the solid lines are the forecasts from February 2023. The forecasts for NSW, Victoria and South Australia have all improved, especially in the short term.
Let’s just take a moment and appreciate Tasmania doing its thing, quietly, avoiding scrutiny or concern.
But, as with anything, AEMO is careful to highlight that because of the unpredictable nature of reliability and security events, unserved energy is still possible, even if the total unserved energy is forecast to remain within the acceptable levels (below the Reliability Standard).
While this Update to the ESOO forecasts expected USE is within the [reliability standard] across the NEM in 2022-23 and 2023-24, there remains risk to consumer outcomes … large USE events are possible even when expected USE is forecast within the [reliability standard]. These large events may be driven by weather uncertainty or other circumstances such as simultaneous generator and/or transmission outages that may erode available supply when it is required.
So no, the actual experts are not forecasting significant blackouts in the near term, but they’re also not silly enough to say that no blackouts will occur.
That second last point in the list above is important — AEMO generates these forecasts on the best known information at the time of forecasting and on the basis that nothing will change between now and then. So while AEMO might forecast the Reliability Standard being breached in 2027, expecting it to eventuate is like standing in the middle of a level crossing and saying “Oh look, there’s a train coming towards me.”
Of course, if we could harness the hot air steaming out of The Australian, we’d never need to worry about unserved energy ever again!
Things Happen
Billionaire
professional griftermining magnate Clive Palmer has enlisted the help of former Federal Attorney General Christian Porter to sue the Australian federal government over a WA law which prevented him from claiming damages when the state blocked his iron ore project in 2020. We think they’re perfect for each other.A trial of wave generation in the Bass Strait has just concluded, with the Australian company saying it has its sights set on the European market. See, maybe we’re not so crazy at all!
With no new news about the Snowy Tunnel Boring Machine, Currently Speaking will be setting up a search party. We’re coming for you Florence!
I say this with every ounce of disrespect I can muster.
Oddly referencing mysterious blackouts which have occurred recently, apparently? Refer to the USE chart.
Forget the disappearance of Harold Holt, this would be a true national tragedy.
This happens much, much more than you think it does.
Still better than that time an elderly Georgian lady digging for scrap copper cut off the internet to most of Armenia.
All those air-conditioned MAFS viewers.
This point seems pedantic, but it’s rather important — if it was expected, then we could have planned for it and the outcomes wouldn’t have been as severe. AEMO actually has a system dealing with these kinds of things via the idea of credible and non-credible contingencies. If a non-credible contingency turns out to be more common then initially thought, it will be reclassified as credible.
There also exists the interim reliability measure, which is even higher — 99.9994% in a given region. But it’s important to note that the interim measure exists for a slightly different purpose, which we may have to come back to another day.
A.k.a. blacking out the poor people suburbs where politicians don’t live.
Definitely not the poor people suburbs, the government would never do that.
In New Zealand the terminology used is Automated Under Frequency Load Shedding or AUFLS, which is one of the best acronyms ever.
Projected Assessment of System Adequacy, obviously.
This AEMO document provides an excellent explanation beyond my glibness.
And they’ve fixed the awful colour schemes of years past — kudos AEMC data viz team!