Eight big questions arising from AEMO report into SA blackout

Print Friendly, PDF & Email

The market operator seems happy to let the blame for the South Australia blackout fall on wind energy, but questions about its own role are warranted, particularly because it seems to have totally ignored a report prepared a decade ago that envisaged a similar scenario.

Print Friendly, PDF & Email

The latest report from the Australian Energy Market Operator again allows wind energy to be blamed as one of the prime causes of the state-wide blackout in South Australia, even if it did blow away some of the classic myths about wind energy.tumblr_static_question-mark3-655x437

But the report raises major questions about the actions of the market operator and the energy market’s rule maker, and the future direction of Australia’s energy transition – because old world thinking is still dominating politics, media and energy management.

Here are eight key questions that arise from the AEMO report:

Why is Josh Frydenberg still banging on about “intermittent generation”

The AEMO report made it clear that neither the “intermittency” of wind generation, nor its ability to operate in high speeds, were an issue in the September 28 blackout. And if wind energy was at fault it was because of the settings of “fault ride-through” mechanisms, which can be, and have been, easily fixed. But that did not stop energy minister Josh Frydenberg in his tracks. In an interview with the ABC’s Chris Uhlmann, Frydenberg insisted that intermittent generation was the real culprit, saying that South Australia’s mix of  40 per cent intermittent generation (wind and solar) “had clearly failed”, and he used this as a call to rein in the “unrealistic” state-based targets, particularly those of Queensland and Victoria.

Frydenberg was kept on his toes by Uhlmann’s forensic questioning: “Isn’t it an inconvenient truth,” Uhlmann suggested, “that we can get to renewable energy (even the federal government’s 23 per cent target) … but it is going to be more expensive?” No point bothering with Frydenberg’s answer. But, no, it doesn’t have to be more expensive at all, and the Queensland expert panel – and any number of other government inquiries into RETs – made that clear in their report on the state’s proposed 50 per cent target.

But electricity prices will almost certainly be more expensive if the federal energy minister continues to apply 20th century thinking to 21st century challenges, and responds with the most expensive options: more gas and more poles and wires. It certainly won’t be any incentive for the principal agencies – the regulator (AER), the policy maker (AEMC) and the operator (AEMO)– to think differently about the energy future if their leader is stuck in last century thinking.

hallett wind farm

Was it really the fall in output from wind farms that triggered the failure of the interconnector?

The AEMO investigation into its own actions has found itself not guilty (more on that below), but left just enough evidence for the anti-wind brigade to find wind energy guilty and sentence it to a thousand disapproving columns. All Andrew Bolt had to do for his column today was to cut and paste from the AEMO report and slap on a headline declaring: “Wind power caused the SA blackout.”

Indeed, AEMO’s second report claims the trigger point of the separation of South Australia from the Heywood interconnector was the loss of 445MW of generation after “self protection” devices in nine out of 13 wind farms turned the facilities off, or down, after six faults in 80 seconds (caused by the collapse of three main transmission lines).

The wind industry and energy experts are surprised (and more than just a little angry and frustrated) by such conclusions, particularly considering AEMO’s own reports and graphs that show a massive loss of voltage that may have preceded the loss of generation and the disconnection.

They point to the huge loss in voltage just before the Heywood interconnector “opened”. As one wind farm engineer asked: “Why is the market operator intent on discounting the physics of what has happened in the network while focussing on the performance of the wind turbines (in this extreme event).” This view is reinforced by AEMO’s own observations:

“The rapid decline in voltage across the SA network was observed immediately prior to the opening of the Heywood Interconnector. This rapid voltage decline was consistent across the SA transmission network from the South East to the North. Once separated from the rest of the NEM, network voltage within SA momentarily returned to normal levels before the rapid frequency fall led to the Black System.

“This observed reduction in network voltages is consistent with a loss of synchronism between the SA power system and the remainder of the NEM. AEMO will be conducting further analysis to confirm that these changes in network voltages are fully understood and were as expected given the circumstances.”
In other words, it could be that the Heywood interconnector protected itself in the same way as the wind farms did. As AEMO states further:Data gathered to date indicates the protection mechanisms on the Heywood Interconnector operated as designed.”

Why didn’t AEMO already know about the wind farms’ self-protecting settings?

AEMO says it was unaware of the settings of the self-protecting “fault ride-through” mechanisms. That seems an extraordinary admission from a market operator who presumably would have a close look at this important mechanism when making its assessment of proposed wind farm connections.

It’s even more surprising because a report conducted for AEMO’s predecessor, NEMCO, looked into this very issue when it was asked, in 2005, to imagine a very similar scenario to that which played out on “Black Wednesday”. The report, by German-based consultants DIgSILENT, and titled “Assessment of Potential Security Risks due to High Levels of Wind Generation in South Australia” looked at what would happen if there was 800MW of wind generation in South Australia and one of the main transmission lines fell over. (Sound familiar?)

It placed particular importance on the very fault ride-through mechanisms that have suddenly become the focus of attention now. It also said that the market operator should expect major variations in output as the wind farms rode through various faults.

This particular paragraph is of interest: “As part of assessing proposed wind farm connections, studies are conducted to verify fault ride-through capability. If necessary appropriate measures are taken to ensure the wind farm satisfies this requirement. These measures may include installation of additional reactive support at the connection point or restricting the output of the wind farm. It is therefore expected that necessary measures will be taken if the problem identified in these studies was identified during the connection studies for this wind farm.”

The result was that onerous conditions were put on ride-through mechanisms on Australian wind farms, with which the wind generators say they complied. But the rules never envisaged multiple line failures, and nor did they address the issue of voltage collapse, which is what occurred in South Australia. The view of AEMO was that the generators were simply required to “ride through” even if it means that they sustained damage. That is an interesting and highly contentious interpretation.

Were the wind farms even supposed to ride through the transmission line failures?2909_sastorm_sp

It’s all very neat for AEMO to allow the blame to fall on wind farms, but the problems may actually be in the rules. It also turns out that the National Energy Rules do not require transmission-connected generation to stay connected for a 3-phase fault (such as three transmission lines falling). This is different to other international standards possibly because three-phase faults are rare in Australia, and historically this level of requirement has not been an issue.

This seems to be confirmed by AGL Energy, which issued this statement yesterday pointing out that its wind farms were operating according to market rules: “AGL Energy Limited (AGL) today confirmed that it had operated its gas and wind generation assets in accordance with Australian Energy Market Operator (AEMO) requirements and the Essential Services Commission of South Australia and ElectraNet safety and connection requirements.”

This could be an important issue. Because as the Murdoch media hyperventilates about the possibility of large energy users suing wind farm operators, they might be better served taking action against the regulators and policy makers who either had lax regulation or did not provide proper oversight. That would be the responsibility of Frydenberg and his predecessors.

Why did AEMO do nothing ahead of the storm?

AEMO said it didn’t do anything ahead of the storm because it did not see a “credible” threat to generation or transmissions lines. But that claim is a little hard to defend when its own document notes that many wind farms had wind speed limits of 90kmh and it noted that winds were forecast to jump to 120kmh or more. As the NER notes, AEMO is required to take into account the “operation within all plant capabilities of plant on the power system.”

Things it could have done: Rely a little less on the Heywood interconnector for generation so it would have some room to move if there were problems in South Australia. Put at least some local FCAS (frequency and ancillary services) on standby in the state. And in the future, have some battery storage. As we noted soon after the event, even tapping into the battery storage of 10,000 homes could have made the difference. And that would be a lot cheaper than building a new interconnector.

Was AEMO more concerned about wholesale prices than energy security?

One of the justifications used in the AEMO report to not take preemptive action – apart from the fact that it did not see any credible threat – was that it did not want to affect market prices. If it had commissioned local FCAS supplies, or wound back generation on Heywood and got the Adelaide-based gas generators to provide more, that would have increased costs. And, to be sure, if nothing failed there would be some grumbling about the need for additional costs. But exactly what is the role of AEMO here? To protect energy security or to be a price-setter?

Should an independent body investigate the role and actions of AEMO, rather than AEMO itself?


Do we really need a new interconnector? And do we really need to double down on gas?220px-Torrens_Island_Power_station_from_the_river_-_portrait

All the proposed “solutions” to South Australia’s high renewable energy penetration involve investing in the two most expensive technology alternatives (and the two primary reasons why South Australia has one of the highest consumer electricity prices in the world): more gas and more poles and wires. Gas, it is said, would provide more “synchronous generation” to ensure system security. A new link to NSW would provide another fall-back.

Both ideas warrant some scrutiny. Another interconnector to NSW would have had to use the very same transmission lines that fell over during the storm, so that wouldn’t have been much use. As the AEMO report makes clear, gas and other fossil fuel generators are so damn slow in responding, they wouldn’t have had much impact either.

So why not battery storage? Storage experts say that just using the battery storage from around 10,000 South Australian homes might have been able to stabilise the grid long enough for the market operator to react. Grid-based storage and micro-grids are the technologies the US and other places have turned to after their centralised networks were smashed by their own storms.

So why don’t we do that here? Because the likes of AEMO and Frydenberg insist that battery storage is still a long way from commercial pricing. But such assessments are based on that 20th Century thinking and the assumption that batteries, like fossil fuel generators and poles and wires, cannot do more than one thing at a time.

As New Zealand network operator Vector points out today, battery storage has numerous potential uses and revenue applications: time shifting solar, shifting loads, shaving peaks, implementing demand response, “firming” renewables (providing constant power, for instance, when a cloud passes over a solar farm), as a back-up power supply when the grid fails (read these stories about consumers unaffected by blackout), as the centrepiece of a microgrid, for ancillary services (providing FCAS), and as a cheaper alternative for poles and wires, upgrades and new transformers.

That’s quite a list. Frydenberg should wake up to new technologies, and along with other state energy ministers give the policy rule maker a sharp kick up the backside and tell it to get cracking, and not to be so focused on the whining of ageing and out-dated fossil fuel generators and grid operators who seem more intent on holding consumers to ransom.

Print Friendly, PDF & Email

  1. Rob Campbell 3 years ago

    It’s very true to say, if you collect enough research from enough sources, you find data that can reach any predetermined outcome. This seems to be more than applicable to the National Electricity Market. Historically essential services where built, owned and maintained to serve bodies corporate, that is a group of people who formed local governments and elected office bearers, which in turn collected and applied funds for the common needs of a community, at least cost. Since state governments have had control of most of these community assets they have been evolving as cash cows, all justified by volumes of carefully collated technical data, cherry picked to suit a government’s current agenda.

    The problem has been getting progressively worse since the start of the new millennium, and it has nothing to do with chronology, more to do with the introduction of the G.S.T. In order to get all of the states to embrace ‘a new tax system’, John Howard pulled off the coup of the century, by first starving the states of federal funding (remember the $40 billion surplus?) and then offering up a honey pot of cash, guaranteed to get bigger, the G.S.T.. The states didn’t do the responsible thing with this new revenue, something that any normal frugal person would do in budgeting and spending it when it came in, no, they did what Morgan Stanley did, borrow against it, and heavily. The result is that many states have much of their income spent before it is earned, leaving little for new critical investment. Many have decided to arrest this situation by ‘selling the farm’ another strategy championed by a recently cash-strapped federal government.

    South Australia has the highest wholesale prices in the country for one reason and one alone, interconnectors. The whole concept that we were sold on with the N.E.M. was competition, this was to be brought about by joining the states together with links that would mean that any generator could virtually sell electricity to any grid connected customer wherever they were. Whilst the federal government has pulled back from its renewables blame for S.A.’s power woes, the very thing they are now blaming the states for is what that they should be responsible for managing. The problem is the N.E.M. is not a Federal government entity, it is a state collaboration born of legislation domicile in South Australia and here’s the rub, the federal government will now likely try to take over the N.E.M., as a means of bringing about the energy security they feel the states have failed to achieve and there is a good argument for it. The recent report from AEMO on the size and state of interconnectors , shows the true inadequacy of links that are doing two things to stifle competition and push up power prices. Firstly, the abundance of renewables that South Australia has installed often has now where to go due to lack of interconnector capacity and secondly, the perfect supplement for these renewables which is base load and peaking gas, lies in the hands of big operators who don’t know what their revenue will be from day to day so endeavor to maximize that revenue whenever the opportunity arises.

    The blackout in S.A. was a cascading event, a deliberate action of the network to react to a sudden drop in frequency, similar to throwing a stick in a bicycle wheel. In order to prevent customers in areas other than where the actual failure occurs, the failing part of the network is cut loose. It just so happens that there was so much damage going down in S.A. at the same time the cascading tracked all the way to the Victoria border. Its no a failure as such but extra money could and should have provided additional isolation points, but this is academic in the situation that S.A. found itself in.

    Despite the multiple layers of bureaucracy and costs of regulating this so called ‘world’s best practice’ system, what it is in reality is chaos and crises management of a system that is piecemeal and fragmented. This situation all but neuters the concept of real competition, and combined with the veil brought about by retailer competition which is now subsidized by a network charge in most jurisdictions, one has to wonder what the situation would be if the N.E.M. was never instigated, in this scenario, the states would have to look locally at the impact of renewables and the supplemental gas supplies required to ensure their effectiveness. In order to provide the promises made when the N.E.M. was sold to us, the need for federal intervention is clear. Alas, the states that continue to profit from the costs of managing the chaos, are unlikely to seek a solution that brings any financial benefit to consumers, given that revenue streams are the main focus of these debt ridden entities, and don’t expect the newly privatized incumbents to do anything noble, the whole premise of their purchases of these fragmented assets was “business as usual” as a minimum.

    • MaxG 3 years ago

      Thank you for your effort and posting this…makes sense to me.

    • Ray Miller 3 years ago

      Thanks Rob as well for your perspective, it seems the game of chess is still playing out with a newly materialized Renewable White Queen upsetting the whole game plan of the Fossil Black King.

  2. john.boland 3 years ago

    As you say, there are lots of questions about AEMO’s performance and also the operational rules and methods. One that you point to that I had already questioned I will restate this way: AEMO says that the ride through provisions were not provided to it by some wind farm operators prior to accreditation by the NEM. My question is whether that is a requirement for accreditation, if not, why not, and if it is, then why didn’t AEMO demand it. As you say, it seems extraordinary that they do not have this knowledge at hand and if they deem it inadequate why did they not do something to change it.

    One question I have is that according to the report the Under-Frequency Load Shedding
    (UFLS) was not invoked which means that sequentially feeders are dropped when frequency drops. The reason AEMO gives is that the frequency did not fall below 49 Hz prior to the dropout of Heywood since after the drop of supply from the wind farms the supply from Heywood kept the frequency above 49 Hz, “in effect preventing the UFLS from seeing the major deficit of generation that existed in SA.” I am astounded that the system was not set up to recognise a major deficit of supply in SA independent of what the interconnector was supplying and instigate UFLS.

    • Malcolm M 3 years ago

      This is a crucial system design fault, made worse by the inter-connector upgrade. There should be a load shedding system within SA that is controlled by loads into SA exceeding a threshold. This system needs to work quickly enough that loads are shed before the inter-connector trips.

    • DJR96 3 years ago

      The UFLS system is only able to react to a rate of change to the frequency of 3Hz/sec or less. This event saw twice that. It happened too quickly for the system to be able to be effective. (Figure 5, page 15 of the preliminary report)

      Furthermore, the generators only have to cope with RoCoF of up to 4Hz/sec. Anything more and they can simply disconnect! So much for fault ride through….

      This all seems like good justification for a network that has a master inverter setting the frequency and all generators can only provide power via a DC element (inverter etc) Then frequency control is a non-issue. Generators don’t have to be so precise. The network doesn’t then need to depend on the frequency of the big generators, or keeping them synchronised to each other. Much more resilient and reliable system.

  3. Alan S 3 years ago

    Do we need some spinning reserve in SA to provide frequency stability? Someone with a better understanding of generation and transmission systems than me may be able to comment. When I worked at Tumut 3 hydro station, its six 250 MW generators would sometimes run in motor mode (powered from the NSW/Vic grid and no water in the turbine) to provide a rotating mass to improve frequency stability and power factor correction. You don’t need a hydro station for this as a flywheel storage system would do the same.

    • Chris Fraser 3 years ago

      Are you sure you need whirling machinery ? Couldn’t you get the same stability of the grid with a battery and inverter ?

      • Nick 3 years ago

        The spinning mass provides inertia that a battery doesn’t.

        • DJR96 3 years ago

          And as soon as you increase the load, that load will try to slow it down. Governing systems are required to quickly react to changes in load to be able to maintain frequency. Even then it can only operate in a small range as opposed to being set at a constant 50Hz that an inverter can do regardless of changes to load.

      • DJR96 3 years ago

        Yes, absolutely you could. An inverter can maintain a constant 50Hz regardless of load.

        • Chris Fraser 3 years ago

          Thanks for that. I’m getting hints from readers that spinning inertia (by courtesy of a direct connection between a generator and the grid) is becoming an obsolete paradigm.

    • Rod 3 years ago

      Spinning reserve has always meant to me being synchronised and available to ramp up but no or very little load. I have my doubts if it would have made any difference in this instance. Thermal coal or gas just can’t ramp up quickly enough.
      There is some disagreement as to what spinning reserve actually means:

      Page 7 has a list.

      • Chris Fraser 3 years ago

        And what inertia means, and perhaps whether or not a spinning machine really matters ….
        If the grid you’re generating to is not completely ‘black’, then isn’t it true you’re trying to match the existing waveform with correct Voltage, correct Frequency and correct Phase ? My PV inverter does it why not others .. ?

        • Alan S 3 years ago

          Your grid interactive inverter is electrically locked to the grid so is frequency and phase angle synchronised. Its output voltage must be sufficiently above the grid to drive power into it. This happens only when the grid is healthy and why the inverter shuts down if grid voltage is low.

          Grid-independent (standalone) inverters presumably have a precision frequency reference in their oscillators to maintain 50 Hz.

      • Alan S 3 years ago

        Good article thanks. The term was used fairly loosely by unit controllers at the power station but it referred mainly to adding rotational inertia to the system.

  4. trackdaze 3 years ago

    I watched the chris and joshy interview on the abc and have to say i was appaled.

    That interview belonged on the home shopping channel.

  5. Nick 3 years ago

    Giles, can you provide some defense of this statesment: “particularly considering AEMO’s own reports and graphs that show a massive loss of voltage that may have preceded the loss of generation and the disconnection.”

    The AEMO report shows voltage at 1.0 pu, then the wind plants trip, then the voltage collapses, then Heywood opens. Can you explain how you’re seeing the voltage collapse before the loss of wind generation?

    • Mick 3 years ago

      Hey Nick – where does the AEMO report say the voltage was at 1.0 pu then wind farms trip?

      There were two major reductions in wind farm output – second of which (the most significant, namely the “Group B” wind farms) occurred at 16:18:15.1 (see page 24).. This is quite clearly *after* substantial voltage dips (to 40% p.u) at 16:18:13 and 16:18:14 (i.e fault no. 5, see table 3, page 18).

      According to the report, these voltage reductions most certainly occurred before the loss of Group B wind generation.

      (Similarly – the other ‘Group A’ wind generation was lost *after* fault 4)..

      So I guess I would be interested to know how you’re seeing the voltage reductions *before* the loss of wind generation!!

      • Nick 3 years ago

        Mick, I’m looking at Figure 1 on page 15 that shows healthy voltage until just past 16:18:15.

        Figure 8 on page 23 shows the sustained reduction in wind groups B and C occurring just past 16:18:15 when the 275 kV voltage was still healthy.

        The voltage dips described in Table 3 are faults, not a “massive loss of voltage” as Giles describes it.

        “So I guess I would be interested to know how you’re seeing the voltage reductions *before* the loss of wind generation!!”

        This is exactly what I’m saying. There is no massive loss of voltage until after the loss of wind output.

        • Mick 3 years ago

          Ahh I see – but figure 1 only shows the voltage on *one* of the three phases for each of the different transmissions lines.. (c.f. figures 4-6 or so, which show the voltage dips on all three phases).

          I personally don’t find the figures overly helpful (due to either resolution, or selective choice of phases) – hence why I was looking at the tables and text – which show the voltage dips occurring before wind farms trip off.

          The voltage dips in table 3 are indeed faults – and triggered the wind farms to trip off. I suspect that if you plotted the C phase of the Davenport line on rather blown up figure 1 (rather than A phase) – you would see a similar “massive loss of voltage” before the group B wind farms tripped off..

          I guess my main point is the text / data in the tables make it clear the voltage dips occurred before wind farms tripped off.

          (..And I meant to say … I would be interested to know how you’re seeing the loss of wind generation before the voltage reductions).

  6. Mike Dill 3 years ago

    So what happened to the ‘black start’ generation that was under contract?

    • DJR96 3 years ago

      This was explained in the 2nd report, chapter 5, page 28.

      SRAS 1 had a circuit breaker that wouldn’t remain closed. Some dodgy control signal kept re-opening it. Being investigated.
      SRAS 2 was deemed insufficient to black start due to locations of downed transmission lines. It also crapped out anyway.

      I assume SRAS 1 is the Torrens Island station. The small (backup) generator was able to provide power to ancillary services such that it could restart once Heywood was re-connected.

  7. Ray Miller 3 years ago

    Yes Giles many questions need to be asked.
    Seems to be much talk about “spinning reserve” but last time I looked wind turbines spin and have considerable amount of stored kinetic energy, I’m not aware of the response time, but its contribution would be still considerable?
    I would like to introduce the term “resilience” in much broader terms as opposed to “security”. Fundamentally the end user should be striving for appropriate amounts of energy service resilience dependent on predetermined needs and costs. No matter how much we spend on the centralized grid infrastructure the end user and community will always have a level of exposure to loss of energy services.

    Some very simple techniques and choices can provide the resilience energy services delaying the negative consequences of the inevitable loss of grid supply.
    Our society is increasing adding more and more fragile critical elements and anti resilient components and practices into our complex networks without regard to the inevitable consequences.

  8. DJR96 3 years ago

    I think I’ve got a grasp of this.
    Ultimately the black system occurred due to the disconnection at Heywood from Victoria. This occurred due to loss of synchronisation between SA and Vic. That is it. (Last paragraph page 15)
    So it was the old tech that let SA down.
    Speaks volumes about the inherent difficulty and non-resilience of trying to keep multiple synchronous AC systems running in parallel.
    Note the HVDC Murraylink performed flawlessly. Perhaps there should only be DC connections between individual AC generators?

    The wind farms all did exactly as they were programmed to do. Unfortunately they were programmed to reduce output or disconnect when a number grid disturbances were detected. In reality, they should be riding through any event until a loss of grid is detected to provide as much support as possible.

  9. DJR96 3 years ago

    Just looking at some numbers.
    Both Ladbroke gas turbines were operating a full capacity, fine. But only 3 of 5 turbines at Torrens Island were operating at the time, and they were only running at just over half capacity (248/480MW).
    And the Murraylink was only at half capacity too (114/220MW).

    So my question is, why didn’t either of these increase capacity in the lead up to the black system event?
    There is no mention of how they reacted other than to say they rode through the disturbances in the lead up to event.
    It seems that as some of the wind farms decreased output, the Heywood connector was relied upon for all of the short fall.
    Can this be right?
    Surely Torrens Is can react as quickly as Loy Yang in Victoria?

  10. DJR96 3 years ago

    And the Heywood connector was carrying 499MW prior, 840MW when attempting to re-close the Davenport-Mt.Lock HV line, and eventually tripped out at 900MW when the network lost synchronous control. That re-close attempt appears to be much of the cause of lost synchronous control.

    Looking at figure 3, page 12 of the preliminary report, it appears the network takes about 4 seconds to recover from a disturbance. And more so, during that recover period things oscillate at about 1Hz as it recovers, which coincides with the time that re-close attempt was for. It exacerbated the disturbance created by attempting the re-close, contributing to losing synchronous control.

    Perhaps easy to say in hindsight, but this re-close attempt should not have been attempted while the Heywood connector was already running over-capacity and the network was extra sensitive and vulnerable to extra disturbances. Synchronicity was very likely not optimal at the time either.

    Thoughts anyone?

Comments are closed.

Get up to 3 quotes from pre-vetted solar (and battery) installers.