AEMO says wind farm changes mean SA blackout won't be repeated | RenewEconomy

AEMO says wind farm changes mean SA blackout won’t be repeated

Print Friendly, PDF & Email

AEMO says basic changes to ride-through settings on wind farms will be enough to prevent a repeat of the “system black” experienced in South Australia last September. But will the Coalition tone down its scare campaign about renewables?

South Australia’s wind energy is providing secure energy to the state. Shuttershock
Print Friendly, PDF & Email

The Australian Energy Market Operator says it is confident that adjustments made to wind farm software means there is no risk of the South Australia blackout being repeated in the future.

AEMO chairman Tony Marxsen told more than 100 energy experts at a presentation under the auspices of the Electrical Energy Society of Australia last week that the “system black” event in South Australia in September – which has set off a huge debate about renewable energy across the country – would not be repeated.

South Australia’s wind energy is providing secure energy to the state. Shuttershock

According to several people who attended the talk, Marxsen said AEMO had analysed the failures that contributed to the blackout, and their possible remedies, such as making power poles more resistant to strong winds and better demand response programs.

So far, he said, only one remedy had been implemented – adjusting the ride-through settings on the wind farms in the state. Extensive AEMO modelling, Marxsen told the audience, showed that this would be enough to prevent a repeat of the “system black” event that put the whole state in darkness for several hours.

“If the same sequence of events happened today the system black would not occur,” Marxsen told the audience, according to one source.

This is an important concession from AEMO. It suggests that South Australia, even with around 40 per cent wind energy and a further 6 per cent from rooftop solar, is not at risk of a system-wide shut-down that affected the state late last year.

Of course, there is little the AEMO can do to prevent isolated blackouts, such as the storm-related events that affected more than 100,000 customers in Western Australia last year, and blackouts in Queensland, where there is no large-scale renewables.

In its reports after the blackout, AEMO said it was not the nature of wind energy (i.e. its variability) that contributed to the blackout, but the ride through settings on the wind farms.

Apparently unbeknown to AEMO, many wind farms had ride through settings that forced the turbines to shut down as a self-protection mechanism after a handful of major faults or voltage changes in a short period of time.


This issue was identified and addressed a decade ago in Europe, where no blackouts have been reported despite the high reliance on wind energy in some countries. There is some controversy about AEMO’s claims that it could not have known about the nature of those settings.

Indeed, Marxsen told the audience that in the age of the internet software changes could be downloaded by the manufacturers without the market operator knowing. He likened it to an update of an Apple iPhone, where the user has no idea what is being changed.

This, he made pains to point out, wasn’t an attack on wind, but a recognition that we are in a new world, so we need new ways of dealing with that.

Indeed, culture as well as technology will be critical components of managing the grid in the future. If power engineers, like many in the mainstream media, refuse to believe that wind and solar can play a dominant role in the grid, then the task will be difficult.

But that cultural change is likely to come soon, with AEMO’s appointment of a new CEO, Audrey Zibelman, who has been running the state of New York’s ambitious Reforming the Energy Vision program, which looks to a renewables-based distributed grid in response to the blackouts caused by Hurricane Sandy.

New York intends to reach 50 per cent renewable energy by 2030 – a target proposed for Australia by Labor that has caused so much controversy in political circles and the mainstream media.

But it is not just Zibelman pushing this line. Chief scientist Alan Finkel, in his draft report prompted by the SA blackout, said while there were challenges to incorporating high levels of variable renewables, there were ready technology solutions.

He also pointed out that the wind farm settings had been identified and solved in Europe a decade ago.

This view was echoed and re-inforced by a report from the CSIRO and Energy Networks Australia, which represents the nation’s grid owners, which said that Australia could, and should, aim for high levels of renewable energy – both for climate and environmental reasons and because it was cheaper.

It also envisioned a grid in South Australia with more than 80 per cent renewables, a level the state appears to be heading to given the number of proposed and confirmed renewable energy projects.

AEMO is not due to release its final report into the blackout until March. After that, it will be the turn of the Australian Energy Regulator to investigate AEMO’s actions.

The AER report will likely focus on two contentious aspects. One is the AEMO’s lack of knowledge about the fault ride-through mechanisms, another is its decision to take no action as the storms approached South Australia and swept across the state with increased ferocity.

Marxsen again defended AEMO’s actions at last week’s presentation, saying there had been no forecast for the tornadoes that tore down three main transmission lines that set off the sequence of events that caused the system black.

But energy experts also note that since that time – and previously – AEMO has been taking a much more cautious approach, announcing contingency measures in Queensland during recent summer storms.

There is a large body of experts that say AEMO could, and should, have taken a series of measures such as putting more gas plant on standby, or reducing the reliance on the interconnector to Victoria.

Marxsen, however, told the ESAA audience last week that there was no evidence that having more thermal capacity (coal or gas plants) could have avoided the sequence of events, given the amount of wind capacity that was lost when the fault ride through mechanisms were triggered.

Print Friendly, PDF & Email

  1. Peter F 4 years ago

    Marxsons comment at the end may is correct if you ignore two fairly obvious solutions solutions,
    1. replace gas turbines with reciprocating engines which can go from stopped to full power in 40-50 seconds or from 15% to 100% power in about 15 seconds. Thus the interconnector could have run at 120-130% capacity for long enough for the reciprocating engines to ramp up.
    The reciprocating engines also use 25-30% less gas than OC turbines at full load and half the gas at half load so it is a win win
    2) About 200MW of batteries which would have reduced the load on the interconnector for long enough to allow
    a) gas turbines/diesels enough time to ramp up
    b) some controlled load shedding

    • Chris Fraser 4 years ago

      Impressive ramp figures. I wonder why we have open cycle turbines at all ? Perhaps cheaper to install for a given MW ?

    • Chris Fraser 4 years ago

      Impressive ramp figures. I wonder why we have open cycle turbines at all ? Perhaps cheaper to install for a given MW ?

      • David Mitchell 4 years ago

        Open cycle turbines are cheap which fits the “baseload and peak” model. What we need now is the “balance the renewable” model for which reciprocating engines are perfect.

      • Peter F 4 years ago

        I have seen an article where they claimed the capital costs per MW for reciprocating engines was actually lower than OCGT for small plants.
        However you need 10 engines instead of one gas turbine so it is in some ways a bit more complicated, but on the other hand you can have one engine on 20% power to provide some almost instant backup and then have the whole plant up and running within a minute. Gas turbines don’t do less than about 30-35% power.
        As continuous power providers CCgas turbines are better (at least in cool climates) because the reach about 60%+ efficiency in winter vs 53% for the reciprocating plant.
        On the other hand at part load in hot weather the turbine plant might fall to 40%+ or even 25% if the steam section is not working, whereas the reciprocating plant will still be in the high 40’s

        Yet another benefit for SA in particular is that because of the small size, plants from 6MW to 250MW can be located throughout the grid, reducing transmission losses and vastly improving storm resilience

        • solarguy 4 years ago

          A gas turbine only takes seconds to start though.

  2. howardpatr 4 years ago

    Don’t expect any back-down from Turnbull, frydenberg and the RWRNJs but credit where credit is due, at least Turnbull had a scientist, Alan Finkel, review the matter, rather than a tame economist.

    • solarguy 4 years ago

      But will Turdbull take any notice of Finkel’s report?

    • Michael Porter 4 years ago

      I think you have spelled frydenberg wrongly. It should be fryingbird.

    • DJR96 4 years ago

      Actually, even an economist would go for anything but coal now too.

      • DevMac 4 years ago

        That’s when you know the war’s been won, although there are still on-going battles.

    • Chris Fraser 4 years ago

      I think Turnbull gets fear-mongering scripts in a fax from Bernardi every morning. If Bernardi forms a new Party I hope Turnbull gets a new fax number at least.

      • Rod 4 years ago

        Just heard Bernardi pulling the pin in the Senate.
        Hopefully the other RWRNJs will join him.
        They will only be diluting One Nation’s vote so will fail miserably.

        • Chris Fraser 4 years ago

          He’s got hopes if he imagines he can build the conservatives on protest votes.

  3. Rod 4 years ago

    So, now I’m really confused.
    Now they are saying the wind farms were at least partially to blame for the system black?
    Where the hell were they supposed to send their electrons if 22 transmission towers were on the ground?

    • trackdaze 4 years ago

      Im guessing the interconnector would have tripped anyway. No?

      • Rod 4 years ago

        My lay mans take on it was:
        Worst storm in 50 years so wind farms going flat out and near max coming from Vic over interconnect, Torrens Island doing squat. (Priced out probably)
        Towers fall down, wind power no where to go.
        Interconnector tries to make up for shortfall, maxes out. Oops.

        • trackdaze 4 years ago


        • Brian Tehan 4 years ago

          I think that the whole scenario was a lot more complex. According to previous articles on this issue and the information above, there were lots of glitches in the network, such as frequency variations as things went offline, lines went down, which caused the wind farms to disconnect after a certain number of issues were detected. This was actually unnecessary and settings should have been changed as this particular issue had been known about for 10 years in Europe. It would be like having a known security problem for 10 years but not fixing it in Australia. The question seems to be why didn’t the regulator know about this and/or why didn’t the manufacturers advise? In any case, it was really human error, not due to having renewable energy. Presumably, if the turbines hadn’t unnecessarily gone offline because of software settings, there would have been power available to some of SA. That’s my understanding.

          • Rod 4 years ago

            Yes, my overly simplistic response was probably trying to make light of the frustration caused by RE being blamed for this event.
            DJR has gone to great pains to clarify the situation for us.
            I seem to recall in one of the reports (Maybe Finkel’s) that regardless of generation source the downed transmission lines would probably have led to a blackout anyway.
            If Northern had been operating, such a large voltage/frequency fluctuation (caused by the majority of links to Adelaide being down) would have led to units tripping IMO.
            Which means wind turbines, with their ride through capacity enabled, are superior to FF baseload generators in such events?
            I haven’t heard that from Frightenberg or Trumbull.

    • DJR96 4 years ago

      Unfortunately yes. But not because they couldn’t produce power. The software settings mentioned meant that some disconnected when faults were detected on the grid. Had they just continued to provide power for as long as the network was live, all would have been fine.

      • Rod 4 years ago

        I thought I read somewhere that only a small number of turbines tripped. About 20MW worth.

        • DJR96 4 years ago

          Report and links in new post above…..

          • Rod 4 years ago

            Thanks, I had read your initial posts on it but not seen your update.
            I’ll have a read when I get a chance.

  4. DevMac 4 years ago

    Baseload this super-critical that “unreliable” whatever. Get stuffed, renewables weren’t to blame! I bet this great piece of news gets shunted to the bottom right hand corner of page thirty of the main papers, if it even gets that much publicity.

    And what is the AEMO’s role if not to do scenario modelling, and how can they do that accurately if they don’t know the characteristics and settings of the plants actually generating the power? AEMO asleep at the wheel. This is as bad a failure as the Census debacle – who’s head is going to roll for this one?

    “Marxsen, however, told the ESAA audience last week that there was no evidence that having more thermal capacity (coal or gas plants) could have avoided the sequence of events, given the amount of wind capacity that was lost when the fault ride through mechanisms were triggered.”

    Weren’t there two failed black starts on gas plants once the wind power was removed from the network? More plants means a higher percentage of success but at a very high cost. Also, two failures out of two smacks of unreliability on the part of non-renewables.

  5. DJR96 4 years ago

    I wrote the following report last year covering this whole thing. It includes the links to the AEMO reports as well.

    Summary of the SA storm and black system event 28/9/2016

    I assume you’ve all read the reports so far:-

    So after analysing all that I can provide some interesting details.

    Yeah the wind was seriously strong. Seven tornadoes with wind speeds as high as 260kmh developed during the storm, some crossing paths with transmission lines. Little wonder some pylons blew over.

    Yes some of the wind turbines reduced output or disconnected altogether.
    There is a few things going on here.
    Wind turbines need about a 30kmh breeze to generate at full capacity. They can keep generating at full capacity up to around 90kmh at which point they have to reduce to be able to not over-stress the pylon structure it stands on. This likely occurred at some units but it’s not been publicised yet just which ones or by just how much they reduced due to high wind speeds.
    They can tolerate wind gusts over 200kmh, so they’re pretty tough. And in fact none of them suffered any structural damage as a result of the storm.
    Almost all of the wind turbines detected disturbances on the grid (voltage sags/surges) and went into a ‘fault ride-through’ mode. When they do this they also reduce output to keep a buffer up in case of any further irregularities, but soon build back to normal generation if it senses the grid being ok.
    Unfortunately, many of them were also programmed to disconnect altogether if they go into this ride-through mode a number of times. Many only after just two occasions. Others 6 or more.
    So every time there was a fault in the transmission lines, either shorting, disconnecting or even re-closing after a trip, a disturbance occurs that the wind turbines detected.
    Also not made very clear is which wind turbines (if any) had to disconnect because there was no active grid to feed-in to due to failed transmission lines. Just like a home solar inverter, if they’re islanded they can’t operate.
    So yes, a fair chunk of wind capacity was lost due to this. But not due to excessive wind velocity!

    The reports say that the synchronous generation (Torrens Island and Ladbroke Grove), and the Murraylink HVDC transmission line all did what they were supposed to do and weren’t damaged at any time. What it does not say though is whether they were called upon to increase capacity as a result of some of the wind generation dropping off. Because it appears all of the shortfall was being brought in via the Heywood interconnector.

    Now this is the bit the reports don’t cover but if you analyse the graph on page 12 of the first report, you can see the disturbance caused by the 2nd trip of the Davenport-Belalie line at 16:18:09, and the loss of the first group of wind turbines (their 3rd detected fault). Then for 4 seconds after there’s a nice wavy line in the graph. This is due to the synchronous generation in Victoria and Torrens Island nearly losing synchronism. They were starting to fight each other and that puts huge extra loads on the transmission lines connecting them (Heywood). But it recovers after 4-5 seconds. Had that been all the disturbances it would have been all ok despite the Heywood line running more than 100MW over its rated capacity. It was coping.

    But then the Davenport-Mt.Lock line is tripped, and more wind generation is lost, creating another major disturbance which also messed with synchronisation. The system had not recovered from this (needing 4-5 seconds) when an attempt to re-close that line is made, totally screwing up synchronism, this time causing the load through Heywood to spike over 850MW which causes it to trip, disconnecting it from Victoria. Without all that energy coming from Victoria, the remaining generation could not supply demand, voltage and frequency plummeted which tripped all other generation off-line. Black system.

    The ONE thing that could have prevented the black system, was to not attempt re-closing that Davenport-Mt.Lock line. Just when the system urgently needed to be shedding some load, it tried to connect more. The Heywood interconnector was already running over-capacity and the system was struggling to stay synchronised.

    The irony here is that all grid connected inverters, whether that be your home solar system or the big wind turbines, have to by design specified in Australian Standards regulations, check for the presence and quality of the grid before feeding-in any energy.
    I don’t think it is too much to expect that the national grid operators should have to check beforehand whether conditions are safe and capable of connecting transmission lines and other elements before doing so. This would appear to be a monstrous gaping hole in safe operating procedures that need remedial action to ensure a more reliable network. These re-closers need to be centrally controlled and not independent units so that the network monitoring and control computers can manage them.

    AEMO seemed to not be aware of some of the software settings built into the wind turbines. In particular, the ‘number of fault ride-through events before disconnecting’ one.
    The wind turbine operators have all since adjusted them to 20 events.
    Significant generators like these are depended upon and every effort needs to be made to prevent any sudden unexpected losses in generation. It’s fine to ramp up and down at a reasonable rate that allows enough time for other generators to be able to adjust to. If the grid is live, let them generate at any time voltage and frequency is within operating limits. Inverters are capable of operating safely over a much wider voltage and frequency range than synchronous generators. Need to take advantage of that.

    Furthermore, I don’t know whether they already are or not, but they ought to comply with the latest AS4777 standard that requires they have reactive power control. This would see them providing leading phase energy when frequency/voltage on the grid is reduced, and vice-versa. This would provide some active grid support when synchronous generation is struggling. This may be achieved with a software upgrade, or may require a retrofitted control board from the manufacturer. Either way it should not be too difficult to implement, but the benefits very significant. The same also applies to the inverters at each end of the Murraylink HVDC. It too technically could provide the same support.

    So there you have it. Some insights you may not see elsewhere.
    It will be interesting to see if AEMO recognises these issues too.
    We’ll await their final report later this month.


    Having analysed the third report from AEMO ( ), which has significantly more information and observations, I have concluded these points:-

    • As a result of the sixth disturbance, when the Davenport-Mt.Lock transmission line tripped, a group of wind turbines reduced their output.
    • This significant drop in generation not only caused a drop in voltage, but started a divergence in frequency between SA and Vic. This can be seen in Figure 12, p.43.
    • The Torrens Island generators indicated a brief increase in output that is attributable to inertia response. That response comes at the cost of frequency reduction. Inertia can not provide extra power without losing frequency. It is a problematic characteristic too relied upon.
    • Torrens Island was not participating in FCAS with the NEM so it was not even required to have a governor response to attempt correcting frequency. Meaning it would not have contributed much, if anything, to increasing the frequency back towards 50Hz. (Unless load was shed.)
    • At this point the frequency diverged between Victoria and SA.
    • The frequency at South East followed most closely with Heywood in Victoria, not SA.
    • The Ladbroke Grove generators being located close to South East followed the frequency of the inter-connector. Meaning their inertia was not contributing much towards maintaining frequency in the rest of SA.
    • This diverging frequency is indicated in Figure 13, p44 as phase angle difference. This is exactly as I described beforehand.
    • Note too that the phase angle had not fully re-aligned after the previous (5th)disturbance at 16:18:08. Previously I had said it needed 4-5 seconds for this to occur. But without governor response from Torrens Island this was going to take much longer to correct itself.
    • Most of the wind turbines that were still contributing power at the time were injecting some of their energy as reactive power which would provide some level of network assistance. But I suspect the settings for this are too wide to provide the sensitivity required to be effective soon enough. ie. The network has to be quite bad before it’ll start helping.
    • When the phase angle difference became too great, the synchronism relay at South East tripped the circuit breaker disconnecting the inter-connector from Heywood. Note that the actual trip and disconnection was due to the protection of the synchronism relay, not because of over-current.
    • Quote from p45:- “It was the combination of high currents and low voltages that resulted in activation of Heywood loss of synchronism relay rather than the sheer size of current (over-load).” AEMO doesn’t seem to comprehend the severe effect diverging phase angle has. That they start fighting against each other causing the low voltage/high currents. The low voltage/high current state is only occurring due to loss of synchronism.
    • This separation/disconnection has occurred a number of times in the past. Always as a result of loss of synchronism, not just over-current. In prior events there have been at least 2.5 times more synchronous generation at the time, providing the inertia to make the phase divergence time longer, allowing the UFLS time to take effect, and allow SA to run as an island after the inter-connector trips.
    • Once separated from the rest of the NEM, Torrens Island was the only means of running SA in island mode, it was the only remaining facility capable of providing a frequency reference. But it was both ungoverned and its frequency (phase angle) was diverged from Ladbroke Groves generators. Meaning it was impossible for it to continue. It never stood a chance. SA could never run in island mode that day!
    • AEMO not requiring Torrens Island to perform FCAS when it was the only operating facility capable of providing it is simply unfathomable. What were they thinking!?
    • Note that all the operating wind turbines and the Murraylink are “followers” to the network. Without an operational network they can do nothing and must disconnect.
    • Many of the elements that make up this event have occurred before. AEMO had prior experience. But it would seem they did not do anything to mitigate another event despite this knowledge and experience.

    As many would have you believe, there needs to be considerably more local synchronous generation to provide better stability and reliability. Which IS true for the current way the NEM is operated.

    But we know the NEM IS in a transition towards more renewable generation, and the SA situation is simply the leading edge to it. A technological solution will need to be implemented to reduce or eliminate the dependence of synchronous generation and the “inertia” that it provides. To make any significant improvements to the network in the future this is essential.
    I am confident such a solution already exists, but AEMO’s mindset seems stuck in 20th century tech and unwilling to consider alternatives. It is unwilling to even allow the existing renewable technology to provide support to the grid despite the tech being able to do so much more than it currently is. A grossly under-utilized opportunity simply because they don’t fully comprehend its abilities.

    I believe this black system event and with Hazelwood closing next year, they’ll be forced into thinking more laterally. One of their current reports calls for more transmission lines between Victoria and SA, at something like a $1.4billion cost. This seems pointless if there is going to be less energy generated in Victoria to transfer. The money would clearly be better invested in another solution.

    Do I have that solution? Well, I am working on something. A work in progress that will need modelling and verification, but I believe it should work. But that is for another time to elaborate.

    • Chris Fraser 4 years ago

      Doesn’t this snafu highlight the following needs ? ;-
      1. Load shedding will in future be considered acceptable, because black starts are only for a crisis (such as loss of transmission towers). Many users will have seconds to a few hours of instant energy to tide themselves over. The whole network should be manageable in smaller pieces.
      2. All generators need to connect to grid through a load levelling device like a capacitor (to put off, temporarily, the additional worry of variable rotor speeds from overloaded generators) and use grid-connect inverters to match to 3. below.
      3. Phase-setting could be done via ‘synchronicity delegation’ – that is – understanding who is still putting out the right frequency and phase on an operable transmission line ? Or else create a low-powered signal, and make others follow.

      • Giles 4 years ago

        That would be a logical conclusion, but witness the bro-ha-ha when BHP was load shed when the interconnector fell down in December. It pretended it lost all power, and the coalition and the murdoch media, and many in fairfax, went crazy. when i pointed out it was just a load shed, BHP complained and the AFR accused me of nit-picking.

        • DevMac 4 years ago

          If you’re not pissing off the AFR then you’re doing something wrong.

    • MaxG 4 years ago

      Nice analysis… thank you for posting!

    • nakedChimp 4 years ago

      I hadn’t thought that all these turbines are just following and not supporting.

      I mean, wit those hybrid inverters at home it’s ‘simple’.. as soon as you disconnect them from the mains they run the island on their own.

      Baffles me that modern Turbines shouldn’t be able to do that (together with solar) and some battery backup I guess..

      Anyhow, thanks for the article in an article! 😉
      Very nicely written.

    • DevMac 4 years ago

      Great and interesting information, thanks for sharing. I very much appreciate the level of detail included. Whilst a lot of it goes over my head, it’s good to have this info in a public forum.

    • DJR96 4 years ago

      Thank you all for your words of encouragement.

      Amongst your replies Chris Fraser and nakedChimp, I think lies the end game solution.
      I envisage we’ll see some large inverter banks with battery storage in each region. When I say large I mean in the order of 20-25% of the regions typical daily peak capacity. First one would be in SA of course and could be 400-500MW, but would not need too much storage. 200-300MWh perhaps. As the proportion of renewable generation increases, more storage capacity can be added.
      Yes this would make them some of the biggest in the world, but the point is that they would be big enough to be properly grid forming. ALL generators would then follow the inverters set 50Hz frequency, even the existing synchronous units. It would be big enough to smooth out all the variations between supply and demand. The synchronous generators no longer have to control the frequency – which they’re not very good at. Frequency would no longer be a variable at all, making the whole notion of ‘inertia’ obsolete.
      Only then will we have a really reliable, stable and secure network.

      But you’re right Ray Miller, it is a bit of a leap for industry to get their head around. Incremental changes won’t cut it. Time for some bold steps.

      • Les Johnston 4 years ago

        Thanks for your input. It highlights the failed attempt of the regulator to be forward thinking and apply risk management. Phase and frequency regulation coupled with reactive power is a challenge for the 21st century. It has been done in Europe and could be done in Australia.

  6. Ray Miller 4 years ago

    Thanks DJR96 most enlightening and appreciate the post.
    One of the difficulties/challenges are problems with the engineering work force and training, while their are examples of exceptions, in general the bulk of engineers are experts at solving the wrong problem. Many engineers are not creative and have a very narrow field of view and lack assumption questioning, unless steps are taken to change the thinking process, improve the lateral thinking and identify what are the ‘real’ problems which need solving, expect more of the same throughout the energy transition.

  7. George Papadopoulos 4 years ago
  8. jukin 4 years ago

    I feel sorry for the sane people in SA.

Comments are closed.

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