What really happened in South Australia. And what we learned | RenewEconomy

What really happened in South Australia. And what we learned

There is a lot of noise about South Australia and not enough analysis. And there are some important lessons from the last few weeks of high prices.

Figure 2 Baseload futres financial year averages, Source: ASX

What you will learn from this article

  1. There is a shortage of “efficient” base load electricity supply in South Australia.
  2. The shortage is mainly due to the second half of Pelican Point been effectively withdrawn from the market. The AEMO was aware of the withdrawal and mentioned it in the October 2015 update on the South Australian “Electricity Statement of Opportunities”. Still, we can’t find any public announcement. That combined with, as it turns out, nearly simultaneous closure of Northern Power Station and the mismatch between those closures and the upgrades and the announced closures of the first half of Pelican Point have made South Australia dependent in the short term on very inefficient and very high cost open cycle gas generators.
  3. Had Pelican Point been operating at full capacity pool prices would have been at least $100 MWh lower in July and, we expect, possibly even more. Pelican Point’s owner Engie is perfectly within its rights to sell the gas and withdraw Pelican Point from service, if that is what it has done, but it could have been far more public about it, at least in our view. For what its worth, the AEMO report shows Pelican Point back in operation at 50% of capacity from 2017. This remains a poor outcome from South Australia’s point of view on the assumption that Pelican Point can ramp up and down as efficiently as say TIPS B (Torrens Island B). This is a technical point that your author has yet to resolve.
  4. The high pool prices are largely irrelevant to the prices South Australian consumers pay for electricity because the majority of the pool sales are hedged in one form or another. However, futures prices in South Australia are $30-$40 MWh higher than in the Eastern States at least for the next couple of years. That means household consumer prices could be 10-15% higher than those in Victoria. Industrial electricity prices in South Australia will also be higher than in Victoria by the same 30-40 $MWh but a bigger %. Industrial demand for electricity in South Australia is small (less than Tasmania) so from a national perspective this isn’t that important.
  5. For the renewable industry South Australia does throw up the challenge that it needs a lot of wind, far more than South Australia has right now, a lot of rooftop PV and some dispatchable renewables if South Australia is to remain a good example of high renewables penetration.
  6. The existing wind farms in South Australia are statistically highly correlated with total wind output. You would have to build a lot more wind farms in the area to significantly reduce the volatility of output. Almost certainly its better to build new wind in a different State or to build a different form of renewable energy less correlated with wind, or alternatively to store the wind energy and better correlate output with demand. Also wind output and demand are not correlated. In fact the correlation coefficient was negative .18 for the period.

Prices are high in the pool
In our view there is a lot of noise about South Australia and not enough analysis. It’s undeniable that pool  prices in South Australia have been high recently. Pool prices have been high across the NEM for months now despite flattish demand, but they have been especially high in South Australia.

Pool prices act mainly as a supply and demand indicator

Higher prices in the pool actually don’t mean very much for consumers in the first instance. Pool prices are basically just the mechanism for determining which generators will run. In ITK’s opinion at least 75% of pool revenues will be covered by hedging contracts of one sort or another. Those hedging contracts will either be baseload contracts for differences. Ie if retailer A and generator B agree to buy and  sell electricity for say $50 MWh  and the pool price is $200 MWh then retailer pays $200 and recovers $150 from the generator.

Alternatively the generator receives $200 and pays $150 to the retailer. Retailers generally cover the baseload or predictable demand in this way. But demand is neither completely predictable or constant so retailers also buy “caps”. Under these contracts generators and retailers agree that some price typically $300 MWh is the most the retailer will pay for a given number of MW of demand.  Typically the supplier of caps is an open cycle gas turbine which may in some cases run on diesel or aviation gas or hydro. This backup generation has low capital costs but high variable costs.

Retailers will typically remain exposed to some small fraction of demand as it may be too expensive to cover hedge cover 24 hours a day 365 days a year for that one or two half hours  a year when demand really peaks . A better description can be found at  hedging explained  .

Figure 1: Illustrative hedging product. Source: Productivity commission
Figure 1: Illustrative hedging product. Source: Productivity commission

So the real message from the high pool prices in South Ausralia is that at the moment there is a shortage of supply relative to demand, particularly when wind generator is low.

Higher pool prices have driven up futures prices

And the signal from the pool market is passed through to futures prices. Futures prices do drive the price that consumers will pay in South Australia. This chart is from our weekly note and shows that market participants expect that baseload electricity in South Australia will cost around $100 MWh for the next 2-3 years about $30-$40 MWh more than in other States.

Figure 2 Baseload futres financial year averages, Source: ASX
Figure 2 Baseload futres financial year averages, Source: ASX

Probably not much can be done about that because it takes 2-3 years to build new supply.

Why are pool and futures prices in South Australia high and why have they risen?

The reason why prices have risen are straightforward. (i) Northern power has stopped producing AND   the low cost gas producer Pelican Point has stopped producing.  This has made the electricity price in South Australia even more sensitive to the gas price. And the gas price has gone up.

Figure 3: Pool price and gas price compared in South Australia

Many of the gas generators in South Australia are very high cost (fuel intensive).

Whereas the closure of Northern Power was done in an orderly and well signalled fashion, we argue that Pelican Point was withdrawn from service in a “beneath the radar” fashion.  Still, contrary to what Tony Wood at the Grattan Institute seems to push, it’s a free market and companies can arrange their affairs within the law as best they see fit. No one can compel Engie to buy gas to start up Pelican Point.

The withdrawal of those two generators has meant that when wind is “off” generation has to come from higher cost gas  generators.

Gas generators are not profiteering at least not in South Australia

Its worth recalling that AGL recently announced that its gas profits would be down in FY17 as compared to FY16 and one reason is that they have had to pay more for gas to run their power stations in South Australia, TIPS A and TIPS B.

The following Figure shows the gas generation output, revenue and gas costs on a per generator basis in South Australia over the period 1 July to 17 July. Although the table shows “pool profits” in fact some generators will make less because they will have sold futures contracts. Specifically we won’t be surprised if generators such as Quarantine and TIPS and Osborne and Ladbroke Grove actually make less profit than shown below because of hedging and depending on how much their gas costs are also hedged.


Basically, if Pelican Point had operated at full capacity over the period it would  have displaced from Snuggery to Hallett and part of Quarantine.  Pool electricity prices would likely have been at least $100 MWh lower and possibly even less.

How has wind performed in South Australia

Unsurprisingly the wind farms in South Australia tend to run at the same time. There are 18 operating wind farms, but Hornsdale has only just commissioned. Over the brief Mar 1 – July 18 2016 period the average output was 501 MW with a standard deviation of 375 MW or 75%.

leitch replacement wind and prices

The  percentage standard deviation is smaller for the total than it is for any of the individual wind farms as expected. That is the portfolio variance is less than individual wind farm variance, but not by much. The average of the individual wind farms standard deviations is 100%. That is there is a 66% chance that the wind farm ouput is 100% above or below its mean at any point  in time.

Another way to look at this is to look at the correlation of the individual wind farms with the total wind farm output as shown in the figure below. In no case is it below 70%. Combined these two statistics tell us that it will take an awful lot of wind farms to reduce the volatility of the wind output in South Australia and that’s another reason why more transmission interconnections are going to be needed.

Figure 5: Wind farm correlation with total output in South Australia
Figure 5: Wind farm correlation with total output in South Australia

And wind ouput is not correlated with demand. Electricity demand doesn’t go up when or because the wind is blowing and vice versa. No surprises there.


What could have been done better with hind sight?

As far back as October 2015 it was clear to the AEMO that there was a potential problem in South Australia. They showed this graph in their Oct 2015 Statement of Opportunities update:


The obvious things that could have been done better are:

  • Made sure the transmission upgrades were fully complete before the closure of Northern Power.
  • The State Govt. could have bought some gas that could have been made available to Pelican Point.
  • The State Govt. could actually buy Pelican Point and outsource the running of it to any number of firms that would provide that service.
  • In the longer term more transmission could be built.
  • Most importantly some renewable energy that is dispatchable whether lithium storage (our preferred choice) or CSP could be built. Obviously not in the time frame to solve this year’s issues but going forward. The fact is that South Australia doesn’t have much coal and gas is going to be tight. Renewables are the answer and South Australia has a great opportunity to use this problem as the spur to move forward.


David Leitch is principal of ITK. He was  formerly a Utility Analyst for leading investment banks over the past 30 years. The views expressed are his own. Please note our new section, Energy Markets, which will include analysis from Leitch on the energy markets and broader energy issues. And also note our live generation widget, and the APVI solar contribution.

Print Friendly, PDF & Email

  1. aggri1 4 years ago

    Can I make a complaint? Why are the dark graphs attributed to NEM Review so terrible? Look at the South Aus’ demand and wind output graph: no scale on the vertical axis, double-spaces in the heading, and a totally uninformative caption. And the graph titled “wind output and demand are not correlated” is totally meaningless with no axis labels anywhere to be seen, and no caption at all.

    When presenting data it’s important to make sure that it can be understood. These graphs are very poor examples of how to present graphical data.

    • David leitch 4 years ago

      If there are faults they are mine and not those of NEM Review. I’m surprised you say the caption in Fig 4 is uninformative. Its true there are no axis labels on the wind output and correlation graph, and you might have added no figure number and a spelling mistake in the title. I didn’t think the axis labels (MW in both cases matter) since even if they were apples and oranges the point is there is no correlation.

      Can I invite you to volunteer to write a few articles and show me how its done.

      • aggri1 4 years ago

        Hi David. Could you shed some light on the “South Aus’ demand and wind output” graph: what are the units? I’m intrigued because the ‘wind’ doesn’t go below 50, and exceeds the ‘demand’ by a considerable fraction almost the whole time. And since the scale ends at 100, it leads one to suspect some sort of percentage, but what would the ‘demand’ line represent then? The periodicity of the ‘wind’ line is also suspect – almost weekly in fact…

        Other than some of the details of your graphs, I quite enjoyed your article. Thank you.

    • WR 4 years ago

      I’m pretty sure that the labels should be reversed in Figure 4. Wind output should be the orange line and demand should be the blue line.

      I have no idea what the values on the vertical axis in Figure 4 are meant to represent but they are proportional to the output and demand for that period. Basically, the 100 mark is equivalent to approximately 2300MW of production or demand. So the 50 mark is equivalent to half of that, etc.

      The vertical axis on the correlation graph is wind production in MW while the horizontal axis represents demand in MW. The point of the graph is that if the data produces a rectangular block like that shown, then that means that there is no correlation between the two variables.

      • David leitch 4 years ago

        Regarding Fig 4. On the reversal of the axes youu are correct. I will try to get that changed. I The vertical axis is a 48 half hour moving total of demand, so it effectively represents daily energy demand and wind output * 2. I will have strong words with the person that prepared that graph.

        • Giles 4 years ago

          Fixed now

        • juxx0r 4 years ago

          Do you reckon you can have strong words with the bloke that used milli dollars in the table?

        • juxx0r 4 years ago

          Exhibit C

  2. Tim Forcey 4 years ago

    Pumped hydro is another energy storage option, often forgotten. See: https://reneweconomy.com.au/2014/pumped-hydro-the-forgotten-storage-solution-47248

  3. David Pethick 4 years ago

    As always, another very worthwhile read.

    I don’t have time to pull together the data, but given weather patterns in Australia I would guess:
    * solar and wind within a state are either uncorrelated or negatively correlated
    * wind across SA and VIC is probably negatively correlated (given the predictable passage of high and low pressure systems across SA)
    * solar irradiation on the time scale of days and weeks is predictable (seasonal and diurnal patterns), wind is predictable on the scale of months and years.

    If that is all correct, technology diversification and regional diversification would seem to be a sensible transition path to renewables.

    The good news is, the market is working. I don’t think much more wind is going to be built in SA in the short term. The business cases for transmission operators is strengthened by these types of price events. The economics of “bolt on” batteries for 1-5 MW renewable generation units will improve over the next decade, and that will transform the generation stack.

    In short – it’s lumpy, messy and contradictory in the short term, but the market is working.


    Dave P.

  4. Farmer Dave 4 years ago

    David, thank you for your interesting articles, including this one, which I found very clear. When you present data for the NEM, why do you omit Tasmania?

  5. MrCyberdude 4 years ago

    Hello David Leitch,

    I have found your recent articles very interesting as well as informative in some areas.
    I take into account your understated disclaimers about assumptions posted throughout.

    Do you have an active Twitter or linked account I can follow?
    Did you previously work for UBS?

    Thanks for your articles, as you are one of the few people that can articulate the larger picture and some of the related issues, even if I may not agree with some of the conclusions.

    Kind regards

  6. Ursula Theinert 4 years ago

    Shortage of transmission capacity between SA and Vic/NSW is the primary cause of the extreme price volatility in SA. This has been a known issue for many years. Maximum import / export capacity on these ties determine the spot price volatility in SA.

    • Mike Dill 4 years ago

      Doubling the amount of solar and wind will also drag down the pricing, and the pricing is actually more stable than it was in the past.

  7. Peter F 4 years ago

    Thanks for the now corrected graphs.
    Another big opportunity is power to heat firstly in the form of expanded storage hot water. If each dwelling/hotel etc. had 100L of hot water per bed the extra cost is about $100-$200 per house compared to a standard system and it represents 2-3 days storage. At maximum demand they can draw about 300-500MW of excess renewables and in SA alone can store 6GW.hrs. This does not include high temperature storage for process heat or low temperature heat for space heating or ice for air conditioning. In addition to absorbing This flattening of demand tilts the shape of the demand curve not only in favour of renewables but combined cycle gas in comparison to OC.

  8. AllanO 4 years ago

    David, thanks for the article. If I could add a couple of observations:

    – a key proximate cause for the extremely high SA spot prices in early July was the fact that works related to the interconnector capacity upgrade were being undertaken on transmission lines in SA; these works resulted in significant short term *reductions* in the amount of power that could be imported from Victoria, just when wind output hit low levels and gas prices spiked

    – in a longer term context, the level and volatility of SA spot prices up to 30 June 2016 were not unprecedented – I’ll try to post a chart of rolling annual spot prices showing this. Obviously the causes of previous events and those most recently are different

    – the same chart shows equally strong percentage rises in NSW and Vic spot prices from their post-carbon tax lows, and also that up to 30-Jun-16 annualised spot prices in Qld were just as high as in SA – despite the very different supply mixes in those regions

    – finally, a slightly picky observation about wind farm output volatility. The distribution of output from windfarms is highly right-skewed and non-Gaussian, so standard deviation can be a misleading metric. In particular, if the calculated s.d. is 100% of the mean, that does *not* imply that there is a 66% chance of output being within +/- 100% of its mean, and especially not a ~1/6 chance of wind output being lower than 100% below the mean, ie zero or negative! I don’t have analysis to hand but I suspect the actual frequency of windfarm output being close to zero is more like 5%.

  9. neroden 4 years ago

    It is actually possible for Parliament to force Engie to buy gas and operate Pelican Point. This is how British-derived legal systems work. I’m not recommending it, but is is possible.

Comments are closed.

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