We’ve noted a few tweets today from people who’ve noted that wind production in South Australia yesterday supplied over half of South Australia’s demand. Given interest in this topic, we thought it would be of value to put yesterday’s production in the context of what’s been happening over the past 6 months or so.
We used NEM-Review to produce the following trend of the output of both demand in SA and the aggregate output from all the wind farms in South Australia.
For completeness we also included the price range of SA spot prices on each of these days, as well.
From this chart we can make a number of observations, including the following:
1) We can see the wind production yesterday was indeed high – and that these high levels of production seem to occur every week or so, on average. In between these spikes in wind production are times of very low production (previously discussed here and here, etc…).
2) Because demand has been tapering off in SA in the past couple weeks (on the tail end of winter) we see that the percentage supplied (from the same peak volume) is higher now. This was the subject of a number of tweets.
3) We can see that the production of wind is somewhat reverse-correlated with demand in SA – as is also shown in the following correlation:
Update added on 18th September
Allan O’Neil added a question about the type of demand we used in performing this analysis, and in particular with respect to the second chart (above). This followed a similar query we received offline from people at AEMO.
For those who are interested, you can read more about three different measures of demand in this “Regional Demand Definition” document from AEMO – in summary, there are three different measures of demand = Native Demand, Operational Demand and Scheduled Demand.
The demand used in the three original charts in this post is Scheduled Demand, with the key point of interest here being that the Scheduled Demand does not take into account Non-Scheduled Generation (which is mainly wind generation).
Because of this fact, the apparent reverse correlation shown in the chart above is not as clear when we add back in the amount of Non-Scheduled Generation for each trading interval over the same date range to produce an approximation of Operational Demand – as shown in the chart below (in blue, for clarity):
From the chart above, we see that wind production and demand are still not positively correlated – but that the degree of negative correlation is not as strong.
4) We also see the reverse correlation between wind production and price (i.e. higher production from wind puts downward pressure on prices in SA – sometimes to the cost of the wind farm operators (or PPA holders) who have to pay back to AEMO whenever prices drop below zero overnight.
(a) In the chart above, we’ve capped the axis range from +$300/MWh to –$300/MWh in order that the scale of the numbers can be seen better, but keep in mind that daily peak prices rose over $4000/MWh and dropped to almost –$1000/MWh
(b) This is easy to understand, as the wind is dispatched with zero marginal cost and hence squeezes out the more thermal plant, which has a higher short-run marginal cost (and hence bid price).
(c) This reverse correlation can also been seen here, over the same date range: