Origin’s wind energy analysis based on hot air

Last month, Grant King, CEO of Origin Energy, gave a speech to CEDA – the committee for the economic development of Australia. In his speech, he was highly critical of the Renewable Energy Target (RET), and of the capabilities of wind energy. The inaccuracies of many of his assertions are detailed here, in a post by Giles Parkinson. I’d like to focus on one specific claim:

“So this is a day I’ve referred to in other presentations, in November last year in Victoria where Victoria had a record system peak for November. It was a very hot day, obviously, and you can see hopefully from this chart that at the time that demand peaked the amount of generation from wind was at a minimum. It went from 14 per cent to two per cent of energy provided at that time.”

King is referring to slide 9 of his presentation, copied above. This was indeed a hot day – a maximum of 39.6 degrees Celsius in the Melbourne CBD. And from the chart presented by King, it seems that wind performed abysmally, as Victorian trading interval price shot violently upwards.

As most analysts know, proffering a small period of time as a representative example of the long-term performance of a technology is questionable. It’s a technique usually summoned for Boltian analysis.

Let’s put this aside for now, and focus on the chart. Going by the terms ‘Market Intermittent’ and ‘Total
Intermittent Generation Levels’, it seems that the analyst involved has taken the total intermittent
generation across all states, and compared them to Victorian demand. I used a piece of software called ‘Nemsight’ to try and replicate their chart:

In Origin’s chart, the first peak seems to peak at around 14% in the early hours of the morning – this seems, tentatively, to match up with our data. The late-evening peak of ~8% also seems to roughly match. Price and demand seem to harmonise perfectly. Yet, the dip during the price spike only seems to go down to ~6%, as opposed to the ~1.7% shown in Origin’s chart.

‘Intermittent’ generation can mean a few things – it’s difficult to determine exactly what’s been used to create Origin’s chart. For ‘Intermittent Generation’, Nemsight uses a field from the AEMO database named ‘TOTALINTERMITTENTGENERATION’ – essentially, a mix of non-scheduled wind generators and a few other non-scheduled fuel types. Leaving out all semi-scheduled wind farms doesn’t seem like an honest representation of wind energy. What happens if we have a look at both non-scheduled and semi-scheduled Victorian wind farms?

Now, our dip matches up almost perfectly with Origin’s. But, our two morning peaks are lower than Origin’s data.

What’s going on? As we’ve seen, metrics of generation can be confusing, and hunting down the sub-calculations behind some of AEMO’s pre-processed data points can be perplexing. Yet, considering the gravity of the conclusion reached by King, we ought to know how they derived their chart.

The question is left open to Origin – whilst we await their response, we can investigate what actually
happened with wind energy on that day. To be sure, King is indeed referring exclusively to wind energy:

“You can see hopefully from this chart that at the time that demand peaked the amount of generation from wind was at a minimum.”

Let’s have a look at the amount of generation from wind, for that day:

Not bad – a total of 21,703 MWh over the course of the day. At the time demand peaked, 16:00 AEST (9386 MW), wind was producing a total of 864.6 megawatts – that’s 864.6 megawatts away from being ‘at a minimum’. It’s worth noting that wind was indeed muted in Victoria for those intervals – but as we already know, the NEM is interconnected, so focusing purely on a single state can be misleading.

The Bigger Picture

Let’s venture from single-state, short-period Boltian analysis. Let us tread instead to summer, as a whole – is wind energy always generating at a low level, as King shows in his single-day example?

The graphic below shows us all the half-hour trading interval periods between 00:00 29/11/2012 through to 23:59:59 28/02/2013, with Grant King’s chart period highlighted purple:

The indefinite nature of emphasising a single day is highlighted pretty clearly by this scatter plot. If you narrow your focus sufficiently, you will always get the answer you desire. In this case, King has found maximum demand, and minimum wind. As we widen our view, our concept of the trends in wind power over summer becomes thankfully clearer, but irritatingly more complex.

Looking at all wind and total demand across all states and taking the top 10% of demand intervals from summer (maximum demand of 31,774.69 MW), we can get an idea of how wind was generating during the summer peaks (with some inspiration for the formatting):

42 of the 100 trading intervals shown here saw wind pumping out more than 1,080 megawatts – more than 60% of installed capacity across the NEM. Variability does not equate to insignificance.

Wind varies with the machinations of the roiling atmosphere. Variability is managed by the market operator – a shortfall in intermittent generation is met with other fuel types.

Wind farms do not provide the market service of ‘capacity’ – they cannot be ramped up to meet the fierce extremes of Australian summer demand. Rather, wind farms contribute ‘energy’ – for instance, see page 16 of this AEMO electricity Report [PDF] – wind generated 26% of total generation in 2011-2012, in SA. This is a non-trivial contribution to total generation in FY12. Modelling [PDF] commissioned by AEMO suggest that a 100% renewable energy mix may be a viable possibility – variable fuel sources are not automatically disqualified from prominence in electricity supply.

King is quite correct in his assertion that a wind turbine is not the same as a coal-fired power station. This is no reason to sweep the technology off the table. The ability of wind power to contribute huge sums of energy to the national electricity market, combined with its minimal impact on environment and atmosphere is reason enough for us to pay close attention to those seeking to draw us backwards, rather than bring us forwards.

Data Files

Some of these files are quite large – be careful if you have a slow computer, or an old version of Microsoft Excel.

Origin Chart Analysis (Raw + working) [4.4MB]

Victorian Wind Generation and Electricity Demand (Raw) [30MB]

Victorian Wind Generation and Electricity Demand (Processed, Scatter plot) [43MB]

Total Wind Generation and Total Demand (Processed, bar chart) [0.7MB]

Ketan Joshi is a Research and Communications Officer at Infigen Energy. The views expressed above are his own, and not those of Infigen Energy