Victoria’s power system is proving more resilient than people thought

“Predictions are very difficult – especially about the future”, or “Past performance is no guarantee of future returns”.

Two well known sayings that remind us that whatever is predicted below may turn out to be wrong. But what is true – on the evidence so far – is that the predictions of imminent disaster with Victoria’s electricity supply have been wide of the mark.

In 2019, Victoria had 3 days of record or near record heat where the power system was highly stressed. On the first January 25, rolling blackouts were imposed, but in spite of the publicity they were relatively minor, about 800MWh was not supplied against total daily demand of 181,000¹.  i.e. an unserved energy level of 0.44% over the day.

It is true that a cool change moving through in the afternoon reduced peak evening demand, but by about 2:30pm on that day all power had been restored anyway.

Fast forward to the current summer, and on December 20 and 30 temperatures were also near records for the time of year, but although emergency conditions were declared (LOR2), there were no rolling blackouts.

On each of these days there were significant lack of availability of dispatchable plants (between 20 and 30 per cent of thermal capacity on both days), inter-connectors were offline for various times and there were significant constraints on delivery of wind and solar power from the West and Northwest of the state.

In the end the USE (Unserved energy demand) for the year was less than 0.002%, or less than 5 minutes per customer across the state. 95% of customer loss of power incidents are due to transmission and distribution faults, accidents and maintenance, so typical customers have 150 minutes of downtime per year.

Doubling generation reliability will move the 150 minutes to 147.5 minutes, a rounding error, but at a potential cost of billions of dollars every year across the state

For example:

On December 20, peak demand was slightly higher than January 25 at just over 9,100 MW. Moreover, it occurred around 6 pm when utility solar had fallen from a peak of 250 MW around the middle of the day to less than 110 MW, and wind was only supplying about 400 MW.

Earlier in the day exports from SA had been severely restricted and wind and solar output from the west of the state had been constrained well below potential output. At the same time Loy Yang A2 and Mortlake 2 were completely offline.

By 8 pm, when solar had completely disappeared, wind had only picked up to about 640MW and demand remained at just under 8,500MW² (vs 6,400 MW on January 25th) yet demand was met.

On December 30 at 3 PM when tracking solar should have still been not far off peak 320 MW of capacity, it was only supplying 88 MW, Brown Coal 3,600 MW of 4,720 MW nominal capacity, hydro 590 out of 2,100 MW and gas 1,320 out of 2,400 MW and net imports were 1,230 MW out of 1,900 MW, so there was still significant spare capacity.

By 5:25 pm when grid demand peaked at about 8,300 MW ³, there were two large coal generators offline, LYA2 and LYB 1, Mortlake 2 was only running at half power, in total 7,100 MW of coal and gas capacity could only manage 5,070 MW and the Dederang link was restricted. Still, AEMO data shows Victoria exporting to 570 MW to NSW and yet still there were no blackouts when most of the state was still well over 40C.

On both days, faults with interconnectors, and lack of capacity on coal generators caused Lack of Reserve notices, but they were managed without blackouts and at no stage did combined gas and hydro reach even 90% of capacity.

The interconnector and or generator failures that did occur could easily have been mitigated with about 300 MW/3 hours more storage/demand response which would have provided a buffer to allow idle gas turbines to power up.

Update: In spite of all the prognostications about Victorian vulnerability, on 4^th of January when the link between NSW and Victoria was interrupted by bushfires, it was NSW which was nearly short of power with prices reaching  $14,700/MWh at around 3:30pm NEM time vs -$3 to -$230/MWh in Victoria at the same time.

After 16:25 when the link was partially restored ,Victoria was exporting almost 20% of its generation to NSW, SA and Tasmania, and spot prices had risen to $34 in Victoria even though Victorian  coal plants were still off line, but remained at $14,700 in NSW with all import links running at capacity and two potlines at Tomago offline and a public appeal resulting in a further 200-300MW reduction in demand⁴. NSW Hydro was zero so apparently bushfires were impacting transmission from the Snowy north as well as to Victoria.

There are some pertinent observations

1. Long distance transmission assets have proved vulnerable to storms, bushfires and thermal derating. It would seem that their value as backup is regularly overstated, while their annual operating and maintenance costs are usually completely ignored.
3. More generation and storage close to loads combined with expanded demand response would provide more system resilience at less cost than more transmission.
5. Compared to international practice we still only have a little over half the demand response we should have, while most similar grids are significantly expanding their share of controllable loads, as we should be.
7. A very large share of demand is temperature dependent. For example, peak Victorian demand on Friday December 20th was over 9,100 MW even when many educational and industrial facilities were phasing down.
9. On Friday, December 13, when most of those loads were still at full scale operation, peak demand only reached 4,800 MW.  This indicates that a strong push for both building efficiency and cooling system efficiency/demand response would have a significant impact on peak demand.
11. Milan is hotter than Melbourne in summer yet peak demand per person in Italy is roughly half that of Victoria.
13. The system survived near record demand, (after allowing for the reduction of industrial demand in the last 10 years) even when 20-35% of the dispatchable generation/interconnectors were unavailable, how much more security do you want?
15. By next summer there will be another 600 MW of solar and about 1,800 MW of wind, and 50-100 MW of storage further reducing reliance on thermal generation.
17. The addition of new conventional power sources would make little difference to customer experience, because even another 1,000 MW of conventional capacity would have reduced average lost time in Victoria in 2019 by less than 3 minutes.
19. There is a real possibility that the peak price suppression effects of 220 MW new gas capacity proposed for Victoria combined with increasing renewable output will reduce the annual profits of coal plants by enough to make continued operation of one or even two units at Yallourn uneconomical within the next 3-4 years, thus actually reducing dispatchable capacity.
21. Transmission issues from the north and west of the state need to be resolved, this may by new links or by storage at both ends of the existing lines or just replacing the existing conductors and transformers or even just better understanding of stability issues.
23. There are many techniques for improving stability of inverter-based networks which have not yet been fully proven but which appear to have far better response rates than traditional, inertia/governor systems on gas and steam turbines.
25. As more confidence is built in these techniques, which can be implemented quite quickly, minimum output from wind and solar plants will rise, further improving system resilience.
27. Continued additions of high capacity factor wind turbines and east/west or tracking solar will shorten the periods of low renewable supply and increase minimum renewable output faster than the nominal increase in capacity. Example, Coonooer Bridge farm generates above 50% of capacity for twice as many hours per year as McArthur.
29. At 6 pm on January13,  large scale output -mostly tracking types were still at 60% of peak output whereas rooftop solar had fallen to 38%

The only substantial risk is a short-term simultaneous loss of at least 30-35% of coal, gas and interconnector capacity on a very high demand day. There is no period in Victoria’s history even with Hazelwood and Anglesea open, where a loss of supply of that magnitude would not have caused severe blackouts.

However, due to the demonstrated unreliability of those assets recently, such a failure is possible but unlikely. But even that risk is being ameliorated by the month as more renewables come online and power system management improves.

This is not to suggest that there is not a lot of work to be done as the power system transitions, but simply to state that we are not at panic stations and the transition is manageable. After all, the UK went from 43% coal in 2012 to 2% last year, while nuclear also declined.

Germany, with far less diverse grid less wind and sunlight, already generates more energy from wind and solar than the entire production from all sources in Australia, and this year it will reach almost 50% renewables.

The transition can be managed.

In our case an indirect indicator of the adequacy of the system is the rapid fall in Victorian spot prices from $91/MWh in December 2018 to $61 in December 2019 and from $146 in January 2019 to $83 so far this January [i] .

In conclusion, the Victorian grid is already much more resilient that most commentators claim or foresaw.

The ongoing additions of wind, solar and small amounts of storage and minor strengthening of transmission and improvement in operating procedures will only further improve the system.

1. Reference:
3.  https://www.aemo.com.au/Electricity/National-Electricity-Market-NEM/Data-dashboard#operational-demand
4. https://anero.id/energy/wind-energy/2019/December/20
6. https://www.smh.com.au/politics/nsw/every-megawatt-counted-how-nsw-dodged-major-outages-amid-record-demand-20200105-p53owm.html
8. https://www.aemo.com.au/Electricity/National-Electricity-Market-NEM/Data-dashboard#average-price-table