May 2024 Variable Renewable Energy Drought and the ISP in 2040

We recently published an article on the 20-27 May 2024 Variable Renewable Energy (VRE) drought. For that article we estimated the hourly capacity factors of utility scale wind and solar by NEM region. The severity of this drought on wind generation particularly in NSW, Victoria and SA invited the question as to how the future grid would perform under the same conditions when gas powered generation (GPG) and hydro are the only genuinely dispatchable sources of electricity. To obtain an indication of this we have utilised AEMO’s 2024 Integrated System Plan (ISP) eight-day renewable drought in 2040 scenario for the NEM, excluding Queensland. 

Figure 1 illustrates how AEMO’s VRE drought scenario for the year 2040 turned out (hereafter referred to the 2024 ISP VRE drought scenario).[i] The first point to note is the figure shows 10-days of which the first two days are not a drought. The actual drought period is 21-28 June and this is the period our analysis will cover. The start of the drought is obvious because wind output drops significantly after 20 June and doesn’t show signs of recovery until 27 June. The other key point to note is the scenario does not result in any unserved energy (USE).

Figure 1: 2040 operability through eight-day renewable drought, NEM except Queensland

To explore these results further (see Table 1) we have derived solar, wind and rooftop MW capacities for 2040 using 2024 ISP data. This provides capacity data by financial year and the average of FY40 and FY41 capacities are taken as the 2040 estimates. Based on these capacities and the output in the 2024 ISP VRE drought scenario, we have estimated the capacity factors for each generation type. The average capacity factors for 2040 in the 2024 ISP are also shown for comparison purposes noting that Queensland is included in this calculation.

Under the 2040 ISP VRE drought scenario, wind’s capacity factor averages 14.7 per cent which is less than half of the annual average of 36.2 per cent whereas the differences for solar are relatively trivial for both utility and rooftop solar. Comparing the 2024 ISP VRE drought scenario with the May 2024 VRE drought shows almost no difference between both the utility (17.5 versus 17.4 per cent) and rooftop (9.3 versus 9.4 per cent) solar capacity factors. The rooftop output is interesting because while it shows a reduced capacity factor of 9.4 per cent versus the average annual of 14.2 per cent there does not appear to be any self-consumption of the output. Estimates we have seen indicate self-consumption is anywhere from 25 per cent for large systems and up to 70 per cent for the smallest systems.[ii]

Unlike solar, the 2024 May VRE drought wind capacity factor of average 9.4 per cent is significantly lower than the 2024 ISP VRE drought scenario average assumption. It is a third less than the 2024 ISP VRE drought scenario and nearly three quarters below the annual average in the ISP for 2040. A 9.4 per cent capacity factor translates to average hourly wind output of 4.02 GW compared with 6.29 GW in the 2024 ISP VRE drought scenario and total wind generation decreases by 435 GWh. Despite this substantial reduction in wind output, a footnote in Appendix 4 of the 2024 ISP states:

“Recent VRE low events in April and May 2024 showed extreme low wind productions across southern states. However, the solar outputs were still relatively high. As new solar capacity is forecast to increase in the future, the severity of the effects of these recent wind droughts is similar to the example conditions explored in this section.”[iii]

Nevertheless, we believe it is worth exploring further and our results are not reassuring because there is load shedding with both hydro and gas stretched to their limits.

Table 1: Inputs and Outputs for 2024 ISP 8-day VRE drought in 2040 and May 2024, 8-day VRE drought (Queensland excluded)

Sources: AEMO 2024 ISP Chart Data, AEMO 2024 IASR, AEC analysis and AEC Energy Insider Dunkelflaute writ large - May 2024?[v]

Gas and hydro work hard in 2024 ISP VRE drought scenario

Table 2 sets out the inputs and outputs for gas and hydro. The maximum hourly generation for each in AEMO’s data is assumed to be the total capacity for each in the NEM with Queensland excluded. Over the eight days, 16.5 PJ of gas are required to generate 1.5 TWh with a maximum daily consumption of 2,459 TJ/day. Hydro also runs hard with an average capacity factor of 79 per cent.

Table 2: Gas and hydro inputs and outputs under 2024 ISP VRE drought scenario

Figure 2 illustrates the gas and hydro capacity factors on an hourly basis. It also shows the one day moving average of gas consumption. As can be observed, hydro is performing base load duties and gas is the same during the drought but very rapidly turns off when not required. In Appendix 4 of the 2024 ISP, AEMO conducts analysis which assumes there will be adequate gas reserves but no increases in storage and pipeline capacity. This means that gas supply for generation is limited to 1,500-1,800 TJ/day and any generation above this limit will have to be powered by liquid fuels (diesel and kerosene). Accordingly, AEMO assumes new entrant gas powered generation (GPG) has enough liquid fuel storage to generate for 14 hours. Whether this would be adequate to get through the drought we do not know.

Figure 2: Gas and hydro capacity factors one day moving average of gas consumption under 2024 ISP VRE drought scenario

9.4 per cent average capacity factor for wind scenario

The data that underpins Figure 1 is all hard coded, which makes it difficult to analyse and stress test such as incorporating a more extreme average capacity factor for wind.  To incorporate the 9.4 per cent wind capacity factor profile and obtain results we need to make the data dynamic.[vii] To calculate usable generation we sum storage, gas, hydro, rooftop PV, utility solar, imports, demand side participation (DSP) and subtract storage charging. Generation is then subtracted by load. Interestingly, after these modifications the calculations produce 3.21 GWh of USE with a maximum half hourly interval of 0.17 GW USE. However, there are also intervals with excess energy (after adjusting for 24 GWh of exports) totalling 12.8 GWh with a maximum half hourly interval of 0.95 GW.

Figure 3 illustrates the impact of the 9.4 per cent wind capacity factor. Small and large solar output has not been changed because they have no significant difference from the May 2024 VRE drought. As one would expect, Figure 3 can only be described as ugly, with 401 GWh of USE, a half hour maximum of 5 GW and no exports are possible. Because the storage charge and generation profile has not been altered and remains static there are some intervals where storage is charging and creating USE, which is clearly illogical.

Figure 3: 2024 ISP VRE drought scenario: non-wind generation and load fixed and wind average capacity factor reduced to 9.4 per cent

Thankfully, there is scope to increase gas and hydro output to reduce the level of USE and Figure 4 illustrates how this impacts the results. The calculation for each interval where there is USE is as follows:

1. Spare hydro capacity is dispatched to eliminate USE;

2.If there is still USE then spare gas capacity is dispatched to eliminate USE; and

3.If there is still USE then it is recorded for that interval.

Hydro and gas dramatically reduce the USE to 66 GWh with a maximum half hour of 2.4 GW. This is still a large amount of USE and represents 1.2 per cent of the total 8-day load.

Figure 4: 2024 ISP VRE drought generation and load fixed except gas and hydro and wind capacity factor reduced to 9.4 per cent

Hydro runs extremely hard at an average capacity factor of 96 per cent and operates at 100 per cent for extended periods (see Figure 5 and Table 3). This increases its hydro output by over 200 GWh. Gas averages 82 per cent and like hydro runs at 100 per cent for extended periods and its output increases by over 100 GWh. Daily gas usage stays at 2,500 TJ/day for extended periods and nearly 18 PJs are consumed. As described above, any daily consumption above 1,500 -1,800 TJ/day will have to be supplied by liquid fuels. As noted previously, it is uncertain as to whether there would be enough liquid fuels to run this hard.

Figure 5: Gas and hydro capacity factors and one day moving average of gas consumption under 9.4 per cent average capacity factor for wind scenario

Table 3 Gas and hydro inputs and outputs under 9.4 per cent average capacity factor for wind scenario

While this analysis is based on a simple modelling exercise and has many limitations or we may have missed something, it has used empirical capacity factor data. Table 4 sets out some factors that could alter the results by either increasing or decreasing USE.

Table 4: 9.4 per cent wind capacity factor modelling assessment

Conclusion

In our view, what is presented in the 2024 ISP VRE drought scenario does not transparently demonstrate how the grid would perform in a VRE drought. This is because it is not possible to replicate the ‘black box’ results and test these in a relatively simple higher level ex post model. In a similar manner to how we assess historical NEM performance where we can calculate generation by fuel type and region, regional VRE capacity factors, load and interconnector flows. This is not a complete list but some things that could improve transparency when modelling a VRE drought scenario:

• Hourly or half hourly utility wind and solar generation by NEM region.
• How is the load determined? For example, is it an average winter profile, something more extreme, etc.
• How CER resources are determined. For example, rooftop PV capacity factors by region and self-consumption assumptions.
• Probability distributions for regional VRE capacity factors as opposed to impenetrable weather analysis.
• Storage capacity information by region.
• The response of hydrogen load. The ISP forecasts 35 TWh of hydrogen load for 2040 and as set out in Appendix 4 it is expected to vary its demand across the day and provide demand side response. However, we cannot locate any useable data to incorporate this.
• The modelling is silent on interconnector flows within the NEM excluding Queensland.

[i] AEMO 2024 ISP Figure 24.

[ii] https://www.sunwiz.com.au/solar-pays-its-way-on-networks-its-no-free-rider/#:~:text=way%20on%20networks.-,It's%20no%20free%20rider,other%20country%20in%20the%20world.

[iii] https://aemo.com.au/-/media/files/major-publications/isp/2024/appendices/a4-system-operability.pdf?la=en, p24.

[iv] Derived by applying ratio of utility PV average capacity factor in ISP with May drought average and then applying the proportional difference to the average small PV capacity factor.

[v] https://www.energycouncil.com.au/analysis/dunkelflaute-writ-large-may-2024/

[vi] Heat rate is assumed to be 11 GJ/MWh and is constant with no start stop gas.

[vii] The 9.4 per cent average wind capacity factor series is created by multiplying the original wind output series by 0.64 which is the of 9.4 per cent ratio to 2024 ISP VRE drought scenario average capacity factor of 14.7 per cent. It is shown in Table 1.