During the year ending in June 2021, Australia's biggest electricity grid (the NEM) saw wind and solar each provide approximately 11% of electricity demand. What would have happened if we built much more wind and solar, so that they provided ~60% and 45% of our electricity respectively? How much storage would we need to manage the variability of wind and solar?
That's what this simple NEM (National Electricity Market) model is designed to test. Actual wind and solar data is scaled up by the required amounts. Short-term storage of 24 GW / 120 GWh (approx 5 hours at average demand) is used to help match supply and demand, which is also assisted by existing hydro. The aim is to see what fraction of our demand is able to be met by this combination of wind, solar, hydro and storage, and how much supplementary generation is also required. The images below show the results of the weekly simulations, in 4-week graphs in reverse order.
A mostly renewable model of the NEM
This simple NEM model takes the last 7 days of generation and demand data from OpenNEM, and rescales the wind by a factor of about 5.3x, utility solar by about 5x and rooftop solar by 3.3x. Why these scale factors? Well, that is what it would take to scale up the wind, utility and rooftop solar to provide 60%, 20% and 25% of our demand over the last 12 months, which is the target of this model. Note that the scale factors reduce slightly each week as more wind and solar are installed. Demand is left unchanged.
I aim to run this model each week, illustrating how close the wind, solar, storage and existing hydro could have to gone to satisfying demand for the week. Each week I show the results on Twitter https://twitter.com/DavidOsmond8 and every 4 weeks I update this website. The results also indicate how much "other" generation would have been needed to supplement the wind, solar, storage and existing hydro, and what the capacity of that "other" would needed to have been. In the sections below I also provide an estimate of the cost and CO2 intensity of the simulated grid. The figures below shows the results to date.
This is a highly simplified model of the NEM. It doesn't consider transmission requirements or congestion, it doesn't consider system strength or inertia requirements. It is a simple model that just tries to match supply and demand. Ignoring transmission constraints makes the model optimistic. Simply scaling the wind & solar without reweighting individual states makes the model conservative. In particular, reweighting the model towards QLD wind would be greatly assist, as it's currently underrepresented but is slightly negatively correlated to wind in other states.
A more complicated model that I built using 3-years of data that looks at each state in isolation, allowing inputs and exports from neighbouring states consistent with current or soon to be upgraded interconnection limits is described here:
But if you want to see the gold standard for a simulation of a mostly renewable NEM, then have a look at AEMO's 2022 ISP. AEMO is Australia's electricity market operator. Their ISP does consider transmission requirements and costs, congestion, system strength and inertia, all the while trying to minimimse costs while maintaining reliability and security. AEMO's step change scenario gets to 83% renewable in 2030-31 with 17 GW / 71 GWh of storage plus Snowy2.0's 2 GW / 350 GWh. It then rises to 97% renewable in 2040-41 with 43 GW / 221 GWh of storage plus Snowy2.0, but has demand approximately 50% higher than the present.
Costs
The following table & figure gives the approximate cost of electricity of this simulation, currently $86/MWh. It includes the cost of the wind, solar, hydro & "other", together with the cost of storage and additional transmission (beyond what is currently installed). The $/MWh figures show the assumed LCOE of each technology. For storage cost, I've assumed all 120 GWh of storage is in the form of battery storage, ignoring that the NEM already has approximately 22 GWh of pumped hydro storage from Tumut3, Wivenhoe & Shoalhaven. The battery capex cost was assumed to be $303/kWh, which was obtained from the 2021-22 CSIRO GenCost report, and is the average of the 4-hour and 8-hour battery cost over the 12 year period 2022-2033 in the Global NZE scenario. The battery is assumed to have a 12-year life, and the $/MWh figure in the table has been calculated by dividing the capex cost by the estimated throughput over its 12-yr life.
The annual transmission cost is based from AEMO's 2022 ISP. Its step-change scenario has an average annual spend on REZ augmentation and Flow Path augmentation over the years 2023-2050 of $1.11b to allow 292 TWh/y of additional utility wind & solar. This has been reduced in a pro-rata fashion for the 139 TWh/y of additional utility wind & solar used in this study.
Hydro was priced at a 30% premium to the price received by hydro over the 5-year period 2017-21. "Other" was priced at $600/MWh, based on the estimated LCOE for a peaker gas plant running at 4% CF with fuel cost of $15/GJ.
The final $/MWh figure was obtained by dividing the total annual cost by the annual demand. Note that excess generation is included in the annualised cost, but is not included in the annual demand.
Emissions
The following table shows two estimates of the emission intensity of the simulated grid. The first estimate includes only direct (scope 1) emissions from each technology, and arrives at a grid intensity of 8 kg CO2-e per MWh. In this calculation, "Other" is the only technology with direct emissions, assumed to be 700 kg CO2-e / MWh (appropriate for a modern reciprocating engine used as a peaking plant).
The second figure is an estimate of the full lifecycle emissions, including the emissions associated with building each technology. The table assumes the assumed kg CO2-e / MWh for each technology. The figure for the battery is based on assumed embodied emissions of 65 kg/kWh, divided by the estimated electricity throughput over a 12-year life. The resultant estimated full lifecycle emission intensity of the grid is 28 kg CO2-e / MWh.
FAQs
How did I choose the wind and solar scale factors? They are chosen to get wind, utility and rooftop solar to 60%, 20% & 25% of annual demand, which are my targets for this model. When this study commenced in Aug 2021, the annual penetration rate of wind, utility & rooftop solar on the NEM was 11.1%, 4.0% & 7.3% respectively, so the scale factors were 5.4x, 5.0x & 3.4x respectively. As each week passes and we install more wind and solar, the required scale factors will reduce. At week 50, they'd reduced to 5.0x, 4.0x & 2.9x respectively. When combined with ~8% from existing hydro and a yet to be determined fraction from 'other', I suspect the model will have 15%-20% excess generation over the year.
How many GW of wind and solar will it take to reach this target penetration? Using current demand levels as used in this simulation, it will take approximately 41 GW of wind, 19 GW of utility PV and 39 GW of rooftop PV, based on assumed capacity factors of 34%, 25% and 15% respectively. However it is likely demand will more than double over the next couple of decades as we electrify sectors such as transport and heating.
Why do your results so far have wind, utility and rooftop PV penetrations not equal to the targeted values of 60%, 25% and 20% respectively? The target values are for a full year. Some weeks and months are more or less windy than others. This simulation was started in late August, which usually corresponds to a more windy time of the year, so it is expected that this simulation will have higher than targeted wind amounts in the early months. As we approach summer then solar penetration will also be above average. Even after 12 months of simulations, the renewable penetrations will likely differ from the target values, as I cannot know in advance if the year will be more or less windy or sunny than average. But it is hoped they will be relatively close.
How have storage, ‘other’ and hydro been dispatched? It is assumed all 3 are highly flexible. They have been dispatched in way to minimise ‘other’ generation, subject to certain limits for hydro. Hydro is limited to between 200MW & 6,000MW, which reflect historical extremes. Weekly hydro is kept above 168 GWh. I’m aiming to keep annual generation to within the range of 10,000 to 19,000 GWh, though hopefully closer to 15,000 GWh or ~7.5% of NEM demand, again reflecting historical levels.
What is ‘other’? ‘Other’ is deliberately left undefined. It could be gas generation, or more preferably it could be a highly flexible dispatchable generator running on renewable fuels (biofuels, green hydrogen etc). For the purposes of CO2 intensity calculations, I’ve assumed “other” has an intensity of 700 kg/MWh which is typical of peaking gas or reciprocating engines. When calculating the renewable percentage of the simulation, I have excluded "other", even though it is hoped that in the future "other" does become renewable.
Does your model assume perfect foresight? For the first 6 days of each 7 day run, my model assumes 24 hours perfect foresight. That is, it knows exactly what the wind, solar and demand will be 24 hours in advance. AEMO is able to reasonably accurately predict these quantities 24 hours in advance (and beyond), though not perfectly. For the last day of my 7-day simulation, the length of the perfect foresight linearly reduces to zero hours.
Didn’t your other study only have 81 GWh of short-term storage? Yes, but my other study also included the 2 GW / 350 GWh Snowy2, which reduces short-term storage requirements. Short-term storage has also gone up due to this model having a slightly higher bias to solar, and due to this model being under-represented in wind from QLD, which is very slightly anticorrelated to wind in the other states.