AUSTRALIAN POWER PLANNING DEBACLE
FAILING TO CONSIDER WINDLESS NIGHTS AND THE COST OF GRID-SCALE STORAGE
The Australian Energy Market Operator is responsible for planning the transition to a grid dominated by wind, solar and batteries to reach the legislated target of net zero emissions by 2050.
Every two years it releases the latest version of the integrated system plan and ISP 2026 launched last week.
Previous iterations of the plan have been subjected to incisive criticism by James Taylor and Associates and others, including Aidan Morrison at the Centre for Independent Studies.
The plan runs over 100 pages with a great deal of impenetrable technical detail. There are also hundreds of pages in ten appendices with a great deal more in supporting information and it will take some time for people to mount considered criticisms. The situation will become more complicated when the Australian Energy Market Commission (AEMC) completes a review of the ISP framework of rules and regulations.
I will focus on a single aspect of the plan, the amount of storage required to ride through severe and prolonged wind droughts across Southeastern Australia.
THE BOTTOM LINE OF THE BATTERY COST
$130 Billion for one night and we have to be prepared for two or three in a row. And think about replacing the batteries every ten years.
The AEMO Calculation of Storage Requirements (extracted using Gemini)
AEMO does not use a single “windless night” static formula. Instead, it determines the necessary storage volume using a highly sophisticated, data-driven approach known as the Rolling Reference Years methodology.
Chronological Mathematical Optimisation
AEMO uses specialised energy market software (called PLEXOS) running two primary components: the Single-Stage Long-Term Model (SSLT) and the Detailed Long-Term Model (DLT).
The DLT model runs a chronological, time-sequential optimisation at half-hourly intervals. The model calculates required capacity by trying to minimise total system capital and operational costs while meeting strict reliability standards.
The Input: 16 Years of Real Weather Data
To capture true weather diversity, the 2026 ISP combines historical data from a rolling sequence of 16 real weather reference years.
The model looks at actual, historical half-hourly wind speeds, solar irradiance, and temperature-driven consumer demand across the entire National Electricity Market (NEM).
By simulating the grid against these real historical sequences, the model automatically encounters real-world wind droughts and consecutive windless nights (such as historical cold, stagnant winter high-pressure systems).
The Storage “Sizing” Logic
When the model simulates a windless night within these weather traces, it calculates storage requirements by balancing the following equation for every half-hour interval ():
To determine the depth (MWh) and power (MW) of storage required to ride through a windless night, the model optimizes based on:
1. The Day-Ahead Surplus: Can the storage fully recharge during the day using cheap, excess solar?
2. The Maximum Coincident Peak: How high is consumer demand during that specific windless block?
3. The Economic Trade-off: At what point is it cheaper to build a 4-hour battery versus an 8-hour battery, or rely on a gas-peaker to cover the final hours of the night?
Through millions of iterations over these 16 weather years, the model establishes the “Optimal Development Path”—ensuring there is mathematically enough storage capacity to survive typical windless nights without risking widespread blackouts.
The Alternative Approach – the Single ”Windless Night” Static Formula
The AEMO model is incomprehensible to politicians and everyone else outside the modelling unit in AEMO. In contrast, the alternative approach enables the storage required to be calculated on the back of a medium-sized envelope.
Consider a windless night with no coal power in the system.
16 hours of darkness and an overnight base load of 20 GW call for 320 GWh of energy to come from hydro, gas, and storage.
The demand for power is supposed to double by 2050. but sticking with the current demand for the calculation for the moment.
Hydro has to be used carefully to avoid draining the dams. Allow 4 GW through the night which is about twice the average use. That is 80 GWh.
Gas is too expensive to burn all night and the availability of gas in future is anyone’s guess.
Allow some wind because it is never completely still for sustained periods.
Bring the demand down to 200 GWh to make the arithmetic easier.
That comes to $130 Billion for one night and we have to be prepared for two or three in a row.
And think about replacing the batteries every ten years or so.
For comparison, the Federal Government spent $240 Billion on health goods and services in 2021-22, and $180 Billion on welfare.
The estimate can easily be improved by checking the relevant numbers to bring them up to date. It is a game that anyone can play, unlike the AEMO’s Chronological Mathematical Optimisation.
It looks as though the cost is not feasible. That is even before we encounter McBratney’s law that states you can build as much storage as you can afford, but you will never have enough spare power in a wind, solar, and battery system to charge it beyond a tiny fraction of the full capacity.
The wind-drought problem made the front page of The Australian late last year, thanks to Chris Uhlmann. He took advice from consultants in the field that there should be a national conversation about the adequacy of storage provisions, but this has not happened.
Indeed, it should have happened over a decade ago in light of the work by Paul Miskelly and Anton Lang. They clearly signalled the need for a 100% backup of wind power in a paper in the technical literature in 2012 and in literally thousands of blog posts by Anton Lang.
Despite this, the consultants did not indicate any awareness of the work by Lang and Miskelly, which apparently remains unknown in official circles, despite the efforts of Joe Nova and the Energy Realist of Australia to bring them to a wider audience.
Through 2020-22 we sent briefing notes to 800+ state and federal MPs and 100 journalists.
We gave up on the politicians but we are still circulating the list of journalists.
Rage, I’d be interested in why there is such a large gap between what the AEMO planners and others get in terms of storage and your straight forward analysis. I’ve seem comments in the past that the AEMO assumes perfect foresight and all storages are full at the start. I also wonder whether they account for the real variability and load carrying capacity of wind and solar. I suspect it will be a brave State Premier who allows closure of baseload plants without demonstrable evidence they aren’t needed. I very much doubt an AEMO model that suggests she’ll be right will do it, although you never know.
Much of this modelling sound like hyperventilating nonsense, based on unquantifiable assumptions. I note your concluding comment about having given up on the politicians. Quite understandable. Therein however lies the tragedy of this whole farcical debate.