An estimate with precise knowledge – glad with that?

By Mike O’Ceirin

Energy Storage: An estimate based on actual data


Much has been written about the use of energy storage systems to stabilize renewable energies. This article uses current data to provide a theoretical answer on what is achievable. Both the demand pattern and wind power come from actual Australian data. This was put on a website for public access and should be referenced in this article. The conclusion is that in the Australian area in our eastern grid with wind and pumped storage power plants (PHES) 500 MW of wind and storage of 22 GW hours are required for every TW hour so that the stability corresponds to the existing demand pattern.

Requirements for maintaining the status quo

Australia faces five closings by 2034 due to the age of coal-fired power plants.

Asset name year Power MW Shipping 2020 State
Liddell 2023 2000 4.1% NSW
Vales point B 2028 1320 3.6% NSW
Yallourn W 2028 1480 4.5% Vic
Eraring 2030 2880 7.2% NSW
Bayswater 2034 2640 7.6% NSW

Since Australia’s electricity demand in the eastern grid was 203 TW hours in 2020, renewable energies have to deliver an additional 55 TW hours annually to replace them. This is not only possible on average. The same pattern of electricity demand needs to be satisfied. For this purpose, wind and PHES were theoretically chosen.

This estimate shows that 27 GW of wind generation and 1182 GW hours of energy storage are required to provide this amount of electricity under Australian conditions on the east coast.

The cost of capital for wind power would be $ 54.3 billion at $ 2,000 per kWh, according to Gencost 20-21 (page 17), plus the high cost of transmission lines. The costs for PHES can be found in the appendix storage costs.

Energy storage is nowhere near as determined as renewable energies. PHES is a tailor-made item with extreme price fluctuations. Certainly there will be a long timeframe for PHES to be built and sufficient production to meet demand. According to this assessment, the task of producing such a large amount of electricity storage can be quite prohibitive. For the Gencost estimate, the PHES would cost $ 62 billion and a Kidston (see attachments) would cost the equivalent of $ 456 billion. Note that these costs have been minimized by assuming readiness to operate at critical levels. In a practical world this would be unacceptable.

PHES estimate

This is only required with variable renewable energy, otherwise no energy storage is required. It applies in particular to wind and solar power generation. This article focuses on electricity from wind power. The technology for storage is limited to PHES. The reason for choosing the power storage system is its capacity and cost.

The estimate is based on data (see appendix data source) on the author’s website Spasmodic Energy. This estimate uses the electricity generation and consumption patterns derived from this data.

First, the demand must be known. The electricity demand on the Eastern Grid was 203 TW hours in 2020, an average of 23 GW hours per hour. This is not sufficient for calculating the energy requirement.

The electricity demand diagram (Figure 1) shows the pattern. Every day, demand is greatest in the early afternoon and lowest in the early morning hours. This is to be expected and must be taken into account when calculating the energy storage requirement. It is recommended that such estimates be made for a full year as there is a clear annual pattern. The lows and highs of the year for wind are included.

Using actual data, an estimate was made by stepping through each actual hour.

This is shown graphically on the Energy Storage page (Figure 2). In the graphic below, the course of the electricity output is shown overlaid with the actual wind power for the selected period. For each step, it is assumed that the starting percentage can be achieved. The steps to solve this problem are as follows.

  1. Is there enough electricity from wind to cover the demand for this hour?
  2. If so, add the excess to memory minus 15% (to account for memory leaks).
  3. If not, pull enough from memory (again, there is a 15% loss) to meet demand.
  4. If the memory has dropped to zero or below, the run has failed. Reduce the demand percentage by one and start over.

The performance is shown in Figure 2. This is a run for all of 2020 starting with an initial charge of 350 GW hours and an expected demand of 10%. Demand is an expected percentage of total demand for that year. As above, it can be seen (right) that 10% failed and 9% eventually settled at 8%. The result is then displayed in the two diagrams. The top one shows the charge in the storage tank throughout the year. In the example it is stable until the end of January, then it falls and continues to fall until the beginning of March, when it is practically zero for many days. The memory gradually builds up to be fully charged in early May.

Please note that the example shown is static, but the actual website is not. There the whole picture can be seen all year round by scrolling the display. This turned out to be very important because trying to display this amount of information on a static graph doesn’t show what is happening. In August, the charge drops back to below 50 megawatt hours, only to increase to a full charge at the end of August, which will last until the end of the year.

The second diagram shows the result. It is clear that energy storage reduces the amount of electricity that can be used. In 2020 the wind was sending 19.68 TW hours, but it was not stable. When stabilized by energy storage, the usable wind energy drops to 16.22 TW hours. The cost of stability in this case is 18%. There is the energy storage overhead of 30% and when it is fully charged, potential power generation is lost. This can be seen in the energy storage graph in Figure 2 in September (when scrolled).

Why is that? It seems very extreme. The point is, wind droughts are common and can last a long time. These can be found on the website under “Wind droughts”. In 2020, it is easy to find many cases where the capacity factor drops below 5%. There is one extreme case where the 33 hour average was 5.5%. There are others where the drop is down to 2%. Although the Australian east grid is very large, covering many thousands of square kilometers, the wind patterns are larger. The whole of Australia can easily have too little wind, but also too much wind.


In order to stabilize the wind energy generated in 2020 as required, 350 GWh of energy storage were required. That will cover 8% of the demand with the existing wind infrastructure in 2020. The wind capacity was 8 GW and the output of 8% is 16.22 TWh. For 1 TWh, 0.5 GW of generation is required. As a rule, 22 GW storage hours are required to stabilize 500 MW wind.

As already mentioned, five large coal-fired power plants on the Australian east network will be closed before 2034. Together they currently generate 27% of the electricity in the eastern grid. That is 55 TW hours, so that if the rule is applied, 1181 GW hours of electricity storage are required. At a minimum cost of $ 52 million per GW hour, that would add up to $ 61.5 billion.


Source data

Australia has a network that connects generators along the entire east coast. This network is monitored by the Australian energy marketing operator AEMO, who records and publishes price and performance data every five minutes. These data are used in this estimate. In order to enable the use of this data on a website, it was converted into hourly data. There are currently 150 million data points. Wind is used as a source of renewable energy that is widespread in Australia.

Pumped Storage Power Plants (PHES)

This estimate assumes that this can be provided with sufficient capacity. The Gencost 20-21 study (page 19) can be used to estimate the costs. A large PHES is to be built in Australia, known as Snowy Mountains 2.0 (SM 2). It is forecast to be able to deliver 2000 MW for 175 hours. This results in a capacity of 350 GW hours. The default test value is the same, but other values ​​can of course be applied to the website’s memory page.

Gencost gives costs per kW and these costs are valid for 48 hours. The cost of 2000 MW on this basis is $ 5 billion. Adjusting to 350 GW-hours means it needs to be multiplied and the result is $ 18 million from $ 2 million per GW-hour. As already mentioned, the Snowy Mountains 2.0 has the same capacity as this one. The often cited cost is $ 10.4 billion. That makes it a lot cheaper than the Gencost study would allow, although the cheapest from that study was chosen as much of the work is already done. It may be an exception.

Kidston is being built in Queensland in an old gold mine. There are clear differences compared to SM 2. It’s a lot smaller and a lot more expensive. The capacity is 2000 MW hours. The cost is $ 386 million per GWh.

Like this:

Like Loading…

Comments are closed.