Pumped hydro more expensive than batteries: battery replacement(s)

Even when accepting the flawed calculations of Ronald Brakels, it provided only a fragile win for the battery scenario. This prompted him to start a new calculation in order to justify the battery/solar setup. It involves a battery system that is currently being built: the Victorian Big Battery. His reasoning is that battery prices drop rapidly and he also proposes a way to set aside some the initial investment in order to replace the battery at the end of its economical life.

Let’s just jump right in. This is the information he gathered about the Victorian Big Battery:

• Capacity: 300 MW
• Storage capacity: 450 MWh
• Price: AU$180,000,000
  (again no justification for this price tag, just his hunch)

Then he repeats the flawed calculation by calculate the price:

• Per kW: AU$180,000,000 / 300,000 kW = AU$600/kW
• Per kWh: AU$180,000,000 / 450,000 kWh = AU$400/kWh

This price per kWh is a pretty meaningless number … yada, yada, yada (if you read previous posts, you already heard that a couple times by now).

He now proposes three different scenarios. Let’s start with the simplest one. He again compares per AU$5,000 worth of Snowy Hydro 2.0 and this amount will buy a battery with a:

• Capacity of AU$5,000 / AU$600/kW = 8.33 kW
• Storage capacity of AU$5,000 / AU$400/kWh = 12.5 kWh

He also proposes a battery/solar scenario, similar to the Hornsdale/Snowy Hydro 2.0 comparison:

• 2 kW of solar at AU$1,000 = AU$2,000
• This means 5,000 – 2,000 = AU$3,000 left and this is enough budget for a battery having a:
  • Capacity of AU$3,000 / AU$600/kW = 5 kW
  • Storage capacity of AU$3,000 / AU$400/kWh = 7.5 kWh.

His reasoning is that this setup will bring in 5 times more capacity than Snowy Hydro 2.0, almost double the storage and an average of 10 kWh of solar (to store in the battery or sell the surplus on the market). He again tries to dimension the system using averages. A storage system should be dimensioned using the maximum needed capacity to meet demand. If the operational capacity is 40 GWh for 2 GW (20 kWh per kW), then this battery setup is insufficient to meet this demand.

The third scenario is a new type of scenario with only half of the budget, setting aside the other half for replacement when the battery reaches its end of life:

• AU$2,500 is enough budget for a battery having a:
  • Capacity of AU$2,500 / AU$600/kW = 4.17 kW
  • Storage capacity of AU$2,500 / AU$400/kWh = 6.25 kWh
• AU$2,500 left “to put towards future, even cheaper, replacements”.

All those calculations have the same issues as the previous. The storage capacity of Snowy Hydro 2.0 and the Victorian Big Battery are vastly different. Although the battery has a better storage capacity/capacity ratio (1.5 versus 1.29), that still pales in comparison with Snowy Hydro 2.0 (between 20 and 100+).

Not taking lifespan into account
This setting aside for future replacement is an interesting scenario that allows Brakels to not take lifespan into account. The lifespan of both installations is vastly different. It is projected that Snowy Hydro 2.0 would have a lifespan of 100 years, but Brakels decided that 70 years would be more than enough (again without giving arguments why). Batteries don’t last that long (he assumes 15 years), so they will have to be replaced (several times) over the lifespan of Snowy Hydro 2.0. Therefor it is not possible to compare Snowy Hydro 2.0 with only the current investment of batteries.

Yet, that is exactly what Brakels does. He justifies this by claiming that the Hornsdale Power Reserve as well as the Victorian Big Battery setups could generate enough money for their own replacement(s) in the future, which I think could be the second meaning of “return” and “fail” in his claim that he “can’t be certain that the additional return from the battery setup will be enough to replace them when they fail” (for the first meaning see previous post). In that sense, he could not declare a winner because of the possibility that the additional revenue generated by the battery might not be enough to replace it when it comes to the end of its economical life.

Replacement in the Hornsdale/solar scenario

Comparing batteries having a different role
In case of the Hornsdale battery comparison, he claims that it could (maybe) generate enough money for its own replacement every 15 years by:

• Take advantage of brief periods of high electricity prices by supplying up to four times more power
• Take advantage of periods of low electricity prices by charging up to four times faster
• Lower losses due to higher efficiency
• Making revenue from its solar component selling energy directly to the grid.

The first two reasons seem taken from the services that the Hornsdale Power Reserve is providing (frequency control and arbitrage). I understand that the Hornsdale Power Reserve earns quite a buck from these services, but I doubt that the same mechanism will also apply in the new situation. Batteries are surely king when it comes to frequency control and arbitrage, but in this scenario the batteries will take the job of bulk balancing. Not sure whether the super fast reaction speed will be much of an advantage here. Also, batteries will have the disadvantage of the very limited storage capacity compared to pumped hydro.

Currently the Hornsdale battery eats the lunch of gas-fired power plants that managed this frequency control until then. Once that frequency control niche is filled, any additional battery capacity will lead to less revenue. It is not guaranteed that batteries will earn as much as the Hornsdale Power Reserve.

Over-estimating “efficiency”
When it comes to efficiency, I agree that a battery has a higher efficiency than hydro, but I doubt that the difference is that large. I recently came across an IEA article observing that the efficiencies of both utility-scale batteries and pumped hydro are around 80% (more specifically 79% for pumped hydro and 82% for utility -scale batteries). Brakels uses the term “efficiency” for both, but what exactly does he mean by that? In the report he refers to, it is stated that this 76% for hydro is “cycle efficiency” (round-trip efficiency) which is the ratio of the amount of electricity that can be retrieved from storage to the amount that was originally put in storage. In this case, when 100 MWh would be put into storage, about 76 MWh could be retrieved. The other 24 MWh are (charging as well as discharging) losses.

My guess is that this 90+ figure is only the one-way efficiency (charging or discharging efficiency). When charging and discharging efficiency are both 90%, then the round-trip efficiency is about 81%. This matches the observations of Harley Mackenzie and Jonathon Dyson who analyzed the output versus consumption ratios of the Hornsdale Power Reserve and found a round-trip efficiency of ±80%.

Granted, 80% is still more than 76%, but the difference is not that big anymore.

Concerning the last reason, selling surplus solar power can indeed earn money for the battery setup, but that might be less than expected. When there is a lot of solar capacity installed, then the generated electricity might lead to a surge of production around noon, meaning lower (maybe even negative) prices when demand is low. Remember, the battery setup would add quite some solar into the mix: the Hornsdale scenario would add an additional 2.88 GW of solar to the grid, the Victorian Big Battery scenario 4 GW.

Even with his flawed calculation, it is indeed difficult to declare a winner.

Replacement in the Victorian Big Battery/solar scenario

Over-estimating the price fall of utility-scale batteries
In the Victorian Big Battery scenario another strategy is used. Half of the amount is set aside for future replacement in the assumption that the battery prices will be much lower than at the time of installation. He cautiously assumes a price fall of around 20% per year (my emphasis):

In US dollars, they’ve declined 80% from 2013, which is an average fall of 20% per year. While there’s no guarantee their price will fall as fast in the future, they will get cheaper.

A price fall of 20% per year seems very optimistic. He bases this 20% price fall on this graph representing the price fall of car batteries from 2013 until 2020 (in 2020 US$/kWh):

Sure, the average price fall is a tad below 20%, but it is clear that this is not the trend going forward. The price fall is getting smaller each passing year. So yes, I agree fullheartedly with his statement that “there is no guarantee their price will fall as fast in the future”. If this trend continues, then prices will more or less stabilize in the not that distant future.

This reminds me of this post about two Australian scientists who extrapolated an increasing solar and wind energy trend of about 20% per year, yet the trend in their reference period was getting smaller each year, just as in this case. They got to their 20% number by averaging the (decreasing) trend of the last 5 years, therefor inflating the extrapolation of that trend into the future, just as Brakels did. There must be something in the water in Australia… 😉

The graph is US data on car batteries. This made me wonder whether the data on car batteries equally applies to utility-scale batteries. It is one thing to make a car battery pack, it is another to make a big battery, make it weatherproof, adding a battery management system, account for more intensive usage and so on.

Brakels didn’t need to find the prices of car batteries in the US to show the price trend for utility-scale batteries. That was also in the report he linked to, in the chapter on battery costs. In the introduction of that chapter, it was stated that storage battery prices fell from AU$2020 in 2009 to AU$450 in 2020. That is almost 80%, but now over 12 years. That is an average of just below 7% per year.

Later in that chapter, a graph is shown depicting the trend from 2017 and also a projection until 2040:

The first years see a substantial drop in prices, but the price fall is getting smaller each year and is expected to level out in the future. Looking at the trend of the graph, the storage battery price in 2017 is somewhat over AU$1,200/kW, lowering to somewhat above AU$1,000 in 2020. That is an average 5.5% drop per year, well below the drop of the car battery packs over the same period and well below the 20% assumed by Brakels. This is just an average. It ends with almost 4% in 2020 (and ever lowering into the future). So, he is seriously over-estimating the price fall of the storage batteries.

To conclude this post, the new calculation has the same flaws as the previous one, plus some additional ones. He is now assuming that the new role of the battery will bring in a similar amount of revenue, he is comparing round-trip efficiency of hydro with one-way efficiency of batteries and over-estimated the price fall of battery prices by using the average price fall of a different type of battery. Whether this new battery setup could match the performance of Snowy Hydro 2.0 is not really clear. Although the numbers of the Victorian Big Battery scenario are better than the Hornsdale scenario, it is still not an apples-to-apples comparison and he is again dimensioning the system using averages, which is a pretty meaningless thing to do.