Electricity supply part 4 - Costs

In Part 1 of this series, we saw how electricity demand changes on an hourly basis, and that if we want to power the electricity grid using renewable energy sources, we are going to need energy storage.

Part 2 showed a way of estimating how much storage we would need, if we just  nicely meet our average energy demand using renewables without wasting any. (It was a lot).

In Part 3 we started working with the idea that if we're prepared to build more renewable energy generation than is strictly necessary, we could get by with a less storage - a lot less.

Let's have a go at costing this up.

Assumptions

In 2021, the electricity price paid by consumers is about 2/3 network, environmental and retailing costs; and about 1/3 generation costs [1].  What we spend through our electricity bill has to cover not only the construction costs for solar and wind farms, the cost of coal and gas; but also all of the office costs,  advertising, insurance,  network maintenance and management. And a gazillion other things you've probably never thought of.

I'm going to assume that the non-generation components of residential electricity price remain constant at about 20c / kWh ($200 / MWh). This is unlikely, because wiring a whole lot more renewable generation and storage into the grid will almost certainly require network  upgrades. So this assumption will result in an underestimate of the costs. I'm also assuming that this residential value applies to every user over the network.

I'm going to assume that the currently installed generation capacity in NSW is, on average, neither insanely profitable, nor loss making. Electricity generation companies don't pay fantastic dividends. They make a small profit. It's a competitive industry. 

If assumption this is true, the income earned by the different generation types in NSW is what's required to make those generators acceptably profitable. 

Making this assumption, in 2021 in NSW, black coal earned $80 per MWh produced, gas earned $185, wind earned $64 and solar $45 per MWh produced. Note that at any given  moment in the grid, all generators earn the same price for the electricity they're producing. The reason why the average price earned is different between the different fuel types is because price varies during the day, and the generators produce at different times of the day. Solar generates power in the middle of the day when the price is low (when the electricity is least useful). Gas generators operate during the evening peak when the price is high. So the price is not only a measure of the cost of generating the electricity, it's also a measure of the economic value of the electricity which is produced by that technology.

The lifetime cost of renewable energy projects don't depend on how much electricity they produce. ie, the "wear and tear" factor isn't significant. I'm assuming that a wind farm that sits idle for its 25 year service life costs about the same to build and maintain, compared to one that generates power whenever the wind blows. That allows me to use the 2021 generation costs to estimate future costs with different levels of excess generation. Suppose that the capacity factor for wind in  2021 was about 20%, and that had resulted in the reasonably profitable generation cost of $64 per MWh. If we build so much wind power that it has an average capacity factor of 10%, it would need an average selling price of $128 per MWh produced to stay profitable. At 5%, it would be $256, and so on. The more we over build renewable generation infrastructure, the more expensive the renewable electricity becomes.

Storage needs are fulfilled by lithium batteries. I know this is not realistic, but "batteries!" is one of the stock responses you get whenever the variability of renewable energy is mentioned so I thought it would be interesting as a case study. Also, batteries are the most popular storage technology that is going into the grid right now and cost data is becoming more reliable.

Mongird et al. (2019) [2] put the cost of lithium battery grid energy storage at US$469/kWh, falling by 25% by 2025. We'll just use their 2025 estimates, converted to Australian dollars at 1.0AUD = 0.70USD. This gives us AUD$517 per kWh of installed capacity including power conversion system, buildings and engineering, with a battery life of 10 years [2., table ES.1].

Costing up some systems

In previous posts in this series, we just scaled "renewables", keeping the same ratio of solar to wind that's currently installed in NSW, in order to work out how much storage is needed to just barely prevent a blackout.

In 2021, utility scale solar was about 40% of (solar plus wind) generation, so let's call the cost of generation 0.4 x $45 /MWh + 0.6 x $64 /MWh = $56.4 /MWh. We'll assume that this is constant up until the point we start spilling renewable generation, because having storage allows us to soak up generation whenever the resource is available, more or less giving us the same cost structure as now. We can scale up  the price by the spill fraction we calculated in previous posts. For example, if we spill 50% of potential renewable generation, the price per MWh will have to be twice as high because the remaining 50% of actual generation has to cover all of the costs and generate that same acceptable profit.

The quoted cost of battery storage needs to be expended every 10 years because of the limited lifetime of lithium cells. Dividing the total storage cost by the total electricity generated over 10 years gives us a storage cost spread over each MWh used in the state. Both figures presented in Figure 1 below.

Figure 1 - estimated costs for storage and generation, corresponding to the Figure 4 of Part 3.

A couple of things to notice here. I've had to use a log axis to be able to sensibly show both costs on the same graph. The optimum point is not, as I had supposed, somewhere around the 30-40% spill point. It's far to the right of that because even at 60% spill (corresponding to more than twice the generation capacity that we nominally need), the storage costs per MWh used are still nearly 10 times as high as the generation costs.

What does this mean?

Taking the best case scenario I've got so far, at the right hand side of Figure 1, this means:

• Building 20 times the solar and wind generation that we currently have installed in NSW. The wasted generation opportunity (because of the times when the energy has nowhere to go) means that the renewables cost $160 per MWh (similar to current gas fired power costs, more expensive than coal)
• Building and replacing every 10 years, about 162,000 MWh of batteries at an ongoing cost of about $8.5 billion per year (just in NSW). For context that's about 1.5% of NSW GDP, ballpark the same as we spend on the military. Just for batteries.
• Retail energy costs ($160 /MWh generation, $1363 storage, $200 network + other) = $1723 /MWh or about 172 c /kW. Expect your electricity bill to rise by a factor of at least six

Bear in mind the limitations I outlined with the assumptions made, plus we're ignoring anything like limits on resources or manufacturing (or international supply chains or inflation, for that matter). You can see it's not as simple as just installing batteries. Using some gas for peaking power generation on a few occasions through the year would greatly reduce the costs involved.

For a future post, perhaps.

References cited

1. AEMC, Residential Electricity Price Trends 2021, Final report, 25 November 2021

2. Mongird, K., Viswanathan, V., Balducci, P., Alam, J., Fotedar, V., and Hadjerioua, B., Energy Storage Technology and Cost Characterization Report, PNNL28866, US Department of Energy, July 2019