Electricity supply part 2 - how much storage for 100% renewables?

 Let's "spill" that excess renewable generation into storage

This is part 2 of a short series of posts thinking about energy storage needs for a renewable electricity grid. We'll develop the ideas first, then explore what happens when we change various things, making the results more realistic at the end.

In Part 1, we saw that, for the first week of April 2022 in NSW Australia, once we expand renewable generation much more than about 5 times current generation, there are times we would be generating more electricity than is required by the state. In order to not cause a grid failure, we have to switch off some renewable generation at those times.

There was another time, just before dawn with no solar yet available, when there was very little wind  power being generated across the entire state.

It seems clear that we need to store that energy when it's available in excess, in order to use at times when the sun isn't shining and the wind isn't blowing. 

I'm going to redraw the graphs we had previously, to show what's going on more clearly as we start thinking about storage. Figure 1 shows just two data series, not stacked. The orange is the demand level, and the green is the total renewable generation. This is the same data as shown in Figure 1 of part 1, except we are just showing the renewables total, not the breakdown by type.

Figure 1 - total electricity demand and total renewable (total of solar and wind) generation in NSW Australia for the first week of April, 2022. The difference between the two curves is currently met by fossil fuel generation, but would need to be withdrawn from energy storage in a fully renewable scenario.

Suppose for simplicity that we just keep the current proportions of solar and wind the same, but multiply the renewable output by 4. We can see in Figure 2 below that supply exceeds demand in the middle of the day, and falls below at night. Whenever the green generation curve is higher than the orange demand one we are charging our energy storage system (instead of spilling the electricity), and when the green curve is below the orange curve, we need to discharge from storage to keep the  lights on.

Figure 2 - First week of April, imagining that we've scaled up renewable generation by a factor of four. When the green generation curve is above the orange demand curve we send excess energy to storage, when it is below the renewable generation is not enough to satisfy demand and needs to be withdrawn from storage.

How much storage is needed?

First, lets work out the power flows needed to and from storage, by subtracting the demand curve from the supply curve. If this is positive, it's a measure of the excess power from renewable generation. If it's negative, power is being withdrawn from storage to support the grid. I've shown this in Figure 3, with the area shaded to make it obvious when it's above or below zero.

Figure 3 - storage power required to balance renewables at 4 times the current generation rate in NSW, for the first week of April, 2022

Remember that energy is power multiplied by time is energy. So although the height on the graph represents power (in MW), the shaded area represents energy in MWh. It looks to me like the total energy withdrawn from the storage (area of the blue shading below the zero mark) might be a bit larger than the energy stored (area of blue shading above zero), but we can't be sure just eyeballing the graph.

What we need to do is add all those areas together - to integrate that storage curve to show how much energy we've accumulated since the beginning of the week. Setting the beginning of the week at zero arbitrarily, we can see the energy storage deficit in MWh below in Figure 4. You can see that the energy stored keeps dropping because renewable generation is not keeping up with demand. 

Figure 4 - Cumulative energy storage in NSW for the first week of April 2022, with four-fold increase of renewable energy supposed. Four-fold is clearly not enough to keep up with demand over this week, as the storage level continues to deplete through the week. Compare to Figure 3 - during the day when renewable generation exceeds  energy demand, the storage level increases, during the night it decreases.

This is easily fixed by building more renewables, If we assume a five-fold increase in renewable generation instead of four-fold then the energy stored accumulates over time. If we go with a 4.7 fold increase in renewable generation then we get the result shown in Figure 5, where we end the week with about the same level of stored energy as we started with ("zero"). This means that we have just enough generation such that the average generation over the week matches the average demand over the week, with sufficient storage allowing us to soak up the variability in both.

Figure 5 - Cumulative energy storage in NSW for the first week of April 2022, with 4.7-fold increase of renewable energy supposed. The storage ends the week at the same storage level at the beginning, indicating that, averaged over the week, 100% of demand was met by renewables. The difference between the highest and lowest stored energy states tells us the absolute minimum size for the energy storage - 96,000 MWh, equivalent to about 14 hours of power stored, at state average power consumption.

Where are we at?

So we're getting a picture of what's needed for a fully renewable grid:

• About 4.7 times the renewable generation infrastructure that we currently have installed, so that we generate enough power over the week (Figure 5)
• At least 96,000 MWh of energy storage - able to power the whole state, by itself, for 14 hours (difference between  highest and lowest values on Figure 5)
• Energy storage able to handle discharge rates of 9,000 MW (highest point on Figure 3)

But there are a few problems with what we've done here:

• This analysis is only for the first week of April 2022. I don't expect that week to be special or unusual in any way, but special and unusual times do occur regularly and the grid needs to be able to deal with them. It also needs to deal with seasonal variations that don't show up in a single week.
• We've just scaled the renewable generation all together, we  might be able to do much better if we build proportionally more wind and less solar, for example
• We've assumed that the storage should be able to absorb all of the renewable energy we throw at it, and that the renewable generation should be enough to meet 100% of demand. This might be an expensive solution. It might be cheaper, for example, to over-build the generation by 20% (which means we waste 20% of what's generated) so that we can make do with a smaller amount of storage.

Next time, we'll develop the analysis to answer these questions and get some more nuanced results.

Electricity supply, part 1 - the need for storage

Keeping the lights on isn't easy

Electricity can't really be stored, as such.

It can also be quite dangerous, and so our electricity delivery systems have been designed with a lot of safety features to prevent electrocution, fires and equipment damage. These things do happen from time to time of course, but mostly, the safety features activate to shut down the affected circuits before any damage is done. This is critical to recovery, as preventing equipment damage means that faults can be corrected and service restored in  minutes or hours rather than the days, weeks or months it would take to replace damaged equipment.

To keep the system safe and the lights on, the delivery sytstem - "the grid" - needs to stay within quite tight tolerances for frequency and voltage (50Hz and 240V in Australia). If things get too far out of whack, or stay out of whack for too long then the safety systems start to activate and shut down parts of the grid to prevent damage. Managing the grid and keeping things from failing is actually a fantastically complex task. A great illustration of the sorts of things that can go wrong are the story of the 2003 blackout in the northeastern US, and the Texas power grid failure of 2021.

This means that supply and demand must be exactly balanced at all times. In the National Electricty Market (NEM) for Australia, which covers the eastern states and South Australia, the market operator (AEMO) operates a planning and bidding process every 5 minutes to ensure this balance is maintained on a minute by minute basis. 

This delicate balance between supply and demand has historically been achieved by letting people use electricity when they want it, and then the market operator AEMO organises the dispatch of supply to meet whatever that demand happens to be at that point in time. This means that total demand varies according to the time of day with our wake / sleep cycle, and also with the weather as we turn heaters and air conditioners on or off, use more or less lighting, watch TV in the evening or go for a walk in the park.

Figure 1, below, shows the total NSW demand for the first week of April 2022. All the data used in this series of posts are freely available to the public via the Australian Energy Market Operator (AEMO) although they are not particularly accessible or in a user friendly format. The NEMOSIS software is free and open source window onto that data that improves accessibility considerably, but I used the commercial package NEMsight, which makes things even easier. The OpenNEM project is the easiest of all, but unfortunately has only the most recent week of data available.

Figure 1 Electricity demand in NSW from 00:01 Friday 1st April to 23:59 Thursday 7th April. Total demand is highest during the day, and lowest in the wee small hours of the morning. Average for the state is about 7,000 MW (megawatts), which is about the same as having around 3 million electric kettles, or 3 thousand electric trains running simultaneously.

The demand shown here includes that which is met through rooftop solar panels "behind the meter" so is closer to the the electricity being used, than to the demand which is "seen" by the market operator (anything that happens behind the electricity meter is essentially invisible to AEMO).

Renewable energy sources don't make it any easier

We have some needs to move away from fossil fuel use for electricity production. Yes climate change, but also fossil fuel depletion. We aren't making any new coal, ever. The rate that we're making new oil and natural gas is so slow as to be zero for all intents and purposes. So for both of these reasons and more, we would really like to stop relying on fossil fuels for electricity generation in the grid.

This causes a real problem because as noted above, supply and demand is balanced by letting people use what they like, and then dispatching the supply to meet that. Fossil fuels are dispatchable, in the sense that you can  (more or less) turn them on or off, up or down, in order to meet the electricity demand and keep the whole grid stable.

Renewable energy generation isn't like that. It relies on energy flows in the environment, like the wind and sun. You can't dispatch those sources to the grid if the natural flow isn't available at that point in time, and so there is no guarantee that they will be available to help generate the electricity demanded. The only thing you can do with renewable sources is to "spill" them - turn them down or off when they are not needed, forgoing electricity generation in order to maintain the required balance.

Figure 2 shows how this renewable generation worked in the first week of April. The graph shows electricity supply from different sources. You can see that solar generators, both domestic rooftop units and utility scale commercial plants, only generate during the day and even then that depends on the season and the weather. Wind power can make electricity at night, but is not consistent over the week. Note that on the 5th April there was almost no wind power generated anywhere within the state of NSW.

Figure 2 - Electricity generation over the same period as Figure 1. "Other" generation is mostly fossil fuels, used to fill in the gaps between the total demand, and the generation from these renewable sources.

Let's go to 100% renewable generation

Figure 1 shows us the instantaneous power requirement for NSW in the period 1 - 7th April 2022. Power multiplied by time (the total area of blue shading in the graph) represents the energy requirement for that week - 1.23 million megawatt-hours, with a wholesale value of about $270 million. Of that energy, about 21% was generated from either solar or wind, the remaining 79% mostly from fossil fuels  (74%) and hydro (5%).

We can do some experiments with the data to test what might happen as we build more renewable generation. Perhaps we can even get to the goal of 100%. For example, suppose that we go on a massive program of wind farm building, so that we have 5 times as much wind farm capacity installed as we currently have in NSW. What would that first week of April have looked like? See Figure 3 below.

Figure 3 - Electricity generation over the same period with a five-fold increase in wind power capacity. The vertical red line is 5:30 AM on 5th April. The sun is still below the horizon and almost no wind power is being produced across the state. To meet demand at this time using wind, we would have to expand the NSW wind generation capacity by more than 120 times what's already installed!

Now we are up to 57% renewable generation but if you look closely, you can see a couple of occasions on the third day, and on the 6th and 7th evenings,  when we had to spill some renewable generation because the total exceeded the demand. This electricity has to be wasted because there is nowhere it can go. If it stays on the grid the voltage will start to rise and then the safety systems will start blacking out parts of the network. So we have to turn off some wind turbines before that happens.

In the scenario shown in figure 3, we have to spill about 3000 MWh of electricity, worth perhaps half a million dollars.

We get a similar situation with the solar - expanding commercial solar farms by a factor of 5 also gets us close to spilling electricity (shown in Figure 4).

Figure 4 - Electricity generation with a five-fold increase in commercial solar farms. No spill, but only 39% of energy generated renewably.

This shows why a renewable electricity grid needs to be able to store energy somehow. There are periods (such as 5:30 AM on the 5th April - see the vertical red line in Figure 3) when none of the renewable resources are available. We will need to be able to draw from hydro, from batteries, or from some other form of storage at those times. 

But exactly how much storage do we need? That's a question we'll explore next time.

P.S. I've made quite a few simplifcations here, in the interests of not obscuring the big picture. We'll address some of those in future posts.