Electricity supply part 3 - balancing storage capacity with generation capacity

This is part 3 of a short series of posts thinking about energy storage needs for a renewable electricity grid in NSW Australia.

In part 1, we saw that electricity demand and renewable electricity generation both fluctuate over time. To generate enough renewable power on average for the first week of April 2022, we need about 4.7 times the currently installed renewable generation. There's a big problem though, because the electricity grid must be balanced at all times, not just on average. At 5:30am on 5th April 2022 the sun hadn't yet risen and there was almost no wind blowing in the state. We would have needed more than 120 times the currently installed wind farms to avoid a blackout!

Clearly, there is a need for energy storage to avoid over-building renewable generation equipment to the point where most of it is doing nothing, most of the time. We had a look at this in part 2, where we worked out the minimum amount of storage required for that first week in April - assuming that we had just exactly enough renewables to meet the average demand. We needed a minimium 96,000 MWh, or about 14 hours at average state power consumption. 

There's still an obvious problem here though - we don't necessarily need so much storage if we're happy to waste  (not generate) some of the renewable electricity. So far we've only considered two extreme cases - no storage and massive over generation (in part 1) and maximal storage with no over generation (in part 2). Both of these extremes will be very expensive. Likely a cheaper solution will be to have some level of over generation - which would require less storage to avoid a blackout. The right balance would be the combination with the lowest overall cost.

Working out storage needs with some over-generation

We're going to need a slightly different approach than what we used previously. As in part 2, we'll start the storage number at zero, at the start of the week. When we withdraw energy from storage it goes down, and when we store electricity it goes up. However, when the storage reaches the full state, any additional generation has nowhere to go and gets wasted ("spilled" - which in  practice means not generated in the first place, although it could have been if there was somewhere for it to go).

Then, we once again adjust the generation high enough such that the energy storage ends the week at about the same place it started - "zero". We end up with a pair of numbers - a required energy storage value in MWh, and an energy generation value - some of which is used and some of which is spilled. 

Let's further suppose that "zero" on the energy storage graph corresponds to half full (this isn't quite right but probably minimises the error in guessing how much energy we might have remaining on any given midnight in April).

Figure 1 below shows what happens with the storage level. Storage fills from sunrise on day 1, reaching the maximum level shown as 30,000 MWh on the chart. The storage minimum of -45,912 MWh occurs in that early morning of the 5th of April.

Figure 1 - Electricity storage level for NSW in the first week of April 2021, with 7 times currently installed renewable power generation (solar and wind). Renewable generation is curtailed when the storage is full, which happens in the middle of each day. Total storage required to avoid a blackout is the  maximum storage level minus the minimum (30,000MWh - -45,912MWh = 75,912MWh)

This has resulted in 31% of renewable generation being curtailed, mainly solar power in the middle of the day that has nowhere to go because the storage is full and generation exceeds demand. 

As we generate more renewable energy, we expect a smaller required storage. Although we see that here, the effect is not very strong because of that night of 4th - 5th of April when there was not much wind generation. The size of the storage, just looking at this week, is that required to get the state through that particular night.

Widen the lens to a whole year of data

One week looks nice on a graph. Although we've captured an interesting event on the 5th of April, we don't know much about those other types of variability. Let's consider how this looks for the whole year of 2021, applying exactly the same ideas but probably without the graphs now because they'll look like cat fur.

I've had trouble locating the data for rooftop PV, so I'll leave that out for now (that actually helps a bit because it  means we have  proportionally less solar and more wind, which helps us get through the nights). I'll repeat the analysis at the end if I can  find the data again.

Figure 2, below, shows how this looks for the whole of 2021, now showing "End of Month" on the abscissa, instead of "End of Day". I've scaled up the generation up enough to just meet demand over the year (7.05 times the currently installed wind and utility solar plant). There's no energy spill, but we need at least 4.88 million MWh of energy storage (equivalent to meeting the state's entire energy demand for a 29 days running with no generation). We'll get to costing that out in a future post.

Figure 2 - 100% renewable power, no excess generation over the year (no spill) - requiring 4.88 million MWh of energy storage. That's 4,880 GWh, or 29 days of storage at state average power consumption.

You can see the problem here. Most of the draw down happens in winter from the end of April to the end of July. Most of the storage capacity is only used once per year, for that big winter dip. Figure 3 shows how the situation looks if we go to 20 times currently installed solar and wind. Now we waste 65% of the renewable energy we could potentially generate, but we only need 162,000 MWh of storage (equating to about 1 day of energy storage, instead of about a month).

Figure 3 - 20x current renewable generation. Only 162,000 MWh of storage are needed, and the storage is charged and discharged more frequently (that's good). However, we waste about 65% of the renewable energy that we could potentially have generated because there isn't anywhere to use it.

Figure 4 shows how the required storage varies as a function of overgeneration. The left hand side corresponds to just barely enough generation (no wastage) but large amounts of storage required, with just over 7 times currently installed renewable generation. The right hand side corresponds to 20 times currently installed renewables, 65% of which is wasted (curtailed, or spilled), but much less storage is needed.

Figure 4 - Storage requirement vs overgeneration (presented as % of generation which is wasted). High levels of overgeneration are wasteful in terms of renewable generation infrastructure, but efficient in terms of storage infrastructure, and vice-versa.

A trade-off is required

Hopefully this makes it clear that a trade-off is required. Efficient use of renewable generation infrastructure requires excessive and inefficient storage. Minimising the amount of required storage requires over-building of renewable infrastructure and wasting of the potential to generate energy, particularly during the summer.

The best trade-off is the one that minimises overall system cost - probably at between 20 and 40% energy wastage,  going from the shape of the curve in Figure 4, but we'll try to put some dollar figures on that next time.