In recent posts and comments there have been a number of back-of-the-envelope estimates – including some from yours truly – of how much pumped hydro storage would be needed to bridge some of the low-wind periods that have been registered in the UK. Here I take a closer look at the question of how much wind power storage would be needed at the high-penetration grid scale.
And I find that estimating how much storage is needed is not a trivial exercise. It is in fact a very complicated one, and we get quite different results depending on what it is we want to achieve and how we go about achieving it. Within limits (usually high ones) we can make the storage requirement pretty much what we want it to be. For any given scenario there is in fact no correct answer.
Here I consider the following scenario. There are many others:
By February 20XX the UK is generating enough wind power to supply an average of 25GW to the grid. The goal is to use pumped hydro storage to convert all of this wind power into dispatchable baseload generation, whereupon wind will supply almost 60% of UK electricity consumption for the month.
The pumped hydro system consists of a lower and upper reservoir of the same size. It loses and gains no water, has no charge/discharge limitations and is 100% efficient (the results can be factored to allow for lower efficiencies).
Demand in February 20XX is the same as in February 2013. Wind generation in February 20XX is increased by a factor of twelve relative to 2013 values (25GW divided by the 2.08GW average generation in February 2013) so that it averages 25GW in 20XX, which works out to 100GW of installed wind capacity at a 25% overall load factor. (I chose February 2013 partly because I’ve used it as an example before, partly because it’s a fairly typical winter month and partly because analyzing the data for all of 2013 was too burdensome. Other months will, however, give different results, and the numbers provided here should be considered in that light.)
Economic considerations are ignored.
Figure 1 summarizes the position in February 20XX (data from Gridwatch). Demand is the same as in February 2013 and wind generation is factored up from 2013 as described above. I’ve assumed that load-following gas plants will supply the variable demand not covered by wind. The requirement is to convert the erratic blue wind generation curve into the flat line at 25GW:
Figure 1: UK wind generation and demand, February 20XX
Let’s look at how we might do this.
The 2,750GWh Option:
We will begin with the simplest case – an empty upper storage reservoir which fills when wind generation exceeds 25GW and empties when it falls below 25GW. If we’re lucky this option will supply enough power to maintain wind generation at the 25GW level all through the month. Figure 2 shows what happens:
Figure 2: Upper reservoir storage and output to grid, 2,750GWh option, February 20XX
Because of the surplus of wind power during the first week the upper reservoir fills rapidly and by February 6th/7th it has accumulated 2,750GWh of storage. After that, however, it’s all downhill until the reservoir runs dry on February 27th, whereupon the system is no longer able to deliver 25MW to the grid and it’s lights out.
So let’s try starting with the upper reservoir full instead of empty:
Figure 3: Upper reservoir storage and output to grid, 2,750GWh option, reservoir starts full, February 20XX
It makes no difference. All that happens is that the surplus wind generation in the early part of the month gets wasted, or “curtailed”, because the upper reservoir is already full and it has nowhere to go, and after that there simply isn’t enough wind generation to keep the reservoir topped up.
The 3,100GWh Option:
What next? We could try increasing reservoir capacity, but there’s only enough wind power to put 2,750GWh of storage into the upper reservoir to begin with, so increasing capacity above this level won’t achieve anything. We could, however, increase the size of the upper reservoir and start with it full instead of empty, and when we do this we find that the system is capable (barely) of supplying 25GW all through the month when the storage capacity is increased to 3,100GWh:
Figure 4: Upper reservoir storage and output to grid, 3,100GWh option, reservoir starts full, February 20XX
But we still finish the month with the upper reservoir only 10% full. Can we guarantee that there will be enough wind in March to fill it back up again? No, we can’t. So to be on the safe side we have to increase the capacity of the upper reservoir further. Let’s make it 5,000GWh, a good, round and hopefully safe number:
The 5,000 GWh Option:
Figure 5: Upper reservoir storage and output to grid, 5,000GWh option, reservoir starts full, February 20XX
Now we have 2,000GWh left in storage at the end of the month. But we’ve lost 3,000GWh during the month, and if this rate of decline continues the upper reservoir will dry up around the middle of March. To be on the safe side it seems that we might need even more than 5,000GWh …..
Or maybe we don’t need anything like as much. There’s another way of doing it.
The 1,200 GWh Option:
Instead of beefing up storage we increase installed wind capacity. We can in fact eliminate the dip in output shown in Figures 2 and 3 by increasing it by only 3%, but this still leaves us with a near-empty upper reservoir at the end of the month. Clearly we won’t have solved the storage problem until we have a combination of installed capacity and storage that keeps the lights on and maintains the upper reservoir at a safe and reasonably stable level. How much capacity do we have to add to get it?
It turns out that we need to increase installed wind capacity by 50%, from 100GW to 150GW, before we achieve the desired result (Figure 6. Note that I start with the upper reservoir empty to avoid having to curtail the surplus wind energy during the first week). We’ve lowered the storage requirement back down to 1,200GWh and the month ends with the upper reservoir full, but we’re still sailing close to the wind. On February 19th the upper reservoir almost dries up. Obviously we need a reserve margin:
Figure 6: Upper reservoir storage and output to grid, 1,200GWh option (150GW installed wind capacity), February 20XX
But how large should the reserve margin be? Enough to cover a day without wind (24 hours times 25GW = 600GWh)? Or should it be two days? Or three days? A week?
The 700 GWh Option:
If we double installed wind capacity from 100 to 200GW we can cut the storage requirement even farther to 700GWh, although still with no allowance for a reserve margin. Doubling capacity leads to a high level of wind power curtailment, as shown in Figure 7:
Figure 7: Upper reservoir storage and output to grid, 700GWh option (200GW installed wind capacity), February 20XX
Time to sum up. We have identified five different options for storing wind power that give pumped hydro storage requirements of anywhere between 700GWh and 5,000GWh. Which is the best?
There isn’t one. In all probability none of them is even feasible, if only because it’s highly unlikely that the UK will have access to this much pumped hydro storage at any time in the foreseeable future, if ever. And even if it did all of the options would be to a greater or lesser extent hostage to the UK weather, which as residents will tell you is not to be trusted.