I usually write this blog in finnish, but since these thoughts originated from the comments to a recent post in Carbon Commentary this one will be in english. (See especially the comments by J. M. Korhonen and Dominic Hofstetter.) I want to understand some general issues regarding attractiveness of electricity storage and also how storage schemes are likely to differ between intermittent power sources and baseload power sources.
Since storage would pay for itself from the spread of electricity prices at different times, as a first step, I wanted to know how much do the prices actually vary. I checked this for the Swedish market and show the result in the figure. In Sweden the average range for daily variation in price is about 18 euros/MWh. Data implies that the price varies from about 0.75 times the daily mean price (typically) at night to about 1.25 times the mean price during peak demand.
So each stored kWh should make a profit from this spread and cannot therefore cost more than this. So it would seem that for large scale electricity storage to be interesting, the cost should certainly be less than about 10 euros/MWh. Unfortunately storing electricity is costly.
Over the lifetime the prices per kilowatt hour can vary from hundreds of dollars to thousands. If we optimistically assume a cost of 100 euros/kWh, to get the cost for a single cycle to the 10 euros/MWh range implies around 10000 cycles. If there is one cycle per day, this implies a lifetime of around 30 years. For battery systems both the cost assumption as well as the lifetime assumption are quite unrealistic. Pumped hydro is somewhat more realistic (although it has its share of problems…see here), but it is geologically limited resource and implies flooding large areas of land.
What kind of efficiency should we expect from our storage system? Let us say we have an efficiency E so that to get 1 kWh of stored electricity for sale, we need to spend 1/E kilowatt hours during storage and lose “minimum price/E” worth of sales. Selling the stored electricity will gain us (at best) “maximum price” and in order for the process to make sense efficiency must be much better than about “minimum price/maximum price”. With Swedish figures anything that has a roundtrip efficiency less than around 60% makes no sense. Naturally, less efficient the system is less it should cost so that it can still make a profit within that 18 euros/MWh range.
Increasing amount of wind power in the grid will increase price volatility. Why this would be desirable is unclear. Typically increased uncertainty is a thing to avoid since it increases the likelihood of proverbial shit reaching all the way up to the proverbial fan. Here it would seem that we first pay someone subsidies for extra volatility and then someone else to get rid of it. Maybe there is beauty in here, but it does seem well hidden. In any case increased volatility can make some storage schemes more viable, but who would build such storage? Currently wind power producers are guaranteed a market AND the price so they do not have an incentive to move their production to some other time periods using storage. For storage the system should be changed so that wind power producers do not get guaranteed prices and have to find actual customers for their product. However, even then the volatility increases to make (theoretical) large scale storage viable would have to be very dramatic and storage schemes would still have to compete with generators burning stored energy contained in fossil fuels.
How would a storage scheme coupled to say wind power compare with a scheme coupled to a baseload power plant? Even if the size of the storage would be the same, storage for wind power would have to live with much larger swings in power levels. A 1 GWe baseload power plant could store night time production at a rate from zero 0 to 1 GWe. If same amount of electricity is produced with wind power storage would be fed with anything between zero and 3 GWe. This would add to the costs. Similar conclusions apply to the rate at which the storage is discharged. If it has to cope with baseload plant dropping off grid, output of 1 GWe is enough. If it has to compensate for swings in wind power production, much more might be required. (Also, it might not be a good idea to put large mass of water moving upwards and then suddenly stop. Reliable pump power seems more fit for this purpose, but someone with required engineering skills can correct me if I am wrong.)
Since the output of a baseload power plant is also predictable, it seems clear that anyone who has storage to spare will find it easier to make a profit by coupling this storage to a baseload power plant. Finally, it should be noted that with wind power storage for about 12 hours is not enough to create a wholly renewable system. Not even close. With such a short time storage we would still need reliable (fossil fuel based most likely) power plants with capacity that is enough to cover all electricity consumption. On the other hand, 12 hours of storage coupled to a baseload power plants would go some way towards removing the need for peaking power plants.
All of the above is of course purely theoretical. I do not see any reason to believe that the cost for electricity storage would become so low as to make it economically viable in anything but fringe applications. Perhaps more viable route is to use night time production to produce heat and fuels that would reduce emissions outside the electricity sector?
Updated on 19.3.2013: Anders Örbom raised an issue in Twitter as to how representative Sweden is due to price levelling effects of their hydro power. I hand’t thought of that and I used swedish data mainly since that I had readily available. From the link in the text one can also get the spot price in Germany. It turns out that in Germany the daily variation is stronger and amounts to about 39 euros/MWh on average. So there the storage can cost more than in Sweden and still be interesting. Nevertherless, to get the price low enough is still a huge challenge even in Germany.