The key item is, “The headline “Offshore wind now cheaper than nuclear power” is very much in the UK news following the latest offshore wind auction in the UK where the lowest bids came in at £57.50 / MWh, well below the Hinkley C strike price of £92.50. But baseload nuclear, which delivers all the time, can’t be compared directly with intermittent wind, which delivers only when the wind blows. To make an apples-to-apples comparison we have to convert wind generation into baseload generation by storing the surpluses for re-use during deficit periods. Doing the math, offshore wind works out to be 6 times more expensive than nuclear power.”
The real strike price of offshore wind
September 20, 2017 by Roger Andrews
Hinkley still scores on reliability and low carbon ….. but the extent to which its costs are obscene is now plainer than ever. In Monday’s capacity auction, two big offshore wind farms came in at £57.50 per megawatt hour and a third at £74.75. These “strike prices” ….. are expressed in 2012 figures, as is Hinkley’s £92.50 so the comparison is fair. As for the argument that we must pay up for reliable baseload supplies, there ought to be limits to how far it can be pushed. A nuclear premium of some level might be justified, but Hinkley lives in a financial world of its own, even before battery technology (possibly) shifts the economics further in favour of renewables …..
Thus spake the Guardian in a recent article entitled Hinkley nuclear power is being priced out by renewables.
What the Guardian says is, of course, nonsense. Comparing non-dispatchable wind directly with dispatchable baseload nuclear is not in the least “fair”. Barring Acts of God baseload nuclear is there all the time; wind is there only when the wind blows. We can level the playing field only by comparing baseload nuclear generation with baseload wind generation, and the only way of converting wind into baseload is to store the surpluses generated when the wind is blowing for re-use when it isn’t. To compare offshore wind strike prices directly with nuclear strike prices we therefore have to factor in the storage costs necessary to convert the wind into baseload, and this post shows what happens to wind strike prices when we do this using the “battery technology” favored by the Guardian. It finds that battery technology does not “(shift) the economics further in favor of renewables”. It prices wind totally out of the market instead.
The two offshore wind farms in question are Hornsea Project 2 (1,386MW) and Moray East (950MW). The project cost for Moray is given as £1.8 billion, or £1,895/kW installed. The project cost of Hornsea Project 1 is given as £3.36 billion, which relative to the 1,218 MW capacity gives £2,759/kW installed. N0 project cost is given for Hornsea Project 2. Moray is 77% owned by EDP Renewables (EDPR) and 23% by Engie. Hornsea is 100% owned by DONG. The locations of Moray and Hornsea are shown below:
To conduct an analysis we have to estimate how much storage will be needed to convert the wind generation from Hornsea and Moray into baseload generation, and to do this we need to know what wind output from these wind farms will be. There are no readily-accessible data for operating UK offshore wind farms, but on the other side of the North Sea are Denmark’s offshore wind farms, and the P-F Bach data base provides hourly generation data for them. So I used Bach’s Denmark data to simulate generation from Hornsea, the larger of the two wind farms, assuming that the results would be reasonably representative. I picked January 2015 as an example month and factored the generation from Danish wind farms in that month up to Hornsea levels relative to installed capacities, which in this case aren’t very different (1,271MW total in Denmark at the beginning of 2015 and 1,386MW at Hornsea). The results are shown in Figure 1:
Figure 1: Hourly generation from Denmark’s wind farms in January 2015 factored up to match Hornsea 2
Strong winds during the first half of the month were largely responsible for the overall 60% capacity factor during the month – respectable for a wind farm. However, the wind blew less strongly in the second half and died away almost to nothing on the 21st and 22nd.
The next step was to convert the spiky wind output into baseload, which requires that surplus generation during windy periods be stored for re-use during deficit periods so that the generation curve comes out flat. Surpluses and deficits were quantified relative to an 825 MW threshold, which is the amount of continuous baseline power Hornsea generates when generation is flat-lined. Figure 2 shows wind generation surpluses and deficits relative to this threshold:
Figure 2: Hourly wind generation surpluses and deficits relative to 825MW of constant baseload output, January 2015
How much storage, which according to the Guardian will be supplied by batteries, will be needed to flatten out these surpluses and deficits? I estimated this in two ways. First I simply accumulated the surpluses and deficits, starting with the batteries discharged, and came up with the battery charge status plot shown in Figure 3. Driven by the generation surpluses in the first half of the month the batteries charge up, reaching a maximum capacity of 95,800 MWh on January 18. Thereafter the deficits set in and the discharges begin, and by the end of the month the batteries are back to being 100% discharged:
Figure 3: Hornsea hourly battery charge status based on accumulation of hourly surpluses and deficits, January 2015
Next I ran the hourly wind generation data through Dave Rutledge’s more sophisticated storage balance algorithm, which starts with the batteries fully charged. The resulting battery charge status plot is shown in Figure 4. 95,800 MWh of battery charge – the same amount as before – is needed at the beginning of the month to keep the batteries charged up until the end of the month, although by the time the end of the month arrives they again have no charge left:
Figure 4: Hornsea hourly battery charge status based on Rutledge storage balance algorithm, January 2015
Beginning with the batteries fully charged, however, creates a complication. During the first half of the month the batteries remain fully-charged for most of the time, and any surplus generation when they are fully-charged has to be curtailed because there’s nowhere to put it. Figure 5 shows the impacts. The curtailment that occurs during the first half of the month amounts to 16% of total monthly generation, and as a result Hornsea delivers an average of only 693MW to the grid instead of the 825MW it would have delivered if the batteries had been discharged rather than charged at the beginning of the month:
Figure 5: Hourly wind generation sent to grid and curtailed based on Rutledge algorithm, Hornsea, January 2015
How to handle this complication? Strictly I should go back and tweak the algorithm until I get an optimum combination of baseload output and battery storage, but in this case it isn’t worth the effort. Why not? Because as we shall shortly see the impacts of the added cost of battery storage on the strike price are so large that even crude approximations are meaningful. So I will run with the 95,800 MWh storage estimate (although it’s almost certainly an underestimate. It assumes 100% charge-discharge efficiency and no battery degradation with time and there is also a high probability that it would increase if time-frames longer than a month were considered.)
Now to economics, and another approximation.
A wind farm gets its fuel for free and maintenance costs are comparatively low; the lion’s share of downstream costs comes from servicing the debt on the initial investment. Here I assume that effectively all of these costs come from debt service, meaning that there will be a direct relationship between the strike price and the initial investment. With this assumption all we have to do to estimate a “batteries included” strike price is add the cost of the batteries to the initial investment and factor the strike price up in proportion. When we do this for Hornsea this is what we get:
Initial wind farm investment = £3.9 billion: I factored the Hornsea Project 1 cost (2,759/kW installed) up in proportion to the increase in installed capacity (1,396 MW for Hornsea 2 vs. 1,218 for Hornsea 1). This gave a total project cost for Hornsea 2 of £3.85 billion, which I rounded up to £3.9 billion.
Cost of battery storage = £35.4 billion: 95,800 MWh of lithium-ion batteries at current prices of around US$500/kWh – £370 at current exchange rates – gives a total cost of 95,800,000 kW * £370/kWh = £35.4 billion.
Cost of wind + battery storage = £3.9 + £35.4 = £39.3 billion
Strike price with batteries included = £579.42/MWh: The strike price increases in proportion to the increase in total investment, i.e. from £57.50/MWh to 39.3/3.9 * £57.50 = £579.42/MWh.
Since as noted earlier the 95,800 MWh storage requirement is almost certainly an underestimate – and quite possibly a large one – we can reasonably conclude that Hornsea’s strike price will be at least six times higher than Hinkley’s £92.50/MWh when the two are compared on an apples-to-apples basis using the Guardian’s battery storage option.
What does this factor-of-six difference tell us? Actually not much, because the comparison is academic. No one is ever going to outlay £35.4 billion to install battery storage at a £3.9 billion wind farm. Backup gas, not battery storage, is presently the only option for smoothing out erratic wind generation, and estimating how much this might add to the Hornsea strike price would be a complex undertaking, although I might give it a shot in a later post.
What it does tell us is that adding even a comparatively small amount of battery storage to a wind (or solar) project could kill it economically, which is probably what motivated the Guardian to make the comment about putting limits on how much “we” have to pay for “reliable baseload supplies”. And in the clean, green, environmentally-conscious, demand-managed, smart-meter-monitored, grid-interconnected, one-hundred-percent renewable world of the future the Guardian envisions we won’t need reliable baseload supplies anyway.