Two thousand five hundred pounds per MWh was the price that electricity briefly reached in the early evening of Wednesday November 4th 2015 here in the UK. If you prefer another currency, that’s US$ 3,750 per MWh.
Now as one of the news articles reporting it points out, this price refers to one accepted offer to generate by one particular (gas-fired) power plant – it certainly doesn’t mean such extreme prices were common or sustained for any period of time. (In fact it’s not clear whether the offer needed to be taken up). However it is an interesting case study of extreme, peak-demand market price fluctuation, brought about by specific circumstances in balancing supply with demand. Peak power is a subject I’ve written about before, so this incident prompted me to briefly spend a bit of time gathering more data on the issue.
For those of you unfamiliar with the UK market, our peak demands occur in the winter months and in the early-evening hours – typically within the 17:00 to 18:30 timeframe. Hence this is when prices peak too.
To illustrate: here’s a year’s worth of demand data (for 2014), running from January 1st on the left to December 31st on the right. Weekday demand is higher than at weekends (hence the latter show as short white stripes between the weekday peaks on this compressed chart).
For an individual day – let’s choose the day in question, November 4th 2015 – here’s the daily demand curve, plotted from midnight on the left (to next midnight on the right). It shows that evening demand peak. Both these charts use data sourced from the excellent Gridwatch site.
Now let’s get back to prices.
Using data from APX Power Spot Exchange (one of the electricity trading platforms here in the UK) it’s possible to strip out the insanely-high-but-transient price that I shamelessly grabbed your attention with in the headline. Instead we can look at prices over half-hourly periods. (Note that there are other markets and types of time period you’ll see pricing quoted from, so don’t worry if these numbers don’t exactly match others – it’s the trends from any consistent set that are of most interest).
The chart below looks at half-hourly prices for the three days prior to November 4th, the 4th itself, and the following two days:
(By the way, every half-hour period during the day is plotted – the horizontal axis only shows text for alternate ones in order to remain legible).
November 4th (and to a lesser extent the preceding two days) still stands out as a substantial spike price compared to normal circumstances! £360/MWh ($540/MWh) is still very expensive electricity!
Note that as well as high prices, on November 4th the National Grid also triggered its “demand-side balancing” mechanism, whereby energy intensive users were paid to reduce their demand.
I can better illustrate how unusual November 4th was from a price perspective, this time by plotting maximum half-hourly prices each day (along with averages on those same days) throughout the year to date:
It’s also noticeable here how the summer months are “quieter” in terms of price spikes than are the high-demand winter ones. That isn’t surprising, given that summer peaks are lower, so there’s plenty of spare supply available in the system to meet it.
So why did this week’s particular spike happen?
It was nothing to do with any particular spike in demand – indeed the demand peak on the 4th was slightly lower than those on the previous two days. Instead it was a supply-side issue.
In particular: “a number of unplanned outages, including the loss of some major coal-fired capacity, as well as some pumped-storage facilities that would usually be called upon to fill unexpected generation gaps”. How much capacity went missing would be interesting to know (but sadly, as far as I know, not published).
Unhelpfully for renewable generation fans, it was also an unusually calm day and an absence of wind power failed to help matters!
At the time of the price peak over the crucial 17.00 to 17.30 period, total wind generation averaged a feeble 281 MW (from a UK installed capacity of 13,386 according to Renewable UK)! Compare that to the same time periods on a) the 1st November (1,593 MW of wind, price £91/MWh) and b) 5th November (2,500 MW of wind, price £52/MWh). Without knowing detail of the conventional power plants that went missing, it’s impossible to know how much wind generation might have been able to create how much price reduction – but it certainly would have helped.
What else might have helped?
Interconnectors between the UK and other markets certainly would. Those we have with France and the Netherlands were operating (to the tune of about 1.5GW and 1GW imports respectively). The National Grid has project plans to create several GW more.
So too would more storage. Again without knowing quite how much capacity was missing compared to usual, it’s impossible to know how much.
As mentioned above, demand response was triggered in this case. Extending the scope of demand management could become a more important mechanism in future. After all, it doesn’t involve expenditure, space, planning and other barriers to building new infrastructure – so it’s potentially the cheapest and quickest way to avoid such supply/demand crunches.
In the end of course, the lights didn’t go out. Sufficient supply could be made available; it was just expensive for a brief time. It was also an unusually rare set of circumstances, as the size of the spike illustrated.
However here in the UK it did reopen some of the debate around whether such supply/demand imbalances might become more frequent in future – particularly as conventional generators (particularly coal) shut down and increasing quantities of renewable generators such as wind enter the system.
And, if that is the favoured direction of travel, how best to manage them in future.
At the other extreme: negative prices!
A few days after writing about extreme high pricing above, there was another interesting pricing example; this time in Germany. Rather than high pricing, caused by limited availability of supply to meet demand, this one was the opposite.
The chart below, taken from this excellent site by Frauhofer ISE, shows electricity production in Germany, during the week starting November 2nd. The different colours indicate different sources of electricity generation, adding up to a total production which fluctuates between a low of about 50GW and a peak just above 80GW. (Note that in this case production is not the same as demand, at least not in Germany, since some of this power is exported – more on that below).
Similarly to the UK, you can pick out the weekend, the two days at the end of the week when less electricity production is needed. Unlike the UK there is a lot of solar energy (in yellow) arriving into the mix during the day. Wind, in light green, was very light early in the week, but blew strongly towards the end.
So what about prices?
Luckily for us, that same site shows prices too:
This shows exactly the same time period as the last chart but slightly different data – which is why the power (GW) axis on the left has slightly different totals. Now only solar and wind are highlighted and grouped together in grey are large-scale “conventional sources” (so, for example with the exclusion of biomass and smaller-scale capacity). In purple is shown the “import balance” (i.e. how much electricity is being exported into other markets, or imported into Germany).
What we are really interested in though is that now added to this chart are prices (with the scale on the right axis). Shown here are those offered a day ahead (in red), based on forward predictions of demand and supply, and the averages actually measured within the day itself (in blue).
Here are a few observations from the chart
Firstly you can see that each day when the sun is shining, there is a dip in prices. This effect, of solar PV driving down market prices from their usual daylight peaks, is one that’s well established in Germany. It is a source of much annoyance to conventional power plants, who used to make good money when these prices stayed high. It even annoys conventional power plants in neighbouring countries: you can see, certainly for the first three days, that bumps in electricity export coincide with peaks in solar electricity production too.
Secondly, not unlike the UK, you can see three particular blue (and red) spikes in price coinciding with the early evening demand peak. Look particularly at the second of these and you’ll find that this demand peak occurs after the sun has gone down and also at a time when there’s essentially no wind. So, not unlike the spike we looked at in the UK earlier, this demand peak would have been met by calling on rather expensive conventional plants. The chart shows an average, but it would be interesting to know how high individual bids got…
However what should really have grabbed your attention is what happened at the weekend, either side of midnight between the 7th & 8th of November. Both predicted and actual prices dipped not just to zero but into negative territory. It’s basically saying there was so little demand for electricity, you couldn’t even give it away! More accurately, there was far too much generation supply available at a time when people simply didn’t need it, even in neighbouring countries (and you can see from the chart that exports were being made).
It’s the opposite end of the extreme pricing spectrum to what we talked about earlier.
The culprit here is obvious, appearing as a big green stripe. Yes, just when everyone was turning off the lights and going to bed, the wind was blowing merrily!
In fact the solutions to this problem are little different to those we highlighted earlier to mitigate against prices spiking upwards:
- Demand response, in this case not turning down demand but creating more of it. It happens already that big energy users can be paid to turn up their usage. Smart homes could take advantage of cheap power to heat their water tanks or turn on their washing machines.
- More interconnection would make a difference. Even though you would imagine neighbouring countries were also turning off the lights and sleeping, they may still find it economic to take Germany’s excess (and cheap) power. It would depend what they were using instead. Spread the electricity market far enough and you can reach places were demand isn’t dipping at the same time, either due to time zone or cultural differences (i.e. staying up and partying until later!)
- Putting excess wind energy into storage when there’s too much and selling it at peak times instead. You can get a sense of Germany’s pumped storage capacity from the first chart (the light blue), where it is put to use at peak times – but you can also see that there simply isn’t enough of it to swallow up energy on the scale of that green stripe during the extreme negative pricing event. In future one obvious way to store excess electricity at midnight would be to put it into the batteries of a large fleet of electric vehicles. Note that storage is just another demand response mechanism – it creates a demand that wasn’t previously there (by turning on a car charger, for example).
So there we are. In the space of one November week, two examples of extreme pricing – one high and one low. There were different causes but there are surprisingly similar ways to evolve electricity systems to avoid them in future.