One south of England wind farm faces default on its bank loans because wind speeds have been as low as ever recorded and electricity output has therefore been much less than expected. So far this year the power delivered by the turbines has been less than two thirds of what is predicted in a typical year, meaning is cash flow is failing to meet even the most pessimistic projections. Statistical analysis suggests that electricity output this low should only be achieved once every several hundred years. Another wind farm in southern England reports similar results, with electricity delivery this year so limited that it would only be expected once a century. 2009’s figures were almost as bad. These results will make it much more difficult to get bank finance for English wind farms. If the rules-of-thumb for wind speeds are turning out to be incorrect, financial institutions will be much more cautious about lending. This note looks at the likely variability of wind farm output and compares it with solar photovoltaic power. Although the returns on PV are likely to be lower, does the recent decline in southern English wind speeds make solar an easier investment to finance?
(At the end of this note I ask UK PV owners to forward me their output records so that I can assemble a central database – needed if we are to persuade banks to lend large sums on solar installations. Erratic output makes financing much more difficult.)
When wind farms are being planned, the developer erects a wind speed meter, usually for one year at the site of the proposed turbines. This was thought to give investors and banks a firm indication of likely electricity output levels. But recent extraordinarily low average wind speeds have made this measurements seem much less reliable.
The consulting engineers who measure wind speeds usually provide developers with figures for average speeds across the year. They translate this into typical power output for the turbines chosen for the site. They also provide a ‘P90’ figure, an estimate of the electricity output that will be exceeded in nine out of ten years. This figure is, of course, lower than the average expected output.
I have looked at the business plans for three wind farms in the UK that were constructed in the last ten years. The P90 output figure is about 87% of the mean expected output. In other words in 90% of all years, the electricity delivered will be at least 87% of the estimated average. (For those of a statistical bent, the expected standard deviation in power output is about 10%).
Two of these wind farms publish details of the actual output to shareholders. In one case, output so far this year has been well below the P90 figure. I have carried out some simple statistical analysis, suggesting that the output levels from the turbines should only have been achieved once every several hundred years. Similar, but less extreme, results are seen at the other wind farm.
These wind levels may just be exceptional – the unprecedented persistence of Artic high pressure over the 2009/10 UK winter certainly reduced typical wind speeds. Some people have suggested that the summer melting of Artic sea ice is affecting wind patterns. It could also be that general meteorological conditions have militated against high winds over the last few years. Or it may that the engineers have simply hugely underestimated the underlying degree of natural annual variability in southern UK winds In other words, the average estimated power outputs for these wind farms are correct over a period of decades, but the chance of the actual number being much higher or much lower in any particular year is significantly greater than has been projected.
In all of these cases the implications for English wind farm developers are potentially severe – the banks will be willing to put up a much smaller fraction of the total cost. Their financial models, which require wind farms to be able to pay back the interest and loan principal even in unusually bad years, will have to be revised.
There are few month-to-month records of the output of solar PV installations in the UK. We know that the expected output has a much tighter distribution than wind power because solar insolation levels are more stable year on year. But just how much more reliable is PV? I did some simple statistical analysis of the monthly output from the installation on my roof to provide some estimates for those looking to get banks to help finance their own purchases of PV panels. This is not great data but will provide a reasonable figure for those trying to guess at just how variable the cash flow from PV will be.
I have six years of monthly figures from our house. (In one month the failure of one of the inverters affected the figures for fourteen days and I have adjusted for this). My analysis shows that the annual P90 figure for a small installation in central southern England is about 3% below the expected average output. To give a simple illustration, if the expected output (which can be obtained from several databases on the web) is 1500 kilowatt hours, then in nine years out of ten the actual output will be greater than 1455 hours. (1) In other words, the natural variability of solar PV output each year is only about one quarter as much as wind power, at least as far as I can measure it in Oxford.
I also looked at the difference between the best 12 month period during the six years during which we have had PV and the worst. The worst ever yearly output (March 2009 to February 2010) was 1388 kilowatt hours and the best was 1552 (May 2007 to April 2008), with the lower figure being 89% of the higher. This variability is dwarfed by the figures from the English wind farms for which I have data.
In order to lend significant sums to PV developers, banks will need good records of the degree of variability of output. They cannot be expected to lend cash to installations if there is a reasonable prospect of cash flows being insufficient to service the loan. If you have any data on the weekly, monthly or yearly output from your PV installation in the UK, please may I have a copy in order to keep a centralised database tracking the variability of electricity output? Thank you.
(1) Anybody wanting the actual six year output figures from my house to carry out their own modelling is welcome to email me. (firstname.lastname@example.org)