Home page Bellringing  Talks and slide lectures  Fell walking  Settle to Carlisle  Metal sculpture  Brickwork  Journeys  Ergonomics  The rest  Site map

Solar power

Total generation Seasonal variation Daily power variation Import & export Ecotricity Pictures

Solar panels on roofWe've always tried to be efficient, to use resources and energy sparingly, and to generate minimum waste. In October 2010, we took the next logical step by putting our south facing roof to good use  with PV (photo-voltaic) panels that generate solar electricity. The panels have a peak rated output of 3.3kW, but the actual output is generally less than that, and varies with time of day, time of year and the weather. Using published monthly averages for sunshine in this part of the country, I estimated that the panels would generate around 2.7MWh per year, which is about 10% more than our annual consumption. Of course that doesn't make us 'self sufficient', because we still need to import energy when we aren't generating, eg to power our lights at night, but we should export more than we import.

Installation was straight forward – we used Ardenham Energy . The panels are mounted on brackets that go under the tiles, and are fixed to the rafters. They feed DC power to an inverter mounted in the loft (see pictures  below) which sends AC power down to the total generation meter in the cupboard with the normal meter (see pictures  below). When the panels are generating, the light on the meter flashes at a rate proportional to the power. We wanted to be able to see the power without going to the cupboard and timing the flashes, so we use an Owl electricity monitor  that sits on the kitchen worktop (see pictures  below). It communicates with a sensor clipped round the power cable. We have another Owl meter to show the power that we are using (see below).  

Results (up to end 2017)

Total generation

Energy generatedThe first graph shows energy generated per week since installation in October 2010 (but for tidiness the graph starts with January as week 1). The grey line is what I expected as an average output through a typical year. I derived it from published monthly averages for sunshine in Berkshire, normalised so that the total energy was the same as the figure derived from the manufacturer's data,

The graph shows a lot of fluctuation from week to week, which just reflects the variability of our weather. On a day when the panels would be generating well over 2kW in the sun, they only generate a few hundred W when it is cloudy. The pattern varies from year to year, for example in 2011 we had a lot of sun in April/May, whereas in 2012 we had the highest weekly output at the end of March, and then quite a low output in April/May. 2013, 2014, 2015 and 2016 were different again. 2013 saw the highest amount of energy generated in any week since installation – 132kWh in early June.

Cumulative generationThe second graph shows cumulative energy generation through each complete calender year. Like the first graph, it starts in January and it uses the same colours for each year. The end of 2010 wasn't a complete year, and so is not shown.

In every year energy generation has been at least 10% more than my predicted figure of 2.7MWh – for the seven years: 3.17MWh, 3.14MWh, 3.00MWh, 3.24MWh, 3.17MWh, 3.15MWh, 3.00MWh,). The early part of 2013 started significantly lower than other years, caught up during the July heat wave and ended the year only slightly lower than the other years, but still above my prediction. 2015 was unusual – cumulative generation by the autumn was higher than all other years but very dull weather in the rest of the year brought the total lower. 2017 was quite high early in the first half of the year but a dull summer meant it finished almost as low as 2013. So far, 2014 has been the best year for total energy generation.

Seasonal variation

 Seasonal variation To see the underlying seasonal pattern, you have to remove the short term fluctuations due to the weather. I did that with rolling 5 week averages (ie the average of a period from 2 weeks before to 2 weeks after the week in question. That removes a lot of the variability, and averaging the values across more than one year removes much of the remaining variability to give the thick black line.

As you would expect, the output in winter is a lot lower than the rest of the year, but there is relatively little change in output from Spring through to early autumn, whereas you might expect a lot more to be generated in midsummer with longer days and the sun higher. There are two reasons this is not so. One is that however long the day, the sun can't shine directly on the panels for more than 12 hours, because during the other 12 hours, even if the sun is above the horizon and shining on the house, it is behind the panels. They still generate from the glow of the blue sky, but at much lower power – like when it is cloudy. The other reason is the angle of our roof, which is about 35° – too steep for maximum midsummer generation and better for the slightly lower spring and autumn sun. In fact 35° is ideal for maximising the overall output during the whole year (see: The Best Angle And Orientation For Solar Panels In The UK

Daily power variation

Power during the day The power varies during the day (even with a clear sky), and the daily pattern varies round the year. In summer, the days are long and the sun rises high, while in the winter the days are short and the sun is low in the sky, with spring and autumn in between. You might expect to see a similar pattern in what is generated., but there is a difference. The curve in winter is indeed lower and narrower, but there is relatively little variation between spring, summer and autumn. The curve in summer is not significantly higher than the curves in spring and autumn because the sun is behind the panels at both ends of the long summer days. The limited difference in height is because of the roof angle is optimal for spring and autumn but not for summer (as explained above).

These graphs are based on readings in 2011 on sunny days in spring (end April), mid summer, autumn (mid September) and winter (mid December). To reduce the effect of fluctuation due to passing clouds I took readings at or near each hour over several days, and then used the highest figure for each hour. NB – All times are shown in BST, so midday is at 13.00 for all curves.

 Import and export

 Before we bought the panels, I predicted that we would export about 85% of what we generate. I had produced a model for generation at different times of day round the year using various published data, and a similar model for consumption based on knowing our approximate pattern of using various devices, and I had then compared the two models.

I can't measure exported energy directly, because I opted not to pay extra for a certified export meter. (We could have been paid a bit more than we do without a meter - where 50% is assumed - but the few pence per kWh extra did not justify the cost of the meter.)

I still wanted to see how my model matched reality, and I can work out how much energy we export by comparison of three separate things: the energy we generate (from the generation meter). the energy we import (from the normal meter) and the energy we consume. To measure consumption, I use an Owl electricity monitor  that has a cumulative energy readout as well as the power display. That introduces a small error, because the Owl doesn't measure power directly but measures current and calculates power by assuming a fixed voltage and power factor of 1. That method is adequate for the Owl's intended educational use, but not quite accurate enough for my purposes. Quite a few of our devices, notably fluorescent lights and switched-more power supplies in electronic equipment, have a power factor less than 1. As a result the display reads a bit higher than true power consumption. I worked out an approximate correction factor by comparing consumption measured with the Owl with that measured by the Consumer Unit during times when there was no generation so the two should have been equal. This correction factor is used in the figures plotted below. (NB - The anomalous dip in GenUse (pale blue) in late summer 2016 was caused by the sensor around the power cable coming partly unclipped so the readings were lower than they should have been.)

Generation, use and import 

Averge import/exportIn the graphs, energy generated is shown above the axis (in kWh per week). The green part is exported, and the pale blue part is consumed by us. The dark blue part below the axis is imported energy. So taken together, the two blue areas represent the total energy that we consumed. I started measuring consumption in April 2011, about 6 months after installation, so the graph runs from then to the end of 2016. During this 69 month period we exported 78% of the energy we generated, and we only needed to import 71% of the energy that we consumed. The graph on the right is averaged over all years, which smooths out much of the week to week variation caused by weather in individual years


We sell the power that we generate through Ecotricity's Microtricity  scheme, which is backed by the Government's Feed-In Tariff. We also buy our gas and electricity from Ecotricity  because they invest all of their profit in building new green energy sources, while charging less than the dominant suppliers. When I last checked, Ecotricity was investing far more per customer in new green production facilities than the large suppliers, including those that offer 'green tariffs'.


Click to enlarge.

Photovoltaic panels on the roof
Inverter in the loft
Total generation meter in the cupboard
Monitors for generation and consumption

  (Page updated 19-January-2018 )  

Back to top Back to Various Other things Return to Home page

Site search: