Understanding Lower Project Performance in 2016-17


  • Overall our project portfolio performed 5.4% below target in the year April 2016 to March 2017, with the Portworthy project performing 10% below target.
  • Whilst there have been some minor operational performance issues, that are now largely resolved, the most significant reason for the lower performance has been lower than average sunshine levels.
  • Analysis suggests that sunshine levels during the last two years have been significantly below average at all our sites.
  • Evidence suggests that longer term performance should return to predicted levels.

Performance against targets

Our total project portfolio performed 5.4% below target during the financial year to end March 2017 compared to a just below target the previous year (Fig. 1). Fig. 2 breaks down performance for last year by location.









As you would expect the BWCE Board are concerned about the causes of the underperformance and any long term implications, given the impact underperformance has on BWCE’s income.

Operational performance

We have had some minor inverter and shading issues during the year on our smaller roof projects which have been, or are in the process of being, resolved. These issues have not had a material impact on the overall performance of the portfolio.

The performance of the actual panels at all three larger sites, Wilmington, Crewkerne and Portworthy has been good. The performance ratio, or the efficiency with which the panels are converting solar irradiation into electricity, has been above target for all three projects. So the panels are working well.

A reduction in the efficiency of the solar panels themselves of 0.5% per year has been taken into account within the target figures.

We have had grid outages at both Crewkerne and Portworthy. This is where the network operator shuts down the solar system to allow them to do work on the local grid. At Crewkerne this was for around 2.5 days during the winter months and so didn’t have a material impact. At Portworthy the impact was a bit greater and amounted to nearly 6 days during spring, summer and autumn months. Since December 2015 this has resulted in just under 2% loss in output.

We would expect there to be continuing grid outages but we do not think they will be a major risk to future output.

Solar Irradiance

All project assessments, including the modelling of future generation and income, are based on assessments of solar irradiance which is calculated using PVGIS, a European Commission database. At our larger sites solar irradiance is remotely monitored on a constant basis.

Where data is available, there is a close correlation between lower than average solar irradiance and lower than forecast generation. This demonstrates that irradiation is the dominant factor, rather than technical or operational factors.

However a key question is, do we expect these low levels of sunshine to continue?

Whilst there are no widely available long-term records of solar irradiance for our specific sites, we have been reviewing our performance against publicly available data on sunshine levels, measured in sun hours[1], to get a sense of how abnormal our experience might be.

The data in Fig. 3 below, compares variation in sunshine levels against the 8 year average in Bath and at the Portworthy and Crewkerne sites. This is modelled data, that comes from www.worldweatheronline.com, allowing the comparison of recent sunshine levels with longer term averages to be made.


Because the publicly available data is modelled rather than actual observed data we also wanted to check how accurate the modelled data might be compared to the real world. So the graph also provides data recorded at the Meterological Office Station at Yeovilton, which is in the same area as the Crewkerne site.

The actual observed data at Yeovilton shows a strong correlation with the modelled data we are using at Crewkerne. The strong correlation between the observed and modelled data here increases our confidence in the modelled data for our other sites as well.

The long-term data from Yeovilton, illustrated in Fig. 4 below, emphasises the unusual nature of the last couple of years. The data suggests that the last time sunshine levels were as low as this was in the early ‘90s. The long term average from actual observations at Yeovilton, is within a couple of percent of the 8 year average shown in Fig. 2.

The overall trends in the modelled data summarised in Fig. 3 shows a good correlation with our project generation performance over a longer period outlined in Fig. 1.

Overall, this means that the last two years have delivered significantly below average levels of sunshine at all three locations, suggesting that in the long term we should see better performance.


[1] Sunshine hours are defined by the met office as the amount of time that direct solar radiation exceeds 120 w/m2. This is not as accurate as measuring actual solar irradiance, which is what we do at our larger sites. Sunshine hours do not take into account indirect solar radiation, of which we get a lot. However, it has not been possible to source historical irradiance data for comparison.


Comments are closed.