More and more people have solar panels. That poses problems for energy network managers because the networks are not designed to cope with big peaks and dips in the electricity supply.
In a project called Every ray counts, energy supplier Alliander and WUR are now working to improve sunshine forecasting. At present, solar energy forecasting is mainly based on what is being generating at the moment, and what historical data might lead you to expect for the rest of the day, says project leader Chiel van Heerwaarden of the Meteorology and Air Quality chair group. This can be improved on by using meteorological data to anticipate the peaks and dips to come. Clouds and cloud formation are essential to this. That may seem obvious but it is not entirely so. A cloud that hides the sun does of course cause a dip in the production of solar energy. But when the same cloud is not in front of the sun, it can cause over-production, explains Van Heerwaarden. ‘That happens because clouds reflect some of the sunshine that falls on them onto the solar panels. This makes the production level higher than it is when the sky is cloudless.’ This effect has not yet been integrated into the models for forecasting solar energy production.
MSc student Esther Peerlings is figuring out the relation between solar radiation and clouds. To this end, she is analysing the fluctuations in solar radiation ‘on a minute scale’ at WUR’s Veenkampen weather station near Wageningen. Linking this data to information from Alliander should provide a better understanding of the effects of particular weather types on the grid. But there is more to it than that. The sunshine forecasts themselves could be a lot more precise. ‘We are working with the KNMI (Royal Netherlands Meteorological Institute) to look at how we can get better solar energy forecasts using the data from Harmonie, the weather model used in the Netherlands,’ explains Van Heerwaarden. In the long term there are even models in the pipeline that can predict individual clouds. ‘But that will not be for another 10 years. What we are going to do now is simulate clouds and see whether that produces better forecasts of solar energy.’
The project with Alliander will last four years and is funded by the Netherlands Organization for Scientific Research NWO. Alliander is contributing funding, as well as PhD candidate Frank Kreuwel, who comes from its ranks.