Solution & Technology
How a Solar Energy Forecasting solution can Improve Powergrid Management in Spain
December 31st, 2015
The Spanish electric system includes 4.4 GWc of photovoltaic plants in its continental network.
The figure 1 shows the solar energy forecasts provided in the morning of December 29th, 2015 for the last day of year 2015. The uncertainty is relatively high, the gap between P10 % and P90 % is almost 800 MW at noon.
Figure 1: Actual production (in blue) vs Forecasted (in red/orange for the percentiles) – 2 days ahead, Time step: 15 minutes
The following day forecasts are more accurate (see figure 2), and also closer to the energy level which will be finally produced, with a lower level of uncertainty. As a matter of fact, the gap between P10% and P90% is reduced to 500 MW, and more generally all the curves have narrowed around P50%.
On a grid management viewpoint, the operational consequence is a reduced need of reserves to tackle the variability of the photovoltaic production.
Figure 2: Actual production (in blue) vs Forecasted (in red/orange pour les quantiles) – 1 day ahead, Time step: 15 minutes
This is a usual result: the closer the time horizon, the more reliable the forecast, and consequently the uncertainty is reduced.
Nevertheless, it is not always the case. The solar forecasts produced in the morning of the D day are showing similar confidence intervals and accuracy level to those of the day before (see figure 3).
Figure 3: Actual production (in blue) vs Forecasted (in red/orange pour les quantiles) – Morning D day, Time step: 15 minutes