Solar Energy Forecasting Used to Manage a Solar PV Plant with Storage in an Insular Grid

June 7th, 2017

In the Martinique Island (French oversea territory in the Caribbean), the grid stability can be seriously affected by the variability of the solar PV production. The larger-scale integration of this energy, will go with the addition of storage capacities and an adapted management system which integrates forecasts of the solar production over several time scales.

For this 100 KWc PV plant, Steadysun has deployed a comprehensive solution based on weather models and satellite images.

On June 7th, 2017, the forecast provided the day before has under-estimated the production.
It is a day presenting a strong variability of production bound to numerous cloudy spells and a fast evolution of the cloud layer.

Figure 1: actual (in blue) vs SteadyMet forecasts of the day before

During the day, this forecast is updated every quarter thanks to satellite images allowing to refine the production profile for the next 30 minutes. The Figure 2 shows that this impoves strongly the accuracy of the forecasts.

Figure 2: actual (in blue) vs intraday satellite based forecasts (in red) – 60 minutes ahead – minute time step

When averaging this profile at a 15 minutes time step (Figure 3), the accuracy performance appears more clearly as compared with Figure 2 which is at one minute time step.

Figure 3: actual (in blue) vs intraday satellite based forecasts (in red) – 60 minutes ahead, averaged on a 15 minutes time step

With such a display, available through Steadysun customer interface, the storage system can be managed with more efficiently, by defining the day before the optimized level of state of load to start the operation on the next day as well as the injection profile of the next day to be announced to the grid operator.
During the day, the asset manager may appeal the stored energy at best to maximize the income.