Analysis and Prediction of the spatial variability of PV generation in distribution grids
Electricity distribution grids are to use a number of options in the future in order to balance variable feed-in from renewable energies and energy demand at the local level. This enables optimal network management, which at the same time makes the best possible use of the energy fed in from renewable resources. Measures such as the targeted curtailment of photovoltaics, the anticipatory adjustment of electricity consumers and industry to the amount of electricity available (demand-side management) and the control of future storage options require high-quality information about all components of the overall system. In order to be able to plan feed-ins from spatially distributed photovoltaic systems, the operators need accurate, spatially and temporally high-resolution forecasts of solar radiation.
For this purpose, scientists in the VariDist project are developing an analysis and forecasting tool that uses the images from cloud cameras and irradiation measurements from the Eye2Sky measurement network. The monitoring network (Figure 1) is currently being set up by the DLR Institute for Networked Energy Systems in northwest Germany. It consists of more than 20 cloud cameras, multiple ceilometers and receives data from over 10 measuring stations that record solar irradiation (direct, diffuse, global radiation) (Figure 2) and other meteorological parameters. Eye2Sky covers an area of 110 x 100 kilometers.
Fig. 2: Measurement station in the Eye2Sky network.
Left: cloud camera and sensors for solar radiation (DLR-VE)
Based on these data sources, the developed tool predicts the feed-in of electricity from distributed photovoltaic systems for the imminent future with high temporal and spatial resolution.
It compares and combines observations and derived measurements from densely arranged cloud cameras, especially in the city of Oldenburg (Figure 3), in order to enable the most accurate predictions.
The practical knowledge from VariDist also serves to estimate the possible added value of such short-term forecasts for distribution network applications. The evaluation of the forecast and analysis data of the tool enables fundamental knowledge about the spatiotemporal variability of the solar radiation and the PV generation capacity in different meteorological situations. In addition, the high-resolution cloud observations from the cloud cameras clarify whether satellite data alone could obtain higher-resolution information on spatial variability without ground-based measurements.
Based on this project, the network status could eventually also be optimally determined and forecast at short notice.