All-sky imager based ramp rate prediction for PV

Wed, 21/06/2017 - 12:10

The solar resource shows variabilities, which influence the frequency, voltage and overall stability of electrical grids with high photovoltaic penetrations. Facing this challenge, legal limitations of the fluctuations of generated electricity (ramp rates regulations) are already effective or under discussion. As expected, small island grids with high solar penetration were the first to introduce ramp rate regulations: Puerto Rico has specified a maximum negative 1-minute ramp rate of 10 % of the rated capacity and Hawaii has defined a maximum negative ramp rate of 1 MW/min during certain times.

Figure 1: Photo of an all-sky imager (ASI). Transient clouds are the main reason for sudden drops in the dispatched electricity of solar power plants. Cameras can be used to predict these ramp rates.

For short-term periods between zero and 15 minutes, the origins of these ramp rates for PV plants are predominantly transient clouds. Forecasting such cloud induced variability can be achieved by all-sky imager (ASI) based systems (Figure 1). Besides legal obligations, the forecasting of ramp rates helps to optimize battery management for combined PV-battery systems. Thus optimized operations might increase the lifespan of electrical storages and auxiliary devices. The low spatial and temporal resolutions of satellite derived forecasts make it difficult to predict ramp rates using them.

Figure 2: Working principle of the WobaS system. All-sky imagers are used to derive future irradiances in high spatial and temporal resolutions.

Founded by the German Federal Ministry for Economic Affairs and Energy and in cooperation with CSP Services and TSK Flagsol, the Institute of Solar Research within the German Aerospace Center (DLR) has developed the four-camera nowcasting system ‘WobaS’. Three WobaS systems are currently operational: One is installed at the Plataforma Solar de Almería, Spain, another is operational at the commercial 50 MW solar power plant La Africana (near Cordoba, Spain) and a third is set up at a solar test site at the University of Evora, Portugal. WobaS systems use the inputs of four off-the-shelf surveillance cameras and at least one ground measurement station for Direct Normal Irradiance (DNI) or Global Horizontal Irradiance (GHI). Without available ground measurements, modelled clear sky irradiance data can be used. Every 30 seconds, four ASI images are taken. In these images, clouds are detected and the 3-D positions of all visible clouds are derived via a technique called voxel-carving. By tracking clouds over multiple timestamps, cloud velocities are determined and used to predict future cloud movements. With the 3-D shapes and positions of all clouds known, their shadows are projected on a ground model. If available, reference ground measurements are used to determine cloud transmittances. With transmittances known, the shadow maps are transformed to irradiance maps. The WobaS systems predict GHI, DNI and GTI maps for an area of 8x8 km2. Spatial and temporal resolutions for these maps are 25 m2 for every 30 seconds for predictions up to 15 min ahead. The working principle is illustrated in Figure 2.

Figure 3 The validation of the WobaS system was conducted on 30 days. Special consideration was given to the effects of spatial and temporal aggregations.

The WobaS system was validated on 30 days. For the validation of the nowcasted irradiance maps (see example in Figure 3), the effects of spatial aggregations are considered: For larger plants, deviations between the WobaS predictions and reference measurements are significantly reduced. Furthermore, the effects of temporal aggregations were studied, which determine the size of the batteries needed to fulfill ramp rate regulations. The conducted validation found the WobaS system to be reliable for operation in industry. The WobaS system and its validation will be presented at SolarPACES 2017 (speaker: Bijan Nouri) as well as at EUPVSEC 2017 (speaker: Pascal Kuhn).

Authoring Information


Deutsches Zentrum für Luft- und Raumfahrt e.V. (DLR),
Institut für Solarforschung

P. Kuhn, B. Nouri, Robert Pitz-Paal, S. Wilbert

DLR, Deutsches Fernerkundungsdatenzentrum
M. Schroedter-Homscheidt

CSP Services GmbH
T. Schmidt

TSK Flagsol Engineering GmbH
Z. Yasser