Uncertainty analysis of Rotating Shadowband Irradiometers

Kurzfassung

Concentrating solar power projects require an accurate knowledge of direct normal irradiance in order to predict the electricity costs and a plant´s operation performance. Several years of satellite data ara available for many locations to estimate irradiances, but to otbtain the required high level of accurycy, ground measured data is necessary. For the ground measurement acquisition thermopile sensors are widely accepted. Due to their sensor properties, those sensors can obtain high levels of accuracy. However, those sensors need an expensive tracking device. Besides, thermopile sensors are prone to soiling, so that the sensors must be cleaned on a regular basis to keep the required level of accuracy. An alternative are Rotating Shadowband Irradiomters which use silicium sensors to measure the irradiance. Those sensors are less prone to soiling and cheaper. Escpecially at remote sites this option is much cheaper as the power deman of a measureing station can be supplied with a photovoltaic system. Better irradiance measurement results at remote sites can be obtained with Rotatin gShadowband Irradiometers due to the difficulty of maintenance and operation which is a problem for thermopile sensors. Silicium sensors have systematic errors that are caused by termperature, incidence angle and spectral effects. The spectral respose of a seilicium sensor is nonhomogeneous but varies in the wavelength region of about 400 to 1200 nanometer. This nonhomogeneous response can result in measurement errors. Radiation outside this wavelength interval is not measured by silicium sensors. Hoever, thos systematic errors can be compensated by correction functions. the resulting uncertainties of the measurement are so far mostly derived by comparison to reference pyranometer. An approach to derive the emasurement uncertainty for a Rotating Shadowband Irradiometer without the use of reference pyranometers is presented in this thesis. The Guide to the expression of Uncertainty in Measurements is used to state an internationally comparable and accepted uncertainty. To understand the origin of Rotating Shadowband Irradiometer´s uncertainties, the solar spectrum and influences on the solar spectrum ar explained in section 2.1. Extinction processes in the atmosphere are described in this section. Those extinction processes cuase a separation of irradiance in different components. Also models to calculate those effects on solar radiation are introduced. This section is fundamental as the nonhomogeneous spectral response of the sensor causes a spectral uncertainty which is later calculated. Ins ection 2.2 used irradiance sensors are explained. The difference between thermopile sensors and silicium sensors are explained and a closer look on the functionality of a Rotating Shadowband Irradiometer is presented. The correction funcitons that are used to reduce the systematic errors of silicium senosres are explained in section 2.3. After that, section 2.4 reports an existing standard and an approach to calculate uncertainties for pyranometers and pyrheliometersas in (JCG08). This is done to get an idea of other approaches that derive uncertainties for a sensor and to see which contributions should be taken into consideration. Results of other uncertainty analyses for Rotating Shadowband Irradiometers are evaluated to see which values can be expected from own calculations. Section 2.5 explains briefly, how an uncertainty can be drevied and stated after (JCG08). The evaluation fo the uncertainty calculation is started in section 3 where the spectral uncertainty is calculated as this uncertainty parameter is the most difficult one. An own approach is used to caluclate this spectral uncertainty of a Rotating Shadowband Irradiometer for one location which is explained in section 3.3. Atmospheric parameters are selected for the Plataforma Solar de Almeria in souther Spain for which a Rotating Shadowband Irradiometer shall be evaluated. Section 3.4 summarizes the approach to derive a spectral uncertainty. Having specified the most difficult uncertainty parameter, other uncertainty parameters are described and evaluated in section 4.1. Those uncertainties can be combined according to (JCG08) as described in section 4.2 and calulated for the direct normal irradiance as in section 4.3. The caluclated uncertainty can be sted in different ways. A second option is stated in section 4.4. The results of the uncertainty calculation are stated in section 5. First one day is evaluated and analyzed in section 5.1. The contribution of each uncertainty parameter is analyzed in section 5.2. Secondly, to validate the uncertainty analysis, a dataset of one year is evaluated in section 5.3 and compared to reference pyranometer at the same site as well as to uncertainties of other researches. Finally, the outcome of this thesis is summarized in section 6 and an outlook for further research is presented.