Auswirkungen von Verschmutzung auf konzentrierende solarthermische Kraftwerke

The soiling of solar mirrors compromises significantly the efficiency of concentrating solar power (CSP)-plants. It causes a partial loss of incoming radiation when being reflected at the mirror surfaces of the solar field. This reduces the output of the power plant or causes increased cleaning costs. Many potential CSP-sites lack information on soiling. In this thesis measurement methods are presented to quantify the effect of soiling. The collected data is applied in a financial yield analysis study and a statistical analysis concerning the meteorological causes of soiling. A novel measurement device called Tracking Cleanliness Sensor (TraCS) is presented. It is an accessory to an existing irradiance measurement station with a solar tracker. It consists of a second pyrheliometer that is installed to measure the direct normal irradiance as reflected by a sample mirror. Comparing the reflected to the directly measured DNI is a measurement for the reflectivity of the sample mirror. The parameter cleanliness is calculated by comparing the reflectivity of a soiled mirror to its reflectivity in the clean state. In a next step, the soiling-rate is calculated. The soiling-rate is defined as the daily loss of cleanliness with time. TraCS is a low cost instrument that requires little additional maintenance. Further advantages compared to currently employed measurement methods are its high time resolution, the use of the current sun spectrum and the measurement of a large mirror surface area of 30 cm2. A soiling-rate dataset of 1, 5 years measured at Plataforma Solar de Almería (PSA) in Spain in daily time resolution is presented and analyzed. The mean soiling-rate is - 0, 0052 /Tag. The dataset is applied in a yield analysis study for two types of power plants at two different sites. The focus lies on the exact reproduction of cleaning activities and costs. Adjustment of cleaning activities to a daily measurement of the soilingrate can increase the profit of a power plant project over its lifetime. Candidate cleaning strategies are compared to a constant cleaning activity with one vehicle (reference). The increase in profit ranges up to 2, 6 %. In existing yield analysis software packages the mean cleanliness of the solar field (xFeld) is assumed as constant. Setting xFeld equal to the yearly mean value of xFeld resulting from the reference strategy and calculating the yearly monetary yield results in a 0, 58 % higher profit than that resulting from a time resolved xFeld. The assumption of a constant xFeld overestimates the project’s revenues in this case because the different weighting of daily DNI-sums by xFeld is not accounted for. Time resolved soiling-rate measurements therefore are an important pre-requisite for the correct estimation of yield analysis studies and optimization of cleaning activities. In a next step, the soiling-rate dataset is compared to the simultaneously measured weather parameters airborne particle mass concentration, diffuse irradiance, dew, wind speed, wind direction, relative humidity and combinations of the latter. It follows, that the particle mass concentration for particles up to 10 �m (PM10) correlate best with the soiling rate. The adherence of particles of this size class to the mirror surface is potentiated by the occurence of dew. On days where particles impact mainly perpendicular to the mirror surface, the correlation coefficients between PM10 and the soiling-rate are higher for many subperiods of two to four vii months. The slopes of linear fits between soiling-rate and the investigated weather parameters vary with time. This can be explained by alternating properties of the adhered particles. For the definition a general model, further measurement parameters have to be included. The novel instrument TraCS enables measurements of time-resolved soilingdatasets. These are necessary for a refinement of yield analysis and the development of improved cleaning strategies as shown here. The influence of several weather parameters on the soiling-rate has been identified