A clustering approach for the analysis of solar energy yields: A case study for concentrating solar thermal power plants

The design of Concentrating Solar Thermal Power (CSTP) systems requires a detailed knowledge of the dynamic behavior of the meteorology at the site of interest. Meteorological series are often condensed into one representative year with the aim of data volume reduction and speeding-up of energy system simulations, defined as Typical Meteorological Year (TMY). This approach seems to be appropriate for rather detailed simulations of a specific plant; however, in previous stages of the design of a power plant, especially during the optimization of the large number of plant parameters before a final design is reached, a huge number of simulations are needed. Even with today’s technology, the computational effort to simulate solar energy system performance with one year of data at high frequency (as 1-min) may become colossal if a multivariable optimization has to be performed. This work presents a simple and efficient methodology for selecting number of individual days able to represent the electrical production of the plant throughout the complete year. To achieve this objective, a new procedure for determining a reduced set of typical weather data in order to evaluate the long-term performance of a solar energy system is proposed. The proposed methodology is based on cluster analysis and permits to drastically reduce computational effort related to the calculation of a CSTP plant energy yield by simulating a reduced number of days from a high frequency TMY.