Aportaciones al modelado de la generación fotovoltaica para la planificación de la red de distribución
- Autores: Miguel Ángel Ramón Marin
- Directores de la Tesis: Roberto Villafáfila Robles (dir. tes.), Andreas Sumper (dir. tes.)
- Lectura: En la Universitat Politècnica de Catalunya (UPC) ( España ) en 2012
- Idioma: español
- Tribunal Calificador de la Tesis: Sergio Herraiz Jaramillo (presid.), Oriol Gomis Bellmunt (secret.), Joan I. Frau Valentí (voc.)
- Materias:
- Enlaces
- Tesis en acceso abierto en: TDX
- Resumen
In this thesis it have been explained the problems that the electric distribution companies in Spain have to plan and operate their power networks, having no detailed measurements of instant active power of a large number of non-programmable generators connected to them. The main objective is to reduce errors and difficulties that this lack of information generates. It has been analyzed the specific case of photovoltaic generators (PV), representing at the end of 2011, a 94.6 % of the special regime generators connected to the power networks in Spain, and have an average power of 72 kW per plant at the end of 2011. This means that are mainly connected to the distribution network, so we study the specific issues regarding the planning of the distribution network. It is defined that prediction models of solar irradiance are divided into parametric and decomposition. The ASHRAE parametric model prediction of solar irradiance is described, checking that its instant error levels are not negligible when compared with real data, since it does not include components that provide greater instant solar irradiance variability, such as atmospheric phenomena. It has been described a model based on the usability for the prediction of instantaneous power of PV generation, which is based on an hourly rate of sunlight (Kt) to transfer the extraterrestrial solar irradiation to a horizontal surface on Earth ¿s surface. It has been seen that the definition of the statistical function that defines Kt, makes the difference between the different models based on usability. It has been developed a proposed solution, which allows the development of models using predictors measured at the Earth¿s s urface, which incorporate the variability of atmos pheric phenomena. It has been evaluated the available data for the distribution companies, as well as specified the conditions necessary to develop specific models to be applied in distribution network planning. The proposed solution consists in a systematic methodology for developing models, based on historical data, which provides a response to a wide range of applications for distribution network analysis. It has been shown that the geographic scope of validity of the developed models exceeds the average needs for distribution network planning. It have been developed three case studies, where application of the methodology has been developed, based on real data of a distribution network in the province of Lleida, Spain. Also, it has been found that the models needed for photovoltaic plants with solar follower are different from those needed for photovoltaic plants without follower, because they need to be developed from different predictors. Contributions have been developed based on examples of application of the developed models to network analysis, based on another set of real data from the distribution network in Lleida, corresponding to a medium voltage line (25 kV). Based on this example it have been described the possible options when applying the methodology, as well as several different situations, drawing conclusions. Furthermore, it have been evaluated the possible seasonality of the models, and therefore, the contribution that would get from the development of specific models for each quarter, against the development of a single model for the prediction of the whole year. Barcelon
