Nuevas estrategias de planificación de producción en plantas termosolares con almacenamiento térmico

Abstract

In response to current energy problems, electricity generatiOn based on intermittent renewable energy, such as solar and wind energy, has grown significantly in recent years thanks to the decrease in costs. This type of generation has an intermittent, variable and difficult prediction character, which makes it difficult to integrate into the electricity grid. However, solar thermal energy has certain characteristics that can partially compensate for the above disadvantages. This technology captures solar radiation in the form of thermal energy by heating a fluid, to later convert it into electricity. By using thermal energy as a form of intermediate energy, this technology is complemented very well with thermal energy storage systems. Thanks to this energy storage system, this type of plants has a certain degree of dispatchability, with the possibility of regulating production. This property favors its participation in the electricity market, where the objective of the electricity producer is to maximize the economic benefits derived from the sale of electricity. This objective can be achieved when production is scheduled based on the electricity sales price profile. Therefore, an optimal generation scheduling problem may arise. Considering only theday-ahead energy market, the resolution of this optimization problem allows obtaining the daily generation schedule, which must be sent to the market normally the day befare. It follows from the above that in the case of renewable plants, in addition to the need to have a prediction of electricity prices, a forecast of the natural resource is required to address the problem. The main objectíves of this Thesis are the design of new strategies for optímal generation scheduling applied to a concentrating solar power plant, and the study by simulation of the economic performance of each strategy when considering the participation of the plant in the day-ahead energy market. The optimal generation schedule is obtained using mixed integer linear programming, which is the most used mathematical tool to solve these types of problems. The first strategy performs an hourly rescheduling of the generation, considering the revenues derived from the sale of electricity in a certain time horizon and the possible penalties for deviatíons from the generation schedule already committed. This strategy allows to include into the problem the new information available every hour, thus addressing the uncertainty present in the predictions and in the modeling of the problem itself. The second strategy includes a mechanism that penalizes changes in generation. In addition, this mechanism penalizes variations differently according to the state of the power block: normal operation, startup and shutdown. In this way, it is possible to increase the number of degrees of freedom of the problem in seárch of better solutions. This reduction of the variability in the generation has as advantages an extension in the lifetime of the elements of the power block, a reduction of its maintenance costs andan easier plant operability. A methodology is proposed to estimate the maximum level of penalty for variations that does not affect economic performance. Finally, another strategy was developed that combines hourly rescheduling with the penalization of variations. The economic impact of the above strategies has been evaluated through simulations on a 50 MW parabolic trough collector plant. Realistic data for the solar resource, electricity prices, penalty costs and predictions of all these data have been used. The results confirm the expected improvements in each of the strategies.