Development of an ANN model for forecasting time series of cooling performances in CSP plants - Solar tower power plant (Gemasolar) as a case study-

Abstract

Unlike conventional thermal power plants which are generally erected in the coastal regions near water bodies, where water resource availability is not a critical factor, concentrating solar power (CSP) plants are usually installed in high sunny zones, where water resources are hard to find and/or to employ. This from a side, on the other side, the hourly simulations of cooling performances in CSP plants, including power required for cooling and water consumption, are very complex and require many calculations, and time consuming. Therefore, the aim of this paper is to develop an artificial neural network (ANN) model to estimate the hourly cooling performances of a solar thermal power plant based on its hourly generated power, ambient temperature and wind speed. In this regard, the commercial power plant Gemasolar with solar tower technology has been chosen to perform the study. The obtained results of the statistical analysis show that ANN can be used as a good option to forecast these two parameters in a large-scale solar tower power plant without passing through detailed modelling. However, it should be pointed out that this model can be only used for plants that are simulated in locations with similar meteorological conditions and solar resources as the site where this study has been performed.