Solar power plants are very interested in the prediction of power net generation and the instabilities that can affect the production of the plant. System Advisor Model (SAM) is widely used to simulate the behavior of concentration solar power plants, where different inputs parameters can be considered for simulations. Reliable short-term forecasting of the solar resource has significant positive consequences since it helps the planning of power plant operators for power dispatch and assess the plant reliability and the market trends. In this work, we quantify the accuracy of the electricity predictions of a solar power plant at different hour-scale, by using the solar irradiance forecasting obtained from Meteosat satellite images. For that, we have simulated the power generation of a solar power plant taking as input the solar irradiance forecasting with the purpose to determine the best forecast horizon time. The main novelty in this work is the analysis on the impact of short-term forecasting of DNI in the plant energy output at different forecasting horizons in solar tower plants with large thermal storage systems.