Exploring wind energy for small off-grid power generation in remote areas of Northern Brazil

This paper proposes simple and deterministic model for generating an hourly synthetic wind speed time series based on a typical meteorological year. It considers daily averages, the maximum wind speed of the day, and its time of occurrence. The model was tested for a location in the northern region of Brazil and is crucial for dispatching energy generation in remote areas such as indigenous communities where demand is seasonal. The model developed in this work employs the ARMA model and Weibull distribution to generate the hourly synthetic time series. For the validation of the model, MERRA-2 reanalysis data were used. A simulation of hourly energy generated by two commercial turbines was performed. The results showed that the northeast wind direction predominates in the studied region of Raposa Serra do Sol, with most wind speed data within the limits of 2 and 8 m/s. Additionally, the data revealed that winds are stronger between December and March, with averages exceeding 6 m/s. An adjacent stationary series was used with the Gaussian space mapping method, obtaining a series with normal distributions that can be easily modeled using autoregressive methods. Furthermore, an hourly modeling approach is employed to predict the intraday behavior of wind speed. The synthetic hourly time series for a typical year was then generated and compared with values from 2010, revealing a high degree of congruency between the model and real data.