A combined probabilistic/time series model for wind diesel systems simulation

Abstract This article describes a new simulation model for wind/diesel systems. It involves a combined time series and statistical approach to estimate the fuel use of diesel generators. In addition to provision for modeling non-identical diesels, the model allows the inclusion of multiple, non-identical wind turbines whose output may or may not be correlated. Three diesel dispatching strategies are provided. One assumes no storage, and when storage is employed, either a peak shaving or cycle charge control option can be used. The storage module uses a flexible battery model specially designed for time series simulation codes. A key assumption for the main analytical model is that, within each time step, the load and wind power are assumed to be normally distributed. The mean net load is the mean load less the mean wind power and its variance is found from the variance of the load and the wind power. A loss of load probability is used to find the maximum and minimum anticipated values of the net load. In addition to summarizing the overall analytical model, this article presents the results of a number of simulations demonstrating the performance prediction (diesel fuel usage) capabilities of the model. For one of these cases (a no storage system), the results show excellent correlation between the model and actual data. Other cases summarized show that the use of the model greatly facilitates the integration of storage into the control scheme, and gives the fuel saving potential for several different wind/diesel system configurations.