Temporal Forecast Uncertainty for Ramp Events

Abstract

As the penetration of wind energy continues to increase around the world, the impact of wind energy on the management of electrical grids is becoming increasingly evident. The challenge for the grid operator of integrating wind energy, or for the energy trader to maximise the market value of the energy, is toughest during periods of rapid change in wind farm production, or ramp events. These events are also tough to forecast accurately, and it is therefore essential to understand the uncertainty of forecasting such events. To date the majority of work on the uncertainty of wind energy forecasts has been focused on the possible amplitude of wind production that might occur at a given time. However, there has been limited investigation into effectively defining the possible timing of significant wind energy events. This paper aims to focus on methodologies for generating temporal forecast uncertainty for rapid changes in wind farm production. The first challenge is to define ramp events, secondly the forecast uncertainty needs to be calculated and finally this information needs to be presented clearly to the end user. This paper covers these three areas, with a focus on the method of calculating forecast uncertainty using multiple NWP inputs, statistical processing and adaptive algorithms. The results are based on GH Forecaster services for both individual and portfolios of wind farms. The outcome of the investigation demonstrates that temporal forecast uncertainty can be calculated and clearly presented to indicate the likely timing and amplitude of wind energy ramp events.