Wind Power Forecast Accuracy in Scandinavia:Analysis of Forecast Errors Using TAPM
Abstract: Wind power has different characteristics compared to conventional energy sources. The main difference is that wind power fluctuates under the influence of meteorological variables. This gives rise to problems related to grid stability. It is therefore important to understand these variations and their predictability in order to improve the integration of wind power into grids. Wind power forecasts play an important role. In this study, the accuracy of Numerical Weather Prediction (NWP) wind speed forecasts over the period 1st Sept. 2013 – 31st Dec. 2016 has been investigated. This was done for four Scandinavian windfarms; Rødsand II, Kårehamn, Jokkmokksliden and Storliden. The NWP used was The Air Pollution Model (TAPM). The main aim of the study was to examine if the implementation of bias correction methodologies could improve the accuracy of uncorrected TAPMforecasts. In addition to this main aim, the study was also intended to investigate the differences in the performance of TAPM between Australian and Scandinavian weather conditions. The results indicated that TAPM has a tendency to under-predict against actual observations. It was also observed that the accuracy varied throughout the year, with the best performance during summer and the worst during winter. Additionally, the performance varied between turbines within the wind farms. The magnitude of forecast errors was lowest for turbines which experience high degrees of wake effects from upstream turbines. Furthermore, the results showed that the implementation of certain bias correction methodologies improved the accuracy. The greatest enhancements were achieved by the implementation of two correction methodologies, both based on the combination of bias correction and timing correction. The NRMSE of wind speed was reduced by almost 50% for “hour ahead” forecasts for Rødsand II and Kårehamn when full bias correction was applied and by nearly 70% for Jokkmokksliden and Storliden. A reduction in the wind speed magnitude error by as much as half was also reached. Based on these results, the overall conclusion is that TAPM can be applied for sites with completely different weather conditions with moderately good accurateness, especially if bias correction is applied. However, due to the limited time frame and other delimitations of this study, further studies are necessary in order to draw deeper conclusions.
AT THIS PAGE YOU CAN DOWNLOAD THE WHOLE ESSAY. (follow the link to the next page)