Evaluation of Price Forecasts for Hydropower Expansion Planning Applications
MetadataVis full innførsel
This Master Thesis deals with the importance of forecasting as exact power prices as possible for use in expansion planning. The work is a continuation of the Project Thesis finalized in December 2007 (Lind & Eggerud, 2007), which indicated certain deviations between the forecasted price fluctuation and the observed market price fluctuation. Consequently, the hypothesis of this report is that the forecasting tool has difficulties simulating under certain conditions. The analysis of prices for year 2000 to 2006 confirms the hypothesis, and states that there are significant deviations between simulated and observed price fluctuation. The deviations seem to be seasonally correlated, and the simulated prices tend to deviate more strongly for the depletion season than for the filling season. Another result is however more important, revealing major deviations for both seasons when considering the difference between the two segments of highest prices. The report shows that the EMPS model used for forecasting undervalues the prices for the heavies loaded hours, while it tends to overvalue the prices for the hours of low load. In addition, a study of monthly average prices throughout the years 2000 to 2003 shows that the model has difficulties forecasting prices during periods of frequent and important changes in price. How this affects generation and income for the power plants used as case studies in this report is investigated through simulations with the EOPS model. Two price records are used as input in order to point out the importance of forecasting the market prices. The first price record is the simulated output from the EMPS model. The second price record is adjusted according to the study of price deviations, in order to represent the market prices to a larger extent. By improving the price forecasts (represented by adjusted price records) simulations result in considerable increase of income, although production is weakly changed. For the three power plants used in this study, annual income is increased by 0.28 % to 3.25 %. Other parameters have shown to be relevant with respect to the possibility of increasing income, and the power plant s utilization time seems to be the most important parameter, since this provides larger flexibility. Despite the little change of accumulated production, income is considerably increased. Details on the simulated production pattern shows that this is due to a reallocation of production, from hours of low price to hours of more elevated price. The report also states that having additional capacity may increase the income per unit of energy produced. This is valid for both price records, although the income is somewhat higher when based on the adjusted price record. Simulations show that a capacity development would give a significantly increased income, and when combined with the improved forecasting, simulated income would be considerably higher. In a wider perspective, the report indicates that the extended interconnection with Continental Europe will influence the Nordic power prices. Based on the study of the German and the Dutch power exchanges, it is anticipated that the Nordic countries will experience more fluctuating prices with a larger range within each day. The interconnections will introduce a larger share of thermal and wind generation, causing new challenges for the forecasting tool. Consequently, in order to evaluate a possible hydropower expansion in the future, an improvement of the forecasting will probably be required.