Approaches for Long-term Regulating Prices Modeling in the Nordics

Sammanfattning: This thesis investigates methods to forecast the long-term regulating power prices (RPP)evolution in the Nordics. During the operating hour the transmission system operatorsbuy balancing power on the regulating power market, and the RPP are the clearingprices. Each market player having caused imbalance is charged an imbalance cost dependingon the RPP. Hence forecasting these prices in the long-term provides valuableinformation for strategic decisions such as wind power investment, power purchase agreementor for market players willing to have revenues from the RPM.First, fundamental approach is investigated. Nord Pool data and probabilistic distributionsare used to forecast the evolution of balancing needs. These volumes are then fedinto models of the balancing price bids ladder, based on the bottom-up model EMPS(EFI1's Multi-area Power Simulator), initially developed to model the long term dayaheadprices. Models appeared to perform poorly: results were underestimated and farfrom observed values.In the second part of this work, a computational intelligence approach using EmpiricalMode Decomposition (EMD) and Articial Neural Networks (ANN) is investigated. Thetrend of regulating prices is extracted with the EMD and forecasted separately withANN. Good results are achieved with the statistical approach, showing the superiorityof this method over the fundamental approach. In the nal part, the results of thestatistical approach are analyzed and conclusions regarding the long-term regulatingprices evolution are drawn for 2017-2020.