Joint Use of Hydrological Modeling and Large-scale Stochastic Optimization Techniques Applied to the Nordic Power System
详细信息 查看全文
- 作者：Joachim Dahl Jensena ; joachim.dahl.jensen@gmail.com" class="auth_mail" title="E-mail the corresponding author ; Torjus Folsland Bolkesjø ; b ; Bjø ; rn Sø ; nju-Moltzauc
- 关键词：Stochastic Dual Dynamic Programming (SDDP) ; Hydrological modelling ; Power markets ; Benchmarking.
- 刊名：Energy Procedia
- 出版年：2016
- 出版时间：January 2016
- 年：2016
- 卷：87
- 期：Complete
- 页码：19-27
- 全文大小：534 K
文摘
The Nordic power market is a hydro-dominated one, having in a normal hydrological year about 50% of its power production from hydropower. The hydro scheduling optimization that is being done by the large hydro producers in Norway and Sweden has strong impact on power prices in the Nordic area. Hence, methodologies that can better predict future inflow levels and power prices are of great value to traders, risk managers and others that are taking positions based on expectations of current and future hydro supply. The objective of this paper is to (i) present a modeling approach for forecasting power prices in hydro-based power systems and corresponding uncertainty levels that are caused by hydrological uncertainty, and (ii) to benchmark the forecasting modeling system towards market data for the period 2010-1015. The hydrological models are built on the well-known principles of the hydrological HBV model. We have developed HBV models for the Scandinavia area, dividing the catchments into 23 homogenous hydrological regions. Using precipitation and temperature actual and forecasts for a large number of meteorological stations, we simulate for net precipitation (precipitation minus evapotranspiration), inflow, snow, soil- and ground water and hydrological temperature, measured in energy units. The fundamental power market model (SDDP) uses these data to simulate market-clearing prices assuming profit-maximizing producers. The benchmark analysis shows that the market price forecasts generally have a good accuracy and on average has higher predictive power than the financial market when it comes to forecasting the actual, delivered prices for front week and front month power futures.