| dc.description.abstract | Hydropower bidding in physical short-term markets is a complex task. It requires careful
modelling of sufficient technical details and good description of the market uncertainty. This
thesis contributes to the field both in terms of scenario generation for electricity market prices,
and with regards to modelling of the bidding optimisation problem.
The thesis show how modelling using stochastic short-term prices requires new coupling
between different planning horizons. Due to autocorrelation in price, the scenario path in
the short-term problem influences the probability for different price scenarios in the long-term
problem.
Regarding scenario generation, it is shown how day-ahead market prices in the Nordic
market has hour-specific variability and dependency on external price drivers. Further, it has
been shown that balancing market prices are hard to predict through time series models. However,
probabilistic forecasts of decent quality can be made when the methodology separates
between the probability of a regulation and the magnitude of the balancing event.
Bidding in the day-ahead market requires sufficient detailed modelling of hydropower
in order to reap benefits from bidding optimization. Variable efficiency, minimum production
levels and start-up cost are important features. Due to the need for high degree of modelling,
decomposition becomes difficult. This thesis shows how stagewise decomposition fails as a
solution technique to the bidding optimization problem because of the shape of the cost curves.
The gains of coordinated bidding in multiple markets are investigated. Coordinated
bidding in day-ahead and balancing markets is found to have negligible impact on profits,
although the bidding pattern is slightly altered in order to position the producer for more
flexibility in the balancing market. Since the balancing market is bound to have limited volumes
in order to preserve system security, the gains from coordinated bidding will stay limited. | nb_NO |
