Coordinated Multimarket Bidding for a Hydropower Producer using Stochastic Programming
Abstract
With the connection of new regions into the Nordic power markets competition intensifies. The incentive of power producers to exploit their portfolios more effectively is increasing. Expecting a higher share of non-dispatchable energy production in the years to come, supply of ancillary services (reserves) will be divided between those producers who possess flexible portfolios. The question then arises whether a hydropower producer should take the reserve markets into account when submitting bids for the day-ahead auction.This thesis considers the bidding problem of a Norwegian hydropower producer bidding into the day-ahead, taking the primary reserve and balancing market into account. A stochastic program is proposed to investigate the value of coordinating the day-ahead bid with the reserve market opportunities.Stochastic programs require good scenarios to produce high-quality solutions, and hence a great effort is put into generating scenarios that do well at predicting market prices and volumes, and representing the associated uncertainty. A comprehensive scenario generation framework that captures the dynamics of each market as well as their inter-dependencies is proposed, tested and found to perform well.
Modeling choices and assumptions are made with the goal of attaining reasonable solution times, while at the same time reflecting the actual planning procedures of the producer. Short solution times allow for a comprehensive case study to be conducted for 250 days in 2016. The case study is performed under three control variables: price deviation from water value, planning horizon granularity and the number of watercourses in the portfolio. First, the testing is done with one watercourse at disposal. Coordination is found to yield a small gain of about 1 %. Next, the profitability of coordinated bidding is further investigated with respect to price deviation from water value. The results show that the gain associated with coordination is higher when this deviation is low. The gain tends to zero when the deviation increases. Next, we investigate the effect of adding more watercourses to the planning problem. Gains decrease, but seem to stabilize around a value of 0.5 %.