Abstract
Bidding of flexible reservoir hydropower in day-ahead (spot) auctions needs to be done under uncertainty of electricity prices and inflow to reservoirs. The presence of reservoirs also means that the short-term problem of determining bids for the next 12–36 hours is a part of a long term problem in which the question is whether to release water now or store it for the future. This multi-scale challenge is usually addressed by using several models for hydropower planning, at least one long-term model and one short-term model. We present a multistage stochastic mixed integer programming model that has a fine time resolution on near term, and a coarser resolution going forward. It handles price as a stochastic parameter and assumes deterministic inflow as it is intended for use in the winter season.
