Abstract
Model-based analysis of future renewable electricity systems is highly complex, primarily because models must simultaneously evaluate both short- and long-term flexibility. This requires fine time resolution and detailed representations of the physical properties. Short-term flexibility can be provided by resources such as batteries, run-of-river hydro, pumped hydro, or demand-side flexibility. Long-term flexibility can be provided by hydro storages.
SINTEF has been developing hydro-thermal market models for many decades. In the Nordic market-based electricity system, these models are used for a range of purposes. The newest of these models, called FanSi, employs formal optimization to calculate the optimal utilization of short- and long-term flexibility resources. The model has performed well in various analyses, with one notable exception: the simulated operation of long-term hydro storages has been considered too risky by hydro producers during prolonged dry and wet periods. Specifically, reservoir storages tend to be emptied too early in the winter or spring in dry years. Since Norwegian electricity production is 100% renewable, with hydro storage playing a critical role in the system and its security of supply, overly risky reservoir operation undermines the reliability of the results.
This paper shows how a newly implemented risk-aversion functionality can be used to give reservoir operation more in line with producers wishes. The results are shown for a detailed model the Northern European electricity system. The risk-aversion functionality is based on Conditional Value at Risk (CVaR) and has been adapted to the optimization methodology used in FanSi. The simulation results include comparisons of reservoir operations and the probability distributions of simulated market prices for cases with and without the new functionality.