We model pumped storage, run-of-river, and reservoir hydro separately
Different types of hydro power are used across GB and Europe, and each type has different operational characteristics and market behaviours. The buildout in countries with lots of hydro power is shown below.
- Pumped storage hydro — A flexible asset that can generate like a normal hydro plant, and also pump water uphill to refill its reservoir when prices are low
- Run-of-river — Captures the natural flow of water without a dam. Cannot store water, so must generate continuously
- Reservoir hydro — More flexible than run-of-river, with the reservoir allowing operators to hold back generation for times when prices are highest
Pumped storage provides flexibility similar to batteries
Pumped hydro has traditionally provided flexibility in day-ahead and intraday markets. In the model, pumped hydro uses similar logic to battery storage assets, with two key differences:
- Lower round-trip efficiency: More energy is lost in the pump-generate cycle
- Longer duration: Typically 8-12 hours of storage compared to 2-4 hours for batteries
Run-of-river plants generate continuously
Run-of-river plants generally cannot be curtailed. They must bid into the market at all times, so their bids are set at the market floor price in the generation stack.
Reservoir hydro plays a key role in setting prices
Reservoir hydro is highly flexible and plays a significant role in setting market prices across much of Europe.
In Iberia, reservoir hydro is a significant fraction of the generation mix and key to building the price shape. In Sweden, Norway, Switzerland, and Spain, reservoir hydro accounts for more than 10% of installed generation capacity. In Germany, reservoir hydro often sets prices indirectly via interconnector imports from these hydro-abundant Scandinavian and Alpine regions.
Seasonal variations affect available output of reservoir hydro
Hydro power is highly seasonal due to variations in rainfall and meltwater throughout the year. Operators must balance generating revenue against maintaining dam levels within safe limits.
The model captures this using historical data to extract the portion of hydro flow that runs constantly (the “baseline flow”), regardless of electricity price. The remainder responds to price signals. This is shown below for Austria.
Bidding behaviour of reservoir hydro is calibrated to historical data
In Spain and Portugal, seasonality affects not only baseline generation, but also how hydro generation bids in the different markets. During spring, when there is usually abundant rain and reservoir levels are high, hydro generation bids at near zero prices leading to very low prices in the day-ahead market, in contrast to other moments of the year, when they bid like a CCGT generator would.
To model this behaviour, reservoir hydro bidding in Spain and Portugal is estimated based on historical patterns and seasonality by determining a water-value factor which is then multiplied by the short-run marginal cost of CCGTs (the technology being substituted by hydro generation in Spain and Portugal). When reservoir levels are high and there is a risk of spillage, the water-value factor drops to close to zero, while during summer, when reservoir levels are low, this factor can go above 1, reflecting a scarcity premium.
For other countries without transparent bid data, SRMCs are calibrated using historical prices and load factors. The calibration is validated by comparing modelled price shapes against historical patterns.
Four European countries have hydro representing more than 10% of installed generation capacity: Sweden, Norway, Switzerland, and Spain. The model shows a good match to historical price shapes in these markets.