Pandora - a helping hand for intra day power plant optimisation
The energy transition is changing the working world of intra day power plant optimisation at a breathtaking pace. While the predictability of thermal base load power plants used to limit volatility, the growing share of renewable energies is now leading to short-term oversupply or undersupply, which is causing the quarter-hourly wholesale prices to fluctuate significantly.
December 2023
Brit Gericke - Head of Algorithms & Optimisation
One of the main tasks of intraday power plant optimisation is to adjust generation in line with market signals. Thanks to the efficient market design, the economically motivated decisions of the market participants ideally lead to a balance of supply and demand without stabilising interventions by the grid operator. What makes short-term wholesale electricity trading so special is that market participants and grid operators share an interest in a stable electricity supply. Intraday power plant optimisation is therefore highly motivated to use the power plants in a way that serves the grid.
Manual power plant optimisation becomes a challenge
Vattenfall's power plant portfolio includes the largest pumped storage power plants in Germany, which play an important role in covering peak loads. As a result, dozens of quarter hours had to be managed simultaneously by the power plant optimisation department and the manual placement and processing of buy and sell orders soon exceeded the limited possibilities. The idea of a fully automated solution was born!
Fully automated power plant optimisation with Pandora
In 2018, Vattenfall analysts developed a trading algorithm that they named Pandora. The solution for the fully automated placement of flexibility on the market consists of the following components
- a mixed-integer linear optimisation model that calculates the profit-maximised power plant deployment using market, weather and power plant data,
- a logic for creating bids based on the optimal schedule,
- a logic for placing and pricing bids (machine learning model),
- an infrastructure for secure connection to the exchanges (NPS, EPEX and ETPA) and
- a new run-through of the optimisation model to distribute the contracted trading volume to the available power plants in the portfolio.
The process is repeated continuously, is monitored without interruption and was set up in accordance with MiFID and REMIT market guidelines.
Algorithm optimises a wide range of power plants and portfolios in seven countries
The various instances of the Pandora trading algorithm currently manage a considerable part of Vattenfall's customer and power plant portfolio in seven European countries:
- Hydropower portfolio in Finland and Germany
- Wind power portfolio in Sweden, Denmark, Finland, Norway, Germany, the UK and the Netherlands
- thermal power plant portfolio in Germany and the Netherlands
- Hydrogen storage in Sweden
- End customer portfolio in the Netherlands and Sweden
- from 2024: battery storage in Germany, the Netherlands and Sweden
- From 2024: Hydropower portfolio in Sweden
The attachment can be scaled as required and is suitable for any energy source thanks to the generalised optimisation model. Vattenfall offers market solutions that enable external utilisation of the proven trading infrastructure.
However, the vision remains the same in all cases: Vattenfall guarantees stable and fair wholesale electricity prices by automating the marketing of our flexibilities using standardised trading algorithms and optimisation models.
Do you have any questions about power plant management with the Pandora algorithm? Please feel free to contact us.
<a class="arrow">renewables@vattenfall.de</a>