Modelling of the balancing market

Context

The increased market share of renewable energy introduces a higher level of variability and unpredictability in the generation of power. This variability in power supply has a significant impact on the different electricity markets. Today, the day-ahead market and the intraday market have been using a unit commitment dispatch model to simulate the market. In this model, we optimize the usage of different production units, to meet the demand whilst obliging to some constraints (e.g., load on the electricity grid). In the past, there was little to no need for a complicated modelling technique for the balancing market since the classic power generation techniques were rather static and reliable. To comply with the changing market, this will need to change. 

Approach

We started from the following problem statement “Can we create a perfect foresight through the modelling of the Balancing Market, similar to what we have today in the day-ahead market?”.

At first, we analysed if the current methodologies, which are being used for the day-ahead market, can be extrapolated to the balancing market. We assessed the current constraints for the unit commitment dispatch model and its scalability to the balancing market.  

Secondly, we assessed the evolution of the balancing market and the difficulties we see today and will see in the future. Here we assessed the status and pain points of the implementations of European wide Electricity Market programs such as IGCC, MARI, PICASSO, TERRE. 

Thirdly, we conducted a benchmarking study with partners from the TSO sector, research institutes, commercial organizations, and balancing responsible parties (BRPs). This study involved multiple interviews to obtain a comprehensive understanding of current activities across Europe.

Lastly, we performed a brief analysis to evaluate the feasibility of developing a modelling solution in-house. This assessment included a cost-benefit analysis specific to our company.

Conclusions

• The models which are being used currently for short-term predications cannot be used for long-term modelling as the results of these become unreliable over the long-term. Long term models will inherently not be able to proceed with the same complexity as the short-term ones. Their simplicity will allow them to stay correct, but therefore also less accurate. 
• Almost all market players (TSO´s, BRP´s, commercial actors) focus on short term forecasts and are therefore not looking into more in depth modelling of the balancing market.
• A lot of the current forecasting is being done with the use of historical data, which is, certainly for the balancing market, not a good representation of future market dynamics.
• There is currently no academic focus on the topic of modelling the balancing area, this is mainly driven by the lack of interest from the market players due to a lack of use cases for the modelling  .

Partners