As renewable energy becomes an increasingly important part of our electricity system, grid operators face a growing challenge: how can the power system remain secure and efficient when renewable generation, electricity demand and market conditions are uncertain?
The DISCRETE project – Data Driven Optimization Models for Secure Real-Time Operation of Renewable Dominated Power Systems – has explored new data-driven optimisation and uncertainty-modelling approaches to support secure and cost-efficient power system operation.
To mark the conclusion of the project, EnergyVille/Etch and the University of Mons are organising a final webinar bringing together two perspectives from the research carried out within DISCRETE.
Programme
- 09:30 – Welcome and introduction
Brief introduction to the DISCRETE project and its key objectives. - 09:35 – Risk-Based Congestion Management Under Non-Gaussian Uncertainty in Hybrid AC/DC Grids by Kaan Yurtseven, KU Leuven
The increasing share of wind and solar power brings greater uncertainty into transmission system operation. This can lead grid operators to maintain larger operational margins, reducing the transmission capacity available to the market and increasing reliance on costly measures such as generation redispatch and renewable curtailment.
In this presentation, Kaan Yurtseven will present his PhD research within the DISCRETE project, which developed a stochastic optimisation framework for congestion management in hybrid AC/DC grids.
The approach uses Polynomial Chaos Expansion to represent and propagate non-Gaussian uncertainty in renewable generation, demand and imbalance prices. It also explores how non-costly actions, such as transmission switching and power flow control through HVDC, can reduce the need for costly redispatch and curtailment while maintaining N–1 security.
A full-year study using the open-source Belgian transmission grid showed a reduction in operational costs of around 15%. - 09:55 – Safe Reinforcement Learning for Battery Energy Storage Participating in the Real-Time Imbalance Market by Jean-François Toubeau, University of Mons
The increasing penetration of renewable energy has intensified uncertainty in power systems, leading to more frequent imbalances and highlighting the value of Battery Energy Storage Systems (BESS) as a source of flexibility.
In this presentation, Jean-François Toubeau will discuss a fast and safe reinforcement learning framework that enables battery energy storage systems to operate efficiently in the Imbalance Settlement market under uncertain conditions.
The approach combines data-driven market learning with physics-based feasibility guarantees through a differentiable projection layer that enforces BESS operational constraints. Using real Belgian market data, the proposed framework achieves safer operation, faster convergence and higher profitability than both conventional reinforcement learning and model-based approaches.
- 10:15 – Q&A and discussion
- 10:30 – End of webinar
Join the webinar
The webinar is free to attend, but registration is required.
Register for the webinar
About DISCRETE
DISCRETE investigates new approaches to uncertainty quantification and data-driven optimisation for secure power system operation. The project aims to support better operational decisions in increasingly renewable-dominated electricity systems, with the ultimate goal of reducing operational costs and CO₂ emissions while maintaining grid security.
The project is supported by the Energy Transition Fund.
With the support of:
