Risicogebaseerd congestiebeheer onder niet-Gaussiaanse onzekerheid in hybride AC/DC-netten

The increasing integration of renewable energy sources is changing the way transmission systems are planned and operated. Wind and solar generation reduce dependence on conventional generation, yet their weather-dependent nature increases the effect of forecast uncertainty on congestion management, balancing, and real-time system security. In addition, demand forecasts also carry errors. The deviations that remain appear as imbalances of the market parties, which are covered by balancing energy and settled at imbalance prices that are not known when operational decisions are taken, so the balancing cost that follows a decision is uncertain as well. At the same time, the deployment of high-voltage direct current transmission is leading to hybrid AC/DC grids, where controllable converter set-points provide additional flexibility but also change how uncertainty propagates through the system. These developments create a need for congestion management methods that account for continuous non-Gaussian uncertainty, coordinate multiple flexibility sources, and quantify both operational and economic risk. This thesis develops a modular stochastic optimization framework for congestion management in hybrid AC/DC grids under uncertainty. The framework is based on Polynomial Chaos Expansion, which enables continuous uncertainty in renewable generation, demand, and imbalance prices to be propagated through optimal power flow formulations without relying on discrete scenario sampling. Within this setting, chance constraints are used to represent operational risk, while Conditional Value-at-Risk is introduced to account for unfavorable economic outcomes. The proposed formulations provide both optimized decisions and probabilistic quantities of interest, including distributions of power flows, voltages, renewable curtailment, redispatch, and operating cost. The first of these formulations is a chance-constrained stochastic optimal power flow model for hybrid AC/DC grids, which jointly optimizes conventional generation, renewable curtailment, and HVDC converter set-points under continuous non-Gaussian uncertainty and treats curtailment as an active decision variable coordinated with grid-side flexibility. The framework is then extended along several modeling dimensions. Transmission switching is incorporated as a non-costly congestion management action through a relaxation strategy compatible with intrusive Polynomial Chaos Expansion. In an N-1 security-constrained formulation, decision variables are assigned an admissible dependence on contingency states and forecast error realizations. A risk-aware extension introduces economic risk into the objective function, so that the trade-off between expected operating cost and unfavorable cost realizations can be examined directly in coefficient space. The final contribution integrates these developments into a risk-based balancing-aware congestion management framework, which links day-ahead redispatch and transmission switching decisions with the balancing cost that follows them once forecast errors are realized. The decision of the operator remains the preventive action that removes the overloads, while the balancing activation that follows it is anticipated through merit-order curves constructed from published market data, with the imbalance price entering as an additional source of uncertainty. The balancing cost is settled among the market parties and is not a cost of the operator, but the preventive decision of the operator influences its size. A preventive decision that leaves the cheaper balancing energy available for the imbalances of the market parties therefore preserves the social welfare of the balancing market. By evaluating the Conditional Value-at-Risk of the balancing cost, the framework makes this effect visible before the decision is taken, so that the operator can account for it when selecting the decision. Application to the Belgian transmission grid over a full study year lowers the total system cost, which is the congestion management cost of the operator together with the balancing cost settled in the market, by about 15%. A perfect selection of the risk-aversion level would raise the saving to about 25%. Congestion management under uncertainty can thus be formulated as a risk-aware decision problem that provides a methodological basis for decision-support tools for transmission system operators.