| dc.description.abstract | The current deterministic day ahead market clearing in wholesale electricity markets, which was constructed around dispatchable and controllable electricity from conventional producers, is in need of adaptation to address the increased uncertainty resulting from the integration of variable renewable energy sources. Through their stochastic nature, variable renewable energy sources are non-dispatchable and output cannot be controlled, potentially resulting in grid imbalances in cases of forecast errors. Two-stage stochastic optimization and Adaptive Robust optimization utilizing the Column-and-Constraint generation algorithm are proposed for the day ahead market clearing in a case study on the Dutch, French and German wholesale electricity markets. A framework of performance indicators is constructed to evaluate- and compare the models on their ability to maximize the social economic welfare, increase system security of supply and integrate variable renewable energy sources in the energy mix. The results of the case studies indicate an increased social economic welfare through improved system security by the proposed models, drastically reducing the occurrence of load shedding events. The stochastic model outperforms the robust formulation during the in-sample scenarios, however proving vulnerable to unexpected stochastic output realizations during the out-of-sample scenarios. The Robust model, showing the highest degree of system security of supply because of its conservative nature, significantly decreases the integration of variable renewable energy sources in the energy mix while requiring the highest amount of upward reserve capacity. Both proposed models indicate a need for increasing the upward reserve capacity in energy systems with high penetration of variable renewable energy sources, while decreasing the utilization of installed stochastic producer capacity. Results of the sensitivity analyses indicate an increased system security to unexpected output realizations of the stochastic model by increasing the in-sample size, although increasingly extended computational time was observed for solving the problem. The sensitivity analysis on the budget of uncertainty of the robust model showed a direct trade-off between integration of variable renewable energy sources and system security of supply, while a relationship between the distribution of installed stochastic producer capacity among the stochastic producers and the budget of uncertainty revealed potential market inequalities in real-world applications of this formulation. | |
