Too Much, Too Soon? Exploring Trade-offs in Generation Expansion Planning

In this paper, we compare and evaluate four model formulations for solving the Generation Expansion Planning (GEP) problem, including static and dynamic models and models, with and without the transmission network. Each formulation is evaluated on three key metrics at the optimal solution: (1) the percentage of load shedding, (2) the utilization rate of newly installed units, and (3) CPU time as a measure of computational complexity. The models are tested in two settings: a toy, two-node power system and three large benchmark systems (39 - 118 bus). The two-node system allows for detailed analysis of the solutions, while the larger systems enable evaluation of scalability and performance of each formulation. Results show that including the network constraints entirely reduces the potential load shedding in future network operation, while the dynamic models show a better utilization of newly installed units compared to static models.