Robust and exible planning of power system generation capacity.

摘要

The successful evolution of a power system is achieved when its future growth path is visualized. Visualizing and interpreting the future are crucial to understand the risks to which the power system is exposed. These are mostly caused by the interdependencies between the power system and other systems (e.g., transportation sector, fuels sector, industry, etc.); and the resulting uncertain environment where these systems perform. Then, the objectives of planning are to reduce the risks of uncertainties and to gain some control over the future by linking it with the past; otherwise risks might materialize in catastrophic consequences.;In particular, motivated by the need of mitigating future risks in power systems, this work focuses on finding robust and flexible investment strategies in the generation capacity expansion planning problem under exposure to multiple uncertainties. They are present in different sources and types such as fuel costs, investment and operational costs, demand growth, renewables variability, transmission capacity, environmental policies, and regulation. The problem when considering multiple uncertainties is much harder, not only because the increased computational effort, but also because it is hard to model the combination of their occurrences in a single optimization problem.;Since each uncertainty deserves special treatment, they are grouped into two categories. Those (categorical) uncertainties that really impact the portfolio investment decisions are classified as global; whereas those that quantitatively describe the intrinsic imperfect knowledge of the categorical are considered local uncertainties. So, to effectively account for robustness, defined as the ability to perform well under unforeseen situations, and flexibility, defined as the ability to adapt cost-efficiently to different situations, modern tools are illustrated and implemented in a computationally tractable manner, resulting in promising planning tools under uncertainty.