UC Berkeley

Unlike conventional generators, the output of a renewable generator is only partially controlable and unpredictable. Consequently, its output level exhibits wide variability. The increasing amount of such kind of renewable resources in the generation mix of the electric power infrastructure poses new challenges to system operations. Due to the limitations of storage in current power systems, demand and supply of electricity must be matched instantaneously. Moreover, the flow of power is governed by Kirchhoff's laws and is prone to congestion. In order to serve load reliably in power systems with large-scale renewable integration, extreme ramping requirements on conventional generation resources will be imposed. The system operators could mitigate those requirements by deploying more reserves or increased amount of flexible generation resources which could undermine both the economic and environmental objectives of deploying renewable resources. Instead of utilizing the flexibility provided by conventional generation, we will focus on exploiting flexibility from the transmission system to mitigate the uncertainty and variability of renewable generation for power systems day-ahead scheduling.

In this dissertation, we focus on harvesting flexibility from the transmission system to mitigate the uncertainty of renewable generation. There are two aspects of the such flexibility. First, we allow actively controlling the topology of the transmission network by switching on or off transmission lines as a recourse option. Second, the transmission lines can temporally adopt higher ratings to avoid the underutilization of the line capacity caused by the conventional conservativeness in line rating calculations. We adopt the two-stage stochastic unit commitment model for the power system day-ahead scheduling problem. Both transmission switching decisions and flexible line rating decisions are modeled as binary decisions in the second stage when renewable generation is realized. A scheme that decomposes an interconnected commercial system into zones is proposed to solve this large-scale stochastic mixed integer programming problem within a reasonable amount of time. In the sub-problem for each zone, conventional generators are separated into a set of slow-ramping generators and a set of fast-ramping generators. The commitment decisions of slow generators are first-stage decisions which are made before the realization of renewable generation. The commitment decisions of fast generators and the dispatch decisions are second-stage decisions. Topology control decisions through switching on/off transmission lines and flexible line rating decisions are modeled as recourse actions. Such recourse actions provide the possibility of actively control the transmission system in response to the variability and the uncertainty of renewable generations. Comparing with deploying more flexible conventional generators which are expensive, flexible transmission network recourse incurs no additional cost other than the cost associated with the abrasions of breakers.

We provide demonstration studies based on the IEEE 118 system and a network representing the Central European system where there are over 650 conventional generators, 679 buses, and around 1000 transmission lines.We observe that the operating cost can be reduced by over 10% for the IEEE 118 system with purely topology control recourse. The cost reduction for the Central European system is over 3%, and substantial reduction is observed for some control areas with topology control recourse. The operating cost can be reduced if flexible line rating recourse is included.