The full text of this item is not available at this time because the student has placed this item under an embargo for a period of time. The Libraries are not authorized to provide a copy of this work during the embargo period, even for Texas A&M users with NetID.
Data-Driven Hazard and Disaster Mitigation in Power System Abnormal Conditions
Abstract
This work is motivated by the imminent need to strengthen the power grid against various disturbances under increasingly complicated operating conditions introduced by various emerg-ing factors that were not present in conventional power system design and operation principles: climate change brings more frequent extreme weather events that threatens the physical integrity of grid infrastructures; deepening penetration of renewable energy adds significant amount of un-certainty into planning and real-time operation; Volatility in the global fuel production, trade and transport disturbs the electricity market stability and thus impose additional constraints on physical operation; Demand-side technologies including electric vehicle, distributed generation, battery storage and demand response fundamentally transformed the characteristic of power flow patterns in power distribution networks during both regular and abnormal status. In this dissertation, we utilize a combination of data-driven approach and domain knowledge on mitigating power system disturbances of different scales and impacts: very-large-scale disastrous event mitigation and cascading failure prevention after small hazards through a reinforcement learning powered protection design.
A recent and representative example of extreme natural disasters, the Texas 2021 winter storm, have severely disrupted the basic functions of the Texas power system which has endured a long-lasting state-wide power outage due to a variety of problems directly and indirectly caused by the freezing temperature. Fuel shortage, snow and icing caused outage or significant de-rating for all types of generator units across the state; historic low temperature have induced record-breaking peak demand; non-ideal outage management and high market prices have aggravated the energy shortage problem even months after winter is over. In our study, we first construct an accurate open-source synthetic model using various publicly-available data, and re-produce the outage timeline through iterative simulation. Then, we propose and perform quantitative analysis of potential corrective measures and design scenarios using our model, aiming to mitigate the damage in future similar extreme events. Additionally, we focus on the macroscopic impact of demand flexibility in reducing outage damage through optimal allocation of load shedding among affected customers. Data manipulation, fitting and analysis techniques are used throughout this work to produce the model and dataset used in simulation.
In contrast to extreme events, minor and routine hazards in relatively low capacity distribution networks can also evolve into undesirable blackouts if not properly cleared by protection equipment. The presence and increasing penetration of various grid-edge technologies begins to pose a significant challenge on existing conventional protection mechanisms, which has become a bottleneck for utility companies to further promote the low-carbon electrification process. To overcome this roadblock, we develop and present a robust reinforcement-learning based protective relay control design that can remain reliable under the presence of high-level distributed renewable generation. This proposed design is implemented and validated through comprehensive simula-tion under realistic setting, in distribution networks with vastly different capacities, configurations and complexities. Generalizing the experiment implementation effort, a ML-friendly simulation platform is built to effectively produce large amount of simulated data and interface with various data-driven decision making frameworks.
Citation
Wu, Dongqi (2022). Data-Driven Hazard and Disaster Mitigation in Power System Abnormal Conditions. Doctoral dissertation, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /197922.