Resilience-oriented power system restoration considering complex load behaviors

• Main Author: Xie, Dunjian
• Other Authors: Xu Yan
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Engineering; Power system; Resilience; Stability; Uncertainty
• Online Access: https://hdl.handle.net/10356/174753
• Institution: Nanyang Technological University

Extreme natural disasters have caused significant power outages, resulting in massive economic and social losses. System resilience, defined as the ability of an entity to anticipate, resist, absorb, respond to, adapt to, and recover from a disturbance, has emerged as a critical need for modern power systems. Power system restoration is one of the most important techniques to improve system resilience. Recently, the development of distributed generation and information technologies provides the potential in achieving efficient and fast power system restoration, but there are still some challenges. First, the dynamic characteristics of the system components during restoration greatly differ from the normal conditions, which could have great impacts on the system's security and stability. More comprehensive modeling techniques for the dynamic components are urgently needed. Second, due to the weakness of the power system during restoration, the stability requirement when is much more critical compared with normal systems. How to address the stability requirement during restoration is an essential problem to be solved by a power system restoration technique. Third, because of the lack of sufficient historical data on emergency cases, the behaviors of uncertain resources are much more difficult to predict and estimate. The diverse uncertainties from the resources, such as loads and renewable energy generators, need to be fully considered and well addressed for a secure power system restoration plan. Fourth, since the power system restoration is a multi-stage process of sequential operation of a large number of components, it is either efficient or reliable to design power system restoration by separately considering a single stage or component. A comprehensive method that could coordinate available resources in multiple stages is essential. This thesis is intended to solve the above issues in the following aspects. • A comprehensive dynamic load model considering multi-phase cold load pickup effects is firstly proposed to represent the load dynamics during restoration. The proposed dynamic load model describes the load behaviors in the inrush phase and enduring phase, which could be conveniently integrated into the power system restoration model. • A trajectory-sensitivity-based method for dynamic frequency-constrained load restoration is developed to address the frequency stability issues when the disturbance of load pickup is introduced to the system. The frequency constraints can be linearized based on trajectory sensitivity analysis and solved iteratively. A model-reduction method is utilized to improve the speed of the time-domain simulation. • A transparent data-driven method for stability-constrained load restoration is proposed. All the stability constraints including frequency, voltage, and rotor angle stability are considered and linearized based on an interpretable machine-learning method, i.e., optimal decision tree. The proposed method could provide a transparent and thus easy-to-understand way to linearize stability constraints through data training. Based on the off-line training for the stability constraints, the proposed load restoration can achieve fast decision-making online. • A robust stability-constrained load restoration model is proposed considering the uncertain dynamic loads. The loads are modeled by a composite load with both static and dynamic components with uncertain compositions. The two-stage load restoration model is proposed, and the stability constraints of frequency and voltage are considered. The proposed method could achieve a robust load restoration scheme against any realization of uncertain load dynamics. • A coordinated power system restoration method is proposed considering the uncertainty of load pickup and renewable generator connection. The proposed method sequentially models and coordinates the stages of power supply black start, network reconstruction, and load restoration for distribution networks. The developed method contributes to a fast and efficient power system restoration strategy. All the proposed methods have been tested based on the IEEE benchmark system and could help to improve system resilience in case of potential outages resulting from large-scale contingency.