Enhancing the security of integrated electricity-gas systems : low-carbon, resilient, and distributed perspectives

Abstract

Natural gas has grown into one of the most critical energy sources in electricity generation and is predicted to remain a bellwether 30 years later in the U.S., making it progressively competitive. To empower efficient transmission of natural gas to gas-fired generators, mass integrated electricity and gas systems (IEGSs) are constructed, which, on the other hand, inevitably incurs security concerns. Particularly, the uncertainty stemming from the high penetration of renewable power generation in IEGSs challenges their operation security. In addition, IEGSs are more liable to suffer from extreme events due to both exogenous and endogenous factors, e.g., an increasing number of extreme weather events and the expanded system scale, foregrounding the significance of their infrastructure security. Moreover, unlike power systems, an IEGS usually consists of multiple agents, e.g., power system agents and gas system agents, and operates without a centralized coordinator. Given the increased emphasis on data security and privacy, conventional centralized operation paradigms may not apply to multi-agent IEGSs. This thesis focuses on the aforementioned security problems and, accordingly, conducts the following research works. First, in order to enhance the operation security of low-carbon IEGSs, this thesis proposes a tailored operation schedule for wind power penetrated unit commitment optimal energy flow (UC-OEF) problems using a two-stage sample robust optimization approach. The objective is to obtain a UC decision by minimizing the sum of the (first-stage) UC cost and the mean of the (second-stage) worst-case operation cost over multiple uncertainty sets. The affine relations between second-stage decision variables and random variables are explored to simplify the UC-OEF model. A duality-based solution method is developed to solve the simplified model. Second, in order to enhance the infrastructure security of IEGSs, i) this thesis proposes resilience enhancement strategies against sequential extreme weather events (SEWEs) using a data-based robust optimization (RO) approach. The unique property of SEWEs, sequentially endangering certain regions of an IEGS, is incorporated to reduce the model conservativeness; ii) this thesis develops resilience enhancement strategies against both deliberate human attacks and wind power uncertainties using a hybrid two-stage robust-stochastic optimization approach. The wind power uncertain-ties are captured by a Markov process-based stochastic optimization approach, and the destructive deliberate human attacks are modelled by an RO approach. Both models are solved by a nested column and constraint generation algorithm. Last but not least, in order to enhance the data security and privacy of multi-agent IEGSs, i) this thesis proposes an extended convex hull (ECH)-based distributed operation framework for OEF problems. The proposed model is convexified by the ECH, provided that each agent is considered as a block. The convexified model is solved by a Jacobi-Proximal ADMM algorithm; ii) this thesis develops a hidden convexity-based distributed framework for OEF problems. Each power/gas node is regarded as a block to empower a flexible distributed operation paradigm. The hidden convexity of the multi-block OEF model is explored to equivalently reformulate each nonconvex constraint as a semi-definite program. An ADMM algorithm is employed to address this model.