@phdthesis{fan-20-aa-powsec,
  title = {Improved Methodologies for Security of Electricity Supply of Future Power System},
  author = {Fang, Duo},
  school = {University of Edinburgh},
  year = 2020,
  pages = {218},
  month = nov,
  doi = {10.7488/era/929},
  note = {Supervisors Djokic, Sasa and Harrison, Gareth},
  comment = {Combines AA with probabilistic info to compute power and thermal loads in distribution grids},
  abstract = {[...] demand-responsive controls and technologies [...] real-time thermal ratings for system components [...] The first contribution of this thesis is the development of advanced computational tools to strengthen the decision-making capabilities of system operators and ensure secure and economic operation under high uncertainty levels. It initially evaluates the hosting capacities for wind-based generation in a distribution network subject to operational security limits. In order to analyse the impacts of variations and uncertainties in the wind-based generation, loads and dynamic thermal ratings of network components, both deterministic and probabilistic approaches are applied for hosting capacity assessment at each bus, denoted as "locational hosting capacity", which is of interest to distributed generation (DG) developers. Afterwards, the locational hosting capacities are used to determine the hosting capacity of the whole network, denoted as "network hosting capacity", which is of primary interest to system operators. As the available hosting capacities change after the connection of any DG units, a sensitivity analysis is implemented to calculate the variations of the remaining hosting capacity for any number of DG units connected at arbitrary network buses. The second contribution of this thesis is a novel optimisation model for the active management of networks with a high amount of wind-based generation and utilisation of dynamic thermal ratings, which employs both probabilistic analysis and interval/affine arithmetic for a comprehensive evaluation of related uncertainties. Affine arithmetic is applied to deal with interval information, where the obtained interval solutions cover the full range of possible optimal solutions, with all realisations of uncertain variables. However, the interval solutions overlook the probabilistic characteristics of uncertainties, e.g. a likely very low probabilities around the edges of intervals. In order to consider realistic probability distribution information and to reduce overestimation errors, the affine arithmetic approach is combined with probabilistic (Monte Carlo) based analysis, to identify the suitable ranges of uncertainties for optimal balancing of risks and costs. Finally, this thesis proposes a general multi-stage framework for efficient management of post-contingency congestions and constraint violations. This part of the work uses developed thermal models of overhead lines and transformers to calculate the maximum lead time for system operators to resolve constraint violations caused by post-fault contingency events. The maximum lead time is integrated into the framework as the additional constraint, to support the selection of the most effective corrective actions. The framework has three stages, in which the optimal settings for volt-var controls, generation re-dispatch and load shedding are determined sequentially, considering their response times. The proposed framework is capable of mitigating severe constraint violations while preventing overheating and overloading conditions during the congestion management process. In addition, the proposed framework also considers the costs of congestion management actions so that the effective corrective actions can be selected and evaluated both technically and economically.}
}