Abstract
Electric power systems present both challenges and solutions to the ambitious goals for reducing greenhouse gas emissions, particularly through the transition from fossil fuels to renewable energy sources. The digitalization of power systems enhances system state observability via increased monitoring, paving the way for more automated operational control. Traditional reliability management of the power system, based on the deterministic ‘N-1’ criterion, fails to utilize these new monitoring and control capabilities. However, new reliability management methods based on risk can capitalize on these advancements to ensure a reliable and affordable power supply. This thesis specifically examines short-term operational planning using security-constrained optimal power flow (SCOPF). SCOPF models the power system flow and includes remedial actions to maintain operational security limits. These remedial actions can be preventive (pre-contingency) or corrective (post-contingency). Comprehensive SCOPF formulations are complex and computationally intensive, especially for large power systems, due to the increased problem size and complexity when accounting for multiple contingencies. Research have often used the same contingencies as the ‘N-1’ criterion for probabilistic operational planning. However, during adverse weather conditions, the probability of multiple outages can be too high to ignore. These ‘N-k’ contingencies, which include many system-splitting scenarios, significantly increase the complexity of a SCOPF model. Therefore, there is a need for scalable probabilistic SCOPF models and solution algorithms to optimize socioeconomic cost benefits in operational planning. This thesis aims to explore socio-economic short-term operational planning for the safe and economical provision of electric power in large-scale systems, focusing on the Nordic power system. It addressed three main focus areas: 1. How to model optimal preventive and corrective actions in a large-scale power system for short-term operational planning? A power transfer distribution factor (PTDF)-based corrective SCOPF based on the DC power flow approximation was formulated and shown to be useful for large-scale systems. The optimization aims to maximize socio-economic welfare, and the model accounts for balancing markets and emergency limits in two post-contingency states. Multi-branch contingencies were modeled with post-contingency system splitting, assuming all islands continued to deliver power if a power balance was found. It was shown that the ‘N-1’ criterion may provide insufficient security under adverse weather conditions. The risk index used in the proposed C-SCOPF was the amount of load shed times its probability. The risk had a limited maximum load shed per contingency and restricted probability using a contingency set. The results indicate that the risk index effectively guides system operation within an acceptable risk level with some economic penalty for low probability events. 2. Which solution methodologies can be used to find optimal preventive and corrective actions in a large-scale power system for short-term operational planning? An iterative solution procedure was developed to solve the C-SCOPF by separating the pre- and post-contingency states. This methodology was successfully tested on systems with up to 10,000 buses, including a real model of the Norwegian power system. The choice of iterative solution strategy – whether to add one or more postcontingency constraints at a time – depends on the number of constraints that are binding at the optimum, which increases with large imbalances between areas. 3. How do probabilistic methods perform as a means toward better socioeconomic operational planning of power systems compared to the deterministic ‘N-1’ criterion? Probabilistic methods were found to offer lower costs and use more corrective actions under normal weather conditions compared to deterministic methods. A case study suggested that probabilistic methods could provide secure system operation at lower operational costs. Moreover, these methods need to carefully consider the distribution of socio-economicwelfare among stakeholders, as the distribution can change and be controlled by the method. The thesis concluded that probabilistic planning, supported by advancements in technology and methodologies, can improve operational planning.