A Parallel Hybrid Method for Fast Analysis of Power System Dynamics

Abstract

Computational simulations are currently extensively applied for analysis of power systems in order to ensure a secure and stable operation of the network. However, the actual trend in the power system operating environment shows several transformations in the grid structure as a result of increasing operation of large interconnected networks, growth in electricity demand, and the integration of renewable energy sources in the energy transition context. Such changes directly impose additional requirements to the stability analysis process, whereby the time-domain simulations widely used for dynamic stability studies are faced with an increase in computational burden due to the increasing complexity of the system under analysis. Nevertheless, the continuous changes in the system's operating point owing to variations in operation conditions shows a need for continuous analysis during network operation. This therefore necessitates advanced methods to cope with the introduced complexity in the analysis process. The present thesis describes a new parallel hybrid computational method for the fast analysis of power system dynamics in large networks in order to address the above mentioned challenges. As a first step, mathematical models of power system components are described for representing the dynamic behavior of the system. New models of renewable energy sources -- solar photovoltaic and wind power -- are described for the functional representation of grid integrated distributed generation in system stability analysis. A phasor time-domain computational method is then presented for analyzing network dynamics in the electromechanical transient phenomena. In this case, a method is described for the conventional balanced transients' analysis and extended to a new method that includes analysis of unbalanced transients. To address the complexity in the analysis of large networks, a parallelization approach is proposed for the time-domain computations. For this, a grid partitioning strategy is presented for dividing networks into simplified parallelizable subnetworks, which are applied in a parallel-in-space computation approach. In a further step to account for continuous analysis including all scenarios in the analysis process, a direct-method is described for fast assessment of system dynamic stability. The direct-method introduces the ability of fast identification of critical network contingencies that require detailed analysis. With this property, the direct-method is combined with the parallel time-domain simulation approach to develop a hybrid computational method for fast, detailed and continuous analysis of power system stability during network operation. The models and methods proposed in the present thesis are benchmarked against available open source and commercial software packages. As part of system integration, the methods are integrated into a modeling, simulation and visualization software framework which facilitates application of the methods in an interactive stability analysis environment. The results from the analysis in the integrated software framework show that the new methods provide an important contribution for setup of online power grid dynamic stability assessment in the smart grid context.