Fast Assessment of Dynamic Behavior Analysis with Evaluation of Minimum Synchronous Inertia to Improve Dynamic Security in Islanded Power Systems

Abstract

Over the last decades, renewable energy sources (RES) participation into the electricity supply mix has been constantly increasing not only in interconnected power systems but also in isolated power systems. This power supply transition seeks to accomplish renewable-based electricity generation targets and policies as well as the de-carbonisation of societies for a more sustainable energy future. Such transition should be achieved through investments and consequent substantial installation of wind and solar farms in power systems, not only because of their obvious environmental benefits but also because of their technological maturity and consequent steady cost declining.Despite renewable energy penetration growth in several power grids, there are some technical challenges to deal with when we are in presence of isolated power systems with variable renewable energy sources. Those technical issues are identical to the ones faced by larger and interconnected systems but they could intensify in these less robust type of systems. Therefore, issues like frequency control and spinning reserve management become even more important to guarantee acceptable levels of stability and security of the system. Moreover, the increasing participation of wind and photovoltaics in the generation mix, unlike conventional generators, leads to a significant reduction in the amount of synchronous inertia present in the system, which is essential to avoid a rapid rate of change of frequency (RoCoF) and large frequency deviations after a contingency. Thus, higher RoCoF and frequency deviations will be observed, which might trigger the protection devices resulting in a cascading outage and a blackout.This thesis presents a preventive control tool capable of evaluate power system stability and identify the minimum synchronous inertia required to maintain system stable for a certain operation scenario (characterized by its dispatch and demand) and, if necessary, to support the decision maker to perform a new power dispatch or consider the activation of synchronous condensers.For that purpose, a small and isolated power system with a significant participation of RES in the generation mix was considered. By performing a power dispatch, several operation scenarios were created and used as input in a MATLAB/SIMULINK model, which was used to study the dynamic response of the system when a disturbance occur. In this work two different disturbances were considered: active power output loss by the biggest thermal unit in the system and the loss of 50\% in both wind and PV active power output. Therefore, a dataset was created to train two different neural networks (one for each contingency) so that they could emulate the dynamic response of the system. By applying a sensitivity analysis by the neural networks it is possible to identify the minimum synchronous inertia required by the system to keep it secure and stable.