Advancing Resilience in Clusters of Grid-Forming and Grid-Following Inverters

This dissertation focusses on developing a hybrid multi-time-scale semi supervised distributed control architecture for a network made up of several clusters of grid forming (GFMI) and grid following inverters (GFLI) connected in a modern active power distribution network. The considered modern power distribution network is made up of several clusters of grid forming and grid following inverters. In such a heavily inverter-dominated power system, ensuring reliable and resilient operation of all these inverter-interfaced generation units of limited capacities, is of paramount importance and severity. The system's voltage and frequency must be maintained within permissible limits, thus ensuring an active power balance between demand and supply while providing uninterrupted power to system loads. This dissertation proposes a control architecture that achieves the resilient operation of inverters subject to steep disturbances originating in the source as well as load side. A self-reconfiguring coordinated mode selection mechanism has been proposed that runs a consensus among all interfaced inverter agents in a cluster. The proposed mechanism elects new (or additional) GFMI source, to support the system’s V-f whenever additional active-reactive power is demanded. It ensures that sufficient GFMI support is available in the network to serve the system loads. Next, a cluster coordinator module has been implemented to regulate the power operation set-points of GFMIs and GFLIs within the cluster. The cluster coordinator aims at optimally dispatching the power set-points to support the system frequency in response to steep disturbances. Moreover, a machine learning-based artificial neural network approach has also been investigated to further enhance the system performance and offer fast frequency restoration. A synchronization mechanism has also been discussed to enable speedy disconnection and interconnection of inverter clusters when commanded by upper layer controllers or in response to grid disturbance events. Finally, a supervisory coordinator has been proposed to enable seamless synchronization and dynamic power sharing among the interconnected inverters. The power optimizers deployed in the cluster coordinator and the supervisory coordinator command the GFMI/GFLI power set-points to attain various control objectives at different time scales. A detailed parametric analysis comparing the performance of the proposed multi-time scale semi-supervised distributed (MT-SSD) control approach with other state-of-the art control approaches is also performed. The comparative analysis results demonstrate an improvement in the overall resiliency of the IBR cluster achieved by implementing the proposed MT-SSD control approach. Next, the implementation feasibility of the discussed MT-SSD control scheme on real-world microgrid systems and the associated communication infrastructure requirements have also been reviewed. Finally, the dissertation concludes by summarizing the key achievements and discussions pertaining to potential future research directions.