Master's thesis and Internship : Adaptive Under-Frequency Load Shedding Based on RoCoF and System Inertia for Transmission Grid Defence Plans

Abstract

The integration of inverter-based resources (IBRs) in modern power systems has reduced the amount of rotational inertia available in the grid. As a consequence, frequency deviations following severe disturbances may become faster and more difficult to control. This thesis investigates an adaptive under-frequency load shedding (UFLS) strategy based on the equivalent system inertia and the rate of change of frequency (RoCoF), with the objective of adapting the shedding decision to the dynamic conditions of low-inertia power systems. The proposed method uses offline dynamic simulations to determine the minimum load shedding percentage required to keep the frequency nadir, defined as the minimum frequency reached during the transient response following a disturbance, above a predefined security threshold. These results are then converted into a practical lookup matrix using the equivalent system inertia and the measured RoCoF as input variables. During online operation, the UFLS controller can directly select the required shedding level from this precomputed matrix. Compared with conventional UFLS schemes, which usually rely on fixed frequency thresholds and predefined load shedding stages, the proposed approach introduces an additional level of adaptivity. The RoCoF is used as an indicator of the severity of the active-power imbalance, while the equivalent system inertia represents the dynamic condition of the system and its ability to withstand frequency deviations. By combining these two quantities, the adaptive matrix can distinguish between operating conditions that may have similar frequency values but require different shedding actions. This makes it possible to adjust the amount of disconnected load more selectively than in conventional frequency-only schemes. The method is assessed through Power System Simulator for Engineering (PSS/E) dynamic simulations on the PSS/E 23-bus SAVNW system and the IEEE 39-bus New England system, using 126 and 330 disturbance scenarios, respectively. The results show the potential of the proposed approach to adapt the amount of shed load to different inertia and disturbance conditions. However, they also highlight that its practical performance depends on fast RoCoF estimation, reliable inertia assessment, and low-latency communication and actuation.