The electrical power system is a critical infrastructure that provides a reliable energy supply for daily businesses. Designing resilient and reliable power grids is of paramount importance to prevent blackouts and their adverse economic impact on society.

Resiliency is fundamentally defined as the ability of a system to respond to high-impact disturbances with a low-probability of occurrence. Evaluating resiliency in power systems is usually done in three stages. The first phase is the disturbance progress. During the first phase, the resilience level deviates from its pre-disturbance level. This can be observed by analyzing different metrics in the network. Secondly, in the case of effective primary control actions, a new steady state operating condition is reached, which differs from the pre-disturbance operating condition. Finally, the system reaches the restorative stage. The recovery starts and the system returns to normal operation.

The assessment of resiliency is a combination of assessing all three previously mentioned stages (during-disturbance, post-disturbance, and restorative). Depending on the focus of the study, different technical aspects of a system are assessed. This thesis focuses on the assessment of the during-disturbance phase because, in future grids with lesser reserves available and limited control capabilities, the initial response of the system following a disturbance becomes more critical.

This thesis presents a basic qualitative study of the dynamic performance due to an active power imbalance during a disturbance on the AC and DC sides of hybrid power systems with an emphasis on an MTDC interconnected offshore-onshore system. The widely used RoCoF (Rate of Change of Frequency) is adopted as a performance metric to assess the active power-frequency response from the perspective of the AC side. In addition, a modified quantification of the Rate of Change of Voltage (RoCoV), which is usually applied in the design of protection schemes, is suggested as an attempt to better capture the response of the DC voltage.

The suitability of these metrics to properly reflect the resulting dynamics is analyzed by considering different disturbances, such as generator outages, line outages, converter outages, and faults like line-to-line and line-to-ground short-circuits, at either the AC or the DC sides. The different disturbances are executed using real-time digital simulations on the EMT model of the CIGRE BM1 DC-AC test system in RSCAD\textsuperscript{\textregistered} FX. The performance metrics are well able to capture the impact of different disturbances on the response of the system. However, the performance metrics are not able to capture oscillating responses.

A study on the parametric sensitivity of the control parameters in the outer control loop of the converters is executed to see the influence of these parameters on the dynamic response and whether the performance metrics are able to capture the influence. The parameters of the outer control loop determine the reference currents for the converter, and thus directly influence the output of the converters. The results show that, for a DC line-to-line short-circuit, the adjustment of the control parameters in the outer control loop has no influence on the response of the DC voltage. For each control setting, the DC voltage still immediately drops to 0. This is also reflected in the performance metrics.

Whereas, adjusting the control parameters in the outer control loop influences the DC voltage response when subjected to an AC 3-lines-to-ground short-circuit. The proportional gains of the controllers mainly influence the overshoot of the DC voltage and have a small influence on the settling time. This corresponds to the role of the proportional gain in PI control, to respond quickly to faults. On the other hand, the integral gain responds slower and integrates the error over time to eliminate the residual error. Therefore, the results show that the integral gain mainly has an impact on the settling time of the response, and almost none on the overshoot. This is not captured by the calculated performance metrics. The performance metrics capture the response after the initial overshoot and this results in an unfair comparison between performance metrics.