Search

Modular multilevel converter (MMC) stands out among converter topologies for medium/high-power applications. With salient features like modularity and scalability, MMC can meet the requirement of theoretically any voltage level with higher efficiency and superior harmonic performance, using converter components with reduced ratings. Unlike conventional converters, MMC has a higher switching frequency due to its higher number of switching components. While MMC has been mainly considered for transmission-level applications such as HVDC, it still needs improvements for medium-voltage systems and grid integration of distributed energy resources. The contribution of this thesis is twofold. First, modulation techniques are designed to reduce switching frequency through novel advanced submodule sorting algorithms, which provide a level of controllability over two conflicting constraints – switching frequency reduction and capacitor voltage deviation – for HVDC applications. Second, a topology design based on full-bridge submodules is proposed to use MMC in grid integration of solar photovoltaic systems.

Voltage Collapse is the system failure to obtain acceptable voltage levels in significant part of the power system, and it is often due to system failure to satisfy reactive power demand. Voltage Collapse can lead to blackout like the one occurred in 2003 in North America. Methods for on-line voltage stability monitoring were established, and indices to quantify it were proposed. However, estimations of voltage collapse point based on these indices are often inaccurate or time consuming. A well-established method of voltage collapse point estimation is the Continuation Power Flow (CPF). CPF is considered accurate but, it is very computationally expensive for large systems. This work aims to speed up the predictor-corrector process by using a VSI called P-index. An initial prediction is made, corrected using a continuation technique, and then updated after correction. The results are relatively accurate and it makes a significant improvement to the CPF computational time.

Online voltage stability assessment of power system is important to ensure reliability and security. Over the years, many voltage monitoring indices have been documented in the literature. These indices are designed to quantify proximity to voltage collapse. However, they all have inaccuracies when predicting the voltage collapse point. A recent research project resulted in a new robust voltage stability monitoring tool called the P-index. A method that utilizes linear approximation of Voltage-Load relationship was introduced alongside the P-index, providing enhanced precision compared to most recognized indices. However, this method generates a high level of inaccuracy because the actual Voltage-Load behavior in general is non-linear, but rather more complex. This study proposed two new models that take into account the nonlinearity of Voltage-Load relationship, providing an enhanced representation of its complexity. Moreover, these models have a higher accuracy in predicting voltage collapse points than the linear method.

When an Out of Step (OOS) event occurs between two connected areas in the power system, the two areas should be separated before the OOS results in generation loss, equipment damage, and eventually a total black-out. This separation is achieved via the OOS function that implemented at specific transmission lines. One of the commonly used methods to achieve the OOS tripping functionality is the Dual-Blinder Scheme, which compare the locally measured impedance to a set value. In this research, a Phasor-Based OOS tripping function, based on the derivatives of the voltage phase angle difference between the two areas, is evaluated against the Dual-Blinder method using a real-time digital simulator, a protective relay, and a Synchrophasor Vector Processor (SVP). The evaluation shows that the Phasor-Based function predicted the OOS before the Dual-Blinder OOS function when the power system was having damping problems. In other cases, the Dual-Blinder OOS function tripped faster in most of the considered cases. Finally, this research suggests some improvements for implementing the Phasor-Based OOS function in the future.