Abstract:
The scarcity of the natural resources, environmental issues and rising population in the
world, demands for innovative concepts, such as microgrids to the modern power
system. Nowadays, microgrids are becoming very popular and the most appropriate
options to enrich the power system with renewable generation. In addition to that, rapid
growth in the DC nature of the loads within the power system is apparent due to the
popularity of power electronic devices and recent trends in electrified transportation
systems. Hence, researchers are introducing direct current microgrid concepts to the
power system, and it has become a highly emerging and trending research area at
present. DC microgrids can operate under two main operating modes: grid-connected
and islanded operation. The main difficulties of implementing the concept in this
concept are the lack of proper international standards, safety features, and protection
issues within the systems. Islanding detection is the most challenging and vital
requirement in microgrid protection to ensure the safety of the personnel and microgrid
equipment and to maintain a smooth and reliable operation of the DCMG. Islanding
detection is used to detect the disconnection of the DC microgrid from the utility and
switch to proper controls to serve critical loads in the power island.
This thesis presents a novel method of islanding detection for DC microgrids by using
Fast Fourier Transform based analysis of DC-link voltage. Further, testing was carried
out adopting a 10-kW low voltage DC microgrid with a single-phase bidirectional
inverter interface. In addition, a DC microgrid consisting of photovoltaic model with
maximum power point tracking, DC loads, AC loads and a battery module with stateof-
charge based multi-mode battery management system was modeled. All the
modeling and simulations were carried out considering several network configurations
and network conditions with the EMTDC/PSCAD v4.2 environment. Simulations
were evaluated according to the IEEE 1547-2018 standard. The probabilistic approach
was applied to show the robustness of the experimental results of the proposed method.
