Study of a highly constricted electrical arc for a resonance DC circuit interrupter

With the increased demand for Direct Current (DC) grids there is an increased need for DC circuit breakers. Due to the lack of an innate zero crossing, DC interruption is a complicated process. Many DC circuit breaker concepts have been studied in the literature. These range from purely electromechanical to purely solid state, along with hybrids of the two. One of the promising concepts for DC breakers is the constricting breaker. Such a breaker constricts the arc and changes its characteristics to enable current interruption. In this thesis, the viability of such a breaker for a passive resonance DC breaker is investigated. The purpose of this thesis is twofold: Firstly, it aims to find a method of identifying the arc characteristics of the constricting breaker with black box models. Secondly, it investigates the viability of the constricting breaker for the passive resonance topology. To characterize the arc, multiple black box models are applied and tested. The models are evaluated by simulating a testing setup and comparing simulated results to measurements. The models are adjusted using numerical least squares optimization, identifying the strengths and weaknesses of each model to conclude the best fit. The best fit was decided to be a semi-discrete function combining the Cassie and Mayr arc models. The arc model is then applied to the passive resonance topology to investigate the breakers viability. When applied to the passive resonance topology, the constricting breaker shows good promise. As expected, the breaker is able to induce resonance into the topology which in turn forces an earlier arc extinction. For the case with low resonance inductance, the derived arc model shows good conformity with experimental results. However, it is also shown that the model is not responsive to an increase in resonance inductance whereas the experimental results show a clear change. This points to a shortcoming in the model, which likely stems from the limitations of the chosen black box model.