Vermogenelektronische Betrouwbaarheid van Meerpoortsomvormers: Rekening Houdend met het Ontwerp en de Werking in Hernieuwbare Energietoepassingen

Electrification is put forward as an important strategy to reduce greenhouse gas emissions. Hence, both energy sources and sinks need to be electrified as much as possible. All these new applications often have different voltage and power levels, requiring power converters at the interconnections. These converters should be cost effective, efficient, and reliable to make large scale adoption possible. Currently, mostly standardised converters are used which have a single input and output. When more than two terminals need to be combined, multiple converters are interconnected to a common bus. This provides an expandable solution which is easy to design. However, other solutions that provide even better characteristics might exist. Instead of using one converter per port, components could be shared between ports, leading to a final solution with less components. Next to reducing the cost, this might also improve the efficiency and reliability. Unfortunately, designing these multiport systems requires a lot of skill and creativity. The solution is also application specific, leading to the need for a redesign when ports need to be added. This makes the use of these converters impractical in most applications. In order to solve this, the converter design should be automated. This work focuses on the first step, deriving the circuit topology. The proposed methodology makes use of the limited amount of valid connections that can be made between power electronic components. At different abstraction levels, the possibilities are explored and invalid configurations are eliminated, transferring only potentially viable solutions to the next round. Finally, the voltage relations and current flows in the remaining candidate converters are evaluated, allowing to identify the final valid solutions. The methodology is tested on a PV-battery use case and valid topologies are found, using less components than the reference topology with individual converters. The reliability improved as well due to the reduced power fluctuation in individual components caused by combining different port power flows. Less fluctuations also allow for the selection of smaller heat sinks, however, it could be decided to keep the heat sinks larger in order to retain a good efficiency. Although the proposed methodology already allows to construct converters with an increased reliability, it remains an important goal to improve the reliability further. Failures in renewable energy applications do not only increase the overall maintenance cost, but also lead to costs linked to the loss of energy produced. Next to selecting reliable topologies during design, it is also possible to increase the reliability in the operational phase, by eliminating a strongly degraded component before catastrophic failure occurs. In order to make these decisions, the condition of the converter components needs to be known. To this end, damage sensitive parameters can be measured, linking an electrical parameter to the amount of damage present. Establishing this relation is difficult in applications with a highly varying mission profile, as these parameters are often also dependent on temperature. Hence, in this work a dual approach is proposed, estimating both the temperature and degradation level using two independent measurements. A PV-battery application is used to demonstrate that it is possible to eliminate a component and form new circuits using the topology derivation methodology. The resulting back-up converter potentially has reduced capabilities, however, operation can be resumed.