Lifetime Modelling of Electrical Machines using the Methodology of Design of Experiments

Abstract

In the coming years, electromobility will be confronted with increasing demands regarding the reliability of electrical machines. In this paper a modeling methodology is presented, which allows to estimate the reliability and lifetime of the insulation system of electrical machines. Different statistical and physical modeling methods are presented, which are transformed for the later multiple regression. The methodology of Design of Experiments (DoE) is used to describe the insulation system. Since the effort for the experimental design of the DoE varies strongly with the number of effects to be investigated and the statistical accuracy, different experimental designs are presented, which can be considered for different numbers of factors. Depending on the research question, a suitable experimental design can be selected. For the calculation of the lifetime, Miner’s rule is used in addition to the multiple regression, so that the percentage lifetime consumption due to a load spectrum can also be calculated. Introduction In the future, technologies such as electrified aircraft, trolley wire trucks or autonomous driving will be established in the field of electric mobility. Reliability, durability and safety are important criteria for the acceptance of new technologies in society. In addition to increased safety requirements, these technologies must also be able to cope with new, more challenging boundary conditions. Whereas bearing damage used to be the most frequent cause of failure of electrical machines, the increased requirements and new boundary conditions lead to a more varied error pattern [1], [2]. A deep understanding of the causes of failure and the relevant damage mechanisms is necessary to design electrical machines for these new applications. In addition to the design of durable machines that operate under increased environmental conditions, lifetime models, on the other hand, can help electrical machines achieve a minimum target lifetime for a given load. This offers potential to save resources and reduce costs. Lifetime models and reliability analyses are therefore becoming increasingly important in the design process of electrical machines. 1 Damage mechanisms The damage of electrical machines is caused by various mechanisms from different physical disciplines and there are various modelling approaches to model these damage processes. Since the use of wide bandgap semiconductors and the increasing electrical load, the cause of failure of electrical machines is increasingly based on faults and breakdowns of the electrical insulation system [2]. Because failures of the electrical insulation system will occure more frequently in the future, research concentrates on lifetime models of these insulation systems. Basically, with regard to the aging effects that damage the insulation system, a distinction can be made between constant stresses and transient stresses. The probability of a fault in the insulation system at constant stresses is proportional to the number of operating hours and at transient stresses proportional to the number of transient effects. Figure 1 shows as examples for constant stresses the ambient temperature and the voltage slope oval and for transient stresses partial discharges rectangular. Figure 1: Examples of different stresses and their interactions. A breakdown in the insulation system often occurs due to different combinations of the individual stresses or the interactions of these stresses. ARGESIM Report 59 (ISBN 978-3-901608-93-3), p 319-325, DOI: 10.11128/arep.59.a59045