Hygrothermal Degradation of Epoxy Electrical Insulating Material—Testing and Mathematical Modeling

• Published in: Polymers Vol. 16; no. 14; p. 2026
• Main Authors: Leffler, Jan, Kaska, Jan, Kadlec, Petr, Prosr, Pavel, Smidl, Vaclav, Trnka, Pavel
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 16.07.2024
• Subjects: Bayesian analysis; Degradation; Design of experiments; Design parameters; Dielectric strength; Dissipation factor; Electrical engineering; Epoxy resins; Failure; High temperature; Humidity; Insulation; Machine learning; Manufacturers; Mechanical analysis; Mechanical properties; Methods; Neural networks; Permittivity; Tensile strength; Bayesian experimental design; multi-factor stress; epoxy resin; degradation models; hygrothermal degradation; material testing
• ISSN: 2073-4360; 2073-4360
• DOI: 10.3390/polym16142026

The degradation of electrical insulating materials has been a subject of interest for decades as they are commonly applied in many fields of electrical engineering. Suitably modeling such a process is important since the known and well-described degradation process reveals the effect of ambient conditions, and this allows us to possibly estimate a material’s remaining useful life. However, not many studies are dealing with the effect of the hygrothermal degradation of impregnating mono-component epoxy resins in the context of electrical engineering. Therefore, this study deals with this issue and discusses both the dielectric response (based on the measurement of relative permittivity, dissipation factor, and dielectric strength) and the mechanical response (based on measurements of tensile strength and Shore D hardness) to a hygrothermal degradation experiment. In addition, the results of thermal analyses are presented for the evaluation of the pristine specimen manufacturing process and possible post-curing processes. Furthermore, this study presents several methodologies for modeling the degradation process, including a novel methodology in this area based on Bayesian experimental design. As an outcome, mechanical parameters are proven to be specific in terms of the actual condition of the material and the Bayesian enhanced degradation model seems to be superior to the conventional evaluation methods in this particular study.