Development of a predictive simulation platform for high-frequency insulation cracking analysis in power electronics devices

The accelerated progress in power electronics technology, notably the pervasive implementation of Insulated Gate Bipolar Transistors in critical domains such as renewable energy generation and flexible HVDC transmission systems, has resulted in unparalleled demands on insulation systems. These systems are subject to extreme operational stresses that include high-frequency, high-amplitude voltage stresses generated by pulse width modulation techniques, coupled with significant Joule heating effects. This leads to the formation of severe electro-thermal multi-stress coupling conditions, which challenge the long-term reliability of insulating materials such as epoxy resin. In such conditions, microscopic defects can initiate an electrical treeing phenomenon, a destructive occurrence that can significantly reduce dielectric strength. Nevertheless, the microscopic evolution of electrical trees is abstract and complex, rendering it difficult for students to comprehend the fundamental physical mechanisms. The primary objective of the present research is to address the prevailing educational challenge through the development of an integrated teaching and research platform. This platform is designed to integrate experimental observation with simulation methodology. This platform aims to transform the abstract process of insulation degradation into a tangible, visible phenomenon, thus enabling a more profound understanding of failure mechanisms under high-frequency and high-temperature conditions.

A comprehensive “experiment-simulation” methodology was employed in this study. The experimental component incorporates a platform that is integrated with four specialized modules: a high-voltage pulse power supply, a precise temperature control system, an observation unit, and an imaging system. The configuration of the setup utilizes a needle-plane electrode configuration within a polytetrafluoroethylene (PTFE) mold to simulate electric field concentration. The utilization of a high-speed CCD camera in conjunction with a high-magnification stereomicroscope facilitates the capture of the real-time morphology of electrical trees in epoxy resin. In the domain of simulations, the research introduces a sophisticated multi-physics model grounded in the phase-field method. In contrast to conventional models that prioritize electric fields, this model incorporates five critical energy components: phase separation free energy, gradient energy, electrostatic energy, Joule heating energy, and strain energy. The model solves heat balance equations to account for temperature effects on material properties, such as dielectric loss. This allows for the prediction of early-stage tree growth under coupled electro-thermal fields.

The experimental results indicated the presence of distinct treeing behaviors in response to varying stress conditions. When subjected to high-frequency stress at ambient temperature (10 kHz, 25 ℃), trees exhibited accelerated growth, forming distinctive “ball-chain-like” carbonized channels as a result of substantial localized energy deposition. In contrast, when subjected to low-frequency (50 Hz) and high-temperature (90 ℃) conditions, a phenomenon of “growth stagnation” emerged. This phenomenon was characterized by the facilitation of bond breaking by heat, while the low frequency allowed for sufficient time for space charge accumulation to shield the electric field, thereby inhibiting its propagation. Under conditions of synergistic high frequency (10 kHz) and high temperature (90 ℃), the most severe degradation was observed, resulting in rapid, dense branching and the formation of large, bubble-like carbonized areas. The simulation model effectively predicted initial propagation paths that aligned with experimental observations. However, as these organisms developed, discrepancies became apparent: while actual trees demonstrated complex, stochastic three-dimensional branching patterns, simulations remained two-dimensional and lacked this complexity. This divergence underscores the limitations of idealized homogeneity assumptions in modeling complex physical-chemical processes.

The present study successfully validates an integrated platform that serves to bridge the gap between theoretical modeling and practical observation in the domain of high-voltage insulation education. The platform successfully visualizes the synergistic effects of frequency and temperature on electrical treeing. It reveals mechanisms ranging from rapid carbonization to charge-induced stagnation. While phase-field simulations offer valuable insights into early-stage driving forces, a comparison with experimental data underscores the complexity of material inhomogeneity in real-world scenarios. This “experiment-simulation” approach significantly enhances students’ comprehension of multi-physics coupling and cultivates essential engineering competencies for insulation system design and reliability assessment for modern power electronics.