Source-linked AI summary
Accelerated Insulation Aging Due to Fast, Repetitive Voltages: A Review Identifying Challenges and Future Research Needs
Mona Ghassemi
TL;DR
Accelerated aging of insulation systems exposed to fast, repetitive voltage pulses from WBG-conversion systems is a significant barrier to realizing and benefiting from high-voltage, high-power-density conversion systems. The paper reviews insulation degradation and identifies technical gaps and future research needs, including realistic waveform modeling for apparatus insulation. The review concludes that existing tests cover only a few kHz and a few tens of kV/μs and use simple geometries, leaving results insufficient for complex realistic situations.
Problem
Accelerated aging of insulation systems exposed to fast, repetitive voltage pulses from WBG-conversion systems is a significant barrier to realizing and benefiting from high-voltage, high-power-density conversion systems.
Source: §1 · p. 1Method
The paper reviews insulation degradation and identifies technical gaps and future research needs, including realistic waveform modeling for apparatus insulation.
Source: §1 · p. 2Results
The review concludes that existing tests cover only a few kHz and a few tens of kV/μs and use simple geometries, leaving results insufficient for complex realistic situations.
Source: §5 · p. 8Takeaways & Limitations
The paper provides a framework for future research on insulation materials and systems exposed to fast, repetitive WBG-converter voltage pulses.
Source: §1 · p. 2Water-tree models are valid only up to 30 kHz, and the cited experiments used sinusoidal rather than square waveforms.
Source: §2.2 · p. 3
Abstract
from arXiv · showhide
Although the adverse effects of using power electronic conversion on the insulation systems used in different apparatuses have been investigated, they are limited to low slew rates and repetitions. These results cannot be used for next-generation wide bandgap (WBG) based conversion systems targeted to be fast (with a slew rate up to 100 kV/us) and operate at a high switching frequency up to 500 kHz. Frequency and slew rate are two of the most important factors of a voltage pulse, influencing the level of degradation of the insulation systems that are exposed to such voltage pulses. The paper reviews challenges concerning insulation degradation when benefitting from WBG-based conversion systems with the mentioned slew rate and switching frequency values and identifies technical gaps and future research needs. The paper provides a framework for future research in dielectrics and electrical insulation design for systems under fast, repetitive voltage pluses originated by WBG-based conversion systems.
1 INTRODUCTION
Fast, repetitive WBG-converter pulses pose an insulation-aging barrier, but available evidence and models do not cover target slew rates, frequencies, or realistic apparatus geometries. The review identifies research gaps across solid insulation and gels and proposes future research directions.
1 INTRODUCTION
- Accelerated insulation aging from fast, repetitive WBG-converter pulses is a major barrier to high-voltage, high-power-density conversion; cited conditions are 50 kV/μs and 100 kHz, and experiments remain below PEBB 1000 targets.The review identifies technical gaps and offers a framework for future insulation-material and system research.Source: §1 · pp. 1–2
- Realistic insulation waveforms differ from converter-terminal pulses because of impedance mismatch and uneven voltage distribution, requiring apparatus EMT models spanning frequencies up to a few or few ten MHz.The review also calls for physically based models validated experimentally to simulate realistic geometries and multi-stress conditions.Source: §1 · p. 2
2.1 INSULATION DEGRADATION DUE TO PD IN SOLID INSULATION
- Under repetitive surges, shorter rise times are associated with larger PD magnitudes and shorter lifetime, while lifetime also decreases as frequency increases.Prior tests were limited to 70 kV/μs and 5 kHz, with high-slew tests using only one peak-to-peak voltage value in the cited examples.Source: §2.1 · pp. 2–3, Figure 1 · p. 2
2.2 INSULATION DEGRADATION DUE TO WATER TREEING IN SOLID INSULATION
- Water-tree growth depends on voltage zero-crossings, but prior tests used sinusoidal rather than square waveforms; reported log L versus log N behavior deviates from linearity between 30 and 447 kHz.Water-tree models are valid up to 30 kHz, and the review identifies the waveform mismatch as a research gap.Source: §2.2 · p. 3, Figure 2 · p. 3
2.3 INSULATION DEGRADATION IN GELS
- Silicone gel encapsulants in WBG modules are prone to degradation, yet the review reports little basic research on failure mechanisms under fast, repetitive pulses.Reported PD-origin hypotheses in modules are inconsistent, and optical detection in transparent gels has been demonstrated with sensitivity below 1 pC.Source: §2.3 · p. 3
3 MULTIPHYSICS MODELING
The section argues that insulation design needs physically based multiphysics modeling that accounts for pre-breakdown mechanisms and realistic multistress conditions. It identifies limitations in experimental and computational approaches, plus unresolved cases for solid insulation with gas voids and gels.
3 MULTIPHYSICS MODELING
- Experiments often use uniform or nonuniform field gaps instead of realistic geometries, describe results qualitatively, and may lead to unreliable conclusions and insulation overdesign.Numerical modeling has also often been limited to single-physics electric-field calculations, without modeling underlying pre-breakdown phenomena.Source: §3 · p. 4
- A physically based “theory” approach considers all physical mechanisms behind pre-breakdown phenomena and supports optimal insulation designs under multistress conditions.The stated stresses are electrical, thermal, mechanical, and environmental.Source: §3 · p. 4
- Multiphysics modeling can support optimal designs, but physically based approaches involve complex coupled equations, numerical challenges, and substantial computing resources.The paper contrasts this approach with time-consuming, expensive tests of materials and fillers that may fail to represent realistic multistress conditions.Source: §3 · p. 4
- For solid insulation with gas voids, the equations and associated parameters remain unknown; modeling gels is described as more difficult than modeling solid insulation.The paper identifies this as an unresolved boundary for physically based breakdown modeling.Source: §3 · p. 4
4 HIGH-FREQUENCY EMT MODELS
Steep-fronted pulses produce apparatus-specific voltage distributions and insulation stresses. The section reviews modeling needs and research gaps for WBG-era operation, including winding bandwidth, cable interactions, thermal effects, and solid-state transformer insulation.
4 HIGH-FREQUENCY EMT MODELS
- Reflections at cable interfaces can create motor-terminal overshoot; a 50 ns rise-time pulse reaches a 2× overshoot at 4 m of cable.Steep-fronted surges can also create transient voltage distributions and hot spots in stator windings, stress-grading systems, and cable terminations.Source: §4 · p. 4
4.1 HIGH-FREQUENCY EMT MODELS FOR MACHINE STATOR WINDING
- Winding models must extend toward GHz frequencies for WBG surges, while high-order models face passivity-enforcement challenges and existing loss models reach only a few MHz.Earlier case studies were limited to 68 kV/μs and 20 kHz; high-frequency modeling also requires frequency-dependent winding losses and cable/converter integration.Source: §4.1 · p. 5
- 43 p.u. overvoltages were simulated for a 20 m cable, and an 11 kV/230 V transformer with a 27 m cable experienced 24 p.u. on its open low-voltage side.These cases illustrate cable-transformer resonance concerns; the section calls for a wideband converter-cable-motor model and expanded study zones for short-cable systems.Source: §4.1 · pp. 5–6
4.3 THERMAL MODELING OF CABLE TERMINATIONS
- Cable-termination thermal models agree with tests, but experiments and models cover only a few kHz; projected temperatures at 500 kHz depend on an assumed trend.Measured temperature rise from 2 to 4 kHz is nonlinear and depends on voltage amplitude; the 1200°C estimate for 12 kV assumes that trend continues to 500 kHz.Source: §4.3 · p. 7
4.4 INSULATION MODELING OF SOLID-STATE TRANSFORMERS
- In an SST study, transformer insulation was the most stressed part, and a two-level full-bridge topology reduced maximum medium-frequency RMS electric field by 35% versus a three-level NPC bridge-leg half-bridge.The SST used 8 kHz for the DC-DC stages.Source: §4.4 · p. 7
- SST dielectric losses depend strongly on switching frequency, while voltage amplitude and material parameters also matter; duty cycle and modulation index have minor impact.The cited loss expression treats rise-time dependence as logarithmic in reciprocal rise time; the authors note approximations and a need for materials suited to higher temperatures and frequencies.Source: §4.4 · p. 7
- Resistive shielding can reduce capacitive disturbances but cannot suppress conducted EMI, leaving shielding design unresolved across transformer types and DC, medium-frequency, and high-frequency cases.A 21.2 kV/cm peak surface field was calculated for an MV, medium-frequency transformer in an SST demonstrator.Source: §4.4 · pp. 7–8
5 CONCLUSIONS
The conclusion identifies accelerated insulation aging under fast, repetitive WBG-converter voltage pulses as a major barrier. It calls for more representative experiments, multiphysics modeling, and high-frequency EMT models to guide insulation design.
5 CONCLUSIONS
- Accelerated aging under pulses reaching 100 kV/μs and 500 kHz is identified as a major barrier to benefiting from next-generation WBG conversion systems.Frequency and slew rate influence insulation degradation.Source: §5 · p. 8
- Reviewed experiments reach only a few kHz and a few tens of kV/μs, and simple-electrode results do not transfer to complex realistic situations.The conclusion identifies these limits as gaps in evidence for next-generation systems.Source: §5 · p. 8
- Multiphysics modeling is proposed to explain pre-breakdown mechanisms and support electrical insulation design.Source: §5 · p. 8
- Very high-frequency EMT models are needed to calculate voltage and electric-field distributions across power-system components under fast, repetitive pulses.The paper presents this research agenda as a framework for future dielectric and insulation design.Source: §5 · p. 8