Partial Discharge Phenomena in Electrical Machines for the More Electrical Aircraft. Part II: Impact of Reduced Pressures and Wide Bandgap Devices

Luca Lusuardi, Alberto Rumi, Andrea Cavallini, Davide Barater, Stefano NuzzoIEEE Access 2021 · DOI

Introduction

As the aviation industry transitions toward the "More Electrical Aircraft" (MEA), traditional hydraulic and pneumatic systems are being replaced by high-power electrical actuators. To achieve the necessary power density and efficiency, engineers are increasingly turning to Wide Bandgap (WBG) semiconductors, such as Silicon Carbide (SiC). These devices offer extremely high switching frequencies and fast voltage transitions ($dV/dt$).

However, these technological advancements introduce new reliability concerns. High-speed switching creates intense electrical stress on motor insulation. Furthermore, aircraft operate at low atmospheric pressures (typically 150–200 mbar at cruise), where gases break down more easily according to Paschen’s Law. This combination makes "Partial Discharges" (PD)—localized electrical breakdowns that gradually erode insulation—a significant threat to motor longevity. This research investigates whether the synergy of SiC switching and low-pressure environments significantly lowers the Partial Discharge Inception Voltage (PDIV).

Experimental Methodology

To simulate the critical contact points between motor windings, the researchers utilized "twisted pairs" of enameled copper wire as insulation models. These samples were tested in a specialized vacuum chamber designed to replicate flight conditions, with pressures ranging from sea level (1 bar) down to 5 mbar and temperatures up to 120°C.

Schematic of the experimental setup showing the vacuum cell, twisted pair EUT, and multiple detection systems including Optical (photon counter), UHF, and HFCT sensors.

Detection and Modeling

Detecting PD in SiC-driven systems is challenging because the inverter's high-speed switching generates substantial electromagnetic noise that can mask conventional electrical signals. The team evaluated three distinct detection methods:

  1. High-Frequency Current Transformers (HFCT): Standard electrical detection.
  2. Ultra-High Frequency (UHF) Antennas: Capturing electromagnetic emissions from the discharge.
  3. Optical Photon-Counting: A system that detects UV light emitted during discharge, which proved particularly effective at low pressures.

To predict PDIV across varying environments, the researchers developed an empirical quadratic model based on the gas number density ($n$), which accounts for the simultaneous effects of pressure and temperature.

Key Insights: The Impact of Switching Speed

A primary objective of the study was to determine if the incredibly fast rise times of SiC inverters (as low as 7 ns) inherently lower the PDIV. The results revealed a surprising conclusion: the inception of the discharge depends almost entirely on the peak voltage reached, rather than the speed of the pulse or the switching frequency.

Voltage waveforms for different rise times (7 ns, 15 ns, and 150 ns) showing the characteristic voltage overshoot associated with faster switching transitions.

The "Overshoot" Trap

While rise time does not change the fundamental physics of the discharge, it does influence the actual voltage seen by the motor. Faster rise times often lead to significant voltage "overshoot" due to impedance mismatches and cable reflections. If an engineer monitors only the DC bus voltage and ignores these transient peaks at the motor terminals, they may incorrectly conclude that the PDIV has decreased, when in fact the insulation is simply being subjected to higher-than-expected peak voltages.

Key Results and Discussion

  • Frequency and Rise Time Independence: PDIV remained constant across switching frequencies up to 100 kHz and rise times ranging from 7 ns to 150 ns.
  • Pressure Sensitivity: The PDIV follows a characteristic "U-shaped" curve relative to pressure, reaching its most vulnerable point (the minimum voltage required for discharge) between 50 and 70 mbar.
  • Optical Superiority at Altitude: As pressure drops, discharges transition from "streamer" to "Townsend" types. While this makes UHF detection less effective, optical detection becomes more sensitive because the thinner air absorbs less UV light.
  • Model Accuracy: The proposed empirical model (Equation 4) successfully predicted PDIV for various wire diameters and insulation thicknesses with an error margin of less than 10%.

Conclusion and Future Outlook

The study confirms that while SiC inverters and low-pressure environments present challenges, the PDIV can be reliably predicted using gas density models. The stability of the dielectric constant in the tested PAI/PEI insulation suggests that environmental density is the dominant factor in discharge inception.

Future work will explore whether even faster Gallium Nitride (GaN) devices might eventually reach a frequency threshold where "permittivity mismatch" between air and solid insulation begins to influence PDIV. Additionally, research is needed to understand how these high-speed pulses propagate through the complex, global winding structures of full-scale aircraft motors.

Summary updated

BibTeX
@article{lusuardi2021partial,
  author = {Luca Lusuardi and Alberto Rumi and Andrea Cavallini and Davide Barater and Stefano Nuzzo},
  title = {{Partial Discharge Phenomena in Electrical Machines for the More Electrical Aircraft. Part II: Impact of Reduced Pressures and Wide Bandgap Devices}},
  journal = {IEEE Access},
  year = {2021},
  doi = {10.1109/access.2021.3058089},
}
Figure