The research, published in Frontiers in Electronics by a team from China Southern Power Grid and its Electric Power Research Institute, tackles a blind spot in transformer diagnostics. Conventional phase-resolved partial discharge (PRPD) analysis rests on a quiet assumption: that discharge behavior is governed solely by the power-frequency electric field. In a vibrating transformer, that assumption breaks down. Mechanical excitation can shuffle gas bubbles through the insulating oil, drag conductive particles across the gap, and force electrodes into periodic relative motion, modulating the local electric field in ways no purely electrical model predicts. To capture these coupled dynamics, the team built an electro-mechanical test platform in which high voltage and controlled vibration could be applied simultaneously to realistic defect models submerged in transformer oil.

Two defect geometries served as the experimental protagonists. The first was a needle–plate configuration—a sharp needle electrode with a 0.2 mm tip radius hovering 1 mm above a grounded copper plate covered by a 1 mm pressboard—mimicking tip-dominated discharges caused by burrs, protrusions, or conductive contamination. The second was a floating electrode pair: two facing needles, one energized and one electrically isolated, simulating suspended conductors or poorly fixed leads. Both models sat in a cylindrical oil-filled tank bolted to a vibration exciter, so that when sinusoidal shaking was applied, only the grounded electrode moved while the high-voltage electrode stayed fixed—reproducing the relative motion real insulation defects experience in service. Vibration was swept across 100, 200, and 300 Hz at acceleration levels from 0.25 g to 1 g, matching the dominant vibration signature of operating transformers.

The headline result concerns the floating electrode defect, and it is striking. Under 300 Hz vibration at just 0.25 g, the partial discharge inception voltage (PDIV) dropped by up to 73.8 percent—meaning the insulation began discharging at barely a quarter of its normal threshold. At the same time, the average discharge magnitude fell by as much as 88.9 percent under 300 Hz and 1 g excitation, while the discharge repetition rate climbed steeply, in some conditions by more than an order of magnitude. In other words, vibration makes floating defects fire earlier, more often, and with smaller individual pulses—a combination that could easily be misread by conventional monitoring as a benign, low-energy pattern when it actually signals dangerously intensified discharge activity.

The key to understanding this behavior is the new analytical tool the team introduced: the Vibration Phase-Resolved Partial Discharge (V-PRPD) pattern. Instead of mapping discharge pulses only against the electrical cycle, V-PRPD plots each pulse against the phase of the mechanical vibration waveform itself. For the floating defect under 100 Hz, 1 g excitation, ultrasonic discharge pulses clustered tightly around the vibration peaks—dense near a vibration phase of 90°, where the needle tips swing closest together and electric field distortion peaks, and sparse near 270°, where the gap is widest. The discharges were literally dancing to the rhythm of the machine’s mechanical motion.

The physics behind this sensitivity lies in the floating electrode’s freedom to move. Because the floating needle is not rigidly constrained, vibration-induced displacement directly compresses and relaxes the 1 mm oil gap, distorting an already severe local field. The team’s analysis shows that under maximum vibration the electrode displacement reaches roughly 4 μm—only about 0.4 percent of the static gap—yet this tiny geometric change is enough to slash the voltage needed to trigger discharge. The reduced gap lowers the effective breakdown distance, while the periodic field fluctuation triggers discharges prematurely, before the field can build to levels that would produce large, energetic pulses. The result is a swarm of frequent, low-magnitude discharges whose repetition rate saturates at higher frequencies as the discharge channels themselves become unstable.

The needle–plate defect told a completely different story—one that defied the intuitive expectation that more vibration always means more discharge. At low vibration levels, PDIV actually rose, peaking at a 5.6 percent increase under 300 Hz and 0.25 g. The explanation lies in bubble dynamics. In a stationary needle–plate gap, the intense field at the needle tip drives dissolved gas out of the oil, and electric field forces migrate these microbubbles along field lines until they aggregate into localized weak spots where discharge preferentially ignites. Gentle vibration disrupts this choreography: it scatters bubbles toward the edges of the insulating pressboard and prevents stable bubble clusters from forming, effectively healing the weak regions and raising the discharge threshold.

Push the vibration harder, however, and the picture flips. Once acceleration exceeded 0.75 g, PDIV dropped below its no-vibration value, falling by up to 11.1 percent at 100 Hz and 1 g. At these intensities, the direct electric field distortion from electrode displacement overwhelms the bubble-scattering benefit, making discharge initiation easier. The V-PRPD pattern for the needle–plate defect confirmed its fundamentally different character: discharge pulses spread in a roughly uniform band across the vibration cycle, with only a modest peak between 180° and 240°, showing that electric field distribution—not mechanical phase—remains the dominant trigger. Conventional PRPD patterns were similarly resilient, with pulses still concentrated around voltage peaks in a triangular distribution, though low-amplitude discharges appeared in the negative half-cycle once vibration was applied.

The discharge development metrics for the needle–plate defect were equally nuanced. Repetition rate rose overall with vibration but dipped non-monotonically at certain amplitude transitions, and average discharge magnitude increased by 34.3 percent under 100 Hz vibration while decreasing under 200 and 300 Hz—because rapid electrode motion splits the available energy into many small, frequent pulses rather than fewer large ones. Maximum discharge magnitude trended upward with vibration intensity, as the highly non-uniform field near the needle tip amplifies electron ionization during gap compression. None of these changes approached the dramatic sensitivity of the floating defect, underscoring that vibration’s fingerprint on partial discharge is written differently for every defect geometry.

For the engineers who keep the grid running, the implications are immediate. A floating conductor shaken loose inside a transformer will not announce itself with the large, classic discharge signatures that PRPD-based diagnostics expect; it will whisper with frequent, tiny pulses that begin at voltages far below design thresholds. Meanwhile, a needle defect under moderate vibration may temporarily look healthier than it is. The study’s authors argue that mechanical operating conditions—and vibration-phase features of the kind their V-PRPD method exposes—must be folded into partial-discharge-based condition assessment. As transformers age and grids push equipment harder, the hum of the machine may turn out to be one of the most important diagnostic signals of all.

Subject of Research: Effect of mechanical vibration on partial discharge behavior in oil–paper insulation defects of power transformers

Article Title: Influence of mechanical vibration on typical oil–paper insulation partial discharge defects in power transformers

Article References: Li, H., Zhang, S., Ding, J., La, Y., Lin, X., Lin, W., Wang, S., & Zou, L. (2026). Influence of mechanical vibration on typical oil–paper insulation partial discharge defects in power transformers. Frontiers in Electronics, 7, Article 1869509. https://doi.org/10.3389/felec.2026.1869509

Image Credits: AI Generated

DOI: 10.3389/felec.2026.1869509

Keywords: power transformers, partial discharge, oil-paper insulation, mechanical vibration, floating electrode defect, needle-plate defect, PDIV, V-PRPD, bubble dynamics, electric field distortion, insulation diagnostics, electro-mechanical coupling