The whole section, many years of lightning
Each run throws thousands of years of random flashes at the ten towers and nine spans from T16 to T25. Every flash gets a random current, position and power-frequency angle. A flash to the shield wires near a tower flashes it over if its current exceeds that tower's critical current from the Backflashover Lab model. A flash that reaches phase S, from the span screening, flashes it if it exceeds 5.7 kA and no phase-S arrester covers that point. Change the scheme, run again, and compare.
Where the flashovers happen
per year, simulated vs recordedCurrents that caused a flashover
share of simulated flashoversDoes the estimate settle?
running rate against simulated yearsRuns so far
| Scheme | BF | SF | Total | Phases |
|---|
Rates per 100 km per year. BF backflashover, SF shielding failure of phase S. Phases: which phase flashed in the backflashovers.
Four lessons
Backflashover dominates, at the bare towers
Shield-wire strikes are hundreds of times more frequent than strokes that slip through to phase S. About three quarters of the simulated outages come from T22 and T24, the two towers without any arresters (T24 also has an assigned 45 Ω footing). At the equipped towers the residual backflashovers fall on phase S, mostly at T20, T23, T19 and T21.
Arresters on S fix both routes
Phase-S arresters at a tower that already has R/T remove its backflashovers too, because all three phases then follow the tower; at both ends of a span they also remove its shielding failures. "T20 and T21" takes out both mechanisms on the valley span. The largest single gain, though, is fitting the bare towers T22 and T24: tick the box and compare.
The record is far above the model
T16–T25 recorded 16 lightning faults in 2022–2025, about 125 per 100 km per year, several times the simulated rate. Fault locations attributed to the nearest tower, multiple strokes, unrecorded weak strokes and the assigned footing values of five towers can each explain part of the gap. The model is a screen, not a reconstruction.
Use it to rank, not to predict
Absolute rates depend on the stroke-to-line formula, the current distribution and the "60% at the tower" rule. Differences between schemes, run with the same inputs, are much more robust than any single number.
Model behind the page
Shield-wire strokes
Strokes to phase S
Recorded faults
Not included: shielding failures of phases R and T, mid-span shield-wire strokes, multiple strokes, induced overvoltages and positive flashes.