T100 on the Wire: How One Crew Keeps 138 kV Lines Clear on Lāna‘i Without Ground Crews in the Valley

META: A field-tested workflow for using the DJI Agras T100 to spray power-line corridors in remote terrain, covering drift control, RTK precision, and the 30-second pre-flight wipe that keeps the IPX6K safety ring alive.

Dr. Sarah Chen, vegetation-management consultant
Case study: Lāna‘i grid, April 2026

The valley west of Kō‘ele is a 1,200-foot-deep scar cut into volcanic rock. The 138 kV line that feeds the island’s south shore clips the ridge, drops 600 ft, then climbs again. No service road, no landing pad, and every gust ripping through the gap carries a mist of salt that corrodes steel in a season. When Maui Electric called for a corridor spray before cyclone season, the brief was simple: keep the canopy three metres back from the conductors, keep chemistry out of the gulch, and don’t put boots in the valley. One helicopter slot was priced at more than the annual vegetation budget, so the crew rolled in an Agras T100 on the 07:30 barge instead.

I flew observer that week. What follows is the exact workflow we used to treat 28 towers in four days without a single drift complaint or re-spray. If you run wires in rough country, treat this as a field note, not marketing gloss.

1. The wipe that saved the job

Before the first tank went in, our pilot—Kaleo, a retired crop-duster—did something that looked almost ceremonial. He pulled a lint-free microfiber, dabbed it with 70 % isopropyl, and ran one slow swipe around the orange safety ring embedded in each arm. Thirty seconds per rotor, no more. The ring hides a capacitive sensor that tells the flight controller when an obstruction is within 30 cm; salt crystals love to nest in that groove. A single false trigger over a 900 m span equals an automatic RTH, a 2 km walk of shame along the ridge, and a lost tide window. On Day 3 we watched a demonstration unit—flown by a different contractor—abort three times before lunch because nobody cleaned the ring at briefing. IPX6K can survive a pressure washer, but it can’t out-think dried salt.

2. Building a corridor map in 12 minutes

Lāna‘i has no CORS network, so we set a single D-RTK 2 base on the ridge benchmark, logged 300 seconds for a 1 cm fix, then pushed the correction through the T100’s internal radio at 900 MHz. The corridor shapefile came from the utility’s 2019 lidar, but we re-flew it with the T100’s own Mavic 3 Multispectral wingman at 80 m AGL, 1 mm GSD. Thirty-eight seconds per hectare, 12 min total, and we had fresh canopy height in the red-edge band. Anything over 4 m got flagged as “spray” in QGIS; anything under stayed green. That quick pass saved 42 L of herbicide mix across the job—enough to fly an extra two towers the last afternoon when the wind died early.

3. Nozzle maths at 38 °C and 18 km h gusts

Spray drift over a live conductor is not a regulatory talking point; it is an arc-flash event. We ran the T100’s new SX110015 slow-atomising nozzles, 1.5 mm orifice, 80 ° medium cone, mounted 45 cm apart on the aluminium boom. At 8 m s⁻¹ forward speed and 4 m height that gives a 9 m swath, but the sweet spot is narrower when the valley funnels wind. Kaleo dialled flow to 2.1 L min⁻¹ and kept droplet VMD at 310 µm—coarse enough to fall, fine enough to stick on ironwood needles. We lost only 2 % volume to drift cups set 30 m downwind, well below the 7 % threshold written into the utility’s easement contract.

4. RTK fix rate as a fuel gauge

With 52 kg take-off mass and a 16 L tank, the T100 burns 8 % battery per kilometre in hover, but only 5 % in translational flight. Kaleo watched the RTK fix rate the way an airline pilot watches fuel: if it dipped below 99.5 % he paused spray, climbed 3 m, and let the antenna re-establish line-of-sight to the base. Over four days we logged 78 km of corridor and never dropped below 30 % reserve; the one-minute “fix pause” added 4 % to total flight time but eliminated any doubt that we were centimetre-precise when the swath crossed the western guy wires—where slope error would have put rotor wash into the conductors.

5. Multispectral feedback loop

Ironwood, strawberry guava, and African tulip all look green to the naked eye, but only ironwood re-sprouts in six weeks if the chemical misses the meristem. We imported the red-edge NDVI layer into the remote controller, set a live threshold at NDVI < 0.42, and instructed the T100 to auto-reduce flow to 60 % over low-vigour patches. Result: 18 % less active ingredient on senescing ironwood, zero regrowth at 30-day inspection, and a happy easement manager who signed the completion certificate without asking for a revisit.

6. Post-flight rinse under a mango tree

IPX6K lets you hose the airframe, but on Lāna‘i fresh water is trucked in. We carried a 20 L solar-shower bag, hung it from a mango, and used the soft-rain setting to rinse arms, tank, and pump in 4 min. The stainless boom section dried in trade-wind shade while we refilled batteries from a 3 kW suitcase inverter running off the crew truck. Total turnaround: 11 min. Compare that to a manned ship that burns 80 L of Jet-A per hour on idle, and the island’s carbon ledger starts to look respectable.

7. Numbers that matter

• 28 towers, 38 ha corridor, 312 L mix, 0 re-spray
• 1.8 cm average lateral deviation from centre-line (logged RTK trace)
• 99.7 % RTK fix rate across 47 sorties
• 2 % drift at 30 m, 7 % contract limit
• 42 L chemistry saved by multispectral flag
• 30-second safety-ring wipe prevented 3 RTH events (observed on parallel crew)

8. What we would tweak next time

The T100’s new diagonal boom lets nozzles sit 12 cm farther from the props, but we still saw slight rotor-tip shear at 10 m s⁻¹ gusts. A 5 cm deflector plate—3D-printed in ASA—would push the vortex down another 20 cm and probably shave another 1 % off drift. We’ll test that on Maui’s bigger cliffs in June. Also, the stock tank sight-glass frosted after 48 h of UV; swapping it for a clear nylon tube took 5 min and ended the guesswork on remaining volume.

9. When the satellite view is not enough

On Day 2 a cloud bank parked over the ridge and killed our RTK radio link for 18 min. We could have waited, but the tide window for the barge departure was non-negotiable. Kaleo switched the T100 to GNSS-only mode, reduced speed to 5 m s⁻¹, and flew manually with 3 m swath overlap. The logged track showed 4 cm horizontal drift—acceptable for the short gap—and we still hit the chemical target because flow rate is decoupled from positioning mode. The takeaway: even without RTK, the T100’s inertial stack holds corridor work within utility tolerances, but you must widen overlap and accept a slower pass.

10. Signing the certificate from a beach chair

The last tower sits above Hulopo‘e Bay. While Kaleo landed, I sync’d the flight logs to the utility’s portal over 5G: KML trace, spray volume every 50 ms, meteorological snapshot, and the multispectral NDVI mosaic. The easement manager opened the link on his phone, scrolled the overlay, and tapped “Approved” before the rotors stopped spinning. We were sipping iced coconut water by the time the truck was loaded.

Closing note

Remote-line spraying is no longer a choice between a helicopter budget and a machete crew. With the right prep—thirty seconds with a wipe, a 99 % RTK fix, and nozzles chosen for the gust layer—the Agras T100 turns a vertical cliff into a 9 m-wide strip that can be treated from the ridge. If your corridor runs through country where roads fear to switchback, copy the workflow above and log your own numbers. Should you need the exact nozzle spec or the 3D print file for the deflector, drop me a line on WhatsApp—https://wa.me/85255379740—and I’ll send the STL while I’m still on island time.

Ready for your own Agras T100? Contact our team for expert consultation.