Drone-based LiDAR and photogrammetry for storage tanks, flare stacks and elevated piping

The assets that justify putting an aircraft in the air

The case for flying is not that drones are modern. It is that a specific set of industrial assets are expensive to reach and cheap to overfly. Storage tank shells and roofs, elevated flare stacks and tips, cooling tower decks and casings, elevated pipe racks and stack structures all share a profile: the surface that needs looking at is thirty to four hundred feet in the air, spread over a large area, and reaching it means building something first.

OSHA sets the trigger. Fall protection is required at four feet above a lower level in general industry under 29 CFR 1910.28, and six feet in construction under 29 CFR 1926.501. Once you cross those thresholds the job acquires a rescue plan, a competent person, harness inspection, anchor points and a permit. Scaffolding a tank shell or a flare stack turns a two-day inspection into a multi-week mobilization with its own safety exposure.

The drone changes the arithmetic on capture, not on judgement. A two-person crew reaches the same surface in an afternoon with nobody at height, and produces a measurable, geo-referenced record rather than a set of photographs someone took while holding a ladder. What it does not do is form an opinion about fitness for service, which is why the deliverable feeds an inspector rather than replacing one.

What aerial capture measures, and what it does not

Aerial LiDAR returns direct range measurements to a surface. DJI publishes Zenmuse L2 detection range at 450 m against a 50 percent reflectivity target at 0 klx, and 250 m against 10 percent reflectivity at 100 klx; point rate at a maximum of 240,000 points per second on single return and 1,200,000 on multiple returns; and system accuracy of 5 cm horizontal and 4 cm vertical at 150 m — measured with a Matrice 350 RTK at 150 m relative altitude, 15 m/s, gimbal at −90°, RTK in FIX status, post-processed in DJI Terra with point cloud optimization enabled. Ranging accuracy is stated as 2 cm at 150 m, RMS 1σ, at 25 °C against 80 percent reflectivity.

Those figures set expectations honestly. Centimetre-class aerial data answers geometry questions — is the shell plumb, is the roof ponding, where is the low point, what is the clearance envelope, what is the stack verticality. It does not answer millimetre questions. For comparison, Leica publishes RTC360 3D point accuracy of 1.9 mm at 10 m at a 68 percent confidence level per JCGM 100:2008, which is why terrestrial scanning still owns tie-in and fabrication work.

And no optical sensor measures wall thickness. Thickness is a through-wall measurement that requires an ultrasonic transducer coupled to clean metal. Aerial capture tells an inspector where to look and documents what the surface looked like on the day; it does not tell anyone how much steel is left. Being explicit about that line is what separates a credible programme from a pitch.

Part 107, stated as the rules that actually bind a plant flight

The single most useful rule is the structure exception. 14 CFR 107.51(b) states that the altitude of the small unmanned aircraft cannot be higher than 400 feet above ground level unless the aircraft is flown within a 400-foot radius of a structure and does not fly higher than 400 feet above the structure's immediate uppermost limit. The EPA's Air Pollution Control Cost Manual covers elevated flares with stack heights from 30 to 500 feet; a 350-foot stack yields 750 feet of legal altitude inside that radius, with no waiver.

The rest of §107.51 constrains the flight envelope: groundspeed no more than 87 knots, minimum flight visibility of three statute miles observed from the control station, and cloud clearance of 500 feet below and 2,000 feet horizontally. Section 107.41 requires prior ATC authorization for Class B, C or D airspace and surface-area Class E — routine for Gulf Coast plants sitting under Houston or Beaumont approaches, and handled through the FAA's LAANC service rather than a paper application.

Two more apply to every flight. Part 89 Remote ID broadcast is required; the FAA's discretionary enforcement policy ended on 16 March 2024. And §107.29 permits night operation without a waiver, provided the aircraft carries anti-collision lighting visible for at least three statute miles with a flash rate sufficient to avoid a collision. Night flight is useful on flare work, where a lit tip against a dark sky records tip condition that daylight washes out.

Where a waiver is genuinely required

Section 107.205 lists what can be waived: §107.25 operation from a moving vehicle or aircraft, §107.29(a)(2) and (b) anti-collision lighting, §107.31 visual line of sight, §107.33 visual observer, §107.35 multiple aircraft, §107.37(a) right of way, §107.39 operation over human beings, §107.41 airspace, §107.51 operating limitations, and §107.145 operation over moving vehicles. Anything not on that list cannot be waived at all.

Two bite in practice. Visual line of sight under §107.31 requires the remote pilot in command, the visual observer and the person manipulating the controls to see the unmanned aircraft throughout the entire flight, unaided except by corrective lenses. A tank farm where a 200-foot tank sits between the pilot and the aircraft breaks that, and the fix is usually observer repositioning rather than a waiver. Section 107.39 prohibits flight over human beings unless they are directly participating, are under a covered structure or inside a stationary vehicle, or the operation complies with Subpart D — and plant contractors working below are not directly participating.

The BVLOS picture is still moving. The FAA published its proposed Part 108 rule, "Normalizing Unmanned Aircraft Systems Beyond Visual Line of Sight Operations", in the Federal Register on 7 August 2025; the comment period closed 6 October 2025, reopened on 28 January 2026 and closed again on 11 February 2026. Until that rulemaking concludes, BVLOS runs on a §107.31 waiver. For fenced-line industrial work inside a single site, that rarely matters — the aircraft stays in sight.

Storage tank shells and roofs

API 653 sets the cadence that drives tank work: external visual inspection by an authorized inspector at least every five years, or RCA/4N years, whichever is less, where RCA is the remaining corrosion allowance and N is the shell corrosion rate in mils per year — on top of routine in-service inspection by owner personnel at intervals not exceeding one month. Every one of those external inspections is an access problem on a tank of any height.

Aerial capture delivers shell geometry, plumbness and roundness data, coating condition imagery across the full circumference, and weld seam appearance at resolutions that a ground-based photograph from fifty feet away cannot produce. On roofs it delivers plate deformation, ponding low points, rafter and pontoon condition on floating roofs, and the drainage geometry that explains where water sits. That is a substantial share of what an external inspection is looking for, gathered without a single person crossing the four-foot threshold.

What it does not deliver is the thickness data or the inspector's judgement. Shell and roof plate UT readings, seal condition on a floating roof, and the close visual determination that API 653 assigns to an authorized inspector all remain hands-on. Where our ASNT Level III consulting work fits is in writing the procedure that defines which findings the aerial data can support and which it cannot — because an undefended boundary there is a finding waiting to be challenged.

Flare stacks and elevated flare tips

The flare tip is the hardest thing on a refinery to look at. It sits at the top of a 30 to 500-foot stack, per the height range the EPA cost manual covers, it is burning, and 40 CFR 60.18 requires the flare to be operated with a flame present at all times. Getting eyes on it conventionally means a full-height scaffold, a rope access team, or taking the flare out of service — and taking a flare out of service is a plant-level event, not a maintenance task.

This is where the §107.51(b) structure exception matters most. Operating within a 400-foot radius of the stack, the aircraft is legal to 400 feet above the stack's uppermost limit, which puts the tip comfortably inside the envelope for almost every elevated flare in US service. The capture produces tip erosion and burn-back imagery, pilot and igniter condition, stack verticality against plumb, guy wire and cable condition, ladder and platform integrity, and riser geometry for any planned modification.

Thermal and plume conditions set the real operating limit, not the regulation. An operating flare produces heat, buoyant plume and turbulence, and the flight plan works the approach angle and standoff around it rather than through it. Flight planning here is a site-specific engineering exercise informed by wind, flare load and stack geometry — which is why the pre-flight walk-down carries as much weight as the flight itself.

Cooling towers, elevated racks and the rest of the elevated envelope

Cooling towers are large, structurally repetitive and unpleasant to inspect. The fan deck, the casing, the distribution basins and the fan stacks are all elevated, wet and in some cases chemically treated. Aerial capture returns deck and casing geometry, fan stack alignment references, distribution layout and drift eliminator condition from above without cell-by-cell entry — and the tower keeps running while it happens. Fill condition, internal structural members and mechanical drive inspection stay internal and stay hands-on.

Elevated pipe racks are the awkward case. They are the assets where tie-in routing decisions get made, which means someone will eventually want fabrication-grade geometry off them, and aerial centimetre-class data does not reach that standard. What aerial capture does well on racks is the overview: support positions, clearance envelopes, and which rack levels are congested enough to require a terrestrial scan. Used that way it scopes the terrestrial work rather than competing with it.

Visual line of sight is the practical constraint across all of it. Rack structure, tower casing and tank shells block the pilot's view of a small aircraft quickly, and §107.31 requires unaided sight throughout the entire flight. Observer positioning, flight leg design and sortie planning around the structure are what keep the operation legal — details that get skipped in a demo flight and matter on a real site.

Turning the flight output into inspection planning

The output of a good flight is not a video. It is a registered, geo-referenced point cloud with co-registered orthoimagery, tied to the same control network as any terrestrial capture on the site, delivered with a record of what was flown, when, in what conditions and to what accuracy. That is the difference between a deliverable an inspector can build a scope on and a folder of photographs nobody can locate on the asset six months later.

Located findings are what change the programme. A pit recorded as a coordinate on a specific plate in a specific shell course can be re-found, re-measured and trended. Once findings live on geometry, RBI visualization and damage mechanism mapping shows whether corrosion clusters where the model predicts — at the water draw, on the windward course, under the ponding low point — or somewhere the damage mechanism review missed entirely.

The same geometry carries thickness data. Corrosion monitoring locations plotted on a real shell rather than in a spreadsheet make thickness trending on asset geometry legible as a spatial pattern, and that is the point at which scan data has become a digital twin rather than a survey. Our 3D scanning services page covers the capture options across terrestrial and aerial; get in touch for a scope on a specific tank farm, flare or tower.

Request a consultation