The economic case for drone bridge inspection is not image quality; it is access. A payload that can hold a stable line of sight under a soffit, in a GNSS shadow, at 15 m standoff replaces a snooper truck, a lane closure and a shift of rope access. Thermal adds the one thing visual inspection cannot do at all: seeing delamination before it becomes a spall.
Key takeaways
- Thermal adds subsurface information to a concrete inspection — delamination and moisture ingress that a visual survey cannot see.
- Required resolution follows from defect type; the smallest defect you must catch sets the flight distance.
- Inside tunnels and under decks GNSS is unavailable, which changes both the aircraft and the georeferencing method.
- The survey replaces rope access and lane closure. That, not equipment cost, is where the economics are decided.
On this page
What thermal adds to a concrete inspection
Delamination is a subsurface separation between concrete layers, usually caused by rebar corrosion. It is invisible until the cover spalls off. Because the debonded layer is thermally isolated from the mass behind it, it heats and cools faster than sound concrete — so under solar loading it reads warm, and during evening cooling it reads cool.
That inversion is diagnostic and useful. A defect that is warm at 14:00 and cool at 21:00 is almost certainly delamination; a warm patch that stays warm in both passes is more likely a stain, a repair patch or a different material.
On soffits and in tunnels there is no solar loading, which removes the mechanism entirely. Thermal contributes far less there, and visual GSD becomes the whole survey.
Resolution required by defect type
| Defect | Sensor | Target GSD | Typical standoff | Notes |
|---|---|---|---|---|
| Hairline crack (< 0.3 mm) | Visible zoom | 0.1–0.2 mm/px | 5–12 m | Lighting angle is decisive |
| Structural crack (> 1 mm) | Visible zoom | 0.5 mm/px | 10–20 m | Width measurement needs scale reference |
| Spalling / exposed rebar | Visible zoom | 1–2 mm/px | 15–30 m | Easy target, prioritise coverage |
| Delamination | Thermal | 20–40 mm/px | 10–25 m | Requires solar loading cycle |
| Efflorescence / leaching | Visible | 2–5 mm/px | 15–30 m | Colour matters; avoid heavy AGC |
| Bearing and joint condition | Visible zoom | 0.5–1 mm/px | 8–20 m | Often the hardest access on the structure |
| Tunnel lining water ingress | Thermal + visible | 20–40 mm/px | 5–15 m | Evaporative cooling gives the signature |
Flying where GNSS is not
Under a deck or inside a tunnel, satellite positioning degrades or vanishes. Three consequences follow. First, the aircraft must be flyable in a non-GNSS mode with vision or lidar positioning — that is an airframe decision, not a payload decision. Second, geotagging fails, so your defect location has to come from structured flight planning and reference marks rather than from EXIF. Third, gimbal stabilisation carries more of the burden because you lose position-hold assistance.
This is where pointing stability earns its specification. A gimbal holding ±0.01° keeps a 12 m standoff shot usable in gusty conditions under a deck; a looser gimbal produces motion blur that destroys the GSD you planned for. The relationship is worked through in what 0.01° accuracy means at range.
Recommended UAVThermal payloads
MV-4P — quad-sensor micro pod. Light enough for the small, agile airframes that soffit and tunnel work demands, with thermal and visible in one unit.
OP-80P — 40x zoom spherical pod for deck, pier and cable inspection where standoff is available and GSD is the constraint.
AX-20D — stabilized gimbal for integrators building a confined-space inspection aircraft around a specific camera and lighting package.
Related reading
Technology: gimbal stabilization technology and payload mechanical integration.
Field practice: bridge and infrastructure inspection with zoom payloads, building envelope thermal surveys and the complete guide to drone inspections.
Adjacent missions: dam and levee inspection and railway corridor inspection.
Cost, access and what replaces what
The comparison that decides a bridge drone programme is against access equipment, not against another camera. An under-bridge inspection unit plus traffic management plus a crew is a substantial daily cost and usually requires lane closure; a drone crew of two with no closure is a fraction of that.
The honest limit is that drones replace routine and screening inspection, not hands-on principal inspection. Codes in most jurisdictions still require tactile examination at defined intervals. What the drone changes is the interval between hands-on visits and the targeting of them — you arrive knowing which three spans need the snooper truck instead of booking it for all twelve.
Data volume is again the hidden cost. A medium bridge at 0.5 mm/pixel produces very large image sets. The programmes that work invest in a defect-tagging workflow early; the ones that do not end up with terabytes nobody has looked at and an inspection record that cannot survive an audit.
Where this goes wrong
Concrete inspection punishes optimistic planning.
- Planning GSD from the datasheet, not from the geometry. A 4K sensor does not guarantee 0.2 mm/pixel; distance and focal length do. Run the arithmetic before the site visit, not after the imagery disappoints.
- Thermal on a soffit. No solar loading means no delamination contrast. Thermal under a deck or in a tunnel mostly finds water ingress, not debonding — plan the sensor to the surface.
- Losing position reference under the deck. Without GNSS, EXIF geotags are wrong or absent. If defect location depends on them, the survey is not defensible. Use structured flight paths and physical reference marks.
- Ignoring wind funnelling. Gorges, valleys and river crossings produce gusts far above the forecast. Image stability collapses before flight safety does, so the first sign of trouble is unusable frames.
FAQ
Can a drone detect delamination as reliably as hammer sounding?
For accessible deck surfaces under good solar loading, aerial thermal correlates well with sounding and covers vastly more area per hour. It is weaker on shaded soffits, on wet surfaces and where cover depth is large. Most agencies use thermal to target sounding rather than replace it.
What time of day should we fly a deck delamination survey?
Mid-afternoon during the heating phase, roughly 12:00–16:00 on a clear day, or during evening cooling around 19:00–22:00. Flying both and comparing the inversion is the highest-confidence method. Overcast days with no thermal cycle produce nothing.
Do we need a radiometric payload?
Not usually. Delamination detection is a relative-contrast task, so a non-radiometric thermal camera is sufficient. Radiometric becomes worthwhile if you also inspect bearings, electrical assets or want to trend absolute values across survey cycles.