Oil and Gas Drone Inspection: Where Drones Add Evidence

On This Page

• Match the sensor to the inspection question
• Understand what the sensors actually measure
• Give the crew the inputs before mobilization
• Carry observations into the maintenance workflow
• Specify deliverables that can be checked
• Keep missing evidence visible
• Questions to settle before choosing a provider

Match the sensor to the inspection question

Start with the asset and the suspected condition, then choose the collection method. The table separates outputs that may look similar in a presentation but support different decisions.

Source basis: BINDT's visual-inspection and ultrasonic guidance, FLIR's thermal and gas-camera documentation, EPA's OGI fact sheet, the methane studies, and USGS calibration guidance listed below. The matching of buyer questions to deliverables is editorial synthesis, checked September 6, 2026.

For visual work, BINDT's HOIS guidance announcement identifies coating assessment and flare-tip or stack inspection among the applications considered. It emphasizes image quality suitable for the intended engineering assessment, including resolution, viewing direction, and lighting. That is a stronger specification than a camera megapixel count.

Define the required surfaces individually. “Inspect tank T-04” leaves open whether the roof, external shell, fittings, and obscured areas are included. For a pipeline corridor, separate observations of accessible aboveground features from any claim about the buried pipe itself.

Understand what the sensors actually measure

Visible images locate indications; thickness needs a measurement method

A photograph can preserve the appearance and location of a coating discontinuity or visible surface change. It does not directly measure remaining metal thickness. BINDT explains that ultrasonic thickness gauging derives thickness from pulse travel time and requires calibration or knowledge of sound velocity in the material.

If a provider offers drone-carried ultrasonic testing, ask for the actual probe arrangement, suitable surfaces, calibration procedure, measurement locations, and retained readings. Adding a probe changes the inspection method and its evidence requirements. A visual survey and a thickness survey should have separate scope lines even if the same contractor supplies both.

Thermal cameras need surface context

Thermal interpretation depends on how the surface emits and reflects infrared radiation. Emissivity describes its ability to emit that radiation. FLIR's emissivity explanation shows why reflective metal can display radiation from surrounding objects, producing an apparent temperature that misleads the viewer.

Ask for the surface assumptions, reflected-temperature treatment, capture time, viewing geometry, and process operating state with each significant thermal finding. A colored screenshot without these inputs is weak support for a maintenance decision. Record a suspected anomaly first; assign its cause only when the thermographer or other responsible specialist has enough information.

Gas imaging and methane quantification are different jobs

OGI uses a camera sensitive to relevant gas absorption bands to make a plume visible under suitable conditions. FLIR's UAS gas-camera guidance says general-purpose thermal cameras do not reliably detect these gases. A supplier's claim of “thermal leak detection” therefore needs the gas species, sensor, and detection method spelled out.

EPA's 2023 Appendix K fact sheet describes an OGI operating envelope using wind speed, gas-to-background temperature difference, and viewing distance. It also identifies camera checks, operator training, and survey records. These are concrete questions for a proposed OGI survey. The fact sheet is a description of a particular U.S. protocol, not blanket authorization for an airborne method or every oil and gas facility.

Quantification adds another step. In the mass-balance method examined by a 2025 drone-methane uncertainty study, an aircraft samples methane concentration across a vertical plane downwind of a source. Concentration and wind measurements are combined to estimate the mass passing through that plane per unit time. Flight-line spacing, plume capture, wind, and post-processing affect the answer.

Thus a location pin, gas image, concentration trace, and result in kilograms per hour are separate outputs. Require the method that connects them. The study's findings apply to the examined method and deployment conditions; they are not a universal accuracy rating for methane drones.

Give the crew the inputs before mobilization

Use a short assignment sheet shared by the inspection lead, flight team, and person accepting the report. Our recommended starting fields are:

• Asset identity and coverage: equipment tags, drawings, required faces or sections, previous findings, and inaccessible areas already known.
• Decision and method: whether the task is visual screening, condition assessment, temperature interpretation, gas detection, quantification, or dimensional measurement.
• Collection conditions: relevant process state, allowable access window, required views, sensor settings and checks, and a rule for stopping when conditions no longer support the measurement.
• Operating constraints: site-approved flight areas and separation, nearby work, access permissions, communications, and emergency coordination.
• Handover: file formats, asset identifiers, location reference, recipient, review responsibility, and deadline for reporting significant findings.

Agree which permissions apply to the actual flight, including its jurisdiction and intended operating area. Site access permission should not be treated as the complete flight authorization.

Hazardous-area suitability deserves a specific answer at oil and gas facilities. In U.S. general industry, OSHA 1910.307 addresses electrical equipment in classified locations and distinguishes intrinsic safety, location approval, and suitability for the location. Have the site's responsible authority establish the permitted equipment and operating area. A drone's weather rating or camera specification does not answer that question.

For maps or models, add the coordinate reference, units, control method, and checks required by the recipient. USGS calibration guidance separates image ground sample distance from geometric accuracy and identifies the importance of control and tie points. A sharp model can still have positioning errors that matter when comparing repeated surveys.

Carry observations into the maintenance workflow

The following sequence is a recommended handover design. Adjust the named roles and urgency rules to the operator's maintenance system.

1. Approve a representative sample. Before broad capture, have the inspection lead judge whether a sample shows the required feature clearly enough. Confirm the file can be opened and associated with the correct asset.
2. Collect against the coverage list. Record completed views, unusable captures, and missed areas separately. Keep the original files alongside any annotated copies.
3. Review findings by method. Route visual observations, thermograms, and methane results to the people qualified to interpret each. Keep the original observation separate from the diagnosis.
4. Create actionable records. Give each finding a stable identifier linked to its equipment tag, overview, detail, timestamp, location, interpretation, and proposed next action.
5. Assign and close follow-up. Record who will investigate, what confirmatory work is needed, and how the repair or disposition will be checked. Retain the link between the initial finding and the closure record.

For example, suppose an image shows a surface change near a pipe support. The initial entry can locate and describe it, flag limited visibility beneath the support, and request closer assessment. The record should not become “wall loss confirmed” merely because a reviewer adds a red outline. This is an illustrative reporting example, not a field finding.

A useful receiving test is whether a maintenance planner can identify the asset and understand the next action without calling the pilot. For the broader distinction between observation, measurement, and diagnosis, see what drone-inspection evidence can prove.

Specify deliverables that can be checked

Purchase a report with defined contents and a way to assess completeness. These suggested checks make supplier proposals comparable without inventing a universal performance target.

These are editorial procurement checks derived from the method limits above, not a claim that one standard mandates this table.

Measure useful coverage against the agreed list. One straightforward project metric is usable scoped views divided by total required views, expressed as a percentage. Keep inaccessible views in the denominator unless the buyer explicitly revises the scope, and report the exclusions alongside the result. This measures completion of the specified capture, not the probability of finding every defect.

For recurring work, track rejected records, recollection, time to reviewed findings, and completed follow-up alongside flight time. Keep their definitions consistent so a change in reporting practice does not look like an improvement in inspection performance.

Compare quotations over that same scope. Ask whether processing, specialist interpretation, raw-data delivery, software access, travel, recollection, and confirmatory inspection are included. Any claimed reduction in access work or downtime needs a comparison with the method the operator would otherwise use for the same accepted task.

Keep missing evidence visible

A non-detection is a result for the sampled place, period, conditions, and method. It is not proof that no leak exists anywhere on the site. Similarly, an obscured surface should remain uninspected in the register. Treating either as a clean bill of health hides the information the next inspector needs.

A 2026 blinded controlled-release study at the TADI facility evaluated commercial methane technologies, including drone systems, against undisclosed release rates. It found substantial differences in quantification performance and identified wind conditions as an important factor. Its controlled setting does not reproduce every operating facility, but it supports asking for testing of the proposed method rather than accepting a sensor specification as system-level proof.

Ask how the provider reports missed detections, false positives, rejected surveys, and uncertainty under conditions similar to yours. An impressive demonstration of one plume says little about these failure cases.

When the required surface or measurement cannot be obtained, change the collection window, use an appropriate alternative inspection method, or narrow the conclusion. Adding automated defect labels or a polished model does not supply an observation that was never captured.

Questions to settle before choosing a provider

Before selecting a provider, close six questions:

1. What specific decision will each sensor's output support?
2. Which surfaces, components, gases, and operating conditions are inside the scope?
3. What demonstrates that the proposed method can collect the required detail or measurement at this site?
4. What remains unobserved, and how will the report identify it?
5. Who interprets the findings, accepts the data, and commissions confirmatory work?
6. What files, identifiers, uncertainty information, and follow-up records will the operator retain?

Choose a visual survey when located surface observations answer the immediate question. Add thermal or methane work when those measurements serve a defined maintenance or emissions task. If the decision requires thickness, internal condition, or another property the proposed sensor cannot establish, commission the corresponding inspection method. The drone earns its place by supplying usable observations that the next person can act on.

Source notes

• BINDT: HOIS guidance for drone-based external visual inspection. Institutional announcement describing image-quality considerations and oil and gas applications; not the full guidance document.
• BINDT: Ultrasonic thickness gauging. Technical explanation of pulse timing, thickness, and calibration.
• FLIR: How emissivity affects thermal imaging. Manufacturer explanation of emission and reflection effects on thermal interpretation.
• FLIR: UAS gas leak detection. Manufacturer guidance distinguishing general thermal cameras from gas imaging equipment.
• EPA: Appendix K technical fact sheet, December 2023. Government summary of OGI operating conditions and documentation within that protocol.
• Atmospheric Measurement Techniques: Drone methane measurement uncertainty, 2025. Peer-reviewed analysis of mass-balance collection and processing uncertainties.
• OSHA: Hazardous classified locations, 1910.307. U.S. general-industry electrical equipment requirements for classified locations.
• USGS: Guidelines for calibration of uncrewed aircraft systems imagery. Government technical report on geometric quality, calibration, and comparability.
• Atmospheric Measurement Techniques: TADI controlled-release testing, 2026. Blinded experimental evaluation of methane technologies and conditions affecting quantification.

Last checked: September 6, 2026.