A radiometric thermal drone surveys every bushing, connector and transformer in a substation from outside the energised perimeter — turning a shutdown-dependent inspection into a routine 40-minute flight.
- 40 minTypical station survey flight
- QuarterlyCommon baseline interval
- 640×512Radiometric thermal baseline
- LoadThe variable that invalidates comparison
Key takeaways
- Radiometric is non-negotiable here. Findings feed maintenance priority systems, and a delta in degrees is actionable where “looks warm” is not.
- Temperature findings are only comparable at known load — log load conditions at time of survey or the data cannot be trended.
- Fly perimeter-in and maintain utility minimum approach distances; the drone removes the ladder crew, not the electrical hazard.
- Quarterly baseline with pre-summer and pre-winter peaks is a common programme; critical stations justify monthly passes.
On this page
What This Guide Covers
This is the operational side: what to point the camera at, how to interpret what comes back, and the procedure around energised plant. For the business case — why aerial survey displaces outage-dependent inspection, and what that is worth across a fleet of stations — see the substation drone inspection application page.
The short version of the case is that most substation thermography historically required either an outage or a crew inside the fence with a ladder. A drone at standoff needs neither, which changes inspection from an event into a routine.
Target Checklist
Five categories cover the great majority of findable defects, and working through them in a fixed order is what makes surveys comparable between visits.
| Target | What you are looking for | Interpretation note |
|---|---|---|
| Transformer tanks and radiators | Cooling anomalies — blocked or inactive radiator sections | Compare sections against each other, not against a threshold |
| Bushings | End-seal heating | Grade against the same bushing on the other phases |
| Disconnects and clamps | Resistance heating at the joint | The classic finding — a loose or corroded connection |
| Capacitor banks | Unit-to-unit deltas within the bank | An outlier unit matters more than the absolute value |
| Battery rooms via open doors | Cell and connection heating | Only where authorised — this is not a routine aerial target |
Notice how many entries say “compare against peers”. Substation thermography is overwhelmingly a comparative discipline: three phases doing the same job under the same load should look the same, and the one that does not is your finding. That is far more robust than any absolute temperature threshold, because it cancels out ambient conditions, emissivity and load in a single step.
Why Radiometric Is Non-Negotiable Here
Substation findings feed maintenance priority systems. A bushing running 18 °C above its phase peers is actionable data that a planner can rank against other work; “looks warm” is not, and it will not survive the conversation with the asset owner.
Calibrated spot measurement and temperature alarms come standard on thermometry payload versions. What they give you is a number attached to a location, repeatable enough that next quarter’s survey can be compared against this one — which is the entire point of a monitoring programme as opposed to a one-off look.
The decision has to be made at purchase. Radiometry depends on factory calibration of the specific core and cannot be added by firmware afterwards, as radiometric vs non-radiometric sets out. A non-radiometric payload bought for a substation programme is a payload that will need replacing.
Emissivity discipline matters here more than in most thermal work, because substations are full of shiny metal. Polished aluminium and galvanised steel are low-emissivity surfaces that under-report their temperature dramatically and reflect their surroundings enthusiastically — why shiny metal lies to your camera covers the correction, and thermal reflections covers the false positives.
Recommended Payloads
Station footprint and standoff drive the choice. All three options below should be ordered in their thermometry configuration.
| Payload | Zoom | Thermal | Ranging | Best fit |
|---|---|---|---|---|
| OP-90A | 10x optical / 30x hybrid | 640×512, <50 mK | 5–1,200 m | Zoom plus radiometric thermal with standoff logging |
| LX-9B | 30x optical / 120x hybrid | 640×512, <50 mK | 5–2,000 m | Large-footprint stations from a single hover |
| MV-2P | Digital only | 640×512, <40 mK | — | Compact option for distribution-class stations |
The LX-9B earns its place on large transmission stations because 30x optical lets one hover position cover several bays, which cuts flight time and — more importantly — keeps the aircraft further from energised plant. The MV-2P is the distribution-class answer: the most sensitive published NETD in the line at <40 mK, in 130 g, where the station is small enough that standoff is not the constraint.
The rangefinder is worth more here than it first appears. Distance-tagging each finding ties the observation to a specific element for the report, and standoff logging documents that the aircraft stayed outside the approach distance.
Load: The Variable That Invalidates Everything
Log load conditions at the time of survey. Temperature findings are only comparable at known load, and this is the single most common reason substation thermal programmes produce data that cannot be trended.
The physics is straightforward: resistance heating scales with current squared, so a defective connection at 30% load may look entirely normal and the same connection at 90% load may be alarming. A survey flown on a mild spring morning at low load will under-report every resistive defect in the station, and comparing it against a summer peak survey will suggest deterioration that is really just load.
Two practical consequences. First, record load with every survey and grade findings in that context. Second, schedule at least some surveys at meaningful load — a station surveyed only during low-demand windows is a station whose resistive faults have never been visible. This is why power line hotspot inspection makes the same point about conductors, and why the pre-summer and pre-winter peak surveys exist in most programmes.
A cold survey is not a clean station. Finding nothing at low load tells you very little. If the programme reports “no findings” after a survey flown at 25% load, the correct conclusion is that the survey had limited diagnostic power, not that the plant is healthy.
Safety and Procedure Notes
Maintain utility minimum approach distances. The drone removes the ladder crew; it does not remove the electrical hazard, and an aircraft that drifts inside the approach distance is a fault current path waiting for an excuse. Treat the published distance as a hard boundary in the flight plan rather than a guideline.
Fly perimeter-in. Establish the aircraft outside the fence, work inward along a planned sequence, and keep an escape direction available at every point. This also produces a consistent survey order, which is what makes visit-to-visit comparison possible.
- Confirm authorisation, the day’s switching state, and the current load figure before flight.
- Establish outside the perimeter and verify GPS lock and compass health away from large steel structures.
- Work the target checklist in a fixed sequence, capturing radiometric imagery with the rangefinder tagging each element.
- Grade findings against phase peers and comparable units rather than against an absolute threshold.
- Record load, ambient conditions and time alongside the imagery — without them the survey cannot be trended.
Compass calibration away from steel and maintaining GPS lock remain standard practice. Modern UAVs handle substation electromagnetic interference at survey distances, but the environment is dense with steel and high current, and a compass calibrated next to a transformer tank will produce heading errors that show up as an aircraft that will not hold position where you want it.
Related reading
- Radiometric Temperature Measurement on UAV Payloads
- Shutters, Blackbodies and Two-Point Correction
- Radiometric vs Non-Radiometric Thermal
- Emissivity in Drone Thermography
- Thermal Reflections and False Hot Spots
- Power Line Hotspot Inspection with Thermal Drones
- Isotherms and Temperature Alarms
- What a Thermal Inspection Report Should Contain
- OP-90A — zoom, radiometric thermal and ranging
- LX-9B — 30x optical multi-sensor pod
- MV-2P — 640×512 thermal at 130 g
- Substation Drone Inspection: The Commercial Case
FAQ
How often should substations be flown?
A quarterly baseline with additional pre-summer and pre-winter passes is a common programme, and critical stations justify monthly surveys. The seasonal passes matter because they catch the station approaching peak load, which is when resistive defects are most visible. A programme flown only on convenient mild days will systematically under-report the faults it exists to find.
Can EMI affect the drone near transformers?
Modern UAVs handle substation electromagnetic interference at survey distances without difficulty. What does cause trouble is the sheer density of steel: compass calibration performed near a transformer tank or structural steel produces heading errors that show up later as an aircraft reluctant to hold position. Calibrate away from steel, verify GPS lock before entering the survey, and treat position-hold behaviour as a health indicator during the flight.
Why does load have to be recorded?
Because resistance heating scales with the square of current, so the same defective connection looks unremarkable at low load and alarming at high load. Without a load figure attached, a temperature reading cannot be compared against last quarter’s survey or against a severity standard — the number is unanchored. This is the most common reason substation thermal programmes accumulate data they cannot actually trend.
Questions about the technology? Talk to our engineers — we reply within 2 business days.