Thermography of an electrical installation is only meaningful under two conditions: the equipment must be energised and carrying representative load, and the internal connections must be visible. Both were arranged in advance. The client’s own contractor removed the protective dead fronts from all eleven boards before the inspection and remained present throughout, attesting the load state at each board, so the internal terminations, busbar connections and outgoing conductors were in view rather than hidden behind a cover. On one board a measured load current was taken at the time of capture, confirming that the panel was genuinely working rather than idling. Two boards, serving a kitchen and a staff kitchenette, could only be surveyed with some equipment out of service. That reduces the sensitivity of the survey on those boards, and the limitation is recorded openly against every image concerned rather than left for the reader to infer.

Environmental conditions were recorded at each board position and the radiometric parameters set per thermogram accordingly, which matters more than it might appear. The premises spanned a public entrance lobby, bars, kitchens and a plant room, and the recorded ambient varied by nearly seven degrees across those positions. Reflected apparent temperature was measured at each board rather than assumed, and a single emissivity value appropriate to the dominant surfaces, painted steel enclosures, moulded device casings and thermoplastic insulated conductors, was applied consistently across the job. Industry-grade thermal imaging and environmental measurement equipment was used throughout, and all data collection on energised equipment ran under the documented safety protocol.

Classification is deliberately not based on absolute temperature. Each measurement is assessed as a temperature difference above a baseline: preferably a comparable device or conductor under similar load within the same image, which cancels the shared measurement error between the two, and where no comparable component exists, the recorded ambient air, with the wider bands that a weaker reference demands. Differences below two degrees are treated as instrument tolerance and normal variation. A measurement becomes a classified anomaly at four degrees above its baseline. Severity and operational consequence are then assessed separately, because a large temperature difference does not automatically mean a serious problem, and a modest one does not automatically mean none.

The installation was operating normally at the time of inspection. Across eleven boards and fifty thermograms, nothing carried a safety consequence, and no measurement came close to the temperature limits that apply to terminations and thermoplastic insulation, the highest reading on the site sitting at roughly two thirds of that limit. Most of the heat recorded was exactly what a working installation should produce. Residual current devices ran warmer than plain protective devices, as they always do, because their internal detection circuitry draws current continuously whenever the circuit is energised. Conductors ran warmest where they grouped and bunched and cooled along their separated lengths. Devices stepped evenly across an assembly rather than one standing apart from its neighbours. That distributed, symmetrical pattern is the signature of load, not of a fault, and it is recorded as normal operation with the measurement stated so the reader can see it.

Two conditions did warrant attention, both at the board serving the main kitchen, which unsurprisingly returned the highest device temperatures of any board on site. The first concerned three live conductors leaving that enclosure and passing through the building fabric to a circuit that could not be traced from the accessible aspect. They ran consistently warmer than comparable conductors within the same enclosure, with the warmth continuous along their length and no concentration at any termination, which places the heat in the current being carried rather than in resistance at a joint. Classified Medium severity with an operational continuity consequence, this finding arises from what could not be established rather than from what was measured: the conductor sizing, the installation method and the circuit loading were all unknown, so no assessment against design capacity was offered, and verification was referred onward.

The more significant finding was an ancillary control panel mounted immediately adjacent to that board, a unit that proves the kitchen ventilation is running before gas is supplied. A localised hotspot on its front casing stood more than twenty degrees above the recorded ambient, confined to a discrete area rather than spread across the casing, with visible discolouration of the surface coincident with the thermal signature. The measured value is the outside of a closed assembly, so the component generating the heat inside it is necessarily hotter by a margin that cannot be determined without access. Classified High severity with a degradation trajectory consequence, the recommendation was that the unit be examined rather than simply monitored, and that the manufacturer’s guidance be sought on its continued serviceability.