Thermal Camera vs Tightness Check vs Load Test: Hotspot Method Decision Matrix

Why this matters

Heat at a connection is the single most common precursor to an electrical fire, and it has one root cause: resistance where there should be none. A loose lug, a corroded splice, an overloaded conductor, or a failing breaker all show up as temperature rise, but they reveal themselves under different conditions. A thermal camera sees the heat but only while the circuit is loaded. A tightness/torque check finds the mechanical cause but tells you nothing about whether the connection is overheating right now. A load test forces the fault to appear when intermittent. Knowing which method to lead with stops you from torquing a lug that was fine and missing the overloaded conductor next to it, or from clearing a panel that fails the next time the dryer runs.

The options

Thermal (infrared) camera. A non-contact imager that maps surface temperature. Used live, under load, with the panel cover off. It localizes the hottest point and lets you compare identical phases or identical breakers side by side. It measures effect (heat), not cause.

Tightness / torque verification. A calibrated torque screwdriver or wrench applied to terminations to confirm they meet the manufacturer's installed torque. De-energized work. It addresses the most common cause (loose connection) directly but cannot find an overloaded conductor or a marginal connection that is technically tight but oxidized.

Load test. Deliberately loading the circuit to a known current (running the actual appliances, or a resistive load bank) to force a marginal or intermittent fault to dissipate heat and reveal itself. Live test. It surfaces faults that are invisible at idle.

When the thermal camera wins

Lead with thermal imaging on any "warm panel," "burning smell," or periodic-maintenance survey where the system can run under real load. With the dead-front removed and the building under normal or deliberately elevated load, scan every termination, breaker, lug, and splice. Interpret by temperature rise above a reference, not absolute temperature: compare a suspect connection to an identically loaded one on another phase. A rise of roughly 1 to 10 C over the reference is a connection to monitor; 11 to 20 C over reference is a probable deficiency to schedule; over 20 C, and especially a phase-to-phase difference where ambient and load are equal, is a serious connection requiring prompt correction per established infrared survey severity practice. Thermal is also the fastest way to spot an overloaded conductor (the whole length is warm, not just the connection) versus a loose lug (a sharp local hotspot). Its limit: no load, no heat. A camera scan of an unloaded panel finds nothing, and emissivity errors on shiny bus and lugs can fool a careless read, so target the insulated or oxidized surfaces.

When the tightness check wins

Lead with torque verification when thermal has localized a hotspot at a termination, or during commissioning and post-fault rework where connections were disturbed. De-energize, lock out, and re-torque to the manufacturer's value marked on the breaker, lug, or terminal block. Aluminum terminations and large-conductor lugs are the prime suspects: they relax and cold-flow over time, and a connection that was correct at install loosens. Do not simply "snug" everything; over-torque crushes strands and is as damaging as under-torque. If a lug will not hold torque, shows discoloration, or the conductor is annealed (loses springiness), the termination is past tightening and needs the conductor cut back and re-landed or the device replaced. The tightness check is decisive for the local hotspot but blind to overloaded runs and to connections that read tight yet are internally oxidized.

When the load test wins

Lead with a load test when the complaint is intermittent and nothing is hot during a routine scan. A breaker that trips "sometimes," a connection that arcs only when the heat pump and water heater coincide, or a feeder warm only on the hottest afternoons will not reveal itself at idle. Bring the circuit to its real working current (run the actual loads, or apply a calibrated load) and then re-scan with the thermal camera while loaded. This combination, load to force the fault plus thermal to localize it, catches the marginal connection that passed a cold inspection. The load test also distinguishes an overloaded circuit (current at or above rating, conductor warm end to end) from a defective connection (current normal, heat at one point). Its limit: it requires you to safely impose and sustain load, and it does not by itself identify the mechanical cause once located.

Field decision flow

1. Can the system run under real or elevated load right now? If yes, scan with the thermal camera first and rank hotspots by rise above an identically loaded reference.
2. Is the hottest point at a single termination (sharp local spike)? De-energize, lock out, and verify torque; replace the conductor end or device if it will not hold or shows annealing/discoloration.
3. Is the heat spread along a whole conductor (not a point)? You have an overloaded or undersized run, not a loose lug. Confirm load current with a clamp against conductor ampacity; correct by load-shedding or upsizing, not torquing.
4. Is the complaint intermittent with nothing hot at idle? Impose a load test to working current, then re-scan thermally to force and localize the fault.
5. Document baseline temperatures so the next survey has a trend, not a single snapshot.

References

• NFPA 70B, Standard for Electrical Equipment Maintenance (infrared thermography surveys, connection inspection, and temperature-rise criteria).
• NFPA 70E, Standard for Electrical Safety in the Workplace, Article 130 (energized work, arc-flash boundary, PPE selection).
• NFPA 70, National Electrical Code, Article 110.14 (electrical connections and termination torque) and Table 310.16 (conductor ampacity).
• NETA MTS (Maintenance Testing Specifications), thermographic survey severity guidelines.
• UL 486A-486B, Wire Connectors (connector torque and termination performance).