Thermal-Tracking Lag and Motor Hotspot Development

A motor hotspot does not announce itself. Winding insulation degrades as a function of temperature and time, and the damage accumulates whether or not anyone is watching. The failure mode examined here is thermal-tracking lag: the interval between successive temperature observations is long enough that a developing hotspot reaches a damaging temperature before the next reading is taken. The asset does not fail because the sensor was wrong. It fails because the sampling window was too wide to resolve the trend.

The mechanism

Consider a 3.3 kV induction motor driving a SAG mill pinion, or a conveyor drive gearbox on a decline. Under normal load the winding temperature sits at some steady value determined by ambient, load, and cooling. A hotspot begins when local conditions change: a blocked cooling duct, a degraded insulation patch, a bearing beginning to run warm, a phase imbalance raising copper loss in one winding. The local temperature rises, but the bulk of the machine does not change much. A single embedded PT100 or a thermal overlay reports the average, not the peak.

The physics is a first-order thermal step. A hotspot with a time constant of several minutes to tens of minutes climbs toward a new equilibrium. If it is sampled once per survey round — say once per shift, or once per route walk — the samples are separated by hours. Between two readings the temperature can rise through the insulation class limit and into the region where the rate of degradation accelerates. The next reading may show an elevated value, but by then the cumulative thermal dose has already been delivered. The hotspot is not detected; it is discovered after the fact.

The same lag applies to control valves and their actuators, where a sticking stem raises motor current and winding temperature slowly, and to conveyor idler bearings where a failing grease charge produces a gradual rise that a monthly thermographic survey will miss entirely.

Why conventional alarms do not cover this

A high-temperature trip is a demand-mode function. It acts when a threshold is crossed. It does not act on the rate of change, and it does not act on the shape of the trend. In IEC 61511 terms, a threshold alarm has diagnostic coverage against the gross fault — the sudden short, the total loss of cooling — but poor coverage against the slow, cumulative rise that characterises hotspot development. The dangerous failure is the one that never crosses the trip point until it is too late, or crosses it so briefly between samples that no alarm is annunciated.

Diagnostic coverage is the fraction of dangerous failures detected by the diagnostic function. A periodic survey with a long interval has low coverage against a slow ramp because the diagnostic is only active at the instant of the survey. The proof-test interval and the diagnostic coverage are coupled: a diagnostic that runs once per shift cannot cover a failure mode whose time constant is shorter than a shift. This is the gap. It is not a sensor accuracy problem. It is a coverage problem.

Thermal mapping as a diagnostic function

The engineering response is to treat thermal mapping as a diagnostic function with a defined proof-test interval, not as a monitoring convenience. That means three things.

First, the temperature field must be resolved spatially and temporally. A single bulk sensor is a point measurement. A mapped field — multiple sensing points, or a model that infers the hotspot from load, current, and cooling state — gives the diagnostic something to compare against. The comparison is what detects the deviation.

Second, the diagnostic must run at an interval shorter than the hotspot time constant. If the hotspot reaches its limit in twenty minutes, a survey every eight hours provides no coverage. The interval is set by the failure mode, not by the maintenance roster.

Third, the diagnostic must be proof-tested. IEC 61511 requires that the proof-test verify the diagnostic function detects the failure mode it claims to cover. For thermal mapping, that means injecting a known thermal perturbation — a calibrated heat source, a simulated load step — and confirming the mapping resolves it within the specified interval. A proof-test that only confirms the sensor reads ambient proves nothing about hotspot detection.

What this preserves

The purpose is not to claim the sensor is a safety device. It is not. The purpose is to give the safety lifecycle a diagnostic function with quantified coverage against a specific failure mode, so that the residual risk calculation reflects reality. When thermal-tracking lag is closed, the hotspot is detected while it is still a trend, not a failure. Maintenance is planned against a known rate of change rather than triggered by a trip. Uptime improves because the intervention happens before the winding fails, and safety improves because the failure mode is covered rather than assumed away.

The same logic applies across the plant: SAG mill motors, conveyor drives, gearbox bearings, control valve actuators. Each has a thermal time constant. Each has a survey interval. The engineering task is to make the second shorter than the first, and to prove it.

Getting started

Identify the rotating and actuated assets where a hotspot would be consequential. For each, estimate the thermal time constant of the hotspot failure mode. Compare it to the current survey interval. Where the interval exceeds the time constant, the diagnostic coverage is inadequate regardless of sensor quality. Define a proof-test that injects a known thermal step and confirms detection within the interval. Record the coverage in the safety lifecycle documentation. The gap is usually not in the sensor. It is in the interval.

Source: internal-backlog — grounded in S5 engineering design notes, not a current market event.

Talk to S5 System about your site

S5 System designs and manufactures ultra-low-power industrial IoT sensors for mining, agriculture, infrastructure and utilities.

Discuss your IoT monitoring project — book a call.