Most maintenance teams hear “thermal monitoring” and picture a handheld IR camera pointed at a gearbox housing. That picture is incomplete.
A sump RTD, a bearing-cap thermocouple, and a housing-skin infrared reading produce three different numbers from the same gearbox, with three different alarm thresholds. Treating any single reading as “the gearbox temperature” is the most common mistake I see in condition-based programs.
What Thermal Monitoring Means on an Industrial Gearbox
Thermal monitoring is the practice of taking temperature readings at defined sensor locations on a gearbox, on a defined frequency, and using the trend to flag friction, lubrication, or cooling problems before they become failures. A single number on a dashboard is not monitoring; a tracked trend at a known location is. The practice sits next to vibration and oil analysis as one of three core condition-based maintenance modalities.
Thermal monitoring is also not the same as a thermal rating, and not the same as the nameplate temperature limits. The rating is the design envelope: how much continuous heat the housing dissipates at rated load.
The limits are the target numbers — typically a 49-82 °C sump-oil normal band, 95 °C alarm, 105 °C shutdown on mineral-oil-filled industrial units. Monitoring is the operational practice that tells you, every shift, where the gearbox sits inside that envelope and which way it’s drifting.
Three Sensor Methods in a Gearbox Thermal Monitoring Program
One gearbox can carry three temperature sensors that all report “gearbox temperature” and all read different numbers — by design, not by error. A sump RTD reads bulk lubricant.
A bearing-cap thermocouple reads bearing-race friction. A housing-skin IR camera reads cast-iron surface temperature. Their alarm setpoints are not interchangeable.
Sump RTD: Bulk Lubricant Temperature
The sump RTD sits in the oil pool. On a heavy-duty industrial unit such as the MTH/MTB series, the right home is a port near the gear-mesh zone.
Insert the probe from the side through the housing near where the ring and pinion engage, and use a long-neck probe that reaches into the bulk fluid. A short probe bolted into the sump drain plug reads the metal skin at the sump wall, already cooled by convection — 5-10 °C of error is normal. The sump RTD is what you compare against the 49-82 °C normal band and the 95 °C alarm setpoint.
Bearing-Cap Thermocouple: Bearing-Race Friction
A bearing-cap thermocouple mounts on the bearing housing of the high-speed input shaft. Bearing temperatures run 11-22 °C higher than sump oil temperatures due to direct rolling friction and load. The alarm setpoint for a bearing-cap probe therefore sits roughly 15 °C above the sump alarm.
A bearing-cap reading of 95 °C on a unit whose sump is at 75 °C is normal; the same 95 °C on the sump probe is the alarm threshold. Do not apply one setpoint to both sensors.
Housing-Skin IR Thermography: Cast-Iron Surface Reading
IR thermography is the third option, and the one to be honest about. Surface temperatures on an enclosed cast-iron housing read 17-28 °C below the internal oil temperature, because the wall mass damps the signal.
A handheld IR scan showing 60 °C on the housing skin can correspond to 80 °C inside the sump, close to where mineral oil starts oxidizing faster. The camera must stay in the same strategic position from scan to scan, in closed rooms without wind flow, or the trend is meaningless. IR works as a fleet-level screening tool and fails as a substitute for an embedded sump probe.
How Thermal Monitoring Thresholds Trigger Action
Two threshold conventions need to coexist in any serious program: a fixed absolute alarm from the nameplate, and a rate-of-rise alarm from the trended baseline. The fixed alarm is the safety net; the trend slope is the leading indicator.
The fixed alarm follows a simple derivation. Pre-alarm sits at 80 percent of the critical value; alarm at the critical value; shutdown at the next step.
If the nameplate critical sump-oil temperature is 100 °C, pre-alarm is 80 °C and immediate-intervention is 100 °C. For most industrial units on mineral oil, this lines up with the 95 °C alarm and 105 °C shutdown convention in published gearbox temperature limits.
On a bearing-cap probe, shift those numbers up by 15 °C. Do not use IR for the trip-the-machine threshold.
A gearbox jumping from 65 °C to 78 °C in under an hour at steady load is in trouble, even though 78 °C is still below the 95 °C absolute alarm. The 13 °C rise is an abnormal-friction signal, weeks earlier than a fixed limit would catch it.
The Arrhenius rule cited in STLE and AGMA guidance is unforgiving: lubricant life halves for every 10 °C rise in temperature for mineral oil. A slow climb from 80 °C baseline to 90 °C operating temperature halves your oil drain interval before any alarm fires. Each threshold crossing maps to a specific maintenance action:
- A rate-of-rise hit triggers an oil sample and an inspection.
- A pre-alarm crossing triggers a load review.
- An alarm crossing triggers a planned shutdown.
That mapping is what makes the data CBM rather than a number on a dashboard.
What Most Thermal Monitoring Programs Get Wrong
The single most common mistake is treating one temperature number as “the gearbox temperature” — running a handheld IR sweep on the housing and calling it monitoring. On an enclosed industrial gearbox, that scan can read 20 °C cooler than the bulk oil inside. You can pass an inspection while the lubricant degrades at twice its design rate.
A working program uses a sump RTD as the primary reading with a known normal band. A bearing-cap thermocouple covers the high-speed shaft on critical assets. IR thermography handles fleet-level screening, and a trending system catches the 13 °C-per-hour climb before it becomes the 30 °C-per-shift failure.
