Motor and gearbox life

Life is not a property of a motor. It is the shortest of five separate wear processes, and duty decides which one you get.

Two mechanisms at the same torque can differ in service life by more than an order of magnitude. The difference is never in the torque figure.

The five clocks

They run at the same time. The first one to finish is your life figure, and it is usually not the one that was specified against.

Brush wear
In a brushed motor the brush is the consumable. Wear is driven by sliding distance, current density at the contact, commutation quality, and atmosphere. Sliding distance is the whole story on speed: a motor at 6,000 rpm wears its brushes at roughly twice the rate of the same motor at 3,000 rpm for the same running time. Carbon brushes depend on adsorbed moisture to form the lubricating film, so a dry or sealed atmosphere can multiply the wear rate. Precious-metal brushes behave differently again and are intolerant of high current. Every start and every reversal adds a burst of arcing that no hours figure captures.
Grease life
In a gearhead running well inside its torque rating, the grease usually ends before the gears do. Base oil is lost by evaporation and oxidation, and both accelerate sharply with temperature. As a working rule for a lubricant, the rate roughly doubles for every 10 to 15 degrees C of rise, so a gearhead sitting next to a heat source can lose most of its expected life without ever seeing an unusual load.
Gear tooth fatigue
Two distinct mechanisms: bending fatigue at the tooth root, and surface pitting on the flank. Both count stress cycles per tooth, not output revolutions. A first-stage pinion in a 100:1 train sees roughly 100 times as many engagements as the output gear, which is why the small early gears are usually the ones found damaged, and why a peak torque that appears for a fraction of a second still matters.
Bearing life
For a ball bearing, rated life scales with the inverse cube of the load: double the load and rated life falls to about an eighth. The load usually comes from the mechanism rather than the motor, through belt tension, a lead screw preload, a gear reaction or a side load on the output shaft. A sleeve bearing does not fail by fatigue at all; it wears, and it wears faster with contamination, side load and marginal lubrication.
Insulation and magnet
Winding insulation has a thermal life that falls steeply with temperature, roughly halving for every 8 to 10 degrees C above its rating. This clock only matters when the duty is aggressive enough to heat the winding, but when it does matter it ends the motor abruptly rather than gradually.

Why torque alone never predicts life

Torque sets current. Current sets loss. But none of the five clocks above runs on torque.

A torque figure tells you the motor will not stall. It does not tell you how far the brushes will slide, how hot the grease will get, how many stress cycles a tooth will see, or how many arcing events the commutator will survive. Those come from speed, duty, ambient, direction changes and total operating time. Consider two applications at an identical continuous output torque: one runs two seconds in forty at 200 rpm at the output, the other runs continuously at 200 rpm in a 50 degree C enclosure. The second accumulates roughly twenty times the sliding distance and runs its lubricant far hotter. The same motor is a twenty-year part in one and an annual replacement in the other, and the specification sheet says nothing about the difference.

This is why a life requirement stated as hours is nearly useless on its own, and why a requirement stated in cycles of the mechanism, with the duty that produces them, can be screened seriously.

T_rms = √( Σ T_i² t_i / Σ t_i )

T_rms
root mean square torque over the cycle, the value to compare against the continuous rating
T_i
torque during segment i
t_i
duration of segment i, including the off segments at zero torque
Include the idle segments with T = 0 and their real duration. Leaving them out is the most common way a duty calculation produces a motor twice the size it needs to be.

P_out = T × 2πn / 60

P_out
mechanical output power, W
T
output torque, N·m
n
output speed, rpm
Divide by the gearhead efficiency and then by the motor efficiency at that point to get electrical input. Working the other way round, from an input power figure, is where most sizing errors begin.

Building a duty profile

Half an hour with a stopwatch and a current probe replaces a year of argument. Build the table below and the screening becomes arithmetic.

The shape of a usable duty profile. One row per distinct segment of the cycle.
Segment Duration s Output speed rpm Output torque mN·m Direction Notes
Breakaway0.20PeakForwardTorque to start from rest, which is usually several times the running torque
Accelerate0.30 to fullHighForwardInertia plus load. Sizes the peak, not the thermal case
Run4.0ConstantRunningForwardThe segment that sets temperature if the duty is high
Decelerate and stop0.3Full to 0BrakingForwardNote whether it stops against a hard stop
Return4.8ConstantLowerReverseCount reversals per hour separately; they wear brushes disproportionately
Idle50.400Include it. This is what makes intermittent duty possible
  1. Measure the breakaway torque, not just the running torque

    From rest, cold, at the worst position in the mechanism and the worst end of the tolerance stack. Breakaway commonly exceeds running torque by three to five times, and it decides whether the machine starts on a cold morning at the low end of the supply.

  2. Count the events, not just the time

    Starts per hour, reversals per hour, and stalls per year. Each of these has its own effect on brush and commutator wear that no hours figure represents.

  3. Record ambient at the motor, not in the room

    Inside the enclosure, next to the motor, after an hour of running. Every one of the five clocks is temperature-sensitive, and the enclosure is usually 15 to 30 degrees hotter than the room.

  4. Convert the life target into mechanism cycles

    Ten years at forty cycles a day is 146,000 cycles. Multiply by the output revolutions per cycle and by the ratio and you have motor revolutions, which is the number every wear process actually cares about.

  5. Record the supply as it behaves under load

    Nominal voltage is not enough. A battery that sags to 80 percent during breakaway changes the whole calculation, and a PWM drive changes the commutation picture as well.

If units are already failing, start with the evidence

Send the duty profile and the description. The plausibility review is free and it says which of the five clocks your symptom is consistent with before anything is shipped or torn down.

Motor failure analysis