The usual answer, repeated in a great many supplier FAQs, is that plastic gears are for light loads and that high torque means metal. That answer costs design engineers real options, because plastic gears do carry substantial torque. What they cannot do is carry it in the same envelope, at the same speed, for the same number of cycles and at the same temperature as steel. Torque on its own is not the variable that settles the question, and treating it as one produces both unnecessary metal and unexpected failures.

The Material Is Weaker, the Geometry Is Not Fixed

The material difference is large and easy to state. A design allowable for acetal in tooth bending sits around 20 to 30 N/mm² depending on grade and the cycle count it is read at. A hardened steel gear of comparable size allows roughly 220 N/mm². Tooth for tooth, the plastic is about ten times weaker.

That looks decisive until the geometry is allowed to move, which in gear design it always is. For a bending-limited tooth, the torque a gear carries rises with the face width and with the square of the module. Double the face and capacity doubles. Double the module and it quadruples. The ten-to-one material penalty can therefore be recovered in a gearbox that is willing to be larger, and in a small reducer there is often room to recover most of it.

A worked check makes the scale concrete. Take a module 1.5 acetal gear with 20 teeth, a 10 mm face and a 30 mm pitch diameter, carrying 0.8 N·m at moderate speed. The tangential load is about 53 N, and with ordinary service factors the root sees roughly 16 N/mm² against an allowable near 25, a safety factor of about 1.6. Now double both the module and the face width. The same tooth stress carries about eight times the torque, near 6 N·m, because torque rises with face width and with module squared. Six newton metres from a plastic gear is not a marginal figure in a gearbox this size.

So the honest answer is that plastic gears handle high torque by being large. The constraint is the volume the housing can spare, not the material alone. Surface stress and running temperature usually stop the design before the bending limit does, so treat that figure as an upper bound rather than a target.

Speed Converts Torque Into Heat

The same gear at the same torque is a different problem at twice the speed. At low speed the tooth root governs, and the calculation above applies. At high speed the limit becomes the temperature of the mesh itself, because the sliding friction at the tooth flank has less time to escape and the surface of the polymer softens. Once the surface softens, friction rises, which generates more heat, and the tooth fails by thermal runaway rather than by stress.

That is why a torque figure quoted without a speed is not a rating. Acetal is commonly limited to a bulk or surface temperature around 60 degrees Celsius, and PEEK to about 80, and a gearbox bolted to a warm motor has less margin to that ceiling than the ambient temperature suggests. The practical rule is to size the tooth for bending at low speed, and to size the mesh for heat at high speed, which are two different calculations with two different answers.

Time Is a Limit as Well

Steel has an endurance limit: below a certain stress it can run indefinitely. Plastics generally do not. Their stress-life curves keep falling at very high cycle counts, so a plastic gear has a finite life at every stress level, and the allowable has to be read at the number of cycles the product will actually accumulate. A gear turning at 100 rpm for an eight-hour shift collects about 48,000 cycles a day and several million in a year, which takes even a light-duty mechanism into the high-cycle region quickly.

Sustained load adds a second mechanism that a torque test will not reveal. Thermoplastics creep, so a tooth held under constant load slowly changes shape. A mechanism that holds a load correctly on the bench can flow open over a long weekend, and the failure will be attributed to the motor or the linkage rather than to the gear.

Shock Is What Actually Breaks Them

A steady torque rating says very little about what happens when the mechanism jams or the load arrives with a bang. Standard practice applies an application factor to the required torque, from about 1.0 for uniform drive to more than 2.0 for heavy shock, and plastic gears sit at the unforgiving end of that table. The teeth strip, the motor then spins freely, and there is no warning between correct operation and failure.

How to Raise the Ceiling

Four measures, in order of cost. Increase the module, which raises capacity with its square and is usually the cheapest change at the design stage. Increase the face width, which raises capacity in proportion and costs axial space. Change the grade, since a glass-filled or PEEK material roughly doubles the allowable surface stress and raises the temperature ceiling with it. And put metal only where the stress is highest, at the output stage or on the motor pinion, keeping the rest of the train plastic for quietness and cost.

What to Send

Four numbers decide this: the output torque, the output speed it is needed at, the duty in cycles per hour and hours per year, and the temperature the gearbox will actually see rather than the room temperature. Add the space the housing can give the gear train, because that is what sets the practical ceiling. With those four, a tooth can be sized to a stress it will survive for the life of the machine, rather than to a torque figure that says nothing about speed, time or heat.