Plastic gears tend to be based on the same geometry as conventional gears. The question is, do they behave the same when subjected to similar stresses and strains?

This article, linked below, in Gear Solutions magazine by W.Kassem, M.Oehler and O.Koch investigates this issue. Tooth root fillet geometries with trochoidal, elliptical, fully-rounded and bionic profiles for steel-plastic pairings were examined on two different gear sizes under a variety of simulations. So, what were the conclusions?

Unreinforced polyamide (PA46) using an electro-plastic material model and short fiber reinforced polyamide (PA46GF30) using an anisotropic electro-plastic material model were used.

A study was performed for each of the profiles to identify the optimum parameters of every fillet geometry for low bending stress. A significant stress reduction was achieved for all profiles through optimization, although no clear optimum root profile could be determined.

A clearly different behaviour in the optimum parameters between the fiber reinforced and the non-fiber reinforced material was detected. The two must therefore have a different geometry depending on the material and gearing variant, the highest stress reduction was achieved with the help of the fully rounded and bionic tooth root.

Investigation wear, the finite element analysis was coupled with transient wear simulation on gear variant made of short fiber reinforced polymer. For this variant, the fully rounded tooth root exhibited significantly higher stress over the running time. Hardly any difference was found between the other tooth root geometries. This was due to the highest stress shifting to the wear notch – which was almost identical between the other designs.

Concluding, the researchers reported that ‘ Analytical methods for gear design reach their limits when inhomogeneous materials are considered. For this reason, it is recommended that numerical methods be used to optimize the tooth root geometry of gears made of fiber-reinforced plastic, where complex material behaviour can be represented. It should also be noted these are purely simulative studies, and the results should be supported with experimental verifications in the future. 

The link here contains the full account of the methods, simulations and findings.