Plastic Gears

Material Modeling

The analysis of plastic gears requires material models capable of capturing high compliance, large deformations, temperature sensitivity, and rate-dependent behavior. MD-Lab develops automated finite-element workflows and calibration tools to accurately represent polymer material behavior in loaded tooth contact analysis and gear-pair simulations.

For loaded tooth contact analysis, the Lab has developed neural-network surrogate models trained on finite-element simulations of polymeric gears. The workflow combines automated geometry generation and meshing with Abaqus simulations to calculate root and contact stresses, force distribution, and static transmission error throughout the meshing cycle.

A fully connected neural network trained on finite-element STE curves achieves a mean absolute percentage error of only 0.49% on previously unseen datasets. This accuracy is approximately one order of magnitude better than conventional analytical solvers, while maintaining computational costs comparable to simple polynomial approximations and several orders of magnitude lower than direct finite-element simulations.

In parallel, MD-Lab develops automated material-parameter identification workflows for polymer constitutive models. Experimental measurements are used to automatically generate and execute CAE simulations, extract response curves, compare them against experimental data, and iteratively optimize material parameters using Bayesian optimization techniques.

The current workflow includes linear elasticity, Voce plasticity, and Perzyna rate-dependent constitutive modelling. The calibrated models are subsequently employed in gear-pair simulations to evaluate different polymer materials and manufacturing processes.

Publications

• Papalexis, C., Sakaridis, E., Terpos, K., Kalligeros, C., Tsolakis, A., & Spitas, V. (2025). Neural network surrogates for finite element models in loaded tooth contact analysis of polymeric gears. Mechanism and Machine Theory, 214, 106127. https://doi.org/10.1016/j.mechmachtheory.2025.106127

NVH Performance

MD-Lab investigates the dynamic behavior of plastic gear pairs using a surrogate-assisted dynamic modelling framework. Neural-network predictions of static transmission error are transformed into time-varying mesh stiffness and integrated into lumped-parameter dynamic models, enabling efficient evaluation of dynamic transmission error, gear-body acceleration, and dynamic load amplification across broad operating conditions.

Research activities focus on polymer materials such as PA66, POM, and PEEK, and compare their performance with equivalent metallic gearsets. Simulations show that plastic gears may exhibit higher static and dynamic transmission-error amplitudes because of their lower stiffness; however, they also produce lower vibration levels and reduced dynamic load amplification.

In several operating conditions, metallic gearsets exhibited dynamic factors up to four times higher than those of polymer gearsets, highlighting the inherent damping advantages of polymeric materials.

Publications

• Papalexis, C., Sakaridis, E., Terpos, K., Kalligeros, C., Tsolakis, A., & Spitas, V. (2025). Neural network surrogates for finite element models in loaded tooth contact analysis of polymeric gears. Mechanism and Machine Theory, 214, 106127. https://doi.org/10.1016/j.mechmachtheory.2025.106127

Additive Manufacturing

Additive manufacturing is becoming an increasingly important approach for plastic gear research, enabling rapid prototyping, low-volume customization, complex tooth geometries, non-involute and asymmetric gear designs, topology optimization, and lightweight lattice structures.

For polymer gears, these opportunities must be balanced against challenges related to surface finish, dimensional accuracy, shrinkage, and load capacity compared with conventional manufacturing methods.

MD-Lab investigates the metrology and functional accuracy of polymer spur gears produced through various additive-manufacturing techniques. Research results show that FDM-printed gears exhibit significantly larger dimensional deviations than metallic gears, typically corresponding to ISO quality classes around Q11-Q12, compared with the tighter tolerances commonly achieved in conventional metallic gear manufacturing.

Similarly, gears produced using material extrusion (MEX-TRB/P/ABS) and powder bed fusion (PBF-LB/P/PA22) demonstrated ISO quality classes around Q12 or higher, remaining well below the accuracy levels typically required for high-precision metallic gear applications.