The selection and simplification of physical models for simulation of abrasive flow machining uniformity

Objectives: With the rapid development of electric vehicles (EVs), the maximum rotational speed of EV motors has reached up to 20 000 r/min. Precision polishing of gear surfaces after grinding has become a promising method for improving the noise, vibration, and harshness (NVH) performance of EVs. Abrasive Flow Machining (AFM) is one of the key technologies for efficiently polishing complex gear tooth surfaces. Fixture design plays a critical role in achieving process objectives, reducing surface ripple and roughness, and minimizing damage to the tooth surface accuracy. This article addresses the trade-off between selecting physical models and balancing the accuracy and computational cost of simulation results. It analyzes the impact of different simulation models on the results, providing guidance for AFM fixture design and offering practical experience for fixture optimization in AFM gear processing. Methods: Simulations are conducted using media with different viscosities, viscosity models, and flow models, within the simplest and most typical slit model. Fluid pressure distribution, velocity vectors, wall shear, and streamline distribution cloud mapsare analyzed to reflect machining uniformity. Based on the conclusions drawn from slit model simulations, the simplest Newtonian fluid—water—is selected as the medium for AFM gear shaft processing simulations. The focus is on the uniformity of streamline distribution in the machining area to optimize fixture design. Results: The analysis of slit model simulation results reveals that different physical models have varying impacts on the outcomes: (1) The selection of viscosity models decisively affects the pressure distribution of low-viscosity media. The type of viscosity and turbulence models has little impact on pressure distribution, but it significantly affects the velocity vector, wall shear, and streamline distribution within the abrasive cylinder. (2) For low-viscosity media: implementing a non-Newtonian fluid model has a significant impact on the pressure distribution. Different flow models show marked differences only in wall shear force distributions. Various viscosity models yield different cloud map distributions, but they produce numerically similar values. (3) For high-viscosity media: simulations with non-Newtonian and Newtonian fluid models show consistent results. However, different flow models greatly influence the results, while various viscosity models lead to changes in all simulation results, except for pressure distribution and streamlines within the slit. Despite these variations, the streamline distribution in the processing area remains largely unchanged. Based on the consistency of streamline distribution, fixture design optimization for the AFM gear shaft is carried out, successfully achieving the goal of eliminating gear "ghost frequencies". Conclusions: Despite variations in the physical models, the simulation results exhibit similar trends in distribution, enabling consistent streamline distribution in the processing area. For low-viscosity media, a non-Newtonian fluid viscosity model with laminar flow simulation can be used, and the selection of viscosity models can be simplified based on the rheological characteristics of the actual abrasive flow medium. For high-viscosity media, setting appropriate viscosity values and using laminar flow simulation with a Newtonian fluid model yields consistent pressure and streamline distribution in the processing area, similar to adding viscosity and turbulence models. The slit model simulation results and AFM gear shaft processing tests both demonstrate that streamline information derived from simple physical models can significantly assist in AFM fixture design. In cases where the physical properties of the abrasive flow medium are uncertain—especially in complex flow paths prone to divergence—using the simplest Newtonian fluid, such as water, with laminar flow simulation can provide a reasonable streamline distribution in the processing area. This approach aids in the analysis of processing uniformity, significantly reduces simulation difficulty and costs, and accelerates the fixture design cycle, ultimately enhancing optimization efficiency.