On the Development of CFD Methodology for Free-Falling Varnish Stream Modeling to Support EV Motor Manufacturing

When manufacturing the stators in EV motors, stator wires are first coated with a layer of resin to provide primary insulation. After winding, impregnating varnish fills all voids within the windings and between the windings and lamination. In addition to electrically insulating the copper wires, another function of the varnish fill is to mechanically secure the copper wires from movement. The process is not complicated in terms of physics. In essence, the mechanics of the varnish flow is the balance of inertia force, viscous force, gravity and surface tension. However, understanding the fluid dynamics of the varnish flow is critical to predicting the quality of the varnish fill, which has a tremendous impact on motor performance. With the advancement of computational fluid dynamics (CFD), the industry can benefit greatly if the varnish trickling process can be tuned, without physical tryouts, to achieve optimal fill. The intent of this research is to develop the computational methodology to study the varnish flow, so that its flow pattern can be simulated and predicted in the varnish trickling process. Because varnish flow by nature is a two-phase flow, special attention is needed to treat the large density difference between air and varnish, as well as the surface tension acting on the phase boundary. Furthermore, the well-known Plateau-Rayleigh instability associated with free-falling streams poses great challenges to the numerical method. In this research, two distinctive computational methodologies, FVM (Finite Volume Method) and SPH (Smoothed-Particle Hydrodynamics), are utilized to model the varnish free-falling streams. The end results are compared with analytical solutions and experimental observations to verify their accuracy. Also included is the comparison of numerical artifacts introduced by the two CFD methods. Another valuable achievement is the utilization of experimental results to reduce the costly computation time for reaching steady-state solution.