This paper was produced for the 2019 NAFEMS World Congress in Quebec Canada
Obtaining a better understanding of the fluid dynamics involved in washing processes in automotive industry allows the identification of design constraints and development of improved processes for high precision parts manufacturing. Simulation of such processes are advantageous to save time and cost involved in the design stage. The current study focuses on the introduction of processes for washing manufactured parts using a particle-based method known as Smoothed Particle Hydrodynamics (SPH). The developed advanced Computational Fluid Dynamic (CFD) solvers are useful in better understanding the fluid dynamics and help to identify design improvements and process optimizations at a reduced cost.
As a mesh-free Lagrangian-based method, the SPH method tracks particle behaviour in the computational domain at each instant of dynamic simulations. In contrast to grid-based methods, SPH is well suited for simulating fluid dynamic problems involving free-surfaces, multi-phase flows, and moving objects with large deformations. The example presented in this study involves a manufactured part that is being washed with high-speed water jet to remove any residual metal chips from the inside. As a primary step in developing the process, knowledge of the flow dynamics through the channel is important. The part design and/or process parameters such as inlet pressure maybe updated based on observations from such simulations. For such problems, the SPH framework presents an excellent option to simulate the process and accurately compute the physical properties.
The presented solver is implemented using Graphics Processing Units (GPUs), which are much more affordable in cost and energy consumption than traditional computations on Central Processing Units (CPU). In addition, the SPH framework is easily parallelizable, allowing simulations to run on multiple GPU devices for problems involving higher number of particles. In addition to the ease of scaling simulations with the problem size, this feature also permits computations at higher particle resolutions as needed for specific physical constraints of a problem. The current study presents results from simulations on a manufactured cooling jacket part, performed using the Predictive-Corrective Incompressible (PCI-SPH) formulation of SPH, and implemented on multiple-GPU devices.
|Date||18th June 2019|