GPU-accelerated meshfree computational framework for modeling the friction surfacing process

This study presents a meshfree framework for modeling the friction surfacing (FS) process using the smoothed particle hydrodynamics (SPH) method. The framework leverages GPU computing to address the computational demands of SPH, incorporates optimization techniques such as particle switching and sub-domain division to enhance simulation time efficiency, and integrates artificial viscosity, artificial stress, and kernel correction for simulation stability. A novel criterion for material separation based on joining temperature and critical shear stress is proposed for the rod material, providing accurate results in terms of the deposited material to the substrate during FS. Furthermore, the model is successfully validated to experimental observations of FS of the aluminum alloy AA5083 in terms of axial force, temperature profiles, and deposit geometries, proving the main dependencies of process parameters on deposit width and thickness. The SPH model provides in-depth insight into the deposition mechanisms, particularly illustrated in terms of material flow, deposited material distribution, and rod flash formation, aligning well with experimental findings. The simulations confirm the deposit shift toward the advancing side, where the maximum temperature is also observed. High plastic strain is concentrated in the rod flash and deposit, with higher values on the advancing side than the retreating side. The validated 3D SPH model provides a robust tool for predicting the thermo-mechanical behavior in FS processes, offering insights to advance the understanding and optimization of this deposition technique.