mfree_iwf-ul_cut_gpu

This is the source code to the publication Metal Cutting Simulations using Smoothed Particle Hydrodynamics on the GPU to be published in the Journal of Advanced Manufacturing Technology (JAMT). The purpose of this package is to simulate orthogonal metal cutting operations using Smoothed Particle Hydrodynamics (SPH), parallelized on the GPU using CUDA. This package features some unique characteristics not available in commercial meshfree solvers like the ones available in LSDYNA or ABAQUS, including:

  • Stabilization of the solution using the complete array of techniques presented in Gray, J. P., J. J. Monaghan, and R. P. Swift. "SPH elastic dynamics." Computer methods in applied mechanics and engineering 190.49-50 (2001): 6641-6662.
  • Thermal solver using either Particle Strength Exchange or the Brookshaw Approximation, including thermal contact between the tool and the workpiece.
  • Parallelization on the GPU

Parallelization on the GPU allows for dramatic performance benefits compared to conventional solver packages. Some typical timings on the Tesla P100 are given below. The timings are for 1mm of cutting at a cutting speed of 70m/min.

Double Precision

Resolution Adiabatic Thermal Thermal Contact
~6'000 457 459 807
~150'000 6'489 9'661 11'372

Single Precision

Resolution Adiabatic Thermal Thermal Contact
~6'000 424 465 465
~150'000 4'793 6'450 7'248

Runtimes are in seconds and valid for the thermal solver using Brookshaw approximation.

Result frames can be viewed using ParaView using the legacy VTK format. Some typical result frames are shown below:

Preliminary Benchmark, Rubber Rings, Color is Von Mises Stress:

rings

Preliminary Benchmark, Plastic-Plastic Impact, Color is Equivalent Plastic Strain:

impact

Orthogonal Metal Cutting, about 6'000 Particles, Color is Equivalent Plastic Strain::

lores_cut

Orthogonal Metal Cutting, about 500'000 Particles, visualization using SPLASH, Temperature Field:

hires_cut

mfree_iwf-ul_cut_gpu was tested on various versions of Ubuntu Linux. The only dependency is GLM. Make files for both a Release version and a Debug build are provided. Tested under NVCC with the GCC suite as host compiler, but any C++11 compliant host compiler should suffice. mfree_iwf-ul_cut_gpu was devleloped at ETHZ by the following authors

mfree_iwf-ul_cut_gpu is free software and licensed under GPLv3