Numerical study of the occurrence of adiabatic shear banding without material failure
Authors
Advisors
Issue Date
Type
Keywords
Citation
Abstract
Segmented chips that arise as a result of formation of adiabatic shear bands are common in high-speed machining of harder alloys. Analytical models describing the mechanics of segmented chip formation have been developed, and these models do not assume or require failure in the material. Most of the finite element (FE) studies of segmented chip formation in the literature always include failure in the material. This thesis presents a numerical study of the occurrence of adiabatic shear banding without material failure. Here, the mechanical properties of hardened 4340 steel was used as a base line, and Johnson and Cook material model for this material were modified to make the material more prone to shear banding. Simulations were performed in two different commercial FE packages: ABAQUS/Explicit and LS-DYNA. The arbitrary Lagrangian-Eulerian (ALE) approach in ABAQUS/Explicit showed shear banding when mass scaling factor was 100. The same simulation did not show shear banding once the mass scaling was removed. Also, the simulations with accelerated thermal softening kept crashing; likely due to the onset of shear banding causing too much mesh distortion. Since steady state was not achieved with any of these simulations, it is hard to say exactly how the intensity of shear banding changes with material properties. Finite element studies in the recent literature, where researchers have shown shear banding through strain softening without material failure using the updated Lagrangian approach, were repeated with ALE in ABAQUS. With the reduction in the parameter controlling the curvature based mesh refinement, shear banding has been successfully simulated in ABAQUS. ALE approach in ABAQUS did not produce the same chip geometry shown in literature because the simulations failed since the mesh was not able to reform to vii a large extent and produce concave shapes as we see in serrated chips. Mass scaling with a factor of 100 did not affect the result. It should be noted that the strain seems to diffuse with the increase in the element size. So, smaller elements are needed to better capture the strains along the PSZ. Lagrangian analysis with a parting-layer approach in LS-DYNA showed shear banding for different plastic properties of the material and no shear banding for others. Observations of the effect of different material and mesh parameters on the shear banding intensity have been made from the simulations and future research are proposed.