FE-Simulation of Metal Cutting Processes

Licentiatavhandling, 2021

This thesis deals with the finite element (FE) simulation of machining processes. Realistic simulation of metal cutting processes enables a more resource-efficient machinability assessment for a given material in terms of cutting forces, chip shape and tool wear at different ranges of cutting conditions. However, the material behaviour during machining needs to be presented properly in the simulations in order to make realistic FE-predictions. Implementation of an appropriate material model with well-tuned parameters is crucial for obtaining reliable FE-simulation results. The performance of different material models including various strain and strain-rate hardening and thermal softening characteristics is investigated for cutting simulation of carbon steels. In order to determine the material model parameters, a new calibration method is proposed. The method uses data from machining experiments - measured forces and chip thickness - to predict the stress, strain, strain-rate and temperature distributions in the primary shear zone during machining. By using these distributions, the parameters of the material model can be calibrated. Since this approach benefits from a semi-analytic model that directly incorporates the experimental results of machining tests, the calibrated parameters are more suitable for machining simulations as compared with those obtained using other methods, for example, conventional tensile/compression or split-Hopkinson pressure bar (SHPB) tests.

Chip formation is governed by the thermo-mechanical properties of the workpiece material, tool geometry and cutting conditions. Hence, the chip can take different shapes such as continuous or serrated depending on the severity of the cutting process for a given material. In addition to a well-defined material model, the reliable prediction of chip shape in machining demands the implementation of an appropriate damage model. In this work, two different damage models are investigated - referred to as local and nonlocal damage models. The difference between these two models is that one (non-local damage model) includes the gradient effect into the formulation influencing the progression rate, whereas the other one does not. The performance of these damage models is evaluated for simulation of damage evolution during tensile and SHPB tests.