Dislocation density based material model applied in FE-simulation of metal cutting

Sammanfattning: Simulation based design enables rapid development of products with increased customer value in terms of accessibility, quality, productivity and profitability. However simulation of metal cutting is complex both in terms of numeric and physics. The work piece material undergoes severe deformations. The material model must therefore be able to accurately predict the deformation behavior for a large range of strain, strain rates (>50000 s-1) and temperatures. There exist a large number of different material models. They can be divided into empirical and physically based models. The far most common model used in simulation of metal cutting is the empirical Johnson-Cook plasticity model, JC model. Physically based models are based on the knowledge of the underlying physical phenomena and are expected to have larger domain of validity. Experimental measurements have been carried out in order to calibrate and validate a physical based material model utilizing dislocation density (DD) as internal variable. Split-Hopkinson tests have been performed in order to characterize the material behavior of SANMAC 316L at high strain rates. The DD model has been calibrated in earlier work by Lindgren et al. based on strain rate up to 10 s-1 and temperatures up to 1300 °C with good agreement over the range of calibration. Same good correspondence was not obtained when the model was extrapolated to high strain rate response curves from the dynamic Split-Hopkinson tests. These results indicate that new deformation mechanisms are entering. Repeating the calibration procedure for the empirical JC model shows that it can only describe the material behavior over a much more limited range. A recalibrated DD model, using varying obstacle strength at different temperatures, was used in simulation of machining. It was implemented in an implicit and an explicit finite element code.Simulation of orthogonal cutting has been performed with JC model and DD model using an updated Lagrangian formulation and an implicit time stepping logic. An isotropic hardening formulation was used in this case. The results showed that the cutting forces were slightly better predicted by the DD model. Largest error was 16 % compared to 20 % by the JC model. The predicted chip morphology was also better with the DD model but far from acceptable. Orthogonal cutting was simulated using an updated Lagrangian formulation with an explicit time integration scheme. In this case were two hardening rules tested, isotropic hardening and a mixed isotropic-kinematic hardening. The later showed an improvement regarding the feed force prediction. A deviation of less than 8% could be noticed except for the feed force at a cutting speed of 100 m/min. The time stepping procedure in combination with the mesh refinement seems to be able to capture the chip segmentation quite well without including damage evolution in the material model.Further works will mainly focus on improving the DD-model by introducing relevant physics for high strain rates.