A Model for Process-Based Crash Simulation

Manufacturing of a bumper system from aluminium extrusions often involves series of forming operations performed in the soft W temper condition, and then artificially age-hardening the components to the material's peak hardness T6 condition. It is perceptible that correct numerical representation of the crash performance of the resulting systems must rely upon a geometry obtained from a model following the process route, i.e. including simulation of all major forming operations. However, the forming operations also result in an inhomogeneous evolution of some internal variables (among others the effective plastic strain) within the shaped components. Here, results from tensile tests reveal that plastic straining in W-temper results in a significant change of the hardening curves (alloy and ageing-dependent increase or decrease in strength) as a function of plastic pre- straining. In addition, the tests revealed that the plastic deformation led to a reduction of the elongation of the T6 specimens. These data were obtained by uniaxial stretching of plates in the W temper to different levels of plastic deformation, sub-sequent artificial ageing to obtain T6 characteristics, machining of uniaxial tensile test coupons and, finally, testing until failure. In the present work, these process effects have been included in a user-defined elasto-viscoplastic constitutive model incorporating a state-of-the-art anisotropic yield criterion, associated flow rule, non-linear isotropic and kinematic hardening rules, a strain-rate hardening rule as well as some ductile fracture criteria. To demonstrate and asses the modelling methodology, a 'through-process analysis' of the uniaxial tensile tests is performed. The pre-stretching of the plates in W temper is modelled with shell elements having an initial random Gauss- distributed thickness and stretched to different levels of plastic strain – comparable to the experimental ones - using the explicit solver of LS-DYNA. Then, uniaxial tensile specimens are trimmed from the deformed plates using the trimming option available in LS-DYNA. Next, strain dependent T6 properties are specified, after which the resulting specimens are stretched until instability and failure. Finally, the model assumptions are assessed by comparing engineering stress-strain curves obtained from the simulations and experiments.