One of the most widely used constitutive model for describing the mechanical behavior at high strain rate of ceramics is the Johnson-Holmquist damage material model (JH-2). JH-2 is the second version of the model proposed by the authors and accounts for strain rate and pressure effects on the yield surface such as the effect of damage on residual material strength and the resulting bulking during the compressive failure. The model requires the knowledge of a number of material dependent parameters many of which cannot be determined directly, and must be inferred.
In the present work, the identification of the material model parameters for the fused silica is presented. The identification procedure consists in an inverse calibration technique where numerical simulations of the experimental tests are iteratively performed, varying the constitutive model parameters, until an agreement between numerical and experimental results is reached. The proposed identification technique is based on the use of the structural optimizer modeFRONTIER. Drop weight test results, relative to different plate geometries, impact velocities and set-up (stacked or un-stacked plates configuration), have been used as reference configurations for the optimization procedure. For each configuration multiple objective functions have been defined and used to determine most of the optimum JH-2 constitutive model parameters by means of a multi-objective genetic algorithm.
Because, for all the drop weight tests, the same range of strain rate is achieved (order of magnitude of 102 s-1), a different test had to be used for the calibration of the related parameter. For this purpose qualitative results of Taylor tests performed at different impact velocities, for which the strain rate is of the order of magnitude of 104 s-1, have been used.
The optimized parameters set has been used to simulate the fused silica behavior in an independent test in which a tile is impacted at a given velocity by a steel spherical projectile and for which a strain rate of the order of magnitude of 103 s-1 is achieved. The good agreement between numerical and experimental results obtained for all the tests demonstrates the validity of the identified set of model parameters. |