COMPUTATIONAL MODELING OF METAL-MATRIX COMPOSITE-MATERIALS .3. COMPARISONS WITH PHENOMENOLOGICAL MODELS

The mechanical behavior of particulate reinforced metal matrix composites, in particular an SiC reinforced Al-3 wt% Cu model system, was analyzed numerically and analytically. In this, the third article in a series, the results of the computational micromechanics modeling are compared with those of simpler and more approximate analytical/numerical models. The simpler approaches considered use phenomenological theories of plasticity and power-law strain hardening. Models that predict overall composite behavior make use of a result, valid for incompressible materials in small strain, that both pure matrix material and composite harden with the same hardening exponent. Results of micromechanical simulations, with power-law slip system hardening, show that in a very approximate sense over restricted strain regions, power-law hardening is preserved with the power-law exponent tending to increase with volume fraction. The results of the computations presented in the previous article are compared with the predictions of one such analytical/numerical model, where the matrix hardening function is fitted to the unreinforced polycrystal stress-strain response. This model employs the self-consistent method to quantify strengthening. There is good agreement between the computed and predicted results. Simulations are performed using existing reinforcement geometry but replacing the physically based crystal plasticity theory with the phenomenologically based J2 flow theory. The results are in good qualitative agreement with those of the original crystal plasticity simulations at both the microscale and the macroscale. Deformation patterns in the J2 flow theory composites are smoother and tend to be less localized than those in the crystal plasticity composites; however, these features depend strongly on volume fraction and morphology. The J2 flow theory composites display power-law exponents whose dependence on overall strain, volume fraction and morphology are much more easily characterized than in the crystal plasticity case.