Multiple strain localization is a first step towards the fragmentation of ductile structures subjected to dynamic stretching. Difficulty in the prediction of this phenomenon arises from its intrinsic random nature and the lack of objective characterization methodologies and criteria.
In the present work, a fundamental statistical framework, based on the work of S.O. Rice (1944), is developed to analyze the neck pattern development in metallic bars subjected to dynamic stretching. For that purpose, finite element simulations are conducted on a round bar with initial cross-section fluctuations modeled via a white noise.
As important outcomes of the present work, it is shown that phases of each perturbation mode remain independent random variables until strain localization occurs. As a consequence, the following features have been characterized in an objective way: (i) time evolution of the number of local minima of the cross-section area, (ii) strain localization time, (iii) number of necks.
From simulations, the effects of the loading rate and of the initial defect amplitude have been investigated with velocities in the range [150 – 3000m/s] and surface roughness Ra in the range [0.01 – 0.65 µm]. It is clearly observed that the initial defect amplitude has an important effect on the strain localization time but not on the neck pattern.