Ice vs. steel: Ballistic impact of woven carbon/epoxy composites. Part II – Numerical modelling

Abstract

Part I of this study covered ballistic tests, in which both solid (steel) and fragmenting (ice) spherical projectiles were fired at specimens of carbon-fibre-reinforced polymer (CFRP) composite. The velocity of impacts varied from 70 to 90 m/s for the solid projectiles and from 300 to 500 m/s for the fragmenting projectiles, resulting in three consistent and comparable levels of structural damage for both types of projectiles. The observed dynamic deformation behaviour and resultant damage were examined using a combination of non-invasive analysis techniques. The focus of this Part II is on the development of a mesoscale modelling strategy for CFRP employing a phenomenological continuum-damage approach. The developed model was validated against the data from original ballistic-impact experiments, demonstrating accurate predictions of both deformation behaviour and observed resultant damage of tested specimens for the various experimental loading conditions without modification of the modelling parameters.

Impact damage modelling in composite laminates – numerical implementation of a strain rate dependent damage model

“…The constitutive relations containing the strain rate effects have been implemented in Abaqus/Explicit using the VUMAT subroutine written in FORTRAN programming language. The failure initiation theory, introduced in [5], has been selected as the damage initiation model. The model is based on modified Hashin's criteria that have been adjusted to be applicable for woven composite plies and to include the strain rate effects.…”

“…The increase of strength values in the elevated strain rate conditions is varying for different loading directions and orientations, leading to the necessity of establishing different curve fitting parameters for separate strengths. This could be observed from Table 1, containing all the fitting parameters used along with the material strength data, taken from [5] for the woven CFRP plies T300 K3/IMP530R and T300 K12/ IMP530R. Due to the lack of extensive and detailed experimental data and for simplicity reasons, values of the 𝐾 2 fitting parameters were set to be equal to zero throughout this study.…”

Numerical modelling of ice: Mechanical behaviour of ice under high strain rates

“…However, determining the parameters of EOS is difficult because planetary impact tests need to be performed, which require hyper impact velocity of ice projectiles and an extremely accurate monitoring system [29,30]. Besides, advanced numerical methodologies, such as the particle-subdomain method [31] and smoothed particle hydrodynamics [32], which always use rate-dependent material properties as input, are more appropriate for replicating the mechanical behaviour of ice impact, especially the failure process. Meanwhile, micromechanics also provide a new insight for the modelling of ice, as they can provide reliable mechanical behaviour considering the effect of temperature and strain rate with a microscale kinematic and constitutional relationship [33].…”

Assessment of head injury risk caused by impact using finite element models

“…This means that the main constituents (fibers and matrix) are not explicitly modeled as separated materials but gathered together in homogenised solid elements. The associated computational expense is dramatically reduced and the correspondence with experimental tests leads to high correlations, as presented in previous studies on composites submitted to impact [174,[180][181][182]. At the same time, it is necessary a certain discretisation along the thickness direction of the plate, as the inter-ply interaction plays a relevant role in energy absorption and structural integrity.…”

“…Furthermore, proper numerical study of the impact damage on composite materials should include these non-linearities and also the dynamic effects of the problem [163]. This approach, which has been extensively used in several FE studies on composite plates undergoing impact [164,165,170,173,174], will be adopted for the numerical simulation of the low-velocity tests conducted in this study.…”