Advanced numerical methods for modelling impact in composite materials

Abstract

This thesis presents the development of novel computational tools to overcome the limitations associated with damage modelling in laminated composite materials such as (i) high computational cost of accurately modelling stiffness and damage and (ii) high pre-processing effort for pre-inserting delamination and matrix-crack planes. A well-known example is the limiting mesh size requirement in cohesive zone modelling (CZM). To overcome these limitations a novel higher order cohesive element, discretising the numerical cohesive zone with multiple integration points within the element is developed, thereby enabling the use of large meshes. These cohesive segments are initiated adaptively between compatible continuum elements alleviating the necessity to pre-insert cohesive elements in a mesh. This method is termed higher order Adaptive Mesh Segmentation (AMS) and enables on-the-fly delamination modelling with large meshes. Implementation is performed with explicit time integration scheme in the commercial finite element solver LS-Dyna; this implementation provides easy access of these developments by industry. This method is verified using, (i) quasi-static double cantilever beam and (ii) soft body beam bending impact examples; a 50% reduction in the global number of degrees of freedom is obtained against conventional modelling methods. AMS is extended to linear elements for modelling delamination and matrix cracks as geometric discontinuities using mixed-time integration for efficient analysis with small elements forming ahead of a crack-tip. As an alternative, higher order elements are combined with an improved continuum damage mechanics method for modelling in-plane damage without mesh orientation bias. This method is demonstrated with ply-level discretisation by modelling damage propagation in (i) a laminate with an open-hole (i) a laminate subjected to a rigid body impact. The large mesh capability is further extended in the through thickness direction by modelling global damage behaviour at the sublaminate scale. This is verified by modelling damage propagation under, (i) low velocity rigid body impact, (ii) high velocity rigid body impact and (iii) high velocity soft body impact in large composite specimens. The computational method developed is successful in modelling damage propagation in multiple examples with good correlation to experiments and computational benefits as compared to the more conventionally used linear elements.

Date of Award	25 Jan 2022
Original language	English
Awarding Institution	University of Bristol
Supervisor	Luiz F Kawashita (Supervisor), Antonio R Melro (Supervisor) & Stephen R Hallett (Supervisor)