On modelling the constitutive and damage behaviour of highly non-linear bio-composites – Mesh sensitivity of the viscoplastic-damage law computations

2019 
Abstract The large strain fracture of non-linear complex solids concerns a wide range of applications, such as material forming, food oral processing, surgical instrumental penetration as well as more recently, the design of biodegradable composites for packaging and bio-medical use. Although simulations are a powerful tool towards understanding and designing such processes, modelling ductile fracture in materials such as soft natural composites imposes a new challenge, particularly when the fracture patterns cannot be pre-defined. Here we bring to light new information on these aspects of benefit to the multidisciplinary community, by characterising and modelling the deformation and fracture of short cellulose fibre starch extruded composites. Hyperviscoelastic-Mullins damage laws show merits in modelling such complex systems. Yet they are inferior to a viscoplastic-damage law able to capture exactly their highly non-linear, rate dependent and pressure dependent pseudo-plastic stress-strain response. The viscoplastic-damage law also predicts fracture based on experimental toughness values without pre-specifying the crack path in a Finite Element (FE) model, displaying superiority over the conventional cohesive zone approach. Yet, despite using a toughness parameter to drive crack propagation, spurious mesh dependency is still observed while other previously unreported sources of error imposed by the finite element aspect ratio are also highlighted. The latter is rectified by developing a novel numerical strategy for calculating the characteristic element length used in the damage computations. Inherent mesh dependency suggests that non-local damage models may be essential to model this newly investigated class of natural composites.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    98
    References
    7
    Citations
    NaN
    KQI
    []