| Summary: | <p>Following the general trend in the aviation sector, aerospace turbine engine manufacturers are making increasing use of advanced composite materials for key structures and components. Of these, the outcome of this work is specifically meant to aid the design and analysis process of a fan blade made of carbon-fibre-reinforced plastic (CFRP).</p> <p>The simulation-based design and analysis process of a fan blade is driven by its designcritical load case, which is <em>bird strike</em>. There are two key challenges faced when modelling this impact event: (i) a very high finite element count arising from a <em>ply-by-ply</em> representation of the composite laminate, which, in turn, can result in uneconomically high model run times, and, (ii) the wide range of strain-rates experienced by a fan blade over the course of its deformation, failure, and damage evolution process, which, in turn, necessitates the composite's rate-dependency to be adequately accounted for.</p> <p>To date, no model exists that addresses this set of challenges. The aim of this work, thus, is to develop a constitutive model that reflects a composite laminate's ratedependent deformation, failure, and damage evolution behaviour across a multitude of plies within a single <em>representative volume element</em> (RVE). The model's key premise is that delamination between those individual plies is entirely driven by the RVE's outof- plane stresses σ33 , σ23 , and σ31 , whereas any failure and damage evolution within the plies is exclusively linked to the RVE's in-plane stresses σ11 , σ22 , and σ12 . This, in turn, then allows for the RVE's treatment on the basis of two separate 'submodels': a <em>delamination model</em> for its out-of-plane, and a ply model for its in-plane behaviour. Particular emphasis is placed on allowing for a realistic strain-rate history experienced by the finite elements without sacrificing the simulation's numerical stability. This is achieved through the development of a <em>low-pass filter</em> -like strain-rate smoothing algorithm, which is shown to exceed the performance of a conventional <em>running average</em> approach.</p> <p>In addition to the modelling work, this study also makes contributions to selected ratedependent datasets required for the model's calibration. The contributions made to the <em>in-plane transverse compressive deformation mode</em> consists of: (i) a comparative specimen design study, demonstrating that a <em>waisted block</em> should be preferred over a <em>block</em> design, (ii) its characterisation across two loading rate regimes, and, (iii) a comprehensive study investigating the influence of key split-Hopkinson bar setup and testing parameters on the achievability of valid high-rate testing conditions for this deformation mode.</p> <p>The contributions made to the <em>in-plane transverse tensile deformation mode</em> consist of: (i) its characterisation across two loading rate regimes, and, (ii) a study similar to the one for the in-plane transverse compressive deformation mode investigating the achievabiliy of valid high-rate testing conditions. In particular, this study demonstrates the general infeasibility to achieve a specimen deformation under a (nearly) constant strain-rate at failure for this deformation mode.</p> <p>The contributions made to the <em>out-of-plane tensile deformation mode</em> consist of: (i) the development of a form-fit specimen design, particularly with high-rate testing in mind, and, (ii) its characterisation across two loading rate regimes.</p> <p>The contributions made to the <em>energy release rate under tensile loading in fibre direction</em> are as follows: (i) characterisation under high-rate loading based on valid force history data, which has not been achieved before, showing a pronounced drop in G<sup>f+</sup></p>Ic at an elevated loading rate, and, (ii) the utilisation of a modified split-Hopkinson bar setup for the corresponding material characterisation process, involving refinements on aspects of the experimental setup, data acquisition, and data reduction. <p>In the final part of this study, the modelling and experimental strands are drawn together through a model calibration and validation exercise, which also includes a direct comparison between the proposed constitutive model and a state-of-the-art ply-by-ply model (<em>LS-Dyna MAT 261</em> ). It is ultimately shown that the developed model is able to achieve a sufficiently accurate representation of a composite laminate's mechanical behaviour, while at the same time clocking drastically lower computation times in comparison to a ply-by-ply modelling approach.</p>
|
|---|