Permeability of flax fibre mats: Numerical and theoretical prediction from 3D X-ray microtomography images

https://doi.org/10.1016/j.compositesa.2021.106644Get rights and content

Abstract

Flax fibre mats are promising and versatile biosourced reinforcements that can be used in composite parts obtained using various processing routes. To optimise their impregnation and the end-use properties of composites, it is crucial to better understand the process-induced evolution of their microstructure and their permeability. In this study, flax fibre mats were subjected to in situ X-ray microtomography compression experiments. The resulting 3D images enabled the evolution of several key descriptors of their microstructure under compression to be determined, and the evolution of their permeability to be quantified by direct fibre scale CFD simulations. The microstructural data were also used as input parameters of a modified directional Kozeny-Carman model, accounting for the anisotropy and heterogeneity of mats. Only one unknown directional parameter was identified by inverse method from permeability calculations performed on numerically generated 3D realistic fibre networks. The predictions of the proposed model were consistent with numerical simulations.

Introduction

Biosourced composites reinforced with plant-based fibres represent a credible alternative to composites reinforced with glass or other synthetic fibres that are commonly used as structural or semi-structural parts in many industrial applications [1], [2]. Several architectures of plant-based fibres are encountered: woven fabrics, knitted fabrics, and non-woven materials such as unidirectional veils of fibres and fibre mats. Fibre mats are versatile fibrous materials that consist of an intricate network of individualised fibrous elements such as discontinuous fibres, or discontinuous fibre bundles or both types of elements. The ease of their manufacturing offers the possibility to obtain various fibrous architectures varying for instance their fibre content, areal density and fibre orientation [3], [4]. In addition, they offer a good compromise with respect to other types of reinforcements because of their good processability (e.g. large deformation properties) in several composite manufacturing processes, and their ability to provide good reinforcement effect to polymer matrices [5], [6].

The manufacturing of composites parts with biosourced fibre mats can be done using either wet, e.g. Liquid Composite Moulding (LCM), or dry forming, e.g. compression moulding, processes [2], [7], [8]. These processes involve the deformation of these reinforcement materials and their simultaneous or subsequent impregnation by a fluid polymer matrix to obtain either a prepreg material or a composite part with the desired shape. For plant-based reinforcement materials such as flax fibre woven fabrics, a poor control of the deformation mechanisms of these materials in their dry state is known to induce several defects such as wrinkling, buckling or tearing that affect the reinforcement architecture and integrity [9], [10], [11]. Similarly, the impregnation of the same reinforcements by fluids such as filled thermoset resins or thermoplastics may result in voids inside and between tows [11], [12]. It has been established that the impregnation of fibrous reinforcements is mainly controlled by their anisotropic permeability properties [13], [14], [15] which are also strongly coupled with their deformation state [16]. At the flow front, capillary effects also play an important role on the impregnation phenomena [17]. Far from the flow front, in saturated zones, the impregnation phenonema occurring in composite forming processes are usually modelled by assuming the Darcy’s law [18], which is, strictly speaking, valid for the flow of incompressible Newtonian fluids through rigid porous media at negligible Reynolds number. Vast research efforts are still ongoing to reveal the links between the components of the permeability tensor K of the Darcy’s law and the deformability of various natural fibrous materials with more or less disordered architectures [19]. For biosourced fibre mats, the prediction of the permeability components is all the more difficult to establish as the fibrous microstructure of these materials is strongly disordered and fibres exhibit large morphological variations [16], [20], [21].

Several experimental studies dealt with the determination of the out-of-plane and in-plane permeability of biosourced fibre mats [22], [21], [23], [24]. The effect of the compaction was also determined and allowed highlighting the effect of the porosity or conversely the fibre volume fraction on the permeability properties. Some authors reported that the evolution of the permeability components could be empirically fitted using either empirical power-law functions of the fibre volume fraction ϕ¯s [14] or the Kozeny-Carman model [25], [26], [27]. This equation is widely used to estimate the permeability K of isotropic porous media:K=1-ϕ¯s32cSv2τ2

where Sv is the specific surface area, τ is the tortuosity (defined as the ratio between the mean flow path length and a characteristic length of the porous media). The Kozeny-Carman model assumes that the porous medium is equivalent to an assembly of parallel tortuous capillaries with equal length and diameter and circular cross sections. The assembly of capillaries has the same equivalent fibre volume fraction ϕ¯s and specific surface area Sv as the porous medium. The parameter c can be seen as a phenomenological corrective term to account for flows in cylinders with any cross section geometry [25], [26], [27], [28], [29], [30]. The Kozeny-Carman model was used by Bizet et al. [23] to fit the evolution of the out-of-plane permeability component of flax fibre mats by determining the best value for c for fitting experimental data. The parameter c was shown to depend non-linearly on the fibre volume fraction.

The Kozeny-Carman equation thus relates the permeability to some key microstructure descriptors of the porous media and c, the parameter coupling the fluid flow with the structure. It is thus crucial to determine accurately these descriptors and their evolution with the deformation. Several studies used 3D X-ray microtomography imaging techniques to characterise the porous and fibrous architecture of biosourced mats [21], [31] and paper-like materials [32], [33], [34]. Thanks to the analysis of 3D images, the authors could measure several crucial descriptors such as the mean porosity ϕ¯p, the specific surface area Sv and tortuosity τ. They also used these parameters in the Kozeny-Carman equation and compared the prediction of this model for the out-of-plane permeability of wood-based fibre mats with results obtained experimentally or by numerical simulation performed on 3D X-ray microtomography images [21], [35]. The chosen value of the parameter c did not allow a good prediction of the out-of-plane permeability of the studied materials [35]. Their choice was governed by considering the geometry of the pore cross sections (circular or flat cross section of pores) that is presumably far from the real geometry of the pores in the investigated anisotropic fibrous networks. This shows that this parameter is critical and also difficult to estimate using uniquely image analysis of 3D fibrous networks. To circumvent this difficulty, Koponen et al. [36] performed flow simulations through the thickness of numerically generated realistic fibrous networks with planar fibre orientation representative of the structure of paper-like materials. Hence, they determined the out-of-plane permeability of these materials, thereby estimating the value of c by an inverse method for this particular direction. They also proposed an empirical non-linear law for the evolution of c as a function of the fibre volume fraction ϕ¯s.

Using high-resolution 3D X-ray microtomography images of fibre networks appear to be a powerful approach to estimate the permeability of fibrous materials with complex architectures [32], [33], [34], [37], [38], [39], [40], [41]. The current progress made in 3D X-ray microtomography imaging allow acquiring 3D images during in situ and in real time experiments that mimic the real forming conditions the composite fibre reinforcements are subjected [42], [43], [44].

Hence, the objectives of this work were to investigate the microstructure and its evolution during transverse compression of thermolinked flax fibre mats, mimicking compaction phenomena that occur in many composite forming processes. For that purpose, in situ compression experiments were performed using synchrotron X-ray microtomography. Image analysis allowed quantifying the evolution of several key microstructure descriptors of the mats during their compaction. The components of the permeability tensor were estimated by direct numerical simulations on the 3D images for the various compression stages. Then, the microstructural data were used as input parameters of a modified anisotropic Kozeny-Carman model. To account for the anisotropy of the flax fiber mats, the tortuosity and an equivalent term to the aforementioned c parameter of the Kozeny-Carman model were seen as directional parameters. Following the approach proposed by Koponen et al. [36], the directional values of c were identified by an inverse method from permeability calculations performed using numerically generated 3D fibre networks. Finally, the relevance of the modified Kozeny-Caman model was discussed and tested for another type of biosourced mat.

Section snippets

Flax fibre mats

The first type of mats was made of 90 wt% of flax fibres combined with 10 wt% of polypropylene fibres. These mats, denoted M1 in the following sections, were fabricated by the Gemtex laboratory using carding, overlapping and needle punching nonwoven technologies [45]. These mats were also consolidated using thermolinking. During thermo-linking, the polypropylene fibres melted, allowing them to bond flax fibres after cooling. This process is usually used to increase the mechanical strength of

Specific surface area Sv

A stereological technique [55], [52] based on intercept lines was used to estimate the specific surface area Sv of the solid phase, i.e. flax fibre phase plus the polypropylene phase. The specific surface area was calculated as follows [56]:Sv=2PL¯

where PL¯ is the mean number of intercepts per unit of intercept lines. PL¯ was calculated from the measurements of the number of intersection with the solid phase of 500 intercept lines the directions of which were randomly distributed in the

Evolution of the pore volume fraction

Fig. 4a shows the variation of the local volume fraction of pores ϕp along the thickness of the sample for the different stages of compression shown in Fig. 2. This figure shows that the volume fraction of pores decreases with the increase in the compression loading and increases after unloading. However, in the initial state and during the deformation of the sample, the volume fraction of pores varies slightly along the thickness. After unloading, the volume fraction of pores is different from

Proposition of an anisotropic Kozeny-Carman permeability model

Based on the previous studies devoted to the prediction of the permeability properties of fibrous materials with disordered fibrous architectures such as papers [36], boards or reinforcements for composites made of discontinuous fibres [21], [35], [38], [39], [40], we propose the following adaptation of the Kozeny-Carman (KC) model for anisotropic porous materials where the principal components of the dimensionless permeability tensor Ki are written as follows:Ki=1-ϕ¯s32ciSv2τi21r¯s2

with i=I,I

Conclusion

In situ out-of-plane compression tests were performed on a thermolinked flax fibre mat, using a micro-press installed on a synchrotron microtomography beamline. This technique allowed us to show that (i) the porous phase evolved significantly as shown by the measurements of the mean porosity, pore size distribution, and the directional tortuosities, (ii) the cross sections of the flax fibres remained almost unchanged, and (iii) the specific surface area increased with increasing the compaction.

CRediT authorship contribution statement

T.A. Ghafour: Conceptualization, Investigation, Writing – review & editing. C. Balbinot: Conceptualization, Investigation, Writing – review & editing. N. Audry: Conceptualization, Investigation, Writing – review & editing. F. Martoïa: Conceptualization, Investigation, Writing – review & editing. L. Orgéas: Conceptualization, Investigation, Writing – review & editing. P.J.J. Dumont: Conceptualization, Investigation, Writing – review & editing. P. Vroman: Conceptualization, Investigation, Writing

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

C. Balbinot gratefully acknowledges the French Ministry of Higher Education and Research for her PhD research grant. We acknowledge EcoTechnilin (Valliquerville, France) and would like to thank Karim Behlouli for supplying mat M2. The experiments done on mat M1 were performed on beamline ID19 at the European Synchrotron Radiation Facility (ESRF) in the framework of the long term project “Heterogeneous fibrous materials” (MA 127), Grenoble, France. We acknowledge the Paul Scherrer Institut

References (70)

  • R. Umer et al.

    The effect of yarn length and diameter on permeability and compaction response of flax fibre mats

    Compos Part Appl Sci Manuf

    (2011)
  • X. Chen et al.

    On the variability of the Kozeny constant for saturated flow across unidirectional disordered fiber arrays

    Compos Part Appl Sci Manuf

    (2006)
  • H.L. Liu et al.

    Permeability prediction of fibrous porous media with complex 3D architectures

    Compos Part Appl Sci Manuf

    (2012)
  • J. Viguié et al.

    Finding fibres and their contacts within 3D images of disordered fibrous media

    Compos Sci Technol

    (2013)
  • S. Rolland du Roscoat et al.

    Estimation of microstructural properties from synchrotron X-ray microtomography and determination of the REV in paper materials

    Acta Mater

    (2007)
  • B. Caglar et al.

    Permeability of textile fabrics with spherical inclusions

    Compos Part Appl Sci Manuf

    (2017)
  • P. Soltani et al.

    Effect of 3D fiber orientation on permeability of realistic fibrous porous networks

    Powder Technol

    (2014)
  • P. Soltani et al.

    Prediction of permeability of realistic and virtual layered nonwovens using combined application of X-ray μCT and computer simulation

    Chem Eng Res Des

    (2017)
  • S.E. Hamdi et al.

    X-ray computed microtomography and 2D image analysis for morphological characterization of short lignocellulosic fibers raw materials: a benchmark survey

    Compos Part Appl Sci Manuf

    (2015)
  • F. Loix et al.

    Woven fabric permeability: From textile deformation to fluid flow mesoscale simulations

    Compos Sci Technol

    (2008)
  • T. Laurencin et al.

    3D real-time and in situ characterisation of fibre kinematics in dilute non-Newtonian fibre suspensions during confined and lubricated compression flow

    Compos Sci Technol

    (2016)
  • D. Ferré Sentis et al.

    3D in situ observations of the compressibility and pore transport in Sheet Moulding Compounds during the early stages of compression moulding

    Compos Part Appl Sci Manuf

    (2017)
  • P. Latil et al.

    Towards the 3D in situ characterisation of deformation micro-mechanisms within a compressed bundle of fibres

    Compos Sci Technol

    (2011)
  • F. Martoïa et al.

    Cellulose nanofibril foams: Links between ice-templating conditions, microstructures and mechanical properties

    Mater Des

    (2016)
  • J. Lux et al.

    Macroscopic thermal properties of real fibrous materials: Volume averaging method and 3D image analysis

    Int J Heat Mass Transf

    (2006)
  • F. Martoïa et al.

    Crumpled paper sheets: Low-cost biobased cellular materials for structural applications

    Mater Des

    (2017)
  • Waldir L. Roque et al.

    A plugin for computing the pore/grain network tortuosity of a porous medium from 2D/3D MicroCT image

    Appl Comput Geosci

    (2020)
  • J.L. Auriault

    Heterogeneous medium. Is an equivalent macroscopic description possible?

    Int J Eng Sci

    (1991)
  • Claude Boutin

    Study of permeability by periodic and self-consistent homogenisation

    Eur J Mech - ASolids

    (2000)
  • F. Chalencon et al.

    Homogeneous and heterogeneous rheology and flow-induced microstructures of a fresh fiber-reinforced mortar

    Cem Concr Res

    (2016)
  • Dominique Jeulin

    Morphology and effective properties of multi-scale random sets: a review

    Comptes Rendus Mécanique

    (2012)
  • T. Abdul Ghafour et al.

    The importance of taking into account behavior irreversibilities when simulating the forming of textile composite reinforcements

    Compos Part Appl Sci Manuf

    (2019)
  • C. Baley et al.

    Average tensile properties of French elementary flax fibers

    Mater Lett

    (2014)
  • A. Bourmaud et al.

    Relationships between micro-fibrillar angle, mechanical properties and biochemical composition of flax fibers

    Ind Crops Prod

    (2013)
  • D.U. Shah

    Developing plant fibre composites for structural applications by optimising composite parameters: a critical review

    J Mater Sci

    (2013)
  • Cited by (7)

    • Compression moulding

      2023, Design and Manufacture of Structural Composites
    • Deep learning accelerated prediction of the permeability of fibrous microstructures

      2022, Composites Part A: Applied Science and Manufacturing
      Citation Excerpt :

      However, this approach lacks the representation of the inherent variability such as yarn deformation, nesting etc., as well as the random distribution of fibers within yarns [20,21]. To overcome this limitation, an alternative route emerged based on collecting the micro- and meso-structural 3D geometric information via X-ray computed microtomography scans and performing flow simulations within those domains [22–26]. Despite the aforementioned advances, these simulations still require access to high computational power as they typically require hours even in highly parallelized systems [27].

    View all citing articles on Scopus
    View full text