Chen Li

Doctorant - Institut Universitaire de Technologie de Tarbes (IUT de Tarbes)
Membre du groupe Matériaux et Structures Composites (MSC)
IUT, 1 rue Lautréamont
65016 Tarbes
Activités de recherche :

Simulation numérique
de la fabrication de composites à  renfort fibreux 3D par le procédé d’infusion

Doctorant: Chen LI1

Xiaojing GONG2, Arthur CANTAREL3


2 , 3

Résumé : La fabrication de pièces composites structurales aéronautiques
nécessite l'utilisation de renforts complexes et notamment des renforts tridimensionnels.
Le procédé d'infusion permettrait la fabrication de ces pièces tout en
maitrisant leur qualité et les coûts. L'objectif de la thèse est de développer
un modèle, à  différentes échelles, de l'écoulement de résine à  traverse le
réseau fibreux du renfort et de proposer une simulation numérique plus fine du
procédé d'infusion. L’influence des structures de renforts 2D et 3D, sera ainsi
étudié aux différentes échelles: micro (intra-mèches), méso (inter-mèches)
et macro (renfort, pièce).

Abstract "” The
manufacture of composite aeronautical structural parts requires the use of more
complex reinforcements, especially the 3D reinforcements. Resin infusion is the
key process of Liquid Composites Molding (LCM) which will determine the quality
and cost of the parts. The target of this thesis is to develop models at
different scales of fibrous network of the reinforcement and to propose a finer
numerical simulation of the infusion process. The influence of the 2D and 3D
reinforcement structures on the infusion process will be studied at the different
scales: micro (intra-tow), meso (inter-tows) and macro (reinforcement, part).

1. Introduction

LCM is
a promising method for composites manufacture with various advantages. It has
four main steps including preform set, resin infusion, curing and disclosing
process. Among all these processes, infusion is the most important step which
will determine the final quality of the part. Thus, numerical simulation of the
resin infusion is meaningful [1]. The work performed and the results obtained
will be presented in the second part, which mainly focus on the permeability
estimation on random fiber models [2] and dual scales models and then the
future work is listed in the perspective.

2. Permeability estimation intra tow in the
section (2D)

For porous media, permeability tensor
is the most significant parameter to describe the property of the reinforcement
for infusion processing. As an intrinsic property of the porous material, K
can be calculated by Darcy's law based on the velocity field obtained by solving
Navier-Stokes equation and continuity equation together. Hence, the Stokes
equation can be used for free flow with the consideration of incompressible
fluid and the inertial term can be neglected in micro, meso and macro scales.

random position and size of the fiber producing codes are developed to set up
realistic geometrical models. Numerical simulation has been realized based on
these models.

In fact,
we have studied the influences of several parameters on the flow of the resin
during infusion. It includes 3 micro-structural parameters: δmin, L,
Δr and a
macro parameter: ε. Here δmin is defined as
the minimum distance between two neighbor fibers among all the fibers; ∆r is the range of the fiber radius, L is the length of the square domain and
ε is the porosity. We have also used
Morris sensitivity method to analyze the influences of these parameters.
Porosity ε has the most obvious
influence on permeability while L has
a significant effect on average velocity compared with other parameters. These
influences observed in a Representative Elementary Volume (REV) can be
generated to whole structure.

3. Permeability estimation of dual scales in 3

In order to estimate the permeability in 3 directions
using dual scales: intra and inter tows, the model with random position codes
have been developed in this work. The idea of this model is to generate the
real position of the fibers, the determination of the permeability of 3
directions can be achieved by applying the calculation method mentioned above for
intra as well as inter tow. The results from this model seem very promising,
because the permeability obtained in the transverse section (Fig. 1) correlate
well with the results published in the reference [3]. Moreover, the
permeability in the longitudinal direction obtained from
this model is much closed to that calculated by traditional method. Actually,
to our knowledge no
any other works has been realized in the literature on the determination
of 3D permeability at dual scales.

4. Perspective

(1) Adaptable
3D geometrical models of braided and woven fiber will be studied to set up for
numerical simulation.

(2) The two phases coupling free flow and porous flow will be studied to
detect the infusion character on both of 2D and 3D models. Parameters of
structure which may influence the infusion process will be defined. Their
interactive influences on the infusion will also be explored.

(3) A final method will be developed for flow front prediction at macro
scale  with various local permeability on different position 
of the part(Fig. 2 for 3D simple part and Fig. 3 for 3D car bonnet).

Figure. 2. Macro porous flow at part level (3D simple part)

Figure. 3. Macro infusion process at part level (3D car bonnet)


Gantois, A. Cantarel, B. Cosson, G. Dusserre, J.-N. Félices, F. Schmidt, BEM-based
models to simulate the resin flow at macroscale and microscale in LCM
processes, Polymer Composites, 34 (8), 2013

Zhang, B. Cosson, S. Comas-Cardona, C. Binetruy, Efficient stochastic
simulation approach for RTM process with random fibrous permeability
Composites Science and Technology,
71, 2011.

[3]W.R. Hwang, S.G. Advani. Numerical simulations of Stokes-Brinkman equations for permeability
prediction of dual scale fibrous porous media. Physics of Fluids, 22 (11), 2010