JEC COMPOSITES MAGAZINE - Issue #113 - June 2017 - 64

TECHNOLOGY simulation
resolutions down to a few microns can be easily reached with appropriate laboratory equipment. Such a technique can provide real 3D images of the microstructure including local flaws. Therefore, IMT Lille
Douai (formerly Mines Douai) suggested a mixed experimental and
numerical approach with a view to:
1) Integrating local microstructure heterogeneities more accurately in
critical zones such as weld lines.
2) Providing experimental data that can be used to develop and assess
local damage criteria.
This article presents a summary of the main results obtained within
the context of the doctoral thesis of A. Ayadi [3] conducted at IMT
Lille Douai. As illustrated in Figure 1, the first part of the project includes two major contributions based on the use of X-ray µCT. The
first objective is to develop a strategy that can take into account real
microstructure heterogeneities accurately to evaluate local mechanical
properties in zones of abrupt geometrical changes where heterogeneities are the least controllable during the injection moulding process.
The second objective is to provide more physically based damage evolution interpretations based on in-situ mechanical tests in the X-ray
micro-tomographic synchrotron facilities. An appropriate flowchart
was developed to post-process 4D big data and to quantitatively assess
experimental damage evolution laws that can later serve for the development of more accurate local damage criteria.
Material and semi-structural samples
The material considered is a polyamide 6.6 reinforced with 35 wt.% of
short E-glass fibres (Zytel®70G35 HSLX, DuPont™). The fibres have
an average length of 250 µm and a nominal diameter of 10 µm. The
selected grade is generally used in automotive parts located near the
car engine such as intake manifolds, inlet gas compressor exit and fuel
injection rails.
A variety of semi-structural parts was considered to reproduce configurations similar to those encountered in industrial parts (Table 1).
In this context, functional parts such as the inlet gas compressor exit
and mild notched plates obtained by injection moulding were first
tested. The mild notched plates have a particular configuration that
induces a strongly in-plane anisotropic fibre orientation distribution
[7]. Moreover, it causes the formation of a hot weld line due to a cir-

cular opening located at the centre. In addition, an original "stair-step"
shaped plate previously optimized at IMT Lille Douai was used to generate tangential and frontal flow variations during the injection moulding process. This form has the advantage to be equipped with multiple
injection gate configurations [8, 9].
Assessment of mechanical properties
in an opposing flow weld line
In the first case, the open-hole plate was considered. Particular interest
was given to the zone where the material fronts meet just after the open
hole and where a hot weld line forms. A µCT scan was performed using an UltraTom X-ray µCT system (ISIS 4D facilities, Lille, France)
equipped with a CCD camera (4000×2624 pixels). More details about
the parameters used are provided in [7].
The effect of the fibre density distribution was incorporated in a finite
element model based on analytical conversions of gray levels to mechanical properties. A numerical homogenization procedure was used
to evaluate the elastic mechanical properties at a macroscopic scale assuming the hypothesis of isotropy. Further analysis of the numerical
results, namely the distribution of stress profiles along the thickness of
the sample, confirmed the reliability of the μCT to analyze the local
fibre arrangement. The stress profiles showed that the microstructure
has a five-layer arrangement in the plate thickness and three layers in
the vicinity of the central open hole. However, an evaluation of the anisotropy index related to the macroscopic elastic properties demonstrated low sensitivity to the local variation of fibre orientations. To overcome this limitation, the μCT data was subjected to a complementary
segmentation step to separate the fibres from the matrix. The resulting
binary data were used as entries for a second two-scale numerical homogenization method that combines the results of two elastic computations at the fibre scale (micro) then at the analyzed samples scale
(macro). Thus, the mechanical parameters of each sample were calculated for hundreds of sub-domains (micro-scale) that form the composite structure (macro-scale) [10]. The results of the micromechanical
model were then introduced in a second macroscopic model as a set of
elementary orthotropic material models of the material's heterogeneous structure (Figure 2). This approach yielded a higher sensitivity of
the macroscopic mechanical properties to local fibre orientations in the
vicinity of the open hole and justified the assumption of orthotropic
symmetry for the studied material in this area.

Tab. 1: Some of the investigated technical parts and experimental techniques
used for the appropriate investigation levels
DIC: Digital image correlation ,SEM: Scanning electron microscopy, µCT: X-ray
computed tomography, SR-µCT: Synchrotron radiation computed tomography

Fig. 2: Overview of the multi-scale finite element homogenization procedure;
characteristic element sizes are provided in appropriate length units [11]

jec composites magazine / No113 June 2017

Table of Contents for the Digital Edition of JEC COMPOSITES MAGAZINE - Issue #113 - June 2017