Beruflich Dokumente
Kultur Dokumente
DOI 10.1007/s12289-012-1116-5
ORIGINAL RESEARCH
Abstract This experimental work concerns the study of the Keywords Composites . RTM . Preforming . Sheet
preforming of a specific highly double curved geometry forming . Case corner . Defaults
with a triple point (case corner) by the sheet forming process
using powdered interlock reinforcement (G1151®). Three
different punches (square box, prism, tetrahedron) were Introduction and context of the study
used in this study, each of them presenting highly double
curved geometry with a case corner. A specific sheet form- Transportation industries are willing to replace metallic mate-
ing device specially designed for the preforming of textile rials by composites materials on structural parts subjected to
reinforcement was used. The expected shapes with the three severe mechanical solicitations with equal mechanical per-
punches have been obtained with an optimized blank-holder formances. Composite materials are proposed because they
pressure. No classical defaults such as wrinkling or yarn are able to provide credible answers to the optimization of
damage are present in the useful zone of the preforms. large and thick structural parts. Their good strength/weight
However, a new default, not observed for spherical or hemi- ratio and especially their anisotropy which can be adapted to
spherical shape has been identified. It concerns the out of the mechanical solicitation of the structure are particularly
plane buckling of yarns. This phenomenon not observed on interesting. Resin Transfer Molding (RTM) is one of the most
the square box is visible on some faces and edges of the important processing methods in the LCM processes family
prismatic and tetrahedron shapes. For the square box, it is [1, 2]. It is adequate for mass production of composite pieces
easily possible to control the orientation of the yarn within with moderate cost compared to the traditional hand-layup or
the preform in the faces, whereas this is not possible for autoclaving and high fiber volume fractions, up to 60 %, are
triangular faces of the prismatic and tetrahedron shapes. The possible. Besides, the closed mould technique prevents haz-
square box punch is therefore more adapted to preform the ardous styrene emission [3, 4]. During the forming step of
highly doubled curved shape with the case corner. very complex shapes significant deformations may occur
resulting in local variation of the textile reinforcement geom-
etry. These modifications of the reinforcement (homogeneity,
fiber volume fraction, deformation state, ..) strongly influence
S. Allaoui (*) : G. Hivet : D. Soulat : A. Wendling : P. Ouagne the resin flow impregnation by the modification of the perme-
Laboratoire PRISME, UPRES EA 4229, Université d’Orléans, ability components (longitudinal and transverse) [5–7]. They
Polytech’Orléans, also influence the mechanical properties of the final composite
8 rue Léonard de Vinci,
part [3, 8, 9]. Numerical simulations have already been exten-
45072 Orléans Cedex 2, France
e-mail: samir.allaoui@univ-orleans.fr sively used to investigate the viability of fabric forming by
optimization of process control parameters, material selection
S. Chatel and tool design to avoid production of costly prototypes.
EADS France Innovation Works Structure Engineering Production
Some of these simulations describe the evolution of the dry
and Aeromechanics, Centre Technocampus EMC2,
Allée du Chaffault, reinforcement behaviour by mechanical approach [10–13].
F44340 Bouguenais, France These numerical works must be validated, compared and
Int J Mater Form
Fig. 1 Case corner part used for aeronautical application Fig. 3 The sheet forming device
Int J Mater Form
system has been designed so that the number and the shape markers. Yarn strains, yarn positions, shear angles …etc
of the blank-holders can be easily changed to investigate the can be computed. The evolution of the macroscopic behav-
influence of the position and the pressure values on the iour, 3D geometry, appearance of defects …etc can also be
preform properties. The second system used to apply tension investigated [24]. Whole of these measurements can be
to the fabric (not visible on Fig. 3) is based on the drawbead obtained for the final shape but their evolution during the
principle. A low tension is applied on the yarn extremities so forming process steps can also be measured.
that the fabric does not show any wrinkles and any shear
before applying the blank-holder pressure and before actu- Presentation of the fabric used
ating the punch.
At the end of the preforming test, the realized dry pre- The fabric used in this study is a powdered woven rein-
form can be fixed by applying a spray of resin on its surface forcement used in aeronautics. It is denoted G1151® and
so that the preform can be removed from the tools and constituted by an interlock weaving of 6 K carbon yarns
stocked in the state corresponding to the end of the test. (630 g/m2, 7.5 yarns/cm) (Fig. 5). The fabric is manufactured
Global aspect (wrinkle, global dimensions …) can be by the Hexcel Company and its mechanical characteristics
analyzed. (tensile test, picture frame tests, bending tests, friction, …
On complement and associated to these global informa- etc) have been studied [30–35].
tion at the scale of the preform, it is important to measure
local quantities such as the membranous positions, the ori- Test conditions
entations of the yarns but also the 3D strains of the fabric. In
this intention, the device is equipped by 3D Digital Image As indicated previously, this study concerns the feasibility
Correlation system which is a noncontact measurement of forming a case corner part by using three different
method [25, 26]. Two numerical cameras are located at the punches (Fig. 2). The first punch is a Prism with 280 mm
top of the device. They can be positioned as a function of the length and 280 mm width. His total height is 175 mm. The
specific zone of the fabric that has to be analyzed. For the base height is 40 mm. The dimensions of the square box
computation of the displacement field, marks tracking tech- punch are 280×280×280 mm3. For these two punches, an
nique is used [8, 27]. A speckle pattern needs to be applied open square die (300×300 mm) is used. The third punch is a
on the fabric surface, (Fig. 4) before the test. They are tetrahedron form with 280 mm sides. His total height is
carried out using white painting for dark fabrics and black 128 mm and the base height is 20 mm. The die used with
painting for clear fabrics. A special care is given to this this punch is a triangular open die (314×314×314 mm).
preparation, because markers must be as circular as possible The edges and vertices possess 10 mm radius for the three
and their diameter lower than the width of the yarns in order punches and 20 mm for the dies. Due to the low value of the
to avoid their deterioration during the test and thus disturb edges radius, the expected shapes should be highly double
measurement [29]. The three space co-ordinates and thus the curved and large strains are expected to take place along the
displacement field of these markers can be computed, using edges. For each preforming test, one layer of G1151® dry
the marker tracking software Deftac 3D [28], from each fabric (≈0.9 m2) is cut out. Markers are then positioned on
digital image taken during the process and from the image the layers as indicated previously (Fig. 4). The initial posi-
captured at the reference state. With this 3D displacement tioning of the fabric is of importance. This partly conditions
field the membranous strains can be easily computed. Post the final yarns orientation within the part and as a conse-
processing treatments can be used to evaluate the meso- quence its mechanical stiffness. In the cases of the prismatic
scopic behaviour of the fabric by considering only local
Fig. 4 Carbon reinforcement with markers at the initial state Fig. 5 Woven interlock carbon reinforcement (G1151®)
Int J Mater Form
900 mm
and the square box punches, the ply has been positioned
with the weft and warp network parallel to the edges of the Prismatic Punch Die
punches faces (detailed on Figs. 6 and 7). In the case of the
tetrahedral punch, due to the geometry, it’s more complex to
Blank-
place the fabric in order to foresee a specific orientation of holders
the yarns in the final shape. For the tests presented in this
1000 mm
paper, it was chosen (Fig. 8) to align the warp and weft
yarns with a symmetry plane of the tetrahedron. To ensure Weft
proper positioning of the fabric, one side of the outer three
sides is taken as reference. The considered side is one that is
parallel to the sides of the fabric ply (at left on the figure).
When positioning the fabric, the weft yarns are aligned on
this side with respect to the open die, when the warp yarns
are perpendicular and passé through the opposite corner. A
set-square is used to ensure this position.
Fabric ply Warp
However for all the tests realized with different punches, the
angle between the weft and warp yarns are controlled using set-
Fig. 7 Initial positioning of the interlock fabric for the prismatic punch
square and the optical measurement in order to ensure that the
fabric handling do not generates shearing before forming. The
measured initial angles are around 90° ±0.5. Results and discussion
Before applying the blank-holder pressure, and before
starting the movement of the punch, the fabric can bend Optimization of the blank-holder pressure on the preforming
under its own weight. It is therefore necessary to apply some step
tensions at the yarn extremities using the drawbead system.
Rectangular or square blank-holder can be placed along First, tests at different blank-holder pressures were carried
each edges and corners of the punch. Height rectangular out in order to determine the optimized pressure to use for the
blank-holders were used for the prismatic and square box interlock fabric forming. Figure 9 presents the different
punches whereas only six of them were used for the tetra- shapes obtained with the prismatic punch that is highly
hedron (Figs. 6, 7 and 8). The pressure was then applied on non-expandable with small curvature radii (10 mm). The
each of them. For these tests the punch velocity is 30 mm/ global shape (Fig. 9a, b, c) looks globally correct. In the
min and its stroke 160 mm. For each punch, at least of three useful zone, with a low blank-holder pressure (0.2 bar) some
tests were carried out. The results and discussions obtained
on the different shapes are exposed and discussed below. 900 mm
900 mm
Warp
Square box Punch Die Die
Weft
Blank-
holders
1000 mm
1000 mm
Weft
d) Zoom on the useful zone at 0.2 bar e) Zoom on the useful zone at 1 bar
few wrinkles (Fig. 9d) appear. However, these ones disap- deformation. It was shown that the G1151® fabric is partic-
pear once the pressure reaches a value of 1 bar (Fig. 9e). ularly adapted to this deformation mode, with a high value
Wrinkles are visible around the useful zones and especially of the locking angle (~55–60°) [31, 32]. However, the only
near the position of the blank-holders. These non-useful notion of locking angle is not sufficient to anticipate the
zones are cut after the process. Consequently, these wrinkles presence of wrinkles. Indeed, a rise of the blank-holder
have no consequences on the quality of the performs, but, pressure has the tendency to delay the appearing of wrin-
these results confirm first the influence of the tension in kles. This means that it is possible to obtain high shear
yarns, due to the blank-holder pressure, on the presence of deformation where the shear angle is higher than the locking
wrinkle and second the necessity to apply some load to angle if tension is applied to the fabric. This confirms the
obtain a preform without default [13]. importance of the coupling between shear and tension as
For a constant blank-holder pressure of 1 bar, Fig. 10 already discussed in [32, 36].
shows the comparison of the macroscopic shapes obtained In the global analysis of the realized preforms, no
with the three different punches. The global three shapes defaults such as broken yarns have been noticed even for
obtained do not show any obvious default such as wrinkling the highest blank-holder pressure tests. From a uniform
or yarn damage as long as a minimum blank-holder pressure pressure P applied on a number of yarn by unit length,
is applied. denoted d, positioned on a width l, it can be deduced, in
To form complex shapes such as the one presented in first approximation, the tension load, f, if the friction coef-
Fig. 10, the fabric may accommodate large in-plane shear ficient between the tool/fabric, μ is known by Eq. 1:
Fig. 10 Drawing of a
carbon interlock weave
with three different punches
(1 bar pressure)
300
Yarns Before Forming yarns. To confirm that fact, tensile tests were carried out on
Yarns after Forming
250 yarns extracted from the dry fabric before and after shaping.
Yarns extracted after forming were pull out from those passing
200 by the shapes because they are subjected to highest mechanical
Force (N)
Warp yarns
Warp yarns
Weft yarns
Warp yarns
c) Tetrahedron shape
Int J Mater Form
Curved yarn
±0.35
0.8 21.5 ±0.5
20 ±1
5.3 ±0.8
±1
2.5 24 ±1 24.5 ±1.5
50°-60° 50°-60°
Fig. 15 Shear angles after a square box forming of the interlock Fig. 17 Shear angles after a tetrahedron forming of the interlock
weave. Red weft yarns; Blue warp yarns; yellow shear zones weave. Red weft yarns; Blue warp yarns; yellow shear zones
(Fig. 15 for the square box; Fig. 16 for the prismatic punch). change of the punch geometry and is submitted to the
Since the weaving network remains parallel, the shear angle pressure of the blank-holder. For the fabric tested, this high
remains low with a maximum of 5.3° for the square box and shear deformed state does not conduct to default, in theses
2.6° for the prismatic punch. Controlling the homogeneity location, but will affect the injection step.
of the yarns position and of the shear deformation state The last zones to analyze concern the triangular faces of
allow, in these zones, to predict a good homogeneity of the prismatic shape (two lateral faces) and the three tetrahe-
the fiber volume fraction and consequently of the mechan- dron faces. The yarns networks do not remain parallel to the
ical properties. On the lateral edge, for the square box, or at edges of the punches (Figs. 16 and 17). For the prismatic
the base corners for the three shapes, the shear deformations shape the weft yarns network initially parallel to the edges
are more complex. Due to the little edges radius value for (Fig. 14b) become curved, and for the tetrahedron the phe-
the square punch and the very deep shaping (Fig. 15), high nomenon is the same but concerns the weft or the warp
shear angles have been reached (~50–60°) and are close to yarns depending on the observed faces. These specific
the locking angle of the G1151® identified during a shear obtained orientations depend on the geometry of the shape,
test [31, 32]. High levels of shear deformation were also the initial positioning of the fabric, but also the position of
measured for the prismatic shape 40° to 50° and 50° to 60° the blank-holder. The measurement of the shear angle on these
for the tetrahedron one at the basis of each corner (Fig. 17). triangular faces, for the two shapes (Figs. 16 and 17) confirm
In these small zones the fabric must accommodate the brutal that the shear deformation in the fabric is more important
comparatively to others faces, due to the curved yarns. On
both sides of the yarns passing by the top of this shape, two
Curved yarn symmetrical zones can be visible where the shear angle is
relatively the same (respectively ~24.5° for the tetrahedron
25
20
2.6 ±1.06
Shear angle (°)
15
41.2 ±2 10
40.9 ±0.6 0
0 20 40 60 80 100 120 140 160
40-50° Stroke of the punch (mm)
Fig. 16 Shear angles after a prismatic forming of the interlock weave. Fig. 18 Shear angle evolution during the tetrahedron forming of the
Red weft yarns; Blue warp yarns; yellow shear zones interlock weave
Int J Mater Form
shape and ~41.2° for the prismatic one). For the tetrahedron triangular faces, at the middle of the face but also on the
punch, the evolution of the fabric shear angle on a triangular diagonal edges, separating the triangular to the rectangular
face during the process is presented in Fig. 18. It can be faces. At the reinforcement level, buckles are one out-of
concluded that the maximum angle is reached and remains plane default. It can be compared to the wrinkling phenom-
constant quite before the end of the process. enon but this one takes place at the scale of the fabric,
Controlling the final yarns orientation is easily possible whereas buckling takes place at the scale of yarns. It could
on rectangular or square faces with the prismatic or square be assimilated to the buckling of yarns and is not the same
box punches. For the triangular faces of the prismatic shape like that has been described for NCF reinforcement like an in-
as well as for the face of the tetrahedron faces, the orienta- plane buckling [17].
tion of the yarns is more difficult to control with regard to The experimental observations show that yarns passing
the desired final shape and which should meet the specifi- by the triple point are subjected to consequent tensile defor-
cations in terms of mechanical solicitation. mation. The yarns perpendiculars to these ones are not so
much tightened. On both sides of the buckles zones, espe-
Out of plane buckles cially on triangular faces, the yarns perpendicular to the one
passing by the triple point are curved as already mentioned in
When using the tetrahedron and the prismatic punches, Section Position of the yarn after forming. These yarns try to
another type of defect denoted by out of plane buckles accommodate bending in their plane. It seems that this is not
appears on faces and on some edges of the preforms possible and out of plane bending takes place. The buckles
(Fig. 19). Locally these buckles constitute a consequent over localization and size depends on the tension of the yarn
thickness, which can reach 3 mm of height near the triple perpendicular to the one passing by the triple point and
point. Consequently the homogeneity of the thickness’s therefore on the blank-holder pressure. The influence of each
preform is not controlled. As described on Fig. 19a, b theses process parameters used during the forming of the consid-
buckles are localized on zones of about 15 to 35 mm width. ered reinforcement on the apparition and the quantification
On the tetrahedron, these buckles zones are present, at least, of these buckles will be the subject of futures works. The
on each triangular faces and edges which converge to the architecture of the reinforcements should probably be inves-
triple point (top of the shape) from the bottom of the shape. tigated with the goal to optimize the constitution of the fabric
On the prismatic shape, the buckles are localized on the to avoid the generation of buckles.
Fig. 19 The out of plane
buckling default at the end of
Curved yarn
the tetrahedron and prismatic
forming of G1151® layer
bukles
Buckles
Buckle
width
a) Buckles zone on tetrahedron shape b) Buckles zone on prismatic shape
11. Peng XQ, Rehman ZU (2011) Textile composite double dome 24. Allaoui S, Launay J, Soulat D, Chatel S (2008) Experimental
stamping simulation using a non-orthogonal constitutive model. tool of woven reinforcement forming. Inter J Mater Form 2
Compos Sci Tech 71(8):1075–1081 (1):815–818
12. Hamila N, Boisse P (2008) Simulations of textile composite rein- 25. Dumont F, Hivet G, Rotinat R, Launay J, Boisse P, Vacher P
forcement draping using a new semi discrete three node finite (2003) Field measurements for shear tests on woven reinforce-
element. Compos B 39:999–1010 ments. Mec Indust 4:627–635
13. Boisse P, Hamila N, Vidal-Sallé E, Dumont F (2011) Simulation of 26. Willems A, Lomov SV, Verpoest I, Vandepitte D (2008) Optical
wrinkling during textile composite reinforcement forming. strain fields in shear and tensile testing of textile reinforcements.
Influence of tensile, in-plane shear and bending stiffnesses. Compos Sci Tech 68:807–819
Compos Sci Tech 71(5):683–692 27. Launay J, Lahmar F, Boisse P, Vacher P (2002) Strain measure-
14. Kelly PA, Bickerton S (2009) A comprehensive filling and tooling ment in tests of fibre fabric by image correlation method. Adv
force analysis for rigid mould LCM processes. Compos A 40 Compos Lett 11(1):7–12
(11):1685–1697 28. Bretagne N, Valle V, Dupré JC (2005) Development of the marks
15. Lin H, Wang J, Long AC, Clifford MJ, Harrison P (2007) traking technique for strain field and volume variation measure-
Predictive modelling for optimization of textile composite form- ments. NDT & E International 38(4):290–298
ing. Compos Sci Tech 67:3242–3252 29. Bremand F, Dupre JC, Lagarde A (1995) Mesure de déformations
16. Mohammed U, Lekakou C, Bader MG (2000) Experimental stud- sans contact par analyse d’images. Proceedings of Photomechanics:
ies and analysis of the draping of woven fabrics. Compos A 31 171–177
(12):1409–1420 30. Cao J, Akkerman R, Boisse P, Chen J et al (2008) Characterization
17. Lee J, Hong S, Yu W, Kang T (2007) The effect of blank holder of mechanical behavior of woven fabrics: experimental methods
force on the stamp forming behavior of non-crimp fabric with a and benchmark results. Compos A 39(6):1037–1053
chain stitch. Compos Sci Tech 67(3–4):357–366 31. Hivet G, Duong AV (2011) A contribution to the analysis of the
18. Sidhu RMJS, Averill RC, Riaz M, Pourboghrat F (2001) Finite intrinsic shear behaviour of fabrics. J Compos Mater 45(6):695–
element analysis of textile composite perform stamping. Compos 717
Struct 52(3–4):483–497 32. Launay J, Hivet G, Duong AV, Boisse P (2008) Experimental anal-
19. Cherouat A, Borouchaki H, Giraud-Moreau L (2010) Mechanical ysis of the influence of tensions on in plane shear behaviour of woven
and geometrical approaches applied to composite fabric forming. composite reinforcements. Compos Sci Tech 68(2):506–515
Int J Mater Form 3(2):1189–1204 33. Anh Vu Duong (2008) Étude expérimentale du comportement
20. Woven Composites Benchmark Forum. http://www.woven mécanique de renforts composites tissés lors de la mise en forme
composites.org sur géométrie non développables. PhD thesis, Université d’Orléans
21. Khan MA, Mabrouki T, Vidal-Sallé E, Boisse P (2010) Numerical 34. De Bilbao E, Soulat D, Hivet G, Gasser A (2010) Experimental
and experimental analyses of woven composite reinforcement study of bending behaviour of reinforcements. Exp Mech 50:333–
forming using a hypoelastic behavior: application to the double 351
dome benchmark. J Mater Process Tech 210:378–388 35. Allaoui S, Hivet G, Wendling A, Ouagne P, Soulat D (2012)
22. Vanclooster K, Lomov SV, Verpoest I (2009) Experimental Influence of the dry woven fabrics meso-architecture on fabric/
validation of forming simulations of fabric reinforced poly- fabric contact behaviour. J Compos Mater 46(6):202–210
mers using an unsymmetrical mould configuration. Compos A 36. Lomov SV, Verpoest I (2006) Model of shear of woven fabric and
40:530–539 parametric description of shear resistance of glass woven rein-
23. Allaoui S, Boisse P, Chatel S, Hamila N, Hivet G, Soulat D, Vidal- forcements. Compos Sci Tech 66(7–8):919–933
Salle E (2011) Experimental and numerical analyses of textile 37. Sachs U, Fetfatsidis K, Schumacher J, Ziegmann G, Allaoui S,
reinforcement forming of a tetrahedral shape. Compos A 42 Hivet G, Vidal-Sallé E, Akkerman R (2012) A friction-test bench-
(6):612–622 mark with Twintex PP. Key Eng mater 504–506:307–312