Structurally stitched preforms: experimental characterization, geometrical modelling, and FE analysi

1
Structurally stitched preformsexperimental
characterization, geometrical modelling, and FE
analysis

• Vitaly KOISSIN, Jan KUSTERMANS, Stepan V. LOMOV,
  Ignaas VERPOEST
• Department MTM, Katholieke Universiteit Leuven
• Kenta HAMADA, Yujiro MOMOJI, Hiroaki NAKAI,
  Tetsusei KURASHIKI, Masaru ZAKO
• Department of Management of Industry and
  Technology, Osaka University

2
Contents

• Introduction
• Materials
• Experimental
• Geometrical modelling
• FE modelling
• Conclusions

3
Introduction

• Structural stitching presumably improves
• lay-up process
• transversal stiffness
• interlaminar strength
• But results also in
• fibre misalignment in the preform plies
• (resin-rich openings)
• That can lead to
• lower in-plane stiffness
• earlier damage onset
• ?

double locked stitching
tufting
dual-needle stitching
4
Contents

• Introduction
• Materials
• Experimental
• Geometrical modelling
• FE modelling
• Conclusions

5
Materials
Fabric quasi-UD hybrid woven (warp 24K carbon
tows alternated with thin polysulfone yarns
weft the same PSU yarns) with the total areal
weight of 226 g/m2.
Preform 28 plies with symmetric stacking
90?/45?/0?/0?/-45?/90?/-45?/0?/0?/45?/0?/-45?/0?/
45?s, where 90? corresponds to the warp (carbon
fibre) direction in the surface plies.
Structural stitching 1K carbon yarn (Tenax).
Tufting with a 5x5 mm piercing pattern. The
machine direction coincides with 0? direction of
the preform. Results in 6.41.2mm (face) or
6.60.7mm (back) openings (average values)
6
Materials
Thickness after impregnation 5.32mm. Average Vf
58 (without openings) or 62 (with
openings). Openings 3.80.8mm (average values),
almost equal in all layers obviously due to
severe compaction in the mould. Nesting very
prominent the ply waviness amplitude sometimes
exceeds the average ply thickness (5.32 / 28
0.19mm).
7
Contents

• Introduction
• Materials
• Experimental
• Geometrical modelling
• FE modelling
• Conclusions

8
Experimental in-plane tension

• Procedure
• uniaxial tension, 3mm/min
• 25030mm specimen size, 130mm gauge length
• series of 6 specimens are tested for 0? (along
  the structural stitching) and 90? directions
• full-field stain mapping (Limess)
• acoustic emission registration (Vallen)
• since the AE sensors should be removed before the
  failure, the test is not completely monotonic but
  is paused at a certain load level.

9
Experimental in-plane tension
Figure Surface strains ex under loading in 0?
(a,b) and 90? (c,d) directions. The load is
applied in x-direction. The average strain level
is 0.75 or 1.35, respectively. The presumable
positions of the stitching yarns are shown with
black lines. Conclusion tufting causes strain
concentrations at the stitching sites.
10
Experimental in-plane tension
Measured stiffness (average, stitched /
unstitched) 0? direction (along the structural
stitching) 73.4 / 87.7 increased due to higher
overall Vf and additional stitching yarn
fibres. 90? direction 46.9 / 38.9 decreased
(apparently due to the fibre misalignment)
11
Experimental in-plane tension
Figure typical history of AE energy in normal
(left) and logarithmic (centre) scales. Right
cumulative sum of AE event counts. Conclusion
depending on the load direction, tufting can
precipitate the mass crack onset (MCO, sdam) and
increase the quantity of cracks. However their
total energy can be lower if compare with the
unstitched material.
12
Experimental in-plane tension
90? direction no distinct features at the
stitching sites distribution of the event energy
along the specimens is similar for virgin and
stitched materials.
13
Experimental in-plane tension
0? direction (along stitching) events
concentrate at the stitching sites distribution
of the event energy is more stepwise for the
stitched material.
14
Experimental 3PBT

• Procedure
• loading 2mm/min, 80mm span length, 10mm diameter
  indenter
• series of 2 specimens are tested for 0? (along
  the structural stitching) and 90? directions
• full-field stain mapping (Limess)

15
Experimental 3PBT

• Conclusion
• failure mode does not change (ply buckling,
  tensile failure, delamination)
• stitching reduces the bending stiffness (both for
  0? and 90? ) and
• presumably reduces strength for 90? (across the
  structural stitching) more specimens should be
  tested

16
Contents

• Introduction
• Materials
• Experimental
• Geometrical modelling
• FE modelling
• Conclusions

17
Geometrical modelling
WiseTex package ? StitchTex software ?
homogenized properties
18
Geometrical modelling
Figure The woven fabric (left) and its WiseTex
geometrical model (right). Absence of nesting
results in too high (gt100) local Vf inside the
yarns, if the nominal thickness (5.32mm) is
preserved. Therefore the thickness is increased
to keep local Vf below 90.7 (the ultimate
packing). The warp and weft crimp is chosen to
minimize the fabric thickness. When calculating
the homogenized elastic properties in TexComp,
the thickness is reduced artificially to the
nominal one, thus nesting the inclusions into
the volume having the correct average Vf.
19
Geometrical modelling

• The preform is built by 1) multiplication and
  rotation of the ready fabric model exported from
  WiseTex and 2) appending the stitching loops
  (optionally).
• Preform assumptions
• no openings, since WiseTex approach does not
  allow to split a yarn
• the piercing pattern is regular (constant
  stitching length and distance between seams)
• seams are straight and parallel

20
Geometrical modelling

• Stitching loop assumptions
• the stitching yarn consists of a single strand.
  This is accepted due to a strong randomization of
  the strand positions and shapes along the yarn
  path
• the cross-section is circular along the
  through-the-thickness path. The yarn flattening
  is optionally modelled only at the preform
  surface, to avoid segmentation of the surface
  plies in an FE model
• the backside loops have equal height and are bent
  regularly onto the preform surface (this is
  random in reality).

21
Contents

• Introduction
• Materials
• Experimental
• Geometrical modelling
• FE modelling
• Conclusions

22
FE modelling

• MeshTex / Sacom / M3 are used
• Assumptions
• textile structure is neglected and substituted
  for UD mats having the same average Vf of the
  carbon fibres. This is reasonably due to compact
  placement and low crimp of the carbon tows
• polysulfone yarns are not modelled also, since
  their stiffness/strength are similar to these of
  the matrix material
• homogenized properties are determined using the
  Chamis formulae
• periodic BC.

23
FE modelling

24
FE modelling

• Assumptions
• the opening and zone of re-oriented fibres are
  rhomboidal
• the local fibre orientation around the opening
  is symmetric with respect to its axes
• Vf is uniform (in reality e.g. local Vf can
  decrease towards the edge of the opening).

25
FE modelling
90/45/0/0/-45/90/-45
/0/0/45/0/-45/0/45s
type A
type B
type D
type C
26
FE modelling
90/45/0/0/-45/90/-45
/0/0/45/0/-45/0/45s
type A
type B
type D
type C
27
FE modelling
90/45/0/0/-45/90/-45
/0/0/45/0/-45/0/45s
type A
type B
type D
type C
28
FE modelling
unstitched
stitched, face
stitched, back
Figure Damaged elements in the surface plies
(mode L, marked white) under 90? loading at 0.8
average strain. The Hoffmans criterion is
used. Conclusions the opening results in a
prominent disturbance of the stress field
introduction of the stitching loop causes even
greater stress concentrations. the stitching loop
triggers earlier damage onset. The damage growth
is more extensive. The same is observed in the
tests.
29
Results
Elastic properties and damage onset (non-stitched
/ stitched)
30
Contents

• Introduction
• Materials
• Experimental
• Geometrical modelling
• FE modelling
• Conclusions

31
Conclusions

• the structural stitching has small influence on
  the in-plane components of the homogenized
  stiffness matrix. The stitching loop is important
  only for the transversal stiffness and related
  constants. Thus, simple models without openings
  can be used to obtain the stiffness matrix
• under bending, the difference can be larger
• however, the stress-strain fields are sensitive
  to a local geometry, which can play the role of a
  stress concentrator and trigger damage.
  Therefore, presence of the stitching yarn and
  openings, as well as specifics of their modelling
  is important for a correct computation of the
  damage onset and propagation
• for a typical structurally stitched composite,
  FE simulation results and theoretical estimations
  (method of inclusions) are compared with
  experimentally measured properties. They agree
  well thus showing efficiency of the developed
  models.
• Future work
• fatigue response
• ?

32
Acknowledgements
the study is done within the I-TOOL (Integrated
Tool for Simulation of Textile Composites)
project funded by the European Commission
Thanks to Dassault Aviation for lay-up and
impregnation IFB, Stuttgart University for
structural stitching.
33
Thank you for your attention!