This paper presents an investigation of voids in carbon fibre reinforced epoxy composites manufactured by resin film infusion using X-ray computed tomography (XCT). Two panels were investigated, one formed with a high viscosity resin, the other with a lower viscosity resin. The study focusses on the characterisation of the 3D distribution of voids in the panels. A new approach to the measurement of defect distribution demonstrated that in both panels, the voids were located close to the binder yarn. When the low viscosity resin was employed, the void distribution was more uniform throughout the panel thickness whereas for the high viscosity resin, the voids were mainly localised in the central part of the panel. Both qualitative and quantitative data were obtained giving extensive, three dimensional information which aids a better understanding of the manufacturing process.

1. Outline Introduction Experimental Results Conclusion
3D Characterisation of Pore Distribution in
Resin Film Infused Composites
Fabien L´eonard1
, Jasmin Stein2
, Arthur Wilkinson2
,
and Philip J. Withers1
1
Henry Moseley X-ray Imaging Facility, The University of Manchester
2
Northwest Composites Centre, The University of Manchester
Industrial Computed Tomography
September 20 th
2012 Wels, Austria

2. Outline Introduction Experimental Results Conclusion
Outline
1 Introduction
Overview
2 Experimental
Materials
X-ray Computed Tomography
3 Results
Void overview
Sample L (low viscosity resin)
Sample H (high viscosity resin)
Sample comparison
4 Conclusion

3. Outline Introduction Experimental Results Conclusion
Overview
Void assessment in composites
The properties of a composite are detrimentally affected by voids
introduced during the manufacturing process.
The common method of determining the void volume fraction is acid
digestion for carbon fibre reinforced composites and matrix burn off for
glass fibre reinforced composites.

4. Outline Introduction Experimental Results Conclusion
Materials
Specimen Manufacturing
Epoxy resins:
triglycidylaminophenol TGAP (Araldite MY0510, Huntsman) and
tetraglycidyl-4,4’-diaminodiphenylmethane
TGDDM (Araldite® MY721, Huntsman)
Hardener:
4,4’-diaminophenyl sulfone, DDS (Aradur 976-1, Huntsman)
Reinforcement:
unidirectional carbon fibre fabric supplied by Sigmatex
12 k carbon tows were bound with a fine glass fibre yarn (spacing of
approximately 6 mm to give a fabric of 445 g.m−2
)

5. Outline Introduction Experimental Results Conclusion
Materials
Specimens
Two composite panels were investigated in this study:
Panel L was made from unmodified resin (low viscosity resin)
Panel H was made from the same resin modified with
polyethersulfone, PES (Virantage VW-10300 FP, Solvay) to improve
fracture toughness (which results in a higher viscosity resin).
Each panel (approximately 5 mm thick) was cut into five 25 mm wide
strips, along the 0◦
direction.

6. Outline Introduction Experimental Results Conclusion
Materials
Resin film infusion
Composite laminates were manufactured by resin film infusion (RFI)
which is suitable for high viscosity resins.
Resin films were degassed at 130 ◦
C for 1 hour and then were
immediately frozen.
The laminates were cured for 2 hours at at 130 ◦
C, 165 ◦
C, and 200 ◦
C
Vacuum was applied throughout the cycle for the infusion to take place.

7. Outline Introduction Experimental Results Conclusion
X-ray Computed Tomography
Acquisition
Scanning was performed at the Henry Moseley X-ray Imaging Facility
on the Nikon Metrology 225/320 kV Custom Bay system:
Voxel size: 17.3 µm
Target: Cu
Voltage: 75 kV
Current: 130 µA
Filter: none
Exposure time: 2000 ms
Number of projections: 3142
Acquisition time: 1h 45’

8. Outline Introduction Experimental Results Conclusion
X-ray Computed Tomography
Visualisation
The visualisation software employed was Visualisation Science Group
proprietary software Avizo Fire version 7.0.1.
The voids, matrix and yarns were segmented as follows:

9. Outline Introduction Experimental Results Conclusion
X-ray Computed Tomography
Visualisation
The visualisation software employed was Visualisation Science Group
proprietary software Avizo Fire version 7.0.1.
The voids, matrix and yarns were segmented as follows:
Pores
[0;60]
voids only

10. Outline Introduction Experimental Results Conclusion
X-ray Computed Tomography
Visualisation
The visualisation software employed was Visualisation Science Group
proprietary software Avizo Fire version 7.0.1.
The voids, matrix and yarns were segmented as follows:
Pores
[0;60]
Yarns
[150;255]
binder yarn and voids

11. Outline Introduction Experimental Results Conclusion
X-ray Computed Tomography
Visualisation
The visualisation software employed was Visualisation Science Group
proprietary software Avizo Fire version 7.0.1.
The voids, matrix and yarns were segmented as follows:
Pores
[0;60]
Composite
[60;150]
Yarns
[150;255]
all labels y

12. Outline Introduction Experimental Results Conclusion
X-ray Computed Tomography
Visualisation
The porosity distribution measurements in 3D are focussed on the
position of the voids relative to both the edges of the panel and the yarns.
The void-to-yarn distance and the void-to-edge distance can be obtained
from Avizo by combining the distance maps with the thresholded voids.
raw 2D slicep

13. Outline Introduction Experimental Results Conclusion
X-ray Computed Tomography
Visualisation
The porosity distribution measurements in 3D are focussed on the
position of the voids relative to both the edges of the panel and the yarns.
The void-to-yarn distance and the void-to-edge distance can be obtained
from Avizo by combining the distance maps with the thresholded voids.
raw 2D slicep
segmented voids

14. Outline Introduction Experimental Results Conclusion
X-ray Computed Tomography
Visualisation
The porosity distribution measurements in 3D are focussed on the
position of the voids relative to both the edges of the panel and the yarns.
The void-to-yarn distance and the void-to-edge distance can be obtained
from Avizo by combining the distance maps with the thresholded voids.
raw 2D slicep
segmented voids
yarn distance map edge distance map

15. Outline Introduction Experimental Results Conclusion
X-ray Computed Tomography
Visualisation
The porosity distribution measurements in 3D are focussed on the
position of the voids relative to both the edges of the panel and the yarns.
The void-to-yarn distance and the void-to-edge distance can be obtained
from Avizo by combining the distance maps with the thresholded voids.
raw 2D slicep
segmented voids
yarn distance map
void-to-yarn distance map
edge distance map

16. Outline Introduction Experimental Results Conclusion
X-ray Computed Tomography
Visualisation
The porosity distribution measurements in 3D are focussed on the
position of the voids relative to both the edges of the panel and the yarns.
The distance of every voxel segmented as void to the closest yarn and the
closest edge can be measured over the entire 3D volume.
raw 2D slicep
segmented voids
yarn distance map
void-to-yarn distance map
edge distance map
void-to-edge distance map

17. Outline Introduction Experimental Results Conclusion
X-ray Computed Tomography
Visualisation
raw 2D slicep
segmented voids
yarn distance map
void-to-yarn distance map
edge distance map
void-to-edge distance map

18. Outline Introduction Experimental Results Conclusion
Void overview
Porosity Overview
Table 1: Summary of the void equivalent diameter1
.
Specimen Minimum Maximum Mean Standard deviation
label mm mm mm mm
L-1 0.027 0.398 0.078 0.055
L-2 0.027 0.404 0.075 0.050
L-3 0.027 0.290 0.073 0.049
H-1 0.027 0.406 0.078 0.055
H-2 0.027 0.458 0.074 0.051
H-3 0.027 0.419 0.083 0.059
1The equivalent diameter is the diameter the void would have if it was perfectly
spherical.

19. Outline Introduction Experimental Results Conclusion
Void overview
Porosity Overview
Table 1: Summary of the void equivalent diameter1
.
Specimen Minimum Maximum Mean Standard deviation
label mm mm mm mm
L-1 0.027 0.398 0.078 0.055
L-2 0.027 0.404 0.075 0.050
L-3 0.027 0.290 0.073 0.049
H-1 0.027 0.406 0.078 0.055
H-2 0.027 0.458 0.074 0.051
H-3 0.027 0.419 0.083 0.059
Minimum just below 0.03 mm and limited by resolution of XCT.
1The equivalent diameter is the diameter the void would have if it was perfectly
spherical.

20. Outline Introduction Experimental Results Conclusion
Void overview
Porosity Overview
Table 1: Summary of the void equivalent diameter1
.
Specimen Minimum Maximum Mean Standard deviation
label mm mm mm mm
L-1 0.027 0.398 0.078 0.055
L-2 0.027 0.404 0.075 0.050
L-3 0.027 0.290 0.073 0.049
H-1 0.027 0.406 0.078 0.055
H-2 0.027 0.458 0.074 0.051
H-3 0.027 0.419 0.083 0.059
Maximum equivalent diameter are around 0.40 mm.
1The equivalent diameter is the diameter the void would have if it was perfectly
spherical.

21. Outline Introduction Experimental Results Conclusion
Void overview
Porosity Overview
Table 1: Summary of the void equivalent diameter1
.
Specimen Minimum Maximum Mean Standard deviation
label mm mm mm mm
L-1 0.027 0.398 0.078 0.055
L-2 0.027 0.404 0.075 0.050
L-3 0.027 0.290 0.073 0.049
H-1 0.027 0.406 0.078 0.055
H-2 0.027 0.458 0.074 0.051
H-3 0.027 0.419 0.083 0.059
Mean void diameter ranges from 0.075 mm up to 0.084 mm.
1The equivalent diameter is the diameter the void would have if it was perfectly
spherical.

22. Outline Introduction Experimental Results Conclusion
Sample L (low viscosity resin)
Void size distribution
KS test on the three equivalent diameter distributions reveals that the
data are from the same continuous distribution.

23. Outline Introduction Experimental Results Conclusion
Sample L (low viscosity resin)
Void spatial distribution
Homogeneous distribution of voids through the panel thickness, the
evenly spaced maxima reflect the regular positions of the yarns.

24. Outline Introduction Experimental Results Conclusion
Sample L (low viscosity resin)
Void spatial distribution
Single sharp and intense peak at a distance close to 0.12 mm which
indicates that the voids are mainly located around the yarns.

25. Outline Introduction Experimental Results Conclusion
Sample H (high viscosity resin)
Void size distribution
KS test on the three equivalent diameter distributions reveals that the
data are from the same continuous distribution.

26. Outline Introduction Experimental Results Conclusion
Sample H (high viscosity resin)
Void spatial distribution
Peak intensities increase with increasing distance from the panel edge,
indicating a higher distribution of voids close to the centre of the panel.

27. Outline Introduction Experimental Results Conclusion
Sample H (high viscosity resin)
Void spatial distribution
Single sharp and intense peak at a distance close to 0.12 mm which
indicates that the voids are mainly located around the yarns.

28. Outline Introduction Experimental Results Conclusion
Sample comparison
Void size distribution
K-S test rejects the hypothesis that the two void distributions are from
the same distribution.

29. Outline Introduction Experimental Results Conclusion
Sample comparison
Void size distribution
K-S test rejects the hypothesis that the two void distributions are from
the same distribution.

30. Outline Introduction Experimental Results Conclusion
Sample comparison
Void spatial distribution – qualitative
Homogeneous void distribution for panel L and higher void distribution
close to the centre of panel H.

31. Outline Introduction Experimental Results Conclusion
Sample comparison
Void spatial distribution – qualitative
Similar sharp peak around 0.12 mm which indicates that the voids are
mainly located around the yarns for both specimens.

32. Outline Introduction Experimental Results Conclusion
Sample comparison
Void spatial distribution – quantitative
50 % of the voids lie within the middle 2.5 mm of panel L
50 % of the voids lie within the middle 0.5 mm of panel H

33. Outline Introduction Experimental Results Conclusion
Sample comparison
Void spatial distribution – quantitative
90 % of the voids are located within 0.25 mm from the yarns in panel L
90 % of the voids are located within 0.50 mm from the yarns in panel H

34. Outline Introduction Experimental Results Conclusion
Summary
Summary
Summary:
Composite panels have been manufactured by resin film infusion
The pore distribution in the panels has been fully characterised
Highlights:
No major difference between specimens with standard pore analysis
The void-to-yarn and void-to-edge distance maps showed major
differences between the panels:
Homogeneous void distribution for panel L and higher void
distribution close to the centre of panel H.
In both panels, the pores are mainly located around the yarns.
Both qualitative and quantitative analyses can be performed

35. Outline Introduction Experimental Results Conclusion
Discussion
Discussion
Questions ?
Fabien L´eonard
fabien.leonard@manchester.ac.uk