23.02.2014 Views

Image-based modelling of carbon/carbon composites

Image-based modelling of carbon/carbon composites

Image-based modelling of carbon/carbon composites

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>Image</strong>-<strong>based</strong> <strong>modelling</strong> <strong>of</strong><br />

<strong>carbon</strong>/<strong>carbon</strong> <strong>composites</strong><br />

J. Ali and P. M. Mummery<br />

School <strong>of</strong> Materials,<br />

University <strong>of</strong> Manchester, UK<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester


Introduction<br />

Objective <strong>of</strong> study<br />

2D woven <strong>composites</strong><br />

Methodology <strong>of</strong> image-<strong>based</strong> (IB) <strong>modelling</strong><br />

– X-ray<br />

Microtomography to FE meshes.<br />

3D Unit cells<br />

FE <strong>modelling</strong> <strong>of</strong> mechanical and thermal behaviour<br />

– Nanoindentation for input mechanical data<br />

– Micro-thermal properties from literature<br />

– Compression test (validation)<br />

– Laser flash test (validation)<br />

Future work on irradiation<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester


Objective<br />

Predicting mechanical and thermal properties<br />

– Function <strong>of</strong> architecture, irradiation and temperature (end goal)<br />

– Greater focus on the complex material architectures<br />

– Framework for all classes <strong>of</strong> material with similar structures<br />

ExtreMat<br />

– European Integrated programme<br />

– New materials for extreme environments<br />

Irradiation, temperature, heat flux<br />

C/C Composites<br />

– Restraints and control rods in fission reactors<br />

– Divertor structures in fusion reactors<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester


2D Woven C/C Composites<br />

Toyo Tanso CX-27<br />

Graphitised & Un-graphitised conditions<br />

Lamina (X)<br />

(Ishihara, JAEA)<br />

Perpendicular (Y)<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester


What is X-ray X<br />

Microtomography?<br />

Non-destructive technique for imaging the<br />

internal structure <strong>of</strong> samples in 3D<br />

Set <strong>of</strong> radiographs <strong>of</strong> sample at many<br />

orientations obtained by rotating it within beam<br />

Read by reconstruction s<strong>of</strong>tware<br />

3D attenuation contrast image <strong>of</strong> object using<br />

standard filtered back-projection algorithm<br />

[1]<br />

[1] Babout et al.: Carbon, 2005, 43, 765-774.<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester


What is X-ray X<br />

Microtomography?<br />

3-D D data set<br />

Ability to visualise internal<br />

features non-destructively<br />

Characterise structures<br />

– “Virtual” metallography<br />

Study deformation<br />

processes in situ<br />

FE Modelling<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester<br />

Tooth and Filling


X-ray Microtomography Data<br />

Imaging conditions at a low voltage <strong>of</strong> ∼60-70kV and<br />

90-210<br />

210µA, with pixel resolution <strong>of</strong> ~20µm<br />

Graphitised C/C:<br />

Grey Scale image<br />

Porosity<br />

Transverse<br />

shrinkage cracks<br />

Matrix<br />

Fibre(s)<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester<br />

200µm


Creating IB FE Meshes<br />

Deduce volume fraction <strong>of</strong> composite phases:<br />

Porosity % - Pore-Master (Porosimeter(<br />

Porosimeter)<br />

– Archimedes (water) displacement test - compared<br />

Fibre % and matrix % - SEM <strong>Image</strong>s<br />

Work out image threshold representative to<br />

volume fraction <strong>of</strong> phase(s)<br />

Mesh generated via Simpleware LTD<br />

Mechanical properties <strong>of</strong> fibres and matrix<br />

determined by Nanoindentation<br />

Thermal properties taken from literature<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester<br />

[2]<br />

[2] Ali et al.: ‘<strong>Image</strong>-<strong>based</strong> <strong>modelling</strong> <strong>of</strong> stress/strain behaviour<br />

in <strong>carbon</strong>/<strong>carbon</strong> <strong>composites</strong>’, Energy Materials, 2006.


Microstructural Representation<br />

Transverse<br />

Un-graphitised<br />

C/C composite<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester


FE Mesh<br />

Red = Porosity, Blue = Matrix, Yellow = fibres<br />

1mm<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester


Unit-cell for comparison<br />

Upgraded to curve architecture [3]<br />

0.5mm<br />

[3] Farooqi et al.: Comp. Mater. Sci., 2006, 37, 361-373.


Micromechanical data<br />

Properties <strong>of</strong> fibres and matrix:<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester<br />

GRAPHITISED<br />

Region Indented<br />

Side-<strong>of</strong>-fibres<br />

End-<strong>of</strong>-fibres<br />

Matrix<br />

UN-GRAPHIT.<br />

Region Indented<br />

Side-<strong>of</strong>-fibres<br />

End-<strong>of</strong>-fibres<br />

Matrix<br />

No. <strong>of</strong> Indents<br />

11<br />

3<br />

3<br />

No. <strong>of</strong> Indents<br />

10<br />

3<br />

3<br />

Mean Modulus<br />

(GPa)<br />

7.1 (±1.6)<br />

24.9 (±1.9)<br />

11.4 (±2.7)<br />

Mean Modulus<br />

(GPa)<br />

5.5 (±1.0)<br />

22.8 (±2.1)<br />

7.5 (±1.1)


<strong>Image</strong>s <strong>of</strong> Indents<br />

Matrix less elastic recovery compared to fibres<br />

End-<strong>of</strong>-fibres<br />

(parallel)<br />

Indent marks<br />

in matrix<br />

(Remaining)<br />

Atomic Force Microscope<br />

(AFM)


Compressive Testing<br />

Validate the models with experimental data<br />

Loading<br />

Extensometer<br />

Platen 1<br />

Platen 2<br />

Sample<br />

Cubic samples:<br />

6mm x 6mm x 6mm<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester


Stress/strain results<br />

Graphitised C/C composite – lamina direction<br />

80<br />

70<br />

60<br />

Stress (MPa)<br />

50<br />

40<br />

30<br />

20<br />

10<br />

0<br />

Experimental<br />

<strong>Image</strong>-<strong>based</strong><br />

Analytical<br />

Unit-cell<br />

0 0.001 0.002 0.003 0.004 0.005<br />

Strain


Stress/strain results<br />

Graphitised composite – perpendicular direction<br />

100<br />

90<br />

80<br />

70<br />

Stress (MPa)<br />

60<br />

50<br />

40<br />

30<br />

20<br />

10<br />

0<br />

Experimental<br />

<strong>Image</strong>-<strong>based</strong><br />

Analytical<br />

Unit-cell<br />

0 0.04 0.08 0.12 0.16 0.2<br />

Strain


Stress/strain results<br />

Un-graphitised C/C composite – lamina direction<br />

70<br />

60<br />

50<br />

Stress (MPa)<br />

40<br />

30<br />

20<br />

10<br />

0<br />

Experimental<br />

<strong>Image</strong>-<strong>based</strong><br />

Analytical<br />

Unit-cell<br />

0 0.001 0.002 0.003 0.004 0.005 0.006<br />

Strain


Stress/strain results<br />

Un-graphitised composite – perpendicular direction<br />

160<br />

140<br />

120<br />

Stress (MPa)<br />

100<br />

80<br />

60<br />

40<br />

20<br />

0<br />

Experimental<br />

<strong>Image</strong>-<strong>based</strong><br />

Analytical<br />

Unit-cell<br />

0 0.03 0.06 0.09 0.12 0.15<br />

Strain


Shear failure<br />

Lamina direction<br />

Perpendicular direction


Analysis <strong>of</strong> stress/strain curves<br />

Predictions <strong>of</strong> Young’s s modulus, E (GPa(<br />

GPa):<br />

Derived from stress/strain curves<br />

Loaded<br />

Direction<br />

Graphitised<br />

lamina<br />

Graphitised<br />

perpendicular<br />

Un-graphitised<br />

lamina<br />

Un-graphitised<br />

perpendicular<br />

Experimental<br />

22.0<br />

0.6<br />

15.4<br />

1.3<br />

<strong>Image</strong>-<strong>based</strong><br />

16.3<br />

0.5<br />

13.2<br />

1.1<br />

Unit-cell<br />

16.7<br />

2.3<br />

13.0<br />

1.5<br />

Analytical<br />

13.3<br />

8.7<br />

11.2<br />

6.2<br />

Best predictions highlighted in red<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester


Summary <strong>of</strong> stress/strain results<br />

Good correlation between both FE models and<br />

experimental curves (valid)<br />

<strong>Image</strong>-<strong>based</strong> models better than analytical and<br />

unit-cells<br />

Unit-cell comparable in lamina direction but<br />

linear – neglect <strong>of</strong> porosity<br />

Comparison <strong>of</strong> image-<strong>based</strong> predictions:<br />

– Un-graphitised model gave better agreement with<br />

experiment than graphitised<br />

– More representative area chosen for mesh<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester<br />

Computational power


Thermal Conductivity:<br />

Measurement Principle<br />

Thermal conductivity:<br />

λ = α ρ C p<br />

– α: : thermal diffusivity (Laser flash)<br />

– : : average density<br />

[4]<br />

– C p<br />

: specific heat<br />

Laser flash<br />

– heat pulse on a sample front face<br />

– temperature raise recorded on rear face<br />

T (ºC)<br />

T max<br />

α =<br />

0.136* L<br />

t<br />

1/ 2<br />

[4] Chawla: ‘Ceramic Matrix<br />

t 1/2 time (s) Composites’, Chapman & Hall, 1993.<br />

2


Schematic <strong>of</strong><br />

Laser Flash Rig


Thermal Analysis<br />

Thermal diffusivity measurements converted to<br />

thermal conductivity<br />

– Used to validate FE predictions<br />

Best representative input data taken from<br />

literature<br />

[3]<br />

Steady state FE analysis<br />

Fourier’s s law<br />

Heat flux marks thermal conductivity<br />

Predictions made using:<br />

– <strong>Image</strong>-<strong>based</strong> models<br />

– Unit-cells<br />

Sub-models for 4 classes <strong>of</strong> porosity (A-D) designed by<br />

Farooqi et al.<br />

[3] Farooqi et al.: Comp. Mater. Sci., 2006, 37, 361-373.


Porosity classes<br />

4 Classes <strong>of</strong> porosity identified by optical and<br />

scanning electron microscopy<br />

[5]<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester<br />

[5] Puglia et al.: Composites Part A: Applied Sci.<br />

and Manufac., 2004, 35, 223-230.


Thermal results<br />

Graphitised C/C<br />

Composite<br />

Experimental<br />

results (Wm -1 K -1 )<br />

<strong>Image</strong>-<strong>based</strong><br />

model (Wm -1 K -1 )<br />

Unit-cell model<br />

(Wm -1 K -1 )<br />

Un-graphitised C/C<br />

Composite<br />

Experimental<br />

results (Wm -1 K -1 )<br />

<strong>Image</strong>-<strong>based</strong><br />

model (Wm -1 K -1 )<br />

Unit-cell model<br />

(Wm -1 K -1 )<br />

Lamina direction<br />

13.94<br />

15.31<br />

16.81<br />

Lamina direction<br />

9.42<br />

11.32<br />

11.47<br />

Perpendicular<br />

direction<br />

5.04<br />

5.85<br />

7.01<br />

Perpendicular<br />

direction<br />

2.85<br />

3.44<br />

3.95


Optimisation<br />

Optimum input data<br />

Material direction<br />

Graphitised composite:<br />

Lamina<br />

Perpendicular<br />

Average:<br />

Un-graphitised composite:<br />

Lamina<br />

Perpendicular<br />

Average:<br />

Thermal conductivity, k (Wm -1 K -1 )<br />

Fibre parallel<br />

33.50<br />

35.44<br />

34.47<br />

21.92<br />

18.57<br />

20.25<br />

Fibre transverse<br />

3.35<br />

3.54<br />

3.45<br />

2.19<br />

1.86<br />

2.03<br />

Matrix<br />

8.38<br />

8.88<br />

8.63<br />

5.48<br />

4.64<br />

5.06<br />

Input data should be the same for both directions<br />

– Therefore best representative input data is in theory<br />

average <strong>of</strong> the above values for each constituent


Comparison <strong>of</strong> input data<br />

Literature <strong>based</strong> input data versus optimised<br />

Graphitised C/C<br />

Composite<br />

Expt. . (Wm -1 K -1 )<br />

Literature (Wm -1 K -1 )<br />

Optimised (Wm -1 K -1 )<br />

Lamina direction<br />

13.94<br />

15.31<br />

13.22<br />

Perpendicular<br />

direction<br />

5.04<br />

5.85<br />

4.90<br />

Un-graphitised C/C<br />

Composite<br />

Expt. . (Wm -1 K -1 )<br />

Literature (Wm -1 K -1 )<br />

Optimised (Wm -1 K -1 )<br />

Lamina direction<br />

9.42<br />

11.32<br />

8.70<br />

Perpendicular<br />

direction<br />

2.85<br />

3.44<br />

3.10<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester


Summary <strong>of</strong> thermal results<br />

<strong>Image</strong>-<strong>based</strong> models showed superior agreement with<br />

experimental data<br />

– Slight over-prediction<br />

– Literature input data questionable, optimised input data better<br />

<strong>Image</strong>-<strong>based</strong> models gives truer representation <strong>of</strong><br />

composite structure<br />

Unit-cell <strong>based</strong> on idealisations <strong>of</strong> composite<br />

microstructure<br />

Difference lies in consideration <strong>of</strong> porosity<br />

Unit-cell main weakness in perpendicular direction<br />

– Thin laminate structure


Application <strong>of</strong> framework<br />

Materials under investigation:<br />

NB-31 3D C/C Composite<br />

B0-27 2D SIC/SIC Composite<br />

SGL 2D C/C Composite<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester


Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester<br />

Conclusion<br />

Comparison <strong>of</strong> FE techniques for woven <strong>composites</strong><br />

Providing useful initial results<br />

Overall results indicate image-<strong>based</strong> models better at<br />

predicting mechanical & thermal behaviour<br />

<strong>Image</strong>-<strong>based</strong> meshes account for complex network <strong>of</strong><br />

pores in structure<br />

FE analysis complicated due to stress shielding large pores<br />

More difficult to assign truer properties<br />

Unit-cells <strong>based</strong> on idealisations <strong>of</strong> composite<br />

microstructure<br />

Truer properties<br />

Thinness main weakness


Future work<br />

Un-irradiated predictions completed<br />

Shows potential for predicting irradiation<br />

behaviour using IB FE meshes<br />

Tomography scans <strong>of</strong> irradiated materials<br />

Other alternatives similarities with thermal oxidation<br />

Need irradiation input data:<br />

Fibres & Matrix<br />

History<br />

Porosity and constituent dimensional changes<br />

under irradiation already accounted<br />

Irradiation sub-routine developed for graphite<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester<br />

May eventually be applicable to <strong>carbon</strong> <strong>composites</strong>


References<br />

[1] L. Babout, , P. M. Mummery, T. J. Marrow, A. Tzelepi,<br />

P. J. Withers: Carbon, , 2005, 43, , 765-774.<br />

774.<br />

[2] J. Ali, C. Berre and P. M. Mummery: ‘<strong>Image</strong>-<strong>based</strong><br />

<strong>modelling</strong> <strong>of</strong> stress/strain behaviour in <strong>carbon</strong>/<strong>carbon</strong><br />

<strong>composites</strong>’, , accepted for publication in Energy<br />

Materials<br />

[3] J. K. Farooqi, , M. A. Sheikh: Comp. Mater. Sci., , 2006,<br />

37, , 361-373.<br />

373.<br />

[4] K. K. Chawla, ‘Ceramic Matrix Composites’, , Chapman<br />

& Hall, 1993.<br />

[5] P. Del Puglia, M .A. Shiekh, , D. R. Hayburst:<br />

Composites Part A: Applied Sci. . and Manufac., , 2004,<br />

35, , 223-230.<br />

230.<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester


Acknowledgements<br />

EPSRC - Funding<br />

AMEC NNC - Funding<br />

Dr Ishihara, JAEA - Supplying samples<br />

Combining the strength <strong>of</strong> UMIST and<br />

The Victoria University <strong>of</strong> Manchester

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!