C

Calculattion of ellectrical conducttivity annd electro

othermall
analysiss of multtilayeredd carbonn reinforcced commposites:
appllication tto damage detecction

Athanassopoulos Nikolaos
N
Departmennt of Mechaanical Enginneering & Aeronautics
A s
Univversity of Paatras

A thesis subm
mitted for tthe degree oof
Doctor of Philossophy
 

This page intentionally left blank

 
 Advisors - Examination Committee
Professor Vassilis Kostopoulos Supervisor
Associate Professor Thrasivoulos Panidis Advisor
Professor Dimosthenis Polyzos Advisor
Professor Georgios Papanicolaou Examiner
Associate Professor Nikolaos Siakavellas Examiner
Assistant Professor Georgios Psarras Examiner
Lecturer Theodoros Loutas Examiner

 
Department of Mechanical Engineering and Aeronautics
University of Patras
6/2/2013
 Authorr: Athanasoppoulos Nikollaos
[email protected]
[email protected]

All rights are reeserved by Athanasopoul

A los Nikolaoss, and contennt may not be
reproduuced, dissemminated, published in anyy form or byy any means,, except with h the
prior written permisssion of the Author.
A

Some parts of the manuscript have been published

p priior in full or partial form
m in
Internattional Researrch Journals. All the apppropriate perrmissions havve been prov vided
Copyright Clearance Cennter (CCC), iin order to reepublish or ddisplay conteent in
by the C
print and electronnic form for the specific thesis/disserrtation and oonly, under the
t
wing confirm
follow mation numbeers (110705448 / 110705449 / 110705550 / 1107055 51 /
11070555 / 11070559).

C
Copyright 2013 by Athaanasopoulos Nikolaos

"http://ccreativecomm
mons.org/licennses/by-nc-ndd/3.0/deed.en__US"

6/2/20013
Acknowledgments i
 

This study is dedicated to my parents, because of their sincere support over the years
ii Acknowledgments
 

Acknowledgments
The present thesis was carried out in the Applied Mechanics Laboratory, Mechanical
Engineering & Aeronautics Department, University of Patras during 2008 – 2013.

Now, having completed this thesis, I would like to sincerely thank:

my supervisor Professor Vassilis Kostopoulos for the collaboration, his guidance and the
opportunities he has offered me during all these years,
Associate Professor Thrasivoulos Panidis for his guidance and cooperation during this thesis
Professor Dimosthenis Polyzos for his interest, the corrections and the participation in the
advisor committee,
Associate Professor Nikolaos Siakavellas for his valuable help in thermography, but mainly
for the knowledge he has provided me with, through his lectures on “Numerical methods in
partial differential equations”,
Professor Georgios Papanicolaou for his help when I was undergraduate student and his
participation in the examination committee,
Assistand Professor George Psarras for his participation in the examination committee,
Lecturer Theodoros Loutas for his participation in the examination committee and his
corrections.

I would like to sincerely thank Ms. Polyxeni Souridi for her support and valuable help in the
English manuscript. Also, my friends D. Roulias, Dr. A. Papacharalampopoulos, Dr. D.
Sikoutris, Dr. D. Mazarakos for their support during all these years.

Finally, I would to thank my intimate friend Angelos Vassiliou for his sincere friendship
during the past 17 years, and the endless discussions over every subject (science, philosophy,
arts, life) which have contributed in the shaping of my personality and way of thinking (more
specific in synthesis, analysis, controversy, acceptation and in transformation of everything).
 iii
 

Study hard what interests you the most in the most undisciplined,
irreverent and original manner possible.
Richard P. Feynman
iv
 

 
 v
 

Calcuulation of electricall conducttivity and electroth

hermal
analyysis of muultilayered carbon reinforceed compoosites:
application to damagge detectiion
 

Ph.D. Thesiss
Departmeent of Mechaanical Engineering and Aerronautics,
University of Paatras
6/2/2013
vi
 

Περίληψη

Κατά τη διάρκεια της παρούσας διδακτορική διατριβής, αποδείχθηκε ότι η ηλεκτρική

αγωγιμότητα των πολύστρωτων και ηλεκτρικά ανισότροπων υλικών με ίνες άνθρακα, μπορεί
να εκφραστεί από έναν ισοδύναμο τανυστή δεύτερης τάξης, ο οποίος είναι το άθροισμα των
τανυστών κάθε στρώσης. Ο ισοδύναμος τανυστής ισχύει υποθέτοντας ότι το πάχος
συγκριτικά με τις υπόλοιπες διαστάσεις του υλικού είναι πολύ μικρό. Η μαθηματική έκφραση
με την οποία μπορεί να προβλεφθεί η ηλεκτρική αγωγιμότητα ενός πολύστρωτου υλικού για
οποιαδήποτε αλληλουχία στρώσεων αποδείχτηκε με συστηματικό τρόπο και βασίζεται στην
αρχή διατήρησης του ηλεκτρικού φορτίου. Η ηλεκτρική αγωγιμότητα κάθε στρώσης
μελετήθηκε πειραματικά στις δύο κύριες διευθύνσεις. Κάθετα στη διεύθυνση των ινών
αναπτύχθηκε ένα εμπειρικό μοντέλο πρόβλεψης της ηλεκτρικής αγωγιμότητας συναρτήσει
του πάχους της στρώσης, της περιεκτικότητας σε ίνες άνθρακα και της θερμοκρασίας. Σε
όλες τις περιπτώσεις μελετήθηκαν πολύστρωτα υλικά ινών άνθρακα χωρίς μήτρα (πορώδης
μορφή-CF preforms) και με πολυμερική μήτρα (CFRPs).
Η επιβεβαίωση της εγκυρότητας του ισοδύναμου τανυστή έγινε με τρεις διαφορετικούς
τρόπους: α) μέσω μετρήσεων της ηλεκτρικής αντίστασης, για διαφορετικές αλληλουχίες
στρώσεων, β) μέσω του φαινομένου Joule, καταγράφοντας και συγκρίνοντας το
αναπτυσσόμενο θερμοκρασιακό πεδίο με το θερμοκρασιακό πεδίο που υπολογίζεται
αριθμητικά, γ) μέσω τρισδιάστατων αριθμητικών μοντέλων όπου τείνουν στην αναλυτική
λύση του δισδιάστατου προβλήματος.
Στη συνέχεια αναπτύχτηκαν ηλεκτροθερμικά μοντέλα με τη μέθοδο των πεπερασμένων
διαφορών με σκοπό τη μελέτη του θερμοκρασιακού πεδίου για διαφορετικές αλληλουχίες
στρώσεων. Το ηλεκτρικό πρόβλημα εκφράζεται από μία ελλειπτική διαφορική εξίσωση όπου
το υλικό είναι ηλεκτρικά ανισότροπο και ομογενές ή μη ομογενές ενώ το θερμικό πρόβλημα
είναι θερμικά ανισότροπο και ομογενές. Χρησιμοποιώντας τον ισοδύναμο τανυστή το
τρισδιάστατο πρόβλημα μετατρέπεται σε ένα δισδιάστατο πρόβλημα με αποτέλεσμα να
απαιτούνται λιγότεροι πόροι για την επίλυση του προβλήματος.
Η συγκεκριμένη εργασία μπορεί να χρησιμοποιηθεί σε μία πληθώρα εφαρμογών όπως
στην ανάπτυξη και στη λειτουργία θερμαινόμενων στοιχείων ενισχυμένων με ίνες άνθρακα
(καλούπια όπου το θερμαντικό στοιχείο το αποτελούν οι ίνες άνθρακα) αλλά και στην
ανίχνευση βλάβης συνθέτων υλικών με αγώγιμη ενίσχυση.
 vii
 

Abstract

During this thesis, it has been proved that the electrical conductivity of multilayered and
electrically anisotropic carbon fiber materials can be expressed by an equivalent second order
tensor, which is equal to sum of each layer’s electrical conductivity tensor. The
aforementioned equivalent electrical conductivity tensor is valid assuming that the material’s
thickness is negligible compared to the other dimensions of the body. The mathematical
expression for the prediction of the electrical conductivity of a multilayered material for any
stacking sequence, is based on the electric current conservation, and was validated using
different methods. Each layer’s electrical conductivity was experimentally studied at the two
principal directions. Transverse to the fibers’ direction, an empirical model was developed for
the prediction of the electrical conductivity as a function of the layer’s thickness, of the fibre
volume fraction and of temperature. All cases involved the study of multidirectional and
unidirectional carbon fiber materials without the presence of matrix (porous form – CF
preform) as well as in the presence of polymeric matrix (CFRP).
The validation of the equivalent tensor was achieved through three different ways: a)
through the measurement of the electric resistance, for various stacking sequences, b) through
the Joule heating effect, by recording and comparing the developing temperature field to the
respective numerically calculated, c) through 3D numerical models which approximate the
analytical solution of the 2D domain problem.
Moreover using the finite difference method, certain electrothermal models were
developed in order to study the temperature field for different stacking sequences. The
electrical problem can be expressed by an elliptic PDE, for the case where the material is
electrically anisotropic and homogeneous, or non-homogeneous. On the other hand, the
transient heat transfer problem involves the case where the material is thermally anisotropic
and homogeneous. Using the equivalent tensor, the 3D domain problem is simplified to a 2D
domain problem resulting in less computational requirements for the solution of the problem.
The present research study could be used in a plethora of application, such as the
development of carbon fibre reinforced heating elements (direct heating CFRP molds) as well
as damage detection in multidirectional composite materials with electrical conductive
reinforcement.
viii
 
Contents ix
 

Contents

Contents…………………………………………………………………………… ix
Table of Figures…………………………………………………………................ xi
List of Tables……….…………………………………………………………....... xvi
Nomenclature………………………………...…………………………………..... xvii

1. Introduction…………………...……………………………………………….. 1
1.1. Electrical conductivity of the continuous CF media………....…..……… 2
1.2. Heat transfer and Joule effect in anisotropic media………..……...…….. 4
1.3. A review on applications……………………………………...….…..….. 4
1.4. Thesis structure……………………………………………........……...… 6
1.5. Originality………………………………………………………....…….. 7

2. Governing equations............................................……………………..……..... 9
2.1. Steady current in anisotropic media…………………....……….….......... 10
2.2. Joule effect…………………………………………...……………….…. 14
2.3. Heat transfer in anisotropic media…...........………....………..……….... 15

3. Electrical conductivity of the unidirectional CF media….................................. 19

3.1. Electrical conductivity of the single CF………………………...……...... 20
3.2. CFRPs’ electrical conductivity models………………………...………... 21
3.2.1. Percolation model………………………………………...……... 22
3.2.2. Bueche model…………………………………………..………... 22
x Contents
 

3.2.3. Nielsen model……………………………………………..……... 23

3.2.4. Berger - McCullough model……….....………………..……....... 24
3.2.5. Ondracek model…………………………………………...…….. 24
3.2.6. Effective medium theory………………………….……………... 25
3.2.7. Fiber contact model………………………………….…………… 25
3.3. Experimental study and prediction of the electrical conductivity of
unidirectional layers………………...…………….……………………... 27
3.3.1. Experimental setups for the EC measurements.............................. 28
3.3.1.1. Experimental setup for the dry CF preform.................... 29
3.3.1.2. Experimental setup for the CFRPs.................................. 30
3.3.2. Ohmic behavior of the CF preform and CFRPs........……...…….. 32
3.3.3. Prediction of electrical resistivity for dry unidirectional layers…. 33
3.3.3.1. Resistivity as a function of thickness and direction.…... 34
3.3.3.2. Resistivity as a function of thickness and fiber volume
fraction………………………………………………… 47
3.3.4. Prediction of electrical resistivity for unidirectional
CFRPs...................………………………………………..……... 58
3.3.4.1. Resistivity as a function of thickness………….………. 58

4. Equivalent EC tensor for multidirectional layered media….....……………….. 65

4.1. Calculation of the equivalent EC tensor……….............…..……………. 66
4.2. Measurements and validation of the EC tensor…………….……………. 70
4.2.1. Microstructural characterization…………………….....……..…. 71
4.2.2. Validation of the equivalent EC tensor via electrical resistance
measurements……………………….…………………………… 75
4.2.3. Sensitivity of measurements……………………………...........… 80
4.3. 3D domain numerical modeling of multidirectional CF media......…....… 81
4.3.1. Rectangular domains – Mixed boundary conditions……...…...… 81
4.3.2. Circular domains – Dirichlet boundary conditions……….……... 85

5. EC and thermal properties as a function of temperature……..........…………… 89

5.1. Experimental apparatus and measurements consistency for the
unidirectional CF preform………………….………………...………….. 90
5.2. EC measurements as a function of temperature……..…………………… 92
5.3. Derivation of equivalent EC tensor as a function of temperature……….. 95
5.4. Thermal properties as a function of temperature………………………… 97
5.4.1. Specific heat capacity measurements……………...…………….. 97
5.4.2. Thermal conductivity measurements……………………….….... 98
5.5. Validation of measurements…………………..…………..…………..…. 98
Contents xi
 

6. Numerical analysis of the electro - thermal problem of the unidirectional and

multidirectional media………………………………………...………………. 101
6.1. Electrical homogeneous media…………………………………..….…… 102
6.1.1. Numerical solution of the electrical problem using FDM and
valiadation using FEM………………………………….……..... 105
6.1.2. Electrothermal analysis in porous CF media…...……………..... 115
6.1.2.1. Symmetric laminates…………………………………… 116
6.1.2.2. Anti-symmetric laminates……………………………… 122
6.1.2.3. Asymmetric laminates……………………….……...…. 123
6.1.3. Electrothermal analysis in CFRP media………………………... 125
6.1.3.1. Symmetric laminates…………………………………… 125
6.1.3.2. Anti-symmetric laminates……………………………... 130
6.1.3.3. Asymmetric laminates……………………………….… 131
6.1.4. Electrothermal analysis in circular CFRP domains..…………… 131
6.2. Electrical heterogeneous media………………….……………...…..….... 133

Application on multifunctional structural heating elements:

7. Damage detection of CFRPs via Joule effect....……………..……………...…. 139
7.1. Interlaminar damage detection (delamination)..………………………… 140
7.2. Fiber and matrix fracture..............................................….………..…….. 147

Conclusions…………..……………………….……………………….…………… 151

References…………………………………………………………………………. 155

Appendix I…………………………………………………………………………. 165

Appendix II……………………………………………………………………...… 169
xii
 

Table of Figures
FIG. 1.1 Unidirectional carbon fiber preform (T700SC) in different magnification 2
levels: (a), (b) unidirectional layer, (c) cross section detail.
FIG. 2.1 Coordinate system (x1, x2) and principal axis (ξ1, ξ2) of the unidirectional 11
CF medium.
FIG. 2.2  Electrical current flow and equepotential lines for (a) isotropic and (b) 13
anisotropic medium.
FIG. 2.3  Boundary conditions (Γd = Dirichlet, Γs = Neumann). 14
FIG. 3.1 Fiber volume fraction as a function of the medium thickness for 28
unidirectional materials.
FIG. 3.2  (a), (b) Experimental setup, (c) UD – inlay T700SC 160gr/m2. 30
FIG. 3.3  Thermography of the CFRP specimens in order to ensure good electrical 31
connections. (a), (b) very good connections, (c), (d) deficient connections.
FIG. 3.4 Ohmic behavior of the porous CF and CFRP media at (a) 90o fiber 32
direction and (b) 0o fiber direction.
FIG. 3.5  Length to width ratio (λ) and specimens geometry. 36
FIG. 3.6  VR: experimental and theoretical results as function of length to width 37
spatial ratio (λ) and the fiber direction (θ) of the UD preform for thickness
h = 0.18mm.
FIG. 3.7 Critical length to width ratio (λcr) as a function of fiber direction (θ). 38
FIG. 3.8  VR (ρ11) as a function of direction (θ) for length to width ratio (λ=7.67). 39
(a) thickness h1 = 0.18mm, (b) thickness (h2 = 5.40mm).
FIG. 3.9  Volume resistivity (ρ11) as a function of direction (θ) and thickness (h) for 40
length to width ratio (λ = 7.67).
FIG. 3.10 (a) Volume resistivity ρ11 as a function of carbon fiber direction (θ) for 41
two different values of thickness (h) and length to width ratio (λ>> λcr),
(b) Theoretical VR ρ11 as a function of carbon fiber direction (θ) and the
thickness.
FIG. 3.11  (a) Volume resistivity ρ11 as a function of direction (θ) and thickness (h) 43
for λ >> λcr, (b) absolute difference Δρ as a function of direction (θ).
FIG. 3.12  Analytical formulation and the experimental results of the volume 44
resistivity as a function of length to width ratio λ and fiber direction for
thickness H1 = 0.18mm and H2 = 5.4mm.
FIG. 3.13 Volume resistivity as a function of ratio (λ) and thickness for direction θ = 45
45ο.
FIG. 3.14  (a) Electric resistivity as a function of time, (b) variation of electric 47
resistivity as a function of time.
FIG. 3.15  (a) Preform thickness as a function of the applied pressure, (b) preform 48
thickness per layer as a function of the applied pressure.
FIG. 3.16 (a) Thickness per layer for different layers number as a function of the 51
fiber volume fraction, (b)
FIG. 3.17  Electrical resistance measurements as a function of pressure and number 52
 xiii
 

of layers.
FIG. 3.18  Volume resistivity as a function of pressure and (a) number of layers, (b) 53
medium thickness.
FIG. 3.19 (a) VR as a function of pressure and number of layers, (b) schematic 54
presentation of the VR calculation.
FIG. 3.20  Volume resistivity as a function of fiber volume fraction. 56
FIG. 3.21  a) Experimental results of pressure – thickness curves and b) experimental 57
results of pressure - fiber volume fraction curves versus the calculated
results using Chens model.
FIG. 3.22 Calculated and experimental curves of VR as a function of fiber volume 58
fraction using the empirical relation and the Chens model.
FIG. 3.23 Volume resistivity transverse to the fibers as a function of number of plies 59
and pressure using different tools and manufacturing technics.
FIG. 3.24 Volume resistivity transverse to the fibers as a function of pressure for 60
different number of plies using CFRP tool.
FIG. 3.25 Rich resin layer in CFRP microstructure material. 60
FIG. 3.26 (a), (b), (c) CFRP microstructure using constant pressure (1bar) and (d), 61
(e), (f) pressure between (1 bar with extra pressure) alternatively).
FIG. 3.27 CFRP microstructure using pressure between (1 bar -1.2bar). 62
FIG. 3.28 (a), (b), (c) CFRP microstructure using constant pressure (5bar). 63
FIG. 4.1  (a) Geometry of the domain and boundary conditions, (b) detail of the thin 66
multilayer medium.
FIG. 4.2 Representation of the EC tensor of each CFRP layer and the equivalent 69
EC tensor for the multidirectional CFRP for different stacking sequences,
(a) (0/30/30/0), (b) (45/30/30/45), (c) (±45), (d) (90/60/-45/90/-45/60/90).
(ξ1, ξ2 are the principal axes of the equivalent material).
FIG. 4.3  Optical microscopy at multidirectional laminates as a function of 74
manufacturing pressure. (a), (b) 0.01MPa, (c), (d) 0.1MPa, (e), (f)
0.8MPa.
FIG. 4.4  Thermal camera images at two different pressure levels for (±45) stacking 72
sequence. a) 0.01 MPa and b) 0.8 MPa.
FIG. 4.5 Dimensionless temperature of (±30) stacking sequence at L1, L2 and L3 75
regions for two different pressure levels (0.1MPa and 0.8MPa).
FIG. 4.6  Differences between current density distribution and electric potential 76
field for a unidirectional laminate (45o) and a multidirectional laminate
(±45). (a) orthogonal domain with boundary conditions, (b) current
density streamlines at 45o, (c) electric potential field at 45o, (d) current
density streamlines at (±45), (e) electric potential field at (±45).
FIG. 4.7  Current density distribution through the material thickness. (a) isotropic, 77
(b) thick anisotropic multidirectional laminate, (c) thin anisotropic
multidirectional laminate.
FIG. 4.8 Relative error between the electrical measurements and theory (Eq. 8) for 79
various manufacturing pressure levels.
FIG. 4.9  (a) Measured electrical resistance error as a function of stacking sequence 80
xiv
 

(±θ) and the misalignment of the layer direction (±γ) during the
manufacturing stage of the specimen, (b) relative error as a function of
fiber direction deviation at (±30, ±45, ±60) stacking sequences.
FIG. 4.10  Relative error of the measured electrical resistance for different anisotropy 81
levels and misalignment in fiber direction (γ) – (1ο, 2ο, 3ο, 5ο).
FIG. 4.11 Equipotential surfaces on a stacking sequence (±45) for different layer 82
thickness (a) 20mm/layer, (b) 5mm/layer, (c) 0.125mm/layer.
FIG. 4.12  Dimensionless current density as a function of dimensionless thickness for 83
different stacking sequences (±θ).
FIG. 4.13  Dimensionless current density as a function of dimensionless thicknesses 84
for different stacking sequences (a) ±15, (b) ±45, (c) ±60.
FIG. 4.14 Numerical solution of the 3D problem at the worst case and at the middle 84
of the specimen. Comparison with the analytical solution.
FIG. 4.15 Equipotential surfaces for circular domains and Dirichlet boundary 85
conditions for different ratios (μ) and thicknesses.
FIG. 4.16  (a) Dimensionless electric potential field for circular domains as a 86
function of layers thickness, (b) absolute error between the numerical
solution (3D space) and the (2D space) analytical solution.
FIG. 4.17  Comparison between numerical and analytical solutions as a function of 87
ratio Μ = Rmax/Rin for layer thickness h = 0.05mm.
FIG. 4.18 Comparison between numerical and analytical solutions as a function of 87
ratio μ = Rmax/Rin for layer thickness h = 0.125mm.
FIG. 4.19  Absolute error between the analytical and numerical solution for different 88
ratios μ. a) h = 0.125mm, b) h=0.05mm.
FIG. 5.1  (a) Manufactured apparatus resembling black body environment, (b) 91
apparatus schematic as described in the (ASTM Standard (1999),
FIG. 5.2 Volume resistivity as a function of temperature at fiber direction (0o) for 93
the T700S -12K and HTA40-3K carbon fiber tows (Athanasopoulos et. al
(2012)).
FIG. 5.3  Volume resistivity as a function of temperature transverse to fiber 94
direction (90o) for the T700S -12K under pressure of 0.1MPa.
FIG. 5.4 Specific heat capacity Cp as a function of temperature for the T700S-12K, 97
HTA40-3K carbon fiber tows (Athanasopoulos et. al (2012)), and a
common CFRP material
FIG. 5.5  Temperature response of the T700S-12K tow to the applied voltage pulses 99
at the tow center. Heating and cooling cycles
FIG. 6.1 Configuration types and numerical examples of CF multilayer media. 103
FIG. 6.2  FIG. 6.2. Rectangular domain Ω and boundary conditions. 105
FIG. 6.3 a) solution after 2100s for 1step algorithm (6601 nodes), (b) solution after 106
366.2s for the multistep algorithm (26001 nodes).
FIG. 6.4  Electric potential field using FDM and FEM for different number of 107
nodes. (a) 51681 nodes, (b) 13041 nodes, (c) 231 nodes, (d) 66 nodes.
FIG. 6.5 Electric potential field using FDM and FEM for different number of 108
nodes. (a) 51681 nodes, (b) 13041 nodes, (c) 231 nodes, (d) 66 nodes.
 xv
 

FIG. 6.6  Electric field using FEM and FDM at the edges of the specimen (a) 13041 108
nodes, (b) 205761. [45o], L = 400mm, W = 200mm, σ11 = 22169 S/m, σ12
= 22119 S/m, σ22 = 22169 S/m), Δφ = 10V.
FIG. 6.7 Electric potential field using FDM and FEM for a CFRP material with 109
stacking sequence (0/30/30/0) at the edge of the domain.
FIG. 6.8  Electric field using FDM and FEM (a) FEM, (b) FDM. Detail of the error 109
at the edge of the domain.
FIG. 6.9 Electric field using FDM and FEM at the edges of the domain. 110
FIG. 6.10  Electric potential field for a high anisotropic material using different 111
number of nodes.
FIG. 6.11 Electric potential and electric field using FEM at the edges of the 111
specimen (a) electric potential as a function of anisotropy level using 861
nodes, (b) electric potential as a function of number of nodes -861, 13041,
207561 nodes-, (c) electric field as a function of anisotropy level using
861, (d) electric field as a function of number of nodes -861, 13041,
207561 nodes-. [45o], L = 400mm, W = 200mm, Δφ = 10V.
FIG. 6.12  Flow diagram of the finite difference algorithm. 114
FIG. 6.13 Numerical results of the symmetric unidirectional – specially orthotropic 117
medium, (04) or (904) in dimensionless form. (a) Current density, (b)
electric potential field, (c), generated heat, (d) temperature field, (e)
measured temperature field (04), (f) measured temperature field (904). λ =
3, L/H =611; (04): σ22/σ11 = 1.3 10-3, σ12/σ11 = 0; (904): σ22/σ11 = 666.7,
σ12/σ11 = 0.
FIG. 6.14  Numerical and experimental results of the symmetric unidirectional - 117
generally orthotropic medium for three different fiber directions, (304), (a)
(304): σ22/σ11 = 0.335, σ12/σ11 = 0.576.
FIG. 6.15 Numerical and experimental results of the symmetric unidirectional - 118
generally orthotropic medium for three different fiber directions, (454),
(604) in dimensionless form. (λ = 3, L/H = 611, σ1/σ2 = 666.7). (a) (454):
σ22/σ11 = 1, σ12/σ11 = 0.997; (b) (604): σ22/σ11 = 2.988, σ12/σ11 = 1.722.
FIG. 6.16 Numerical and experimental results of the symmetric multidirectional - 119
generally orthotropic medium in dimensionless form. λ = 3, L/H =611,
σ1/σ2 = 766.7. (a) (0/30)s: σ22/σ11 = 0.144, σ12/σ11 = 0.247; (b) (45/30)s:
σ22/σ11 = 0.601, σ12/σ11 = 0.744; (c) (60/30)s: σ22/σ11 = 1, σ12/σ11 = 0.863.
FIG. 6.17  Numerical and experimental results of the symmetric multidirectional - 120
generally orthotropic medium in dimensionless form. λ = 3, L/H =611,
σ1/σ2 = 766.7. (90/30)s: σ22/σ11 = 0.144, σ12/σ11 = 0.247.
FIG. 6.18 Numerical results of the anti-symmetric multidirectional – cross ply or 121
angle ply or quasi isotropic medium in dimensionless form. (a) Current
density (b) electric potential, (c) generated heat, (d) temperature field
(90/0)s, (±30)s, (±45)s, (0/±60)s. λ = 3, σ1/σ2 = 766.7. (e) Measured
temperature field of (90/0)s: L/H =611, σ22/σ11 = 1, σ12/σ11 = 0, (f)
measured temperature field of (±30)s: L/H =611, σ22/σ11 = 0.334, σ12/σ11 =
0; (g) measured temperature field of (±45)s: L/H =611, σ22/σ11 = 1, σ12/σ11
= 0; (h) measured temperature field of (0/±60)s: L/H =407, σ22/σ11 = 1,
σ12/σ11 = 0.
xvi
 

FIG. 6.19  Numerical results and thermal camera image of the asymmetric 122
multidirectional medium (45/60/-45/-60) in dimensionless form. λ = 3,
L/H =611, σ1/σ2 = 766.7. (a) Current density (b) electric potential, (c)
generated heat, (d) temperature field, (e) measured temperature field
σ22/σ11 = 1.664, σ12/σ11 = 0.
FIG. 6.20 Numerical results of the asymmetric multidirectional medium (-30/60) or 124
(-20/70). (a) Current density (b) electric potential, (c) generated heat, (d)
calculated temperature field. λ = 3, L/H =1222, σ1/σ2 = 766.7. (e) thermal
camera image (-30/60): σ22/σ11 = 1, σ12/σ11 = 0; (f) thermal camera image
(-20/70): σ22/σ11 = 1, σ12/σ11 = 0.
FIG. 6.21  Numerical results and thermal camera image of the asymmetric 124
multidirectional medium (-30/45) in dimensionless form. (λ = 3, L/H
=1222, σ1/σ2 = 766.7). (-30/45): σ22/σ11 = 0.601, σ12/σ11 = 0.053.
FIG. 6.22 Numerical results of the symmetric unidirectional – specially orthotropic 126
medium, (04) or (904) in dimensionless form. (a) Current density, (b)
electric potential field, (c), generated heat, (d) temperature field, (e)
measured temperature field (04), (f) measured temperature field (904). λ =
3.13, L/H =420; (04): σ22/σ11 = 1.129 10-3, σ12/σ11 = 0; (904): σ22/σ11 =
885.7, σ12/σ11 = 0.
FIG. 6.23  Numerical and experimental results of the symmetric unidirectional - 127
generally orthotropic medium for three different fiber directions, (304),
(454) in dimensionless form. (λ = 3.13, L/H = 420, σ1/σ2 = 885.7). (a)
(304): σ22/σ11 = 0.334, σ12/σ11 = 0.576; (b) (454): σ22/σ11 = 1, σ12/σ11 =
0.997.
FIG. 6.24 Numerical and experimental results of the symmetric generally orthotropic 128
media in dimensionless form. (a) (60/30)s: σ11 = 22169, σ12 = 19155, σ22 =
22169; (b) (45/30)s: σ11 = 27698, σ12 = σ21 = 20637, σ22 = 16639.
FIG. 6.25  Numerical and experimental results of the symmetric generally orthotropic 129
media in dimensionless form. (0/30)s: σ11 = 38757, σ12 = 9577.6, σ22 =
5579.6.
FIG. 6.26 (a), (b), (c), (d) Numerical results of the multidirectional anti-symmetric 129
angle ply or quasi isotropic medium in dimensionless form. (e) (±30)s: σ11
= 33228, σ12 = σ21 = 0, σ22 = 11109; (f) (±45)s: σ11 = σ22 = 22168.5, σ12 =
σ21 = 0; (g) (±60)s: σ11 = 11109, σ12 = σ21 = 0, σ22 = 33228; (h) (0/±60)s:
σ11 = σ22 = 22168.5, σ12 = σ21 = 0.
FIG. 6.27  (a), (b), (c), (d) Numerical results and thermal camera image of the 130
multidirectional anti-symmetric and the asymmetric media in
dimensionless form. (e) (45/60/-45/-60): σ11 = 16639, σ12 = 0, σ22 = 27698;
(f) (-30/60): σ11 = σ22 = 22168.5, σ12 = σ21 = 0.
FIG. 6.28 (a), (b), (c), (d) Numerical results and thermal camera image of the 131
multidirectional anti-symmetric and the asymmetric media in
dimensionless form. (-30/60): σ11 = σ22 = 22168.5, σ12 = σ21 = 0.
FIG. 6.29  Numerical results and thermal camera images for two circular 132
multidirectional CFRPs using the equivalent EC tensor (Rmax/Rin = 4, H =
0.726mm, h = 0.121mm). (a), (b) current density, (c), (d) electric potential
field (e), (f) calculated temperature field, (g), (h) measured temperature
 xvii
 

field.
FIG. 6.30 Algorithm using FDM for the electrical non-homogeneous medium. 134
FIG. 6.31 Calculated and measured temperature distribution at the edge and at the 135
middle of the coupon (0/30/30/0).
FIG. 6.32 Calculated and measured temperature distribution at the diagonal of the 136
coupon (0/30/30/0).
FIG. 6.33 Calculated temperature distribution at the edge of the coupon (0/30/30/0) 136
for two different electric potential differences.
FIG. 6.34 Specific heat capacity and EC as a function of space for two different 137
electric potential differences at steady state.
FIG. 7.1  (a) Illustration of a CFRP domain with an interlaminar damage 141
(delamination) and an optical microscope image of the delaminated area
(non-electrically conductive area between the layers), (b) through the
thickness illustration at the delamination region, (c) tensor ellipse at the
healthy material (i.e., ±45 stacking sequence) and tensor ellipses at the
damage area.
FIG. 7.2 Numerical and experimental results in dimensionless form for a (±45) 144
stacking sequence with dimensions (L = 128mm, W = 45mm). (a) healthy
material, (b) artificial rectangular damage in the middle of the specimen
(28mm x 45mm, white dashed area) “enhanced online”, (c) artificial
rectangular damage at the edge of the specimen (32mm x 20mm)
“enhanced online”. In all cases the thickness of the laminate is 0.25mm.
Material properties: ρm=1092kg/m3, ρf=1800kg/m3, cm=1187J/kgK,
cf=750J/kgK, (1-φ)=0.7, k11=k22=5.42W/mK, k33=0.65W/mK,
σ1=44287S/m, σ2=50S/m, σ3=σ33=50S/m.
FIG. 7.3  (a) Measured temperature field as a function of time (along the width 145
delamination between a (±45) CFRP medium), (b) measured and
calculated temperature field as a function of time (edge delamination
between a (±45) CFRP medium).
FIG. 7.4 (a) calculated temperature field during the heating of the CFRP for a 146
delamination at the center of a (±45) stacking sequence, (b) thermography
image in dimensionless form of an interlaminar damage with rectangular
geometry (28mmx28mm) at the center of a (±45) stacking sequence
specimen (110mmx128mm) (Δφ=0.82V, Ι=3.10Α, Τmax=41oC), (c)
thermography image in dimensionless form of a second interlaminar
damage with the same geometry and stacking sequence.
FIG. 7.5  (a) measured temperature field during the heating of the CFRP specimen 147
with 2 layers of 0o fiber direction, (b) thermography image of a circular
damage (diameter of the artificial interlaminar damage=40mm) at a CFRP
plate (400mm x 200mm) with stacking sequence (0/90/45/-45/-
45/45/90/0) and thickness hCFRP≈1mm (the magnified image has been
captured at time t2≠t1), (Δφ=3V, Ι=35Α).
FIG. 7.6 (a) CFRP specimen with 4mm crack, (b) measured temperature field, (c) 148
calculated temperature field.
FIG. 7.7  (a) CFRP specimen with 9mm and 3mm cracks, (b) measured temperature 148
field, (c) calculated temperature field.
xviii
 

FIG. 7.8 Measured temperature field in the case of matrix cracking (2mm crack). 149

List of Tables
TABLE 3.1 Volume resistivity values of various single CFs at room 21
temperature
TABLE 3.2 Experimental results of CF tow electrical resistance for the two 30
different methods.
TABLE 3.3 Experimental results of UD T700SC electric resistivity ρo in fiber 33
direction.
TABLE 3.4 Experimental results of UD T700SC VR transverse to the fiber 34
direction ρ90.
TABLE 3.5 Results of calculated and measured VR for various directions, 46
aspect ratio (λ) and thickness.
TABLE 4.1 Electrical resistance measurements of multidirectional CFRPs at 78
different manufacturing pressure levels.
TABLE 5.1 Carbon fiber tow VR at room temperature and geometric details. 92
TABLE 5.2 Thermal coefficients of the VR for CFRPs and CF preforms along 94
and transverse to the fiber direction.
TABLE 5.3 Thermal properties of the molds materials 98
TABLE 7.1 Governing equations, generated heat for the different regions and 142
EC tensor’s values for each area.

Nomenclature
J Current density A/m2
ρ Electrical resistivity
σf Electrical conductivity of the fiber
σm Electrical conductivity of the matrix
φ Porosity
(1-φ) Fiber volume fraction
vf Fiber volume fraction
E Electrical field
m1 cosθ
m1 sinθ
a Thermoelectric tensor
φ  Electric potential field
θ Fiber direction
σ EC tensor
σik Coefficient of EC tensor
 xix
 

Τ Temperature
t time
Q Generated heat
k Thermal conductivity tensor
q Heat flux
kik Coefficient of thermal conductivity tensor
kf CF thermal conductivity
km Matrix thermal conductivity
c Specific heat capacity
ρ Density
R Electrical resistance
nf Number of CFs
L Length
W Width
H Total thickness of the laminate
h Thickness of each layer
N Number of layers
Nk Critical number of layers
df CF diameter
Α Cross section area
λ Ratio length to width or external to internal radius
P Pressure
I Electrical current
S Cross section area
Sn Cross section area of each layer
r Internal radius
R External radius
σΤ Equivalent EC tensor as a function of temperature
ε Total hemispherical emissivity
σ Stefan Boltzmann constant
Pr Prantl number
v Kinematic viscosity
β Coefficient of thermal expansion
δ Characteristic length
Ra Rayleigh number
Gr Grashof number
Nu Nusselt number
h Heat transfer coefficient
g Gravitational acceleration

Subscripts and Superscripts

f Fiber
m Matrix
H Healthy
D1,2 Damaged area
xx
 

Τ Transpose matrix
Τ Temperature

Abbreviations
CF Carbon fiber
CFRP Carbon fiber reinforced plastic
VR Volume resistivity
EC Electrical conductivity
CTE Coefficient of thermal expansion

 
Introduction 1
 

1. start
Chapter 1
Introduction

Carbon fibres (CFs) are used in many industrial applications, mainly as a reinforcing
material in low weight applications. Further to their excellent mechanical properties,
CFs are also electrically conductive. Scientific interest has been triggered towards the
study of carbon fibres electrical properties, due to the fact that they offer themselves
as a potential material in a wide variety of applications. The expansion of areas of
application for CFs is stimulated by their attractive properties, not found in other
materials, such as electrical conductivity, low density, low-to negative coefficient of
thermal expansion. Also, CFs can be used as multifunctional materials in a plethora of
applications of various scientific fields.
In particular, CF laminates present a significantly high electrical anisotropy, the
prediction of which is of paramount importance as far as multilayered materials are
concerned. Especially in self-heating structural materials, the determination of the
electrical conductivity (EC) of the multilayered medium is the most crucial factor for
the calculation of the generated heat and the prediction of the temperature field.
2 Chapter 1

1.1. Electrical conductivity of the continuous CF anisotropic

media
The highly anisotropic behavior of the volume resistivity (VR) of unidirectional CFs
occurs due to the two, different electron pathways. Along the fiber direction, the
current flows through the fibers and the VR of the unidirectional (UD) medium
depends upon the resistivity of the CFs and upon fiber volume fraction. In transverse
to the fiber direction, the current is assumed to follow a stochastic zigzag path across
the fiber-to-fiber contacts, Fig. 1.1 (Knibbs & Morris (1974), Weber & Kamal (1997),
Greenwood et al. (1975), Tse et al. (1981)).
Several models have been presented to predict the electrical conductivity of the
UD medium along and in transverse to the fibres' direction, as a function of the fibre
volume fraction. But, according to Weber & Kamal (1997), most of them cannot be
experimentally validated. As it will become clear in further chapters, the above occurs
due to the fact that the experimental measurements had not involved the real
microstructure of the material.

FIG. 1.1. Unidirectional carbon fiber preform (T700SC) in different magnification levels:
(a), (b) unidirectional layer, (c) cross section detail.

The electrical conductivity (EC) of the unidirectional carbon fiber reinforced

plastics (CFRPs) and dry CF preforms has been thoroughly studied and is dependent
upon the a) carbon fibre EC (Shindo (2000)), b) fibre direction (Knibbs & Morris
Introduction 3
 

(1974), Weber & Kamal (1997), Greenwood et al. (1975), Tse et al. (1981)), c) fibre
volume fraction (Weber & Kamal (1997)), d) temperature (Shindo (2000),
Athanasopoulos et al. (2012), Takahashi & Hahn (2011)), e) as well as the plies’
thickness (Athanasopoulos & Kostopoulos (2011)). It has been proven that the
specimens' dimensions with respect to the fibres' direction, should be studied only in
the case where the specimens' length is significantly bigger compared to the width
(Athanasopoulos & Kostopoulos (2011)). Dimensions should be accounted only in the
case where the measurements are taken along or transverse to the fibres' direction.
The material's electric conductivity is affected by the specimen's width (Tse et al.
(1981)) and thickness (Athanasopoulos & Kostopoulos (2011)).
In real structures, CFRPs are used in the form of multidirectional laminates. The
determination of the EC of the anisotropic multilayered laminates is the most crucial
factor for the calculation of the electric field and the current density distribution. The
EC of multilayered materials had been studied by Volpe (1980), who proposed an
empirical expression for estimating the EC of a laminate made of a family of
(0/90/45) plies. Another reference has been tracked at the piezoresistivity and
piezoconductivity of orthotropic materials on the basis of electrostatic theory and the
classical thin laminate theory (Xiao et. al (1998)). Most scientific papers involving the
study of CFRP electrical conductivity, have focused on CFRP damage detection
(Angelidis et. al (2004), Ueda & Todoroki (2007), Todoroki et. Al (2003), Kupke et.
al (2000), Todoroki & Tanaka (2001), Todoroki et. al (2001), Angelidis & Irving
(2006), Wen et. al (2010), Schuelr et. al (2000)), on CFRP damage monitoring (Park
et. al (2002), Kostopoulos et. al (2009), Ιrving & Thiagarajan (1998), Wang & Chung
(1996), Wang & Chung (1997), Kaddour et. al (1993)) and on CFRP piezoresistivity
(Todoroki et. al (2009)). All the above concern stacking sequences of either (0), (90)
or (0/90), where in all cases the coordinate system coincides with the EC tensor's
principal axes. Same can be observed in the following studies (Schueler et. al (2000),
Ueda & Todoroki (2008), Park et. al (2006)), where the equation does not involve
cross derivatives. On the other hand, the study of (Menana & Feliachi (2009), Menana
& Feliachi (2011), Megali et. al (2010)) was conducted under the assumption that
each layer is a 3D material of a certain fibre direction and that layers are electrically
insulated between them. Finally, Piche et. al (2011) has presented a model which
examines the electric field distribution taking into account the fibre contacts between
the layers, which constitute a 3D problem. A thorough study is presented in the work
of Athanasopoulos & Kostopoulos (2012)e, according to which, the EC of dry fabrics
4 Chapter 1

can be calculated for each stacking sequence by using an equivalent EC tensor in the
case where the material's thickness is negligible compared to the other dimensions.
The above can be validated for CFRP materials also by the work of (Athanasopoulos
& Kostopoulos (2013), Athanasopoulos & Kostopoulos (2012)d).

1.2. Heat transfer and Joule effect in anisotropic media

Joule’s first-law dictates that when electric current passes through a conductor, heat is
generated, and this is known as the Joule heating effect or (ohmic heating, resistive
heating, direct heating). The determination of the EC of the multilayered medium is
the most crucial factor for the calculation of the generated heat and the prediction of
the temperature field.
The multilayered media are categorized in unidirectional laminates (UD) or
multidirectional laminates (MD) (Datoo (1991)). The calculation of the equivalent EC
tensor in the case of dry CF layered preforms and CFRPs is provided, since it is the
major parameter that controls the generated heat distribution within the CF medium.
The distribution of the generated heat in the case of multidirectional laminates depend
upon the direction of the CF layers. Thus, a particular stacking sequence may provide,
or not provide, a uniform temperature distribution under certain boundary conditions.
Analytical calculations of the EC tensor have been provided, and based on these, the
coupled electro-thermal problem was numerically solved and concluded to the
calculated temperature distribution (Athanasopoulos & Kostopoulos (2012)e).

1.3. A review in applications

The usage of CFs in the Heat Transfer field is of great significance, because of the
unique CF properties. CFs also find use as electrodes of large surface and of excellent
corrosion resistance, and as anti-static components. Conductive textile products and
composites have been considered for electromagnetic shielding and antistatic
purposes in various applications for electrical, electronic industries and the Defense
(Šafárová & Grégr (2010)).
In the composite manufacturing industry, by applying potential difference on CF
preforms, the generated heat and thus the preform temperature is easy to control, it is
only being affected by the applied electric power (Athanasopoulos et. al (2010)). Each
Introduction 5
 

fibre behaves as a thin, ohmic resistor with a diameter of (d ≈ 7μm). Thousands of

fibres compose a surface in the form of CFRP or dry preform. Therefore the surface,
itself, is an ohmic resistor in the form of tow, woven fabric, UD, 3D woven fabric and
multi oriented ply laminates. Various applications involve CFs as heating elements,
using the Joule heating effect. These applications are not widely spread. In
aeronautics, CFs have been used for the de-icing of aircraft propellers. Also, studies
have been carried out concerning the repair of aviation components with CF patches,
cured by resistive heating of carbon fibre reinforcement (Dimoka (2012)). The Joule
heating effect is also used to heat up the dry CF preform, leading to reduced resin
viscosity and thus promoting preform impregnation (Athanasopoulos et. al (2010),
Zhu & Pitsumani (2000), Mawardi & Pitsumani (2003)). After the impregnation
phase is completed, the composite remains electrically conductive, thus further
exploitation of the resistive heating can be obtained (curing, post curing) (Joseph &
Viney (2000)). The use of embedded resistive heating elements provides significant
improvement in cure cycle time and cure uniformity (Zhu & Pitsumani (2000)).
Liquid composite molding (LCM) and autoclave processes for manufacturing CFRPs
are highly temperature sensitive. So far, this problem was addressed using complex
mold setups and large ovens for resin acceleration and laminate curing. All these
procedures are energy-inefficient and increase total cost. The current approach omits
the need of ovens and complex heating systems as it uses the CFRP mold surface as
the heating element. In essence, the heating of the mold is achieved through the use of
CFs (Funke & Brandes (2011), Athanasopoulos et. al (2012)b, Koutsoukis (2011)).
Accordingly, the CFs are used as heating elements covering the whole surface of the
mold, resulting in a uniform heating source. This advantage combined with the very
low CTE of the CFRP, makes this material an excellent choice for the mold
manufacturing industry. The use of CFs as primary heating elements presents
excellent results in terms of uniformity in temperature distribution and geometric
adaptability. CF preforms show very good geometry adaptability and promote
manufacturing of complex geometry parts that require heating. Also, Stephen &
Donald (2008) propose the use of CF tows as heating elements to cure thermosets in
inflatable space structures. CFRPs can be used as multifunctional heating elements
(manufacturing, curing monitoring and damage detection using Joule effect). CF
preforms can be used as flexible heating elements because of the very good
geometrical adaptability, which sets them ideal material to heat up complex
geometries (Athanasopoulos & Kostopoulos (2012)e).
6 Chapter 1

1.4. Thesis structure

After the presence introduction the following chapters are presented:
The second chapter is an analytical description of the governing equations that
constitute the electrothermal problem: a) in electrically homogenous and non-
homogenous anisotropic materials, b) in thermally homogenous, anisotropic
materials.
The third chapter includes a brief presentation of various approaches for the
prediction of the electric conductivity at the two main directions (σ1, σ2) of CFRPs.
Furthermore, the EC has been studied experimentally as a function of thickness and
fibre volume fraction for the case of UD carbon fibre plies. The measurements
involved in the aforementioned study produced diagrams which relate the material's
VR to the material's thickness and the applied pressure. Transverse to the fibers’
direction, an empirical model was developed for the prediction of the electrical
conductivity as a function of the layer’s thickness. Additionally, the model developed
by Chen et. al (2000) was used in order to relate the VR to the carbon fibre volume
fraction.
The fourth chapter includes the proof of the mathematical expression for the
prediction of the electrical conductivity of a multilayered material for any stacking
sequence. It has been proved that the electrical conductivity of multilayered and
electrically anisotropic carbon fiber materials can be expressed by an equivalent
second order tensor, which is equal to sum of each layer’s electrical conductivity
tensor. The aforementioned equivalent electrical conductivity tensor is valid assuming
that the material’s thickness is negligible compared to the other dimensions of the
body. The mathematical expression for the prediction of the electrical conductivity of
a multilayered material for any stacking sequence, is based on the electric current
conservation, and was validated using different methods. Moreover, the composite
materials are categorized by structure, and electrical resistivity measurements of
multilayered materials for various stacking sequences are presented, in order to
validate the equivalent EC tensor. Finally, the electric potential was calculated
through 3D numerical models, assuming that the layers are in total contact with each
other.
The fifth chapter studies the EC and the thermal properties of the medium, as a
function of temperature. The equivalent EC tensor was related to the temperature as
Introduction 7
 

well as to the thermal coefficients of the material's electrical conductivity, along and
transverse to the fibers' direction.
In the sixth chapter, certain numerical models were developed using finite
difference method (FDM), in order to solve the electrothermal problem on an
electrically anisotropic, homogenous or non-homogenous material. The numerical
models use the equivalent EC tensor of various stacking sequences. By means of the
calculated EC tensor of the multilayer preform, the elliptical form of the governing
equation can be solved numerically. Based on this, the generated heat can be
determined. The determination of the EC of the multilayered medium is the most
crucial factor for the calculation of the generated heat and the prediction of the
temperature field. The generated heat in a known domain (Ω), with given boundary
conditions, changes according to the EC tensor of the multilayered medium. Different
stacking sequences lead to different EC tensors. Thus, the resulting generated heat in
the known domain is also different. Introducing the generated heat into the heat
transfer equation, the temperature field over the CF medium can be predicted. For the
experimental verification, a thermal camera was used to record the temperature field
developed on a CF multilayer preform under given electric potential field. The
experimental results were compared to the respective numerical calculations of the
temperature field, where the EC tensor was calculated analytically based on the
proposed methodology. The results were confirmed experimentally via thermal
camera images, for a large number of representative stacking sequences. Also, the
FDM solutions were compared to the FEM solutions. Finally the electrothermal
problem was solved for the electrical non-homogeneous anisotropic material.
In conclusion, the final chapter studies the case of damage detection in CFRPs
using the Joule effect. Numerical models were developed in order to predict the
material's temperature field. The numerical results were compared to the respective
experimental measurements, recorded with a thermal camera.

1.5. Originality
The present work aspires to fill the literature, which concern the prediction of the
direct current EC of multilayered materials, by using a simple mathematical formula.
Specifically, the following have been studied thoroughly for the first time:
8 Chapter 1

i. The EC of a porous unidirectional medium (CF preform) and the derivation of

empirical formulas, along and transverse to the fibres' direction, as a function of
thickness, temperature and fibre volume fraction. Also, the EC of the CFRP
unidirectional medium, along and transverse to the fibers' direction, as a function of
thickness and the microstructure of the material.
ii. The EC of the porous and CFRP multilayered media for any stacking sequence,
in the case where there is no electric gradient along the plies' thickness.
iii. Electrical, electrothermal numerical models and experimental validation for
multilayer anisotropic media using the proposed equivalent EC tensor. The
calculation of the equivalent EC tensor can be used in order to avoid largely memory-
allocating three-dimensional models.
iv. Damage detection using the Joule effect. The mechanism of the phenomenon is
elucidated and validated against experimental results by comparing the measured to
calculated temperature field.
Governing equations 9
 

2. start
Chapter 2
Governing equations

The solution of the electric problem, must satisfy the continuity equation of the
electric current, Eq. 2.1. Taking into consideration that the present study focuses on
the steady state problem, it is obvious that the factor describing the transient state,
must be omitted.
ρ
0
ρ t
 J     J  0 (2.1)
t
In order to solve the above equation, it is necessary that the EC tensor is known.
In the present study, the material might be both electrically and thermally anisotropic.
After having found the electric field and the current density distribution, it is possible
to calculate the generated heat and the temperature field at both transient and steady
state. In this study the problem is assumed, at first, electrically homogenous and
anisotropic and then, electrically non-homogeneous and anisotropic.
10 Chapter 2
 

2.1. Steady current in anisotropic media

CFRPs can be divided into unidirectional and multidirectional laminates (Datoo
(1991)). In the case of the unidirectional laminates, it is known that the electric
current flows inside the material following two different paths, thus creating electrical
anisotropy (Greenwood et. al (1974), Knibbs & Morris (1974), Tse et. al (1981),
Weber & Kamal (1997)). At fibre direction, the current flows along the carbon fibres
(in the ideal case), Fig. 1.1(b). In the case where the material around the fibers is an
insulator, then the electrical conductivity of the unidirectional (UD) preform depends
upon the EC of CFs. Thus, the EC of a CF medium along the fiber direction can be
expressed by Eq. 2.2.
 m 0
 1   f (1   )   m   f v f (2.2)

where (σm) is the EC of the surrounding material, (σf) is EC of the fiber and (vf = 1 -
φ) is fiber volume fraction. The volume resistivity (resistivity, specific electrical
resistance, or volume resistivity) of the material is the inverse of the EC and is
expressed by Eq. 2.3.
1 1
1   ,   0o (2.3)
1  f vf
Transverse to the fibers, the current flows through the random fibre contacts Fig.
1.1(c). Due to the two different flow paths, a material of high electric anisotropy is
occurs (σ1 >> σ2). The EC of a CF medium, and therefore the degree of anisotropy, is
dependent upon the carbon fibers’ EC (σ1, σ2), the CF fibre volume fraction (1 - φ)
(Weber & Kamal (1997)), the material’s thickness (h) (Athanasopoulos &
Kostopoulos (2011)), as well as the temperature (T) (Shindo (2001), Takahashi &
Hahn (2011), Athanasopoulos et. al. (2012)) and the fibers direction (θ) (Knibbs &
Morris (1974), Tse et. al (1981), Xiao et. al (1999)). Using the hypothesis of a
continuum medium, the EC of the CF medium can be expressed by a symmetric
second order tensor (Tse et. al (1981), Xiao et. al (1999), Athanasopoulos &
Kostopoulos (2012)), symmetric due to the Onsager theorem (Landau & Lifshitz
(1984)). Hence for the unidirectional laminates the EC tensor depends upon the
previously mentioned parameters, Eq. 2.4.

σ = σ(σ1, σ2, T, vf, h, θ) (2.4)

Governing equations 11
 

The electric conductivity tensor as a function of the conductivities at the two

main directions, is expressed by the following, well-known formula, Eq. 2.5.

 σ1m12 +σ2m22 (σ1 -σ2 )m1 m2 

σ =  m  m = 
T
(2.5)
2 2 
(σ1 -σ2 )m1 m2 σ1m2 +σ2m1 

where (m1 = sinθ), (m2 = cosθ) and (θ) is the fibers' direction, Fig. 2.1.

FIG. 2.1. Coordinate system (x1, x2) and principal axis (ξ1, ξ2) of the unidirectional CF
medium.

In a steady current, the spatial distribution of (J) is independent of time, and

satisfies the equation, Eq. 2.1. The electric field in the conductor in which a steady
current flows is constant satisfies the Eq. 2.6.

E 0 (2.6)

The relation between the current density and the electric field for an anisotropic
conductor expressed by Eq. 2.7, (Landau & Lifshitz (1984)).

J σE (2.7)

where ( a ) is the thermoelectric tensor and is a characteristic of the material. The

electric field (E) is the gradient of the electric potential field, provided by Eq. 2.8.

E   (2.8)

Eq. 2.7 includes the thermoelectric phenomenon, as well. This means that it takes into
consideration the current distribution as a function of both the electric field and the
temperature. In the present work the thermoelectric effect is considered negligible,
compared against the Joule heat. Therefore, the current density is expressed by the
simplified form, Eq. 2.9.
12 Chapter 2
 

    
Jx1    11  12  13 
 x1 x2 x3 
    
Jx2    21  22  23  (2.9)
 x1 x2 x3 
    
Jx3    31  32  33 
 x1 x2 x3 
or by the Eq. 2.10.


J i  ik (2.10)
x k
Introducing Eq. 2.9 into Eq. 2.1 the PDE which describes the electrical problem is
expressed by Eq. 2.11.

           
 11 12 13   21 22 23 
x1  x1 x2 x3  x2  x1 x2 x3  (2.11)
     
 31 32 33 0
x3  x1 x2 x3 
In the case of the anisotropic and homogeneous medium, the EC tensor is
symmetric (σ12 = σ21, σ23 = σ32, σ13 = σ31) and can be expressed by the following
elliptic PDE, Eq. 2.12. As we have mentioned before, the symmetry of the EC tensor
is a consequence of the symmetry of the kinetic coefficients (Onsager theorem).

2 2 2 2 2 2

11  22  33  212  213  223 0 (2.12)
x12 x22 x32 x1x2 x1x3 x2x3

The representation of Eq. 2.12 for homogeneous anisotropic medium is given by

Eq. 2.13.

  σ    0  σ :     0 (2.13)

Assuming a 2D body, the elliptic PDE can be expressed as follows, Eq. 2.14.
For speed's sake, whenever the EC tensor (σ) is mentioned we will assume that it is
dependent upon the parameters already mentioned, Eq. 2.4.
Governing equations 13
 


0
x 3
2  2 2
σ :      0  11  212   22 0 (2.14)
x12 x1 x 2 x 22

When the coordinate system is transformed, Fig. 2.1 from the (x1, x2) to another
set of axes (ξ1, ξ2), which are the principal axes, the Eq. 2.14 simplifies to Eq. 2.15.

2 2
1 2  2 2  0 (2.15)
1 2
In the case of the anisotropic and electrically non-homogeneous medium the EC
tensor is symmetric again and can be expressed by the following equation, Eq. 2.16.

  σ    0    σ     σ :     0 (2.16)

The equation changes because the tensor (σ) depends on its spatial coordinates,
which means that the material is electrically heterogeneous. Therefore, Eq. 2.16 in 2D
becomes Eq. 2.17.

  11 12    21 22  2 2  2

       
 11 2  212  22 0 (2.17)
x1  x1 x2  x2  x1 x2  x1 x1x2 x22
Apparently, the current density lines are not perpendicular to the equipotential
lines, as in the well - known case of the isotropic medium, Fig 2.2. This is a
significant difference between the electric behavior of the anisotropic medium
compared against the respective isotropic.

FIG. 2.2. Electrical current flow and equepotential lines for (a) isotropic and (b) anisotropic
medium.
14 Chapter 2
 

The typical boundary conditions for the general anisotropic medium are
described by a combination of a first kind boundary condition (Dirichlet boundary
condition) at the Γd regions, Eq. 2.18, as shown in Fig. 2.3.

 = f(x1 ,x2 ,t) at Γd (2.18)

and electrical insulated regions at the Γs regions, Eq. 2.19, (Jn = 0).

Jn  0 at s (2.19)

FIG. 2.3. Boundary conditions (Γd = Constant electric potential, Γs = Electrical insulated).

2.2. Joule effect1

When an electric current flows through a solid or liquid of finite conductivity,
the electric energy is converted to heat through resistive losses in the material. The
heat is generated when the conduction electrons transfer energy to the conductors’
atoms through collisions. the thermoelectric effect has been omitted, then the
generated heat per unit time and volume is the dot product of the electric field and the
current density, Eq. 2.21.

  E J
Q (2.20)
el

The generated heat per unit time and volume is the dot product of the electric
field and the current density, Eq. 2.21.

                                                            
1
The correct terminology is Joule - Thomson effect.
Governing equations 15
 

  E J =Eσ E
Q (2.21)
el

The dot product ( E J ) in a known domain (Ω), with given boundary

conditions, changes according to the EC tensor of the layered medium. Thus, the
resulting temperature field in the known domain is also different.

2.3. Heat transfer in homogeneous anisotropic media

The heat flux can be calculated by the dot product of the thermal conductivity and the
temperature gradient, Eq. 2.22.

q   k  Τ (2.22)
In the case where the medium is isotropic, the thermal conductivity (k) is a
scalar. When the medium is thermally anisotropic, the thermal conductivity (k) can be
expressed by a second order tensor. Expanding Eq. 2.22 the heat flux in each direction
is expressed by Eqs. 2.23.
T

T
    
qx1    k11  k12  k13 
 x1 x2 x3 
T

    
qx2    k 21  k22  k23  (2.23)
 x1 x2 x3 
T

    
qx3    k 31  k32  k33 
 x1 x2 x3 

The thermal conductivity along the fibres' direction differs from the thermal
conductivity in the transverse to the fibers' direction (kf// ≠ kf┴). As far as porous
media or composite materials are concerned, the thermal conductivity depends on the
fibre density, on the material's microstructure as well as on the thermal conductivity
of the matrix material. If the heat conduction occurs in parallel with in the fibers and
the matrix material, then the overall conductivity (k) is the weighted arithmetic mean
of the conductivities (kf) and (km), Eq. 2.24, (Nield & Bejan (2000)).

k  (1  )k f  k m (2.24)
The Joule effect in porous medium can be calculated using Eq 2.25.
16 Chapter 2
 

  (1  )Q
Q   Q
 (2.25)
el s m

The overall heat capacity per unit volume and overall thermal conductivity Eqs.
2.27.

(c)c  (1  )(c)f  (c) m (2.26)

The porosity or matrix volume fraction (ϕ) of a medium is defined as the ratio
of the voids or resin in the medium over the total volume. Thus (1 – ϕ) is the fraction
that is occupied by solid (carbon fibers).
Considering an elementary volume, the heat transfer energy equation of dry or
CFRP anisotropic media expressed by Eq. 2.27.
T

T

 c ( T ) c    kc     kc :       E  J  

T
t    k c    0
T

(2.27)
T


  (1  )(c) f  (c) m  T  k c :      E  σ  E 
t

where (kc) is the thermal conductivity tensor of the CF medium. This is the general
heat transfer equation, which includes the dependence of the coefficient of thermal
conductivity on the temperature.
In this thesis, we have assumed that for all temperatures studied, the variation of
thermal conductivity is negligible and is therefore considered independent of
temperature (Pradere et. al (2008)). Hence, the material is considered thermally
anisotropic and homogenous. Therefore, Eq. 2.27 for the case of CFRP is simplified.
In the case where the material's thickness is too small compared to the other
dimensions (2D space problem), Eq. 2.27 should include the losses due to convection
and radiation (Qlosses). Therefore, it becomes Eq. 2.28.
T


 (1  )(c)f  (c)m  T  kc :      E  σ  E   Q losses (2.28)
t
The present work studies the electric problem of porous and CFRP materials at
the steady state, at constant current. We assume the material homogenous or
heterogeneous, and electrically anisotropic. The thermal problem is studied at the
transient and steady state under the assumption that the material is thermally
anisotropic and homogenous. The solution of both electric and thermal problems is
Governing equations 17
 

primarily based on the examination and determination of the material's electric and
thermal properties, in relation to the aforementioned assumptions, Eq. 2.4.
18 Chapter 2
 
EC of unidirectional CF and CFRP media 19
 

3. start
Chapter 3
Electrical conductivity of unidirectional CF and
CFRP media

Most models concerning the prediction of the electric conductivity are fairly
complicated, and do not converge to the experimental results. As it has already been
mentioned, all these models do not describe the electrical conductivity with precision,
and therefore the difference between the predicted electric conductivity and the
experimental measurements is significant.
This chapter presents the factors that affect the electrical conductivity of the
unidirectional carbon fiber layers as well as certain empirical models for the
prediction of the electrical conductivity along to fiber direction and transverse to fiber
direction. The volume resistivity of the unidirectional porous and the CFPR layers can
be described as a function of preform thickness and fiber volume fraction or the
applied pressure.

 
20 Chapter 3
 

3.1 Electrical conductivity of the single CF

CFs are transformed from organic matter by 1000 – 1500°C heat treatment, which are
the substance with imperfect graphite crystalline structure, arranged along the fiber
axis. A single filament is a thin tube with a diameter of 5 – 8 micrometers and consists
almost exclusively of carbon. The atomic structure of carbon fiber is similar to that of
graphite, consisting of sheets of carbon atoms arranged in a regular hexagonal pattern
(Shindo (2000)).
Based on precursor fiber materials, CFs are classified into: PAN - based, pitch-
based, mesophase pitch-based, isotropic pitch-based, rayon-based, gas - phase grown
carbon fibers. Electrical conductive fibers can be classified into two general
categories, those that are naturally conductive and those that are specially treated to
create conductivity (Chung (1994)). Specially treated fibers represent metal coated
fibers, fibers filled or loaded with carbon of metallic salts. Naturally conductive fibers
or metallic fibers are development from electrically conductive metals such as ferrous
alloys, nickel, stainless steel, etc. Other naturally conductive fibers are carbon fibers.
The factors that affect the fibre's EC, are the following.

i. Electrical conductivity of various CFs: according to Shindo (2000) variations in

composition and structure of CFs, related to the conditions of their production and to
impurities present in their structure, are the cause of changes in electrical resistivity of
these materials, resulting in properties ranging from those of conductors to
semiconductors, Table 3.1. In general, the electric conductivity of CFs is controlled in
three principal ways: a) by the time-temperature regime during fiber formation, b)
orientation stretching, c) the introduction of chemical elements into structure of the
initial raw materials.

ii. Temperature dependency: in contrast to metals, the CF electric conductivity

rises, as the material’s temperature rises. The resistivity decreases with increasing
temperature for each type of fiber. This is because the carrier density increases with
temperature, just as for carbons and graphites in general.

iii. CFs electrical conductivity as a function of frequency: the hopping conduction

increases with the frequency, according to a power law, predominating over the band
conduction. The conductivity above 100 kHz is governed by hopping conduction

 
E of unidirectiional CF and CFRP
EC C media 2
21
 

which probab
w bly results ffrom the disordered staccking structuure of the micrographit
m tic
laayers. It is sppeculated thaat this is duee to the largee skin depth (Imai & Kanneko (1992))).

TABLE 3.1. Volume

T V resistivity values of various single CFs at room tempeerature (Shinddo
(22000), Toho Tenax
T (2008), Torayca (20005)).

ivv. Gas adso orption dep pendency: exposure off (CFs) to (O2) gas leeads to cleaar
onductivity, indicating thhat the domiinant chargee carrier is a hole. The EC
inncrease of co E
inncreases rapidly upon exxposing to (SO( 2). On thhe contrary, adsorption of (NO), annd
(NN2O) lowerss the electricaal conductivvity (Imai & Kaneko (1992)).

v. Young modulus
m dep
pendency: att a given tesst temperatuure, the volume resistivitty
decreases witth increasingg tensile moodulus. This is because an increase in the tensile
m
modulus is accompanied
a d by a decreease in the concentratio
c n of defectss, and defeccts
caause carrier scattering. The more and a aligned thet structuraal elements (microfibrils)
arre, the largeer is free patth for electrrons movemment. As the resistivity decreases,
d thhe
teensile modullus and the density
d increeases (Shindoo (2000)).

3
3.2 CFR
RPs’ electtrical condductivity models
Many models have beenn developed in order too predict thee electric co
M onductivity oof
C
CFRPs, along
g or transveerse to the fibres' direcction. Most of these models are noot

 
22 Chapter 3
 

experimentally validated (Clingerman et. al (2002), Weber & Kamal (1997)). This
thesis thoroughly describes the electric conductivity along or transverse to the fibres,
for both cases of porous CF materials and CFRPs. The material's microstructure is of
great significance, and is directly dependent on the manufacturing process. In order to
verify the effect of the microstructure on the electrical conductivity, certain
measurements were taken first from dry fabrics (no presence of resin), and then from
CFRPs. Below, certain models are present for the prediction of the EC of CFRPs,
which have been suggested in the literature (Sections 3.2.1 – 3.2.7), (Weber & Kamal
(1997)).

3.2.1 Percolation Model

The percolation theory has been used to interpret the behavior in a mixture of the
conducting and non-conducting components. A simple power law, Eq. 3.1 describes
the relation between composite EC and the electrical conductive filler concentration
in the vicinity of the percolation threshold.

c  f  vf  vcrit 
u
(3.1)

where (σc) is the composite conductivity, (σf) is the conductivity of conductive

reinforcement, (vf) is the fiber volume fraction, (vcrit) is the percolation threshold, (u)
is the critical exponent. This equation is valid at concentrations above the percolation
threshold, when (vf > vcrit). (Weber & Kamal (1997)).

3.2.2 Bueche Model

Bueche (1973) considered the problem of conducting particles in a non-conducting
matrix. The resistivity of the composite expressed by the Eq. 3.2.

mf
c  (3.2)
(1 vf )f  vf gm

where (ρf) is the VR of the fiber, (ρm) is the VR of the matrix, (vf) is the volume
fraction of fibers, (ωg) is the fraction of the conductive phase incorporated in an
infinite cluster given by Eq. 3.3.

 
EC of unidirectional CF and CFRP media 23
 

(1)2 y
g  1 (3.3)
(1 y)2 

the variable (y) is the smallest root of the Eq. 3.4.

(1  )f 2  y(1  y)f 2 (3.4)

where (f) is the maximum number of contacts per particle, (α) is the probability of
contact between neighboring particles, Eq. 3.5
The (α) term is defined as

vf
 (3.5)
vmax

where (vmax) is the maximum packing fraction of particles. A value of (f) may be
calculated from the fact that, at the percolation threshold, (α = (f - l)-1) (Weber &
Kamal (1997)).

3.2.3 Nielsen Model

According to Nielsen (1974) the EC is defined following the Eq. 3.6.

 1  ABvf 
c  m   (3.6)
1  Bvmax 

where (B) and (ψ) are given by Eq. 3.7 and Eq. 3.8 respectively.

f / m 1
 (3.7)
f / m  A

 1 v 
  1     2 max  vf (3.8)
 vmax 

(A) is a function of the aspect ratio of the particles. For polymer composite systems,
(B) is essentially equal to 1. The (ψ) term can be calculated by knowing the maximum
packing fraction of particles, while (A) is found from tabulated values (Weber &
Kamal (1997)).

 
24 Chapter 3
 

3.2.4 Berger - McCullough Model

According to Berger & McCullough (1984)) the Eq. 3.9 predicts the composite
conductivity in either the longitudinal, transverse, or normal directions.

  v v (  m )2 
cr  vf f  vmm   i f m f  (3.9)
 vfi f  vmi m 

where (Vfi) and (Vmi) are defined as, Eqs. 3.10.

Vfi  (1  i )vf  i vm
(3.10)
Vmi  (1  i )vf  i vm

The (λi) term is a chain length factor which is dependent on the aspect ratio of the
particles (Weber & Kamal (1997)).

3.2.5 Ondracek Model

Ondracek (1987) derived a model for field properties of multiphase materials which
are at equilibrium and whose microstructure is homogeneous. The equation for
composite conductivity is expressed by Eq. 3.11.

Y
  1 
r  c  f 1  
f  c  m    2FD  cos (1  3FD )  
2
(3.11)
1  vf   
f  m  c    1 
 m  f 1  
 2FD  cos (1  3FD )  
2


where (θ) is the angle of fiber direction, (FD) is a shape factor. The exponents (r) and
(Y) are defined using Eqs. 3.12.

 FD 1  2FD  
r 
1  2FD  cos (1  3FD ) 
2

(3.12)
FD 1  2FD  2FD 1  FD 
  1
1  2FD  cos (1  3FD ) 2FD  cos2 (1  3FD )
2

 
EC of unidirectional CF and CFRP media 25
 

The shape factor (FD), varies from 0 for infinitely wide fibers to 0.5 for
infinitely long fibers (Weber & Kamal (1997)).

3.2.6 Effective Medium Theory

According to Taya & Ueda (1987) for a conductive composite, the original lattice
consisting of randomly distributed conductance’s, is replaced by a lattice of similar
symmetry.

 vf  f m   f m  S11  S33   2m  

c  m 1  (3.13)
 2 f m  1 vf  S11S33 m  f m  2  vf  S11  S33  
2

For fiber composites we can calculate the (S11, S22 and S33) using Eq. 3.14.

d2    2 1/2 
S11  2 2 3/2   2 1  cosh1  ,S22  S11 ,S33  1 2S22 (3.14)
2(  d )  d  d  d
 

where (  ) is the fiber length, (d) is the fiber diameter (Weber & Kamal (1997)).

3.2.7 Fiber Contact Model

According to Weber & Kamal (1997) three different cases of particle to particle
contact were considered: a) the particles are separated by a thin layer of matrix, b) the
particles are in contact at a single point, c) the particles have a flat circle of contact.
The first case is analogous to quantum mechanical tunneling, where heat or electricity
may be conducted in particles that are nearly touching. The second and third cases are
likely to be found in a typical composite. For composites, where the conductivity of
the inclusion is large compared with that of the matrix, essentially all of the current
flows through the inclusions. The potential gradient within a particle is very small,
except near points of contact with other particles. In the vicinity of these points, the
magnitude of the current density and the gradient of potential are large compared to
values far from a contact point. Therefore, the conditions near the contact points
determine the total current through the particle.
The dependence of the resistivity on fiber length, orientation, and volume
fraction, as well as the area of contact is expressed by the following expressions, Eq.
3.15 and Eq. 3.16.

 
26 Chapter 3
 

d2f X
c_long  (3.15)
4vpdc cos2 

d2f X
c_ trans  (3.16)
4vpdc sin2 
where (d) is the diameter of circle of contact and (X) can be calculated by Eq. 3.17.

1
X (3.17)
0.59  0.15m
where (m) is the number of contacts. The volume fraction of fibers participating in
strings, can be written using Eq. 3.18.

vp  v (3.18)
where (β) is equal to zero below the percolation threshold, (vcrit), and to 1 at high
values of (v). The (β) term is therefore dependent on (v) and can be written as, Eq.
3.19.

  f(v  vcrit ) (3.19)

It is assumed that above a “threshold” volume fraction (vcrit), all fibers participate in
strings. Therefore, at volume fractions greater than (vu), (β) is 1. At volume fractions
below (vcrit), (β) is zero. For concentrations in the range (vcrit < v < vu), (β) can be
calculated by Eq. 3.20:

v  vcrit
 (3.20)
vu  crit
A value of (φt) at which (β) will be equal to 1 is assumed, and (ϕcrit) is obtained from
the experimental data. (β) is then found, and (vp), is calculated.

 vp 
mvp  mmax   (3.21)
 vpm 
 max 
The majority of the above models do not accurately predict the composite
electrical resistivity. Only the percolation theory and Bueche model take account of
the percolation threshold: the resistivity decreases with any addition of fibers for the

 
EC of unidirectional CF and CFRP media 27
 

other theories. Only the percolation theory yields results which are in agreement with
the experimental data (Weber & Kamal (1997)).
Literature being deficient, at first the EC of dry carbon fabrics and then, the EC
of CFRPs were studied. Therefore we could first examine the ideal case and then,
compare it to the respective CFRP, where the mechanism is more complicated.

3.3 Experimental study and prediction of the EC of

unidirectional CF layers
As it has already been mentioned, the above models do not describe the EC with
accuracy, resulting in a significant variation between the predicted and the measured
electric conductivity. Even the models that satisfactorily predict the EC of CFRPs,
require the knowledge of certain coefficients calculated empirically or they require
certain experimental measurements, beforehand.
The EC of the unidirectional layers depends on certain factors. The present
chapter studies experimentally the dependence of the UD layers' EC upon the fibers'
direction (θ), the material’s thickness (h), the length/width ratio (λ), the fibre volume
fraction (vf), as well as on the material's microstructure. Hence, for UD laminates the
EC tensor depends upon the previously mentioned parameters, Eq. 3.22.

σ = σ (σ1, σ2, vf, h) (3.22)

Weber et al (1997) have shown the dependence of the measured conductivity on the
length to width ratio (λ) of CFRP samples. In the proposed model, two discrete
regions are concluded with regard to the length to width ratio. This aspect ratio
dependence was clearly confirmed and a critical threshold (λcr) is proposed after
multiple measurements, in the vicinity of which none of the previously mentioned
regions apply. This aspect ratio threshold (λcr) depends upon the fiber direction. Also,
for every fiber direction a specific value of the ratio (λ) is quantified, above which the
(λ) ratio has negligible impact on the VR measurement. Above this value the material
VR tends to obtain the value extracted using the VR tensor. Tse et al. (1984) have
shown dependence of the measured electrical conductivity on the width of CFRP
samples transverse to the fiber direction. In the present work it is confirmed that for
fiber direction (θ = 90ο) as well as for various fiber directions, in the UD layer, having
a width greater than a specific width, the layer VR (ρ22) remains constant. Finally a

 
28 Chhapter 3
 

mathemaatical expression that corrrelates voluume resistiviity to the preeform geom metrical
propertiees (thicknesss, aspect ratiio, fiber direection) for a given fiberr volume fraaction,
was extraacted.

3.3.1 Experiment
E tal setups for
f the EC measurem
ments
Two diffferent meassuring setupps were devvised, in orrder to calcculate the electric
conductivvity of the porous
p and the CFRP m material resppectively. Thhese setups differ
from each other due to the separate nature off each of thee materials: oone is porouus (dry
medium)), whereas th he other is ssolid (CFRP medium). This
T thesis ddoes not studdy the
electric cconductivityy in low vallues of the fibre volumme fraction, hhaving taken into
consideraation that thhe industrially producedd fabrics havve a fibre volume
v fraction of
approximmately > 300%, and byy no meanss lower thann that. Theerefore, as far as
compositte materials are concernned, there iss no point ini studying low fibre volume
fractions.

FIG. 3.1. Fiber volum

me fraction as
a a function of the mediium thicknesss for unidirectional
materials.
Figgure 3.1 dem
monstrates thhe fibre volumme fraction of a commoon UD fabricc, as a
function of the fabrric's thicknesss. It is obvvious that most
m manufaacturing processes
involve ffibre volumee densities off over (45%). Considerinng the abovee, it is pointlless to
study casses of lowerr fibre voluume densities, since they are not used in real CFRP
structures.

 
EC of unidirectional CF and CFRP media 29
 

3.3.1.1 Experimental setup of the dry CF preform

In order to measure the electrical resistance of the dry carbon fiber UD preform, an
apparatus was developed to ensure uniform pressure distribution on the surface of the
preform specimens. Measurements taken with this setup, proved to be very accurate
and close to the theoretical electrical resistance along the fiber direction. The effect of
the contact resistance and the electrodes at the measurements were practically
negligible. The specimens were placed parallel to each other, on an electrical, non-
conductive, rigid, flat surface. CF preforms were sealed airtight by using sealant tape
and a vacuum bagging film. A vacuum pump extracts the air from the inside of the
setup. The vacuum level was monitored using two vacuum gauges. The applied
pressure was constant in all tests, and equal to (P = 0.96 atm or 97.27kPa).
At the edges of the preform specimens, two copper, braided electrodes were
placed as shown in Fig. 3.2(a) and 3.2(b). Also in the opposite side, two silver tapes
were placed (Appendix I). The pressure was applied on both the preform and copper
braid electrodes. Copper cables were soldered on the braid electrodes and through the
airtight volume they were connected to a precision multimeter (Keithley 2002).
Measurements taken using the braided copper electrodes are compared to another
measurement setup and theoretical value. The preform – braided copper contact
interface affects less than 4% the measurements. In the next section of this work the
measurement verification procedures are presented in full extent.
The material tested was the T700SC UD preform (fiber diameter df = 7μm), is
shown in Fig. 3.2(c). According to the manufacturer the VR of the fiber is (ρf = 1.6
10-5 Ωm) at room temperature (Shindo (2000), Torayca (2000)) and a (0.3%) weight
content of non-conductive stitches normal to the carbon fiber direction that do not
affect the overall electrical behavior of the material.
Firstly, a (T700SC/12K) carbon fiber tow was selected and its electrical
resistance was measured following two different methods, using five specimens for
each one. The first method is the one described before, whilst the second one uses
silver paint at the edges of carbon fiber tow and then copper wires soldered at the
edges (Kupke et.al (2001)). The theoretical resistance of the carbon fiber tow is given
by the following expression, Eq. 3.23.

Lf
R f = f ,   0o
df2 (3.23)
nf
4

 
30 Chhapter 3
 

where (nnf) is the nummber of fiberrs inside thee tow, (df ) is the diametter of the T7
700SC
carbon fiiber, (Lf) is the
t specimenn length andd (ρf) is the V VR of carboon fiber. Tabble 3.2
presents the results ofo the two measuring
m appproaches annd the theoreetical value of the
CF tow rresistance.

FIG. 3.2. (a), (b) Experrimental setupp, (c) UD – innlay T700SC 160gr/m2.

1st Meethodology 2nd M

Methodologyy
Length (m)
( 0.1 0.1
No. of fib
bers 12000 12000
No. of Speccimens 5 5
Experimeental
R0  3.563  0.091 R0  33.600  0.092
2
M
Mean Value  S.D (Ω)
Theeoretical Resiistance (Ω) R 0  3.465
Relative Differeence δ(%) 2.828 3.896

TABLE 3.2.
3 Experimeental results of CF tow resiistance for thee two differennt methods.

3.3.1.2 Experimental
E l setup of thee CFRP mateerial
The voluume resistivitty of CF preepreg was stuudied in relaation to the llaminate thicckness
(number of layers) foor fiber orienntations of 0o and 90o respectively. The
T dimensions of

 
E of unidirectiional CF and CFRP
EC C media 3
31
 

thhe specimens were (30m mm x 240mm m). The speccimens weree heated at 130oC for 2.55h
unntil the curinng process was
w completted and thenn they were lleft to cool down d to roomm
teemperature (Sigrafill
( / Sigratex).
S Ussing a very sensitive
s muultimeter (Keeithley 20022),
thhe specimenns resistivityy was measuured. The specific
s he material is
resisstivity of th
caalculated by y the followwing formulla, Eq. 3.244. It has too be mentiooned that thhe
m
measurement ts were carrieed out usingg the four-eleectrode methhod.

R(298K)A
(298K)  (3.244)
L
w
where t length off the specimeen and (A) is the cross section area.
(L) is the

FIIG. 3.3. Thermography of the CFR RP specimenns in order to ensure good g electriccal
coonnections. (aa), (b) very goood connectioons, (c), (d) deficient connections.

 
32 Chhapter 3
 

As far the speccimens’ processing prioor to the meeasurements is concerneed, the

tips weree scratched,, and cleaneed with aceetone. Then,, three layers of silverr paint
followedd by two layeers of silver tape were applied
a on thhe tips. Finally, the speccimens
were plaaced in a cu ustom-made device wheere high pressure was aapplied to alll four
electrodees (four elecctrodes – tw
wo point meeasurements)) that are inn contact wiith the
specimenn (Athanasoopoulos et. al (2013), KupkeK et. al
a (2001)). The methodd was
indirectlyy validated. By applyying electriic potential difference, the speciimen's
temperatuure rises. Inn the case where
w the coonnections bbetween the copper elecctrodes
and the C
CFRP are fauulty, the temmperature distribution on the materiall's surface will
w not
be uniforrm.

3.3.2 Ohmic
O beha
avior of thee CF prefoorms and thhe CFRPs
Before anny measurem ments were taken, we considered
c u
useful to exaamine whethher the
material behaves ass a linear electric condductor due to t its defauult microstruucture.
Accordinng to the literrature, carboon fibres are classified as semi-condductors and behave
b
as ohmicc resistors. Hence, the electric currrent is a liinear functioon of the appplied
voltage.

FIG. 3.4. Ohmic behavvior of the poorous CF and CFRP media at (a) 90o fibber direction and
a (b)
0o fiber diirection.

Duee to lack off relative refferences conncerning UD

D fabrics, wee collected certain
c
measurem
ments of thee electric coonductivity, in order to study the ohmic
o behavvior of

 
EC of unidirectional CF and CFRP media 33
 

both dry fabrics and CFRPs, along and transverse to the fibers' direction. The results
are presented in the following diagrams. It is obvious that the electric current is a
linear function of the voltage (the standard deviation for all the cases R2 = 1).
Therefore, we can safely conclude that the conductor is linear, Fig. 3.4.

3.3.3 Prediction of the VR for dry unidirectional media

The applied pressure results in a ( v f  0.5 ). The nominal thickness of each layer of
the UD preform used was (h = 0.18mm), Fig. 3.2(c) and had a nominal weight of (160
gr/m2). The number of fibers of this UD preform can be estimated by the approximate
fiber content at this pressure, with the mathematical expression given by Eq. 3.25.

Af 4v f WH
vf   nf  (3.25)
A df2
where (W) and (H) are the width and the thickness respectively of the UD preform
samples. Thus, using Eq. 3.24 and Eq. 3.25 the electrical resistance and VR of the
T700SC UD preform for (θ = 0o) can be calculated. The results for both analytical
calculations and experimental measurements of the VR (ρ11) of the UD carbon fiber
preform (T700SC), are shown in Table 3.3.

Experimental Results Theoretical Results

ρ(10-5 Ωm) ρ(10-5 Ωm)
Dimensions (m) L = 0.29m, W = 0.018m
Thickness - h(mm) 0.180 - 5.400
No. of Specimens 5

Mean Value  S.D 0  3.166  0.056 0  3.199

Relative Difference δ (%) 1.032

TABLE 3.3. Experimental results of UD T700SC electric resistivity ρo in fiber direction.

As it is shown, and based on the results presented in Tables 3.2 and Table 3.3,
the relative difference between the measurements and the theoretical results, is (δ <
4%) either using the silver paint or the copper braid at the edges of the test samples.
Furthermore, the electrical resistance of the CF tow has almost the same value
regardless of the methodology used and the experimental measurements are very
close to the theoretically calculated value.

 
34 Chapter 3
 

As expected, the resistivity (ρ11) remains constant regardless of the number of

layers or thickness of the carbon fiber preform used, Table 3.3. This means that in the
case of angle (θ = 0o), all layers act as parallel resistors. The parallel connection does
not apply to UD carbon fiber preforms with fiber directions (θ ≠ 0ο), for the given
applied pressure, as it will be explained in following sections.
In transverse to the fibers' direction (θ = 90ο), the electrical resistivity of the UD
carbon preform depends upon the random contacts between the fibers. The applied
pressure on the preform, determines the final value of resistivity. Tests were carried
out, to investigate the correlation between resistivity and specimen width. Tse et al.
(1984) have shown that the electrical conductivity of the material depends upon the
width of the specimen. The electrical conductivity increases when the specimen width
is increased.
The EC reaches a constant value for a specimen width of (W > 0.6cm). In the
present study it was found that the specimen conductivity remains constant for all
(length to width ratios λ) with specimen width of (W > 0.8cm). Table 3.4 summarizes
the results for a carbon fiber layer preform in the case of (θ = 90ο) for different length
to width ratios (λ). The width of all samples was ( W  0.8cm ) and the thickness was
(h = 0.18mm). The VR was constant in all cases. For thickness (h = 0.18mm), and
transverse to fiber direction the electrical volume resistivity was (ρ22 = 0.0263 Ωm).

Length to width ratio (λ) 0.686 - 11

No. of Specimens 8
h(mm) 0.18

Experimental Mean Value  S.D 90  2.632  0.1488

10-2 ρ (Ωm)

TABLE 3.4. Experimental results of UD T700SC VR ρ90 perpendicular to the fiber

direction.

3.3.3.1 Resistivity as a function of thickness and direction

Several tests were carried out for the following scenarios:
i) Experimental measurement and theoretical calculation of UD preforms electrical

conductivity for fiber directions (   0o ,30o , 45o , 60o and 90o ), as a function of (λ).

 
EC of unidirectional CF and CFRP media 35
 

ii) Experimental measurement of UD preform VR for directions (

  0o ,30o , 45o , 60o and 90o ), as a function of the preform thickness. Comparison
between experimental and the extracted analytical results.
iii) Generalized approach of VR as a function of fiber direction, length to width ratio
and preform thickness.

The VR can be expressed as a function of length to width ratio in the case of UD

carbon preforms with   0 o . The EC in the three principal axes (1,2) is expressed by,
Eq. 3.26.
σ
 0
 1  (3.26)
0 2 
The off-axis VR is then calculated by rotating the in plane principal coordinate
system 1-2 around the 3-axis by an angle of (θ). Assuming 2D conditions, the
resistivity tensor in the (x1 x2) plane, which form an angle (θ) with the principal
directions is given by Eq. 3.27.

 ρ1 m12 + ρ 2 m 22 (ρ 2 - ρ1 )m1 m 2   ρ11 ρ12 

ρ =  m    m 
T
   (3.27)
(ρ 2 - ρ1 )m1m 2 ρ1 m 22 + ρ 2 m12  ρ 21 ρ 22 
The electrical conductivity tensor is the inverse resistivity tensor and is given by the
following expression, Eq. 3.28.
ρ

σ σ12   σ1 m12 + σ 2 m 22 (σ1 - σ 2 )m1 m 2 

σ  1
 σ   11   (3.28)
σ 21 σ 22   (σ1 - σ 2 )m1m 2 σ1 m 22 + σ 2 m12 
it should be noted that( m1 = cosθ) and (m2 = sinθ).
In the present work the length to width ratio (λ, aspect ratio) is divided into three
different regions. Critical aspect ratio (λcr) is the ratio where the fibers start from one
corner of the specimen and reach the exactly opposite corner and it expresses the
tangent of the fiber direction, (λcr = tanθ) as shown in Fig. 3.5(d). As the aspect ratio
increases ( λ > λ cr ), then the fibers have the arrangement shown in Fig. 3.5(b) and the
dominant mechanism of the electrons path is the statistical zigzag path along the
fiber-to-fiber contact and the transverse volume resistivity is based on this
mechanism.

 
36 Chhapter 3
 

Forr any fiber direction

d andd also for thhe extreme cases
c of (θ=0ο and θ=90 0ο) the
VR along the length h of the UD D preform can
c be calcuulated using Eq. 3.29. As A the
aspect raatio increasess ( λ >> λ cr ), the value of VR (ρ11) approaches
a tthe solution of the
EC tensoor, Eq. 3.27, where (    ).

c )
( L  W cot
R11  (  2 sin 2   1 coos 2  ) 
hWW
L ( L  W cot  )
11  (  2 sin 2   1 cos 2  ) 
hW hW
(3.29)
(
c  ) 2  1
( L  W cot
11  (  2 sin 2   1 coos 2  ) 
L
( L  W cot  )  cot c 
11  (  2 sin 2  )    2 sin 2    1 
L   

When (   cr ), then

n more thann one fiber “bridges”
“ thee two electrrodes as sho
own in
Fig. 3.5(cc). As a resu
ult the valuee of the elecctrical conduuctivity depeends upon th
he CFs
conductivvity, and thee equation thhat describes the phenom menon is Eq. 3.30.
When asppect ratio (λλ) becomes tooo small, connductivity iss reduced annd reaches a steady
s
state valuue.

FIG. 3.5. Length to widdth ratio (λ) and

a specimens geometry.

 
E of unidirectiional CF and CFRP
EC C media 3
37
 

L L
R11  1 cos   1 cos  
h cos  (W  AB ) h cos  (W  L tan  )
L L (3.300)
11  1 
hW h cos 2  (W  L tan  )
W 1
11  1  1
cos  (W  L tan  )
2
cos  (1   tan  )
2

f
w
where 1  .
vf
Due to the fact thaat only one of
o the two mechanisms
m (longitudinaal / transversse
diirection) is dominant
d every time, thee shaded areeas of Fig. 3..5 do not con ntribute to thhe
phhenomenon.. In the plotss given in off Fig. 3.6, thhe analyticall and experimmental resullts
foor the three (λ)
( regions are
a presentedd. A gap is observed arouund the ( λ = λ cr ) region..

FIIG. 3.6. VR: experimentall and theoretiical results ass function of length to wid
dth spatial rattio
(λλ) and the fibeer direction (θθ) of the UD preform
p for thhickness h = 0.18mm.

 
38 Chhapter 3
 

Thee resistivity in this regioon can be exttracted by m making the asssumption th hat the
value off resistivity should be bbetween neigghboring vaalues (linearr interpolatioon). It
must be noted
n here th
hat all the pllots in Fig. 3.6
3 corresponnd to thickneess (h = 0.18 8mm).
For thickkness (h > 0.18mm) the diagram
d is different, as presented
p in the next secction.
As it is evidentt, based on Fig.
F 3.6, the VR of UD dry preform m in differennt fiber
directionns for high leength to widtth ratios (   cr ) can be
b calculatedd by Eq. 3.227. For
example for fiber dirrections (θ = 60ο) the VR
V can be caalculated forr (   10 ) annd Eq.
3.27 is valid. Figure 3.7 shows the
t value of the critical length
l to width ratio (λcr
c ) as a

function of the fiber direction (θ)), in the casee of UD carbbon fiber preforms.

FIG. 3.7. Critical lengtth to width rattio (λcr) as a function

f of fibber direction ((θ).

Aloong the fibeer direction,, for the saame fiber volume

v fracttion (vf =0.55) the
electricall conductivitty is indepenndent of the preform thickness. The calculation of the
overall reesistance is similar to thhe case on parallel
p resistors. This is
i not valid in the
fiber directioon (   0 ). As
case of fi o
A the prefoorm thicknesss increases tthe VR decrreases.
R h h
The absolute differrence of VR 1 k
d
depends uppon the fibeer directionn. The
maximum
m absolute difference
d off the VR wass observed foor (θ = 90o).
Furrthermore h h is relaated to the fiber volum
1 k
me fraction or to the appplied
pressure on the prefoorm. The aim
m of this paart of experim
mental meassurements wasw the
investigaation of the VR at consstant pressurre as a funcction of thicckness at diffferent

 
E of unidirectiional CF and CFRP
EC C media 3
39
 

fiiber directioons of UD carbon

c prefoorm. Two different
d length to width h ratios werre
seelected (λ1 = 28, λ2 = 7.67).
7 The electrical
e coonductivity of
o five grouup of samplees
having differeent number of layers (naamely 1, 5, 10, 20 and 30 3 layers) at five differennt
o o o o o
fiiber directionns (namely 90
9 , 60 , 45 , 30 and 0 ) were measuured.
In Fig. 3.8 the expeerimental and theoreticall results are presented foor (λ > λcr). It
caan be seen that there is good corrrelation beetween the analytical
a curve and thhe
exxperimental results. Fuurthermore, when Eq. 3.27 3 is usedd (length too width ratiio
iggnored) there is significant deviationn from the experiment
e ((> 12%), whhile using Eq.
3.29 that acccounts for thhe length too width ratio, the differrence betweeen analytical
reesults and exxperimental measuremen
m nts is reduced (< 6%).

FIIG. 3.8. VR (ρ11) as a function

f of direction
d (θ) for length too width ratioo (λ=7.67). (a)
thhickness h1 = 0.18mm, (b) thickness (h2 = 5.40mm).

 
40 Chhapter 3
 

Figgure 3.9 pressents the expperimental reesults of the resistivity vversus thicknness as
a function of the diffferent fiber directions
d in the case of UD
U carbon ffiber preform m. The
VR decreases, when n the numbeer of UD caarbon fiber layersl increaase and reacches a
constant value after a large numbber of carbonn fiber layerss.

FIG. 3.9. Volume resisstivity (ρ11) ass a function of

o direction (θθ) and thickneess (h) for lenngth to
width ratio (λ = 7.67).

Forr (λ >> λcr) the VR exppressed by Eq. E 3.29 apprroaches the expression of Eq.
3.27 andd in this casee the resistivvity is indeppendent of the
t length too width ratioo. The
experimeental and annalytical ressults are prresented in Fig. 3.10(aa). The diffeerence
between the experim mental meassurements annd analyticaal results is smaller thann 2%.
Fig. 3.100(b) presentss the VR ass a functionn of the thicckness and the t fiber dirrection
ρ(θ,H) foor (λ >> λcr).
Figgure 3.11 shoows the expperimentally measured variation
v of the VR alonng the
length off the UD carbbon fiber specimen prefform, versus the number of the UD carbon
c
fiber pliees used. The experimenttal data havee been fitted using the poower low givven in
Eq. 3.31..
Thee experimenttal data in Fiig. 3.11 are fitted
f with a power curvve:

 
E of unidirectiional CF and CFRP
EC C media 4
41
 

N  b  aN  b
ρ θ,N  aN (3.311)

where (N) is the

w t number of
o carbon layyers used. T
The absolute difference of
o the VR is
giiven by Eq. 3.32.

ρθ,N1  ρθ,NNcr   N1  Ncrc (3.322)

FIIG. 3.10. (a)) Volume ressistivity ρ11 as a a functionn of carbon fiber directio on (θ) for twwo
diifferent valuees of thicknesss (h) and lenggth to width ratio
r (λ>> λcrr), (b) Theoreetical VR ρ11 aas
a function of carbon fiber ddirection (θ) annd the thickneess.

N1=1 and Ncrr is the numbber of layers above whichh the VR rem
mains almosst constant.
For Nk > Ncr, Eq. 3.33.

 
42 Chapter 3
 

ρθ,Ncr  ρθ,Nk  c 

c  aN k b a
  Nk  b 
b
ln a
c   
c ln  N k 
ρ 
ln  θ,1
a ρθ,1
 ρθ,Nk 
b 
c=ρθ,Nk ln  N k  (3.33)

Hence, Eq. 3.31 concludes the following form, Eq. 3.34.

 
ρ 
ln θ,1 

 ρ
 θ,Nk  (3.34)
  
ln Nk 
ρθ,N  ρθ,1 N  

Based on Eq. 3.34 it is concluded that the value of the resistivity as a function of
thickness depends upon the constant initial value of the VR of the single ply and the
number of plies power to a constant exponent. As it can be seen in Fig. 3.11, the
absolute difference (Δρ) is a linear function of fiber direction θ. For (θ=90o) this
variance reaches the maximum value  ρ 190 N  m ax and for (θ=0o) this difference cr

equals to zero  ρ 10 N  0 .

cr

Then, the VR for an amount of (Nk) layers is expressed by Eq. 3.35.

ρ θ,N k  ρ θ,1  ρ1θ N k  ρ θ,1   (3.35)

where parameter (μ) is the slope of Fig. 3.11(b). The final equation for the VR versus
fiber direction (θ) and number of layers (N) is given by the Eq. 3.36.
 
 ρ 
ln  θ,1 
ρ   
  ,1  (3.36)
  

ln  N k 
ρ θ,N  ρ θ,1 N  
, N  N k , 

Expressing Eq. 3.36 in terms of total thickness H, H  Nh , the final expression

takes the form of Eq. 3.37:

ρ 
ln ρ θ,1 
 
ρθ,1   ,1  (3.37)
ρθ,H  b
H b , b   
, H ,
h ln Nk 
 

 
E of unidirectiional CF and CFRP
EC C media 4
43
 

Equatio
on 3.37 exprresses the VR
R ρ(θ,h) as a function oof the fiber direction annd
thhe total thick
knesses undeer constant pressure
p of P  100 kPa
a , which conncludes in thhe
caase of dry prreforms to a v f  0.5 .

FIIG. 3.11. (a) Volume resisstivity ρ11 as a function off direction (θ)) and thickneess (h) for λ >>
>
λccr, (b) absolutte difference Δρ
Δ as a functiion of directioon (θ).

 
44 Chhapter 3
 

Combining Eq.
C E 3.30 andd Eq. 3.37, generalized expressionss of the electrical
resistancee of a UD carbon fibeer preform along the specimen’s llength and of the
respectivve VR are concluded foor different length
l to wiidth ratios (λλ), fiber dirrection
and nummber of carboon fiber pliees. These ressults are sum
mmarized in Eq. 3.38 an nd Eq.
3.39 resppectively.
( L  W cot  )
R11    2 (h) sin 2   1 cos 2   
HW
(3.38)
(

 ρ   ( L  W cot  )
R11   2,1
b
H  b  siin 2   1 coos 2  
 h   H
HW
and
 ρ 2,1    coot  
11   H  b  sin 2   1 cos
c 2   1  (3.39)
(
 h
b
    

Theese equatio
ons can caalculate thee electricall resistancee or the VR
V (
 , , H  0.18mm ). Figure 3.112 shows pllots resultedd by Eq. 3.338 as well as the
experimeental results of the VR as a functioon of lengthh to width rratio (λ) and
d fiber
directionn for total thickness (H1 = 0.18mm) and
a (H2 = 5.44mm).

FIG. 3.122. Analytical formulation

f a the experrimental resullts of the voluume resistivity as a
and
function oof length to width
w ratio λ and fiber dirrection for thhickness H1 = 0.18mm andd H2 =
5.4mm.

 
E of unidirectiional CF and CFRP
EC C media 4
45
 

In Fig. 3.13 the VRR is presenteed versus rattio λ and thiickness for fiber
f directioon
ο
(θθ = 45 ). Whhen ratio (λ < λcr), then the VR is the t same forr different thhickness sincce
thhe dominant resistivity is the resistivvity along thhe fibers direection and alll layers act aas
parallel resisttors. Eq. 3.30 is independent of thhe laminate thickness. On the otheer
hand, for (λ > λcr) the callculation of the VR is prrovided by Eq. E 3.38 andd incorporatees
thhe thickness term (H).

`
FIIG. 3.13. Vollume resistivity as a functioon of ratio (λλ) and thickness for directioon θ = 45ο.

More thhan 150 testss were carrieed out on drry UD carboon fiber prefforms in ordeer
too confirm thhe theoretical relations and to extrract the maathematical formulas that
coorrelate the single UD carbon
c fiberr layer electrrical propertties of (0o annd 90o) to thhe
ellectrical properties for every
e directiion, thicknesss and aspecct ratio. The experimenttal
seetup proved to be flexibble and reliaable, but thee handling aand cutting of the fabriccs
m be very careful in orrder to minim
must mize fiber-laayout disruption.
The exttracted equaations can be b used to predict the VR of dryy UD carboon
prreforms. Thhe major connclusion of thet present w work is thatt any UD caarbon preform m
o
caan be fully characterized
c d by perform
ming only onne measurem ment at a 90 direction foor

 
46 Chapter 3
 

a single layer at a fiber volume fraction ( v f  0.5 ). In Table 3.5 some of the obtained
results of the calculated and measured VR for various fiber directions, length to width
ratio (λ) and number of plies are presented.

Spatial Total Fiber Volume resistivity Volume resistivity Relative

Ratio Thickness Direction (Experimental) (Theoretical) Difference
λ Η (mm) (θ) ρ11 (Ωm) ρ11 (Ωm) δ (%)

4.795 1.80 45 0.009536 0.008832 7.97

4.795 1.80 60 0.015212 0.014724 3.32
4.137 0.18 30 0.003611 0.003837 5.88
3.068 0.18 30 0.002728 0.002874 5.12
6.534 0.18 60 0.017939 0.018050 0.62
5.438 0.18 60 0.016897 0.017698 4.53
4.342 0.18 60 0.016216 0.017167 5.54
3.246 0.18 60 0.015663 0.016278 3.79
2.151 0.18 60 0.014730 0.014485 1.69
6.575 0.18 45 0.010663 0.011193 4.73
5.397 0.18 45 0.010732 0.010754 0.21
4.301 0.18 45 0.009554 0.010131 5.70
3.205 0.18 45 0.008569 0.009082 5.65
2.109 0.18 45 0.006613 0.006942 4.76
1.561 0.18 45 0.005025 0.004747 5.86
7.671 1.80 30 0.004259 0.004258 0.014
7.671 5.40 30 0.004188 0.003987 5.04
7.671 0.18 45 0.010831 0.011479 5.64
7.671 1.8 45 0.009339 0.009566 2.38
7.671 5.4 45 0.009081 0.008957 1.38
7.671 0.18 60 0.018049 0.018310 1.43
7.671 0.9 60 0.015217 0.016229 6.24
7.671 2.7 60 0.014712 0.014495 1.50
7.671 5.4 60 0.015134 0.014287 5.93
28 0.18 30 0.006214 0.006192 0.36
28 5.4 30 0.004950 0.004831 2.45
28 0.9 45 0.011298 0.011282 0.14
28 2.7 45 0.010092 0.010077 0.15
28 5.4 60 0.015139 0.015131 0.05
0.849 0.18 30 0.000086 0.000084 2.38
1.370 0.18 30 0.000213 0.000206 3.40
0.714 0.18 45 0.000243 0.000225 8.00
0.547 0.18 45 0.000141 0.000143 1.40
TABLE 3.5. Results of calculated and measured VR for various directions, aspect ratio (λ)
and thickness.

 
E of unidirectiional CF and CFRP
EC C media 4
47
 

3.3.3.2 Resiistivity as a function

f of thickness
t andd fiber volum
me fraction
The electric conductivity
T c y of the poroous material is dependennt on the app plied pressurre,
hence the fibrre volume frraction. The fibre volum me fraction ass a function of the applieed
prressure, cann be expresssed by the following
f em
mpirical formmula Eq. 3..40 (Saunders
(11999), Chen et al. (2000), Chen et all. (2001)).

v f  aPb (3.400)

The queestion is, whhether the EC

C could be expressed
e as a function of
o pressure oor
ass a function of the fibre volume
v fraction and thicckness of thee porous matterial.

FIIG. 3.14. (a) Electric resisstivity as a funnction of time, (b) variatioon of electric resistivity ass a
fuunction of tim
me.

 
48 Chhapter 3
 

In oorder to linkk all the aboove with eacch other, cerrtain assumpptions were made.
The mateerial's visco--elastic behaavior, underr certain connditions, cauuses deformaations.
We assuume that th he electric conductivityy was meaasured at thhe final staage of
deformattion (at the steady
s state)). In other words,
w we assume that thhe material cannot
c
further deform.
d The measuremennts were takken after a certain
c periood of time. Figure
F
3.14(a) demonstrates
d s the electriic resistivityy (R) as a function
f of time, for various
v
number of o plies, hen nce for variious values of thicknesss. Figure 3..14(b) depiccts the
variationn of resistivitty as a functiion of time.

FIG. 3.155. (a) Preformm thickness ass a function of

o the appliedd pressure, (b) preform thiickness
per layer aas a function of the appliedd pressure.

 
EC of unidirectional CF and CFRP media 49
 

We can observe that at all cases, after the passing of approximately 200s there is
no significant change in the measurements of the electric resistivity. This, becomes
clearer in Fig. 3.14(b). Measurements were taken for each pressure level (0.1Mpa,
0.2Mpa, 0.4Mpa, 0.6Mpa, 0.8Mpa, 0.9Mpa) applied to (2, 6, 9, 24) number of layers.
For each specimen, the measurement was repeated three times. The electric resistivity
measurements were not significantly different from each other.
It can be observed in Fig. 3.15(a), logically, that the layers' thickness (H) tends
to become fixed at a steady value, whereas in Fig. 3.15(b) it is demonstrated that the
thickness per ply (h) value is greater for the cases where the ply number is smaller
((h2 > h24). This is a typical behavior of porous, fibrous materials and is thoroughly
explained in the following reports (Chen et. al (2000), Chen et. al (2001)).
It should also be noted that the h(P) curves can be very well and accurately
represented by a logarithmic curve, Fig. 3.15(b).
The diagrams occurred by calculating the fibre volume fraction per ply, as a
function of pressure. Fig. 3.16(a) demonstrates the fibre volume fraction as a function
of the ply's thickness, for various number of plies. It is obvious that it is easier to
achieve higher fibre volume fraction, for increasing number of plies. The results were
compared to the manufacturer's data. Figure 3.16(b) presents the occurring curves of
the fibre volume fraction as a function of the applied pressure. These are typical
curves that can be described by the aforementioned empirical equation Eq. 3.40.
A research paper published in 2001 links the pressure to the fibre volume
fraction for various numbers of layers, taking into account the natural properties and
not using the empirical formula. Therefore the thickness and the fibre volume fraction
are directly linked to the applied pressure. A micromechanical conceptual model has
been established by (Chen et. al (2001)). The bulk compressibility (Cb) is expressed as
a function of porosity and five other parameters (Cpo, Cs, φo, φf and k), Eq. 3.41.

(1  v f )(v ff  v f )k  (1  v f )(v ff  v f )k 
Cb (v f )  Cb0  2  v f   Cs (3.41)
(1  v0f )(v ff  v0f )k  (v ff  vof )k 

where ( vof = 1 - φ0) and ( v ff = 1 - φf) are respectively, the initial and final fiber volume
fraction. The fiber compressibility (Cs) can readily be identified or interpreted from
the slope of the thickness-pressure curve in Regime 3, Fig. 3.15. The final preform
fiber volume fraction ( v ff ), the initial preform bulk compressibility (Cbo), and the

 
50 Chapter 3
 

empirical index (k) need to be determined from experiments. We suppose that ( v ff ),

(Cbo) and (k) are related to fabric cross-sectional geometry and weaving pattern and
thus are fairly independent of fiber material properties. In terms of preform thickness
(Η), we have.

1  (v of H o H )  1  H f H 
Cb ( H )   C 
1  v of  1  H f H  b0
 o 
(3.42)
 vo H  vo H  1 H f H  
k

  2  f o  1  f o     Cs
 H  H 1  H H  
   f o  

Thus, from Eq. 3.42, the thickness - pressure relation is given by Eq. 3.43.

1  H 
P 1   (3.43)
Cb ( H )  H o 

Considering Eq. 3.42, an equivalent relation between pressure (P) and fiber
volume fraction (vf) can be expressed by Eq. 3.44.

1  vf
 o

P  1  (3.44)
Cb (v f )  v f

The above two equations, together with Eq. 3.41 and Eq. 3.42, allow the
constitutive law to be presented in terms of pressure vs. thickness or pressure vs.
fabric fiber volume fraction. In this case, the constitutive law is defined with only five
parameters. These parameters have clear physical meaning and are directly
measurable except for the empirical index (k). In the previous paragraph, we linked
the VR to the number of plies as well as to the preform's thickness and direction.
Neglecting the direction and focusing on the most important part of the equation, the
VR can be expressed as a function of the thickness or the number of plies, by the
following equations Eq. 3.45.
 
 
ln ρo   
ρ 
ρ



NF  ln ρ o 

(3.45)
ln NF  ρo b  NF 
ρ N  ρo N  
or ρH  H ,b   

hb ln NF 
 

 
E of unidirectiional CF and CFRP
EC C media 5
51
 

FIIG. 3.16. (a)) Thickness per

p layer forr different layyers number as a functio
on of the fiber
voolume fractioon, (b)

The VRR is also deppendent on thet applied pressure.

p Thhis dependennce occurs fofor
wo reasons a) the layerrs' thickness decreases as
tw a the presssure increasees (non-lineaar
fuunction of appplied presssure), b) the number of fibre contaccts increases,, as describeed
a (2001), annd the contacct surface off the fibers bbecomes largger. Electrical
byy Chen et. al
reesistance meeasurements were taken from specim mens of varioous number of layers, annd

 
52 Chhapter 3
 

for varioous pressure levels Fig. 3.14, using the aforem mentioned settup, enclosed in a
pressurizzed containerr. The followwing diagram m demonstraates the prevviously mentioned
measurem ments, Fig. 3.17. It is oobvious that the electric resistance is
i dependentt upon
the numbber of layers, and that it decreases
d (aas expected) and tends too stabilize.

FIG. 3.17. Electrical reesistance meaasurements as a function off pressure andd number of laayers.

In order to caalculate the specific ressistance, it is necessaryy that the layers'

l
thicknesss be known. Due to the setup
s being enclosed in a container, it was impoossible
to retrievve thicknesss measuremeents during the test. Thherefore, thee layers' thicckness
was meaasured as a function off applied prressure, usinng a mechannical comprression
setup, as described inn Appendix II.I
Knoowing the sppecimens' geeometrical characteristiccs for every ppressure lev vel and
number of o plies, it is possible too determine the specificc resistivity for all casess. The
followingg diagrams demonstrate
d e these meassurements ass a function of thicknesss, Fig.
3.18(b), or
o as a function of num mber of plies,, Fig. 3.18(aa), for variouus pressure levels.
l
The empirical formula Eq. 3.31 is i valid for every
e pressurre level.

 
E of unidirectiional CF and CFRP
EC C media 5
53
 

Inn all cases, the VR teends to becoome constannt, at a cerrtain value. It has to be b
m
mentioned th
hat other currves were tessted (logaritthmic, etc.), but they faiiled to satisffy
thhe electric measurements
m s.

FIIG. 3.18. Vollume resistiviity as a functiion of pressuure and (a) nuumber of layers, (b) mediuum
thhickness.

 
54 Chhapter 3
 

FIG. 3.199. (a) VR as a function of pressure and number of laayers, (b) schhematic presen
ntation
of the VR
R calculation.

Thee VR of twoo plies can be expresseed as a funcction of presssure for diffferent

number oof plies in a similar wayy. If the VR is expressedd as a functiion of pressuure (or

 
EC of unidirectional CF and CFRP media 55
 

dimensionless pressure P/Pmax), Fig. 3.19(a), were (Pmax) is the pressure value that if is
exceeded, the thickness remains approximately constant, the following empirical
equation occurs,

 P 
ρ (P)    (3.46)
 Pmax 
where (P) is the applied pressure.
For P / Pmax = 1 the VR is:

ρ (P)  ρ PF   (3.47)

ρPF is measured, therefore it is known.

 ρ Po 
 ln N
 PF  
PF  Po  ρ N  Pmax   ρN 
Po

N  ρN 
ρ Po   PF      (3.48)
 Pmax  ρ N  Po   Pmax 
ln  
 Po 
Hence, the final expression of the VR as a function of pressure is given by Eq. 3.49.

 ρ Po 
ln  N PF 
 ρ N 
P  P  (3.49)
ρ(P,N)  ρ PNF  max  ln  max 
 Po 
 P 

Using Eq. 3.45, it is possible to calculate the specific resistance for various number of
layers (N), Fig. 3.11. Then, the calculated specific resistance is introduced in Eq. 3.49
in order to calculate the VR as a function of the applied pressure, Fig. 3.19(b).
The calculation of the VR as a function of the number of plies and the applied
pressure, demands that we experimentally measure the VR of four extreme cases, as
shown in Fig. 3.19(b).
For example, if we want to calculate the VR at pressure of 0.4 MPa, for a
material of 8 layers, we can use Eq. 3.45, for (Ν = 8) , while (ρο), (ρΝF) are known for
pressure of 0.1 MPa.
 
ln  0.031 
0.012 
 
ln  30  (3.50)
(8)  0.031*8
ρ 0.1  0.01735m
 

The VR can be calculated using Eq. 3.45, for (Ν = 8) , while ( oPF ), ( NF
PF
) are known
for pressure of P > 0.8 MPa.

 
56 Chhapter 3
 

 
ln  0.014 
0.005 
 
(3.51)
(
ln  30 
ρ1(88)  0.014 *88  
 0.00746m
Using Eqq. 3.49, we calculate
c the VR for presssure of P = 0.4 MPa.

 0.1 
ln ρ8 
 
 ρ18  ln 0.01735 
0.007746 

 
(3.52)
(
P  
ln Pmax

  0.8  
ln 0.8


ρ (0.4,8)  ρ1(8)  max  P 
 o   0.000746    0.1 
  0.009888m
 P   0.4 

Thee VR can allso be expreessed as a fuunction of thhe fibre voluume fractionn. The
fibre voluume fractionn can be calcculated from
m Eq. 3.53.

NA
vf = (3.53)
(
H f

where (N N) is the num

mber of layeers, (A) is thhe areal denssity of the m
medium, (H) is the
total thickkness and (ρ
ρf) is the dennsity of the CF.
C
Figgure 3.20 preesents the exxperimental measuremennts of the VR R as a functtion of
the fibre volume fracction (vf) as a function off the numberr of plies.
Considerring that the layers' thickkness is linkeed to the pressure througgh Eq. 3.43, and to
the fibre volume den c be expreessed as a function
nsity throughh Eq. 3.44, thhen the VR can
of the fibbre volume fraction
fr by thhe followingg Eq. 3.54.

FIG. 3.20. Volume resiistivity as a fuunction of fibber volume fraaction.

 
E of unidirectiional CF and CFRP
EC C media 5
57
 

 ρ Po 
lnn  N P 
 ρ NF 
  P 
ln  max 
   Po 

PF  Pmax  (3.544)
ρ (v  ρN 
f ,N)
1  v of 
 1  
 Cb (v f )  v f
  

w
where vof is the
t uncomprressed fiber volume
v fracction, (Cb) is the bulk compressibilityy,
Po = is the applied
a presssure for low
w ratio (P / Pmax), (Pmaxx) is the appplied pressurre

oN
w
where the thickness remaains constantt, NPF is the VR
V at pressuure (Pmax), (  ) is the V
VR
att pressure (P
Po).

FIIG. 3.21. a) Experimental

E results of pressure – thickkness curves and b) experrimental resullts
off pressure - fiiber volume fraction
f curvees versus the calculated
c results using Chhens model.

 
58 Chhapter 3
 

Figgure 3.21 preesents the exxperimental reesults of presssure – thicknness curves and
a the
experimenntal results of pressure - fiber
f volume fraction curvves versus thhe calculated results
using Cheens model.
Finnally Fig. 3.2
22 presents calculated
c andd experimenttal curves of V
VR as a function of
fiber voluume fraction using
u the aforrementioned empirical
e relaation and the C
Chens model..

FIG. 3.222. Calculated and experimeental curves of VR as a function

fu of fibber volume fraction
fr
using the empirical relaation and the Chens modell.

3.3.4 Prediction
P o electricaal resistivity of unidirrectional CFRPs
of C
The preddiction of thee VR is moree complicateed in the casee of CFRPs.. This is due to the
manufactturing procedure, whicch ultimateely defines the fibre distribution at a
microscoopic level. The
T parameteers that affect the VR of o a CFRP material
m are: i) the
constructtion mold (appearance
( of cracks due to diff fferent CTE)), ii) the appplied
pressure and temperaature profile (formation of
o rich-in-resin areas bettween the lay yers)
thereforee, the mold must have approximateely the samee CTE as thhe layers, an nd the
pressure and temperaature profile should be suuch, that no resin rich laayers are form
med.

3.3.4.1 Resistivity as
a a functionn of thicknesss
The presssure and teemperature profile
p signiificantly afffect the micrrostructure of the
material. Low applieed pressure during the manufacturin
m ng procedurre, can lead to the
formationn of resin riich areas beetween the plies.
p Due tto that, the mmaterial's sp
pecific

 
E of unidirectiional CF and CFRP
EC C media 5
59
 

reesistivity in transverse tot the fibers'' direction can

c be assum med similar tot connectinng
reesistors in paarallel, Fig. 3.23.
3
This, coontradicts what
w we prevviously menntioned abouut the dry, porous
p mediia.
B if the exccess of resinn could be diminished,
But d t specific resistivity will
the w behave aas
thhe respectivee in dry, porrous materiaals. The folloowing imagee demonstrattes the resullts
foor various cases,
c Fig. 3.23. The specific resistivity of CCFRPs com mpared to thhe
reespective off dry, porouss materials, is higher for the same aapplied presssure level, aas
exxpected.

FFIG. 3.23. Voolume resistivvity transversse to the fiberrs as a functiion of numbeer of plies annd
ppressure usingg different toools and manuffacturing techhnics.

The eleectric resistivvity in transsverse to thhe layers cann be calculaated from thhe
emmpirical equuations, for various
v valuues of thicknness and fibrre volume frraction. In thhe
caase of CFRP Ps, the VR depends on other factorrs as well, ssuch as the mold and thhe
m
manufacturin ng proceduree. Furthermoore, the mateerial's anisottropy is greaater compareed
too the respecttive of dry, porous
p materrials.
Figure 3.24
3 presents the VR traansverse to fiber
fi directioon as a functiion of applieed
prressure and CFRP thickknesses. Wee can obserrve the samee behavior as in dry C CF
prreforms.
The microstructure
m e of a CF FRP materrial is stroongly depennded by thhe
m
manufacturin ng process annd the technnic. Impropeer technic annd pressure losses durinng
thhe manufactuuring processs or impropper temperatture and preessure profile can cause a
riich resin layyer between the
t CFRP laayers. The EC E of the lam minate in thiis case is verry

 
60 Chhapter 3
 

different,, especially transverse and throughh the thicknness directioons. Throug gh the
thicknesss, the rich resin
r layer forms
f an ellectrical insuulated layer. In this case the
electricall conductivitty its matchh less than thhe EC in thee transverse direction (σσ33 <<
σ22), Fig. 3.25. It must
m be notedd that similaar CFRP miicrostructuree can be obsserved
when theermal press is used improoperly.

FIG. 3.24. Volume resiistivity transvverse to the fibbers as a funcction of pressure for different
number off plies using CFRP
C tool.

RP microstructture material..1
FIG. 3.25. Rich resin layer in CFR
                                                            
1
This figurre has been kinndly provided by
b C. Kostagiaannakopoulou.

 
E of unidirectiional CF and CFRP
EC C media 6
61
 

Figuree 3.26 and Fig. 3.27 presents

p the CFRP miccrostructure using optical
m
microscopy for
fo using diffferent techniics and appliied pressure at manufactturing stage.
U
Using constannt applied prressure of 1bbar the CFRRP structure presents reaach resin layeer
between the structure,
s airr inclusions, non-uniform fiber volume fraction, Fig.
m and low fi
3.26(a-c). Usiing pressuree between (1 and 1.2 bar) alternatively the rich reesin layer annd
thhe air inclusiions were reduced, Fig. 3.26(d-f).
3
Final Fig.
F 3.28 prresents the CFRP miccrostructure transverse to the fibeer
t case wheere the applied pressuree is 5bar, Fig. 3.25. Thee interlaminaar
diirection in the
annd intralamiinar regionss have unifoorm structurre and the air a inclusionns have beeen
reeduced.

FIIG. 3.26. (a), (b), (c) CFRP

P microstructture using connstant pressurre (1bar)..

 
62 Chhapter 3
 

FIG. 3.27. CFRP microostructure usiing pressure between (1 baar and 1.2bar).

 
E of unidirectiional CF and CFRP
EC C media 63
 

FIIG. 3.28. (a), (b), (c) CFRP

P microstructture using connstant pressurre (5bar).

 
64 Chapter 3
 

 
Equivalent EC tensor for multidirectional layered CF media 65
 

4. start
Chapter 4
Equivalent EC tensor for multidirectional layered
media

The electric conductivity (EC) of the unidirectional CF preform and CFRP laminates
presents strong electric anisotropy and can be expressed by a symmetric, second order
tensor. The EC of multidirectional laminates can also be expressed by a different
second order tensor (equivalent EC tensor). The calculation of the equivalent EC
tensor is valid assuming that the multidirectional material is homogenous at the
interlaminar and intralaminar regions. Also, the plies' thickness is negligible in
comparison to the other dimensions of the body. Two methodologies were used in
order to validate the equivalent EC tensor. The first explores the validity of the
equivalent EC tensor by measuring the electrical resistance of the multidirectional
laminates, using a high-precision multi-meter. The second is based on the numerical
solution of the coupled electrical / thermal problem, Chapter 6. Also the electric
potential field and the current density distribution for 3D multilayer problem was
studied, assumed that the contact area between the interlaminar area is perfect. When
the thickness of the material is very small in comparison to the other dimensions of
the body, the electrical gradient through the thickness is negligible.
66 Chhapter 4
 

4.1. Electrical
E conducttivity off CF muultidirectiional lay
yered
media
Assuminng that the body
b is thinn (2D space approach), the EC tennsor can be easily
determinned by usingg the continuuity equatioon for anisottropic continnuous mediaa. The
followingg equation expresses
e thhe current coonservation for the anissotropic bodyy, Eq.
4.1.

 ( J)dV   J  dS
V S
(4.1)

The curreent density in anisotropic media exppressed by thhe Eq. 4.2.

  
   
 
J  12  x1
J  σ    1     11   
J2   21  22    
 x  (4.2)
 2
        
 J      11   12  i    21   22  j
 x1 x2   x1 x2  

Thee total curreent that passees through the

t surface S (total crosss section arrea), is
equal to the sum of the electric current thatt passes throough the cross section of each
( n), Fig. 4.1, where (n)) is the outw
layer of tthe surface (S ward unit norrmal vector of the
surfaces (S) and (Sn). Hence, thee integral of the dot prodduct of the cuurrent densitty and
the surfacce of the cro
oss section exxpressed by Eq. 4.3.

FIG. 4.1.. (a) Geomettry of the doomain and boundary

b connditions, (b) detail of thee thin
multilayeer medium.
Equivalent EC tensor for multidirectional layered CF media 67
 

N N
I   In   J  ndS    Jn  ndSn 
S Sn
n1 n1
N
(4.3)
   J1i  J2 j   ndS     Jn1i  Jn2 j   ndSn
S Sn
n1

where (J1) is the current density that crosses the representative volume in the direction
(x1), (J2) is the current density that crosses the representative volume in the direction
(x2), (Jn1) is the current density that crosses the nth CFRP layer in the direction (x1),
and (Jn2) is the current density that crosses the nth CFRP layer in the direction (x2).
Integrated Eq. 4.3 over the thickness (H) and the first side (ΔW) from (–ΔW/2 to
ΔW/2) of the representative volume and dividing by the total surface (S=H ΔW) the
current density at direction (x1) is given by Eq. 4.4.
N

S  J1i  J2 j   idS   S  Jn1i  Jn2 j   idSn 

n1 n

    N
   
  11  12 dS    11n  12n  dSn 
S
 x1 x2  n1
Sn
 x1 x2 
(4.4)
    N  n   
 S 11  12   Sn 11  12n 
 x1 x2  n1  x1 x2 
    1 N  n   
 J1  11  12   Sn 11  12n 
 x1 x2  S n1  x1 x2 
where (Sn = hn ΔW) is the cross section surface of each layer.
Following the same procedure and integrating over the other side of the
representative volume (from –ΔW/2 to ΔW/2) then it is concluded that the current
density at direction (x2) is given by Eq. 4.5.
N

S  J1i  J2 j   jdS  

n1
S  Jn1i  Jn2 j   jdSn 
n

    N
 n   
   21   22 dS      21   22n  dSn 
S
 x1 x2  n1
Sn
 x1 x2 
(4.5)
    N  n   
 S  21   22   Sn  21   22n 
 x1 x2  n1  x1 x2 
    1 N  n   
 J2   21   22   Sn  21   22n 
 x1 x2  S n1  x1 x2 
68 Chapter 4
 

The total current density (J) is equal to J=|J1|i+|J2|j. Based on Eq. 4.4 and Eq. 4.5 the
final expression of the equivalent EC tensor (σ) for the multidirectional laminate can
be expressed by Eq. 4.6.

       
  11   12  i    21   22 j
 x1 x2   x1 x2 
(4.6)
1 N     1 N  n  n  
  Sn   11n   12n  i   Sn   21   22 j
S n 1  x1 x2  S n 1  x1 x2 

In matrix form Eq. 4.7 can be expressed by the following equation.

    N N
   
 11   S n
 12   x1  1  n 1 n 11
S n
  x 
n 12

    n 1
 1 (4.7)
 21  22     S  N N
n    
  S n 21  S n 22
n
 x    x 
 2  n 1 n 1  2

Hence the equivalent EC tensor ( σ ) is the sum of the EC tensor of each layer ( σn ).

1 N 1 N
σ  
S n1
Snσ n  σ 
N
σ
n 1
n (4.8)

where (N) is the number of plies of the multidirectional CFRP laminate, considering
plies of constant thickness.
By combining plies of different fibre direction, a new material is produced
whose EC depends upon its layers’ electric conductivity. Figure 4.2 shows the EC
ellipse of each ply of the laminate as well as the equivalent EC tensor of the laminate
for four different stacking sequences. Further on, by using Eq. 4.9, one may define the
two eigenvalues (λ1,2) of the ellipse.
2
      22 
1,2  11 22
  11    12
2
(4.9)
2  2 
The directions of the principal axes have the following slopes, Eq. 4.10.

 
tan 1  12
, tan  2  12
(4.10)
1   22
2   22

where (θ1) and (θ2) are the angles between the x1-axis and the principal axes of the EC
tensor. Later on we will thoroughly analyze the fact that certain stacking sequences
(special cases) lead to an electrically isotropic material. As shown in Fig. 4.2(c), in the
cases where the fibre direction between two successive layers is Δθ = 90o (  1 ,  2
E
Equivalent EC tensor
t for multtidirectional layyered CF mediia 6
69
 

w
where 1  2 ) the values (σ12 = σ21 = 0) and (σ11 = σ22). Thiis configurattion results to
t
be the equivaalent EC tensor of the multidirection
m nal laminate a diagonal tensor, whicch
m be writteen in the form
may m of a scalarr quantity, Eq.
E 4.11.

1   11
  90
12   21
 90

9   90
11  22
σ   11 
θ
/
θ
+
9
0
9 
2  21   21
  900   90
 22   22  
12  90
 21
(4.111)
1   
 
0 
  11 22  σ

θ
/
θ
-
9
0
2 0  22  11 


In ordeer to make the aforem mentioned more

m compreehensible, we
w present an a
exxample of thhe calculatioon of the EC
C tensor, in the
t case wheere the stackking sequencce
iss (0/30/30/0)), Fig. 4.2(a)).

FIIG. 4.2. Reprresentation off the EC tenssor of each C

CFRP layer annd the equivaalent EC tensoor
foor the multiidirectional CFRP for different
d staccking sequennces, (a) (0 0/30/30/0), (bb)
(445/30/30/45), (c) (±45), (dd) (90/60/-455/90/-45/60/90). (ξ1, ξ2 arre the princip
pal axes of thhe
eqquivalent matterial).
70 Chapter 4
 

i.e. the EC tensor for each layer can be calculated using Eq. 4.12.

 cos 2    2 sin 2  ( 1   2 ) cos  sin  

σ ( )   1  (4.12)
 ( 1   2 ) cos  sin   2 cos    1 sin  
2 2

where (σ1 = 44287 S/m) and (σ2 = 50 S/m) is the EC along the fibre direction and
transverse to the fibre direction. Hence at (0o) the EC tensor can be calculated by Eq.
4.13.

 cos 2 0   2 sin 2 0 ( 1   2 ) cos 0 sin 0   44287 0 

σ  1  S/m
0

2  (4.13)
50 
︵ ︶
 ( 1   2 ) cos 0 sin 0  2 cos 0   1 sin 0   0
2

and the EC tensor at (30o) can be calculated by Eq. 4.14.

 cos2 30   2 sin 2 30 (1   2 ) cos 30 sin 30  33227.75 19155.18

σ  1  S /m
3
0

︵ ︶ (4.14)
 (1   2 ) cos30sin 30  2 cos 30  1 sin 30 19155.18 11109.25
2 2

In the case where the thicknesses of the CFRP layers are equal to each other and using
the Eq. 4.8, the equivalent EC tensor of the (0/30/30/0) CFRP stacking sequence can
be calculated.

1  2 11  2 11 2 120  2 1230  38757.37 9577.59 

N 4 0 30
σ   S /m
0
/
3
0

︵ ︶  0 30  (4.15)
s

4  2 21  2 21
30
2 22
0
 2 22   9577.59 5579.62 
Then the eigenvalues are λ1 = 41323.68 S/m, λ2 = 3013.32 S/m and the slopes tanθ1 = 
0.2679 (θ1 =15ο), tanθ2 = -3.7320 (θ2 = -75ο).
It has to be noted that the aforementioned formula is applicable to other
composite materials fields. Specifically, as far as flow in porous multidirectional
media is concerned, the equivalent permeability tensor can be calculated similarly
(Advani et. al (1991), Greenkorn (1983)). Another reference has been tracked at the
piezoresistivity and piezoconductivity of orthotropic materials on the basis of
electrostatic theory and the classical thin laminate theory (Xiang et. al (1999)).

4.2. Measurements and validation of the equivalent EC tensor

The unidirectional laminates consist of CF plies of the same direction whereas in the
case of the multidirectional laminates, at least one layer is of different direction from
the others. The multidirectional laminates can be categorised as symmetric, anti-
Equivalent EC tensor for multidirectional layered CF media 71
 

symmetric and non - symmetric. A variety of different cases were studied in order to
verify the applicability of Eq. 4.8 for the calculation of equivalent EC of a CFRP
laminate.

4.2.1 Microstructural characterization

The homogeneity of the CFRP laminates under investigation is of great importance
for the EC of the CFRPs. During the manufacturing process of the CFRP laminates,
the material is compressed, resin flows through the fibres and the excess resin is
captured inside an appropriate cloth (breather cloth). In order to consider a
multilayered material as homogenous, it is necessary to examine the intralaminar and
interlaminar regions. In the case where a rich resin layer is formed between the CF
plies of the CFRP laminate, it results in significant reduction of the fibre volume
fraction between the plies. Therefore the material lacks its macroscopic homogeneity
at the interlaminar areas and the presence of a thin resin film restricts the contact
between the conductive CFs and could electrically insulate the laminate through the
thickness.
The presence of a rich resin film between the plies and its thickness is dependent
upon the manufacturing procedure and it is mainly driven by the pressure applied on
the surface of the CFRP laminate, the temperature profile and the resin viscosity. The
degree of inhomogeneity of the multidirectional laminates is dependent upon the
applied pressure during the manufacturing procedure. Figure 4.3 shows optical
microscope pictures of CFRP laminates manufactured by using different pressure
values of (P1 = 0.01 MPa, P2 = 0.1 MPa and P3 = 0.8 MPa). The manufacturing of the
specimens involved CF prepregs supplied by SGL (Sigrafill / Sigratex, E022 epoxy,
UD-prepreg, 242 g/m2) and the use of an autoclave, in order meet the appropriate
temperature profile (3h at 130oC).
Application of low pressure causes in the development of an inhomogeneous material,
resulting in:
a) Resin rich areas at interlaminar regions,
b) Non-uniform fibre distribution (volume fraction) at intralaminar regions,
c) Imperfections between and inside the plies.
The problems mentioned above are dependent upon the manufacturing process
and mainly upon the pressure applied. The imperfections in the CFRP structure and
the non-uniform CF distribution disturbs the fibre contacts and the pressure at the
72 Chhapter 4
 

contact areas,
a increasses the contaact resistance between thhe fibres andd results in a non -
macroscoopically hom mogeneous electrical conductive
c material. B
By increasinng the
applied ppressure, the thickness of the resin riich film is reeduced. For high pressurre, the
resin richh layer is alm
most diminished.
Figgures 4.3(a) and
a (b) dem monstrate twoo adjacent CF C plies, whiich resulted under
manufactturing pressure of (P1 = 0.01MPa)). We can cclearly obserrve the richh resin
areas bettween the fib bres as well as
a a (≈ 26μm m) thick resinn layer betwween the pliees. It is
clear thatt the fibre voolume fractioon is not connstant at all the
t areas.

FIG. 4.3.. Optical miccroscopy at multidirection

m nal laminates as a functionn of manufaccturing
MPa, (c), (d) 0.1MPa, (e), (f) 0.8MPa.
pressure. (a), (b) 0.01M
Equivalent EC tensor for multidirectional layered CF media 73
 

It is obvious that the applied pressure during the manufacturing process affects the
fibre volume fraction and its homogeneous distribution, the resin-rich layer
characteristics as well as the micro-structural imperfections. In the discussed case of
(P1 = 0.01MPa) manufacturing pressure, the resin-rich layer thickness is of about 6 to
7 times the CF fibre diameter.
Figures 4.3(c) and (d) demonstrate two adjacent CF plies resulted under (P2 =
0.1 MPa) manufacturing pressure. The resin-rich layer’s thickness has been reduced
to (≈ 15μm) and the fibre concentration has increased.
Finally, Figs 4.3(e) and (f) show the resulted adjacent CF plies under
manufacturing pressure of (P3 = 0.8 MPa). The resin rich layer in this case is almost
diminished and the fibre volume fraction is higher and almost homogeneous, causing
the formation of almost perfect periodic hexagonal array. According to these figures,
one expects different electrical conductivity results for each case.
As it is already discussed earlier, higher fibre volume fraction results to higher
number of fibres-in-contact. Low volume fraction areas are much less conductive.
Despite the fact that the resin-rich film is thin, the CF layers are insulated in the same
way as if a thin plastic film was placed between them. However, even at the lowest
manufacturing pressure, statistically only certain number of CFs will actually be in
contact. In any case, the imperfections and the low fibre volume fraction regions
significantly contribute to the disturbance of the homogeneity of electric field at a
microscopic level, thus affecting the macroscopic electrical behaviour of the laminate.
The application of an electric potential difference between the edges of an
electrically conductive material leads to a temperature rise due to the Joule heating
effect. Both the electric field and the temperature distribution developed in the case of
a conductive laminate, are seriously affected by the material’s microstructure. The
non-uniformity of the developed temperature profile can be considered as a measure
of the laminate's conductive non - homogeneity. The current study is focused on the
approximation of the ideal case where the resin-rich layer is almost diminished, while
the fibre volume fraction is assumed homogenous and macroscopically constant.
Before we present the numerical results of the temperature field, it is imperative
that we compare the results of the homogeneous CFRP structure (detail of Fig. 4.4(b))
to the respective non-homogeneous (detail of Fig. 4.4(a)), in the case where electric
potential is applied. CFRP laminates with a stacking sequence of (±45) were used for
two cases of applied manufacturing pressure (0.01 MPa and 0.8 MPa respectively).
74 Chhapter 4
 

The resuulted temperature field, due to an electric

e potential at the free edges of the
specimenns is differennt for each caase, as it is shown
s in Figg. 4.4.

FIG. 4.4. Thermal cam

mera images att two differennt pressure levvels for (±45)) stacking seqquence.
a) 0.01 M
MPa and b) 0.88 MPa.

Hoowever, according to Eq. 4.8, thhe concludeed EC shoould lead to t the

developm ment of a unniform temperature fieldd at the full surface of tthe specimenn. The
monitoreed temperatuure discrepanncies demonstrated in Fiig. 4.4(a) aree not in agreeement
with the theoreticallly expected temperaturee profiles and
a they steem from thee non-
homogenneity of the EC tensor,, which apppears in thee case of loow manufaccturing
pressure (0.01 MPaa). This disccrepancy is not observved after thee increase of o the
manufactturing pressuure to (0.8 MPa),
M Fig. 4.4(b).
Prooceeding ontto two CFR RP specimens of stackking sequencce (±30) fo or two
different cases of appplied pressuure during thheir manufaacturing stagges, this tim
me first
case of aapplied presssure being 0..1 MPa and second casee of applied ppressure bein ng 0.8
MPa, wee can observve that the teemperature non-uniform
n mity is greateer in the firsst case
than in thhe second caase, Fig. 4.5..The maximmum temperaature deviatioon in the firsst case
is (ΔΤ = Τmax-0.86Τmaax), while in the second case
c is (ΔΤ = Τmax-0.93Τm
max).

Thee following g diagram demonstraates the temperature

t distributio
on in
dimensioonless form along the specimens’’ length (0.1MPa, 0.8MPa) over three
different regions, L11, L2, L3. T
The two casees of appliedd pressure aare depicted in the
followingg diagram as a red and a black linee, first case being the appplied pressure of
(0.1 MPaa) and second case beingg the appliedd pressure off 0.8 MPa, reespectively. As far
E
Equivalent EC tensor
t for multtidirectional layyered CF mediia 7
75
 

ass the first caase is concernned (0.1 MP

Pa), the tempperature non--uniformity over the threee
diifferent areaas of the speccimen, is obvvious.
It has too be noted that the tem mperature gradient at thee tips of thee specimen is
caaused by heaat transfer duue to conducction throughh the copperr connectors..

FIIG. 4.5. Dimeensionless tem

mperature of (±30) stackinng sequence at
a L1, L2 and L3 regions fo
for
tw
wo different pressure
p levels (0.1MPa annd 0.8MPa).

44.2.2 Validdation of the

t equivaalent EC tensor
t via electricall resistancce
m
measureme ents
The bassic assumptiion of the annalysis provided above is that Eq. 4.8 4 is valid in i
thhe case wheere the CFR RP laminate is homogenneous and iits thicknesss is relativelly
smmall (Athannasopoulos & Kostopoullos (2012)d)), comparedd to its otherr dimensionns.
T contacts between thee adjacent pliies of differeent fibre direection withinn the laminatte
The
crreate a neww, anisotroppic material.. Thus, the electric cuurrent is connducted in a
diifferent way
y compared to t the case of o the singlee lamina or tthe UD lam minate. For thhe
validation of the Eq. 4.8 small thickkness CFRP (high lengthh to thicknesss ratio (L/H H)
annd width to thickness
t (W
W/H) ratio haas been applied).
Measuriing the electtrical resistaance of an orrthogonal CF FRP laminatte, Fig. 4.6(aa)
having a lenggth of (L = 210 mm), a width of (W W = 70 mm m) and a thickness of (h =
o
0.125 mm) which
w consistts of only onne ply of (455 ) directionn, we retrievee an electrical
reesistance meeasurement oof (R11 ≈ 1660 Ω) accorrding to refeerences (Weber & Kam mal
76 Chhapter 4
 

(1997), Athanasopou
A ulos & Kostoopoulos (2011)). The m measured resiistance will be the
o
same in the
t case whhere the direcction is (-455 ). In the caase where alll of the pliees of a
CFRP lam o (45o), it iss useless to measure
minate are of m the EC of the U UD laminatee since
the curreent density vector
v is not normal to thhe equipotenntial lines, Fig.
F 4.6(b) annd (c).
Using thhe CF preprreg plies inn a (±45) manner m for constructing a (±45) CFRP
laminate,, the measurred resistance along the length
l directtion is (R11 = 0.54 Ω). As
A it is
already presented
p earrlier the off diagonal terrms of the EC
E tensor havve to be zero o, Fig.
4.6(d) annd (e).
e for every stacking seqquence (θ/-θθ) where thee fibre direcction betweeen two
i. e.
successivve layers is Δθ = 90o thhe off diagoonal terms oof the conduuctivity tensor are
vanishedd since       as shhown in Eq. 4.12. Hence the electriical resistancce can
12

12

be calcullated using Eq.

E 4.16.
 
 /  1 L Eq.8 1 L 11 11 1 L  455 45/ 45 1 L
R11      R11  45
11 /  WH 11  111  WH 11 WH 111 WH (4.16)
(
2
where (H
H) is the totaal thickness of the CFRP
P laminate, (L) and (W)) is the lengtth and
the widthh of the lamiinate respecttively.

FIG. 4.6. Differences between currrent density distribution aand electric ppotential field d for a
o
unidirectiional laminatte (45 ) and a multidirectiional laminatte (±45). (a) orthogonal domain
d
with bounndary conditions, (b) curreent density strreamlines at 445o, (c) electrric potential field
f at
45o, (d) cuurrent densityy streamlines at (±45), (e) electric potenntial field at (±
±45).
E
Equivalent EC tensor
t for multtidirectional layyered CF mediia 7
77
 

Any deeviation of the electric field leadss to a non-uniform currrent densityy.

A
Assuming thaat each layerr of a CFRPP laminate iss homogenouus and that it has a giveen
ellectrical anissotropy, the only factorrs that affectt the distribuution of the electric fielld
annd the currennt density arre the geomeetry of the laaminate and the
t boundaryy conditionss.
Eq. 4.8 is valid in the
t case wheere the thickkness is relattively small,, compared to t
thhe other dimmensions. Thee contact of the two layeers create a nnew, anisotropic materiaal.
T
Therefore, th
he current isi conductedd in a diffeerent way. T The electrons now flow
thhrough the (M)
( surface (interlaminar region) F Fig 4.7(b) annd (c), as a result of thhe
chhange of thee EC tensor of the materrial. For thicck CFRP pliies, Fig 4.7(bb) the currennt
density value is low on thhe surfaces (U)( and (L) of o the specimmen, therefoore the currennt
'pprefers' to be
b conducteed along thee interlaminnar region oof the CF layers wherre
m
maximum coonductivity is
i observed, resulting inn high curreent density at a the surfacce
m
mentioned, as well as at the areas inn close proxximity. In thhe case of a thin material
w
which approximates thee ideal, the current deensity is coonstant alon ng each plyy's
thhickness, Figg 4.7(c).
Table 1 presents a complete
c sett of measureements for ddifferent CFR RP laminatees,
m
manufactured d by using ddifferent presssure levels.. The theoreetical results (Eq. 4.8) arre
coompared agaainst the eleectric measuurements takken from thee same lamin nates, and foor
diifferent mannufacturing pressure
p vallues (which concluded to different fibre volum me
frractions at th
he resulted laaminate).

FIIG. 4.7. Currrent density distribution

d thhrough the material
m thicknness. (a) isotrropic, (b) thicck
annisotropic mu
ultidirectionall laminate, (c)) thin anisotroopic multidireectional lamin
nate.
78 Chapter 4
 

The relative error between the experimental measurements and the analytical
calculations / results for relatively high manufacturing pressure is very small, and the
measurements are in excellent agreement with the theory.
It has to be mentioned that the off diagonal values of the equivalent EC tensor for all
the electrical resistance measurements of the CFRP specimens were equal to zero. In
all cases, a high-precision multimeter (Keithley - 2002) was used for measuring the
electrical resistance.
Figure 4.8 presents the mean error between the theoretical results and the
measured valued of the electrical resistance of CFRP laminates for the various
stacking sequences tested and the different manufacturing pressure levels applied. In
the case where the applied pressure is (P1 = 0.01 MPa) the mean error is higher than
(100%), mainly due to random fibre contacts and the material’s inhomogeneity. For a
pressure level at (P2 = 0.1 MPa), the mean error is of about (25%) with a relatively
significant standard deviation. In this case, the fibres are closer to each other, but still
the resin-rich layer between the plies is thick enough to insulate the CF layers.
Experimental
thickness -

Analytical

error (%)
Length to

Length to
Sequence
Pressure

Stacking

Relative
Applied

H (mm)
width –

h (mm)
(MPa)

(L/W)

(L/H)

(Ω)

(Ω)

(±30) 6.05 0.138 0.276 652.2 1.069 0.939 13.84

(±45) 5.94 0.134 0.268 671.6 1.709 1.403 21.80
0.1
(±60) 6.66 0.129 0.258 697.7 4.370 2.914 49.97
(±75) 6.07 0.128 0.256 703.1 15.360 11.095 38.44
(±30) 5.90 0.122 0.244 963.1 0.873 0.896 2.53
(±45) 5.90 0.122 0.244 963.1 1.470 1.344 9.42
0.6
(±60) 5.90 0.122 0.244 963.1 2.740 2.680 2.24
(±75) 5.90 0.122 0.244 963.1 9.360 10.010 6.49
(±30) 3.13 0.113 0.226 929.2 0.400 0.416 3.82
(±45) 3.13 0.113 0.226 929.2 0.583 0.624 6.51
(±60) 3.13 0.113 0.226 929.2 1.260 1.245 1.20
0.8
(-30/60) 3.13 0.113 0.226 929.2 0.615 0.624 1.38
(45/60/-
45/-60) 3.13 0.113 0.452 464.6 0.410 0.416 1.44
(0/±60)s 3.13 0.113 0.678 309.7 0.215 0.208 3.13

TABLE 4.1. Electrical resistance measurements of multidirectional CFRPs at different

manufacturing pressure levels. 
E
Equivalent EC tensor
t for multtidirectional layyered CF mediia 7
79
 

For presssure values over (P > 0.6 MPa), it is observed tthat the EC measuremen
m nts
teend to agreee with the thheoretical reesults, thus eliminating
e tthe error an
nd minimizinng
thhe standard deviation.
d Thhe fibre voluume fractionn of the laminnate is relatively high, thhe
pllies are in co ontact with eeach other annd there are no observedd imperfectioons within thhe
laaminate struccture. In the case of dry CF media, wherew there iis no presentt of resin, thhe
errror is signifficantly low,, reaching 4.44% for presssure levels of
o (P = 0.1 MPa).
M
It has to o be mentiooned that thee measuremeents were caarried out ussing the fouur-
ellectrode method. Conceerning the processing
p o the samplles prior to the tests, thhe
of
eddges of the samples werre scratchedd and cleanedd with acetoone. Then, thhree layers oof
siilver paint followed
fo by two layers of o silver tappe were appllied at the edges. Finallyy,
thhe specimens were placeed in a custoom-made devvice where high h pressure was applieed
too the electroddes that are in
i contact with
w the speciimen.
The eleectrical condductivity of the CF preepreg along its principaal axes of thhe
unnidirectional CFRP mateerial are (σ0 = σ1 = 442887 S/m) and (σ90 = σ2 = 50 5 S/m) wheen
thhe applied manufacturing
m g pressure iss equal to (0.8MPa).
In orderr to confirm the validity of the Eq. 4.8,
4 as well aas to point ouut the cases in i
w
which it cannnot be applied, we manuffactured appproximately ((80) specimeens of variouus
sttacking sequ uences.

FIIG. 4.8. Relaative error bettween the eleectrical measuurements andd theory (Eq. 8) for variouus
m
manufacturing pressure leveels.
80 Chhapter 4
 

4.2.3 Sensitivity
S of
o the meassurements
It has too be mentiooned that anny deviationn in the anggle could caause a signiificant
deviationn in the finall measuremeent. Therefore, the plies’ placement should be carried
c
out attenntively. The EC measurrements deppended on the t placemeent misalignnments
during thhe manufactuuring proceddure of the sppecimens.
In the case off (±45) staccking sequeence, the relative error of the electrical
resistancee measurem ments for one degree deviiation “±1o, (±46) real sttacking sequ uence“
is equal to 5%. Incrreasing the angle
a of the fiber directtion of the stacking
s seqquence
(±θ) the relative erroor of the meeasurement increases
i siggnificantly aand in a nonnlinear
way, Fig. 4.9(a).
In Fig.
F 4.9(b) presents
p thee dependence of the relative error forf fiber dirrection
deviationn (γ) at (±30,, ±45, ±60) stacking
s sequuences.

FIG. 4.9. (a) Measured d electrical reesistance errorr as a functioon of stackingg sequence (±
±θ) and
the misaliignment of thhe layer direction (±γ) during the manuufacturing stagge of the specimen,
(b) relativve error as a function off fiber directiion deviationn at (±30, ±4 45, ±60) staacking
sequencees.

Finnal Fig. 4.10 presents thee relative errror of the meeasured electtrical resistance as
a functioon of the anisotropy
a level of the material (rratio ρ22 / ρ11) for diffferent
misalignm ment directions (γ). For this specificc range of annisotropy levvels (CF preeforms
and CFR RPs) the relattive error is very
v small.
Thee manufactu uring proceddure and thee accurate pllacement acccuracy of thhe CF
layers arre both verry importantt in order to minimizee the error of the electrical
resistancee measurem ments.
E
Equivalent EC tensor
t for multtidirectional layyered CF mediia 8
81
 

FIIG. 4.10. Rellative error of the measureed electrical rresistance forr different annisotropy leveels
ment in fiber ddirection (γ) – (1ο, 2ο, 3ο, 5ο).
annd misalignm

4
4.3. 3D numerical
n l modelinng of multtidirectionnal CF media
m
Inn order to validate
v thee equivalent EC tensor,, 3D modells were devveloped usinng
C
COMSOL sooftware. An electrical 3D 3 domain problem
p cann be considered as a 2D
domain probllem when thhe thicknesss of the boddy is negligiible in comp parison to thhe
otther dimenssions of the body and therefore,
t thhe electricall gradient in
n direction x3
(tthrough the thickness) is negligible. Especiallyy in the casee of the anissotropic bodyy,
thhe electrical gradient thrrough the thiickness of thhe laminate depends on the EC of thhe
m
material at thhe x3 axis (σσ33). We studdied orthogoonal and circcular domain ns of differennt
thhickness layeers and diffeerent boundaary conditionns using CO OMSOL, in order
o to studdy
thhe electric pootential fieldd and the currrent densityy distributionn of the 3D domain.
d

4
4.3.1. Rectaangular doomains – Mixed
M bounndary condditions
Applying an electric poteential at the edges of a multilayer
A m m
material (±θ),, Fig. 4.11 thhe
ellectrical field changes according
a to the thicknesss of each laayer. Figure 4.11 presennts
thhe electric pootential fieldd as a functioon of thickneess (layer thhickness h1 = 20mm/layeer,
h2 = 5mm/layyer, h3 = 0.1225mm/layer)).
82 Chhapter 4
 

 
FIG. 4.11. Equipotentiial surfaces on
o a stacking sequence (±
±45) for differrent layer thiickness
(a) 20mm mm/layer, (c) 0.125mm/layyer.
m/layer, (b) 5m

Thee equipotenttial surfaces tend to aliggn with the fiber direction because of the
t material.. The interlaaminar regioon presents a sharp trannsition
high anissotropy of the
due to diiversion of the
t fiber direction (+45 to -45), Figg. 4.11(a) annd (b). In the case
where thhe thickness of each layyer is very small, the equipotential
e l surfaces teend to
parallelizze between them.
t This faact appears when
w the offf-axes values of the EC tensor
are equaal to zero (tthe same apppears in isootropic mateerials). An exception can c be
E
Equivalent EC tensor
t for multtidirectional layyered CF mediia 8
83
 

obbserved at the
t edges off the 3D dom main. In reaality, the elecctrical gradiient is alwayys
“tthere”. Howwever the inffluence is very
v small iff we considder that real CFRP layers
vary from 0.005mm to 0.25mm.
Figure 4.12 presennts the currrent density distributionn as a funcction of (±θθ)
sttacking sequuence for thiickness (h = 20mm). Ass it was expected, the cuurrent densitty
diistribution at
a the thicknness directioon tends to be uniform m in the casee of (±89.955)
sttacking sequuence.

FIIG. 4.12. Diimensionless current dennsity as a fuunction of dimensionless thickness fo

for
diifferent stacking sequences (±θ).

Figure 4.13
4 presentts the dimennsionless currrent densityy distributionn through thhe
thhickness of the 3D dom main for diffferent stackking sequencces and thiccknesses, Figg.
4.13(a) (±15)), Fig. 4.13(bb) (±45), Figg. 4.13 (c) (±
±60). In eachh case when the thicknesss
of the layer isi very smalll, then the current density distribuution tends to t be uniform m
thhrough the thhickness.
In ordeer to validatte the numeerical resultss, the analyytical solutioon of the 2D
domain probllem was com mpared to thee numerical solution for two differennt thicknessees
(00.5mm, 0.1225mm) at thee region wheere the maxim mum deviatiion appears, Fig. 4.14(a)).
One maay observe that the num merical soluution of thee electric pootential at thhe
innterlaminar region coinncides with the analyticcal solution. At the edg ge of the 3D
domain the error
e is maxximum and tends
t to beccome zero inn the case ofo a very thiin
laayer, Fig. 4.1
14(b).
84 Chhapter 4
 

FIG. 4.133. Dimensionnless current density as a function off dimensionleess thicknesses for
different sstacking sequuences (a) ±155, (b) ±45, (c)) ±60.

 
FIG. 4.144. Numerical solution of thhe 3D probleem at the worrst case and aat the middle of the
specimen.. Comparisonn to the analyttical solution.
E
Equivalent EC tensor
t for multtidirectional layyered CF mediia 8
85
 

4
4.3.2. Circuular domaiins – Diricchlet bounddary condittions
Two layers with EC tensors off σ45 and σ-45 respecctively, prod
T duce a new
m
multidirection nal material of (±45) with
w EC tensor of σ±45. IIf the layerss’ thickness is
hiigh, we can observe fluuctuations off the electricc potential fiield, Fig. 4.1
15(a) and (bb).
A
Also in the circular dom mains, the electric
e poteential gradieent through the thicknesss
teends to vanissh, Fig. 4.15(d).
Consideering that thee new materrial σ±45 can be expressedd by a scalarr quantity annd
not a tensor, we can calculate the eleectric field distribution
d iin polar coorrdinates usinng
thhe Eq. 4.17.

 r / Rmax 
  r   o ln
l   (4.177)
 Rin / Rmax 

 
FIIG. 4.15. Equuipotential suurfaces for cirrcular domainns and Dirichhlet boundary conditions fo
for
diifferent ratioss (μ) and thickknesses. (K=eexternal radiuus / layer thickkness = R/h)
86 Chhapter 4
 

Thee previouslyy mentionedd analytical solution is compared to the resppective

numericaal solution inn 3D space,, for variouss thickness llayers and vvarious ratios (μ =
Rmax / Rin) where (μ)
( is the exxternal to innternal radiius of the domain.
d Fig. 4.16
demonstrrates the electric field diistribution as a function of radius (rr), for a ratioo of (λ
= 8) and for various ply
p thickness.

FIG. 4.166. (a) Dimenssionless electtric potential field for cirrcular domainns as a functtion of
layers thicckness, (b) absolute
a errorr between thee numerical ssolution (3D space) and thhe (2D
space) anaalytical solutiion.

It can
c be obserrved that, ass the thickneess reduces, the numericcal solution at the
middle oof the laminnates approxximates the analytical solution. It iis obvious inn Fig.
4.16(b) thhat the errorr between thhe two curvees for an inteernal radius ((Rin) tends to zero
for high rratios (μ = Rmax / h, wheere (μ) is the external raddius to the layyer thicknesss).
Figgs. 4.17 and 4.18 demonnstrate the ellectric potenntial in dimennsionless forrm, as
a function of radius (r)
( for a layeer thickness ofo (h = 0.05m mm), ratios (μ( = 8, 16, 24,
2 32)
and for a layer thicknness of (h = 0.125mm), ratios of (μ μ = 8, 20, 60 ,80) respecttively.
These diagrams depiict the electrric potentiall at the regioon where maaximum dev viation
from the analytical solution
s cann be observeed. The curvves coincidee for ratios μ > 8.
Thereforee Eq. 4.8 is valid when the layer thhickness is smalls and it is safe to saay that
the error tends to zero.
Thee absolute error betweeen the num merical andd the analyttical solutio on for
different thicknesses, is presentedd in Fig. 4.19 and Fig. 44.20
E
Equivalent EC tensor
t for multtidirectional layyered CF mediia 8
87
 

FIIG. 4.17. Com mparison betwween numerical and analyytical solutionns as a function of ratio μ =
Rmax/Rin for layyer thickness h = 0.05mm.

FIIG. 4.18. Com mparison betwween numerical and analyytical solutionns as a function of ratio μ =
Rmax/Rin for layyer thickness h = 0.125mm
m.
88 Chhapter 4
 

FIG. 4.19. Absolute errror between the

t analyticall and numericcal solution foor different raatios μ.
a) h = 0.125mm, b) h=0.05mm.

It can be observed thhat, for a certtain small arrea the error is significannt. In that areea, we
expect a variation in the temperaature field.
i) For thhe case wheere the thickkness is neglligible comppared to the other dimennsions,
the 3D prroblem can be b simplifiedd to a 2D prooblem, and tthus solved.
ii) The electric
e fieldd distributionn tends to beecome uniforrm along thee thickness but b not
along maaterials edgees, in the caase of the second type of boundaryy conditionss, Fig.
4.11(c).
EC and thermal properties as a function of temperature 89
 

5. start
Chapter 5
EC and thermal properties as a function of
temperature

In order to perform this work, the behavior of CF electrical conductivity, specific heat
capacity and conductivity thermal coefficient with respect to temperature changes had
to be evaluated. Understanding the behavior of these properties resulted in accurate
prediction and control of the CF-preform temperature. The EC of the porous CF
medium (preform) and the CFRP medium as a function of temperature and fiber
direction were studied. Having taken into consideration the aforementioned
measurements, a modified equivalent EC tensor was extracted.
The thermal conductivity and the specific heat capacity were measured. The
thermal conductivity was assumed constant and the specific heat capacity as a
function of temperature was measured, using Differential Scanning Calorimetry
method (DSC). The extracted values of the DSC were compared to the literature.
Finally, a transient 1D model was developed and compared to experimental
measurements in order to validate all the measured coefficients (c, ρ, a1, k, ε). The
accumulated knowledge can be utilized in applications requiring accurate temperature
control.
90 Chapter 5
 

5.1. Experimental apparatus and measurement consistency for

the unidirectional CF preform
The testing apparatus described in Fig. 5.1 was designed and manufactured according
to the rules dictated by the ASTM C835-95 Standard (ASTM (1999), Taylor &
Rimbrough (1970)). Two different PAN based CF tows, each having a different
number of fiber, were tested. The first CF tow was the (HTA40J/E-3K-E13 / E13 type
with ca. 1.3%, sizing based on epoxy resin) consisting of 3000 fibers and the second
one was the (T700S-12K-60E / sizing type and amount 60E 0.3%) consisting of
12000 fibers. Both types of fibers feature the same volume resistivity equal to 1.6x10-
5 Ωm at room temperature (Torayca (2005), Toho Tenax (2000)). The length of the
tested tows was 0.1m. All fibers in a tow should be in contact with the copper
electrodes to ensure a smooth distribution of the current and to avoid errors in
measurements. To this end, the tow edges were infused with a small amount of silver
paint. Subsequently, using Sn99.5/Cu0.5/Co soldering material, the copper electrodes
were locally soldered. Voltage was applied using a stabilized DC power supply, able
to regulate either voltage or current. Vacuum was applied using a vacuum pump and
measured with a vacuum gauge. A thermocouple was placed in the middle and at the
centre of the CF tow. The measurements were stored in a PC via a USB data receiver.
Every test was performed with constant voltage input, having different amplitude. All
measurements were obtained once the steady state conditions were reached, assuming
that no temperature gradient is present through the tow thickness.
After every measurement session the tow was left to cool to room temperature
and its resistance was measured again. This step was used as a verification
measurement ensuring that no chemical reaction took place during the test, which
could alter the tow composition, and that all fibers were still in contact with the
copper electrodes. The circuit’s net resistance was measured using a multimeter
(Kithley 2002) and the value was subtracted from the measurements. Details for the
manufactured apparatus can be found in (Athanasopoulos et. al (2012)c).
This calorimetric test method covers the determination of total hemispherical
emittance of metal surfaces, graphite surfaces and coated metal surfaces in the range
of 300K to 1700K. According to the standard the size of the thermocouple wire
should be the minimum possible, suggesting that wire diameter less than 0.13mm
provides acceptable results. In this work because of the specimen size, the 0.13mm
E and thermal properties as a function of teemperature
EC 9
91
 

thhermocouplee wires induced substanttial error. Ussing K-type thermocoupples with wirre
diiameter of 0.13mm
0 resuults to recorrded temperrature of almmost 40oC loower than thhe
reecorded tem
mperature reesulted by the use of J-type therrmocouples having wirre
diiameter 25μm. The last ones
o were evventually used for all thee measuremeents.

FIIG. 5.1. (a) Manufactureed apparatus resembling bblack body eenvironment, (b) apparatuus
scchematic as described
d in thhe (ASTM Staandard (1999), Athanasopooulos et. al (22012)c).

This discrepaancy of the results

T r proviided by the use of K-tyype thermocoouples occuurs
because the thhermocouplee’s size is coomparable too the carbonn fiber tow size (thicknesss
annd width) reesulting to a significantt thermal grradient at the contact arrea leading to t
errroneous meeasurements. Electrical insulation of o the therm mocouples iss also cruciaal,
siince a smaall current leak l can allter the meeasurement significantlyy. Failure oof
innsulation caan be deteccted as spikkes in the measured ssignal and non-repetitiv
n ve
m
measurement ts. Another indication of o insulation failure is byb the fact that
t when thhe
cuurrent of thee CF-tow is reversed, thhe spikes meentioned prevviously havee the opposite
diirection. Eleectrical insullation of thee thermocoupple tip can bbe achieved by using low
viiscosity and high thermaal stability liiquids that solidify in rooom temperaature or withh a
sllight temperrature rise, for examplee liquid releease agents used in com mposite parrts
m
manufacturin ng. During thet tests, thee measured signal was smooth, sppike-free, annd
sttable in the steady
s a repeatabbility was achieved.
state and
92 Chapter 5
 

The crucial parameter that influences measurement repeatability/reliability is the

way the tow fibers are connected to the copper electrodes. Verification of the
electrode - tow connections together with the validity check of the assumption that all
fibers conduct current, was performed by measuring the resistance of every tow and
comparing them against the theoretical ones. Assuming that the tow is intact (all
fibers are bridging the electrodes) its theoretical resistance at room temperature is
given by Eq. 5.1.
L 4L
R (298K)  (298K)  (298K) (5.1)
A n f df2
In Table 5.1 the measured mean values and standard deviation of the tow
resistance are presented. It can be observed that the relative error is less than 3%, thus
supporting the reliability of the adopted methodology.

CF Tow Theoretical Experimental Relative Geometrical

Value R(Ω) Value R(Ω) ± S.D Error Characteristics
(L=0.1m) (L=0.1m) δ(%) of a tow

100mm x 2.5mm
HTA40J/E-3K-E13 13.500 13.435  0.052 0.5 x 0.27mm
100mm x 6mm x
T700S-12K-60E 3.465 3.563  0.091 2.83 0.5mm
TABLE 5.1. Carbon fiber tow VR at room temperature and geometric details.
(Athanasopoulos et. al (2012)c).

5.2. EC measurements as a function of temperature

Porous CF medium: The value of the volume resistivity with respect to temperature
was calculated from the voltage and current measurements at different tow
temperatures. As shown in Fig. 5.2 there is a linear correlation between volume
resistivity and temperature. Using the following linear relation, Eq. 5.2.

el ( T)  elo 1 a  T  To  (5.2)

 
The temperature resistivity coefficient calculated from the experimental results for the
two types of CF tow were a1 = aHTA40= -0.000308 K-1 and a1 = aT700S= -0.000247 K-1.
As a comparison the amorphous carbon temperature resistivity coefficient is a=-
0.0005 K-1 (Jewett & Serway (2008)). The negative value of this coefficient is a
characteristic of semiconductors. The coefficient of thermal expansion (CTE) of the
E and thermal properties as a function of teemperature
EC 9
93
 

toows are -0.1x10-6 K-1 and -0.38xx10-6 K-1 foor HTA40 aand T700S respectivelyy.
C
Considering these low values, thee length chhange (shrinnkage) of th
he tows waas
neglected.

FIG. 5.2. Voluume resistivityy as a functioon of temperaature at fiber direction

F d (0o) for the T7000S
-112K and HTAA40-3K carboon fiber tows (Athanasopouulos et. al (20012)c).

The maaximum voluume resistivity change iin the investtigated tempperature rangge
w approxim
was mately 10%, indicating thatt this effeect cannot bee neglected and
a should be b
coonsidered duuring calculaations. As already
a menttioned, the whole
w test was
w performeed
unnder vacuum m atmospheeric conditioons to preveent fiber oxxidation. Mo oreover, afteer
evvery heatingg / cooling cycle
c the tow w resistance remained unchanged,
u inndicating that
toow integrity was maintaiined and thatt all the initiial fibers werre intact.
In the case
c of constant volumee, the electriccal resistancce transversee to the fibers
(σσ2) as a fu unction of temperature
t is presenteed in Fig. 5.3. The values v of thhe
-1
teemperature resistivity
r cooefficient traansverse to thhe fibers aree a2 = -0.001131 K for 220
-1
laayers and a2 = -0.00152 K for 30 layers.
l The temperature
t coefficientss transverse tot
thhe fibers seeem to have a weak relattion to the number
n of laayers (or the thickness) oof
thhe CF preforrm.
94 Chhapter 5
 

FIG. 5.3.V 0o) for

Volume resisttivity as a funnction of tempperature transsverse to fiberr direction (90
the T700SS -12K under pressure of 0.1MPa.
0

CFRP medium:
m acccording to Hahn et. all the tempeerature resisttivity coeffiicients
calculated from the experimentaal results forr an epoxy CFRP
C materrial (HM K13C2U)
-1 -1
were a1 = -0.000528K and a2 = -0.00138
- K .
must be noteed that the body assum
It m med to be thhermally thinn (no tempeerature
gradient tthrough the thickness off the body).

Throughh the
Matterial 0o (x10-3) 90o (x10
( -3
)
thickness
-0.247 (T7700SC) -1.31 (20L
L - (T700SC))
CF prreform -
-0.308 (H
HTA40) -1.52 (30L
L -(T700SC))

CF
FRP -0.528 (HM K13C2U) -1.38 (8L) -0.941 (8L)
(

TABLE 5.2. Thermaal coefficientss of the VR

R for CFRPss and CF prreforms alonng and
transversee to the fiber direction.

Thee temperaturre resistivityy coefficient transverse tot the fibers, is approxim mately
four timees higher thaan the resistiivity coefficient parallell to the fiberrs. Also the values
v
EC and thermal properties as a function of temperature 95
 

between the CFRP and CF preform media, transverse and parallel to the fibers’
direction, are very close.

5.3. Derivation of the equivalent EC tensor as a function of

temperature
The EC tensor of the multilayered medium can also be expressed as a function of
temperature. In contrast to metals, the CF electric conductivity rises, as the material’s
temperature rises. As far as the thermal coefficient is concerned, the value at fibre
direction (0o) is different from the respective at the transverse direction ( a0  a90 ).
The EC along and transverse to the fibres' direction can be expressed by Eq. 5.3.

1Τ  1o [1  a1 ( Τ  Τ o )]
(5.3)
2Τ  2o [1  a2 ( Τ  Τ o )]
where  1o ,  2o are the principal conductivities at room temperature and 1Τ ,  2Τ are
the principal conductivities at temperature (T). Hence the EC tensor as a function of
temperature for a unidirectional layer follows the Eq. 5.4.
Τ

 1 m12  2 m22 (2  1 ) m1 m2 

Τn

ρ   (5.4)
Τ

(2  1 ) m1 m2 1 m22  2 m12  n

where m1 = cosθ and m2 = sinθ.

Introducing Eq. 5.3 to Eq. 5.4 the EC tensor for each layer (n) can be expressed
as a function of the temperature coefficient and the principal conductivities at room
temperature, Eq. 5.5.

m m2   1 0 a1 0   ρ1o 0   m1 m2 

Τn

ρ  1    T  T 
o   0 ρo  m2

m2 m1  n  0 1  0 a2  n  2 n
m1  n (5.5)

 
Τn

1
 σ T  ρ   
1
  mn  I    a n ( T  To ) ρn  mn 
T
n

 
σn

T
Finally introducing the EC tensor of each layer of the medium ( ) in Eq. 4.8, the
EC tensor of the multidirectional medium can be expressed by the following equation,
Eq. 5.6.
96 Chapter 5
 

 S  m   I    a  
N N
1 1 1
σT   S nσ T  σ T  
( T  To ) ρ n  m n
T
(5.6)

n
n n n
S n 1 S n 1

In the case where the thickness of each layer is the same, the equivalent EC tensor as
a function of temperature and thermal coefficients is given by Eq. 5.7,

  m   I    a  
N N
1 1 1
σT  σ T  σ T  
( T  To ) ρ n  m n
T
(5.7)
n

n n
N n 1 N n 1

where  m n denotes the directional cosines and  mn is the transpose matrix of the
T

directional cosines,  a n is the matrix of the thermal coefficients and  n is the VR

tensor of each layer at principal directions.
i.e. for the calculation of the equivalent EC tensor at T = 230oC of the (0/30)s laminate
with thermal coefficients (a1 = - 0.3 10-3 1/K, a2 = - 1.3 10-3 1/K) the EC tensor at 0o
fiber direction is given by the Eq. 5.8.

cos 0  sin 0  1 0 0.3 0  

σ 1T         230  30 
 sin 0 cos 0   0 1  0 1.3 
1
1/ 44287 0   cos 0 sin 0  
 0  
 1/ 50  sin 0 cos 0 
47113.8 0 
 σ 1T   S/m
 0 67.6 (5.8)

Τhe EC tensor at 30o fiber direction is given by the Eq. 5.9.

cos30  sin 30  1 0 0.3 0  

σ 2T         230  30 
 sin 30 cos30   0 1  0 1.3 
1
1/ 44287 0   cos30 sin 30  
 0  
 1/ 50  sin 30 cos30 
35352.2 20371.6
 σ 2T   S /m
20371.6 11829.1  (5.9)

Hence, the equivalent EC tensor at T = 230oC can be calculated by Eq. 5.10.

E and thermal properties as a function of teemperature
EC 9
97
 

N
1 1
σ T(0 /30 ) 
s
N
 σ nT 
4
 2σ 1T + 2σ 2T  
n 1

 412333 10185.8 
 σ T( 0 /30
/ )s   S /m (5.10))
101855.8 5948.44 

5
5.4. Theermal prooperties ass a functioon of tem
mperature
5
5.4.1. Speccific heat capacity
c m
measuremen
nts
The variationn of specificc heat capaccity versus teemperature was measurred with DS
T SC
teests on the fibers
f underr investigatioon. The procedure was the one desscribed in thhe
opperation maanual of thee DSC appaaratus (DuPond (1989))). At first the t aluminum m
saample pans were run em mpty and thhe heat flow w was measuured. This curve
c was thhe
reeference curvve. Then thee sample tin was filled with
w a small aamount (appproximately 66-
10mg) of carbbon fibers, it was sealedd with a lid and
a put insidde the chambber. The sam me
ass before heaating profile was appliedd and the heaat flow was measured. The
T differencce
between the twot heat flow curves (reeference emmpty pans andd fiber filled
d), along witth
thhe calibratioon coefficient of the apparatus,
a w
were used for
fo the deriv vation of thhe
reelationship of
o specific heeat capacity versus tempperature.

FIG. 5.4. Speccific heat capaacity Cp as a function of teemperature foor the T700S-12K, HTA400-
3K
K carbon fibeer tows (Athaanasopoulos et. al (2012)c), and a comm mon CFRP maaterial.
98 Chapter 5
 

The results indicate that a linear relation exist between specific heat capacity and
temperature, as shown in Fig. 5.4. The experimental results are in good agreement
with the reference (Pradere et. al (2005)).

5.4.2.Thermal conductivity measurements

For the CF layered material the following thermal properties were measured; thermal
effusivity, thermal conductivity and specific thermal capacity, Table 5.3. For that
reason, samples representative for each material were prepared. The thermal
properties were measured using the Mathis TCi device. This measurement procedure
is a non-destructive one, and the samples remain intact and can be used again.
The thermal conductivity was assumed constant and equal to the value at room
temperature. Pradere et. al (2008) have proved that the thermal conductivity changes
at high temperatures ( > 750K).

Conductivity
e(J/m2s1/2K)

Cp (J/kgK)
Effusivity

k(W/mK)
Direction

Capacity
Material

ρ(kg/m3)

Sp. Heat
Density

k22, k33 90o 1490.00 1012.70 0.64 1075.50

UD CFRP
k11 0o 1490.00 2560.00 4.09 1075.50

TABLE 5.3. Thermal properties of the molds materials (Athanasopoulos et. al (2012)c).

5.5. Validation of measurements

A finite difference scheme was developed, in order to predict and quantify the
transient phenomenon and is verified against experimental results. In order to exploit
in-depth the phenomenon and achieve very good agreement between numerical and
experimental results, temperature dependence of thermal and electrical properties
(specific heat capacity and volume resistivity) was examined under controlled vacuum
environment, ensuring no fiber oxidation. Numerical predictions and experimental
results were in excellent agreement.
E and thermal properties as a function of teemperature
EC 9
99
 

Assuming thermal conductivity

c y of the carboon fibers to bbe constant and
a due to thhe
abbsence of surrounding
s air (or anyy fluid in general
g sincee the presennt study waas
performed under
u vacuuum conditions), heat is transferrred only by radiatioon
(AAthanasopou ulos et. al (2012)b). As a result the governing
g paartial differential equatioon
iss expressed by
b Eq. (5.11).

σ
T 2T J2 A
cp (T)
t
 k 2  el  rad (T)
x f (T) V
T 4
 T4  

T 2T Arad
cp (T)  k 2  elf (T))J 2  (T)  T 4  T4  (5.11))
t x df2
nf L
4

FIIG. 5.5.Tempperature respoonse of the T7700S-12K tow w to the applieed voltage pu

ulses at the tow
w
ceenter. Heatingg and coolingg cycles (Athaanasopoulos eet. al (2012)c)).

An expllicit numericcal scheme with

w 200 noddes was developed, com mbining all thhe
afforementionned experimmentally meaasured therm mal and eleectrical paraameters. Thhis
soolution schem
me was usedd for the inveestigated CFF tows.
T results of
The o temperatuure versus time
t at the central point of the fiiber tows arre
prresented in Fig.
F 5.5 for the
t 12K tow, and are com mpared againnst the experimental datta.
T numericaal results agrree very welll with the exxperimental values.
The v
100 Chapter 5
 

Prediction and control of the Joule heating response of CFs, requires knowledge
of the thermal and electrical properties as a function of temperature, since these
properties influence the phenomenon in great extent.
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 101
and multidirectional CF media

6. start
Chapter 6
Numerical analysis and experimental validation of
the electrothermal problem: unidirectional and
multidirectional CF media

The EC of continuous CF layers is highly anisotropic and is expressed by a symmetric

second order tensor. Using the continuity equation for anisotropic media, the EC
tensor of a CF multilayer medium can be predicted by Eq. 4.8. Hence, the EC tensor
of the CF medium can be calculated for any stacking sequence. By means of the
calculated EC tensor of the multilayered preform, the elliptical form of the governing
equation can be solved numerically. Based on this, the generated heat can be
determined. It is well known that Joule’s first-law dictates that when electric current
passes through a conductor, heat is generated. The latter is also known as Joule effect
(resistive heating or ohmic heating). The determination of the EC of the multilayered
medium is the most crucial factor for the calculation of the generated heat and the
prediction of the temperature field. The generated heat in a known domain (Ω), with
given boundary conditions, changes according to the EC tensor of the multilayered
medium. Different stacking sequences lead to different EC tensors. Thus, the resulting
102 Chapter 6

generated heat in the known domain is also different. Introducing the generated heat
into the heat transfer equation, the temperature field over the CF medium can be
predicted. For the experimental verification, a thermal camera was used to record the
temperature field developed on a CF multilayer preform under given electric potential
field. The experimental results were compared to the respective numerical
calculations of the temperature field, where the EC tensor was analytically calculated
based on the proposed methodology. The results were confirmed experimentally via
thermal camera images, for a large number of representative stacking sequences.
The basic assumptions made for the current work are the following: the
multidirectional preform is a 2D body (thermal and electric potential gradient through
thickness, were not taken into consideration). Apart from the study of the
electrothermal problem, the validity of the equivalent EC tensor was also proved
(indirect validation of the EC tensor). As far as the aforementioned indirect method is
concerned, the analytical calculations of the EC tensor have been provided and based
on these, the coupled thermo - electric problem was solved numerically and
concluded to the calculated temperature field in multidirectional laminates. These
results were confirmed experimentally via the thermal camera images for a plethora of
representative stacking sequences. In chapter 6.2 the material was considered as a
non-homogeneous material. The equivalent EC tensor can be expressed as a function
of temperature, Chapter 5, Eq. 5.7. The EC is different from point to point σ(x1,x2)
and the PDE is the governing equation, Eq. 2.17.
In both cases (electrical homogeneous and heterogeneous medium) each
problem (electrical or electrothermal) was solved using a FDM scheme. In order to
validate the accuracy of the numerical scheme the results were compared to a
commercial FEM software (COMSOL).

6.1 Electrical homogeneous media

The multilayered conductive preforms are categorized in unidirectional laminates and
multidirectional laminates (Datoo (1991)). Regarding the direction of the fibres,
carbon fibre layered structures are divided into symmetric, anti-symmetric and
asymmetric laminates. These laminates are categorized into SOPL and MOPL, Fig.
6.1. In parallel to the above, appropriate numerical models were developed in order to
solve the electrical problem of an electrically anisotropic medium, Eq. 4.8 using the
finite difference method in order to calculate the electric field and the current density
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 103
and multidirectional CF media

distribution. The thermoelectric effect being omitted, the generated heat per unit time
and volume is the dot product of the electric field and the current density ( E  J ).
Since the generated heat is known, the solution of the heat transfer problem provides
us with the temperature field developed in the investigated samples. The solution of
the electrical problem, and consequently the developed temperature field are directly
connected to the EC tensor of the multidirectional laminate. Different stacking
sequences lead to different EC tensors.

FIG 6.1. Configuration types and numerical examples of CF multilayer media (Datoo (1991)).

The application of an electric potential difference between the edges of an

electrically conductive material leads, as it has already been discussed above, to a
temperature rise due to the Joule heating effect. Both the electric field and the
temperature distribution developed in the case of a conductive laminate, are seriously
affected by the material’s microstructure. The non-uniformity of the developed
temperature profile (concerning a reference one for each specific case of laminates)
can be considered as a measure of the conductive laminate's non - homogeneity. The
current study is focused on the approximation of the ideal case, where the resin-rich
layer is almost diminished, while the fibre volume fraction is assumed homogenous
and macroscopically constant.
104 Chapter 6

Electrical model: as a first step the equivalent EC tensor was calculated as a

function of fiber volume fraction, thickness and temperature for various stacking
sequences. A numerical algorithm was developed to solve the elliptic PDE Eq. 2.14
for various stacking sequences in rectangular domains, Fig. 6.2.
Further to the governing equation that defines the mathematical problem, the
formulation of the physical problem and its solution requests also the appropriate set
of boundary conditions. Two kinds of boundary conditions can be identified.
Electrical insulated regions (Γs) and a first kind boundary condition on the known
electric potential imposed on (Γd) regions.
The known electric potential has been applied between the boundaries Γd as
shown in Fig. 6.2, and given by Eq. 6.1(a) and (b).
(6.1a)
f at x1  0
0 at x1  L (6.1b)

The Γs boundaries are electrically insulated and there is no current flow in the
normal to the boundary direction (Jn = 0), where (n) is the unit vector which is normal
to the boundary. The boundary conditions in this case are expressed by Eq. 6.2.

   12   (6.2a)
11  11   0 at x2  0
n '  x1 11 x 2 
   12  
11  11   0 at x2  W (6.2b)
n '  x1 11 x 2 

The problem solved assumes a rectangular domain with aspect ratio (λ = L/W).
θ denotes the angle between the horizontal direction and the principal direction of
each layer.
A finite difference scheme was created in MATLAB environment for a specific
rectangular domain and given boundary conditions. An iterative method (Gauss -
Seidel) was used for the convergence of the solution of Eq. 2.14.
Solving the problem numerically, results have been extracted for the electric
potential field, electric field and the current density. Furthermore, using the results of
current density and electric field as inputs, the function of generated heat is
concluded. Finally, using the generated heat as input in the heat transfer equation, the
resulted temperature field can be predicted.
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 105
and multidirectional CF media

FIG. 6.2. Rectangular domain Ω and boundary conditions.

6.1.1 Numerical solution of the electrical problem using FDM and

validation using FEM
A second order accuracy finite difference scheme (FDM) was programmed using
MATLAB in order to calculate the electrical and the electrothermal problem for
electrical and thermal anisotropic materials with high level of EC anisotropy. Gauss-
Seidel iteration method was used in order to solve the problem. The solution and the
accuracy of the FDM were compared and validated with finite element method (FEM)
of COMSOL software. Also, in this chapter we present the difficulties which
appeared because of the high anisotropy of the material. Especially in UD materials
where the anisotropy of the material is too high (i.e. +30, +45). As it will be discussed
later in this chapter, a second order accuracy FDM scheme is capable to calculate with
very good approximation the electric potential field. In the case of very high
anisotropic levels (σ11 / σ22 > 5000 -10000) this scheme is unsuitable. FEM and FDM
are not capable to predict with high accuracy sharp and large electrical gradients.
Because of the high anisotropy the number of iterations increased (iterations >
30000) and the solution time is high. This problem appears when the number of nodes
is high. In the case where the number of nodes is small, the solution is fast the
106 Chapter 6

accuracy is decreased. For this reason a multistep algorithm was developed in order to
increase the accuracy of the solution and reduce the number of iterations.
An initial guess was used (φ(x1,  x2) = 0) for a coarse mesh for the first step.
After the 1st solution of the electric potential field, the element size was reduced to
half, having taken into consideration the previous solution as an initial guess. After
the 2nd solution of the electric potential field, the element size reduced to half the
previous having taken into consideration the previous solution, as an initial guess.
This is repeated continuously until the desired accuracy.
i.e. in the case of a domain 800mm x 200mm: 1st step (126 nodes), 2nd step (451
nodes), 3rd step (1701 nodes), 4th step (6601 nodes) , 5th step (26001 nodes) the total
solution time was approximately 366 sec, Fig. 6.3(b).

FIG. 6.3. (a) solution after 2100s for 1step algorithm (6601 nodes), (b) solution after 366.2s
for the multistep algorithm (26001 nodes).

In the case of one step algorithm, the number of nodes was 6601 and the
solution does not convergence after 2100 sec, Fig. 6.3(a). The explanation of low
speed convergence consists in the small numerical fraction at the FDM equation “we
can say that the flow of information from iteration to iteration transverse to the fiber
direction is low”. The solution is fast at the fiber direction (1st principal axis) but low
at the other principal axis.
In order to validate the solutions accuracy of the FDM scheme, the results were
compared with the FEM code solution provided by COMSOL. Two different cases
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 107
and multidirectional CF media

were studied: the first example was a high anisotropic material (454) and the second
example was a material with a stacking sequence of (0/30)s. The element dimensions
using FEM and FDM were the same. (type the element quadrilateral)
Figures 6.4 and 6.5 present the solution of the electric potential field at the right
and left edge of the orthogonal domain using the FEM and FDM for different element
sizes (51681 nodes and 13041 nodes).

FIG. 6.4. Electric potential field using FDM and FEM for different number of nodes. (a)
51681 nodes, (b) 13041 nodes, (c) 231 nodes, (d) 66 nodes.

For a fine mesh the, Fig. 6.4(a) and (b), Fig. 6.5(a) and (b) the two solutions are
in very good agreement. In contrast using a coarse mesh Fig. 6.4(c) and (d), Fig.
6.5(c) and (d) the solutions do not coincide (231 nodes and 66 nodes).
Figure 6.6 present the FDMs solutions as a function of element size and FEMs
solution as a function of elements size ((domain size = 400mm x 200mm), 1.25mm –
51681 nodes, 2.5mm – 13041 nodes, 20mm – 231 nodes, 40mm – 66 nodes).
108 Chapter 6

FIG. 6.5. Electric potential field using FDM and FEM for different number of nodes. (a)
51681 nodes, (b) 13041 nodes, (c) 231 nodes, (d) 66 nodes.

FIG. 6.6. Electric field using FEM and FDM at the edges of the specimen (a) 13041 nodes,
(b) 205761. [45o], L = 400mm, W = 200mm, σ11 = 22169 S/m, σ12 = 22119 S/m, σ22 = 22169
S/m), Δφ = 10V.

In the second case a (0/30)s stacking sequence was used in order to compare and
validate the FDM code. The equivalent EC tensor was calculated using the proposed
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 109
and multidirectional CF media

Eq. 4.8. The electric potential difference was again 10V at the edges of the orthogonal
domain. We observe that the FDM and FEM solutions coincide perfectlly, Fig. 6.8.

FIG. 6.7. Electric potential field using FDM and FEM for a CFRP material with stacking
sequence (0/30/30/0) at the edge of the domain.

In the case of finite element method, the electric field (Ex2) the solution is
discontinuous at the region near the corner, detail of Fig. 6.8(a). The electric field
solution (Ex2) in the case of FDM is smooth, Fig. 6.8(b).

FIG. 6.8. Electric field using FDM and FEM (a) FEM, (b) FDM. Detail of the error at the edge
of the domain.
110 Chapter 6

This problem appears in a very small region of the domain and it does not affect
the overall temperature field. Figures 6.9 and 6.10 present the electric filed (Ex2) at
the edges of the domain for the FEM and FDM.

FIG. 6.9. Electric field using FDM and FEM at the edges of the domain.

In the case where the EC anisotropy is too high (>5000-10000) both FDM and
FEM are unable to calculate with high accuracy the electric potential field. As we can
observe in Fig. 6.10, Fig. 6.11 the electric potential solution is inaccurate at the region
where the sharp solution appears.
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 111
and multidirectional CF media

FIG. 6.10. Electric potential field for a high anisotropic material using different number of
nodes.

FIG. 6.11. Electric potential and electric field using FEM at the edges of the specimen (a)
electric potential as a function of anisotropy level using 861 nodes, (b) electric potential as a
function of number of nodes -861, 13041, 207561 nodes-, (c) electric field as a function of
anisotropy level using 861 nodes, (d) electric field as a function of number of nodes -861,
13041, 207561 nodes-. [45o], L = 400mm, W = 200mm, Δφ = 10V.
112 Chapter 6

The values of the EC tensor for the above cases were:

Anisotropic Mater. 1: σ11 = 22169 S/m, σ12 = 22119 S/m, σ22 = 22169 S/m.
Anisotropic Mater. 2: σ11 = 22169 S/m, σ12 = 22168 S/m, σ22 = 22169 S/m).
Anisotropic Mater. 3: σ11 = 22169 S/m, σ12 = 22118.8 S/m, σ22 = 22169 S/m).

Thermal model: after the calculation of the generated heat (Q(x1,x2)) the
transient heat transfer equation was used to predict the temperature field for the
aforementioned stacking sequences. The specific FDM explicit code solves 2D
electrothermal problems.
Referring to the heat transfer problem, a second kind boundary condition
(Neumann boundary condition) that will be imposed on the thermal insulated regions
and a first kind boundary condition on the known temperature regions (Dirichlet
boundary condition) imposed on (Γd) regions.
Considering a medium electrically and thermally homogeneous the EC and the
thermal conductivity are independent of temperature. The specific heat capacity of the
material has a linear relation to temperature, Fig. 5.4.
Heat losses due to natural convection, can be incorporated into the code. In order to
calculate the heat transfer coefficient the following physical constants must be
calculated: film temperature Τf and the thermal properties of air in the specific
temperature: thermal conductivity of air (kair), Prandtl number (Pr), Rayleigh number
(Ra), kinematic viscosity (ν), volumetric coefficient (β), characteristic length (δ),
Nusselt number (Nu).
The following calculations are valid at steady state conditions. If we assume that the
temperature filed does not rapidly change, the heat transfer coefficient can be
calculated for each time step using the following expressions.
The film temperature is the average temperature of the surface and environment
temperature, Eq. 6.3.
Tav  T
Tf  (6.3)
2
The thermal conductivity of the air can be expressed as a function of temperature
using the following polynomial, Eq. 6.4.

kair  4.641x1012 xTf3  2.902x108xTf2  9.044x105xTf  1.002x103 (6.4)

Numerical analysis and experimental validation of the electrothermal problem: unidirectional 113
and multidirectional CF media

Also, Prantl number and kinematic viscosity can be expressed using polynomial
fittings, Eq. 6.5., Eq. 6.6.

Pr  2.504x1020 xTf6  2.490x1016 xTf5  9.852x1013 xTf4 

1.998x109 xTf3  2.187x106 xTf2  1.168x103 xTf  (6.5)
1
9.288x10

v=4.928x1011xTf2  8.348x108 xTf1  1.457x105 (6.6)

The volumetric coefficient is

1
 (6.7)
Tf
and the characteristic length and Rayleigh number

 LW 
 (6.8)
 2L  2W 
g  Ts -T  3
Ra  Gr Pr  Pr (6.9)
2
Then the Nusselt number and the heat transfer coefficient for the upper and
lower surface of the thin body can they calculated.

Nu up  0.54Ra 0.25
(6.10)
Nu down  0.27Ra 0.25
The heat transfer coefficient for the upper surface is expressed by Eq. 6.11a and
the heat transfer coefficient for the opposite surface is expressed by Eq. 6.11b.
h up  Nu up  k air /   (6.11a)
h down  Nu down  k air /  
(6.11b)
h av 
h up +h down 

2 (6.11c)
Note that the surface temperature (T) of the body, can be assumed equal to the
body temperature, for the case of a thin body. Finally the specific heat capacity can be
calculated using Eqs. 6.12a and 6.12b.
114 Chapter 6

CF_ preform 
Cp( T )  2.847T 140.83 (6.12a)
CFRP 
Cp( T )  2.960T  32.15 (6.12b)

The following figure presents the structure of the FDM code: a) calculation of
the equivalent EC tensor and the thermal properties of the material, b) numerical
solution of the electrical problem, c) numerical solution of the transient heat transfer
problem with heat source, Fig. 6.12.

FIG. 6.12. Flow diagram of the finite difference algorithm.

i. Electrical conductivity along and transverse to the fibers according to Eq.

3.36, Eq. 3.49, Eq. 3.51.
ii. Stacking sequence
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 115
and multidirectional CF media

iii. Thermal conductivity of the material along and transverse to the fibers,
density, specific heat capacity, fiber volume fraction, room temperature an
initial temperature.
iv. calculation of the equivalent EC tensor.
v. initial and final number of nodes (elements)
vi. solution of the governing equation, Eq. 2.14 using the multistep algorithm and
convergence check for each step.
vii. Calculation of the electric field, current density and Joule effect take into
consideration the final solution of the governing equation.
viii. Determination of maximum time step and the duration of the phenomenon.
ix. Time step determination and stability check for each value of specific heat
capacity.
x. Solution of the transient heat transfer with heat source.

The measured temperature profile of each investigated laminate case was

compared to the respective numerical results derived from the solution of the
respective electrical / heat conduction problem. For easier interpretation, the results
will be presented in dimensionless form. More precisely, the spatial co-ordinates have
been expressed as x 1*  x 1 L and x *2  x 2 W . Furthermore, the electric field and the
electric potential are also presented in dimensionless form, following the
expressions E *x1  E x1 E max
x1
, E *x 2  E x 2 E max
x2
, and *    max . Finally, the generated heat
and temperature field are presented in dimensionless form too, following the
expressions * Q
Q  Q

max
and T *   T  Tenv   Tmax  Tenv  respectively, (Tenv ≈ 29oC).

6.1.2 Electrothermal analysis in porous CF media

The electric potential was applied to the defined porous CF medium and the resulted
temperature distribution is measured using a thermal camera. The comparison
between the calculated and the measured temperature profiles is the validation step for
the calculated EC tensor of the multidirectional CF preform. The material tested was
made of the T700SC UD CF layers with CF electric resistivity of ρf = 1.6 x 10-5 Ωm
(areal weight equal to 80 gr/m2, thickness (h) equal to 0.09mm at 0.1MPa applied
pressure). After a few seconds, using a thermal camera (Flir SC660) the temperature
116 Chapter 6

field image was captured for each case studied. The maximum temperature for all the
following results is 65oC (338K) with a deviation of ±5oC.
Figure 6.1 shows the typical grouping of the continuous fibre laminates, while
the last column provides the different lay ups studied within the frame of the present
work. The EC tensor for symmetric, anti-symmetric and asymmetric laminates was
calculated using Eq. 4.8. The values of the electrical conductivity at principal axes of
the unidirectional dry CF layer are σ1 = 31585 S/m and σ2 = 42 S/m.
The numerically resulted temperature field was compared against the
experimental results, which were monitored using the thermal camera. For the
numerical results, the streamlines of current density, the electrical potential field, the
electric field, the generated heat and the temperature distribution are presented, as
well as the temperature distribution monitored with the thermal camera. The indirect
experimental validation of the equivalent EC tensor is based on the Joule effect and
includes the use of a thermal camera for having a global, detailed comparison between
numerical and experimental results. For this step of validation, specimens of
orthogonal geometry, Fig. 6.2, were used again, with length to width ratio equal to (L
/ W = 3) and length to layer thickness ratio (L / h = 2444). Known electric potential
difference was applied at the edges of the samples. Due to the Joule effect, the
temperature of the samples increased, and the temperature field was recorded with a
thermal camera.

6.1.2.1 Symmetric laminates

According to Fig. 6.1 the symmetric layers are divided into unidirectional layered
media and multidirectional layered media. Unidirectional plies consist of one or more
layers with the same fibre direction. Multidirectional plies consist of two or more
layers with different fibre directions.

i. Unidirectional layered medium

Specially orthotropic: in the case of the special orthotropic carbon fibre preforms
(04 or 904), the generated heat is uniform because the off-diagonal terms of the EC
tensor are equal to zero. Although the calculated generated heat and temperature field
for both cases are uniform ( Q * (x1 , x 2 )  1 , T* (x1 , x2 )  1 ), Fig. 6.13(c) and (d), the
experimental results present small temperature deviations (ΔΤ ≈ ±2οC) because of the
slight local variation of the fibre volume fraction ( Timage
*
(x1 , x 2 )  1 ), Fig. 6.13(e) and (f).
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 117
and multidirectional CF media

FIG. 6.13. Numerical results of the symmetric unidirectional – specially orthotropic medium,
(04) or (904) in dimensionless form. (a) Current density, (b) electric potential field, (c),
generated heat, (d) temperature field, (e) measured temperature field (04), (f) measured
temperature field (904). λ = 3, L/H =611; (04): σ22/σ11 = 1.3 10-3, σ12/σ11 = 0; (904): σ22/σ11 =
666.7, σ12/σ11 = 0. (Athanasopoulos & Kostopoulos (2012)).

Generally orthotropic: in the case where all the layers have the same fibre
direction (for θ ≠ 0ο or θ ≠ 90ο), the medium is a unidirectional - generally orthotropic
(304), (454), (604). Because of the high anisotropy of the medium, the equipotential
lines tend to be directed along the fibres direction, Fig. 6.14 and 6.15.

FIG. 6.14. Numerical and experimental results of the symmetric unidirectional - generally
orthotropic medium for three different fiber directions, (304), (a) (304): σ22/σ11 = 0.335, σ12/σ11
= 0.576. (Athanasopoulos & Kostopoulos (2012)e).
118 Chapter 6

Apparently, the current density lines are not perpendicular to the equipotential lines,
as in the well-known case of isotropic medium. Particularly in this case, one may
notice that the current density in some regions of the medium is parallel to the
equipotential lines. This is a significant difference between the electric behavior of the
anisotropic bodies compared against the isotropic ones. This also explains the results
in Fig. 3.5 and Fig. 3.6. In Fig. 6.15, it is shown that the above discussed numerical
results and the numerically resulted temperature field are in excellent agreement with
the thermal camera temperature images.

FIG. 6.15. Numerical and experimental results of the symmetric unidirectional - generally
orthotropic medium for three different fiber directions, (454), (604) in dimensionless form. (λ
= 3, L/H = 611, σ1/σ2 = 666.7). (a) (454): σ22/σ11 = 1, σ12/σ11 = 0.997; (b) (604): σ22/σ11 =
2.988, σ12/σ11 = 1.722. (Athanasopoulos & Kostopoulos (2012)e).

ii. Multidirectional layered medium

Generally orthotropic: in the case of multidirectional - generally orthotropic
preforms, obviously the number of layers has to be N ≥ 4. In order to validate the
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 119
and multidirectional CF media

calculated EC tensor by the proposed methodology, four different stacking sequences

were selected and studied as well, namely: (0/30)s, (60/30)s, (45/30)s, Fig. 6.16
(90/30)s, Fig. 6.17.

FIG. 6.16. Numerical and experimental results of the symmetric multidirectional - generally
orthotropic medium in dimensionless form. λ = 3, L/H =611, σ1/σ2 = 766.7. (a) (0/30)s: σ22/σ11
= 0.144, σ12/σ11 = 0.247; (b) (45/30)s: σ22/σ11 = 0.601, σ12/σ11 = 0.744; (c) (60/30)s: σ22/σ11 = 1,
σ12/σ11 = 0.863. (Athanasopoulos & Kostopoulos (2012)e).
120 Chapter 6

In this case the calculated generated heat, hence the resulted temperature field, is not
uniform, Fig. 6.16 and Fig. 6.17. The spatial distribution of the generated heat in each
case is different and is directly related to the concluded EC tensor. Then, as it is
expected, the analytical calculation of the EC tensor leads to the generated heat
profiles.

FIG. 6.17. Numerical and experimental results of the symmetric multidirectional - generally
orthotropic medium in dimensionless form. λ = 3, L/H =611, σ1/σ2 = 766.7. (90/30)s: σ22/σ11 =
0.144, σ12/σ11 = 0.247.

Figure 6.16 and Fig. 6.17 shows that the calculated temperature profiles concluded by
using the analytical calculated EC tensors in the case of the studied stacking
sequences, are in excellent agreement with the experimentally captured thermal
camera temperature results.

Cross ply, angle ply, quasi isotropic: Considering now the cross ply, angle ply,
quasi isotropic layers, Fig. 6.18(a), (b, (c) and (d) present the numerical results in
dimensionless form for the current density, the electric potential field, the electric
field, the generated heat and the temperature field for all the aforementioned cases. In
the case of multidirectional - cross ply, the off-diagonal terms of EC tensor that
correspond to the multidirectional - cross ply laminates (0/90)s are equal to zero.
Hence, the calculated generated heat and the resulted temperature distribution is
uniform, Fig. 6.18(c) and (d).
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 121
and multidirectional CF media

FIG. 6.18. Numerical results of the anti-symmetric multidirectional – cross ply or angle ply or
quasi isotropic medium in dimensionless form. (a) Current density (b) electric potential, (c)
generated heat, (d) temperature field (90/0)s, (±30)s, (±45)s, (0/±60)s. λ = 3, σ1/σ2 = 766.7. (e)
Measured temperature field of (90/0)s: L/H =611, σ22/σ11 = 1, σ12/σ11 = 0, (f) measured
temperature field of (±30)s: L/H =611, σ22/σ11 = 0.334, σ12/σ11 = 0; (g) measured temperature
field of (±45)s: L/H =611, σ22/σ11 = 1, σ12/σ11 = 0; (h) measured temperature field of (0/±60)s:
L/H =407, σ22/σ11 = 1, σ12/σ11 = 0. (Athanasopoulos & Kostopoulos (2012)e).

The numerical results are validated experimentally by the thermal camera images for
the same stacking sequence as shown in Fig. 6.18(e). Any symmetric stacking
sequence of 0o and 90o plies, independently of the number of plies in each direction,
leads to the same result. All the above are valid, as it has already been noticed, in the
case where the medium is considered ‘thermally thin’ and no thermal gradient through
the thickness exists. In the case of multidirectional - angle ply, again the off-diagonal
terms of the conductivity tensor are vanished since  12   12  , as shown in Eq. 6.12.

1   11 12  12  1 11  11

     
0 
σ θ /-θ   11    
     (6.12)
2  21   21  22   22  2  0  22   22 

Figure 6.18(c) and (d) present again the results in dimensionless form of the
numerically calculated generated heat and temperature distribution
( Q * (x1 , x 2 )  1 , T* (x1 , x2 )  1 ) based on the EC tensor resulted for two angle ply laminates
(-30/30)s and (-45/45)s, and it is compared against the experimentally monitored
temperature distribution ( Timage
*
(x1 , x 2 )  1 ), Fig. 6.18(f) and (g). Finally, in the case of

multidirectional - quasi isotropic plies the off diagonal terms are equal to zero ( 
122 Chapter 6

quasi isotropic stacking sequence). In the case of MOPL quasi isotropic medium, the
angle of the typical nth layer follows the relation, Eq. 6.13.

n  180 n 1 N (6.13)

As a consequence the resulted calculated temperature profile is uniform, Fig. 6.18(c)

and (d). This is confirmed again by the by the thermal camera temperature results,
Fig. 6.18(h).

6.1.2.2 Anti-symmetric laminates

The anti-symmetric laminates are multidirectional plies that follow the stacking
sequences presented in Fig. 6.19. Two categories exist: cross plies and angle plies.
The calculated temperature field of the multidirectional - anti-symmetric cross plies is
the same as the respective of the symmetric multidirectional cross plies. In the case of
the multidirectional anti-symmetric angle plies [(θ1/θ2/.../θN)/(-θ1/-θ2/.../-θN)], the
calculated EC tensor leads to uniform temperature distribution
( Q * (x1 , x 2 )  1 , T* (x1 , x2 )  1 ) and this is also experimentally validated ( Timage
*
(x1 , x 2 )  1 ).

FIG. 6.19. Numerical results and thermal camera image of the asymmetric multidirectional
medium (45/60/-45/-60) in dimensionless form. λ = 3, L/H =611, σ1/σ2 = 766.7. (a) Current
density (b) electric potential, (c) generated heat, (d) temperature field, (e) measured
temperature field σ22/σ11 = 1.664, σ12/σ11 = 0. (Athanasopoulos & Kostopoulos (2012)e).
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 123
and multidirectional CF media

This is due to the fact that, again, the off-diagonal terms are equal to zero as shown in
Eq. 6.14.

 N ( i /   i ) N
  N ( i /   i ) 
  11
1   1
  12( /  ) 
i i
   11 0 
σ [(θi / -θi ),...,(θN / -θN )] 
 1
  1   1 
N  N ( i /   i ) N
 N N
( i /  i )  (6.14)
   21   22(i / i )   0   22 
  1  1    1 

The numerical and the experimental results concluded in the case of (45/60/-45/-
60) layered medium are presented in Fig. 6.19.

6.1.2.3 Asymmetric Laminates

An infinite number of combinations of UD plies may lead to asymmetric laminates.
Only few of them conclude to a EC tensor that finally results in a uniform temperature
profile. Some representative cases of this group of laminates will be studied within the
frame of the present work.
Direction verticality: In the case where the angle between two successive layers
is Δθ = 90o ( 1 ,  2 where 1   2 ) the values σ12, σ21 = 0 as shown in Eq. 6.15. As a
result the generated heat is uniform and the medium studied concludes to a uniform
temperature,
  90
1   11 12   21
 90
 1 11   22

0 
σ θ / θ+90   11  90   90      
 σ θ / θ-90
2  21   21  22   22  2  0  22   11 
(6.15)

Fig. 6.20(d) shows the calculated temperature distribution in the case of asymmetric
laminates (-30/60) and (-20/70), which matches perfectly with the experimentally
measured temperature distribution for the same multilayer medium, Fig. 6.20(e) and
(f). In general (assuming thin body approach), for any combination of pairs of layers
[(θ1/θ1+90)/ (θ2/θ2+90)/… (θΝ/θΝ+90)], the off-diagonal terms of the total
conductivity tensor will be equal to zero, Eq. 6.16.

 N (i /i  90) N

  N (i /i 90) 
  11
1   1
  ( i / i  90)
12  1    11 0 
σ (θ1 / θ1 +90),... 
 1
    1 
N  N (i /i  90)  N (6.16)
( i / i  90) 
N N

   21  
,(θN / θN +90)
 22(i /i 90)   0  22 
  1  1    1 
124 Chapter 6

FIG. 6.20. Numerical results of the asymmetric multidirectional medium (-30/60) or (-20/70).
(a) Current density (b) electric potential, (c) generated heat, (d) calculated temperature field. λ
= 3, L/H =1222, σ1/σ2 = 766.7. (e) thermal camera image (-30/60): σ22/σ11 = 1, σ12/σ11 = 0; (f)
thermal camera image (-20/70): σ22/σ11 = 1, σ12/σ11 = 0.

Ratio σ12/σ11 tends to zero: the final case studied in the present work is an
asymmetric stacking sequence of the form (-30/45). In this specific case, the angle
between the layers is not 90o. However, this 15o difference leads now to small values
of the off-diagonal terms ( 12 /11  0 ), which consequently provides small deviation in
the generated heat and results in an almost uniform temperature distribution as shown
in Fig. 6.21.

FIG. 6.21. Numerical results and thermal camera image of the asymmetric multidirectional
medium (-30/45) in dimensionless form. (λ = 3, L/H =1222, σ1/σ2 = 766.7). (-30/45): σ22/σ11 =
0.601, σ12/σ11 = 0.053. (Athanasopoulos & Kostopoulos (2012)e).
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 125
and multidirectional CF media

Laminates with more than three layers and uniform generated heat: for

asymmetric layers  1 /.../N where the sum of off diagonal terms can be equal to
N N
zero   12    21

 0 , where the number of layers are N > 2 with fibre directions
 1  1

which can be 1  2  ...   , Eq. (6.17).

 N  N
  N  
  11
1   1
  
12    11 0 
  /.../ 
 1
  1   1  (6.17)
N  N    N  
1 N
N N

   12   22   0   22 
  1  1    1 
In this case again the resulted temperature distribution is uniform due to the generated
heat of the body.

6.1.3 Electrothermal analysis in CFRP media

The material tested was made of the Sigrafill / Sigratex UD CF prepregs with
thickness (h) equal to 0.121mm at 0.8MPa applied pressure. After a few seconds,
using a thermal camera (Trotech IC80L) the temperature field image was captured for
each case studied.
The EC tensor for symmetric, anti-symmetric and asymmetric laminates was
calculated using Eq. 4.8. The values of the electrical conductivity at principal axes of
the unidirectional dry CF layer are σ1 = 44287 S/m and σ2 = 50 S/m. The numerically
resulted temperature field was compared against the experimental results monitored,
using the thermal camera. Again, for the numerical results, the streamlines of current
density, the electrical potential field, the electric field, the generated heat and the
temperature distribution are presented, as well as the temperature distribution
monitored by the thermal camera.

6.1.3.1 Symmetric laminates

i. Unidirectional layered medium
Specially orthotropic: in the case of the special orthotropic CFRPs (04 or 904),
the generated heat is uniform because the off-diagonal terms of the EC tensor are
equal to zero. Although the calculated generated heat and temperature field for both
cases are uniform ( Q * (x1 , x 2 )  1 , T* (x1 , x2 )  1 ), Fig. 6.22(c) and (d), the experimental
126 Chapter 6

results present small temperature deviations (ΔΤ ≈ ±3.5οC) because of the slight local
variation of the fibre volume fraction ( Timage
*
(x1 , x 2 )  1 ), Fig. 6.22(e) and (f).

FIG. 6.22. Numerical results of the symmetric unidirectional – specially orthotropic medium,
(04) or (904) in dimensionless form. (a) Current density, (b) electric potential field, (c),
generated heat, (d) temperature field, (e) measured temperature field (04), (f) measured
temperature field (904). λ = 3.13, L/H =420; (04): σ22/σ11 = 1.129 10-3, σ12/σ11 = 0; (904):
σ22/σ11 = 885.7, σ12/σ11 = 0.

Generally orthotropic: in the case where all the layers have the same fibre
direction (for θ ≠ 0ο or θ ≠ 90ο), the medium is a unidirectional - generally orthotropic
(304), (454), (604). Because of the high anisotropy of the medium, the equipotential
lines tend to be directed along the fibres direction, Fig. 6.23. Apparently, the current
density lines are not perpendicular to the equipotential lines, as in the well-known
case of isotropic medium. Particularly in this case, one may notice that the current
density in some regions of the medium is parallel to the equipotential lines. This is a
significant difference between the electric behavior of the anisotropic bodies
compared against the isotropic ones.
In Fig. 6.23, it is shown that the above discussed numerical and the numerically
resulted temperature field results one more time are in excellent agreement with the
thermal camera temperature images. The local deviation of temperature field at the
edges of the specimens appears because of the poor quality trimming of the sample.
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 127
and multidirectional CF media

FIG. 6.23. Numerical and experimental results of the symmetric unidirectional - generally
orthotropic medium for three different fiber directions, (304), (454) in dimensionless form. (λ
= 3.13, L/H = 420, σ1/σ2 = 885.7). (a) (304): σ22/σ11 = 0.334, σ12/σ11 = 0.576; (b) (454): σ22/σ11
= 1, σ12/σ11 = 0.997.

ii. Multidirectional layered medium

Generally orthotropic: in the case of generally orthotropic laminates, obviously
the number of layers has to be N ≥ 4. In this case by definition, the laminate has at
least two layers corresponding to two different fibre directions. In order to validate the
calculated equivalent EC tensor by the proposed methodology, three different
stacking sequences were selected and studied, namely: (0/30)s, (60/30)s, (45/30)s.
In this case the numerical results for the different laminates studied are presented
in the first six columns of Fig. 6.24 and 6.25. In this case the calculated generated heat
128 Chapter 6

and hence the resulted temperature field shown in the sixth of Fig. 6.24 and 6.25 is
not uniform.

FIG. 6.24. Numerical and experimental results of the symmetric generally orthotropic media
in dimensionless form. (a) (60/30)s: σ11 = 22169, σ12 = 19155, σ22 = 22169; (b) (45/30)s: σ11 =
27698, σ12 = σ21 = 20637, σ22 = 16639. (Athanasopoulos & Kostopoulos (2013)).

The spatial distribution of the generated heat in each case is different and is
directly related to the concluded equivalent EC tensor. The last column of Fig. 6.14
presents the experimental results of the temperature profile developed in the above
laminates, after the application of predetermined potential difference at the edges of
the laminates. As it is shown for all the studied stacking sequences, the experimental
results monitored with the thermal camera are in excellent agreement with the
respective numerical ones.
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 129
and multidirectional CF media

FIG. 6.25. Numerical and experimental results of the symmetric generally orthotropic media
in dimensionless form. (0/30)s: σ11 = 38757, σ12 = 9577.6, σ22 = 5579.6. (Athanasopoulos &
Kostopoulos (2013)).

Angle ply, quasi isotropic: Figures 6.26(a), (b), (c) and (d) present the numerical
results in dimensionless form for the current density, the electric potential field, the
electric field, the generated heat and the temperature field for the cases of (±30)s,
(±45)s and (±60)s. The concluded numerical results are valid in the case where the
medium is considered ‘thermally thin’ and no thermal gradient through the thickness
exists.

FIG. 6.26. (a), (b), (c), (d) Numerical results of the multidirectional anti-symmetric angle ply
or quasi isotropic medium in dimensionless form. (e) (±30)s: σ11 = 33228, σ12 = σ21 = 0, σ22 =
11109; (f) (±45)s: σ11 = σ22 = 22168.5, σ12 = σ21 = 0; (g) (±60)s: σ11 = 11109, σ12 = σ21 = 0, σ22
= 33228; (h) (0/±60)s: σ11 = σ22 = 22168.5, σ12 = σ21 = 0. (Athanasopoulos & Kostopoulos
(2013)).
130 Chapter 6

In the case of multidirectional angle ply laminates, the off diagonal terms of the
conductivity tensor are vanished since   12   12  . The calculated temperature field
of the angle ply laminates (±30)s and (±45)s, (±60)s is compared against the
experimentally monitored temperature profiles of each case, which are presented in
Figs 6.26(e),(f) and (g) respectively.
In the case of multidirectional quasi isotropic plies (0/±60)s the off diagonal
terms are equal to zero (  quasi isotropic stacking sequence). As a consequence the
resulted calculated temperature profile is uniform, Fig. 6.26(d). This is confirmed,
again, by the thermal camera temperature results, Fig. 6.26h).

6.1.3.2 Anti-symmetric
Considering the anti-symmetric laminates two categories exist: cross plies and angle
plies. a) In the case of anti-symmetric angle plies laminates [(θ1, θ2...θN)/(-θ1,-θ2,...-
θN)] the calculated EC tensor leads to uniform temperature distribution and this is also
experimentally validated. The above happens is due to the fact that again the off
diagonal terms are equal to zero. The numerical results concluded in the case of
(+45,+60,-45,-60) layered medium presented in Figs 6.27(a),(b),(c) and (d) and the
corresponding experimental results concerning temperature profile are presented in
Fig. 6.27(e).

FIG. 6.27. (a), (b), (c), (d) Numerical results and thermal camera image of the
multidirectional anti-symmetric and the asymmetric media in dimensionless form. (e) (45/60/-
45/-60): σ11 = 16639, σ12 = 0, σ22 = 27698; (f) (-30/60): σ11 = σ22 = 22168.5, σ12 = σ21 = 0.
(Athanasopoulos & Kostopoulos (2013)).
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 131
and multidirectional CF media

6.1.3.3 Asymmetric Laminates

Direction verticality: in the case where the angle between two adjacent plies is Δθ =
90o the generated heat is uniform and the medium studied concludes to a uniform
temperature.

FIG. 6.28. (a), (b), (c), (d) Numerical results and thermal camera image of the
multidirectional anti-symmetric and the asymmetric media in dimensionless form. (-30/60):
σ11 = σ22 = 22168.5, σ12 = σ21 = 0. (Athanasopoulos & Kostopoulos (2013)).

This is validated also experimentally for an asymmetric laminate of (-30/60). The

concluded temperature field in this case is presented in Fig. 6.28(d) and (e). In general
(assuming thin body approach), for any combination of pairs of layers [(θ1/θ1+90)/
(θ2/θ2+90)/… (θΝ/θΝ+90)], the off-diagonal terms of the total conductivity tensor will
be equal to zero.

6.1.4 Electrothermal analysis in circular CFRP domains

Finally we studied the electric field distribution in multi-layered materials of circular
geometry, by applying potential difference between the two concentric circles of Fig.
6.29(a) (using Dirichlet boundary conditions). The boundary conditions for the second
case of the following problem are described by a first kind boundary condition
(Dirichlet boundary condition) at the Γd regions and by second kind boundary
conditions (Neumann boundary condition) at the Γs, Fig. 6.29(b).
132 Chapter 6

FIG. 6.29. Numerical results and thermal camera images for two circular multidirectional
CFRPs using the equivalent EC tensor (Rmax/Rin = 4, H = 0.726mm, h = 0.121mm). (a), (b)
current density, (c), (d) electric potential field (e), (f) calculated temperature field, (g), (h)
measured temperature field. (Athanasopoulos & Kostopoulos (2012)d).

The indirect method for the validation of the tensor involved the use of a
thermal camera. We studied a circular domain with a symmetric stacking sequence
(0/30/45)s and ratio (λ = Rmax / Rin = 4). The electric potential difference was applied
at the edges of the circular domain.
The temperature field of the multilayered medium depends on, the EC tensor
(stacking sequence of the CFRP) and the boundary conditions. The calculated
temperature field is in excellent agreement to the temperature measurements, using
the thermal camera.
Considering a 2D body and using the continuity equation for anisotropic media,
the electrical conductivity tensor of a CF multilayered preform can be predicted. The
EC tensor of the CF medium can be calculated for any stacking sequence based on the
proposed equation, Eq. 4.8. Taking into consideration the calculated EC tensor of the
CF multilayer medium, the elliptical PDE can be solved numerically. Therefore, the
electric field, the current density and the generated heat, can be calculated.
Introducing the generated heat into the heat transfer equation, the temperature
distribution over the CF preform can be predicted. The experimental results were
compared against the numerical calculations of the temperature field for each of the
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 133
and multidirectional CF media

unidirectional and multidirectional carbon fiber preforms. In all the tested cases the
calculated EC tensor leads to a numerical model which is in excellent agreement with
the experimental results. This is a strong indirect validation of the analytical EC
tensor calculation of any CF multilayered medium.
In an electrically isotropic body, the vector of current density is orthogonal to
the equipotential lines, whereas in an anisotropic body, the current density vector may
not be orthogonal to the equipotential lines. Also, the equipotential lines are not
necessarily perpendicular to the adiabatic surfaces.
As mentioned, the concluded temperature field depends upon the EC tensor. In
the case of rectangular domains and the given boundary conditions, the temperature
field of the CF preform is uniform when the values of σ12, σ21, are equal to zero. If the
ratio (λ = L/H) is high enough, the assumption that the body in question is a thin body
(2D), can be made. In the case of thick preforms with high L/H ratios, the EC tensor is
still valid.
The electric conductivity of any thin multidirectional CFRP medium can be
expressed by a symmetric second order tensor which derives from the combination of
the EC tensor of each layer. In the case where the material has been manufactured
with high quality standards, it is absolutely reliable to assume that the material is
homogeneous. Thus, under certain manufacturing conditions, the interlaminar
inhomogeneity could be eliminated, then the material imperfections could be
minimized, and eventually the fibre volume fraction would be uniform everywhere
inside the material. Then the laminate can be assumed homogenous and anisotropic,
and therefore the experimental measurements of the EC are in excellent agreement
with the theoretical prediction of Eq. 4.8. The validity of the proposed equivalent EC
tensor's calculation has also been confirmed by comparing the numerical results to the
respective experimental. Finally, the calculation of an equivalent EC tensor can be
used in order to avoid largely memory-allocating three-dimensional models.

6.2 Electrical heterogeneous media

Equation 2.17 was solved numerically take into consideration that the EC is a
function of temperature. In this case the material is electrical non-homogeneous and
anisotropic. In this case, all the material properties except for the thermal conductivity
134 Chapter 6

are temperature dependent. Natural convection has been applied in each node using
the previous equations (Eq.6.3 – Eq. 12).
The following diagram presents the FDM code steps in order to solve Eq. 2.17 and
Eq. 2.29, Figure 6.30.

FIG. 6.30. Algorithm using FDM for the electrical non-homogeneous medium.

Electric potential was applied at the edges of an orthogonal domain with

(0/30/30/0) stacking sequence, (detail Fig. 6.31). The applied electric potential was
1V and the coupon was of the following dimensions (L=225mm, W=68mm,
H=0.5mm). The calculated results were compared against thermal camera
measurements Figure 6.31 presents the calculated and the measured temperature at the
edge, Fig. 6.31(a) and at the middle, Fig. 6.31(b) of the domain. The black line
represents the numerical results for the ideal case where they are no copper
connections. The blue line represents the numerical results in the case where the
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 135
and multidirectional CF media

copper connections had also been modeled. The red dotted line represents the
measured results using the thermal camera.

FIG. 6.31. Calculated and measured temperature distribution at the edge and at the middle of
the coupon (0/30/30/0).

In all the cases the calculated temperature is very close to the experimental
results. It must be mentioned that the total current that passes through the coupon is (I
= 4.22A) and the calculated is (I = 4.43A). The electrical current in the numerical
scheme can be calculated integrated the current density at the edge of the domain
(copper connection region).
136 Chapter 6

FIG. 6.32. Calculated and measured temperature distribution at the diagonal of the coupon
(0/30/30/0).

Finally, two different cases were calculated in order to compare the temperature
distribution in the cases of a) constant and b) temperature dependent EC tensor. The
results are present in Fig. 6.33 for two different power levels. In the first power level
the temperature distribution for the two cases is almost the same. In the second case
the temperature distribution is slightly different. The temperature dependency of the
EC tensor does not impact in the final temperature field at low temperature levels.
This happens because the electrical resistivity coefficient of the CF medium is very
small. The governing equation could be applied in the case of very high temperatures
> 600oC.

FIG. 6.33. Calculated temperature distribution at the edge of the coupon (0/30/30/0) for two
different electric potential differences.
Numerical analysis and experimental validation of the electrothermal problem: unidirectional 137
and multidirectional CF media

Figure 6.34 presents the specific heat capacity and the EC conductivity (σ11) in 2D
space at the steady state temperature.

FIG. 6.34. Specific heat capacity and EC as a function of space for two different electric
potential differences at steady state.
138 Chapter 6
Damage detection of CFRPs using Joule effect 139
 

7. start
Chapter 7
Damage detection of CFRPs via Joule effect

Any change of the microstructure of the composite laminate due to an interlaminar

damage, locally changes the electrical conductivity tensor of the medium. Applying
electric potential difference, the temperature of the medium rises, due to the Joule
effect. In the presence of interlaminar damage or CF breakage, the developed
temperature field changes locally. Following the coupled electrical/thermal solution of
the problem, the mechanism of the phenomenon is elucidated and validated against
experimental results, by comparing the measured to calculated temperature field. In
nondestructive inspection, a lot of work has been carried out using electrical methods
(Kupke et. al (2001), Baltopoulos et. al (2012), Angelidis et. al (2004)).
Unfortunately, the above present certain difficulties in detecting damage in CFRP
structures and the research is grounded in small CFRP specimens (Suzuki et. al
(2012)). Via the Joule effect (resistive heating) and by using thermography imaging,
Guzmán et. al. (2011) have reported the feasibility to detect cracks in nano - doped
woven composites. Also, Suzuki et. al. (2004) have reported the feasibility of
detecting damages caused by indentation loads in CFRPs, in the special case where a
thick rich resin film exists between layers with direction (0/90).
140 Chapter 7
 

7.1. Interlaminar damage detection (delamination)

All the aforementioned applications require, primarily, the determination of the
multilayered CFRP's electric conductivity (EC). The determination of the EC of the
anisotropic multilayered material is the most crucial factor for the calculation of the
electric field and the current density distribution.
In most composite structures reinforced with continuous carbon fibers, CFRP
laminates are made of multidirectional layers. Each multidirectional laminate is
formed of at least two layers of different fiber direction. Using the hypothesis that the
laminate's thickness is very small compared to the other laminate dimensions and
using the continuity equation, it has been proven that the EC tensor of dry
multidirectional media (Athanasopoulos & Kostopoulos (2012)f) as well as the EC
tensor of multidirectional CFRP media (Athanasopoulos & Kostopoulos (2012)d), can
be expressed by an equivalent symmetric, second order tensor, Eq. 4.8. Hence, the EC
tensor of the CFRPs can be calculated for any stacking sequence.
In order to consider a multi-layered material macroscopically homogenous, it is
necessary to examine the intralaminar and interlaminar regions. During the
manufacturing process of the CFRP laminate, a specific temperature profile as well as
a specific pressure profile is applied. A rich-in-resin layer, which may be formed in-
between the carbon fibre plies of a CFRP, results in a significant local reduction of the
fiber volume fraction between the plies. Therefore the material could lack in
homogeneity, since the presence of even a thin resin layer could locally electrically
insulate the laminate. In the case where the material has been manufactured with high
quality standards, it is absolutely reliable to assume that the material is homogeneous.
Thus, under certain manufacturing conditions, the interlaminar inhomogeneity could
be eliminated, the material imperfections could minimise and eventually the fibre
volume fraction would be uniform everywhere inside the material.
A common mode of failure in CFRPs is delamination (detail Fig. 7.1(a) and Fig.
7.1(c)). In the presence of an interlaminar damage (delamination) the material
separates into three different domains: the healthy material (H, σH), the upper

delaminated area (D1, σD2) and the lower delaminated area (D2, σD2), Fig. 7.1. The
EC tensor of the healthy material is different from to the EC tensors of the damaged
area σH ≠ σD1, σD2. Whenever electric potential difference is applied on an

 
D
Damage detection using Joulee effect 1441
 

unndamaged laminate, thee electric fieeld as well as the curreent density distribution
d is
dependant on n the EC tennsor of the material. Onn the other hand, in thee presence of o
damage, the material's E EC tensor presents
p a llocal variation at the damaged
d areea
because of the t resin maatrix breakaage (detail in i Fig. 7.1((a)). Therefo
fore, both thhe
ellectric field and the currrent density distribution are affectedd by the locaal variation of
o
E tensor, Fig. 7.1(c).
EC
In otherr words, Fig.. 7.1(c) illusttrates a locally damagedd laminate, which
w consissts
of two layerss (±45). The EC of the healthy
h dom
main (H) can be expresseed by a scalaar
quuantity σ±455 = σH becaause the off diagonal terrms are equaal to zero (σ
σ21 = σ12 = 0)
0
annd (σ11 = σ222). The EC tensor
t at the domains (D
D1) and (D2)), are expresssed by an EC
E
teensor (σ45=σσD1 and σ-455=σD2) respeectively. Wee note that thhe (ξ1, ξ2) aree the principal
axxes of the material,
m correesponding foor each EC tensor.
t

FIG. 7.1. (a) Illustration off a CFRP dom main with an interlaminar damage (delaamination) annd
ann optical miicroscope im mage of the delaminated area (non-ellectrically coonductive areea
between the layers), (b) thhrough the thhickness illusstration at thhe delaminatiion region, (cc)
teensor ellipse at the healthyy material (i.e., ±45 stackking sequencee) and tensor ellipses at thhe
damage area. (Athanasopou
( ulos & Kostoppoulos (2012))a).

All in all, differennt stacking sequences lead to diffferent EC tensors. t As a

coonsequence,, the generatted heat, whhich is equaal to the inneer product of o the electrric
fiield and currrent density ( E  J ) variies from poiint to point aaccording too the stackinng
seequence botth in the case of the health mateerial and thhe damaged d material, aas
prresented in Table
T 7.1. Thhe temperatuure field in the
t different domains is also
a differennt.

 
142 Chapter 7
 

Introducing the electrically generated heat into the heat transfer equation, the
temperature field over the CFRP laminate can be obtained.
i.e. the EC tensors of the (σ45) and the (σ-45) can be calculated using Eq. (7.1).

σ
 σ m 2 + σ 2 m 22 (σ1 - σ 2 )m1m 2 
n  1 1  (7.1)
 (σ1 - σ 2 )m1m 2 σ 2 m12 + σ1m 22 
where m1=cos(45), m2=sin(45) in the case of the (σ45) EC tensor and m1=cos(-45),
m2=sin(-45) in the case of the (σ-45) EC tensor. (σ1) and (σ2) are the EC along and
transverse to the fibers direction respectively. Hence, using Eq. (1) for the multilayer
material (σ±45) the equivalent EC tensor can be calculated by Eq. (7.2).
σ1
=
-
σ
42
5

- 42
51
=
1 σ11 + σ22 
45 -45
1 0
σ ±45  σ 45 +σ -45   -45  (7.2)
σ1
=
σ
41
5

- 42
52

45
N 2 0 σ22 + σ11 
where (N = 2) is the number of layers, Eq. (4.8).

Healthy material,(H) Damaged area Damaged area

(upper surface), D1 (lower surface), D2
σH

σD

σD
Governing     0      0     0
1

2
equation
σ

σ
Generated E E E
H

D
1

D
2
E E E
heat
σH
σ

σD

σD

EC tensor 22168.5 0  22168.5 22118.5  22168.5 22118.5

1

     
(S/m)
45
 0 22168.5 45 
22118.5 22168.5
45 
22118.5 22168.5 

TABLE 7.1. Governing equations, generated heat for the different regions and EC tensor’s
values for each area. (Athanasopoulos & Kostopoulos (2012)a).

The above stated approach can be applied for the detection of damage in the
form of delamination. For the experimental verification of the proposed damage
detection method, an electric potential was applied using a DC power supply
(QPX1200L). The electrodes were attached at the tips of the specimen according to
Kupke et. al. (2001). After curing, the top resin rich layer was removed from the
edges of the samples and a conductive top layer of silver-paint was applied to them.
The specimen tips were cleaned using acetone and the copper electrodes were welded
on the specimens with the aid of a highly conductive solderable tape (silver-tape). It
has to be noted that the application of pressure on the silver tape is achieved with the
use of clamps. A thermal camera (Trotech IC80L) was used to record the temperature
field developed on the CFRP multilayer laminates. Accuracy in thermal-camera

 
Damage detection using Joule effect 143
 

temperature measurements is achieved by spraying black mat paint (with total

hemispherical emissivity ε ≈ 0.96) on the surface of the specimens. The room
temperature for all measurements was at approximately (28οC).
The CFRP laminates were made again of a high standard prepreg material
(Sigrafill / Sigratex, epoxy UD-prepreg, 242gr/m2) using the autoclave manufacturing
process. The applied pressure at manufacturing stage of the CFRP laminates was (P =
0.8MPa). The curing cycle of the material was 3h at 130oC. The material was
artificially damaged, by placing a thin plastic film of (hfilm = 25μm) thickness between
the (±45) CFRP layers. The experimental results were compared against the numerical
calculations of the temperature field, where the EC tensor was calculated based on the
proposed mathematical formula, Eq. (4.8). In all the tested cases, the calculated EC
tensor leads the numerical model and results in excellent agreement of the numerical
with the experimental findings.
For comparison, the results will be presented in dimensionless form. More
precisely, the spatial co-ordinates have been expressed as x 1*  x 1 L and x *2  x 2 W ,
where (L) and (W) are the length and the width of the specimen respectively. The
electric potential and the temperature field presented in dimensionless form, following
the expression  *    m ax and T *   T  T e n v   T m a x  T en v  respectively.
Two rectangular specimens were manufactured, each of them containing a
rectangular artificial damage at a different location on the specimen. Therefore, two
cases of delamination were examined according to Fig. 7.2, (white dashed lines
denote the geometry of the delaminated area). In both cases, the material’s EC tensor
values at the different areas, are presented in Table 7.1.
Figure 7.2(a) represents the calculated current density, the electric potential, the
temperature field and the measured temperature field of the healthy material, which
consists of a (±45) stacking sequence.
As far as the first case of damage is concerned, Fig. 7.2(b), the potential was
(Δφ  = 1.73V), the electric current was (Ι = 2.02Α) leading to a maximum recorded
temperature of (Τmax = 70oC). The local change of the electrical properties of the
material at the delaminated area causes the variation of the electric field (detail, Fig.
7.2(b)). This fact, results to a variation of the maximum temperature developed
locally at the middle of the damaged area, as well as at the corners of the damaged
region, Fig. 7.2(b). Following a careful observation of Fig. 7.2(b), it is possible to
define the extension and the exact location of the damage between the layers.

 
144 Chhapter 7
 

Furtherm
more, a mulltimedia of Fig. 7.2(b)) provides tthe compariison betweeen the
calculated and the meeasured trannsient temperrature field.
As far as the second casee of damagee is concernned (edge deelamination)), Fig.
7.2(c), thhe potential was
w (Δφ = 1.60V),
1 the electric
e curreent was (Ι = 2.43Α) leadding to
a maximmum recordeed temperatuure of (Τmaxx = 59oC). The observvations are almost
a
similar aand the bouundaries off the damagge are definned by the developed local
maximum m of temperaatures, Fig. 7.2(c).
7

FIG. 7.2. Numerical and

a experimeental results in dimensionnless form foor a (±45) staacking
sequence with dimenssions (L = 128mm,
1 W = 45mm). (a)) healthy maaterial, (b) arttificial
rectangular damage inn the middlee of the speccimen (28mm m x 45mm, white
w dashedd area)
“enhanced online”, (c) artificial reectangular dam
mage at the edge of the specimen
s (322mm x
20mm) “enhanced online”. In all cases the thickness of thee laminate is 0.25mm. Material
M
propertiess: ρm=1092kg/m3, ρf=11800kg/m3, cm=1187J/kggK, cf=750JJ/kgK, (1-φ φ)=0.7,
k11=k22=55.42W/mK, k33=0.65W W/mK, σ1=44287S/m,
= σ2=50S/m, σ3=σ33=550S/m.
(Athanasoopoulos & Koostopoulos (22012)a).

Thee same electtric potentiall difference was appliedd to both thee numerical model
m
and the experiments
e s. The compparison betw ween the num merical and the experim mental
results, lleads to thee conclusionn that the numerical
n m
models are cconfirmed by
b the
thermal ccamera meassurements. AllA the necesssary thermaal and electrrical propertiies are
providedd in Fig. 7.2.
Figure 7..3(a) and (b) presents thee temperaturre field at diffferent time steps.

 
D
Damage detection using Joulee effect 1445
 

FIG. 7.3. (a) Measured

F M tem
mperature fieldd as a functioon of time (aloong the widthh delaminatioon
between a (±45) CFRP meddium), (b) meeasured and calculated
c temmperature fieldd as a functioon
of time (edge delamination
d between a (±
±45) CFRP medium).

Finally, the case of a delaminattion detection, which is at

a the centerr of the CFRRP
laaminate, cann be similarlly defined. The
T followinng thermal camera
c imagge, Fig. 7.4(aa)
demonstrates the aforem mentioned ressults at the steady state. Additionallly, two morre
caases were exxamined: a) a CFRP plaate of (02) sttacking sequuence with a delaminatioon
att the center of the speciimen, Fig. 7.5(a),
7 and b) a CFRP plate of stack king sequencce
(00/90/45/-45/-45/45/90/0)), with (1 mm)
m thickneess, (0.4m) llength and (0.2m)
( widthh,

 
146 Chhapter 7
 

Fig. 7.5(bb). Thereforre we can asssess the maaterial’s conddition, the loocation as well
w as
the geommetrical traitss of the damaage.

FIG. 7.4. (a) calculated temperaturee field duringg the heating of the CFRP P for a delamiination
at the cennter of a (±45)) stacking seqquence, (b) thhermography image in dim
mensionless foorm of
an interlaaminar damag ge with rectanngular geomeetry (28mmxx28mm) at thee center of a (±45)
stacking sequence sp pecimen (110mmx128mm m) (Δφ=0.822V, Ι=3.10Α Α, Τmax=41oC),C (c)
thermograaphy image in dimensionlless form of a second inteerlaminar dam mage with thee same
geometryy and stacking g sequence. (A Athanasopoulos & Kostopooulos (2012)aa).

Thee mechanism m of the phenomenon

p n for the damage deteection is veerified.
Equationn 4.8 is validd and can bee used in ordder to solve the
t electricaal problem annd the
transient heat transfeer problem inn multidirecttional CFRP media. The local tempeerature
maxima define the extension
e off the damage, as well as a the locatioon of the daamage

 
D
Damage detection using Joulee effect 1447
 

between the layers.

l Furthhermore, studdying the moorphology off the temperature field w
we
caan identify the trough thickness
t loocation of thhe delaminatted area (beetween whicch
laayers is the damage
d locatted).

FIIG. 7.5. (a) measured

m tem
mperature fielld during thee heating of tthe CFRP speecimen with 2
laayers of 0o fiiber directionn, (b) thermoggraphy imagee of a circulaar damage (d
diameter of thhe
arrtificial interllaminar damaage=40mm) at a CFRP plate p (400mmm x 200mm) with stackinng
seequence (0/90 0/45/-45/-45/445/90/0) and thickness hCFFRP≈1mm (thee magnified im
mage has beeen
caaptured at timme t2≠t1), (Δφ==3V, Ι=35Α). (Athanasopooulos & Kosttopoulos (2012)a).
 

7 Fiber and matrrix fracturre

7.2.
The electric potential
T p fielld in the CFRRP material changes beccause of the fiber fracturre.
T alteration
The n of the electtric potentiaal leads to a different
d tem
mperature fieeld (especiallly
att the crack region)
r becaause of the Joule
J effect. This changge can be obbserved via a
thhermal cameera.
Two diffferent casess have been studied
s in orrder to validaate the possiibility of fibeer
frracture: a) a UD specim men with a crrack of 4mm m length, Figg. 7.6, b) a UD U specimeen
w two craccks of 3mm and
with a 9mm lenngth respectively, Fig. 77.7. The num merical modeels
have been com mpared and validated ussing a thermaal camera.
Figure 7.6(b) presents thee measured temperature
t field in the ccase of the single
s crack at
thhe center off the specim men and Fig. 7.6(c) presents the calculated tempperature fielld
duuring the heating stage.

 
148 Chhapter 7
 

FIG. 7.6. (a) CFRP speecimen with 4mm

4 crack, (b)
( measured temperature field, (c) calcculated
temperatuure field.

Alsso, Fig. 7.7(bb) presents thhe measuredd temperature field of thee 2nd case annd Fig.
7.7(c) preesents the caalculated tem
mperature fieeld during thhe heating sttage. All in all,
a the
maxima temperaturee regions inddicate the crack tips andd the numeriical models are in
excellentt agreement with
w the therrmal cameraa measuremeents.

FIG. 7.7. (a) CFRP speecimen with 99mm and 3mm cracks, (b)) measured tem
mperature fieeld, (c)
calculatedd temperature field.

 
D
Damage detection using Joulee effect 1449
 

In the case of matrixx cracking thhe electric potential

p fieldd alteration is
i smaller annd
thhe impact onn the temperaature field iss less Fig. 7.88.

FIIG. 7.8. Meassured temperaature field in the

t case of matrix
m crackingg (2mm crackk).

In concclusion, dam mage can be detected ussing a therm mal camera via the Joule
efffect for diffferent kind of
o damage (ddelaminationn, fiber fractture, matrix fracture).
f It is
immportant thaat we stress the
t significaance of the area
a on whicch the potenttial differencce
iss applied. Thhis can be exxplained by the followinng; by attachhing the electtrodes on tw wo
reegions on th he material, it is possible that the damage
d migght not be easily
e locatedd,
coompared to the case whhere the elecctrodes are attached
a on a different set
s of regionns.
T geometry
The y of the dammaged area affects
a the ttemperature profile. Furrthermore, thhe
laayers’ thickknesses, aloong with thhe dimensioons of the specimen are of great
immportance, as well. It should be noted thatt the most challenging g part of thhe
exxperimental procedure was the atttachment off the electroodes on the CFRP platte.
B
Because of thhis technicaal problem thhe connectioons may cauuse small diifferentiationns
between the calculated
c annd the measuured temperaature field.
Further research is rrequired in order
o to defiine the marggins where thhe Eq. (4.8) is
valid (as far as the overaall thickness of the lamiinates concerrns) as well as where thhe
prroposed metthod can bee used. This methodologgy has severral advantagges over otheer
m
methods usedd for damagge detection in the case where curvved and com mplex surfacees
arre involved.

 
150 Chapter 7
 

 
Conclusions 151
 

8. start
Conclusions

The EC has been studied as a function of thickness and fibre volume fraction
along and transverse to the fibre direction for the case of unidirectional CF plies in
two different forms: porous (dry preform) and CFRP. The measurements involved in
the aforementioned study produced diagrams and empirical formulas which relate the
material's VR to the material's thickness and the applied pressure. Additionally, the
model developed by Chen et. al (2000) was used in order to relate the VR to the
carbon fibre volume fraction. Using this methodology, only four measurements are
necessary in order to calculate the EC for any thickness, fibre volume fraction (or
pressure level) and fibre direction.
Using the electrical current conservation, it has been proved that an equivalent
EC tensor can be used to predict the EC of any stacking sequence. The equivalent EC
tensor is valid for the porous CF layers and for the CFRP layers. The following
mathematical expression was validated using three different routes: a) electrical
resistivity measurements of multilayered materials for various stacking sequences, b)
using simple 3D models in order to study the hypothesis that the electrical gradient
through the thickness is negligible when the layers are in total contact, c) comparing
electro thermal numerical models with thermal camera measurements for different
stacking sequences. The equivalent EC tensor was also related to the temperature as
well as to the thermal coefficients of the material's electrical conductivity. The
specific heat capacity as a function of temperature was measured in order to develop
more accurate electrothermal models
152 Conclusions
 
The numerical models were developed using the finite difference method
(FDM), in order to solve the electrothermal problem on an electrically anisotropic,
homogenous or non-homogenous material. The numerical models use the equivalent
EC tensor of various stacking sequences. By means of the calculated EC tensor of the
multilayer preform, the elliptical form of the governing equation was solved
numerically. Based on this, the generated heat was determined for each case. The
determination of the EC of the multilayered medium is the most crucial factor for the
calculation of the generated heat and the prediction of the temperature field. The
generated heat in a known domain (Ω), with given boundary conditions, changes
according to the EC tensor of the multilayered medium. Different stacking sequences
lead to different EC tensors. Thus, the resulting generated heat in the known domain
is also different. Introducing the generated heat into the heat transfer equation, the
temperature field over the CF medium can be predicted. For the experimental
verification, a thermal camera was used to record the temperature field developed on a
CF multilayer preform under given electric potential field. The experimental results
were compared to the respective numerical calculations of the temperature field,
where the EC tensor was calculated analytically based on the proposed methodology.
The results were confirmed experimentally via thermal camera images, for a large
number of representative stacking sequences. Also, the FDM solutions were
compared to FEM solutions in order to validate the electrical potential field, for the
case of the electrical homogeneous and non-homogeneous materials. The models can
predict very well the electrical potential field but higher order schemes are necessary
for high accuracy solutions, especially in the high anisotropy cases.
The behaviour of the EC along and transverse to the fibre direction, can be study
further using numerical models and take into consideration the microstructural
characteristics of the material. The numerical solutions of the elliptic PDE (electrical
problem) are accurate but higher order finite difference schemes are required in order
to approximate better the sharp gradients of the electric potential. In the case where
the CFRP material is thick, the 2D numerical models are not valid. 3D numerical
models are required in order to solve the electrical problem. The structure of the
material at interlaminar and intralaminar regions is crucial and defines the validity of
the equivalent EC tensor.
The damage detection using Joule effect is an attractive method because of its
applicability in carved and complex CFRP structures. It is important that we stress the
significance of the area on which the potential difference is applied. The most
challenging part of the procedure is the attachment of the electrodes on the CFRP
structure.
Conclusions 153
 
Finally, this study can be used in manufacturing of composite materials (using
direct heating) in order to develop multifunctional materials, like structural heating
elements (light CFRP moulds), which are capable of controlling the CFRP elements
temperature. In the case of damage the use of Joule heating can make it visible using a
thermal camera.
154 Conclusions
 
References 155
 

References

Angelidis, N. & Irving, P.E. (2007), “Detection of impact damage in CFRP laminates
by means of electrical potential techniques”, Composites Science and Technology
67(3-4), 594–604.

Angelidis, N., Khemiri, N. & Irving, P.E. (2004), “Experimental and finite element
study of the electrical potential technique for damage detection in CFRP laminates”,
Smart Materials and Structures 14(1), 147-54.

Angelidis, N., Wei, C.Y. & Irving, P.E. (2004), “The electrical resistance response of
continuous carbon fibre composite laminates to mechanical strain”, Composites Part
A: Applied Science and Manufacturing 35(10), 1135–47.

Arpacy, V.S. (1966), “Conduction heat transfer”, Addison-Wesley Publishing

Company.

ASTM C 835-95. Standard test method for total hemispherical emittance of surfaces
from 20 to 1400oC. West Conshohocken: ASTM.

Athanasopoulos, N. & Kostopoulos, V. (2011), “Prediction and experimental

validation of the electrical conductivity of dry carbon fibre unidirectional layers”,
Composites Part B: Engineering 42(6):1578–87.

Athanasopoulos, N. & Kostopoulos, V. (2012)a, “Damage detection via Joule effect

for multidirectional carbon fiber reinforced composites”, Applied Physics Letters
101(11): 114109-5.
156 References
 

Athanasopoulos, N. & Kostopoulos, V. (2012)d, “Calculation of an Equivalent

Electrical conductivity tensor for multidirectional carbon fiber Reinforced Materials”,
Progress In Electromagnetics Research Symposium Proceedings, Moscow, Russia,
August, 1013-1018.

Athanasopoulos, N. & Kostopoulos, V. (2012)e, “Resistive heating of multi-

directional and unidirectional dry carbon fibre preforms”. Composites Science and
Technology, 72(11), 1273-1282.

Athanasopoulos, N. & Kostopoulos, V. (2013), “Strong validation of the equivalent

electrical conductivity tensor of thin multidirectional carbon fibre reinforced plastics”,
Under review.

Athanasopoulos, N. & Kostopoulos, V. (2013), “Strong validation of the equivalent

electrical conductivity tensor of thin multidirectional carbon fibre reinforced plastics”,
Under review.

Athanasopoulos, N., Koutsoukis, G., Vlahos, D. & Kostopoulos, V. (2012)b,

“Lightweight composite molds using carbon fibres as heating elements”, Composites
Part B: Engineering (Accepted).

Athanasopoulos, N., Sikoutris, D., Panidis, T. & Kostopoulos, V. (2012)c,

“Numerical investigation and experimental verification of the Joule heating effect of
polyacrylonitrile-based carbon fiber tows under high vacuum conditions”, Journal of
Composite Materials 46(18), 2153–65.

Athanasopoulos, N., Sotiriadis, G. & Kostopoulos, V. (2010), “A study on the effect

of Joule - heating during the liquid composite molding (LCM) process and on the
curing of CFRP composite laminates”, 10th International Conference on Flow
Processes in Composite Materials, Paper No 32, Ascona, July 2010.

Baltopoulos, A., V., Polydorides, N., Pambaguian, L., Vavouliotis, A. & Kostopoulos.
V. (2012), “Damage identification in carbon fiber reinforced polymer plates using
electrical resistance tomography mapping”, Journal of Composite Materials
(Accepted).

Baltopoulos, A., Vavouliotis, A., Kostopoulos, V., Polydorides, N. & Pambaguian, L.

(2012), Electrical tomography as a tool for non-destructive assessment of composite
References 157
 

structures”, Emerging Technologies in Non-Destructive Testing V - CRC Press, 389-

99.

Bensaid, S., Trichet, D. & Fouladgar, J. (2005), “3-D Simulation of induction heating
of anisotropic composite materials”, IEEE Transactions on Magnetics 41(5), 1568-71.

Bensaid, S., Trichet, D. & Fouladgar, J. (2006), “Electromagnetic and thermal

behaviors of multilayer anisotropic composites materials”, IEEE Transactions on
magnetics 42(4), 995-98.

Bruschke, M.V., & Advani, S.G. (1990), “A Finite Element/Control Volume

Approach to Mold Filling in Anisotropic Porous Media”, Polymer Composites, 11(6),
398-405.
Bueche, F. (1972), “Electrical resistivity of conducting particles in an insulating
matrix”, Journal Applied Physics 43(11), 4837.

Carslaw, H. S. & Jaeger, J. C. (1959), “Conduction oh heat in solids, 2nd Edition”,

Oxford University Press.

Chen, B., Cheng, A.H.-D. & Chou, T.-W. (2001), “A nonlinear compaction model for
fibrous preforms”, Composites Part A: Applied Science and Manufacturing 32(5).
701-7.

Chen, B., Lang, E.J. & Chou, T.W. (2001), “Experimental and theoretical studies of
fabric compaction behavior in resin transfer molding”, Materials Science and
Engineering 317(1-2), 188–96.

Christopher, L., Holloway, M., Sarto, S. & Johansson, M. (2005), “Analyzing carbon-
fiber composite materials with equivalent-layer models”, IEEE Transactions on
Electromagnetic Compatibility 47(4), 833-44.

Chung, D.D.L. (1994), “Carbon Fiber Composites”, Butterworth-Heinemann.

Clingerman, M.L, King, J.A., Schulz, K.H. & Meyers J.D. (2002), “Evaluation of
Electrical Conductivity Models for Conductive Polymer Composites”, Journal of
Applied Polymer Science 83(6), 1341–56.

Datoo, M.H. (1991), “Mechanics of fibrous composites”, Elsevier applied science, pp.
153-55.
158 References
 

Dimoka, P. (2012), “Repairing of aircraft structures using Joule effect”. Diploma

thesis, University of Patras.

Filippov V.V. (2008), “A procedure for determining components of the specific

electricsal conductivity tensor of anisotropic semiconductors films”, Instruments and
Experimental Techniques, 51(4), 630-33.

Funke, H. & Brandes, J. (2011), “Device for electrically conductive laminates made
of carbon-fiber-reinforced plastics (CFRP laminates)”, Patend.

Galehdar, A., Nicholson K.J., Rowe, W.S.T. & Ghorbani, K. (2010), “The
conductivity of unidirectional and quasi isotropic carbon fiber composites”
Microwave Conference Proceedings, European Melbourne, Australia 882 – 85.

Galehdar, A., Rowe, W. S. T., Ghorbani, K., Callus, P. J., John, S. & Wang, C.H.
(2011), "The effect of ply orientation on the performance of antennas in or on carbon
fiber composites," Progress In Electromagnetics Research 116, 123-36.

Galehdar, A., Rowe, W.S.T., John, S., Callus, P.J. & Ghorbani, K. (2010), “The
anisotropic conductivity of unidirectional carbon fibre reinforced polymer laminates
and its effect on microstrip antennas”, Microwave Conference Proceedings,
Melbourne, VIC, Australia. 1470–1473

Gigliotti, M.., Grandidier, J.C., Lafarie-Frenot, M.C. & Marchand D. (2010),

“Development of experimental and modeling tools for electro-mechanical fatigue
tests in composite materials for aircraft applications”, 14th European Conference on
Composite Materials, Budapest, Hungary.

Greenkorn, R.A. (1983), “Flow Phenomena in Porous Media”, New York: Marcel
Dekker.

Greenwood, J.H., Lebedat, S. & Bernasconi, J. (1975), “The anisotropic electrical

resistivity of a carbon fibre reinforced plastic disc and its use as a transducer”,
Journal of Physics E: Scientific Instruments 8(5), 369-70.

Hale, D.K. (1976), “Review the physical properties of composite materials journal of
materials science”, Journal of Materials Science, 11(11), 2105-41.
References 159
 

Imai, J. & Kaneko, K. (1992), “Electrical Conductivity of a Single Micrographitic

Carbon Fiber with a High Surface Area under Various Atmospheres”, Langmuir 8(7),
1695-97.

Irving, P.E. & Thiagarajan C. (1998), “Fatigue damage characterization in carbon

fibre composite materials using an electrical potential technique”, Smart Materials
and Structures 7(4), 456-66.

Jewett JW and Serway RA. Electricity and magnetism, Physics for scientists and
engineers with modern physics. Vol. 758, th ed. Pacific Groove, California:
Thompson, 2008.

Joseph, C. & Viney, C. (2000), “Electrical resistance curing of carbon – fibre/epoxy

composites”, Composites Science and Technology 60(2), 315-19.

Kaddour, S., AI-Salehi, F.A.R., A1-Hassani S.T.S. & Hinton, M.J. (1994), “Electrical
resistance measurement technique for detecting failure in CFRP materials at high
strain rates”, Composites Science and Technology 51(3), 377-85.

Knibbs, R.H. & Morris J.B. (1974), “The effect of fibre orientation on physical
properties of composites”, Composites 5(5), 209-18.

Kostopoulos, V., Vavouliotis, A., Karapappas, P., Tsotra, P. & Paipetis, A. (2009),
“Damage Monitoring of Carbon Fiber Reinforced Laminates Using Resistance
Measurements. Improving Sensitivity Using Carbon Nanotube Doped Epoxy Matrix
System”, Journal of Intelligent Material Systems and Structures 20(9), 1025-34.

Koutsoukis, G. (2011), Ανάπτυξη ελαφρών καλουπιών από σύνθετα υλικά με τη

χρήση ινών άνθρακα ως θερμαντικά στοιχεία. Πανεπιστήμιο Πατρών.

Krauss, J.D. (1991). “Electromagnetics, 4th Edition”, Mcgraw-Hill Series in Electrical

Engineering, 185-218.

Kupke, M., Shulte, K., & Shuler, R. (2001), “Nondestructive testing of FRP by dc and
ac electrical methods”, Composites Science and Technology 61(6), 837-47.

Landau LD and Lifshitz EM. “Steady current, electrodynamics of continuous media,

2nd ed”, Oxford: Pergamon Press, Vol. 8, pp. 86–104.
160 References
 

Mawardi, A. & Pitchumani, R. (2003), “Optimal temperature and current cycles for
curing of composites using embedded resistive heating elements”, Journal of Heat
Transfer 125(1), 126–36.

Megali, G., Pellicano, D., Cacciola, M., Calagno, S., Versaci, M. & Morabito F.C.
(2010), “EC modeling and enhancement signals in CFRP inspection”, Progress in
Electromagnetics Research 14, 45-60.

Menana, H. & Feliachi, M. (2009), “3-D eddy current computation in carbon fiber
reinforced composites”, IEEE Transactions on magnetics 45(3), 1008-11.

Menana, H. & M. Feliachi, (2011), “An integro differential model for 3d current
computation in carbon fiber reinforced polymer composites”, IEEE Transactions on
magnetics 47(4), 756-63.

Michael A. Berger, & R.L. McCullough (1984), “Characterization and analysis of the
electrical properties of a metal-filled polymer”, Composites Science and Technology
22(2), 81-106.

Nield, D.A. & Bejan, A. (2006), “Heat Transfer through a Porous Medium,
Convection in Porous Media,. 2nd ed», Springer, pp. 27-28.

Nielsen, L.E. (1974), “The Thermal and Electrical Conductivity of Two-Phase

Systems”, Industrial and Engineering Chemistry and Fundamentals, 13(1), 17-20.

Ondracek, G. (1987), “The quantitative microstructure-field property correlation of

multiphase and porous materials”, Reviews on Powder Metallurgy and Physical
Ceramics 3, 205-322.

Ozisik, M.N. (1993), “Heat conduction”. Wiley.

Ozisik, M.N. (2002), “Boundary Value Problems of Heat Conduction”, Dover

Publications.

Park, J.B., Hwang, T.K., Kim, H.G. & Doh, Y.D. (2007), “Experimental and
numerical study of the electrical anisotropy in unidirectional carbon-fiber-reinforced
polymer composites”, Smart Materials and Structures 16(1), 57-66.
References 161
 

Park, J.B., Okabe, T. & Takeda N. (2003), “New concept for modeling the
electromechanical behavior of unidirectional carbon-fiber-reinforced plastic under
tensile loading”, Smart Materials and Structures 12, 105-14.

Piche, A., Revel, I. & Peres, G. (2011), “Experimental and numerical methods to
characterize electrical behavior of carbon fiber composites used aeronautic industry”,
Advances in Composite Materials – Analysis of Natural and Man Made Materials.

Pradere, C., Batsale, J.C., Goyhénèche, J.M., Pailler, R. & Dilhaire, S. (2009),
“Thermal properties of carbon fibers at very high temperature”, Carbon 47(3), 737-47.

Pradere, C., Goyheneche, J.M., Batsale, J.C., Dilhaire, S. and Pailler, R. (2005).
Specific heat measurement of single metallic, carbon, and ceramic fiber matrix at very
high temperature, Review of scientific instruments, 76, 064901.

Safárová, V. & Grégr, J. (2010), “Electrical conductivity measurement of fibers and

yarns”, 7th International conference – TEXSCI, Liberec, Czech Republic.

Saunders, R.A. Lekakou, C. & Bader, M.G. (1999), “Compression in the processing
of polymer composites 1. A mechanical and microstructural study for different glass
fabrics and resins”, Composites Science and Technology 59(7), 983-93.

Schueler, R., Joshi, S. & Schulte, K. (2001), “Damage detection in CFRP by electrical
conductivity mapping”, Composites Science and Technology 61(6), 921–30.

Schulte, K. & Baron, C.H. (1989), “Load and failure analyses of CFRP laminates by
means of electrical resistivity measurements”, Composites Science and Technology
36(1), 63-76.

Shen, L., Li, J., Liaw, B.M., Delale, F. & Chung, J.H. (2007), “Modeling and analysis
of the electrical resistance measurement of carbon fiber polymer–matrix composites”,
Composites Science and Technology 67(11-12), 2513–20.

Shindo, A. (2000), “Polyacrylonitrile (PAN)-based carbon fibers” In: Kelly A and

Zweben C (eds) Comprehensive composite materials, Amsterdam: Elsevier Science,
Vol. 1, pp.1–33.

Stephen, A.S. & Donald, J.L. (2008), “Consolidation of U-Nyte Epoxy-Coated

Carbon-Fiber Composites via Temperature-Controlled Resistive Heating”, Journal of
Composites Materials 42(24): 2551-66.
162 References
 

Suzukia, Y., Todoroki, A., Matsuzakia, R., & Mizutania, Y. (2012), “Impact-damage
visualization in CFRP by resistive heating: Development of a new detection method
for indentations caused by impact loads”, Composites Part A: Applied Science and
Manufacturing 43(1), 53-64.

Takahashi, K. & Hahn, H.T. (2011), “Investigation of temperature dependency of

electrical resistance changes for structural management of graphite/polymer
composite”, Journal of Composite Materials 45(25), 2603–11.

Taya, M | Ueda, N, (1987), “Prediction of the In-Plane Electrical Conductivity of a

Misoriented Short Fiber Composite: Fiber Percolation Model vs. Effective Medium
Theory”, Journal of Engineering Materials and Technology (Trans. ASME). 109(3),
252-56.

Todoroki, A., Samejima, Y., Hirano, Y. & Matsuzaki, R. (2009), “Piezoresistivity of

unidirectional carbon/epoxy composites for multiaxial loading”, Composites Science
and Technology 69(11-12), 1841–46.

Todoroki, A., Tanaka, M. & Shimamura, Y. (2002), “Measurement of orthotropic

electric conductance of CFRP laminates and analysis of the effect on delamination
monitoring with an electric resistance change method”, Composites Science and
Technology 62(5), 619–28.

Todoroki, A., Tanaka, M. & Shimamura, Y. (2003), “High performance estimations

of delamination of graphite/epoxy laminates with electric resistance change method”,
Composites Science and Technology 63(13), 1911-20.

Toho Tenax. Tenax HTA/HTS filament yarn. Technical Data Sheet, 2008.

Torayca. T700S Data sheet. Technical Data Sheet, No. CFA-005, 2005.

Tse, K.W., Moyer, C.A. & Arajs, S. (1981), “Electrical Conductivity of Graphite
Fibre-Epoxy Resin Composites”, Materials Science and Engineering 49(1), 41-6.

Ueda., M. & Todoroki, A. (2008), “Delamination monitoring of CFRP laminate using

the two-stage electric potential change method with equivalent electric conductivity”,
Engineering Fracture Mechanics, 75(9), 2737-50.
References 163
 

Volpe, V. (1980), “Estimation of electrical conductivity and electromagnetic shielding

characteristics of graphite/epoxy laminates”, Journal of Composite Materials 14(3),
189-98.

Wang, S., Chung, D.D.L. & Chung, J.H. (2005), “Impact damage of carbon fiber
polymer–matrix composites, studied by electrical resistance measurement”,
Composites Part A: Applied Science and Manufacturing 36(12), 1707–15.

Wang, X. & Chung, D.D.L. (1998), “Self-monitoring of fatigue damage and dynamic
strain in carbon fiber polymer-matrix composite”, Composites Part B: Engineering
29(1), 63-73.

Wasselynck, G., Trichet, D., Ramdane, B. & Fouldagar, J. (2010), “Interaction

between electromagnetic field and CFRP materials: a new multiscale homogenization
approach”, IEEE transactions on magnetics 46(8), 3211-80.

Weber, M. & Kamal M.R. (1997), “Estimation of the volume resistivity of electrically
conductive composites”, Polymer Composites 8(6), 711-25.

Wen, J., Xia, Z. & Choy, F. (2011), “Damage detection of carbon fiber reinforced
polymer composites via electrical resistance measurement”, Composites Part B:
Engineering 42(1), 77–86.

Xiao, J., Li, Y. & Fan. W.X. (1999), “A laminate theory of piezoresistance for
composite laminates”, Composites Science and Technology, 59(9), 1369-73.

Zhu, L. & Pitchumani, R. (2000), “Analysis of a process for curing composites by the
use of embedded resistive heating elements”, Composites Science and Technology
60(14), 2699-2712.
164 References
 
Appendix I 165
 

Appendix I
A) Except for the validation that was presented in Chapter 3, two different electrode
configurations have been tested in order to validate the results. In the case of the dry
CF preform: a) a silver tape and a copper braid electrode were placed at the bottom
and at top of the edge, also silver tape was placed between the CF layers. b) silver
tape and a copper braid electrode were placed at the bottom and at top of the edge,
also silver paint was infused at the edge of the specimens. The electrical resistance
measurements were very close in all the cases.

 
FIG. 1. Two different cases of the electrical connections in the case of the dry CF preforms.
(a), (c), (d) silver tape, CF preform, silver tape, CF preform, braid copper electrode. (b), (d),
(f) silver tape, silver paint and copper braid electrode.

In the case of the CFRP material: a) the specimen was scratched and cleaned, some
layers of silver paint were applied at the edge of the specimen, silver tape were placed
166 Appendix I
 

according to Fig. 2. Final pressure clamps were used in order to measure with
accuracy and eliminate the contact resistance. b) silver tape and a copper braid
electrode were placed at the bottom and at top of the edge during the manufacturing
procedure, Fig. 3.

FIG. 2. 1st type of the electrical connections in the case of the CFRP material. Scratching,
cleaning, 2-3 layers of silver paste (around the material and at the edge), solderable silver
tape and pressure clamps.
 

 
Appendix I 167
 

B) Detailed figures of the dry CF preform using optical microscope. 

FIG. 4. Structure of the dry CF preform on the surface of the specimens. (a), (b) CF in
random directions, (c), (d), (e), (f) non uniform fiber distribution at the edge of the specimens.
 

FIG. 5. Thermal camera images of the specimens using Joule effect for different specimens
dimensions and thicknesses.
 
 
 

 
 
Apppendix II 169
 

A
Append
dix II

 
FIIG. 1. Measurement setup of the CF preeform electriccal resistance as a functionn of applied
prressure and nuumber of layeers.
 

 
 
 

 
 
 
 

 
 
 
 

 
 
 
 
 

T
The presentt thesis waas carried out in thee Applied Mechanics Laboratory
ry,
M
Mechanical E
Engineering
g & Aeronaautics Department, Uniiversity of Patras
P durin
ng
2008 – 2013
 

D
During this thesis, it haas been prooved simplyy that the eelectrical coonductivity of
o
m
multilayered and electriccally anisotrropic carbonn fiber materrials can be expressed by
b
ann equivalent second ordder tensor, which
w is equual to sum oof each layeer’s electricaal
coonductivity tensor. Thiss equivalent electrical conductivity
c tensor is vaalid assuminng
thhat the mateerial’s thickn
kness is neglligible comppared to thee other dimeensions of thhe
boody. The ma
athematical expression
e f the prediiction of the electrical coonductivity of
for o
a multilayereed material for
f any staccking sequennce, is baseed on the eleectric currennt
coonservation,, and was vaalidated using different methods.
m
Moreovver using thee finite differrence methodd, certain eleectrothermall models werre
de
developed in order to studdy the tempeerature fieldd for differennt stacking seequences. Thhe
ellectrical prooblem can be
b expresseed by an elliptic PDE, for the casse where thhe
m
material is electrically
e a
anisotropic and homogeeneous, or non-homoge
n eneous. Usinng
thhe equivalen
nt tensor, thee 3D domainn problem iss simplified to a 2D dom
main problem
reesulting in leess computational requirrements for tthe solution of the probllem.
The preesent researcch study couuld be used in a plethorra of applica
ation, such aas
thhe developm
ment of carbon fibre reiinforced heaating elemennts (direct heating
h CFR
RP
m
molds) as weell as damaage detectioon in multiddirectional ccomposite materials
m witth
ellectrical connductive reinf
nforcement.
N. Athanasopouloos