Vous êtes sur la page 1sur 196

ROYAUME DU MAROC ‫المملكة المغربية‬

UNIVERSITE IBN TOFAIL ‫جامعة ابن طفيل‬

BAGHAD Abd
Résumé : CENTRE D’ETUDES DOCTORALES ‫مركز دراسات الدكتوراه‬
KENITRA ‫القنيطرة‬
Dans l’industrie aéronautique, le procédé autoclave est devenu un équipement indispensable pour la
fabrication des composantes structurelles de haute qualité à partir des composites à matrice polymère.
Les cycles de polymérisation en autoclave sont des étapes clé quant à la fabrication de composites de
bonnes qualités à base du préimprégné. Ces cycles définissent la microstructure et propriétés physiques,
qui contrôlent à leurs tours les propriétés mécaniques et thermomécaniques finales de la pièce. La
présente thèse s’inscrit dans le cadre d’un projet de recherche et développement qui vise à étudier une
nouvelle préimprégné aéronautique 8552S/AS4. L’objectif est d’identifier les paramètres clés de la FACULTE DES SCIENCES
fabrication, afin d’améliorer les propriétés mécaniques tout en préservant les propriétés physiques. Dans
un premier temps, l’objectif était de déterminer la cinétique de la réaction de polymérisation de la résine FORMATION DOCTORALE : PHYSIQUE ET APPLICATIONS
époxyde, et de valider un modèle cinétique permettant de décrire son comportement dans un cycle de SPÉCIALITÉ : SCIENCE DES MATÉRIAUX

Fabrication et évaluation des stratifiés carbone/époxy pour l’aéronautique : Effets


température et de temps. Par la suite, les cycles thermiques de fabrication en autoclave ont été stimulé

des paramètres de fabrication en autoclave et les conditions opérationnelles


en utilisant la DSC, afin de démonter le temps minimal de maintien à un palier de température, et bien
évidement de déterminer l’évolution du degré de polymérisation au cours du temps. Ensuite, les effets THÈSE DE DOCTORAT
des paramètres de fabrication en autoclave (température, pression et vide) sur les propriétés
microstructurales, physiques, thermomécaniques et mécaniques ont été évalués. Comme la température Par
de fonctionnement des nacelles des avions est de 120°C, il était important d’étudier l’effet de la
température d’opération et du temps de maintien sur les propriétés mécaniques et thermomécanique. Mr. Abd BAGHAD
Mots-clés : préimprégné époxy, composites stratifiés, cinétique de polymérisation, paramètres de Sous le thème
fabrication en autoclave, propriétés mécaniques, conditions opérationnelles.

Fabrication et évaluation des stratifiés carbone/époxy pour


l’aéronautique : Effets des paramètres de fabrication en
autoclave et les conditions opérationnelles
Absract :
Autoclave process has become an indispensable equipment in the aeronautical industry for the
manufacturing of high quality structural components from polymer matrix composites. Autoclave cure Soutenue le …. 2022 devant le Jury composé des :
cycles are key steps in the manufacture of high quality prepreg-based composites. These cycles
determine the microstructure and physical properties, which in turn affect the final mechanical and
thermomechanical properties of the part. The present thesis is part of a research and development Nom et Prénom Grade Statut Etablissement
project studying a new aerospace prepreg 8552S/AS4. The objective is to identify the key Pr. PES Président
manufacturing parameters in order to improve the mechanical properties while maintaining the physical
properties. First, the objective was to determine the curing kinetics of the epoxy resin, and to develop a Pr. Mohammed Assouag PES Rapporteur Ecole Nationale Supérieure d’Arts et Méties – Meknès
kinetic model to describe its behavior in a temperature and time cycle. Then, the thermal cycles in
autoclave were stimulated by using the DSC in order to determine the minimum holding time at Pr. Pr. Abdellah Laazizi PES Rapporteur Ecole Nationale Supérieure d’Arts et Méties – Meknès
isothermal temperature, as well as to evaluate the evolution of the degree of cure over time. The effects
of the autoclave manufacturing parameters (temperature, pressure and vacuum) on the microstructural, Pr. PES Rapporteur
physical, thermomechanical and mechanical properties were then evaluated. Since the operating
temperature of aircraft nacelles is 120°C, it was important to study the effect of operating temperature Pr. PES Examinateur
and holding time on mechanical and thermomechanical properties. Since the operating temperature of
Nom du to
aircraft nacelles is 120°C, it was important Lboratoire d’accueil
study the effect of operating temperature and holding Pr. PES Examinateur
time on mechanical and thermomechanical properties.
Pr. Khalil El Mabrouk PES Co-directeur Université Euromed de Fès – Fès
Keywords: Epoxy prepreg, composite laminates, curing kinetics, autoclave process parameters,
mechanical properties, operating conditions. Pr. Khalid Nouned PH Directeur Faculté des Sciences – Kénitra
N° d’ordre : …/…..

CENTRE D’ETUDES DOCTORALES | UNIVERSITE IBN TOFAIL KENITRA


Adresse : B.P 242, Kénitra – Maroc | https://ced.uit.ac.ma ANNEE UNIVERSITAIRE : 2021/2022
Je dédie cette thèse à mon frère, Mr. Iazza Baghad (1973-2021), qui a
compté et compte encore beaucoup pour moi. Merci beaucoup "Baba
Iazza", je n’oublierai jamais les sacrifices que tu aies fait pour me
voir réussir.
Remerciements
La réalisation de cette thèse a été pour moi un défi personnel et professionnel mais surtout une
magnifique et riche expérience humaine. Elle m’a permet d’apprendre beaucoup de choses sur moi-
même et sur le monde. J’en garderai, malgré les difficultés rencontrées, de plaisants souvenirs et
de vrais ami(e)s. C’est aussi l’occasion de remercier toutes les personnes qui ont contribué, tout au
long de mes études, à leur réussite et surtout à l’ouverture de mon esprit sur le monde, sur notre
communauté, sur les autres et sur les questions qui les préoccupent.

« Il n'y a guère au monde un plus bel excès que celui de la reconnaissance. »

Jean De La Bruyère

Tout d’abord, je tiens à remercier mon directeur de thèse, le professeur Khalid Nouneh, pour son
soutien et ses précieux conseils tout au long de mes études de doctorat. En second lieu, je tiens à
exprimer ma sincère reconnaissance à mon co-directeur de thèse, le professeur Khalil El Mabrouk,
qui m'a donné l'occasion de travailler sur de nombreux projets passionnants dans son équipe. Il m’a
fait découvrir le monde des matériaux composites, m’a transmis la philosophie de la recherche
scientifique, et il m’a notamment formé avec beaucoup d’attention à la maintenance et la gestion
des infrastructures et les équipements de recherche. Il m'a toujours donné la liberté d'explorer
différentes idées et d'utiliser différents instruments dans différentes infrastructures du campus, et
son mentorat m'a permis de sortir de ma zone de confort et d'aborder des problèmes de recherche
intéressants. Tout cela a abouti à la naissance d’un chercheur indépendant dans le domaine des
polymères et leurs matériaux composites, et je ne pourrai jamais le remercier assez pour le temps
et les efforts qu'il a consacrés à ces formations précieuses.

Je suis également reconnaissant pour les conseils, le soutien et les leçons transmises par le
professeur Sebastien Vaudreuil, de l’Université Euromed de Fès. Il m’a permis d'acquérir une
expérience précieuse qui me sera utile pour les prochaines étapes de ma carrière. Je remercie très

i
sincèrement le professeur Salim Bounou, directeur de l'École d'Ingénieurs BiomedTech de
l'université Euromed, sur sa disponibilité, accueil, et particulièrement pour sa confiance. Toute ma
reconnaissance va également au professeur Hicham Zaitan, de la Faculté des Sciences et
Techniques de Fès, dont sa grande rigueur scientifique, son amitié et sa disponibilité ont été des
facteurs déterminants pour la réussite de mes études supérieures.

Je voudrais également exprimer toute ma gratitude à M. xxxx, professeur à l’université xxxx, pour
l’honneur qu’il m’a fait d’accepter la présidence de mon jury. Les professeurs xxx, de xxx et xxxx,
de xxx, pour l’intérêt critique qu’ils ont porté en acceptant de juger ce travail. Je tiens à remercier
aussi xxx, professeur à xxx, xxxx, professeur à xxxx, pour l’honneur qu’ils m’ont fait en acceptant
de participer à ce jury.

Ensuite, je tiens à souligner que mes travaux de recherches ne seraient pas possibles sans le soutien
financier important et reconnaissant du Safran Nacelles, l’Académie Hassan II des Sciences et
Techniques et l’Université Euromed de Fès. Je remercie tout particulièrement le professeur
Mostapha Bousmina, président de l’Université Euromed de Fès qui n’a ménagé aucun effort pour
le bon déroulement de mes travaux de recherche.

Je remercie tous mes collègues et les membres de mon équipe de recherche au sein de l’Université
Euromed de Fès, et plus particulièrement Dr. Zakaria Tabia, Mme. Sihame Akhtache, Mme. Salwa
El Baakili, Mr. Noussaire Elfihry, Mr. Yassine El Ansary, Mme Afaf Hrichi et Pr. Meryeme Bricha.
Ils ont été d'excellents collègues, mentors et amis, et je leurs souhaite le plus grand succès, la santé
et le bonheur dans le futur. Je remercie tout particulièrement ma formidable et meilleure amie,
Mme. Samia Taghbalout, qui m'a soutenu tout au long de mes études supérieures. Les
encouragements de Samia m'ont aidé à survivre aux moments les plus difficiles ; son soutien a été
la chose la plus précieuse pour moi au cours des neuf dernières années.

Je suis profondément reconnaissant au staff de l’Université Euromed de Fès, et plus


particulièrement le Pr. Anas Bakouri, Pr. Abdellah Arrahli, Pr. Nabil El Brahmi, Mr. Tarek
Eddryouch, et Mme. Hind El Jai pour leur amitié. Mes remerciements très spéciaux vont à Mr.
Brahim Seddouki, directeur de la Direction de la communication de l’Université Euromed de Fès,
ainsi qu’à Mme Yasmine Bahaji, directrice de la Direction des affaires estudiantines et culturelles
de l’Université Euromed de Fès, pour leur extrême gentillesse, amitié extraordinaire et pour avoir
ii
rendu mon expérience au sein de l’université amusante et mémorable. Je tiens également à
remercier Mr. Mustafa Maaizi, électricien de la plateforme de recherche de l’Université Euromed
de Fès, pour les longues heures que nous avons passées ensemble à résoudre les pannes de
l’autoclave.

Finalement, je serai toujours redevable à ma famille d'être arrivée là où je suis maintenant. Mes
parents n'ont jamais eu la possibilité de faire des études supérieures, mais ils m'ont inculqué le rêve,
dès mon plus jeune âge, de poursuivre la meilleure formation académique possible. Mes frères
Iazza et M’bark et ma soeur, Fatima, ont été un soutien constant de mon potentiel pour atteindre
cet objectif. Une pensée pour mon grand-père, Lahoucine Baghad qui nous a quittés le 20 août
2021.

Abd Baghad

Fès, le 28 févr.-22

iii
Résumé

Dans l’industrie aéronautique, le procédé autoclave est devenu un équipement indispensable pour
la fabrication des composantes structurelles de haute qualité à partir des composites à matrice
polymère. Les cycles de polymérisation en autoclave sont des étapes clé quant à la fabrication de
composites de bonnes qualités à base du préimprégné. Ces cycles définissent la microstructure et
propriétés physiques, qui contrôlent à leurs tours les propriétés mécaniques et thermomécaniques
finales de la pièce. La présente thèse s’inscrit dans le cadre d’un projet de recherche et
développement qui vise à étudier une nouvelle préimprégné aéronautique 8552S/AS4. L’objectif
est d’identifier les paramètres clés de la fabrication, afin d’améliorer les propriétés mécaniques tout
en préservant les propriétés physiques. Dans un premier temps, l’objectif était de déterminer la
cinétique de la réaction de polymérisation de la résine époxyde, et de valider un modèle cinétique
permettant de décrire son comportement dans un cycle de température et de temps. Par la suite, les
cycles thermiques de fabrication en autoclave ont été stimulé en utilisant la DSC, afin de démonter
le temps minimal de maintien à un palier de température, et bien évidement de déterminer
l’évolution du degré de polymérisation au cours du temps. Ensuite, les effets des paramètres de
fabrication en autoclave (température, pression et vide) sur les propriétés microstructurales,
physiques, thermomécaniques et mécaniques ont été évalués. Comme la température de
fonctionnement des nacelles des avions est de 120°C, il était important d’étudier l’effet de la
température d’opération et du temps de maintien sur les propriétés mécaniques et
thermomécanique.

Mots-clés : préimprégné époxy, composites stratifiés, cinétique de polymérisation, paramètres de


fabrication en autoclave, propriétés mécaniques, conditions opérationnelles.

iv
Abstract

Autoclave process has become an indispensable equipment in the aeronautical industry for the
manufacturing of high quality structural components from polymer matrix composites. Autoclave
cure cycles are key steps in the manufacture of high quality prepreg-based composites. These cycles
determine the microstructure and physical properties, which in turn affect the final mechanical and
thermomechanical properties of the part. The present thesis is part of a research and development
project studying a new aerospace prepreg 8552S/AS4. The objective is to identify the key
manufacturing parameters in order to improve the mechanical properties while maintaining the
physical properties. First, the objective was to determine the curing kinetics of the epoxy resin, and
to develop a kinetic model to describe its behavior in a temperature and time cycle. Then, the
thermal cycles in autoclave were stimulated by using the DSC in order to determine the minimum
holding time at isothermal temperature, as well as to evaluate the evolution of the degree of cure
over time. The effects of the autoclave manufacturing parameters (temperature, pressure and
vacuum) on the microstructural, physical, thermomechanical and mechanical properties were then
evaluated. Since the operating temperature of aircraft nacelles is 120°C, it was important to study
the effect of operating temperature and holding time on mechanical and thermomechanical
properties. Since the operating temperature of aircraft nacelles is 120°C, it was important to study
the effect of operating temperature and holding time on mechanical and thermomechanical
properties.

Keywords: Epoxy prepreg, composite laminates, curing kinetics, autoclave process parameters,
mechanical properties, operating conditions.

v
Table des matières

Remerciements .................................................................................................................................. i

Résumé ............................................................................................................................................ iv

Abstract ............................................................................................................................................ v

Table des matières ........................................................................................................................... vi

Liste des tableaux ...........................................................................................................................xii

Liste des figures............................................................................................................................. xiv

INTRODUCTION GENERALE................................................................................................... 1

CHAPITRE 1 REVUE DE LITTÉRATURE ............................................................................ 5

1.1. Généralités sur les matériaux composites stratifiés ........................................................... 5

1.1.1. Constituants d’un matériau composite ....................................................................... 6

1.1.2. Procédés d’élaboration ............................................................................................. 12

1.2. Effet du cycle de fabrication en autoclave sur les propriétés des composites stratifiés .. 14

1.2.1. Optimisation et modélisation du cycle thermique .................................................... 15

1.2.2. Effet de la pression et du vide sur les propriétés des composites stratifiés .............. 27

1.2.3. Conclusions .............................................................................................................. 31

1.3. Porosité au cours du cycle de fabrication en autoclave ................................................... 32

1.3.1. Origine et développement des vides ......................................................................... 33

1.3.2. Influence de la porosité sur les propriétés mécaniques ............................................ 34

vi
CHAPITRE 2 THE ISOTHERMAL CURING KINETICS OF A NEW CARBON
FIBER/EPOXY RESIN AND THE PHYSICAL PROPERTIES OF ITS AUTOCLAVED
COMPOSITE LAMINATES ...................................................................................................... 39

2.1. Avant-propos ................................................................................................................... 39

2.2. Résumé ............................................................................................................................ 39

2.3. Abstract ............................................................................................................................ 40

2.4. Introduction...................................................................................................................... 41

2.5. Materials and experimental protocol ............................................................................... 42

2.5.1. Materials ................................................................................................................... 42

2.5.2. Thermal analysis and curing kinetics ....................................................................... 43

2.5.3. Autoclave processing of composite laminates ......................................................... 45

2.5.4. Characterization of autoclaved composite laminates ............................................... 45

2.6. Results and discussion ..................................................................................................... 46

2.6.1. DSC curve analysis .................................................................................................. 46

2.6.2. Curing kinetics ......................................................................................................... 49

2.6.3. Microstructure and physical properties of laminated composites ............................ 50

2.7. Conclusions...................................................................................................................... 54

2.8. Acknowledgments ........................................................................................................... 54

CHAPITRE 3 CURE KINETICS AND AUTOCLAVE-PRESSURE DEPENDENCE ON


PHYSICAL AND MECHANICAL PROPERTIES OF WOVEN CARBON/EPOXY
8552S/AS4 COMPOSITE LAMINATES ................................................................................... 56

3.1. Avant-propos ................................................................................................................... 56

vii
3.2. Résumé ............................................................................................................................ 56

3.3. Abstract ............................................................................................................................ 57

3.4. Introduction...................................................................................................................... 58

3.5. Experimental procedure ................................................................................................... 60

3.5.1. Materials ................................................................................................................... 60

3.5.2. Thermal analysis....................................................................................................... 60

3.5.3. Fabrication and curing .............................................................................................. 61

3.5.4. FTIR ......................................................................................................................... 62

3.5.5. Density and void content .......................................................................................... 62

3.5.6. Optical microscopy................................................................................................... 63

3.5.7. Mechanical properties .............................................................................................. 63

3.6. Results and discussion ..................................................................................................... 63

3.6.1. DSC .......................................................................................................................... 63

3.6.2. FTIR analysis of laminated composites processed by autoclave ............................. 68

3.6.3. Effect of curing pressure on microstructure ............................................................. 69

3.6.4. Effect of curing pressure on mechanical properties ................................................. 73

3.7. Conclusions...................................................................................................................... 76

3.8. Acknowledgments ........................................................................................................... 77

CHAPITRE 4 EFFECT OF VACUUM-PRESSURE ON MICROSTRUCTURE AND


MECHANICAL PROPERTIES OF AUTOCLAVED EPOXY/CARBON COMPOSITE
LAMINATES ................................................................................................................................ 79

4.1. Avant-propos ................................................................................................................... 79

4.2. Résumé ............................................................................................................................ 79

viii
4.3. Abstract ............................................................................................................................ 80

4.4. Introduction...................................................................................................................... 81

4.5. Experimental procedure ................................................................................................... 83

4.5.1. Fabrication of composite laminates .......................................................................... 83

4.5.2. Microstructure analysis ............................................................................................ 84

4.5.3. Mechanical tests ....................................................................................................... 85

4.5.4. Statistical analysis .................................................................................................... 86

4.6. Results and discussions.................................................................................................... 87

4.6.1. Influence of vacuum-pressure on microstructure and mechanical properties .......... 87

4.6.2. Statistical analysis and relationship between microstructure and mechanical


properties ................................................................................................................................. 92

4.7. Conclusions...................................................................................................................... 99

4.8. Acknowledgements........................................................................................................ 100

CHAPITRE 5 APPLICATION OF RESPONSE SURFACE METHODOLOGY FOR


INVESTIGATION THE IMPACT OF AUTOCLAVE PROCESS PARAMETERS ON THE
PHYSICAL PROPERTIES OF EPOXY/CARBON LAMINATES...................................... 102

5.1. Avant-propos ................................................................................................................. 102

5.2. Résumé .......................................................................................................................... 102

5.3. Abstract .......................................................................................................................... 103

5.4. Introduction.................................................................................................................... 104

5.5. Methodology .................................................................................................................. 106

5.5.1. Materials ................................................................................................................. 106

5.5.2. Hand lay-up and autoclave curing .......................................................................... 106

ix
5.5.3. Characterization...................................................................................................... 108

5.6. Results and discussions.................................................................................................. 109

5.6.1. Experimental data ................................................................................................... 109

5.6.2. Analysis of variance and quadratic model fitting ................................................... 110

5.6.3. Estimation of quantitative effects of process parameters and regression models .. 111

5.6.4. Main and interaction effects ................................................................................... 113

5.6.5. Linear correlations and model effectiveness .......................................................... 117

5.6.6. Contour and surface plots for responses ................................................................. 119

5.7. Conclusions.................................................................................................................... 123

5.8. Acknowledgments ......................................................................................................... 124

CHAPITRE 6 EFFECTS OF HIGH OPERATING TEMPERATURES AND HOLDING


TIMES ON THERMOMECHANICAL AND MECHANICAL PROPERTIES OF
AUTOCLAVED EPOXY/CARBON COMPOSITE LAMINATES ..................................... 126

6.1. Avant-propos ................................................................................................................. 126

6.2. Résumé .......................................................................................................................... 126

6.3. Abstract .......................................................................................................................... 127

6.4. Introduction.................................................................................................................... 128

6.5. Materials and methods ................................................................................................... 130

6.5.1. Materials ................................................................................................................. 130

6.5.2. Hand lay-up and autoclave curing .......................................................................... 130

6.5.3. Dynamic mechanical properties ............................................................................. 131

6.5.4. Mechanical properties ............................................................................................ 131

6.6. Results and discussion ................................................................................................... 133

x
6.6.1. Dynamic mechanical properties ............................................................................. 133

6.6.2. Mechanical properties ............................................................................................ 134

6.7. Conclusions.................................................................................................................... 148

6.8. Acknowledgements........................................................................................................ 149

CONCLUSION GENERALE ................................................................................................... 151

LISTE DES RÉFÉRENCES BIBLIOGRAPHIQUES ................................................................. 156

xi
Liste des tableaux
Table 1. Propriétés des fibres à haute résistance [9] ........................................................................ 6
Table 2 : Résultats de propriétés physiques des quatre cycles [22] ............................................... 16
Table 3 : Effet de la porosité sur la résistance au cisaillement interlaminaire (RCIL) pour les
composites époxy/fibres de carbone............................................................................................... 36
Table 4. Designed autoclave cure cycles........................................................................................ 45
Table 5. The time corresponding to the maximum heat flow for different isothermal temperatures
........................................................................................................................................................ 47
Table 6. The curing kinetics parameters of 8552S epoxy resin prepreg at different isothermal
temperatures ................................................................................................................................... 52
Table 7. Curing kinetics parameters for 8552S/AS4 resin ............................................................. 67
Table 8. Characteristic bands related to epoxy/amine resin ........................................................... 69
Table 9. Descriptive statistics of composite properties .................................................................. 92
Table 10 : Pearson’s correlation coefficient and respective p value between microstructure and
mechanical properties ..................................................................................................................... 93
Table 11 : One Way ANOVA and means comparisons by Scheffé and Tukey tests between density
and mechanical properties .............................................................................................................. 95
Table 12 : One Way ANOVA and means comparisons by Scheffé and Tukey tests between fiber
volume fraction and mechanical properties.................................................................................... 96
Table 13 : One Way ANOVA and means comparisons by Scheffé and Tukey tests between void
content and mechanical properties ................................................................................................. 97
Table 14 : Effect of autoclave process parameters on the physical properties of laminated
composites .................................................................................................................................... 105
Table 15 : Factors and levels considered for design .................................................................... 106
Table 16 : Box-Behnken for the experimental study and experimental responses ...................... 110
Table 17 : Analysis of variance (ANOVA) table for the validation of the model ....................... 112
Table 18 : Estimated regression coefficients (as coded factors) and P-values for the degree of cure,
density, fiber volume fraction, and void content .......................................................................... 114

xii
Table 19 : Full Factorial Design for 8552S composite laminates ................................................ 132
Table 20 : Effect of experimental variables on mechanical properties ........................................ 135
Table 21 : Analysis of variance for laminate compressive modulus ............................................ 136
Table 22 : Analysis of variance for laminate compressive strength ............................................. 140
Table 23 : Analysis of variance for interlaminar shear strength .................................................. 143
Table 24 : Results of the curve fitting of the experimental LCMN, LCSN, and ILSSN into the model
presented in Equations (3), (4), and (5) ........................................................................................ 147

xiii
Liste des figures
Figure 1. Évolution de l'utilisation des matériaux composites dans les avions Airbus [8] ............. 2
Figure 2. Fibres discontinus(a) et fibres continus(b) ........................................................................ 5
Figure 3 : Préimprégné ................................................................................................................... 10
Figure 4 : Stratifiés unidirectionnels (a) et quasi-isotropes (b) [19] .............................................. 12
Figure 5 : Composants d'ensachage................................................................................................ 13
Figure 6 : Four autoclave de polymérisation à Casablanca, Safran Nacelles Maroc ..................... 14
Figure 7 : Différents cycles de durcissement du préimprégné AS4/8552 et l’évolution
correspondante de la viscosité durant le cycle [22] ........................................................................ 15
Figure 8 : Cycle thermique conventionnel pour les préimprégnés utilisés dans l’aéronautique [23]
........................................................................................................................................................ 17
Figure 9 : Effet de la vitesse de chauffage sur le degré de durcissement (a) et la viscosité (b) [24]
........................................................................................................................................................ 18
Figure 10 : Cycle à un seul palier de température [27] .................................................................. 19
Figure 11 : Schéma des réactions époxy/amine (1-3) et d’éthérification (4) pour les amines
aliphatiques [30] ............................................................................................................................. 22
Figure 12 : Cycle de pression conventionnel utilisé dans le procédé autoclave [23] ..................... 27
Figure 13 : Corrélation entre la pression de l'autoclave et la teneur en vide pour un composite
stratifié fibres de carbone/époxy [59] ............................................................................................. 29
Figure 14 : (a) Variation de la viscosité en fonction du temps, (b) Mesure de la teneur en vide en
fonction du premier instant d’application de la pression [51] ........................................................ 30
Figure 15 : Structure and chemical composition of 8552S: (i) Tetraglycidyl Methylenedianiline,
(ii) Triglycidyl p-aminophenol, (iii) 4,4’-diaminodiphenylsulfone, and (iv) 3,3’-
diaminodiphenylsulfone ................................................................................................................. 43
Figure 16 : Extraction of fibers by acid digestion. The color of the sulfuric acid solution
progressively changes from transparent (a) to light brown (b), followed by dark brown (c), then
black (d), and finally the obtained fibers. ....................................................................................... 47

xiv
Figure 17 : (a) Heat flow versus time and (b) total heat of polymerization/crosslinking reactions at
different isothermal temperatures ................................................................................................... 48
Figure 18 : Evolution of the degree of conversion (a) and the reaction rate (b) as a function of time
........................................................................................................................................................ 48
Figure 19 : Variation of the maximum conversion degree as a function of the isothermal
temperature ..................................................................................................................................... 49
Figure 20 : Modeling of polymerization/cross-linking kinetics using different models: (a) 120 °C,
(b) 140 °C, (c) 160 °C, (d) 180 °C, (e) 200 °C, and (f) 220 °C. ..................................................... 51
Figure 21 : Properties of laminated composites as a function of isotherm temperatures: (a) degree
of cure, (b) density, (c) fiber volume fraction, and (d) void content .............................................. 53
Figure 22 : Curing cycles used to fabricate 8552S/AS4 composites ............................................. 61
Figure 23 : Autoclave and vacuum bags assembly used in this study ........................................... 62
Figure 24 : Dynamic DSC analysis: Heat flow vs temperature (a) and time (b) ........................... 64
Figure 25 : DSC curve simulating the autoclave curing cycle (heating to 180°C at 3°C/min and
hold), (b): Temperature, degree of cure, and reaction rate as a function of time in the autoclave
curing cycle .................................................................................................................................... 65
Figure 26 : DSC dynamic curve (3°C/min) obtained in the second run after the first cycle (heating
to 180°C at 3°C/min and hold) ....................................................................................................... 66
Figure 27 : Reaction rate vs. degree of curing plot with a model provided by Cole et al .............. 67
Figure 28 : FTIR spectra of prepreg and composite laminates ...................................................... 68
Figure 29 : Influence of pressure on (a) density, thickness, (b) fiber volume fraction, resin volume
fraction, and void content ............................................................................................................... 70
Figure 30 : Optical microscope images through the thickness of composite laminates taken at 50×
magnification: (a) at 0.1MPa, (b) at 0.3MPa, (c) at 0.5MPa, and (d) at 0.7MPa ........................... 70
Figure 31 : Mechanical properties as a function of pressure at room and 120°C: (a) LCM, ILSM,
and (b) ILSS and LCS .................................................................................................................... 72
Figure 32 : Correlation between physical and mechanical properties: ILSM and LCM vs. density
(a), ILSS and LCS vs. density (b), ILSM and LCM vs. fiber volume fraction (c), ILSS and LCS
vs. fiber volume fraction (d), ILSM and LCM vs void content (e), and ILSS .............................. 75
Figure 33 : Correlation between LCM and ILSM (a), and LCS as a function of ILSS (b) ............ 76

xv
Figure 34 : The experimental and fitted values of composites thickness and density as a function of
vacuum-pressure ............................................................................................................................. 87
Figure 35 : Evolution of fiber volume fraction and resin volume fraction (a), and void content as a
function of applied vacuum-pressure (b)........................................................................................ 88
Figure 36 : Typical optical micrographs of cross-sections of composite laminates manufactured
under different vacuum-pressure taken at x50 magnification: (a) at 0MPa, (b) at -0.01MPa, (c) at -
0.02MPa, (d) at -0.03MPa, (e) at -0.04MPa, (f) at -0.06MPa, and (g) at 0.08MPa. ...................... 89
Figure 37 : Laminate compressive modulus (LCM) as a function of vacuum-pressure (a),
interlaminar shear strength (ILSS), and laminate compressive strength (LCS) as a function of
different cure vacuum (b) ............................................................................................................... 91
Figure 38 : Relationship between density and mechanical properties: (a): Interlaminar shear
strength (ILSS) and laminate compressive strength (LCS) as a function of density, and (b):
Laminate compressive modulus (LCM) as a function of density. Pearson’s correlation test was
employed for each relation. The results indicate; ILSS (r = 0.99706 and p<0.00001), LCS (r =
0.99634 and p<0.00001), and LCM (r = 0.78348 and p = 0.03714). ............................................. 93
Figure 40 : Relationship between fiber volume fraction and mechanical properties: (a): Interlaminar
shear strength (ILSS) and laminate compressive strength (LCS) as a function of fiber volume
fraction, and (b): Laminate compressive modulus (LCM) as a function of fiber volume fraction.
Pearson’s correlation test was employed for each relation. The results indicate; ILSS (r = 0.99892
and p<0.00001), LCS (r = 0.98473 and p<0.00001), and LCM (r = 0.67703 and p = 0.09479). .. 94
Figure 40 : Relationship between void and mechanical properties: (a): Interlaminar shear strength
(ILSS) and laminate compressive strength (LCS) as a function of void content, and (b): Laminate
compressive modulus (LCM) as a function of void content. Pearson’s correlation test was employed
for each relation. The results indicate; ILSS (r = -0.99622 and p<0.00001), LCS (r = -0.99844 and
p<0.00001), and LCM (r = -0.66544 and p = 0.10281). ................................................................ 98
Figure 41 : Autoclave process schematic ..................................................................................... 108
Figure 42 : Main effect plots of process parameters for the degree of cure, density, fiber volume
fraction, and void content ............................................................................................................. 115
Figure 43 : Classification of effects on the degree of cure (a), density (b), fiber volume fraction (c),
and void content (d) ...................................................................................................................... 117

xvi
Figure 44 : The experimental values plotted against the predicted values derived from the empirical
models of responses: (a) degree of cure, (b) density, (c) fiber volume fraction, (d) and void content
...................................................................................................................................................... 118
Figure 45 : 3D response surfaces and contour plots of the degree of cure with respect to autoclave
process parameters; (a) Temperature vs. pressure, (b) temperature vs. vacuum, and (c) pressure vs.
vacuum. The rest of the process parameters in each plot were set to level 0. .............................. 119
Figure 46 : 3D response surfaces and contour plots of density with respect to autoclave process
parameters; (a) Temperature vs. pressure, (b) temperature vs. vacuum, and (c) pressure vs. vacuum.
The rest of the process parameters in each plot were set to level 0. ............................................ 120
Figure 47 : 3D response surfaces and contour plots of fiber volume fraction with respect to
autoclave process parameters; (a) Temperature vs. pressure, (b) temperature vs. vacuum, and (c)
pressure vs. vacuum. The rest of the process parameters in each plot were s .............................. 122
Figure 48 : 3D response surfaces and contour plots of void content with respect to autoclave process
parameters; (a) Temperature vs. pressure, (b) temperature vs. vacuum, and (c) pressure vs. vacuum.
The rest of the process parameters in each plot were set to level 0. ............................................ 123
Figure 49 : Autoclave cure cycle .................................................................................................. 131
Figure 50 : Dynamic mechanical properties of composite laminate ............................................ 134
Figure 51 : Pareto chart illustrating the effect of operating temperature and holding time on LCM
of composite laminates at 95% significance level, designed as a vertical dashed line ................ 137
Figure 52 : Main effects plot for the main laminate compressive modulus. The reference line is
designed as a longitudinal dashed line ......................................................................................... 138
Figure 53 : Interaction plot for laminate compressive modulus ................................................... 139
Figure 54 : Pareto chart illustrating the effect of operating temperature and holding time on laminate
compressive strength at 95% significance level, designed as a vertical dashed line ................... 140
Figure 55 : Main effects plot for the main laminate compressive strength. The reference line is
designed as a longitudinal dashed line ......................................................................................... 141
Figure 56 : Interaction plot for laminate compressive modulus ................................................... 142
Figure 57 : Pareto chart illustrating the effect of operating temperature and holding time on
interlaminar shear strength at 95% significance level, designed as a vertical dashed line .......... 143

xvii
Figure 58 : Main effects plot for the main laminate compressive strength. The reference line is
designed as a longitudinal dashed line ......................................................................................... 144
Figure 59 : Interaction plot for interlaminar shear strength ......................................................... 145
Figure 60 : Normalized laminate compressive modulus (a), normalized laminate compressive
strength (b), and normalized interlaminar shear strength (c) as a function of holding time and
operating temperature ................................................................................................................... 148

xviii
INTRODUCTION GENERALE

Au cours des 60 dernières années, la demande de matériaux structuraux à haute résistance, plus
grande rigidité et basse densité a augmenté rapidement. De ce fait, la rigidité élevée et le rapport
résistance/poids des composites polymères renforcés de fibres (PRF) en font des matériaux plus
souhaitables pour les applications structurelles, par rapport aux matériaux conventionnels [1]. De
plus, les composites PRF présentent une excellente résistance à la fatigue et à la corrosion et une
bonne flexibilité de conception [2]. Les PRF sont appliqué dans plusieurs domaines tels que
l'aéronautique et l'aérospatiale, le génie civil, le génie maritime, les équipements sportifs, les
instruments médicaux et de nombreuses autres industries [3].

Le développement des matériaux utilisés dans le domaine de l’aéronautique remonte au premier


vol en 1903, lorsque la cellule était une structure en bois. En 1927, les alliages à base d'aluminium
ont atteint la position dominante dans les matériaux aéronautiques avec le développement de
technologies de revêtement et d'anodisation [4]. Au cours de ces dernières années, l’utilisation des
matériaux composites a connu une augmentation significative. La Figure 1 montre le pourcentage
massique des matériaux utilisés dans les avions Airbus. De même dans les avions Boeing, les PRF
sont largement utilisés ; Par exemple, les PRF représentent 50% du poids du Boeing 787-9, qui
consomme 20% de carburant en moins que tout autre avion de sa catégorie [5], réduisant ainsi les
émissions de CO2 [6]. Les composites polymères renforcés de fibres, aident donc, à rendre les
avions plus légers, durables et économiques en carburant, donnant lieu à de plus longs vols sans
escale [7].

Le niveau de développement actuel des PRF, en termes de qualité et quantité, explique que les
thèmes de recherche et de développement concernent, non plus seulement la qualité du produit,
mais aussi l’amélioration de la productivité. En effet, l’amélioration de la rentabilité peut être
obtenue soit par amélioration de chacune des phases d’un procédé existant, ou par un changement
global du procédé de fabrication. L’objectif globale de ce présent projet est d’augmenter la
productivité des opérations de fabrication des pièces en matériaux composites des nacelles. Il s’agit

1
de remplacer les éprouvettes de suivi de fabrication par la surveillance des paramètres clés du
processus de fabrication en autoclave.

Cette thèse s’inscrit donc dans le cadre d’un projet de recherche et développement intitulé « Mise
en place d’une nouvelle matière époxyde », financé par l’Académie Hassan II des Sciences et
Techniques et impliquant le groupe SAFRAN, ainsi que l’Université Euromed de Fès. Il consiste
à analyser le nouveau préimprégné aéronautique Hexply 8552S/37%/280H5/AS4-3K/1170 mm
fabriqué par la société Hexcel Composites France.

Figure 1. Évolution de l'utilisation des matériaux composites dans les avions Airbus [8]
Plus particulièrement, les conditions de mise en œuvre des composites à base du préimprégné
représentent une étape critique dans le procédé de fabrication. Cependant, ces conditions
contrôlent la microstructure et le degré d’avancement des différentes réactions de polymérisation
et de réticulation de la résine époxyde. Le degré d’avancement et la microstructure à leurs tours
sont en relation directe avec les propriétés mécaniques finaux de la pièce. Les paramètres de
cuisson en autoclave ayant un impact direct sur la qualité des pièces sont les suivantes : (1) Vitesse
de montée en température, (2) température et durée du cycle intermédiaire, (3) température et durée
de cycle de cuisson en autoclave, (4) vitesse de refroidissement, (5) moment et l’intensité de mise
sous vide (6) et moment et intensité de mise sous pression.

2
Dans un premier temps, une étude de la cinétique de durcissement permet de mieux comprendre la
relation structure-propriétés, d’optimiser le cycle thermique, et de déterminer les différentes
variables en validant un modèle cinétique adéquat. Après, pour identifier les paramètres de
fabrication à considérer, il est nécessaire d’étudier l’effet de la variation de l’un ou plusieurs
paramètres lors du procédé de fabrication en autoclave sur les propriétés mécaniques suivantes :

 Le cisaillement interlaminaire à température ambiante et à une température de 120 °C


correspondante à celle qui peut subir le matériau dans les nacelles. L’objectif c’est d’évaluer
l’influence de chaque paramètre sur la tenue de la jonction fibre matrice.
 Essais de compression à 120 °C pour évaluer le comportement dans un mode résine pur.

Les présents travaux de recherche visent à acquérir le plus grand nombre possible d’informations
fondamentales et académiques sur la nouvelle résine époxyde et son comportement lors de la
cuisson en autoclave, et d’appréhender les relations entre les différents paramètres de durcissement
et les propriétés mécaniques dans le but d’offrir une large fenêtre de processualité. Ainsi, Safran
Nacelles a entamé en 2012 un processus de qualification de cette matière afin de pouvoir
l’introduire sur les programmes séries et les programmes en développement. Cette étude porte sur
la fabrication des pièces stratifiés à base de la résine époxy en présence des amines et renforcée par
les fibres de carbone.

3
CHAPITRE 1

REVUE DE LITTÉRATURE

Ce chapitre a pour objectif de faire l’état des connaissances dans les domaines liés à la fabrication
des stratifiés carbone/époxy en autoclave. Il se scinde en quatre axes : la définition des matériaux
composites renforcés de fibres de carbone (CFRP), la mise en forme des stratifiés carbone/époxy
et l’effet des paramètres de fabrication en autoclave et de la microstructure sur les propriétés
mécaniques du matériau considéré.

1.1.Généralités sur les matériaux composites stratifiés

Un matériau composite peut être défini comme étant une combinaison de deux matériaux ou plus
qui donne, de meilleures propriétés que lorsque les composants individuels sont utilisés seuls.
Contrairement aux alliages métalliques, chaque matériau conserve ses propriétés chimiques,
physiques et mécaniques distinctes [9]. Certaines des propriétés qui peuvent être améliorées en
formant un matériau composite sont la rigidité, la résistance, la réduction de poids, la résistance à
la corrosion, les propriétés thermiques, la durée de vie en fatigue et la résistance à l'usure [10].

Figure 2. Fibres discontinus(a) et fibres continus(b)


Les deux constituants sont généralement une fibre et une matrice. Les fibres typiques comprennent
le verre, l'aramide et le carbone, qui peuvent être continus ou discontinus. Les matrices peuvent
être des polymères, des métaux ou des céramiques. Les renforts continues comprennent des fibres

5
unidirectionnelles, des tissus et des enroulements hélicoïdaux, tandis que les renforts discontinus
comprennent des fibres coupées et un tapis aléatoire (Figure 2).

Table 1. Propriétés des fibres à haute résistance [9]

Fibre Densité Résistance Module Déformation Diamètre Coefficient


(g/cm3) à la d’élasticité à la rupture (μm) de dilatation
traction (GPa) (%) thermique
(MPa) (ppm/°C)

Verre E 2.491 3447.380 75.845 4.800 9.144 5.040

Verre S 2.547 4481.593 86.874 5.600 9.144 2.340

Quartz 2.187 3378.432 68.948 5.000 8.900 1.800

Aramide -1.980
(Kelvar 1.439 3792.118 131.000 2.800 11.938
49)
Spectra -1.800
0.9689 3102.642 172.369 0.700 25.400
1000
Carbone -0.360
1.799 3654.222 227.527 1.500 8.128
(AS4)
Carbone -0.360
1.772 5033.174 282.685 1.800 5.080
(IM-7)
Graphite -0.540
2.159 2413.166 737.739 0.300 10.922
(P-100)

Bore 2.574 3585.275 399.896 0.900 101.600 4.500

1.1.1. Constituants d’un matériau composite


1.1.1.1. Fibres

Le rôle principal des fibres est de fournir la résistance et la rigidité au matériau composite.
Cependant, en tant que classe, les fibres à haute résistance sont cassantes [9];

6
 Possède un comportement contrainte-déformation linéaire,
 Une faible déformation à la rupture (1-2% pour le carbone),
 Et présentent des variations de résistance plus importantes que les métaux. Le Tableau 1
présente un résumé des principaux types de fibres de renforcement.

Les fibres de verre sont les plus utilisés en raison de leurs rapport coût/propriétés mécaniques ;
environ 90% de composites polymères renforcées de fibres [11]. Les fibres de verre E (ou fibres
de verre électrique), sont les fibres les plus courantes, et elles sont largement utilisées dans les
composites commerciaux [9]. Elles se distinguent par leurs faibles coûts, une densité élevée, un
faible module d’élasticité, ainsi que par leurs résistances à la corrosion, leurs résistivités électriques
et de bonnes caractéristiques de maniabilité [9,12]. Par contre, le verre S (verre structurel) a été
développé pour répondre à des besoins spécifiques. Sa densité, son niveau de performance et son
coût se situent entre ceux du verre E et du carbone. La fibre de quartz est utilisée dans de
nombreuses applications électriques en raison de sa faible constante diélectrique ; cependant, elle
est très cher [9].

La fibre d'aramide (par exemple, le kevlar) est une fibre organique extrêmement résistante, de faible
densité et présente une excellente tolérance aux dommages. Bien qu’elle présente une résistance
élevée à la traction, il fonctionne mal en compression. Elle est également sensible au rayonnement
ultraviolet, et son utilisation est limitée à des applications spécifiques avec des températures
d’opérations inférieures à 180°C.

Une autre fibre organique est fabriquée à partir de polyéthylène de poids moléculaire ultra élevé
(Spectra). En raison de sa faible densité, elle présente une résistance spécifique et un module très
élevés à température ambiante. Cependant, étant un polyéthylène, son utilisation est limitée à des
températures inférieures à 140°C.

La fibre de carbone dispose de la meilleure combinaison de propriétés, mais elle est également plus
chère que le verre ou l'aramide [9]. Elle a une faible densité, un faible coefficient de dilatation
thermique et elle est conducteur [9,11]. Elle est structurellement très efficace et présente une
excellente résistance à la fatigue. Il est également fragile (déformation à la rupture inférieure à 2%)
et présente une faible résistance aux chocs. La résistance à la compression axiale représente 10 à

7
60% de leur résistance à la traction [13,14] et la résistance à la compression transversale représente
12 à 20% de leur résistance à la compression axiale [15]. La fibre de carbone est disponible dans
une large gamme de résistance (2000-7000 MPa) et de rigidité (module entre 200-1000 GPa) [9].
Ces propriétés innées des fibres de carbone les rendent idéales pour des applications dans les
secteurs de l'aérospatiale, de l'électronique et de l'automobile.

Les termes carbone et graphite sont souvent utilisés pour décrire le même matériau. Cependant, les
fibres de carbone contiennent environ 95% du carbone et sont carbonisées à 1000-1500°C, tandis
que les fibres de graphite contiennent environ 99% de carbone et sont d'abord carbonisées puis
graphitisées à des températures comprises entre 2000 et 3000°C [9]. En général, le processus de
graphitisation se traduit par une fibre avec un module plus élevé.

Les fibres de carbone et de graphite sont fabriquées à partir de polyacrylonitrille (PAN) ou de brai
à base de pétrole [9,11]. Le polyacrylonitrille (PAN) est l'un des précurseurs les plus couramment
utilisés dans la production de fibre de carbone, qui offre une résistance à la traction et un module
d’élasticité plus élevés. Selon la température de durcissement finale, différentes classes de fibres
de carbone à savoir les fibres à haute ténacité, les fibres à module intermédiaire, les fibres à haut
module et les fibres à ultra-haut module sont formées avec des précurseurs PAN [11]. Les fibres
de brai à base de pétrole ont été développées en tant qu'alternative à moindre coût au PAN, mais
sont principalement utilisées pour produire des fibres de graphite à module élevé et ultra-élevé, et
elles sont utilisées pour fabriquer des composants de haute rigidité, et qui nécessitent une
conductivité thermique et électriques élevées [9]. Les fibres de carbone et de graphite sont produites
sous forme de faisceaux non torsadés appelés câbles. Les tailles de câble courantes sont lk, 3k, 6k,
12k et 24k, où k = 1000 fibres. Immédiatement après la fabrication, les fibres de carbone et de
graphite sont normalement traitées en surface pour améliorer leur adhérence à la matrice polymère.

Les fibres en général, jouent un rôle crucial dans la décision de la propriété finale des systèmes
composites renforcés de fibres. Ainsi, une sélection appropriée du matériau fibreux et de son
orientation relative peut conduire à des composites avec des propriétés sur mesure pour répondre
aux exigences spécifiques de l'application. Cependant, au cours des études comparatives, il faut
tenir compte des changements des propriétés résultantes de la différence dans les techniques de

8
traitement, de la variation des matériaux précurseurs et des traitements post-fabrication, car tous
ces facteurs influencent les propriétés finales des fibres résultantes [11].

1.1.1.2. Matrices

Les matrices utilisées dans les matériaux composites ont pour rôle de maintenir les renforts dans
une forme compacte, de transférer les sollicitations mécaniques aux renforts et de protéger la
surface des matériaux. En se basant sur l’origine de la matrice, les matériaux composites sont
classés en trois catégories : (1) Composites à matrice polymère, (2) composites à matrice
métallique, (3) et composites à matrice céramique.

Dans le cadre de cette étude, nous nous s’intéresserons aux composites à matrice polymère. Il existe
deux catégorie de polymères : Les thermoplastiques et les thermodurcissables. Les résines
thermoplastiques est un ensemble de macromolécules, qui passe de l’état liquide à l’état fondu sous
l’action de la chaleur de façon réversible. Elles sont constituées de chaines linéaires ou ramifiées
à liaisons covalentes. Ces chaines sont liées entre elles par des liaisons faibles (Van der Waals et
hydrogène).

Par contre, les résines thermodurcissables sont constitués d’une seule macromolécule représentée
par un réseau tridimensionnel infusible dont la rigidité augmente avec la densité. Elles sont
généralement constituées d'une résine et d'un agent de durcissement compatible. Lorsque les deux
sont initialement mélangés, et sous l’effet de la chaleur, les réactions de polymérisation et de
réticulations se produisent, ce qui entraine la formation des chaines réticulés entre-elles. Ces
chaines sont liées par des liaisons fortes (covalentes). Contrairement aux thermoplastiques, cette
transformation est irréversible. Elles sont les plus utilisées pour la mise en œuvre de matériaux
composites. Les résines les plus répandues utilisées sont les polyesters, les esters vinyliques, les
époxydes, les bismaléimides, les polyamides et les composés phénoliques.

Dans le domaine de l’aéronautique, les matrices époxydes sont les plus répandues [16]. Le degré
élevé de réticulation et la nature des liaisons interchaînes confèrent aux époxydes durcis de
nombreuses caractéristiques souhaitables. Ces caractéristiques comprennent une excellente
adhérence à de nombreux substrats, une résistance élevée (traction, compression et flexion), une

9
résistance chimique, une résistance à la fatigue, une résistance à la corrosion et une résistance
électrique très importantes [17]. Les propriétés des matériaux composites dominées par la matrice
(résistance interlaminaire, résistance à la compression) sont réduites lorsque la température de
transition vitreuse est dépassée, et limite donc l'application des plus hautes composites époxy
thermodurcissables à des températures qui ne dépassent pas 120 °C [9].

1.1.1.3. Préimprégné

Les préimprégnés (Figure 3) sont depuis longtemps les éléments de base des structures composites
à hautes performances. Ils sont des produits semi-finaux composé des fibres unidirectionnel ou
sous forme de tissus, d’une résine, de son durcisseur, et un certain nombre d’adjuvants.
L’imprégnation est faite à un rapport résine/fibres bien déterminé. Ils sont destinés à un moulage à
chaud sous vide, et ils permettent d’accéder aux plus haute niveaux de performance et de fiabilité,
notamment pour des applications aéronautiques.

Figure 3 : Préimprégné

Les préimprégnés sont livrés généralement sous forme de rouleaux. Les caractéristiques d’un
préimprégné sont les suivantes [18]:

- Rapport massique fibre/résine,


- Type de résine et de renfort,
- Durée de vie à température ambiante,
- Conditions de stockage,
- Propriétés mécaniques obtenues sur stratifié,

10
- Procédé de mise en œuvre.

Les préimprégnés à matrice thermodurcissable doivent être conservés dans des réfrigérateurs à
basse température pour éviter l'achèvement du durcissement de la résine. Dans ces préimprégnés,
après avoir imprégné les fibres avec la résine thermodurcissable, la réaction de réticulation est
suspendue en maintenant le matériau à des températures inférieures à zéro. De ce fait et
contrairement aux préimprégnés à matrice thermoplastique, les préimprégnés thermodurcissables
présentent toujours une durée de conservation limitée [19].

Comme le rapport massique fibre/résine est bien contrôlé, l’utilisation des préimprégnés amène à
la fabrication d’un composite de haute qualité, plus homogène, et qui présente des propriétés quasi-
constantes sur l’ensemble du matériau.

1.1.1.4. Composites stratifiés

La géométrie largement utilisée pour les composites à fibres continues est le stratifié (Figure 4). Ils
sont des matériaux dans lesquels les couches individuelles, les plis ou les préimprégnés sont
orientés dans des directions qui améliorent la résistance dans la direction de la charge principale.
Comme les fibres présentent des résistances mécaniques élevées par rapport à celles des matrices
(Par exemple, la résistance en traction des fibres est 50-100 fois supérieure à celle de la matrice),
les stratifiés unidirectionnels (Figure 4a), sont extrêmement solides et rigides dans la direction (0°),
et très faibles dans la direction (90°). Cependant, dans cette direction, la charge doit être portée par
la matrice.

Comme l’orientation des fibres présente un impact direct sur les propriétés mécaniques, il
semblerait logique d'orienter autant de couches que possible dans la direction principale de charge.
Bien que cette approche puisse fonctionner pour certaines structures, il est nécessaire d'équilibrer
la capacité de la charge dans un certain nombre de directions différentes, telles que les directions
0°, ±45 et 90 (Figure 4b). Un stratifié équilibré avec un nombre égal de plis dans les directions 0°,
±45 et 90 est appelé stratifié quasi-isotrope, car il porte des charges égales dans les quatre
directions.

11
Figure 4 : Stratifiés unidirectionnels (a) et quasi-isotropes (b) [19]

1.1.2. Procédés d’élaboration

Les procédés d’élaboration des composites à base des préimprégnés sont des procédés par voie
sèche. Ils existent deux procédés de mise en œuvre de pièces de qualité aéronautique : Procédé de
mise sous vide et procédé autoclave.

1.1.2.1. Procédé de mise sous vide

Le procédé de mise sous vide est une méthode qui utilise la pression atmosphérique pour maintenir
en place les déférentes composants d’un matériau stratifié. Le composite stratifié est compacté par
l’intermédiaire d’un sac étanche déposé sur un moule rigide (Figure 5). Le rôle de chaque
composant d’ensachage est :

- Suce à vide : Assurer le niveau de vide au sein du sac ;


- Mastique : Assure l’étanchéité du sac ;

12
- Film perforé : Retient la résine par ses micros pores tout en laissant passer l’air autour si le
vide est maintenu ;
- Breather : Uniformise le vide au sein du sac et absorbe la résine en excès qui passe à travers
le film perforé ;
- Sac de cuisson : Sac à haute résistance thermique qui assure une bonne cuisson du stratifié.

Figure 5 : Composants d'ensachage

Lorsque le sac est scellé au moule, la pression à l'extérieur et à l'intérieur de ce sac est égale à la
pression atmosphérique. Lorsqu'une pompe à vide évacue l'air, la pression de l'air à l'intérieur de
l'enveloppe est réduite tandis que la pression de l'air à l'extérieur de l'enveloppe reste à 1atm. La
pression atmosphérique force les côtés du sac et tout ce qui se trouve à l'intérieur de l'enveloppe,
exerçant une pression égale et uniforme sur la surface de l'enveloppe. La différence de pression
entre l'intérieur et l'extérieur de l'enveloppe détermine la force appliquée sur le stratifié.
Théoriquement, la pression maximale possible qui peut être exercée sur le stratifié, s'il était possible
d'obtenir un vide parfait et d'éliminer tout l'air de l'enveloppe, est d'une atmosphère.

En maintenant le vide, le sac est placé à l'intérieur d'un four ou sur un moule chauffé, et le composite
est produit après application d'un cycle de durcissement à température appropriée.

1.1.2.2. Procédé autoclave

Le traitement en autoclave est généralement la méthode de choix pour fabriquer des composants
composites en plastique renforcé de fibre pour des applications à hautes performances, en

13
particulier de grandes structures complexes [20]. Il permet de fabriquer des composites stratifiés
avec un taux de porosité inférieur à 1% et présentent des propriétés mécaniques élevées [21]. Ce
procédé consiste à mettre le sac sous vide (Figure 5) dans une enceinte (Figure 6) sous des pressions
élevées.

Figure 6 : Four autoclave de polymérisation à Casablanca, Safran Nacelles Maroc

1.2.Effet du cycle de fabrication en autoclave sur les propriétés des composites stratifiés

Les propriétés et les performances des matériaux thermodurcissables renforcée par des fibres
dépendent généralement du pourcentage volumique de la résine thermodurcissable, de sa
composition chimique, du mécanisme et le degré d’avancement de la réaction de réticulation, et
des paramètres du processus de fabrication comme la durée, la température, la pression et le niveau
de vide. Bien que les fabricants des préimprégnés commerciaux suggèrent généralement des cycles
de durcissement pour des applications personnalisées, ces cycles ne peuvent pas être optimaux pour
des applications spéciales.

La détermination et l’optimisation des cycles de durcissement en autoclave se base généralement


sur l’étude de l’évolution de la cinétique de réticulation de la résine, ce qui amène à un contrôle
parfait du degré de réticulation, et de réduire le temps de fabrication. En effet, il est crucial que

14
l’optimisation de la durée de fabrication puisse satisfaire la fraction volumique des fibres, les micro
défauts, et principalement les propriétés mécaniques.

Figure 7 : Différents cycles de durcissement du préimprégné AS4/8552 et l’évolution


correspondante de la viscosité durant le cycle [22]

1.2.1. Optimisation et modélisation du cycle thermique


1.2.1.1. Cycles thermiques de durcissement en autoclave

Hernández et al. [22] ont fabriqué des composites à matrice polymère AS4/8552 par moulage à
compression sous une pression de 2 bar et ceux, à différents cycles thermiques. Tous les cycles

15
thermiques avaient atteint une valeur maximale de température de 180°C à la fin du procédé comme
le montre la Figure 7. Le tableau 1.2 résume les résultats empiriques relatives aux différents cycles
utilisés.

Table 2 : Résultats de propriétés physiques des quatre cycles [22]

Cycle ∆𝐻𝑟𝑒𝑠 (J/g) α Tg (°C) Vv (%) Δd (J/g)


C-1 19,1 0.891 207.6 2.9 973±286
C-2 18.4 0.895 210.4 0.4 1075±375
C-3 18.3 0.896 210.6 1.1 1276±330
C-4 15.7 0.910 211.9 2.7 1131±275

D’après ce tableau, le degré de durcissement (α) et la température de transition vitreuse (Tg) pour
les différents cycles restent pratiquement constant, mais la teneur en vide enregistre un changement
allant de 0,4 à 2,9%. La teneur en vide la plus basse (0,4%) est obtenue pour le C-2 qui présente
deux pentes de chauffage et deux paliers isothermes. Le faible pourcentage en vide est dû au
premier palier isotherme pour lequel, la viscosité était minimale ce qui a permis de maintenir une
long durée d’application du vide.

Généralement, le cycle thermique représenté ci-dessous (Figure 8) est le cycle conventionnel de


durcissement fréquemment rencontré pour les préimprégnés utilisés dans le domaine de
l’aéronautique [23]. Ce cycle thermique de durcissement contient deux pentes de chauffage, deux
paliers isothermes, et finalement, une pente de refroidissement. Durant le premier chauffage, la
viscosité de la résine diminue rapidement jusqu’à atteindre une viscosité minimale sans réticulation
des monomères de la résine. Dans la deuxième partie, premier palier isotherme, la viscosité de la
résine reste dans son état le plus faible et ce, pour une longue durée ce qui permet d’évacuer le
maximum du vide. Durant le cycle suivant (chauffage), la viscosité varie rapidement à cause la
réaction de réticulation qui commence dès les premières instances de chauffage. La réticulation va
se terminer au cours du deuxième palier isotherme avant de procéder au refroidissement du
matériau.

16
Figure 8 : Cycle thermique conventionnel pour les préimprégnés utilisés dans l’aéronautique [23]

En ce qui concerne l’effet de la vitesse de chauffage sur les propriétés du matériau, la Figure 9
représente l’impact de la vitesse de chauffage sur le degré de réticulation et la viscosité. Comme le
montre la figure 9a, après avoir atteint la température de durcissement initiale, la vitesse de
chauffage lente a montré un degré de durcissement légèrement supérieur, mais toujours inférieur à
0,10. Il était logique que l'échantillon soit maintenu pendant un temps plus long à la vitesse de
chauffage lente pendant la première période de montée en température, ce qui a amélioré le degré
de réticulation de la résine. Au cours de la phase de durcissement initiale, une tendance similaire
d'évolution du degré de durcissement a été observée pour les vitesses de chauffage rapides et lentes.
Les résultats suggèrent que la vitesse de chauffage rapide était plus efficace et économisait 35% de
temps de durcissement par rapport à la vitesse de chauffage lente. La résine peut également être
complètement durcie. Indépendamment de la vitesse de chauffage, la viscosité a présenté une
tendance similaire pendant l'étape de chauffage et de durcissement initial (Figure 9b). Les deux
viscosités étaient adaptées à l'infiltration de la résine. Après le premier palier de température, la
résine présentait toujours une viscosité relativement plus faible et il n'y avait aucun signe de
gélification, ce qui a contribué à l'infiltration de résine.

Dong et ses collaborateurs [24] ont étudié l’effet de la température du premier palier et sa durée sur
les propriétés de composites stratifiés. Une différence notable de la porosité a été observée : (1) A
faible température (100°C), la viscosité de la résine n’était suffisamment pas basse pour infiltrer
chaque couche de préimprégné, ce qui a entravé l’évacuation de l’air et conduit à la formation des

17
vides intra-laminaires (3,3%). (2) Au contraire, la viscosité à 140°C était trop faible, mais la
réaction de gélification s’est produite trop vite, ce qui a conduit à une imprégnation de résine
insuffisante. De ce fait, des vides inter- et intra-laminaires ont été observé (2,8%). (3) Lorsque le
composite a été initialement durci à 120°C, il n’y avait aucun vide dans le stratifié, et il présente
des propriétés mécaniques élevées. (4) La durée du palier dépend du processus de gélification. Par
exemple, lorsque la température atteignait 140 ° C, la gélification se produisait après seulement 45
minutes en raison de la réactivité plus élevée de la résine.

Figure 9 : Effet de la vitesse de chauffage sur le degré de durcissement (a) et la viscosité (b) [24]

Vo et al. [25] ont examiné l’effet du deuxième palier de température sur les propriétés physiques
et mécaniques des matériaux composites. Ils ont observé que la porosité ne dépend pas de la
température du deuxième palier, tandis que le degré de durcissement final, la température de
transition vitreuse et la résistance au cisaillement interlaminaire augmentent en augmentant la
température du deuxième palier. Il est aussi préférable de choisir une durée optimale pour ce palier
isotherme de telle sorte que la résine atteindra un durcissement maximal.

En raison des couts élevés du processus de fabrication en autoclave, des chercheurs ont utilisés un
cycle à un seul palier de température (Figure 10) pour réduire la durée de fabrication [26–28].
L’équipe de Koushyar a montré que degré de durcissement et la température de transition vitreuse,

18
ainsi que la résistance au cisaillement interlaminaire diminuent en diminuant la température du
palier [26], par contre, la résistance à la compression ne dépend pas de cette température [28].

Figure 10 : Cycle à un seul palier de température [27]

1.2.1.2. Modélisation du cycle thermique

Sous l’effet de la chaleur, les matériaux thermodurcissables subissent un processus chimique


irréversible appelé réticulation. Ce processus désigne la formation des liaisons covalentes suivant
les trois directions de l’espace. Au cours de la réaction, les matériaux thermodurcissables passent
d’un liquide peu visqueux à un caoutchouc viscoélastique, et en fin, à un solide vitreux insoluble.

La détermination de la cinétique de réticulations est la première étape dans le processus


d’optimisation du cycle de réticulation. Cependant, elle est importante et essentielle à l’échelle
industriel pour la mise en place d’un cycle de durcissement optimal. En particulier, l'étude de la
cinétique de réaction des systèmes thermodurcissables en fonction des conditions de traitement
thermique, d'un point de vue macro-cinétique, est très importante pour l'analyse et la conception
des opérations de traitement. Comme démontré dans la partie précédente, si la température et la
durée de cycle de traitement thermique ne sont pas optimales, la résine peut être moins ou trop

19
durcie ; les résines trop durcies deviennent généralement cassantes et fragiles, tandis que les résines
moins durcies peuvent être caoutchouteux, manquant ainsi de résistance mécanique.

L’optimisation du cycle thermique peut améliorer la rentabilité, soit en réduisant la durée globale
du traitement thermique, soit en modifiant le profil thermique imposée. En optimisant le cycle
thermique, il est également possible d’augmenter l’efficacité globale de production. Par exemple,
un manque de connaissance de la durée de réticulation optimale peut prolonger la durée de la
fabrication du produit, et donc avoir des conséquences économiques.

Dans l’intérêt de la qualité et de la fiabilité des composants, il devient de plus en plus important de
surveiller la réticulation et le durcissement de chaque nouvelle résine avant son application au
niveau industriel afin d’atteindre les conditions de durcissement optimales pour un processus de
fabrication donné. Cette surveillance permet à l’utilisateur de suivre les changements physiques,
chimiques et structurels qui se produisent pendant la réticulation. Ainsi, il permet de détecter l’état
du matériau à chaque instant, y compris toute anomalie pendant le processus de durcissement et
d’assurer la cohérence des propriétés et des performances du produit final.

De point de vue financier, les composites renforcée de fibres de carbone sont relativement chers,
et ne peuvent pas être recyclés en raison de la nature irréversible de la réaction de réticulation de
la résine thermodurcissable. Sans l’aide d’une autre technique d’étude, le produit final ne peut pas
être mis au rebut, ce qui entraine des couts de production élevés. La connaissance de la cinétique
chimique permet de réduire le volume de ferraille en simulant le processus de fabrication, et en
limitant la plage des paramètres de contrôle avant le début de la fabrication. De ce fait, l’étude la
cinétique peut garantir une quantité minimale de déchets pouvant être causés par des fluctuations
accidentelles des paramètres de contrôle pendant la fabrication en autoclave, et la destruction
intentionnelle des matériaux composites afin de vérifier leurs propriétés mécaniques.

Une bonne compréhension de la cinétique de réticulation est donc, l’une des exigences vitales pour
le développement d’un système de surveillance fiable. L’objectif de cette partie, est dans un premier
temps, de décrire les résines époxydes thermodurcissables, et de comprendre le mécanisme de la
réaction de réticulation des résines époxydes par des durcisseurs amines. Dans un second temps,
cette partie abordera les différents modèles cinétiques existant dans la littérature, pour prédire le
degré d’avancement de cette réaction au cours du temps et en fonction de la température.
20
(i) Mécanisme de réticulation époxy amine

Généralement, l’ouverture du cycle de l’époxyde pourra être réalisée soit par homopolymérisation
ou par addition d’un agent de réticulation coactif (Polyaddition). L’homopolymérisation est une
réaction d’élimination de l’oxygène du cycle époxy en utilisant des catalyseurs acides ou basiques
et la réaction est souvent activée par rayonnement. La deuxième méthode, est celui qui nous
intéresse dans notre étude, s’agit d’une attaque nucléophile de l’un des carbones formant le cycle
par une amine ou un composé anhydre. Les propriétés finales du réseau époxyde dépend de l’agent
de réticulation [17]. Lorsque le durcisseur est une amine aliphatique, le processus de durcissement
se produit fréquemment à température ambiante, mais il est lent et incomplet. Par conséquent, les
réseaux obtenus présentent de basses températures de transition vitreuse (Tg) et une grande capacité
à carbonater et à absorber l'eau. D'un autre côté, les résines époxydes durcie avec des amines
aromatiques présentent généralement une bonne résistance thermique et chimique, tandis que les
avantages des durcisseurs anhydrides sont leur faible contraction, leur viscosité et leur excellente
résistance thermique.

La réaction de polyaddition pourrait s’effectuer en deux, trois ou quatre étapes en fonction des
proportions des réactifs [17,29,30]. La Figure 11 illustre le mécanisme réactionnel de
polymérisation/réticulation de la résine époxyde par l’amine.

La réaction de l’amine primaire du durcisseur avec l’époxyde pour former une amine secondaire
(réaction 1), qui réagit à son tour avec un autre époxyde pour former une amine ternaire (réaction
2) sont les principales réactions chimiques qui se produisent. Le groupe amine tertiaire exerce un
effet catalytique et provoque l’auto-polymérisation du groupe époxyde pour former un polyéther
(réaction 3). Bien que la basicité des amines primaires et différente de celle des amines secondaires,
la constante de vitesse (k) de la réaction 1 est différente de celle de la réaction 2. Si on suppose que
k1 = k2, des branches seront formés dès le début de la réaction. Par contre, si k1>k2, il y a d’abord
une formation d’une chaine linéaire avant de procéder à la formation des branches [31]. Il est à
noter que les deux hydrogènes des amines primaires ont initialement la même réactivité, mais
lorsque l’un d’entre eux se met à réagir (Réaction 1), le deuxième hydrogène (amine secondaire)
devient mois réactive).

21
Le groupement hydroxyle formé lors de l’ouverture de cycle époxy joue un rôle catalytique
important dans la cinétique du processus de durcissement, en fournissant une viscosité plus élevée
qui dépend de sa concentration [30,32]. Or, les groupes hydroxyles accélèrent la réaction époxy
amine, ce qui entraine un comportement autocatalytique. L’homopolymérisation des résines époxy
et l’éthérification entre les groupes époxy et hydroxyle (Équation 4) sont d’autres réactions
possibles. Cependant, l’homopolymérisation des groupes époxy est généralement considérée
négligeable en l’absence d’un catalyseur acide ou basique [30].

Figure 11 : Schéma des réactions époxy/amine (1-3) et d’éthérification (4) pour les amines
aliphatiques [30]

En outre, la réaction d’éthérification est généralement beaucoup plus lente que les réactions époxy-
amine et ne devient significative que s’il y a excès de groupements époxy lorsque l’amine primaire
est suffisamment appauvri [33]. Si les constituants du préimprégné contiennent plusieurs groupes
réactifs (oxirane ou amine), l’addition se fait aux extrémités de la molécule ramifiée, ainsi qu’avec
des monomères frais. Pendant la réaction chimique, le poids moléculaire et la polydispersité

22
augmentent jusqu’à ce qu’une seule molécule soit formée, le système devient un gel. Les diamines
communes avec un volume moléculaire faible, agissent comme point de réticulation [32].

(ii) Cinétique de la réaction de réticulation

Les réactions chimiques qui ont lieu pendant le durcissement déterminent la structure moléculaire
des matériaux thermodurcissables qui, à leur tour, détermine les propriétés physiques, électriques
et mécaniques du matériau composite. Les réactions de réticulation des résines thermodurcissables
tel que les époxydes, sont parmi les réactions les plus difficile à suivre. Cela est dû au fait que :

i. Les matériaux thermodurcissables commencent généralement par des résines, qui se


réticulent ensuite pour former des matériaux vitreux insolubles et rigides. Par
conséquent, la cinétique de réticulation est compliquée par l'interaction relative entre la
cinétique chimique et les propriétés physiques changeantes.
ii. Les réactions de réticulations des matériaux thermodurcissables est exothermique, et
comme ces polymères présentent une faible conductivité thermique, la température et
la vitesse de la réaction peuvent varier considérablement en fonction de la masse de la
résine.
iii. Les systèmes époxydes-amines subissent généralement plus d’un type de réaction.
iv. Les préimprégnés sont à base d’au moins deux types de résines époxydes et deux
durcisseurs amines, empêchant la description de la réaction et la détermination de la
structure du matériau final. De même, les fondateurs des préimprégnés ne divulguent
pas d’informations suffisantes, ni de résine pure pour une étude approfondie.
v. Comme nous le verrons dans les chapitres suivants, dans les formulations industrielles,
une variété d’additifs sont inclus dans le système pour assurer une affinité chimique
entre les fibres de carbone et la résine, résultant en une cinétique de réticulation
complexe.
vi. De plus, après le processus de polymérisation/réticulation, les polymères sont
polydispersés ou hétérogènes en poids moléculaire.

23
L’exigence globale des matériaux thermodurcissables après réticulation, est que les propriétés
physiques, chimiques et techniques restent dans les limites spécifiées. Par conséquent, il est
nécessaire de mieux comprendre le processus de réticulation et l'évolution de la relation structure-
propriété du matériau final, afin de contrôler économiquement la réticulation et d'optimiser les
propriétés physiques des produits finaux.

(iii)Perspectives

La compréhension du mécanisme et de la vitesse de la réaction (cinétique) est la première étape


essentielle du processus d’évaluation de la relation traitement-morphologie-propriétés-durabilité
des résines thermodurcissables et de leurs composites. La connaissance du degré de réticulation et
de la façon dont ce degré varie avec la vitesse de chauffe, la température et le temps est utile et
important de plusieurs points de vue :

i. Tout d'abord, de s'assurer que la résine est traitée thermiquement de manière optimale,
et compatible avec son utilisation finale. L'analyse cinétique aide le scientifique,
l'ingénieur ou le responsable du contrôle des processus à décider des conditions
température-temps les plus efficaces pour la génération d'un composant fabriqué à partir
d'une résine thermodurcissable qui a les propriétés optimales pour toutes les
applications souhaitées.
ii. Le degré de réticulation définit la durée de conservation du matériau pour un ensemble
donné de conditions de stockage et sa durée de vie en contrôlant ses propriétés
d'adhésivité, de mouillage et d'écoulement.
iii. Les informations cinétiques sont utiles pour équilibrer la consommation d'énergie par
rapport à l'intégralité du durcissement d'un produit thermodurcissable.
iv. Pour un préimprégné commercial, le cycle de durcissement recommandé par les
fabricants donnera, en général, un matériau durci adéquatement. Cependant, il est
souvent souhaité de durcir le préimprégné dans un ensemble de conditions différentes.
Cela se produit, par exemple, lorsque plusieurs matériaux d'un assemblage doivent être
durcis simultanément ou si des composants sensibles à la température doivent être liés.
La cinétique de durcissement sous la forme d'une loi de vitesse phénoménologique
24
(discutée dans les parties suivantes) permet d'optimiser le cycle de durcissement pour
une situation de drapage donnée. Si, en outre, des corrélations entre le degré de
réticulation et la force de cohésion du préimprégné ont été établies, elles permettent de
s'assurer que le joint adhésif n'a pas été compromis par les changements de cycle de
durcissement.

(iv) Cinétique chimique et diffusion

Pour qu'une réaction chimique puisse se produire, deux molécules doivent diffuser des distances
comparables aux distances intermoléculaires avant que leurs sites réactifs puissent entrer en
collision et finalement se lier. Par conséquent, la cinétique de la réaction est sensible à la fois à la
vitesse intrinsèque du processus de réaction chimique entre les molécules en contact et aux
limitations de transport des molécules portant les sites réactifs. Si les réactifs peuvent
considérablement diffuser, la vitesse de la réaction ne dépend que des concentrations et des
réactivités intrinsèques des sites réactifs, avec des constantes de vitesse indépendantes de l'étendue
de la réaction (réaction contrôlée chimiquement). Si le transport des réactifs est limité par un faible
coefficient de diffusion, la vitesse de réaction est contrôlée principalement par la probabilité de
rencontre entre les paires réactives (réaction contrôlée par la diffusion). Il arrive souvent qu'une
transition d'une réaction contrôlée chimiquement à une réaction contrôlée par diffusion se produise
progressivement à mesure que le produit de la réaction diminue les diffusivités dans le système.
C’est le cas de plusieurs systèmes époxydes-amines [20,29,33–37].

Durant le processus de durcissement, les monomères sont combinées pour former un réseau rigide
tridimensionnel. Dans l’état initiale, les monomères sont à l’état fluide (faible viscosité) et peuvent
se déplacer librement [37]. En effet, la vitesse de la réaction chimique est plus lente que la diffusion
moléculaire, et donc c’est cette vitesse qui détermine la vitesse à laquelle la structure du réseau se
développe. Au cours des derniers stades de réticulation, la mobilité moléculaire diminue du fait
que, la structure de plus en plus réticulée peut entraver le mouvement des réactifs : La vitrification
peut se produire si la Tg est supérieur à la température de traitement [34]. Le système se transforme
donc en un état vitreux. Au fur et à mesure que le système se transforme en un état vitreux
(augmentation de la viscosité du système), la vitesse de la réaction diminue significativement, et la

25
réaction devient contrôlée par la diffusion des réactifs. Bien que le passage d'une cinétique
chimique à une réaction contrôlée par diffusion ne se produise pas ni à la gélification ni à la
vitrification [37,38].

Fondamentalement, dans la dernière étape de la réticulation, la réaction n'est plus longtemps


associée à la migration de petites molécules, mais elle implique des déplacements de molécules
longues ramifiées et une mobilité segmentaire du réseau. Par conséquent, les entraves stériques et
la diffusion segmentaire lente abaissent la vitesse de réticulation et conduisent à une réaction
contrôlée par la diffusion. De plus, la mobilité segmentaire est considérablement diminuée par la
vitrification.

(v) Caractérisation cinétique par DSC

Dans la littérature, plusieurs méthodes ont été utilisées pour étudier le processus de réticulation des
résines et des préimprégnés époxydes. La DSC est la technique fréquemment la plus utilisée
[20,28,29,33,34,39–45]. Dans cette technique, un échantillon est chauffé à une vitesse constante
(balayage dynamique) ou à une température particulière (balayage isotherme), et le flux de chaleur
vers ou depuis le matériau est mesuré. Étant donné que la réaction de réticulation dans les
polymères thermodurcissables est exothermique et génère de la chaleur, l'achèvement de la
réticulation totale est alors indiqué par le manque de génération de chaleur.

L'objectif de l'étude cinétique d'un thermodurcissable de durcissement est, en général, basé sur la
construction d'un modèle cinétique, phénoménologique ou mécanique, qui peut prédire les
caractéristiques du système : la vitesse de la réaction et le degré de réticulation pour un profil temps-
température donné. Par conséquent, de point de vue de la modélisation mathématique, la corrélation
et la représentation des données expérimentales sous la forme d'une simple expression cinétique
est d'une importance considérable. À cet égard, la DSC a été largement utilisée comme moyen
pratique pour évaluer les équations de vitesse et pour estimer les paramètres cinétiques associé.

26
1.2.2. Effet de la pression et du vide sur les propriétés des composites stratifiés

Généralement, la pression est appliquée pour : (1) compacter le composite, (2) contrôler la fraction
volumique de fibre souhaitée, (3) et réduire les vides qui peuvent se développer pendant le
processus de cuisson de la résine. Le vide permet d’éliminer tout air emprisonné pendant le
drapage, et les substances volatiles formées pendant la réaction de polymérisation/réticulation de
la résine. La pression du vide doit être inférieure à la pression de l’autoclave pour assurer le
compactage du stratifié [46]. Le cycle de pression représenté dans la Figure 12 est le cycle de
pression conventionnel de durcissement fréquemment utilisé dans le procédé autoclave.

Figure 12 : Cycle de pression conventionnel utilisé dans le procédé autoclave [23]

Lors de la première pente de chauffage et le premier palier isotherme, le matériau est uniquement
sous vide. La mise sous vide accompagnée d’une montée de température abaisse la viscosité de la
résine qui migre entre les plis du stratifié. Une pression supplémentaire de l'autoclave aurait un
effet négatif, piégeant l'air et autres volatils [23]. Avec le début de la deuxième rampe de chauffage,
la pression dans l'autoclave augmente tandis que le vide dans le sac à vide diminue ; l’augmentation
de la pression évacuera l’excès de la résine, il s’en suit un gel rapide de la résine qui achève sa
polymérisation/réticulation. Or, le transfert de la pression de l’autoclave à la résine contenue dans
le composite stratifié ne se produit pas d’une manière hydrostatique, car la résine n’est pas enfermée
dans un système à volume constant. L’écoulement peut se produire en premier temps verticalement
(direction de l’épaisseur) et horizontalement. En outre, le sac sous vide et l’ensemble de purge qui
27
transfère la charge vers le réseau de fibres qui peut servir à un réseau de ressorts. Cette charge est
ensuite transportée à travers ce dernier à la surface avec une légère perte par frottement. Lorsque
le réseau de fibres est chargé et comprimé, la résine s’écoule davantage. A ce stade, la viscosité
augmente et la perméabilité décroit, ce qui rend l’écoulement de la résine plus difficile et entraine
un niveau de pression élevé dans tout le stratifié [47].

1.2.2.1. Effet du vide appliqué

L’utilisation d’un système d’ensachage sous vide permet de réduire la quantité du vide pigée dans
la résine. Par ailleurs, ce procédé souffre de deux facteurs : Le premier facteur concerne la pression
partielle des matières volatiles dans le système de résine (eau) et le deuxième, c’est le changement
de viscosité durant le processus de durcissement [48].

Durant l’application du vide, les bulles d’air sont physiquement transportées hors du système de la
résine. Pour cette raison, l’élimination des vides dépend à la fois de la viscosité de la résine et du
réseau de fibres. A des faibles viscosités, il s’est avéré que l’évacuation des bulles d’air est aisée.
En effet, durant le processus de chauffage, la viscosité de la résine va diminuer avec la température
pour atteindre un minima (avant la gélification). En continuant le chauffage, le système atteindra
la gélification avant que la viscosité commencera à nouveau sa croissance due aux phénomènes de
réticulation forte (voir la Figure 8). Durant ces cycles thermiques, il existe une période pour laquelle
l’opération d’évacuation du vide est très efficace [48]. En effet, l’évacuation du vide ne devrait pas
être réalisée durant tout le cycle. Une évacuation excessive peut même conduire à une augmentation
réelle de la teneur en vides [23,48]. Un très petit nombre d’études ont traité l’effet de l’intensité du
vide appliqué sur les propriétés des matériaux composites [26,49,50].

Sudarisman et al. [50] ont étudié l'influence de cinq pressions de vide différentes sur la
microstructure des composites polymères renforcés par des fibres de carbone unidirectionnels. Ils
ont constaté que l'augmentation de la pression de vide de -0,15 à -0,20 MPa induisait une
diminution d'environ 5,3% de la fraction volumique de fibre. Cependant, l'augmentation
supplémentaire du niveau de vide de -0,20 à -0,30 MPa a produit une fraction volumique de fibre
d'environ 58%. Cette fraction volumique reste constante malgré l’augmentation du niveau du vide

28
(-0,65 MPa). Boey et al. [49] ont étudié l'influence du vide d'ensachage sur la teneur en vides, le
module et la résistance à la flexion. Ils ont découvert que l'augmentation du vide d'ensachage
conduisait à une réduction supplémentaire significative de la teneur en vides et à une amélioration
de la contrainte et le module de flexion. Koushyar et al. [26] ont exploré la variation de la
microstructure, des propriétés thermiques et mécaniques sous deux pressions de vide et durées
différentes, comprenant le vide recommandé et un vide maintenu tout au long du cycle de
durcissement. La variation de la durée de l'application sous vide a un effet minimal sur la
température de transition vitreuse et le degré de durcissement. Au contraire, la teneur en vides s'est
avérée être augmentée de manière significative, tandis que la densité diminuait lorsque le vide était
maintenu tout au long du cycle.

Figure 13 : Corrélation entre la pression de l'autoclave et la teneur en vide pour un composite


stratifié fibres de carbone/époxy [59]

1.2.2.2. Effet de la pression appliquée

Généralement dans la littérature, la pression de durcissement est utilisée pour créer différents
niveaux de vides dans les stratifiés [26,51–54] dans une tentative de trouver une relation entre les
propriétés mécaniques et la porosité [26,53,55–57] et/ou la pression [26,57].

29
Ces études ont confirmé qu'une diminution de la pression de durcissement augmente la porosité,
donnant à son tour des propriétés mécaniques réduites. Liu et al. [51] ont observé que la teneur en
vide diminue en augmentant la pression ; Elle passe de 3,2% à 0,6% lorsque la pression appliquée
passe de zéro à 0,6 MPa, tandis que la résistance au cisaillement interlaminaire à température
ambiante a augmenté de 18%. Costa et al. [55] ont rapporté une tendance similaire pour la
résistance au cisaillement interlaminaire des stratifiés, où une diminution de 34% est observée
lorsque la teneur en vides passe de 0,55% à 5,60%. Di landro et al. [58] ont remarqué une baisse
de la résistance au cisaillement interlaminaire de 25% (porosité de 7%) par rapport aux composites
stratifiés fabriqués sans vides.

Les travaux d’olivier et al. [59] ont trouvé que quel que soit les conditions de pression, le degré de
durcissement reste pratiquement constant entre 0,97 et 0,98, alors que la température de transition
vitreuse varie de 155°C à 160°C. Après le point de gélification, la viscosité de la résine augmente
rapidement et toute augmentation de la pression n’a aucun effet sur la diminution de la teneur en
vide. Pour les cycles de durcissement dans lesquels la pression a été appliqué durant les premières
instances, ils ont observé une diminution de la teneur en vide, lorsque la pression de l’autoclave
passe de 0,1 à 0,7 MPa, mais peu de changement a été observé entre 0,7 et 1 MPa ce qui est exprimé
dans la Figure 13.

Figure 14 : (a) Variation de la viscosité en fonction du temps, (b) Mesure de la teneur en vide en
fonction du premier instant d’application de la pression [51]

En ce qui concerne le moment de l’application de la pression et son effet sur les propriétés des
composites stratifiés, Liu et al. [51] ont cherché le moment optimal pour appliquer la pression.
30
Pour atteindre cet objectif, ils ont augmenté la pression de l’autoclave dans la plage de viscosité
minimale, après différents temps de séjour dans le premier palier de température : 30 (cycle a), 60
(cycle b), 90 (cycle c) et 120min (cycle d) (Figure 14a). La Figure 14b montre les résultats du temps
de l’application de la pression en fonction du pourcentage de vides cumulé.

1.2.3. Conclusions

Le principal défi du procédé autoclave est de déterminer le cycle de cuisson optimal, qui produira
une pièce entièrement durcie, sans vide et sans distorsion dans les plus brefs délais et de manière
plus économique. Comme le comportement chimique et rhéologique est régi par le cycle de la
température, la qualité du matériau est régie entre autre par le processus de compactage qui consiste
à exercer un cycle de pression et de vide en même temps que celui de la température. Afin de
fabriquer un matériau de haute qualité, trois conditions doivent être réunies :

- Durcissement parfaite du matériau stratifié en appliquant la pression au stade de viscosité


minimum afin de chasser l’excès en résine. Ceci conduit à augmenter le pourcentage
volumique des fibres, et de réduire la porosité [26,44,59].
- Le degré de réticulation de la résine doit être proche à 1. Bien que la réaction de réticulation
est exothermique, elle peut causer une dégradation thermique dans le composite lors de la
fabrication [60].
- La durée doit être minimale à cause des couts élevés de l’autoclave. La réduction de la
durée de fabrication entraine un bénéfice assez important dans la production massique.

Plusieurs points sont à prendre en considération dans le procédé autoclave :

- Un vide trop élever peut entrainer un dégazage de la résine (évaporation ==> présence de
zones sèches ==> porosité) ;
- Une pression prématurée implique une évacuation trop intense de la résine, et par
conséquent l’augmentation de la fraction volumique des fibres ce qui peut aboutir à la
présence des zones sèches.
- Une faible température entraine un degré de réticulation faible, tandis qu’une température
élevée provoque un dégazage de la résine et par conséquent, la présence de zones sèches.

31
Bien que la pression et la température sont contrôlées suivant un cycle, il existe trois modèles de
contrôle en autoclave [61]:

- Le modèle empirique est uniquement basé sur la fabrication d’un grand nombre
d’échantillons suivant des cycles différents, et d’analyser les propriétés physiques et
mécaniques des matériaux obtenus.
- Le modèle analytique est basé sur les lois théoriques de transport, permettant d'évaluer
l'évolution de la température et de la pression en fonction des propriétés voulues.
- Le modèle expert est une combinaison des deux modèles précédents et fait appel à une
régulation automatique des valeurs actuelles de température et de pression en fonction des
erreurs et écarts de mesures entre les propriétés actuelles (expériences) et celles attendues.
C'est un modèle auto régulant.

1.3.Porosité au cours du cycle de fabrication en autoclave

Les défauts peuvent être classés, en fonction de leur emplacement, en défauts de matrice, de fibre
et d'interface [62] :

- Les défauts de fibre comprennent l'ondulation et le désalignement des fibres et les fibres
cassées (en raison de la courbure des fibres pendant la fabrication, de la friction dans la
machine textile, …),
- Les défauts d'interface comprennent la déliaison initiale fibre/matrice et la délamination
interlaminaire,
- Et les défauts de matrice incluent la réticulation incomplète de la matrice et les vides.

Les vides désignent les régions non remplies ni de résine, ni de fibre ; Ils sont l’un des défauts les
plus importants. L’importance du taux de vides est due à son effet considérable sur une large
gamme de propriétés, ainsi qu’à leurs fortes probabilités de formation. Bien que le taux de vide a
été identifié dans de nombreuses études comme un paramètre influençant les propriétés
mécaniques, il a été prouvé qu'une analyse précise des effets des vides doit également tenir compte
d'autres caractéristiques des vides telles que leurs formes, leurs tailles et leurs emplacements.

32
Pour mieux comprendre les effets des vides, il faut évaluer leurs caractéristiques en correspondance
avec leur formation. L'évaluation des effets des vides a commencé dans les années 1960, mais ce
n'est que dans les années 1980 que les chercheurs ont commencé une analyse systématique de la
formation des vides [63]. Cela a plusieurs raisons notables : Premièrement, la formation et les effets
des vides ne sont pas encore complètement compris. Deuxièmement, les techniques de fabrication
modernes telles que le durcissement hors autoclave qui visent des coûts de production et une durée
de fabrication faible avec une grande précision sont confrontées au vide comme l'un de leurs
principaux problèmes. Troisièmement, avec la viscosité accrue des polymères modifiés et la
complexité plus élevée des pièces dans les structures composites récentes, l'évacuation des vides
devient plus difficile [64]. Enfin et surtout, la production d'un composite sans défaut est très
coûteuse alors qu'elle n'est souvent pas nécessaire. Cela nécessite que le processus de fabrication
soit quantifié avec des paramètres qui peuvent être modifiés pour minimiser le coût, tout en
répondant aux exigences mécaniques.

1.3.1. Origine et développement des vides

La formation des vides dans les composites polymères renforcés par les fibres à base des
préimprégnés, est principalement étudiée en mettant l’accent sur les étapes et le cycle de
fabrication. Bien que l’étape d’imprégnation est généralement effectuée par le fournisseur du
matériau avant la fabrication du composite stratifié, la formation de vides à ce stade est moins
discuté dans la littérature.

Les mécanismes de formation et de croissance de vides pendant le processus de fabrication en


autoclave ne sont pas encore entièrement connus [63,65]. Les études disponibles se concentrent
principalement sur la compréhension de la physique générale, ainsi sur les corrélations entre le taux
de vides dans le matériau final et les paramètres de fabrication. Les principales sources de vides
sont : l’air emprisonné pendant l’imprégnation (vide intra-laminaire) ou pendant le drapage (vide
interlaminaire) [62], les substances volatiles formées lors de la réaction de réticulation de la résine,
et l’humidité dissoute dans la résine [47,65,66].

33
L’air emprisonné lors de l’étape de drapage, et le niveau de substances volatiles sont les facteurs
qui peuvent être contrôlés dans la production des composites stratifiés à base des préimprégnés,
afin de réduire la porosité du matériau final. Les facteurs déterminants dans la fabrication de pièces
composites sont le stockage et le conditionnement des préimprégnés, le processus de drapage
(contrôle du niveau de piégeage entre les plis) et les conditions de durcissement.

Pour permettre une analyse systématique de leurs effets, les vides sont généralement introduits dans
le composite de manière contrôlée. Les vides artificiels peuvent être créés par:

- L’application d’une pression ou d’un vide inapproprié [26,51,53,59,67–70],


- Profil de température incorrect [57,71],
- Introduction de l’humidité entre les plis lors du drapage[67].

1.3.2. Influence de la porosité sur les propriétés mécaniques

L'intérêt principal de l'analyse des vides dans les composites structuraux est la quantification de
leurs effets sur le comportement mécanique. Les études ont montré que, le taux de vides influence
le comportement mécanique, en particulier les propriétés dominées par la matrice telles que la
résistance au cisaillement interlaminaire, la résistance à la compression longitudinale et la
résistance à la traction transversale.

Des méthodologies analytiques, statistiques et plus récemment numériques ont été utilisées dans
les études sur les effets des vides, tandis que les travaux expérimentaux apportent la plus grande
contribution dans ce domaine. De nombreuses études ont corrélé la dégradation des propriétés
mécaniques avec l'augmentation de la teneur en vides, et certaines d'entre elles ont signalé une
teneur en vides critique [55,67,72], en dessous de laquelle le vide a un effet négligeable sur les
performances. Cependant, dans un certain nombre d'études [59], il a été observé que la teneur en
vide n'est pas suffisant pour tirer des conclusions précises ; Il est nécessaire de prendre en
considération leurs formes, leurs tailles, leurs emplacements et leurs distributions.

Dans la littérature, la propriété la plus étudiée est la résistance au cisaillement interlaminaire. Ceci
est simplement dû au fait que le cisaillement interlaminaire est dominé par les propriétés de la

34
matrice et de l'interface fibre/matrice, et qu'il est donc sensible à la présence de vides. Les études
explorant l'effet des vides sur la résistance au cisaillement interlaminaire sont résumées dans le
Tableau 3. Cependant, une augmentation du taux de vides entraine la diminution de la résistance
au cisaillement interlaminaire.

35
Table 3 : Effet de la porosité sur la résistance au cisaillement interlaminaire (RCIL) pour les composites époxy/fibres de carbone

Structure du Drapage Porosité(%) Observations Référence


PI
0.4 – 4.0 Réduction de RCIL avec l’augmentation du coefficient [73]
Unidirectionnel
d’atténuation (équivalent au taux de vides).
0 – 10.0 Importance de la taille et de l’emplacement des vides, en [59]
Unidirectionnel
plus de son pourcentage volumique.
0 – 3.5 Chaque augmentation de 1% de la porosité concorde avec [51]
Unidirectionnel
une réduction de 6% de RCIL.
0.6 – 12.3 Importance de l’interface fibre/matrice lors de la présence [74]
Unidirectionnel
des vides dans la détermination de RCIL
Unidirectionnel
0.3 – 10.3 Proposition d’une équation qui définit la variation de RCIL [75]
Unidirectionnel
en fonction de la porosité.
0.4 – 2.9 Proposer un modèle de section nette simple pour corréler la [71]
Unidirectionnel réduction de RCIL avec l'augmentation de Vv, en tenant
compte de la morphologie des pores.
[0/902 /0]4𝑠 0.2 – 27.3 Ajustement de la réduction de RCIL sous l’effet des vides [26]
avec une équation exponentielle; Importance des vides
dans la propagation des fissures à travers les plis.

36
0-2 La présence de grands vides dans des endroits critiques est [70]
Unidirectionnel le responsable sur la réduction de RCIL, et non pas le taux
de vides.
0.8 – 3.4 En cas de taux de vides similaire, la dégradation est plus [76]
Unidirectionnel sévère dans le composite avec peu de grands vides que dans
celui avec beaucoup de petits vides.
0 – 5.1 Même tendance de réduction de RCIL avec Vv que pour les [77]
Non applicable stratifiés UD, mais sensibilité plus faible: réduction de 25%
avec une augmentation de 4% du vide.
[(±45)4 /(0,90) 0.2 – 8.0 Plus grande dispersion dans RCIL pour des contenus vides [53]
/(±45)2 ]𝑠 plus élevés.
Tissu [(±45)/04 /(0,90)/02 ]𝑠

[(0,90)/(±45)/(0,90) 0-8 Réduction de 25% avec une augmentation de 6% du taux [58]


/(0,90) de vides.
/(±45)
/(0,90)]𝑛

37
CHAPITRE 2

THE ISOTHERMAL CURING KINETICS OF A NEW


CARBON FIBER/EPOXY RESIN AND THE PHYSICAL
PROPERTIES OF ITS AUTOCLAVED COMPOSITE
LAMINATES

2.1.Avant-propos

Dans un premier temps, il est nécessaire de déterminer la cinétique de la réaction de


polymérisation/réticulation de la résine époxyde, et de valider un modèle cinétique permettant de
décrire son comportement dans un cycle de température et de temps. Par la suite, il est important
de stimuler les cycles thermiques de fabrication en autoclave en utilisant une technique simple,
dans ce cas il s’agit de la DSC, afin de démonter le temps minimal de maintien à un palier de
température, et bien évidement de déterminer l’évolution du degré de réticulation au cours du
temps. Il était par la suite important d’évaluer l’effet de la température sur les propriétés
microstructurales et physiques et de déterminer les conditions permettant de fabriquer un matériau
avec des propriétés optimales.

2.2.Résumé

Cette étude examine les effets de différentes températures isothermes sur le comportement
thermique et les propriétés physiques (degré de polymérisation, la densité, la fraction volumique
des fibres et la teneur en vide) des composites stratifiés carbone/époxy quasi isotropes. Tout

39
d'abord, la cinétique de durcissement d'un nouveau préimprégné commercial, Hexply®8552S, a
été étudiée par calorimétrie isotherme à balayage différentielle (DSC). Le modèle d'ordre n, le
modèle autocatalytique et le modèle de Kamal & Sourour ont été utilisés pour déterminer les
paramètres cinétiques de durcissement qui s'adaptent aux données expérimentales. Les résultats ont
montré que le modèle phénoménologique de Kamal & Sourour était le plus approprié, comme
l'indiquent les valeurs relativement élevées du R2 ajusté. Sur la base de ce résultas, des composites
stratifiés ont été fabriqués en autoclave à différentes températures isothermes (160, 180, 190 et
205°C). Les résultats ont montré que la température isotherme influençait de manière significative
le degré final de polymérisation, la densité et la teneur en vides. Cependant, un degré de
polymérisation et une densité optimales et la faible teneur en vides ont été obtenues à 180 °C.

Mots-clés : préimprégné époxy, composites stratifiés, cinétique de polymérisation, réticulation,


microstructure, propriétés physiques.

2.3.Abstract

This study investigated the effects of different isotherm temperatures on the thermal behavior and
physical properties (i.e., degree of cure, density, fiber volume fraction, and void content) of quasi-
isotropic carbon fiber reinforced polymer composite laminates. First, the curing kinetics of a new
commercial prepreg, Hexply®8552S, was investigated by isothermal differential scanning
calorimetry (DSC). The nth-order, autocatalytic, and Kamal & Sourour models were used to
determine curing kinetic parameters that fit the experimental data. The results showed that the
phenomenological model of Kamal & Sourour was the most appropriate, as indicated by the
relatively high values of the adjusted R square. Based on the kinetics results, autoclave processing
was used to produce laminated composites. The isothermal curing temperatures were 160, 180,
190, and 205°C, respectively. The results showed that the isotherm temperature significantly
influenced the final degree of cure, density, and void content. At 180 °C, the highest values of
degree of cure and density were obtained, while the lowest values of void content were remarqued.

Keywords: Epoxy prepreg, composite laminates, curing kinetics, crosslinking, microstructure,


physical properties.

40
2.4.Introduction

Carbon/epoxy composite laminates offer advantages such as high static fracture resistance,
excellent fatigue strength, high mechanical strength, electrical conductivity, and high operating
temperature [4]. As a result, these materials are widely used in many fields (marine, automotive,
medical, sports, …), particularly in the aerospace industry [63]. The final properties of a composite
material depend on the structure, proportion, and distribution of its three main constituents: the
polymer matrix, the carbon fibers, and the voids. Voids occur either from moisture dissolved in the
resin [78] or from air bubbles trapped in the matrix during autoclave processing [79]. However, the
voids correlate directly with the mechanical properties of the laminated composites, particularly
the interlaminar shear strength, compressive modulus, and laminate compressive modulus [80].

The quality requirements of today's aerospace industry are stringent. Moreover, there is an
immediate need to enhance the efficiency and cost-effectiveness of aircraft structural systems while
providing reliable and consistent processing methods. Notwithstanding the development of new
processes, the technique used in the aeronautical industry remains the autoclave process. However,
the autoclave designer must adequately consider various criteria and parameters involved in the
processing and developing of autoclave systems. Before autoclaving the composites, it is
imperative to determine the curing kinetics of the resin matrix. Understanding the mechanism and
curing kinetics is the essential first step in the process of evaluating the processing-morphology-
properties-durability relationship of thermoset resins and their composites. From several
perspectives, knowledge of the degree of curing and how this degree varies with temperature and
time is valuable and essential: (i) First, ensure that the resin is optimally heat-treated and compatible
with its end-use; (ii) the degree of curing defines the material's shelf life for a given set of storage
conditions and its life span by controlling its tackiness, wetting, and flow properties; and (iii)
kinetic information helps balance energy consumption against the overall cure of a thermoset
product.

Currently, the curing kinetics of resin matrix composites is mainly based on the mechanism and
phenomenological methods [81]. There are various methods for studying the curing kinetics of
epoxy systems. Differential scanning calorimetry (DSC) was a successful method to investigate
41
exothermic curing reactions [39,44,45]. Through DSC, kinetic analysis can be performed by both
isothermal and dynamic methods. Using isothermal DSC analysis, it is evident that the reaction
rate and the extent of curing are monitored simultaneously during the reaction. The kinetic
parameters are calculated using a simple method, although it is time-consuming [42]. Kashani et
al. [82] investigated the effect of isotherm curing temperatures on the viscoelastic properties of
two commercial plain-weave carbon fiber prepregs and reported that the degree of cure, glass
temperature transition, storage modulus, vitrification time, and gel time depends on the isotherm
temperature. Lee et al. [83] used different isothermal cure temperatures to fabricate laminates with
different curing levels, ranging from 0.6 to 0.96, and observed that the laminate compressive
strength was improved by increasing the degree of cure and that the compressive strength modulus,
tensile strength, and tensile modulus remained unchanged. The curing kinetics will also
significantly influence microstructural characteristics such as void content, fiber volume fraction,
and the resulting composite mechanical properties. Sudarisman et al. [50], the holding temperature
considerably affects the void content, fiber volume fraction, and fiber misalignment of
unidirectional carbon fiber-reinforced polymer (CFRP). Koushyar et al. [26,28] investigated the
effects of variations in autoclave temperature on the microstructure, physical, and mechanical
properties of a carbon fiber/epoxy composite. They found that the degree of curing and the glass
transition temperature depended on the autoclave manufacturing temperature.

Since the autoclave curing process parameters have a crucial influence on the quality of CFRP
composite laminates, this study investigates the curing kinetics of the epoxy resin matrix. The
curing kinetic parameters were determined by isothermal DSC analysis, and three kinetic models
were tested. Additionally, the autoclave process prepared the composite laminates, and their
microstructures were evaluated.

2.5.Materials and experimental protocol


2.5.1. Materials

The specimens were made using sheets of pre-impregnated Hexply®8552S, a toughened epoxy
resin system provided by Hexcel (Figure 15). The prepreg description and storage conditions are

42
described in our previous studies [80,84]. The analytical-grade chemicals used in this study were
purchased from Sigma-Aldrich and used without further purification.

Figure 15 : Structure and chemical composition of 8552S: (i) Tetraglycidyl Methylenedianiline,


(ii) Triglycidyl p-aminophenol, (iii) 4,4’-diaminodiphenylsulfone, and (iv) 3,3’-
diaminodiphenylsulfone

2.5.2. Thermal analysis and curing kinetics

The thermal and kinetic data were secured using a DSC Q20 differential scanning calorimeter (TA
Instrument, USA). Prepreg samples of approximately 5-10 mg were placed in aluminum specimen
holders, and the reference was heated under a nitrogen atmosphere. To describe the kinetic model,
samples were tested isothermally at 120, 140, 160, 180, 200, and 220 °C, respectively, based on
the results of our previous study [80]. The DSC runs were carried out at a heating rate of 20 °C/min.
Once the isothermal cure was achieved, the samples were rapidly cooled to 30°C, and the residual
enthalpy (Hres) was then determined using dynamic analysis of 20°C/min up to 300°C until the
samples were fully cured. The degree of cure (α) at a time (t) during isothermal curing is defined
as the following:

43
1 𝑡 𝑑𝐻𝑖𝑠𝑜
𝛼 (𝑡) = ∫ ( ) 𝑑𝑡 (1)
𝐻𝑇 0 𝑑𝑡
Where HT (J/g) is the total enthalpy of the reaction calculated by the following formula:

𝐻𝑇 = 𝐻𝑟𝑒𝑠 + 𝐻𝑖𝑠𝑜 (2)


Where 𝐻𝑖𝑠𝑜 is the total isothermal curing heat (J/g) obtained by integrating the exothermic peak in
the isothermal cure curves. The reaction rate (dα/dt) is defined as:

𝑑𝛼 1 𝑑𝐻𝑖𝑠𝑜
= (3)
𝑑𝑡 𝐻𝑇 𝑑𝑡
The reaction rate can be described by the following Eq. 4:

𝑑𝛼
= 𝑘𝑓(𝛼) (4)
𝑑𝑡
Where k is the rate constant expressed by the Arrhenius equation (Eq. 5), and f(α) is a function of
the kinetic model.

𝐸𝑎
𝑘 = 𝐴𝑒 (−𝑅𝑇) (5)

A is the pre-exponential factor, Ea is the activation energy, and R is the universal gas constant. A
simple reaction model of nth order, such as that given by Eq. 6, is applied for a curing system that
shows no autocatalytic phenomena and no complexity in the reaction mechanism. Kinetic formulas
describing autocatalytic behavior are also discussed in the literature (i.e., the maximum reaction
rate dα/dt is observed for t > 0 in an isothermal analysis). The simplest autocatalytic kinetic model
is given by Eq. 7.

𝑑𝛼
= 𝑘(1 − 𝛼)𝑛 (6)
𝑑𝑡
𝑑𝛼
= 𝑘𝛼 𝑚 (1 − 𝛼)𝑛 (7)
𝑑𝑡
Where n and m are the reaction orders. Autocatalytic reaction models consider that the curing
process is governed by more than one rate constant, which, in turn, implies the presence of more
than one reaction mechanism and activation energy in the system. However, k1 represents the rate
constant of the primary amine reaction with the epoxide ring and the formation of a secondary
amine. In contrast, k2 represents the rate constant of the secondary amine with another epoxide ring

44
and the formation of a ternary amine. The most classic autocatalytic model is Kamal & Sourour
(Eq. 8) [85]. Several authors have used this model [34,37,86], and it gives good results up to the
diffusion phase of the reactive species at high cure degrees. However, from this moment onward,
the model deviates from the experiment. Consequently, this model does not consider the
phenomenon of species diffusion and is unsuitable for high conversion rates when vitrification
occurs.

𝑑𝛼
= 𝑘1 + 𝑘2 𝛼 𝑚 (1 − 𝛼)𝑛 (8)
𝑑𝑡

2.5.3. Autoclave processing of composite laminates

Square prepreg plies of 20 × 20 cm2 were prepared with different carbon fiber orientations (0°C
and 45°C). Then, panels consisting of 8 prepreg plies and the following [0/90/-45/+45]s stacking
sequence was laid up by hand, following the same steps as Baghad et al. [80]. Finally, the laminated
composites were cured in an autoclave (Tain Techtop Industries, China) using the cure cycles
shown in Table 4.

Table 4. Designed autoclave cure cycles

Cure cycle Ramp up Cure pressure Dwell temperature Dwell time Vacuum-pressure Ramp down
(°C/min) (kPa) (°C) (min) (kPa) (°C/min)
C-1 3 500 160 120 -40 3
C-2 3 500 180 120 -40 3
C-3 3 500 190 120 -40 3
C-4 3 500 205 120 -40 3

2.5.4. Characterization of autoclaved composite laminates

Composite densities were measured using standard test methods for density and specific gravity
(relative gravity) of plastics by displacement, ASTM D792-13 [87], using water as an immersion
liquid. Three specimens for each panel were measured, and the samples were approximately 1g in

45
mass. In addition, the final degree of cure of laminated composites (α’) was measured using
dynamic DSC analysis (3°C/min) performed immediately after the end of the cure cycle:

𝐻′ 𝑇 − 𝐻′𝑟𝑒𝑠
𝛼′ = (9)
𝐻′ 𝑇
Where 𝐻′ 𝑇 = 211,61 𝐽/𝑔 is the total heat released during dynamic cure [5] and 𝐻′𝑟𝑒𝑠 the area
under the heat flow curve.

The constituents' content was determined according to standard test methods for constituent content
of composite materials, ASTM D3171-15 [88]. This technique is the most commonly used in the
industrial sector and has an uncertain level of ±0.2 vol.%. Briefly, samples and sulfuric acid were
placed in a beaker and heated until the solution fumed. During this step, the transparent sulfuric
acid solution (Fig. 16a) changes color from light brown (Fig. 16b) to dark brown (Fig. 16c), and
finally to black (Fig. 16d). This step requires 3 to 4 hours. Once the solution becomes dark,
hydrogen peroxide is added to accelerate the oxidation reaction of the resin matrix. The solution
then appeared clear, and if there was an undigested portion, this implies that the residence time in
the sulfuric acid was insufficient to achieve complete digestion. Next, the beaker was removed
from the hotplate to cool down, and suction filtration was employed to extract fibers. The obtained
fibers were washed three times with distilled water and acetone to remove any remaining sulfuric
acid and hydrogen peroxide. Finally, fibers were placed in an oven at 100 °C for 1h, and their
weight was recorded (Fig. 16e).

2.6.Results and discussion


2.6.1. DSC curve analysis

To investigate the polymerization/crosslinking kinetics at each isothermal temperature and to avoid


biasing the calculations, the change in heat flux during heating is not considered. The heat flux
versus time (t) at different isothermal temperatures for the 8552S system is shown in Figure 17a.
The curves reported in this figure are typical during an isothermal reaction of thermosetting
polymers. In fact, during the initial stages, the concentration of the epoxy/amine reactants is at its
maximum, which explains the rapid increase in heat flow as a function of time; in other words, a
high reaction rate. The shift in the maximum reaction rate observed is associated with the

46
autocatalytic behavior of epoxy systems and their composites [41]. The time corresponding to the
maximum reaction rate (maximum heat flux) for different isothermal temperatures is shown in
Table 5. The second experiment was used as a baseline to calculate the total heat at each isothermal
temperature. However, we recorded a small shift from the expected results during the second scan,
probably due to the evaporation of some resin constituents during the first scan, which reduced the
sample mass in the capsule. Therefore, the curves are shifted vertically to allow the calculation of
the end of cure critical time. The heat of cure is thus defined by the area between the two graphs.
Figure 17b shows the variation of the heat of polymerization/crosslinking as a function of
isothermal temperatures.

Table 5. The time corresponding to the maximum heat flow for different isothermal temperatures

Tiso (°C) 120 140 160 180 200 220


Tp (s) 291.13 342.63 413.13 464.13 657.43 597.53

Figure 16 : Extraction of fibers by acid digestion. The color of the sulfuric acid solution
progressively changes from transparent (a) to light brown (b), followed by dark brown (c), then
black (d), and finally the obtained fibers.

As the heat of reaction increases with the increase in temperature, three zones are recorded, as
shown in Figure 17b. The first zone corresponds to the area between 120 and 140 °C. In this zone,
the heat released by the reaction is the lowest, with a slight increase in the function of the isothermal
temperature. Therefore, we can say that the polymerization reaction is the predominant reaction in
this zone. In the area between 140 and 180 °C, there is competition between the polymerization
reaction and the crosslinking reaction. At the last zone (above 180 °C), there is an apparent increase

47
in the total curing heat flux as a function of the isotherm temperature, reaching a value of 541,053
J/g of resin around 220 °C.

0.6
(a) Iso120°C - 1st run 600
0.5 Iso140°C - 1st run (b)
Iso160°C - 1st run 500
0.4 Iso180°C - 1st run
Heat Flow (W/g)

Iso200°C - 1st run


0.3 400
Iso220°C - 1st run

ΔHIso (J/g)
0.2 300
0.1
200
0.0
100
-0.1
0
0 1000 2000 3000 4000 120 140 160 180 200 220
Time (s) TIso (°C)
Figure 17 : (a) Heat flow versus time and (b) total heat of polymerization/crosslinking reactions at
different isothermal temperatures

1.0 3.5x10-3
(a) (b) Iso120°C
3.0x10-3 Iso140°C
0.8 Iso160°C
2.5x10-3 Iso180°C
Iso200°C
da/dt (s-1)

0.6
2.0x10-3 Iso220°C
a

0.4 Iso120°C 1.5x10-3


Iso140°C
Iso160°C 1.0x10-3
0.2 Iso180°C
Iso200°C 5.0x10-4
Iso220°C
0.0
0 5000 10000 15000 20000 25000 30000 0.0
2000 4000 6000 8000
Time (s)
Time (s)

Figure 18 : Evolution of the degree of conversion (a) and the reaction rate (b) as a function of time
Figure 18 represents the evolution of the degree of conversion and the reaction rate as a function
of time. Figure 18a shows that, for each isothermal temperature, the value of α stabilizes around a
limit value at the end of the analysis. This value, which is noted αmax, represents the state of
progress of the polymerization/crosslinking reactions at a given temperature, whose value increases
with the temperature considered. Indeed, the isotherm performed at 220 °C reaches a degree of
conversion of 87% for an experimental duration of 42 min. In contrast, at this time, the composite

48
subjected to a temperature of 120 °C reaches only 2% of polymerization. The evolution of the
maximum degree of conversion as a function of the isothermal temperature displays an exponential
trend (Figure 19). Figure 18b shows that the rated maximum appears after the start of the isothermal
analysis, highlighting the autocatalytic mechanism of the reaction. However, as the isothermal
temperature increases, the polymerization and crosslinking kinetics become faster.

1.00
0.95
0.90
0.85
0.80
R2 = 0.999
amax

0.75
0.70
0.65
0.60
0.55
0.50
100 120 140 160 180 200 220
TIso

Figure 19 : Variation of the maximum conversion degree as a function of the isothermal


temperature

2.6.2. Curing kinetics

Figure 20 represents the evolution of the reaction rate as a function of the degree of cure, while
Table 6 shows the curing kinetics parameters. The maxima of the reaction rates are obtained for a
degree of cure of around 20%, which is independent of the isothermal temperature. The same
behavior was observed for other epoxy/amine systems [31]. The reaction rates become very low
for advanced conversion states due to the decrease in accessible functions (the epoxides and
amines) and the possible effects of diffusion during the transition to vitrification.

The n-order model considers that the epoxy/amine system can be represented by a single reaction;
an amine group and an epoxy group. Furthermore, this model does not consider the catalytic or
autocatalytic aspect of the reaction. The results show that the K values increase with isothermal
temperature, as predicted by Arrhenius' equation. Similarly, the values of n increase with

49
isothermal temperature. The R2 values demonstrate the data’s good correlation for the evolution of
dα/dt as a function of α from an isothermal temperature of 160 °C. From Figure 20, the model
shows an elevated rate for zero degrees of cure at a given isothermal temperature, with a linear
decrease in dα/dt as a function of α. This behavior is not valid for a system that exhibits an
autocatalytic aspect, such as the system involved in our study. As a result, the nth-order model is
therefore not the most appropriate.

The autocatalytic model is based on the fact that the autocatalytic mechanism dominates over the
uncatalyzed mechanisms of the epoxy/amine system. The results reported in Table 6 highlight a
significant R2, which shows a good correlation of the experimental data for the evolution of dα/dt
as a function of α. The values of the parameters show that the values of K, n, and m increase with
isothermal temperature. The shape of the modeling curves is satisfactory compared with the results
of the nth-order model. The last model concerned by our study is the model of Kamal & Sourour,
which is also an autocatalytic model. The values of R2 and the shape of the modeling curves are
reasonable compared with the two previous models.

2.6.3. Microstructure and physical properties of laminated composites

The variation in autoclave manufacturing parameters leads to a change in laminated composites'


constituent content and physical properties, namely, resin and fiber content, density, and void
content, which affect the mechanical properties [25,26]. Therefore, laminated composites'
microstructure and physical properties were investigated to evaluate the relationship between resin
curing kinetics and final part quality. Figure 21 shows the final degree of cure, density, fiber volume
fraction, and void content for panels cured at different isotherm temperatures. For autoclave cure
cycles C-1 and C-2, insignificant changes in the final degree of cure were observed. However,
increasing the isothermal temperature (C-3) reduces the degree of cure from 0.897 to 0.831. Then
a slight decrease was observed by applying an isothermal temperature of 205°C (C-4). As a result,
the degree of cure decreases by exceeding the recommended temperature (180 °C).

50
Model OrdreN (User)
Equation K*(A-x)^n
Model OrdreN (User)
Reduced 3.92707E-10
Equation K*(A-x)^n
Chi-Sqr
2.71501E-10
-5 Reduced
-4 0.79354
6x10 Chi-Sqr
2.0x10 Adj. R-Square
Value Standard
(a) Iso120°C Adj. R-Square 0.42505
Value
(b)
Standard Error Iso140°C K 1.3807E-4 7.63
K 5.07354E-5 5.10794E-7 da/dt n 0.44524
nth order model da/dt n 0.37642 0.00593 nth order model A 0.64376
-5
5x10 Autocatalytic model
A 0.52099 0

Autocatalytic model
Kamal & Sourour model 1.5x10-4 Kamal & Sourour model
4x10-5
da/dt (s-1)

da/dt (s-1)
3x10-5 1.0x10-4

2x10-5
5.0x10-5
1x10-5

0
0.0 0.2 0.4 0.6 0.8 1.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0
a
a

4.0x10-4 1.0x10-3
(c) Iso160°C (d) Iso180°C
nth order model nth order model
Autocatalytic model 8.0x10-4 Autocatalytic model
3.0x10-4 Kamal & Sourour model Kamal & Sourour model
da/dt (s-1)

da/dt (s-1)

6.0x10-4
-4
2.0x10
4.0x10-4

1.0x10-4
2.0x10-4

0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
a a

2.0x10-3 4.0x10-3
(e) Iso200°C (f) Iso220°C
nth order model 3.5x10-3 nth order model
Autocatalytic model Autocatalytic model
1.5x10-3 Kamal & Sourour model 3.0x10-3 Kamal & Sourour model
da/dt (s-1)

da/dt (s-1)

2.5x10-3
1.0x10-3 2.0x10-3
1.5x10-3
5.0x10-4 1.0x10-3
5.0x10-4
0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
a a
Figure 20 : Modeling of polymerization/cross-linking kinetics using different models: (a) 120 °C,
(b) 140 °C, (c) 160 °C, (d) 180 °C, (e) 200 °C, and (f) 220 °C.

51
Table 6. The curing kinetics parameters of 8552S epoxy resin prepreg at different isothermal temperatures

nth-order model Autocatalytic model


TIso (°C)
K (s-1) n R2 K (s-1) m n R2

120 5.07x10-5 ± 5.11x10-7 0.38 ± 5.93x10-3 0.43 2.19x10-4 ± 6.22x10-6 0.43 ± 9.11x10-3 0.77 ± 10.63x10-3 0.60
140 1.38x10-4 ± 7.63x10-7 0.45 ± 3.87x10-3 0.79 5.00x10-4 ± 3.30x10-6 0.46 ± 2.36x10-3 0.86 ± 2.97x10-3 0.97
160 3.68x10-4 ± 1.82x10-6 0.91 ± 4.75x10-3 0.83 0.00160 ± 1.12x10-5 0.47 ± 2.35x10-3 1.60 ± 4.17x10-3 0.97
180 8.58x10-4 ± 2.76x10-6 0.68 ± 3.02x10-3 0.95 0.00162 ± 7.81x10-6 0.25 ± 1.81x10-3 0.96 ± 2.84x10-3 0.99
200 1.87x10-3 ± 4.33x10-6 0.92 ± 2.57x10-3 0.97 0.00332 ± 9.52x10-6 0.22 ± 1.05x10-3 1.23 ± 2.00x10-3 0.99
220 3.85x10-3 ± 9.19x10-6 1. ± 1.32x10-3 0.98 0.00632 ± 2.13x10-5 0.20 ± 1.32x10-3 1.33 ± 2.44x10-3 1.00

Kamal & Sourour model


TIso (°C)
K1 (s-1) K2 (s-1) m n R2

120 0.00 ±2.51x10-7 1.99x10-4 ± 6.31x10-6 0.40 ± 10.36x10-3 0.72 ± 14.51x10-3 0.60
140 0.00 ±1.56x10-7 8.88x10-4 ± 3.73x 10-6 0.44 ± 2.81x10-3 0.85 ± 4.21x10-3 0.97
160 4.34x10-6 ± 1.83x10-7 1.76x10-3 ± 1.49x10-5 0.50 ± 2.76x10-3 1.69 ± 6.04x10-3 0.97
180 0.00 ± 5.11x10-7 1.60x10-3 ± 8.25x10-6 0.24 ± 1.95x10-3 0.95 ± 3.68x10-3 0.99
200 0.00 ± 3.15x10-7 3.31x10-3 ± 9.97x10-6 0.22 ± 1.09x10-3 1.22 ± 2.32x10-3 0.99
220 3.27x10-5 ± 5.84x10-7 6.70x10-3 ± 2.03x10-5 0.22 ± 1.16x10-3 1.41 ± 2.47x10-3 1.00

52
The laminate density shows a slight decrease as the isotherm cure temperature increases, which
.can be attributed to the variation of the degree of cureComposite laminates with low void contents
were obtained at 180 °C, which is explained by the significant degree of conversion achieved at
this isotherm cure temperature (0.897). Increasing the isotherm temperature leads to reduced void
content, generally by 91%, when the isotherm temperature goes from 160 to 180 °C. This reduction
was accompanied by a decrease in fiber volume fraction and an increase in density. Similar
behavior was reported by Sudarisman et al. [50]. However, the porosity of all panels does not
exceed 1 vol.%, and it’s very acceptable in the aeronautical and aerospace industries (less than 2.5
vol.%) [89]. The negligible porosity of all cured panels indicates that cure pressure (500 kPa) was
appropriately applied [80].

2.0
1.0 0.897 (a) (b)
0.887
0.831 0.827 1.612 1.618 1.602 1.602
0.8 1.5
Density (g/cm3)

0.6
a'

1.0
0.4
0.5
0.2

0.0 0.0
160 180 190 205 160 180 190 205
Temperature (°C) Temperature (°C)

80
(c) 0.8
(d)
67.175 66.699 66.259 65.112 0.699
Fiber volume fraction (%)

60
0.311
Void content (%)

0.6

40 0.028
0.4 0.278

20 0.2

0 0.0
160 180 190 205 160 180 190 205
Temperature (°C) Temperature (°C)
Figure 21 : Properties of laminated composites as a function of isotherm temperatures: (a) degree
of cure, (b) density, (c) fiber volume fraction, and (d) void content

53
2.7.Conclusions

Quasi-isotropic laminates composed of an 8552S/AS4 prepreg were manufactured by autoclave


under 500 kPa of pressure, -40 kPa of vacuum pressure, and different isotherm temperatures were
designed according to the thermal behavior of the prepreg. Similar thermal behavior was observed
for all the cycles. However, an optimal degree of curing was achieved by applying an isothermal
temperature of 160 or 180 °C. By exceeding this temperature, the final degree of cure decreases.
The same results were obtained after manufacturing the composites in the autoclave. However,
composites manufactured at an isothermal temperature of 180 °C have excellent physical properties
such as a high degree of curing, high density, and low void content.

2.8.Acknowledgments

The first author (Mr. Abd Baghad) would like to express his sincere gratitude to his deceased
brother, Mr. Iazza Baghad, who passed away on August 7th, 2021, for his numerous sacrifices and
support. Furthermore, financial support from Safran Nacelles (France), the Hassan II Academy of
Sciences and Technologies (Morocco), and Euromed University of Fes (Morocco) is gratefully
acknowledged.

54
CHAPITRE 3

CURE KINETICS AND AUTOCLAVE-PRESSURE


DEPENDENCE ON PHYSICAL AND MECHANICAL
PROPERTIES OF WOVEN CARBON/EPOXY
8552S/AS4 COMPOSITE LAMINATES

3.1.Avant-propos

Dans un second temps, afin de donner suite aux travaux de recherche présentés dans le chapitre
précédent, ce présent chapitre vise à vérifier l’impact de la pression appliquée lors de la fabrication
en autoclave sur les propriétés physiques et mécaniques des composites stratifiés. Cette étape est
particulièrement importante afin de déterminer la pression optimale pour la fabrication des
composites de haute qualité. Cependant, l’obtention de l’évolution de différentes propriétés en
fonction de la pression permettra de saisir l’importance de la pression appliquée en autoclave et
son effet sur les propriétés physiques, et d’évaluer la relation propriétés physiques – propriétés
mécaniques. Il est à noter que les résultats de ce présent chapitre seront exploités pour déterminer
les niveaux de pression à évaluer lors l’optimisation des paramètres à travers des plans surfaciques
(Chapitre 5).

3.2.Résumé

Les propriétés mécaniques de la pièce à base des composites stratifiés dépendent fortement de leur
cycle de durcissement en l'autoclave. Au cours de ce cycle, la température, la pression, le vide et
le temps de traitement vont influencer la qualité des pièces fabriquées. En effet, la teneur en vides
est considérée comme le défaut le plus nuisible dans les stratifiés carbone/époxy car elle affaiblit

56
les propriétés mécaniques dominées par la matrice telles que les résistances au cisaillement
interlaminaire et à la compression. Dans le présent travail, la DSC est utilisée pour caractériser
l'influence du temps et de température sur le comportement de la résine époxyde. Ensuite, une série
de composites stratifiés [0/90/-45/+45]s ont été fabriqués en autoclave sous différentes pressions
pour évaluer l’effet de la pression de fabrication en autoclave sur la microstructure et les propriétés
mécaniques. Le module de cisaillement interlaminaire, la résistance au cisaillement interlaminaire,
le module de compression et la résistance à la compression à température ambiante et à la
température de fonctionnement des moteurs des avions ont été évalués. La corrélation entre la
microstructure et les propriétés mécaniques a également été étudiée. Il s'avère que les propriétés
mécaniques des stratifiés carbone/époxy fabriqués dépendent de la pression et de la microstructure.
Ces résultats sont explorés pour établir une voie optimale de pression d'autoclave qui minimiserait
la porosité sans contrebalancer les propriétés mécaniques.

Mots-clés : Composites à matrice polymère, pression de durcissement, propriétés à haute


température, propriétés mécaniques, microstructure.

3.3.Abstract

The final mechanical properties of composite laminates are highly dependent on their curing cycles
in the autoclave. During this cycle, the temperature, pressure, vacuum, and treatment time will
influence the quality of manufactured parts. The void content is considered the most harmful
defects in carbon/epoxy laminates since they weaken the matrix-dominated mechanical properties
such as interlaminar shear and compressive strengths. In the present work, differential scanning
calorimetry is used to characterize the influence of time/temperature on the behavior of the epoxy
resin. Then, a series of [0/90/-45/+45]s laminates composites are autoclave-cured under various
applied pressures to evaluate their impact on microstructure and mechanical properties. The
interlaminar shear modulus, interlaminar shear strength, laminate compressive modulus, and
laminate compressive strength at room and operating engine temperature were measured. The
correlation between microstructure and mechanical properties was also studied. The mechanical
properties of manufactured carbon/epoxy laminates are found to be dependent on pressure and

57
microstructure. These results are explored to establish an optimal autoclave pressure route that
would minimize porosity without counterbalancing mechanical properties.

Keywords: Polymer-matrix composites, curing pressure, High-temperature properties,


Mechanical properties, microstructure.

3.4.Introduction

Carbon fiber reinforced polymers (CFRP) are increasingly used in aerospace applications because
of their higher thermal and mechanical properties compared to conventional composites. They are
also superior to their metallic counterpart regarding corrosion and fatigue resistance [2].
Widespread application of lightweight materials with high mechanical properties as aircraft
materials yield improvement in fuel efficiency, increased payload, and flight range, all of which
directly reduce an aircraft operating cost [7]. By improving and optimizing the mechanical
properties of CFRP, the time between maintenance can also be lengthened to reduce repair costs.
These benefits have allowed the use of carbon fiber-reinforced polymer in modern passenger
aircraft such as Boeing 787 and Airbus A350 (up to 50% of their weight in CFRP) [19].

Laminate composites are manufactured using several methods. The most frequently used are the
autoclave process and the out-of-autoclave process [56]. Autoclave processing yield structures with
less than 1% of porosity and high mechanical integrity [3,27]. However, the autoclave-curing
parameters need to be chosen carefully to achieve good results, among which are curing
temperature and time, pressure, and bag vacuum level. For instance, the curing temperature and
time affect the polymerization/crosslinking reactions of the epoxy resin. The reaction mechanism
and curing process determine the final composite characteristics and its mechanical properties [90].
In epoxy/amine-based resin, polymerization/crosslinking reactions occur at specific temperatures
and result in the opening of the epoxy ring, resulting in the formation of ethers and alcohol groups
and the conversion of primary amines to secondary and tertiary amines [30]. The pressure applied
on the laminate surface inside the autoclave allows the CFRP to conform to the mold shape and to
be compacted to the desired fraction of fiber volume, thus reducing the voids likely to form during

58
the curing cycle. The vacuum is used to remove any entrapped air and volatiles during the curing
cycle.

In earlier studies, the curing pressure was used to create different levels of voids in laminates
[26,51,53–55] in an attempt to find out a relation between mechanical properties and voids
[26,53,55–57] or pressure [26,57]. These studies confirmed that a decrease of curing pressure
increases porosity, yielding in turn reduced mechanical properties. Composites produced by Liu et
al. [51] yielded decreasing void content from 3.2% to 0.6% when applied pressure was increased
from zero to 0.6 MPa, while ILSS increased by 18%. Costa et al. [55] reported a similar trend for
ILSS of carbon epoxy laminates, where a 34% decrease is observed when void content increases
from 0.55% to 5.60%. Di Landro et al. [58] noticed a decline of ILSS up to 25% in composite with
the highest porosity (7%), compared to composite laminates with no void content.

While the influence of pressure on manufactured parts is more documented, the current literature
available is poor on the impacts of temperature [28,83,91,92]. This is compounded by the fact that
the void formation mechanisms within composite laminates during autoclave curing are not fully
understood. While some studies point to dissolved moisture in the resin as a possible cause of void
formation [47], where autoclave pressure could force moisture to remain in solution thus preventing
the formation and expansion of voids [65], other studies point to entrapped air and volatiles as the
primary source of voids [79].

Engine aircraft nacelles currently incorporate many significant composite components, where
carbon/epoxy prepregs account for almost half of the total volume of the nacelle structure. Among
the possible composites for engine aircraft nacelles where a continuous service temperature of
120°C is to be expected, the woven Hexply 8552S/AS4 is the object of this study. This work aims
to evaluate the impact of input autoclave conditions on the final microstructure of laminate
composites and optimize the resulting mechanical properties by adjusting pressure. This will be
achieved first by evaluating through Differential Scanning Calorimeter (DSC) the behavior of
epoxy resin during time-temperature treatment. The goal is to predict the time/temperature
treatment cycle enabling a total polymerization/crosslinking and, to some extent, help achieve
energy saving during autoclave processing. Once established, this cycle is used to prepare in an
autoclave under various pressures laminate composites, which will be assessed in terms of density,

59
thickness, and voids content. Mechanical properties, namely laminate compressive strength (LCS),
laminate compressive modulus (LCM), interlaminar shear modulus (ILSM), and interlaminar shear
strength (ILSS), will be established at room temperature and the expected continuous service
temperature of 120°C. Correlations between microstructure and mechanical properties such as
LCM, ILSM, LCS, and ILSS will be discussed. The relationship between LCM and ILSM, and
between LCS and ILSS will also be established.

3.5.Experimental procedure
3.5.1. Materials

The material used in this study is a woven carbon fiber/epoxy prepreg (Hexply 8552S/AS4/280H5)
manufactured by Hexcel France. It is a high-performance epoxy matrix for use in primary
aerospace structures. Hexply 8552S is an amine cured, toughened epoxy resin (37% resin content)
reinforced by woven carbon AS4 PAN-based carbon fibers (3000 filament counts tows) with 280
g/m2 areal fiber weight. Hexply 8552S uses a 180°C curing epoxy matrix [93]. The prepreg was
stored in hermetically sealed packing at -18°C. DSC analysis and autoclave manufacturing were
performed within the 12 months’ shelf life limit (from date of manufacture). Samples under its
packaging were placed at room temperature for one day before their uses; their out-life was 17-18
days.

3.5.2. Thermal analysis

A differential scanning calorimeter (DSC Q20, TA Instruments) was used to simulate the heat flow
change during the autoclave curing cycle and evaluate the behavior of Hexply AS4/8552S. Before
each DSC run, a neat prepreg sample (10mg) was encapsulated in a standard aluminum sample
pan. The encapsulated sample was then put in the DSC cell, and all experiments were carried out
under nitrogen environment. For the dynamic measurements, the DSC cell was heated at a constant
rate of 3°C/min over a temperature range of 35 to 320°C. The resulting sample was the second run
specimen, and the test was performed using the same conditions. Combined analysis between
isothermal and non-isothermal was performed at a heating rate of 3° C/min to 180°C, after which

60
the sample is held at that temperature for 170 min. The resulting sample was analyzed using a
3°C/min dynamic scan until the specimen was fully cured to determine the residual enthalpy.

0,8 -0,10
P0,1
200
P0,3 0,7
P0,5 -0,08
P0,7 0,6
150
Temperature (°C)

Pressure (MPa)

Vacuum (MPa)
0,5
-0,06

0,4
100
0,3 -0,04

0,2
50
-0,02
0,1

0 0,0 0,00
0 50 100 150
Time (min)

Figure 22 : Curing cycles used to fabricate 8552S/AS4 composites

3.5.3. Fabrication and curing

CFRP 8 layers samples, having a dimension of 20 by 20 cm, were laid up by hand to achieve quasi-
isotropic [0/90/-45/+45]s stacking sequence. The samples were then autoclave-cured, using one of
the cycles shown in Figure 22 that were derived from DSC results and manufacturer recommended
cycles (MRC). The cycle of thermal curing proposed by Ng et al. [94] was adopted; the heating
rate for all testing was set at 3oC/min to reach the dwell temperature of 180°C during 90 min (dwell
time was determined based on DSC results), while a 3°C/min ramp down was used. Various
pressures (0.1, 0.3, 0.5, and 0.7 MPa) were chosen to study the effect of curing pressure variation
while the vacuum was fixed at -0,02 MPa. Sample curing was performed in an autoclave (Taian
Techtop Industries, China), using vacuum bags prepared with a specific arrangement illustrated in
Figure 23. The role of the components used is as follow:

61
- Vacuum connectors: Quick and removable way to connect the bag to the vacuum system;
- Vacuum bagging film: Film having a high thermal resistance and used to apply pressure
uniformly across the entire laminate;
- Breather: Uniformizes the vacuum pressure within the bag while absorbing any excess resin
passing through the perforated release film;
- Release film: retains more resin on the laminate surface.
- Sealant tape: ensures the tightness of the bag.

Figure 23 : Autoclave and vacuum bags assembly used in this study

3.5.4. FTIR

FTIR analysis of prepreg AS4/8552S and composite laminates before and after autoclave curing
was performed in ATR mode using a Thermo Nicolet iS50 FTIR (Thermo Fisher Scientific Co.,
Waltham, MA, USA) spectrometer. Analysis conditions were set at 64 scans from 400 to 4000 cm-
1
, using a resolution of 4 cm-1. The sample’s smoother surface was chosen to achieve the best
contact possible with the ATR crystal.

3.5.5. Density and void content

62
The density of the laminate was measured following ASTM D792-13, using water as the immersion
liquid [87]. The void content was measured by the digestion method, following the ASTM D3171-
15 [88], using heated concentrated sulfuric acid to dissolve the resin.

3.5.6. Optical microscopy

The void shape, location, and sizes were examined using optical microscopy, Nikon Eclipse
LV100ND (Japan), equipped with a DS-Ri2 camera. The specimens for porosity analysis were
polished using a polishing machine (Unipol-820) with 400, 600, 1000, and 2000 grit size silicon
carbide abrasive paper.

3.5.7. Mechanical properties

The combined loading compression (CLC) and short beam shear (SBS) tests were conducted on a
Universal Testing System (MTS Criterion model 45, MTS Systems Corporation (USA)) according
to ASTM D6641/6641M–16 [95] and NF EN 2563 [96], respectively. The dimensions of CLC and
SBS samples were 140×13×e and 20×10×e, respectively (e is the thickness). As this is a quasi-
isotropic laminate, the mechanical properties are equal in all directions within the plane of the
laminate. Therefore, the tests were carried out in only one direction. Testing at service temperature
was achieved by the mean of an environmental chamber (Thermcraft Lab-Tem LBO, Thermcraft
incorporated, USA) allowing the sample’s temperature to reach the consign value without influence
on the load cell. SBS tests were performed at room temperature and 120°C, whereas CLC tests
were conducted at 120°C. At elevated temperature, a delay of 10 min was imposed before testing.
Five to seven specimens were tested for each test and temperature condition.

3.6.Results and discussion


3.6.1. DSC

DSC was used to characterize the influence of time/temperature on the behavior of the epoxy resin.
For this purpose, heat flow is recorded using either dynamic or a combination of dynamic and

63
isotherm measurements depending on the curing conditions. The dynamic DSC curves are shown
in Figure 24. Two consecutive runs are performed to evaluate any changes caused by the first
heating cycle. As seen by the shift in Figure 24a of the heat flow signal to higher values, an
exothermic reaction attributed to the polymerization process begins from ~155°C and reaches a
maximum at ~190°C. A second peak at ~236°C, found in the tail end of the first, is attributed to
the crosslinking reaction. This phenomenon is observed only in the first run, as the second run does
not show any significant heat flow exchange. This difference is explained by the complete curing
of the epoxy during the first run. Figure 24b represents changes in the heat flow of the specimen as
a function of time during the first run. The area under the exothermal peak is calculated by drawing
a linear baseline between onset and end. This value corresponds to the heat of curing. As the
obtained value of 211,61 J/g is for prepreg containing 37 wt.% of resin, the curing heat thus
obtained amount to 571,92 J/g.

(a) 1st run (b) 1st run


0.2 0.2
2nd run Baseline

0.1 0.1
Heat Flow (W/g)

Heat Flow (W/g)

0.0 0.0 211,61 J/g of prepreg

-0.1 155 °C -0.1

-0.2 -0.2

-0.3 -0.3

50 100 150 200 250 300 1000 2000 3000 4000 5000
Temperature (°C) Time (s)

Figure 24 : Dynamic DSC analysis: Heat flow vs temperature (a) and time (b)

To achieve maximum curing during the autoclave cycle while minimizing the required time, a
combined analysis between dynamic and isothermal was performed at a heating rate of 3° C/min
to 180°C, after which the sample is held at that temperature for 170 min. The goal of the test is to
evaluate through DSC the degree of curing and the rate of reaction during the curing cycle as well
as the minimum time duration necessary to achieve the optimum cure (Figure 25). The DSC profile
simulating the autoclave cycle (Figure 25a) shows a heat reaction (∆H) equal to 522,89 J/g of resin
(193.47 J/g of prepreg). This value is comparable to the total heat (∆Ht) obtained from our previous

64
measurement by dynamic test (571,92 J/g). The DSC curve becomes horizontal near the end,
meaning that curing is close to complete. The final degree of curing is equal to 0,914. This value
is confirmed by cooling down the sample and performing a second heating run in dynamic mode.
The resulting DSC curve (Figure 26) yields a value of residual enthalpy equal to 44.62 J/g of resin
(16,51 J/g of prepreg), implying a residual conversion of 0,078. However, it has been shown After
a degree of cure of 0.9, the mechanical properties remain practically unchanged [83].

1.0 200
1st run a
0.1 193,47 J/g of prepreg Baseline -4 180
da/dt 8.0x10
0.8 Temperature 160
Heat flow (W/g)

0.0 -4 140
6.0x10

Temperature
da/dt (s-1)
0.6 120

-0.1 a 4.0x10-4
100
0.4 80
60
-0.2
0.2 2.0x10-4 40

(a) 20
-0.3 (b)
0 2000 4000 6000 8000 10000 12000 0.0 0.0 0
0 1000 2000 3000 4000 5000 6000
Time (s)
Time (s)

Figure 25 : DSC curve simulating the autoclave curing cycle (heating to 180°C at 3°C/min and
hold), (b): Temperature, degree of cure, and reaction rate as a function of time in the autoclave
curing cycle

To predict the reaction progress over time, the degree of conversion at each instant t is calculated
by integrating the area under the exothermic curve as follows:

𝑡
1 𝑑𝐻
𝛼 (𝑡) = . ∫ ( ) 𝑑𝑡 (10)
𝐻𝑡 0 𝑑𝑡

The reaction rate is then determined by the differential of the curve α = f(t), which is assumed
proportional to the heat generation and expressed as:

𝑑𝛼 1 𝑑𝐻(𝑡)
= . (11)
𝑑𝑡 𝐻𝑡 𝑑𝑡

65
Variation of temperature, degree of curing, and reaction rate as a function of time are represented
in Figure 25b, where it can be observed clearly that the reaction rate is low at the beginning
(~155°C). Once temperature approaches 180°C, a sharp increase in reaction rate is observed,
reaching its maximum value of 8.14E-4 s-1. After ~16 min at isothermal temperature, the reaction
rate decreased to 2.92E-4 s-1 while the degree of cure reached ~0.68. The maximum degree of
conversion (0,914) is attained after about 1h at isothermal temperature. It can be concluded from
these results that composites laminates must be maintained at 180°C for 1 hour during the autoclave
curing cycle.

2nd run
Baseline

16,51 J/g of prepreg


-0.2
Heat flow (W/g)

-0.3

-0.4

50 100 150 200 250 300


Temperature (°C)

Figure 26 : DSC dynamic curve (3°C/min) obtained in the second run after the first cycle (heating
to 180°C at 3°C/min and hold)

To gain a better understanding of the behavior of the resin, the development of cure during
autoclave curing cycle simulate by DSC is investigated using the model proposed by Cole et al.
(Eq.3) [35], and used and tested by Garstka et al.[97]:

𝑑𝛼 𝑘𝛼 𝑚 (1 − 𝛼)𝑛
= (12)
𝑑𝑡 1 + 𝑒 𝐶(𝛼−(𝛼𝑐0 +𝛼𝑐𝑇 .𝑇))

66
Where α is the degree of cure, t is the time (s), m and n are reactions orders, C is the diffusion
constant, 𝛼𝐶0 is the critical degree of cure at 0 K, 𝛼𝐶𝑇 is the constant accounting for the increase in
critical resin degree of cure with temperature. k is the reaction constant with an Arrhenius type of
dependence with temperature.

Table 7. Curing kinetics parameters for 8552S/AS4 resin


Parameters A (s-1) Ea (J/mol) m n C 𝛼𝐶0 𝛼𝐶𝑇 (K-1)
Values 74612.43627 62834.88354 0,83585 2.07289 33.78520 -0.61511 0.00702

In our case, Figure 27 showed the model predictions which indicates that the Cole et al. model is a
good model fit for the epoxy resin studied in this work (R2 = 0,973). The corresponding values of
the parameters are summarized in Table 7.

Experimental
8,0x10-4 Model Fit

6,0x10-4
da/dt (s-1)

4,0x10-4

2,0x10-4

0,0
0,0 0,2 0,4 0,6 0,8 1,0

Figure 27 : Reaction rate vs. degree of curing plot with a model provided by Cole et al

67
3.6.2. FTIR analysis of laminated composites processed by autoclave

FTIR is a sensitive method for monitoring the evolution of polymerization / crosslinking reactions
during autoclave curing. The FTIR spectra of 8552S prepregs before autoclave curing and of
laminated composites manufactured at different pressures are found in Figure 28. Table 8 lists the
characteristic bands of the different molecular groups.

Prepreg Prepreg
P0,1 P0,1
P0,3 P0,3
P0,5 P0,5
3368 P0,7 P0,7
Absorbance

Absorbance
940
1624
3463 2996 1455 906
3057 1598
3238 1411

3600 3400 3200 3000 2800 1600 1400 1200 1000


Wavenumber (cm-1) Wavenumber (cm-1)

Figure 28 : FTIR spectra of prepreg and composite laminates

It can be noticed in Figure 28 the full opening and total reaction of the epoxide groups present
normally found before autoclave curing. This is illustrated by the total disappearance of the 906
cm-1 and 3057 cm-1 characteristics bands, which are respectively attributed to C-O-C deformation
and C-H tension of the methylene group of the epoxy ring. FTIR also highlighted the total
disappearance of the 2996 cm-1 and 1455 cm-1, corresponding respectively to the stretching
vibrations and fundamental deformation of the -CH2 in the epoxy group. The disappearance of the
940 cm-1 and 1411 cm-1 bands are also noted, corresponding respectively to the antisymmetric
deformation of epoxy ring and deformation of C-H of epoxy. Several changes are also observed in
the amine content, where FTIR spectra show the disappearance of the characteristic 3463 cm-1 and
3368 cm-1 symmetrical and antisymmetrical stretching bands of the primary amine group, as well
as the distinctive band of the primary amine at 3238 cm-1. These changes are accompanied with the
appearance of a very strong and broad signal between 3700 cm-1 and 3100 cm-1, attributed to the
stretching of the O-H and N-H groups. The broad range indicate the existence of intermolecular
68
hydrogen bonds [98]. A reduction in the 1624 cm-1 band can also be observed. This band is a sum
of the bands attributed to the stretching vibrations of phenyl ring and deformation vibration of -
NH2 in primary amine. A sharp decrease in the broad band around 1598 cm-1 is also observed. This
band is attributed to the deformation vibration of -NH group in secondary amine. The changes in
the 1400 – 700 cm-1 region are attributed to emerging secondary and tertiary amine groups and O-
H group [29]. As these signals are overlapped, they are difficult to investigate as far as the observed
changes are concerned.

Table 8. Characteristic bands related to epoxy/amine resin

Wavenumber (cm-1) Functional group Wavenumber (cm-1) Functional group

3700-3100 O-H (Hydroxyl group) 1624 NH2 (Primary amine)

3463 N-H (primary amine) 1598 N-H (Secondary amine)

3368 N-H (primary amine) 1455 C-H (Epoxy group)

3238 N-H (primary amine) 1411 C-H (Epoxy group)

3057 C-H (Epoxy group) 940 C-O-C (Epoxy group)

2996 C-H (Epoxy group) 906 C-O-C (Epoxy group)

3.6.3. Effect of curing pressure on microstructure

The manufactured laminate composites were carefully controlled in terms of carbon fiber and
epoxy resin. A visual inspection of vacuum bags showed no loss of fiber during the autoclave
curing process. A loss of the resin was however observed depending on the applied pressure; as we
will see, this is proven by optical microscopy images and acid digestion. FTIR analysis confirmed
results obtained by simulating the thermal cycling through DSC. Disappearance after curing of the
bands associated with the epoxy ring and the primary amine is explained by the total reaction of
these functional groups with each other (Figure 28).

69
100
Fiber (b)
2.15 90
0.3
(a) 1640 Resin 0.2

2.10 80 Void content 0.1


0.0

Volume fraction (%)


1620 70
2.05

Density (kg/m3)
Thickness (mm)

60
2.00 1600
50
1.95
1580 40
1.90 30
1560
1.85 Thickness 20
Density 10
1.80 1540
0.0 0.2 0.4 0.6 0.8 0
0.1 0.3 0.5 0.7
Pressure (MPa) Pressure (MPa)

Figure 29 : Influence of pressure on (a) density, thickness, (b) fiber volume fraction, resin volume
fraction, and void content

Figure 30 : Optical microscope images through the thickness of composite laminates taken at 50×
magnification: (a) at 0.1MPa, (b) at 0.3MPa, (c) at 0.5MPa, and (d) at 0.7MPa

The influence of cure pressure on microstructure has been presented in Figure 29, whereas Figure
30 illustrates optical microscopy images of each composite. Figure 29a presents the average
thickness and density of composites vs. curing pressures. As clearly illustrated, the thickness

70
decreases with increasing curing pressure while density increased. According to Figure 30, there is
a reduction in the thickness of each layer, resulting from the increase in cure pressure. These results
provided good support for the theoretical hypothesis that sample thickness and density depend
greatly on applied pressure. As a result, the cure pressure ensures good adhesion between different
prepreg layers of a composite laminate. Thickness variation was found to present a logarithmic
fitting curve (R2 = 0,999) as a function of applied pressure, whereas the density shows a 3 rd
polynomial curve fitting (R2 = 1). At 0,1 MPa, the laminate composites have an average thickness
and density of 2.10 mm and 1556 kg/m3, respectively. When curing pressure is increased to 0.3
MPa, thickness reduces to 2 mm while density reaches 1572.76 kg/m3. Increasing pressure further
to 0.5 MPa yield another 2.75% decrease in thickness and a 2.57% increase in density. Applying a
0.7 MPa pressure yield composites with a thickness of 1.916 mm and a density of 1632.85 kg/m3.

The influence of cure pressure on the fiber volume fraction, resin volume fraction, and void content
has been presented in Figure 29b. As the imposition of more pressure leads to remove a large
amount of resin (Figure 30), a decrease in resin volume fraction and an increase in fiber volume
fraction were observed (Figure 29b). By comparing the various optical microscope images, we
notice that the frequency and size of resin-rich areas found in composites decrease with increasing
pressure; they are usually found between layers, which constitute the laminated composite. Figure
29b also showed that the void content decreased as the cure pressure increased from 0.1MPa to
0.7MPa, behavior reported by multiple authors [51,54,99]. For a curing pressure of 0.1 MPa,
laminates exhibit a high void volume content. The voids are very large (Figure 30a), having an
irregular shape (meso-void) [63]. Many reports identified low pressure as a highly detrimental
parameter for laminate composites, leading to high void content and lower mechanical properties
[26,76,99]. Li et al. [100] found that when the applied pressure is low, an important number of
large voids mostly at ply interfaces are created, of which the number and size reduce significantly
when the cure pressure is increased. Increasing the curing pressure of the autoclave to 0.3 MPa
yield a 25% decrease in void content to 4.19% when compared to samples prepared at 0.1 MPa. At
0.5 MPa, the void content is significantly reduced to attain a value of 0.05%, representing a 99.1%
reduction from the level measured at 0.1 MPa.

71
Laminate Compressive Modulus (MPa)
30000 6000
SBS (a)
Interlaminar shear modulus (MPa)

(b)
SBS-120
25000 5000

20000 4000

15000 3000

10000 2000

5000 1000

0 0
0.1 0.3 0.5 0.7 0.1 0.3 0.5 0.7
Pressure (MPa) Pressure (MPa)

90 460
(c)

Laminate Compressive Strength (MPa)


SBS
85 440
Interlaminar Shear Srength (MPa)

SBS-120
CLC
80 420

75 400

70 380

360
65
340
60
320
55
300
50
280
45
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Pressure (MPa)

Figure 31 : Mechanical properties as a function of pressure at room and 120°C: (a) LCM, ILSM,
and (b) ILSS and LCS

At high curing pressure (0.7 MPa), a negative void volume percentage was calculated (-0.17%).
This is explained by the fact that calculation of void content assumes a constant density for filler
and matrix during the autoclave curing process [88], both of which vary slightly. Based on optical
microscope images (Figures 30a, 30b, and 30c), the voids generally are located between the layers.
Researchers who have studied the effect of curing cycle pressure on void formation observed that
voids occurred mostly at the ply interface and concluded that the voids are much larger, flattened,
and elongated at low curing pressure while exhibiting a spherical shape at higher pressure [26,51].

72
When applying high pressure (Figure 30d), adjacent prepreg layers would become bonded together
at a faster rate, and thus inhibit any entrapped voids between layers [50]. By the action of pressure,
voids are encouraged to disperse and dissolve within the resin consequently causing a lower void
content and high fiber volume fraction [49].

3.6.4. Effect of curing pressure on mechanical properties

Autoclave pressure significantly affects the thickness and density of laminate composites, as well
as voids content and fiber volume fraction. Samples with minimal porosity, high density, minimal
thickness, and high fiber volume fraction should thus produce the best mechanical behavior. This
hypothesis was tested by measuring the ILSM, ILSS, LCM, and LCS (Figure 31). The ILSM and
LCM strongly depend on the applied pressure (Figure 31a). When curing pressure is increased from
0.1 MPa to 0.5 MPa, the corresponding ILSM at room temperature and 120°C increases of 41.01%
and 49.37%, respectively. When pressure is increased to 0.7 MPa, a slight decrease in the ILSM
can be observed. Equally, an increase in LCM from 4360.731 MPa to 5206.301 MPa is observed
when the cure pressure increases (Figure 31b), representing an improvement of 19.39%. The
increase in LCM is particularly significant when cure pressure increase from 0.5MPa to 0.6MPa
with a relative variation of 10.53% between these pressures.

The LCS and ILSS of the composite laminates are shown in Figure 31c. For a curing pressure of
0.1 MPa, laminates exhibit low mechanical properties in terms of LCS and ILSS. Many reports
identified low pressure as a highly detrimental parameter for laminate composites, leading to high
void content and lower mechanical properties [26,99]. The fracture mechanism of composites cured
at low pressure is mostly delamination, causing the cracking of the matrix. The presence of
interfacial porosity and in-between fibers may be a contributing factor [68]. Increasing pressure of
the autoclave to 0.3 MPa yield a 5.96% increase in LCS to 350.17 MPa when compared to samples
prepared at 0.1MPa. The ILSS and ILSS-120 also show improvement by 10.04%, and 4.34%,
respectively. At 0.5 MPa, the LCS, ILSS, and ILSS-120 all show improvement to reach their
maximum values at 360.34 MPa, 79.39 MPa, and 58.88 MPa respectively.

73
The mechanical properties of laminated composite materials are affected by their microstructure.
To be able to correctly interpret the mechanical behavior, it is necessary to study the relationship
microstructure-mechanical properties (Figure 32). Figure 32a and Figure 32b depict the
mechanical properties as a function of density. They argued that improved density should increase
mechanical properties. This is generally due to the elimination of the porosity between the layers,
and the thickness reduction under the influence of pressure.

As the mechanical properties of carbon fiber are higher than those of the matrix, the mechanical
properties of CFRP composite laminates would be mainly determined by their fiber volume
fraction. This is enshrined clearly in Figure 32c and Figure 32d, which represent the correlation
between fiber volume fraction, and LCM, ILSM, LCS, and ILSS. As already outlined, a higher
fiber volume fraction implies a small matrix volume fraction and, as a consequence, a reduction in
the amount of matrix surrounding the carbon fibers and decreased structural integrity of the
composite laminate [50].

Correlations between void content and mechanical properties are presented in Figure 32e and
Figure 32f. They show that the mechanical properties are inversely proportional to the void content,
especially between 0.05% and 5.56%. At low porosity, other structural defects such as fiber damage
or the prepregs sheet sliding because of high pressure could explain the decrease in composite
resistance. The low matrix volume fraction probably destroys the mechanical properties of
composite laminates. Carolina et al. [101] highlighted a decline of 27% in ILSS caused by the
presence of macroporosity that induced crack initiation and connection of different delamination
plies, causing the speeding up of crack propagation and jump of the interlaminar layer.

Our results are in good agreement with the theoretical hypothesis excepted at high pressure. Similar
behavior was reported by others [51,53,55,71]. Hernandez et al. [71] and Zhu et al. [53] concluded
that neither the resin nor the interface strength can explain the ILSS decrease of composites with
low voids content, and is probably caused by the difference in void morphology. Chang et al. [54]
stated that when the applied pressure is too high, the resin will flow too fast and will induce local
defects such as poor adhesion, fat rat, and distorted fibers.

74
Laminate Compressive Strength (MPa)
Laminate Compressive Modulus (MPa)
6000 500
26000 SBS 85 SBS
Interlaminar Shear Modulus (MPa)

Interlaminar Shear Strength (MPa)


SBS-120 SBS-120
CLC 80 CLC 450
24000
5500
75
22000 400
5000 70
20000
65 350

18000 4500 60
300
16000 55
(a) (b)
4000 250
1560 1580 1600 1620 1640 1560 1580 1600 1620 1640

Density (kg/m3) Density (kg/m3)

Laminate Compressive Strength (MPa)


500
6000 Laminate Compressive Modulus (MPa) 85 SBS

Interlaminar Shear Strength (MPa)


26000 SBS
Interlaminar Shear Modulus (MPa)

SBS-120 SBS-120
CLC 80 CLC 450
24000
5500
75
22000 400
70
5000
20000 350
65

18000 4500 60
300
55
16000
(c) (d)
4000 250
65 66 67 68 69 70 71 72 73 66 67 68 69 70 71 72
Fiber volume fraction (%) Fiber volume fraction (%)

Laminate Compressive Strength (MPa)


500
Laminate Compressive Modulus (MPa)

6000
26000 85
Interlaminar Shear Modulus (MPa)

SBS SBS
SBS-120 SBS-120
Short-Beam Strength (MPa)

CLC 80 CLC 450


24000
5500
75
22000 400
70
5000
20000
65 350

18000 4500 60
300
55
16000
(e) (f)
4000 250
0 2 4 6 0 2 4 6
Void content (%) Void content (%)

Figure 32 : Correlation between physical and mechanical properties: ILSM and LCM vs. density
(a), ILSS and LCS vs. density (b), ILSM and LCM vs. fiber volume fraction (c), ILSS and LCS vs.
fiber volume fraction (d), ILSM and LCM vs void content (e), and ILSS

75
The determination of the relationship between the various mechanical properties of 8552S/AS4
composite laminates will allow us to predict its behavior without conducting the particular test. As
the SBS test method is simpler than the CLC method, the LCM is plotted against ILSM (Figure
33a), and LCS is plotted as a function of ILSS (Figure 33b). LCM was found to present an
exponential fitting curve (R2 = 0,989) as a function of ILSM, whereas the LCS increased
proportionally with the ILSS ; this result being similar to that previously reported by Vo et al. [25].

5400 400
y = 4356.235 + 2.983*10-6 exp (7.722*10-4 x) y = 58.013 + 44.249 x
Compressive modulus (MPa)

R2 = 0.989
5200

Compressive strength (MPa)


R2 = 0.931
375
5000

4800
350
4600

4400
325
4200
(a) (b)
4000 300
15000 20000 25000 52 54 56 58 60
Interlaminar shear modulus (MPa) Interlaminar shear strength (MPa)

Figure 33 : Correlation between LCM and ILSM (a), and LCS as a function of ILSS (b)

3.7.Conclusions

The influence of applied pressure during the autoclave curing process on the quality of laminate
composites was studied. The curing pressure was fixed at either 0.1 MPa, 0.3 MPa, 0.5 MPa, and
0.7 MPa, while the temperature was constant at 180 oC and vacuum set to -0.02 MPa. At low curing
pressures, the laminate composites presented a higher void content than what is measured at high
pressure. The presence of voids, density, and fiber volume fraction was found to impact
significantly the mechanical properties. Increases in LCS, ILSS, and ILSS-120 of 9.04%, 18.22%,
and 10.52%, respectively, are observed when curing pressure is raised from 0.1 MPa to 0.5 MPa.
A comparison of ILSS at room and service temperatures shows the sensitivity of mechanical
properties to temperature variation.

76
Plots of mechanical properties as a function of porosity and pressure enable the selection of an
optimal curing pressure route, which minimizes porosity while maximizing the mechanical
properties of composite laminates of woven carbon/epoxy AS4/8552S.

3.8.Acknowledgments

The authors acknowledge the financial support received, respectively, from the EuroMed
University of Fes (Morocco), the Hassan II Academy of Sciences and Technologies (Morocco),
and Safran Nacelles (France).

77
CHAPITRE 4

EFFECT OF VACUUM-PRESSURE ON
MICROSTRUCTURE AND MECHANICAL
PROPERTIES OF AUTOCLAVED EPOXY/CARBON
COMPOSITE LAMINATES

4.1.Avant-propos

Après la détermination de la température et la pression optimales, il était par important par la suite
d’évaluer l’effet de la pression du vide appliquée lors de la fabrication en autoclave sur les
propriétés physiques et mécaniques des pièces composites. En effet, Il est toujours utile d'étudier
et de déterminer à quel niveau de variation des paramètres de traitement en autoclave est acceptable
sans changement significatif des propriétés mécaniques. Les résultats de ce présent chapitre seront
exploités dans le chapitre 5, traitant l’optimisation des paramètres de fabrication en autoclave en
utilisant les plans d’expérience.

4.2.Résumé

L'influence du vide appliqué lors de la fabrication en autoclave sur la microstructure et les


propriétés mécaniques a été étudiée pour un stratifié composite fibre de carbone/résine époxy,
AS4/8552S. Les stratifiés ont été fabriqué à des pressions de vide variant de -0,01 à -0,08 MPa et
leur épaisseur, densité, fraction volumique de la résine, fraction volumique de fibre et la teneur en
vides ont été établies en fonction de la pression du vide. Les tests de compression combinée et de
cisaillement interlaminaire ont été évalués à une température d’opération de 120°C. Les résultats
ont montré que l'épaisseur, la densité, la résistance à la compression et la résistance au cisaillement

79
interlaminaire varient de façon exponentielle avec la pression du vide appliquée et atteignent un
état stable à une pression de vide de -0,04MPa. L'analyse de corrélation de Pearson a révélé une
variation linéaire de la résistance au cisaillement interlaminaire et de la résistance à la compression
avec la densité, la fraction volumique des fibres et la teneur en vide, ainsi qu'entre le module de
compression et la densité. Les résultats de l'analyse statistique ANOVA unidirectionnelle ont
montré que les paramètres les plus importants affectant la résistance au cisaillement interlaminaire
sont la densité et la teneur en vides. Il a également été démontré que tous les paramètres de
microstructure physique affectent la résistance à la compression, tandis que la densité et la fraction
volumique des fibres sont les paramètres les plus significatifs qui affectent le module de
compression.

Mots-clés : composites stratifiés, pression de vide, microstructure, propriétés mécaniques, ANOVA


unidirectionnelle.

4.3.Abstract

The influence of autoclave cure vacuum on the microstructure and mechanical properties was
studied for a woven carbon fiber/epoxy resin composite laminate, AS4/8552S. The laminates were
cured at vacuum-pressures varied of -0.01 to -0.08 MPa and their corresponding thickness, density,
resin volume fraction, fiber volume fraction, and void content were established accordingly as a
function of vacuum-pressure. The combined loading compression and short beam shear mechanical
tests were determined under 120°C. The Results showed that thickness, density, laminate
compressive strength (LCS), and interlaminar shear strength (ILSS) varied exponentially with
applied vacuum-pressure and reached a steady-state at the applied vacuum of -0.04MPa. Pearson's
correlation analysis revealed linear variation in ILSS and LCS with density, fiber volume fraction,
and void content, as well as between laminate compressive modulus (LCM) and density. The results
of One Way ANOVA statistical analysis showed that the most important parameters affecting the
ILSS are the density and void content. It has also shown that all physical microstructure parameters
affect the LCS, whereas density and fiber volume fraction is the most significant parameters for
affecting the LCM.

80
Keywords: Composites laminates, Autoclave cure vacuum, Microstructure, Mechanical properties,
One Way ANOVA

4.4.Introduction

Composite materials based on thermosetting resins have been widely used over the past 40 years.
Large manufacturers such as Airbus and Boeing have shown interest in these materials. For example,
carbon fiber reinforced polymer composite (CFRP) makes up 50% of the Boeing 787-9’s weight,
which consumes 20% less fuel than any other aircraft in its category and thus reduced CO2 emissions
[6]. Besides, CFRP is adapted to complex design forms that are difficult to obtain with conventional
metals and alloys. The vast majority of the resins used are thermosetting resins: the crosslinking of
the chains during polymerization gives them an infusible character. In addition to their improved
mechanical properties, they are also characterized by undeniable disadvantages such as the need for
low-temperature storage, long curing time, and manual drapery generating the most irreversible
defects.

The mechanical properties of CFRP represent a critical role in different engineering applications,
specifically in the aeronautical industry. As composite laminate is commonly carbon fibers, resin
matrix, and void contents, the determination of microstructure seems to be important for many
reasons:

1) Carbon fibers are stronger than resin. As a result, the final mechanical properties of composite
materials depend on the fiber/resin ratio;

2) The compression loads are carried by the carbon fibers, while the resin distributed the loads
between the fibers in tension and prevented the fibers from buckling in compression. The resin is
also the primary load carrier for interlaminar shear [9];

3) Voids are one of the major defects affecting the mechanical properties of composites, particularly
compressive and interlaminar shear strength [63].

Although the aircraft parts are not subject to the same thermomechanical constraints; in the design
phase and according to the intended function, the materials are chosen for their physical-chemical
and mechanical properties. The recommended service temperature of CFRP composite laminates is
below their glass temperature Tg. In nacelles, structures housing aircraft engines, some parts can

81
reach 120°C. It is important to clarify the effect of elevated temperature on microstructure and
mechanical properties.

Autoclave processing is an important technique for manufacturing CFRP. In this process, a prepreg
lay-up is transformed into a very stiff structure. The properties of the final composite are highly
dependent on manufacturing conditions; cure pressure, vacuum pressure, holding temperature, and
time. The curing temperature and time influence the polymerization/crosslinking reactions of the
resin matrix. The applied pressure on the laminate surface inside the autoclave allows the CFRP to
conform to the mold shape and to be compacted to the desired fraction of fiber volume, thus reducing
the voids likely to form during the curing cycle. Vacuum pressure is used to remove any entrapped
air and volatiles during the curing cycle. In the literature review, many works are dedicated to
evaluating the impact of cure pressure [51,53,66] and holding temperature [60,102,103] on the
mechanical properties of autoclave-prepared composites; whereas, few studies have been focused
on the effect of vacuum pressure on the same properties. Sudarisman et al. [50] investigated the
influence of five different vacuum pressure on the microstructure of unidirectional CFRP
composites. It was found that the increase of vacuum pressure from -0.15 to -0.20MPa induced a
decrease of approximately 5.3% in fiber volume fraction. However, further increasing the vacuum
level from −0.20 to −0.30MPa produced a fiber volume fraction of approximately 58% which then
remained essentially constant as the vacuum level increased to −0.65MPa. Boey et al. [49] carried
out the influence of the bagging vacuum on the void content bending strength and the modulus.
They found that increasing the bagging vacuum led to a further significant reduction in void content
and improved flexural stress and flexural modulus. Koushyar et al. [26] explored the variation of
microstructure, thermal, and mechanical properties under two different vacuum pressure durations,
comprising vacuum vented at recommended pressure and vacuum held throughout the cure cycle.
The variation of the duration of vacuum-application has a minimal effect on the glass temperature
and degree of the cure. Contrariwise the void content was found to be increased significantly,
whereas the density decreased when the vacuum was maintained throughout the cycle.

Concerning the mechanical properties of CFRP, several studies have evolved, which assume that
strength and modulus can be attributed to physical microstructure parameters such as density and
fiber volume fraction, as well as void content. Up to date, the only studies that investigate the
relationship between the cured laminate density and mechanical properties are that of Koushyar et
82
al. [26] and Baghad et al. [80]. Their results showed an exponential increase in the interlaminar shear
strength (ILSS) with the density. As mentioned above, a various study has shown that the mechanical
properties are related to fiber volume fraction. Consequently, the strength can be improved with
increasing fiber volume fraction [104]. Void content has also been shown to affect mechanical
properties. It was reported that mechanical properties decrease with increasing void content
[54,58,105,106]. Several researchers have not found a clear relationship between void content and
mechanical properties [57,68,99].

This research aimed to elucidate the relationship between vacuum pressure and microstructure and
to clarify the relationship between vacuum pressure and mechanical properties. The other objective
of this study is to perform an analysis of the microstructure-mechanical properties correlation of
CFRP. To achieve these objectives, CFRP was manufactured using different vacuum-pressure. The
variation in microstructure and mechanical properties at 120°C was investigated based on measured
thickness, density, fiber volume fraction, void content, interlaminar shear strength, and laminate
compressive strength, as well as the laminate compressive modulus. Moreover, the correlation
between the microstructure and mechanical properties using Pearson correlation analysis, One Way
analysis of variance (ANOVA), and Scheffé and Tukey tests are examined and discussed.

4.5.Experimental procedure
4.5.1. Fabrication of composite laminates

The commercial prepreg manufactured by Hexcel Composites, HexPly 8552S/AS4 was used in this
study. It’s a 180°C curing epoxy matrix reinforced by carbon woven (37%) and designed for the
autoclave process. The woven carbon reinforcement used in this prepreg was a polyacrylonitrile
(PAN) high strength carbon fiber, AS4. This prepreg is designed for primary aerospace structures.

The stacking sequence of [0/90/-45/+45]2S was adopted to fabricate panels using 200 x 200 mm2
plies. The plies are laid up by hand to form composite laminates and cured in an autoclave
(manufactured by Taian Techtop Industries, China) under five different vacuum-pressure; 0MPa, -
0.01MPa, -0.02MPa, -0.03Pa, -0.04MPa, -0.06MPa, and -0.08MPa, with a pressure of 0.5MPa, a
holding temperature of 180°C, and a holding time of 90 min.

83
4.5.2. Microstructure analysis

Specimens for microstructural characterization were carefully cut from the composite panels.
Composite density was measured using the standard test methods for density and specific gravity
(Relative gravity) of plastics by displacement, ASTM D792-13 [87], using water as an immersion
liquid. Three specimens for each panel were measured and samples were approximately 1g in mass
(20 x 20 x t mm3). The density of a sample at 23°C can be calculated by:

a
𝜌𝑐 = 𝜌𝑤 ( ) (13)
𝑎−𝑏
Where:

- a: apparent mass of specimen in the air (g),


- b: apparent mass of specimen immersed in water (g), and
- 𝜌𝑤 : density of water (g/cm3)

The void contents, fiber volume contents, and resin volume content in the composite laminates
were measured according to standard test methods for constituent content of composite materials,
ASTM D3171-15 [88]. This method shows an uncertainty level of ±0.2 vol.%. three specimens
were tested from each panel. The void volume fraction (void content) of a composite laminate is
calculated by:

𝑊𝑟 𝑊𝑓
𝑉𝑉 = 100 − (𝑉𝑟 + 𝑉𝑓 ) = 100 − 𝜌𝑐 ( − ) (14)
𝜌𝑚 𝜌𝑓
Where:

- 𝑉𝑟 : resin volume fraction (%),


- 𝑉𝑓 : fiber volume fraction (%),
- 𝑊𝑟 : resin weight fraction (%),
- 𝑊𝑓 : fiber weight fraction (%),
- 𝜌𝑐 : density of the sample (g/cm3) calculated using Eq. 13,
- 𝜌𝑚 : density of the 8552S resin matrix (𝜌𝑚 = 1.3 𝑔/𝑐𝑚3 ), and
- 𝜌𝑓 : density of the carbon fiber (𝜌𝑓 = 1.77 𝑔/𝑐𝑚3 ).

The fiber weight fraction and the resin weight fraction are calculated as follows:

84
𝑀𝑓
𝑊𝑓 = ∗ 100 (15)
𝑀𝑖

𝑊𝑟 = 100 − 𝑊𝑓 (16)
Where Mi is the initial mass of the sample (g), and Mf is the final mass of the sample after digestion
(g).

The shapes and sizes of voids were measured using optical microscopy, Nikon Eclipse LV100ND
(Japan), equipped with a DS-Ri2 camera., and the images were analyzed through the open-source
image analysis software, Image J (National Institutes of Health, USA). The specimens for image
analysis were polished using a polishing machine with 400, 600, and 800 grit size silicon carbide
abrasive paper.

4.5.3. Mechanical tests

The mechanical tests were conducted on mechanical testing machine MTS Criterion model 45,
manufactured by MTS Systems Corporation, and equipped with a furnace thermcraft Lan-Tem
LBO (Thermcraft incorporated, USA). The tests were performed at a temperature of 120°C, and a
time of 10 min was imposed before starting each test according to our previous work [84]. Five to
seven tested specimens of each panel were analyzed.

The Short Beam Shear tests were performed according to NF EN 2563 [96]. The tests were
conducted with 10 kN load cells. The apparent interlaminar shear strength ILSS (MPa) of
composite samples was determined following Eq. 17.

3 PR
ILSS = (17)
4 w. t
Where PR (N) is the maximum applied load on the force-displacement curve, b is the specimen
gage width (mm) and h is the specimen gage thickness (mm).

The Combined Loading Compression (CLC) tests were conducted following the ASTM D6641
[95], and the tests were performed with 100 kN load cells. The laminate compressive modulus

85
(LCM) was calculated between 1000 and 3000 microstrain, and the Laminate Compressive
Strength LCS (MPa) was calculated according to:

Pf
LCS = (18)
w. t

4.5.4. Statistical analysis

Statistical analysis was conducted in Origin Lab program software using the ANOVA module (One
Way analysis of variance) with a significance level set to 0.05 for all analyses. In the first stage,
One-way analysis of variance (ANOVA) was employed for multiple comparisons of the properties
of composite laminates. If the overall p-value is less than 0.05, at least one of the group means is
different from the others. Since this does not tell us which groups are different from each other, we
performed a post-hoc test capable of controlling the error rate by family; Tukey and Scheffé tests
were used. The Scheffé test is mathematically related to the ANOVA F-test, while the Tukey test
is based on Student's range statistics to perform all group comparisons. Pearson's correlation test
was employed to evaluate the relationship between the density and the void content, fiber volume
fraction and void content, density and mechanical properties, fiber volume fraction, and mechanical
properties, as well as between void content and mechanical properties. The Pearson's correlation is
based on the covariance method which makes it the best method to measure the association between
two variables, and it gives information about the magnitude of the correlation, as well as the
direction of the relationship. The Pearson correlation coefficient (r) is expressed by the following
equation:

∑(𝑥𝑖 − 𝑥̅ )(𝑦𝑖 − 𝑦̅)


r= (19)
√∑(𝑥𝑖 − 𝑥̅ )2 ∑(𝑦𝑖 − 𝑦̅)2
Where 𝑥𝑖 is the x variable sample (ρ, 𝑉𝐹 , or 𝑉𝑉 ), 𝑦𝑖 is the y variable sample (ILSS, LCS, or LCM),
𝑥̅ and 𝑦̅ are the mean of values in x and y variables, respectively.

86
4.6.Results and discussions
4.6.1. Influence of vacuum-pressure on microstructure and mechanical properties

The laminated composites manufactured were carefully controlled in terms of resin and carbon
fiber. A visual inspection of vacuum bags showed no loss of fiber during the autoclave curing
process, while a loss of resin was observed depending on the applied vacuum-pressure; As we can
see, this is proven by optical microscopy images and acid digestion results.

3.5 1.7
R2 = 0.983

3.0 1.6

Density (g/cm3)
Thickness (mm)

Thickness
Exponential Fit of Thickness 1.5
2.5
Density
Exponential Fit of Density
1.4
2.0

R2 = 0.977 1.3
1.5
0.00 -0.02 -0.04 -0.06 -0.08
Vacuum-pressure (MPa)

Figure 34 : The experimental and fitted values of composites thickness and density as a function of
vacuum-pressure

Figure 34 shows the effect of vacuum-pressure on the thickness, as well as the cured laminate
density. The density is inversely proportional to the thickness. According to this Figure, the
thickness and density have an exponential function to vacuum-pressure. The results demonstrate
that the thickness decreased gradually from 3.3000mm to 1.9108mm when the vacuum-pressure
increased from 0MPa to -0.04MPa. When the vacuum-pressure increased from -0.04MPa to -
0.08MPa, the thickness of composites decreased smoothly from 1.9108mm to 1.8334mm. It can be
seen that the vacuum-pressure in the range of 0 and -0.04MPa reduced the thickness significantly

87
(by 42.10%), while thickness had a slight effect from the vacuum of -0.04MPa and beyond. On the
other hand, the composite density first improved progressively and then increased slightly with an
increase in vacuum pressure from 0MPa to -0.08MPa. When the vacuum pressure was applied (-
0.01MPa), the density increased rapidly from 1.3016g/cm3 to 1.5701g/cm3, an improvement of
20.63% due to the reduction of the thickness and removal of entrapped air during layup and
volatiles during cure. This ensures better contact between the layers. Doubling the vacuum-pressure
(-0.02MPa) raised the density by 21.39% as compared to that found without a vacuum bag. Density
of laminates continues to improve, moving from 1.5800g/cm3 at -0.02MPa to 1.6023g/cm3 at -
0.03MPa, and to 1.6180g/cm3 at -0.04MPa. Then, the density increased slowly from 1.6180g/cm3
to 1.6266g/cm3 when the vacuum-pressure increased from -0.04MPa to -0.08MPa. Therefore, the
effect of vacuum-pressure on the composite density was significant in the range 0 and -0.04MPa,
but haven’t significant effect (increased by 0.53% as compared to that found at -0.04MPa) in the
range of -0.04 and -0.08MPa. This is explained by the fact that vacuum-pressure conditions the
reduction of space separating the adjacent prepreg layers, and consequently the thickness (refer to
Figure 36), which increases the density of the laminates.

Figure 35 presents the measured void content, AS4 carbon fiber volume fraction, and 8552S resin
volume fraction of all the specimens according to the ASTM standard ASTM D3171-15 [88] as
described in section “Experimental procedure”. Figure 36 represents optical microscopy images of
each laminate.

100 20
(a) Fiber (b)
Resin
Volume fraction (%)

80 0.06
Void content (%)

15
0.04
60
10
0.02
40
0.00
5 -0.02 -0.03 -0.04 -0.06 -0.08
20

0 0
0 -0.01 -0.02 -0.03 -0.04 -0.06 -0.08 0 -0.01 -0.02 -0.03 -0.04 -0.06 -0.08
Vacuum-pressure (MPa) Vacuum-pressure (MPa)
Figure 35 : Evolution of fiber volume fraction and resin volume fraction (a), and void content as a
function of applied vacuum-pressure (b)

88
Figure 36 : Typical optical micrographs of cross-sections of composite laminates manufactured
under different vacuum-pressure taken at x50 magnification: (a) at 0MPa, (b) at -0.01MPa, (c) at
-0.02MPa, (d) at -0.03MPa, (e) at -0.04MPa, (f) at -0.06MPa, and (g) at 0.08MPa.
89
As can be seen, optical microscopy provides valuable additional information regarding the sizes,
distribution, and shapes of voids, in addition to resin and fiber distributions. Increasing the vacuum-
pressure decreased the resin volume fraction and void content, which in turn increased the fiber
volume fraction. The values of void content were found to vary from 0.022% at -0.08MPa to
17.306% at 0MPa, whereas the values of fiber volume fraction are situated between 29.218% at
0MPa and 34.492% at 0.08MPa. This suggests the importance of an appropriate vacuum level for
the production of high-quality laminates.

Without vacuum application, the fiber weight fraction (34.45%) and the resin weight fraction
(65.55%) of the laminate composite are almost equal to those of the commercial prepreg before
autoclave manufacture (37% and 63%, respectively); The composite retains the same starting mass
fraction. Since the thickness and density are too small (see Figure 34), the sum of fiber volume
fraction and resin volume fraction would be much less than 100% (Eq. 2), which will generate a
composite with high void contents (Figure 35). The study of composite by optical microscopy
(Figure 36a) confirms these results. The voids are large, having an irregular shape (meso-void)
[63], and are localized in resin-rich regions and between the plies. It also appears that the resin is
piled up between the plies.

The application of low vacuum-pressure causes a strong increase in the fiber volume fraction and
a sharp decrease in the resin volume fraction and the void volume fraction. For example, applying
the vacuum-pressure of -0.02MPa was seen to lead to a decrease of approximately 99.71% and
12.44% in void content and resin volume fraction, respectively. These decreases are accompanied
by an increase in fiber volume fraction by about 44.70%; The vacuum-pressure, therefore, helps to
reach an optimal fiber volume fraction. This shows that, even with low vacuum pressure, adjacent
prepreg layers would become bonded together (increase in density), which results in the elimination
of a significant amount of excess resin from the prepreg arrangement (increase in the fiber volume
fraction), and thus prevent any entrapped air during layup and between the layers (where most of
the voids were found in the composite produced without vacuum-pressure (Figure 36a)) from being
evacuated out of the laminate system. This behavior is similar to that previously observed by Boey
et al. [48]. However, further increasing the vacuum-pressure from -0.02MPa to -0.08MPa produced
composites with almost the same fiber volume fraction and resin volume fraction, with a small

90
decrease in void content. The microscopic images show that the void morphology remains constant
after -0.02MPa, displaying a small void content. These results were in contrast to those previously
reported by Sudarisman et al. [50] who reported that the fiber volume fraction and void content
remain constant from a vacuum level of -0.03MPa to -0.065MPa.

80
8 (b) R2 = 0.9967 400
(a) 70

LCS (MPa)
60 300

ILSS (MPa)
LCM (GPa)

R2 = 0.9976
50
4
200
40
ILSS
2 LCS
30 Exponential Fit of ILSS 100
Exponential Fit of LCS
0 20
0 -0.01 -0.02 -0.03 -0.04 -0.06 -0.08 0.00 -0.02 -0.04 -0.06 -0.08

Vacuum-pressure (MPa) Vacuum-pressure (MPa)

Figure 37 : Laminate compressive modulus (LCM) as a function of vacuum-pressure (a),


interlaminar shear strength (ILSS), and laminate compressive strength (LCS) as a function of
different cure vacuum (b)

The influence of vacuum pressure on LCM is presented in Figure 37a. The values of LCM were
found to vary from a minimum of approximately 3.6206GPa without vacuum-application to
6.6703GPa at -0.06MPa. The LCS and ILSS at 120°C for different cure vacuums are summered in
Figure 37b. From these results, it appears that increasing the cure vacuum had the effect of
increasing composite laminate strengths. It was observed that the LCS and ILSS increase with
increasing vacuum-pressure and then reach a plateau up to -0.03MPa. As displayed in the figure,
the LCS and ILSS have an exponential function to vacuum cure. The parameters for the fitting
function were determined using Origin software. The fitting coefficient (R2) reaches 0.996 and
0.997, respectively, which justifies that the exponential function fits the ILSS and LCS well. In
fact, this function may be used to predict the behavior of composite laminates. Without vacuum-
application, the LCS and ILSS recorded values of 92.2915MPa and 25.0647MPa respectively, and
then increased to reach 355.9448MPa and 56.3339MPa successively at an applied vacuum level of
-0.02MPa. The increase in LCS and ILSS by 2.86% and 2.96% are also observed when vacuum-
pressure is raised from -0.02MPa to -0.03MPa. Finally, the variation of LCS and ILSS starts to
91
reach their steady-state around the vacuum level of -0.04MPa. The strength improvement may be
attributed to a decline in the void content and the increase in fiber volume fraction and density.

Table 9. Descriptive statistics of composite properties

Properties Mean Standard deviation SE of mean


Density (g/cm3) 1.56008 0.11599 0.04384
Fiber volume fraction (%) 66.46008 8.33346 3.14975
Void content (%) 2.87153 6.43849 2.43352
ILSS (MPa) 52.88337 12.78311 4.83156
LCS (MPa) 321.94052 102.37175 38.69288
LCM (GPa) 5.52824 1.03341 0.39059

4.6.2. Statistical analysis and relationship between microstructure and mechanical


properties

Statistical descriptions of composite properties used in the analysis can be found in Table 9. In the
same way, Figure 38 shows the correlation between the density and ILSS and LCS, and LCM as a
function of density at operating temperature. One Way ANOVA statistical analysis and means
comparisons by using Tukey test and Scheffé test can be seen in Table 11. From Fig. 38, there is
clearly with the increased cured laminate density of composite laminates leads to an increase in
mechanical properties. Although the study conducted by Koushyar et al. [26] shows that ILSS
increased exponentially with density, the mechanical properties of the material concerned by our
study increased linearly with increasing density. Pearson’s correlation test revealed that a
statistically strong correlation was observed between the cured laminate density and ILSS and
between the density and LCS; Pearson’s correlation coefficient and respective p-values are
presented in Table 10. From results analysis One Way ANOVA can be noted the mean values of
all levels are equal, where at the 0.05 level, the population means are significantly different. It’s
also indicated from means comparisons of both tests that the significance number (Sig) equals 1,
which implies that the difference of the means is significant at the 0.05 level. Density is, therefore,

92
an excellent predictor of the quality of a composite laminate and a major characteristic taken into
account in the evaluation of the material for its use.

80 7.5
ILSS 400 LCM
7.0 Linear Fit of LCM
Linear Fit of ILSS
LCS 6.5

LCM (GPa)
60 Linear Fit of LCS 300 6.0

LCS (MPa)
ILSS (MPa)

5.5
5.0
200
40 4.5
4.0
100 3.5
(a) (b)
20 3.0
1.30 1.35 1.40 1.45 1.50 1.55 1.60 1.65 1.30 1.35 1.40 1.45 1.50 1.55 1.60 1.65
Density (g/cm3) Density (g/cm3)
Figure 38 : Relationship between density and mechanical properties: (a): Interlaminar shear
strength (ILSS) and laminate compressive strength (LCS) as a function of density, and (b):
Laminate compressive modulus (LCM) as a function of density. Pearson’s correlation test was
employed for each relation. The results indicate; ILSS (r = 0.99706 and p<0.00001), LCS (r =
0.99634 and p<0.00001), and LCM (r = 0.78348 and p = 0.03714).

Table 10 : Pearson’s correlation coefficient and respective p value between microstructure and
mechanical properties

Mechanical properties
Microstructure ILSS LCS LCM
r p r p r p
ρ 0.99706 <0.00001 0.99634 <0.00001 0.78348 0.03714
VF 0.99892 <0.00001 0.98473 <0.00001 0.67703 0.09479
VV -0.99093 <0.00001 -0.99844 <0.00001 -0.66544 0.10281

Figure 39 refers to the relation between mechanical properties and fiber volume fraction of
8552S/AS4 composite laminates. It can be clearly seen that, as the fiber volume fraction increased,
the LCS, ILSS, and LCM all showed an increasing tendency; This indicates that fiber volume
fraction variation might have a positive effect on the optimization of the mechanical properties.

93
Pearson correlation analysis (Pearson’s correlation coefficient and respective p-values can be seen
in Table 10) revealed a strong correlation between fiber volume fraction and ILSS (r = 0.99892 and
p<0.00001) , and between fiber volume fraction and LCS (r = 0.98473 and p<0.00001). The
statistical results of fiber volume fraction for LCM did not support completely the linear correlation
( r = 0.67703 and p = 0.09473). The reason for increasing mechanical properties is that AS4 carbon
fibers are stronger than the 8552S resin matrix. The effect of fiber volume fraction on mechanical
properties was also investigated in previous studies, and it was found that the strengths of the
composites increased with increasing fiber volume fraction [104,107]. The mechanical properties
of composite material will increase in proportion to the amount of fiber present. But when the
amount of fibers in a laminate is too large, the strengths will degrade [104]. Table 12 shows One
Way ANOVA statistical analysis and means comparison by using the Tukey test and Scheffé test
of the mechanical properties for the composite laminates with various AS4 fiber volume fractions.
From One Way ANOVA results turn out that the means of all levels are equal, where at the 0.05
level, the population means are significantly various. Based on Scheffé and Tukey tests, the
significance number (Sig) of all comparisons equals 1, indicating that the difference of the means
is significant at 0.05 level.

80 7.5
ILSS 400 LCM
7.0
70 LCS Linear Fit of LCM
Linear Fit of CLS 6.5
Linear Fit of ILSS
LCS (MPa)
ILSS (MPa)

LCM (GPa)

60 300 6.0
5.5
50
5.0
200
40 4.5
4.0
30 100
(a) 3.5 (b)
20 3.0
45 50 55 60 65 70 75 45 50 55 60 65 70 75
Fiber volume fraction (%) Fiber volume fraction (%)

Figure 39 : Relationship between fiber volume fraction and mechanical properties: (a):
Interlaminar shear strength (ILSS) and laminate compressive strength (LCS) as a function of fiber
volume fraction, and (b): Laminate compressive modulus (LCM) as a function of fiber volume
fraction. Pearson’s correlation test was employed for each relation. The results indicate; ILSS (r
= 0.99892 and p<0.00001), LCS (r = 0.98473 and p<0.00001), and LCM (r = 0.67703 and p =
0.09479).

94
Table 11 : One Way ANOVA and means comparisons by Scheffé and Tukey tests between density and mechanical properties

DF Sum of Squares Mean Square F-value p-value


Model 1 9219.2808 9219.2808 112.82828 1.85875E-7
Error 12 980.52875 81.71073
ILSS - ρ
Total 13 10199.80955
𝑹𝟐 = 𝟎. 𝟗𝟎𝟑𝟖𝟕
Model 1 359252.69232 359252.69232 68.55975 2.63890E-6
Error 12 62879.93180 5239.99432
LCS - ρ
Total 13 422132.62412
𝟐
𝑹 = 𝟎. 𝟖𝟓𝟏𝟎𝟒
Model 1 55.11204 55.11204 101.92887 3.22981E-7
Error 12 6.48829 0.54069
LCM - ρ
Total 13 61.60033
𝑹𝟐 = 𝟎. 𝟖𝟗𝟒𝟔𝟕
Means Comparisons
Scheffé test Tukey test
F-value Prob Sig q Value Prob Sig
ILSS - ρ 112.82828 1.85875E-7 1 15.02187 3.23510E-7 1
LCS - ρ 68.55975 2.6389E-6 1 11.7098 2.79467E-6 1
LCM - ρ 101.92887 3.22981E-7 1 14.27788 4.64000E-7 1

95
Table 12 : One Way ANOVA and means comparisons by Scheffé and Tukey tests between fiber volume fraction and mechanical properties

Overall ANOVA
DF Sum of Squares Mean Square F-value p-value
Model 1 645.1441 645.1441 5.54118 0.03645
Error 12 1397.12723 116.42727
ILSS - VF
Total 13 2042.27133
𝑹𝟐 = 𝟎. 𝟑𝟏𝟓𝟗𝟎
Model 1 228445.90218 228445.90218 43.30965 2.60882E-5
Error 12 63296.53028 5274.71086
LCS - VF
Total 13 291742.43246
𝑹𝟐 = 𝟎. 𝟕𝟖𝟑𝟎𝟒
Model 1 12994.4097 12994.40970 368.56013 2.24808E-10
Error 12 423.08677 35.25723
LCM - VF
Total 13 13417.49648
𝑹𝟐 = 𝟎. 𝟗𝟖𝟖𝟒𝟕
Means Comparisons
Scheffé test Tukey test
F-value Prob Sig q Value Prob Sig
ILSS - VF 5.54118 0.03645 1 3.32902 0.03645 1
LCS - VF 43.30965 2.60882E-5 1 9.30695 2.62556E-5 1
LCM - VF 368.56013 2.24808E-10 1 27.14996 1.25437E-7 1

96
Table 13 : One Way ANOVA and means comparisons by Scheffé and Tukey tests between void content and mechanical properties

Overall ANOVA
DF Sum of Squares Mean Square F-value p-value
Model 1 8754.14666 8754.14666 85.46376 8.31008E-7
Error 12 1229.17322 102.4311
ILSS - VV
Total 13 9983.31988
𝑹𝟐 = 𝟎. 𝟑𝟏𝟓𝟗𝟎
Model 1 356317.58244 356317.58244 67.73178 2.80980E-6
Error 12 63128.57627 5260.71469
LCS - VV
Total 13 419446.15871
𝑹𝟐 = 𝟎. 𝟕𝟖𝟑𝟎𝟒
Model 1 24.70348 24.70348 1.16191 0.30226
Error 12 255.13277 21.26106
LCM - VV
Total 13 279.83624
𝑹𝟐 = 𝟎. 𝟗𝟖𝟖𝟒𝟕
Means Comparisons
Scheffé test Tukey test
F-value Prob Sig q Value Prob Sig
ILSS - VV 85.46376 8.31008E-7 1 13.07392 9.78513E-7 1
LCS - VV 67.73178 2.80980E-6 1 11.63888 2.96600E-6 1
LCM - VV 1.16191 0.30226 1 1.52441 0.30226 1

97
80 7.5
ILSS 400
7.0
70 Linear Fit of ILSS
LCS 6.5
Linear Fit of LCS

LCS (MPa)
ILSS (MPa)

LCM (GPa)
60 300 6.0
5.5
50
5.0
200
40 4.5
4.0
30 100
(a) 3.5 (b)
20 3.0
0 4 8 12 16 20 0 4 8 12 16 20
Void content (%) Void content (%)

Figure 40 : Relationship between void and mechanical properties: (a): Interlaminar shear strength
(ILSS) and laminate compressive strength (LCS) as a function of void content, and (b): Laminate
compressive modulus (LCM) as a function of void content. Pearson’s correlation test was employed
for each relation. The results indicate; ILSS (r = -0.99622 and p<0.00001), LCS (r = -0.99844 and
p<0.00001), and LCM (r = -0.66544 and p = 0.10281).

The mechanical properties of AS4/8552S composite laminates with various void content are shown
in Figure 40. As can be expected, the mechanical properties are very sensitive to the void content,
and the maximum modulus and strengths correspond to the minimum porosity. Pearson correlation
analysis (Table 10) revealed very high linearity between void content and ILSS (r = -0.99622 and
p<0.00001) and between void content and LCS (r = -0.99844 and p<0.00001), whereas the linear
correlation between porosity and LCM is less important (r = -0.66544 and p = 0.10281). There are
several models in the literature to explain the effect of void content on the strengths of composites
laminates. Although studies have shown that strengths decrease exponentially as a function of
porosity [26], others have found that strengths decrease linearly with porosity [51]. Researchers
found that, below a critical void content level, the laminate strengths does not affect by the porosity
[55]. For example, Costa et al. [55] reported a critical void of 0.9% for carbon/BMI laminates and
epoxy carbon laminates. Chang et al. [108] show that when the porosity is less than 1%, the changes
of ILSS are negligible. Table 13 shows One Way ANOVA statistical analysis and means
comparison by using the Scheffé test and Tukey test for the mechanical properties of all specimens
with various porosity. Overall, ANOVA results of ILSS and LCS, the means of all levels are equal,
where at 0.05 level, the population means are significantly different. While, for LCM, the p-value
(0.30226) is higher than 0.05, which implies that the population means are not significantly
98
different and indicates strong evidence for the null hypothesis. The means comparisons by the
Scheffé test and Tukey test notes that the significance number (Sig) equals 1 for ILLS and LCS,
indicating that the difference of the means is significant at the 0.05 level. On the other hand, for
LCM, Sig equals 0, which implies that the difference of the means is not significant at the 0.05
level.

Based on One Way ANOVA and p-value results comparing mechanical properties to
microstructure (Table 10), we can conclude:

- For the ILSS, the p-values of density and void content are <0.00001, while the p-value of
fiber volume fraction is equal to 0.03645. It can be found that density and void content are
the most significant parameters for affecting the ILSS.
- For the LCS, the p-values of density, fiber volume fraction, and void content are <0.00001.
This implies that all parameters affect the LCS.
- For the LCM, the p-values of density and fiber volume fraction are <0.00001, while the p-
value of void content is equal to 0.30226. It can be found that density and fiber volume
fraction is the most significant parameters for affecting the LCM.

4.7.Conclusions

Under a given curing process, CFRP's were manufactured using different vacuum-pressure. The
variation in microstructure and mechanical properties at the tested vacuum-pressures was
investigated based on measured thickness, density, fiber volume fraction, void content, ILSS, LCS,
as well as the LCM. To study the relationship between microstructure and mechanical properties,
Pearson correlation analysis, and One Way ANOVA was performed, and Scheffé and Tukey tests
were utilized to analyze groups. Results showed that thickness, density, LCS, and ILSS have an
exponential function to vacuum-pressure. It was found that they increase with increasing vacuum-
pressure and then reach a plateau up to -0.04MPa. It can be concluded that this value represents the
optimal vacuum pressure. Pearson correlation analysis revealed very high linearity between ILSS
and LCS and different parameters defining the microstructure of the composite and between LCM
and density. In contrast, the correlation between LCM and fiber volume fraction and between LCM

99
and void content is less significant. Based on the results of One Way ANOVA statistical analysis
and means comparison by used Tukey and Scheffé tests:

- The most important parameters affecting the ILSS are the density and void content.
- All physical microstructure parameters (density, fiber volume fraction, void content) affect
the LCS.
- The most important parameters affecting the LCM are the density and fiber volume fraction.

4.8.Acknowledgements

This research was supported by Safran Nacelles (France), Hassan II Academy of Sciences and
Technology (Morocco), and Euromed University of Fes (Morocco).

100
CHAPITRE 5

APPLICATION OF RESPONSE SURFACE


METHODOLOGY FOR INVESTIGATION THE
IMPACT OF AUTOCLAVE PROCESS PARAMETERS
ON THE PHYSICAL PROPERTIES OF
EPOXY/CARBON LAMINATES

5.1.Avant-propos

Après la détermination de l’effet de chaque paramètre sur les propriétés physiques et mécaniques
des composites stratifiés, nous avons exploités les résultats des chapitres précédents (Chapitres 2-
5) pour déterminer les niveaux de chaque paramètre offrant des résultats optimaux. Dans le présent
chapitre, les plans de surface de réponse (Box-Behnken) ont été utilisés pour modéliser la courbure
de chaque paramètre et de déterminer les modèles empiriques des propriétés physiques des
composites stratifiés concernés par notre étude.

5.2.Résumé

Les paramètres de fabrication en autoclave sont décisifs pour contrôler les défauts et les propriétés
physiques et mécaniques des composites stratifiés. L'effet de la température, de la pression et du
vide sur le degré de polymérisation, la densité, la fraction volumique des fibres et la teneur en vide
des composites stratifiés a été étudié par un plan de surface de réponse et la méthode de Box-
Behnken, ainsi que par l'application de l'analyse de la variance (ANOVA), afin d'examiner
l'importance des modèles de régression. Les résultats ont montré que le degré de polymérisation
dépend fortement de la température. En même temps, pour la densité, l'effet linéaire de la pression

102
était le terme le plus significatif, suivi par l'interaction entre la température et la pression, les termes
quadratiques du vide et de la pression, et les termes linéaires de la température et du vide. Pour la
fraction volumique des fibres, tous les termes linéaires, les termes quadratiques de la pression et
du vide, et l'interaction entre la température et la pression étaient significatifs. Cependant, la teneur
en vide n'a que cinq termes significatifs (l'effet linéaire de la pression suivi de tous les termes
quadratiques et du terme linéaire du vide). Les réponses prédites à l'aide des modèles quadratiques
étaient en bonne concordance avec les valeurs expérimentales (les valeurs R2 étaient supérieures à
98 % pour toutes les réponses).

Mots-clés : Paramètres du processus autoclave, stratifiés composites en polymère renforcé de


fibres de carbone, plan de surface de réponse, propriétés physiques.

5.3.Abstract

Process parameters for the autoclave cure method are critical to controlling manufacturing induced
microstructure and composite laminates' physical and mechanical properties. The effect of
temperature, pressure, and vacuum on the degree of cure, density, fiber volume fraction, and void
content of laminated composites was investigated by response surface methodology (RSM) and
Box-Behnken Design (BBD), along with the application of analysis of variance (ANOVA), to
examine the importance of regression models. The results showed that the degree of cure highly
depends on the temperature. At the same time, for density, the linear effect of the pressure was the
most significant term, followed by the interaction between the temperature and pressure, the
quadratic terms of vacuum and pressure, and the linear terms of temperature and vacuum. For fiber
volume fraction, all the linear terms, quadratic terms of pressure and vacuum, and interaction
between temperature and pressure were significant. However, void content has only five significant
terms (the linear effect of the pressure followed by all the quadratic terms and the linear term of
vacuum). The predicted responses using quadratic models were in good concordance with the
experimental values (R2 values were above 98% for all responses).

Keywords: Autoclave process parameters, Carbon fiber reinforced polymer (CFRP) composite
laminates, Response surface methodology, physical properties.

103
5.4.Introduction

The autoclave process is established as one of the most common manufacturing processes of high-
performance structural carbon fiber reinforced polymer (CFRP) due to high applied pressure [63].
CFRP is used in many vital industries such as automotive and military [109]. However, their
development has traditionally come from aerospace, where laminated composites based on prepreg
and autoclaves are typically used [110]. Some of this interest is related to their high strength/weight
ratio, good fatigue life, high specific stiffness, high corrosive resistance, and low coefficient of
thermal expansion [2]. Furthermore, by reducing the weight of an aircraft using lightweight
materials, there will be an increase in fuel economy, improve flight range, and increase payload,
which will reduce operating costs and harmful emissions [7]. Some of the modern aircrafts such as
Boeing 787 and Airbus A350XWB contain more than 50 wt.% carbon fiber reinforced composite
laminates and sandwich structures [19].

To describe the mechanical behavior of laminated composites, researchers have to determine their
physical properties. Several studies have concluded that the mechanical properties are strongly
dependent on void content [26,51,53,55,57,58,80], fiber volume fraction [26,80,104], density
[26,80], and degree of cure [83]. Furthermore, these properties depend on the autoclave process
parameters, namely temperature, pressure, and vacuum [26,28,111–
113,49,50,53,54,59,73,80,108]. Grunenfelder et al. [65] announced that laminated composites
could be manufactured with almost no voids with an appropriate temperature, pressure, and
vacuum. This is mainly thanks to high pressure leading to the removal of volatiles, dissolved
species, and eliminating any entrapped air. In this context, a statistical approach that considers the
effects of autoclave process parameters on physical properties can be beneficial in fabricating
laminated composites with optimized performances.

Over the past few decades, research has concentrated on each parameter's effect on the physical
properties. However, a thorough analysis of their combined effect is still missing. In summary, we
have highlighted studies that examine the influence of each parameter in the autoclave process and
the responses analyzed by various researchers (Table 14). Stone et al. [73] evaluated the influence
of pressure and the vacuum-bag internal pressure on void content. Zhu et al. [53,111] showed that
void content increased with decreasing autoclave pressure (void content ranging from 0.4 to 9%),

104
and the voids are spherical and small at a pressure of 0.4 MPa and become more prominent and
elongated as the pressure goes from 0.4 to 0.0 MPa. However, voids' shape, size, and distribution
depend on autoclave process parameters [54,59,108]. Liu et al. [51] reported that pressure is
imperative to reducing the void content and decreasing exponentially.

Similarly, Gu. et al. [112] concluded that the void content of laminated composites decreased nearly
exponentially with an increase of pressure applied to two different prepregs. They also reported
that voids tended to grow at a higher temperature. However, low temperature should be used to
minimize the void content. Koushyar et al. [26,28] investigated the effects of variation in autoclave
parameters on physical properties. They found that the autoclave temperature does not affect void
content, and the degree of cure increased with increasing temperature. They concluded that density
dropped as the pressure decreased, and this drop was more pronounced for laminates, which the
vacuum was held throughout the cure cycle. The evolution of fiber volume fraction was quite
similar to that of density. One other conclusion from both studies is that the void content increased
significantly when the vacuum was maintained throughout the cure cycle. However, Boey et al.
[49] reported that the void content decreased with the increasing vacuum.

Table 14 : Effect of autoclave process parameters on the physical properties of laminated


composites

Autoclave process
Physical properties References
parameters
Pressure Void content [51,53,59,111]
Pressure Fiber volume fraction and void content [54,108]
Pressure Density, fiber volume fraction, resin volume [80]
fraction, and void content.
Pressure and temperature Void content [112]
Pressure and vacuum Void content [49,73,113]
Pressure, temperature, and Glass transition temperature, degree of cure, [26,28]
vacuum duration density, fiber volume fraction, and void content
Pressure, temperature, and Fiber volume fraction and void content [50]
vacuum

Although several investigations in open literature describe the effect of autoclave process
parameters on the physical properties of CFRP composite laminates, this study is the first to use
the RSM and BBD techniques and report about the interaction between the autoclave process

105
parameters. This study aims to determine the linear, quadratic, and interactions effects of three
independent parameters (temperature, pressure, and vacuum) on the physical properties (degree of
cure, density, fiber volume fraction, and void content) by the application of the RSM and three-
level BBD. In addition, the application of analysis of variance (ANOVA) was also used to examine
the importance of regression models at a 0.05 level of significance.

5.5.Methodology
5.5.1. Materials

The autoclave prepreg being used in this work consists of a 5-harness satin weave carbon fiber
(AS4-3k PAN) and toughened epoxy resin (8552S). This prepreg is acquired from Hexcel
Composites (France). The nominal resin content is 37% by weight, and the areal fiber weight is
280 g/m2. The 8552S epoxy resin is designed for curing at 180°C. The densities of resin and fiber
content are 𝜌𝑚 = 1300 𝑘𝑔/𝑚3 and 𝜌𝑓 = 1770 𝑘𝑔/𝑚3 , respectively. Storage conditions were
applied identically to those reported in our previous study [80]. Analytical grade chemicals used in
this study were purchased from Sigma-Aldrich and used without further purification.

Table 15 : Factors and levels considered for design

Coded parameter levels x1, x2, x3*


Parameters (Xj) Unit ∆𝑋𝑖
-1 0 +1
𝑋1 = 𝑇𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒 °C 160 180 200 20
𝑋2 = 𝑃𝑟𝑒𝑠𝑠𝑢𝑟𝑒 kPa 300 500 700 200
𝑋3 = 𝑉𝑎𝑐𝑢𝑢𝑚 kPa -40 -30 -20 10
* 𝑥1 = (𝑋1 − 180)/20; 𝑥2 = (𝑋2 − 500)/200; 𝑥3 = (𝑋3 + 30)/10.

5.5.2. Hand lay-up and autoclave curing

According to our previous work, the laminated composites were prepared using the Hand lay-up
method followed by the autoclave curing technique (Figure 41) [80,84]. In this study, a three-factor,
three-level Box-Behnken design (BBD) and response surface methodology (RSM) were used to
investigate the influences of autoclave process parameters on physical properties find the optimum

106
conditions in the preparation of laminated composites. A preliminary study has been carried out to
recognize main factors and choose appropriate ranges of the parameters. The main independent
variables in this study were the cure temperature (x1; 160-200 °C), the pressure (x2; 300-500 kPa),
and the vacuum-application (x3; -40--20 kPa). Based on preliminary studies using isothermal DSC
analysis, the holding time at isothermal temperaturew was 120 min. The Minitab version 18
statistical analysis software was used to create the Box-Behnken experimental design and analyze
the results. Three levels of each variable were coded as -1 (Lowest value of the variable), 0
(Medium value of the variable), and +1 (Highest value of the variable). The process parameters
and the corresponding levels are provided in Table 15. The relation between the coded values and
the real values of the independent variables were obtained using Eq. 20 [114]:

𝑋𝑖 − 𝑋𝑖0
𝑥𝑖 = (20)
∆𝑋𝑖
Where xi and Xi are a coded value and a real value of the variable, respectively; Xi0 is the real value
of the Xi at the center point, and ΔXi is the step change of the variable. Fifteen experiments,
including three points for the central point, were conducted, as shown in Table 16. The total number
of experiments can be calculated using the following formula [115]:

𝑁 = 𝑘 2 + 𝑘 + 𝑐𝑝 (21)
Where k is the number of factors (k = 3 in our study) and cp is the replicate number of the central
point. This study's responses were degree of cure, density, fiber volume fraction, and void content.
The regression model between the main independent variables (Xi) and the responses (Y) can be
generally written as [116]:

3 3 2 3
2
𝑌 = 𝛽0 + ∑ 𝛽𝑖 𝑋𝑖 + ∑ 𝛽𝑖𝑖 𝑋𝑖 + ∑ ∑ 𝛽𝑖𝑗 𝑋𝑖 𝑋𝑗 + 𝑒 (22)
𝑖=1 𝑖=1 𝑖=1 𝑗=𝑖+1

Where Y is the response; β0, βi, βii, and βij were the regression coefficients of variables for intercept,
linear, quadratic and interaction terms, successively; Xi and Xj are the coded levels of independent
variables, and e an unknown constant error vector. All experiments were carried out in triplicate,
and the average degree of cure, density (kg/m3), fiber volume fraction (%), and void content (%)
were taken as the response successively.

107
Figure 41 : Autoclave process schematic

5.5.3. Characterization

Differential scanning calorimeter (DSC) analyses were conducted with a DSC Q20 (TA
Instruments, USA). Dynamic analysis was performed from 35 to 300°C at a heating rate of
3°C/min. The final degree of cure of laminated composites is given by:

𝐻𝑇 − 𝐻𝑟𝑒𝑠
𝐷𝑂𝐶 = (23)
𝐻𝑇
Hres is the residual enthalpy of the laminated composite, and HT is the total enthalpy of the uncured
prepreg. As shown in our previous study, the uncured 8552S prepreg shows a total enthalpy of
211.61 J/g of prepreg [80].

108
Composite densities were measured using the standard test methods for density and specific gravity
(Relative gravity) of plastics by displacement, ASTM D792-20 [117], using water as an immersion
liquid. This technique is the must use in the industrial sector and shows an uncertainly level of ±0.2
vol.%. The sample size is 200 mm x 200 mm. The following equation was used to calculate the
cured laminate density (kg/m3):

𝑚𝑎
𝜌𝑐 = × 997.5 (24)
𝑚𝑎 − 𝑚𝑏
Where ma (kg) is the apparent mass of sample in air and mb (kg) is the apparent mass of sample
wholly immersed in water.

The samples were then treated by the acid digestion method, according to the standard test methods
for constituent content of composite materials, ASTM D3171-15 [88]. The objective is to determine
the fiber volume fraction and the void content. However, the constituent content is required to
evaluate the quality of a fabricated composite laminate and the cure cycle used during fabrication.
Therefore, the fiber volume fraction Vf (%) and the void content Vv (%) were determined according
to the following equations:

𝑚𝑓 𝜌𝑐
𝑉𝑓 = × 100 (25)
𝑚𝑎 𝜌𝑓
100𝜌𝑐
𝑉𝑣 = 100 − [(𝑚𝑎 − 𝑚𝑓 )𝜌𝑓 + 𝑚𝑓 𝜌𝑚 ] (26)
𝑚𝑎 𝜌𝑚 𝜌𝑓
Where mf (kg) is the final mass of the sample after acid digestion. ρm and ρf (kg/m3) are the densities
of the matrix and the carbon fibers, respectively.

5.6.Results and discussions


5.6.1. Experimental data

A three-factor, three-level BBD was utilized in that study to optimize autoclave process parameters
with improved density and fiber volume fraction and minimize the void content. Table 16 shows
the 15 numbers of the experiment design matrix and the experimental responses (degree of cure,
density, fiber volume fraction, and void content). As shown in this table, negative values were
obtained for the void content because the calculation assumes that densities of resin and fiber
109
remain constant during the autoclave curing, both of which vary slightly [80,88]. Therefore, the
data present in this table will be better assessed in the BBD described in the following sections.
The principle of this study is as follows: Firstly, the analysis of variance (ANOVA) and the
verification of the quadratic models fit are investigated. Then, the regression equations of the
experimental responses are secured after analyzing ANOVA results. This will enable us to identify
the most significant terms regarding confidence levels (95%, 99%, or 99.9%). In this research, we
consider the variables having p-values smaller than 0.05 as significant and those having p-values
smaller than 0.001 as very significant variables for the autoclave process. Finally, the response
surfaces plots will be generated to determine the expected optimal process parameters.

Table 16 : Box-Behnken for the experimental study and experimental responses

Variables Responses
Run
X1 X2 X3 Density Fiber volume Void content
no. DOC 3
(°C) (kPa) (kPa) (kg/m ) fraction (%) (%)
1 200 500 -20 0.99 ± 0.03 1604.5 ± 1.8 66.376 ± 0.406 0.572 ± 0.012
2 180 700 -40 0.90 ± 0.01 1625.8 ± 1.2 68.463 ± 0.264 -0.311 ± 0.002
3 160 500 -40 0.81 ± 0.01 1612.9 ± 1.2 67.437 ± 0.257 0.312 ± 0.003
4 180 300 -40 0.87 ± 0.03 1584.8 ± 0.9 71.944 ± 0.182 4.105 ± 0.005
5 200 700 -30 0.99 ± 0.02 1603.6 ± 0.8 65.717 ± 0.198 0.404 ± 0.009
6 200 300 -30 0.97 ± 0.02 1586.4 ± 1.3 74.665 ± 0.257 4.965 ± 0.009
7 180 500 -30 0.91 ± 0.02 1603.0 ± 0.6 64.564 ± 0.682 0.022 ± 0.011
8 180 300 -20 0.89 ± 0.02 1581.4 ± 4.4 71.461 ± 1.636 4.187 ± 0.256
9 200 500 -40 0.99 ± 0.02 1608.1 ± 1.3 66.325 ± 0.279 0.278 ± 0.005
10 180 500 -30 0.92 ± 0.01 1603.6 ±1.1 64.762 ± 0.700 0.041 ± 0.137
11 180 700 -20 0.91 ± 0.01 1624.4 ± 0.7 68.559 ± 0.683 -0.171 ± 0.213
12 160 500 -20 0.81 ± 0.02 1611.5 ± 1.3 67.758 ± 0.260 0.532 ± 0.008
13 160 700 -30 0.78 ± 0.02 1631.8 ± 2.5 71.640 ± 0.554 0.376 ± 0.011
14 160 300 -30 0.74 ± 0.03 1570.0 ± 2.3 70.830 ± 0.518 4.837 ± 0.012
15 180 500 -30 0.91 ± 0.02 1602.8 ±2.4 64.498 ± 0.420 0.071 ± 0.009
Experimental results are presented as mean ± standard deviation (SD).

5.6.2. Analysis of variance and quadratic model fitting

To find the significant process parameters affecting the density, fiber volume fraction, and void
content of laminated composites, Analysis of Variance (ANOVA) was performed. This statistical

110
technique can draw many essential conclusions based on the analysis of experimental results.
Indeed, the significance of the quadratic model and its conformity with experimental data are
checked by this statistical method [118]. In an ANOVA table, the degree of freedom (DF)
represents the experimental data amount. The adjusted sums of squares (Adj. SS) are the data used
to calculate the F-value (Fisher’s variance ratio) for a term. The adjusted mean squares (Adj. MS)
measure how much variation a term has. F-value is the amount of variation in the data about the
mean, and it is used to calculate the p-value. When the p-value is known, it is possible to decide
the statistical significance of each parameter [119]. Table 17 shows the results of the ANOVA
analysis. This table shows that ANOVA results divide the total sources into two sub-components:
sources concerning the quadratic model and sources for experimental error. By comparing these
sub-components, the significance of the quadratic model can be determined [120]. In Table 17, the
F-values obtained for the degree of cure, density, fiber volume fraction, and void content were
determined as 47.79, 664.63, 147.84, and 1399.45, respectively. F-values for all regressions are
higher than the critical values (27.24 at 99.9% confidence level) and the corresponding P-values
are very low (p-values < 0.001), indicating high prediction ability and adequacy of the quadratic
models [121,122]. A p-value of the model equal to 0.000 indicates that at least one term had a
significant effect on the quadratic model [119].

Another important parameter that should be considered is the lack of fit, defined as the data
variation around the fitted quadratic model. The lack of fit F-values was 9.25, 6.02, 8.43, and 11.56
for the degree of cure, density, fiber volume fraction, and void content, respectively. In contrast,
relevant p-values were determined to be 0.099, 0.146, 0.108, and 0.081, respectively. This indicates
that the lack of fit was statistically not significant; F-values < critical F-value (4.77 at 95% level
confidence) and p-values > 0.05. This implies that the quadratic model is appropriate for the
experimental results [123].

5.6.3. Estimation of quantitative effects of process parameters and regression models

The statistical significance of the linear, quadratic, and interaction effect of the process parameters
in the regression equation was determined by applying the p-values. However, for any of the terms
in the model, a lower p-value indicated a more significant effect on the physical properties of
111
laminated composites [124,125]. The regression model results, including coded coefficients and p-
value for each term, were reported in Table 18 for the degree of cure, density, fiber volume fraction,
and void content. For degree of cure, except for the temperature (p-value <0.001) and the pressure
(p-value = 0.04), as well as the pressure square (p-value = 0.018) all other terms are not significant.
However, the linear (temperature, pressure, and vacuum), square (pressure × pressure and vacuum
× vacuum), and the interaction temperature × pressure of the three autoclave process parameters
based on p-value are significant on the density. For fiber volume fraction, we notice that the linear
(temperature and pressure), square (temperature × temperature, pressure × pressure, and vacuum ×
vacuum), and interaction (temperature × pressure) coefficients of the three process parameters are
significant. Similarly, for void content, the linear effects of pressure (p-value <0.001) and vacuum
(p-value = 0.012), as well as all the quadratic interactions (p-values <0.05) are significant.

Table 17 : Analysis of variance (ANOVA) table for the validation of the model

Response Source DF Adj SS Adj MS F-Value P-Value Remark


Model 9 0.085302 0.009478 47.79 0.000 ***
Error 5 0.000992 0.000198
Degree of
Lack-of-Fit 3 0.000925 0.000308 9.25 0.099 ns
cure
Pure Error 2 0.000067 0.000033
Total 14 0.086293
Model 9 4159.26 462.14 664.63 0.000 ***
Error 5 3.48 0.70
Density Lack-of-Fit 3 3.13 1.04 6.02 0.146 ns
Pure Error 2 0.35 0.17
Total 14 4162.74
Model 9 137.145 15.2384 147.84 0.000 ***
Fiber Error 5 0.515 0.1031
volume Lack-of-Fit 3 0.478 0.1592 8.43 0.108 ns
fraction Pure Error 2 0.038 0.0189
Total 14 137.661
Model 9 56.3734 6.2637 1399.45 0.000 ***
Error 5 0.0224 0.0045
Degree of
Lack-of-Fit 3 0.0212 0.0071 11.56 0.081 ns
cure
Pure Error 2 0.0012 0.0006
Total 14 56.3958
*
Significant at p < 0.05; **Significant at p < 0.01; ***Significant at p < 0.001; nsNot significant

A quadratic polynomial model was chosen to describe the relationship between autoclave process
parameters and responses (Eq. 22). The mathematical models were fitted according to a confidence

112
level of 95% (Table 18). Empirical models obtained by the coded values for degree of cure, density,
fiber volume fraction, and void content are given in Eq. 27-30. In these equations, X1, X2, and X3
represent temperature, pressure, and independent vacuum variables, respectively. A negative sign
in front of the term expresses a synergistic influence, while a negative sign expresses an
antagonistic influence [114]. For example, it can be seen in Eq. 27 that temperature and pressure
had positive effects on the degree of cure, so higher parameters setting would result in a higher
degree of cure. The temperature had the greatest effect on the degree of cure among all significant
parameters due to its highest coefficient.

𝐷𝑂𝐶 = 0.91333 + 0.10000𝑋1 + 0.01375𝑋2 − 0.02542𝑋2 2 (27)


𝜌𝑐 = 1603.130 − 2.950𝑋1 + 20.375𝑋2 − 1.225𝑋3 − 5.167𝑋2 2 + 6.133𝑋3 2
(28)
− 11.15𝑋1 𝑋2
𝑉𝑓 = 64.608 − 0.573𝑋1 − 1.815𝑋2 + 1.486𝑋1 2 + 4.619𝑋2 2 + 0.880𝑋3 2
(29)
− 2.440𝑋1 𝑋2
𝑉𝑣 = −2.2245𝑋2 + 0.0920𝑋3 + 0.5359𝑋1 2 + 2.0649𝑋2 2 − 0.1571𝑋3 2 (30)

5.6.4. Main and interaction effects

Several process parameters such as temperature, pressure, and vacuum affect the physical
properties of laminated composites. By applying BBD and RSM methodology, it is possible to
optimize these process parameters and evaluate their possible interactions. The main effect of
process parameters for studied responses is visualized in Figure 42. As seen in this figure, the effect
of temperature is positive for the degree of cure, with an optimum value of up to 1. These results
go beyond previous studies, showing that the degree of cure increased with increasing the
isothermal temperature [26,28].

113
Table 18 : Estimated regression coefficients (as coded factors) and P-values for the degree of cure, density, fiber volume fraction, and
void content

DOC Density Fiber volume fraction Void content


Term Coded P-value Coded P-value Coded P-value Coded P-value
coefficients coefficients coefficients coefficients
Constant 0.91333 0.000*** 1603.13 0.000*** 64.608 0.000*** 0.0447 0.300ns
𝑋1 0.10000 0.000*** -2.950 0.000*** -0.573 0.004** 0.0202 0.431ns
𝑋2 0.01375 0.040* 20.375 0.000*** -1.815 0.000*** -2.2245 0.000***
𝑋3 0.00375 0.485ns -1.225 0.009** -0.002 0.987ns 0.0920 0.012*
𝑋1 2 -0.01792 0.058ns -0.017 0.971ns 1.486 0.000*** 0.5359 0.000***
𝑋2 2 -0.02542 0.018* -5.167 0.000*** 4.619 0.000*** 2.0649 0.000***
0.00458 0.559ns 6.133 0.000*** 0.880 0.003** -0.1571 0.006**
𝑋1 𝑋2 -0.00500 0.509ns -11.150 0.000*** -2.440 0.000*** -0.0250 0.488ns
𝑋1 𝑋3 -0.00000 1.000ns -0.550 0.244ns -0.067 0.692ns 0.0185 0.604ns
𝑋2 𝑋3 -0.00250 0.737ns 0.500 0.284ns 0.145 0.409ns 0.0145 0.683ns
*
Significant at p < 0.05; **Significant at p < 0.01; ***Significant at p < 0.001, nsNot significant

114
T emperature (°C) Pressure (kPa) Vacuum (kPa)
1.00
0.95
DOC

0.90
0.85
0.80
160 180 200 300 500 700 -40 -30 -20

T emperature (°C) Pressure (kPa) Vacuum (kPa)


1620
Density (kg/m3)

1610
1600
1590
1580
160 180 200 300 500 700 -40 -30 -20
Fiber volume fraction (%)

T emperature (°C) Pressure (kPa) Vacuum (kPa)


71
70
69
68
67
66
65
64
160 180 200 300 500 700 -40 -30 -20

T emperature (°C) Pressure (kPa) Vacuum (kPa)


Void content (%)

5
4
3
2
1
0

160 180 200 300 500 700 -40 -30 -20

Figure 42 : Main effect plots of process parameters for the degree of cure, density, fiber volume
fraction, and void content

Moreover, the density and fiber volume fraction variation is slight, characterized by a negative
slope for density and fiber volume fraction. In contrast, the temperature has a negligible effect on
the void content. These results agree well with the study by Sudarisman et al. [50], showing that
the void content was relatively constant. The fiber volume fraction firstly decreased and then
increased with increasing temperature. A similar conclusion was reached by Alavi-Soltani et al.

115
[28] concerning the variation of void content as a function of temperature. The high temperature
(200°C) causes the resin evaporation, which explains the slight increases in fiber volume fraction.

The pressure (second factor) proves to be a crucial process parameter, which determines the
physical properties of laminated composites, with positive effects for density and void content and
adverse effects for fiber volume fraction. These observations are attributed to the fact that the
adjacent prepreg layers would become bonded together at a faster rate with increasing pressure,
resulting in a loss of resin, inhibiting entrapped voids between layers, and improving the density of
laminated composites [50,51,54,80,99]. Figure 42 shows that starting from 500 kPa, the value of
fiber volume fraction and density increased with increasing pressure due to removing excess resin;
this result was similar to that previously reported by Sudarisman et al. [50]. On the other hand, an
increase in pressure does not lead to a significant variation in the degree of cure. This found
evidence that the degree of cure depends only on the temperature.

For the last process parameter (vacuum), the results in Figure 42 generally confirm the data
obtained from the estimated regression coefficients (Eqs. 27-30), including an inconsiderable
influence on the degree of cure, with a negligible effect on density, fiber volume fraction, and void
content. This is consistent with what has been found in previous studies investigating the effect of
vacuum on physical properties of laminated composites [26,49,50].

To find the most influential linear, quadratic, and interaction effect of the autoclave process
parameters on laminated composites' physical properties, the Pareto diagram at a 95% significant
level is illustrated in Figure 43, where the studied effects are shown. Those that exceeded the
reference line corresponding to the critical t-value are significant. In contrast, the ones that did not
exceed the line are insignificant and do not influence the degree of cure, density, fiber volume
fraction, and void content. The temperature was the most pronounced parameter influencing the
degree of cure (Figure 43a). Therefore, to improve the degree of cure, it is necessary to act on this
parameter. However, the linear and the quadratic terms of pressure also affected the degree of cure.
For density (Figure 43b), the linear effect of the pressure was the most significant term, followed
by the interaction between the temperature and pressure, the quadratic terms of vacuum and
pressure, and the linear terms of temperature and vacuum. The other terms were not significant.
Similarly, fiber volume fraction (Figure 43c) has six significant terms out of nine: linear terms,

116
quadratic terms of pressure and vacuum, and interaction between temperature and pressure. On the
other hand, void content presents only five significant terms as follows (Figure 43d): the linear
effect of the pressure followed by all the quadratic terms, and the linear term of vacuum, and all
interaction terms are not significant.

2.57 2.57
X1 X2
X22 X1X2
X2 X23
2
X 1
X22
Term

Term
X3 X1
X1X2 X3
X23 X1X3
X2X3 X2X3
X1X3 Y = DOC (a) X21 Y = Density (b)
0 5 10 15 20 0 10 20 30 40 50 60 70
Standardized Effect Standardized Effect

2.57 2.57
X22 X2
X2 X22
X1X2 X21
X 2
1
X23
Term

Term

2
X 3
X3
X1 X1
X2X3 X1X2
X1X3 X1X3
X3 Y = Fiber volume fraction (c) X2X3 Y = Void content (d)
0 5 10 15 20 25 30 0 10 20 30 40 50 60 70 80 90 100
Standardized Effect Standardized Effect

Figure 43 : Classification of effects on the degree of cure (a), density (b), fiber volume fraction (c),
and void content (d)

5.6.5. Linear correlations and model effectiveness

The predicted values of responses were obtained using the regression Eqs. 27-30 by substituting
the process parameters (X1, X2, and X3). Figure 44 shows graphs comparing the predicted and
experimental values of the degree of cure, density, fiber volume fraction, and void content. From
the results, the experimental and predicted values were not significantly different, indicating a high
correlation between them and that the proposed models adequately describe the experimental range
studied.

117
2
The adjusted coefficient of determination (𝑅𝐴𝑑𝑗 ) can be used as another way to verify the accuracy
2
of the regression models. A value of 𝑅𝐴𝑑𝑗 closer to 100% indicates that the regression model is
2
more valid [114]. However, the 𝑅𝐴𝑑𝑗 values for degree of cure, density, fiber volume fraction, and
void content are 98.76%, 99.91%, 99.60%, and 99.96%, respectively, which are close to 100%.
This implies that 98.76%, 99.91%, 99.60%, and 99.96% of the specimen variation is attributed to
the independent variables and that only 1.24% (degree of cure), 0.09% (density), 0.40% (fiber
volume fraction), and 0.04% (void content) of the total variation cannot be described by the
2
regression models [122]. As a result, the high correlation observed in Figure 44 and the higher 𝑅𝐴𝑑𝑗
values indicate that the regression models have great predictive ability, and are adequate to describe
the relationship between the physical properties and the autoclave process parameters.

Equation y = a + b*x
Weight No Weighting
Residual Sum of 9.80271E-4
Squares
Pearson's r 0.99424 1640
1.00 Adj. R-Square 0.98762
Value Standard Error
Predicted Data
(DOC)
Intercept
Slope 1630 0.01026
0.98851
0.02648
0.02956
Predicted Data (Density)
Predicted Data (DOC)

0.95 1620
R2Adj = 98.76% R2Adj = 99.91%

0.90 1610

1600
0.85
1590
0.80
1580
0.75 1570
(a) (b)
0.75 0.80 0.85 0.90 0.95 1.00 1570 1580 1590 1600 1610 1620 1630 1640
Experimental Data (DOC) Experimental Data (Density)
Equation y = a + b*x
Weight No Weighting
Residual Sum of 0.51344
Squares
76
Predicted Data (Fiber volume fraction)

Pearson's r 0.99813
Adj. R-Square 0.99597

Predicted Data Intercept


Value
0.25582 5 Standard Error
1.15858
Predicted Data (Void content)

(Fiber volume Slope 0.99626 0.01694


74 fraction)

4
72 R2Adj = 99.60% R2Adj = 99.96%
3
70
2
68
1
66
0
64 (c) (d)
-1
64 66 68 70 72 74 76 -1 0 1 2 3 4 5
Experimental Data (Fiber volume fraction) Experimental Data (Void content)

Figure 44 : The experimental values plotted against the predicted values derived from the empirical
models of responses: (a) degree of cure, (b) density, (c) fiber volume fraction, (d) and void content

118
700
DOC
(a) < 0.80
600 0.80 – 0.85

Pressure (kPa)
0.85 – 0.90
0.90 – 0.95
1 .0
500 > 0.95
DOC 0 .9
750 400
0.8
600
4 50 P ressure (kP a)
160
180 300
200 300
Temperature (°C) 160 170 180 190 200
Temperature (°C)

-20
DOC
(b) < 0.80
-25 0.80 – 0.85

Vacuum (kPa)
0.85 – 0.90
0.90 – 0.95
1.0 -30 0.95 – 1.00
DOC 0.9
> 1.00
-20 -35
0.8
-30
160 Vacuum (kP a)
180 -4 0
200 -40
Temperature (°C) 160 170 180 190 200
Temperature (°C)

-20
DOC
(c) < 0.88
-25 0.88 – 0.89
Vacuum (kPa)

0.89 – 0.90
0.90 – 0.91
0.92 -30 0.91 – 0.92
DOC 0 .90
> 0.92
-20 -35
0 .88
-3 0
300 Vacuum (kP a)
450 600 -40
750 -40
P ressure (kP a) 300 400 500 600 700
Pressure (kPa)

Figure 45 : 3D response surfaces and contour plots of the degree of cure with respect to autoclave
process parameters; (a) Temperature vs. pressure, (b) temperature vs. vacuum, and (c) pressure
vs. vacuum. The rest of the process parameters in each plot were set to level 0.

5.6.6. Contour and surface plots for responses

To visualize and validate the interaction influences of autoclave process parameters on the physical
properties of laminated composites, the 3D surface response plot and contour plot were obtained
for the hold values of temperature (180°C), pressure (500 kPa), and vacuum (-30 kPa). Figures 45
to 48 show the obtained results for degree of cure, density, fiber volume fraction, and void content,
119
respectively. Both the surface response and contour plots are based on coefficients given in Table
18.

700
Density
(a) (kg/m3)
600 < 1570

Pressure (kPa)
1570 – 1580
16 40
1580 – 1590
500 1590 – 1600
Density 1620 1600 – 1610
(kg/m3) 1600 1610 – 1620
750
1580 400 1620 – 1630
6000
4 50 P ressure (kP a)
> 1630
160
180 300
200 300
Temperature (°C) 160 170 180 190 200
Temperature (°C)

-20
Density
(b) (kg/m3)
-25 < 1602
Vacuum (kPa)

1602 – 1604
1604 – 1606
1612 -30 1606 – 1608
Density 1608 1608 – 1610
(kg/m3)
1604
1610 – 1612
-2 0 -35 > 1612
16 00 -3 0
1160
60 Vacuum (kP a)
180 -40
2 00 -40
Temperature (°C) 160 170 180 190 200
Temperature (°C)

-20
Density
(c) (kg/m3)
< 1580
-25
Vacuum (kPa)

1580 – 1590
1590 – 1600
1620 -30 1600 – 1610
Density 1610 – 1620
16 00 > 1620
(kg/m3) -2 0 -35
1580
-3 0
300
30 Vacuum (kP a)
450 600 -40
75 0 -40
P ressure (kP a) 300 400 500 600 700
Pressure (kPa)

Figure 46 : 3D response surfaces and contour plots of density with respect to autoclave process
parameters; (a) Temperature vs. pressure, (b) temperature vs. vacuum, and (c) pressure vs.
vacuum. The rest of the process parameters in each plot were set to level 0.

From the surface plot and contour plot of the degree of cure (Figure 45), it can be identified that
the most significant change range in the value of the degree of cure is caused by the temperature,
which confirms that this physical parameter largely depends on the temperature level. However,
120
the degree of cure was improved initially with increasing the pressure until reaching a maximum
at 500 kPa (> 0.92) and then was deteriorated with further increasing the pressure (Figure 45c).
This can be explained by the significant decrease of resin volume fraction with decreasing pressure.
The vacuum does not show any effect of degree of cure in the experimental range studied (Figure
45b). Further, it is also observed that the interaction effect between temperature and vacuum on the
degree of cure has little effect (Figure 45b). Compared to other interaction effects, the interaction
effect of pressure and temperature on the degree of cure is the closest to being significant. Low
vacuum (around -20 kPa) and medium pressure (around 500 kPa) lead to an improved degree of
cure (Figure 45c).

The effects of autoclave parameters on density are shown in Figure 46. The effect of pressure is
very intense when it is plotted against temperature (Figure 46a). At low pressures and low
temperatures, the density drops significantly, suggesting that the effect of temperature is probably
due to a low degree of cure in low temperature. When the pressure reaches approximately 500 kPa,
the effect of temperature becomes negligible. After exceeding this pressure, the density
significantly improved at low temperatures; the highest density was obtained at a temperature of
160°C and pressure of 700 kPa. It is maybe due to the viscosity of the resin. However, Hubert et
al. [20] showed that the gel time and the initial minimum viscosity decrease as the cure temperature
increases. Accordingly, if the pressure is too high, the resin would flow faster than expected, which
leads to unexpected defects in the local area such as poor glue, rich fat, and fibers distorting [108].

Similarly, from the surface plot and contour plot of fiber volume fraction (Figure 47), it can be seen
that the interactive effects of temperature and pressure have a significant role in fiber volume
fraction (Figure 47a). The highest fiber volume fraction was obtained at high temperature and low
pressure or low temperature and high pressure. At high pressure, more significant quantities of
excess resin were removed from the laminated composites. In contrast, a higher temperature will
increase resin evaporation, resulting in a higher value of fiber volume fraction [50]. Figures 46b-c
and 47b-c presenting the interaction of vacuum, respectively, with temperature and pressure
confirm no relationship between those interactions and the density and fiber volume fraction.
Figure 48 confirms the dependence of the void content on the pressure, and all the interactions are
statistically and physically significant to influence the void content.

121
700
Fiber
(a) volume
600 fraction

Pressure (kPa)
(%)
75
< 66
500 66 – 68
Fiber vo lume 70 68 – 70
fractio n (%) 70 – 72
750 400 72 – 74
65 6000
4 50 P ressure (kP a)
> 74
1160
60
180 300
200 300
Temperature (°C) 160 170 180 190 200
Temperature (°C)

-20
Fiber
(b) volume
-25 fraction (%)

Vacuum (kPa)
< 65.0
68
65.0 – 65.5
-30 65.5 – 66.0
Fiber vo lume 67 66.0 – 66.5
fractio n (%) 66 66.5 – 67.0
-2 0
65 -35 67.0 – 67.5
-3 0 > 67.5
1160
60 Vacuum (kP a)
180 -40
200 -40
Temperature (°C) 160 170 180 190 200
Temperature (°C)

-20
(c) Fiber
volume
-25 fraction
Vacuum (kPa)

(%)
72.5
< 66
-30 66 – 68
Fiber vo lume 70.0 68 – 70
fractio n (%) 67.5 70 – 72
-2 0 -35
65.0
> 72
-30
300
30 Vacuum (kP a)
450 600 -40
750 -40
P ressure (kP a) 300 400 500 600 700
Pressure (kPa)

Figure 47 : 3D response surfaces and contour plots of fiber volume fraction with respect to
autoclave process parameters; (a) Temperature vs. pressure, (b) temperature vs. vacuum, and (c)
pressure vs. vacuum. The rest of the process parameters in each plot were s

122
700
(a) Void
content
600 (%)

Pressure (kPa)
< 0
0 – 1
4.5 500 1 – 2
Vo id 3 .0 2 – 3
co ntent (%) 1.5 3 – 4
750 400
0.0 > 4
600
00
450 P ressure (kP a)
160
180 300
200 300
Temperature (°C) 160 170 180 190 200
Temperature (°C)

-20
Void
(b) content (%)
-25 < -0.2

Vacuum (kPa)
-0.2 – 0.0
0.0 – 0.2
0 .6 -30 0.2 – 0.4
Vo id 0 .3 0.4 – 0.6
co ntent (%) > 0.6
0.0 -20 -35
-30
160 Vacuum (kP a)
180 -40
200 -40
Temperature (°C) 160 170 180 190 200
Temperature (°C)

-20
Void
(c) content
-25 (%)
Vacuum (kPa)

< 0
0 – 1
4 .5
-30 1 – 2
Vo id 3 .0 2 – 3
co ntent (%) 1.5 3 – 4
-2 0 -35
0.0 > 4
-30
300 Vacuum (kP a)
450 600 -40
750 -40
P ressure (kP a) 300 400 500 600 700
Pressure (kPa)

Figure 48 : 3D response surfaces and contour plots of void content with respect to autoclave
process parameters; (a) Temperature vs. pressure, (b) temperature vs. vacuum, and (c) pressure
vs. vacuum. The rest of the process parameters in each plot were set to level 0.

5.7.Conclusions

Box Behnken Design (BBD) associated with the Response Surface Methodology (RSM) was
successfully used to design the experiments and evaluate the effects of autoclave process
parameters: temperature (160-200 °C), pressure (300-700 kPa), and vacuum (-40--20 kPa) on the
physical properties of epoxy/carbon composite laminates. From this study, the degree of cure
123
highly depends on the temperature. At the same time, for density, the linear effect of the pressure
was the most significant term, followed by the interaction between the temperature and pressure,
the quadratic terms of vacuum and pressure, and the linear terms of temperature and vacuum. For
fiber volume fraction, all the linear terms, quadratic terms of pressure and vacuum, and interaction
between temperature and pressure were significant. However, for void content, the linear effect of
the pressure was the most significant term, followed by all the quadratic terms and the linear term
of vacuum. The analysis of variance shows a high significance of the established models, with high
values of the coefficients of determination, thus ensuring a satisfactory fit of the quadratic model
with the experimental data.

5.8.Acknowledgments

The authors gratefully acknowledge the financial support received, successively, from the
EuroMed University of Fes (Morocco), the Hassan II Academy of Science and Technology
(Morocco), and Safran Nacelles (France).

124
CHAPITRE 6

EFFECTS OF HIGH OPERATING TEMPERATURES


AND HOLDING TIMES ON THERMOMECHANICAL
AND MECHANICAL PROPERTIES OF AUTOCLAVED
EPOXY/CARBON COMPOSITE LAMINATES

6.1.Avant-propos

Dans le cadre de la présente thèse financé par Safran Nacelles, l’Académie Hassan II des Sciences
et Techniques et l’Université Euromed de Fès, il était important d’évaluer l’effet de la température
de fonctionnement des moteurs des avions et le temps de maintien sur les propriétés mécaniques et
thermomécaniques, tout en respectant la température de transition vitreuse comme limite à ne pas
dépasser. Ce chapitre décrit alors les résultats obtenus au cours de cette étude.

6.2.Résumé

La température de fonctionnement élevée est considérée comme un facteur essentiel affectant les
performances des structures composites stratifiées. Cependant, dans les conditions d'utilisation
prévues dans les moteurs et nacelles d'avion, les polymères renforcés de fibres de carbone sont
exposés à des températures allant de la température ambiante à 120°C. Le présent travail se
concentre sur les propriétés thermomécaniques et mécaniques d'un stratifié composite fibre de
carbone/époxy tissé soumis à différentes températures de fonctionnement (60, 120 et 180°C) et
temps de maintien (10, 30 et 60 min). Les essais thermomécaniques ont montré que la rigidité et la
raideur de la matrice en résine époxy diminuent avec l'augmentation de la température. Cependant,

126
toutes les propriétés mécaniques (module de compression, résistance à la compression et résistance
au cisaillement interlaminaire) diminuent progressivement avec la température de fonctionnement.
La résistance à la compression du stratifié et la résistance au cisaillement interlaminaire ont diminué
jusqu'à 43,74 % et 30,60 % à la température de fonctionnement de 180 °C, respectivement, par
rapport aux valeurs mesurées à 60 °C. Cette étude indique que seules les propriétés de compression
sont affectées par le temps de maintien, tandis que l'interaction entre la température de
fonctionnement et le temps de maintien affecte le module de compression du stratifié.

Mots-clés : composites à matrice polymère, propriétés mécaniques, propriétés thermomécaniques,


Propriétés à haute température ; Temps de maintien.

6.3.Abstract

The elevated operating temperature is considered an essential factor affecting the performance of
laminated composite structures. Under the intended conditions of use in aircraft engines and
nacelles, carbon fiber reinforced polymers are exposed to temperatures ranging from room
temperature to 120°C. The present work concentrates on the thermomechanical and mechanical
properties of autoclaved woven carbon fiber/epoxy composite laminate subjected to different
operating temperatures (60, 120, and 180°C) and holding times (10, 30, and 60 min).
Thermomechanical results showed that the rigidity and stiffness of the epoxy resin matrix decrease
with increasing temperature. However, all mechanical properties (laminate compressive modulus,
laminate compressive strength, and interlaminar shear strength) decreased progressively with
operating temperature. Laminate compressive strength and interlaminar shear strength were seen
to decline up to 43.74% and 30.60% on the operating temperature of 180°C, respectively, compared
to measured values at 60°C. This study indicates that only the compression properties were affected
by the holding time, while the interaction between the operating temperature and holding time
affects the laminate compressive modulus.

Keywords: Polymer-matrix composites; Mechanical properties, Thermomechanical properties;


High-temperature properties; Holding time.

127
6.4.Introduction

Carbon fiber reinforced polymer composite (CFRP) has many applications in advanced technology
areas due to a very high strength/weight ratio. Composite materials based on thermosetting resins
have been widely used over the past 40 years. Large manufacturers such as Airbus and Boeing have
shown interest in these materials. For example, CFRP makes up 50% of the Boeing 787-9’s weight,
which consumes 20% less fuel than any other aircraft in its category [5], and thus reduced CO2
emissions [6]. In addition, CFRP is adapted to complex design forms that are difficult to obtain
with conventional metals and alloys. The vast majority of the resins used are thermosetting resins.
The resulting material is therefore not recyclable. Although they have interesting mechanical
properties, they are also characterized by undeniable disadvantages such as the need for low-
temperature storage, long curing time, and manual drapery generating the most irreversible defects
[19,23,79].

The aircraft parts are not subject to the same thermomechanical constraints. According to the
intended functions, the materials are chosen for their physical-chemical and mechanical properties
in the design phase. The recommended service temperature of CFRP composite laminates is below
their glass temperature Tg. In nacelles, structures housing aircraft engines, some parts can reach
120°C. In fact, it is crucial to clarify the relation between mechanical properties and elevated
temperature. However, the effect of high temperatures has already been investigated in many tests
studying the impact of water [126]. In addition, several works have been specifically conducted to
evaluate the influences of high operating temperatures on mechanical properties.

Tsotsis [127] evaluated the residual mechanical properties of CFRP exposed to high temperatures
in air and suggested that the compressive properties decreased significantly than tensile properties.
Later, Tsotsis [128] and Tsotsis et al. [129] revealed that weight loss could lead to significant errors
in estimating the degradation of mechanical properties. Akay et al. exposed CFRP to high
temperatures, evaluated interlaminar shear strength (ILSS) and impacted performance. This study
suggested that high temperatures led to significant degradation of the mechanical properties, mainly
because of a progressive loss of the resin matrix. Aktas et al. [130] have experimented with the
tensile, compressive, and shear properties of glass-epoxy composites at high temperatures. The
results show that the longitudinal modulus, transverse modulus, shear modulus, shear strength, and

128
interlaminar shear strength decrease by about 52%, 59%, 74%, 45%, 32%, respectively, when
increasing temperature from 20°C to 100°C. They also reported that transverse compressive
strength decreases slowly with increasing the operating temperature. By increasing the temperature,
the laminates become less stiff and failure occurs in a ductile manner [131]. Hawileh et al. [102]
experimentally investigated the mechanical properties of carbon, glass, and carbon-glass fiber
reinforced polymer for temperatures ranging from 25°C to 300°C. It was observed that elastic
modulus and tensile strength were degraded when samples were exposed to elevated temperatures.
Aklilu et al. [103] reported that CFRP's interlaminar shear strength and stiffness decreased when
the temperature increases. Properties deterioration can be attributed to chemical and/or physical
damages in the resin matrix, loss of adhesion at the fiber/matrix interface, and/or minimization of
fiber strength and stiffness [132]. Cao et al. [133] studied the effect of elevated temperatures
(ranging from 20 to 120°C) on the tensile strength of CFRP. They concluded that tensile strength
exhibits stable behavior below the glass transition temperature but drops rapidly during the glass
transition phase and finally reaches a plateau. Recently, Jia et al. [134] exposed CFRP to
temperatures ranging from -100°C to 100°C and evaluated their mechanical properties under static
and dynamic three-point bending tests. It was found that composites provide enhanced flexural
strength at lower temperatures (-60 and -100°C) while relatively poor strength at an elevated
temperature (100°C). Baghad et al. [80] studied the effect of operating engine temperature at 120°C
on interlaminar shear modulus, interlaminar shear strength, laminate compressive modulus, and
laminate compressive strength. Very limited number of studies were carried out to study the
performance of fiber reinforced polymer composite laminates over time at high temperatures [135–
139]. Li et al. [135] investigated the mechanical properties of glass fiber reinforced polymer
(GFRP) exposed to operating temperatures ranging from 100 to 350°C for a holding time of 0, 60
or 120min. Ashrafi et al. [136] studied the influence of elevated temperature and holding time (20
and 120 min) on tensile properties of CFRP. Badawy et al. [137] investigated the influence of
exposure temperature (-10 to 80°C) and exposure time (60 and 18min) on the impact behavior of
GFRP. The results showed that the mechanical properties are dependent on temperature. Moroever,
there have also been studies conducted to study the time-temperature creep behavior on CFRP
[140–142], while other researchers investigated the numerical simulation of the thermal stress of
the CFRP laminated composites [143–145]. However, Benli et al. [146] showed that the

129
contribution of thermal stress to impact damage increases with decreasing temperature, and that
the stress at low temperatures have a significant influence of the impact damage and parameters of
unidirectional CFRP composite laminates.

The present work aims to use a statistical design for the first time to evaluate the interactions
between the operating temperature and holding time of autoclaved composite laminates with a
commercial woven prepreg (8552S). Moreover, short beam shear and combined loading
compression tests were performed at temperatures ranging from 60 to 180°C and holding times
between 10 and 60 min. The changes in laminate compressive modulus (LCM), laminate
compressive strength (LCS), and interlaminar shear strength (ILSS) were investigated, and then
fitting the results into the temperature-mechanical properties dependent model proposed by Ha and
Springer.

6.5.Materials and methods


6.5.1. Materials

The commercial prepreg supplied by Hexcel Composites (France), HexPly 8552S/AS4 was used
in this study. It’s an autoclave prepreg with a 180°C curing epoxy matrix reinforced by carbon
woven (37% by weight), and designed for primary aerospace structures. The prepreg was stored in
hermetically sealed packing at -18°C. Before its use, prepreg under its packing was placed at room
temperature for 24h to remove moisture.

6.5.2. Hand lay-up and autoclave curing

The laminates have been fabricated in two steps; hand lay-up and autoclave curing process. Firstly,
the stacking sequence of [0/90/-45/+45]S was used to manufacture a composite laminate using
200*200 mm2 plies. Then, the plies are laid up by hand, and placed on an aluminum plate to obtain
a surface finish similar to the one of the lower plane in contact with the tool plate according to the
procedure reported by Dos Santos et al. [147]. Finally, the laminate was cured in an autoclave
(Taian Techtop Industries, China) using the curing cycle shown in Figure 49.

130
180 500
-60
Temperature (°C) 400

Pressure (kPa)

Vacuum (kPa)
120
300 -40

200
60
-20
100

0 0 0
0 50 100 150
Time (min)

Figure 49 : Autoclave cure cycle

6.5.3. Dynamic mechanical properties

The viscoelastic properties of composite laminates like storage modulus, loss modulus, and tan δ
were tested in a cantilever using DMA 850 (TA Instruments) with the frequency of 1 Hz and
amplitude of 2 µm according to ASTM D7028 [148]. The specimens' size was 60 x 13 x t mm3 (t
is the thickness). The temperature was increased at a rate of 5°C/min, from 30°C to 300°C, and
then cooled.

6.5.4. Mechanical properties

Samples were tested using mechanical testing machine MTS Criterion (model 45, MTS Systems
Corporation) equipped with a furnace thermcraft Lan-Tem LBO (Thermcraft incorporated, USA).
Combined loading compression (CLC) tests were performed basically according to ASTM D6641
[95]. The load cell used was 100 kN with a feed rate of 1.3 mm/min. The laminate compressive
strength LCS (MPa) was calculated according to:

131
PR
LCS = (31)
b. t
Where PR (N) is the maximum applied load on the force-displacement curve, b is the sample width
(b = 13 mm), and t is the sample thickness (mm). Six specimens were tested from each panel.

The short beam shear tests were conducted with a 10 kN load cell, a feed rate of 1 mm/min, and a
support span of 10mm according to NF EN 2563 [96]. The interlaminar shear strength ILSS (MPa)
of the laminate was determined following Eq. 32.

3 PR
ILSS = (32)
4 b. t
Where PR (N) is the maximum applied load on the force-displacement curve, b is the specimen
gage width (b = 10 mm), and t is the specimen gage thickness (mm).

Table 19 : Full Factorial Design for 8552S composite laminates

Experimental Operating Holding Experimental Operating Holding


condition temperature (°C) time (min) condition temperature (°C) time (min)
1 180 30 6 60 60
2 180 60 7 120 30
3 60 10 8 120 10
4 180 10 9 60 30
5 120 60

A full Factorial Design has been used to evaluate the effects of operating temperature (60, 120 and
180 °C) and holding time (10, 30 and 60 min) on the laminate compressive modulus (LCM),
laminate compressive strength (LCS), and interlaminar shear strength (ILSS) of composite
laminates. The design results in 9 experimental conditions, as shown in Table 19. Since the aircraft
engine nacelles can be subjected to a temperature up to 120°C, the mechanical properties under high
operating temperature are realized at the same condition. At the same time, however, we considered
the glass transition temperature of composite laminates as limits not to be exceeded. This is due to
the fact that the stiffness and strength decreased dramatically when exceeding the glass transition
temperature [149]. To determine the range of holding time, some preliminary tests were performed

132
at the middle value of the operating temperature (120°C). The experiments were conducted for
holding times of 10, 30, 60, 120 and 180 min, based on the procedure reported by Ashrafi et al [136],
who exposed samples to two different holding times; 20 min and 120 min. The results revealed that
the influence of the holding time on mechanical properties was more remarkable in the range of
holding time from 10 to 60 min.

In order to estimate the experimental error, three samples have been tested for each experimental
condition in two different runs [147,150], totaling 54 specimens. Experimental data were analyzed
using Minitab software v. 18, including Analysis of Variance (ANOVA) to establish the statistical
significance of the effects of holding temperature and time and to obtain their interactions on LCM,
LCS, and ILSS using a confidence level of 95% [147,150].

6.6.Results and discussion


6.6.1. Dynamic mechanical properties

A sample is subjected to Dynamic Mechanical Analysis (DMA) to measure their storage modulus
(E’), loss modulus (E”), and damping factor (tan δ) as a function of temperature (Figure 50). The
corresponding measured elements (E’; E’’ and tan δ) provide well information about stiffness,
degree of crosslinking, and carbon fiber/epoxy matrix interfacial bonding [151]. According to Figure
50, it can be seen that laminates are in the glassy state at a temperature below 150°C and the rubber
state above 200°C [152]. Thus, the glass transition region for laminate is from about 150 to 200°C,
with a glass temperature Tg of 200.23°C as a local maximum of the tan δ.

The E’ expresses the elastic portion of viscoelasticity and described the stiffness and rigidity of
the laminate [153,154]. The storage modulus curve is composed of three parts (Figure 50). The
laminate is very stiff and rigid in the first part due to the rigid polymeric chains [155]. In the second
part (glass transition phase), the decrease of E’ was more significant due to the mobility of the
polymeric chains matrix [156,157]. In addition, movement in the polymeric chains generally affects
the stiffness and carbon fiber/epoxy matrix adhesion [158]. Finally, during the last phase (rubber
state), E’ was almost constant because of the high polymeric chain movement at higher temperatures
[157]. Figure 50 also noticed that E’ remained stable between 30 and 60°C. On the other hand, a

133
slight reduction of about 1.8% to 11.49% of elastic modulus was observed in the temperature range
of 120°C and 180°C, respectively.

The E” is a viscous response of the composite. It is regarded as a laminate tendency to dissipate


energy applied to it [159]. From Figure 50, the E” increases, attained a maximum value in the glass
transition phase, and then declined. The damping factor also shows a similar trend as in the case of
loss modulus. Figure 50 shows that the increase of temperature from 120°C to 180°C enhanced the
E” by 30.24 and 278.66% successively than its value at 60°C. This enhancement is attributed to the
improved carbon fiber/epoxy matrix adhesion.

2000
0.10 30000
1800
1600

Storage modulus (MPa)


0.08
25000

Loss modulus (MPa)


1400
Tan()

0.06 1200
20000
1000
0.04 800
15000
600
0.02
400
10000
200
0.00
50 100 150 200 250
Temperature (°C)

Figure 50 : Dynamic mechanical properties of composite laminate

6.6.2. Mechanical properties


6.6.2.1. Experimental data

Table 20 illustrates the mean values and standard deviation of the LCM, LCS, and ILSS obtained
for autoclaved 8552S/AS4 laminate samples. As a general comment, it may be observed that the
LCS is six times higher than ILSS at different temperatures. This behavior is due to the fact that
the interlaminar shear loads are carried by matrix, while the compressive loads are carried by matrix

134
and carbon fibers. The resin matrix distributes the stress between the fibers in tension and stabilizes
them. It should be noted that the mechanical properties of carbon fibers are significant than those
of epoxy matrix.

Table 20 : Effect of experimental variables on mechanical properties

LCM (GPa) LCS (MPa) ILSS (MPa)

EC Replicate 1 Replicate 2 Replicate 1 Replicate 2 Replicate 1 Replicate 2

Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD

1 4.431 0.230 4.385 0.188 224.916 10.381 240.548 6.979 48.711 0.505 51.545 3.432

2 4.358 0.044 4.424 0.084 223.707 5.792 218.651 3.346 47.938 2.067 51.938 4.003

3 6.594 0.314 6.614 0.290 432.432 12.941 454.498 15.924 69.753 5.856 79.813 1.865

4 4.445 0.090 4.561 0.093 279.095 10.922 273.317 7.379 51.741 6.324 57.507 2.561

5 4.961 0.040 5.099 0.113 378.973 2.362 383.607 7.429 60.552 0.821 62.068 3.297

6 6.297 0.043 6.187 0.107 424.321 12.630 422.433 8.524 73.007 1.746 75.423 1.724

7 5.212 0.055 5.168 0.019 398.697 7.507 387.401 10.192 59.980 1.414 64.916 1.281

8 5.162 0.055 5.304 0.027 367.532 4.431 365.378 9.331 59.803 5.979 69.613 2.740

9 6.351 0.051 6.467 0.168 420.892 11.989 440.836 4.581 72.587 2.045 75.181 2.964

6.6.2.2. Statistical design


a. Laminate compressive modulus (LCM)

The results have been analyzed statistically using ANOVA in order to identify the parameters that
significantly affect the LCM of laminated composites. As seen in Table 21, the probability value for
the interaction between the operating temperature and holding time was 0.177, which is more than
0.05. Thus, this term did not influence LCM while the operating temperature and holding time were
adequate with a probability value greater or equal to 99%. The effect of operating temperature and
holding time on LCM of a composite laminate is quantified with a Pareto analysis at a 95%
significant level, as shown in Figure 51. This analysis provides the essential variable of both
individual effects and the interaction of all studied impacts [160]. The results show a reference line
on the Pareto chart corresponding to the critical t-value (t = 2.26). However, this analysis revealed

135
that, in the range of operation conditions studied, both operating temperature and holding time were
the influential variables (i.e., exceeding the reference line). Therefore, the effect of operating
temperature on LCM was more pronounced than that of holding time (Figure 51). Li et al. [135]
observed the same behavior for elastic modulus of GFRP exposed to operating temperatures ranging
from 100 to 350°C for a holding time of 0, 60 or 120min. They reported that the higher operating
temperature was, the more degradation influence of holding time on GFRP was. Ashrafi et al. [136]
conducted an experimental study on the tensile properties of GFRP at temperatures up to 300°C and
two different exposure times. The results showed that the elastic modulus decreased as the
temperature and holding time increased, and that the effect of exposure time was less important than
the effect of temperature. Another finding reported from this study showed that the effect of holding
time becomes important as the temperature increased, which is in accordance with our results (Table
20 and Figure 60). These results are discussed in detail later in the paper. Correia et al. [161] showed
that compressive properties such as as stiffness were remarkably reduced with temperature. This can
be attributed to the degradation of the matrix and the fiber-matrix interface [133].

Table 21 : Analysis of variance for laminate compressive modulus

Source DF Adj SS Adj MS F-Value P-Value


Model 8 12.3056 1.53819 318.37 0.000
Linear 4 12.2668 3.06670 634.73 0.000
Operating temperature 2 12.1142 6.05708 1253.66 0.000
Holding time 2 0.1526 0.07632 15.80 0.001
Operating temperature x holding time 4 0.0388 0.00969 2.01 0.177
Error 9 0.0435 0.00483
Total 17 12.3490

The interactive impact between maintaining time and using temperature was not significant, as
evident from Pareto charts. Those results agree with the main effect of operating conditions on the
LCM, as shown in Figure 52. Referring to this Figure, the line connecting 60 to 180°C has a steeper
slope than connecting 10 and 60 min. This phenomenon suggests that the operating temperature has
a significant role on LCM, and an increase of holding time is not affected much on LCM. It can be

136
seen from Figure 6.4a that the mean of LCM is 6.418 MPa with an operating temperature of 60°C.
The mean of LCM decreases with increasing the operating temperature and reaches 5.151 MPa at
120°C. There is a decrease of 19.74% in the mean of LCM when the operating temperature goes
from 60 to 120°C. When the operating temperature increased from 120 to 180°C, the mean of LCM
decreases from 5.151 MPa to 4.434 MPa, which the decline rate reached 30.74% to its values at
60°C. Thus, the modulus is typically related to the storage modulus (E’) [157].

2.26

Operating temperature
Term

Holding time

Opearting temperature
x
Holding time

0 10 20 30 40
Standardized Effect

Figure 51 : Pareto chart illustrating the effect of operating temperature and holding time on LCM
of composite laminates at 95% significance level, designed as a vertical dashed line

On the other hand, it can be seen from Figure 52b that the mean of LCM did not change significantly
with increasing holding time. As reported in the Figure, there is a decrease of 2.04% in the mean of
LCM when the holding time goes from 10 to 30 min. This decreases for laminate to 4.14% when
the holding time goes from 10 to 60 min. In conclusion, every 1min increase in holding time reduces
the mean of LCM by 0.069% for the 8552S/AS4 composite laminate, and their relationship is
approximately linear (See Figure 52b). This generally related to the time required for reaching
thermal equilibrium [162]. At low holding time, the target temperature is reached on the surface but
at high holding time, the core of the specimen reached the operating temperature [136].

137
6.5 19.74% 30.96%

Mean of LCM (MPa)


6.0

2.04% 4.14%
5.5
5.334

5.0

4.5
(a) (b)
60 120 180 10 30 60
Operating temperature (°C) Holding time (min)

Figure 52 : Main effects plot for the main laminate compressive modulus. The reference line is
designed as a longitudinal dashed line

Figure 53 illustrates the interaction effect plot for the mean of LCM. The interaction plot gives the
possible interaction between the mean of LCM of the studied factors. This Figure reveals that the
lines are nearly parallel. This interaction effect suggests that the change in the mean of LCM from
60 to 180 °C does not depend on the level of the holding time [163]. Thus, the Figure shows no
significant increase in LCM, even after 60 min of holding time. These results support well the
previous observations about interaction between the factors under evaluation. However, Figure 53
shows also that the effect of holding time declined slightly with increasing the operating
temperature. A behavior already observed by Li et al. [135] for the elastic modulus, in a holding
time between 0 and 60 min. Since the operating temperature may not reach the softening temperature
of carbon fibers, the degradation only occurred for the epoxy resin and the interface adhesion
between the fibers and the resin in laminated composites [135].

138
7.0
10 min
30 min
6.5 60 min
Mean of LCM (MPa)
6.0

5.5

5.0

4.5

4.0
60 120 180
Operating temperature (°C)

Figure 53 : Interaction plot for laminate compressive modulus

b. Laminate compressive strength (LCS)

Table 22 illustrates the ANOVA results for laminate compressive strength. According to these
results, It can be observed that operating temperature, holding time, and interaction between them
was effective on LCS with a probability value of more than 99%. The Pareto chart highlights the
most influential factors for the LCS (see Figure 54). The effect of operating temperature, holding
time, and interaction between both factors extend beyond the reference line. The operating
temperature is the most significant factor controlling LCS for the composite laminates, followed by
interaction between holding time and operating temperature. The effect of holding time is on the
limit of significance. These results agree with those of the mean of LCS and interaction effect plots
as shown in Figures 55 and 56, respectively. The same behavior was reported previously by Liu et
al. [138] who found that both the longitudinal and transverse compressive strength decreased
observably with the increasing temperature and time. However, the LCS is dependent on the
modulus of the epoxy resin [164].

139
Table 22 : Analysis of variance for laminate compressive strength

Source DF Adj SS Adj MS F-Value P-Value


Model 8 119031 14878.9 199.08 0.000
Linear 4 115752 28938.1 387.20 0.000
Operating temperature 2 114541 57270.5 766.30 0.000
Holding time 2 1211 605.7 8.10 0.010
Operating temperature x holding time 4 3279 819.7 10.97 0.002
Error 9 673 74.7
Total 17 119704

2.26

Operating temperature

Operating temperature
Term

x
Holding time

Holding time

0 10 20 30 40
Standardized Effect

Figure 54 : Pareto chart illustrating the effect of operating temperature and holding time on
laminate compressive strength at 95% significance level, designed as a vertical dashed line

In Figure 55, the mean of LCS was found to decrease with operating temperature and holding time.
This result confirms well the negative impact on the compressive strength of composite laminates.
Moreover, it is observed that the effect of holding time at constant temperature is less significant,
whereas the operating temperature has a substantial role on LCS at the 0.05 level. Referring to Figure
55a, the line connecting 120°C has a steeper slope than the line connecting 60 and 120°C. The loss
140
of strength reached 12.09% and 43.74% at 120 and 180°C, respectively. When subjecting to
operating temperature below Tg, the resin surface will remain rough similar to unconditioned
specimen, and some micro cracks will be appeared in the resin, whereas subjecting to temperature
slightly above Tg, the resin starts to soften, revealing the fibers position [165]. This confirms that
operating temperature around Tg affect the resin and result reduction in compressive strength of
laminated composites. In the same sense, but with a less critical effect, the performance values of
the laminate decrease with the holding time (Figure 55b). There is a decrease of 2.71% in mean of
LCS when the holding time goes from 10 to 30 min. When the holding time increased from 30 to
60 min, the mean of LCS decreases from 352.215 MPa to 341.949 MPa, which the decline rate
reached 5.55% to its values at 60°C. The compression test yields the mechanical properties of each
elementary ply-laminate; compressive loads are carried by carbon fibers while the resin spread the
charge between the fibers, stabilize them and while preventing their formation. As presented above,
high holding time results in a sample with a homogeneous temperature, which will push the
degradation of the resin and the interface adhesion between the fibers and the resin, and
consequently, the strength degradation of laminated composites.

450
12.09% 43.74%

400
Mean of LCS (MPa)

2.71% 5.55%

350 352.1

300

250
(a) (b)
60 120 180 10 30 60
Operating temperature (°C) Holding time (min)

Figure 55 : Main effects plot for the main laminate compressive strength. The reference line is
designed as a longitudinal dashed line

141
The operating temperature and holding time are said to interact with each other if the effect of
holding time on the LCS is different at different levels of the operating temperature [163]. For
understanding the interaction between factors under consideration, it is crucial to construct the
interaction plot. Figure 56 shows the mean of LCS versus the operating temperature for each level
of holding time. The above plot shows that lines are not parallel, and the influence of the holding
time is different at different operating temperature levels. This indicates that there is an interaction
between these two factors. The results show that the minimum in LCS is achieved when the
operating temperature is fixed to 180°C and the holding time kept 60 min. At 180°C, the interaction
between these factors leads to strength degradation of laminate, as shown in Figures 54 and 56. In a
study conducted by Tsotsis [127], the effect of increasing operating temperature was considered a
degrading factor of compressive properties. However, the compressive strength decreases as the
matrix stiffness is essential for avoiding fibre kinking [149]. Aktas et al. [130] have obtained that
compressive strength for unidirectional glass/epoxy decreased with increasing the temperature.
They reported that the compressive failure occurs in the composite due to fiber micro buckling
independent of changing temperature, and the epoxy matrix will be more ductile with increasing
temperature.

10 min
450 30 min
60 min
400
Mean of LCS (MPa)

350

300

250

200
60 120 180
Operating temperature (°C)

Figure 56 : Interaction plot for laminate compressive modulus

142
Table 23 : Analysis of variance for interlaminar shear strength

Source DF Adj SS Adj MS F-Value P-Value


Model 8 1591.03 198.878 12.18 0.001
Linear 4 1579.97 394.993 24.19 0.000
Operating temperature 2 1550.10 775.048 47.46 0.000
Holding time 2 29.88 14.939 0.91 0.435
Operating temperature x holding time 4 11.05 2.764 0.17 0.949
Error 9 146.97 16.330 0.17
Total 17 1738.00

2.26

Operating temperature
Term

Holding time

Operating temperature
x
Holding time

0 2 4 6 8
Standardized Effect

Figure 57 : Pareto chart illustrating the effect of operating temperature and holding time on
interlaminar shear strength at 95% significance level, designed as a vertical dashed line

c. Interlaminar shear strength (ILSS)

As the interlaminar shear loads are carried only by the matrix, the resin degradation is evaluated
throught interlaminar shear tests. Table 23 displays the ANOVA for interlaminar shear strength.

143
From this Table, it can be seen that the probability value for holding time and interaction between
the operating temperature and holding time were 0.435 and 0.949, respectively, which are more than
0.05, so these terms had no effect on ILSS. The operating temperature was the only significant term
with P ˃ 99%, implying that the loss of ILSS was almost the same whatever the holding time
between 10 and 60 min. Pareto charts in Figure 57 illustrate the most crucial variable of different
terms under evaluation.

75 15.44% 30.60%

70
Mean of ILSS (MPa)

3.94% 4.46%
65
62.89
60

55

50
(a) (b)
60 120 180 10 30 60
Operating temperature (°C) Holding time (min)

Figure 58 : Main effects plot for the main laminate compressive strength. The reference line is
designed as a longitudinal dashed line

In Figure 57, since the effect of the operating temperature is beyond the reference line, it is apparent
that operating temperature dominates the ILSS of composite laminates. This results is consistent
with that obtained by Aktas et al. [130]. Akay et al. [166] evaluated the ILSS and impact
performance of CFRP laminated composites at elevated temperatures and reported that high
temperatures could cause a progressive loss of matrix, and thus the degradation of the laminate.
Similarly, Li et al. [139] characterized the interlaminar shear properties of three-dimensional
orthogonal woven composites, and showed that the ILSS decreased with increasing temperature
ranging from 25°C to 150°C. This was associated with degradation of the epoxy resin and the

144
carbon/fiber/resin interface performane at high temperature. On the other hand, the holding time has
little influence on ILSS, while the interaction between the operating temperature and holding time
is almost negligible. These results are consistent with those of ANOVA (Table 23). A similar
conclusion was reached by Badawy [137], who studied the effect of exposure temperature (-10 to
80 °C) and exposure time (60 and 180 min). She found a marginal effect of holding time on the
impact strength of laminates for all temperature range.

10 min
75
30 min
60 min
70
Mean of ILSS (MPa)

65

60

55

50

60 120 180
Operating temperature (°C)

Figure 59 : Interaction plot for interlaminar shear strength

Main and interaction effects of the ILSS are visualized in Figures 58 and 59, respectively. As seen
in Figure 58, the main impact of the ILSS is similar to LCM and LCS presented above. The highest
ILSS value is provided at low temperature for a short holding time. A 30.60% reduction in ILSS is
observed at 180°C (Figure 58a). Like the LCM and LCS, the ILSS is not affected by holding time
(Figure 58b) since the operating temperature mainly dominates the mechanical properties. The
interaction effect reveals a considerable reduction in ILSS over holding time at 120 and 180°C,
while no substantial change occurs at 60°C (Figure 59).

145
d. Mechanical properties comparison

In order to compare our results, the mechanical properties have been normalized (Ratio between the
value at temperature T and the value obtained at room temperature). The normalized laminate
compressive modulus (LCMN), normalized compressive strength (LCSN), and normalized
interlaminar shear strength (ILSSN) were then fitted into the temperature-mechanical properties
dependent model proposed by Ha and Springer [167] and discussed above to find Tr and n laminate
parameters :

𝑇𝑟 − 𝑇 𝑛
𝐿𝐶𝑀𝑁 = ( ) (33)
𝑇𝑟 − 𝑇0
𝑇𝑟 − 𝑇 𝑛
𝐿𝐶𝑆𝑁 = ( ) (34)
𝑇𝑟 − 𝑇0
𝑇𝑟 − 𝑇 𝑛
𝐼𝐿𝑆𝑆𝑁 = ( ) (35)
𝑇𝑟 − 𝑇0
Where:

- LCMN represents the ratio of LCM at temperature T to the LCM0 = 6.758 GPa at T0 = 25°C,

- LCSN corresponds the ratio of LCS at temperature T to the LCS0 = 489.952 MPa at T0,

- ILSSN is the ratio of ILSS at temperature T to the ILSS0 = 81.788 MPa at T0,

- Tr is a reference temperature (Temperature at which the stiffness becomes negligible), and

- n is a constant value between 0 and 1.

The normalized mechanical properties were regressed into the model (Figure 60), and the
corresponding fitting results are listed in Table 24. We note that each data point in the plots
represents an average of the two replicate means (Table 20). The first observation is that all the
mechanical properties decrease by increasing operating temperature. The second observation that
characterizes the curves is its linear decrease in LCMN and ILSSN with increasing operating
temperature. These results were consistent with previous research findings [103].

Extending the holding time at isotherm temperature could also reduce mechanical properties.
However, as showed in Figure 60, the effect of holding time reduction on LCMN, LCSN, and ILSSN
was very limited. For example, the LCMN, LCSN, and ILSSN decreased from 0.997, 0.905, and 0.914

146
at 10 min to 0.948, 0.879, and 0.903 at 30 min, then reduced slightly to 0.924, 0.864, and 0.907 at
60 min, respectively, when the operating temperature was 60°C. At high operating temperature
(180°C), the LCMN, LCSN, and ILSSN decreased by 2.10%, 15.63%, and 8.23%, respectively, as
the holding time goes from 10 to 30 min. However, we observe a slight change between 30 and 60
min (i.e., the LCMN, LCSN, and ILSSN reduced by 0.46%, 5.05%, and 1.30%, respectively, as the
holding time goes from 30 to 60 min), which indicates that any further increase of holding time
above 60 min will not result in a reduction of the mechanical properties.

Table 24 : Results of the curve fitting of the experimental LCMN, LCSN, and ILSSN into the model
presented in Equations (3), (4), and (5)

Parameters Holding time (min) Tr (°C) n R2adjusted


10 485.654 1.000 0.952
LCMN 30 473.443 1.000 0.982
60 463.905 1.000 0.967
10 300.264 0.690 1.000
LCSN 30 200.712 0.346 0.970
60 198.108 0.351 0.972
10 503.425 1.000 0.949
ILSSN 30 425.571 0.999 0.999
60 420.029 1.000 0.995

Another unusual behavior is that the experimental and predicted results of the LCMN, LCSN, and
ILSSN appear to be in good agreement, as indicated by the relatively high values of the adjusted R
square (Table 24). In addition, the fitting curve parameters (Tr and n) offers insights into the effect
of operating conditions on the mechanical performance of 8552S composite laminates. First, the
values of Tr are generally higher for LCMN and ILSSN than LCSN, as illustrates in Table 24. Another
notable behavior of Tr is its descending behavior as a function of holding time.

The decrease in compressive strength was apparent much sooner than the interlaminar shear
strength. As a finding of fact, the compressive properties of CFRP composite laminates should be
used as a measure of deterioration since they exhibited a more pronounced reduction than

147
interlaminar properties under elevated operating temperatures and holding times. The same results
have been reported by Tsotsis [127] by comparing the degradation of compressive and tensile
properties of the fiber-reinforced polymer during high-temperature exposure, whereas Aklilu et al.
[103] reported a linear degradation in mechanical properties under increasing temperatures.

(a) (b)
1.1 1.1
1.0
1.0
0.9
0.9
0.8
LCMN

LCSN
0.8 0.7
0.6
0.7
0.5
0.6
0.4
10 min 30 min 60 min 10 min 30 min 60 min
0.5 0.3
0 50 100 150 0 50 100 150 0 50 100 150 0 50 100 150 0 50 100 150 0 50 100 150
Operating temperature (°C) Operating temperature (°C)

(c)
1.1
1.0
0.9
0.8
ILSSN

0.7
0.6
0.5
0.4
10 min 30 min 60 min
0.3
0 50 100 150 0 50 100 150 0 50 100 150
Operating temperature (°C)

Figure 60 : Normalized laminate compressive modulus (a), normalized laminate compressive


strength (b), and normalized interlaminar shear strength (c) as a function of holding time and
operating temperature

6.7.Conclusions

This study focused on the influence of operating temperature and holding time on the
thermomechanical and mechanical properties of 8552S/AS4 laminated composites. The DMA
results revealed that the elastic properties decreased with increasing operating temperature. Full
148
experimental design was employed to evaluate the effect of the interaction between the main factors
and their influence on the mechanical properties. All the main factors and their interaction affected
significantly the LCS of laminated composites. The LCM was affected primarily by the main
factors, while the ILSS is a function of operating temperature only. However, LCS and ILSS were
seen to decline up to 43.74% and 30.60% on the operating temperature of 180°C, respectively,
compared to measured values at 60°C. This clearly demonstrates that LCS is more sensitive to
operating temperature compared to the ILSS. Consequently, it is recommended to use the
compressive properties to evaluate the strength degradation.

6.8.Acknowledgements

The authors thank the Safran Nacelles (France), the EuroMed University of Fes (Morocco), and
the Hassan II Academy of Sciences and Technologies (Morocco) for providing the financial
support to undertake this work.

149
CONCLUSION GENERALE
CONCLUSION GENERALE

L’objectif principal de ces travaux de recherche est d’effectuer une analyse physico-chimique,
mécanique et thermomécanique de la résine 8552S dans le but est d’identifier les paramètres clés,
afin d’améliorer les propriétés mécaniques tout en préservant les propriétés physiques. Pour ce
faire, certains sous-objectifs se doivent d’être atteints. Dans un premier temps, l’objectif était de
déterminer la cinétique de la réaction de polymérisation/réticulation de la résine époxyde, et de
valider un modèle cinétique permettant de décrire son comportement dans un cycle de température
et de temps. Par la suite, il était important de stimuler les cycles thermiques de fabrication en
autoclave en utilisant une technique simple, dans cas il s’agit de la DSC, afin de démonter le temps
minimal de maintien à un palier de température, et bien évidement de déterminer l’évolution du
degré de réticulation au cours du temps. Il était par la suite important d’évaluer l’effet de la
température, de la pression et du vide sur les propriétés microstructurales, physiques,
thermomécaniques et mécaniques et déterminer les conditions permettant d’atteindre notre objectif
principal. De ce fait, et afin de comprendre le rôle de chaque paramètre de durcissement, nous
avons variés les paramètres en suivant un plan de Box-Behnken. Comme la température de
fonctionnement des nacelles des avions est de 120°C, il était important d’étudier l’effet de la
température d’opération et du temps de maintien sur les propriétés mécaniques et
thermomécanique. Ces objectifs ont pu être atteint à l’aide des nombreux plans d’expériences et
de travaux de caractérisation physico-chimiques, thermomécaniques et mécaniques qui furent mis
à l’exécution dans le cadre de ce projet de recherche et développement, financé par l’Académie
Hassan II des Sciences et Techniques, Safran Nacelles et l’Université Euromed de Fès.

Dans la première partie, malgré que la composition de la résine est peu connue, la caractérisation
par DSC de la résine 8552S/AS4 permet de dévoiler les divers mécanismes cinétiques intervenant
au cours de la réaction de polymérisation et de réticulation de la résine. Pourtant, l’utilisation des
modèles cinétiques en mode isotherme et/ou en mode dynamique, dédiés aux systèmes
époxy/amines montrent excellentes résultats. Comme les informations dont nous disposons sont
insuffisantes pour l’application de modèles mécanistiques, le modèle phénoménologique de Kamal

151
et Sourour est le plus adéquat. L’exploitation de ces résultats de modélisation cinétique aboutit à la
détermination des températures de durcissement possibles en autoclave. Ainsi, une étude a été
réalisée en appliquant une vitesse de chauffe de 3°C/min et différents paliers de température afin
de déterminer le degré de conversion maximale à chaque température, et évidement, le temps
nécessaire pour atteindre ce degré de conversion.

Dans la deuxième partie, nous avons exploité ces résultats lors de la fabrication en autoclave des
composites stratifiés à base de 8552S/AS4. En effet, et afin de déterminer les pressions possibles à
appliquer lors de la fabrication en autoclave, des composites stratifiés ont été fabriquer à différentes
pressions (0,1 MPa, 0,3 MPa, 0,5 MPa, et 0,7 MPa) et à température (180 °C) et vide (-0,02 MPa)
constants. L’objectif était d’étudier l’effet de la variation de la pression sur les propriétés
structurales et physique dans un premier temps. Par la suite, d’étudier son effet sur les propriétés
thermomécaniques et mécaniques. Les résultats obtenus montrent qu’aux faibles pressions, les
composites stratifiés présentaient une teneur en vides plus élevée que ce qui est mesuré à haute
pression. La présence de vides, la densité et la fraction volumique des fibres ont un impact
significatif sur les propriétés mécaniques. Des améliorations de la résistance à la compression et de
la résistance au cisaillement interlaminaire à température ambiante et à 120 °C de 9,04%, 18,22%
et 10,52%, respectivement, sont observées lorsque la pression de fabrication en autoclave passe de
0,1 MPa à 0,5 MPa. Une comparaison de la résistance au cisaillement interlaminaire à température
ambiante et à 120 °C montre la sensibilité des propriétés mécaniques aux variations de température
de fonctionnement. Les variations des propriétés mécaniques en fonction de la porosité et de la
pression permettent de sélectionner une voie de pression de durcissement optimale, qui minimise
le taux de vides tout en maximisant les propriétés mécaniques des stratifiés composites
carbone/époxy.

Ensuite, après avoir déterminé la température et la pression optimales pour la fabrication en


autoclave des pièces de haute qualité, l’effet de la pression du vide a été examiné. La variation de
la microstructure et des propriétés physiques et mécaniques en fonction de la pression du vide a été
étudiée en se basant sur l'épaisseur des composites stratifiés, la densité, la fraction volumique des
fibres, la teneur en vides, la résistance au cisaillement interlaminaire, la résistance à la compression,
ainsi que le module de compression. Pour étudier la relation entre la microstructure et les propriétés

152
mécaniques, une analyse de corrélation de Pearson et l’analyse de la variance unidirectionnelle ont
été réalisées, et les tests de Scheffé et de Tukey ont été utilisés pour analyser les groupes. Les
résultats ont montré que l'épaisseur, la densité, la résistance au cisaillement interlaminaire et la
résistance à la compression varient exponentiellement avec la pression du vide ; ils augmentent
avec la pression du vide et atteignent ensuite un plateau à -0,04MPa. On peut en conclure que cette
valeur représente la pression de vide optimale. L'analyse de corrélation de Pearson a révélé une très
grande linéarité entre la résistance au cisaillement interlaminaire et la résistance à la compression
et différents paramètres définissant la microstructure du composite et entre le module de
compression et la densité. En revanche, les corrélations entre le module de compression et la
fraction volumique des fibres et entre le module de compression et la teneur en vides sont moins
significatives. Sur la base des résultats de l'analyse statistique : (a) les principaux paramètres
affectant la résistance au cisaillement interlaminaire sont la densité et la teneur en vides, (b) la
résistance à la compression dépend de la densité, la fraction volumique des fibres, et la teneur en
vides, et (c) les paramètres les plus importants affectant le module de compression sont la densité
et la fraction volumique des fibres.

Après la détermination de l’effet de chaque paramètre sur les propriétés physiques et mécaniques
des composites stratifiés, nous avons exploités les résultats obtenues pour déterminer les niveaux
de chaque paramètre offrant des résultats optimaux. Des plans de surface de réponse (Box-
Behnken) ont été utilisés pour modéliser la courbure de chaque paramètre et de déterminer les
modèles empiriques des propriétés physiques des composites stratifiés concernés par notre étude.
D'après les résultats de cette étude, le degré de polymérisation dépend fortement de la température.
En même temps, pour la densité, l'effet linéaire de la pression était le terme le plus significatif, suivi
par l'interaction entre la température et la pression, les termes quadratiques du vide et de la pression,
et les termes linéaires de la température et du vide. Pour la fraction volumique des fibres, tous les
termes linéaires, les termes quadratiques de la pression et du vide, et l'interaction entre la
température et la pression étaient significatifs. Cependant, pour la teneur en vide, l'effet linéaire de
la pression était le terme le plus significatif, suivi de tous les termes quadratiques et du terme
linéaire du vide. L'analyse de la variance montre une grande importance des modèles établis, avec
des valeurs élevées des coefficients de détermination, assurant ainsi un ajustement satisfaisant du
modèle quadratique avec les données expérimentales. Ces résultats donnent à Safran Nacelle une

153
base de données importante permettant de cibler la propriété demandée en fonction du besoin, et
de fixé les paramètres en terme de température, de pression et de vide afin d’obtenir sa valeur
maximale.

Dans la dernière partie, il était important d’évaluer l’effet de la température de fonctionnement et


le temps de maintien sur les propriétés mécaniques et thermomécaniques, tout en respectant la
température de transition vitreuse comme limite à ne pas dépasser. Les principales conclusions sont
les suivantes : (a) Les principaux facteurs (température de fonctionnement et temps de maintien) et
leur interaction affectent la résistance à la compression des composites stratifiés, comme en
témoignent l’analyse de la variance et le diagramme de Pareto standardisé. (b) Les principaux
facteurs affectent le module de compression, alors que l'effet de l’interaction n'était pas significatif.
(c)La résistance au cisaillement interlaminaire est fonction de la température de fonctionnement
uniquement. (d)Toutes les propriétés mécaniques diminuent progressivement avec la température
de fonctionnement. (e)Les propriétés mécaniques en fonction de la température de fonctionnement
s'accordent bien avec le modèle Ha et Springer. (f)Enfin, pour évaluer la dégradation de la
résistance, il est recommandé d'utiliser les propriétés de compression.

Comme programmé lors de la mise en place de ce présent projet, ces travaux de recherche peuvent
être valoriser pour l’amélioration du procédé de fabrication par la mise en place du monitoring in
situ. Un projet qui revient à Safran et l’Académie Hassan II de le piloté.

154
LISTE DES RÉFÉRENCES BIBLIOGRAPHIQUES

[1] Y. Xu, J. Zhu, Z. Wu, Y. Cao, Y. Zhao, W. Zhang, A review on the design of laminated
composite structures: constant and variable stiffness design and topology optimization,
Adv. Compos. Hybrid Mater. 1 (2018) 460–477. https://doi.org/10.1007/s42114-018-0032-
7.

[2] B. Chatterjee, S. Bhowmik, Evolution of material selection in commercial aviation


industry—a review, in: Sustain. Eng. Prod. Manuf. Technol., Elsevier, 2019: pp. 199–219.
https://doi.org/10.1016/B978-0-12-816564-5.00009-8.

[3] C. Soutis, Introduction, in: Polym. Compos. Aerosp. Ind., Elsevier, 2015: pp. 1–18.
https://doi.org/10.1016/B978-0-85709-523-7.00001-3.

[4] X. Zhang, Y. Chen, J. Hu, Recent advances in the development of aerospace materials,
Prog. Aerosp. Sci. 97 (2018) 22–34. https://doi.org/10.1016/j.paerosci.2018.01.001.

[5] A. Baraskar, Boeing 787-9 Dreamliner Basic plane flight performances Report, Moscow,
2017. https://doi.org/10.13140/RG.2.2.15748.12161.

[6] K. Tanaka, T. Takei, T. Katayama, Effect of High Temperature Environment on the


Tensile Strength of Carbon Fiber/Highly Heat Resistant Polyamide Resin, Key Eng. Mater.
774 (2018) 337–342. https://doi.org/10.4028/www.scientific.net/KEM.774.337.

[7] T. Dursun, C. Soutis, Recent developments in advanced aircraft aluminium alloys, Mater.
Des. 56 (2014) 862–871. https://doi.org/10.1016/j.matdes.2013.12.002.

[8] D. Bettebghor, S. Grihon, J. Morlier, Bilevel optimization of large composite structures


based on lamination parameters and post-optimal sensitivities. Part 1: Theoretical aspects,
1 (2018) 1–17.

[9] F.C. Campbell, Chapter 10 - Thermoplastic Composites: An Unfulfilled Promise, in:


Manuf. Process. Adv. Compos., Elsevier, Elsevier Science, 2004: pp. 357–397.

156
https://doi.org/https://doi.org/10.1016/B978-185617415-2/50011-3.

[10] J.N. Reddy, J.N. Reddy, Mechanics of Laminated Composite Plates and Shells Theory and
Analysis, (2003) 840. https://doi.org/10.1007/978-1-4471-0095-9.

[11] P. S, S. KM, N. K, S. S, Fiber Reinforced Composites - A Review, J. Mater. Sci. Eng. 06


(2017). https://doi.org/10.4172/2169-0022.1000341.

[12] T.P. Sathishkumar, S. Satheeshkumar, J. Naveen, Glass fiber-reinforced polymer


composites - A review, J. Reinf. Plast. Compos. 33 (2014) 1258–1275.
https://doi.org/10.1177/0731684414530790.

[13] S. Chand, Carbon fibers for composites, J. Mater. Sci. 35 (2000) 1303–1313.
https://doi.org/10.1023/A:1004780301489.

[14] T. Ohsawa, M. Miwa, M. Kawade, E. Tsushima, Axial compressive strength of carbon


fiber, J. Appl. Polym. Sci. 39 (1990) 1733–1743.
https://doi.org/10.1002/app.1990.070390811.

[15] A.H. Shinohara, T. Sato, F. Saito, T. Tomioka, Y. Arai, A novel method for measuring
direct compressive properties of carbon fibres using a micro-mechanical compression
tester, J. Mater. Sci. 28 (1993) 6611–6616. https://doi.org/10.1007/BF00356404.

[16] J.D. Muzzy, A.O. Kays, Thermoplastic vs. thermosetting structural composites, Polym.
Compos. 5 (1984) 169–172. https://doi.org/10.1002/pc.750050302.

[17] B. Bilyeu, W. Brostow, K.P. Menard, Epoxy Thermosets and Their Applications . Ii .
Thermal Analysis, 22 (2000) 107–129.

[18] P. Bernay, Les preimpregnes dans le nautisme, (1992) 7–9.

[19] K.V.N. Gopal, Product design for advanced composite materials in aerospace engineering,
in: Adv. Compos. Mater. Aerosp. Eng., Elsevier, 2016: pp. 413–428.
https://doi.org/10.1016/B978-0-08-100037-3.00014-6.

[20] P. Hubert, A. Johnston, A. Poursartip, K. Nelson, Cure Kinetics and Viscosity Models of
Hexcel 8552 epoxy resin, Int. SAMPE Symp. Exhib. (2001) 2341–2354.

157
http://scholar.google.com/scholar?hl=en&btnG=Search&q=intitle:Cure+Kinetics+and+Vis
cosity+Models+for+Hexcel+8552+Epoxy+Resin#0.

[21] C. Soutis, Carbon fiber reinforced plastics in aircraft construction, Mater. Sci. Eng. A. 412
(2005) 171–176. https://doi.org/10.1007/BF00605914.

[22] S. Hernández, F. Sket, J.M. Molina-Aldareguí a, C. González, J. Llorca, Effect of curing


cycle on void distribution and interlaminar shear strength in polymer-matrix composites,
Compos. Sci. Technol. 71 (2011) 1331–1341.
https://doi.org/10.1016/j.compscitech.2011.05.002.

[23] H. Lengsfeld, W.-F. Felipe, K. Johannes, L. Javier, A. Volker, Prepregs and Monolithic
Part Fabrication, 1st ed., Hanser, 2016.

[24] A. Dong, Y. Zhao, X. Zhao, Q. Yu, Cure cycle optimization of rapidly cured out-of-
autoclave composites, Materials (Basel). 11 (2018). https://doi.org/10.3390/ma11030421.

[25] T. Vo, K. Vora, B. Minaie, Effects of postcure temperature variation on hygrothermal-


mechanical properties of an out-of-autoclave polymer composite, J. Appl. Polym. Sci. 130
(2013) 3090–3097. https://doi.org/10.1002/app.39541.

[26] H. Koushyar, S. Alavi-Soltani, B. Minaie, M. Violette, Effects of variation in autoclave


pressure, temperature, and vacuum-application time on porosity and mechanical properties
of a carbon fiber/epoxy composite, J. Compos. Mater. 46 (2012) 1985–2004.
https://doi.org/10.1177/0021998311429618.

[27] Z. Bergant, A. Savin, J. Grum, Effects of manufacturing technology on static, multi-


frequency dynamic mechanical analysis and fracture energy of cross-ply and quasi-
isotropic carbon/epoxy laminates, Polym. Polym. Compos. 26 (2018) 358–370.
https://doi.org/10.1177/0967391118798266.

[28] S. Alavi-Soltani, S. Sabzevari, H. Koushyar, B. Minaie, Thermal, rheological, and


mechanical properties of a polymer composite cured at different isothermal cure
temperatures, J. Compos. Mater. 46 (2012) 575–587.
https://doi.org/10.1177/0021998311415443.

158
[29] E. Schab-Balcerzak, H. Janeczek, B. Kaczmarczyk, H. Bednarski, D. Sek, A. Miniewicz,
Epoxy resin cured with diamine bearing azobenzene group, Polymer (Guildf). 45 (2004)
2483–2493. https://doi.org/10.1016/j.polymer.2004.02.027.

[30] G. Nikolic, S. Zlatkovic, M. Cakic, S. Cakic, C. Lacnjevac, Z. Rajic, Fast Fourier


Transform IR Characterization of Epoxy GY Systems Crosslinked with Aliphatic and
Cycloaliphatic EH Polyamine Adducts, Sensors. 10 (2010) 684–696.
https://doi.org/10.3390/s100100684.

[31] C. Paris, Étude et modélisation de la polymérisation dynamique de composites à matrice


thermodurcissable, Institut National Polytechnique de Toulouse (INP Toulouse),
Université de Toulouse, 2011.

[32] M. González-González, J.C. Cabanelas, J. Baselga, Applications of FTIR on Epoxy Resins


- Identification, Monitoring the Curing Process, Phase Separation and Water Uptake,
Infrared Spectrosc. - Mater. Sci. Eng. Techology. 2 (2012) 261–284.
https://doi.org/10.5772/2055.

[33] S. Swier, G. Van Assche, W. Vuchelen, B. Van Mele, Role of complex formation in the
polymerization kinetics of modified epoxy-amine systems, Macromolecules. 38 (2005)
2281–2288. https://doi.org/10.1021/ma047796x.

[34] M. Javdanitehran, D.C. Berg, E. Duemichen, G. Ziegmann, An iterative approach for


isothermal curing kinetics modelling of an epoxy resin system, Thermochim. Acta. 623
(2016) 72–79. https://doi.org/10.1016/j.tca.2015.11.014.

[35] K.C. Cole, J.J. Hechler, D. Noel, A new approach to modeling the cure kinetics of
epoxy/amine thermosetting resins., Macromolecules. 24(11) (1991) 3098–3110.
http://dx.doi.org/10.1021/ma00011a012.

[36] F. Shiravand, J.M. Hutchinson, Y. Calventus, Non-isothermal cure and exfoliation of tri-
functional epoxy-clay nanocomposites, Express Polym. Lett. 9 (2015) 695–708.
https://doi.org/10.3144/expresspolymlett.2015.65.

[37] S. Corezzi, D. Fioretto, G. Santucci, J.M. Kenny, Modeling diffusion-control in the cure

159
kinetics of epoxy-amine thermoset resins : An approach based on con fi gurational entropy,
Polymer (Guildf). 51 (2010) 5833–5845. https://doi.org/10.1016/j.polymer.2010.09.073.

[38] I. Mereu, L. Comez, I. National, S. Corezzi, On the interplay between the slowdown of
dynamics and the kinetics of aggregation : The case study of a reactive binary mixture,
(2015). https://doi.org/10.1063/1.4918743.

[39] A.K. Singh, B.P. Panda, S. Mohanty, S.K. Nayak, M.K. Gupta, Thermokinetics behavior
of epoxy adhesive reinforced with low viscous aliphatic reactive diluent and nano-fillers,
Korean J. Chem. Eng. 34 (2017) 3028–3040. https://doi.org/10.1007/s11814-017-0221-z.

[40] R. Hardis, J.L.P. Jessop, F.E. Peters, M.R. Kessler, Cure kinetics characterization and
monitoring of an epoxy resin using DSC, Raman spectroscopy, and DEA, Compos. Part A
Appl. Sci. Manuf. 49 (2013) 100–108. https://doi.org/10.1016/j.compositesa.2013.01.021.

[41] D. Puglia, L. Valentini, J.M. Kenny, Analysis of the cure reaction of carbon
nanotubes/epoxy resin composites through thermal analysis and Raman spectroscopy, J.
Appl. Polym. Sci. 88 (2003) 452–458. https://doi.org/10.1002/app.11745.

[42] L.F. Yang, K.D. Yao, W. Koh, Kinetics Analysis of the Curing Reaction of Fast Cure
Epoxy Prepregs, J. Appl. Polym. Sci. (1999) 1501–1508.

[43] L. Sun, S.S. Pang, A.M. Sterling, I.I. Negulescu, M.A. Stubblefield, Thermal analysis of
curing process of epoxy prepreg, J. Appl. Polym. Sci. 83 (2002) 1074–1083.
https://doi.org/10.1002/app.10053.

[44] E. Ernault, E. Richaud, B. Fayolle, Thermal oxidation of epoxies: Influence of diamine


hardener, Polym. Degrad. Stab. 134 (2016) 76–86.
https://doi.org/10.1016/j.polymdegradstab.2016.09.030.

[45] J. Parameswaranpillai, A. George, J. Pionteck, S. Thomas, Investigation of Cure Reaction,


Rheology, Volume Shrinkage and Thermomechanical Properties of Nano-TiO 2 Filled
Epoxy/DDS Composites, J. Polym. 2013 (2013) 1–17.
https://doi.org/10.1155/2013/183463.

[46] P. Hubert, G. Fernlund, A. Poursartip, Autoclave processing for composites, Woodhead

160
Publishing Limited, 2012. https://doi.org/10.1016/B978-0-85709-067-6.50013-4.

[47] J.L. Kardos, M.P. Duduković, R. Dave, Void growth and resin transport during processing
of thermosetting — Matrix composites, in: Adv. Polym. Sci., 1986: pp. 101–123.
https://doi.org/10.1007/3-540-16423-5_13.

[48] F.Y.C. Boey, S.W. Lye, Void reduction in autoclave processing of thermoset composites.
Part 2: Void reduction in a microwave curing process, Composites. 23 (1992) 266–270.
https://doi.org/10.1016/0010-4361(92)90187-Y.

[49] F.Y.C. Boey, S.W. Lye, Effects of vacuum and pressure in an autoclave curing process for
a thermosetting fibre-reinforced composite, J. Mater. Process. Tech. 23 (1990) 121–131.
https://doi.org/10.1016/0924-0136(90)90152-K.

[50] Sudarisman, I.J. Davies, Influence of compressive pressure, vacuum pressure, and holding
temperature applied during autoclave curing on the microstructure of unidirectional CFRP
composites, Adv. Mater. Res. 41–42 (2008) 323–328.
https://doi.org/10.4028/www.scientific.net/AMR.41-42.323.

[51] L. Liu, B.-M. Zhang, D.-F. Wang, Z.-J. Wu, Effects of cure cycles on void content and
mechanical properties of composite laminates, Compos. Struct. 73 (2006) 303–309.
https://doi.org/10.1016/j.compstruct.2005.02.001.

[52] M.L. Costa, M.C. Rezende, Effect of Void Content on the Moisture Absorption in
Polymeric Composites, (2006). https://doi.org/10.1080/03602550600609549.

[53] H. Zhu, D. Li, D. Zhang, B. Wu, Y. Chen, Influence of voids on interlaminar shear
strength of carbon/epoxy fabric laminates, Trans. Nonferrous Met. Soc. China. 19 (2009)
s470–s475. https://doi.org/10.1016/S1003-6326(10)60091-X.

[54] T. Chang, L. Zhan, W. Tan, S. Li, Void content and interfacial properties of composite
laminates under different autoclave cure pressure, Compos. Interfaces. 24 (2017) 529–540.
https://doi.org/10.1080/09276440.2016.1237113.

[55] M.L. Costa, S.F.M. d. Almeida, M.C. Rezende, The influence of porosity on the
interlaminar shear strength of carbon/epoxy and carbon/bismaleimide fabric laminates,

161
Compos. Sci. Technol. 61 (2001) 2101–2108. https://doi.org/10.1016/S0266-
3538(01)00157-9.

[56] X. Liu, F. Chen, A review of void formation and its effects on the mechanical performance
of carbon fiber reinforced plastic, Eng. Trans. 64 (2016) 33–51.

[57] S. Hernández, F. Sket, C. González, J. Llorca, Optimization of curing cycle in carbon


fiber-reinforced laminates: Void distribution and mechanical properties, Compos. Sci.
Technol. 85 (2013) 73–82. https://doi.org/10.1016/j.compscitech.2013.06.005.

[58] L. Di Landro, A. Montalto, P. Bettini, S. Guerra, F. Montagnoli, M. Rigamonti, Detection


of Voids in Carbon/Epoxy Laminates and Their Influence on Mechanical Properties,
Polym. Polym. Compos. 25 (2017) 371–380.
https://doi.org/10.1177/096739111702500506.

[59] P. Olivier, J.P. Cottu, B. Ferret, Effects of cure cycle pressure and voids on some
mechanical properties of carbon/epoxy laminates, Composites. 26 (1995) 509–515.
https://doi.org/10.1016/0010-4361(95)96808-J.

[60] A.M. Kudrin, K.S. Gabriel’s, O.A. Karaeva, The Temperature Effect on Mechanical
Properties of Carbon Fiber Reinforced Plastics for Aviation Purposes, Inorg. Mater. Appl.
Res. 9 (2018) 763–766. https://doi.org/10.1134/S2075113318040196.

[61] M.M.L. KOANDA, Optimisation des propriétés physiques d’un composite carbone époxy
fabriqué par le procédé RFI, UNIVERSITÉ DE MONTRÉAL ÉCOLE
POLYTECHNIQUE DE MONTRÉAL Ce, 2012.
http://www.dt.co.kr/contents.html?article_no=2012071302010531749001.

[62] H. Huang, R. Talreja, Effects of void geometry on elastic properties of unidirectional fiber
reinforced composites, Compos. Sci. Technol. 65 (2005) 1964–1981.
https://doi.org/10.1016/j.compscitech.2005.02.019.

[63] M. Mehdikhani, L. Gorbatikh, I. Verpoest, S. V. Lomov, Voids in fiber-reinforced


polymer composites: A review on their formation, characteristics, and effects on
mechanical performance, J. Compos. Mater. 53 (2019) 1579–1669.

162
https://doi.org/10.1177/0021998318772152.

[64] Y. Ledru, G. Bernhart, R. Piquet, F. Schmidt, L. Michel, Coupled visco-mechanical and


diffusion void growth modelling during composite curing, Compos. Sci. Technol. 70
(2010) 2139–2145. https://doi.org/10.1016/j.compscitech.2010.08.013.

[65] L.K. Grunenfelder, S.R. Nutt, Void formation in composite prepregs - Effect of dissolved
moisture, Compos. Sci. Technol. 70 (2010) 2304–2309.
https://doi.org/10.1016/j.compscitech.2010.09.009.

[66] J.P. Anderson, M.C. Altan, Formation of voids in composite laminates: Coupled effect of
moisture content and processing pressure, Polym. Compos. 36 (2015) 376–384.
https://doi.org/10.1002/pc.22952.

[67] Z.S. Guo, L. Liu, B.M. Zhang, S. Du, Critical void content for thermoset composite
laminates, J. Compos. Mater. 43 (2009) 1775–1790.
https://doi.org/10.1177/0021998306065289.

[68] Y. Li, Q. Li, H. Ma, The voids formation mechanisms and their effects on the mechanical
properties of flax fiber reinforced epoxy composites, Compos. Part A Appl. Sci. Manuf. 72
(2015) 40–48. https://doi.org/10.1016/j.compositesa.2015.01.029.

[69] P. Yang, R. El-Hajjar, Porosity Defect Morphology Effects in Carbon Fiber - Epoxy
Composites, Polym. - Plast. Technol. Eng. 51 (2012) 1141–1148.
https://doi.org/10.1080/03602559.2012.689050.

[70] Y. Nikishkov, G. Seon, A. Makeev, Structural analysis of composites with porosity defects
based on X-ray computed tomography, J. Compos. Mater. 48 (2014) 2131–2144.
https://doi.org/10.1177/0021998313494917.

[71] S. Hernández, F. Sket, J.M. Molina-Aldareguı´a, C. González, J. LLorca, Effect of curing


cycle on void distribution and interlaminar shear strength in polymer-matrix composites,
Compos. Sci. Technol. 71 (2011) 1331–1341.
https://doi.org/10.1016/j.compscitech.2011.05.002.

[72] H. Jeong, Effects of voids on the mechanical strength and ultrasonic attenuation of

163
laminated composites, J. Compos. Mater. 31 (1997) 276–292.
https://doi.org/10.1177/002199839703100303.

[73] D.E.W. Stone, B. Clarke, Ultrasonic attenuation as a measure of void content in carbon-
fibre reinforced plastics, Non-Destructive Test. 8 (1975) 137–145.
https://doi.org/10.1016/0029-1021(75)90023-7.

[74] L.W. Davies, R.J. Day, D. Bond, A. Nesbitt, J. Ellis, E. Gardon, Effect of cure cycle heat
transfer rates on the physical and mechanical properties of an epoxy matrix composite,
Compos. Sci. Technol. 67 (2007) 1892–1899.
https://doi.org/10.1016/j.compscitech.2006.10.014.

[75] P.A. Olivier, B.Î. Mascaro, P. Margueres, F. Collombet, CFRP with voids: Ultrasonic
characterization of localized porosity, acceptance criteria and mechanical characteristics,
ICCM Int. Conf. Compos. Mater. (2007).

[76] A.G. Stamopoulos, K.I. Tserpes, P. Prucha, D. Vavrik, Evaluation of porosity effects on
the mechanical properties of carbon fiber-reinforced plastic unidirectional laminates by X-
ray computed tomography and mechanical testing, J. Compos. Mater. 50 (2016) 2087–
2098. https://doi.org/10.1177/0021998315602049.

[77] K.M. Uhl, B. Lucht, H. Jeong, D.K. Hsu, Mechanical strength degradation of graphite fiber
reinforced thermoset composites due to porosity, Rev. Prog. Quant. Nondestruct. Eval. 7 B
(1988) 1075–1082. https://doi.org/10.1007/978-1-4613-0979-6_24.

[78] J.L. Kardos, M.P. Duduković, R. Dave, Void growth and resin transport during processing
of thermosetting — Matrix composites, in: K. Dušek (Ed.), Adv. Polym. Sci., Springer
Berlin Heidelberg, Berlin, Heidelberg, 1986: pp. 101–123. https://doi.org/10.1007/3-540-
16423-5_13.

[79] S.S. Tavares, V. Michaud, J.A.E. Månson, Through thickness air permeability of prepregs
during cure, Compos. Part A Appl. Sci. Manuf. 40 (2009) 1587–1596.
https://doi.org/10.1016/j.compositesa.2009.07.004.

[80] A. Baghad, K. El Mabrouk, S. Vaudreuil, K. Nouneh, Cure kinetics and autoclave-pressure

164
dependence on physical and mechanical properties of woven carbon/epoxy 8552S/AS4
composite laminates, Polym. Polym. Compos. 29 (2021) S903–S913.
https://doi.org/10.1177/09673911211028413.

[81] L. Zhu, Z. Wang, M. Bin Rahman, W. Shen, C. Zhu, The Curing Kinetics of E-Glass Fiber
/ Epoxy Resin Prepreg and the Bending Properties of Its Products, (2021).

[82] P. Kashani, B. Minaie, An ex-situ state-based approach using rheologicla properties to


measure and model cure in polymer composites, J. Reinf. Plast. Compos. 30 (2011) 123–
133. https://doi.org/10.1177/0731684410388441.

[83] S. Lee, G.S. Springer, Effects of Cure on the Mechanical Properties of Composites, J.
Compos. Mater. 22 (1988) 15–29. https://doi.org/10.1177/002199838802200102.

[84] A. Baghad, K. El Mabrouk, S. Vaudreuil, K. Nouneh, Effects of high operating


temperatures and holding times on thermomechanical and mechanical properties of
autoclaved epoxy/carbon composite laminates, Polym. Compos. 43 (2022) 862–873.
https://doi.org/10.1002/pc.26416.

[85] S. Sourour, M.R. Kamal, Differential scanning calorimetry of epoxy cure: isothermal cure
kinetics, Thermochim. Acta. 14 (1976) 41–59. https://doi.org/10.1016/0040-
6031(76)80056-1.

[86] L. Serrano, P. Olivier, J. Cinquin, L. Serrano, P. Olivier, J. Cinquin, Modélisation des


cinétiques de polymérisation de résines destinées à la fabrication Hors Autoclave, in: 2017.

[87] ASTM D792-13, Standard Test Methods for Density and Specific Gravity (Relative
Density) of Plastics by Displacement, 2013.

[88] ASTM D3171–15, Standard Test Methods for Constituent Content of Composite
Materials, 2015.

[89] B. Plank, C. Gusenbauer, S. Senck, H. Hoeller, J. Kastner, Porosity Determination in


CFRP by means of X-ray Computed Tomography Methods, in: 2nd Int. Symp. NDT
Aerospace, We.1.a.2, Hambg., 2010: pp. 1–2.

165
[90] M. Sánchez-Soto, T. P. Pagés, K. Lacorte, Briceño, F. Carrasco, Curing FTIR study and
mechanical characterization of glass bead filled trifunctional epoxy composites, Compos.
Sci. Technol. 67 (2007) 1974–1985. https://doi.org/10.1016/j.compscitech.2006.10.006.

[91] S.K. Gupta, M. Hojjati, Thermal cycle effects on laminated composite plates containing
voids, J. Compos. Mater. 53 (2019) 489–501. https://doi.org/10.1177/0021998318786785.

[92] T. Centea, G. Peters, K. Hendrie, S. Nutt, Effects of thermal gradients on defect formation
during the consolidation of partially impregnated prepregs, J. Compos. Mater. 51 (2017)
3987–4003. https://doi.org/10.1177/0021998317733317.

[93] Hexcel, HexPly 8552 Data Sheet, (2016) 1–6.

[94] Y. Ng, J. Tomblin, Fabrication of NMS 128 Qualification, Equivalency, and Acceptance
Test Panels (for Hexcel 8552 and 8552S prepregs), 2011.

[95] ASTM D6641 / D6641M-16e2, Standard Test Method for Determining the Compressive
Properties of Polymer Matrix Composite Laminates Using a Combined Loading
Compression (CLC) Test Fixture, 2016.

[96] NF EN 2563, Série aérospatiale Plastiques Renforcés de fibres de carbone - Stratifiés


unidirectionnels - Détermination de la résistance apparente au cisaillement interlaminaire,
1997. (n.d.).

[97] T. Garstka, N. Ersoy, K.D. Potter, M.R. Wisnom, In situ measurements of through-the-
thickness strains during processing of AS4/8552 composite, Compos. Part A Appl. Sci.
Manuf. 38 (2007) 2517–2526. https://doi.org/10.1016/j.compositesa.2007.07.018.

[98] G.Z. Xiao, M. Delamar, M.E.R. Shanahan, Irreversible interactions between water and
DGEBA/DDA epoxy resin during hygrothermal aging, J. Appl. Polym. Sci. 65 (2002)
449–458. https://doi.org/10.1002/(sici)1097-4628(19970718)65:3<449::aid-app4>3.3.co;2-
5.

[99] D. Saenz-Castillo, M.I. Martín, S. Calvo, F. Rodriguez-Lence, A. Güemes, Effect of


processing parameters and void content on mechanical properties and NDI of
thermoplastic composites, Compos. Part A Appl. Sci. Manuf. 121 (2019) 308–320.

166
https://doi.org/10.1016/j.compositesa.2019.03.035.

[100] S.J. Li, L.H. Zhan, R. Chen, W.F. Peng, Y.A. Zhang, Y.Q. Zhou, L.R. Zeng, The Influence
of Cure Pressure on Microstructure , Temperature Field and Mechanical Properties of
Advanced Polymer-matrix Composite Laminates, 15 (2014) 2404–2409.
https://doi.org/10.1007/s12221-014-2404-0.

[101] A.C.M.Q.S. Santos, F.M. Monticeli, H. Ornaghi, L.F. de P. Santos, M.O.H. Cioffi,
Porosity characterization and respective influence on short-beam strength of advanced
composite processed by resin transfer molding and compression molding, Polym. Polym.
Compos. (2020) 1–10. https://doi.org/10.1177/0967391120968452.

[102] R.A. Hawileh, A. Abu-obeidah, J.A. Abdalla, A. Al-tamimi, Temperature effect on the
mechanical properties of carbon , glass and carbon – glass FRP laminates, Constr. Build.
Mater. 75 (2015) 342–348. https://doi.org/10.1016/j.conbuildmat.2014.11.020.

[103] G. Aklilu, S. Adali, G. Bright, Temperature effect on mechanical properties of carbon,


glass and hybrid polymer composite specimens, Int. J. Eng. Res. Africa. 39 (2018) 119–
138. https://doi.org/10.4028/www.scientific.net/JERA.39.119.

[104] H.W. He, F. Gao, Effect of Fiber Volume Fraction on the Flexural Properties of
Unidirectional Carbon Fiber/Epoxy Composites, Int. J. Polym. Anal. Charact. 20 (2015)
180–189. https://doi.org/10.1080/1023666X.2015.989076.

[105] Z. Aying, L. Haibao, Z. Dongxing, Research on the Mechanical Properties Prediction of


Carbon/Epoxy Composite Laminates With Different Void Contents, Polym. Compos.
(2016) 1230–1241. https://doi.org/10.1002/pc.

[106] C. Dong, Effects of Process-Induced Voids on the Properties of Fibre Reinforced


Composites, J. Mater. Sci. Technol. 32 (2016) 597–604.
https://doi.org/10.1016/j.jmst.2016.04.011.

[107] M.R.M. Rejab, C.W. Theng, M.M. Rahman, M.M. Noor, A.N.M. Rose, An Investigation
into the Effects of Fibre Volume Fraction on GFRP Plate Abstract — This paper presents
the mechanical properties of Glass Fibre Reinforce Plastic, in: 2008: pp. 136–142.

167
[108] T. Chang, L. Zhan, W. Tan, S. Li, Effect of autoclave pressure on interfacial properties at
micro- and macro- level in polymer-matrix composite laminates, Fibers Polym. 18 (2017)
1614–1622. https://doi.org/10.1007/s12221-017-7384-4.

[109] D.C. da S. Monte Vidal, H.L. Ornaghi, F.G. Ornaghi, F.M. Monticeli, H.J.C. Voorwald,
M.O.H. Cioffi, Effect of different stacking sequences on hybrid carbon/glass/epoxy
composites laminate: Thermal, dynamic mechanical and long-term behavior, J. Compos.
Mater. 54 (2020) 731–743. https://doi.org/10.1177/0021998319868512.

[110] L.G. Stringer, Optimization of the wet lay-up/vacuum bag process for the fabrication of
carbon fibre epoxy composites with high fibre fraction and low void content, Composites.
20 (1989) 441–452. https://doi.org/10.1016/0010-4361(89)90213-9.

[111] H. Zhu, B. Wu, D. Li, D. Zhang, Y. Chen, Influence of Voids on the Tensile Performance
of Carbon/epoxy Fabric Laminates, J. Mater. Sci. Technol. 27 (2011) 69–73.
https://doi.org/10.1016/S1005-0302(11)60028-5.

[112] Y. Gu, M. Li, Z. Zhang, Z. Sun, Void formation model and measuring method of void
formation condition during hot pressing process, Polym. Compos. 31 (2010) 1562–1571.
https://doi.org/10.1002/pc.20944.

[113] J. Kakakasery, V. Arumugam, K. Abdul Rauf, D. Bull, A.R. Chambers, C. Scarponi, C.


Santulli, Cure cycle effect on impact resistance under elevated temperatures in carbon
prepreg laminates investigated using acoustic emission, Compos. Part B Eng. 75 (2015)
298–306. https://doi.org/10.1016/j.compositesb.2015.02.002.

[114] S. Daneshpayeh, F. Ashenai Ghasemi, I. Ghasemi, M. Ayaz, Predicting of mechanical


properties of PP/LLDPE/TiO2 nano-composites by response surface methodology,
Compos. Part B Eng. 84 (2016) 109–120.
https://doi.org/10.1016/j.compositesb.2015.08.075.

[115] H. Chaker, N. Ameur, K. Saidi-Bendahou, M. Djennas, S. Fourmentin, Modeling and Box-


Behnken design optimization of photocatalytic parameters for efficient removal of dye by
lanthanum-doped mesoporous TiO2, J. Environ. Chem. Eng. 9 (2021) 104584.

168
https://doi.org/10.1016/j.jece.2020.104584.

[116] H. Kantrong, C. Charunuch, N. Limsangouan, W. Pengpinit, Influence of process


parameters on physical properties and specific mechanical energy of healthy mushroom-
rice snacks and optimization of extrusion process parameters using response surface
methodology, J. Food Sci. Technol. 55 (2018) 3462–3472. https://doi.org/10.1007/s13197-
018-3271-2.

[117] ASTM D 792-20, Standard Test Methods for Density and Specific Gravity ( Relative
Density ) of Plastics, 2020.

[118] S. Yazici Guvenc, G. Varank, A. Cebi, B. Ozkaya, Electro-activated Persulfate Oxidation


of Biodiesel Wastewater Following Acidification Phase: Optimization of Process
Parameters Using Box–Behnken Design, Water. Air. Soil Pollut. 232 (2021).
https://doi.org/10.1007/s11270-020-04962-8.

[119] D.C. Montgomery, Montgomery Design and Analysis of Experiments Eighth Edition.
Arizona State University, 2013.

[120] A. Aleboyeh, N. Daneshvar, M.B. Kasiri, Optimization of C.I. Acid Red 14 azo dye
removal by electrocoagulation batch process with response surface methodology, Chem.
Eng. Process. Process Intensif. 47 (2008) 827–832.
https://doi.org/10.1016/j.cep.2007.01.033.

[121] A.R. Khataee, M. Fathinia, S. Aber, M. Zarei, Optimization of photocatalytic treatment of


dye solution on supported TiO2 nanoparticles by central composite design: Intermediates
identification, J. Hazard. Mater. 181 (2010) 886–897.
https://doi.org/10.1016/j.jhazmat.2010.05.096.

[122] A. Aziz, A. Driouich, A. Bellil, M. Ben Ali, S.E.L. Mabtouti, K. Felaous, M. Achab, A. El
Bouari, Optimization of new eco-material synthesis obtained by phosphoric acid attack of
natural Moroccan pozzolan using Box-Behnken Design, Ceram. Int. (2021).
https://doi.org/10.1016/j.ceramint.2021.08.203.

[123] C.M. Ewulonu, J.L. Chukwuneke, I.C. Nwuzor, C.H. Achebe, Fabrication of cellulose

169
nanofiber/polypyrrole/polyvinylpyrrolidone aerogels with box-Behnken design for optimal
electrical conductivity, Carbohydr. Polym. 235 (2020) 116028.
https://doi.org/10.1016/j.carbpol.2020.116028.

[124] M.Z. Hassan, S.A. Roslan, S.M. Sapuan, Z.A. Rasid, A.F. Mohd Nor, M.Y. Md Daud, R.
Dolah, M.Z. Mohamed Yusoff, Mercerization optimization of bamboo (Bambusa vulgaris)
fiber-reinforced epoxy composite structures using a Box-Behnken design, Polymers
(Basel). 12 (2020) 1–19. https://doi.org/10.3390/POLYM12061367.

[125] Y. Zhang, C. Li, D. Chu, G. Yan, M. Zhu, X. Zhao, J. Gu, G. Li, J. Wang, B. Zhang,
Process optimization for the preparation of thiamethoxam microspheres by response
surface methodology, React. Funct. Polym. 147 (2020) 104460.
https://doi.org/10.1016/j.reactfunctpolym.2019.104460.

[126] T.Q. Liu, X. Liu, P. Feng, A comprehensive review on mechanical properties of pultruded
FRP composites subjected to long-term environmental effects, Compos. Part B Eng. 191
(2020) 107958. https://doi.org/10.1016/j.compositesb.2020.107958.

[127] T.K. Tsotsis, Thermo-Oxidative Aging of Composite Materials, J. Compos. Mater. 29


(1995) 410–422. https://doi.org/10.1177/002199839502900307.

[128] T.K. Tsotsis, Long-Term Thermo-Oxidative Aging in Composite Materials: Experimental


Methods, J. Compos. Mater. 32 (1998) 1115–1135.
https://doi.org/10.1177/002199839803201104.

[129] T.K. Tsotsis, S.M. Lee, Long-term thermo-oxidative aging in composite materials: failure
mechanisms, Compos. Sci. Technol. 58 (1998) 355–368. https://doi.org/10.1016/S0266-
3538(97)00123-1.

[130] M. Aktas, R. Karakuzu, Determination of Mechanical Properties of Glass-Epoxy


Composites in High Temperatures, Polym. Compos. 30 (2009) 1437–1441.
https://doi.org/https://doi.org/10.1002/pc.20708.

[131] A. Salehi-Khojin, R. Bashirzadeh, M. Mahinfalah, R. Nakhaei-Jazar, Effect of temperature


on impact properties of hybrid fiberglass/kevlar laminate composites, in: Am. Soc. Mech.

170
Eng. Appl. Mech. Div. AMD, 2005: pp. 19–26. https://doi.org/10.1115/IMECE2005-
81530.

[132] E.C. Botelho, L.C. Pardini, M.C. Rezende, Hygrothermal effects on the shear properties of
carbon fiber/epoxy composites, J. Mater. Sci. 41 (2006) 7111–7118.
https://doi.org/10.1007/s10853-006-0933-7.

[133] S. Cao, X. Wang, Z. Wu, Evaluation and prediction of temperature-dependent tensile


strength of unidirectional carbon fiber-reinforced polymer composites, J. Reinf. Plast.
Compos. 30 (2011) 799–807. https://doi.org/10.1177/0731684411411002.

[134] Z. Jia, T. Li, F. pen Chiang, L. Wang, An experimental investigation of the temperature
effect on the mechanics of carbon fiber reinforced polymer composites, Compos. Sci.
Technol. 154 (2018) 53–63. https://doi.org/10.1016/j.compscitech.2017.11.015.

[135] G. Li, J. Zhao, Z. Wang, Fatigue behavior of glass fiber-reinforced polymer bars after
elevated temperatures exposure, Materials (Basel). 11 (2018) 1–16.
https://doi.org/10.3390/ma11061028.

[136] H. Ashrafi, M. Bazli, A. Jafari, T. Ozbakkaloglu, Tensile properties of GFRP laminates


after exposure to elevated temperatures: Effect of fiber configuration, sample thickness,
and time of exposure, Compos. Struct. 238 (2020) 111971.
https://doi.org/10.1016/j.compstruct.2020.111971.

[137] A.A.M. Badawy, Impact behavior of glass fibers reinforced composite laminates at
different temperatures, Ain Shams Eng. J. 3 (2012) 105–111.
https://doi.org/10.1016/j.asej.2012.01.001.

[138] Z. Liu, Z. Guan, F. Liu, J. Xu, Time-temperature dependent mechanical properties of cured
epoxy resin and unidirectional CFRP, 2017 8th Int. Conf. Mech. Aerosp. Eng. ICMAE
2017. (2017) 113–117. https://doi.org/10.1109/ICMAE.2017.8038626.

[139] J. Li, W. Fan, T. Liu, L. Yuan, L. Xue, W. Dang, J. Meng, The temperature effect on the
inter-laminar shear properties and failure mechanism of 3D orthogonal woven composites,
Text. Res. J. 90 (2020) 2806–2817. https://doi.org/10.1177/0040517520927009.

171
[140] H.L. Ornaghi, J.H.S. Almeida, F.M. Monticeli, R.M. Neves, M.O.H. Cioffi, Time-
temperature behavior of carbon/epoxy laminates under creep loading, Mech. Time-
Dependent Mater. (2020). https://doi.org/10.1007/s11043-020-09463-z.

[141] M. Katouzian, S. Vlase, Creep response of neat and carbon-fiber-reinforced peek and
epoxy determined using a micromechanical model, Symmetry (Basel). 12 (2020) 1–20.
https://doi.org/10.3390/sym12101680.

[142] W.K. Goertzen, M.R. Kessler, Creep behavior of carbon fiber/epoxy matrix composites,
Mater. Sci. Eng. A. 421 (2006) 217–225. https://doi.org/10.1016/j.msea.2006.01.063.

[143] A. Elmiladi, V. Petrovic, A. Grbovic, A. Sedmak, I. Balac, Numerical evaluation of


thermal stresses generated in laminated composite structure exposed to low temperatures,
Therm. Sci. 25 (2021) 3847–3856. https://doi.org/10.2298/TSCI200729338E.

[144] H. Cheng, J. Chang, Y. Sun, J. Zhang, X. Wang, Numerical simulation of stress


distribution for CF/EP composites in high temperatures, J. Therm. Stress. 42 (2019) 416–
425. https://doi.org/10.1080/01495739.2018.1469102.

[145] S. Liu, B. Shi, A. Siddique, Y. Du, B. Sun, B. Gu, Numerical analyses on thermal stress
distribution induced from impact compression in 3D carbon fiber/epoxy braided composite
materials, J. Therm. Stress. 41 (2018) 903–919.
https://doi.org/10.1080/01495739.2018.1437000.

[146] S. Benli, O. Sayman, The Effects of Temperature and Thermal Stresses on Impact Damage
in Laminated Composites, Math. Comput. Appl. 16 (2011) 392–403.
https://doi.org/10.3390/mca16020392.

[147] J. César dos Santos, L. Ávila de Oliveira, T.H. Panzera, C.D.L. Remillat, I. Farrow, V.
Placet, F. Scarpa, Ageing of autoclaved epoxy/flax composites: Effects on water
absorption, porosity and flexural behaviour, Compos. Part B Eng. 202 (2020).
https://doi.org/10.1016/j.compositesb.2020.108380.

[148] ASTM D7028-07(2015), Standard Test Method for Glass Transition Temperature (DMA
Tg) of Polymer Matrix Composites by Dynamic Mechanical Analysis (DMA), (2015).

172
http://www.astm.org/cgi-bin/resolver.cgi?D7028-07(2015).

[149] W. Hufenbach, M. Gude, R. Böhm, M. Zscheyge, The effect of temperature on mechanical


properties and failure behaviour of hybrid yarn textile-reinforced thermoplastics, Mater.
Des. 32 (2011) 4278–4288. https://doi.org/10.1016/J.MATDES.2011.04.017.

[150] D.C.A.S.U. Montgomery, Design and Analysis of Experiments Ninth Edition, 2017.
www.wiley.com/go/permissions.%0Ahttps://lccn.loc.gov/2017002355.

[151] M. Jawaid, A.K. H.P.S, O. Alattas, Woven hybrid biocomposites: Dynamic mechanical
and thermal properties, Compos. Part A Appl. Sci. Manuf. 43 (2012) 288–293.
https://doi.org/10.1016/j.compositesa.2011.11.001.

[152] J. Feng, Z. Guo, Temperature-frequency-dependent mechanical properties model of epoxy


resin and its composites, Compos. Part B Eng. 85 (2016) 161–169.
https://doi.org/10.1016/j.compositesb.2015.09.040.

[153] S. Cao, Z. Wu, F. Li, Effects of temperature on tensile strength of carbon fiber and
carbon/epoxy composite sheets, Adv. Mater. Res. 476–478 (2012) 778–784.
https://doi.org/10.4028/www.scientific.net/AMR.476-478.778.

[154] M. Asim, M. Jawaid, M. Nasir, N. Saba, Effect of fiber loadings and treatment on dynamic
mechanical, thermal and flammability properties of pineapple leaf fiber and kenaf phenolic
composites, J. Renew. Mater. 6 (2018) 383–393.
https://doi.org/10.7569/JRM.2017.634162.

[155] N. Jesuarockiam, M. Jawaid, E.S. Zainudin, M.T. Hameed Sultan, R. Yahaya, Enhanced
thermal and dynamic mechanical properties of synthetic/natural hybrid composites with
graphene nanoplateletes, Polymers (Basel). 11 (2019).
https://doi.org/10.3390/polym11071085.

[156] Z. Candan, D.J. Gardner, S.M. Shaler, Dynamic mechanical thermal analysis (DMTA) of
cellulose nanofibril/nanoclay/pMDI nanocomposites, Compos. Part B Eng. 90 (2016) 126–
132. https://doi.org/10.1016/j.compositesb.2015.12.016.

[157] N. Saba, M. Jawaid, O.Y. Alothman, M.T. Paridah, A review on dynamic mechanical

173
properties of natural fibre reinforced polymer composites, Constr. Build. Mater. 106
(2016) 149–159. https://doi.org/10.1016/j.conbuildmat.2015.12.075.

[158] N. Hameed, P.A. Sreekumar, B. Francis, W. Yang, S. Thomas, Morphology, dynamic


mechanical and thermal studies on poly(styrene-co-acrylonitrile) modified epoxy
resin/glass fibre composites, Compos. Part A Appl. Sci. Manuf. 38 (2007) 2422–2432.
https://doi.org/10.1016/j.compositesa.2007.08.009.

[159] M. Jawaid, H.P.S. Abdul Khalil, A. Hassan, R. Dungani, A. Hadiyane, Effect of jute fibre
loading on tensile and dynamic mechanical properties of oil palm epoxy composites,
Compos. Part B Eng. 45 (2013) 619–624.
https://doi.org/10.1016/j.compositesb.2012.04.068.

[160] S.O. Adio, M.H. Omar, M. Asif, T.A. Saleh, Arsenic and selenium removal from water
using biosynthesized nanoscale zero-valent iron: A factorial design analysis, Process Saf.
Environ. Prot. 107 (2017) 518–527. https://doi.org/10.1016/j.psep.2017.03.004.

[161] J.R. Correia, M.M. Gomes, J.M. Pires, F.A. Branco, Mechanical behaviour of pultruded
glass fibre reinforced polymer composites at elevated temperature: Experiments and model
assessment, Compos. Struct. 98 (2013) 303–313.
https://doi.org/10.1016/j.compstruct.2012.10.051.

[162] X. Wu, Y. Gao, T. Jiang, L. Zheng, Y. Wang, B. Tang, K. Sun, Y. Zhao, W. Li, K. Yang,
J. Yu, 3D thermal network supported by CF felt for improving the thermal performance of
CF/C/Epoxy composites, Polymers (Basel). 13 (2021).
https://doi.org/10.3390/polym13060980.

[163] J. Antony, Training for design of experiments using a catapult, Qual. Reliab. Eng. Int. 18
(2002) 29–35. https://doi.org/10.1002/qre.444.

[164] S. Bard, M. Demleitner, R. Weber, R. Zeiler, V. Altstädt, Effect of curing agent on the
compressive behavior at elevated test temperature of carbon fiber-reinforced epoxy
composites, Polymers (Basel). 11 (2019). https://doi.org/10.3390/polym11060943.

[165] M. Bazli, M. Abolfazli, Mechanical properties of fibre reinforced polymers under elevated

174
temperatures: An overview, Polymers (Basel). 12 (2020) 1–31.
https://doi.org/10.3390/polym12112600.

[166] M. Akay, G.R. Spratt, B. Meenan, The effects of long-term exposure to high temperatures
on the ILSS and impact performance of carbon fibre reinforced bismaleimide, Compos.
Sci. Technol. 63 (2003) 1053–1059. https://doi.org/10.1016/S0266-3538(03)00018-6.

[167] S.K. Ha, G.S. Springer, Nonlinear elastic properties of organic matrix composites at
elevated temperatures, J. Eng. Mater. Technol. Trans. ASME. 110 (1988) 124–127.
https://doi.org/10.1115/1.3226019.

175

Vous aimerez peut-être aussi