Fill - Free fillable Thse de doctorat Prsente et soutenue publiquement Universit PDF form

N° :______

Thèse de doctorat

Présentée et soutenue publiquement à

Université Lille I Sciences et Technologies

pour obtenir le grade de

Docteur

Spécialité : Molécules et matières condensées

par

Jeremy CIRET

Thèse dirigée par

Prof. Serge BOURBIGOT et Prof. Sophie DUQUESNE

Soutenue le 16 février 2010 devant la Commission d’Examen composée de :

Prof. Jean-Bernard Vogt, Président du jury

Dr. Jannick Duchet-Rumeau, Rapporteur

Prof. Laurent Autrique, Rapporteur

Dr. Rachel Butler, Examinateur

Prof. Serge Bourbigot, Directeur de thèse

Prof. Sophie Duquesne, Co-directeur de thèse

Investigation of Intumescent

Coatings for Fire Protection -

Application to Jet-Fire

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

N° :______

Thèse de doctorat

Présentée et soutenue publiquement à

Université Lille I Sciences et Technologies

pour obtenir le grade de

Docteur

Spécialité : Molécules et matières condensées

par

Jeremy CIRET

Thèse dirigée par

Prof. Serge BOURBIGOT et Prof. Sophie DUQUESNE

Soutenue le 16 février 2010 devant la Commission d’Examen composée de :

Prof. Jean-Bernard Vogt, Président du jury

Dr. Jannick Duchet-Rumeau, Rapporteur

Prof. Laurent Autrique, Rapporteur

Dr. Rachel Butler, Examinateur

Prof. Serge Bourbigot, Directeur de thèse

Prof. Sophie Duquesne, Co-directeur de thèse

Investigation of Intumescent

Coatings for Fire Protection -

Application to Jet-Fire

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Acknowledgements

Ph-D Report J. Ciret Page 5

Acknowledgements

My first thank goes to Dr. J.M. Lefebvre and to Prof. R. Delobel for having welcomed me in

the LSPES Laboratory (UMR CNRS 8008) and in the PERF group, giving me the opportunity to

work on this interesting project and to benefit of all the knowledge and skills provided either

by the laboratory and CREPIM.

My sincere thanks go to my supervisors in the PERF group, Prof. Serge Bourbigot and Prof.

Sophie Duquesne, who provided me with the best scientific and moral support I could ever

have desired. I am especially grateful to them for their trust and guidance experienced for

the whole period of the PhD, which, combined with their continuous and stimulating

participation, kept me highly motivated and enthusiastic.

During the entire PhD, I was able to take advantage of a strong partnership with my

industrial sponsor. I particularly would like to acknowledge Dr. Paul Jackson for his generous

scientific and technical contribution and his strong impulse in the project. I also would like to

thank Dr Rachel Butler with whom I created a very enjoyable and efficient collaboration. I

finally thank the many other people of the company with whom I collaborate to bring this

project to a successful conclusion.

I also sincerely thank all of the current and former members of the PERF Laboratory for their

help and friendship over the last three years: Gaëlle, Michou, Brigitte, Nadine. To my

labmates Maude, Fabienne, Aurore, Virginie, Mathilde (x2), Oriane, Séverine, Caroline,

Fatima, Yohann, Thomas, Loulou, Damien, Nicolas, Antoine, Jeremie,... thanks for the fun

and support for these three years! I also would like to acknowledge the technicians Pierre,

and Mickaël for their help.

Particular thanks are addressed to Bertrand Revel for his expertise in NMR, to Christophe

Penvern and Maxence Vandewalle for the numerous XRD analyses.

Finally, I would like to thank my parents, my sisters and of course, my girlfriend Dr. Christelle

Réti (met thanks to this thesis) for their never-ending support, patience and affection.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Content

Ph-D Report J. Ciret Page 7

Content

ACKNOWLEDGEMENTS ..................................................................................................................................... 5

CONTENT ........................................................................................................................................................... 7

LIST OF FIGURES ............................................................................................................................................. 11

TABLE CAPTIONS ............................................................................................................................................ 15

RESUME FRANÇAIS ......................................................................................................................................... 17

GENERAL INTRODUCTION ............................................................................................... 25

CHAPTER I: STATE OF THE ART .................................................................................... 31

I. OFFSHORE FIRE RISK ASSESSMENT ...................................................................................................... 36

I. 1. Specific fire scenario on offshore platforms ........................................................................36

I. 2. An overview of the nature of hydrocarbon jet-fire...........................................................38

I. 2. a) Jet-fire length ............................................................................................................. 39

I. 2. b) Radiative and convective heat contribution .............................................................. 42

II. KEY REQUIREMENTS FOR OFFSHORE PROTECTION ............................................................................ 46

II. 1. Passive fire protection ..................................................................................................................46

II. 2. Epoxy based intumescent coatings .........................................................................................48

II. 3. Fire tests specified for offshore fire protection ..................................................................56

II. 3. a) Fire tests based on furnace conditions ...................................................................... 56

II. 3. b) Jet-fire resistance test ................................................................................................ 58

III. CONCLUSIONS .................................................................................................................................... 62

CHAPTER II: EVALUATION OF THE EFFICIENCY OF THE INTUMESCENT

COATINGS FACING TO JET-FIRE ..................................................................................... 63

I. MATERIALS AND EXPERIMENTAL TECHNIQUES.................................................................................. 66

I. 1. Materials ............................................................................................................................................66

I. 2. Jet-fire resistance test ...................................................................................................................68

II. LARGE-SCALE JET-FIRE TEST EVALUATION......................................................................................... 70

II. 1. Intumescent formulation 1 .........................................................................................................70

II. 2. Intumescent formulation 2 .........................................................................................................75

II. 3. Intumescent formulation 3 .........................................................................................................78

II. 4. Intumescent formulation 4 .........................................................................................................85

III. CONCLUSIONS AND DISCUSSION ABOUT THE JET-FIRE BEHAVIOURS ........................................... 88

CHAPTER III: PHYSICAL BEHAVIOURS OF INTUMESCENT COATINGS ......... 91

I. EXPERIMENTAL APPARATUS ................................................................................................................ 94

I. 1. Materials ............................................................................................................................................94

I. 2. Complex viscosity and expansion measurements .............................................................94

I. 3. Mechanical resistance of the char ...........................................................................................96

I. 4. Swelling abilities .............................................................................................................................96

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Content

Page 8 Ph-D Report J. Ciret

I. 5. Thermal stability ............................................................................................................................98

II. RESULTS AND DISCUSSION .................................................................................................................... 99

II. 1. Visco-elastic measurements .......................................................................................................99

II. 1. a) Precaution to use rheological tool to characterize intumescent coatings ................ 99

II. 1. b) Results of IF-1 and role of the intumescent additives .............................................. 100

II. 1. c) Visco-elastic measurements and swelling abilities of IF-2, IF-3 and IF-4 ................ 103

II. 2. Mechanical resistance of the formulations ...................................................................... 106

II. 3. Swelling abilities of the coatings .......................................................................................... 108

III. CONCLUSIONS ABOUT PHYSICAL PROPERTIES ..............................................................................114

CHAPTER IV: CHEMICAL APPROACH OF THE INTUMESCENT PROCESS .. 117

I. MATERIALS AND EXPERIMENTAL DETAILS .......................................................................................121

I. 1. Material ........................................................................................................................................... 121

I. 2. Thermal stability ......................................................................................................................... 122

I. 3. Heat treatment ............................................................................................................................. 122

I. 4. X-Ray diffraction .......................................................................................................................... 123

I. 5. Solid state NMR ............................................................................................................................ 123

II. BASIC MIXTURES BASED ON INTUMESCENT INGREDIENTS AND TIO

............................................125

II. 1. Thermal stability ......................................................................................................................... 125

II. 1. a) Basic mixtures based on APP and different carbon sources .................................... 125

II. 1. b) Basic mixtures based on APP, different carbon sources and TiO

........................... 128

II. 2. X-Ray diffraction analyses of the residue .......................................................................... 130

II. 2. a) Basic mixtures based on APP and different carbon sources .................................... 131

II. 2. b) Basic mixtures based on APP, different carbon sources and TiO

........................... 131

II. 3.

C NMR spectra of the residue .............................................................................................. 133

II. 3. a) Basic mixtures based on APP and different carbon sources .................................... 133

II. 3. b) Basic mixtures based on APP, different carbon sources and TiO

........................... 134

II. 4.

P NMR spectra of the residue .............................................................................................. 135

II. 4. a) Basic mixtures based on APP and different carbon sources .................................... 135

II. 4. b) Basic mixtures based on APP, different carbon sources and TiO

........................... 136

II. 5. Discussion and conclusion ....................................................................................................... 137

III. CHEMISTRY AND REACTIVITY OF THE COMPLETE INTUMESCENT FORMULATIONS ..................139

III. 1. Thermal stability of the intumescent formulations ...................................................... 139

III. 2. Mechanism of degradation of intumescent formulation 2 ........................................ 142

III. 2. a) Investigation of the structure of the intumescent system by X-Ray diffraction .. 142

III. 2. b) Investigation of the evolution of the carbon structure in the intumescent system

C solid-state NMR ............................................................................................................. 144

III. 2. c) Investigation of the evolution of the phosphorus components in the intumescent

system by

P solid-state NMR ................................................................................................. 146

III. 3. Mechanism of degradation of intumescent formulation 3 ........................................ 147

III. 3. a) Investigation of the structure of the intumescent system by X-Ray diffraction .. 147

III. 3. b) Investigation of the evolution of the carbon structure in the intumescent system

C solid-state NMR ............................................................................................................. 149

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Content

Ph-D Report J. Ciret Page 9

III. 3. c) Investigation of the evolution of the phosphorus components in the intumescent

system by

P solid-state NMR ................................................................................................. 150

III. 4. Mechanism of degradation of intumescent formulation 4 ........................................ 151

III. 4. a) Investigation of the structure of the intumescent system by X-Ray diffraction .. 152

III. 4. b) Investigation of the evolution of the carbon structure in the intumescent system

C solid-state NMR ............................................................................................................. 153

III. 4. c) Investigation of the evolution of the phosphorus components in the intumescent

system by

P solid-state NMR ................................................................................................. 155

III. 5. Discussion and conclusion ....................................................................................................... 156

III. 5. a) Influence of the carbon source on the chemical pathway and the formation of

TiP

156

III. 5. b) Chemical investigation of residues from jet-fire resistance test - Correlation with

proposed chemical pathways .................................................................................................. 159

CHAPTER V: DEVELOPMENT OF AN EFFICIENT LABSCALE TEST

MIMICKING LARGE SCALE JET-FIRE TEST ............................................................. 163

I. PRELIMINARY FIRE TEST IN FURNACE CONDITIONS .........................................................................167

I. 1. Experimental apparatus and materials ............................................................................ 167

I. 2. Temperature profiles and results ......................................................................................... 168

II. DEVELOPMENT OF A FIRE TEST MIMICKING JET-FIRE ......................................................................172

II. 1. Experimental settings and optimisation ........................................................................... 172

II. 1. a) Experimental apparatus .......................................................................................... 172

II. 1. b) Determination of the optimal parameters .............................................................. 173

II. 2. Tests on squared panels coated by intumescent formulations ................................ 177

II. 3. Tests on squared panels coated by intumescent formulations reinforced with

mesh 181

II. 4. Tests on flanged panels coated by intumescent formulations ................................. 183

II. 5. Conclusions and discussion about the benefit of a new experimental tool ......... 188

GENERAL CONCLUSIONS ................................................................................................ 191

APPENDIX: EXPERIMENTAL RESULTS OBTAINED WITH 2

CONFIGURATION .......................................197

configuration without mesh .......................................................................................................... 197

configuration with mesh ................................................................................................................. 198

BIBLIOGRAPHY .............................................................................................................................................201

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

List of figures

Ph-D report J. Ciret Page 11

List of figures

Figure 1 : Développement de la couche protectrice intumescente ................................................................................. 19

Figure 2 : Mise en évidence des zones de point faible pour IF-1, IF-2 et IF-3............................................................. 20

Figure 3 : Evolution de la viscosité en fonction de la température pour IF-1, IF-2, IF-3 et IF-4 ........................ 21

Figure 4 : Diffractogrammes des residus des trois formulations intumescentes après un traitement

thermique à 600°C................................................................................................................................................................................. 22

Figure 5 : Profils de température relevées au dos de plaque d'acier revêtue de peintures intumescentes IF-

2, IF-3 et IF-4 dans des conditions specifieés par la norme ISO 834- Feu d’hydrocarbures ................................ 23

Figure 6 : Pictures from Piper Alpha disaster (a) and Mumbai accident (b) .............................................................. 27

Figure 7 : China apparent steel consumption [1] .................................................................................................................... 33

Figure 8 : Reduction factors for steel strength and stiffness at elevated temperatures [5] ................................. 34

Figure 9 : Flame geometry for an expanding, rising fireball [6] ....................................................................................... 37

Figure 10 : Jet-fire lengths versus heat released for different hydrocarbons [27] ................................................... 40

Figure 11: Flame and lift-off diagrams ......................................................................................................................................... 41

Figure 12 : Variation of total heat flux over the surface of a pipe according to distance from the flame

source: (a) pipe at 21m; (b) pipe at 15m; (c) pipe at 9m [27] ........................................................................................... 43

Figure 13 : Temperatures versus time at different locations of a pipe engulfed by jet-fire [13] ....................... 43

Figure 14 : Fraction of radiative power versus total power observed for jet-fire with different

hydrocarbons [27] ................................................................................................................................................................................. 44

Figure 15 : Radiative heat flux as a fraction of total heat flux versus hydrocarbon content for an engulfed

object for a jet-fire [27] ....................................................................................................................................................................... 45

Figure 16 : Reaction and formula of DGEBA [62] .................................................................................................................... 50

Figure 17 : Reduction of the primary amine and opening of the epoxy ring............................................................... 50

Figure 18 : Formation of a 3D-network ....................................................................................................................................... 50

Figure 19 : Schematic representation of the reticulated network between epoxy resin and hardener ......... 51

Figure 20 :Formation of intumescent char ................................................................................................................................. 51

Figure 21 : Hydrolysis of ammonium polyphosphate ........................................................................................................... 53

Figure 22 : Ring closing esterification of the phosphor ester and subsequent reduction to a carbon

network ...................................................................................................................................................................................................... 54

Figure 23 : Development of intumescence (α= conversion degree) ............................................................................... 54

Figure 24 : Standard fire test curves used for materials applied on offshore platforms ....................................... 57

Figure 25 : Jet-fire resistance test in progress on an intumescent coating .................................................................. 59

Figure 26 : Jet nozzle used for jet-fire resistance test [103] ............................................................................................... 59

Figure 27 : Layout of test facility with steelwork test specimen used for evaluating protective fire

materials [103] ....................................................................................................................................................................................... 60

Figure 28 : Construction of structural steelwork test specimen with a central flange [103] .............................. 60

Figure 29 : Thermocouple positions for structural steelwork test specimen [103] ................................................ 61

Figure 30 : Formula of APP, n>1000 ............................................................................................................................................. 66

Figure 31 : Formula of THEIC ........................................................................................................................................................... 67

Figure 32 : Formula of pentaerythritol (a) and dipentaerythritol (b) ........................................................................... 67

Figure 33 : Structure of mesh with carbon and glass fibre ................................................................................................. 68

Figure 34 : Thermocouple positions for structural steelwork test specimen ............................................................ 69

Figure 35 : Pictures taken after jet-fire test of IF-1 ................................................................................................................ 71

Figure 36 : Thickness of non-reacted coating remaining at various locations after jet-fire test on IF-1 ....... 72

Figure 37 : Thickness of char at various locations after jet-fire test on IF-1............................................................... 72

Figure 38 : Temperature measured versus time on the right side of the box (in red on panel schema) for

jet-fire test on IF-1 ................................................................................................................................................................................. 73

Figure 39 : Temperature measured versus time on the left side of the box (in red on panel schema) for jet-

fire test on IF-1 ........................................................................................................................................................................................ 74

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

List of figures

Page 12 Ph-D report J. Ciret

Figure 40 : Temperature measured versus time on the central flange (in red on panel schema) for jet-fire

test on IF-1 ................................................................................................................................................................................................ 74

Figure 41 : Pictures taken after jet-fire test of IF-2 ................................................................................................................ 76

Figure 42 : Schematic diagram showing fire damage to the flange after jet-fire test on IF-2.............................. 76

Figure 43 : Temperature measured versus time on the central flange (in red on panel schema) for jet-fire

test on IF-2 ................................................................................................................................................................................................ 77

Figure 44 : Schematic diagram showing fire damage to the flange after jet-fire test on IF-3.............................. 79

Figure 45 : Pictures taken after jet-fire test of IF-3 ................................................................................................................ 80

Figure 46 : Pictures exhibiting the density and the hardness of IF-3 after jet-fire test .......................................... 81

Figure 47 : Temperature measured versus time on the central flange (in red on panel schema) for jet-fire

test on IF-3 not reinforced by mesh .............................................................................................................................................. 81

Figure 48 : Aspect of the flange reinforced by mesh after jet-fire test on IF-3 .......................................................... 82

Figure 49 : Temperature measured versus time on the central flange (in red on panel schema) for jet-fire

test on IF-2 reinforced by mesh ...................................................................................................................................................... 83

Figure 50 : Temperature measured versus time on back face of the box (in red on panel schema) for jet-

fire test on IF-2 reinforced by mesh .............................................................................................................................................. 84

Figure 51 : Pictures taken after second jet-fire test on IF-3 reinforced with mesh ................................................. 84

Figure 52 : Investigation of chars produced by the intumescent formulations (IF-1, IF-2 and IF-3) after jet-

fire resistance test ................................................................................................................................................................................. 85

Figure 53 : Pictures taken after jet-fire test of IF-4 ................................................................................................................ 86

Figure 54 : Char from IF-4 after jet-fire test exhibiting large voids close to metallic substrate ......................... 86

Figure 55 : Temperature measured versus time on the back face (in red on panel schema) for jet-fire test

on IF-4 ......................................................................................................................................................................................................... 87

Figure 56 : Zones of failure on jet-fire test according to the intumescent formulations ....................................... 89

Figure 57 : Dynamic visco-elastic measurements in a parallel-plate rheometer ...................................................... 95

Figure 58 : Measurement of the char strength in a parallel-plate rheometer ............................................................ 96

Figure 59 : Experimental set-up for measuring the swelling during a mass loss experiment using infrared

camera ........................................................................................................................................................................................................ 97

Figure 60 : IR images of an intumescent coating on steel plate upon heating at t=0s (a) and at the

maximum of expansion (b) ................................................................................................................................................................ 97

Figure 61 : Typical relative expansion as a function of time of an intumescent formulations during a mass

loss calorimeter experiment (external heat flux = 35 kW/m²) ......................................................................................... 98

Figure 62: Correlation between viscosity changes (F=100N) and thermal degradation of the system ......... 99

Figure 63 : TGA Curves of the four systems ............................................................................................................................ 101

Figure 64 : Viscosity measuremant versus temperature for epoxy resin alone with boric acid, with APP and

with both (F=200N) ........................................................................................................................................................................... 101

Figure 65 : Swelling abilities versus temperature for epoxy resin alone with boric acid, with APP and with

both ........................................................................................................................................................................................................... 102

Figure 66 : Viscosity versus temperature for IF-1, IF-2, IF-3 and IF-4 (F=100N) .................................................. 104

Figure 67 : TGA curves of the intumescent formulations in air at 10°C/min .......................................................... 104

Figure 68 : Swelling versus temperature for IF-1, IF-2, IF-3 and IF-4 measured in the rheometer .............. 105

Figure 69 : Mechanical resistance measured at 500°C for char issued from IF-1, IF-2, IF-3 and IF-4 .......... 107

Figure 70 : Comparative aspects of the four chars after 5 minutes of free development in rheometer

furnace at 500°C .................................................................................................................................................................................. 108

Figure 71 : Influence of the normal force applied by rheometer plates on swelling abilities of coatings

(here results of IF-2 are exposed) ............................................................................................................................................... 109

Figure 72 : Expansion measurements of IF-2 exposed to conical heater at two different heat fluxes ......... 110

Figure 73 : Expansion measurements of IF-3 exposed to conical heater at two different heat fluxes ......... 111

Figure 74 : Expansion measurements of IF-4 exposed to conical heater at two different heat fluxes ......... 112

Figure 75 : Experimental setup and temperature profile of the heat treatment in tubular furnace ............. 123

Figure 76 : TGA curves of the pure components in air at 10°C/min ............................................................................ 126

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

List of figures

Ph-D report J. Ciret Page 13

Figure 77 : TGA and derivative TGA curves of the mixtures based on APP and carbon sources in air at

10°C/min ................................................................................................................................................................................................ 127

Figure 78 : TGA and difference weight loss curves of APP/Epoxy resin/TiO

in air at 10°C/min .................. 129

Figure 79 : TGA and difference weight loss curve of APP/PER/TiO

in air at 10°C/min ................................... 129

Figure 80 : TGA and difference weight loss curve of APP/dPER/TiO

in air at 10°C/min ................................. 130

Figure 81 : X-Ray diffractograms of basic mixtures APP/Carbon source after heat treatment at T=450°C

..................................................................................................................................................................................................................... 131

Figure 82 : X-Ray diffractograms of the ternary systems APP/Carbon source/TiO

after heat treatment at

T=450°C .................................................................................................................................................................................................. 132

Figure 83 : X-Ray diffractograms of TiO

(red) and TiP

(blue) ................................................................................ 132

Figure 84 :

C NMR spectra of the binary systems APP/Carbon source after heat treatment at T=450°C 134

Figure 85 :

C NMR spectra of the ternary systems APP/Carbon source/TiO

after heat treatment at

Figure 86 :

P NMR spectra of the binary systems APP/Carbon source after heat treatment at T=450°C 136

Figure 87 :

P NMR spectra of the ternary systems APP/Carbon source/TiO

after heat treatment at

Figure 88 : TGA curves of the intumescent formulations in air at 10°C/min .......................................................... 139

Figure 89 : Derivative TGA curves of the intumescent formulations in air at 10°C/min.................................... 140

Figure 90 : Determination of the five characteristic temperatures .............................................................................. 142

Figure 91 : X-Ray diffraction patterns of IF-2 and of its residues at several characteristic temperatures. 143

Figure 92 : X-Ray diffractograms of APP and of TiO

.......................................................................................................... 144

Figure 93 :

C NMR spectra of IF-2 and of its residues at several characteristic temperatures .................... 144

Figure 94 :

C NMR spectrum of the thermoset cured resin .......................................................................................... 145

Figure 95 :

P NMR spectrum of IF-2 and of its residues at several characteristic temperatures ................ 146

Figure 96 : X-Ray diffraction patterns of IF-3 and of its residues at several characteristic temperatures. 148

Figure 97 :

C NMR spectra of IF-3 and of its residues at several characteristic temperatures .................... 149

Figure 98 : Assignment of bands on

C NMR spectrum of PER ..................................................................................... 149

Figure 99 :

P NMR spectra of IF-3 and of its residues at several characteristic temperatures .................... 150

Figure 100 : X-Ray diffraction patterns of IF-4 and of its residues at several characteristic temperatures

Figure 101 :

C NMR spectra of IF-4 and of its residues at several characteristic temperatures .................. 154

Figure 102 : Assignment of bands on

C NMR spectrum of dipentaerythritol ...................................................... 154

Figure 103 :

P NMR spectra of IF-4 and of its residues at several characteristic temperatures .................. 155

Figure 104 : Comparative

P NMR spectra of residues of IF-2, IF-3 and IF-4 at 300°C ..................................... 157

Figure 105 : Comparative

P NMR spectra of residues of IF-2, IF-3 and IF-4 at 450°C ..................................... 158

Figure 106 : Highlights of TiP

formation at different temperature according to the coating .................... 158

Figure 107 : Piece of IF-2 char from large scale jet-fire test ............................................................................................ 160

Figure 108 : X-Ray diffractograms of IF-2 residues from jet-fire resistance test ................................................... 160

Figure 109 : Piece of IF-4 char from large scale jet-fire test ............................................................................................ 161

Figure 110 : X-Ray diffractograms of IF-4 residues from jet-fire resistance test ................................................... 161

Figure 111 : Industrial furnaces used for UL 1709 standards ........................................................................................ 166

Figure 112 : Jet-fire box test and zones of high temperature and/or high pressure ............................................ 166

Figure 113 : Pictures of the small scale furnace test ........................................................................................................... 168

Figure 114 : Comparison of the time/temperature curves obtained for IF-2, IF-3, IF-4 and for epoxy resin

using the small scale furnace test ................................................................................................................................................ 169

Figure 115 : X-Ray diffractograms of residues of IF-2, IF-3 and IF-4 after small furnace tests ....................... 170

Figure 116 : Experimental setup of the labscale jet-fire test ........................................................................................... 172

Figure 117 : Temperature profiles on the backside of panels coated by IF-1, IF-2, IF-3 and IF-4 in labscale

jet-fire test at two heat fluxes, 50kW/m² (full curves) and 35kW/m² (dotted curves)...................................... 174

Figure 118 : Char aspect after labscale jet-fire test at 35kW/m² (on the left) and at 50kW/m² (on the right)

– Example of IF-2 ................................................................................................................................................................................ 174

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

List of figures

Page 14 Ph-D report J. Ciret

Figure 119 : Temperature profiles on the backside of panels coated by IF-1, IF-2, IF-3 and IF-4 in labscale

jet-fire test according to air jet contribution, with air jet (plain curves) and without air jet (dotted curves)

Figure 120 : Temperature profiles on the backside of panels coated by IF-1, IF-2, IF-3 and IF-4 according

to time when air jet is switching on: air jet is switched on at the beginning of the test (plain curves) or air

jet is switched on 150 seconds after the beginning of the test (dotted curves) ..................................................... 176

Figure 121 : Char aspect after labscale jet-fire test without time delaying (on the left) and with time

delaying (on the right) – Example of IF-2 ................................................................................................................................ 176

Figure 122 : Temperature profiles obtained in labscale jet-fire test for IF-1, IF-2, IF-3 and IF-4 (heat flux of

50kW/m², air speed of 15m/s, delayed time of 300 seconds) ....................................................................................... 178

Figure 123 : Picture of IF-1 char after labscale jet-fire test and thermal image for labscale jet-fire test (heat

flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds) ......................................................................... 178

Figure 124 : Picture of IF-2 char after labscale jet-fire test and thermal image for labscale jet-fire test (heat

flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds) ......................................................................... 179

Figure 125 : Picture of IF-3 char after labscale jet-fire test and thermal image for labscale jet-fire test (heat

flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds) ......................................................................... 180

Figure 126 : Picture of IF-4 char after labscale jet-fire test and thermal image for labscale jet-fire test (heat

flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds) ......................................................................... 180

Figure 127 : Temperature profiles obtained in labscale jet-fire test for IF-2, IF-3 and IF-4 reinforced by

mesh (heat flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds) ................................................ 181

Figure 128 : Pictures of the char obtained with intumescent coatings reinforced by mesh after labscale jet-

fire test (heat flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds) ............................................ 182

Figure 129 : Experimental setup to use flanged panels in labscale jet-fire test ..................................................... 183

Figure 130 : Temperature profiles obtained in labscale jet-fire test for IF-1 coated on flanged panel (heat

flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds) ......................................................................... 184

Figure 131 : Char obtained from IF-1 after labscale jet-fire test on flanged panel (heat flux of 50kW/m², air

speed of 15m/s, delayed time of 300 seconds) ..................................................................................................................... 184

Figure 132 : Pictures of chars obtained from IF-2, IF-3 and IF-4 after labscale jet-fire test on flanged panel

(heat flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds) ............................................................. 185

Figure 133 : Temperature profiles obtained in labscale jet-fire test for IF-2 coated on flanged panel (heat

flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds) ......................................................................... 186

Figure 134 : Temperature profiles obtained in labscale jet-fire test for IF-3 coated on flanged panel (heat

flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds) ......................................................................... 187

Figure 135 : Temperature profiles obtained in labscale jet-fire test for IF-4 coated on flanged panel (heat

flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds) ......................................................................... 187

Figure 136 : Temperature profiles obtained in labscale jet-fire test for IF-1, IF-2, IF-3 and IF-4 (heat flux of

50kW/m², air speed of 15m/s, delayed time of 150 seconds) ....................................................................................... 197

Figure 137 : Picture of chars after labscale jet-fire test and thermal image for labscale jet-fire test (heat

flux of 50kW/m², air speed of 15m/s, delayed time of 150 seconds) ......................................................................... 198

Figure 138 : Temperature profiles obtained in labscale jet-fire test for IF-2, IF-3 and IF-4 reinforced by

mesh (heat flux of 50kW/m², air speed of 15m/s, delayed time of 150 seconds) ................................................ 199

Figure 139 : Pictures of the char obtained with intumescent coatings reinforced by mesh after labscale jet-

fire test (heat flux of 50kW/m², air speed of 15m/s, delayed time of 150 seconds) ............................................ 199

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Table captions

Ph-D report J. Ciret Page 15

Table captions

Tableau 1 : Tableau récapitulatif de la composition des formulations intumescentes .......................................... 18

Table 2 : Examples of components playing a role in intumescent coatings [79, 80] ............................................... 53

Table 3 : Ratio between the two parts used to prepare the intumescent coatings .................................................. 66

Table 4 : Comparison between the active ingredients of 4 formulations ..................................................................... 67

Table 5 : Comparative sum-up of the different beahviours exhibited by the four intumescent formulations

for large-scale jet-fire resistance test ........................................................................................................................................... 88

Table 6 : Experimental protocols used to reach visco-elastic properties ..................................................................... 95

Table 7 : Swelling temperature range ....................................................................................................................................... 105

Table 8 : Swelling abilities obtained with and without normal force and ratio between the two

measurements to quantify sensitivity to the normal force .............................................................................................. 112

Table 9 : Comparison between the active ingredients of 3 formulations .................................................................. 121

Table 10 : APP/Carbon source/TiO

ratio of the basic mixtures .................................................................................. 121

Table 11 : Screening of the different setting parameters used in the labscale jet-fire test ............................... 173

Table 12 : Optimized configurations used to evaluate the four coatings with the labscale jet-fire test ...... 177

Table 13 : Evaluation and correlation of the behaviours of coatings .......................................................................... 189

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Résumé français

Ph-D report J. Ciret Page 17

Résumé français

Cette thèse s’intéresse aux comportements de quatre peintures intumescentes (appelées

dans la suite du mémoire IF-1, IF-2, IF-3 et IF-4) élaborées à base d’une résine époxyde et

destinées à protéger contre le feu les structures métalliques de plateformes offshores

(Tableau 1). L’utilisation de telles peintures protectrices dans un environnement si hostile

que les plateformes pétrolières ou les sites pétrochimiques requiert évidemment des

performances supérieures aux protections thermiques classiques. Nous décrivons dans le

premier chapitre les principaux risques d’incendie possibles sur un site d’hydrocarbures.

Nous avons détaillé les particularités du « jet-fire », l’un des feux les plus puissants et les

plus dangereux pour les hommes et la structure métallique de la plateforme, alimenté par

une fuite continue et directionnelle d’hydrocarbures. Nous nous sommes intéressés

également aux moyens mis en œuvre pour le combattre, non pas en l’évitant (un feu est

toujours lié à un accident et donc évitable mais imprévisible) mais en comprenant son action

et en limitant son effet sur la structure. Pour cela l’utilisation de revêtements protecteurs,

en particulier les peintures intumescentes à base de résine époxyde (intumescence signifie la

formation d'un bouclier protecteur carboné expansé faisant office de barrière thermique et

limitant les transferts de gaz de dégradation combustibles pouvant alimenter la flamme), et

les tests développés pour évaluer leur performance ont été détaillés. Dans le second

chapitre, nous avons étudié le comportement des formulations intumescentes lors d’un test

de résistance au jet-fire. Puis nous avons tenté dans les troisième et quatrième chapitres de

comprendre et d’expliquer ces comportements grâce à des outils expérimentaux permettant

d’étudier les propriétés physiques de nos revêtements intumescents et l’évolution chimique

de ses formulations. Enfin, le dernier chapitre de ce manuscrit s'est attaché au

développement d’un test à l’échelle laboratoire permettant de reproduire les phénomènes

caractéristiques du jet-fire. Cet outil expérimental permet de mimer les comportements des

formulations intumescentes à petite échelle de manière rapide et fiable.

D’un point de vue de la gestion des risques, les plateformes pétrolières sont connues pour

leur haut degré de risque à cause de leur situation insulaire, de leur aménagement

congestionné et bien sur à cause de l’inflammabilité des produits extraits, les hydrocarbures.

Malheureusement il a fallu attendre des drames pour prendre en compte ces circonstances

et l’accident de Piper Alpha en 1988 a permis l’élaboration de plusieurs rapports étudiant les

feux spécifiques aux plateformes offshores. Ces rapports ont permis de mettre en évidence

la particulière dangerosité du « jet-fire ». Le jet-fire est une flamme turbulente résultant de

la combustion d'hydrocarbures libérés en continu dans une direction particulière avec une

importante quantité de mouvement. Ainsi les deux paramètres importants d’un tel

phénomène sont clairement exprimés dans sa définition : une direction et une vitesse

significative dangereuses pour les hommes et les structures qui pourraient être « frappées »

par la flamme. Pour limiter les dommages structurels et les températures élevées,

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Résumé français

Page 18 Ph-D report J. Ciret

l’utilisation de matériaux isolants est donc étudiée. Cependant la force de la flamme et

l’environnement agressif (corrosion, climat) imposent l’utilisation de résines intumescentes

à base époxydes (les résines à base époxy sont les plus résistantes aux conditions

agressives).

Une peinture intumescente utilisée pour la protection des structures métalliques est

généralement constituée de trois composants principaux (Tableau 1):

- La source acide susceptible de libérer un acide inorganique (par exemple l’acide

phosphorique) lorsqu’il est soumis à une élévation de température. Le polyphosphate

d’ammonium (APP) ou des phosphates de mélamine sont généralement utilisés; dans notre

cas les quatre formulations comportent de l’APP. Dans IF-1, de l’acide borique est également

ajouté comme source acide additionnelle.

- La source de carbone contenant des groupes hydroxyle (-OH) susceptibles de réagir

avec la source acide. Dans le cas des résines époxyde, la source carbonée peut être

constituée par le polymère lui-même, c’est le cas pour IF-2, ou bien un composé spécifique

est ajouté, souvent des polyols, du pentaérythritol (PER) pour IF-3 ou bien du

dipentaérythritol (dPER) pour IF-4 ; l’enrobant de l’APP utilisé dans IF-1.

- L’agent d’expansion libérant sous l’action de la chaleur une quantité importante de

gaz ininflammables provoquant l’expansion de la structure phospho-carbonée, comme par

exemple la mélamine, ou dans notre cas le polyphosphate d’ammonium (APP) qui joue un

double rôle (dégagement d'ammoniac).

En plus de ses composés actifs dans le processus de l’intumescence, les formulations sont

complexes et contiennent des additifs et des charges couramment utilisé dans les peintures.

La formulation exacte n’est pas connu néanmoins la présence de TiO

est avérée.

Tableau 1 : Tableau récapitulatif de la composition des formulations intumescentes

Peintures

intumescentes

Matrice polymère

Source acide

Source de carbone

Agent

gonflant

IF-1

Prépolymère de

type DGEBA +

polyaminoamide

APP enrobé +

Acide borique

Résine époxy + THEIC

(enrobant l’APP)

APP

IF-2

Prépolymère de

type DGEBA +

polyaminoamide

APP

Résine époxy

APP

IF-3

Prépolymère de

type DGEBA +

polyaminoamide

APP

Résine époxy + PER

APP

IF-4

Prépolymère de

type DGEBA +

polyaminoamide

APP

Résine époxy + dPER

APP

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Résumé français

Ph-D report J. Ciret Page 19

Le processus de développement de l’intumescence est généralement décrit de la manière

suivante. L’acide libéré lors de la dégradation de la source acide réagit avec l’agent source de

carbone pour produire une structure phospho-carbonée. L’agent gonflant se décompose

ensuite et entraîne un dégagement important de produits gazeux, ce qui provoque le

gonflement du bouclier et la création de la structure intumescente, une couche protectrice

isolante (Figure 1).

Figure 1 : Développement de la couche protectrice intumescente

Le bouclier se forme dans une phase semi-liquide qui coïncide avec la formation des gaz et

l’expansion de la surface, puis il se solidifie ensuite en une couche multicellulaire rigide.

L’efficacité d’un système intumescent dépend de sa capacité à gonfler et à développer une

structure multicellulaire. Le gonflement est dû à la diffusion lente des gaz de dégradation du

système dans une matrice charbonnée partiellement dégradée. Lors de ce phénomène

d’intumescence, l’étude de la viscosité est donc essentielle: si la matrice dégradée est trop

liquide, les gaz ne sont pas piégés et diffusent facilement pour alimenter la flamme. Si au

contraire la matrice dégradée est trop visqueuse, l’augmentation de pression entraînée par

le dégagement de produits gazeux peut conduire à l’apparition de fissures permettant la

diffusion d’oxygène ainsi qu’un échange de chaleur et de matière entre la flamme et le

polymère. Il est également important de caractériser la résistance mécanique de la structure

développée car celle-ci doit conserver ses propriétés protectrices même sous l’action de

contraintes externes telle que l’impact d’une flamme issue du jet-fire. Enfin, il faut bien sûr

caractériser les systèmes intumescents selon leur capacité de résistance au feu. Pour cela

des tests normés ont été développés avec des courbes temps/température spécifiques aux

feux d’hydrocarbures. Cependant aucun de ces tests ne prenait en compte l’aspect

dynamique du jet-fire avec l’impact de la flamme sur la structure. Ainsi un nouveau test a

récemment été développé par des industriels et normé en 2007 pour déterminer la

résistance aux feux propulsés des matériaux de protection passive contre l'incendie.

Ce test à grande échelle réalisé sur nos quatre formulations intumescentes a permis de

déterminer quatre comportements spécifiques face à l’impact d’une flamme et aux

conditions extrêmes de température (Figure 2). Sous les contraintes thermiques et

mécaniques du jet-fire IF-1 réagit et développe un bouclier très dur et résistant à l’impact.

Cependant ce bouclier est peu expansé et n’apporte pas une isolation suffisante du substrat

au point le plus chaud, localisé sur le haut de la partie centrale.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Résumé français

Page 20 Ph-D report J. Ciret

Par contre la seconde formulation intumescente (IF-2) gonfle suffisamment sur toute la

surface de la plaque d’essai. Mais le bouclier expansé présente une très faible résistance

mécanique et est détruit sous l’impact du jet-fire et de la pression inhérente. L’utilisation

d’un grillage pour renforcer la structure expansée et pour maintenir en place la protection

apporte une amélioration significative et permet de meilleures performances face au jet-fire.

Ensuite la formulation intumescente IF-3 est présentée comme un compromis entre les deux

premières formulations. Elle doit permettre une bonne expansion pour isoler suffisamment

et une résistance mécanique face au jet-fire, néanmoins l’utilisation du grillage renforçant

est toujours nécessaire. La structure protectrice résiste donc à la pression et isole la plaque

métallique des températures extrêmes, mais lors du test on observe un écoulement de la

peinture sur les faces latérales de la plaque laissant non protégées de larges zones. Ainsi le

point faible d’IF-3 est observé à un endroit où ni IF-1 ni IF-2 ne se sont révélées sensibles.

Figure 2 : Mise en évidence des zones de point faible pour IF-1, IF-2 et IF-3

Enfin IF-4 échoue au test de résistance aux feux propulsés par manque d’adhésion, la

structure expansée ayant révélé de grandes alvéoles creuses à l’interface métal/revêtement.

Ces alvéoles permettent ainsi le passage des flammes et de la chaleur.

Dans ce travail, nous avons donc tenté d’expliquer ces quatre comportements très différents

en s’intéressant aux propriétés physiques et aux évolutions chimiques de ces revêtements

intumescents. Les propriétés du 1

revêtement (IF-1) ont été particulièrement étudiées par

Maude Jimenez dans ses travaux de thèse qui a pu mettre en évidence l’effet de l’acide

borique sur les propriétés viscoélastiques et le mécanisme de dégradation d’IF-1. De

manière similaire nous nous sommes donc intéressés aux trois autres formulations, sachant

que la source de carbone est respectivement la résine époxyde seule dans IF-2, la résine

époxyde et le pentaérythritol dans IF-3, la résine époxyde et le dipentaérythritol dans IF-4.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Résumé français

Ph-D report J. Ciret Page 21

Tout d’abord l’étude des propriétés mécaniques (mesure de la résistance du char en

compression lorsqu'il subit une contrainte normale) à l’aide d’un rhéomètre plan-plan utilisé

de manière originale selon une technique développée au laboratoire ont permis de

confirmer la différence existant entre IF-1 développant un bouclier peu expansé mais

résistant et les trois autres formulations, qui gonflent beaucoup mais qui présentent une

faible résistance à la contrainte. Une raison avancée pour expliquer cette différence est la

présence de fibres silicates et d’acide borique dans IF-1 (Maude Jimenez a clairement

montré l’effet de ces ingrédients sur les propriétés mécaniques du bouclier intumescent).

Ensuite un effet de la source de carbone a été démontré sur l’évolution de la viscosité

apparente des systèmes en fonction de la température (ici nous mesurons une viscosité

apparente pour nous permettre de quantifier dans des conditions expérimentales données

les propriétés d'écoulement de nos matériaux). En effet, le pentaérythritol provoque un

déplacement vers les basses températures du pic de viscosité observé lors du phénomène

d’intumescence. Ainsi le minimum de viscosité est observé respectivement vers 270°C pour

IF-3 et vers 360°C pour les 3 autres formulations (Figure 3).

Figure 3 : Evolution de la viscosité en fonction de la température pour IF-1, IF-2, IF-3 et IF-4

Ce comportement viscoélastique particulier de IF-3 est clairement attribué au PER comme le

montre le comportement de IF-4 où le PER a été remplacé par du dipentaérythritol. Ce large

éventail de température où la viscosité du système est faible est avancé comme explication

à l’échec d’IF-3 au test de résistance de jet-fire et l’utilisation d’IF-4 plutôt qu’IF-3 permet de

résoudre partiellement ce problème.

De plus l’effet de la source de carbone a également été étudié par une approche chimique.

En effet, des analyses chimiques par diffraction des rayons X et RMN à l’état solide des

noyaux

C et

P ont été réalisées à différentes températures caractéristiques de la

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Résumé français

Page 22 Ph-D report J. Ciret

dégradation des formulations intumescentes afin de comprendre leur processus

d’intumescence. Des mécanismes de dégradation ont été proposés pour les 3 formulations

IF-2, IF-3 et IF-4 en fonction de leurs particularités. Quelque soit la source de carbone, nous

avons démontré que l’APP réagissait dans un premier temps avec la source de carbone pour

former une structure charbonnée et des composés phosphorés issus de sa dégradation. Ces

produits de dégradation et de réaction de l’APP présentent une réactivité importante avec

de nombreux composés thermiquement stable et présents dans les formulations

intumescentes. Lors de la dégradation thermique de nos formulations, nous observons ainsi

une réaction entre ces composés à base de phosphore et le dioxyde de titane TiO

pour

donner une structure cristalline de pyrophosphate de titane TiP

. D’après la littérature,

cette structure apporte des améliorations en termes de résistance mécanique et de

performance au feu du bouclier développé. De plus, nous avons mis en évidence que la

source de carbone intervient sur la température à laquelle les réactions se déroulent. En

effet nous avons montré que l’évolution chimique du revêtement à base de

dipentaérythritol était accélérée, la réactivité entre l’APP et le dipentaérythritol est

supérieure à celle entre l’APP et le PER ou entre l’APP et le réseau époxyde. Cela se traduit

par la formation de la structure protectrice et du TiP

à plus basses températures comme

on peut le voir sur les diffractogrammes à 600°C des trois formulations intumescentes

(Figure 4).

Figure 4 : Diffractogrammes des residus des trois formulations intumescentes après un traitement

thermique à 600°C

Ainsi l’influence de la source de carbone est démontrée. Même si le mécanisme de

dégradation est similaire quelque soit la source de carbone utilisée (résine époxyde,

pentaérythritol et dipentaérythritol) les étapes « clés » de formation de la structure

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Résumé français

Ph-D report J. Ciret Page 23

protectrice et du TiP

se déroulent à des températures moins élevées en présence de

dipentaérythritol. Ce mécanisme a également été mis en évidence lors des tests de

résistance au feu et la formation précoce de la structure cristalline à base de titane pourrait

expliquer les meilleures performances observées lors d’un test dans des conditions de feu

d’hydrocarbures (norme ISO 834) (Figure 5).

Figure 5 : Profils de température relevées au dos de plaque d'acier revêtue de peintures

intumescentes IF-2, IF-3 et IF-4 dans des conditions specifieés par la norme ISO 834- Feu

d’hydrocarbures

Enfin et puisque la réalisation de test à grande échelle est financièrement coûteuse et ne

permet pas l’évaluation de nombreuses formulations, nous avons dans un dernier chapitre

développé un test imitant les conditions d’un jet-fire à l’échelle laboratoire. En associant un

cône irradiant pour apporter la contribution radiative présente lors d’un jet-fire et un jet

d’air pour l’aspect dynamique et la quantité de mouvement, nous avons construit un

dispositif expérimental pour tester nos formulations intumescentes dans des conditions de

températures et de pression permettant des corrélations satisfaisantes entre le test de

résistance à grande échelle et à l’échelle laboratoire. En effet après une phase

d’optimisation des paramètres, nous avons réussi à obtenir des comportements de nos

peintures facilement identifiables. Quelque soit les conditions de test, IF-1 présente le même

comportement, développant un bouclier peu expansé mais résistant. Nous retrouvons ainsi

les conclusions du test à grande échelle d’IF-1. Concernant IF-2 et IF-3, la résistance

mécanique de la structure est également mise en cause à l’échelle réduite puisque le

revêtement est « soufflé » par le jet d’air dès le premier impact. De la même façon que lors

du test normé, l’utilisation de grillage renforçant a permis de maintenir une protection et de

retarder l’augmentation de température. Enfin la substitution de PER par le

dipentaérythritol a été étudiée en comparant le comportement d’IF-3 et d’IF-4 : le test

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Résumé français

Page 24 Ph-D report J. Ciret

réalisé sur IF-4 a montré des différences significatives avec celui d’IF-3, une couche

résiduelle résistante est formée en présence de dipentaérythritol permettant une protection

substantielle du substrat.

Ainsi durant cette thèse, trois approches ont été suivies pour étudier et expliquer le

comportement des peintures intumescentes destinées à la protection de plateformes

offshores et susceptibles de subir un jet-fire. Une approche physique tout d’abord qui nous a

permis de mettre en avant l’importance de la résistance mécanique et du comportement

viscoélastique des peintures intumescentes. L’influence du PER sur la viscosité a été évalué

et son effet négatif a obligé la substitution du PER par du dipentaérythritol dans la

formulation. Une approche chimique a ensuite permis de mettre en avant un mécanisme de

dégradation pour chaque formulation avec la formation d’une structure protectrice et d’un

composé cristallin (TiP

) bénéfique pour les performances feu des revêtements. Le

dipentaérythritol joue également un rôle important sur l’aspect chimique des formulations

puisqu’il présente une meilleure réactivité avec l’APP que le PER ou la résine époxyde,

permettant aux étapes « clés » du mécanisme de se réaliser à plus faible température. Enfin

une approche expérimentale nous a amené à développer un test à l’échelle laboratoire pour

simuler les conditions d’un « jet-fire » et étudier les comportements des formulations

intumescentes.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

General Introduction

Ph-D report J. Ciret Page 25

General Introduction

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

General Introduction

Ph-D report J. Ciret Page 27

In 1988, Occidental's Piper Alpha platform (United Kingdom) was destroyed by explosion and

fire and 167 workers were killed in the blaze. 17 years later, on Mumbai High North platform

(India), a support vessel collided with the structural framework, ruptured a riser causing a

major fire which destroyed the platform and killed 11 persons. In spite of advancement and

progress in fire protection, these dramatic events, picked on a too long list, have widely

opened the mind about fire hazards possible on offshore platforms. Moreover several key

lessons have been learnt following each accident. All major disasters lead to a tightening of

regulations and to the development of new technology. The protection of steel framework

against fire has notably become an important issue in the construction industry. Protection

of residential or commercial building has been regulated since several years but it is only

after 1988 and Piper Alpha disaster that characteristics of the offshore platforms have been

taken into account to propose new fire tests and new materials.

Usual cellulosic fire tests have been substituted by more severe hydrocarbon tests. Use of

active and passive fire proofing have been required and thanks to their properties,

intumescent coatings (intumescence means the formation of an expanded carbonaceous

char upon heating) have become one of the easiest and one of the most efficient ways to

protect steel structure against fire. The coating has the ability to expand by a factor of 40 or

more upon heat exposure that effectively leads to the protection of the substrate against

rapid increase of temperature, thereby maintaining the structural integrity of the building.

Figure 6 : Pictures from Piper Alpha disaster (a) and Mumbai accident (b)

In this context, this dissertation, divided in five parts, is based on the study of an epoxy

based intumescent coating, applied on offshore platforms. These coatings have to ensure

protection of steel in case of “jet-fires”, which means that it has to resist to very severe

conditions combining extremely high heating (fire) and high momentum (velocity of the

jet).

The first chapter of this Ph-D work proposes a state of the art focused on the specific fire

hazards that can occur on offshore platforms and on the new solutions developed to

struggle against them and avoid human and materials losses. Different hydrocarbon fire

situations will be detailed, as pool fires or fireballs but it is jet-fires that will catch particularly

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

General Introduction

Page 28 Ph-D report J. Ciret

our attention because of its severity and consequence on steel structure. Jet-fires represent

an important fire risk in offshore installations and potential jet-fires may vary with respect to

flame length, heat flux and velocities. We propose thus to study jet-fire taken into account

these parameters. Then in a second section of this state of the art, we will deal with the

means developed to limit jet-fire effect and obviously the intumescent coatings. They are

designed to perform under severe conditions and to maintain the steel integrity for up to

three hours when the temperature of the surroundings is in excess of 1100°C. Their

performances are tested using fire test, classically based on time temperature curves. We

will comment typical fire scenario. Furthermore, we will treat of a specific fire test,

developed to simulate jet-fire and to determine the resistance of passive fire protection

materials facing to high load resulting from a gas line rupture on an offshore oil rig.

The second chapter is also dedicated to jet-fires and particularly to the standard developed

by the Health and Safety Executive to determine the efficiency of the passive fire protection

coating facing to jet-fire conditions: OTI 95654 Jet-fire resistance - Test of passive fire

protection materials. After a description of the experimental settings and the parameters

used for severe conditions, we will detail the results exhibited by the four intumescent

formulations that have been investigated in this work. Their weaknesses will be highlighted

by a thermal behaviour and illustrated by an investigation of the specimen after test.

Physical and chemical properties of the intumescent coatings will be then respectively

investigated on a third and fourth chapter. Indeed, the efficiency of the char is closely

related to the physical behaviours, namely its ability to expand and the visco-elastic

properties of the melted degraded matrix during the intumescent process. The swelling of

intumescent coating is due to the slow diffusion of the evolved degradation gases released

into the degraded matrix. The importance of the visco-elastic properties of this layer is then

crucial since it would affect the porosity and the expansion of the resulting char. The

toughness of the char will also be determined thanks to an original test based on a

rheological approach.

In the fourth chapter, we will report the complete investigation of the selected intumescent

formulations using a chemical approach investigating their degradation pathway. Preliminary

analyses on mixtures of the additives contained in the coating have first been carried out. At

the end of this chapter, we will expect to propose a mechanism of degradation for our

formulations, highlighting the main particularities and the role played by the ingredients and

their interactions. Knowing the chemical and physical properties of the coating (Chapters III

and IV), our goal will be to explain their behaviour facing the jet-fire exposed on the second

chapter.

Finally, the evaluation of coatings in hydrocarbon and jet-fire test regimes is both expensive

and time consuming and the companies involved in the research and development of fire

protection coatings are looking to reducing these costs by developing high throughput

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

General Introduction

Ph-D report J. Ciret Page 29

screening using small scale fire tests. Another objective of this PhD work is to examine

whether the performance in large scale industrial fire tests can be correlated with

parameters of the intumescent coating measured using labscale tools. Therefore the last

chapter of this report (Chapter V) will talk about the development of a novel labscale tests

permitting to predict the behaviour of intumescent formulations for the jet-fire test.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Ph-D report J. Ciret Page 31

Chapter I: State of the

art

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Ph-D report J. Ciret Page 33

A century ago, steel was a comparatively unusual building material, mass production of steel

did not begin until 1855 and then it took several decades before it became popular as a

building material [1]. However, steel is today a much appreciated building material all over

the world and is used in a vast array of different types of buildings; from huge sport arenas

to offshore platforms. World steel demand has increased steadily since 1999, but this

increase has been accelerating since 2002 and represents around 50 million tonnes more per

year. This strong surge in steel consumption is the result of the dramatic acceleration of

domestic steel demand in China (Figure 7).

Figure 7 : China apparent steel consumption [1]

The selection of structural steel as a building’s framing system brings numerous benefits to a

project. All other materials are measured against the standard of structural steel and

structural steel is still the material of first choice [2]. These benefits include economic

approach with speed of construction, lower project costs or readily availability but also more

technical ones. Structural steel is a non-combustible material (M-0 rating). It exhibits a large

standard strength in both compression and tension. Moreover structural steel, long

considered the premier green construction material, is continuing to improve its

environmentally friendly position by reducing greenhouse carbons emissions (-37.7%

between 1990 and 2003) [2]. At the same time, the industry remains the world leader in the

use of recycled material, with recycled content now accounting for 95% of the structural

steel produced.

However, structural steel has some disadvantages. Most steel are susceptible to corrosion

when freely exposed to air and water, and therefore must be covered periodically with anti-

corroding paints. The use of weathering steels, however, in suitable applications tends to

eliminate this cost. Furthermore, steel is sensitive to heat and fire since it is an excellent

heat conductor. Although structural parts are incombustible, their strength is detrimentally

reduced at elevated temperatures reaching in fires. The major focus has been on the tensile

and strength properties of the steels at high temperature [3]. Steel begins to lose most of its

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Page 34 Ph-D report J. Ciret

structural properties from 550°C that is defined as critical temperature. The critical

temperature of steel is defined as the temperature at which the materials loses up to 50% of

its strength and can no longer support the design load, that being the maximum load

permitted by the structural provisions of the building codes [4]. Yield strength is the stress at

which material strain changes from elastic deformation to plastic deformation, causing it to

deform permanently. In Figure 8, the ratio between yield strength at temperature θ and the

yield strength at ambient is plotted versus temperature θ, we remark that ratio decreases

from 400°C and reaches 0.5 at 550°C.

Figure 8 : Reduction factors for steel strength and stiffness at elevated temperatures [5]

This point must be taken into consideration by architects and solutions must be developed in

accordance with the structural steel and the locations of the structure. Indeed, regulations

and requirements are not the same for a residential, a commercial building or an offshore

platform where risks are maximum.

Our study is based on materials used for offshore platforms.

The first part of this chapter describes the different risk on offshore platform sites. Offshore

quantitative risk assessments have historically been complex and costly, but here we try to

introduce the main fire risks that could occur in offshore and we focus on jet-fire that

appears as the most severe fire scenario. Its properties and its parameters are investigated

to understand the different modes of heat transfer occurring during the test and to propose

potential safety solutions.

Then the second part deals with the several ways existing for the protection of steel and the

particular requirements for offshore platforms. These systems are called “passive

fireproofing materials”, which means insulating systems designed to decrease the heat

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Ph-D report J. Ciret Page 35

transfer from a fire to the structure being protected. Currently, the intumescent coating

based on epoxy resin appears as the best mean to protect steel framework which requires

high protection against fuel fires and to resist the strong conditions as fuel fires or jet-fires

occurring in offshore platforms, we will detail thus their action and their characteristics in

this part.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Page 36 Ph-D report J. Ciret

I. Offshore fire risk assessment

Offshore oil and gas platforms are well known for their compact geometry, high degree of

congestion, limited ventilation and difficult escape routes. A study carried by the UK Health

and Safety Executive showed that process and structural failure incidents account for almost

80% of the risk to personnel offshore. Potential risks offshore include: blowouts, riser and

process leaks, fires, explosions, vessel collisions, helicopter accidents, dropped objects,

structural failures, and capsizing. An examination of incidents such as Piper Alpha in the

North Sea and the P-36 production semi-submersible off Brazil reveals that most offshore

incidents are in fact process-related and a small mishap under such conditions can quickly

escalate into a catastrophe. However, among all the accidental process-related events

occurring offshore, fire is the most frequently reported [6-8].

I. 1. Specific fire scenario on offshore platforms

This section considers the physical fires types that, in the light of the identified fire event,

need to be appraised in terms of harm. Fire events on offshore result always from a

hydrocarbon escape. This loss of hydrocarbon containment can arise from mechanical

failure, damage or procedural failures. The leakage rates and their duration greatly influence

the nature and extent of a fire. Consequently, offshore hydrocarbon fires can have varied

characteristics and encompass a very extensive range of size, fires might also be preceded by

an explosion causing damage to structures, process plant or fire protection systems and

thereby affect the extent of fire development. The fire events and their ranges had arrived

by considering what could be released, where and under what conditions and the possible

event following the release. To do this, the following information is desirable:

- The range of flammable hydrocarbons on offshore facilities

- How they arrive, are stored and leave

- Where they are processed/stored/used

- Typical transfer/process/storage conditions, pressure and temperatures

- Typical releases for each category of material, release rates and the inventory at risk

- Possible events following release

Based on these considerations above, three main classes of fire events can be defined and

characterized.

A pool fire is a turbulent diffusion fire burning above a pool of vaporizing hydrocarbon fuel

where the fuel vapor has negligible initial momentum. The probability of occurrence of pool

fires on offshore platforms is high due to continuous handling of heavy hydrocarbons

onboard. Liquid fuel released accidentally during overfilling of storage tanks, rupture of

pipes and tanks etc., forms a pool on the surface, vaporizes, and upon ignition, results in a

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Ph-D report J. Ciret Page 37

pool fire. Consequence models for pool fires in open spaces have been well-documented

over the past few years [9-11]. A key feature of these fires is that there is a degree of

feedback between the fire and the fuel in that, to a greater or lesser extent, heat transfer

back from the fire to the pool influences or even controls the rate of evaporation and hence

the fire size and other characteristics. The prime variable other than fuel type that influences

the fire characteristics is the pool size. Wind effects are also significant in deciding certain

parameters such as flame drag, flame tilt and flame length [6].The condition of the substrate

may also be influential via heat transfer into the pool liquid. The hazard consequence of pool

fires are pool spread, flame extent and geometry and external thermal radiation from

combustion products. These parameters determine the potential harm of this kind of fires.

A jet-fire is a turbulent diffusion flame resulting from the combustion of a fuel continuously

released with some significant momentum in a particular direction. Jet-fires represent a

significant element of risk associated with major incidents on offshore installations, with the

fuels ranging from light flammable gases to two-phase crude oil releases. Consequently to

their particularities and to its severity, jet-fire will be fully developed in a specific section.

A fireball (Figure 9) is a rapid turbulent combustion of fuel, usually in the form of a rising and

expanding radiant ball of flame. When a fire such as a pool or jet fire impinges on a vessel

containing pressure-liquefied gas, the pressure in the vessel rises and the vessel wall

weakens. This can eventually lead to catastrophic failure of the vessel with the release of the

entire inventory. This phenomenon is known as a boiling liquid expanding vapor explosion

(BLEVE). In such releases, the liquefied gas released to the atmosphere flashes due to the

sudden pressure drop. If the released material is flammable, it will ignite; in addition to

missile and blast hazards, there is thus a thermal radiation hazard from the fireball

produced.

Figure 9 : Flame geometry for an expanding, rising fireball [6]

It is this thermal radiation, which dominates in the near field. Although the duration of the

heat pulse from a fireball is typically of the order of 10–20 seconds, the damage potential is

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Page 38 Ph-D report J. Ciret

high due to the fireball's massive surface emissive power. Large-scale experiments carried

out by Roberts et al. [12] with propane as the fuel, measured a maximum average surface

emissive power ranging from 270 to 333kW/m

. The engulfment of structures in a fireball is

generally immediately catastrophic to life. The hazard consequences of fireballs that

determine their potential for harm are size, rise duration and radiation of the phenomena.

To conclude about fire types that can happen on offshore site, obviously these classes

overlap and offshore fire events might be a combination of these fire types. Consequently

risks and harms are again increased. In addition with the close proximity of personnel and

the difficulties of escape, hazard and unfortunately dramatic issues with human losses

cannot be eliminated to these industrial facilities.

I. 2. An overview of the nature of hydrocarbon jet-fire

Jet-fire has its own specific properties:

“A Jet-fire is a turbulent diffusion flame resulting from the combustion of a fuel

continuously released with some significant momentum in a particular range of

directions.”[13]

This is one of the most used definitions of jet-fire and it is focused on the continuity of the

gas flow, the important momentum and the direction of the impingement without

mentioning intensity, severity torching, high velocities or localised impact, in spite of the

strong association done with jet-fire in some publications [14].

Jet-fires represent a significant element of risk associated with major incidents on offshore

installations, with the fuels ranging from light flammable gases to two-phase crude oil

releases. This can lead to structural, storage vessel, and pipe-work failures, and can cause

further escalation of the event (i.e. domino effect). A recent study by Gómez-Mares et al.

[15] reports that approximately 50% of the recorded cases of jet-fires are followed by

additional severe events. This is an extremely important factor because it means that the

scale of an accident may increase considerably when a jet-fire occurs. For example, the riser

failure which occurred during the Piper Alpha Incident in 1988 [16], contributed to the

collapse of the platform and a major loss of life.

Prior to the Piper Alpha incident, there was little public domain information concerning the

quantification of fire hazards possible on offshore platforms and the incident led to the

project of a technical report coordinated by the Steel Construction Institute and by UK

Health and Safety Executive [13]. This report contains an appraisal of the knowledge and

ability to predict the hazards consequence of a jet-fire. It permits also to determine the

physical characteristics and the key feature of such phenomenon.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Ph-D report J. Ciret Page 39

I. 2. a) Jet-fire length

There is a considerable body of literature on the geometry of jet flames [17-24]. The

geometry of a jet-fire is important in predicting radiation from the flame and the possibility

of flame impingement on nearby facilities. The most important geometric parameter is the

flame length. The flame length is the distance from the release point to the defined end of

the flame. Whatever the scale is, the definition is not modified and its determination is easily

available. An internet site [25] allows even to calculate it based on several parameters and

using a sophisticated formula commented below.

But its precision is linked with the circumstances taken into account the release conditions

and jet-fire size can be primarily related to the heat released. For gaseous releases, this is

related to the size of the leak (hole diameter) and the pressure (which may vary as a function

of time). For non-gaseous fuels, the liquid content results in relatively higher release rates

for a given aperture and pressure compared to gaseous releases and, when the release is

two-phase (such as may arise from a relatively long pipe connected to a storage vessel

containing a liquid above its boiling point), estimates the release rate is non-trivial.

QAL *

(Eq.1)

Where L is the flame length, in m and Q is the net heat released by combustion, in MW. A and b are

constant

A first correlation (Eq. 1) can be used to present reasonable predictions in spite of its

simplicity. Several studies [18, 19, 26] have optimized it and adapted it, by changing A and b

values, to represent sonic and subsonic release for a range of gaseous fuels and a large range

of stack diameters between 500 and 800mm. For example Figure 10 shows jet-fire lengths

for a range of fuels plotted against the net heat released by combustion. It exhibits good fit

with empirical values of A and b.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Page 40 Ph-D report J. Ciret

Figure 10 : Jet-fire lengths versus heat released for different hydrocarbons [27]

Other correlations more sophisticated and asking more information not easily available have

been also developed. The well known model of Hawthorne et al. [22] (Eq.1) is for instance

used to predict the distance, L, from the orifice to the visible flame tip on the internet

website mentioned above. Hawthorne et al. [22] take the turbulent mixing and buoyancy

processes at the end of the flame into account but in the other hand neglect the jet exit

conditions contrary to previous formula. So the ratio of moles of reactants to products and

the stoichiometry of the gas mixing seem to play a role on the length of the flame and the

latter is thus measured as a distance where mixing has reduced the fuel concentration to

stoichiometric:

2/1

3.5

(Eq.2)

Where B is the flame lift-off distance, m

D is the source hole diameter,

is the molar concentration of fuel in the stoichiometric mixture

and T

are respectively the temperature of the flame and this of the vapour leaving the hole, in K

is the ratio of moles of reactants to products

is the ratio of the molecular weight of the surroundings atmosphere to that of the gas issuing

from the nozzle.

A new notion of flame distance is also introduced here; it is the lift-off distance. Indeed jet

flames are generally lifted i.e. the base of the flame is not attached to the release point

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Ph-D report J. Ciret Page 41

(Figure 11). This occurs because of the high velocities, strain rates and the richness of the

fuel near the source make it difficult to maintain a flame. The actual lift off position is

defined either as the point at which a blue gaseous flame appears or the point further

downstream at which yellow sooty flame first appears.

Figure 11: Flame and lift-off diagrams

These correlations give us information about the length of a perfect flame issued of the

ignition of a gas release. However, some limitations must be underlined. Indeed ignition of

discharges of fluids in open surroundings does not necessarily produce stable jet flames. The

discharge conditions can be such that flames cannot burn back to form a steady flame.

Whether or not a stable jet-fire will arise following the releases of a pressurized hydrocarbon

will depend principally upon the nature of the fuel, the size of the hole from which the

release occurs and the geometry of the surroundings. In practice [27] this means that most

small leaks will be inherently unstable and will not support a flame without some form of

flame stabilization, such as the presence of another fire in the vicinity or any impact onto a

pipe-work, vessels or the surrounding structure. However, in the highly congested

environment offshore, impact within a short distance is very likely, and hence small leaks are

most likely to stabilize on the nearest point of impact. Apart from providing flame

stabilization, impact onto an obstacle may modify the shape of the jet-fire. Objects which

are smaller than the flame half-width at the point of impact are unlikely to modify the shape

or length of the flame to any great extent. However, impact onto a large vessel may

significantly shorten the jet-fire, and impact onto a wall or roof could transform the jet into a

radial wall jet where the location and direction of the fire is determined by the surface onto

which it impacts and its distance from the release point.

Other limitations of the flame length determination are the influence of wind for external

jet-fire. Indeed, in a crosswind, the flame lengths shorten highly due to an additional mixing

and buoyancy. These effects are not as well identified and characterized as in still air, but

empirically flames from subsonic gaseous releases are quite wind and buoyancy affected

beyond the early momentum dominated region of the flame. Air entrainment in low velocity

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Page 42 Ph-D report J. Ciret

jet-fire is not particularly efficient; they are therefore comparatively long, sooty and radiant.

Some studies include these effects by adding a factor inside the expressions [28] (Eq.3):

49.051.0

4.0

eLL

(Eq.3)

Where L is the wind-affected flame length measured in m in a straight line from the release point to

the flame tip

is the still air flame length, in m

V is the wind speed in m.s

-1

Therefore introducing as the most important parameter of the jet-fire, the flame length

appears as a characteristics widely influenced obviously by the size of fuel leakages and also

by the surroundings. If the first effect is directly link with the accident degree (a wider source

hole cause a high release of fuel and thus a large powerful flame), the other parameters

highlight once again the high risks occurring on offshore platforms, due to the highly

congested environment offshore that reinforce the flame and the random wind effect.

I. 2. b) Radiative and convective heat contribution

The thermal load to an engulfed object in a jet-fire will be a combination of radiative load

and convective load.

The radiative behaviour of jet flames is very varied and complex [13]. Radiation from a jet

flame arises from hot soot particles and gases. In most of the offshore jet flames, radiation

from soot dominates. However, in many sonic natural gas flames the soot radiation will be

quite small compared to the molecular emission from CO

and H

O vapour. The

directionality of jet-fires provides the possibility of direct flame impingement onto

surroundings structures or targets. So this is a supplementary source of serious hazard to

offshore facilities. In literature, we find several studies about nature and extent of

impingements and the resulting heat fluxes. But as for radiation, a problem appears with the

soot radiation, the molecular radiation and their ratio. Clearly, the total heat flux which is

imparted to an engulfed object will vary over the surface of the object. In addition, the

relative proportions of convective and radiative heat flux will vary over the surface, with the

highest convective component likely to be experienced close to the point of impact of a

flame where the highest velocities occur, whereas the highest radiative heat load will be

experienced where the more radiative part of the flame (usually towards the end of the

flame) is viewed by the object.

Figure 12 shows total heat fluxes experienced by a horizontal cylinder (pipe) impacted by a

horizontal high pressure gas jet-fire [29]. The cylindrical surface is presented flat by cutting

along the rear. Several configurations can be observed:

- (a) When the pipe was located at 21m (towards the end of the flame) the maximum

heat fluxes were experienced at the point of impact on the front of the pipe.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Ph-D report J. Ciret Page 43

- (b) At 15m the heat loads were relatively uniform around the pipe.

- (c) But at 9m, the heat loads were greatest to the rear of the pipe due to radiation

from the tail of the flame.

Figure 12 : Variation of total heat flux over the surface of a pipe according to distance from the

flame source: (a) pipe at 21m; (b) pipe at 15m; (c) pipe at 9m [27]

As the more radiative part of the flame is closer to the tail, this can result in the highest

overall heat fluxes being experienced on the rear surface of an engulfed object which may

seem counter-intuitive. Other tests with subsonic methane jet flames [30] and subsonic [30]

and sonic [31] gaseous propane flames at different flow rates and exit velocities have

presented similar results and allow to conclude to the non-local impingement of the jet-fire,

that increase again the harm of this phenomena. Indeed, temperature measurements [32]

along a pipeline positioned close to the source show that the rear of the tubular target

received the greater incident flux as evidenced by the greater temperatures, shown in the

Figure 13.

Figure 13 : Temperatures versus time at different locations of a pipe engulfed by jet-fire [13]

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Page 44 Ph-D report J. Ciret

As noted above, the combustion process within a natural gas jet-fire is relatively efficient

and produces little soot (carbon). Consequently, these flames are not as luminous as higher

hydrocarbon flames. The net result is that the radiative heat transfer to the surroundings is

lower than for comparable higher hydrocarbon jet-fires. This is reflected in the fraction of

heat radiated, F, for such fires as can be seen in Figure 14 (F is defined as ratio between

energy released as radiation from the flame surface and net energy of combustion). As can

be seen, F increases with carbon number (from natural gas to crude oil), reflecting the

increased radiative emissions from soot within the higher hydrocarbon jet-fires.

Figure 14 : Fraction of radiative power versus total power observed for jet-fire with different

hydrocarbons [27]

Due to the radiant soot emissions, the radiative heat transfer from higher hydrocarbon jet-

fires is generally greater than that from natural gas flames and the generally lower velocities

arising from flashing liquid releases (such as propane or butane) result in a lower convective

flux to engulfed objects. For the rear surface of an engulfed object, Figure 15 shows that the

fraction of the heat flux which is radiative increases from about 0.5 for natural gas to about

0.8 for fuels containing a large proportion of higher hydrocarbons [33-36].

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Ph-D report J. Ciret Page 45

Figure 15 : Radiative heat flux as a fraction of total heat flux versus hydrocarbon content for an

engulfed object for a jet-fire [27]

Different authors report also several correlations between F factor and the jet exit velocity

[19, 28, 37]. In the case of a pressurized gas–liquid mixture (such as ‘live’ crude oil), the high

gas velocities may still occur and result in a high convective contribution, whilst the higher

hydrocarbon content maintains a high radiative contribution; making these type of jet-fires a

‘worst case’ in terms of total heat flux to engulfed obstacles [27].

Finally, the next section describes the high fire risk assessment on offshore platforms.

Increased by the locations, fire risk on offshore is mainly caused by an accident and lead

always to materials collapse and always to human losses. The dramatic Piper Alpha accident

has a strong influence on this assessment and great efforts have been made to draw up this

hazard identification. Hydrocarbon leaks and subsequent fire and/or explosions are the main

threats to offshore explorations and production of hydrocarbon oil and gas, with fire kinds as

pool fires, fireballs, or cloud fires. But the jet-fire represents the worst fire risk in offshore

installations. In a jet-fire assessment, the two important parameters are the flame

dimensions, the thermal radiation heat flux and velocities field around the flame. The

convective heat transfer rate can be very high increasing by velocities and impingement and

leading to rapid failure of objects engulfed by the flame envelope. Thermal radiation outside

the flame envelope can also lead to equipment failure caused by rapid increase of heat flux.

Therefore, due to these severe conditions occurring on offshore platforms, solutions must

be provided not to avoid unavoidable accident but to reduce consequences of fire event and

safe structure and obviously people.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Page 46 Ph-D report J. Ciret

II. Key requirements for offshore

protection

Several ways exist for the fire protection of steel. There are two categories of prevention and

control of fires; passive protection and active control and protection. Active 'protection'

consists of several systems that may require human intervention to initiate. These include

water deluge, foam systems, sprinklers, monitors, inert systems, fire extinguishers. Passive

fire protection (PFP) is defined as “a coating, cladding or free-standing system which, in the

event of a fire, will provide thermal protection to restrict the rate at which heat is

transmitted to the object or area being protected” [38].

In most cases passive fire protection materials exhibit successful results when they are used

in conjunction with “active” systems. Passive fire protection is the primary measure

integrated within the constructional fabric of a structural steel to raise the fire resistance of

the structure and to maintain the fundamental requirements of structural stability. However

their use on offshore platforms requires additional severe properties that must be kept in

mind.

Therefore, after a global description of the different passive systems in a first part of this

section, we will detail in a second one the key parameters required by intumescent coatings

based on epoxy resin to comply with the jet-fire resistance since this kind of PFP is widely

used on offshore platforms. Then in a third part we will introduce the different methods

existing to evaluate their performance and their efficiency for maintaining steel integrity and

for reducing damage on offshore installations.

II. 1. Passive fire protection

The main property of passive fire protection systems used for steel is the ability to maintain

the steel structure at temperature below 550°C, which is considered as the critical

temperature for structural steel, above which, it is in jeopardy of losing its strength, leading

to collapse.

Different methods can be used to improve fire protection of steel and the main classes of

passive fireproofing materials used are cementitious products, fibrous materials, composites

and intumescent materials. Early in the previous century [39], encasements in concrete,

brick, clay tiles, lath and plaster, and similar materials were commonly used for this purpose.

Now, less expensive forms of protection have been developed, such as cementitious

coatings and sprayed of mineral fibres, which can be applied directly on the steel members

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Ph-D report J. Ciret Page 47

[40]. Furthermore intumescent paints are available in Europe for over 25 years, although

these materials have been known for centuries [41].

Cementitious products include concrete, gunite, lightweight vermiculite based mixes,

gypsum, calcium silicate and magnesium oxychloride. All cementitious fireproofing materials

protect the steel in two ways. Firstly, they contain trapped moisture which is released upon

heating and keep the steel temperature around 100°C until all the water is evolved.

Secondly, the product acts as a thermal insulator [42]. Moreover, these materials are

essentially inorganic and therefore do not burn. No additional smoke or toxic fumes are

produced in a fire which makes them suitable for internal use in living quarters and work

areas [43].

Advantages of these materials are that they have an attractive appearance, are usually hard

and durable, relatively cheap, and easy to install or repair. The cementitious materials can be

sprayed (Gunite, vermiculite-cement, magnesium oxychloride, etc.) or supplied as panels

(Gypsum, plaster, calcium silicate, etc.), which are fixed to the structure with either steel

wire or nailed to a timber cradle [44].

However, although they offer reasonable protection, the traditional cementitious based

materials have been found in a number of instances not to give satisfactory performance in

the highly corrosive offshore environment and, consequently, because of this and also

because of its weight, have been largely replaced by epoxy intumescent systems suitable for

hydrocarbon jet-fires [45].

Fibrous materials include boards and blankets of mineral wool and ceramic fibres. They are

particularly used as passive fireproofing systems when thermal insulation is an additional

requirement. Inorganic binders that do not burn out during the initial stages of the fire are

recommended [46]. Mineral wool is commercialized to resist to 850°C and ceramic fibre to

1150°C [47]. As ceramic fibre is more expensive than mineral wool they are often used in

combination. Mineral wool fibres can be sprayed with an adhesive to provide protection to

structural steelwork. Mineral Wool insulation and fire protection products can be used in all

types of buildings. It is also used in critical applications in offshore oilrigs and petrochemical

refineries as it protects against hydrocarbon fires [48]. The main disadvantage is that fibrous

materials absorb water easily, they are therefore recommended for internal use only except

where they can be effectively clad with metal sheeting. Some of these materials have also

recently been shown to be having human health and safety implications, so their use is

increasingly restricted.

Composite fire protection panels are produced with various types of materials from

different fire resistant classes. Panels may consist of a metallic cladding, often stainless steel,

a cementitious board (typically plaster or gypsum) and mineral or ceramic fibres in between.

These panels may resist cellulosic or hydrocarbon fires as required. They are generally fixed

to structural members with steel binding wire or bolted to a supported frame or cradle.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Page 48 Ph-D report J. Ciret

These three first technologies are currently used to protect structural steel from fire and

heat. However, in this dissertation the building that should be protected is offshore

platforms, sensitive to very aggressive conditions, namely corrosion, chemical attacks and

potential powerful fires. Therefore, although traditional cementitious fireproofing materials

are known for their low cost and ease of installation, they do not resist to corrosion and

aggressive weathering conditions. They do neither resist to jet-fires, which is an essential

requirement in the petrochemical industry and cannot be used for protecting offshore

platforms [45]. Then similar disadvantages can be drawn about the fibrous materials, in

addition with the sensitivity to water and their use on offshore platforms is not extended.

On the other hand, intumescent materials are widely used in passive fire protection

systems, where potential pathways for the spread of flames and fumes, such as conduits,

wall openings and ventilation grilles can be protected with fire retardant products. An

intumescent material is one that undergoes a chemical change when exposed to heat or

flames, becoming viscous then forming expanding bubbles that harden into a dense, heat

insulating and multi-cellular char. Consequently, as shown by an increasing market supplied

by several companies, intumescent systems are the current better technology for the

protection of offshore platforms. Intumescent coatings are generally epoxy based and are

delivered to the site in drums to be sprayed or hand applied [42]. Intumescent epoxy

materials have been recognised as the best means of saving lives and limiting damage when

hydrocarbon fires hit offshore platforms [49]. Therefore the rest of this state of art deals

with the study of different commercial thick intumescent coatings in terms of thermal,

chemical or physical properties.

II. 2. Epoxy based intumescent coatings

The increasing acceptance of the benefits of epoxy intumescent fireproofing materials has

resulted in much increased uses, particularly offshore, of this type of product in recent years

[50].

Epoxies are thermosetting materials and consist in a two-component mixture composed of a

prepolymer and a hardener. Thermosets are, due to the irreversible curing reaction, rigid

three dimensional cross-linked plastics [51]. Due to their two components implementation

and the properties of the cross-linked structure [52], epoxy-based thermosets are choice

elements for varying the characteristics of the end products following a quasi-infinite variety

of recipes [53]. Globally, after curing, epoxies have satisfactory intrinsic mechanical

properties such as suitable weather, chemical and thermal resistance [54]. In addition,

different methods have been studied to modify their network structure and to improve

these properties. One of the most important parameters controlling the network structure is

the crosslink density and could be determined by differential scanning calorimetry and

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Ph-D report J. Ciret Page 49

modified by changing the prepolymer chain length [52]. Furthermore, thanks to an

appropriate selection of formulation chemistry (i.e. not only epoxy prepolymer and hardener

but also catalysts, additives, impact modifiers, etc.), a wide range of properties and

behaviours, required for numerous applications, can be provided to epoxy resins [55]. So,

the many applications of epoxy resins have focused attention on these materials since the

beginning of their commercial use by the end of the 1930’s [56]. Epoxies are applied for

moulding compounds, surface coating and painting materials, composites, microelectronic

encapsulated materials, printed circuit boards, adhesives, etc [53, 54]. Epoxy resins are used

in many industrial applications in which stiff, high glass transition temperature, low creep

and high heat resistant materials are required [57]. All these technical advantages justify

their higher cost in comparison with other thermosetting materials [58].

The versatility of these resins is due to the instability of the epoxy group. Thanks to its great

nucleophilicity, this ring can be opened by reacting with different compounds like

polysulfides, polyamides, aromatic and aliphatic amines and anhydrides [56, 59]. Groβ et al.

[60] have shown that the nucleophilicity of the functional groups is the determining

parameter in curing of epoxy resins in order to provide different mechanical, chemical and

physical properties. The presence of polar groups on the polymer backbone allows good

adhesion to metal substrates and the cured epoxy resin shows high resistance to abrasion,

moisture, thermal shock and vibration damage. High tensile and compressive strengths can

be obtained and weather resistance is excellent, even if they show a poor resistance to UV

exposure.

There are two main categories of epoxy resins, namely the glycidyl epoxy, and non-glycidyl

epoxy resins. The glycidyl epoxies are further classified as glycidyl-ether, glycidyl-ester and

glycidyl-amine. They are prepared by polycondensation between a dihydroxy compound, a

dibasic acid or a diamine and epichlorhydrin. Glycidyl-ether epoxies such as diglycidyl ether

of Bisphenol A (DGEBA) are the most commonly used epoxies [51]. The non-glycidyl epoxies

are either aliphatic or cycloaliphatic epoxy resins, and are formed by peroxidation.

The most studied coatings use a diglycidylether of Bisphenol A (DGEBA). It is synthesized by

reaction between bisphenol A and epichlorohydrin (Figure 16). DGEBA provides several

properties to the epoxy coating according to the polymerization degree [51], namely

adhesive properties, corrosion of hydrolysis sites [61].

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Page 50 Ph-D report J. Ciret

Figure 16 : Reaction and formula of DGEBA [62]

In order to convert epoxy resins from DGEBA to a hard and rigid network, a hardener or

curing agent is necessary. Amines are the most commonly used curing agents for epoxy cure

[63]. Primary and secondary amines are highly reactive with epoxies. Tertiary amines are

generally used as catalysts, commonly known as accelerators for the cure reaction.

Otherwise polyaminoamide can also been used as hardener.

The curing process is a chemical reaction in which the epoxy groups react with a secondary

amine to form a hydroxyl group and to reduce the amine to a primary one (Figure 17).

Figure 17 : Reduction of the primary amine and opening of the epoxy ring

Then in a second step the reaction continues with the reaction of the amine with another

epoxy group. So a three dimensional network is formed, containing ternary amine and

framework issued of DGEBA (Figure 18, Figure 19).

Figure 18 : Formation of a 3D-network

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Ph-D report J. Ciret Page 51

Figure 19 : Schematic representation of the reticulated network between epoxy resin and hardener

However, this hard material does not exhibit fire retardant properties and need additives

and/or chemical modifications in order to develop an acceptable reaction to fire [54]. Indeed

epoxy resins are used in many industrial applications, but flammability of this resin type is a

major limitation in areas requiring high flame retardance. The approach chosen to improve

the flame retardance of composites has been to add flame retardants and smoke

suppressants to the resin [64-66]. However, the addition of some flame retardants could

significantly decrease the mechanical properties of polymers [67-69] and must consequently

be used with care in order to maintain with respectable ratio intrinsic properties of epoxy

resin.

Intumescent coating acts as a thermal barrier for construction materials during fire hazards.

The coating has the ability to expand by a factor of 40 or more during heat exposure that

effectively leads to the protection of the substrate against rapid increase of temperature,

thereby maintaining the structural integrity of the building. Most of the published

information on intumescent coatings is in the patent literature while little is reported on the

chemical-physical mechanism of intumescence [70-74]. The first intumescent coating has

been patented in 1938 [75].

The primary focus of research on intumescent coatings is to identify the combination of

ingredients that results in a coating that develops a controlled volume of cohesive and

insulative char when the system is exposed to a fire (Figure 20) and a system that also

provides all the protection and other properties required of a high quality coating.

Figure 20 :Formation of intumescent char

In 1970, Vandersall [76] published a detailed paper describing the usual and efficient

elements of intumescent coatings. He stated “A quality intumescent paint cannot be visually

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Page 52 Ph-D report J. Ciret

differentiated from a conventional paint. It provides the protective, serviceable and

aesthetic properties of the non fire-retardant system. However, when heated above its

critical temperature, the film begins to melt, to bubble and to swell, forming a thick, non-

flammable, multi-cellular insulative barrier which affords protection for the substrate.” So

the goal of intumescent additives is to provide fire protection without altering other

properties.

Vandersall [76] classified the chemical compounds of intumescent systems in three

categories:

- A carbonisation agent, this means a carbon-rich polyhydric compound: starch and

polyhydric alcohols are mostly used. The number of carbons will influence the

amount of char formed whereas the number of hydroxyls will determine the rate of

char formation. In the studied intumescent epoxy coating, the polymer itself acts as

the carbonific agent.

- An acid source: an inorganic acid, either free or formed in situ from a precursor

during heating.

- A foaming agent: usually halogenated or nitrogenated compounds, which also evolve

copious quantities of non-flammable gases.

Similarly, Jones et al. [77] defined as carbonific the compounds in intumescent formulations

that act as a carbon source for char formation and spumific those which evolve gaseous

products and induce foaming.

Numerous papers have tried to describe the possible mechanism leading to intumescence

[74, 76, 78, 79]. It is generally accepted that first, the acid source breaks down to yield a

mineral acid. This takes part in the dehydration of the carbonisation agent to yield the

carbon char and finally, the blowing agent decomposes to yield gaseous products. The latter

causes the char to swell and hence provides an insulating multi-cellular protective layer. This

shield limits the heat transfer from the substrate to the heat source, resulting in a

conservation of the underlying material. The presence of a compound in each of the above

three classes does not by itself ensure intumescent behaviour of the mixture. In fact, a series

of chemical and physical processes must occur in an appropriate sequence, while the

temperature increases, to produce the intumescence phenomenon.

Examples of the components from the different categories are shown in Table 2.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Ph-D report J. Ciret Page 53

Table 2 : Examples of components playing a role in intumescent coatings [79, 80]

Inorganic acid source

Carbonisation compounds

- Acids (phosphoric, boric, sulphuric)

- Ammonium salts

(Phosphate, polyphosphates, sulphates,

halides, etc.)

- Amine/amide phosphates

(Urea, Guanylurea, Melamine phosphate,

etc.)

- Organophosphorus compounds

(Tricresyl phosphate, alkyl phosphate,

haloalkyl phosphate).

Starch

Dextrin

Sorbitol

Pentaerythritol

Polymers

Phenol-formaldehyde resins

Methylol melamine

Foaming compounds

Amines/amides

Chlorinated paraffins

Tetrachlorophtalic resins

Melamine

Urea

Urea-formaldehyde resins

Dicyandiamide

Melamine

Polyamides

Classical and efficient intumescent formulations use ammonium polyphosphate as acid

source and foaming agent and pentaerythritol as carbon source [81-83]. Used on epoxy resin

based coating at different ratios, this fire retardant mixture alters degradation

characteristics of the materials by increasing the residual mass and the char yield [84, 85].

The incorporation of the APP/PER system in the resin provides also the convenient

mechanical and thermal properties to the protective shield: high flexibility and heat diffusion

abilities [86]. Its mechanism of degradation is widely studied and now well known. At 200°C,

APP reacts with PER to yield ortho- and pyro- phosphate species, via a hydrolysis reaction

(Figure 21). At this step, the first P-O-C bonds have been formed; the key of the efficiency of

the coating will be to keep these bonds, providing beneficial properties to the insulative

structure.

Figure 21 : Hydrolysis of ammonium polyphosphate

Then the phosphor ester formed can undergo a ring closing esterification, in which water

and ammonia is released [87]. A succession of these and similar reactions results in a char

consisting of mainly carbon (Figure 22), but also small amounts of oxygen, phosphorus, and

nitrogen atoms [70]. Besides, the binder takes part in the control of the char expansion and

ensures a uniform foamed structure [88].

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Page 54 Ph-D report J. Ciret

Figure 22 : Ring closing esterification of the phosphor ester and subsequent reduction to a carbon

network

The formation of the effective char occurs via a semi-liquid phase, which coincides with gas

formation and expansion of the surface [89]. Gases released from the degradation of the

intumescent material, and in particular of the blowing agent, have to be trapped and to

diffuse slowly in the highly viscous melt degraded material in order to create a layer with

appropriate morphological properties as shown in Figure 23.

Figure 23 : Development of intumescence (α= conversion degree)

The viscosity of the degraded matrix in the blowing phase is, as a consequence, a critical

factor [90, 91] and the different components of the intumescent formulation can modify this

visco-elastic behaviour. If the degraded matrix has a too low viscosity, easy diffusion of gases

takes place and the gases will not be trapped but rather escape to feed the flame. If the

viscosity is too high, there can be formation of some cracks which allow the oxygen to

diffuse into the char and which could increase the heat transfer from surface to substrate.

Hence the efficiency of the protective layer may decrease. Moreover, several previous

studies have shown good correlation between the different steps of the intumescent

process and the dynamic properties (expansion and apparent complex viscosity) of the

formed char using thermal scanning rheometer measurements on different intumescent

materials [90, 92-94]. Applied on intumescent coatings based on epoxy resin, Jimenez [95,

96] showed that viscosity and swelling are linked as the expansion takes place because of

the decrease in the viscosity of the degrading material combined with the release of volatile

degradation products. In this study, Jimenez showed also the modification and the

synergistic effect provided by the different intumescent additives on the visco-elastic

properties and the swelling abilities of the materials. The degradation of the thermoset resin

at about 350°C creates an important decrease of viscosity detrimental to development of an

efficient intumescent shield. Boric acid and APP seems to be the key additives to reduce this

fall of viscosity and to allow a high expansion and a char with a good mechanical resistance.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Ph-D report J. Ciret Page 55

Furthermore, another important parameter that also has to be taken into account is the

mechanical strength of the intumescent char. This factor is significant because in the

conditions of fire, char destruction can proceed not only by means of ablation and

heterogeneous surface burning but also by means of an external influence such as wind,

mechanical action of the fire or convective air flows. This aspect appears as a key parameter

especially after describing the jet-fire phenomenon in the previous section and knowing the

location where our intumescent formulations are applied. A consideration of the mechanical

stability of intumescent char appeared first in the technical literature in discussions about

the nature of bonds which provide char stability [70].

A simple criterion [97, 98] was subsequently proposed for the mechanical stability of the

intumescent char: mechanical properties of foams depend both on the structure, porosity

and mechanical properties of the foam material; it is assumed that chemical structure is one

of the main factors affecting mechanical stability. Moreover, physical structure data (such as

thickness of the walls inside the char cap, micro-density of the char, density of the char

material) should also been taken into consideration. Finally, mechanical stability probably

depends on the direction of external perturbation and its type (direct, rotation, vibration).

The question for intumescent chars is why and how this perturbation is distributed through

the char cap, from the surface layer and up to “frozen” pyrolysis zone. This distribution

depends on the char porosity and, to a less extent, on the chemical structure.

Reshetnikov et al. used a “Structurometer ST-1”, developed at the Moscow State Food

Academy, to measure the force required to destroy the char [98]. The samples were first

pyrolysed and then a destructive force was applied to the sample. It was concluded that one

of the main factors influencing the mechanical stability is the char porosity: the smaller the

pore size the better the char strength. A similar method was developed in our laboratory at

Lille University [93] in recent years, using the Thermal Scanning Rheometer. The main

interest of our approach is that the measurement of the destruction force is made directly

after swelling of the sample, at high temperature, working as an “in situ” experiment. This is

particularly important in the case of jet fires, as the char has to resist very aggressive

conditions, such as explosions, violent wind.

As a summary based on a careful review of the literature, we may establish parameters

required by the intumescent structure to provide an efficient protection:

- The flame retardancy of modified polymeric materials does not only depend on the

thermal stability of the constituent elements, but also on their rate of degradation,

rate of char formation and the amount of residual char [99].

- The performance is determined by the kinetics of the charring process and the exact

timing of the expansion of the char. If it takes place too early after dehydration, the

char is ineffective and non-insulative. If the expansion occurs too late, the char

hardens before swelling [93].

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Page 56 Ph-D report J. Ciret

- The efficiency of intumescent systems depends on the thermo-physical properties of

foamed char [100].

- The efficiency of the char is closely related to its ability to expand and form a multi-

cellular structure and consequently to its physical properties during the intumescent

process. One of the main aspects of intumescent char physical structure is its

uniformity and its porosity [88].

- A high mechanical stability promoting by a highly cross-linked structure is also

required for the foamed char to resist to fire conditions [97]. This last property is

even considered as the most important because of the aggressive conditions

occurring for a jet-fire event.

To conclude, this section has described the main fire retardants used in epoxy resin focusing

on the intumescent systems which are the most appropriate to provide fire protection in

offshore platform applications. The selection of intumescent flame retardant for a material is

then governed by the chemical and physical properties of the whole material and its

degradation characteristics. We have introduced APP and PER as convenient intumescent

additives for epoxy resin and drawn up the main properties required to such system. These

materials introduced above are materials designed to protect offshore platforms and their

properties are thus developed in agreement with focus on the mechanical resistance and the

thermal insulation. Indeed, fire events possible on offshore platforms can be described as

powerful fire and jet-fire resistance requires high stability and resistance facing to

impingement. Consequently, the physical properties introduced for intumescent coatings

must be considered above all as fundamental for fire protection of offshore platforms and

will be developed in a following chapter.

II. 3. Fire tests specified for offshore fire protection

II. 3. a) Fire tests based on furnace conditions

There are no internationally accepted rules for fire safety of offshore constructions. Instead

there are many national and corporation standards. Required levels of protection are

normally specified in terms of time and temperature on the basis of one or more criteria,

which may include statutory requirements, design considerations and insurance cost

implications. The duration is established by a time rating which is determined by testing in

accordance with an approved standard specifying a time / temperature curve (Figure 24).

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Ph-D report J. Ciret Page 57

Figure 24 : Standard fire test curves used for materials applied on offshore platforms

The hydrocarbon test curve (UL 1709) [101] is meant to duplicate or to be indicative of the

rapid temperature rise seen when a hydrocarbon fuel such as oil or natural gas burns: the

temperature rises rapidly to 900°C within 4 minutes and significantly higher overall

temperatures are reached (between 1100°C and 1200°C). This hydrocarbon fire test curve,

developed by the Mobil Oil Company in the early 1970’s and adopted by a number of

organisations and in particular Underwriters laboratories, UK Dept. Of Energy, BSI, ISO and

the Norwegian Petroleum Directorate is now a common test method for high risk

environments such as petrochemical complexes and offshore platforms.

Derived from the above-mentioned hydrocarbon curve, the French regulators asked for a

modified version, the so called hydrocarbon modified curve (HCM). The maximum

temperature of the HCM curve is 1300°C instead of the 1100°C for standard hydrocarbon

curve. However, the temperature gradient in the first few minutes of the HCM fire is as

severe as all hydrocarbon based fires. This test has been developed to consider jet-fire

scenarios in which leaking high pressure hydrocarbon gases ignite to produce intense,

erosive jet flames that can reach speeds of 150 metres per second.

The RWS (Rijkst Water Staat) curve [102] was developed by the Ministry of Transport in the

Netherlands. This curve is based on the assumption that in a worst case scenario, a fuel, oil

or petrol tanker fire with a fire load of 300MW could occur, lasting up to 120 minutes. The

RWS curve was based on the results of testing carried out by TNO Centre for Fire Research in

the Netherlands in 1979. The difference between the RWS and the hydrocarbon curve is that

the latter is based on the temperatures that are expected from a fire occurring within a

relatively open space, where some dissipation of the heat occurs, whereas the RWS curve is

based on temperature you find when a fire occurs in an enclosed area, such as a tunnel,

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Page 58 Ph-D report J. Ciret

where there is little or no chance of heat dissipating into the surrounding atmosphere. The

RWS curve simulates the initial rapid growth of a fire using a petroleum tanker as the source,

and the gradual drop in temperatures to be expected as the fuel load is burnt off.

The standards listed above are the most commonly used standards. However, it is generally

recognized that the conditions in the standard and hydrocarbon furnace tests do not

represent the characteristics such as heat flux and erosion found in full scale high velocities

jet-fires. After the explosion and fire on Piper Alpha, Lord Cullen [16] highlighted it is clearly

desirable that any fire test used must be realistic. Any information provided by the test must

be useful to the chemists/industrial about the probable behaviour of the fire barrier in real

hydrocarbon fire conditions. With increasing reliance being placed on passive fire protection

materials offshore and the realisation that in many areas jet-fires pose the most onerous

threat, the need for a jet-fire test is now widely accepted.

II. 3. b) Jet-fire resistance test

An initial project, denominated OTI 95634 [103] has been developed jointly by the UK Health

and Safety Executive and the Norwegian Petroleum Directorate for using predominantly on

offshore installations. This jet-fire test (Figure 25) was designed to determine how various

objects and passive fire protection materials perform when exposed to this fire scenario. The

test impinges a high-speed stream of ignited propane fuel onto a substrate. The propane is

delivered at a rate of 0.3kg/s and can consume 1 tonne of fuel per test. Since 2007, this

project is become an international standard [104] that determines the resistance to jet-fires

of passive fire protection materials.

The method provides an indication of how passive fire protection materials perform in a jet-

fire that may occur for example, in petrochemical installations. It aims to simulate the

thermal and mechanical loads imparted to passive fire protection material by large scale jet-

fires resulting from high-pressure releases of flammable gas, pressure liquefied gas or

flashing liquid fuels. Jet-fires give rise to high convective and radiative heat fluxes as well as

high erosive forces.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Ph-D report J. Ciret Page 59

Figure 25 : Jet-fire resistance test in progress on an intumescent coating

To generate both types of heat flux in sufficient quantity, a 0.3 kg/second sonic release of

gas is aimed into a hollow chamber, producing a fire ball with an extended tail. The flame

thickness is thereby increased and hence so is the heat radiated to the test specimen.

Propane is used as the fuel since it has a greater propensity to form soot than natural gas

and can therefore produce a flame of higher luminosity. High erosive forces are generated

by release of the sonic velocity gas jet 1 meter from specimen (bulkhead) surface. The

propane vapour issues from a tapered, converging nozzle of length 200mm, inlet diameter

52mm and outlet diameter 17.8mm (Figure 26).

Figure 26 : Jet nozzle used for jet-fire resistance test [103]

The nozzle is aimed horizontally and normal to the rear wall of the test specimen. The tip of

the nozzle is located 1.00m from the front surface of the test specimen and with its centre

on quarter of the height of the test specimen above the inner surface of the test specimen’s

base (Figure 27). The average heat flux is approximately 240kW/m² and the maximum heat

flux 300 kW/m². The heat fluxes are highest in the upper part of the chamber and lowest in

the corners and at the jet impact zone, as already mentioned in the previous section.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Page 60 Ph-D report J. Ciret

Figure 27 : Layout of test facility with steelwork test specimen used for evaluating protective fire

materials [103]

Four different versions (panel, planar steelwork, structural steelwork and tubular section

tests) of the test procedure can be used according to the materials under consideration. The

structural steelwork test specimen is constituted of an open fronted bow, with the addition

of a central flange (also named web, and note it is this name which appears on the different

figures), with passive fire protection as described on Figure 28. The structural steelwork test

should be used to represent the application of passive fire protection material to steelwork

with corners or edge features, for example “I” beams.

Figure 28 : Construction of structural steelwork test specimen with a central flange [103]

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Ph-D report J. Ciret Page 61

The test duration can be defined in either a specified period of exposure to the jet-fire or the

period required to reach a specified maximum temperature. The test specimen is fixed to

the rear box and the assembly is mounted on lightweight concrete blocks approximately 1.0

m from the ground. The test specimens are instrumented with type K thermocouples (Figure

29). They are fastened to the back of the rear wall by drilling and peeing or other equivalent

methods while sheathed ones are inserted into the central flange.

Figure 29 : Thermocouple positions for structural steelwork test specimen [103]

Little information is currently available about this standard because of the date of

publication. However some new passive fire protection materials has proven to be capable

of resisting to jet-fire for 90 (rating J-90) or 120 minutes (rating J-120) according to this

standard. For instance Chartek 7 is an epoxy intumescent fire protection system featuring

excellent jet-fire resistance for up to two hours [105], Benarx is also a patented jet and

hydrocarbon fire-rated epoxy fire protection material that is tested and certified up to two

hours [106].

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter I: State of the art

Page 62 Ph-D report J. Ciret

III. Conclusions

This current state of the art is based on the hazards onboard offshore platforms. Indeed, the

slightest incident on such locations can quickly become the worst accident. The nature of

their operation (extraction of volatile substances sometimes under extreme pressure in a

hostile environment) means risk, accidents, and tragedies occasionally occur following by

pool fires or fireball. But the worst scenario that can occur on offshore platforms is most

probably the jet-fire. A jet-fire is a turbulent diffusion flame resulting from the combustion

of a fuel continuously released with high momentum in a particular direction. The convective

heat transfer rate can be very high increasing by velocities and impingement and leading to

rapid failure of objects engulfed by the flame envelope. Thermal radiation outside the flame

envelope can also lead to equipment failure caused by rapid increase of heat flux. Such

events are cause of worst accident.

Therefore, new technology has been developed to be specially adapted to fire protection of

offshore platforms. If the oldest insulative methods such as cementitious products or fibrous

materials do not prove great efficiency in the aggressive offshore conditions, intumescent

epoxy coatings have proven to be efficient protection against hydrocarbon fires of offshore

oilrigs and petrochemical refineries. Thanks to fire retardant additives, we have so detailed

the intumescent process and the key parameters required to promote an efficient

protection. Ammonium polyphosphate combined with pentaerythritol appears as the most

used and the most efficient intumescent mixture that exhibits the satisfying properties with

a sufficient swelling rate, an ability to dissipate heat and a great mechanical stability facing

to severe conditions.

The efficiency of epoxy coating is then evaluated by different fire test. Firstly hydrocarbon

fire test are carried out in a furnace environment under time/temperature conditions

specified by relevant curves then evaluated by the jet-fire resistance test. Recently

standardized, this test focuses on the thermal and mechanical loads and provides an

indication of how passive fire protection materials perform in a jet-fire. It leads to ‘J’ ratings

for given length of time. However, a rapid survey of the open literature have shown that the

most of the papers reports comprehensive studies about the jet fire test itself but no work

report as far as we know, detailed studies about jet fire test on steel protected with

intumescent coating. Nevertheless, both these tests are essential since they give a good

simulation of a real fire, all our work in this study will so be based on the behaviours of the

intumescent coatings facing to jet-fire.

That‘s why in the next chapter, we will comment the results obtained by four different

intumescent coatings evaluated at the jet-fire resistance test based on the parameter

discussed above.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Ph-D report J. Ciret Page 63

Chapter II: Evaluation of

the efficiency of the

intumescent coatings

facing to jet-fire

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Ph-D report J. Ciret Page 65

The previous chapter has reviewed the different methods to protect steel framework and

notably offshore platforms. We have also focused on the standardized tests able to measure

the efficiency of an intumescent coating undergoing fire corresponding to different

scenarios, from hydrocarbon fires to most hazardous jet-fires. The purpose of this work is

intumescent coating protecting offshore platforms and the associated fire scenario is the jet-

fire. Usually thick intumescent coatings are firstly evaluated in large furnace test in

agreement with standard UL 1709 [101]. If they pass it, they are then evaluated by the jet-

fire resistance test of passive fire protection materials OTI 95634 [103] or ISO 22899 [104].

Furthermore, several properties of the char have been mentioned in the literature. The

previous state of art shows desirable properties from the point of view of flame retardant

performance: mechanical strength and integrity, adherence of the coating on the steel

plates, porosity and of course abilities to limit heat and mass transfer in order to protect the

substrate. The efficiency of an intumescent coating depends thus mainly on the quality of

the char produced during burning and degradation.

This chapter examines the fire behaviour of four intumescent coatings according to the OTI

95634 report [103]. In a first part, the characteristics of the four intumescent coatings will be

presented, highlighting the main active components responsible for the intumescent process

and whose action has detailed in the first chapter. Then details and settings of the jet-fire

resistance test will be commented according to the requirements and setup specified by the

standard. Finally, the results exhibited by the intumescent formulations will be fully

discussed.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Page 66 Ph-D report J. Ciret

I. Materials and experimental

techniques

I. 1. Materials

This entire dissertation is based on the study of four different epoxy-based intumescent

coatings. The whole formulations are divided into two parts:

- The part A containing the prepolymer and others components

- The part B containing the curing agent (an amide) and others components.

Both parts are stored separately, mixed just before the application of the coating on the

substrate at different ratio (Table 3) and cured at ambient temperature over a 48h period.

Table 3 : Ratio between the two parts used to prepare the intumescent coatings

Intumescent coating

Ratio Part A/Part B

IF-1

2.44/1

IF-2

2.49/1

IF-3

2.31/1

IF-4

2.31/1

Among these four formulations, some ingredients are common. For the four grades of

intumescent coatings, the prepolymer is the same: a DGEBA type liquid prepolymer DER 311

from Dow Chemicals while the curing agent is a polyaminoamide. In addition, fire retardant

additives play an important role; different grades of ammonium polyphosphate have been

used according to the formulations. The coated ammonium polyphosphate used in the IF-1

formulation is a blended mixture of ammonium polyphosphate (APP) (Figure 30), and tris-(2-

hydroxyethyl)isocyanurate (THEIC) (Figure 31) commercialized as IFR 36 from Clariant [105].

APP is an acid source yielding to phosphoric acid upon heating, reacting with THEIC and

epoxy and leading to the formation of an intumescent structure.

Figure 30 : Formula of APP, n>1000

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Ph-D report J. Ciret Page 67

Figure 31 : Formula of THEIC

On the three other formulations an uncoated ammonium polyphosphate (Figure 30), Exolit

AP422 from Clariant, is added as both acid source and blowing agent (because of the release

of ammonia) [76]. In addition, the carbon source is played respectively by the epoxy resin on

IF-2, by pentaerythritol (PER) added to epoxy resin on IF-3 and by dipentaerythritol (dPER)

added on epoxy resin on IF-4 (Figure 32). Note that epoxy resin can also play a role of char

former in IF-3 and IF-4 formulations.

Figure 32 : Formula of pentaerythritol (a) and dipentaerythritol (b)

Furthermore, classical additives used on paint formulations have been added: pigments,

surfactant, and mineral fillers but the exact formulations of the four coatings are not known.

However TiO

is used on IF-2, IF-3 and IF-4 and could play a role for the chemical pathway.

The main active ingredients of the intumescent formulations are reported in Table 4,

highlighting the main differences between the coatings.

Table 4 : Comparison between the active ingredients of 4 formulations

Intumescent

coating

Polymeric matrix

Acid source

Carbon source

Blowing

agent

IF-1

DGEBA prepolymer

+ polyaminoamide

Coated APP +

Boric acid

Epoxy resin + THEIC

from coated APP

APP

IF-2

DGEBA prepolymer

+ polyaminoamide

APP

Epoxy resin

APP

IF-3

DGEBA prepolymer

+ polyaminoamide

APP

Epoxy resin + PER

APP

IF-4

DGEBA prepolymer

+ polyaminoamide

APP

Epoxy resin + dPER

APP

In addition, the use of mesh in intumescent coating provides an additional reinforcement of

the barrier formed when burning. This is crucial when the coating is exposed to high

pressure during the test as in the case of the jet-fire resistance test. The efficiency of this

reinforcement has been demonstrated in the case of the most fragile coatings as IF-2 and IF-

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Page 68 Ph-D report J. Ciret

3 whereas its effect is not significant for a robust coating as IF-1 (see discussion in the next

sections). Mesh (Figure 33) comprises alternating strands of carbon and glass fibre, woven

together with polyester thread running in one direction. Each component is used with an

accurate objective:

- The polyester melts at 140°C leaving the strands of carbon and glass fibre free to

move as the intumescent shield expands.

- The glass fibres melt at about 400°C, it is used to make it easier to handle.

- Finally, the carbon fibres exhibit a high thermal stability and should be preserved

through a standard hydrocarbon fire test but, it cannot resist prolonged exposure to

the pressures involved in a jet-fire test.

Figure 33 : Structure of mesh with carbon and glass fibre

I. 2. Jet-fire resistance test

The different formulations labelled IF-1, IF-2, IF-3 and IF-4 have been evaluated in the

conditions of a jet-fire according to the recommendations established by the Health and

Safety Executive [103] and already described in the first chapter. This test permits to

determine properties of passive fire protection materials. It is designed to give an indication

of how well passive fire protection materials will perform in a jet-fire. In spite of smaller

dimensions than typical offshore or plant structures, the thermal and mechanical stresses

imparted to the passive fire protection materials by large-scale jet-fires resulting from high-

pressure releases of natural gas have been shown to be similar as those from the jet-fire

defined in this procedure.

The structural steelwork test specimen is constituted of an open fronted box divided into

two back faces with the addition of a central flange. To describe test panel, we will refer thus

to back faces (right and left hand sides) and to central flange. For testing passive fire

protection materials applied as coating, the structural steelwork specimens shall have

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Ph-D report J. Ciret Page 69

passive fire protection material applied directly to all inside surfaces according to the

standard specifications. However, in the case of IF-2 and IF-3, some tests have been carried

out on panel where only the central flange is coated by intumescent formulations. This

adaptation of the experimental setup has been motivated by the properties of the coatings

and the low mechanical resistance exhibited by these two formulations. Thanks to these

settings, we have focused our investigation in the zones where the pressure is maximal and

where the coatings should not resist and cause the test failure.

In the jet-fire test, the fuel is commercial propane, delivered as vapour without a liquid

fraction at a steady state rate of 0.30±0.05kg/s. The tip of the nozzle is located 1.00m from

the front surface of the test specimen and with its centre on quarter of the height of the test

specimen above the inner surface of the test specimen’s base.

Temperature measurements are carried out on specified positions thanks to type K

thermocouples. The thickness of the passive fire protection material shall be also measured

at each thermocouples position. To locate event on the test specimen, we refer thus to

thermocouple positions.

Figure 34 : Thermocouple positions for structural steelwork test specimen

Finally, the test duration is defined as the period required to reach the specified maximum

temperature chosen for metallic substrate (i.e. 400°C). The test is over when two

thermocouples have reached this temperature.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Page 70 Ph-D report J. Ciret

II. Large-scale jet-fire test

evaluation

In accordance with all these recommendations and instructions described in the previous

section, the different coatings have been tested according to the OTI jet-fire test. Results

must provide a reliable representation of the behaviour of different coatings facing to jet-

fire and give a certification for the coatings, useful for knowing its good application on

offshore platforms and for marketing commercializing the formulation as a jet-fire rated

intumescent paint.

Results exhibited by the four intumescent formulations described above appear as our

starting point to explain and to understand the behaviour of the formulations. It is expect to

get important information to detect their weaknesses in order to enhance their properties

and consequently their fire performance. As mentioned in the state of the art, we will focus

on determining governing parameters required for developing efficient coatings.

II. 1. Intumescent formulation 1

This part describes a jet-fire resistance test lasting 62 minutes on a structural steelwork test

specimen protected with the passive fire protection materials IF-1 with an average thickness

of 9.7mm.

During test, the flame engulfed the specimen making it difficult to see the coated surfaces

very clearly through the flame. Nevertheless, some remarks can be stated during the test:

- After 5 minutes from the start of the test, pieces of coating were seen to detach from

the surface of the specimen, particularly on the central flange in line with the release

point.

- After 10 minutes from the start of the test, cracks appeared in the coating on

different locations and pieces of coating became detached from the bottom surface

of the box.

- After 30 minutes from the start of the test, pieces of the coating were displaced from

the top left hand side of the back face.

- After 40 minutes from the start of the test, flames could be seen getting behind the

top surface coating on the right hand side of the box.

After cooling the test specimen, a first description of the char can be made (Figure 35). In

spite of high temperature, coating is not completely reacted all along the panel. Indeed, we

can distinguish two kinds of zones:

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Ph-D report J. Ciret Page 71

- A first one where the coating is completely reacted and where a protective char is

formed. This zone corresponds to hottest zone (one single thickness measurement is

available and reported in Figure 37 ; 0mm is reported in Figure 36 about non-reacted

coating thickness).

- A second zone where a deep layer remains non-expanded because of insulation and

low temperature (two thickness measurements are thus available, on the surface

corresponding to formed char and in deep corresponding to non-reacted coating).

In addition, there is a circular region of dark, sooty coating on the left hand side of the

flange, extending on to the flange. This circular region is centred approximately 400mm

above the bottom of the back box and has a radius of around 200mm. This zone is clearly

viewable in Figure 35 and corresponds to the region where the propane jet impinges on the

specimen.

Figure 35 : Pictures taken after jet-fire test of IF-1

To highlight the reaction of the coating and the char formation, we must detail the progress

of the development of the coating. Figure 36 reports the thickness of partially reacted

coating (here we mean the formation of an expanded char), where no char formation has

occurred, knowing that the initial thickness was around 10mm. The coating on the front and

sides of the top two thirds of the flange have completely reacted (0mm non-reacted coating

remaining); on the remaining third the coating has partially reacted with no char formation

(4mm). On the back face of the specimen, the coating has partially reacted, with between

20% and 80% of the original thickness remaining, apart from a small area on the bottom

right hand side which has fully reacted. The reaction of the coating can be directly link with

the surface temperature and we suggest that the flange is the hottest location while on the

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Page 72 Ph-D report J. Ciret

back face convective effect and flame circulation on the box may cold down the coating and

prevent its complete reaction.

Figure 36 : Thickness of non-reacted coating remaining at various locations after jet-fire test on IF-1

In correlation with the reaction of the coating, the char thicknesses are reported on Figure

37. It is noteworthy that char thickness is well uniform all along the back plate and on the

sides of the flange in spite of partially reaction: mean thickness is around 40mm. However,

these different measurements allow to observe the destruction of the protective

intumescent layer on the front of the central flange. Indeed a low thickness is detected on

the middle of the central flange front (6mm) that can be caused by the flame erosion and

the fragility of the char issued of the complete reaction.

Figure 37 : Thickness of char at various locations after jet-fire test on IF-1

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Ph-D report J. Ciret Page 73

Consequently, regarding the behaviour of the coating facing to temperature and flames, end

of the test occurs when two thermocouples have reached 400°C, two times must thus be

highlighted:

- At 48 minutes, when the first thermocouple reaches 400°C: thermocouple at position

14 on the top of the central flange (Figure 29).

- At the end of the test (62 minutes) when a second thermocouple reaches 400°C and

the propane flow ends: thermocouple at position 15 on the centre of the central

flange (Figure 29).

On the right side of the back face of the box (Figure 38), the temperatures are measured by

six thermocouples and begin to increase linearly from 100 seconds after the ignition to the

end of the test. The temperature rise is then approximately linear at all positions to reach a

maximum temperature of 206°C after 48 minutes (when the thermocouple 14 reaches

400°C) and a temperature of 228°C at the termination of the test.

Figure 38 : Temperature measured versus time on the right side of the box (in red on panel

schema) for jet-fire test on IF-1

Similarly, on the left side of the back face of the box (Figure 39), the temperatures are

measured by six thermocouples and begin to increase linearly from 120 seconds after the

ignition to the end of the test. The temperature rise is then approximately linear at all

positions to reach a maximum temperature of 200°C after 48.5 minutes.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Page 74 Ph-D report J. Ciret

Figure 39 : Temperature measured versus time on the left side of the box (in red on panel schema)

for jet-fire test on IF-1

Therefore, temperatures on back face are controlled thanks to the protection of a partially

reacted hard coating. Then temperatures on the central flange have also been investigated

thanks to five thermocouples. The inspection of the coating has already shown that the

reaction is complete yielding to char formation and the thermal protection is consequently

less efficient than on the rest of the panel due to the destruction of a part of the char on the

central flange front. Indeed the temperature evolutions are presented Figure 40 and the

maximum temperature measured on the specimen during the test is 484°C at thermocouple

position 14 located on the flange where the coating is thinner.

Figure 40 : Temperature measured versus time on the central flange (in red on panel schema) for

jet-fire test on IF-1

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Ph-D report J. Ciret Page 75

So, the structural steelwork test specimen, coated with a single layer of IF-1 was subjected

to a propane jet-fire with an average mass flow rate of 0.30kg.s

-1

for a period of

approximately 62 minutes. The thermocouples at position 14 displayed significantly higher

temperatures throughout the test than the other 17 thermocouples. The test was not

terminated until a second thermocouple (at position 15 on the central flange) reached a

temperature of 400°C but the cause of failure appears as clearly the hot conditions on the

flange and the bad efficiency of the protection on this location where heat flux is maximal. A

reason advanced to explain this failure is the low expansion of the coating associated to not

appropriate thermo-physical properties (e.g. heat conductivity) and thus its failure at the

hottest point where low expanded protection is not sufficient to insulate metallic substrate.

II. 2. Intumescent formulation 2

Similar experiments in accordance with the Health and Safety Executive document OTI

95634 have been carried out with intumescent formulation 2 used with mesh located at a

nominal depth of 7mm. The test is undertaken on panel box with central flange but only the

flange is coated in the passive fire protective material because we suspect this flange to be

subjected to higher temperature and higher pressure. The PFP material thickness has been

made at the positions specified in standards on the flange and the global average coating

thickness is 10.8. The flange is instrumented with 6 thermocouples to monitor the specimen

temperature positioned on the flange (position 13 to 18 specified by OTI 95634 standard in

accordance with Figure 29).

During the test, the flame engulfed the specimen making it difficult to see the coated

surfaces very clearly through the flame (same behaviour as for IF-1). The test is over when

successively the temperature at position 15 and then at position 17 reached a value of 400°C

approximately after 35 minutes.

Following the completion of the test, the specimen was allowed to cool down. After this

time a detailed inspection was carried out (Figure 41). The coating on the specimen had fully

reacted, no coating remain without being expanded. There was a section of the flange in the

region where the propane jet impacted on the specimen that had no coating present. The

coating on the rest of flange has fully reacted and remained intact on the top.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Page 76 Ph-D report J. Ciret

Figure 41 : Pictures taken after jet-fire test of IF-2

The description of the flange is shown schematically in Figure 42. The coating on the upper

portion of the flange has reacted but remained intact. There are areas where the top layer of

the coating has been removed but the mesh and the coating beneath the mesh is in place

and continues to insulate steel to flames. The mesh has lifted 130mm from the flange on the

left hand side of the flange (as viewed facing the front of the box). A 170mm width band

where the propane jet impacts on the specimen has no coating present and constitutes a

non protected zone for steel.

Figure 42 : Schematic diagram showing fire damage to the flange after jet-fire test on IF-2

Temperature profiles can be gathered for 6 thermocouples at equivalent position all along

the flange (position 13 to 18 from the top to the bottom). On the flange, plots of

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Ph-D report J. Ciret Page 77

temperature variation with time are presented in Figure 43. Several behaviours can be

determined in accordance with aspect of the coating after test.

- Firstly, at all positions, the temperature rise increases steadily throughout the test up

to approximately 1500 seconds.

- After approximately 1500 seconds, the temperatures measured at thermocouple

positions 13, 14 and 15 continue to rise approximately linearly; these measurements

correspond to the top of the flange where coating is still intact after test. At others

positions, temperatures continue to increase but the rate of temperature reduces

until approximately 2800 seconds, 2000 seconds and 3000 seconds respectively for

position 16, 17 and 18.

- Furthermore, after these respective times, thermocouple measurements exhibit an

chaotic evolution. These rapidly changing rates of temperature rise measured at

thermocouples 16 to 18 can most likely be attributed to the breakdown of the

passive fire protection coating due to erosion from the propane jet.

The test has been terminated when the temperature measured at thermocouple position 17

exceeds 400°C.

Figure 43 : Temperature measured versus time on the central flange (in red on panel schema) for

jet-fire test on IF-2

The flange coated with a layer of IF-2, with an initial thickness of 10.8mm and with a steel

mesh at a nominal depth of 7mm, was subjected to a propane jet-fire for a period of

approximately 35 minutes. Temperatures on the test specimen increased throughout the

test and the maximum temperature rise measured at the termination of the test was 410°C

at thermocouple position 17, located directly where jet impacts. After test the coating is

intact on the top half of the specimen and is significantly eroded on the bottom half of the

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Page 78 Ph-D report J. Ciret

specimen with no coating present in the area of impact of the propane jet, highlighting

clearly the low mechanical resistance of the developed coating facing to jet pressure and

impact.

II. 3. Intumescent formulation 3

About the intumescent formulation 3, jet-fire tests have been carried out in two steps.

Indeed, because the formulation of IF-2 and IF-3 are very close, a first test has been carried

out only with the flange coated by the passive fire protective material including a mesh,

similarly to the previous one introduced for IF-2 (in agreement with OTI specifications except

that only central flange is coated by intumescent paint). This first test should permit an

observation of the char after test and determine the mechanical resistance of the coating

directly facing to the jet impact. Then a second test describes a jet-fire resistance test on a

completely coated panel box with IF-3 reinforced by mesh composed of a hybrid carbon and

glass fibre and mounted in a steel box.

Some remarks can be stated during the test:

- From the start of the test, there were flakes of material blasted from the surface of

the flange. Pieces are smaller in size than for the intumescent formulation 2.

- Approximately 10 minutes from the start of the test, flakes continues to be ejected

from the flange upon impact. There is erosion of the coating in the region of jet

impact and some erosion of the rear wall of the specimen.

- Approximately 25 minutes from the start of the test, although being not easy to

discern due to flame, horizontal cracks appear to be developing where the jet is

impacting the flange and just above where the jet is impacting the flange.

Following the completion of the test, the specimen was allowed to cool and then a detailed

inspection is carried out and is viewable schematically on Figure 44. The coating on the

specimen had fully reacted. There was a section of the flange in the region where the

propane jet impacted on the specimen that had no coating present. The coating on the

upper portion of the flange had fully reacted and remained intact on the flange. This area of

the test specimen was coloured black from the top to about 290mm from the top of the

specimen with the remainder of the test specimen coloured white/grey (Figure 45).

There were areas where the top layer of the coating had been removed but the mesh and

the coating beneath the mesh were in place. At the bottom right hand side of the test

specimen, the coating was pushed away from the metal surface.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Ph-D report J. Ciret Page 79

Figure 44 : Schematic diagram showing fire damage to the flange after jet-fire test on IF-3

Figure 45 shows the flange immediately after the jet-fire test. It was apparent at this point

that the char close to the jet impingement has been eroded away, whereas at other parts of

the flange the char looks intact. The general appearance of the breakdown of the material

and the subsequent char showed similar trends both sides of the flange. The surface of the

top part of the flange is characterised by blackened intumesced char, lower down the char is

grey / white, followed by an area of detachment where no material appears present and at

the bottom most part of the flange of the char is white in colour and has almost completely

detached. Cracking is clearly visible at the top of the flange where the char is black or grey in

surface appearance although the char integrity appears good.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Page 80 Ph-D report J. Ciret

Figure 45 : Pictures taken after jet-fire test of IF-3

Upon destructive inspection, it is apparent that the char on the flange sides is hard and

compact and is difficult to damage (Figure 46). Removal of char from the around cracks

showed that these were only surface deep, suggesting that material ejected from the flange

during test in the early stages is from the surface. Where the surface appearance is black,

the char is not all eroded, the char is extremely compact and hard enough to resist. Where

the surface is grey, all the char above the mesh is mostly eroded either to a point just above

the mesh or flush with the mesh. The char below the mesh is still hard and dense in nature

and resistant to damage.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Ph-D report J. Ciret Page 81

Figure 46 : Pictures exhibiting the density and the hardness of IF-3 after jet-fire test

In accordance with the setup of this test, temperatures are only reported on positions 13 to

18 (Figure 47), corresponding to thermocouples of the flange and placed where the thermal

failure occurs generally.

Figure 47 : Temperature measured versus time on the central flange (in red on panel schema) for

jet-fire test on IF-3 not reinforced by mesh

Increases in temperature are clearly seen at all positions in the period approximately 5 to 10

seconds after the start of the test. From the graph, at 25 minutes there is a rapid increase in

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Page 82 Ph-D report J. Ciret

temperature which is believed to be detachment of material around thermocouple positions

16 and 17 then at 18. Failure occurs at 30 minutes where the jet impinges the flange and

thermocouples positioned at 17 then at 16 measure temperature above 400°C.

As for a previous test with intumescent formulation 2, thermocouple 14 is lower in

temperature throughout the test which seems to be typical of the circulation properties of

the jet around the box.

Finally, the behaviour of IF-3 exhibits some similarities with the results exposed with IF-2:

the failure describing by this first test shows a lower mechanical resistance of the char where

jet impacts.

Furthermore, an additional jet-fire test on a completely coated box has been carried out

using mesh to reinforce the intumesced structure and maintain coating in place. Effects are

clearly viewable and beneficial to the resistance to coating. Indeed on the area of the

previous observed detachment, the thickness and mesh placement have given a good

improvement of resisting the jet-fire test for 60 minutes and a nonattendance of

detachment. By investigating the char aspect and notably the central flange, mesh

reinforcement permits to keep a residual layer on steel under the mesh. So the zone where

usually no coating is present disappears to be substituted by a zone where the residual

resistant layer continues to protect steel substrate as viewable on Figure 48.

Figure 48 : Aspect of the flange reinforced by mesh after jet-fire test on IF-3

Consequently to the presence of this residual layer, temperature profiles taken on flange

positions do not exhibit high temperature and do not cause failure of the test. Indeed, Figure

49 shows a smooth increase of the temperature all along the flange. The temperature rise is

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Ph-D report J. Ciret Page 83

approximately linear at all positions to reach a maximum temperature of 190°C at the end of

the test.

Figure 49 : Temperature measured versus time on the central flange (in red on panel schema) for

jet-fire test on IF-2 reinforced by mesh

The end of the test has been determined by the increase of temperature on the back face.

Indeed, intumescent formulation 3 failure occurs on the back face of the box, an area where

neither IF-1 nor IF-2 fail. The temperature increase occurs on thermocouple 9 positioned at

box back face Figure 50. Increases in temperature are perceptible at all positions in the

period approximately 5 to 10 seconds after the start of the test. The temperature rise

increased steadily throughout the test up to approximately 500 seconds from the start of the

test for all the thermocouples and for the duration of the test for thermocouples at positions

1 to 9 (all placed on the top of the back face). After approximately 500 seconds the

temperature measured at thermocouple positions 9 increased at a higher rate to a

temperature in excess of 400

C after 11 minutes. Quickly this high temperature is conducted

to thermocouples positions 10 and 11 and the test is terminated when the second

thermocouple reaches 400°C: at position 11 after 13 minutes. Thanks to these

thermocouples positions (Figure 29), a zone of failure can be delimited on the bottom of the

back face.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Page 84 Ph-D report J. Ciret

Figure 50 : Temperature measured versus time on back face of the box (in red on panel schema) for

jet-fire test on IF-2 reinforced by mesh

Char inspection after cooling of the box exhibits two large zones on the back bow where

flowing char has been discovered leading to bad protection of the substrate and premature

failure. The destructive flow pattern of the jet should cause a shearing force and

intumescent formulation 3 seems to be highly susceptible to damage from such an action.

Figure 51 : Pictures taken after second jet-fire test on IF-3 reinforced with mesh

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Ph-D report J. Ciret Page 85

Finally, the material resists to the jet-fire test for 13 minutes prior to failure. The material at

the top of the flange gave close to 60 minutes prior to reaching 400°C nevertheless failure

occurs early on a zone where neither IF-1 nor IF-2 fail. Overall, the material exceeds

intumescent formulation 2 performance but is inferior to intumescent formulation 1.

Intumescent formulation 3 is characterised as evolving a char of dual layers. The char at the

heated surface is similar to the char produced by intumescent formulation 2, i.e. with a low

density and a porous char with large hollow cells while the char at the substrate is much

harder, denser and less flexible; approaching the structure of char produced by intumescent

formulation 1. Therefore, the intumescent formulation 3 can be thought of as a compromise

between the char properties of intumescent formulation 1 and intumescent formulation 2

(Figure 52).

Figure 52 : Investigation of chars produced by the intumescent formulations (IF-1, IF-2 and IF-3)

after jet-fire resistance test

II. 4. Intumescent formulation 4

The fourth intumescent formulation has been evaluated on a structural steelwork test

specimen with central flange. The entire panel surface has been coated by the PFP material

(back faces + central flange) according to the jet-fire resistance test.

At first sight, the char formation is significantly different to that of IF-3. Its inspection after

cooling of the box reveals sever damage on the back faces (Figure 53) and the presence of

large voids present at the steel/char interface (Figure 54). This bad contact between

substrate and coating causes a bad adhesion and that the jet had a route behind the char to

peel back.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Page 86 Ph-D report J. Ciret

Figure 53 : Pictures taken after jet-fire test of IF-4

Figure 54 : Char from IF-4 after jet-fire test exhibiting large voids close to metallic substrate

Another significant difference is the appearance of large cracks all long the thickness of the

char. They should be caused by the inflexibility of the char and either by the impinging jet on

the surface layer or by the build up of spumifics and the pressure from the inner layer.

Its behaviour is so consequently modified and first conclusions can be established in

accordance with the temperature evolution (Figure 55).

- Up to 40 minutes, char stays intact.

- Then after 40 minutes, it is noticed that a section of char was going to peel off in the

bottom left hand corner of the box.

- After 41 minutes, we observe an increase in amount of smoke produced by the

coating reaction.

- After 45 minutes, an intense increase of temperature is observed in thermocouples 7

and 9 then in thermocouples 10.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Ph-D report J. Ciret Page 87

- Finally 5 minutes later at 51 minutes after the gas release, failure of the specimen

occurs when temperature exceeds 400°C in thermocouples 10 and 9.

Figure 55 : Temperature measured versus time on the back face (in red on panel schema) for jet-

fire test on IF-4

This failure can directly be linked to large voids observed with the post inspection of the char

and the flame that can be engulfed close to the metallic substrate. Causes of failure seem to

be a bad adhesion and the build-up of spumific components on the lower layer of the char

leading to vacuum zone and cracks all along the thickness.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Page 88 Ph-D report J. Ciret

III. Conclusions and discussion

about the jet-fire behaviours

To resume, each intumescent formulations exhibit different failures (Figure 56) and

regarding those different failures, some explanations can be suggested highlighting the

different swelling abilities or the mechanical resistance (Table 5).

Table 5 : Comparative sum-up of the different beahviours exhibited by the four intumescent

formulations for large-scale jet-fire resistance test

Intumescent

coatings

Test conditions

Time of

failure

Position of failure

Explanations of the

failure

IF-1

Entire panel coated

No mesh

reinforcement

62 minutes

Top of the central

flange (hottest

point)

Bad insulation due

to a low expansion

IF-2

Only flange coated

Mesh reinforcement

35 minutes

Bottom of the

central flange

(flame impact)

Bad resistance to

jet-fire impact

IF-3 (first try)

Only flange coated

Mesh reinforcement

30 minutes

Bottom of the

central flange

(flame impact)

Bad resistance to

jet-fire impact

IF-3 (2

try)

Entire panel coated

Mesh reinforcement

13 minutes

Bottom of the

back faces

Bad protection

caused by flowing of

char

IF-4

Entire panel coated

51 minutes

Back faces

Bad adhesion to

steel substrate

causing voids

Obviously, intumescent formulation 1 produces a very hard and not flexible char that does

not expand very much. As it does not expand enough, it fails at the hottest part of the box

due to lack of insulation. Experimental data given by thermocouples all along the test and in

the state of the art about fire tests have permitted to determine the heat phenomenon that

takes place for jet-fire and to localise the hottest point in the top third of the flange front.

Temperature distribution is consistent with the direct observation of the char after test and

correlates the lowest swelling of the specimen with the hottest point of the panel.

Even if no report is available in this chapter, jet-fire resistance test has been carried out on a

structural panel box completely coated by IF-2 (back faces and central flange) and has

confirmed the test conclusions and the behaviours of IF-2. Intumescent formulation 2 failure

occurs on flange edge where the flame impacts. The char made from this product seems not

to be sufficiently resistant and is destroyed by the pressure due to the flame impact.

Reinforcement provided by inclusion of mesh system before testing can permit a great

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Ph-D report J. Ciret Page 89

improvement of the mechanical resistance. Nevertheless, results commented above show a

resistance weakness of IF-2 directly linked with the impact and pressure of the flame: no

damage is detected and reaction of the coating is completed whatever the position on the

panel except on the bottom of the flange front, where jet impacts.

Then intumescent formulation 3 has been tested as a compromise between the two first

coatings and should solve swelling and mechanical weaknesses. Indeed char issued from IF-3

appears as strong enough to resist on edge of flange and exhibit an expanded insulative

char. However its failure occurs on the back face of the box: an area where neither

intumescent formulation 1 nor intumescent formulation 2 fail. This failure seems due to the

destructive flow pattern of the jet and the shearing force highly susceptible to damage IF-3

on the back face.

Figure 56 : Zones of failure on jet-fire test according to the intumescent formulations

Finally, intumescent formulation 4, based on a slight modification of IF-3, introduces also a

new type of failure caused by appearance of large voids at the steel/coating interface. These

large voids permit the engulfment of flame near substrate leading to an increase of

temperature and a non protection of the steel. Nevertheless, we can report an improvement

about the time of failure between IF-3 and IF-4 in accordance with their behaviour:

expansion of the large voids occurs later than the flowing of the char observed with IF-3.

Therefore, we have shown that the four intumescent coatings exhibit different behaviours

for jet-fire test. Some of these failures are self explanatory with the investigation of the char

aspect. Low swelling abilities of IF-1, low mechanical resistance of IF-2 or viscosity evolution

of IF-3 are clearly implicated, but some questions remain. Furthermore, in spite of all these

interesting results and observations, jet-fire tests and, to a least extent, hydrocarbon fire

tests are very expensive and time consuming. Pieces of information about the coating

behaviours are so limited and the possibilities of testing a large number of formulations are

reduced. New approaches to study the intumescent coatings and to determine their fire

efficiency must be developed at small scale to reduce cost and to make easier access to

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter II: Evaluation of the efficiency of the intumescent coatings facing to jet-fire

Page 90 Ph-D report J. Ciret

substantial experimental data. Nevertheless, the different phenomena that take place during

these kinds of tests are not always well-known and their extrapolations from large to small

scale are risky. Therefore, the development of such experimental setup must be undertaken

with care and with accuracy and will be described in the last chapter.

Our work in the next chapters will be so to understand these behaviours at laboratory scale.

The range of small scale tests should give us a large number of results concerning different

characteristics of the intumescent coatings. But before introducing a small scale fire test, we

have focused our study on the fundamental determination of the physical and chemical

properties that may be responsible for the jet-fire performances. Currently, swelling abilities

and mechanical resistance are parameters easily available thanks to experimental tools

developed previously in our laboratory and that have already proven their efficiency to

characterize intumescent systems [95, 107]. Therefore, the physical approach developed in

the next chapter to study the key properties of the intumescent formulations should permit

to understand further the behaviours of the coating in fire conditions.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Ph-D report J. Ciret Page 91

Chapter III: Physical

behaviours of

intumescent coatings

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Ph-D report J. Ciret Page 93

This chapter deals with the study of the dynamic behaviour of four intumescent coatings

based on epoxy resin, designed for offshore platforms and described on the previous

chapter. Consequently, it means that these coatings have to provide protection on the steel

over a long time period and they also have to resist to aggressive conditions (i.e. severe

hydrocarbon fires).

The state of art about intumescent formulations introduced on the first chapter has

highlighted the crucial properties required to develop efficient coatings in terms of fire

protection: mechanical strength and integrity, cohesion and adherence, porosity. We have

also widely focused on visco-elastic properties exhibited by the partially degraded and

melted materials during the intumescent process. If the degraded matrix has a too low

viscosity upon heating, easy diffusion of gases takes place and the gases will not be trapped

but rather escape to feed the flame. If the viscosity is too high, there can be formation of

some cracks which allow the oxygen to diffuse into the char and thus decrease the efficiency

of the protective layer. As a consequence, the measurement of the apparent viscosity using

a classical rheometer should give information about the ability of complex systems will

behave upon heating (flowing of the condensed matter, expansion of the structure

measuring both the apparent viscosity and the gap between the plates of the rheometer).

Taken into account the different failures exhibited by the four coatings at large-scale jet-fire

test described in the previous chapter, we have suggested failure of IF-1 because of its low

expansion, failure of IF-2 because of its poor mechanical resistance, failure of IF-3 because of

its rheological properties (flowing of the char) and failure of IF-4 because of its poor

adhesion to substrate. That is why, after description of the experimental apparatus used to

study in details these properties, we will discuss the results and the consequence on the

quality of the char obtained during burning and on the resulting efficiency of an intumescent

coating.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Page 94 Ph-D report J. Ciret

I. Experimental apparatus

I. 1. Materials

The materials used in the experiments below have been previously described on the chapter

II (Section I.1 – page 68) but main components can be briefly reminded.

- IF-1 contains boric acid that plays a crucial role on the degradation mechanism and

also a carbon source blended with APP coated by THEIC [105].

- The formulations of the three other coatings (IF-2, IF-3 and IF-4) are similar. One

noticeable difference is the change of carbon source used in each formulation.

Carbon source in IF-2 is played by the thermosetting resin (no additional

incorporation of a specific char former), a DGEBA type epoxy resin while on IF-3 and

IF-4 a specific components is added on formulation respectively pentaerythritol for

IF-3 and dipentaerythritol for IF-4.

- Others additives usually used on paint formulations have been added: pigments,

surfactant and mineral fillers but their compositions are unknown.

I. 2. Complex viscosity and expansion measurements

Visco-elastic measurements were carried out in order to evaluate the rheological behaviour

of the system when the temperature increases. The rheological studies are based on the use

of Thermal Scanning rheometer TSR Rheometric Scientific ARES-20A, in a parallel plate

configuration. This apparatus allows an original approach of the study of the physical

behaviour of an intumescent coating as both the apparent viscosity and the swelling can be

simultaneously measured as a function of temperature and/or time.

In order to measure the viscosity and the expansion, sample pellets (Ø = 25mm, thickness =

3mm) are positioned between the two plates (Figure 57). The sample pellets were produced

by casting the coating between PTFE plates and after curing for 24 hours at ambient

temperature.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Ph-D report J. Ciret Page 95

Figure 57 : Dynamic visco-elastic measurements in a parallel-plate rheometer

Testing is carried out using a “Dynamic Temperature Ramp Test” with a heating rate of

10°C/min in the range 25-500°C, a strain of 1% and a constant normal force of 100N (200Pa)

or 200N (400 Pa) according to the tested sample (Table 6). Indeed, the four formulations

exhibit different mechanical resistance, IF-1 appears as very strong and compact whereas IF-

2, IF-3 and IF-4 are relatively “fragile”. Hence, two experimental protocols must be used to

characterize reliably intumescent formulations. The first ones for “strong” coatings has been

developed for IF-1 and validated by a previous study [96]. This study has compared the

behaviour of intumescent formulations (notably IF-1) using different laboratory analyses

(notably rheometer) and has correlated thanks to Principal Component Analysis the

laboratory and large-scale furnace test results. Results suggest excellent correlation between

the different techniques. The second protocol is derived from the first one, except that the

normal force is reduced in order not to destruct the char.

Table 6 : Experimental protocols used to reach visco-elastic properties

Thermal program

Protocol for

fragile coatings

Protocol for

strong coatings

Initial Temperature

25°C

Ramp Rate

10°C/min

Final Temperature

500°C

Sampling time

Strain

Normal force

100N (200Pa)

200N (400Pa)

Sensitivity

10g

20g

This test leads to the apparent complex viscosity values and allows the measurement of the

swelling corresponding to the gap between plates. The test is a measurement of the ability

of the coating to expand in spite of the load applied upon it during the intumescence. So the

degree of expansion is related to the strength of the developing char [88]. Furthermore,

knowing the behaviour of the formulations for jet-fire test and notably ones of IF-3, this test

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Page 96 Ph-D report J. Ciret

provide us information about the visco-elastic behaviour of formulation and could explain

the potential flow of materials observed at high temperature.

I. 3. Mechanical resistance of the char

Mechanical resistance is evaluated using TSR Rheometric Scientific ARES-20A as we did for

measuring viscosity and swelling (see above) but with a specific test protocol. Samples are

also 25mm-sized pellets but thinner than ones for viscosity measurements: h = 1mm instead

of 3mm in order to reduce the final size of the sample after free swelling and allow reliable

measurements between plates. At t = 0 s, the sample is put into the furnace heated at

500°C, without applying any strain (the upper plate is not in contact with the sample) as

shown in Figure 58. This allows the sample to intumesce without any constraint. After 5

minutes when swelling is completed, the upper plate is then brought into contact with the

intumesced material and the gap between the plates is reduced linearly (0.02mm/sec).

Figure 58 : Measurement of the char strength in a parallel-plate rheometer

The force is followed as a function of the gap between the two plates. The upper plate used

in this experiment has a diameter of 5mm in order to increase the pressure on the whole

sample and to ensure complete destruction of the char.

I. 4. Swelling abilities

To measure swelling abilities and complete the results obtained by rheological experiments,

we have followed using image analysis the development of the char and we have calculated

the relative swelling in pure radiative conditions. We have so carried out experiments on

squared panels of intumescent coatings (5x5 cm², thickness = 2mm) under mass loss conical

heater at two different external heat fluxes (35 and 50kW/m²) following the protocol shown

on Figure 59. The experimental set-up was designed to shoot a mass loss calorimeter

experiment using an infrared camera.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Ph-D report J. Ciret Page 97

Figure 59 : Experimental set-up for measuring the swelling during a mass loss experiment using

infrared camera

Although no information about temperature will be investigated, the use of infrared camera

provides some benefits. Indeed the infrared camera allows to get clear images with a greater

contrast to make image analysis than traditional camera. Typical infrared images at the

beginning of the experiment and at the maximum of expansion of the intumescent coating

are shown in Figure 60.

Figure 60 : IR images of an intumescent coating on steel plate upon heating at t=0s (a) and at the

maximum of expansion (b)

Using image analysis in dynamic conditions (from a movie), swelling of the intumescent can

be measured and quantified (see the arrow on Figure 60). In this approach, it is assumed

during calculation on images that the expansion is homogeneous and occurs in one

dimension. Typical curve exhibits a sigmoidal shape showing first a rapid development of

intumescence and second a pseudo-steady state at longer times (Figure 61). The benefit of

this approach is to get a quantitative phenomenological model which might be included in

further modeling.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Page 98 Ph-D report J. Ciret

Figure 61 : Typical relative expansion as a function of time of an intumescent formulations during a

mass loss calorimeter experiment (external heat flux = 35 kW/m²)

I. 5. Thermal stability

Thermo-gravimetric analyses were carried out at 10°C/min under synthetic air (flow rate =

100mL/min, Air Liquide grade), using a TA SDT Q-600. The samples (approx. 10mg) in powder

form were placed in open alumina pans, using gold sheets in order to avoid reactions of the

phosphorus species of the coating with alumina.

Time (s)

0 200 400 600 800

Swelling (%)

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Ph-D report J. Ciret Page 99

II. Results and discussion

II. 1. Visco-elastic measurements

As introduced in the previous section and discussed in the state of art, apparent viscosity

measurements and swelling abilities are important characteristics of our intumescent

formulations and have been widely correlated with the fire behaviour [90]. Here they can be

estimated simultaneously with an only one experiment.

II. 1. a) Precaution to use rheological tool to characterize

intumescent coatings

Before investigating the apparent viscosity changes of the intumescent systems, the

behavior of the epoxy resin must be studied. Its evolution of the apparent viscosity must

firstly be correlated with thermal stability of the systems (Figure 62). Before 300°C, the

material is not degraded, the system is a solid tri-dimensional reticulated network thus the

viscosity values have not physical signification. Then the main step of degradation of the

resin, which loses 80% of its mass, occurs between 300 and 460°C while the apparent

viscosity measurements of the epoxy resin show an important decrease in the temperature

range [330-350°C] with a minimum at 350°C. At 350°C, the material has the aspect of a black

viscous paste. We can so link these two phenomena. The degradation of the resin causes the

decrease of the crosslink density and therefore a decrease of the apparent viscosity. The

material is quite totally degraded at 500°C. The high viscosity can then be explained by the

carbonization of the material.

Figure 62: Correlation between viscosity changes (F=100N) and thermal degradation of the system

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Page 100 Ph-D report J. Ciret

So, use of rheometer and investigation of apparent viscosity change give interesting

information but must be interpreted correctly. No physical interpretation is possible before

the onset of the degradation. Once the degradation has begun, we can reach the

measurement of an apparent viscosity values. This approach may then be used to

characterize the behaviour of an intumescent coating.

II. 1. b) Results of IF-1 and role of the intumescent additives

First work has been done at our laboratory only with IF-1. It provides reliable correlations

between viscosity and swelling and highlights the effects of active components of the

formulations such as APP or boric acid [95]. The four systems are a thermoset resin based on

a DGEBA prepolymer and an amide, a thermoset resin mixed with boric acid, a thermoset

resin mixed with the APP derivative and IF-1 (intumescent formulation based on epoxy resin

and containing both boric acid and APP). Its study shows how to use a rheometer to

characterize the behaviour of intumescent formulations.

Figure 63 compares the thermal stability of the four systems while Figure 64 compares their

apparent viscosity changes.

When boric acid is added to the resin, a first degradation step is viewable between 100 and

170°C due to dehydration of H

[105]. This first step of degradation has consequence on

the viscosity changes because we can detect a quick variation of the viscosity around 150°C.

Moreover, the apparent viscosity shows also a decreasing peak but over a larger range of

temperature [300-390°C] compared with [330-360°C] in the case of the epoxy resin alone.

When APP is incorporated into the resin, the TGA curves show evidence that the main

degradation step occurs at a lower temperature than epoxy resin alone. As a consequence of

this early degradation, the apparent viscosity measurements decrease at lower temperature

(290°C) and in a larger temperature range [290-370°C] when APP is added to the thermoset

resin.

When both additives are combined into IF-1, there is a first variation of the apparent

viscosity measured around 150°C due to H

dehydration. Then a small decrease of

viscosity is viewable at 330°C and could be correlated with the slow degradation of IF-1

between 250°C and 500°C. The observations show the formation of a pasty intumescent

shield from 330°C. The intumescent additives reduce the fall of viscosity due to degradation

of the polymeric matrix forming a carbonaceous coating and permit trapping of gases inside

this coating. Jimenez [105] explains the behaviour of commercial formulation IF-1

investigating mixtures based on the main components of the formulations (epoxy resin, APP

and boric acid). She showed the role of particular additives and notably silicate fillers which

should reinforce the char and limit the viscosity decrease.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Ph-D report J. Ciret Page 101

Figure 63 : TGA Curves of the four systems

Figure 64 : Viscosity measuremant versus temperature for epoxy resin alone with boric acid, with

APP and with both (F=200N)

Then, the swelling abilities of the same formulations are compared on Figure 65.

Expansion of the thermoset resin is significantly increased when the APP derivative is added.

APP additive degrades to yield ammonia above 200°C [108] and the expansion is attributed

both to ammonia and to the evolution of volatile degradation products, which are trapped in

the structure. APP incorporated into the resin leads to intumescence.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Page 102 Ph-D report J. Ciret

Similarly, boric acid causes the intumescent formulation to swell from 140°C. It was

proposed that the formation of B

due to the dehydration of the boric acid leads to the

formation of an intumesced material [95].

The presence of both boric acid and the APP derivative in the intumescent formulation (IF-1)

leads to a higher expansion (180%) while the main decrease of viscosity is the lowest. The

residue obtained is solid and dense. It is noteworthy that the swelling (main step) starts at

300°C suggesting that the expansion takes place because of the relatively low viscosity of the

char combined with the release of volatile degradation products. As the peak of viscosity of

the intumescent formulation is the smallest, the swelling is the highest because the

decomposition gases can slowly diffuse into the matrix, providing high expansion and

avoiding escape (either from cracks in the case of too high viscosity or as bubbles in the case

of too low viscosity) to feed the flame.

Figure 65 : Swelling abilities versus temperature for epoxy resin alone with boric acid, with APP

and with both

Finally, this short study based on IF-1 and its components illustrate the role of each

intumescent ingredient on the swelling abilities and on the visco-elastic properties of the

systems. Rheological approach is also described as an efficient tool to characterize the

intumescent formulations as soon as the degradation occurs. This method is so validated

and can be used to study the three other formulations

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Ph-D report J. Ciret Page 103

II. 1. c) Visco-elastic measurements and swelling abilities of IF-

2, IF-3 and IF-4

Figure 66 compares the apparent complex viscosity of the three intumescent formulations

IF-2, IF-3 and IF-4 measured with a normal force of 100N (instead of 200N because of the

fragility of the intumescent structure from these coatings).

By observing TG curves of these intumescent formulations (Figure 67), we detect that

degradation begins at 200°C. Therefore, all values taken before 200°C have no physical

signification. Up to this temperature, the viscosity measurements can be investigated. All

formulations exhibit profile with a quick viscosity decrease attributed to matrix degradation

and gases diffusion then with a viscosity increase characteristics of the stiffening of the

structure by carbonization process [90, 95]. In spite of these common phenomena, the

profiles versus temperature are widely different according to the intumescent coatings.

Indeed, viscosity measurements for IF-2 show a decrease at 300°C to reach an apparent

viscosity minimum of 2.5x10

Pa.s at 370°C. So, the temperature range where the viscosity is

relatively low is [300-400°C]. On the contrary, the behaviour exhibited by IF-3 is widely

different. The apparent viscosity decreases from 220°C and the values have remained very

low (around 5x10

Pa.s) for a wider temperature range [220-430°C]. Finally the profile of IF-4

is close to that of IF-2, the apparent viscosity decrease at 270°C to reach apparent viscosity

minimum of 2.9x10

Pa.s at 360°C. The temperature range where relatively viscosity is low is

between 270°C and 400°C for the three formulations.

Therefore, among the three formulations, we can observe two main different behaviours. In

one hand, IF-2 and IF-4 exhibit a small viscosity decrease and a narrow temperature range

where the apparent viscosity remains relatively low. In the other hand, the profile of IF-3 is

specific with a quick viscosity decrease at lower temperature and a wide temperature range

from 220°C to 430°C with low viscosity. To explain these visco-elastic behaviour differences,

we suspect obviously the different carbon sources. Indeed, the main difference between IF-

2, IF-3 and IF-4 is the carbon source used for intumescent process with epoxy resin on IF-2,

epoxy resin and pentaerythritol on IF-3 and epoxy resin and dipentaerythritol on IF-4. The

pentaerythritol is mainly suspected to cause the low viscosity of IF-3.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Page 104 Ph-D report J. Ciret

Figure 66 : Viscosity versus temperature for IF-1, IF-2, IF-3 and IF-4 (F=100N)

Figure 67 : TGA curves of the intumescent formulations in air at 10°C/min

About the expansion, we have previously seen that it is attributed to the evolution of volatile

degradation products, which are trapped in the structure. Thanks to the Figure 68 and to the

Table 7, we observe that expansion measurements are correlated with viscosity changes.

Indeed, it is noteworthy that swelling starts at temperature where the decrease in viscosity

is observed, suggesting that the expansion takes place because of a concomitant gases

release and decrease in the viscosity of the material. Similarly the swelling period

corresponds to the temperature range where the viscosity is low.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Ph-D report J. Ciret Page 105

We can also remark a correlation between the maximal swelling and the minimum of

viscosity: the lowest the minimum viscosity, the highest the swelling. With a viscosity

minimum around 5x10

Pa.s, IF-3 exhibits a higher swelling with a rate above 150% while IF-4

swells only to 110%. This reduction in swelling observed between IF-3 and IF-4 can be

explained by the carbon donor used and especially by the ratio between of OH groups and C

groups. Indeed, the crucial parameter for a carbon source yielding high expansion is the ratio

between OH groups and C groups [83]. Therefore, the pentaerythritol is the compound most

rich in hydroxyl groups per carbon groups (4 OH per 5 C for PER and 6 OH per 10 C) and its

presence in IF-3 can explain the best swelling rate exhibited by IF-3. Moreover, due to its

viscosity decrease at lower temperature, intumescent period of IF-3 occurs at lower

temperature and for a wider temperature range [210–290°C].

Figure 68 : Swelling versus temperature for IF-1, IF-2, IF-3 and IF-4 measured in the rheometer

Table 7 : Swelling temperature range

Intumescent Formulation

Temperature where

viscosity decreases

Swelling temperature

range (°C)

IF-2

300°C

[250 - 315]

IF-3

220°C

[210 - 290]

IF-4

270°C

[245 - 300]

Therefore, the investigation of the intumescence with a rheological tool allows to underline

some characteristics of the formulations. The effect of the carbon source is notably

determined by comparison between IF-2, IF-3 and IF-4. Pentaerythritol causes a decrease of

the apparent viscosity at lower temperature (220°C). As a consequence, swelling and

intumescence process starts earlier for IF-3. Substitution of PER by dipentaerythritol in IF-4

allows to modify this behaviour. IF-4 behaves at high temperature similarly than IF-2 with

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Page 106 Ph-D report J. Ciret

viscosity values around 2.5x10

Pa.s between 300 and 400°C. The role of the carbon donor in

intumescent coatings and these effect on the visco-elastic behaviours have been also

reported by Andersson et al. [109]. Indeed, they have studied apparent viscosity profiles of

formulations containing pentaerythritol, dipentaerythritol and tripentaerythritol and they

showed that the formulation with pentaerythritol starts its intumescent process at a lower

temperature over a wider temperature range than the formulations with dipentaerythritol

and tripentaerythritol.

These observations and explanations must be directly linked with the behaviour of the

coatings observed during jet-fire resistance test. Indeed, IF-3 exhibits phenomena of char

flowing that occur on the back faces and that have firstly been explained by the rheological

issue. Indeed, we have just shown that apparent melt phase viscosity of IF-3 is minimum for

a broad temperature range [210-290°C] and that the development of an expanded structure

occurs at lower temperature. However, the char does not seem very resistant and flows

under the pressure and the temperature of the flame. The char of IF-2 is so subjected to

severe aggressions and cannot withstand for a too long time to high velocity jet. On the

contrary, in similar conditions, IF-4 exhibits better performance in agreement with the visco-

elastic change observed between IF-3 and IF-4. Use of dipentaerythritol would so be more

desirable than pentaerythritol in coatings for an application in severe conditions in order to

limit the potential flowing and cracking and the early destruction of the protective shield.

II. 2. Mechanical resistance of the formulations

When viscosity varies during the temperature range of intumescence process, the internal

structure of the charred material is also modified. It is this structure that governs the

character of the intumescent shield and the efficiency of its protection. Indeed, if the shield

becomes too hard, the creation and propagation of cracks leading to a rapid degradation of

the material occurs. If a char has a good structural, morphological and heat insulative

properties but is easily destroyed under a mechanical action, its efficiency is totally lost in

the turbulent regime of combustion. The mechanical strength of the char is considered as

particularly important in the case of jet-fires while the char has to resist very aggressive

conditions, such as explosions, violent wind, and impact of the flame.

Figure 69 presents the destruction force of the carbonaceous char plotted against the

distance between the plates for the four coatings at 500°C. At first sight, we can divide the

coatings into two categories: IF-1 appears as a more robust and more resistant char than the

three others.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Ph-D report J. Ciret Page 107

Figure 69 : Mechanical resistance measured at 500°C for char issued from IF-1, IF-2, IF-3 and IF-4

The curves corresponding to IF-2, IF-3 and IF-4 are similar: the char is not mechanically

resistant and a very low force is sufficient to destroy it. It is only when the gap between the

plates reaches 3mm that the force increases. This can be explained by the fact that the

residue has been strongly compressed.

The curve corresponding to IF-1 exhibits a complete different behaviour compared to the

three others. When gap between plates is higher than 6 mm, the measured normal force is

low while for 6mm, the force begins to vary and increases quickly which means that the char

is harder. It is because of the presence of boric acid which provides better mechanical

resistance and the presence of mineral fibres and silicates [95, 105].

Globally, the mechanical resistance curves underline the worst resistance of IF-2, IF-3 and IF-

4 by comparison with IF-1. This parameter have been commented in the state of the art as

primordial to maintain an efficient char facing to fire strain and are subjected to be cause of

IF-2 failure for jet-fire resistance test. Indeed, protective structure developed from IF-2 and

IF-3 does not resist where the jet impacts i.e. where pressure is maximal.

To confirm the difference between the mechanical resistances of the coatings, we can

visually observe the internal structure exhibited by the char after reaction and development

(Figure 70). Intumescent development takes place in rheometer at 500°C for 5 minutes

without any strain, similarly to the first phase of the experimental protocol used to evaluate

the mechanical resistance (before destruction between the two plates).

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Page 108 Ph-D report J. Ciret

Figure 70 : Comparative aspects of the four chars after 5 minutes of free development in

rheometer furnace at 500°C

Useful information has been obtained thanks to the previous pictures to explain the

behaviour of the sample (mechanical resistance and swelling). Indeed IF-1 exhibits an

internal structure that is compact and that keep the initial form with few tiny cells in

agreement with the conclusions suggested by mechanical resistance curve. A high density

could explain the high resistance of the char. On the contrary, morphologies of the char

resulting from IF-2, IF-3 and IF-4 exhibit large voids that could weaken the structure. An

additional difference is thus underlined thanks to the internal structure of the char. It is a

compact and carbonised char issued from IF-1 whereas the three others demonstrate a

porous and fibrous structure as the mechanical resistance experiments let firstly suggest.

II. 3. Swelling abilities of the coatings

As we mention in the previous section, experiments made with rheometer need application

of a normal force to obtain reliable measurements. However, the application of this normal

force prevents the free development of the coating and obviously modifies the abilities of

the coatings to expand. An investigation of the swelling with variation of the normal force

illustrates clearly this effect and shows the effects on all coatings (Figure 71).

Figure 71 shows results on IF-2 and shows that swelling abilities vary from 400% to 150%

according to the normal force applied on the char. We can also note the fragility of the

developed char when normal force is 400N. From 300°C, the measurement of the swelling is

erratic and decreases from 150 to 50% due to the crash of the intumesced structure by the

upper plate.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Ph-D report J. Ciret Page 109

Figure 71 : Influence of the normal force applied by rheometer plates on swelling abilities of

coatings (here results of IF-2 are exposed)

Therefore, a new experimental setup has been developed to get a reliable measurement of

the swelling without strain. With this protocol, only the three fragile coatings have been

analysed because IF-1 is sufficiently robust to swell in rheometer with normal force. Results

of IF-2, IF-3 and IF-4 are investigated at two external heat fluxes: 35kW/m² that corresponds

to simulation of a fire in development and 50kW/m² that corresponds to the simulation of a

developed fire with pre-flashover phenomena [110, 111].

About IF-2 (Figure 72), we can observe the influence of the heat flux applied on the swelling

abilities. Under a high heat flux, the expansion of IF-2 is higher at 50kW/m² than at

35kW/m². Without strain, IF-2 exhibits a swelling ratio of 900% at 35kW/m² and 1200% at

50kW/m². Even if the heat conditions are different in rheometer furnace (convective heat at

500°C) and under heat fluxes (radiation), we can compare the two measurements of

swelling, respectively done on rheometer (130% viewable on Figure 68) or in cone

calorimeter. This comparison allows to clearly show the limitation of swelling caused by the

normal force and the strain applied during rheological experiments.

Other change caused by the heat flux variation is the speed of development. Indeed, we can

have a rough estimation of this parameter by comparing the slope of the swelling curve.

Clearly, we observe a faster development at high heat flux because at the beginning of the

test IF-2 growths by 950% in 100 seconds under a 50kW/m² heat flux while it growths only at

450% during the same time at 35kW/m². This quicker growth at high heat flux can be

explained by the degradation rate of the material. Indeed, we can assume that material

degradation is higher at high heat flux and causes then an important release of volatile

products i.e. an important amount of gases trapped into the swollen structure.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Page 110 Ph-D report J. Ciret

Figure 72 : Expansion measurements of IF-2 exposed to conical heater at two different heat fluxes

Same experiments have been carried out on IF-3. The trends, clearly viewable and exposed

with IF-2 in Figure 72, are however more difficult to be stated with IF-3. Indeed, in Figure 73,

we observe that the expansion of the char versus heat flux appears very close: the swelling

ratio of IF-3 at the end of the test is around 1000% whatever the heat flux. We can however

determine a slight difference in terms of swelling speed because the shield develops faster

at high heat flux: the complete expansion is reached after 150 seconds at 50kW/m² versus

250 seconds at 35kW/m². It is also noteworthy that time needed to reach the complete

development of char at 50kW/m² in the case of IF-3 (i.e. 1000% after 160s) is lower than for

a char from IF-2 (i.e. 1200% after 250s). So the behaviour of IF-3 permits to put quickly in

place a structure in severe conditions and consequently to provide a protection earlier than

IF-2. This information is in agreement with the early intumescent process caused by PER and

highlighted by previous rheological experiments.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Ph-D report J. Ciret Page 111

Figure 73 : Expansion measurements of IF-3 exposed to conical heater at two different heat fluxes

Substitution of PER by dipentaerythritol is the main difference between IF-3 and IF-4. We

have already observed the consequence of this modification on visco-elastic properties and

swelling abilities on the previous section. However, normal force modified our

measurements and no conclusion could be reliably drawn.

On Figure 74, we observe the development of the char issued from IF-4 at 50kW/m² and at

35kW/m². Effect of the heat flux is easily viewable on the speed of development, but less

evident on the final swelling. Indeed, whatever the heat flux, a swelling of around 700% is

reached by the sample. Nevertheless, the steady state is obtained after 150 seconds at

50kW/m² while it occurs after 300 seconds at 35kW/m². The speed of development can be

estimated at 650% per 100 seconds at 50kW/m² and at 350% per 100 seconds at 35kW/m².

So we note that this speed is lower than in the case of IF-3. This low expansion and this slow

speed of development confirm the effect of dipentaerythritol compared from

pentaerythritol. Indeed, we have already enounced that the ratio of OH groups is strongly

linked to swelling abilities. Consequently use of dipentaerythritol instead of PER in

intumescent formulations causes a decrease of the swelling abilities and slows down the

growth of the intumescent shield rate because PER has a higher ratio of hydroxyl group than

dipentaerythritol.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Page 112 Ph-D report J. Ciret

Figure 74 : Expansion measurements of IF-4 exposed to conical heater at two different heat fluxes

To resume about swelling abilities (Table 8), we can firstly remark the effect that is caused

by normal force and strain applied by plates on rheological measurements. Even if the heat

phenomena that take place in the two devices are different (convection in rheometer

furnace and pure radiation in cone calorimeter), a direct comparison of the values shows

clearly the limitation implying by the strain. Hence, we have defined a ratio between the

swelling abilities measured without normal force and the ones measured with normal force

in order to highlight the sensitivity of the char to strain. Higher the ratio, more the char

would be affected by strain. Even if the accuracy of this parameter has not been determined,

we can obtain some interesting information about the mechanical resistance of the coating.

Indeed, when we take into account the swelling at 35kW/m², no significant difference is

observed. Whatever the coating, the calculated ratios exhibit values in a narrow range. At

low heat flux, the specificities of the formulations and their differences cannot be

underlined. On the contrary, at 50kW/m², IF-2 appears clearly as the most sensitive char to

strain whereas IF-3 and IF-4 still exhibits values close to the ones at 35kW/m². Thanks to this

experimental parameter, we have so underlined the weakness of IF-2 responsible for its

failure during jet-fire resistance test.

Table 8 : Swelling abilities obtained with and without normal force and ratio between the two

measurements to quantify sensitivity to the normal force

Measurements

with normal force

Measurements

at 35kW/m²

Ratio at 35kW/m²

(Col.2/Col.1)

Measurements

at 50kW/m²

Ratio at 50kW/m²

(Col.4/Col.1)

IF-1

140%

IF-2

130%

900%

6.9

1200%

9,2

IF-3

150%

1100%

7.3

1000%

6.6

IF-4

110%

700%

6.3

750%

6.8

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Ph-D report J. Ciret Page 113

Second, with this experimental setup, the effect of substitution of PER (IF-3) by

dipentaerythritol (IF-4) is also viewable. Indeed, thanks to PER, IF-3 develops its intumescent

shield quicker than IF-4 containing dipentaerythritol. Moreover, IF-4 exhibits a lower

swelling rate than IF-3 whatever the heat flux. This can be explained by a lower ratio of OH

groups in dipentaerythritol than in PER.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Page 114 Ph-D report J. Ciret

III. Conclusions about physical

properties

Thanks to these different experiments, we have determined some fundamental physical

properties of intumescent coatings which are known to play an important role to build a

performing coating. Indeed investigations in the state of the art have demonstrated that the

viscosity of the degraded matrix, the swelling abilities of the formulation and the mechanical

resistance of the developed structure are critical factors.

About our four intumescent coatings, we have firstly shown significant differences in terms

of viscosity provided by carbon sources (epoxy resin alone or with PER or with

dipentaerythritol according to the formulations). Indeed, PER causes the shift of the viscosity

decrease at lower temperature (270°C) in IF-3. Consequently swelling and intumescence

process begin also at lower temperature for IF-3. Substitution of PER by dipentaerythritol (IF-

4) allows to modify the behaviour of IF-3 to a behaviour close to that of IF-2. Indeed, IF-4

exhibits a viscosity decrease and begins to swell at higher temperature than IF-3 and we

suspect this modification to provide better fire properties to IF-4 than to IF-3. These results

must be linked with the behaviours observed for jet-fire tests. Indeed, the IF-3 failure and

the flowing of char observed for jet-fire test can be partially explained by the visco-elastic

properties. Indeed, we can suggest that phenomenon of flowing that occurs on the back

faces can be the consequence of a too low viscosity of the melt phase, as observed at 220°C

in our rheological experiments. In similar conditions in jet-fire test, IF-4 presents a better

performance because this flowing disappears in agreement with the change observed for

rheological experiments with our coatings and in literature when PER is substituted by

dipentaerythritol.

The mechanical resistance is also a crucial parameter for an efficient protective shield and

this parameter could explain the failure of IF-2 and IF-3 without mesh. Our experiments

allow to conclude that IF-1 is the most robust and dense char, its char appears as not

sensitive to any strain. Presence of boric acid and a high content of silicate in IF-1 explain this

property and have been well studied by Jimenez [105]. On the contrary, the three other

coatings do not exhibit any resistance facing to the least force applied on the top of the char

during experiments carried out in the rheometer. This sensitivity to strain is shown by

experiments in rheometer to evaluate the mechanical resistance but also by comparing the

swelling abilities of the coatings in rheometer and in free conditions. Indeed a new

experimental setup has been developed and it permits to reach the swelling abilities in

radiative conditions without application of normal force. This methodology highlights the

limitations of the expansion implying by any strain on coatings. Thanks to an empirical

parameter (ratio between the two swelling measurements with and without normal force),

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter III: Physical behaviours of intumescent coatings

Ph-D report J. Ciret Page 115

we have determined that IF-2 and IF-3 are the most sensitive coating to strain. This confirms

thus previous results observed for IF-2 and for IF-3 during jet-fire test resistance that fail in

the impact zone where pressure is maximal and where no coating resists to impact.

Therefore, physical approach provides us interesting information about the intumescent

process. Some conclusions target the role of boric acid and the effect of the carbon sources

on the visco-elastic behaviours or on the mechanical resistance of the formulations. Strongly

linked with the efficiency of the char of these mixtures, these properties have been shown

by physical approach but could be explained by the investigation of the chemical evolution

that takes place inside these formulations. That is why the next chapter is dedicated to the

chemical approach of the intumescent process focused on the role of the carbon source and

the slight differences known inside IF-2, IF-3 and IF-4.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 117

Chapter IV: Chemical

approach of the

intumescent process

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 119

Intumescent coatings are designed to maintain the steel integrity under severe conditions.

As discussed in Chapter I, we know that intumescent formulations are generally based on

three “active” ingredients:

- A carbon source (polyols such as PER or dipentaerythritol) to provide carbons needed

to form the carbonaceous structure,

- An acid source (usually ammonium polyphosphate APP) to dehydrate the carbon

source from an accurate temperature or in presence of flames.

- A blowing agent producing non-flammable gases that cause expansion of the

carbonaceous structure.

APP-PER systems have been widely studied and their mechanism of action has been

elucidated [70] and detailed in the state of the art. Several synergisms have also been

observed with APP and notably combination of APP with boron compounds (e.g. H

, Zinc

Borate…) and TiO

(ingredients widely used in intumescent paints notably in our

formulations). The reactivity of APP with boron derivatives has shown to be beneficial in

terms of thermal stability and fire protection in intumescent paints [74, 112] because of the

formation of borophosphates. On the contrary with TiO

, the formation of TiP

(reaction

between phosphate and TiO

) is less investigated and conclusions are opposite: some papers

[85, 113] conclude to a beneficial contribution of the interaction between APP and TiO

terms of mechanical properties and fire performance while others [114] do not report any

benefit. One of the purposes of this section will thus be to investigate the reactivity of APP

with TiO

in each formulation and to clarify the potential benefit of such reaction.

In addition the results for jet-fire test (Chapter II) and first investigations of the physical

behaviours (Chapter III) have underlined the influence of the carbon source and notably the

effect of the substitution of PER by dipentaerythritol. We have shown that carbon source

could allow a shift of the temperature range where intumescent process occurs, PER based

systems developing their intumescence at lower temperature than dipentaerythritol based

ones. So the second purpose of this chapter will be to investigate the chemical changes of

the systems due to use of different carbon sources and to elucidate the role of each carbon

donor.

To carry out these investigations, in a first step, basic mixtures based on active components

in our intumescent formulations have been studied. This preliminary investigation should

provide some information about the mechanism of degradation of the intumescent

formulations, the respective role of the carbon sources and the effect of TiO

. To determine

the chemical degradation pathways, we have first studied the thermal stability of the

mixtures then by using relevant techniques, we have analysed the residues of mixtures after

a 3 hours of heat treatment at a characteristic temperature of the intumescent process.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 120 Ph-D report J. Ciret

In a second step, a similar approach has been followed to study IF-2, IF-3 and IF-4 in order to

highlight the interaction and the reactivity between the different components in presence of

the resin and of others ingredients such as silicates for example. IF-1 is excluded of this study

because its formulation is widely different since it contains boric acid. The effect of boric

acid has already been observed on physical properties in the previous chapter and the

chemical investigation is reported in details in the literature [74, 105]. As a conclusion, we

should propose a chemical degradation pathway for IF-2, IF-3 and IF-4.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 121

I. Materials and experimental

details

I. 1. Material

A complete presentation of the intumescent formulations has been made at the beginning

of the second chapter. Nevertheless, it is useful to remind the main active components.

The investigation is based on the study of three different epoxy based intumescent coatings,

IF-2, IF-3 and IF-4. On the three formulations an ammonium polyphosphate, AP422 from

Clariant, is added as both acid source and blowing agent (because of the release of

ammonia) [76]. In addition the carbon source is played respectively by the epoxy resin on IF-

2, by both epoxy resin and pentaerythritol (PER) on IF-3 and by both epoxy resin and

dipentaerythritol (dPER) on IF-4 (Table 9). Furthermore, additives used on paint formulations

have been added (pigments, surfactant and mineral fillers) but the exact composition of the

coatings are not known. One of these additives will be investigated: TiO

that is suspected to

play a role in the chemical degradation pathway of the systems.

Table 9 : Comparison between the active ingredients of 3 formulations

Intumescent

coating

Polymeric matrix

Acid source

Carbon source

Blowing

agent

IF-2

DGEBA epoxy resin

APP

Epoxy resin

APP

IF-3

DGEBA epoxy resin

APP

Epoxy resin + PER

APP

IF-4

DGEBA epoxy resin

APP

Epoxy resin + dPER

APP

First, to focus our investigation on intumescent process we have carried out experiments

with basic mixtures containing the “active” ingredients of intumescence. The systems

studied are mixtures based respectively on APP-Epoxy, APP-PER and APP-dPER containing or

not TiO

. Ratios (Table 10) between these different components have been chosen according

to those used in intumescent formulations.

Table 10 : APP/Carbon source/TiO

ratio of the basic mixtures

Composition

APP

Epoxy Resin

PER

dPER

TiO

APP/Epoxy

APP/Epoxy/TiO

APP/PER

APP/PER/TiO

APP/dPER

APP/dPER/TiO

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 122 Ph-D report J. Ciret

Our investigations have been so carried out firstly on six basic mixtures that could provide a

first idea about the chemical degradation mechanism of the complete intumescent

formulations. Indeed, the use of three carbon sources represents the main (known)

difference observed on IF-2, IF-3 and IF-4.

I. 2. Thermal stability

Thermo-gravimetric analyses were carried out at 10°C/min under synthetic air (flow rate =

100mL/min, Air Liquide grade), using a TA SDT Q-600. The samples (approx. 10mg) in powder

form were placed in open alumina pans, using gold sheets in order to avoid reactions of the

phosphorus species of the coating with alumina.

In order to determine whether a potential increase or decrease in the thermal stability

happens between the studied ingredients, the weight difference curves between

experimental and calculated TGA curves were computed as follows [72]:

- W

intu

(T): values of weight given by the TGA curve of the intumescent ingredients (APP

+ Carbon source),

- W

TiO2

(T): values of weight given by the TGA curve of TiO

- W

exp

(T): values of weight given by the TGA curve of the basic mixture (APP + Carbon

source + TiO

- W

the

(T): theoretical TGA curve computed by linear combination between the values

of weight given by the TGA curve of the intumescent mixtures and TiO

: W

the

(T) =

x.W

TiO2

(T)+(1-x).W

intu

(T) where x is the content of TiO

in the mixtures.

- ΔW(T) = W

exp

(T)-W

the

(T): weight difference curve.

Therefore, the ΔW(T) curves allow us to show a potential increase or decrease in the thermal

stability of the system due to the presence of TiO

I. 3. Heat treatment

Thanks to the thermo-gravimetric curves, we are then able to determine five characteristic

temperatures (heat treatment temperature: HTT) in order to simulate the evolution of the

intumescence structure. Heat treatments have been carried out under synthetic air in

tubular furnace following the temperature-time program shown in Figure 75.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 123

Figure 75 : Experimental setup and temperature profile of the heat treatment in tubular furnace

The coatings are heat treated for three hours at these specific temperatures in a tubular

furnace and the products are then grinded as powder in order to carry out chemical

analyses. The heat treated residues are then analysed by XRD and solid state NMR.

I. 4. X-Ray diffraction

The X-Ray diffraction (XRD) analyses allow us to characterize the crystallized products in the

char. Samples have been analyzed to investigate the structure of the intumescent char and

its composition.

XRD spectra were recorded in the [5°–60°] 2-theta angle range using a Bruker AXS D8

diffractometer with radiation Cu K

(λ

(Cu Kα)

= 1,5418 Å) in configuration 2θ/θ. The acquisition

parameters were as follows: a step of 0.02°, a step time of 2 s. The data are analysed using

the diffraction patterns of Inorganic Crystal Structure Database (ICSD).

The ICSD is a comprehensive collection of crystal structure entries for inorganic materials,

produced by Fachinformationszentrum Karlsruhe, Germany, and the National Institute of

Standards and Technology, US. ICSD includes records of all inorganic crystal structures with

atomic coordinates published since 1913. The data base contains more than 70 000 record.

I. 5. Solid state NMR

The high-resolution solid-state NMR spectroscopy was performed using a Bruker Avance II

100 and a Bruker Avance II 400 spectrometer at a spinning speed of 5kHz and of 10kHz

respectively (exception are mentioned when results are presented). Bruker probe heads

equipped with 7mm (Bruker 100) and with 4mm (Bruker 400) MAS (Magic Angle Spinning)

assembly were used.

The low abundance of some atoms compared to the protons leads to poor absorption of the

radiofrequency pulse in a FT-NMR experiment. This limitation can be overcome by exciting

the protons in a sample followed by a sequence of two series of long-time pulses which

make the protons and the other nuclei resonate at the same frequency. The latter is called

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 124 Ph-D report J. Ciret

the "Hartman-Hahn" condition and the process is called "Cross-polarization" (CP). The time

of cross-polarization is called the "contact time". Cross-polarization leads to a large

enhancement of the excitation of the nuclei. The large number of protons in the sample

however interferes with the decay of the isolated nuclei due to weak interactions of the

spins. This dampening of the signal can be removed by a strong radio frequency signal which

essentially holds the protons in a highly resonating state so they are not capable of

absorbing resonance from the nuclei. This is called “

H Dipolar Decoupling (DD)”.

C NMR measurements were performed at 100.8 MHz (9.4 T) with CP-MAS-DD. All spectra

were acquired with contact times of 1ms and a repetition time of 5s was used for all the

samples. Typically, 1024 scans (or 2048 for samples heat-treated at high temperature) were

necessary to obtain spectra with good signal to noise ratio. Tetramethylsilane was used as

the reference.

P NMR measurements were performed at 40.5 MHz (2.35 T) using MAS. A repetition time

of 120s was used for all samples. H

in aqueous solution (85 %) was used as reference.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 125

II. Basic mixtures based on

intumescent ingredients and TiO

II. 1. Thermal stability

II. 1. a) Basic mixtures based on APP and different carbon

sources

First step of this study investigates the influence of the carbon source on the thermal

stability of the systems without TiO

. In Figure 76, we observe the thermal stabilities of each

component.

From the TGA curves, it can be seen that APP has no weight loss before 300°C. The

mechanism of degradation of pure APP has been widely investigated [76, 87]. The weight

loss around 20% over the range of 290 – 500°C is attributed to APP decomposition releasing

and H

O. The weight loss over the range 500 – 650 °C can be attributed to the

sublimation of P

[115]. The residual weight is 20% at 800°C.

The three carbon sources exhibit different behaviours. The degradation of pentaerythritol

occurs between 220 and 300°C. The weight loss is mainly due to PER dehydrating, then

dehydrogenizing and charring. Similarly, dipentaerythritol begins to degrade at 300°C and is

almost completely degraded at 400°C (a first degradation step can be observed before 300°C

corresponding to degradation of around 5% of PER contained in commercial

dipentaerythritol). Then after 400°C, the degradation rate is slower until 570°C, this last step

should correspond to the rupture of C-O-C bond. Finally, the degradation of the epoxy resin

occurs in two main steps: a first step between 300 and 460 °C, corresponding to a weight

loss of 80% and a second step between 460 and 600°C leading to the total degradation of

the resin. The mechanism of thermal degradation of the cured epoxy resin has been studied

by different authors. According to Grassie et al. [116], the weakest points in the cross-linked

resin network should be the N-C and the O-CH

bonds. However, the high concentration of

hydroxyl groups in the cured resin favours localised intra-molecular hydrogen bonding which

would promote dehydration as it has been found in other studies [117, 118]. In the other

hand, Keenan and Smith [118] postulated that primary degradation occurs at the N-C

bond and most likely at the C

-C(CH

)

-C

bonds by homolytic rupture to produce free

radicals. Complex secondary reactions then occur to lead to the formation of small

molecules such as phenols or cresols found during degradation of cured epoxy

Finally, titanium dioxide exhibits no weight loss up to 800°C and is thermally stable.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 126 Ph-D report J. Ciret

Figure 76 : TGA curves of the pure components in air at 10°C/min

The thermal stability of the basic mixtures without TiO

is reported in Figure 77. Whatever

the composition, degradations occur in three main steps:

- Between 200 and 400°C, two degradation steps occur successively. The determined

temperatures of degradation are correlated with those observed on the TGA curves

of pure products (Figure 76) because we observe degradation of APP/PER from

190°C, APP/dPER from 205°C and APP/Epoxy from 250°C. Those temperatures are

close to the respective degradation onset of pure carbon sources.

- The last step, which could be attributed to the oxidation of the formed char, occurs

at higher temperature (between 600 and 700°C) that is significantly different

according to the carbon source: the degradations of APP/PER and APP/dPER begin at

lower temperature than that of APP/Epoxy.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 127

Figure 77 : TGA and derivative TGA curves of the mixtures based on APP and carbon sources in air

at 10°C/min

Regarding the curves, it is obvious that APP/PER is the less stable, APP/Epoxy exhibits the

higher thermal stability and the behaviour of APP with dipentaerythritol seems to be

intermediate. This stability ranking is in agreement with those of carbon sources but results

also of an interaction between the two different components. Indeed, literature [86, 119]

reports that APP and PER react together at 200°C: phosphoric ester bonds are formed first

by alcoholysis of the polyphosphate chain, then alcoholysis and/or esterification leads to

cyclic pentaerythritol phosphate structures inserted in the polyphosphate chains. Similarly,

first step of APP/dPER degradation occurs on temperature range from 250 to 350°C, in

agreement with dPER TG curve. We could assume that similar interactions/reactions occurs

between dipentaerythritol and APP. The increase of the temperature at which the

interaction happens could be explained because the dehydration ability of the dimer is lower

than the one of pentaerythritol (lower ratio of hydroxyl sites compared to carbons). Finally,

APP and cured epoxy resin react together at 320°C. The high temperature of reaction can

also be explained by the high temperature of degradation of epoxy resin (viewable on Figure

76) and by the lower reactivity of epoxy chain with APP.

The residue obtained after those first steps of degradation is more stable in the case of

dipentaerythritol and epoxy resin than in the case of PER. Finally, the last step of

degradation (oxidation) occurs at similar temperature (from 550°C to 600°C) for APP/PER

and APP/dPER while it is delayed for APP/Epoxy from 600°C to 700°C. This difference of

temperature of oxidation could provide information about the sensitivity of the char to

oxidation. The char resulting from the mixture between APP and epoxy is less sensitive to

oxidation than the two other char. Nevertheless, the final residual weight is very close

whatever the mixtures compositions.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 128 Ph-D report J. Ciret

Thanks to this investigation about thermal stability, three heat treatment temperatures can

lead to interesting information and further analysis at those temperatures should be carried

out. Those temperatures are:

- T=450°C, the evolutions of APP/Epoxy and APP/dPER seem to be similar with a

residual weight (RW) of 65% whereas APP/PER is more degraded (RW=55%).

- T=600°C, the mixtures between APP and PER or dipentaerythritol present a RW=25%

whereas the residue of the APP/Epoxy system is around 60%.

- T=800°C, whatever the carbon source, the degradation is almost completed and the

residual mass is lower than 5%.

Therefore, carbon sources used on these mixtures exhibit different thermal behaviours and

cause change on thermal stability of the mixtures. Among the three main steps of

degradation, we can observe that at low temperature epoxy resin improves the stability of

the mixtures, by comparison with PER and dipentaerythritol. This improvement is also

viewable at high temperature because the onset temperature of last step is also increased of

100°C with epoxy resin, without modifying the final residual weight.

II. 1. b) Basic mixtures based on APP, different carbon sources

and TiO

In addition to these mixtures and to study the influence of TiO

and the potential reactivity

between phosphorus-based components and titanium dioxide, TGA curves of APP/Carbon

source/TiO

systems have been carried out. The modification of the thermal stability is

investigated comparing the theoretical and the experimental TGA curves according to the

method described in the experimental section of this chapter.

For APP/Epoxy/TiO

mixtures (Figure 78), we do not observe any difference between the

experimental and the calculated curves up to 620°C, evidencing that no interaction between

TiO

and APP/Epoxy occurs until this temperature. On the contrary, after 620°C, a significant

improvement is viewable and lead to the assumption that a reaction between TiO

and the

degradation products of APP/Epoxy occurs leading to the formation of a thermally stable

residue.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 129

Figure 78 : TGA and difference weight loss curves of APP/Epoxy resin/TiO

in air at 10°C/min

At the contrary, in the case of APP/PER/TiO

mixtures (Figure 79), a large stabilization is

observed from 220°C up to 800°C. This effect is mainly observable between 500 and 800°C

(between 220°C and 500°C, the difference of residual weight is only 3%). Thus, it can be

concluded that, as observed in the case of epoxy resin, the stabilization provided by TiO

occurs at high temperature when a reaction between TiO

and products of a first step of

degradation between APP and PER could take place.

Figure 79 : TGA and difference weight loss curve of APP/PER/TiO

in air at 10°C/min

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 130 Ph-D report J. Ciret

Finally, the behaviour of APP/dPER/TiO

(Figure 80) is similar to that of APP/PER/TiO

. No

interaction is detected up to 500°C. At higher temperature, a large stabilization is observed

and it suggests a reaction between TiO

and degradation products of APP and/or APP/dPER.

Figure 80 : TGA and difference weight loss curve of APP/dPER/TiO

in air at 10°C/min

About the effect of TiO

, whatever the carbon source we cannot observe any significant

modification on the difference weight loss curves consequently to the addition of TiO

up to

500°C for PER and dPER mixtures (up to 600°C for APP/Epoxy/TiO

). However, from this

temperature, a significant stabilisation is observed and potential interaction and/or reaction

is suggested between TiO

and the products issued from the APP-carbon sources

degradation. Thanks to TiO

, the residual weight of the mixtures at 800°C is doubled, going

from around 10% (corresponding to thermally stable TiO

) to 20% (attributed to a thermally

stable products based on titanium derivatives).

II. 2. X-Ray diffraction analyses of the residue

In order to go further in the understanding of the above observed phenomenon (thermal

stabilisation at high temperature), APP/Carbon sources mixtures and APP/carbon

source/TiO

mixtures have been heat treated at T = 450°C for 3 hours in a tubular furnace in

oxidative conditions, according to the profile exposed in Figure 75. This temperature has

been chosen in order to observe products resulting from the interaction between TiO

and

the degradation products of APP-Carbon sources. Indeed, even if this interaction between

components has been detected in TGA curves from 500°C (or 600°C), previous investigations

lead in the laboratory have shown a kinetic difference between a degradation in tubular

furnace (isotherm heat treatment for three hours) and a dynamic degradation in TGA device

(rate of 10°C/min). We can thus assume empirically that step of degradation observed from

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 131

500°C in TGA curves occurs at lower temperature after three hour-heat treatment in tubular

furnace. The residues have then been analyzed using adapted spectroscopic tools.

II. 2. a) Basic mixtures based on APP and different carbon

sources

The first analytical tool used to identify the product is X-Ray diffraction. Whatever the

carbon source, diffractograms of residue after heat treatment at 450°C (Figure 81) are

similar. The residues exhibit a charred structure and their spectra exhibit a broad band

(centred on 2θ = 22–23°) suggesting the formation of an amorphous product containing

carbons [112, 120]. It is well recognized that the carbon structure obtained when the

intumescent process occurs is poorly organized [74, 121]. It could also be noted that at this

temperature APP (a crystallised compound) is no more observed. It is not surprising

according to the TGA curve of the APP reported in Figure 76 demonstrating that the weight

loss at this temperature is around 20%. Indeed, at this temperature APP decomposes firstly

to yield polyphosphoric acid and evolving ammonia (first step of the TGA curve), then

between 360 and 420 C, the polyphosphoric acid reacts with the carbonaceous compound to

form a phospho-carbonaceous structure, namely char [122].

Figure 81 : X-Ray diffractograms of basic mixtures APP/Carbon source after heat treatment at

T=450°C

II. 2. b) Basic mixtures based on APP, different carbon sources

and TiO

When TiO

is added in the mixture, the behaviour of the ternary systems differs from one

system to another (Figure 82). Indeed, after heat treatment at T = 450°C, a well-defined

diffractogram for the two mixtures composed by APP/Epoxy/TiO

and APP/PER/TiO

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 132 Ph-D report J. Ciret

observed. This well-defined diffractogram may be attributed to the presence of TiP

[114,

123, 124], whereas peaks of TiO

are not viewable anymore thanks to direct comparison

with Figure 83. As a consequence, it can be proposed that a reaction between APP and/or its

degradation products and TiO

leading to the formation of TiP

occurs at high

temperature. In other hand, the addition of TiO

in APP/dPER system does not lead to the

formation of titanium pyrophosphate since only a broad band centred around 2theta=22° is

viewable. This diffractogram is similar to the one obtained previously when the APP/dPER

system was heat treated without TiO

Figure 82 : X-Ray diffractograms of the ternary systems APP/Carbon source/TiO

after heat

treatment at T=450°C

Figure 83 : X-Ray diffractograms of TiO

(red) and TiP

(blue)

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 133

The X-Ray diffraction analyses demonstrate that TiO

modifies the chemical evolution of the

basic mixtures based on APP/Epoxy and APP/PER. The product resulting from a reaction

between TiO

and APP or its degradation products was identified as TiP

. This reaction

does not occur before T = 450°C when the carbon source used in the mixture is

dipentaerythritol. In order to go further in the understanding,

C NMR and

P NMR

analyses were used to follow the changes respectively of the carbon structure and of the

phosphorus species present in the char.

II. 3.

C NMR spectra of the residue

II. 3. a) Basic mixtures based on APP and different carbon

sources

The CP/DD-MAS

C spectroscopy is a powerful tool to characterize the carbonaceous

materials and, thus, to confirm the hypothesis proposed from the previous X-Ray

diffractograms. The

C NMR spectra of the binary systems are presented in Figure 84. They

exhibit a broad band centred at 130ppm. The broadness of the band implies the presence of

several non-magnetically equivalent carbons [74, 86]. Thus, it may be assumed that a not

well-organized structure is observed. This result is in good agreement with the data obtained

from X-Ray diffraction analyses that demonstrate that an amorphous structure is obtained.

Moreover, the chemical shift of this band leads us to assign the carbon present in the char to

unsaturated C=C bonds mainly present in aromatic and polyaromatic species [125].

Furthermore, oxidation of the char can also be proposed since the large band present an

overlap between 150 and 170ppm, in particular in the case of APP/Epoxy and APP/PER.

In addition, we can distinguish a small band between 0 and 50ppm on the spectra of the

APP/PER and of the APP/dPER mixtures. This band can be assigned to aliphatic carbons from

polyol derivatives even if the high temperature suggests its degradation.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 134 Ph-D report J. Ciret

Figure 84 :

C NMR spectra of the binary systems APP/Carbon source after heat treatment at

T=450°C

Whatever the carbon source used in the binary systems (APP/carbon source), the residues

obtained after heat treatment at 450°C present a carbonaceous structure

(aromatic/polyaromatic stacks that is partially oxidised).

II. 3. b) Basic mixtures based on APP, different carbon sources

and TiO

Figure 85 reports the

C NMR spectra of the ternary basic mixtures after heat treatment at T

= 450°C. Similarly, to the results obtained in the case of the binary system, a broad band

centred at 130ppm is observed for the three carbon sources demonstrating that part of the

carbon is included in amorphous aromatic/polyaromatic condensed structure. In the case of

APP/Epoxy and APP/dPER, the spectra obtained with and without TiO

are more or less

similar even if their broadness is higher with TiO

, above all for APP/dPER/TiO

compared to

APP/dPER. However, the spectrum of the residue obtained from APP/PER/TiO

shows, in

addition to the band at 130ppm, additional broad band (with three main peaks) centred at

30ppm that can be assigned to aliphatic carbons. This band is not observed in the case of the

binary system without TiO

and the presence of aliphatic carbons on mixtures at this

temperature is surprising because at 450°C PER is totally degraded. This band can attributed

to alkyl carbon type –CH

- [125]. So, not attributed to PER, such aliphatic groups could be

yielded during the degradation mechanism and the formation of the protective structure

and trapped in the polyaromatic shield. We could suspect a modification of the degradation

mechanism causing not only the formation of an intumescent structure but also aliphatic

species thermally stable and trapped into the structure.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 135

Figure 85 :

C NMR spectra of the ternary systems APP/Carbon source/TiO

after heat treatment at

T=450°C

Therefore,

C solid-state NMR highlights the formation of the carbonaceous char at 450°C.

Whatever the composition of the mixtures, the characteristics broadband centred at

130ppm is observed, demonstrating the development of the condensed structure composed

of aromatic and polyaromatic species, these species can be partially oxidized in particular in

the case of char resulting from the degradation of APP/Epoxy and of APP/PER that appear

more oxidized than char of APP/dPER.

Concerning TiO

, this additive does not modify the global mechanism of degradation for two

systems: APP/dPER and APP/Epoxy. However, we could suspect with APP/PER/TiO

that

stable aliphatic carbons are yielded during the degradation and trapped into the structure.

Indeed,

C NMR allows us to demonstrate the formation of the carbonaceous structure

from 450°C whatever the composition with and without TiO

. This carbonaceous structure is

mainly composed of aromatic and polyaromatic species (mixed with aliphatic carbons for

APP/PER/TiO

) and is issued from a reaction between APP and carbon source that occurs

before 450°C.

II. 4.

P NMR spectra of the residue

II. 4. a) Basic mixtures based on APP and different carbon

sources

P NMR is finally used in order to investigate both the amorphous and crystallised

phosphorus species present in the residual char of the basic mixtures. Indeed, even if X-Ray

diffraction analyses demonstrate the formation of TiP

, it does not give information,

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 136 Ph-D report J. Ciret

regarding the amorphous phosphorus-containing products present in the residue and it is

well recognised that intumescent char can contain amorphous phosphate-based compounds

[74, 86].

Figure 86 and Figure 87 report the

P NMR spectra of the basic mixtures respectively

without and with TiO

. In the case of APP/Epoxy and APP/PER, the residues obtained after

heat treatment exhibit a large band centred around 2ppm. This band can be assigned to

phosphoric acid [86, 88, 121, 126-128]. But in the case of APP/dPER, the spectrum is noisier

and the band is shifted to 0ppm. We can attribute it to orthophosphate species linked to

aliphatic carbons previously determined by

C NMR in Figure 84. Indeed APP degrades to

form ortho-phosphoric and pyro-phosphoric acids. Moreover, in the case of APP/PER (clearly

viewable) and of APP/dPER (detected among a noisy spectrum), a large band centred around

-52ppm is observed and may be attributed to cross-linked phosphates and/or branched

polyphosphoric acid [129-131] which are in an intermediate state between APP degradation

and formation of phosphorus oxides P

Figure 86 :

P NMR spectra of the binary systems APP/Carbon source after heat treatment at

T=450°C

II. 4. b) Basic mixtures based on APP, different carbon sources

and TiO

When TiO

is added in the basic mixture, the degradation pathway of the formulations is

different. For APP/Epoxy/TiO

, a large peak centred at around 0ppm and a collection of

bands between -40 and -50ppm is observed. According to the previous XRD spectrum and

literature [85, 114], this collection of bands could be attributed to TiP

whereas the large

peak at 0ppm is still assigned to phosphoric acid. For APP/PER/TiO

, only

TiP

is viewable

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 137

with a collection of bands between -40 and -50ppm. No peak is detectable at 0ppm and thus

it suggests that no phosphoric acid or orthophosphate species are present. However

knowing the initial content (56% of APP i.e. 17% of phosphorus and 8% of TiO

i.e. 5% of

titanium), we calculate that phosphorus components are in excess in the mixtures. So we

can suggest that not only TiP

derivatives is formed but also branched polyphosphoric acid

is, viewable in the same chemical shift range [131]. Finally, with dipentaerythritol, bands are

observed between 0 and -20ppm whereas a large and not well-defined band is observed

between -40 and -50ppm. The first set of peaks is usually detected in the case of

intumescent systems. Bands at 0ppm can be attributed to phosphoric acid (similarly to band

observed for APP/Epoxy/TiO

) while band at -10ppm is attributed to ortho- and pyro-

phosphate species linked to aromatic carbons and/or to pyrophosphate species, [86, 127].

Then the large band centred around -45ppm can be attributed to titanium pyrophosphate,

probably in a glassy structure because of the broadness of the band not. Those results are in

good agreement with the X-Ray diffraction analysis (Figure 82).

Figure 87 :

P NMR spectra of the ternary systems APP/Carbon source/TiO

after heat treatment at

T=450°C

II. 5. Discussion and conclusion

Those results lead us to propose two different chemical degradation pathways:

- (1) either the APP degrades leading to the formation of phosphoric acid, the acid

reacts with the carbon source leading to a phospho-carbonaceous structure that degrades

and further reacts with TiO

leading to titanium pyrophosphate embedded in a degraded

phosphorus and carbon containing matrix.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 138 Ph-D report J. Ciret

- (2) or the APP degrades leading to the phosphoric acid and competitive reactions

between the acid and the carbon source or TiO

occur leading to the formation either of a

phospho-carbonaceous structure and/or of TiP

It may be proposed that depending on the carbon source used in the system, the first or the

second pathway could occur and thus the influence of the titanium dioxide will be different.

TiO

can react with ammonium polyphosphate (or its derivate products) to form TiP

, a

ceramic-like compound at high temperature [113]. This ceramic-like material is suspected to

enhance the properties of the char structure and to protect the substrate against fire [85].

Indeed the TiP

structure consists of TiO

octahedra and PO

tetrahedra sharing corners in

a three-dimensional network [123] and the Ti-O-Ti structure can better resist to oxidation

and can improve the thermal stability of the intumescent coating at middle and later stage

of a fire [113].

Second in the case of PER, we suspect TiO

to slightly modify the degradation mechanism of

the APP/PER mixture. Indeed, while with epoxy resin or dipentaerythritol, we observe a

degradation of the aliphatic carbons and an aromatization of the residues, with PER and

TiO

, thermally stable aliphatic carbons seems to be formed and trapped in the

carbonaceous structure. Moreover, stabilization of the system by TiO

could explain the

absence of phosphoric acid and why we detect only TiP

at high temperature.

In order to go further in the understanding and to demonstrate that the effect observed in

the basic mixtures can also be detected in the complete intumescent formulations, similar

analyses of heat treated intumescent formulations will then been investigated following the

same protocol.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 139

III. Chemistry and reactivity of

the complete intumescent

formulations

The same experiments and the same techniques as those described in the first section of this

chapter are carried out on the complete intumescent formulations in order to highlight the

interaction and the reactivity between the different components of the complete

formulations. The objective is to propose a chemical degradation pathway for each

intumescent formulation.

III. 1. Thermal stability of the intumescent

formulations

The thermo-gravimetric curves of the four intumescent formulations in air are presented in

Figure 88 and their derivatives in Figure 89. Thanks to these graphs, we can determine the

different steps of degradation of each formulation and compare the stability of the different

intumescent coatings versus temperature in thermo-oxidative conditions.

Figure 88 : TGA curves of the intumescent formulations in air at 10°C/min

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 140 Ph-D report J. Ciret

Figure 89 : Derivative TGA curves of the intumescent formulations in air at 10°C/min

Firstly, the mechanism of degradation of IF-1 is well known and exhibits four main

degradation step as described in a previous work [105]. Two overlapped steps are observed

between 100 and 200°C corresponding to a weight loss of about 10% and to the degradation

of boric acid. Indeed H

successively dehydrates into metaboric acid and then into boron

oxide:

A third step occurs between 180 and 500°C, corresponding to a weight loss of about 35%.

Boron oxide and phosphoric acid obtained from APP degradation react together to yield

borophosphates BPO

and as boron has been added in excess compared to phosphorus,

some B

remain at 600°C.

Then, a fourth step, occurring between 500 and 800°C, leads to a final residue of about 30%

of the initial weight. This step is attributed to degradation of different fillers contained in IF-1

and to the oxidation of the char.

This mechanism of degradation has been well-defined and the performance of the

intumescent formulation has been successfully correlated with the formation of

borophosphates. However, the three other formulations do not contain boron components

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 141

and borophosphate formation become thus impossible. Therefore, the goal of the next

investigations is to determine a chemical degradation pathway for the three other coatings.

Regarding those coatings, we can note that, except the degradation of boric acid (100-

180°C), the two other steps of degradation occur in the same range of temperature: from

200 to 400°C concerning the main step and oxidation of the transient char is observed after

600°C. Nevertheless, some differences provide us additional information. First, the stability

of IF-2 is higher since its degradation begins at 200°C while the mass losses of IF-3 and IF-4

become significant respectively from 170°C and from 180°C. This highest stability of IF-2

could be explained by the addition of respectively PER and dipentaerythritol in IF-3 and IF-4

that causes degradation at lower temperature as observed with basic mixtures: thermal

stability of APP/Epoxy is higher than APP/PER and APP/dPER (Figure 77). Moreover, we could

note that the oxidative degradation at high temperature of IF-2 and IF-4 is faster than IF-3.

Contrary to the basic mixtures where APP/Epoxy residue was less sensitive to oxidation than

the two others, it is the IF-3 (with PER) that seems to be the less sensitive to oxidation and

that exhibits the highest stability. Consequently, at 800°C, the thermal degradation of IF-3 is

not completed and continues above 800°C.

Thanks to these TGA curves highlighting the thermal stability of the formulations (Figure 90),

five temperatures have been chosen corresponding to characteristic steps of the

intumescence process:

- At 250°C, boric acid is completely degraded while degradation of the three other

coatings starts.

- At 300°C, the rate of degradation is maximal.

- At 450°C, the degradation rate is slow, a transient material is obtained.

- The fourth temperature (600°C) corresponds to the material just before its last

degradation step attributed to oxidation.

- At 800°C, a relatively stable residue is formed for the coatings, except for IF-3. An

additional higher temperature (900°C) has been added to investigate the evolution of

this intumescent formulation.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 142 Ph-D report J. Ciret

Figure 90 : Determination of the five characteristic temperatures

According to these temperatures, the intumescent coatings are heat treated for three hours

in a tubular furnace (m = 5g) and the residues are analysed using adapted spectroscopic

analyses.

III. 2. Mechanism of degradation of intumescent

formulation 2

III. 2. a) Investigation of the structure of the intumescent

system by X-Ray diffraction

Figure 91 reports the X-Ray diffraction patterns of IF-2 and of its residue obtained after heat

treatment at the previously defined characteristic temperatures.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 143

Figure 91 : X-Ray diffraction patterns of IF-2 and of its residues at several characteristic

temperatures

From ambient temperature to 300°C, mainly APP and TiO

are detected (see Figure 92 for

comparison) with two intense peaks around 15° and four peaks between 25 and 30°. Then

from 450°C and at higher temperatures, diffractograms become noisy due to the

carbonisation of the system and progressive loss of its crystallinity. Whereas TiO

is still

detected (peaks at 27, 36 and 54°), APP is no longer observed. The disappearance of the

characteristic peaks of APP can be explained by its reaction with epoxy resin and the

formation of a carbonaceous structure as previously observed in TGA curves of APP/Epoxy

(Figure 77) and of IF-2 (Figure 88) showing a step of degradation of systems between 300

and 450°C. This observation is also in agreement with literature [76, 79] that reports

reactivity of APP with carbon source around 300°C and with the degradation of the epoxy

resin in the same temperature range (Figure 76). So we can suggest that in this temperature

range [300 – 450°C] the intumescent process occurs as already observed by rheological

experiments as swelling is important from 300°C. In addition, we detect since 450°C the

progressive growth of a large band between 20 and 25° that can be attributed to amorphous

carbons [120] mainly present in char.

Finally, for a heat treatment at 800°C, TiO

is still detected but three additional peaks at 23,

25 and 27° appear and can be assigned to TiP

(Figure 82).

P NMR spectra are necessary

to confirm those species because of the low resolution of XRD diffractogram.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 144 Ph-D report J. Ciret

Figure 92 : X-Ray diffractograms of APP and of TiO

III. 2. b) Investigation of the evolution of the carbon

structure in the intumescent system by

C solid-state NMR

The

C NMR spectra of IF-2 and of its residue obtained after heat treatment at the

previously defined characteristic temperatures are reported on Figure 93. Only heat

treatments up to 450°C have been reported because additional analyses do not present

further information. At higher temperature, spectra are similar to that at 450°C, exhibiting a

broad band centred at 130ppm with a low signal to noise ratio.

Figure 93 :

C NMR spectra of IF-2 and of its residues at several characteristic temperatures

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 145

From ambient temperature to 300°C, the three first spectra exhibit a signal that is mainly

characteristic of the epoxy resin (see Figure 94 to comparison). Indeed, we assume that, as

far as we know the formulations, IF-2 does not contain other carbon elements, except

DGEBA type resin and amide hardener with unknown carbonyl chains [59, 132]. This

assumption is confirmed by spectra at low temperature where no more additional peaks

have been detected thanks to

C NMR. Thanks to these spectra, we can however remark

that two bands detected at 70ppm and at 170ppm have intensities that decrease

progressively with temperature. At 170ppm, band is attributed to carbons from carbonyl

group of the hardener and has disappeared at 250°C. At 70ppm, band corresponding to CH

O sites (C3 and C10 from epoxy network) decreases up to 300°C. This intensity decrease

should be caused by a scission of this bond in the epoxy chain and thus the formation of φ–

OH. Then we suggest that APP could react with hydrolyzed epoxy resin and form of P-O-C

bonds with aromatic carbons (C4 and C7) whose bands decrease up to 300°C.

Then the spectrum of the material heat treated at higher temperature (T > 300°C) only

shows a noisy signal. The temperature of 300°C could be classified as “a critical

temperature” for the intumescent development. Indeed, as previously demonstrated this

temperature corresponds to the temperature range where the sample swells (observed

during rheological experiments) and corresponds to the beginning of the APP degradation

and its reaction with the carbon source (observed thanks TGA curves and X-Ray

diffractograms).

Figure 94 :

C NMR spectrum of the thermoset cured resin

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 146 Ph-D report J. Ciret

III. 2. c) Investigation of the evolution of the phosphorus

components in the intumescent system by

P solid-state

NMR

Finally, the

P NMR spectra of IF-2 and of its residue obtained after heat treatment at the

previously defined characteristic temperatures are reported on Figure 95.

P NMR appears

as the spectroscopic tool of choice since it provides information regarding the evolution of

the phosphorus containing compounds that are observed in the intumescent shield along

the heat treatment.

Figure 95 :

P NMR spectrum of IF-2 and of its residues at several characteristic temperatures

On Figure 95, the spectra of the residues heat treated up to 300°C show a double band at -

22 and -24ppm which is attributed to the central group of polyphosphates in ammonium

polyphosphate. In that temperature range, we observe broad bands centred at 0ppm and at

-10ppm attributed respectively to phosphoric acid or orthophosphates linked to aliphatic

carbons and orthophosphates linked to aromatic carbons. These two bands are caused by

the degradation of APP and its reaction with the carbon source. Indeed APP degradation

leads to the formation of phosphoric and/or pyrophosphoric acids via a hydrolysis reaction

that may react on –OH sites of hydrolyzed epoxy resin, as suggesting after

C NMR

investigation, to form P-O-C bridges.

At higher temperature, (T = 450°C), a broad band centred around 0ppm and presenting

shoulder at = -8ppm is observed. At this temperature, the degradation of the APP is

complete, phosphoric acid has been mainly detected (band at 0ppm) but they remain also

ortho- and pyrophosphate linked to aromatic groups from epoxy resin degradation (band at -

8ppm).This last band should confirm the formation of thermally stable P-O-C bonds [128].

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 147

The spectrum of the residue collected after a heat treatment of 600°C shows a broad band

between 0 and -25ppm with a singlet at = -20ppm. The singlet at -20ppm is assigned to

polyphosphoric acid often observed in degradation of intumescent systems [90], but the non

complete knowledge of the formulation prevent us to conclude accurately. Broad and low

intense bands between 0 and -25ppm could also be attributed to ortho- and pyro-phosphate

resulting from APP. Lastly, at T = 800°C, a collection of bands similar to those observed in the

case of the basic mixtures and attributed to TiP

is detected. It is thus demonstrated that

at higher temperature the formation of titanium pyrophosphate occurs also in the case of IF-

2 similarly to observation made with APP/Epoxy/TiO

basic mixture.

To conclude, when epoxy resin is used as a single acid source, IF-2 exhibit a mechanism of

degradation leading to the formation of titanium pyrophosphate. Nevertheless, several steps

can be determined to yield crystalline structure. Firstly, APP and epoxy resin may react

together, via a scission of CH

-O bonds. This reaction permits the formation of a phospho-

carbonaceous structure from 450°C. Then TiO

, observed up to 600°C, reacts with

phosphorus components to product TiP

at 800°C, causing therefore scission of P-O-C

bonds.

III. 3. Mechanism of degradation of intumescent

formulation 3

A similar approach than the first one used for IF-2 was used to investigate the mechanism of

degradation of IF-3. The focus is nevertheless on the influence of PER and the consequence

of this modification between the two formulations. Furthermore as already mentioned with

investigation of the thermal stability, last oxidation of IF-3 is not completed at 800°C. We

have so added a higher heat-treatment temperature (900°C) to investigate the chemical

changes of this intumescent system.

III. 3. a) Investigation of the structure of the intumescent

system by X-Ray diffraction

Figure 96 reports the X-Ray diffraction patterns of IF-3 and of its residues obtained after heat

treatment at the previously defined characteristic temperatures.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 148 Ph-D report J. Ciret

Figure 96 : X-Ray diffraction patterns of IF-3 and of its residues at several characteristic

temperatures

Similarly, to IF-2, the first component that we can detect by X-Ray diffraction at low

temperature is APP with intense peaks. However while we are able to detect it from ambient

to 300°C for IF-2, the peaks attributed to APP have already disappeared at 250°C for IF-3 and

only TiO

is clearly viewable at 300°C. Therefore it seems that addition of PER causes the

degradation of APP and/or its reaction with PER as suggested elsewhere at lower

temperature [127, 133]. This degradation at lower temperature is also in agreement with

the TG curve of IF-3 where the onset temperature appears lower than those of IF-2 (Figure

89). Consequently at 300°C, IF-3 exhibits only a noisy spectrum where TiO

peaks are still

viewable (at 27, 35 and 54°) because of the intumescent process and of the carbonization.

This carbonization and the development of an intumescent structure containing amorphous

carbons cause the appearance of a broad band between 20 and 25° as observed with IF-2

from 300°C. It can thus be assumed that the phospho-carbonaceous structure characteristic

of the intumescent shield should be formed at lower temperature when PER is used as co-

carbon source with epoxy resin.

Then up to 800°C, we observe noisy signal. At 800°C, similarly to what occurs in IF-2, we

detect appearance of three peaks at 23, 25 and 27° assigned to TiP

(Figure 82) whose

crystallisation is in progress. Indeed, for a heat treatment of 900°C, we observe a well-

defined spectrum highlighting the formation of well crystallized TiP

. Diffractograms of

complete formulations become as well-defined as those of basic mixtures and reveal the

formation of a more crystallized structure that literature [113] qualifies as a ceramic-like

material enhancing the char structure and protecting the substrate.

P NMR spectrum will

confirm the presence of such specie.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 149

III. 3. b) Investigation of the evolution of the carbon

structure in the intumescent system by

C solid-state NMR

The

C NMR spectra of IF-3 and of its residue obtained after heat treatment at the

previously defined characteristic temperatures are reported on Figure 97. Once again, only

heat treatments up to 450°C have been reported because additional analyses do not present

further information except a broad band centred at 130ppm among a noisy signal.

Figure 97 :

C NMR spectra of IF-3 and of its residues at several characteristic temperatures

The first difference provides by

C NMR is the detection of PER at ambient temperature

with two bands at 50 and 58ppm (Figure 98).

Figure 98 : Assignment of bands on

C NMR spectrum of PER

We note that when the material is heat treated at low temperature (250°C), PER has already

reacted or degraded as suggested by the disappearance of its two bands and the conclusions

drawn using the TG curve of pure products (Figure 76). The epoxy resin degrades in a second

step because the “signature” of the epoxy resin is observed until a heat treatment of 300°C

in the case of IF-3. It should also be underlined that the band at around 70ppm (band

corresponding to CH

-O sites in epoxy network - Figure 94) does not decrease up to 300°C

when the carbonization due to intumescent process occurs and a consequent noisy

spectrum is observed. It is thus possible to assume that in presence of PER, the APP or its

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 150 Ph-D report J. Ciret

degradation products react with PER rather than with the epoxy resin. The transient residue

should thus present a different structure comparing IF-2 and IF-3.

Previous works with basic mixtures and several papers [119, 133, 134] report this interaction

between APP and PER, showing that upon heating APP and PER together phosphoric ester

bonds are formed first by alcoholysis of the polyphosphate chain around 280°C. The formed

phosphor-ester can then undergo a ring closing esterification and a reduction to a phospho-

carbonaceous network. This second step could be viewable from 450°C when spectrum of

the material exhibits only a noisy signal. Similarly, to IF-2, this phenomenon may be

attributed to the aromatic and polyaromatic carbons forming the structural shield.

III. 3. c) Investigation of the evolution of the phosphorus

components in the intumescent system by

P solid-state

NMR

Finally, the

P NMR spectra of IF-3 and of its residue obtained after heat treatment at the

previously defined characteristic temperatures are reported on Figure 99.

Figure 99 :

P NMR spectra of IF-3 and of its residues at several characteristic temperatures

In the case of IF-3, APP is detected up to 300°C. This is demonstrated by the double bands at

-22 and -24ppm while we have suspected its degradation at lower temperature drawn on

the disappearance of characteristic peaks at 15° on X-Ray diffractograms (Figure 96). This

apparent discordance can be explained by a loss of the crystallinity of APP which gives no

signal by XRD. But this sustained detection of APP at 300°C confirms also a similar evolution

of IF-2 and IF-3. Indeed the degradation of IF-3 begins from 250°C since additional bands

appear after a heat treatment at this temperature. At 250°C and 300°C, we detect two bands

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 151

at -4 and -12ppm that are attributed respectively to ortho-phosphates PO

in R

(R being

an alkyl group) and to either PO

in R’R

(with R’ being an aromatic group whereas R is an

alkyl group) or to pyro-phosphates (R

). At higher heat treatment temperature (450°C),

we can distinguish three bands in the region [-1 to -20ppm]. Even if it is difficult to attribute

accurately each one, it may be reasonably assumed that these bands correspond

respectively to orthophosphate and/or phosphoric acid (-1ppm), to pyrophosphates groups

(at -7ppm) and to orthophosphate in Φ

RPO

around -15ppm.

Then from 450°C, a broad band between -40 and -50ppm is also observed. This chemical

shift range corresponds to the range where TiP

is observed. It can be reasonably assumed

that non-crystallised titanium pyrophosphate is formed at this temperature because we

cannot detect it with X-Ray diffraction. Lastly, it can be noticed that whatever the heat

treatment (≤800°C), no evidence of crystallised TiP

is found

as we do not observe any

well-defined collection of bands between -40 and -50ppm at the contrary of the results

obtained in the case of the basic mixture APP/PER/TiO

(Figure 87). This result is however in

agreement with the results obtained from the X-Ray diffraction analyses at 800°C that do not

show clear evidence of the formation of titanium pyrophosphate except only a progressive

outline of the three peaks at 23, 25 and 27° that suggests its formation at higher

temperature.

Nevertheless, it appears that when PER is added as co-carbon source with epoxy resin on

intumescent formulation, it is the additive (PER) that plays mainly the role of carbon source

and that react faster with APP. In the other hand, epoxy network is distinguished up to 300°C

without reacting with APP because their characteristic bands are not modified up to 450°C

(even at 70ppm where APP usually react with epoxy network in the case of IF-2). This change

of carbon source could modify the chemical pathway of IF-3 compared to IF-2. We assume

that the residue obtained after reaction between APP and carbon source should exhibit a

different structure based on ortho- and pyro-phosphates. Then titanium pyrophosphate is

suggested at 800°C and well detected at 900°C. So we remark that main difference between

IF-2 and IF-3 could be the temperature where TiP

is crystallized. In spite of degradation at

lower temperature, PER and intermediate residue seem to delay its formation at higher

temperature.

III. 4. Mechanism of degradation of intumescent

formulation 4

Due to their similarities, investigation of IF-4 will be done in direct comparison with IF-3.

Indeed, PER is completely substituted by dipentaerythritol in order to modify the visco-

elastic behaviour of the coating. We have already shown that this modification causes a shift

of the viscosity decrease and the observed behaviour is close to the one of IF-1 or IF-2.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 152 Ph-D report J. Ciret

However, a decrease of the swelling abilities was also noted. Aim of this investigation is to

study the chemical changes of the system.

III. 4. a) Investigation of the structure of the intumescent

system by X-Ray diffraction

Figure 100 reports the X-Ray diffraction patterns of IF-4 and of its residue obtained after

heat treatment at the previously defined characteristic temperatures.

Figure 100 : X-Ray diffraction patterns of IF-4 and of its residues at several characteristic

temperatures

Firstly, we observe that patterns of IF-4 (Figure 100) are less noisy than those obtained for

IF-3 (Figure 96). It can thus be proposed that since X-Ray diffraction analysis is sensitive to

the crystallized components, the crystallinity of the charring system resulting from IF-4

decomposition is higher than that observed for IF-3.

At ambient temperature, similarly to what observed for previous coatings, the pattern of the

complete intumescent formulation exhibits mainly the bands that are characteristic of APP

and TiO

(Figure 83). The characteristic peaks of APP disappear partially after a heat

treatment at 250°C whereas the peaks attributed to TiO

(an intense peak at 27° and two

low peaks around 36 and 54°) are observed up to 600°C for IF-4. Then, the degradation of

APP is complete at 300°C. We observe the degradation of APP at lower temperature similarly

to IF-3 while APP is still detected at 300°C for IF-2. Consequently, it may be proposed that

PER and dipentaerythritol exhibit a higher reactivity with APP than epoxy resin. This

difference can be explained by the mechanism occurring in each formulation. With PER or

dipentaerythritol, we assume that esterification reaction occurs while with epoxy resin we

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 153

have shown that a first bond scission from epoxy network is needed to form φ–OH sites and

then a reaction with APP could occur. The phospho-carbonaceous structure resulting from

the degradation of APP and characteristic of the intumescent shield should be formed at

lower temperature when dipentaerythritol is used as co-carbon source with epoxy resin in

the complete formulations.

Following the degradation of APP, from 450°C, we detect directly the formation of TiP

demonstrated by the presence of three intense diffraction peaks between 2theta = 20 and

30°. These peaks are detected up to a heat treatment of 800°C. Moreover, the X-Ray

diffraction pattern of IF-4 heat treated at 800°C is very well-defined and presents some

additional peaks particularly at 22° and 24°. Due to unknown formulation and in spite of

investigation using ICSD database, we cannot attribute these peaks. Therefore, as a first

conclusion, X-Ray diffraction analyses show that whereas IF-3 and IF-4 both degrade with the

subsequent formation of TiP

, the temperature range in which this reaction occurs is very

different. Indeed, TiP

may be suspected from 800°C for IF-3 (with pentaerythritol)

whereas it is clearly observable from 450°C in IF-4 (when PER is substituted by

dipentaerythritol).

III. 4. b) Investigation of the evolution of the carbon

structure in the intumescent system by

C solid-state NMR

Thanks to the

C NMR spectra of IF-4 and of its residue obtained after heat treatments

(Figure 101), we should confirm the chemical pathway discussed above and the formation of

the potential carbonaceous structure. Once again, only heat treatments up to 450°C have

been reported because analyses at higher temperature do not present additional

information and any new change in system.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 154 Ph-D report J. Ciret

Figure 101 :

C NMR spectra of IF-4 and of its residues at several characteristic temperatures

On the spectra at ambient temperature, we mainly detect bands attributed to epoxy resin by

direct comparison with Figure 94. Moreover, typical bands attributed to dipentaerythritol

(narrow and intense bands at 50, 58 and 66ppm) are also detected (Figure 102).

Figure 102 : Assignment of bands on

C NMR spectrum of dipentaerythritol

When the material is heat treated at low temperature (250°C), the dipentaerythritol

degrades shown by the disappearance of peaks attributed to dipentaerythritol. This

observation suggests a reaction of dipentaerythritol or of it degradation products and

similarly to what happens with PER (disappearance of bands at 250°C) we assume an

esterification reaction of dipentaerythritol by APP to form phosphoric ester bonds.

Furthermore this reaction is correlated by the degradation of APP at 250°C (lower

temperature than those observed with IF-2 or IF-3) determined using the X-Ray diffraction

patterns. In parallel, we can note that the epoxy resin does not degrade up to 300°C. So if we

consider that both epoxy resin and PER are used as carbon source in the intumescent

formulation, we can observe that APP reacts preferentially with PER (disappearance of its

bands) than with epoxy resin (whose bands are remaining at 250°C). Thus we can assume

that reactivity of PER with APP is higher than ones of epoxy resin with APP and that the

activation energy required to such reaction is lower for PER with APP than for epoxy resin

with APP.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 155

At 300°C, degradation of the matrix is also observed because only a large band centred at

around 130ppm is detected. This broad band is characteristic of intumescent coating and

implies the presence of several non-magnetically equivalent carbons. This band can be

assigned to several types of aromatic and polyaromatic species which can be partially

oxidized. It suggests the condensation of aromatic species to polyaromatic species. This

broadening may also be explained by the fact that a disorganized structure is created. At

higher heat treatment temperature (T = 450°C), the signal is very noisy, similarly that we

have already observed for IF-2 and IF-3.

III. 4. c) Investigation of the evolution of the phosphorus

components in the intumescent system by

P solid-state

NMR

The chemical change of phosphorus components should also be highlighted by

P NMR. The

different spectra of IF-4 and of its residue obtained after heat treatments are introduced on

Figure 103.

Figure 103 :

P NMR spectra of IF-4 and of its residues at several characteristic temperatures

The main phosphorus component that we detect and that degrades is APP shown by its

double band at -22 and -24ppm. However while we detect this double band up to 300°C for

IF-2 (Figure 95) and IF-3 (Figure 99), when dipentaerythritol substitutes PER, band disappears

and APP degrades completely before 300°C. From 250°C, several broad bands centred at 0

and -10ppm assigned to orthophosphates issued from APP degradation and its reaction with

the carbon source. It is also noteworthy that at T = 300°C, the two broad bands centred at 0

and -10ppm are still observed whereas another broad band centred at -30ppm appears. This

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 156 Ph-D report J. Ciret

band was observed neither in the case of IF-2, nor in the case of IF-3 but it is well recognized

that this band appears when an intumescent system degrades [129, 131]. This band may be

attributed to condensed phosphate/polyphosphate species.

At 450°C, we can also see a broad band around -40 and -50ppm, similar to that viewable on

the spectrum of the residue of IF-3 at 800°C. This band could be similarly attributed to non-

crystallised TiP

. This is in agreement with the XRD pattern at the same temperature

(450°C) showing large bands attributed to titanium pyrophosphate. Once again, we can note

the lower temperature at which this phenomenon occurs. By increasing the temperature

treatment up to 800°C, this band transforms progressively. The bands are more and more

well-defined. At 800 °C, a well-defined spectrum exhibiting a collection of bands between -

40 and -50ppm, similarly to what we observe with the basic mixtures (Figure 87) and

attributed to TiP

. In addition a broad band is detected at -30ppm but, similarly to peaks

observed by X-Ray diffraction, we cannot assign it because the composition of the

formulations are not completely known.

Therefore, X-Ray diffraction and NMR analyses carried out on IF-4 show that TiP

formed from 450°C. Similar reaction as that with PER may firstly be assumed between

dipentaerythritol and APP to form a phosphorus ester at 250°C then a phospho-

carbonaceous structure at 300°C. However, reaction between phosphorus component and

TiO

occurs at lower temperature than with PER. Indeed, we detect TiP

from 450°C

instead of 600°C and their bands are well defined at 800°C instead of 900°C. It evidences

that dipentaerythritol speeds up the evolution of the system without modifying the global

chemical pathway observed with PER leading to a well-crystallized TiP

formation.

III. 5. Discussion and conclusion

III. 5. a) Influence of the carbon source on the chemical

pathway and the formation of TiP

In order to highlight the effect of the carbon source, we have carried out investigation on

three intumescent formulations which contain three different carbon sources: epoxy resin

for IF-2, PER and epoxy resin for IF-3 and dipentaerythritol and epoxy resin for IF-4. In the

previous chapter, we have underlined the nature of the carbon source on visco-elastic

behaviours and on swelling abilities of the intumescent coatings. Thanks to this current

chapter, some differences have been highlighted about the chemical changes that take place

on the different intumescent systems thanks to the different carbon sources.

Firstly, we have demonstrated that epoxy resin could play the role of carbon source on IF-2

as well as other polyols but only when no other char former is present. Indeed when PER or

dipentaerythritol is added, no reaction between epoxy resin and acid source (APP) is

detected (or at higher temperatures). Nevertheless, the chemical pathway of IF-2 and the

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 157

reaction between APP and epoxy resin suggest formation of TiP

similarly to what occurs

with PER and dipentaerythritol. Then, whatever the carbon source, the TiP

formation is

observed at high temperature. However, effect of the carbon source can be observed by

comparing the temperature where the steps of degradation occur as carbon source seem to

shift the temperature where APP degrades or when TiP

is formed. To corroborate this

change of temperature in the pathway, we can compare the different systems at different

characteristic temperatures of their degradation.

Firstly, at 300°C, we observe in Figure 104 that APP is completely degraded on IF-4 while it is

still detected for the two other coatings, IF-2 and IF-3 evidenced by the double bands at -22

and -24ppm of APP. So thanks to

P NMR we can suppose a degradation/reaction of APP at

lower temperature with dipentaerythritol than with PER or epoxy resin. The reactivity of the

carbon source with APP plays an important role for the esterification reactions and we

suppose that it is better with dipentaerythritol. This first step yields to the development of a

protective shield but also intermediate products containing residues of APP.

Figure 104 : Comparative

P NMR spectra of residues of IF-2, IF-3 and IF-4 at 300°C

Even if the structure of the intermediate phosphorus products could be different according

to the carbon source, we have shown that these phosphorus based compounds react then

with TiO

to form TiP

. Figure 105 shows the evolution of the phosphorus components at

450°C. The delay highlighted about the comparative analyses of IF-3 and IF-4 is prolonged at

this temperature corresponding to the reaction between phosphorus components and TiO

Indeed, for IF-4 we have already detected a broad band between -40 and -50ppm, which

suggests TiP

formation. This same band can also be observed with difficulties on spectra

of IF-3 but is less intense (highlighting the delay in terms of degradation) and is not viewable

for IF-2.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 158 Ph-D report J. Ciret

Figure 105 : Comparative

P NMR spectra of residues of IF-2, IF-3 and IF-4 at 450°C

Lastly, X-Ray diffraction pattern of the final residue obtained from IF-2 at 800°C, IF-3 at

800°C and 900°C and IF-4 at 800°C can be compared on Figure 106. The diffraction patterns

in particular those of IF-3 at 900°C and IF-4 at 800°C show similar bands in spite of the

difference of temperature. The peaks attributed to TiP

are clearly observed in both cases

whereas for IF-2 and for IF-3 at 800°C the formation of TiP

can only be suspected.

Figure 106 : Highlights of TiP

formation at different temperature according to the coating

Therefore, these last X-Ray diffractions and other NMR analyses carried out on the three

studied coatings show that whatever the carbon sources used, titanium pyrophosphate is

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 159

formed. However, the temperature range at which it is formed is different. Concluding a

systematic investigation of the evolution of the coatings, the chemical approach of the three

coatings highlights the importance of the carbon source and also the role played by TiO

the different formulations. In the study of basic mixtures based on APP, different carbon

sources (epoxy resin, pentaerythritol and dipentaerythritol) and TiO

, we have demonstrated

the formation of TiP

for the three coatings. The titanium pyrophosphate is the product of

a reaction between TiO

and the phosphorus components issued from the reaction between

APP and different carbon sources. Indeed, APP first degrades into ortho- and pyro-

phosphate as evidenced by solid state

P NMR and then its degradation products react with

TiO

to yield TiP

. However, we should note that the temperature range in which this

reaction occurs is different according to the carbon source used. Indeed, TiP

can hardly

be detected from 800°C for IF-3 whereas it is observable from 450°C for IF-4 in diffractogram

with three peaks between 20 and 30°. For IF-2 without additional carbon source added to

the formulation, epoxy resin alone plays a similar role leading to the formation of TiP

around 800°C.

Consequently, it may be proposed that the use of dipentaerythritol instead of PER or epoxy

resin allows to increase the reactivity of APP with the carbon source and thus leads to the

formation of TiP

at lower temperature, notably. Furthermore, we suspect this crystalline

structure to provide a better fire performance to the intumescent coating. Consequently,

the formation of such structure at lower temperature and the shift of the chemical pathway

could explain the different behaviour of IF-4 for jet-fire test and could be a benefit chemical

change for fire protection.

III. 5. b) Chemical investigation of residues from jet-fire

resistance test - Correlation with proposed chemical

pathways

To correlate further these observations with the performance of the coating at the jet-fire

test, we have analysed by X-Ray diffraction residues obtained from the jet-fire resistance

tests described on the second chapter. Only IF-2 and IF-4 analyses have been done because

they were the only residue available.

On IF-2 char resulting from the large scale jet-fire test (Figure 107), three zones can be

distinguished on the piece of char. From the surface to the steel/coating interface, we

observe respectively an upper layer that is dense and compact (thickness = 2cm), then on

the middle this part exhibits large vacuum with lot of bubbles, and finally near steel another

thinner layer (1cm) that is dense and compact.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 160 Ph-D report J. Ciret

Figure 107 : Piece of IF-2 char from large scale jet-fire test

So we have analysed these three samples taken on the piece of char of large scale jet-fire

test (Figure 108). On large scale test, peaks characteristics of TiP

is detected whatever the

layer of the sample analysed. In the upper layer, sharp peaks (attributed to TiP

or TiO

)

are superimposed on a broad band characteristic of amorphous carbons. Amorphous

carbons are mainly located in the upper layer because on the two other samples (middle and

inner layer), all bands viewable on the diffractograms are sharp and can be attributed to

TiP

or TiO

. Indeed the internal structure of the char seems to be mainly formed with

crystalline components and the detection of titanium pyrophosphate is more and more well

defined when the layer is deep. We suggest thus a high concentration of carbons on the

surface and furthermore presence of TiP

all along the depth of the char.

Figure 108 : X-Ray diffractograms of IF-2 residues from jet-fire resistance test

On IF-4 char, it is easier to distinguish two zones on the piece of char (Figure 109). From the

surface to the steel/coating interface, we observe respectively a 1cm top layer that peels off

for the jet-fire test and a harder and more compact second part with a thickness of 3cm. We

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Ph-D report J. Ciret Page 161

have however obtained three samples by dividing into two sections the wide layer near

substrate.

Figure 109 : Piece of IF-4 char from large scale jet-fire test

We have analysed these three samples (Figure 110). On the surface, we find again a broad

peak at 27° attributed to amorphous carbons, their presence at the surface to yield to

carbonaceous structure is underlined. Overlapped with this broad band, other bands

appearing at 25° could be attributed to TiP

. On deeper layers, TiP

could be also

detected on the middle of the sample and bands of TiO

are present on the basis of the

residue, where temperature is lower. Because of thermal protection and temperature

gradient, TiP

is less and less present from the surface to the depth of the coating.

Figure 110 : X-Ray diffractograms of IF-4 residues from jet-fire resistance test

Therefore, we have highlighted the formation of TiP

for IF-2 (all along the depth) and for

IF-4 (less crystallized compound). Furthermore, we have determined a high concentration of

amorphous carbons on the surface. These results should however be put into perspective

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter IV: Chemical approach of the intumescent process

Page 162 Ph-D report J. Ciret

with the wide temperature range reached for jet-fire test. Indeed, locations of specimen

from jet-fire box are not well specified and knowing the temperature gradient existing all

along the box, we can highly think that on some zones temperature reached is higher than

800°C and permits TiP

formation as for IF-2 sample and inversely some zones are cooler

and does not allow the reaction (as we suppose for IF-4 sample).

Nevertheless, with the detection of TiP

in piece taken from jet-fire specimen, our

chemical pathways proposed in this chapter for each coating are validated. We can assume

that whatever the carbon sources used, titanium pyrophosphate is formed. Our analyses

have demonstrated however, that the temperature range at which it is formed is different.

Indeed, reactivity between APP and the different carbon sources are different,

dipentaerythritol react with APP at lower temperature than pentaerythritol. Consequently,

the substitution of PER by dipentaerythritol (IF-3 becomes IF-4) allows chemical steps of

degradation occurring at lower temperature, notably the formation of TiP

We could finally correlate the formation of such structure and the results exposed for jet-fire

resistance test. Indeed, literature shows that formation of TiP

could enhance the

mechanical resistance of the coating. So it would be rather to form such structure as soon as

possible after the beginning of the incident in order to protect reliably the substrate. The

next chapter with the development of a labscale tool mimicking large fire test should show

the consequence of the chemical pathways on the fire performance of each coating.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Ph-D report J. Ciret Page 163

Chapter V: Development

of an efficient labscale

test mimicking large

scale jet-fire test

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Ph-D report J. Ciret Page 165

Thick intumescent coatings are usually evaluated in large furnace tests by the test UL 1709

[101] If they pass this test, they are then evaluated by the jet-fire OTI 95634 [103]. Both

these tests are essential since they give a good simulation of a real fire and give means of

determining whether or not fire protection products present minimum performance criteria

required in an offshore platform. However, as these tests are very expensive and time

consuming, they reduce the possibilities of testing a large number of formulations.

Nevertheless, significant differences have been highlighted about the behaviours of the four

intumescent formulations after large-scale jet-fire resistance test, suspecting their physical

properties, notably mechanical resistance or visco-elastic behaviours, being important

parameters. These influences have been determined and confirmed by an investigation of

the physical behaviours, we have suggested that high mechanical resistance of IF-1 is due to

boric acid while other formulations (without H

) exhibit low mechanical resistance facing

to jet-fire impact. So use of mesh in those intumescent coatings is required to provide an

additional reinforcement of the fragile barrier and to keep in place the protective shield.

Furthermore the viscosity decrease observed for IF-3 has been stated as an explanation to

the jet-fire failure and can be explained by the presence of PER. Indeed, PER is known to

cause modification of the visco-elastic behaviours of the coating at lower temperature. The

substitution of PER by dipentaerythritol (IF-3  IF-4) allows to avoid this prejudicial change.

In a second approach, we have determined the chemical pathways of the three intumescent

formulations IF-2, IF-3 and IF-4, where only carbon sources change (epoxy resin in IF-2,

epoxy resin and PER in IF-3 and epoxy resin and dipentaerythritol in IF-4). Effect of TiO

has

also been investigated, resulting to formation of TiP

at high temperature thanks to a

reaction with the phosphorus components of the char with TiO2. Regarding the results, it

may be proposed that the substitution of PER by dipentaerythritol allows to increase the

reactivity of APP with the carbon source and thus allows occurrences at lower temperature,

notably the formation of TiP

. Similar pathways have been observed for jet-fire resistance

test in IF-2 and IF-4 leading to formation of a carbonaceous structure and of a crystallized

compound of TiP

. This product based on titanium and on phosphorus is described in

literature to enhance greatly the fire performance of intumescent systems and the

mechanical properties of the char, acting as a ceramic binder to reinforce the shield [113].

Knowing all these aspects, the aim of this last chapter is to try to develop experimental and

reliable tools at the lab scale which could provide information about the development and

the efficiencies of the intumescent structure. These tools could constitute a predictive

approach to determine the fire behaviour of the evaluated formulations and give us a low

cost and fast test allowing rapid and reliable results of a large number of different coating

formulations in order to correlate observations made by our physical or chemical

approaches with the results obtained in large scale fire tests.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Page 166 Ph-D report J. Ciret

Therefore, in the first section of this chapter, we introduce a labscale test developed to

mimic industrial furnaces used for UL 1709 standards (Figure 111). Equipped with a powerful

burner this test allows to evaluate samples according to standardized temperature profiles

specified for hydrocarbon fires.

Figure 111 : Industrial furnaces used for UL 1709 standards

Then, previous jet-fire experiments done with the four intumescent formulations and

introduced on the 2

chapter have shown the crucial effect of pressure and temperature

during the test. Careful analysis of the chars after the jet-fire test revealed zones of high

pressure and zones of high temperature (Figure 112).

Figure 112 : Jet-fire box test and zones of high temperature and/or high pressure

The work on the second section of this chapter has so to detail the development of a small

scale test mimicking jet-fire conditions. The setup of this test should take into account these

two parameters (pressure and temperature), provide similar distributions of heat and

pressure and thus permit observation of coating behaviours in agreement with large scale

ones.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Ph-D report J. Ciret Page 167

I. Preliminary fire test in furnace

conditions

The aim of this test is to evaluate the efficiency of the intumescent coating in terms of heat

transfer similarly to industrial standardized furnace tests but at laboratory scale.

I. 1. Experimental apparatus and materials

Contrary to large industrial fire test, our test occurs in 50cm

-sized furnace but its design

remains close to that of bigger furnaces (Figure 113). In addition, advantages of our test are

the small scale and the easy handling of the tool to screen several formulations and to test

quickly the performance of different formulations. As in the case of large furnace test,

time/temperature profiles are measured on the back of a coated steel plate.

In our test, the heat source is a powerful propane burner generating 35kW. Steel plates used

are squares of 10*10 cm², 3.5mm thick, the steel is exactly the same as that used in the

industrial test. About 1mm of the intumescent coating is applied on the surface of the steel

plate. A black coating, provided by Medtherm Corporation (Huntsville, Alabama), resistant to

800°C and having a constant emissivity of 0.92 is applied on the non-heated side of the steel

plates. The constant emissivity of the backside of the plate allows accurate measurement of

the surface temperature of the plate using an infrared pyrometer or infrared camera. The

infrared camera is positioned at a constant distance from the steel plate and focused on the

back surface of the plate. It detects the temperature on the non heated face of the steel

plate and records the time / temperature curve on a computer. The measurement of

temperature without contact has several advantages: it is instantaneous and can be carried

out in places which are difficult to access. It enables measurement over a larger range of

temperatures without corrosion or degradation. However, it is still important to carefully

control various parameters which could affect the measurements, such as for example the

presence of fumes or dust.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Page 168 Ph-D report J. Ciret

Figure 113 : Pictures of the small scale furnace test

The aim of intumescent coatings is to decrease as much as possible the slope of the

time/temperature curve, i.e. reaching 400°C as late as possible. Indeed, steel begins to lose

most of its structural properties between 470 and 550°C. The temperature of 400°C has

been thus unofficially adopted as critical temperature for heavily loaded structural

components. So, time of failure, namely the time when the steel temperature reaches

400°C, is obviously one of the parameters to be improved: the best result is obtained when

the longest time of failure is reached. It means that the coating has the best protective

effect.

I. 2. Temperature profiles and results

Small furnace tests have then been carried out on the different intumescent formulations

comparing virgin panel and a panel coated only with the epoxy resin (DGEBA + hardener).

The time/temperature curves obtained in labscale furnace tests are shown on Figure 114.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Ph-D report J. Ciret Page 169

Figure 114 : Comparison of the time/temperature curves obtained for IF-2, IF-3, IF-4 and for epoxy

resin using the small scale furnace test

Firstly, we can note the good agreement between the specified temperature profile of the

standard and the temperature taken by thermocouple inside furnace during test, even if the

first goal of this test is not a certification but a preliminary observation of the fire behaviour

of the different coatings. The temperature rises rapidly to 900°C within 4 minutes and

significantly higher overall temperatures are reached (between 1100°C and 1200°C).

Then the virgin plate reaches 400°C in 3 minutes and the epoxy resin coating whose quality

is of course not the thermal insulation does not improve the thermal behaviour of the panel

(time of failure is 3 minutes and 30 seconds). However, when the intumescent coatings are

applied, time of failure increases significantly. It is multiplied by 4 whatever the formulation:

IF-2 and IF-3 exhibit a similar behaviour with a time of failure between 750 and 800 seconds

while substitution of PER by dipentaerythritol in IF-4 provides an increase of 25% more of

time of failure up to 1050 seconds i.e. 17 minutes.

This improvement of time of failure can be correlated with the previous chemical properties

introduced in the previous chapter. X-Ray diffractograms of residues are plotted on Figure

115.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Page 170 Ph-D report J. Ciret

Figure 115 : X-Ray diffractograms of residues of IF-2, IF-3 and IF-4 after small furnace tests

Whatever the formulations, analyses have shown the formation of TiP

for a testing time

of 30 minutes. This formation is in agreement with the temperature measured on furnace

and with the chemical investigation of the previous chapter. Indeed at 800°C, TiP

formed whatever the carbon source, nevertheless diffractograms in Figure 115 show clearly

that its crystallinity are not the same in the case of IF-4 with dipentaerythritol than in the

cases of IF-2 and IF-3 with respectively PER or epoxy resin. On diffractogram of IF-4, we

observe three well-defined peaks attributed to TiP

(similarly to what we observe after

tubular furnace heat treatment at 800°C for IF-4) showing the great crystallinity of the

residue. On the contrary, for IF-2 and IF-3, the ratio signal to noise is higher even if the three

peaks of TiP

are detected (similar to what we observe after tubular furnace heat

treatment at 800°C for these two intumescent formulations). Therefore, based on

conclusions demonstrated in the previous chapter about the formation of TiP

at lower

temperature with dipentaerythritol (IF-4), we assume that titanium pyrophosphate have

been yielded earlier for fire test and we can suggest that the better efficiency observed with

IF-4 than with IF-2 and IF-3 is correlated to the chemical changes of the formulation.

Therefore, thanks to this first test according to hydrocarbon fire and UL 1709 standard, we

have highlighted a clear correlation between the chemical evolution and TiP

formation in

a first hand and the thermal efficiency and the time of failure in the other hand. It appears

that the formation at lower temperature of the crystalline structure based on titanium can

provide an additional efficiency of the coating. We suspect thus that TiP

formation

enhances fire protection because of formation of a ceramic-like protective material in char

structure, as suggested elsewhere [113, 124]. Moreover, the thermal properties of such

structure could allow a better reflection and a lower absorption of heat, these characteristics

being responsible for the bulk and surface temperature of the shield.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Ph-D report J. Ciret Page 171

Exhibiting good correlation with previous investigation about physical and chemical

properties of the intumescent formulations, the first experimental setup appears as a

reliable, fast and low cost tool to evaluate fire performance of intumescent coatings in

furnace conditions. However, similar to standard conditions, a second tool must be

developed to take into account the specificity of jet-fires that occur on offshore platforms

and their characteristics due to high velocities, such as heat flux and erosion.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Page 172 Ph-D report J. Ciret

II. Development of a fire test

mimicking jet-fire

The aim of the jet-fire test is to simulate an accidental load from a gas line rupture on an

offshore oil rig. So, similarly to what we have developed for hydrocarbon furnace test, a

small scale test mimicking jet-fire experimental settings will be beneficial to multiply the

data and to minimize the cost. Moreover, a second objective of this test will be to separate

the two phenomena that take place in a jet-fire and the two crucial parameters that we have

highlighted on the second chapter:

- A high momentum of the jet due to the velocity of the jet and flame impact on the

panel causing important pressure notably on the impact zone

- A radiative phenomenon due to the flames and providing the most part of the heat

and causing the greater temperature change.

II. 1. Experimental settings and optimisation

II. 1. a) Experimental apparatus

Tests have been firstly carried out on 10*10cm²-sized squared metallic panels coated with

1.5mm of each coating. Eventually a mesh has been included into the coating to reinforce

the structure similarly to specimen used for large-scale jet-fire test and described in the

second chapter. In a last section, we have also evaluated the performance of intumescent

formulations coated in panels with central flange, similarly to the ones used in large scale

jet-fire test.

An experimental setup (Figure 116) has been home-built using two different heating devices

in order to observe precisely the behaviours of the coating and to control independently and

separately the two main phenomena characterizing a jet-fire.

Figure 116 : Experimental setup of the labscale jet-fire test

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Ph-D report J. Ciret Page 173

In a first hand, an electrical cone heater provides pure radiative heat and in a second hand

an air jet of controlled temperature and pressure providing pure convective heat is used. The

combination of the cone heater and the air jet should then provide a mix of convective and

radiative heat at the same time with accurate controlled parameters (heat flux, pressure or

velocity of air jet, temperature of air jet). Therefore, our work starts with the determination

of the optimal setting parameters to discriminate the behaviours of the four intumescent

formulations according to the previous observations during jet-fire test exposed in the

second chapter.

II. 1. b) Determination of the optimal parameters

First experiments have been carried out to determine the appropriate experimental

conditions: heat flux of the cone heater, velocity of the air jet, time to deliver the air jet (an

early air jet can avoid the expansion of the intumescent coating). The studied parameters

are summarized in Table 11. All these parameters have been chosen in agreement with the

standard of the large scale jet-fire resistance test focusing on nozzle geometry and gas flow

of propane but also based on the literature review about jet-fire highlighting important role

of the jet length (varying with velocity) and heat radiation (taking into account with the heat

flux from conical heater).

Table 11 : Screening of the different setting parameters used in the labscale jet-fire test

Air temperature

Air speed

(Air pressure)

Heat flux of

cone heater

Delayed time to

switch on air jet

Total Time

450°C

15m/s (0.5bar)

35kW/m²

Immediately

15min

30m/s (1bar)

50kW/m²

After 150 seconds

After 300 seconds

Concerning the heat flux from the cone heater, several tests have been carried out at 35 and

50kW/m² on the four coatings. Results are presented in Figure 117. Increasing the external

heat flux of the cone does not have a great effect on the time/temperature profiles (air jet is

switched on at time zero and the air speed is fixed at 15m/s). Whatever the heat fluxes,

temperatures on the plateau (steady state zone) are similar; however the time to reach it is

as expected shorter at 50kW/m² than at 35kW/m². Similar intumescence factors (expansion)

and observations of the hole caused by the air jet are found at the two irradiances (see the

pictures on Figure 118).

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Page 174 Ph-D report J. Ciret

Figure 117 : Temperature profiles on the backside of panels coated by IF-1, IF-2, IF-3 and IF-4 in

labscale jet-fire test at two heat fluxes, 50kW/m² (full curves) and 35kW/m² (dotted curves)

Figure 118 : Char aspect after labscale jet-fire test at 35kW/m² (on the left) and at 50kW/m² (on

the right) – Example of IF-2

However, heat flux of 50kW/m² targets a forced-flaming combustion scenario that is typical

of developed fires. Consequently, we have chosen the higher irradiance (50kW/m²) to

continue the optimisation of the tests.

Typical experimental conditions are then: (i) cone heater supplying an external heat flux of

50kW/m² and (ii) air jet exhibiting a velocity of 15m/s heated at 500°C. In spite of this high

temperature, it is noteworthy that the main heat contribution is provided by the radiative

heat flux (surface temperature of the sample is equal to 730°C) while air jet provides mainly

an impacting force at the surface of the sample (momentum effect) causing damage due to

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Ph-D report J. Ciret Page 175

air speed. Indeed when we compare the profiles with and without air jet (Figure 119), the

temperatures reached are very close, except for IF-3.

Figure 119 : Temperature profiles on the backside of panels coated by IF-1, IF-2, IF-3 and IF-4 in

labscale jet-fire test according to air jet contribution, with air jet (plain curves) and without air jet

(dotted curves)

Another observation is that differentiation between the coatings is not significant. Even with

air jet, no change can be detected to differentiate the char resistance of the coatings. To

improve our protocol, we have examined different setups by varying the time when air jet is

turning on. Without air jet, the coating can expand freely, after full expansion of the

intumescent coating, the air jet is turned on and observations can be made concerning the

char resistance. Figure 120 shows the consequence of delaying the air jet. Indeed, after

switching on the air jet at 150 seconds, we observe a quick increase of the temperature on

the backside of the panel because of partial destruction of the protective coating. This

destruction is evidenced by the pictures taken after the test that show only a print when air

jet and cone heater are switching on together (Figure 121, left picture) while important

damage is observed when air jet is delayed (Figure 121, right picture).

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Page 176 Ph-D report J. Ciret

Figure 120 : Temperature profiles on the backside of panels coated by IF-1, IF-2, IF-3 and IF-4

according to time when air jet is switching on: air jet is switched on at the beginning of the test

(plain curves) or air jet is switched on 150 seconds after the beginning of the test (dotted curves)

Figure 121 : Char aspect after labscale jet-fire test without time delaying (on the left) and with time

delaying (on the right) – Example of IF-2

According to the above results, we have decided to apply a ‘delayed time’ for turning air jet

in order to improve the differentiation of the coatings. Two values of delayed time have

been selected at 150 seconds and 300 seconds. Time delaying allows the char develops

freely and then to be impacted by the air jet.

Finally, the velocity of the air jet or pressure must be determined. At 30m/s, all the coatings,

except IF-1, exhibit severe damage but differences cannot be distinguished (results are not

presented). We have thus fixed the velocity of the air jet at 15m/s. This velocity allows

observing significant difference between the coatings and provides a sufficient impact to

damage the more fragile coating without modifying the behaviour of the most robust one.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Ph-D report J. Ciret Page 177

In Table 12, we have summarized the optimal parameters permitting to discriminate the

behaviour of the four intumescent coatings.

Table 12 : Optimized configurations used to evaluate the four coatings with the labscale jet-fire

test

Air

temperature

Air speed

Heat flux of

cone heater

Delayed time to

switch on air jet

Total Time

configuration

450°C

15m/s

50 kW/m²

After 300 seconds

15min

configuration

450°C

15m/s

50 kW/m²

After 150 seconds

15min

In order to be as clear as possible, we propose to present only the results obtained with the

configuration both when coatings are applied alone on square panels and when mesh is

included to reinforce the intumesced structure. Others results obtained with the 2

configuration can be found in appendix.

II. 2. Tests on squared panels coated by

intumescent formulations

The four coatings have been evaluated using squared panels according to the optimisation

done above. Before presenting the results and thanks to previous analyses notably

determination of the mechanical resistance using our rheological approach, the behaviour of

our coatings can be summarized as follows:

- IF-1 is known to be a robust coating exhibiting a dense and compact char. This

coating is our reference in terms of resistance facing to jet and the preliminary

experiments confirms this information.

- The three other coatings, IF-2, IF-3 and IF-4, are coatings with relatively low

resistance facing to strain, jet and impact.

Temperature profiles obtained with the four coatings on the backside of the panel are

shown in Figure 122.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Page 178 Ph-D report J. Ciret

Figure 122 : Temperature profiles obtained in labscale jet-fire test for IF-1, IF-2, IF-3 and IF-4 (heat

flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds)

Analysing the temperature profiles of the four coatings, we can suggest two behaviours. A

type of coating is sensitive to air jet because IF-2 and IF-3 exhibit increasing slopes on the

time/temperature profiles when turning on air jet. The other types of coating, IF-1 and IF-4,

exhibit smooth time/temperature profiles. It can be explained by the remaining charred

layer on the panel while chunks of intumescent char are blasted off by air jet leaving

unprotected steel at the location of the impact zone in the case of IF-2 and IF-3.

Careful investigation of the char after performing the test provides additional information.

IF-1 char (Figure 123) does not present any trace of the air jet impact. The swelling is

relatively low (0.7cm – 450%) but the char is very dense and it provides a homogeneous

protection characterized by no temperature gradient on the backside.

Figure 123 : Picture of IF-1 char after labscale jet-fire test and thermal image for labscale jet-fire

test (heat flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds)

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Ph-D report J. Ciret Page 179

Concerning the ‘fragile’ coatings (IF-2 and IF-3), damage on IF-2 (Figure 124) caused by air jet

is very significant. After test, different concentric circular zones from the centre to the edge

are observed. At the centre, a 1.2cm-diameter zone is almost uncoated which explains the

increase of the temperature and the ‘halo’ in the middle of the IR image (Figure 124).

Swelling up to 1.3cm with eroded char is observed on the edge of the impact zone.

Figure 124 : Picture of IF-2 char after labscale jet-fire test and thermal image for labscale jet-fire

test (heat flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds)

IF-3 is known to be the most fragile coating. After turning on the air jet, IF-3 exhibits

temperature close to that obtained with virgin panel (Figure 122). By observing the char

after test (Figure 125), the reason is self-explanatory. Indeed due to air jet impact, the

structure is totally destructed and lets a large unprotected zone. In this zone, a quick

increase of temperature is measured since this temperature reaches that of virgin panel.

Obviously, no swelling measurement is available. This behaviour can be explained both by a

bad adhesion and a bad mechanical resistance to air jet. The large zone of unprotected steel

should be compared to the observations done after the jet-fire test where IF-3 failed for the

same reasons. In fact, under strain caused by a jet, the intumescing structure does not

adhere to the metallic substrate and is ejected by the jet.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Page 180 Ph-D report J. Ciret

Figure 125 : Picture of IF-3 char after labscale jet-fire test and thermal image for labscale jet-fire

test (heat flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds)

Concerning IF-4, the difference with IF-3 is very significant. Firstly, on the temperature

profiles (Figure 122) we can observe that the temperature does not change when the air jet

is switched on. This good insulation is confirmed by the char aspect (Figure 126). Indeed, we

can see that a stronger protective layer (thickness = 1cm) remains stuck on the panel all

along the test. Consequently, the protection is homogeneous and the temperature on the

backside of the panel remains lower than 250°C. However, it is noteworthy that the upper

part of the intumescent structure is partially damaged by the air jet although it provides

protection.

Figure 126 : Picture of IF-4 char after labscale jet-fire test and thermal image for labscale jet-fire

test (heat flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds)

As a conclusion with this experimental configuration, the formation of a cohesive structure

and the superior impact resistance of IF-1 are confirmed. Differences between the

formulations can be commented as follows:

- IF-1 does not exhibit any damage caused by air jet impact. Nevertheless, expansion

is relatively low.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Ph-D report J. Ciret Page 181

- The intumescent structure developed from IF-3 is completely destructed while that

of IF-2 is partially destroyed at the location of impact and the rest of the structure

is kept on the panel.

- Substitution of PER by dipentaerythritol provides a significant improvement of the

resistance of the char. Containing dipentaerythritol in IF-4, a residual protective

layer is maintained on the panel even if the upper part is partially damaged while

with PER on IF-3, all the coating is pulled out.

II. 3. Tests on squared panels coated by

intumescent formulations reinforced with mesh

The use of mesh in intumescent coating provides an additional reinforcement of the barrier

formed when burning. The efficiency of this reinforcement has been demonstrated for large

scale jet-fire resistance test with IF-2 and IF-3, particularly because of their low mechanical

resistance whereas its effect is not significant for a robust coating as IF-1. Therefore, to

improve the mechanical resistance and to keep a sufficient insulation, mesh has been

incorporated inside the three “fragile” coatings. Except this incorporation, experimental

configuration remains similar and the temperature profiles are shown on Figure 127.

Figure 127 : Temperature profiles obtained in labscale jet-fire test for IF-2, IF-3 and IF-4 reinforced

by mesh (heat flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds)

The inclusion of mesh improves the resistance of the coating facing to air jet. Although the

temperature profiles of IF-2 and IF-3 still exhibit an increasing slope directly link with the

switching on of the air jet, this phenomenon is attenuated and almost disappeared for IF-2

thanks to the reinforcement provided by mesh.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Page 182 Ph-D report J. Ciret

Concerning IF-4 coating, knowing that its behaviours without mesh are relatively good and

damage does not appear as dramatic, effect of mesh are not significant and have just

confirmed a better resistance of IF-4 facing to air jet than IF-2 and IF-3. No temperature

variation has been detected on its temperature profile. This difference of behaviours

between IF-3 and IF-4 is much more highlighted by the similar evolution of the temperature

before onset of the air jet (curves are superimposed) while a gap appears after applying the

air jet.

Therefore, mesh provides a great improvement for the resistance of the coating. To confirm

information provided by temperature profile, an investigation of the char aspects after test

is shown on Figure 128. First observations of the chars suggest that a higher amount of

residues remain on the steel plate than with the previous settings. Incorporating mesh in the

coating permits to keep the intumescent structure on the panels and to protect them.

IF-2

IF-3

IF-4

Figure 128 : Pictures of the char obtained with intumescent coatings reinforced by mesh after

labscale jet-fire test (heat flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds)

IF-2 develops an intumescent structure with a swelling of 1.7cm with mesh. On this

structure, a 1.3cm diameter central hole is formed by the impact of the air jet which does

not reach the panel. A residual dense layer remains stuck on the panel, this layer is kept

thanks to mesh and its presence is not observed for previous experiments carried out

without mesh.

Char developed from IF-3 exhibits large differences with and without mesh after the test.

Mesh keeps the structure on the panel except on the impact zone where a 1.9cm-diameter

zone of non protected steel is formed which can explain the increase of temperature.

However, by comparison with first experiments we can note that panels remains mainly

protected. Of course thermal conductivity of steel and local weakness of the coating cause a

temperature profile exhibiting quick increase when air jet impacts.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Ph-D report J. Ciret Page 183

Finally, the behaviour of IF-4 is similar with and without mesh. Indeed, the temperature

profiles are similar reaching 250°C in the steady state zone. Similar intumescence factor and

damages are observed. No effect of mesh must however be put into perspective with the

great behaviour exhibited without mesh similarly to IF-1.

Using the experimental parameters defined in the beginning of this section, the benefit of

mesh is shown. It permits to enhance the impact char resistance and the thermal

insulation keeping char at the surface of the substrate. Nevertheless, mesh is not sufficient

to maintain the structure of IF-3 on the impact zone.

II. 4. Tests on flanged panels coated by

intumescent formulations

To mimic the specific setup of the panel during jet-fire test, our panels have been modified

accordingly. The panels have been designed with a flange located in the middle of the steel

plate (Figure 129). This geometry could modify the swelling of the systems especially at the

edge between the flange and the back faces. Holes have been drilled right through the

flange and thermocouples have been embedded in the flange so that the thermocouple is

located at the interface coating/panel during the test. The panel undergoes an external flux

of 50kW/m

and the jet is turned on at 300 seconds after the beginning of the test. Jet

impacts at level of the second hole i.e. in the middle of the flange. However, with this

geometry, it is not possible to measure temperature profiles and to observe thermal

cartography by IR camera, only measurements with thermocouples at three positions is

possible for measuring time/Temperature profiles.

Figure 129 : Experimental setup to use flanged panels in labscale jet-fire test

Using this specific design IF-1 exhibits superior impact resistance against air jet. The

temperature profiles taken in three points (Figure 130) show a constant increase of

temperature except when the air jet is turned on. At this time (300 seconds), we observe a

slight decrease of the temperature and then its stabilization. We have attributed this change

to the variation of the temperature and the contribution of the air jet cooling down the

surface of the coating (convective effect) as previously discussed.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Page 184 Ph-D report J. Ciret

Figure 130 : Temperature profiles obtained in labscale jet-fire test for IF-1 coated on flanged panel

(heat flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds)

Before onset of the air jet, temperature evolutions are very smooth. Radiative effect causes

a temperature gradient from the bottom to the top of the flange. However, we can detect

one inflexion point on the temperature profiles when the air jet is switched on. These

variations cannot really be explained by observing the samples after test. Indeed, visual

investigation of the char (Figure 131) after the test confirms the good resistance of the

coating facing to heat and to impact. No damage is detected on the char surface, a compact

and dense char with some bubbles coats the whole panel.

Figure 131 : Char obtained from IF-1 after labscale jet-fire test on flanged panel (heat flux of

50kW/m², air speed of 15m/s, delayed time of 300 seconds)

Concerning the three other formulations, the intumescent chars are affected facing to jet,

notably on the impact zone, as shown on Figure 132.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Ph-D report J. Ciret Page 185

IF-2

IF-3

IF-4

Figure 132 : Pictures of chars obtained from IF-2, IF-3 and IF-4 after labscale jet-fire test on flanged

panel (heat flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds)

For IF-2 (Figure 132), the central part of the flange exposed to the jet is no longer coated by

intumescent structure at the end of the test while the upper and lower places are still

protected. The surface adjacent to the flange is also damaged by air jet because we can

observe on the left part a partial blast of the coating caused by air flow. However, contrary

to the experiments carried out with flat panels, temperature profiles obtained with

thermocouples on the interface metal/coating does not show this destruction on the flange.

Indeed, in agreement with the observation done on IF-2 char, we expected to observe

significant increase of temperature, notably on the impact zone where coating is destructed.

Instead of this, IF-2 profiles (Figure 133) exhibit curves that are initially superimposed

representing the uniform evolution of the coating facing to heat flux from cone heater. But

when air jet is switched on (after 300 seconds), temperature increases slightly on the

boundaries and more strongly on the impact zone. However, the differences are not

significant to make conclusions.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Page 186 Ph-D report J. Ciret

Figure 133 : Temperature profiles obtained in labscale jet-fire test for IF-2 coated on flanged panel

(heat flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds)

For IF-3 (Figure 132), picture shows that the part of the flange where the jet impacts is

totally unprotected. Furthermore, we can also detect damage at the surface adjacent to the

flange. Jet impacts the flange and damages a central band perpendicular to the flange by

erosive effect. On this band, the residual layer is very thin (not measurable) while on top and

bottom of the side face a swelling of 0.7cm is observed and usually sufficient to provide

protection. On temperature profiles, the effect of the air jet can be detected. Indeed, initially

superimposed until turning on the air jet, the three curves exhibit different temperature

changes. As soon as the air jet is switched on, the temperature on the impact zone increases

rapidly while on the edges of the flange it continues to increase smoothly. To highlight the

larger damage on IF-3 than on IF-2, we can also remark that higher temperatures are

reached on temperature profiles of IF-3 (> 300°C) than those of IF-2. Thus these conclusions

have been stated investigating the char aspect and have been confirmed by the thermal

evolutions, even if temperature gaps are very slight and only trend can be proposed.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Ph-D report J. Ciret Page 187

Figure 134 : Temperature profiles obtained in labscale jet-fire test for IF-3 coated on flanged panel

(heat flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds)

For IF-4 (Figure 132), the coating is damaged on the flange where the air jet impacts the

char. Nevertheless, the char developed on the back faces exhibits an enhanced resistance

facing to air jet. In terms of swelling, we can measure a swelling on the side face of 1.5cm at

the highest point. About the temperature evolution (Figure 135), the temperatures increase

similarly until turning on the air jet. We then observe a higher increase of temperature at the

impact zone than at the top and at the bottom in agreement with major location of the

damage. However, the differences are small and prevent any further discussions.

Figure 135 : Temperature profiles obtained in labscale jet-fire test for IF-4 coated on flanged panel

(heat flux of 50kW/m², air speed of 15m/s, delayed time of 300 seconds)

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Page 188 Ph-D report J. Ciret

To conclude this part and this experimental settings about the use of this specific design of

panel, temperature measurements do not permit any significant discriminations between

the formulations. Our first explanation is the inappropriate location of the thermocouples.

Steel has a high heat conductivity which permits a rapid distribution of the temperature in all

directions (isotropic heat conductivity). Accepting this, the rapid increase of the temperature

on the impact zone (because of the partial destruction of the intumescent char) is

propagated along the flange and a temperature gradient cannot be detected. The design and

the techniques used to measure the temperatures have thus to be improved in order to

obtain results mimicking jet-fire test in terms of geometrical aspect. Nevertheless, we can

observe some consequences of the panel design on the expansion of the intumescent

coatings, showing less homogeneous swelling along the panels due to presence of flange and

complexity of the geometry. Others conclusions can be established about the thermal

insulative provided by coatings and the temperature exhibited by coated panels where IF-2

seems to better insulate metallic panels than IF-3 and IF-4 in spite of superficial damage:

final temperature measured by thermocouples are lower in the case of IF-2 than in the two

others cases.

II. 5. Conclusions and discussion about the benefit

of a new experimental tool

To underline the importance of the chemical difference and to understand the behaviours of

the different coatings facing to jet-fire, we have undertaken to develop a labscale air-jet test.

After optimizing the experimental parameters, a setup was proposed to evaluate the

different coatings and their resulting resistance facing to impact.

The first conclusions and the correlations with large-scale results are very promising. Indeed

if IF-1 appears as a robust coating without equivalent in terms of resistance, the three other

coatings suggest significant and interesting differences. For IF-2, its mechanical resistance is

poor. Localized on the impact zone, damage caused by the air jet provides a large insulation

loss. This low resistance can be improved by incorporating mesh inside coating to avoid

critical damage of the intumescent structure. However, mesh does not completely solve the

insulation loss at the impact. For IF-3, its behaviour is the worst. Indeed facing to air jet, the

intumescent shield is totally blasted off while this coating flows away of the panel due to its

low viscosity during the jet-fire test. The intumescent structure does not resist to impact and

is pushed from the centre to the edge because it does not adhere to panel. Once again,

mesh provides significant improvement to this coating, allowing to keep the structure on the

whole panel except at the impact zone. The best improvement is provided by the

substitution of PER by dipentaerythritol. Indeed using dipentaerythritol in IF-4, a residual

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Ph-D report J. Ciret Page 189

protective layer is maintained on the panel and allow a homogeneous protection of the

substrate, as efficient as IF-1 in terms of insulation and apparent swelling.

So the conclusions of this section are drawn as follows:

- Whatever the conditions, IF-1 exhibits the same behaviour developing a hard and

compact char resistant to air jet.

- Concerning IF-2 and IF-3, correlations have been found between our labscale air jet

test and the jet-fire test about the low mechanical resistance and the destruction of

the structure.

- Substitution of PER by dipentaerythritol causes modification of the behaviour of

the coating. IF-3 and IF-4 react differently facing to impact, with total destruction

for the first formulation while a robust protective layer remains to protect panels

with the second one.

- Mesh reinforces significantly the coatings, notably for IF-2 and IF-3 and appears as a

solution to pass severe test with impact of flames.

Correlations with the large scale jet-fire test can be sum up in the Table 13.

Table 13 : Evaluation and correlation of the behaviours of coatings

IF-2

IF-3

IF-4

Labscale test

Weakness on the front of

the jet, impact causes

destruction on the centre

of the panel

Reaction of the char

facing to the air flow,

structure is destroyed all

along the surface due to

blast

Dipentaerythritol

provides appearance of a

residual layer resistant to

impact and protecting

panel

Large-scale jet-

fire test

Failure due to low

resistance of the char

after 35min

Failure due to low

viscosity, flowing of the

char on the back face

after 13min

The char peels back as

the adhesion is very poor,

failure after 51min

Therefore, thanks to this labscale tool, we are able to observe and to explain the four

different behaviours exhibited by the intumescent coatings. Some of these behaviours are

self explanatory, similarly to large-scale jet-fire test and highlight the weaknesses of the

formulations. Indeed to correlate these two scales, we can find some common points and

similar observation about the intumescent formulations as the low swelling abilities of IF-1

even if its robustness allow a great fire performance for 60 minutes, or the low mechanical

resistance of IF-2 and of IF-3, causing their test failures at both two scales. But the main link

that we can establish between the two tests is mainly due to visual investigation of the char

and thermal consequence that damage can cause on the insulative efficiencies of the

coatings. Effectively, if main heat is provided by cone heater and radiation, the impingement

of an air jet on a flat plate can reproduce the erosive effect on the fire protection material

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Chapter V: Development of an efficient labscale test mimicking large scale jet-fire test

Page 190 Ph-D report J. Ciret

and can damage strongly the most fragile char as observed when IF-3 is completely blasted

by air jet from the beginning of the test.

To conclude about the labscale test, even if no quantitative information have been obtained

after this test in terms of fire-resistance rating, experiments and observations permit to

obtain reliable information about the behaviour of the different coatings. Some aspects are

still different between the two scales tests but we manage to establish some correlations. In

addition with the first tool developed in laboratory to mimic industrial furnaces, this labscale

test mimicking the characteristics of the jet-fire test constitutes a new tool to make ‘high

throughput’ screening of new formulations.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

General Conclusions

Ph-D report J. Ciret Page 191

General Conclusions

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

General Conclusions

Ph-D report J. Ciret Page 193

This Ph-D work was based on the study and explanations of the behaviours of four epoxy-

based thick film intumescent coating, applied on offshore platforms and which has to resist

in very strong conditions, such as jet-fires.

Our studying methods have investigated three different approaches about high fire risks on

offshore locations. These approaches are all based on a starting assessment introduced on

the chapter II about the behaviours of our four studied coatings for large scale jet-fire

resistance test. Each coatings exhibit behaviour that highlights their weaknesses facing to

impact of jet-fire or heat generated by such phenomenon. Clearly, intumescent formulation

1 produces a very hard and not flexible char that does not expand very much. As it does not

expand enough, it fails at the hottest part of the box due to lack of insulation. On the

contrary intumescent formulation 2 failure occurs on flange edge where the flame impacts.

The char evolved from this product seems not to be sufficiently resistant and is destroyed by

the high momentum delivered by the flame impact. Reinforcement provided by inclusion of

carbon and glass fibre mesh system before testing can permit a great improvement of the

mechanical resistance. Then intumescent formulation 3, a compromise between the two

first coatings, appears as strong enough to resist on edge of flange and exhibit an expanded

insulative char. However its failure occurs on the back face of the box: an area where neither

intumescent formulation 1 nor intumescent formulation 2 fail. Finally, intumescent

formulation 4, based on a slight modification of IF-3, introduces also a new type of failure

caused by appearance of large voids at the steel/coating interface.

Regarding these four different failures highlighting mechanical resistance and visco-elastic

properties, we have firstly investigated the physical properties of the intumescent coatings.

Differences in terms of viscosity are shown by the use of different carbon sources. Indeed

effect of the carbon sources (epoxy resin alone or with PER or with dipentaerythritol) are

clearly determined by comparison between IF-2, IF-3 and IF-4. PER causes the shift of the

viscosity decrease at lower temperature (270°C) and consequently swelling and

intumescence process begin also at lower temperature for IF-3. Substitution of PER by

dipentaerythritol (IF-4) allows to modify the behaviour of IF-3. These results must be linked

with the results of jet-fire tests because the IF-3 failure and the flowing of char observed for

jet-fire test can be partially explained by the visco-elastic properties. Indeed we can suggest

that phenomena of flowing that occurs on the back faces can be the consequence of a too

low viscosity of the melt phase, as observed at 270°C in our rheological experiments.

Nevertheless, this flowing disappears for jet-fire test when PER is substituted by

dipentaerythritol. Secondly, the mechanical resistance is also suspected for explaining the

failure of IF-2 and IF-3 without mesh. Our experiments allow to conclude that IF-1 is the

most robust and dense char, not sensitive to any strain while the three other coatings do not

exhibit any resistance facing to the less force applied on the top of the char. This sensitivity

to strain confirms thus results observed of IF-2 and IF-3 for jet-fire test resistance that fail in

the impact zone where pressure is maximal and no coating resist to impact. Therefore

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

General Conclusions

Page 194 Ph-D report J. Ciret

physical approach provides us interesting information about the intumescent process and

main conclusions target the role of the carbon sources in intumescent formulations.

Strongly linked with the efficiency of the char of these mixtures, these properties have been

underlined by physical approach but could be also explained by the investigation of the

mechanism of degradation that takes place inside these formulations. By analysing several

residues from intumescent coatings at different characteristic temperatures mechanism of

degradation as a function of the char former has been proposed. Indeed, investigations have

been focused on the effect of the different carbon sources used on the intumescent

formulations (epoxy resin in IF-2, epoxy resin and pentaerythritol on IF-3, epoxy resin and

dipentaerythritol in IF-4). The chemical interactions that occur between the main

components have then been investigated by thermo-gravimetric analyses, X-Ray diffraction

and solid state NMR. Firstly, we have shown the effect of the carbon source on the chemical

changes. We have demonstrated that epoxy resin could play the role of carbon source on IF-

2 if no other component is present. However when PER or dipentaerythritol is added, no

reaction between epoxy resin and acid source (APP) is detected. In a first step APP and the

carbon source react together to yield to the development of a protective shield but also

intermediate products containing phosphorus residues. Then, whatever the formulation, the

formation of TiP

is observed at high temperature: phosphorus residues react with TiO

yield to a crystalline structure of TiP

. This crystallized compound is reported in literature

to enhance mechanical properties and flame retardant performance. However, the

temperature range at which it is formed is different. Indeed, TiP

can be detected from

800°C for IF-3 whereas it is only clearly observable from 450°C for IF-4. For IF-2 without

additional carbon source added to the formulation, epoxy resin alone plays a similar role

leading to the formation of TiP

at around 800°C. Consequently, it may be proposed that

the substitution of PER by dipentaerythritol allows to increase the reactivity of APP with the

carbon source and thus enables reaction to occur at lower temperature, notably the

formation of TiP

. The formation of TiP

have been shown for furnace jet-fire and linked

between fire performance and chemical pathway are proposed focusing on the formation of

TiP

and its benefit to thermal efficiency of IF-4 compared to IF-3.

Finally, knowing all these properties and chemical changes exhibited by our four

intumescent formulations and explaining their fire performance, the development of an

experimental and reliable tool has been reported. This tool could provide information about

the development and the efficiencies of the intumescent structure. These tools could

constitute a predictive approach to determine the fire behaviour of the evaluated

formulations. Moreover, thanks to these experimental tools, a low cost and fast test would

permit rapid and reliable results of a large number of different coating formulations in good

correlation with the results obtained in large scale standardized fire tests. First test

mimicking hydrocarbon fire standard, we have highlighted a clear correlation between the

chemical pathway and the TiP

formation in a first hand and the thermal efficiency and

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

General Conclusions

Ph-D report J. Ciret Page 195

the time of failure in the other hand. The second test tries to create a similar phenomenon

that takes place for large scale jet-fire. Using a radiative heater and air jet with high

momentum, the setup permits to focus on the two key parameters of hazardous jet-fire.

Thanks to this experimental setup and to optimized settings, we have success to find again

similar behaviours of the four coatings and to explain them. Some conclusions can thus be

drawn to sum up these behaviours. Whatever the conditions, IF-1 exhibits the same

behaviour developing a hard but low expanded char resisting to jet-fire. Concerning IF-2 and

IF-3, correlations have been found between our labscale air jet test and the jet-fire test

about the low mechanical resistance and the destruction of the structure letting

unprotected the metallic substrate. Mesh can however reinforce significantly the

intumesced structure and appears as a solution to pass severe test with impact of flames.

Finally, the substitution of PER by dipentaerythritol allows to modify behaviour of the

coatings. On the contrary, from IF-3, IF-4 forms a robust protective layer that remains to

protect panels and avoids the increase of the temperature.

Therefore, although the labscale jet-fire test is not completely optimised to detect behaviour

of the coating with the specific geometry of the large scale test (with a central flange), the

combination of the fundamental conclusions about the physical properties and the

degradation pathways of the different formulations and these experimental tools permit to

highlight the weaknesses of the intumescent coatings and to propose solutions to improve

this defaults. Indeed substitution of PER by dipentaerythritol and use of mesh appear as two

solutions developed in this report, evaluated with similar experiments and exhibited

satisfying results to enhance the fire performance of IF-2 and IF-3.

Finally, to end this Ph-D work, we can consider some outlooks. Indeed, thanks to

investigation about role of carbon sources, we have highlighted its influence on the chemical

degradation pathway and temperature at which it occurs, using epoxy resin either alone, or

with PER, or with dipentaerythritol. Our conclusions are based on chemical analyses at

different specific temperatures to determine the changes but it would be interesting to

obtain quantitative information about the kinetic parameters. Indeed, the kinetic of

degradation is a major concern and could be simulated using TGA device. A thermo-kinetic

model could be determined in order to calculate kinetic parameters. This study should allow

to refine the degradation mechanism of the material, notably the reaction between APP and

the carbon source. Our study by chemical analyses has shown differences in terms of

reactivity between APP and the different carbon sources, this approach by modelling could

quantify this reactivity and compute the activation energy of each reaction. It would be so

possible to confirm the better reactivity of APP with dipentaerythritol and the formation of

the protective structure at lower temperature. To complete this study, new carbon sources

(as tripentaerythritol) could be added to the main ingredients in order to screen the

reactivity of APP with carbons sources.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

General Conclusions

Page 196 Ph-D report J. Ciret

Second, to go further to explain the effect of mesh and the improvement viewable for

labscale jet-fire test, we could propose a specific study highlighting its role for severe test as

jet-fire resistance one. Indeed, the use of mesh in the intumescent coating was an aspect

quickly broached in this report but not really investigated except by observing its

consequence on temperature profiles. However, in hot and severe conditions, hybrid carbon

and glass fibre mesh show an interesting behaviour and the observation made after the jet

fire tests as well as the composition of the mesh led us to suspect an influence of the oxygen

on the thermal degradation of the mesh. Indeed, if the glass fibres do not degrade until

1000°C whatever the atmosphere, the carbon fibres are greatly sensitive to oxygen. Mesh

can so remain in the zone where the oxygen level is low (pyrolysis conditions) whereas if

oxygen is present the mesh completely degrades. In both cases, the glass fibres will be

melted. Taking into account this information, this difference of behaviours according to the

oxygen level can be a clue to determine the zone of pyrolysis and zone of combustion on the

test panels during the jet-fire resistance test.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Appendix: Experimental results obtained with 2

configuration

Ph-D report J. Ciret Page 197

Appendix: Experimental results

obtained with 2

configuration

In order to better discriminate the coatings, we have used the same protocol as described in

the last chapter of this report except that the delayed time of the air jet was reduced at 150

seconds. By decreasing the time delaying, we hope to observe a lower expanded structure

and consequently a lower resistance of the char. So we hope to better differentiate coatings

and to observe significant differences between the three fragile coatings (IF-2, IF-3 and IF-4),

knowing that IF-1 exhibit the same behaviour whatever the settings used to test it.

configuration without mesh

Firstly, without mesh, we have observed similar results and similar behaviours at 150

seconds and at 300 seconds (Figure 136 compare to Figure 122). The temperature profiles

allow to highlight the potential local destruction of the coating. As observed previously, IF-1

and IF-4 exhibit smooth time/temperature profiles underlining good toughness of the char.

On the contrary, temperature change can already be observed on temperature profiles of IF-

2 and IF-3 when air jet is turning on. At the end of the experiment, IF-2 exhibits temperature

close to virgin panels, confirming that destruction of the structure and stop of the

protection. About IF-2, the consequences of air jet are smaller, but temperature variation

points to potential local damages.

Figure 136 : Temperature profiles obtained in labscale jet-fire test for IF-1, IF-2, IF-3 and IF-4 (heat

flux of 50kW/m², air speed of 15m/s, delayed time of 150 seconds)

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Appendix: Experimental results obtained with 2

configuration

Page 198 Ph-D report J. Ciret

Picture analyses confirm all the information obtained by temperature profiles. Once again,

behaviour of IF-1 is not modified whatever the experimental configuration: the coating

exhibited a low expanded char without sign of any strain.

Then concerning the three other coatings, a good agreement can also be stated between the

temperature profiles and the pictures exposed on Figure 137.

IF-2

IF-3

IF-4

Figure 137 : Picture of chars after labscale jet-fire test and thermal image for labscale jet-fire test

(heat flux of 50kW/m², air speed of 15m/s, delayed time of 150 seconds)

Indeed, we can observe severe damage on structure issued of IF-2 and IF-3. Impact cause

hole on the surface of IF-2 and destruction of the structure, nevertheless this destruction is

not sufficient to observe a total destruction up to the substrate. This explains the gap

already viewable between virgin panel and IF-2 coated panel. Damage are more important

on IF-3 char, structure is totally destroyed from the centre to the edge in a radius of 3cm.

The intumesced structure is pulled out, letting a large non-protected zone where

temperature increases. Finally, for IF-4, an upper fragile intumescent layer that is destructed

and pulled out by air jet, but a lower layer that resists to the jet continues to protect the

panels all along the surface. With a low thickness (0.7cm), IF-4 exhibits behaviour and

resistance close to IF-1, with a low expanded layer avoiding exposition of the panel to heat

and fluxes.

Therefore, this 2

configuration confirms the properties and the differences of each coating.

The resistance of IF-2 and IF-3 and the appearance of a residual layer on IF-4 have been the

main conclusions provided by this new series of experiments.

configuration with mesh

A last series of experiments have been carried out. The delayed time is still reduced at 150

seconds and inclusion of mesh to improve the holding of the char facing to air jet. No new

information have been obtained, however temperature profiles (Figure 138) and pictures

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Appendix: Experimental results obtained with 2

configuration

Ph-D report J. Ciret Page 199

describing the char aspect after test (Figure 139) confirm the previous conclusions and

illustrate the behaviours of the coatings and the role of the mesh.

Figure 138 : Temperature profiles obtained in labscale jet-fire test for IF-2, IF-3 and IF-4 reinforced

by mesh (heat flux of 50kW/m², air speed of 15m/s, delayed time of 150 seconds)

Indeed, by observing temperature profiles, an improvement of the coating resistance facing

to air jet is exhibited thanks to inclusion of mesh. The quick variation of temperature

(viewable usually when air jet is switching on) is less significant for IF-2 and IF-3. Their

behaviours are still modified due to air jet but IF-3 is now clearly differentiated to virgin

panels: residual material remains stuck to protect substrate.

IF-2

IF-3

IF-4

Figure 139 : Pictures of the char obtained with intumescent coatings reinforced by mesh after

labscale jet-fire test (heat flux of 50kW/m², air speed of 15m/s, delayed time of 150 seconds)

Char aspects are consistent with the temperature variations. All damages are superficial

without reaching the panels and cancelling the thermal protection. Nevertheless, we

continue to observe print where jet impacts: pressure caused by air jet and mesh limit the

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Appendix: Experimental results obtained with 2

configuration

Page 200 Ph-D report J. Ciret

swelling on the central point. Swelling on the edge is 8mm while mesh on the central point

limits the development but maintains the lower layer. On IF-3, a maximal swelling is

measured at 1.6cm and mesh is positioned at 0.6cm from the panel. So it remains a residual

layer on the whole panel to protect it. Finally on IF-4, a very low swelling (0.7cm) is viewable

all along the surface without mark of any impact or mesh undiscovered. Large blocks of

materials, initially the upper layer of the structure, are however pulled out.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Bibliography

Ph-D report J. Ciret Page 201

Bibliography

[1] Recent Steel market developments, in Seventh OECD High-Level Meeting on Steel

Issues, 28 June 2004, Paris.

[2] S. Melnick, Structural Steel Brings Environmental Advantages to Construction,

American Institute of Steel Construction, http://www.prleap.com/pr/33400.

[3] F.S. Kelly and W. Sha, A comparison of the mechanical properties of fire-resistant and

S275 structural steels, Journal of Constructional Steel Research, 1999, 50(3): p. 223-

233.

[4] Fourth Revisions and Errata: National Building Code, 2002, National Research Council

Canada.

[5] M. Haller and L.G. Cajot, Fire resistance of steel structures, Arcelor LCS Research

Report, 2006.

[6] R. Pula, F.I. Khan, B. Veitch and P.R. Amyotte, Revised fire consequence models for

offshore quantitative risk assessment, Journal of Loss Prevention in the Process

Industries, 2005, 18(4-6): p. 443-454.

[7] V.P. Molchanov, A.N. Giletich, A.A. Makeev and Y.N. Shebeko, Fire safety of offshore

stationary steelproof oil-producing platforms, Neftyanoe Khozyaistvo - Oil Industry,

2004(9): p. 82-87.

[8] J. Krueger and D. Smith, A practical approach to fire hazard analysis for offshore

structures, Journal of Hazardous Materials, 2003, 104(1-3): p. 107-122.

[9] P.J. Rew, W.G. Hulbert and D.M. Deaves, Modelling of thermal radiation from

external hydrocarbon pool fires, Process Safety and Environmental Protection, 1997,

75(2): p. 81-89.

[10] M.J. Pritchard and T.M. Binding, A new approach for predicting thermal radiation

levels from hydrocarbon pool fires, in Institution of Chemical Engineers Symposium

Series, 1992, Manchester.

[11] A.D. Johnson, Model for predicting thermal radiation hazards from large-scale LNG

pool fires, in Institution of Chemical Engineers Symposium Series, 1992, Manchester.

[12] T. Roberts, A. Gosse and S. Hawksworth, Thermal radiation from fireballs on failure of

liquefied petroleum gas storage vessels, Process Safety and Environmental

Protection: Transactions of the Institution of Chemical Engineers, Part B, 2000, 78(3):

p. 184-192.

[13] OTI 92596: Oil and gas fires: characteristics and impact, 1992, HSE Offshore

Technology Report.

[14] OTI 92597: Behaviour of oil and gas fires in the presence of confinement and

obstacles, 1992, HSE Offshore Technology Report.

[15] M. Gomez-Mares, L. Zarate and J. Casal, Jet fires and the domino effect, Fire Safety

Journal, 2008, 43(8): p. 583-588.

[16] The Honourable Lord Cullen, The Public Inquiry into the Piper Alpha Disaster, H.S.

Office, 1990.

[17] K. Sonju and J. Hustad, An experimental study of turbulent jet diffusion flames,

Norwegian maritime research, 1984, 12(4): p. 2-11.

[18] Recommended Practice 521: Pressure-relieving and depressuring systems. - 5th

edition, 2007, American Petroleum Institute.

[19] R. McMurray, Flare radiation estimated, Hydrocarbon processing, 1981: p. 175-181.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Bibliography

Page 202 Ph-D report J. Ciret

[20] A.L. Suris, E.V. Flankin and S.N. Shorin, Length of free diffusion flames, Combustion,

Explosion, and Shock Waves, 1978, 13(4): p. 459-462.

[21] B.J. McCaffrey, Momentum diffusion flame characteristics and the effects of water

spray, Combustion science and technology, 1989, 63(4-6): p. 315-335.

[22] W.R. Hawthorne, D.C. Weddell and H.C. Hottel, Mixing and combustion in turbulent

gas jets, in International symposium on Combustion, 1998, Pittsburgh.

[23] H.A. Becker and S. Yamazaki, Entrainment, momentum flux and temperature in

vertical free turbulent diffusion flames, Combustion and Flame, 1978, 33(C): p. 123-

149.

[24] G.T. Kalghatgi, Blow-out stability of gaseous jet diffusion flames I. In still air,

Combustion science and technology, 1981, 26(5-6): p. 233-239.

[25] GexCon UK Ltd, Jet Fire - Flame shape, Available from:

http://www.gexcon.com/calculators/new/flame_shape.php (in 2009).

[26] H.G. Wertenbach, Spread of flames on cylindrical tanks for hydrocarbon fluids, Gas

and Erdgas, 1971, 112(8).

[27] B.J. Lowesmith, G. Hankinson, M.R. Acton and G. Chamberlain, An overview of the

nature of hydrocarbon jet fire hazards in the oil and gas industry and a simplified

approach to assessing the hazards, Process Safety and Environmental Protection,

2007, 85(3 B): p. 207-220.

[28] G.A. Chamberlain, Developments in design methods for predicting thermal radiation

from flares, Chemical Engineering Research and Design, 1987, 65(4): p. 299-309.

[29] L.T. Cowley and M.J. Pritchard, Large-scale natural gas and LPG jet fires and thermal

impact on structures, in 14th International LNG/LPG Conference - Gastech 90, 1990,

Amsterdam.

[30] J. Hustad and K. Sonju, Heat transfer to pipes submerged in turbulent jet diffusion

flames, in Eurotherm N°17, 1990, Cascais (Portugal).

[31] Saga Petroleum, Jet-fire explosion test for passive fire protection, The design

specification for the Snorre field, 1987.

[32] L.C. Shirvill, Efficacy of water spray protection against propane and butane jet fires

impinging on LPG storage tanks, Journal of Loss Prevention in the Process Industries,

2004, 17(2): p. 111-118.

[33] J.F. Bennett, L.T. Cowley, J.N. Davenport and J.J. Rowson, Large scale natural gas and

LPG jet fires, final report to EC, Shell Report, 1991, TNER 91.022.

[34] J.N. Davenport, Large scale natural gas/kerosene mixed fuel jet fires, final report to

the API, Shell report, 1994, TNER.94.061.

[35] J.N. Davenport, Large scale natural gas/butane mixed fuel jet fires, final report to EC,

Shell report, 1994, TNER.94.030.

[36] Advantica, Horizontal jet fires of oil and gas, in Unpublished Advantica report, 1997.

[37] D.K. Cook, M. Fairweather, J. Hammonds and D.J. Hughes, Size and radiative

characteristics of natural gas flares I. Field scale experiments, Chemical Engineering

Research and Design, 1987, 65(4): p. 310-317.

[38] ISO 13702: Petroleum and natural gas industries - Control and mitigation of fires and

explosions on offshore production installations - Requirements and guidelines, 1999,

International Organization for Standardization.

[39] H. Shoub, Early history of fire endurance testing in the United States, American

Society for Testing and Materials, Special Technical Publication 301, 1961, Fire Test

Methods, Philadelphia, PA: p. 1-9.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Bibliography

Ph-D report J. Ciret Page 203

[40] G.C. Gosselin, Structural fire protection, predictive methods, Proceedings of building

science insight, 1987.

[41] M.G. Goode, Fire Protection of Structural Steel in High-Rise Buildings, National

Institute of Standards and Technology Report, 2004, NIST GCR 04-872.

[42] G. Ellicott, The passive advantage, Fire protection safety and security, World

hydrocarbon: p. 67-68.

[43] R. Smith, Sprayed mineral coatings and intumescent coatings for the fire protection

of structural steel, Journal of Protective Coatings and Linings, 2003, 20(4): p. 12-21.

[44] P. Barber, Offshore cementitious passive fire protection - A proven and cost effective

solution, in Plastics & Rubber Institution, 1987, London.

[45] D. Mistry, Laboratory testing of hydrocarbon fire protection coatings and effect of

wet environments, in NACE - International Corrosion Conference Series, 2006,

Orlando.

[46] J. Gordon, How to specify intumescent coatings, Building Engineer, 2008, 83(9): p. 12-

13.

[47] E.L. Garrett, Ceramic fibre materials for fire protection, Insulation, 1978, 22(2).

[48] W.R. Parlor, Utilisation of mineral wool for offshore fire protection, in Plastic &

Rubber Institution, 1987, London.

[49] P. Mather, Fire protection gets passive, International Coatings Limited.

[50] J.V. Dunk, Advantages of an epoxy medium as the basis of a durable passive fire

protection, in Plastics & Rubber Institution, 1987, London.

[51] A. Fainleib, J. Galy, J.P. Pascault and H.J. Sue, Two ways of synthesis of polymer

networks based on diglycidyl ether of bisphenol A, bisphenol A, and sulfanilamide:

Kinetics study, Journal of Applied Polymer Science, 2001, 80(4): p. 580-591.

[52] J. Galy, A. Sabra and J.-P. Pascault, Characterization of epoxy thermosetting systems

by differential scanning calorimetry, Polymer Engineering and Science, 1986, 26(21):

p. 1514-1523.

[53] V. Rebizant, A.S. Venet, F. Tournilhac, E. Girard-Reydet, C. Navarro, J.P. Pascault and

L. Leibler, Chemistry and mechanical properties of epoxy-based thermosets

reinforced by reactive and nonreactive SBMX block copolymers, Macromolecules,

2004, 37(21): p. 8017-8027.

[54] S. Horold, Phosphorus flame retardants for composites, in International SAMPE

Technical Conference, 1999, Chicago, IL, USA.

[55] J.T. Carter, G.T. Emmerson, C. Lo Faro, P.T. McGrail and D.R. Moore, The

development of a low temperature cure modified epoxy resin system for aerospace

composites, Composites Part A: Applied Science and Manufacturing, 2003, 34(1): p.

83-91.

[56] H. Lee and K. Neville, Hanbook of epoxy resins, Vol. 1, 1967, edited by McGraw-Hill.

[57] E. Girard-Reydet, J.P. Pascault, A. Bonnet, F. Court and L. Leibler, A new class of epoxy

thermosets, Macromolecular Symposia, 2003, 198: p. 309-322.

[58] J.M. Laza, E. Bilbao, M.T. Garay, J.L. Vilas, M. Rodriguez and L.M. Leon, Thermal

properties and fire behaviour of materials produced from curing mixed epoxy and

phenolic resins, Fire and Materials, 2008, 32(5): p. 281-292.

[59] M.I.B. Tavares, J.R.M. D'Almeida and S.N. Monteiro, 13C solid-state NMR analysis of

the DGEBA/TETA epoxy system, Journal of Applied Polymer Science, 2000, 78(13): p.

2358-2362.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Bibliography

Page 204 Ph-D report J. Ciret

[60] A. GroÃŸ, H. Kollek and H. Brockmann, The nucleophilicity as the determining

parameter in the curing of epoxy resins, International Journal of Adhesion and

Adhesives, 1988, 8(4): p. 225-233.

[61] Dow Liquid Epoxy Resins, Dow Plastics.

[62] National Institute of Industrial Research Board, Polymers And Plastics Technology

Handbook, Vol. 1, edited by Asia Pacific Business Press Inc.

[63] C. Barrere and F. Dal Maso, Epoxy-resins crosslinked with Polyamines: Structure and

Properties, Revue de l'Institut Français du Pétrole, 1997, 52(3): p. 317-335.

[64] Y.F. Shih and R.J. Jeng, Carbon black-containing interpenetrating polymer networks

based on unsaturated polyester/epoxy II. Thermal degradation behavior and kinetic

analysis, Polymer Degradation and Stability, 2002, 77(1): p. 67-76.

[65] R.N. Rothon and P.R. Hornsby, Flame retardant effects of magnesium hydroxide,

Polymer Degradation and Stability, 1996, 54(2-3 SPEC. ISS.): p. 383-385.

[66] G. Matuschek, Thermal degradation of different fire retardant polyurethane foams,

Thermochimica Acta, 1995, 263(C): p. 59-71.

[67] J. Jang, H. Chung, M. Kim and H. Sung, The effect of flame retardants on the

flammability and mechanical properties of paper-sludge/phenolic composite,

Polymer Testing, 2000, 19(3): p. 269-279.

[68] R. Seddon and J.F. Harper, Influence of flame retardant additives on the processing

characteristics and physical properties of ABS, Macromolecular Symposia, 2001, 169:

p. 109-119.

[69] S.R. Owen and J.F. Harper, Mechanical, microscopical and fire retardant studies of

ABS polymers, Polymer Degradation and Stability, 1999, 64(3): p. 449-455.

[70] S. Bourbigot, M. Le Bras, L. Gengembre and R. Delobel, XPS study of an intumescent

coating application to the ammonium polyphosphate/pentaerythritol fire-retardant

system, Applied Surface Science, 1994, 81(3): p. 299-307.

[71] S. Duquesne, M. Le Bras, S. Bourbigot, R. Delobel, H. Vezin, G. Camino, B. Eling, C.

Lindsay and T. Roels, Expandable graphite: A fire retardant additive for polyurethane

coatings, Fire and Materials, 2003, 27(3): p. 103-117.

[72] S. Duquesne, S. Magnet, C. Jama and R. Delobel, Intumescent paints: Fire protective

coatings for metallic substrates, Surface and Coatings Technology, 2004, 180-181: p.

302-307.

[73] C. Drevelle, S. Duquesne, M. Le Bras, J. Lefebvre, R. Delobel, A. Castrovinci, C.

Magniez and M. Vouters, Influence of ammonium polyphosphate on the mechanism

of thermal degradation of an acrylic binder resin, Journal of Applied Polymer Science,

2004, 94(2): p. 717-729.

[74] M. Jimenez, S. Duquesne and S. Bourbigot, Intumescent fire protective coating:

Toward a better understanding of their mechanism of action, Thermochimica Acta,

2006, 449(1-2): p. 16-26.

[75] US-2106938, Fireproofing of wood, patented by H. Tramm, C. Clar, P. Kuhnel and W.

Schuff for RUHRCHEMIE AG in 1938,

http://www.freepatentsonline.com/2106938.html.

[76] H.L. Vandersall, Intumescent coating systems, their development and chemistry,

Journal of Fire and Flammability, 1970, 1: p. 97-140.

[77] US-2452054, Fire-retardant composition and process, patented by J. Grinnell, J.

Walter and S. Samuel for ALBI MFG Inc. in 1948,

http://www.freepatentsonline.com/2452054.html.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Bibliography

Ph-D report J. Ciret Page 205

[78] R. Delobel, M. Le Bras, N. Ouassou and F. Alistiqsa, Thermal behaviour of ammonium

polyphosphate-pentaerythritol and ammonium pyrophosphate – pentaerythritol

intumescent additives in polypropylene formulations, Journal of Fire Sciences, 1990,

8(2): p. 85-108.

[79] G. Camino, L. Costa and G. Martinasso, Intumescent fire-retardant systems, Polymer

Degradation and Stability, 1989, 23(4): p. 359-376.

[80] M. Wladyka-Przybylak and R. Kozlowski, Thermal characteristics of different

intumescent coatings, Fire and Materials, 1999, 23(1): p. 33-43.

[81] R. Delobel, N. Ouassou, M. Le Bras and J.M. Leroy, Fire retardance of polypropylene:

Action of diammonium pyrophosphate-pentaerythritol intumescent mixture, Polymer

Degradation and Stability, 1989, 23(4): p. 349-357.

[82] R. Delobel, M. Le Bras, N. Ouassou and R. Descressain, Fire retardance of

polypropylene by diammonium pyrophosphate-pentaerythritol: Spectroscopic

characterization of the protective coatings, Polymer Degradation and Stability, 1990,

30(1): p. 41-56.

[83] S. Bourbigot, M. Le Bras, S. Duquesne and M. Rochery, Recent advances for

intumescent polymers, Macromolecular Materials and Engineering, 2004, 289(6): p.

499-511.

[84] M. Gao, W. Wu and Y. Yan, Thermal degradation and flame retardancy of epoxy

resins containing intumescent flame retardant, Journal of Thermal Analysis and

Calorimetry, 2009, 95(2): p. 605-608.

[85] J.W. Gu, G.C. Zhang, S.I. Dong, Q.Y. Zhang and J. Kong, Study on preparation and fire-

retardant mechanism analysis of intumescent flame-retardant coatings, Surface and

Coatings Technology, 2007, 201(18): p. 7835-7841.

[86] S. Bourbigot, M. Le Bras, R. Delobel, P. Breant and J.M. Tremillon, Carbonization

mechanisms resulting from intumescence-part II. Association with an ethylene

terpolymer and the ammonium polyphosphate-pentaerythritol fire retardant system,

Carbon, 1995, 33(3): p. 283-294.

[87] G. Camino, L. Costa, L. Trossarelli, F. Costanzi and A. Pagliari, Study of the mechanism

of intumescence in fire retardant polymers: Part VI-Mechanism of ester formation in

ammonium polyphosphate-pentaerythritol mixtures, Polymer Degradation and

Stability, 1985, 12(3): p. 213-228.

[88] S. Duquesne, S. Magnet, C. Jama and R. Delobel, Thermoplastic resins for thin film

intumescent coatings - Towards a better understanding of their effect on

intumescence efficiency, Polymer Degradation and Stability, 2005, 88(1): p. 63-69.

[89] A.R. Horrocks, Developments in flame retardants for heat and fire resistant textiles -

The role of char formation and intumescence, Polymer Degradation and Stability,

1996, 54(2-3 SPEC. ISS.): p. 143-154.

[90] M. Bugajny, M. Le Bras and S. Bourbigot, New approach to the dynamic properties of

an intumescent material, Fire and Materials, 1999, 23(1): p. 49-51.

[91] P. Anna, G. Marosi, I. Csontos, S. Bourbigot, M. Le Bras and R. Delobel, Influence of

modified rheology on the efficiency of intumescent flame retardant systems, Polymer

Degradation and Stability, 2001, 74(3): p. 423-426.

[92] F. Carpentier, S. Bourbigot, M. Le Bras and R. Delobel, Rheological investigations in

fire retardancy: Application to ethylene-vinyl-acetate copolymer-magnesium

hydroxide/zinc borate formulations, Polymer International, 2000, 49(10): p. 1216-

1221.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Bibliography

Page 206 Ph-D report J. Ciret

[93] S. Duquesne, R. Delobel, M. Le Bras and G. Camino, A comparative study of the

mechanism of action of ammonium polyphosphate and expandable graphite in

polyurethane, Polymer Degradation and Stability, 2002, 77(2): p. 333-344.

[94] M.L. Bras, M. Bugajny, J.M. Lefebvre and S. Bourbigot, Use of polyurethanes as char-

forming agents in polypropylene intumescent formulations, Polymer International,

2000, 49(10): p. 1115-1124.

[95] M. Jimenez, S. Duquesne and S. Bourbigot, Characterization of the performance of an

intumescent fire protective coating, Surface and Coatings Technology, 2006, 201(3-

4): p. 979-987.

[96] M. Jimenez, S. Duquesne and S. Bourbigot, Multiscale experimental approach for

developing high-performance intumescent coatings, Industrial and Engineering

Chemistry Research, 2006, 45(13): p. 4500-4508.

[97] I.S. Reshetnikov, A.N. Garashchenko and V.L. Strakhov, Experimental investigation

into mechanical destruction of intumescent chars, Polymers for Advanced

Technologies, 2000, 11(8-12): p. 392-397.

[98] I.S. Reshetnikov, M.Y. Yablokova, E.V. Potapova, N.A. Khalturinskij, V.Y. Chernyh and

L.N. Mashlyakovskii, Mechanical stability of intumescent chars, Journal of Applied

Polymer Science, 1998, 67(10): p. 1827-1830.

[99] E. Kandare, B.K. Kandola and J.E.J. Staggs, Global kinetics of thermal degradation of

flame-retarded epoxy resin formulations, Polymer Degradation and Stability, 2007,

92(10): p. 1778-1787.

[100] I. Reshetnikov, A. Antonov, T. Rudakova, G. Aleksjuk and N. Khalturinskij, Some

aspects of intumescent fire retardant systems, Polymer Degradation and Stability,

1996, 54(2-3 SPEC. ISS.): p. 137-141.

[101] UL 1709: Rapid rise fire tests of protection materials for structural steel, 1994,

Underwriter Laboratories.

[102] RWS Curve, SP Swedish national testing and research Institute.

[103] OTI 95634: Jet fire resistance - Test of passive fire protection materials, 1996, HSE

Offshore Technology Report.

[104] ISO 22899-1: Determination of the resistance to jet fires of passive fire protection

materials - General requirements, 2007, International Organization for

Standardization.

[105] M. Jimenez, Etude des mécanismes de protection et de réaction au feu dans les

revêtements intumescents, Thesis defended at Université des Sciences et

Technologies de Lille, Lille, on 5th October 2006.

[106] Beerenberg Corp. AS, Benarx F Epoxy Box Performance, Available from:

http://www.benarx.com/products/benarx-f-epoxy-box-1/benarx-f-epoxy-box-

performance (in 2009).

[107] C. Reti, M. Casetta, S. Duquesne, S. Bourbigot and R. Delobel, Flammability properties

of intumescent PLA starch and lignin, Polymers for Advanced Technologies, 2008,

19(6): p. 628-635.

[108] G. Camino, G. Martinasso and L. Costa, Thermal degradation of pentaerythritol

diphosphate, model compound for fire retardant intumescent systems: Part I-Overall

thermal degradation, Polymer Degradation and Stability, 1990, 27(3): p. 285-296.

[109] A. Andersson, S. Landmark and F.H.J. Maurer, Evaluation and characterization of

ammonium polyphosphate-pentaerythritol- based systems for intumescent coatings,

Journal of Applied Polymer Science, 2007, 104(2): p. 748-753.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Bibliography

Ph-D report J. Ciret Page 207

[110] V. Babrauskas, CONE CALORIMETER - A VERSATILE BENCH-SCALE TOOL FOR THE

EVALUATION OF FIRE PROPERTIES, in 1986, Luxembourg, Luxemb.

[111] V. Babrauskas and W.J. Parker, IGNITABILITY MEASUREMENTS WITH THE CONE

CALORIMETER, Fire and Materials, 1987, 11(1): p. 31-43.

[112] F. Samyn, S. Bourbigot, S. Duquesne and R. Delobel, Effect of zinc borate on the

thermal degradation of ammonium polyphosphate, Thermochimica Acta, 2007,

456(2): p. 134-144.

[113] Z. Wang, E. Han, F. Liu and W. Ke, Thermal behavior of nano-TiO2 in fire-resistant

coating, Journal of Materials Science and Technology, 2007, 23(4): p. 547-550.

[114] A. Laachachi, M. Cochez, E. Leroy, P. Gaudon, M. Ferriol and J.M. Lopez Cuesta, Effect

of Al2O3 and TiO2 nanoparticles and APP on thermal stability and flame retardance

of PMMA, Polymers for Advanced Technologies, 2006, 17(4): p. 327-334.

[115] G. Camino and L. Costa, Performance and mechanisms of fire retardants in polymers-

A review, Polymer Degradation and Stability, 1988, 20(3-4): p. 271-294.

[116] N. Grassie, M.I. Guy and N.H. Tennent, Degradation of epoxy polymers: Part 4-

Thermal degradation of bisphenol-A diglycidyl ether cured with ethylene diamine,

Polymer Degradation and Stability, 1986, 14(2): p. 125-137.

[117] J.C. Paterson-Jones, V.A. Percy, R.G.F. Giles and A.M. Stephen, The thermal

degradation of model compounds of amine-cured epoxide resins II. The thermal

degradation of 1,3-diphenoxypropan-2-ol and 1,3-diphenoxypropene, Journal of

Applied Polymer Science, 1973, 17(6): p. 1877-1887.

[118] M.A. Keenan and D.A. Smith, Further aspects of the thermal degradation of epoxide

resins, Journal of Applied Polymer Science, 2003, 11(7): p. 1009-1026.

[119] G. Camino, L. Costa, L. Trossarelli, F. Costanzi and G. Landoni, Study of the

mechanism of intumescence in fire retardant polymers: Part IV-Evidence of ester

formation in ammonium polyphosphate-pentaerythritol mixtures, Polymer

Degradation and Stability, 1984, 8(1): p. 13-22.

[120] M. Le Bras, S. Bourbigot, Y. Le Tallec and J. Laureyns, Synergy in intumescence -

Application to Î²-cyclodextrin carbonisation agent in intumescent additives for fire

retardant polyethylene formulations, Polymer Degradation and Stability, 1997, 56(1):

p. 11-21.

[121] S. Bourbigot, M.L. Bras and R. Delobel, Carbonization mechanisms resulting from

intumescence association with the ammonium polyphosphate-pentaerythritol fire

retardant system, Carbon, 1993, 31(8): p. 1219-1230.

[122] A. Marchal, R. Delobel, M. Le Bras, J.M. Leroy and D. Price, Effect of intumescence on

polymer degradation, Polymer Degradation and Stability, 1994, 44(3): p. 263-272.

[123] S.T. Norberg, G. Svensson and J. Albertsson, A TiP2O7 superstructure, Acta

Crystallographica Section C: Crystal Structure Communications, 2001, 57(3): p. 225-

227.

[124] Z. Wang, E. Han and W. Ke, Effect of nanoparticles on the improvement in fire-

resistant and anti-ageing properties of flame-retardant coating, Surface and Coatings

Technology, 2006, 200(20-21): p. 5706-5716.

[125] G.E. Maciel, V.J. Bartuska and F.P. Miknis, Characterization of organic material in coal

by proton-decoupled 13C nuclear magnetic resonance with magic-angle spinning,

Fuel, 1979, 58(5): p. 391-394.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

Bibliography

Page 208 Ph-D report J. Ciret

[126] S. Bourbigot, M. Le Bras, P. Breant, J.M. Tremillon and R. Delobel, Zeolites: New

synergistic agents for intumescent fire retardant thermoplastic formulations - Criteria

for the choice of the zeolite, Fire and Materials, 1996, 20(3): p. 145-154.

[127] S. Bourbigot, M. Le Bras, R. Delobel, R. Decressain and J.P. Amoureux, Synergistic

effect of zeolite in an intumescence process: Study of the carbonaceous structures

using solid-state NMR, Journal of the Chemical Society - Faraday Transactions, 1996,

92(1): p. 149-158.

[128] S. Bourbigot, M. Le Bras, R. Delobel and J.M. Tremillon, Synergistic effect of zeolite in

an intumescence process: Study of the interactions between the polymer and the

additives, Journal of the Chemical Society - Faraday Transactions, 1996, 92(18): p.

3435-3444.

[129] J.R. Van Wazer, C.F. Callis, J.N. Shoolery and R.C. Jones, Principles of phosphorus

chemistry II. Nuclear magnetic resonance measurements, Journal of the American

Chemical Society, 1956, 78(22): p. 5715-5726.

[130] T.M. Duncan and D.C. Douglas, On the 31P chemical shift anisotropy in condensed

phosphates, Chemical Physics, 1984, 87(3): p. 339-349.

[131] M. Bugajny, S. Bourbigot, M. Le Bras and R. Delobel, The origin and nature of flame

retardance in ethylene-vinyl acetate copolymers containing hostaflam AP 750,

Polymer International, 1999, 48(4): p. 264-270.

[132] S.A. Sojka and W.B. Moniz, Curing of an epoxy resin as followed by carbon 13NMR

spectroscopy, Journal of Applied Polymer Science, 1976, 20(7): p. 1977-1982.

[133] S. Bourbigot, M. Le Bras, R. Delobel and L. Gengembre, XPS study of an intumescent

coating II. Application to the ammonium polyphosphate/pentaerythritol/ethylenic

terpolymer fire retardant system with and without synergistic agent, Applied Surface

Science, 1997, 120(1-2): p. 15-29.

[134] G. Camino, L. Costa and L. Trossarelli, Study of the mechanism of intumescence in fire

retardant polymers: Part II-Mechanism of action in polypropylene-ammonium

polyphosphate-pentaerythritol mixtures, Polymer Degradation and Stability, 1984,

7(1): p. 25-31.

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr

INVESTIGATION OF INTUMESCENT COATINGS FOR FIRE PROTECTION - APPLICATION TO JET-FIRE

Résumé : Cette étude s’intéresse aux comportements de 4 peintures intumescentes développées pour

protéger des plateformes offshores et susceptibles de résister aux « jet-fires ». Un jet-fire peut

intervenir sur un site pétrochimique suite à une fuite d’hydrocarbures sous pression et causer de sérieux

dommages de part la chaleur dégagées et surtout la quantité de mouvement générées. Les aspects

physiques et chimiques de ces formulations ont été développés permettant de mettre en avant les

effets du pentaérythritol sur le comportement viscoélastiques et le processus d’intumescence. Par

diffraction des rayons X et par RMN à l’état solide, nous avons montré les interactions entre ammonium

et polyphosphate et différentes sources de carbones (pentaérythritol, dipentaérythritol, réseau

époxyde) permettant la formation d’un char. Les résidus phosphorés réagissent ensuite avec TiO

pour

former une structure cristalline TiP

suspectées d’améliorer la résistance au feu et la résistance

mécanique du char. Des tests feu ont confirmés ces améliorations. Dans un dernier chapitre nous avons

développé un test permettant de reproduire à l’échelle laboratoire les phénomènes radiatifs et

convectifs du jet-fire. Les premiers résultats ont montré de bonnes corrélations entre les observations

faites à grande échelle et celles réalisées au laboratoire.

Mots clefs : Résine époxyde, Dégradation thermique, Intumescence, Polyphosphate d’ammonium,

Pentaerythritol, Dioxyde de titane, RMN du solide, Diffraction des rayons X, Développement de test feu.

Abstract: The aim of this study is to understand and to explain behaviours exhibited by four epoxy-

based intumescent formulations used on offshore platforms facing to jet-fire. A jet-fire is a turbulent

diffusion flame resulting from the combustion of a fuel continuously released with some significant

momentum. It represents a significant element of the risk on offshore installations. Regarding the

formulation studied, we have developed three approaches. Firstly, the visco-elastic behaviour and

mechanical resistance of the formulations have been investigated. The results show that pentaerythritol

causes a viscosity decrease at lower temperature that appears as prejudicial to maintain efficient char

on steel substrate. In a second part, chemical evolutions of the intumescent formulation have been

determined thanks to solid-state NMR and X-Ray diffraction. Interactions between ammonium

polyphosphate and respective carbon sources present in formulations have been assumed, yielding to

the formation of char and production of phosphorus residues. Then these phosphorus residues react at

high temperature with TiO

to form a crystalline structure TiP

suspected to enhance mechanical

properties and flame retardant performance. In a last part, furnace fire tests confirm this enhancement.

Furthermore, a new small-scale experimental setup is developed mimicking large scale jet-fire resistance

test in order to obtain rapidly and at low cost reliable behaviours of a large number of formulations

facing to high load mixing radiative heat and flame impact. First results have been correlated with the

large-scale ones and different geometries have been considered.

Keywords: Epoxy resin, Thermal degradation, Intumescence, Ammonium polyphosphate,

Pentaerythritol, Titanium dioxide, Solid state NMR, X-Ray diffraction, Fire testing development.

Intitulé et adresse du laboratoire: Laboratoire LSPES – Equipe PERF CNRS UMR 8008

E.N.S.C.L, Cité Scientifique, Bât. C7

F-59652 Villeneuve d’Ascq Cedex

Thèse de Jérémy Ciret, Lille 1, 2010

http://doc.univ-lille1.fr