Easby 07

NEWCASTLE UNIVERSITY
----------------------------
205 36772
----------------------------
LIBRARY
Fire Behaviour of Pultruded Composites
Thesis submitted in accordancewith the requirementsof the University of Newcastle

upon Tyne for the degreeof Doctor of Philosophy.
R. C. Easby
Centre for CompositeMaterials Engineering,

School of Mechanical & SystemsEngineering,
University of Newcastleupon Tyne.
May 2007
Fire Behaviour ofPultruded Composites
Acknowledgments
Firstly I would like to express my gratitude to my supervisor, Professor A. G.
Gibson, who has provided me with continual support and guidance throughout my
studies. I would also like to acknowledge the integral part Dr. Yongshu.Wu has
played, providing answersto endlessquestions. Similarly I would like to thank Mr.
Ron Jensen for dealing with all of my last minute requests and ideas with good
humour, and professionalism.
I also wish to thanks my sponsors, the EPSRC and Fiberline Composites, in
particular Mr. Anders Korsgarrd for his constantassistancethroughout this project.
Finally I would like to acknowledgethe assistance,support and endlesspatience of
my colleagues,family and friends and of courseDeb.
P, C Easby
PhD Thesis
Abstract
This thesis describesthe development of a model to predict the failure responseof
loaded pultruded compositesin fire. The model takes an existing thermal model [19] capableof describing temperatureevolution and residual resin content as functions
of time through a pultrusion for fires with heat-fluxesup to and including 125kWm.
Experiments were developedto determinehow tensile and compressivestrength (aT
and ac), and longitudinal and transverse stiffhess (Ei and E2) of the composites
varied with temperature. This required specialised equipment to be designed and
fabricated. The mechanicalproperty data were recordedas functions of temperature
and combined with the thermal model and classical laminate theory. The resulting
failure model, outputs tensile and compressive strength of the pultrusion as a
function of time for fires with heat- fluxes up to and including 125kWm-2. A, B, D
matrix evolution as a function of time is also produced.
The modelling procedure was carried out for polyester and phenolic glass fibrereinforced pultrusions subjectedto a 5OkWm-2heat-flux and verified by a series of
propane burner tests. The modelled tensile and compressiveresults match the data
from the propaneburner tests to a reasonabledegreeof accuracy. It was shown that
the materials were more susceptibleto compressivefailure rather than tensile failure,
when subjectedto a fire.
Work on the model was supplemented by a series of larger scale fire tests on box and
T sections, including flexure tests in a pool fire and temperature controlled furnace.
In both tests it was found that failure occurred on the compressive side of the section,
the
failure
in
time
with a
order of 100 seconds. Compression tests were also carried
out on short box columns to investigate the effect various fire protections systems
had on failure time of the columns, when surrounded by a heat-flux of 50k)V2.
It
was found that protecting the loaded section inside an insulating sleeve proved the
most successful approach.
P, C Easby
PhD Thesis
ACKNOWLEDGEMENTS
ABSTRACT
NOMENCLATURE
1.
INTRODUCTION
1.1.
Introduction to Composite Materials
1.2.
Reinforcement Material
13.
Reinforcement Architecture
1.3.1. Unidirectional
1.3.2. ChoppedStrandMat/Needle Weave
10
10
10
1.4.
Matrix Material
1.4.1. Polyester
1.4.2. Phenolic
11
11
11
1.5.
11
Pultrusion
1.6.
Fire Reaction & Fire Resistance
1.6.1. Fire Reaction
1.6.2. Fire Resistance
13
13
13
1.7.
Test Procedures
1.7.1. Cone Calorimeter (including atmospherecontrolled)
1.7.2. OSU Heat ReleaseRate Technique
1.7.3. Limiting Oxygen Index (LOI) Test
1.7.4. Radiant PanelTest
1.7.5. Flame Propagation& SpreadTests
1.7.6. SmokeDensity
1.7.7. Single Burning Item (SBI) Test
1.7.8. Room Fire Test
1.7.9. FurnaceTests
1.7.10.
Pool-fire Tests
1.7.11.
Burner Tests
Jet-fire Tests
1.7.12.
14
15
15
16
17
17
18
19
19
20
21
22
22
1.8.
Thesis Overview
24
1.9.
Chapter Summary
25
2.
2.1.
LITERATURE REVIEW
26
The 'Slow Burn-Through' Effect
26
2.2.
Fire ResistanceModelling
2.2.1. Responseof Compositesto Fire
Heat Conduction Model
2.2.2.1-D
2.2.3. HendersonModel
2.2.4. Simplified HendersonModel
27
28
30
31
32
R. C Easby
PhD Thesis
2.3.
Fire Behaviour under Load
32
2.4.
Chapter Summary
34
THERMAL MODELLING
3.
3.1.1-D
Thermal Model
36
36
3.2.
Thermogravimetric Analysis (TGA) Parameters
37
3.3.
Propane Burner Test
40
3.4.
Rear Face Temperature Profile
42
3.5.
Chapter Summary
42
4.
MECHANICAL
PROPERTIES
44
4.1.
Materials
44
4.2.
Tensile Strength at Elevated Temperatures
46
4.3.
Compressive Strength at Elevated Temperatures
47
4.4.
Longitudinal (El) & Transverse (E2) Stiffness at Elevated Temperatures
48
4.5.
Chapter Summary
49
5.
FIRE TESTING UNDER LOAD
51
5.1.
Tensile Tests in Fire
51
5.2. '
Compression Tests in Fire
52
53.
Pool Fire Test under Load
55
5.4.
, Furnace Test under Load
57
5.5.
Columns under Compression in Fire
59
5.6.
Chapter Summary
61
6.
RESULTS
63
6.1.
Mechanical Properties
6.1.1. Problemsarising from the Material Cross Section
6.1.2. Tensile Strength
6.1.3. CompressiveStrength
6.1.4. Longitudinal (EI) and Transverse(132)Stiffhess
6.1.5. Mechanical Propertiesvs. Temperature
63
64
65
69
71
71
6.2.
Fire Testing under Load
6.2.1. Tensile and CompressiveTests
6.2.2. Pool Fire test under Load
6.2.3. Fumacetest under Load
78
78
79
81
R. C Easby
PhD Thesis
Columns under compression
6.2.4.
Experimental reliability
63.
6.3.1. Mechanical propertiesat elevatedtemperatures
6.3.2. Fire testing under load
6.3.3. Pool fire test under load
6.3.4. Furnacetest under load
7.
7.1.
MODELLING BEHAVIOUR UNDER LOAD

Laminate Constitutive Equations
82
84
84
86
87
87
89
89
Input of Mechanical Properties vs. Temperature

7.2.
7.2.1. Mechanical Propertiesof CSM Skins
91
91
Results
7.3.
7.3.1. A, B, D Matrix Evolution
7.3.2. StrengthsModelling
7.3.3. Tensile StrengthPrediction
7.3.4. Buckling and CompressiveStrengthPrediction
94
94
99
101
103
8.
CONCLUSIONS
108
Propaneburner test
108
8.2.
Thermal modelling
108
8.3.
Mechanical properties as a function of temperature
108
8.1.
Fire testing under Load

8.4.
8.4.1. Tension
8.4.2. Compressionincluding inturnescents
8.4.3. Pool fire test
8.4.4. Furnacetest
8.4.5. Columns including coatings
108
108
109
109
109
110
Modelling
8.5.
8.5.1. A, B, DMaffix7Evolution
8.5.2. Tensile StrengthPrediction
8.5.3. Compressiveand Buckling StrengthPrediction
110
110
9.
FUTURE WORK
112
9.1.
Thermally Induced Deformations
112
9.2.
Modelling of fire protective coatings
112
93.
Sandwich panels
112
9.4.
Buckling Response
113
9.5.
Structures
113
REFERENCES
R. C Easby
PhD Thesis
115
PUBLICATIONS
122
APPENDIX
123
Visual Basic Code

ABD Matrix calculation (Polyester)
Tensile and Compressivestrengthsas fimctions of time (Polyester)
Stressstrain curve modelling (Polyester)
ABD Matrix calculation (Phenolic)
Tensile and Compressivestrengthsas functions of time (Phenolic)
Stressstrain curve modelling (Phenolic)
123
123
132
136
138
147
151
Thermal Model Outputs
153
Engineering Drawings
164
R. C Easby
PhD Thesis
Nomenclature
pre-exponentialfactor
A matrix
inverted A matrix
breadth
B matrix
inverted B matrix
c buckling constant
CP specific heat capacity
cpg specific heat of volatile gases
D matrix
inverted D matrix
E flexural modulus
E Young's Modulus
E(T) longitudinal modulus as a function of temperature
Ei longitudinal flexural modulus
E2 transverseflexural modulus
E45
modulus at 45* to the longitudinal
Ecsm flexural modulus of the skin material
EF. 11 flexural modulus of the full section of material
Ei activation energy
EuD flexural modulus of the core material
G12 shearmodulus
h enthalpy of the solid phase
hc enthalpy of the composite
hg enthalpy of volatile gases
I secondmoment of area
k relaxation constant(usedin the context of mechanicalpropertiesas
function of temperature)
k,, k thermal conductivity in the through thicknessdirection
mid plane curvatures
TT
thermally induced curvatures
I spanlength
m instantaneousmass
m mass
MO initial mass
Mf
final mass
G
flux
Ihg
mass
of volatile gases
9
bending moment matrix

loading
constant
n
ni order of reaction
A normal load matrix
p property value at any temperature
PR relaxed property value
A,
;i
; it
b
x
t
R. C Easby
PhD Thesis
Pu unrelaxedproperty value
heat
decomposition
of
Qi
QP endothermicdecompositionenergy
Q matrix
,
R universal gas constant(8.3144 Jmol-'K)
R resin content (usedin the context of mechanicalpropertiesas
function of temperature)
t time
T temperature
t thickness
t, thicknessof the core material
t2
thicknessof the full section
T9 glasstransition temperature
Vf fibre volume fraction
w applied load
Greek letters
deflection
appliedstrain
Ef failure strain
WO mid planestrains
thermallyinducedstrains
wor
Poisson'sratio
v density
p
or(T materialstrengthasa functionof temperature
failure strengthof the composite
Orf failure strengthof the fibres
0'. stressin the matrix at the failure strengthof the fibres
UApp
appliedstress
O'buckling buckling
strength
(7cl
longitudinalcompressive
strength
Of, failure
strength
O'max
exaggerated
strengthvalue
UT]
longitudinaltensilestrength
R. C Easby
PhD Thesis
1. Introduction
I. I.
Introduction to Composite Materials
Composite materials have a high strength-to-weightratio when comparedto metals,

making them a desirable structural material. In its most basic form, a composite
material consists of at least two elements,which when combined, form one single
material with mechanicalpropertiesgreaterthan its components.
Most composites consist of a bulk material called the matrix, combined with a
reinforcement material, to increase strength and stiffness of the matrix. The most
common composite materials are fibre reinforced plastics (FRP). These generally
consist of a polymer matrix reinforced with glass,carbonor aramid fibres.
The mechanical properties of a composite will vary according to the proportions of
matrix and reinforcement used. This proportionality is defined by the fibre volume
fraction. The larger the fibre volume fraction, the stiffer and strongerthe composite.
1.2.
Reinforcement Material
In FRP's, reinforcement material can be glass fibre, aramid fibre or carbon fibre.
This thesis is concerned with E-glass fibre reinforcement. On its own, E-glass
exhibits high tensile and compressive strengths and high stiffness propeities. It is
one of the cheapestreinforcementsavailable, making it a common choice for general
engineering applications, accounting for around 90% of the reinforcement used in
structural reinforced plastic applications[10].
Table 1.1. Typical propertiesof E-glass [11].

Specific gravity
Young's
Poisson's ratio
Modulus
Tensile Strength
(MPa)
(GPa)
E-glass
R. C Easby
PhD Thesis
2.55
72
0.2
2400
10
Reinforcement Architecture
1.3.1.
Unidirectional
Unidirectional (UD) reinforcement consists of fibres running in the same direction

(0' direction). The main benefit of using UD reinforcement is that fibres can be
oriented in the exact position to allow any forces to passalong the fibre, thereby take
up any loading the component would experience in service. This is advantageous
when designing structural components. The downside of having all the fibres laying
in the same direction is that all the stiffness and strength is associatedwith that
direction. There is very little strength or stiffness in other directions. Loads acting
parallel to the fibre are easily dealt with, whereasany acting in other directions cause
problems.
Chopped Strand Mat/Needle Weave

1.3.2.
ChoppedStrand mat (CSM)/Needleweave is a random in plane reinforcement
material. It differs from simpleCSMby the way that matis 'needled'together.This
provides the material with additional strength. This enablesthe material to be
pultruded(seeSection1.5),sincethe strengthis sufficientto resistthe tractiveforces
associated
with theprocess.
(i)
(ii)
Figure 1.1. Unidirectional glass fibre reinforcement (i) [12] and CSM/Needle Weave glass fibre
reinforcement (ii) [13].
R. C Eashv
PhD Thesis
II
1.4.
Matrix Material
Polyester
1.4.1.
Thermosetting polyester is the most widely used matrix material due to its easeof
use, low cost and good all round mechanicalproperties.
Phenolic
1.4.2.
Phenolic resin is used as a matrix when improved fire resistanceis required.
Phenolicbasedcompositesretaintheir propertiesat highertemperatures.
Table 1.2. Typical propertiesof polyester and phenolic resins [11].

Matrix
Specific
Young's
Poisson's
Tensile
Compressive
Material
Gravity
Modulus
ratio
Strength
Strength
(MPa)
(MPa)
0.36
60
130
50
(GPa)
Polyester
1.21
3.6
Phenolic
1.15
3.0
Pultrusion
Pultrusion is a manufacturing process by which many composites with structural
applications are made.
The process can vary according to manufacturer but is essentially the same. The
by
begin
Fiberline
Composites
as spools of unidirectional
manufactured
pultrusions,
into
is
fed
The
then
through
a
material
passes
a pre-former.
reinforcement which
heating and curing chamberwhere resin is injected under pressure. The cured profile
length.
is
be
into
The
to
the
sawn
material
pulled
other end
any required
emergesat
through the processby a series of pulling devices(Figure 1.2). The end product has
a constant cross section. A variation of the process called 'pulfonriing' is used to
produce componentswith slight changesin section.
R. C Easby
PhD Thesis
12
Often an enveloping layer of choppedstrand mat (CSM) needleweave mat is applied

to protect the unidirectional core from damage. A surface-veil is usually applied to
the surfaceof the profile to give a high quality surfacefinish.
Rainforciament
Figure 1.2. The pultrusion process [ 14].
Typical applications for pultruded sections include drive shafts, ladders, walkways,
handles etc. Several large structures have been constructed using pultruded sections
foot
bridges
load
bearing
including
bridges,
the
and medium
component
road
main
as
low
The
buildings.
the
cost of manufacture
advantage
of
pultrusion
is
greatest
sized
(Table
1.3).
processes
continuous
associated with
Table 1.3. Advantagesand disadvantagesof the pultrusion process.

Advantages
Disadvantages
Very fast and economical
Limited to constant/nearconstantcross section
Resin content can be accuratelycontrolled
High die costs
High fibre volume fractions can be obtained

Resin impregnationareais minimised, reducing
volatile emissions
Figure 1.3. Typical pultruded profiles. The large profile in the centre was used in the construction of
a road bridge.
R. C Easby
PhD Thesis
13
Fire Reaction & Fire Resistance
In order for a material to be used in a fire risk application it must conform to a series
its
fire.
towards
regarding
reaction
and
resistance
of standards
Fire Reaction
1.6.1.
Fire reactionconcernsa material'sresponseto fire. Typically fire reactionexamines

the following key parameters:
"
Ignitibility.
Time-to-ignition (TTI) is a very important fire reaction property
since it defines the onsetof combustion.

"
Heat release. Heat release rate (HRR) is the single most important fire
reaction property[15] because the heat released by burning material can
provide the additional thermal energy required for the growth and spreadof
fire. Heat releaseis defined as the thermal energy produced per unit area of
in
inflammable
ignite
bum
decomposition
the
and
surface, when
products
in
heat
flux.
fire,
to
of
vicinity
a material
or subjected a
"
Surface spread of flame. The speedof flame-spreadacrossthe surface of

combustiblematerials is an important factor in fiie growth.
"
Smoke and toxicity. Smoke reducesvisibility which is of great concern in a

fire situation. Smoke can also be highly toxic, which obviously posesserious
health hazards.
1.6.2.
Fire Resistance
Fire resistancedescribesthe ability of a material or structureto restrict the spreadof

fire and to retain mechanical and physical integrity. Key fire resistanceparameters
include heat transfer, bum-through resistanceand structural integrity.
P- C Easby
PhD Thesis
1.7.
14
Test Procedures
There are several fire tests in existence, each of which is used to test materials to
determinewhether they conform to one standardor another. An area of concern for
the materials industry is that there is not a single common approachto determinethe
fire responseof composites[16], and it is often necessaryto use a combination of
tests to sufficiently describe the behaviour of a composite material or structure in
fire.
Fire tests vary in scale, from small bench-scaleproceduresto large-scaleroom tests.

Fire resistancetests tend to be towards the large end of this scale. The most popular
fire reaction tests are bench scale tests becausethey are quick, inexpensive and will
usually yield reproducible data. Bench-scaletests are limited becausethey ignore the
fire
indeed
it
is
due
to
growth,
said that they only relate to a 'snapshot' of the
effect
overall fire behaviour [17]. Heat releaserates, air movements and the oxygen/fuel
in
bench-scale
in
fire,
different
tests.
to
those
that
a real
are often very
exist
ratio
Thesecan affect the measuredfire reaction properties [ 18]. One further drawback of
bench-scaletesting is that the entire test sample is consumed. This may not be the
fire
due
in
in
levels
to
the
scenario
enclosed,unventilated
reduced
a
real
oxygen
case
spaces[19].
Intermediate-scaletests provide a link between the bench-scaletests and the more
expensivelarge-scaletests. They can overcome some of the shortcomingsof bench
lean
involving
fire
they
towards
a scale
as
more
scenarios,
usually
scale-testing
real
model or part-sectionof the structureunder investigation.
Large-scale tests are expensive and complex with long set-up times. For these
it
is
they
are
performed
reasons
rarely and only when
unavoidable. For instance,if
there is any uncertainty with scaling up data from an intermediate-sizedtest. Largehave
in
tests
also
scale
a commercial role demonstratingthe fire performanceof new
materials and methods.
R. C Easby
PhD Thesis
15
The size and type of test (fire reaction or fire resistance) for the most common fire
tests are detailed in Table 1.4. The most common fire tests are as follows:
Cone Calorimeter (including atmosphere controlled)

The cone calorimeter is a versatile fire reaction test capable of measuringmost fire
1.7.1.
reaction properties with the exception of flame spread. It forms the basis of many
fire standards, placing it in high regard with universities and research institutes
across the world. The test involves exposing a small sample to a heat flux and
igniting any evolved gases. These gases are then analysed. The sample is also
placed on a load cell to record any changein mass. Somecone calorimetershave the
facility to control the combustion atmosphere, namely the level of oxygen.
Conventional cone calorimeterscan only provide data in atmosphericconditions (i. e.
21% oxygen). Atmosphere controlled apparatusare not so common due to the high
cost involved, and because fire reaction properties are usually required under
fire
in
conditions,
simulating
a
a
scenario.
material's response
real
atmospheric
Laser photometer beam
(i)
(ii)
Figure 1.4. Schematic(i) [20] and generalview (ii) [21] of a cone calorimeter. The prefix'cone'
comesfrom the shapeof the heater.
1.7.2.
OSU Heat Release Rate Technique
The Ohio State University (OSU) heat releaserate techniqueconsistsof an adiabatic

chamber into which a sample is placed and exposedto a constant heat flux ranging
R. C Easby
PhD Thesis
16
from 35-100 kWm-2. The specimenis ignited using a high temperatureflame, then
heat releaserate is monitored and on occasion, smoke release. This method is not as
common as the cone calorimeter since it has been shown to be prone to a greater
it
Moreover
the
than
cone
calorimeter.
cannot measuremassloss.
error
c*xippiy
Figure 1.5. Schematicof OSU calorimeter[22].
Limiting Oxygen Index (LOI) Test

1.7.3.
The Limiting Oxygen Index (LOI) test measuresthe minimum level of oxygen
neededto supportcombustion.This is determinedby igniting a specimenin a glass
chimneyand controlling the oxygen level in the atmosphere.The LOI index is
defmedasthe minimumpercentage
of oxygenneededfor the materialto bum with a
'candle-like' flame for three minutes. Alternatively it can be describedas the
minimum percentageof oxygen neededfor the flame to spread50mm down the
sample. The apparatusconsistsof an oxygen-nitrogenmixing systemand analysis
equipment.
The value of the LOI test has been questionedby many since it bears little
to a real fire scenario[15].Despitethis it usedin the polymerindustry
resemblance
asa methodof describinginflammability.
R. C Easby
PhD 7hesis
(i)
17
(ii)
Figure 1.6. Schematic(i) [23] and generalview (ii) [21] of LOI test.
Radiant Panel Test

1.7.4.
The radiant panel test involves subjecting a panel of the test material to a known heat
flux. The panel itself is angled at 45* to the heater, directed at the panel's top edge.
During the test, the rate at which the flame travels down the panel along with the
temperaturerise are recorded. From these a Flame SpreadIndex is determined and
because
has
been
This
the
to
criticised,
material
with
others.
method
used compare
like the LOI test the direction of flame spread is downwards. This is deemed
fire
flame
is
real
where
of
a
spread
predominately upwards, and therefore
unrealistic
more rapid.
Flame Propagation & Spread Tests

Many other flame spreadtestsare in existencealthough not as common as the radiant
1.7.5.
panel test. NASA has developedan Upward Flame PropagationTest which subjects
-2
flame
heat
flux
kWM
The
is
determined
75
by
to
spread
average
rate
a
of
a sample
.
dividing the length of flame travel by the bum time. A sample is deemedto have
passedif this number is less than 6.12 mms-1. Other tests include the lateral flame
spreadmethod and the fire tunnel test.
R. C Easby
PhD Thesis
18
Radiant panel
gas supply
Blv
for rodiont ponal
cdrsuppty
(i)
(ii)
Figure 1.7. General view (i) [2 1] and schematic view (ii) [24] of the radiant panel test. Note the top
edge of the specimen angled towards the heater.
Smoke Density
1.7.6.
The most widely used smoke density test is the NBS (National Board of Standards)
flux,
This
by
known
heat
to
operates
a
sample
usually
subjecting
a
smoke chamber.
25 kWM-2 and measuringthe density of the generatedsmoke. This involves using a
,
in
This
light
(a
determine
light
to
transmission.
photometric
system)
monochromatic
turn is used to determinethe specific optical density which is inversely related to the
visibility of the smoke.
(i)
(ii)
Figure 1.8. Schematic(i) [24] and general view (ii) [2 11of a NBS smoke chamber.
R. C Easby
PhD Thesis
Fire Behaviour ofPultruded
1.7.7.
Composites
19
Single Burning Item (SBI) Test
The SBI test simulates a burning waste receptacle (the burning item) in a room
comer. It is essentially an intermediate-scale room comer test (see section 1.7.8),
consisting of two wall panels made of the test material.
The burning waste
receptacle is simulated by a triangular-shaped propane burner generating a heat flux

of approximately 50kWm -2. The test is performed inside a fire room with a fume
extraction system in the ceiling. Heat, smoke and gases released from the burning
walls are extracted and analysed, giving data on temperature, heat release rate, smoke
density, 02 and C02 concentrations.
(i)
(ii)
Figure 1.9. Schematic(i) [21] and close up (ii) of SBI test [25].
1.7.8.
Room Fire Test
There are several intermediate to full scale room fire tests in existence. These are
fire
behaviour
to
the
of composites for use in buildings and ship
establish
used
is
The
these
the room comer test, which is essentially a
of
simplest
compartments.
slightly larger SBI test. In this case though, the heat source is a 30 litre pan of
hexanefuel.
The largest room fire test is the room calorimeter test. This consistsof a room with a
doorway at one end. The ceiling, both side walls, and the end wall without the
R. C Easby
PhD Thesis
20
doorway are clad in the material to be tested. The test material is installed in the
end-use condition, with all joints, fixtures and surface coatings applied. The fire
sourceis a propaneburner placed in one of the rear comers. A fume extraction hood
is positionedover the open doorway in the end wall to remove any fumes for analysis
(seeFigure 1.10).
V-,
.-
-,
""..
5uhm I. -
(ii)
(i)
Figure 1.10. Room comer test (i) [26] and schematicof the room calorimeter test (ii) [27].
This test can provide information on a number of fire reaction properties, including
heat release rate, smoke density, toxic gas emission and flame
time-to-ignition,
In addition, certain fire resistance properties can be determined
spread [ 19].
including,
heat penetration through the walls, bum-through
rate and structural
fire.
to
response
1.7.9.
Furnace Tests
The most commonplaceresistancetest consistsof a structural samplein the form of a

panel, one side of which is subjected to the temperature profile of the furnace.
2
2
from
Om
Im
I
Panels measuring
to
can be tested, depending on the size of the
furnace. Fire resistanceis defined by the time taken for the cool side of the panel to
for
hotspot
180*C
140*C
above
ambient
or
a
of
above ambient to appear.
reach
Furnace tests offer a high level of repeatability compared to other reaction tests
provided they are carried out using the same ftimace, since variability between
R. C Easby
PhD Thesis
furnaces exists.
21
This variation can be brought on by having different lining
materials, different thermocouple positions and different insulating materials in the

furnacewalls.
Figure 1.11. Generalview of a large scalefurnace with the test panel removed [28].
1.7.10.
Pool-fire Tests
Pool fire tests consist of placing a section, sometimes under load, over a tray of a
highly inflammable liquid, depending on the application. Heat transfer and structural
integrity are measured throughout the test. The fire is often subject to convection
issue
making
streams,
repeatability
air
an
currents and
Figure 1.12. Typical pool fire test. In this instancea pultruded composite section is under constant
flexural strain whilst in the fire. The point at which the photographwas taken, the pool fire has
reachedthe point of secondaryflashover.
R. C. Easby
PhD Thesis
22
Furthermore,it has been found that pool fires can exhibit two distinct flashovers. An
initial low temperatureflashover is followed by a much hotter secondaryflashover,
heat
fire
begins
back
heating
itself,
the
the
to
the
reflect
of
on
when
rapidly
remaining fuel.
1.7.11.
Burner Tests
Burner tests provide a less expensive option for resistance testing. One such
technique consists of testing samplesin the region of 100 x 100 squaremillimetres,
heat
flux.
In addition to measuringthe integrity of composite
to
a
constant
subjected
laminates in fire, it can also be used to measure heat flux transmitted to an
it
has
Furthermore
been demonstratedthat it can be used to test
substrate.
underlying
loaded structures[29].
1.7.12.
Jet-fire Tests
A jest fire consistsof a high velocity jet of flame directed towards a specimen. Jetfires are a very severe test due to the high heat flux they generateand the erosive
itself.
jet
A small number of jet-fire rigs exist in the United Kingdom,
the
nature of
United Statesand Norway. They are designedprincipally for testing pipes, vessels
and structures,with or without passive fire protection, under conditions relevant to
the oil and gas industries. The largest of theserigs is operatedby British Gas at their
Spadeadamfacility. This rig is capable of directing a 20 metre horizontal flame of
burning natural gas onto a test specimen,subjecting it to heat flux of -3OOkWrrf2and
a jet velocity of -50ms'. The specimen is exposedto the fire for a fixed length of
time, and subsequentlyassessedfor bum-through and functionality.
Jet-fire tests are very expensiveto carry out due their large scale. Small to medium
scale tests have been developed, capable of subjecting test specimensto high heat
fluxes and gasvelocities, but without the high operating costs[29-3 1].
A British Standardfor determining the resistanceof passive fire protection materials

towardsjet fires is currently under evaluation.
R. C Easby
PhD Thesis
23
Figure 1.13. Jet-fire test underway at the British Gas Spadeadam Facility [32]. The rig shown has
the ability to direct a 20 metre horizontal flame, equivalent to a heat flux of -300kWM-2 at a velocity
of -50ms-.
Table 1.4. Details of the scaleand type of the most common fire tests.
Test Size
Test type
Test name
Small 'bench-
Intermediate-
scale'
scale
Large-scale
Fire Reaction
Cone calorimeter
OSU technique
LOI test
Radiant panel test

NBS smoke
chamber
SBI test
Room fire test
Fire Resistance
Room fire test
Fumace test
Bumer test
Pool-fire test
Jet-fire test
Vf
V/
R. C Easby
PhD Thesis
24
The problem with fire resistanceof GFRP (glass fibre reinforced plastics) is seenby
many as the most significant factor hindering the expansionof the material in many
engineeringapplications, particularly those with strict guidelines for performance in
fire.
Thesis Overview
1.8.
This thesis describesthe developmentof a model capableof predicting failure times

of pultruded GFRP when exposedto fire, under load. This is achieved by building
on existing fire resistancemodels[I-8,33] with laminate theory and data describing
mechanicalproperties as functions of temperature. This model is explained in Figure
1.14 below and is describedin detail in chapter 7. The failure model was verified by
a seriesof fire tests on the pultruded material, loaded in tension and compression.
Mechanicalpropertiesvs. temperature
Fire resistancemodel
Thenno-gravirnetric
analysis
Darameters
Laminatetheory
Fire tests for verification
FAILURE MODEL
IN------------
Figure 1.14. Overview of the failure model developedin this thesis.
Data describing mechanicalproperties as functions of temperatureare unavailable in

the literature so experimentsneededto be designedand carried out. This is described
in Chapter 4. The fire tests carried out to verify the model are detailed in Chapter 5
and results of both the fire tests and the mechanical property tests are discussedin
Chapter6.
R. C Easby
PhD Thesis
25
Further to this failure model, investigative work was also carried out to prolong the
failure time of pultruded sections. This work involved flexing pultruded sections in
pool fires and a furnace, as well as carrying out compression tests on columns
heat
flames
from
burner.
to
a
and
propane
subjected
These experiments are
described in Chapter 5 and the results discussed in Chapter 6.
1.9.
Chapter Summary
Composites are used in many industries in a variety of applications utilising their

high strength-to-weight ratio. Due to their organic matrix material, they are usually
highly flammable, prompting a great deal of research into their reaction and
resistanceto fire.
Many fire reaction and resistancetests exist, ranging in scaleand cost from the small
like
jet
laboratory
high
large
tests
the
test,
to
the
cone
calorimeter
cost
scale
scale,
fire resistancetest (see section 1.7). As of yet there is no standardfire reaction or
fire resistancetest, there is instead a selection of tests that are recognisedby the fire
and compositesindustries.
R. C Easby
PhD Thesis
26
2. Literature Review
2.1.
The 'Slow Burn-Through' Effect
Concernshave been raised about the fire reaction and fire behaviour of composites,
brought on by the combustibility of the organic polymer matrix[34]. In spite of the
inflammable nature of composite materials, they exhibit some interesting and
in
fire.
The most important of these properties is the
useful
properties
potentially
'slow bum-through' effect. In compositesabove a certain critical thickness,the bum
through rate is greatly reduced,as are heat releaseand heat transmissionthrough the
material. These result in increasedfire integrity of the composite, increasing their
in
fire
for
protection applications[35].
use
potential
The factors contributing to the 'slow bum-through' effect are:
"
Transport properties of the laminate. Thermal conductivity and diffusivity

of the laminate is low when comparedto steel.
"
Transport properties of the residual glass. The reinforcement depleted of

any resin material, has a lower thermal conductivity and diffusivity than that
of the laminate.
"
Endotherm due to decomposition and vaporisation. Resin decomposition

highly
are
and vaporisation
endothen-nicand therefore temporarily delay heat
conduction through the laminate.
"
Convection of volatiles.
As the gaseous decomposition products diffuse
through the laminate towards the hot surface,they can be expectedto produce
a cooling effect[35].
The responseof compositesin fire is discussedin greaterdetail later in this chapter.
Findings like the slow bum through effect have enabled performance led design
procedures to replace the more conservative approach of relying heavily on
regulations[36]. This new attitude towards material selection has led to composites
P- C Easby
PhD Thesis
27
being used in fire risk situations, provided they can be qualified and their behaviour
predicted[2].
2.2.
Fire ResistanceModelling
Thermal decomposition of fibre reinforced composites is very complex, combining

thermal, chemical and physical processes. The thermal processesinvolved include
heat conduction through the material, heat absorbed through decomposition, heat
ignition
inflammable
from
the
of
any
reactant gases,and any convective
generated
heat loss from the reactant gasesand any water vapour flowing through and out of
the material. Chemical processesinclude melting, pyrolysis and volatilisation of the
fibres,
and the growth and oxidation of char. The physical processes
and
matrix
include,
thermal softening, thermal expansion and contraction,
may
occurring
internal pressurebuild-up caused by the formation of volatile gases and vapours,
thermally-induced strains, delamination damage, matrix and surface cracking, and
the softening, melting and fusion of fibres. In order to appreciatehow composite
is
important
how
fire,
it
they
to
to
these
and
understand
processes
materials respond
interact.
The needfor reliablemodelscapableof predictingthe thermal,thermal-physical

and
thermal-mechanical
responseof compositeswas recognisedin the 1970swhen
began
be
for
Composites
fibre
to
of
applications.
composites
used aerospace
carbon
all typesarenow beingusedin moreindustriesand in moreapplications,increasing
demandfor reliablefire models.
Accurately modelling the fire responseof compositesis beneficial in several ways.
Firstly, models can rapidly assessthe fire resistanceof new materials for composite
fire
This
the
tests that need to be
can reduce
number of expensive
products.
help
improve
fire
Reliable
the
our
can
models
also
understanding
of
conducted.
behaviour of composites,and further the developmentof new fire-safe materials.
There are several mathematical models describing the fire responseof composites,
varying in complexity from the simple where only heat conduction is considered,to
R. C Easby
PhD Thesis
28
the very complex where several processes described earlier are considered.
However, before any of theses models are discussed it is useful to describe the
is
fire.
that
to
a
composite
of
events
occur
when
material
exposed
sequence
Responseof Composites to Fire

2.2.1.
When a heat flux typical of a high temperaturefire is applied to one side of a
first
is
heat
into
to
the
take
the material.
place
event
conduction
material,
composite
Therateof heatconductionthroughthe materialis governedby the incidentheatflux
diffusivity
is
Thermal
diffusivity
low
for
the
the
thermal
of
material.
and
most
compositesparticularly in the through thicknessdirection. This resultsin a steep
temperaturegradientbetweenthe hot front faceandthe cold rear face. The process
is
by
heat
it
the
the
complicated
conduction
anisotropic
nature
giving
of
of
material,
different values for thermal conductivity in different directions. It is further
fact
heat
by
that
thermal
the
conductivity
and
specific
of composites
complicated
large
Although
has
been
temperature.
a
amount
of
work
carriedout on the
vary with
thermalconductivityof composites[37-46],a theoreticalmodelcapableof predicting
it as a functionof temperatureis not yet available. This is alsotrue of specificheat
[47].
Heat conduction through a laminate will cause it to expand. Due to the thermal
in
direction,
be
It
thickness
through
the
this
expansion
will
non-uniform.
gradient
in
hot
face,
be
decreasing
the
towards the cold face.
greatest
will
At temperaturesbelow the decomposition temperature of the matrix material, heat
transfer occurs largely by conduction. When the temperature rises above the
decomposition temperature,the matrix and fibres degradeendothermically, yielding
gaseousproducts. Thesereaction volatiles flow through the char layer to the surface.
If the temperaturereachesabove 100-150'C, then any moisture presentin the matrix
is, vaporised. The endothermic reactions have the effect of temporarily delaying
conduction through the material.
R. C Easby
PhD Thesis
29
The reaction volatiles and moisture vapour are initially trapped in the material due to
the low gas permeability of composites. This leadsto a rapid rise in internal pressure
is
heated
further.
At
this
the
to
the
point
well above
matrix
material
expand
causing
its glass transition temperature(Tg), making it soft and compliant. This allows the
formation of gas filled pores, and delaminationsand matrix cracksto occur.
Eventually the matrix will become sufficiently cracked and porous to allow volatiles
degraded
to
to
the
through
the
surface
escape
region of the
and moisture vapour
has
delaying
This
the
of.
effect
conduction.
process
composite.
Nr
A
-CNa
Laminate
-
I. r
HEAT
- ---.
A
Flow of
volatiles and
vapour
( ,----Ir
Aj-I, r
AE--.
=
Figure 2.1. Schematicdetailing the flow of volatiles and vapour to the hot face of a composite
laminate, subjectedto heat.
Decomposition continuesuntil the reaction zone reachesthe rear face where the last
is
decomposition
The
to
turned
the
material
and
volatiles.
char
combustible
of
process is usually complete at this point unless the temperatureis high enough to
induce pyrolysis reactions between the char and silica network of the degraded
reinforcement (in excessof 1000'C). If this is the case,then considerablemassloss
can occur [48]. Ablation can also occur at high temperatures,which is accelerated
by high velocity air flow over the surface.
describedarelistedin Table2.1.
All of theprocesses
R. C Easby
PhD Thesis
30
Table 2.1. Summaryof the main processesoccurring when a compositeis exposedto fire.
Anisotropic heat conduction through virgin material and char
Thermally induced strains
Decomposition of polymer matrix and organic fibres
Pressurerise due to formation of combustiongasesand vaporisation of moisture
Flow of gasesfrom the reaction zone through the char zone
Formation of delamination,and matrix cracks
Reactionsbetweenchar and fibre reinforcement
Ablation
Complex models [2,8,35,49-64]
consider many of the processeslaid out in Table
2.1. The majority of mathematical models are all based on one-dimensional heat
conduction.
Heat Conduction Model

2.2.2.1-D
in the studyof heattransferit is usualto considerthe threemodesof thermalenergy
transfer:conduction,convectionand radiation. For simplicity of analysis,nearlyall
modelsfor compositesonly considerconduction.
mathematical
The simplestmodel is a I-D approachthat considersheat conductionthrough a
in
(x) directionwhen heatedfrom one side. The
composite the through-thickness
thatthe compositeis a thick slabwith a uniform in-planetemperature
modelassumes
distribution. Therearfaceis alsoassumedto be adiabatic.The I -D heatconduction
[65] is expressed
as:
[k.. Lax
aT
aT I
PCP
=a
at ax
(2.1)
is
T
the temperature,t is the time and x is the distancebelow the hot surfacein
where
the through thickness direction. p and Cp are the density and specific heat of the
composite, respectively, and kx is the thermal conductivity of the composite in the
through-thickness direction.
p, Cp and kx are assumed to be independent of
temperature,although this is not strictly the case.
R. C Easby
PhD Thesis
31
Henderson Model
2.2.3.
The latestmathematical
modelsowea lot to Henderson[7,8,50-54]. This work was
basedon the theoryof the fire responseanddecompositionof wood,particularlythat
is
[67].
Henderson's
I
[66]
Kansa
Kung
et
al.
equation
expressed
and
as:
of
-D
aT
PCP-=k
at
a'T ak aT
-iXT +-
axax
aT ap
(Qj +h- hg)
thgCP9 --
ax at
(2.2)
where i equals I and 2 for the matrix decomposition and carbon-silica reactions,
is
be
k
taken
to
the through thickness thermal conductivity. The
and
respectively,
first term on the right hand side of the equation considersheat conduction. This is
also the casefor the secondterm, although this considersthe influence of changesin
the rate of heat conduction brought on by variation in the transverse thermal
it
is
As
not possible to model the changein thermal conductivity, this
conductivity.
is based on an empirical relationship. The third term on the right hand side of the
equation considers the effect of volatiles flowing through and out of the damaged
final
The
The
this
this
term on the
term.
of
effect
makes
a
cooling
negative
material.
heat
from
hand
to
the
generation
matrix
relates
or
consumption
resulting
side
right
decomposition and any char glass reactions, where Qj, h and hg are the heat of
decomposition, enthalpy of the solid phase, and enthalpy of the volatile gas,
is
This
term
negative for endothermic reactions, and positive for
respectively.
exothermic reactions. In this final term the decomposition reaction rates are
determinedfrom the massloss rate from the Arrhenius kinetic rate equation:
am
_Mf]ni
-Am
0[Mo
at
RT
(2.3)
where Al, E and ni are the pre-exponentialfactor, activation energy and order of the
determined
These
through thermo-gravimetric analysis. R
reaction.
parametersare
is the universal gas constant; mo, mf and m are the initial, final and instantaneous
massof the material, respectively.
R. C Easby
PhD Thesis
Fire Behaviour ofPullruded Composites
32
Simplified Henderson Model

2.2.4.
A simplifiedversionof the Hendersonmodelwasdevelopedby Gibsonet al.[35]. In
this caseit is assumedthat glass-charreactionsdo not occur,makingin inapplicable
to heat fluxes above-125kWm (1000'C). As the majority of real fires are in the
is
justified.
in
is
fact
This
50-75kWm-2
this
the
simplification
reflected
region of
that the testsdescribedin section1.7aredesignedto functionin this region. Further
during
that
thermal
transport
and
gas
are
assume
properties
constant
simplifications
the decompositionof the matrix. The thermal conductivity and specific heat
is
This
to
the
are
assumed
composite
remain
constant.
expressed
as:
of
properties
)--nE
aT
PC p
-mf
!!
(
ajm
k
PA[m
= -2p
at lax ox
mo
]e
RT(Qp +hc -h
G) -
''G
-LhG
(2.4)
ax
The three terms on the right hand side of the equation relate respectively to heat
conduction through the material, endothermic resin
transportationof heat to the hot face,by volatile convection.
decomposition, and
Much like the Hendersonmodel, no account is taken for char formation, believed to
be beneficial in prolonging integrity, nor does it take into account fibres falling away
from the hot surfaceafter a prolonged exposureto fire. In spite of this the model has
in
determining
fire
the
performance of many types of composite
proven accurate
systems[2,4,35,64,68].
2.3.
Fire Behaviour under Load
Gibson et al. developeda failure model[6] basedon a two layer model developedby
This assumesthe laminate consistsof two explicit layers.
The first is the thermally affected region and is assumedto have zero mechanical
Mouritz et al.[34,69,70].
layer
is
The
the undamaged region consisting of untouched,
second
properties.
virginpaterial with room temperaturemechanicalproperties (seeFigure 2.2).
RC Easby
PhD Thesis
33
A major shortcoming of this approach is that it fails to take into account the
in
progressivechange mechanicalproperties of a material as temperatureincreases;it
for
This
need
a
step
change.
accurate material property data has
assumes
simply
been highlighted by other workers[71,72]. Further work has been carried out to try
in
better
how
this[73]
understanding
of
material properties vary
resulting
and achieve
with temperature.
Thermally affected layer
I
HEAT
Figure 2.2. Schematicof a'compositesampledescribedusing the two layer approach. The sampleis
undergoing tensile loading whilst subjectedto a one-sidedheat flux (x denotesdepth of laminate, xC
denotesdepth of char layer).
Large scale testing has been carried out to determine the fire response of large
pultruded structures[74-76]. The main findings of this work are that pultruded
decking when loaded, is capableof maintaining structural integrity provided that the
heat source is on the tensile side of the beam[75]. Further work has investigatedthe
effect internal water-cooling has on extending failure time of composite flooring[74].
R. C Easby
PhD Thesis
34
Again in this case the fire attacked the tensile side of the structure, highlighting the
in
integrity
if
tension.
to
ability maintain structural
2.4.
Chapter Summary
In spite of being inherently combustible they exhibit some interesting properties in

fire, particularly the 'slow bum-through' effect. In composites above a certain
bum
is
through
the
thickness,
rate
greatly reduced, as are heat release and
critical
heat transmissionthrough the material. These result in increasedfire integrity of the
composite,increasingtheir potential for use in fire protection applications.
Thermal decomposition of fibre reinforced composites is very complex, combining
thermal, chemical and physical processes. Theseare detailed in Table 2.1.
In the study of heat transfer it is usual to consider the three modes of thermal energy
transfer: conduction, convection and radiation. For simplicity of analysis,nearly all
mathematical models for composites only consider conduction. The simplest of
these is a 1-D approach that considers heat conduction through a composite in the
through-thicknessdirection when heatedfrom one side (equation2.1).
The latest mathematical models are based on Henderson's 1-D equation (equation
2.2) which considersheat conduction through the laminate, as well as the effect of
volatiles flowing through the damagedlaminate, and any decompositioneffects.
A simplified version of Henderson's equation was developed, suitable for fires with
heat fluxes less than -125kWm-2 (1000T).
There has also been some investigative work into the fire behaviour of composites
developed
load.
layer
A
that assumesthe compositehas
two
whilst under
model was
two explicit layers. The first is the thermally affected region and is assumedto have
zero mechanicalproperties. The secondlayer is the undamagedregion consisting of
untouched,virgin material with room temperaturemechanicalproperties (see Figure
R. C Easby
PhD Thesis
35
2.2). Large scale testing has been carried out to determine the fire responseof large
findings
The
main
of this work are that pultruded decking
pultruded structures.
is
loaded,
integrity
capable
of
maintaining
structural
provided that the heat
when
sourceis on the tensile side of the beam.
P, C Easby
PhD Thesis
36
3. Thermal Modelling
1-D Thermal Model
Heat flowing through a laminate can be describedby a modified version of Laplace's
form
based
its
In
this
relationship
on the Henderson
one-dimensional
equation[I-9].
equationgives,
06P
,
oT
=a
k-M)-p.
at &( &
'M-(Q
+hc-hG) -
AYG
at p
hG
(3.1)
ax
where T, t and x are temperature, time and through thickness co-ordinates,

k
Cp
and are the density, specific heat and conductivity of the
respectively. p,
flux
is
hG
hc
AY,
the
mass
of
volatiles,
and
are the respectiveenthalpies
composite.
Qp
is
decomposition
the
the
the
and
evolved
gas.
endothermic
energy.
composite
of
The three terms on the right hand side of the equation relate respectively to heat
conduction through the laminate, endothermic resin decomposition, and
transportationof heat to the hot face by volatile convection.
Thermal decomposition of the matrix material can be approximated by a single
dependence,
Arrhenius
temperature
reaction with
am
at =-,
4[m-mf)]ne(-%T)
(3.2)
mo
where m, t and T are the mass,time and temperaturevariables respectively,A, E and

n are the rate constant, activation energy and order of the reaction. R is the gas
constant (8.3144 Jmol-'K'). Polyester resin (and indeed many others) can be
describedin this way, whereby after decomposition a relatively small proportion of
solid material is left behind (char), after all the volatiles have burnt off. Phenolic
in
decomposes
a more complex manner. This takesplace in two distinct stages,
resin
R. C Easby
PhD Thesis
37
(often
followed
by
formation
initial
producing
char
water),
primary condensation
an
at a much higher temperature.
In each case, the decomposition parameters need to be determined by thermoloss
it
heated
based
for
(TGA),
the
upon
a
mass
curve
resin
as
at
gravimetric analysis
heating
rate.
a constant
3.2.
Thermogravimetric
Analysis (TGA) Parametqs
In order to use the thermal model (equation 3.1), kinetic inputs are required (see
Figure 1.14.). The necessarykinetic parametersare;
A, the rate factor (s-')
E, the activation energy (Jmol-'K)
Theseparameterswere obtained from the literature and are detailed in Table 3.1 and
Table 3.2 for polyester and phenolic respectively.
Table 3.1. TGA parametersfor polyester resin [77].

Parameter
Value
1.29 x 1013J7F--
E2x
105 Jmor'k7'
Table 3.2. TGA parametersfor phenolic resin [2]. Note that phenolic resin decomposesin two
distinct phases,requiring two setsof parameters. A percentageof remaining massis also neededto
define the boundary betweenthe two phases.
Parameter
Value
A (Phase1)
5 s-'
E (Phase2)
27200 Jmorlk7l
A (Phase2)
68 sl
E (Phase2)
65200 Jmorlk7l
Mass remaining after Phase1
87%
R. C Easby
PhD Thesis
38
A finite difference model based upon the simplified Henderson equation (equation
3.1) was used to produce data describing the temperature evolution and resin
decompositionthrough laminatesof the pultrusions (Figure 3.2 and Figure 3.3). The
has
0.04056
in
time
seconds
and
a
step
of
records
one
every 76 times steps.
model
Typical input data is detailed in Table 3.3.
Table 3.3. Typical input data for the thermal finite difference model. This data would be for an 8mm.
polyester pultrusion.
Input
Resin type
Duration of test
Thicknessof 3 layers
Fibre volume fractions
Room Temperature
Constantheat flux
Gas constant
A
E
H
Resin density
Fibre density
Fibre thermal conductivity
Resin thermal conductivity
Fibre specific heat
Resin specific heat
Specific heat of gases
Value
Polyester
13 mins
1.00 mm, 6.00 mm, 1.00 mm
0.311,0.527,0.311
21'C
50 kWni2
0.8431 Jmor'K7'
1.29 1013s-'
2x 105Jmor'k7'
2344600 Jkg-1
1200kgM-3
2560 kgM-3
1.09Wr'K7'
0.19 Wm-'K7'
760 Jkg-1C-1
1600Jki'C-1
2386 Jkg 1C1
.5
The model predictsa universaltemperaturerise through the pultrusion. Higher

temperaturesare reachednearer the hot face for a fixed time period, similarly
temperatureincreasesat a greaterrate nearerthe hot face. Residualresin content
decreasesat a greater rate nearer the hot face, correspondingwith the higher
temperatures.
The model splits the laminate into the 50 discrete layers with a node at the boundary
of eachlayer making up a total of 51 nodes(I to 5 1). Boundary nodesexist between
the 2 different reinforcement materials, occurring at nodes5 and 47 (Figure 3.1). For
each time step, the model outputs a resin content and temperature for 7 predetermined nodes. Typical output files ibr an 8mm thick polyester pultrusion are
detailed in the appendix (Tables A and B). Files for both the residual resin content
and temperatureevolution are shown.
R. C Easby
PhD Thesis
csm
Unidirectionalcore
CSIVI
39
Mid -point
(Node 26)
151
-rl
Figure 3.1. Section through a pultrusion highlighting the boundary nodesbetweenthe different
reinforcement materials.
800
700
600
0LD
500
400
Lo
W
CL
300
200
100
0
200
400
Time s
(i)
600
Boo
200
400
600
800
Time s
(ii)
Figure 3.2. Temperatureevolution (i) and Residual Resin Content (RRC) (ii) for a polyester
pultrusion subject to a 50kWm-2heat flux.
The model curves describing residual resin content for the phenolic pultrusion
87%
is
in
kink
This
to
the point
the
resin
content.
corresponding
curves
exhibit a
that distinguishesbetween the two different phasesof decomposition associatedwith
phenolic resins
P- C Easby
PhD Thesis
800,
7.5 mm 8 mm
100
Input temp
40
95
700
Hot face
90
600
85
500
80
400
75
E
300
FO
70
Lo
CL
65
60
55
Hot face
50
200
400
600
Time s
(i)
800
200
400
600
800
Time s
(ii)
Figure 3.3. Temperatureevolution (i) and Residual Resin Content (RRC) (ii) for a phenolic
pultrusion subject to a 50kWm-2heat flux.
The accuracyof this model neededto be verified. This was carried out by comparing
the calculated cold face temperature of the laminate with a measured thermal
response. This thermal responsewas obtained by using a propaneburner test.
3.3.
Propane Burner Test
A small scale fire resistance test was developed[29] using a calibrated propane
bumer[77] capable of producing a constant heat flux. This is directed towards the
sample, which measures11Omm x. 11Omm, and is held vertically- in a steel frame,
leaving an area of 100mm by I 00mm exposedto the burner flame (Figure 3.4). The
material is insulated from the frame by a 5mm layer of Kaowool. This minimised
heat conduction through the frame and preventedany volatiles escapingand burning
at the edgeof the sample.
R. C Easby
PhD Thesis
41
Steel Frame
Figure 3.4. Propaneburner test. Note the insulation usedto minimise heat conduction through the
steel frame.
A thermocouplewas usedto measurean indicative field temperatureI Ommfrom the

front of the sample. It was decided to use a constant incoming heat flux throughout
the test. This would addressany problems causedby material flash-over. Flash-over
by
field
50increase
60
temperature
about
seconds
and would
usually occurred after
1OOOC.
(i)
(ii)
Figure 3.5. Propaneburner test underway,(i) before and (ii) during material flashover.
Thermocouples attached to the rear face of the sample, monitored cold face
temperature.The cold face responsedata is subsequentlycomparedwith temperature
3.1.
from
if the two profiles sufficiently matched
data
equation
calculated
evolution
R. C Easby
PhD Thesis
42
then the material constants used in the thermal modelling can be regarded as
acceptable.
Rear Face Temperature Proflle
3.4.
The cold face responsesof the samplestested with the burner were comparedto the
both
In
casesthe modelled responsematchesthe measuredcold
response.
modelled
face responseto a reasonabledegreeof accuracy(Figure 3.6).
800,
700
600 4
0 500Lo
.a
La 4000
cx
E
300Flo
200
100
400
200
600
80C
Time s
(i)
200
400
Time s
600
800
(ii)
Figure 3.6. Comparisonof the real cold face response(dotted line) with the modelled response(solid
line) for both polyester (i) and phenolic (ii) pultrusions. In both casesthe material was subject to a
50kWrn heat flux, the temperatureprofile of which is also shown.
3.5.
Chapter Summary
A finite difference model based upon the simplified Henderson equation (equation
3.1) was used to produce data describing the temperature evolution and resin
decomposition through laminates of the pultrusions (Figure 3.2 and Figure 3.3).
This model usesTGA data taken from literature.
R. C Easby
PhD Thesis
43
A propane burner test was developed to verify the model's accuracy where the
face
be
compared with the measuredresponseform the
response
can
modelled rear
test.
RC Easby
PhD Thesis
44
4. Mechanical Properties
To model the structural behaviour of a composite material in fire, material elastic
data
in
However
functions
this
temperature
of
are needed.
constantsand strengthsas
literature.
Consequently
in
the
a series of experiments were
area are scarce
developed and carried out to deliver this information. These are detailed in this
is
data
dependent
Fitting
to
the
temperature
subsequent
relationship
a
chapter.
discussedin chapter6.
4.1.
Materials
Pultrusions were provided by Fiberline Composites, and supplied in box and 'I'
from
The
beams.
Simple
these
sections
were
also
cut
profiles.
plate
sectioned
both
from
E-glass
a polyester and a phenolic
with
sections were manufactured
CSM
(UD)
Material
a
needle
of
a
core
with
up
unidirectional
was made
matrix.
60%
layer.
UD
The
core
made
of the material thickness, although
up
weave surface
this did vary slightly from section to section (see Figure 4.1). Details of mechanical
(see
4.1).
by
Table
the
manufacturer
provided
properties were
CSM needleweave surfacelayer,

each makingup 20% of material
thickness
Unidirectional(UD) core,
60% of materialthickness
Figure 4.1. Detail view of an T beam highlighting the different layers through the section. Primary
fibre directions are also shown. All of the supplied material was constructedwith this Mayer system.
P, C Easby
PhD 7hesis
45
Table 4.1. Material specificationssuppliedby Fiberline Composites.
Material
UTS (MPa)
Comp. Strength
(MPa)
Polyester(PE)
Phenolic(Ph)
240
216
240
216
Longitudinal
StiffnessE,
(GPa)
23
18.4
Transverse
StiffnessE2
(GPa)
8.5
6.8
The fibre volume fraction of each layer was also determined alongside matrix and
(see
4.2).
Table
content
void
Table 4.2. Volume fraction data.

Material
Polyester UD
Polyester CSM weave
Phenolic UD
Phenolic CSM weave
R. C Easby
PhD Thesis
Fibre volume
fraction %
52.7
31.1
36.1
31.1
Matrix volume
fraction %%
42.1
63.5
38.9
63.5
Void volume fraction

5.1
5.3
24.8
5.3
4.2.
46
Tensile Strength at Elevated Temperatures
Pultruded material was milled into a 'dog-bone' shape (see Appendix) for tensile
testing. Tests were carried out over a range of temperaturesfrom room temperature
up to 400T.
A uniform temperatureof the gauge length was maintained through a
temperaturecontrolled, aluminium heatingjacket (see Figure 4.2). Once the desired

temperature was reached, an increasing tensile load was applied until failure
occurred. Data were recordedand stress-straincurves were plotted.
N
Tensile load
Temperature
controlled heating
jacket
III
Cartridge heater
Figure 4.2. Tensile heating rig assembly. Note that heating only takesplace in the gaugelength of
the sampleto prevent the material slipping in the grips of the test frame.
(i)
(ii)
Figure 4.3. Tensile test carried out at room temperature on a polyester sample (i), and a view of the
assembly with external insulation removed (ii).
R. C Easby
PhD Thesis
47
Compressive Strength at Elevated Temperatures
4.3.
Material was cut to size (see Appendix) and positioned in a compressivetesting rig
designed to both suppressbuckling, and provide uniform heating (see Figure 4.4).
The rig and samplewere heatedto the desired temperature.Once up to temperature,
an increasingcompressiveload was applied until failure occurred. This was carried
from
400T.
for
temperature
to
temperatures
ranging
room
up
out
Data were
recordedand stressstrain curves were plotted.
Anti-buckling devices/heating plates
Cartridgeheaters
Figure 4.4. Compression rig assembly. Note the anti-buckling devices, designed to suppress the first
mode of global-buckling, and also heat up the sample.
Figure 4.5. View of the compression rig A ith external insulation removed for clarity. Note the
thermocouples used to monitor and control the cartridge heaters.
R. C Easby
PhD Thesis
4.4.
Composites
48
Longitudinal (EI) & Transverse (ED Stiffness at Elevated

Temperatures
Stiffness measurementswere carried out according to BS EN ISO 14125: 1998. A 3bend

[78].
16:
1
For
tests
the
these
thickness
to
rig
ratio
of
point-bend with a span
(see
4.6).
Tests
Figure
insulated
temperature
thermally
were
controlled
and
was
desired
Once
the
temperaturewas reached,
temperatures.
of
carried out over a range
load was applied in the form of a dead weight (100g). The temperatureand load
data
during
deflection
10
for
were
period
of
seconds,
which
a
were maintained
load
10
I
the
after
was applied.
seconds
and
recordedat
Span equivalent to 16t,

where t= material thickness
Weight hanger
LOAD
Figure 4.6. Longitudinal and transversestifftiess testing rig. Note the span-to-thicknessratio of 16:1,
and the isothermic chamberto allow teststo be carried out at constanttemperatures.
R. C. Easby
PhD Thesis
49
Deflection data were converted to provide stiffness values (GPa) over the
temperaturerangeby using the equation,
E-
4815
(4.1)
load,
I
length
flexural
W
I
is
E
the
the
the second
the
modulus,
applied
of
span,
where
b
deflection.
data
Once
10s
the
the
and
measured
of
area
was
reached
and
moment
recorded, the load was removed and the sample relaxed. After a short period the
temperature of the rig was slowly increased to the next desired value and the
for
The
transversemodulus, E2was
repeated
procedure
procedurerepeated.
The mechanical properties El, E2.,aT, and acl as functions of temperaturewere all
failure
(see
in
Chapter 7).
the
a
model
of
construction
used
4.5.
Chapter Summary
To model the structural behaviour of a composite material in fire, material elastic

functions
in
data
However
temperature
this
strengths
as
of
are
needed.
and
constants
in
literature.
Consequently a series of experiments were
the
area are scarce
developedand carried out to deliver this information. They consistedof temperature
designed
heating
to maintain a constanttest sampletemperature.
rigs,
controlled
The material testedin this thesis was provided by Fiberline Composites,and supplied
in box and T sectioned beams. Simple plate sections were also cut from these
from
E-glass
both
The
manufactured
with
were
sections
a polyester and a
profiles.
Material
(UD)
was
made
up
of
a
unidirectional
matrix.
core with a CSM
phenolic
needle weave surface layer. The UD core made up 60% of the material thickness,
R. C Easby
PhD Thesis
50
from
(see
4.1).
Figure
Details
did
to
this
section
section
of
slightly
vary
although
by
(see
4.1).
Table
the
manufacturer
provided
mechanicalpropertieswere
R. C Easby
AD Thesis
51
5. Fire Testing Under Load

The simplest,method of assessingthe fire resistance of a composite system under
load is to carry out small scale tests equivalent to stress-rupturetests[3,34,70,71,
79]. Although they cannot be used for qualification purposes, they can provide
failure
information
indications
of
modes,
and
also
provide
about the
useful
effectivenessof fire protection, without the expenseand complexity of a larger scale
test.
Tensile Tests in Fire

Tensile sampleswere fabricatedmeasuring500mm x 75mm with the long side being
direction.
longitudinal
The
loaded
in
the
samples
were
with a constant tensile
cut
load and subjectedto a flame from a propane burner (Figure 5.1). The heat flux of
2
bumer
the
was calibrated to 50kWM, based on hot face temperatureand distance
from the front of the bumer to the front of the sample. The heat flux was kept
by
the
test
throughout
monitoring the temperature from the 'hot-face'
constant
thermocouple. The time taken for the sampleto fail from the moment the burner was
turned on (time to failure) was recorded for several loads. Ultimate tensile strength
determined
failure
time of 1 second.
and
also,
recorded
with
a
was
0
Tensile load
. Pool
N-type'hot-face'
thermocouple-
Tensile sample
Ll.-*
Propanebumer
Tensile load
Figure 5.1. Arrangement for tensile fire test using a propaneburner as a heat source.
RC Easby
PhD Thesis
(i)
Composites
52
(ii)
Figure 5.2. Front view (i) and side view (ii) of a polyester pultrusion undergoing a tensile test whilst
subjected to a 50kWm-' heat flux.
5.2.
Compression Tests in Fire
Compressive sampleswere fabricated measuring 120mm xI 00mm again with the

long side being cut in the longitudinal direction. The samples were held in a
impact
in
Boeing
the
compressionafter
principle with
constrained compressionrig
test frame[80]. This was to suppressglobal buckling of the samplesduring testing
(Figure 5.3), while at the same time allowing samplesof a large surface area to be
load,
loaded
in
Once
and
the
compressive
with a constant
sampleswere
tested.
place
heat
flux
bumer,
from
flame
the
of which was monitored
to
a
propane
subjected a
for
loads.
Ultimate
failure
Time
before.
to
was
several
recorded
as
and maintained
I
failure
determined
time
of
also
a
and
recorded
with
was
compressive strength
it
little
failure
The
when
occurred,
rapid
and
warning.
event,
was
very
with
second.
This test was repeatedfor the two materials with an inturnescentcoating applied to
the hot face.
The inturnescent material was supplied by Clariant, details of which are found in
Table 5.1.
R. C Easbv
PhD Thesis
Composites
53
Sample
Anti-buckling guides
Propane burner
-
Compressive load
Slots milled to sample thickness
Figure 5.3. Arrangementfor compressionfire test using a propanebumer as a heat source. Note the
use of anti-buckling guides. Thesesuppressglobal buckling allowing sampleswith a large surface
areato be tested.
(i)
(ii)
Figure 5.4. Phenolic pultrusion undergoinga compressiontest whilst subjectedto a 50kWM-2heat

flux (i), and a polyester pultrusion after undergoinga compressiontest without to fire (Ii). The
purposeof such a test was to establisha value for ultimate compressivestrength,recordedas a failure
occurring after I secondof exposureto fire.
R. C. Easby
PhD Thesis
54
Table 5.1. Details of the type of Clariant inturnescentsystemstestedusing the compressionfire test.
All systemswere testedon both polyester and phenolic pultrusions
Substrate material
Type of Clariant intumescent coating
Polyester
100 phr Exolit AP740
Polyester
Polyester
Polyester
Phenolic
Phenolic
Phenolic
Phenolic
R. C Easby
PhD Thesis
5.3.
55
Pool Fire Test under Load
A rig was designed to flex beam sections (I and box) to their maximum design
curvature whilst in a pool fire (see Figure 5.5). Load was applied using a hydraulic
cylinder. The hydraulics and sensitive measuringequipmentwere protectedby metal
shielding
(not shown). Load was monitored with a load cell and change in stroke of the
cylinder with and LVDT.
The section under test was clamped at each end between the assembliesas shown,
leaving a gauge length of I m. The section was then flexed to its design
curvature.(see Figure 5.6). This correspondedto a bend ratio of 1:100 (deflection:
gaugelength).
Pool fire tray
Figure 5.5. Pool-fire bending rig.
Oncethe necessaryload wasreachedthe pool fire was lit. This consistedof 9 litres
2in
0.69m
The load was kept constant
of paraffin a tray with an effectiveareaof
during the test and time to failure was recordedalongsideload and deflection,6.
R. C Easby
PhD Thesis
56
Temperaturewas also recorded along the gauge length of the section to determine
how hot the fire was, and how it varied. Failure was sudden,usually in the order of
100s.
Figure 5.6. Details of the type of bend each section was subjectedto. The top skin is in compression,
and the bottom in tension. (N/A = Neutral axis).
Maintaining a steady temperatureproved difficult.
The pool-fire would exhibit an
initial flash-over reaching approximately 100T. This would then causethe un-bumt
heat
to
up sufficiently to cause the much larger, secondary flashover (see
paraffin
Figure 5.7).
Figure 5.7. Typical temperatureprofile of a pool-fire. Note the initial flash-over followed by the
much larger one.
R. C Easby
PhD Thesis
57
Furnace Test under Load
5.4.
A smaller bending rig was designed and fabricated (Figure 5.9) to fit around a
furnace. The furnace in question was capable of following the SOLAS fire curve
(seeFigure 5.8), therefore overcoming the problem of repeatability experiencedwith
the pool fire.
800-1
100
200
300
400
Soo
600
700
800
Time s
Figure 5.8. SOLAS curve (thick line) comparedwith furnace temperature(thin line).
The test section was positioned through the centre,of the oven with both ends
lined
beam's
It
that
the
was
ensured
central
axis
up with that of
clamped as shown.
the lever arms. Once in place the load was slowly applied until the bend ratio of
1:100 was achieved.
The load was controlled by applying a torque to a threadedshaft as shown. As soon
as the load had stabilised the furnace was ignited. Throughout the test the load was
kept constant. Time to failure was recorded.
R. C Easby
PhD Thesis
Composites
58
Torque applied here
Figure 5.9. SOLAS furnace bending rig. The pultruded section is loadedby applying a torque to the
threadedconnectingrod.
(i)
(ii)
Figure 5.10. View through the inside of a section flexed to its design curvature (i), highlighting the
flexibility of the section. View of a section undergoingfurnace testing (ii).
R. C EasbI,
PhD Thests
5.5.
59
Columns under Compression in Fire
Results from the pool-fire and furnace tests demonstratedthat all sectionsfail in the
flexed
design
A
100
to
their
new
curvature.
maximum
seconds
when
region of
by
intention
designed
this
the
of overcoming
problem extending
with
experimentwas
failure time. The new test took the form of a compression test involving short
basis
fire
barriers
The
to
the
test
of
was
and
systems.
various
with
coated
columns,
failure
if
time compared to the virgin
would
prolong
each coating/system
see
heat
flux.
Compression
to
was chosenover
an
enveloping
subjected
when
material,
flexing as the loading method because it meant that a smaller section of material
Compression
fire
be
the
to
was also the
with
relevant
protection.
prepared
needed
failure mechanismin all of the beamstestedin flexure,
Sections of 60x6Ox5mmcolumns were cut to a length of 90mm. These were then
The
between
two
assemblywas put under constantcompressionand
platens.
placed
load.
designated
to
a
compressed
Whilst under compression, the column was

(see
burners
heat
flux
from
to
a pair of opposing propane
subjected a surrounding
Figure 5.11). The burners were set to 5OkWm-2and monitored and maintained as in
the tensile and constrainedcompressiontestsbefore. Time to failure at severalloads
was recorded.
Base plate
Compressiveload
Figure 5.11. Arrangement for column compressiontest using a pair of propaneburners as a heat
source. Note the opposingburner flames, providing an enveloping heat.
R. C Easby
PhD Thesis
60
Ultimate compressivestrengthwas also determined and recorded with a failure time

of 1 second.When failure occurred, it was rapid and with little warning particularly
at higher loads. The columns provided useful basis for testing fire coatings and
insulations. The test was repeatedfor columns with fire protective coatings/systems
applied. (seeFigure 5.12).
Column
i. ) Bare column
ii. ) Coatedcolumn
iii. ) 'Boxed column'
Figure 5.12. Details of column coatings and systemstested,ranging from simple coatedcolumns to
more complex systemsinvolving a pultruded sleeve.
Table 5.2. Details of type of fire protection testedon the pultruded columns.
Column Material
Type of protection system
Description
Polyester
Coating
Clariant intumescent
Polyester
Coating
Kerin, ceramic based
Polyester
Coating
Geo-polymer
Polyester,Phenolic
Coating
'Tinned'
Polyester
'Boxed column'
Air insulation
Polyester
'Boxed column'
Kaowool
The 'boxed columns' included an insulating layer. Both air and Kaowool were
tested as insulation. K-type thermocoupleswere attached at key interfaces on the
'boxed columns', as well as on the inside of the column (see Figure 5.13).
Temperatureevolution at thesepoints was recordedfor the duration of the test.
R. C Easby
PhD 7lesis
61
Figure 5.13. Top down view of a section through a 'boxed column', highlighting the position of the
thermocouples.
(i)
(ii)
Figure 5.14. View of an unprotected polyester column prior to fire testing (i). Note the opposing
bumer arrangement. A coated column undergoing fire testing (ii). Note the area where a section of
the fire protective coating has fallen away, exposing the column beneath.
5.6.
Chapter Summary
Several fire tests were developed to assess the fire resistance of a composite system
under load.
Tensile and compressive samples were subjected to a heat flux of
50kWM-2 from a propane burner whilst loaded in a test frame. Time to failure was
recorded in each case, for different tensile and compressive loads.
Intumescent
coatings were also tested on the compressive samples.
Short columns of the same material were also tested in a similar way. These too,
provided a basis for testing a series of coatings including supplementary fire
protection.
R. C Easbv
PhD Thesis
62
A pool fire test was also developedinvolving a rig that was capableof flexing long
'F sections and box sections to their design curvatures. Several beamswere tested
and their time to failure was recorded. A similar, smaller scale test was developed
using a furnace capable of following a SOLAS fire curve. This improved upon the
pool fire test's repeatability, and test turnaround.
R. C Easby
PhD Thesis
63
6. Results
Mechanica
roperties
To investigate high temperature applications of composites, particularly how they

in
fire,
it
is
be
to
a
mechanically
useful
able to expressthe elastic constants
perform
in
form
function
data
Finding
temperature.
this
a
of
proved very
as
and strengths
difficult.
Therefore a series of high temperature mechanical property tests were
it
fitting
Once
the
results
of
were
obtained
a
case
a suitable
out.
was
carried
temperature dependent relationship.
This was made easier by the fact that
thermosettingresins only go through a single transition phasebefore the resin begins

to decompose. This occurs at the glass transition temperatureTg. This meant that a
suitable temperaturedependentrelationship only really neededto fit this region.
A polynomial in temperature can be used to describe the variation of an elastic
].
in
This
transition
this
region[9,81
method often requires a polynomial of
constant
describe
6
the relationship across the whole transition region.
to
accurately
order
This method is flawed, since the relationship only behavesreliably within the fitted
region of the data.
For this thesis it was decided to use a function basedon the hyperbolic tan function
(tanh),
[2PU
P(T) =P+
+ PR
]JI
T,
)D
tanh[k(T
-
(6.1)
k
is
PR
Pu
the
are
and
property
and
un-relaxed
relaxed
values
a
where
respectively,
constant describing the breadth of relaxation, T the absolute temperature, and Tg the
absolute temperature of the mechanical glass transition (where P is equivalent to the
mean of Pu and PR). This is detailed in Figure 6.1.
R. C. Easbv
PhD Thesis
350
64
pu
300 250 200

CL
0
E
IL 150 -
100
PR
50
0
0
50
150
100
Temperature
200
250
Figure 6.1. Propertyvariation with respectto temperaturedescribedwith a4 parameterrelationship.
6.1.1.
Problems arising from the Material Cross Section
The pultruded material in this thesis consisted of a unidirectional core, with CSM
layers
(described
in
Figure 4.1 and Figure 6.2).
needles weave outer
Figure 6.2. Close-upof a section Of pUltrudedpolyester, highlighting the three separatelayers.
R. C EasbI,
PhD Thesis
65
This causedproblems during the tensile testing phase. The unidirectional material
had a tendency to pull through the CSM needle weave outer layers (Figure 6.3). It
by
decided
this
to
carrying out tensile tests on the
problem
overcome
was
for
full
those
the
to
too,
section of material.
supplement
unidirectional material
(i)
(ii)
Figure 6.3. Close-up of a tensile test sampleof the full sectionmaterial, highlighting the core
intact.
inner
leaving
failing
(ii)
the
(i),
the
core
the
through
and
outer
skins
outer skins
material pulling
Unidirectional samples were also tested in flexure (EI and EA compression tests
this
The
full
tested
In
was
the
material
only.
section
material
on
out
were carried
7.
detailed
in
Chapter
for
which
is
analysispurposes
way
6.1.2.
Tensile Strength
Tensile tests were carried out on both the unidirectional (core) material and the full
thickness material at varying temperatures. Stress-straincurves of the tests were
plotted.
Figure 6.4 shows the stress strain curves for the core material of a polyester
pultrusion.
The curves are in two distinct groups. The first set is the curves for the
tests carried out at lower temperatures below the material's Tg (glass transition
temperature). In this set, tensile failure occurred at higher stresses and strains. The
second set is the tests which were carried out above the material's Tg. Tensile
R. C. Easbv
PhD Thesis
66
strength for this group is much lower. This behaviour is as expectedand is evident
for all the tensile tests(seeFigure 6.5, Figure 6.6, Figure 6.7).
50*C
Generaltrend of
drop In strengthas
temperature
increases
0.01
0.005
0.015
0.02
0.025
0.03
Strain
Figure 6.4. Stress-straincurvesof unidirectional (core) polyester pultrusion undergoingtensile tests

at varying temperatures.Note the two distinct groups of tests,thosecarried out below Tgand those
above.
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0.045
Strain
Figure 6.5. Stress-straincurves of the full section polyesterpultrusion undergoing tensile tests at
varying temperatures.
R. C Easby
PhD Thesis
67
The drop in strength due to the material passing through it glass transition
temperature(Tg) causingthe resin to soften, was greaterthan anticipated. Consider
Equation 6.2 which is basedon the Rule of Mixtures Equation,
a-
vf
(I-vf)
+07.
=C;
(6.2)
where a* is the failure strength of the composite, q; is the failure strength of the
fibres, u,, the stressin the matrix at the failure strain of the fibres, and Vf the fibre
volume fraction. When the resin contribution to composite strength is considereda
much lower drop should be expected.
This highlights that this is not the only processoccurring. This phenomenonis still
is
but
believed to be the 'composite action' effect on the material,
discussion
under
whereby when a material is below its Tg all the reinforcement is subjected to the
levels.
Once the matrix softens (when T> Tg) any waviness or
same strain
misalignment in the fibres becomes apparent, causing fibres to fail at different
in
strains, this casereducing strength. This effect would be processdependent. For
instance,pultruded compositeswould be affected lessby this phenomenonthan those
that were laid up by hand. This is becauseany wavinesscausingthis effect would be
is
the
material pulled through the die. However, the fibres will still snag
reducedas
against each other and twist around each other causing this effect, particularly when
you considerthe large volume of fibres being pulled through the die at once.
At the higher temperatures;both the polyester and phenolic pultrusions;maintain a
high tensile strength. This is becausetensile strength is ultimately dependenton
fibre strength,which doesnot really deteriorateuntil 800'C.
At lower temperatures(below Tg) the phenolic material was has a higher tensile
strength than the polyester material, and is therefore capable of reaching higher
levels of strain, 5-6% comparedto 34% strain for the polyester. This was true for
both the unidirectional (core) material, and the full thicknesspultrusion.
Fire Behaviour ofPullmded Composites
68
600
500
400
0
(4 300
(ix
13
Co
200
100
0.01
0.02
0.03
0.04
0.05
0.06
Strain
Figure 6.6. Stress-straincurves of unidirectional (core) phenolic pultrusion in tensile testsat different
temperatures.
450 -1
17*C
0.01
0.02
0.03
0.04
0.05
0.06
Strain
Figure 6.7. Stress-stain curves of the full section phenolic pultrusion undergoing tensile testsat
varying temperatures.
R. C Easby
PhD Thesis
69
6.1.3.
Compressive Strength
Compression
testswerecarriedout on full thicknesspultrusions.Stress-strain
curves
of the testswere plotted (Figure 6.9 and Figure 6.10). Testson the core material
werenot deemednecessarysincethe CSM/needleweaveouterskinswould provide
negligiblecompressive
strength.
The
stress-strain
compression
tests.
curves
follow
the familiar
saw-tooth
profile
The main feature is the level of compressive
above Tg. This is noticeably
associated
with
strength retained
lower than with the tensile results, highlighting
the resin
dependency of compressive strength.
The failure mechanism of compressive failure is very different to that of tensile.

Failure involves the formation of a band of kinked material[82].
Failure is initiated
in a region where the fibres are misaligned out of plane of the laminate. This region
experiences high levels of shear loading between the fibres, eventually triggering
failure by local shear deformation.
Figure 6.8. Close up of a compressionsampleof phenolic pultrusion, highlighting the band of kinked
material.
R. C Easbv
PhD Thesis
70
350
300
250
to 200
0
Co
150
100
50
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
0.018
Strain
Figure 6.9. Stress-strain curves of the full section polyester pultrusion undergoing compression tests
at varying temperatures.
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
0.018
Stra In
Figure 6.10. Stress-straincurves of the full sectionphenolic pultrusion undergoing compressiontests

at varying temperatures.
P, C Easby
PhD Thesis
6.1.4.
71
Longitudinal (Ej and Transverse (ED Stiffness
Modulus tests were carried out on both the full thickness material and the
in
longitudinal
direction
(EI)
in
the
the
the
core
and
core
material
unidirectional
transversedirection (EA
Mechanical Properties vs. Temperature

6.1.5.
Throughoutthis section,curves have been fitted to the data points using the 4
parameterrelationship(Equation6.1),detailedat thebeginningof this chapter.
Figure 6.11 shows the results of tensile measurements(CYTO
of the polyester based
function
of temperature. Figure 6.12 shows the results for the phenolic
material as a
basedmaterial. Testing was carried out on both the full thickness of material (UD
between
CSNVneedleweave skins), and the UD core itself
core sandwiched
The
material retained a large proportion of its strengthat high temperature. This is due to
the strength of the reinforcement, which in the case of E-glass retains a significant
least
its
800T.
to
up
at
strength
of
proportion
The results for the full thicknessmaterial are not a true representationof the material
tensile strength as a function of temperature. This is due to the skin material failing
before the core material (Figure 6.3).
The results show a much higher tensile strength for the phenolic pultrusion when
comparedto the polyester pultrusion. If the fibre volume fractions of the materials
(Table 4.2) are considered,one w ould expect the polyester material to outperform the
phenolic material. However it is difficult to accuratelypredict the tensile strengthof
a fibre-reinforced material since you can never be sure if all the fibres are carrying
any applied load. Fi bres may follow a wavy or undulating path through the material,
load-carrying
their
reducing
ability. A possible reasonwhy the phenolic material has
a higher tensile strength in spite of its glass content becomes apparent when the
manufacturing processis considered. The phenolic material is manufacturedunder a
R. C Easby
PhD Thesis
72
higher resin injection pressure than the polyester. This may have the effect of
improving
in
fibre
alignment,
material strength.
reducing any waviness
450 -1
400
350
300
250 -1
200
Full section
150
100
50
0
0
50
100
150
200
250
300
Temperature C
Figure 6.11. Tensile strength(GTI)VS.temperatureof polyester pultrusion. Note both unidirectional

(core) and the full section material were tested.
600-1
500
m 400CL
Corematerial
Im
c
B 300U)
.
AA
pp
Full secton
5 200
100
0
0
50
100
150
200
250
300
Temperature C
Figure 6.12. Tensile strength(GTOVS-temperatureof phenolic pultrusion. Note both unidirectional

(core) and the full section material were tested.
R. C Easby
PhD Thesis
Fire BehaviourofPultrudedComposites
73
Figure 6.13 and Figure 6.14 show the compressivestrength(cc,) of the polyester
pultrusion,andthe phenolicpultrusionwith respectto temperature
respectively.The
drop
in
increasesabovethe
compressivestrengthastemperature
curvesshowa steep
on the conditionof theresin.
materials'Tg,stressingits dependency
350
300
250
200
iso
CL
E
0 100
0
50
0
0
0
20
40
60
80
100
120
140
160
180
200
Temperature C
Figure 6.13. Compressivestrength(acl) vs. temperatureof polyesterpultrusion. Note only the full
section material was tested.
300
250
200
150
100
50
0
0
20
40
60
80
100
120
140
160
180
200
Temperature C
Figure 6.14. Compressivestrength (crcl) vs. temperatureof phenolic pultrusion. Note only the full
section material was tested.
R. C Easby
PhD Thesis
74
The phenolic pultrusion has a considerably lower compressive strength than the
larger
in
is
by
This
the
much
void
caused
content the phenolic material,
polyester.
24.8% (in the UD core) comparedto 5.1% (UD core) for the polyester.
Figure 6.15 and Figure 6.16 illustrate the longitudinal stiffness (EI) as a function of
temperaturefor the polyester material and the phenolic material respectively. Data
for both the unidirectional (core) and the full material are shown.
In the caseof the polyester pultrusion the data show the familiar drop in magnitude
does
defies
it
Tg
The
the
through
and
region.
convention
as passes
phenolic material
not demonstratethis behaviour.
Data points for both 1 secondand 10 secondshave been shown. This is gives some
indication of any creepthat is taking place whilst the material is undergoing testing.
35
13
25
20
is
9
0
0
Full section
10
la
5-1
0
0
so
IIII
100
150
200
250
300
Temperature C
Figure 6.15. Longitudinal stiffness (EI) vs. temperatureof polyester pultrusion. Note both
0
denote
full
(0
(core)
I
tested
the
stiffness
and
section material were
at second, at 10
unidirectional
seconds).
R. C Easby
PhD Thesis
75
35
30
13
a
L 25
0
.E
13
43
Im
c
13
0
a
0
a
0
E
Core matedal
a
0
20-
-A
0
Full section
15-
10
.9
5
0
0
50
100
150
200
250
350
300
400
Temperature C
Figure 6.16. Longitudinal stiffness (EI) vs. temperatureof phenolic pultrusion. Note both
0
full
10
(core)
denote
I
(0
thickness
tested
and
stiffness
at
second,
material
were
unidirectional
secondsfor core material, A denote I secondand 0 10 secondsfor full section material).
14
12
13
10
8
6
4
2
0
0
20
40
60
80
100
120
140
160
180
200
Temperature C
Figure 6.17. Transversestiffness (ED vs. temperatureof polyester pultrusion. Note only
unidirectional (core) material was tested(0 denotestiffinessat I second,0 10 seconds).
Figure 6.17 and Figure 6.18 show the transversestiffness (E2) of the unidirectional
(core) polyester material as a function of temperature and that of the phenolic
is
Once
data
for
10
I
second
and
seconds
again
respectively.
shown, and the drop in
stiffness as the material passesthrough the Tgregion is apparentin both materials.
kC Easby
PhD Thesis
76
2.5
1.5
Core matedal
13
200
250
0.5
0
0
50
100
150
Temperature C
Figure 6.18. Transversestiffness (132)vs. temperatureof phenolic pultrusion. Note only

0
(core)
(0
denote
I
tested
material
was
stiffness at second, 10 seconds).
unidirectional
The parametersused to construct the curves using Equation 6.1 to describe all the
functions
of temperatureare listed in Table 6.1.
mechanicalpropertiesas
A value for shear modulus (G12)at room temperaturewas establishedby carrying
out a flexural stiffness test on a sample cut at a 45" angle from the longitudinal. The
(E45)
value
was
was converted into a value for shear modulus
subsequentstiffhess
through Equation 6.3, basedon the compliancematrix,
G12 -"2
E45
22
EI
+
-E2
2v
(6.3)
longitudinal
El,
E2
and
v
are
stiffhess, transversestiffness, and poisons ration
where
respectively. The resultant value for shear modulus (1.2 GPa) was low enough to
assumethat as the temperaturerose it would drop rapidly, and higher temperature
valueswould be next to zero.
R. C Easby
PhD Thesis
77
Table 6.1. Table of parametersused to describemechanicalpropertiesas a function of temperature

(Pu, PR,Tg and k are respectively,un-relaxedproperty value, relaxed property value, mechanicalglass
transition temperatureand a constantdescribing the breadth of relaxation). Shearmodulus has been
omitted.
Material
Phenolic
Polyester
PR
Tlg
PU
PR
Ts
354
242
150
0.03
500
347
100
0.035
(Full material)
CYTI
230
220
150
0.03
400
278
100
0.035
ac, (Full material)
320
60
95
0.045
270
100
100
0.02
E, (UD core)
32
14
150
0.01
26
22
300
0.005
E, (Full material)
13
100
0.025
22
19
300
0.05
15.2
0.7
45
0.025
2.12
1.2
50
0.06
Parameter
PU
or, (UD core)
E2(UD Core)
R. C Easby
PhD Thesis
78
Fire Testing under Load
6.2.
Tensile and Compressive Tests

6.2.1.
Both tensileand compressivefailure would occur with little warning. Onceplies
beganto fail, completefailure occurredvery soonafter. The tensilestrengthcurves
both exhibit a high level of residualstrength,due to glassretainingits strengthat
(Figure
(Figure
6.19).
high
The
temperatures
curves
stress
rupture
compressive
very
6.20) show a more rapid decline in strengthwhen comparedto the tensile stress
fact
This
the
that pultrudedcompositesare particularly
underlines
curves.
rupture
failurewhensubjectedto fire.
to compressive
susceptible
500
450
400
350
300
Phenolic
(a
Polyester
250
-9
U)
4)
Z 200
150
100
50
0
10
100
1000
10000
Time s
Figure 6.19. Tensile stress-rupturecurve of both polyester 0 and phenolic 0 materials. Test
sampleswere subjectedto a heat flux of 50kWm. Note the I secondpoints representthe material
ultimate tensile strength.
RC Easby
PhD 77iesis
79
350
---n
10
100
10000
1000
TIme s
Figure 6.20. Compressivestress-rupturecurve of both polyester 0 and phenolic 0 materials. Test

sampleswere subjectedto a heat flux of 50kWnf2. Note the I secondpoints representthe material
ultimate compressivestrength.
6.2.2.
Pool Fire test under Load
The sectionstestedin the pool fire experimentsall failed in a time periodwithin the
is
6.2).
100
It
beam
(see
'I'
Table
the
the
with
seconds;
exception
of
of
order
thoughtthe rapid failure of the 'I' sectionwas causedby the heatand flamesbeing
ableto attackboth sides,effectivelyremovingthe benefitof a cold face. The inside
(see
face,
failure
box
times
their
the
sections
acts
as
a
cold
extending
noticeably
of
Figure6.21).
Table 6.2. Failure times for sectionstestedin a pool fire. Note the rapid failure of the 'I' section.
Beam description
Time to failure (secs)
60 x 60 x 5mm box (polyester)
42
100 x 100 x 6nim box (polyester)
83
88
120 x 60 x 6mm T (polyester)
15
RC Easby
PhD Thesis
80
Figure 6.21. Section through a polyester box section showing the undamaged 'cold face' on the
inside of the section.
In each case failure occurred suddenly and with little warning. Each section failed
due to local buckling in the compressiveskin (see Figure 6.22). The reason being
that compressive stiffness is heavily dependent on resin depletion. The size of
section had a limited effect on failure time with the larger 100 x 100mmbox sections
fairing better than the smaller 60 x 60mm section. This could perhapsbe due to the
heat taking longer to conduct through the larger sections,therefore taking longer to
reach its T. value, softening the material.
(i)
(ii)
Figure 6.22. Local buckling in the compressiveskin of a polyesterbox section (i), and a side view of
a polyester T section (ii), highlighting the failure point, initiated by local buckling of the top web.
R. C Eashv
PhD Thests
81
6.2.3.
Furnace test under Load
The sectionstestedin the furnaceagainall failed within a similartime period,in the
orderof 100seconds.Thephenolicsectionslastedslightly longer(approximately30
seconds)thanthe polyestersections(seeTable6.3). In eachcasefailurewassudden
little
by
local
buckling
in the compressive
skin of the
warning,
caused
andcamewith
section. Onceagainthis canbe put downto the fact that stiffnessin the compressive
sideof thebeamis heavilydependent
on the conditionof theresin.
Table 6.3. Failure times for sectionstestedin a SOLAS fire curve furnace.
Beam description
Time to failure (secs)

60 x 60 x 5nun box (polyester)
124
60 x 60 x 5mm box (phenolic)
150
60 x 60 x 5mm box (phenolic)
152
120
The consistencyof results highlights the consistencyof the SOLAS fiimace test.
P- C Easby
PhD Thesis
82
Columns under compression

6.2.4.
Compressiontestswere carriedout on both polyesterand phenoliccolumns. For
bothmaterials,thebarecolumnshaddimensionsof 60 x 60 x 5mm,and90mmtall.
Figure 6.23 shows the effect various systems for improving fire performance on a
polyester column. All the systemswere tested at least once at a serviceable load,
equivalentto 46 MPa.
250
200
150
100
50
Figure 6.23. Resultsof compressiontestson polyester box columns. From left to right: A indicates
Clariant coatedcolumn, 0 bare column, 0 Kerin coatedcolumn, X Geo-polymer coatedcolumn, 0
'tinned' column* 'boxed column' with air insulator, + 'boxed column' with Kaowool insulator.
Note that the I secondpoint representsthe ultimate compressivestrengthof the column.
It can be seenthat of all the coatedcolumns testedthe 'tinned' column performed the
best, extending failure time by approximately 110 seconds. Out of the 'boxed
column's tested, the Kaowool insulated one performed the best, improving failure
time by approximately 575 seconds,comparedto the 128 secondsof the air insulated
one.
The temperatureprofiles of the two 'boxed columns'testedare detailedin Figure

6.24andFigure6.25. It is seenthatthe temperature
of the columnincreasesfar more
in
insulated
the
air
case,thanin the Kaowoolinsulatedcase. '
rapidly
R. C Easby
PhD Thesis
50
100
ISO
200
250
83
300
Time 9
Figure 6.24. Temperatureprofiles through a 'boxed column' with air as the insulator.
900-1
Time a
Figure 6.25. Temperatureprofile through a 'boxed column' with Kaowool as the insulator.
Figure 6.26 details the effect the 'tinned' systemhad on a phenolic column. Failure
time was extendedbut not as significantly as with the polyester column. In this case
it was only 23 seconds.
kC Easby
PhD Thesis
Fire Behaviour ofPul"ded
Composites
84
140-1
120
10
100
1000
Time s
Figure 6.26. Resultsof compressiontestson phenolic box columns. From left to right: 0 indicate
bare column, 0 'tinned' column. Note that the I secondpoint representsthe ultimate compressive
strength of the column.
6.3.
Experimental reliability
This section discussesreliability issuesrelating to the range of experiments carried

out in this thesis.
6.3.1.
Mechanical properties at elevated temperatures
During this series of experiments,a great deal of care was taken to ensurethat the
material was at the statedtemperatureof the test. In each case(tensile, compressive,
and flexure), the test utilised temperature controllers to maintain, and monitor the
sample's temperature. During the tests it was ensured that the sample was fully
saturatedat the desired temperatureby leaving it for a period of 20 to 30 minutes
before any load was applied. Further to this, the flexure test rig also contained a
'dummy' sample,inside of which were thermocouplespositioned at the mid point of
its section (Figure 6.27) and at 3 points along its length.
R. C Easby
PhD Thesis
85
r\/\,
thermocouple
Li ------------------I
t/2
II
'dummy sample'
b/2
Figure 6.27. Sectionalview through a 'dummy' flexural sample,indicating the position of the
thermocoupletip in the samplemid-point.
All of the test rigs were well insulated during the heating and saturation periods.
There was potential for the tensile and compressive tests to lose heat when the
into
frame.
Efforts were made to ensurethat
the
test
contact
with
sample/rig came
this period was as short as possible to reduce the heat sink effect. This involved
be
head
frame
to
test
the
to
that
the
was set up and ready go with only
cross
ensuring
load
cell re-calibrated.
moved and
Overall, I feel that this seriesof experimentsare very reliable and repeatabledue to
the use of temperaturecontrolled heating systems,and sound experimentalpractice.
The only major shortcoming of these experiments was that the temperature
had
been
limited
400'C.
higher
If
to
temperature
were
rated
controllers
controllers
available, a more marked decreasein mechanicalproperties of the phenolic material,
in tension and flexure may have beenobserved.
The accuracyof someof the curvesfitted to the resultsmaybe broughtinto question.

The majority of the curves follow the experimentaldata well. Some of them
howeverdo not. In spiteof this it was decidedto continuewith the hyperbolictan
functionfor reasonsof simplicity. In any casethe curvesaretheresimplyto provide
XC Easby
PhD Thesis
86
the failure model with a reasonably accurate trend describing how the mechanical
propertiesdecayat elevatedtemperatures.
6.3.2.
Fire testing under load
Several problems exist with the experiments where the heat source is a propane
burner.
The burner is calibrated using a pressuregauge. This pressureties in with a local

field temperatureat the hot face of the sample. This in turn correspondsto a heat
flux. During the mechanicaltesting with the burner, this pressuregauge facility was
not available. This meant relying on the field temperature of the hot face as an
indication of heat flux. Although in theory this would be sufficient, it is worth
mentioning that this temperatureis susceptibleto the material flashing over. Careful
monitoring of the material during testing is required, to ensurethat the gaspressureis
not reducedto compensatefor the rise in field temperaturedue to flashover. This is
to ensurethat the sampleis subjectedto a consistentheat flux.
The test frame used in these experimentswas an Avery 50 tonne Universal Testing
Machine. This is a hydraulically operated machine, relying on an old fashioned
indicate
dial
to
the load being applied. In spite of the errors associatewith
analogue
the machine's age (worn components, leaking hydraulics etc), the machine was
always reliable and never broke down. indeed, it's worth mentioning that the
physical size of the machine prevented it overheating as a result of the flames from
the propaneburners.
Overall, largely due to the problems associatedwith controlling the propaneburner,
the results of these experiments can be only be regarded as indicative of a general
trend.
R. C Easby
PhD Thesis
6.3.3.
87
Pool fire test under load
The pool fire tests can only be regardedas indicative tests. This is largely due to the
unpredictable nature of the pool fire itself. Its temperatureis extremely susceptible
to thermal currents, and in this case,the elementstoo. The 'double flashover' effect
of the fire also makesexperimentsdifficult.
The rig itself also has errors associatedwith it. The load is applied with a hydraulic
hand-pump.
load
is
kept
by
The
cylinder,
operated
with
a
manual
constant
monitoring the output form a load cell, and adjusting the pumping accordingly. This
method is reasonablyaccuratebut the time delays from the data acquisition system,
human
factor,
to
the
not
mention
results in a constant load curve that looks a little
shaky and noisy.
6.3.4.
Furnace test under load
Similar to the pool-fire test, the fiimace tests can only be regarded as indicative.
Although the SOLAS furnace is more reliable and predictable than the pool fire,
errors are present. Variations in furnace temperaturewill exist dependingon things
like whether or not the fiimace has been used recently and the ambient room
temperature. The test also assumesa constanttemperatureinside the fiimace. This
cannotbe guaranteed.
The largest reliability issue exists with maintaining a constant load. In the case of
the furnace test, hydraulics were deemed unnecessary, as the heat was wholly
contained inside the furnace. This allowed the operator to stand next to the furnace
to apply the load. This allowed a more simple method of applying a torque to a
threadedbar to apply the load. This method did however introduce a more error into
the experiment. This was due to the mechanical 'lag' in the system, something not
experiencedwith hydraulics due to the incompressiblenature of the hydraulic fluid.
Further to this error was that associatedwith the feedback system as in the pool-fire
test.
R. C Easby
PhD Thesis
88
89
7. Modelling Behaviour under Load

A laminate analysis failure model was basedon classical laminate theory [10,83-85].
This is describedin section 7.1 below. The model requires input from the thermal
equation described in section 3.1, namely temperature evolution and residual resin
content (RRC) through the thickness of the material. -It also requires material
mechanicalproperties as functions of temperature,detailed in chapter6. The stepsof
the model are detailed in Figure 7.1.
Mechanicalpropertiesas a function
of temperature(E(T), cr(T))
Temp
Laminateconstitutive
equations
I -D thermal model
Failuremodel
RRC
Figure 7.1. Flow chart describing the stepsthrough modelling behaviour under load.
7.1.
Laminate Constitutive Equations
Laminate theory [10,83-85] is a commonly used tool in the compositesindustry to

'
determine the strength and stiffness of a laminate. For the purpose of analysis, the
pultruded material used in this thesis will be regarded as a laminate with a finite
number of plys. Under isothermal conditions, applied forces and bending moments
are related to the resultant mid-plane strains and curvaturesby,
l FV
-= [2 iil[wol
sf'- ff B
R. C Easby
PhD Thesis
(7.1)
90
where, N and M are matrices of the normal loads and bending moments in the
laminate:
N=
N.,
mx
NY
My
mxy
and
N.,
Y
(7.2)
and, Woand k are the mid-plane strains and curvatures:
k.,
x
Wo
and
= --,
ky
(7.3)
k.,
y
-Y-Y
The A, B and D matrices are defined as,
;i=2:
hk
hk
_n
f Udz
-nh,
Z fUzdz
-,
k=I ht-,
k=I hk-1
F)
2dZ
i&Z
Y,
(7.4)
k=I hk-,
where Q is the matrix of ply stifffiess constantstransformedto the co-ordinate

systemof the laminate:
Q12
Q= Q12Q22Q26
Q16
Q26
(7.5)
Q66
These constantsvary from ply to ply according to the orientation within each layer.
In this case they also vary in the z-direction due to the effects of temperature
variation and resin content.
R. C Easby
PhD Thesis
91
The boundary conditions of. the problem are usually in the form of in-plane loads and
(7.6)
inverted
fully
the
more preferable:
of
equation
version
moments,making
[21
Lo
To -I
7.2.
(7.6)
Input of Mechanical Properties vs. Temperature
As statedearlier, the mechanicalproperties as functions of time are a key input to the

in
6.1.
Table
by
defined
6.1
These
out
set
equation
using parameters
are
model.
Thesevalues are inputted to the model through equation7.7:
P(T) = [Pu + (Pu - PR)(I- tanh(k(T - Tg)))]Rn
(7.7)
i.
being
(100%
1,
decimal
is
R
e. virgin material,
expressed
as
a
resin content
where
for
loaded
in
0,
For
being
50%).
0.5
tension,
compressionand
and
material
n.=
and
flexure, n=1. This was basedupon the notion that compressiveand flexural loading
is
largely
dependent
heavily
tension,
to
which
upon resin condition, as opposed
are
based
decided
figures
fibre
1
0
The
on
upon
strength.
of and were
reliant upon
in
this area[79,86].
work
previous
7.2.1.
Mechanical Properties of CSM Skins
The material tested in this thesis has 3 distinct layers, CSM needle weave on the
surfaces and unidirectional
fibres in the central core (Figure 4.1 and Figure 6.2).
This system caused problems whilst determining the materials' tensile strength as a
function of temperature (see section 6.1.1).
This problem was overcome by treating the material as 3 separate layers, and
determining mechanical properties as a function of temperature for each one. The
data for the core material was determined experimentally using the methods set out
in Chapter 4. This was not possible for the CSM material becausethe layers were
R. C Easby
PhD Thesis
92
too thin to separatefrom the core material. The data for the CSM material was
calculatedby using combination of sandwich-beamtheory and data from literature.
These methods are detailed in Table 7.1. The data describing compressivestrength
as a function of temperaturewas determined entirely experimentally. In this case,
splitting the material into its 3 layers was deemedunnecessary. This was due to the
negligible effect the CSM needle weave layers would have on the materials'
compressivestrength.
Table 7.1. Methods used to determinemechanicalpropertiesas functions of temperature,for both

UD core material and CSM needleweave outer layers. All experimentalmethodsare describedin
Chapter4.
Method adopted
Property
Experimental
GT-UD)ac, El-UD, E2-UD
Sandwich-beam
Ecsm
Literature
CFT-Csm
The sandwich-beammethod considersthe material as a typical sandwichbeam and

utilises the expression,
333
EF,,,,=
EuD
+E2CSM
333
12
Lt2
Ll
(7.8)
t2
where EF,,,,,is the flexural modulus of the full section of material, EuD is the flexural
modulus of the core material and Ecsm is the flexural modulus of the skin material.
The thickness of the full section is tj and the thickness of the core material is tj. This
calculates the flexural modulus for the CSM skins from the flexural modulus of both
the core material EuD, and the full section, EF,,Il (both obtained experimentally). The
resulting flexural modulus Ecsm is the same in both perpendicular and longitudinal
directions.
Equation 7.8 assumes a beam of constant width making EF"Ij a function
of the relevant flexural moduli (EuD, Ecsm) and ratio of the material thicknesses (ti,
t2)-
C
Easby
JR.
PhD Thesis
93
The tensile strength of the CSM material as a function of temperaturewas obtained

through the literature. A tensile strength of 100 MPa [ 11,87] was used as a room
temperature tensile strength, degrading to a negligible strength of I MPa. The
degradationin strength is governedby the sameequation as all the other mechanical
data
(Equation
6.1). The rate of degradation matches that of the core
property
material i. e. the samevalues for k and Tg.
All of the mechanicalproperties for both the core material and skin material can be
describedwith equation 6.1, and fitted using the four parametersPu, PR, Tgand k. A
full set of theseparametersused in the modelling is listed in Table 7.2.
Table 7.2. Parametersused to describemechanicalpropertiesas a function of temperaturefor the

laminate failure model. Parametersare listed for both the uni-directional and CSM material. Note
that the parametersfor compressivebehaviour are the samefor both the UD and skin material.
Phenolic
Material
Polyester
Parameter
PU
PR
Tg
PU
PR
Tg
CFT-LTD
354
242
150
0.03
500
347
100
0.035
GT-Csm
100
150
0.03
100
150
0.03
UC-UD
320
60
95
0.045
270
100
100
0.02
crc-csm
320
60
95
0.045
270
100
100
0.02
El-uD
32
14
150
0.01
26
22
300
0.005
El-csm
6.5
0.015
160
20.5
18.7
230
0.008
E2-UD
15.2
0.7
45
0.025
2.12
1.2
50
0.06
E2-csm
6.5
-3
0.015
160
20.5
18.7
230
0.008
R. C Easby
PhD Thesis
7.3.
94
Results
7.3.1.
A, B, D Matrix Evolution
Figure 7.2, Figure 7.3 and Figure 7.4 show the evolution of the A, B, D, matrix
components for an 8mrn polyester pultrusion using the laminate failure model.
Similarly Figure 7.6, Figure 7.7 and Figure 7.8 show this evolution for an 8mm
phenolic pultrusion.
The A matrix components,relating to in-plane loads and deformations, decline over
time reflecting the decline in overall mechanical properties. This decline is more
is
due
in
This
the
the
to
that
material
of
phenolic.
polyester
when
compared
marked
to phenolic compositesretaining mechanical properties at high temperatures. The B
loads
in-plane
describe
interaction
between
the
the
and out-ofmatrix components
plane bending and twisting. This value is initially zero due to the symmetry of the
CSWneedle
in
direction.
This
the
to
through-thickness
the
as
rises a peak
material
larger
burnt
is
imbalance.
The
peak
causing
second,
away
a
symmetrical
weave skin
is causedby further asymmetry as the UD core material is degraded. Finally the D
influence
bending
decline
The
time.
of
with
resistance
matrix componentsgoverning
the progressive asymmetry can be seen with the shoulders in the curves. These
in
the B matrix curves.
the
peaks
coincide with
1
for 8mm polyester and 8mm.
Figure 7.5 and Figure 7.9 1
show the evolution of
D'11
phenolic pultrusions respectively. This is equivalent to the material's 'flexural
decline
in
[88].
Once
declines
(EI)
the
time
this
reflecting
again
over
stiffness'
overall mechanical properties. As with the D matrix components, a shoulder is
visible on the curves, again reflecting the progressive asymmetry. As with the A
matrix parametersthis decline is far more significant in the polyester material, due to
the phenolic material maintaining its mechanicalproperties at higher temperatures.
R. C Easby
PhD Thesis
95
250-1
200
150-
I-
I Uti
Aii
A22
0
0
100
200
300
400
500
600
700
800
Time s
Figure 7.2. Evolution of the A componentof the A, B, D matrix for an 8mm thick polyester
pultrusion exposedto a one-sidedheat flux of 50kWm-2.
100
200
300
400
Soo
600
700
800
Time 9
Figure 7.3. Evolution of the B componentof the A, B, D matrix for am 8mm.thick polyester
pultrusion exposedto a one-sidedheat flux of 50kWnf2.
R. C Easby
PhD Thesis
96
700
600
500
400
300
200
100
0
Figure 7.4. Evolution of the D componentof the A, B, D matrix for an 8mm thick polyester
pultrusion exposedto a one-sidedheat flux of 50kWm.
700
600
Soo
04
E 400
z
r
E! 300
FlexuralStiffness
200
100
0
0
100
200
300
400
500
600
700
800
Time s
Figure 7.5. Evolution of flexural stiffness (I/D'11) for an 8mm thick polyesterpultrusionexposedto a
2
one-sidedheatflux of 50kWni7
RC Easby
PhD Thesis
97
250
200
Aii
E
150
Coinn
A22
50
0iII
0
100
200
300
400
500
600
700
800
Time s
Figure 7.6. Evolution of the A component of the A, B, D matrix for an 8nun thick phenolic
pultrusion exposed to a one-sided heat flux of 50kWff2.
70
60
50
40
30
20
10
100
200
300
400
500
600
700
800
Time s
Figure 7.7. Evolution of the B componentof the A, B, D matrix for an 8mm thick phenolic
pultrusion exposedto a one-sidedheat flux of 50kWm-2.
R. C Easby
PhD 7hesis
Fire Behaviour ofPultruded Composiles
98
1200,
1000
Dii
800
600
D22
400
200
0iIII
0
100
200
300
400
500
600
700
800
Time s
Figure 7.8. Evolution of the D componentof the A, B, D matrix for an 8mm thick phenolic
pultrusion exposedto a one-sidedheat flux of 50kWrn72.
1000900FlexuralSUffness
8007004.4 600 E
5004UU300200
100
0;
0
100
IIII
200
00
400
500
600
700
800
Time a
Figure 7.9. Evolution of flexural stiffness (I/D'Il) for an 8mm thick phenolic pultrusion exposed to a
one-sided heat flux of 50kWrff2.
All of the A, B, D matrix and flexural stiffness curves exhibit a general trend of
decay as time and hence temperatureprogresses. This is true for both the polyester
based pultrusion and the phenolic. This general decay is more marked in the
polyester pultrusion than the phenolic. This correspondswith the results of the
mechanical property experiments described in chapter 6. Bearing this in mind, it
XC Easby
PhD Thesis
99
might be worth considering running the model for a longer time period for the
indication
This
pultrusion.
would
provide
some
phenolic
of the time at which the
phenolic material would begin to suffer significant decay in its mechanical
properties.
This would however require mechanical property data at higher
temperatures.
Strengths Modelling
7.3.2.
In order to model the strengthof the materialit was necessaryto considerit as an
8mmthick laminatewith 17nodes,1 nodefor each0.5 mm in the throughthickness
direction. Strengthwasmodelledby applyingan increasingstrainto the model. For
eachnode,an appliedstresswas calculatedfrom the appliedstrainandEl valueasa
function of temperature.If this appliedstressexceededthe strengthof the material
(basedon materialstrengthas a function of temperature,(Y(T)),then the materialat
thatnodewasdeemedto havefailed,andhavezerostrength.If the appliedstressdid
not exceedthe strengthof the material,then the stressat that noderemainedat the
level of the appliedstress(Figure 7.10). This processwas repeatedat eachnode.
The stressin the materialat any given strainwas obtainedby averagingthe stressat
each of the 17 nodes. The strengthof the pultrusion as function of time was
determinedby taking the maximumstressvalue on the stress-straincurve for each
time step(Figure7.11).
a(T), based on T and RRC
E(T), basedT and RRC
CyApp
Ply stress = aApp

Applied strain E
Ply is deemedto have failed,

and strengthgoes to zero
Figure 7.10. Algorithm describing the processesinvolved in modelling strengthpredictions. This

method was adoptedfor both tensile and compressivestrengths.
R. C Easby
PhD Thesis
100
300-1
250
0.02
0.04
0.06
0.08
106
200
3bO 400
500
Time s
Strain
(i)
(ii)
Figure7.11. Stress-straincurves for a polyester pultrusion subject to 5OkWnf2 heat flux (i). The
maximum stressvalue on each curve for eachtime step is usedto determinethe strengthof the
pultrusion over time (ii).
By presuming that the material strength drops to zero when the applied_stress
is
followed.
its
that
a saw-tooth stress-straincurve
exceeds strength value, assumes
This is acceptablefor compressivestrengthmodelling as Figure 6.13 and Figure 6.14
is
For
tensile
this
modelling,
strength
not strictly true.
show.
A secondmethod was adopted which was basedupon the actual stress-straincurve
data from the tensile tests. The curves were modelled using the empirical equation,
0" =
(7.9)
max[1_e6]
where a is the modelled stressin the material,
an exaggeratedstrengthvalue, E
is the Young's Modulus and E is an applied strain.
A stress-strain curve is
is
This
curve
strain.
superimposedover each,
over
a
range
of
applied
constructed
for
by
E and
the
altering
values
real stress-straincurve
(Figure 7.12). The result
is a family of constructed stress-straincurves, matching the real stress-straincurves
RC Easby
PhD Thesis
101
for that material. This family of curves is used to model the stress at a particular
As
temperature.
soon the stress-strain curve reached the
node, at any given
failure
temperature,
that
occurred.
at
strength
material
maximum
Strain
Figure 7.12. Empirical stress-straincurve governedby the parametersE and q.,.. af represents
failure strengthand ef failure strain.
In this case it was assumedthat the tensile test results for the full section material
failure
tensile
true
the
of the material.
represented
7.3.3.
Tensile Strength Prediction
Figure 7.13 and Figure 7.14 show tensile strengthas a function of time for a
both
figures
In
the two
and
phenolic
respectively.
pultrusion
polyesterpultrusion
have
for
been
shown.
strength
modelling
methods
alternative
The predictions for the polyester material reflect the downward trend of experimental
points well.
The 'saw-tooth' method (dotted line) for strength prediction
is
improved
by
This
the empirical method
the
upon
material
strength.
underestimates
(solid line), which still underestimatesit slightly. The reasonfor this may be because
(all
in
layers)
full
three
the
material section
were used the
stress-straincurves of
function
These
the
tensile
with
strength
analysis.
combined
as
a
of
were
empirical
temperaturedata (Figure 6.11) for the full material section.
RC Easby
PhD Thesis
102
400,
350
300
250
%Na
200150
I-
100
50
0
0
50
100
150
200
250
300
350
400
450
500
Time 9
Figure 7.13. Model predictions for the relationship betweentime-to-failure and applied tensile stress
along with experimentalpoints for an 8mm thick polyester pultrusion, subjectedto a one-sidedheat
flux of 50kWm-2 Predictions are shown for both methodsof strengthmodelling (Saw tooth method,
.
dotted line, and empirical method, solid line).
50
100
150
200
250
300
350
400
450
500
Time 9
Figure 7.14. Model predictions for the relationship betweentime-to-failure and applied tensile stress
along with experimentalpoints for an 8mm thick phenolic pultrusion, subjectedto a one-sidedheat
flux of 50kWm-2. Predictions are shown for both methodsof strengthmodelling (Saw tooth method,
dotted line, and empirical method, solid line).
RC Easby
PhD Thesis
103
The predictions for the phenolic material reflect the downward trend of data. Once
again the predictions underestimatethe material strength. In this casethe saw-tooth
method provides a more accuratefit in the later stagesof the fire.
More accuratefits could probably be achieved if real data for the tensile strength of
the skins was available rather than relying on the literature. On a similar notion the
stress-strain curves used for the empirical curve fitting did not provide a true
representationof the tensile behaviour of the materials. During tensile testing, both
the polyester and phenolic suffered form the uni-directional core pulling through the
layers
6.3).
CSM
(Figure
Consequentlythe stiffness and strengths gained
of
outer
from the stress-straincurvesmaybe an underestimationof the true values.
The correlation between modelled and actual results for the phenolic material is
weak. The modelled results underestimatethe real data a great deal more than for
the polyester material, suggestingthat there is some other factor causing a greater
underestimationthan those already described. Perhapsa greaterunderstandingof the
fire
to
that
phenolic
resin
would go
changes
undergoes
subject
when
physical
finding
towards
out what causes this underestimation. If mechanical
someway
better,
in
300'C
have
fitted
data
this
of
excess
was
available
curve may
property
since the model is reliant on this data.
7.3.4.
Buckling and Compressive Strength Prediction
Figure 7.15 and Figure 7.16 show the evolution of compressive strength over time
for an 8mm.polyester pultrusion and 8mm.phenolic pultrusion respectively. In each
case the buckling responsehas been modelled too. The reason behind modelling
both compressive and buckling failure is down to the fact that the samples tested
could have failed compressivelyor through buckling, or indeed a combination of the
two. The fundamental difference between the two is that compressive failure is
governed solely by the compressivestrength of the material. Buckling failure on the
is
hand
not only dependenton the material's compressive strength; it is also
other
reliant on the material's geometry. The buckling responseis particularly significant
RC Easby
PhD Thesis
104
largely
beams,
long
that
slender
are
used
as
and
pultrusions
one
considers
when
hencemore susceptibleto buckling failure.
Compressivefailure was modelled using the 'saw tooth' method (Figure 7.10). This
both
if
For
the
stress-strain
materials, the
curves.
method was chosen as reflected
in
degradation
degrades
time,
the
mechanical
strength
over
reflecting
compressive
properties.
The buckling responsewas basedon the equation,
abuckling
VDIID22
-Tb7
(7.10)
b
is
is
buckling
is
the
test
the
the
thickness
t
sampleand
of
strength,
whereqb,,,
kjjg
thebreadth.Di, andD22areboth components
of theD portionof theA, B, D matrix.
The constantc is associatedwith the edgesconstraintsof the sample. This varies
from 3.6 if the loadedsampleis simply supportedon all edges,to 7.5,if it is clamped
(see
[85]
Figure
7.17).
at all edges
The buckling curves in Figure 7.15 and Figure 7.16 have been constructedwith the
fit
for
best
for
figure
7.5
This
the
the
the
provides
c constant.
maximum value of
data, and also correspondswith very snug fit of the buckling rig around the samples,
in
be
fully
Further
this
to
to
a
clamped
carried
out
case.
work
needs
corresponding
figure
for c. This is detailed in Chapter9.
find
to
an exact
area
The buckling responsecurves shown in Figure 7.15 and Figure 7.16 degrade over
time reflecting the overall decline in mechanicalproperties. Both curves also contain
a shoulder similar to those visible in the D matrix parameters. This is unsurprising
is
is
it
buckling
that
the
response
a function of DI, and D22.
considered
when
However, in this case only one shoulder is apparent as opposed to two in the D
by
buckling
has
been
This
the
reduced
secondary
shoulder
matrix.
strength equation
7.10.
P, C Easby
PhD Thesis
50
100
ISO
200
250
300
350
105
400
Time s
Figure 7.15. Model predictions for the relationship betweentime-to-failure and applied compressive
2
stressfor an 8mrn thick polyester pultrusion, subjectedto a one-sidedheat flux of 50kWnf
Predictionsare shown for both constrainedbuckling failure and compressivefailure, along with
0
for
both
bare
0,
Results
Clariant
coatedmaterial are shown.
experimentalpoints.
material
and
50
100
150
200
250
300
350
400
Time s
Figure 7.16. Model predictions for the relationship betweentime-to-failure and applied compressive
2
stressfor an 8mm thick phenolic pultrusion, subjectedto a one-sidedheat flux of 50kWnf .
Predictions are shown for both constrainedbuckling failure and compressivefailure, along with
experimentalpoints. Results for both bare material 0, and Clariant coatedmaterial 0 are shown.
P, C Easby
PhD Thesis
106
Results for the intumescejit coated material are also shown. The coated polyester
long
bare
lasted
120
(Figure
7.15)
the
twice
as
seconds,
as
approximately
material
between
different
little
difference
There
the
the
stress.
was
equivalent
at
material
inconclusive.
for
Three
The
the
coated
phenolic
samples
of
proved
results
coatings.
the four coated samplestested failed before a bare sample tested at an equivalent
improved
The
that
on the bare sample's life-span was coated
single sample
stress.
lasted
for
217
AP740.
This
314
Exolit
150phr
the
to
seconds
secondscompared
with
bare phenolic.
For both materials, the predictions for compressiveand buckling failure accuratelyfit
the experimentalpoints.
Sampleedges
Sampleedges
Figure 7.17. View of the constrainedcompressionrig (detailed in Figure 5.11) highlighting the edges
of the test sample. The way in which theseedgesare constrained,has a direct effect on the buckling
responseof the sample.
Overall, the modelled results reveal the dominant effect of exposing compositesto
fire whilst under load. Exposing a composite to the effects of fire alone will have
little dramatic effect, as will applying a load to a composite. Combining the two
however, results in catastrophicfailure. This is reflected in the modelled results by
their downward trend.
RC Easby
PhD Thesis
107
The results are limited by the experimental reliability of the fire under load tests.
This is discussed in detail in section 6.3. Further limitations exist due to the
underestimation of the tensile strength which is discussed earlier in this chapter.
Theselimitations needto be consideredbefore using this model as anything beyond a
tool to determine a general trend. This position can be improved upon by obtaining
real tensile data as a function of temperatureof the CSM outer skins.
However the model's approachdoes provide an excellent foundation upon which to
build a3 layer failure model for marine composites. The 3 layers used in a typical
marine composite ply usually have thicknessesin the magnitude of cm's, allowing
mechanicalproperty data at elevatedtemperaturesfor all of the separatelayers to be
obtained. This would overcomemany of the problems encounteredwith this model.
9C Easby
PhD Thesis
108
8. Conclusions
8.1.
Propane burner test

burnertest,properlycalibrated
Thepropane
canprovidea cheapandreliable
testing.
methodfor fire resistance
8.2.
Thermal modelling
The thermal model basedon the simplified Hendersonequation(equation
3.1) can accuratelypredict temperatureevolutionand residualresin content
througha pultrudedcomposite.
8.3.
Mechanical properties as a function of temperature
designed
to carryouttheseriesof mechanical
propertytestsat
" Theequipment
differenttemperatures
provedreliable. This is dueto goodexperimental
practiceandwell thermallyinsulated
equipment.
" The empiricaltanh relationship(equation6.1) usedto describehow the
in
to
temperature
most
reliable
properties
varyaccording
proved
mechanical
cases.It wasnotedhoweverthat in somecasesthe fit wasperhapsnot so
decided
in
It
to
tanh
to
the
good. was
continue use
relationship thesecases
for simplicity,besides
thefit wasreasonably
accurate.
in
higher
loss
tensilestrengththanexpected.
" Both materialsexperienced
a
Thiswasdueto thelossof 'composite
action.
" Higherratedtemperature
of
controllers
shouldbeusedin futuretests,capable
in excess
temperatures
reaching
of 400'C.
8.4.
Fire testing under Load

Tension
8.4.1.
"
Tensile failure would occur with little warning. Once plies beganto fail,
failure
occurredvery soonafter.
complete
"
Both materialstested retaineda high residual strength,due to the glass

retainingits strengthat high temperatures.
RC Easby
PhD Thesis
109
Due to reliability issues with controlling the propane burner, these tests
should only really be used to indicate a generaltrend.
Compression including intumescents
8.4.2.
"
Compressive
failure would occurquickly andwith little warning.
"
The failure mechanismwas a combinationof compressiveand buckling

failure.
"
Compressivestrength declines more rapidly when comparedto tensile

strength,underlining the fact that pultruded compositesare particularly
"
failurewhensubjectedto fire.
susceptibleto compressive
Inturnescentcoatingsextendedthe failure time for the polyestermaterial,
with little betweenany of the coatingstested. The resultsfor the phenolic
materialprovedinconclusive.
"
Due to reliability 'issueswith controlling the propaneburner, these tests

shouldonly really be usedto indicatea generaltrend.
Pool fire test
8.4.3.
"
Pultrudedsectionsflexed to their designcurvaturesfail after a time in the

orderof 100seconds.
"
Failure always occurredon the compressiveside and the mechanismwas

always local buckling. This is due to compressivestiffnessbeing heavily
"
"
reliantuponthe conditionof the matrix material.

Box sectionsoutperformed'I' sections. This is becausebox sections
effectively had a cold internal face. 'I' sectionsdid not, as they were
damagedon all sidesby the fire.
Thepool fire proveddifficult to predictandthereforelackedrepeatability.
Furnace test
8.4.4.
The temperaturecontrolledfurnacewascapableof matchingthe SOLASfire

curve.
Thetestwasvery repeatable.
Pultrudedsectionsflexed to their designcurvaturesfail after a time in the
orderof 100seconds.
P, C Easby
PhD Thesis
"
110
Once again, failure always occurred on the compressive side and the
mechanismwas always local buckling. This is due to compressivestiffness
being heavily reliant upon the condition of the matrix material.
"
Phenolic sectionsoutperformedthe polyester sections.
"
The temperatureinside the furnace cannot be guaranteed.
Columns including coatings
8.4.5.
"
Thetwin burnermethodadoptedin the experimentprovidedan enveloping

heatflux.
"
In their barestate,thephenoliccolumnsoutperformedthepolyestercolumns
at a typical designstress.
"
The 'boxed' columnsperformedbestout of all the systemstested,although

the silveredcolumnresultsarealsopromising.
"
Due to 'reliability issueswith controlling the propaneburner, thesetests

be
really
shouldonly
usedto indicatea generaltrend.
8.5.
Modelling
A, B, D Matrix Evolution
8.5.1.
"
The A matrix components, relating to in-plane loads and deformations,

decline over time reflecting the decline in overall mechanical properties.
This decline is more marked in the polyester material when comparedto that
of the phenolic.
"
The B matrix componentsdescribethe interaction between the in-plane loads

and out-of-plane bending and twisting. This value is initially zero due to the
symmetry of the material in the through-thicknessdirection. This rises to a
peak as the CSM/needle weave skin is burnt away causing a symmetrical
imbalance. The second, larger peak is causedby further asymmetry as the
UD core material is degraded.
"
The D matrix componentsgoverning bending resistancedecline with time.

The influence of the progressiveasymmetry can be seenwith the shouldersin
the curves. Thesecoincide with the peaksin the B matrix curves.
P, C Easby
PhD Thesis
The model also shows the evolution of
1
D',
III
which is equivalent to flexural

I
stiffness.
Tensile Strength Prediction
8.5.2.
"
Themodelsomewhatunderestimates
the tensilestrengthof both the polyester
"
andphenolicpultrusions.
In the caseof the 'saw-tooth' modelling method,this may be causedby
"
the strengthof the skin material.

underestimating
In the caseof the empiricalmodellingmethod,using the full sectionstressstraincurvesmaybe the causeof this.
"
It may be worth consideringrunningthe modelfor a longerlengthof time in

the phenoliccaseto determineif and when the strengthpredictiontails off.
Howeverin orderto do this, mechanicalpropertydataat highertemperatures
would be required.
Compressive and Buckling Strength Prediction
8.5.3.
"
Themodel,aftersomeinitial underestimation,
accuratelypredicts
for
both
failure
polyesterandphenolicpultrusion.
compressive
"
Bucklingresponsereflectsthe trendin experimentaldata,but is not fitted

for
the constantc, governingthe edgeconditions.,
value
exact
with an
Overall the 3 layer failure model provides an excellent basis on which to develop a
failure model for 3 layer marine compositestructures(sandwich panels).
RC Easby
PhD Thesis
112
9. Future Work
Thermally Induced Deformations

When a pultrusion or laminate is subject to a heat flux, distortions can occur in the
absenceof any externally applied force. The effect of theses thermally induced
deformationscan be accountedby modifying equation 7.1 to give,
;i
][zo
-
goT ]
- 11--
-T
-T
k
and
06
where,
(9.1)
in
induced
the
thermally
are
strainsand curvaturesrespectively,
the absenceof any externallyappliedload. This would furtherimprovethe accuracy

failure
the
model.
of
9.2.
Modelling of fire protective coatings
It would be possible to develop the thermal model to include the use of fire
do
In
to
so it would be necessaryto determine the thermal
protective coatings. order
behaviour of such coatings. In particular, thermal conductivity, any radiative effects
the material may have, and any volatile gasesthat may be emitted. This data will
help to understandhow a coating would affect temperatureevolution in a pultrusion
or laminate.
9.3.
Sandwich panels
The three layer approach to analysing the pultruded material in this thesis could
integrity
be
in
fire.
Provided
to
the
sandwich
of
model
panels
structural
easily used
that data describing the mechanical properties of the constituent layers, as functions
is
temperature
obtainable,the model should work.
of
RC Easby
PhD Thesis
9.4.
113
Buckling Response
In order to fully determine an accurate buckling responsea new compression rig

(Figure 5.3) would need to be designed. The new rig would need to be capable of
between
different
thus
thicknesses,
c
relationship
a
allowing
accepting material of
and t to be determined.
9.5.
Structures
There is great potential for extending the model to -consider full size structural
in
In
beams
box
beams
flexure
'1'
this
thesis.
tested
the
order
and
as
elementssuch
to develop the model to consider large sections in flexure, severalnew inputs would
be required.
Firstly, the model describing the resin degradationand temperatureevolution would
having
into
This
be
the
take
a closed
of
to
effect
should
account
adapted.
need
like
box
beam.
T
It
like
that
box
beam,
was noted
an
or an open section
a
section
beams outperform T beams in a pool fire due to the cold face on the inside of the
box.
Secondly,the effect the fire hason the stiffnessof the beamneedsto be accounted
for. Stiffnessdegradesas the materialpassesthroughthe transitionregion. This
degradationis more marked in compressionthan tension, due to compressive
is
dependent
heavily
(Figure
9.1).
This
being
problem
uponresincondition
stiffness
highlightedby the flexurestestscarriedout in this thesis,whereeverysectiontested,
failed on the compressive
side.
RC Easby
PhD Thesis
114
NIA_
FIRE
(i)
(ii)
Figure 9.1. Comparisonof compressiveand tensile moduli of a pultruded compositesectionbefore

(i) and durng a fire (ii), highlighting the rapid decline in compressivestiffhess.
RC Easby
PhD Thesis
115
References
1.
Anilturk, D. and Chan, W. S., Structural stability ofcomposite laminated

Composite
Materials,
high
Journal
temperature
to
of
orfire.
column exposed
2003.37(8): p. 687-700.
2.
Dodds, N., Gibson, A. G., Dewhurst, D., and Davies, J.M., Fire behaviour of
3.
689-702.
laminates.
Composites:
A,
2000.31(7):
Part
p.
composite
Gibson, A. G., Wright, P.N. H., Wu, Y. -S., and Evans,J.T., Laminate theory
2006.
load
In
infire.
press,
under
ofcomposites
analysis
4.
Looyeh, M. R.E., Bettess,P., and Gibson, A. G., A one-dimensionalfinite

GRPpanelsfor
thefire-performance
offshore
of
simulationfor
element
&
for
Fluid
Heat
Journal
Methods
International
Numerical
of
structures.
Flow, 1997.7(6): p. 609-625.
5.
Lua, J. and O'Brien, J. Fire simulationfor wovenjabric compositeswith

temperatureand massdependentthermal mechanicalproperties. in
Compositesin Fire 3.2003. Newcastleupon Tyne, England.
6.
Gibson, A. G., Wright, P.N. H., Wu, Y. -S., Mouritz, A. P., Mathys, Z., and
Gardiner, C.P., The integrity ofpolymer compositesduring and afterfire.
Journal of CompositeMaterials, 2004.38(15): p. 1283-1308.
7.
Henderson,J.B. and Wiecek, T.E., A mathematicalmodel to predict the

thermal responseof decomposing, expandingpolymer laminates.Journal of
CompositeMaterials, 1987.21: p. 373-393.
8.
Henderson,J.B., Wiebelt, J.A., and Tant, M. R., A modelfor the thermal

responseofpolymer compositematerials with experimentalverification.
Journal of CompositeMaterials, 1985.19: p. 579-595.
9.
Gibson, A. G., Wright, P.N. H., Wu, Y. -S., Mouritz, A. P., Mathys, Z., and
Gardiner, C.P., Modelling residual mechanicalproperties ofpolymer
10.
81-90.
Composites,
2003.32(2):
Rubber
Plastics,
p.
and
afterfire.
composites
Eckold, G., Design and Manufacture of CompositeStructures. 1994,
Cambridge:WoodheadPublishing Ltd.
11.
CompositeMaterials in MaritimeStructures. Cambridge OceanTechnology

Series4, ed. R.A. Shenoi and J.F. Wellicorne. Vol. 1. FundamentalAspects.
1993, Cambridge: CambridgeUniversity Press.
XC Easby
PhD Thesis
12.
Image courtesy of www.ndt.net.
13.
Image courtesy of www.jytt-intl. com.
14.
Image courtesy ofFiberline
15.
Babrauskas,V. and Peacock,R.D., Heat releaserate: Thesingle most
116
Composites AIS.
important variable infire hazard. Fire Safety Journal, 1992.18(3): p. 255272.
16.
Sorathia,U., Lyon, R.E., Ohlerniller,T., andGrenier,A., A reviewoffire test

SAMPEJournal,1997.33:p. 23-31.
methodsandcriteriafor composites.
17.
Babrauskas,
V. andWickstr6m,U. Therational development
bench
of
scale
fire testsforfull-scalefireprediction.in 2ndInternationalFire Safety
ScienceSymposium.
1989.
18.
V., Effectivemeasurement
Babrauskas,
techniquesforheat,smokeand toxic
fire gases.Fire SafetyJournal,1991.17:p. 13-26.
19.
V. Fire testmethodsforevaluationoffire retardantefficacyin

Babrauskas,
polymericmaterials.in Fire RetardancyofPolymericMaterials.2000.New
York: MarcelDekker,Inc.
20.
21.
Imagecourtesyof www.ps1c.ws.
Imagecourtesyof www.fire-testing.com.
22.
V., Fire testmethodsforevaluationoffire-retardantefficacyin

Babrauskas,
in
Fire RetardancyofPolymericMaterials,A.F. Grand
materials,
polymeric
andC.A. Wilkie, Editors.2000,MarcelDekkerInc.: New York. p. 81-113.
23.
24.
25.
Imagecourtesyof www.bolton.ac.uk.
Greene,E., Marine Composites,
www.marinecomposites.
com.
26.
Imagecourtesyof www.mpauin-stuttgart.de.
Imagecourtesyofwww.rockwooLcom.
27.
ISO 9705:1993Fire testson buildingmaterialsandstructures.
28.
Imagecourtesyof www.annilcfs.com.
Gibson,A. G., Wu, Y.-S.,Naas,A.L., andMcNaught,R.J.A low costburner
29.
techniqueforthedevelopment
andmodellingof laminatesinfire. in
Composites
in Fire 3.2003.NewcastleuponTyne,England.
30.
Folkers,J. Fire testingandperformanceoffiber glasspipe. in Composites

in
Fire. 1999.Newcastle-upon-Tyne.
P, C Easby
PhD Thesis
117
31.
Hill, P.S. and White, G.C. Fire testing andperformance ofcomposite
32.
materials. in Compositesin Fire. 1999.Newcastle-upon-Tyne.

Image courtesy of www.spadeadam.biz.
33.
Gibson, A. G. Basic mechanismsoffire damagein organic matrix composites.

in Compositesin Fire 3.2003. Newcastleupon Tyne, England.
34.
Mouritz, A. P. and Mathys, Z., Post-fire mechanicalproperties ofmarine
35.
polymer composites.Composite Structures,1999.47(1-4): p. 643-653.

Gibson, A. G., Wu, Y. -S., Chandler,H.W., Wilcox, J.A. D., and Bettess,P., A
modelfor the thermalperformance of thick compositelaminates in
hydrocarbonfires. Revue de Unstitut Frangaisdu P6trole, 1995.50(l): p. 6974 (special issue).
36.
The Honourable Lord Cullen, ThePublic Enquiry into the Piper Alpha
Disaster. 1990,HMSO: London.
37.
Kalagiannakis,G., Hemelrijck, D.V., and Assche,G.V., Measurementsofof

thermalproperties of carbonlepoxyand glasslepoxyusing modulated
temperaturedifferential scanning calorimetry. Journal of Composite
Materials, 2004(38): p. 163-174.
38.
Tai, H., Equivalent thermal conductivity oftwo- and three-dimensional

orthogonallyfiber-reinforced compositesin one-dimensionalheatflow.
Journal of CompositesTechnology & Research,1996(18):p. 221-227.
39.
Gowayed,Y., Hwang, J.-C., and Chapman,D., Thermal conductivity of

textile compositeswith arbitrary preform structures. Journal of Composites
Technology & Research,1995(17):p. 56-62.
40.
Havis, C., Peterson,G., and Fletcher, L., Predicting the thermal conductivity
and temperaturedistribution in alignedfibre composites.Journal of
Thermophysics,1989(3):p. 416-422.
41.
James,B., Wostenholm, G., Keen, G., and McIvor, S., Prediction and
measurementof the thermal conductivity ofcomposite materials. Journal of
PhysicsD: Applied Physics, 1987:p. 261-268.
42.
Fanucci, J.P., Yhermal responseofradiantly heatedkevlar and

graphitelepoxy composites.Journal of CompositeMaterials, 1987(21):p.
129-139.
R. C Easby
PhD Thesis
43.
118
Ott, H. -J., Thermal conductivity of compositematerials. Plastics,Rubber

Processingand Applications, 1981(1): p. 9-24.
44.
Han, L. and Cosner,A., Effective thermal conductivities offibrous

1981(103):
387-392.
Journal
Heat
Transfer,
of
p.
composites.
45.
Progelhof, R.C., Throne, J.L., and Ruetsch,R.R., Methodsfor predicting the

thermal conductivity of compositesystems:A review. Polymer Engineering &
Science,1976.16(9).
46.
Springer, G. and Tsai, S., Thermal conductivity of unidirectional materials.

Journal of CompositeMaterials, 1967(l): p. 166-173.
47.
Zalameda,J.N., Measured through-the-thicknessthermal diffusivity ofcarbon

fiber reinforced compositematerials. Journal of CompositesTechnology &
Research,1995(21):p. 98-102.
48.
Ladacki, M., Silicon carbide in ablative chars. Journal of the American

Institute of Aeronautics and Astronautics, 1966(4):p. 1445-1447.
49.
Dimitrienko, Y. I., Thermomechanicalbehaviour ofcomposite materials and

structures under high temperatures:2. Structures. Composites,1997(28A):
p. 463-471.
50.
Tant, M. R., Henderson,J.B., and Boyer, C.T., Measurementand modelling of

the thermochemicalexpansionofpolymer composites.Composites,1985(16):
p. 121-126.
51.
Henderson,J.B. and Tant, M. R., Measurementof thermal and chemical

properties ofa glass-filledpolymer compositeto high temperature.High
Temperatures-HighPressures,1996(18):p. 17-28.
52.
Henderson,J.B. and Doherty, M. P., Measurementofselectedproperties ofa

glass-filledpolymer compositeto high temperature.High Temperatures-High
Pressures,1987(19):p. 95-102.
53.
Florio, J., Henderson,J.B., and Test, F.L., Measurementof the

thermochemicalexpansionofporous compositematerials. High
Temperatures-HighPressures,1989(21): p. 157-165.
54.
Florio, J., Henderson,J.B., Test, F.L., and Hariharan, R., A study of the effects
ofthe assumptionof local-thermal equilibrium on the overall thermallyinduced responseofa decomposition,glass-filledpolymer composite.
International Journal of Heat & Mass Transfer, 1991(34):p. 135-147.
&C Easby
PhD Thesis
55.
119
Sullivan, R.M., A finite elementmethodfor thermochemicallydecomposing

State
University.
1990,
The
Pennsylvania
polymers.
56.
Sullivan, R.M. and Salamon,N.J., A finite elementmethodfor the

thermochemicaldecompositionofpolymeric materials -L Theory.
International Journal of Engineering Science,1992(30):p. 431-441.
57.
Sullivan, R.M. and Salamon,N. J., A finite elementmethodfor the

thermochemicaldecompositionofpolymeric materials - II. Carbon phenolic
939Science,
1992(30):
International
Journal
Engineering
p.
of
composites.
951.
58.
Sullivan, R.M., A coupledsolution methodfor the thermochemicalresponse

ofdecomposing expandingpolymeric composites.Journal of Composite
Materials, 1993(27):p. 408-434.
59.
Pcring, G.A., Farrell, P.V., and Springer, G.S., Degradation of tensile and
shearproperties of compositesexposedtofire or high temperature.Journal
of CompositeMaterials, 1980(14):p. 54-66.
60.
McManus, H.L. and Springer, G.S., High temperaturebehaviour of

thermochemicalbehaviour ofcarbon-phenolic and carbon-carbon
2061992(26):
Analysis.
Journal
Composite
Materials,
L
p.
of
composites,
229.
61.
McManus, H.L. and Springer, G.S., High temperaturebehaviour of

thermochemicalbehaviour of carbon-phenolic and carbon-carbon
230IL
Results.
Journal
Composite
Materials,
1992(26):
p.
of
composites,
255.
62.
Dimitrienko, Y. I., Thermomechanicalbehaviour ofcomposite materials and

high
temperatures:1. Materials. Composites,1997(28A): p.
structures under
453-461.
63.
Dimitrienko, Y. I., Thermal stressesand heat-mass-transferin ablating

compositecompositematerials. Journal of Heat and Mass Transfer, 1997(38):
p. 139-146.
64.
Mouritz, A. P., Gibson, A. G., Wu, Y. -S., Gardiner, C.P., and Mathys, Z.,
Validation ofthe Gibson modelfOr thefire reaction-properties offibrepolymer composites.Fire and Materials, In press.
KC Easby
PhD Thesis
65.
120
Chang, C.I., Thermal effectson polymer compositestructures. Theoretical &

Applied FractureMechanics, 1986(6): p. 113-120.
66.
Kung, H. C., A mathematicalmodel oftoodpyrolysis. Combustion and

Flame, 1972(18):p. 185-195.
67.
Kansa, E.J., Perlee,H.E., and Chaiken, R.F., Mathematical model ofwood

Combustion
Flame,
1977.
in1cuding
internalforced
and
convection.
pyrolysis
29: p. 311-324.
68.
Looyeh, M. R.E. and Bettess,P., A finite elementmodelfor thefire

performance of GRPpanels including variable thermal properties. Finite
Elementsin Analysis & Design, 1998(30):p. 313-324.
69.
Mouritz, A. P. and Mathys, Z., Mechanicalproperties offire-damaged glassreinforcedphenolic composites.Fire and Materials, 2000.34: p. 67-75.
70.
Mouritz, A. P. and Mathys, Z., Post-fire mechanicalproperties ofg1ass2001.

Composites
Science
Technology,
composites.
and
reinforcedpolyester
61(4): p. 475-490.
71.
Bausano,J., Boyd, S., Lesko, J., and Case,S. Compositelife under sustained
compressionand one sided simulatedfire exposure:characterisation and
in
Composites
England.
Fire
3.2003.
Newcastle
Tyne,
in
upon
prediction.
72.
Halverson, H., Bausano,J., Case,S., and Lesko, J. Simulation ofstructural

responseof compositestructures underfire exposure.in Compositesin Fire
3.2003. Newcastle upon Tyne, England.
73.
Mahieux, C.A. and Reifsnider, K. L., Property modelling across transition

temperaturesin polymers: a robust stiffness-temperaturemodel. Polymer,
2001.42(7): p. 3281-3291.
74.
Keller, T., Tracy, C., and Hugi, E., Fire enduranceofloaded and liquidcooled GFRP slabsfor construction. CompositesPart A, 2006.37(7): p.
1055-1067.
75.
Massot, J.J. Glass reinforcedplastic heavy loadflooringfor offshore

in
platforms. CompositeMaterials in the OffshoreIndustry. 1994.Institut
Frangaisdu Petrole, France.
76.
Greene,E., Fire performance ofcomposite materialsfor naval applications.

1993, Structural CompositesInc.: Melbourne, FL, USA.
&C Easby
PhD Thesis
77.
121
Browne, T.N. A., A Modelfor the Structural Integrity of CompositeLaminates

in Fire, in School ofMechanical & SystemsEngineering. PhD Thesis 2006,
University of Newcastle-upon-Tyne:Newcastle-upon-Tyne.
78.
Fibre-reinforcedplastic composites- Determination offlexural properties BS

ENISO 14125: 1998.2003.
79.
Gardiner, C.P., Mathys, Z., and Mouritz, A. P., Tensileand Compressive

Properties ofFRP Compositeswith Localised Fire Damage.Applied
CompositeMaterials, 2002.9(6): p. 353-367.
80.
Boeing Specification Support Standard,Advanced CompositeCompression

TestsBSS 7260.1986.
81.
Kulcami, A. P. and Gibson, k. F. Non-destructive Characterisation ofEffects

EMoisture
Ester
Resin
Vinyl
Temperature
Elastic
Moduli
and
and
of
on
of
in
18th
Annual
Composites,
Composite.
American
Society
Resin
of
glass
Technical Conference.2003. Florida, USA.
82.
Budiansky, B. and Fleck, N. A., CompressiveFailure ofFibre Composites.

Journal of Mech. Phys. Solids, 1993.41(l): p. 183-211.
83.
Agarwal, B.D. and Broutman, L. J., Analysis andperformance offiber

&
Sons,
1990,
Chichester:
Wiley
Inc.
2nd
John
ed.
composites.
84.
Hull, D., An introduction to compositematerials. I st ed. CambridgeSolid

State ScienceSeries. 1981,Cambridge: CambridgeUniversity Press.
85.
Young, W. C. and Budynas,R.G., Roark's Formulasfor Stressand Strain. 7th

ed. 2002, London: McGraw-Hill.
86.
Couchman,L. and Mouritz, A. P., eds.Modeling ofNaval Composite

Structures in Fire. 2006, CooperativeResearchCentre for Advacned
Composite Structureson behalf of the United StatesOffice of Naval
Research.
87.
Murphy, J., ReinforcedPlastics Handbook. 1994, Oxford: Elsevier Advanced

Technology.
88.
Wang, J.S. and Chan, W. S., Effects ofdefects on the buckling load ofrodpack
laminates. Journal of the American Helicopter Society, 2000: p. 216-221.
9C Easby
PhD Thesis
122
Publications
A. G. Gibson, T. N. A. Browne, R. C. Easby, Y-S Wu, Z. Mathys, A. P. Mouritz
Laminate Theory Analysis of Compositesunder Load in Fire
Proc. of The 4ffi International Conference On Composite Materials For Offshore
Operations. Houston TX, USA, October 2005.
T. N. A. Browne, R. C. Easby,A. Elmughrabi, A. G. Gibson, A. P. Mouritz

Composite StructuresUnder Load In Fire: Characterisationand Modeling
Proc. of DURACOSYS (Durability of Composite Systems).
Blacksburg VA, USA, September2006.
P, C Easby
PhD Thesis
Virginia Tech,
Appendix
Visual Basic Code
ABD Matrix calculation (Polyester)
SubABCDvsTimeo
RemCalculatematrix for all times
Forj =I To 39
Dim Time(50)
Worksheets("Tempprofile"). Activate
Timeo) = ActiveSheet.Cellso + 7,1). Value
Worksheets("ABD Matrix"). Activate
ActiveSheet.Cellso + 24,1). Value = Timeo)
RemCalculateQ Matrix for eachnode

For i=1 To 17
Rem Input temperatureprofiles
Dim tem(20)
tem(i) = ActiveSheet.Cellso + 7, i+ 1).Value
Rem Input rrc values
Dim rc(20)
Worksheets("RRC").Activate
Cellso
i+
ActiveSheet.
7,
1).Value
+
rc(i) =
Rem Calculate E1,E2,G12,u2l
Dim EI(20)
Dim E2(20)
Dim G12(20)
Dim ul2(20)
Dim u2l(20)
Dim cE 1(20)
Dim cE2(20)
Dim cG12(20)
RC Easby
PhD Thesis
123
Fire Behaviour ofPultmded Composites
Dim TanhE1(20)
Dim TanhE2(20)
Dim TanhG12(20)
DimTanhEIS(20)
Dim TanhE2S(20)
Dim TanhG12S(20)
Dim cEls(20)
Dim cE2S(20)
Dim cG12S(20)
Worksheets("Model").Activate
Rem UD properties
TgEl = ActiveSheet.Cells(4,2)
TgE2 = ActiveSheet.Cells(5,2)
TgG 12 = ActiveSheet.Cells(6,2)
kE I= ActiveSheet.Cells(4,3)
kE2 = ActiveSheet.Cells(5,3)
kG 12 = ActiveSheet.Cells(6,3)
MuE I= ActiveSheet.Cells(4,4)
MuE2 = ActiveSheet-Cells(5,4)
MuG12 = ActiveSheet.Cells(6,4)
MrEl = ActiveSheet.Cells(4,5)
MrE2 = ActiveSheet.Cells(5,5)
MrG 12 = ActiveSheet.Cells(6,5)
Rem CSM properties
TgE IS = ActiveSheet.Cells(I 1,2)
TgE2S= ActiveSheet.Cells(I 2,2)
TgG I 2S = ActiveSheet.Cells(I 3,2)
kE IS = ActiveSheet.Cells(i 1,3)
kE2S = ActiveSheet.Cells(12,3)
kG 12S= ActiveSheet.Cells(I 3,3)
MuE IS = ActiveSheet.Cells(I 1,4)
MuE2S = ActiveSheet.Cells(I 2,4)
MuG 12S= ActiveShect.Cells(I 3,4)
MrE IS = ActiveSheet.Cells(I 1,5)
MrE2S = ActiveSheet.Cells(12,5)
MrG I 2S = ActiveSheet.Cells(I 3,5)
If i >= 4 And i <= 14 Then
kE
(tem(i)
I
(i)
I*
TgE
1)
=
cE
RC Easby
PhD Thesis
124
125
TanhE I (i) = (Exp(cE 1(i)) - Exp(-cE I (i))) / (Exp(cE I (i)) + Exp(-cE I (i)))
EI (i) = (((I - TanhE I (i)) * (MuE I- MrE 1)) /2+ MrE 1) * (rc(i) / 100) AI
cE2(i) = kE! * (tem(i) - TgE2)
TanhE2(i) = (Exp(cE2(i)) - Exp(-cE2(i))) / (Exp(cE2(i)) + Exp(-cE2(i)))
E2(i) = (((I - TanhE2(i)) * (MuE2 - MrE2)) /2+ MrE2) * (rc(i) / 100)A I
cG 12(i) = kG 12 * (tem(i) - TgG 12)

TanhG12(i) = (Exp(cG12(i)) - Exp(-cG12(i))) / (Exp(cG12(i)) + Exp(-cG12(i)))
100)
AI
GI 2(i) = (((l - TanhGI 2(i)) * (MuG 12 - MrG 12)) /2+ MrG 12) * (rc(i) /
Else
I
S)
kE
(tem(i)
TgE
1S*
1
cE s(i) =
-
TanhE I S(i) = (Exp(cE I s(i)) - Exp(-cE I s(i))) / (Exp(cE I s(i)) + Exp(-cE I s(i)))
EI (i) = (((I - TanhE I S(i)) * (MuE 1S- MrE I S)) /2+ WE I S) * (rc(i) / 100) AI
kE2S
*
(tem(i)
TgE2S)
cE2S(i) =
TanhE2S(i) = (Exp(cE2S(i)) - Exp(-cE2S(i))) / (Exp(cE2S(i)) + Exp(-cE2S(i)))
E2(i) = (((I - TanhE2S(i)) * (MuE2S - MrE2S)) /2+ MrE2S) * (rc(i) / 100) A1
cGl2S(i) = kG12S * (tem(i) - TgG12S)
TanhGl2S(i) = (Exp(cGl2S(i)) - Exp(-cGl2S(i))) / (Exp(cGl2S(i)) + Exp(cGl2S(i)))

G12(i) = (((I - TanhGl2S(i)) * (MuGl2S - MrGl2S)) /2+ MrGl2S) * (rc(i) / 100)
I
End If
u12(i) = ActiveSheet.Cells(3,8). Value

/
*
EI(i)
E2(i)
u2l(i) = u12(i)
Q matrix
RemCalculate
Dim Q11(20)
Dim Q12(20)
Dim Q22(20)
Dim Q66(20)
*
1(i)
u2
u12(i)
uxy =IQll(i)=El(i)/uxy
Q12(i) = u12(i) * E2(i) / uxy
Q22(i) = E2(i) / uxy
Q13=0
Q21= Q12
Q31=0
Q23=0
Q66(i)= G12(i)
Next i
P, C Easby
PhD Thesis
Rem Calculating jEi

Dim thick(20)
Dim yi(20)
Dim Eyi(20)
Y=O
Ey=O
E=O
bd3 0
EIO 0
alpha =0
For i=1 To 17
thick(i) = ActiveSheet.Cells(i + 19,1). Value
Rem Printing
Worksheets("Sheet1").Activate
ActiveSheet.Cells(i + 2,2). Value = thick(i) + 0.5
yi(i) = thick(i) + 0.5
Worksheets("SheetI "). Activate
ActiveSheet.Cells(i + 2,4). Value =EI (i) * yi(i)
Eyi(i) =EI (i) * yi(i)
Ey = Ey + Eyi(i)
E=E+ EI(i)
Next i
Y=Y+(Ey/E)
ActiveSheet.Cellso + 24,2). Value =Y
Rem 2nd Loop
For i=1 To 17
thick(i) = ActiveSheet.Cells(i + 19,1).Value
Rem Printing
P, C Easby
PhD Thesis
126
127

Rem ActiveSheet.Cells(i + 2,3). Value =E1 (i)
bd3 = bd3 + (E I (i) / 12)
ActiveSheet. Cells(19,5). Value =Y
(E
(i)
*
(y
2)
I
A
+
yi(i))
alpha
alpha =
_
Next i
ElO = bd3 + alpha
ActiveSheet.Cellso + 24,3). Value = EIO
Rem CalculateABD Matrix
* (QII(2)+Qll(4)+QII(6)+Qll(8)+
All =(0.5/3) * QII(1)+QII(17+4
Qll(16 +2 * (Qll(3)+ QII(5) + QII(7)+
Qll(10)+Qll(12)+QII(14)+
QI l(9) + QI l(1 1)+ Ql l(13) + QI l(15)

A 12= (0.5/ 3) * Q 12(1)+Q 12(17+4* (Q12(2)+Q 12(4)+Q 12(6)+Q 12(8)+
Q12(10)+Q 12(12)+Q 12(14)+Q 12(16+2* (Q12(3)+Q 12(5)+Q 12(7)+
Q12(9)+Q 12(11)+Q 12(13)+Q 12(15)
A21 = (0.5/ 3) * Q 12(1)+Q 12(17+4* (Q12(2)+Q 12(4)+Q 12(6)+Q 12(8)+
Q12(10)+Q 12(12)+Q 12(14)+Q 12(16+2* (Q12(3)+Q 12(5)+Q 12(7)+
Q12(9)+Q 12(11)+Q 12(13)+Q 12(15)

A22 = (0.5/ 3) * Q22(I) + Q22(I7 +4* (Q22(2)+ Q22(4)+ Q22(6)+ Q22(8)+
Q22(10)+ Q22(I2) + Q22(14)+ Q22(I6 +2* (Q22(3)+ Q22(5)+ Q22(7)+
Q22(9)+ Q22(11) + Q22(I3) + Q22(15)
A66 = (0.5/ 3) * Q66(1)+ Q66(17+4* (Q66(2)+ Q66(4)+ Q66(6)+ Q66(8)+
Q66(I0) + Q66(12)+ Q66(I4) + Q66(I6 +2* (Q66(3)+ Q66(5)+ Q66(7)+
Q66(9)+ Q66(I 1) + Q66(I3) + Q66(15)
A16=0
A61 =0
A26 =0
A62 =0
Bl I= (0.5/ 3) (((Ql l(l) * -4) + (Ql 1(17)* 4)) +4* ((Ql 1(2)* -3.5)+ (Ql 1(4)
1.5)
(Q
11(8)
*
*
(Q
11(6)
(Q
II
(10)
0-5)
(Q
I
1(12)
*
1.5)
+
+
+
+
+
-0.5)
-2.5)
(QI 1(14)* 2.5)+ (QI l(l 6) * 3.5))+2* ((Q11(3)* -3) + (Q11(5)* -2) + (Q11(7)
0)
(Ql
(Ql
1(9)
1(11)
*
(Ql
1(13)
*
(Ql
1)
2)
1(15)
*
3)))
+
+
+
+
-1)
B 12= (0.5/ 3) (((Q12(l) * -4) + (Q12(17)* 4)) +4* ((Q12(2)* -3.5)+ (Q12(4)
*
*
(Q12(10)
*
(Q12(6)
(Q12(8)
0.5)
(Q12(12)
*
1.5)
+
+
+
+
+
-1.5)
-0.5)
-2.5)
(Q12(14)* 2.5)+ (Q12(16)* 3.5))+2* ((Q12(3)* -3) +Q 12(5)* -2) + (Q12(7)
(Q12(15)
(Q12(13)
*
2)
*
(Q12(1
*
(Q12(9)
0)
1)
1)
3)))
+
+
+
+
-1)
B21 = (0.5/ 3) (((Q12(l) * -4) + (Q12(17)* 4)) +4* ((Q12(2)* -3.5)+ (Q12(4)
*-1.5)
(Q
*
(Q
12(6)
*
1
2(l
0)
0.5)
(Q
12(8)
(Q
12(12)
*
1.5)
+
+
+
+
+
-0.5)
-2.5)
(Q12(14)* 2.5) + (Q12(16)* 3.5))+2* ((Q12(3)* -3),+ (Q12(5)* -2) + (Q12(7)
(Q
*
*
12(9)
0)
(Q
1)
*
1)
12(13)
2)
12(15)
12(l
*
1)
3)))
+
+
+Q
+Q
RC Easby
PhD Thesis
B22 = (0.5 / 3) (((Q22(l) * -4) + (Q22(17) * 4)) +4* ((Q22(2) * -3.5) + (Q22(4)
(Q22(6)
(Q22(8)
*
(Q22(10)
*
+
0.5)
+
(Q22(12)
*
+
1.5)
+
+
-2.5)
-1.5)
-0.5)
(Q22(14) * 2.5) + (Q22(16) * 3.5)) +2* ((Q22(3) *
(Q22(5)
*
(Q22(7)
+
+
-3)
-2)
(Q22(9)
*
0)
(Q22(l
*
1)
1)
(Q22(13)
*
+
+
2)
(Q22(15)
*
3)))
+
+
-1)
B66 = (0.5 / 3) (((Q66(l) * -4) + (Q66(17) * 4)) +4* ((Q66(2) *
(Q66(4)
+
-3.5)
(Q66(6)
(Q66(8)
*
(Q66(10)
*
0.5)
(Q66(12)
*
+
+
1.5)
+
+
+
-2.5)
-1.5)
-0.5)
(Q66(14) * 2.5) + (Q66(16) * 3.5)) +2* ((Q66(3) *
(Q66(5)
*
(Q66(7)
+
+
-3)
-2)
(Q66(9)
*
0)
(Q66(l
1)
*
1)
(Q66(13)
+
*
+
2)
(Q66(15)
*
3)))
+
+
-1)
B16 =0
B61 =0
B26 =0
B62 =0
DI I =(0.5/3)*(((Qll(l)*
16)+(QII(17)*
16))+4*((QII(2)*
12.25)+
(Q 11(4) * 6.25) +,(Q 11(6) * 2.25) + (Q 11(8) * 0.25) + (Q I1 (10) * 0.25) + (Q I 1(12)
* 2.25) + (Q I 1(14) * 6.25) + (Q I 1(16) * 12.25)) +2* ((Q 11(3) * 9) + (Q 11(5) * 4)
+ (Ql 1(7) 1) + (Ql 1(9) * 0) + (Ql l(l 1) * 1) + (Ql 1(13) * 4) + (Ql 1(15) * 9)))
D 12= (0.5 3) * (((Q12(l) * 16)+ (Q12(17)* 16))+4* ((Q12(2)* 12.25)+

(Q12(4)* 6..25) + (Q12(6)* 2.25)+ (Q12(8)* 0.25)+ (Q12(l0) * 0.25)+ (Q12(12)
* 2.25)+ (Q12(14)* 6.25)+ (Q12(16)* 12.25))+2* ((Q12(3)* 9) + (Q12(5)* 4)
+ (Q12(7) 1) + (Q12(9)* 0) + (Q12(l 1) * 1) + (Q12(13)* 4) + (Q12(15)* 9)))
D21 = (0.5 3) * (((Q12(l) * 16)+ (Q12(17)* 16))+4* ((Q12(2)* 12.25)+
(Q 12(4) * 6.25) + (Q 12(6) * 2.25) + (Q 12(8) * 0.25) + (Q 12(l 0) * 0.25) + (Q 12(12)

* 2.25) + (Q 12(14) * 6.25) +Q 12(16) * 12.25)) +2* ((Q 12(3) * 9) +Q 12(5) * 4)
+Q 12(7) 1) +Q 12(9) * 0) +Q1 2(l 1) * 1) +Q 12(13) * 4) +Q 12(15) * 9)))
D22 = (0.5 3) * (((Q22(l) * 16) + (Q22(17) * 16)) +4* ((Q22(2) * 12.25)+
(Q22(4) * 6.25) + (Q22(6) * 2.25) + (Q22(8) * 0.25) + (Q22(10) * 0.25) + (Q22(12)
* 2.25) + (Q22(14) * 6.25) + (Q22(16) * 12.25))+2* ((Q22(3) * 9) + (Q22(5) * 4)
+ (Q22(7) * 1) + (Q22(9) * 0) + (Q22(l 1) * 1) + (Q22(13) * 4) + (Q22(15) * 9)))
D66 = (0.5 / 3) * (((Q66(l) 16) + (Q66(17) * 16)) +4* ((Q66(2) * 12.25)+
(Q66(4) * 6.25) + (Q66(6) 2.25) + (Q66(8) * 0.25) + (Q66(10) * 0.25) + (Q66(12)
* 2.25) + (Q66(14) * 6.25) + (Q66(16) * 12.25)) +2* ((Q66(3) * 9) + (Q66(5) * 4)
+ (Q66(7) * 1) + (Q66(9) * 0) + (Q66(l 1) * 1) + (Q66(13) * 4) + (Q66(15) * 9)))
D16=0
D61 =0
D26 =0
D62 =0
ActiveSheet.Cells(43,2). Value = Al 1
ActiveSheet.Cells(44,2). Value =A 12
ActiveSheet.Cells(43,3). Value = A21
P, C Easby
PhD Thesis
ActiveSheet.Cells(43,6). Value =BII

ActiveSheet.Cells(44,6). Value =B 12
ActiveSheet.Cells(43,7). Value = B21
ActiveSheet.Cells(48,6). Value =DII
ActiveSheet.Cells(49,6). Value = D12
ActiveSheet.Cells(50,6). Value =D 16
Rem Inputing Inverted ABD matrix into ABD Matrix worksheet
Dim Aldash(20)
Dim A2dash(20)
Dim A6dash(20)
DimBldash(20)
Dim B2dash(20)
Dim B6dash(20)
Dim C1dash(20)
Dim C2dash(20)
Dim C6dash(20)
Dim DI dash(20)
Dim D2dash(20)
Dim D6dash(20)
For i= I To 3
RC Easby
PhD 7hesis
129
130
Aldash(i) = ActiveSheet.Cells(54, i+ 1).Value
A2dash(i) = ActiveSheet.Cells(55, i+ 1).Value
B1dash(i) = ActiveSheet.Cells(54, i+ 5).Value
Cldash(i) = ActiveSheet.Cells(59, i+ 1).Value
C2dash(i)= ActiveSheet.Cells(60, i+ 1).Value
C6dash(i)= ActiveSheet.Cells(61, i+ 1).Value
Dldash(i) = ActiveSheet.Cells(59, i+ 5).Value
D2dash(i) = ActiveSheet.Cells(60, i+ 5).Value
ActiveSheet.Cellso + 24, i+ 3).Value = Aldash(i)
ActiveSheet.Cellso + 24, i+ 6).Value = A2dash(i)
ActiveSheet.Cellso + 24, i+ 12).Value =BI dash(i)
Act iveSheet.Cellso + 24, i+ 15).Value = B2dash(i)
,
Cellso + 24, i+ 18).Value = B6dash(i)
ActiveSheet.
ActiveSheet.Cellso + 24, i+ 21).Value =CI dash(l)
ActiveSheet.Cellso + 24, i+ 24).Value = C2dash(i)
ActiveSheet.CellsO + 24, i+ 27).Value = C6dash(i)
ActiveSheet.CellsO+ 24, i+ 30).Value =DI dash(i)
ActiveSheet.Cellso + 24, i+ 33).Value D2dash(i)
ActiveSheet.Cellso + 24, i+ 36).Value = D6dash(i)
Next i
ActiveSheet.Cellso + 24,43). Value =DII
ActiveSheet.Cellso + 24,44). Value = D22
ActiveSheet.Cellso + 24,53). Value = Al I
ActiveSheet.Cellso + 24,54). Value =A 12
ActiveSheet.Cellso + 24,55). Value A21
ActiveSheet.Cellso + 24,59). Value A 16
RC Easby
PhD Thesis
ActiveSheet.Cellso + 24,60). Value = A61

ActiveSheet.Cellso + 24,6 1).Value = A66
ActiveSheet.Cellso + 24,62). Value =BII
ActiveSheet.Cellso + 24,63). Value =B 12
ActiveSheet.Cellso + 24,64). Value = B21
ActiveSheet.Cellso + 24,7 1).Value =D 12
Nextj
End Sub
P, C Easby
PhD Thesis
131
Tensile and Compressive strengths as functions of time (Polyester)

Sub StrengthsVsTimeo
Rem Carry out strengthcalculations for all time intervals
Forj = l'To 39
Dim Time(50)
Worksheets("TempProfile"). Activate
Worksheets("Strengthsvs Time"). Activate
ActiveSheet.Cells +30,4). Value = Timeo)
ActiveSheet.Cells + 77,4). Value = Timeo)
Rem Calculate EI per node
For i=I To 17
Rem input temperaturevalues
Dim tem(20)
Rem input RRC values
Dim rc(20)
i+
Cellso
ActiveSheet.
7,
1).Value
+
rc(i) =
Rem calculate El
Dim E1(20)
Dim cE 1(20)
Dim TanhE1(20)
Dim cE I s(20)
Dim TanhElS(20)
Rem UD paramters
TgE I= ActiveSheet.Cells(4,2)
kE 1= ActiveSheet.Cells(4,3)
MuE I ActiveSheet.Cells(4,4)
MrE I ActiveSheet.Cells(4,5)
Rem CSM NEedle weave paramters
TgE IS = ActiveSheet.Cells(I 1,2)
RC Easby
PhD Theiis
132
133
kE IS = ActiveSheet. Cells(I 1,3)

MuE IS ActiveSheet. Cells(I 1,4)
MrE IS ActiveSheet. Cells(I 1,5)

(tem(i)
1)
kE
TgE
(i)
I*
I
cE
=
-
TanhEl(i) = (Exp(cEI(i)) - Exp(-cEI(i))) / (Exp(cEl(i)) + Exp(-cEI(i)))

EI (i) = (((I - TanhE I (i)) * (MuE I- WE 1)) /2+ WE 1) * (rc(i) / 100) A1
Else:
cEI s(i) = kE IS* (tem(i) - TgE 1S)
TanhE I S(i) = (Exp(cE I s(i)) - Exp(-cE I s(i))) / (Exp(cE I s(i)) + Exp(-cE I s(i)))
EI (i) = (((I - TanhE I S(i)) * (MuE IS- WE I S)) /2+ WE I S) * (rc(i) /I OO)A I
End If
Rem Lamina node strength
Dim sT(20)
Dim sC(20)
Rem UD properties
TgsT = ActiveSheet.Cells(7,2)
TgsC = ActiveSheet.Cells(8,2)
ksT = ActiveSheet.Cells(7,3)
ksC = ActiveSheet.Cells(8,3)
MusT = ActiveSheet.Cells(7,4)
MusC = ActiveSheet.Cells(8,4)
MrsT = ActiveSheet.Cells(7,5)
MrsC = ActiveSheet.Cells(8,5)
Rem CSM properties
TgsTS = ActiveSheet.Cells(I 4,2)
TgsCS = ActiveSheet.Cells(I 5,2)
ksTS = ActiveSheet.Cells(I 4,3)
ksCS = ActiveSheet.Cells(I 5,3)
MusTS = ActiveSheet.Cells(l 4,4)
MusCS = ActiveSheet.Cells(I 5,4)
MrsTS = ActiveSheet.Cells(I 4,5)
MrsCS = ActiveSheet.Cells(15,5)
csT = ksT * (tem(i) - TgsT)
TanhsT = (Exp(csT) - Exp(-csT)) / (Exp(csT) + Exp(-cS*T))
sT(i) = (((I - TanhsT) * (MusT - MrsT)) /2+ MrsT) * (rc(i) /I
csC = ksC * (tem(i) - TgsQ
P, C Easby
PhD Thesis
OO)A
134
TanhsC= (Exp(csC) - Exp(-csC)) / (Exp(csC) + Exp(-csC))

/2+
/
*
(rc(i)
100)
^0
*
(MusC
MrsQ
MrsQ)
(((I
TanhsC)
sC(i) =
Else:
csTS = ksTS * (tem(i) - TgsTS)
TanhsTS = (Exp(csTS) - Exp(-csTS)) / (Exp(csTS) + Exp(-csTS))

sT(i) = (((l - TanhsTS) * (MusTS - MrsTS)) /2+ MrsTS) * (rc(i) /1
OO)A
csCS= ksCS * (tern(i) - TgsCS)
TanhsCS = (Exp(csCS) - Exp(-csCS)) / (Exp(csCS) + Exp(-csCS))

*
(MusCS
/2+
(rc(i)
*
(((I
TanhsCS)
MrsCS))
MrsCS)
sC(i) =
-
100) AI
End If
Next i
Rem Stressat a point (tensile and compressiveloading)

Dim
Dim
Dim
Dim
TStress(20)
TStren(80)
CStress(20)
CStren(80)
For k=0 To 70
For i=I To 17
k/
1000
strain=
TStress(i) = strain * EI(i)
CStress(i)=strain* EI (i)
1000
1000
If TStress(i) >-- sT(i) Then TStress(i)

if CStress(i) >= sC(i) Then CStress(i)
Next i
0
0
TStren(k) = (0.5 /3* ((TStress(l) + TStress(17))+4* (TStress(2)+ TStress(4)+

TStress(6)+ TStress(8)+ TStress(l 0) + TStress(l2) + TStress(l4) + TStress(l 6)) +2
* (TStress(3)+ TStress(5)+ TStress(7)+ TStress(9)+ TStress(l 1) + TStress(13)+
TStress(15))))/8
CStren(k) = (0.5 /3* ((CStress(l) + CStress(17))+4* (CStress(2)+ CStress(4)+
CStress(6)+ CStress(8)+ CStress(10)+ CStress(12)+ CStress(14)+ CStress(16))+
2* (CStress(3)+ CStress(5)+ CStress(7)+ CStress(9)+ CStress(l 1) + CStress(13)
+ CStress(15))))/8
ActiveSheet.Cellso + 30, k+ 5).Value = TStren(k)
ActiveSheet.Cellso + 77, k+ 5).Value = CStren(k)
Next k
Nextj
P, C Easby
PhD Thesis
End Sub
P, C Easby
PhD Thesis
135
Stress strain curve modelling (Polyester)

SubStressStrainCurveso
Forj =I To 39
Dim Time(50)
Dim tem(20)
Dim TStress(20)
Dim sT(20)
Dim TStren(200)
Workshects("Tempprofile"). Activate
Worksheets("Output").Activate
Fork= 0 To 70
For i=1 To 17
tem(i) = ActiveSheet.CellsO+ 7, i+ 1).Value
If 0< tern(i) <= 33.5 Then E= 12

If 0< tern(i) <= 33.5 Then sig = 1250
'Alter valueshere
'Alter valueshere
If 33.5 < tem(i) <= 70 Then E= 10.2

If 33.5 < tem(i) <= 70 Then sig = 1200
'Alter values here

'Alter values here
If 70 < tem(i) <= 105 Then E= 12.7

If 70 < tem(i) <= 105 Then sig = 1100
'Alter values here

'Alter valueshere
If 105 < tem(i) <-- 135 Then E= 11.5

If 105 < tem(i) <= 135 Then sig = 1250
'Alter values here

'Alter values here
If 135 < tem(i) <-- 180 Then E= 11.35

If 135 < tem(i) <= 180 Then sig = 750
'Alter values here

'Alter values here
If 180 < tem(i) <= 230 Then E= 13

If 180 < tem(i) <= 230 Then sig = 575
'Alter values here

'Alter values here
it C Easby
PhD Thesis
136
If 230 < tem(i) <= 275 Then E= 12.8

If 230 < tem(i) <= 275 Then sig = 615
'Alter values here

'Alter values here
If 275 < tem(i) <= 300 Then E= 12.5

If 275 < tem(i) <= 300 Then sig = 575
'Alter values here

'Alter values here
137
If 3 00 < tern(i) Then E= 1000000

If 300 < tem(i) Then sig = 1000000
strain= k/ 1000
TStress(i) = sig * (I - Exp((-(1000 * E) * strain) / sig))
TgsT = 195 'Alter values here
ksT = 0.03 'Alter valueshere
MusT 320 'Alter valueshere
MrsT 220 'Alter values here
csT = ksT * (tem(i) - TgsT)
TanhsT = (Exp(csT) - Exp(-csT)) / (Exp(csT) + Exp(-csT))

(MusT
/2+
*
MrsT))
MrsT)
(((l
TanhsT)
sT(i) =
-
If TStress(i) >= sT(i) Then TStress(i) =0

Next i
'Tmpezium rule
TStren(k) = ((0.5 / 2) * ((TStress(l) + TStress(17))+2* (TStress(2)+ TStress(3)+

TStress(4)+ TStress(5)+ TStress(6)+ TStress(7)+ TStress(8)+ TStress(9)+
TStress(10)+ TStress(l 1) + TStress(12)+ TStress(13)+ TStress(14)+ TStress(15)+
TStress(16))))/8
Next k
Nextj
End Sub
RC Easby
PhD Thesis
ABD Matrix calculation (Phenolic)
Sub ABCDvsTimeo
Rem Calculate matrix for all times

Forj =I To 39
Dim Time(50)
Worksheets("Temp profile"). Activate
Timea) = ActiveSheet. Cellso + 7,1). Value

RemCalculateQ Matrix for eachnode

For i=I To 17
Rem Input temperatureprofiles
Dim tem(20)
Rem Input rrc values
Dim rc(20)
rc(i) = ActiveSheet.Cellso + 7, i+ 1).Value
Rem CalculateE1,E2,G12,u2l
Dim E1(20)
Dim E2(20)
Dim G12(20)
Dim ul2(20)
Dim u2l(20)
Dim cE 1(20)
Dim cE2(20)
Dim cG12(20)
Dim TanhE1(20)
Dim TanhE2(20)
Dim TanhG12(20)
DimTanhEIS(20)
Dim TanhE2S(20)
Dim TanhGl2S(20)
6
RC Easby
PhD Thesis
138
139
Dim cE I s(20)
Dim cE2S(20)
Dim cGl2S(20)
Rem UD properties
TgE I= ActiveSheet.Cells(4,2)
TgE2 = ActiveSheet.Cells(5,2)
TgG 12 = ActiveSheet.Cells(6,2)
kE I= ActiveSheet.Cells(4,3)
kE2 = ActiveSheet.Cells(5,3)
kG12 = ActiveSheet.Cells(6,3)
MuE I= ActivcShcet.Cells(4,4)
MuE2 = ActiveSheet.Cells(5,4)
MuG12 = ActiveSheet.Cells(6,4)
MrE 1= ActiveSheet.Cells(4,5)
MrE2 = ActiveSheet.Cells(5,5)
MrG 12 = ActiveSheet.Cells(6,5)
Rem CSM properties
TgEl S= ActiveSheet.Cells(I 1,2)
TgE2S = ActiveSheet.Cells(12,2)
TgG 12S = ActiveSheet.Cells(I 3,2)
kE 1S= ActiveSheet.Cells(i 1,3)
kE2S = ActiveSheet.Cells(I 2,3)
kG12S = ActiveSheet.Cells(13,3)
MuE IS = ActiveSheet. Cells(I 1,4)
MuE2S = ActiveSheet. Cells(l 2,4)
MuG 12S= ActiveSheet. Cells(I 3,4)
MrE IS = ActiveSheet.Cells(I 1,5)

MrE2S = ActiveSheet.Cells(12,5)
MrG 12S= ActiveSheet.Cells(l 3,5)
if i >= 4 And i <-- 14 Then
kE
(tcm(i)
1*
TgE
1)
(i)
1
cE
=
-
TanhE 1(i) = (Exp(cE I (i)) - Exp(-cE I (i))) / (Exp(cE I (i)) + Exp(-cE I (i)))
E1 (i) = (((I - TanhE 1(i)) * (MuE 1- MrE 1)) /2+ MrE 1) * (rc(i) / 100) A1
*
(tem(i)
kE2
TgE2)
cE2(i) =
TanhE2(i) = (Exp(cE2(i)) - Exp(-cE2(i))) / (Exp(cE2(i)) + Exp(-cE2(i)))
100)A
E2(i) = (((I - TanhE2(i)) * (MuE2 - MrE2)) /2+ MrE2) * (rc(i) /
P- C Easby
PhD Thesis
140
kG12
*
(tem(i)
TgG12)
cG12(i) =
TanhG12(i) = (Exp(cG12(i)) - Exp(-cG12(i))) / (Exp(cG12(i)) + Exp(-cG12(i)))
G12(i) = (((I - TanhG12(i)) * (MuG12 - MrG12)) /2+ MrG12) * (rc(i) / 100) AI
Else
I
S)
(tem(i)
TgE
kE
IS*
I
s(i)
cE
=
-
TanhE 1S(i) = (Exp(cE I s(i)) - Exp(-cE I s(i))) / (Exp(cE I s(i)) + Exp(-cE I s(i)))
EI (i) = (((I - TanhE I S(i)) * (MuE IS- MrE 1S)) /2+ WE 1S) * (rc(i) / 100) A
*
(tem(i)
kE2S
TgE2S)
cE2S(i) =
TanhE2S(i) = (Exp(cE2S(i)) - Exp(-cE2S(i))) / (Exp(cE2S(i)) + Exp(-cE2S(i)))
E2(i) = (((I - TanhE2S(i)) * (MuE2S - MrE2S)) /2+ MrE2S) * (rc(i) / 100)A I
*
(tem(i)
TgG12S)
kG12S
cGl2S(i) =
TanhGl2S(i) = (Exp(cGl2S(i)) - Exp(-cGl2S(i))) / (Exp(cGl2S(i)) + Exp(cGl2S(i)))
G 12(i) = (((I - TanhG12S(i)) * (MuG 12S- MrG I 2S)) /2+ Milb I 2S) * (rc(i) /I OO)A
End If
Cells(3,8).
Value
ActiveSheet.
u12(i) =
/
*
EI(i)
E2(i)
u2l(i) = u12(i)
Q matrix
RemCalculate
Dim Q11(20)
Dim Q12(20)
Dim Q22(20)
Dim Q66(20)
*
u2l(i)
u12(i)
uxy =1Qll(i)=El(i)/uxy
Q12(i) =ul2(i) * E2(i)/uxy
Q22(i) = E2(i) / uxy
Q13=0
Q21= Q12
Q31=0
Q23=0
Q66(i)= G12(i)
Next i
Rem Calculating Ei
Dim thick(20)
Dim yi(20)
P, C Easby
PhD Theiis
Dim Eyi(20)
Y=O
Ey=O
E=O
bd3 0
EIO 0
alpha =0
For i=1 To 17
thick(i) = ActiveSheet.Cells(i + 19,1). Value
Rem Printing
Worksheets("Sheetl"). Activate
Worksheets("SheetI ").Activate
ActiveSheet.Cells(i + 2,4). Value =EI (i) * yi(i)
Eyi(i) =EI (i) * yi(i)
Ey = Ey + Eyi(i)
E=E+ EI(i)
Next i
Y=Y+(Ey/E)
ActiveShect.Cellso + 24,2). Value =Y
Rem 2nd Loop
For i= I To 17
Worksheets("Model"). Activate
thick(i) = ActiveSheet. Cells(i + 19,1). Value
Rem Printing
Worksheets("Sheet1"). Activate
Worksheets("Sheetl"). Activate
Rem ActiveSheet. Cells(i + 2,3). Value
RC Easby
PhD Thesis
EI (i)
141
142
bd3 = bd3 + (E I (i) / 12)

ActiveSheet. Cells(19,5). Value =Y
alpha = alpha + (E I (i) * (y _ yi(i)) A 2)
Next i
1310= bd3 + alpha
ActiveSheet.Cellso + 24,3). Value = EIO
Rem CalculateABD Matrix
Al I= (0.5/ 3) * ((Ql l(l) + Ql 1(17))+4* (Ql 1(2)+ Ql 1(4)+ Ql 1(6)+ Ql 1(8)+

QII(10)+QII(12)+ QII(14)+ Qll(16))+2 * (Qll(3) +QII(5)+QII(7) +
Ql 1(9)+ Ql l(l 1)+ Ql 1(13)+ Ql 1(15)))
Al 2= (0.5/ 3) * ((Q12(l) +Q 12(17))+4* (Q12(2)+Q 12(4)+Q 12(6)+Q 12(8)+

Q12(l0) +Q 12(12)+Q 12(14)+Q 12(16))+2* (Q12(3)+Q 12(5)+Q 12(7)+
Q12(9)+Q 12(l 1) +Q 12(13)+Q 12(15)))
A21 = (0.5/ 3) * ((Q12(l) +Q 12(17))+4* (Q12(2)+Q 12(4)+Q 12(6)+Q 12(8)
Q12(l 0) +Q 12(12)+Q 12(14)+Q 12(16))+2* (Q12(3)+Q 12(5)+Q 12(7)+
Q12(9)+Q1 2(l 1) +Q 12(13)+Q 12(15)))
A22 = (0.5/ 3) * ((Q22(l) + Q22(17))+4* (Q22(2)+ Q22(4)+ Q22(6)+ Q22(8)+
Q22(10)+ Q22(12)+ Q22(14)+ Q22(16))+2* (Q22(3)+ Q22(5)+ Q22(7)+
Q22(9)+ Q22(l 1) + Q22(13)+ Q22(15)))
A66 = (0.5/ 3) * ((Q66(l) + Q66(17))+4* (Q66(2)+ Q66(4)+ Q66(6)+ Q66(8)+
Q66(10)+ Q66(12)+ Q66(14)+ Q66(16))+2* (Q66(3)+ Q66(5)+ Q66(7)+
Q66(9)+ Q66(11) + Q66(13)+ Q66(15)))
A16=0
A61 =0
A26 =0
A62 =0
BI 1= (0.5/ 3) (((Ql l(l) * -4) + (Ql 1(17)* 4)) +4* ((Ql 1(2)* -3.5)+ (Ql 1(4)
(Q
*
11(8)
(Q
(10)
*
1.5)
(Q
11(6)
11
0.5)
(Q
*
I
1(12)
1.5)
+
+
+
+
+
-0.5)
-2.5)
(Q1l(l 4) * 2.5)+ (QI l(I 6) * 3.5))+2* ((Q11(3)* -3) + (Q11(5)* -2) + (Q11(7)
(Q
(11)
*
0)
11
1)
(Q
*
(Q
11(13)
2)
(Q
I
1(15)
*
11(9)
3)))
1)
+
+
+
+
B 12 = (0.5 / 3) (((Q 12(l) * -4) + (Q 12(17) * 4)) +4 ((Q 12(2) * -3.5) + (Q 12(4)
* -1.5) + (Q12(8) * -0.5) + (Q12(10) 0.5) + (Q12(12) * 1.5) +
(Q12(6)
+
-2.5)
(Q12(14)* 2.5) + (Q12(16)* 3.5))+2* ((Q12(3)* -3) + (Q12(5)* -2) + (Q12(7)

(Q
*
(Q
12(13)
*
(Q
*
0)
12(l
1)
1)
2)
12(15)
(Q
*
12(9)
3)))
1)
+
+
+
+
B21 = (0.5/ 3) (((Q12(l) * -4) + (Q12(17)* 4)) +4* ((Q12(2)* -3.5)+ (Q12(4)
(Q12(10)
(Q12(8)
*
*
0.5)
(Q12(6)
(Q12(12)
*
1.5)
+
+
+
+
+
-1.5)
-0.5)
-2.5)
(Q12(14)* 2.5)+ (Q12(16)* 3.5))+2* ((Q12(3)* -3) + (Q12(5)* -2) + (Q12(7)
(Q
*
(Q
*
1)
12(13)
2)
12(15)
*
0)
12(l
1)
*
(Q
12(9)
3)))
+Q
1)
+
+
+
B22 = (0.5/ 3) * (((Q22(l) * -4) + (Q22(17)* 4)) +4* ((Q22(2)* -3.5)+ (Q22(4)
(Q22(10)
*
0.5)
(Q22(8)
*
(Q22(12)
*
*
(Q22(6)
1.5)
+
+
+
+
+
-1.5)
-0.5)
-2.5)
(Q22(14)* 2.5)+ (Q22(16)* 3.5))+2* ((Q22(3)* -3) + (Q22(5)* -2) + (Q22(7)
*
*
(Q22(13)
2)
(Q22(15)
0)
(Q22(l
1)
1)
(Q22(9)
*
3)))
+
+
+
+
-1)
B66 = (0.5/ 3) (((Q66(l) * -4) + (Q66(17)* 4)) +4 ((Q66(2)* -3.5)+ (Q66(4)
*
(Q66(6)
(Q66(8)
*
(Q66(10)
0.5)
(Q66(12)
+
*
+
+
1.5)
+
+
-1.5)
-0.5)
-2.5)
P, C Easby
PhD Thesis
(Q66(14)* 2.5)+ (Q66(16)* 3.5))+2* ((Q66(3)*

(Q66(5)
*
(Q66(7)
+
+
-3)
-2)
(Q66(9)
*
0)
(Q66(l
1)
*
1)
(Q66(13)
+
*
+
2)
(Q66(15)
* 3)))
+
+
-1)
B16=0
B61=0
B26=0
B62 =0
DI I =(0.5/3)*(((Qll(l)*
16)+(Qll(17)*
16))+4*((Qll(2)*
12.25)+
(Q 11(4) * 6.25) + (Q 11(6) * 2.25) + (Q 11(8) * 0.25) + (Q 11(10) 0.25) + (Q I 1(12)

* 2.25) + (Q I 1(14) * 6.25) + (Q I l(l 6) * 12.25)) +2* ((Q 11(3) 9) + (Q 11(5) * 4)
+ (Ql 1(7) 1) + (Ql 1(9) * 0) + (Ql 1(11) * 1) + (Ql 1(13) * 4) + (Ql 1(15) * 9)))
D 12 = (0.5 3) * (((Q 12(l) * 16) + (Q 12(17) * 16)) +4* ((Q 12(2) * 12.25)+
(Q 12(4) * 6.25) + (Q 12(6) * 2.25) + (Q 12(8) * 0.25) + (Q 12(l 0) * 0.25) + (Q 12(12)
* 2.25) + (Q 12(14) * 6.25) + (Q 12(16) * 12.25)) +2* ((Q 12(3) * 9) + (Q 12(5) * 4)
+ (Q12(7) 1) + (Q12(9) * 0) + (Q12(l 1) * 1) + (Q12(13) * 4) + (Q12(15) * 9)))
D21 = (0.5 3) * (((Q 12(l) * 16) + (Q 12(17) * 16)) +4* ((Q 12(2) * 12.25)+
(Q 12(4) * 6.25) + (Q 12(6) * 2.25) + (Q 12(8) * 0.25) + (Q 12(l 0) * 0.25) + (Q 12(12)
* 2.25) + (Q 12(14) * 6.25) + (Q 12(16) * 12.25)) +2* ((Q 12(3) * 9) + (Q 12(5) * 4)
+ (Q 12(7) 1) + (Q 12(9) * 0) + (Q 12(l 1) * 1) + (Q 12(13) * 4) + (Q 12(15) * 9)))
D22 = (0.5 3) * (((Q22(l) * 16) + (Q22(17) * 16)) +4* ((Q22(2) * 12.25)+
(Q22(4) * 6.25) + (Q22(6) * 2.25) + (Q22(8) * 0.25) + (Q22(10) * 0.25) + (Q22(12)
* 2.25) + (Q22(14) * 6.25) + (Q22(16) * 12.25)) +2* ((Q22(3) * 9) + (Q22(5) * 4)
+ (Q22(7) 1) + (Q22(9) * 0) + (Q22(l 1) * 1) + (Q22(13) * 4) + (Q22(15) * 9)))
D66 = (0.5 3) * (((Q66(l) * 16) + (Q66(17) * 16)) +4* ((Q66(2) * 12.25)+
(Q66(4) * 6.25) + (Q66(6) * 2.25) + (Q66(8) * 0.25) + (Q66(10) * 0.25) + (Q66(12)
* 2.25) + (Q66(14) * 6.25) + (Q66(16) * 12.25)) +2* ((Q66(3) * 9) + (Q66(5) * 4)
+ (Q66(7) * 1) + (Q66(9) 0) + (Q66(l 1) * 1) + (Q66(13) * 4) + (Q66(15) * 9)))
D16=0
D61 =0
D26 =0
D62 =0
ActiveSheet.Cells(43,2). Value =AII
ActiveSheet-Cells(45,2).Value =A 16
ActiveSheet-Cells(45,4).Value = A66
ActiveSheet.Cells(45,7). Valuc = B26
RC Easby
PhD Thais

Rem Inputing Inverted ABD matrix into ABD Matrix worksheet
DimAldash(20)
Dim A2dash(20)
Dim A6dash(20)
Dim BI dash(20)
Dim B2dash(20)
Dim B6dash(20)
Dim CI dash(20)
Dim C2dash(20)
Dim C6dash(20)
Dim DI dash(20)
Dim D2dash(20)
Dim D6dash(20)
For i= I To 3
AI dash(i) = ActiveSheet.Cells(54, i+ 1).Value
it C Easby
PhD Thesis
144
BI dash(i) = ActiveSheet.Cells(54, i+ 5).Value

Cldash(i) = ActiveSheet. Cells(59, i+ 1).Value
C2dash(i) = ActiveSheet. Cells(60, i+ 1).Value
C6dash(i) = ActiveSheet. Cells(61, i+ 1).Value
Dldash(i) = ActiveSheet.Cells(59, i+ 5).Value

ActiveSheet.Cellso + 24, i+ 3).Value =AI dash(i)
ActiveSheet.Cellso + 24, i+ 6).Value = A2dash(i),
ActiveSheet.Cellso + 24, i+ 12).Value =BI dash(i)
ActiveSheet.Cellso + 24, i+ 15).Value = B2dash(i)
ActiveSheet.Cellso + 24, i+ 18).Value = B6dash(i)
ActiveSheet.Cellso + 24, i+2 1).Value =CI dash(i)
ActiveSheet.Cellso + 24, i+3 0).Value =DI dash(i)
Next i
ActiveSheet.Cellso + 24,43). Value =DII
ActiveSheet.Cellso + 24,53). Value =AII
ActiveSheet-Cellso+ 24,59). Yalue = A16
ActiveSheet-Cellso+ 24,6 1).Value = A66
ActiveSheet.Cellso + 24,62). Value =BII
RC Easby
PhD Thesis
145

ActiveSheet.Cellso + 24,7 1).Value =D 12
ActiveSheet.Cellso + 24,75). Value =D 16
Next
End Sub
XC Easby
PhD Thesis
146
Tensile and Compressive strengths as functions of time (Phenolic)

Sub StrengtlisVsTimeo
Rem Carry out strengthcalculations for all time intervals
Forj =1 To 39
Dim Time(50)
Rem CalculateEI per node
For i=I To 17
Rem input temperaturevalues
Dim tem(20)
Rem input RRC values
Dim rc(20)
i+
Cellso
7,
1).
ActiveSheet.
Value
+
rc(i) =
Rem calculateEI
Dim E1(20)
Dim cE 1(20)
Dim TanhEI(20)
Dim cEls(20)
DimTanhEIS(20)
Activate
Worksheets("Model").
RemUD paramters
Cells(4,2)
TgEI= ActiveSheet.
Cells(4,3)
kE1= ActiveSheet.
Cells(4,4)
MuEI ActiveSheet.
Cells(4,5)
MrE 1 ActiveSheet.
RemCSMNEedleweaveparamters
Cells(I 1,2)
TgEIS = ActiveSheet.
it C Easby
PhD Theiis
147
kE IS = ActiveSheet.Cells(I 1,3)
MuE IS ActivcSheet.Cells(l 1,4)
WE IS ActiveSheet.Cells(I 1,5)
If i >= 4 And i <-- 14 Then
cE I (i) = kE I* (tem(i) - TgE 1)
TanhE I (i) = (Exp(cE I (i)) - Exp(-cE I (i))) / (Exp(cE I (i)) + Exp(-cE I (i)))
EI (i) = (((I - TanhE I (i)) * (MuE I- MrE 1)) /2+ MrE 1) * (rc(i) / 100) /%I
Else:
kE
(tem(i)
TgE
I
S)
IS*
1
s(i)
cE
=
-
TanhElS(i) = (Exp(cEls(i)) - Exp(-cEls(i))) / (Exp(cEls(i)) + Exp(-cEls(i)))

EI (i) = (((I - TanhE I S(i)) * (MuE IS- WE I S)) /2+ WE I S) * (rc(i) /I OO)A
End If
Rem Lamina node strength
Dim sT(20)
Dim sC(20)
Rem UD properties
TgsT = ActiveSheet.Cells(7,2)
TgsC = ActiveSheet.Cells(8,2)
ksT = ActiveSheet.Cells(7,3)
ksC = ActiveSheet.Cells(8,3)
MusT = ActiveSheet.Cells(7,4)
MusC = ActiveSheet.Cells(8,4)
MrsT = ActiveSheet.Cells(7,5)
MrsC = ActiveSheet.Cells(8,5)
Rem CSM properties
TgsTS = ActiveSheet.Cells(I 4,2)
TgsCS = ActiveSheet.Cells(I 5,2)
ksTS = ActiveSheet.Cells(l 4,3)
ksCS = ActiveSheet.Cells(I 5,3)
MusTS = ActiveSheet.Cells(I 4,4)
MusCS = ActiveSheet.Cells(I 5,4)
MrsTS = ActiveSheet.Cells(I 4,5)
MrsCS = ActiveSheet.Cells(I 5,5)
*
(tem(i)
ksT
TgsT)
csT =
-

sT(i) = (((l - TanhsT) * (MusT - MrsT)) /2+ MrsT) * (rc(i) / 100) A0
(tem(i)
ksC
*
TgsQ
=
csC
it C Easby
PhD Thesis
148
149
TanhsC= (Exp(csC) - Exp(-csC)) / (Exp(csC) + Exp(-csC))

/
/*
0
/2+
*
(rc(i)
100)
*
(MusC
MrsQ
MrsQ
(((I
TanhsC)
sC(i) =
Else:
TgsTS)
*
(tem(i)
ksTS
csTS =
-
TanhsTS = (Exp(csTS) - Exp(-csTS)) / (Exp(csTS) + Exp(-csTS))

/
(rc(i)
100)
AI
/2+
*
*
(MusTS
MrsTS))
MrsTS)
(((I
TanhsTS)
sT(i) =
-
(tern(i)
TgsCS)
*
ksCS
csCS=
TanhsCS= (Exp(csCS)- Exp(-csCS)) / (Exp(csCS) + Exp(-csCS))
/
/2+
*
(rc(i)
*
(MusCS
MrsCS))
MrsCS)
TanhsCS)
(((I
sC(i) =
End If
Next i
100)
A1
Rem Stress at a point (tensile and compressive loading)
Dim TStress(20)
Dim TStren(80)
Dim CStress(20)
Dim CStren(80)
For k0 To 70
To 17
For iI
k/
1000
strain=
TStress(i)= strain * El(i) * 1000
CStress(i)=strain* EI (i) * 1000
If TStress(i) >-- sT(i) Then TStress(i)
If CStress(i) >-- sC(i) Then CStress(i)
Next i
0
0
TStren(k) = (0.5 /3* TStress(1) + TStress(17 +4* (TStress(2)+ TStress(4)+

TStress(6)+ TStress(8)+ TStress(10) + TStress(12) + TStress(14) + TStress(16 +2
* (TStress(3)+ TStress(5)+ TStress(7)+ TStress(9)+ TStress(11) + TStress(13)+
TStress(15 /8
CStren(k) = (0.5 /3* CStress(1)+ CStress(17+4* (CStress(2)+ CStress(4)+
CStress(6)+ CStress(8)+ CStress(10) + CStress(12) + CStress(14) + CStress(16 +
2* (CStress(3)+ CStress(5)+ CStress(7)+ CStress(9)+ CStress(11) + CStress(13)
+ CStress(15 /8
ActiveSheet.Cellso + 30, k+ 5).Value TStren(k)
ActiveSheet.Cellso + 77, k+ 5).Value CStren(k)
Next k
Nextj
XC Easby
PhD Thesis
End Sub
RC Easby
PhD Thesis
150
. Fire Behaviour Pultruded Composites

of
Stress strain curve modelling (Phenolic)

Sub StressStrainCurveso
Forj =1 To 39
Dim
Dim
Dim
Dim
Dim
Time(50)
tem(20)
TStress(20)
sT(20)
TStren(200)
Activate
Worksheets("Output").
Cellso+ 7,1).Value= Timeo)
ActiveSheet.
For k0
To 70
For iI
To 17
If 0< tem(i) <-- 48.5 Then E= 13.25

If 0< tem(i) <= 48.5 Then sig = 700
'Alter values here

'Alter values here
If 48.5 < tem(i) <= 90 Then E= 12.9

If 48.5 < tem(i) <= 90 Then sig = 800
'Alter values here

'Alter values here
if 90 < tem(i) <-- 140 Then E= 12.1

if 90 < tem(i) <-- 140 Then sig = 790
'Alter values here

'Alter values here
If 140 < tem(i) <= 215 Then E= 11.3

If 140 < tem(i) <= 215 Then sig = 780
'Alter values here

'Alter values here
If 215 < tem(i) <= 275 Then E= 11.5

If 215 < tem(i) <= 275 Then sig = 800
'Alter values here

'Alter values here
If 275 < tem(i) <= 300 Then E= 10.8

If 275 < tem(i) <= 300 Then sig = 1150
'Alter values here

'Alter values here
P, C Easby
PhD Thesis
151
152
If 300 < tern(i) Then E= 1000000

If 300 < tern(i) Then sig = 1000000
k/
1000
strain=
TStress(i) = sig * (I - Exp((-(1000 * E) * strain) / sig))
TgsT=100 'Alter values here
ksT = 0.035 'Alter valueshere
MusT 400 'Alter valueshere 500
MrsT 278 'Alter valueshere 347
(tem(i)
TgsT)
*
ksT
csT =
/2+
*
(MusT
MrsT))
(((I
TanhsT)
MrsT)
sT(i) =
If TStress(i) >= sT(i) Then TStress(i) =0
Next i
'Trapezium rule
TStren(k) = ((0.5 / 2) * ((TStress(l) + TStress(17))+2* (TStress(2)+ TStress(3)+

TStress(4)+ TStress(5)+ TStress(6)+ TStress(7)+ TStress(8)+ TStress(9)+
TStress(10)- TStress(l 1) + TStress(12)+ TStress(13)+ TStress(14)+ TStress(l 5) +
TStress(16))))/8
Next k
Nextj
End Sub
9C Easby
PhD Thesis
153
Thermal Model Outputs

Table A. Typical temperatureevolution output file for an 8mm polyester pultrusion. The table from
left to right reads;Time is seconds,temperature(Tfur) in *C, followed by the temperaturesin OCat
nodes, 1,3,5,40,47,49 and 5 1.
time
00
3.11
6.23
9.34
12.46
15.57
18.69
21.8
24.92
28.03
31.15
34.26
37.37
40.49
43.6
46.72
49.83
52.95
56.06
59.18
62.29
65.41
68.52
71.63
74.75
77.86
80.98
84.09
87.21
90.32
93.44
96.55
99.67
102.78
105.89
109.01
112.12
115.24
118.35
121.47
124.58
127.7
P, C Easby
PhD Thesis
Tfur
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
ND1
ND2
135
17
209.8
265.6
303.4
332.4
355.9
375.5
392.2
406.5
418.5
428.4
436.4
442.8
447.9
452
455.3
458.1
460.6
462.8
464.8
466.6
468.3
470
471.7
473.3
474.9
476.5
478.1
479.7
481.4
483.1
484.8
486.6
488.5
490.4
492.3
494.4
496.5
498.6
500.8
503
505.2
17
105
160.6
200.2
231.2
256.8
278.5
297.3
313.6
327.7
339.6
349.6
357.8
364.6
370.1
374.8
378.8
382.2
385.2
388
390.4
392.8
394.9
397
398.9
400.9
402.7
404.5
406.3
408.1
409.9
411.8
413.6
415.4
417.3
419.2
421.1
423.1
425.1
427.2
429.3
431.4
ND3
17
46.7
86.7
120.5
149.1
173.8
195.5
214.7
231.9
247.2
260.8
272.8
283.2
292.3
300.1
307
313
318.3
323.1
327.4
331.4
335
338.4
341.5
344.5
347.3
350
352.6
355.1
357.5
359.9
362.2
364.4
366.6
368.7
370.9
372.9
375
377.1
379.1
381.1
383.1
ND4
ND5
ND6
ND7
40
47
49
51
17
17
17
17.1
17.4
18.3
19.7
21.8
24.6
27.9
31.7
36
40.7
45.8
51.1
56.6
62.4
68.2
74.2
80.2
86.2
92.2
98.1
104.1
109.9
115.7
121.4
127
132.4
137.8
143.1
148.2
153.3
158.2
163
167.7
172.2
176.7
181
185.3
189.4
193.4
17
17
17
17
17.1
17.3
17.9
18.9
20.4
22.4
25
28
31.5
35.4
39.7
44.4
49.3
54.4
59.7
65.2
70.8
76.4
82.1
87.7
93.3
98.9
104.5
109.9
115.3
120.6
125.8
130.9
135.8
140.7
145.4
150.1
154.6
159
163.2
167.4
171.5
175.4
17
17
17
17
17
17.2
17.5
18.2
19.2
20.8
22.9
25.4
28.5
31.9
35.8
40
44.6
49.4
54.4
59.6
64.9
70.2
75.7
81.1
86.5
91.9
97.3
102.5
107.8
112.9
117.9
122.8
127.6
132.3
136.9
141.3
145.7
149.9
154
158
161.9
165.7
17
17
17
17
17
17.1
17.4
18
19
20.4
22.4
24.8
27.7
31
34.7
38.8
43.1
47.8
52.6
57.5
62.6
67.8
73
78.2
83.4
88.6
93.7
98.8
103.7
108.6
113.4
118.1
122.6
127.1
131.4
135.6
139.7
143.7
147.6
151.4
155
158.6
130.81
133.93
137.04
140.15
143.27
146.38
149.5
152.61
155.73
158.84
161.96
165.07
168.19
171.3
174.41
177.53
180.64
183.76
186.87
189.99
193.1
196.22
199.33
202.45
205.56
208.67
211.79
214.9
218.02
221.13
224.25
227.36
230.48
233.59
236.71
239.82
242.93
246.05
249.16
252.28
255.39
258.51
261.62
264.74
267.85
270.97
274.08
277.19
280.31
283.42
286.54
289.65
292.77
RC Easby
PhD Thesis
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
507.5
509.7
511.8
513.9
516
517.9
519.8
521.7
523.5
525.3
527.1
528.9
530.7
532.5
534.4
536.2
538.1
539.9
541.7
543.5
545.2
546.9
548.5
550.1
551.7
553.2
554.7
556.2
557.7
559.2
560.7
562.1
563.5
564.9
566.2
567.5
568.7
569.8
570.9
572
573.1
574.1
575.1
576
577
578
578.9
579.9
580.8
581.8
582.7
583.6
584.6
433.5
435.7
437.8
439.9
442
444.1
446.2
448.3
450.4
452.5
454.7
456.9
459.1
461.4
463.6
465.9
468.2
470.5
472.7
474.9
477
479.1
481.1
483.1
485
486.9
488.8
490.7
492.6
494.4
496.3
498.1
499.8
501.6
503.2
504.8
506.3
507.8
509.2
510.5
511.8
513.1
514.3
515.6
516.8
518
519.2
520.5
521.7
522.9
524
525.2
526.4
385
386.9
388.8
390.6
392.4
394.2
395.9
397.6
399.3
401
402.7
404.3
406
407.7
409.3
411
412.7
414.4
416.1
417.8
419.5
421.2
422.9
424.5
426.2
427.9
429.6
431.3
433
434.7
436.5
438.2
439.9
441.6
443.2
444.9
446.5
448.1
449.6
451.1
452.6
454.1
455.6
457
458.5
460
461.4
462.9
464.3
465.7
467.2
468.6
470
197.3
201.1
204.8
208.4
211.9
215.3
218.6
221.8
224.9
228
230.9
233.8
236.6
239.3
241.9
244.5
247
249.4
251.8
254
256.2
258.4
260.4
262.4
264.3
266.2
268
269.8
271.5
273.1
274.7
276.2
277.7
279.1
280.5
281.8
283.1
284.3
285.5
286.7
287.8
288.9
289.9
290.9
291.9
292.8
293.7
294.6
295.4
296.3
297.1
297.8
298.6
179.2
183
186.6
190.1
193.6
196.9
200.1
203.3
206.3
209.3
212.1
214.9
217.7
220.4
223
225.5
228
230.3
232.6
234.8
237
239.1
241.1
243
244.9
246.7
248.4
250.1
251.8
253.4
254.9
256.4
257.8
259.2
260.5
261.8
263
264.2
265.4
266.5
267.6
268.6
269.6
270.6
271.5
272.4
273.3
274.1
274.9
275.7
276.5
277.2
277.9
154
169.4
172.9
176.4
179.8
183
186.2
189.3
192.3
195.2
198
200.7
203.4
206.1
208.7
211.2
213.6
215.9
218.2
220.3
222.4
224.4
226.4
228.3
230.1
231.9
233.6
235.3
236.8
238.4
239.9
241.3
242.7
244.1
245.3
246.6
247.8
249
250.1
251.2
252.2
253.2
254.2
255.1
256
256.9
257.8
258.6
259.4
260.1
260.8
261.6
262.2
262.9
162
165.3
168.6
171.7
174.8
177.7
180.6
183.3
186
188.6
191.1
193.8
196.4
198.8
201.1
203.3
205.4
207.5
209.5
211.4
213.3
215.1
216.8
218.5
220.1
221.7
223.2
224.6
226.1
227.4
228.7
230
231.2
232.4
233.5
234.6
235.7
236.7
237.7
238.6
239.6
240.4
241.3
242.1
242.9
243.7
244.4
245.1
245.8
246.5
247.1
247.7
248.3
295.88
299
302.11
305.23
308.34
311.45
314.57
317.68
320.8
323.91
327.03
330.14
333.26
336.37
339.49
342.6
345.71
348.83
351.94
355.06
358.17
361.29
364.4
367.52
370.63
373.75
376.86
379.97
383.09
386.2
389.32
392.43
395.55
398.66
401.78
404.89
408.01
411.12
414.23
417.35
420.46
423.58
426.69
429.81
432.92
436.04
439.15
442.27
445.38
448.49
451.61
454.72
457.84
P, C Easby
PhD Thesis
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
585.5
586.4
587.2
588.1
589
589.8
590.7
591.5
592.3
593.1
593.9
594.6
595.4
596.2
596.9
597.6
598.3
599.1
599.8
600.4
601.1
601.8
602.5
603.1
603.8
604.4
605
605.7
606.3
606.9
607.5
608.1
608.7
609.2
609.8
610.4
610.9
611.5
612
612.5
613.1
613.6
614.1
614.6
615.1
615.6
616.1
616.6
617.1
617.6
618
618.5
618.9
527.5
528.7
529.8
530.9
532
533.1
534.1
535.2
536.2
537.2
538.2
539.2
540.2
541.2
542.1
543
544
544.9
545.8
546.7
547.5
548.4
549.2
550.1
550.9
551.7
552.6
553.4
554.1
554.9
555.7
556.5
557.2
558
558.7
559.4
560.1
560.8
561.5
562.2
562.9
563.6
564.3
564.9
565.6
566.2
566.9
567.5
568.1
568.7
569.3
569.9
570.5
471.3
472.7
474.1
475.4
476.7
478
479.3
480.5
481.8
483
484.2
485.4
486.5
487.7
488.8
490
491.1
492.2
493.2
494.3
495.4
496.4
497.4
498.5
499.5
500.4
501.4
502.4
503.3
504.3
505.2
506.1
507
507.9
508.8
509.7
510.6
511.4
512.3
513.1
514
514.8
515.6
516.4
517.2
518
518.7
519.5
520.3
521
521.7
522.5
523.2
299.3
300
300.7
301.4
302
302.6
303.3
303.9
304.4
305
305.6
306.1
306.7
307.2
307.7
308.2
308.7
309.2
309.7
310.2
310.6
311.1
311.6
312
312.5
312.9
313.3
313.8
314.2
314.6
315
315.4
315.8
316.2
316.6
317
317.4
317.8
318.2
318.6
319
319.4
319.7
320.1
320.5
320.9
321.2
321.6
322
322.4
322.7
323.1
323.4
278.6
279.3
279.9
280.6
281.2
281.8
282.4
282.9
283.5
284
284.5
285
285.5
286
286.5
287
287.4
287.9
288.3
288.8
289.2
289.6
290
290.5
290.9
291.3
291.6
292
292.4
292.8
293.2
293.5
293.9
294.3
294.6
295
295.3
295.7
296
296.4
296.7
297.1
297.4
297.7
298.1
298.4
298.7
299.1
299.4
299.7
300
300.4
300.7
155
263.5
264.2
264.8
265.4
265.9
266.5
267
267.5
268
268.5
269
269.5
269.9
270.4
270.8
271.3
271.7
272.1
272.5
272.9
273.3
273.7
274.1
274.5
274.8
275.2
275.5
275.9
276.2
276.6
276.9
277.3
277.6
277.9
278.3
278.6
278.9
279.2
279.5
279.8
280.2
280.5
280.8
281.1
281.4
281.7
282
282.3
282.6
282.9
283.1
283.4
283.7
248.9
249.5
250
250.6
251.1
251.6
252
252.5
253
253.4
.
253.8
254.3
254.7
255.1
255.5
255.9
256.3
256.6
257
257.4
257.7
258.1
258.4
258.7
259.1
259.4
259.7
260
260.3
260.6
260.9
261.2
261.5
261.8
262.1
262.4
262.7
263
263.3
263.6
263.8
264.1
264.4
264.7
264.9
265.2
265.5
265.7
266
266.3
266.5
266.8
267
460.95
464.07
467.18
470.3
473.41
476.53
479.64
482.75
485.87
488.98
492.1
495.21
498.33
501.44
504.56
507.67
510.79
513.9
517.01
520.13
523.24
526.36
529.47
532.59
535.7
538.82
541.93
545.05
548.16
551.27
554.39
557.5
560.62
563.73
566.85
569.96
573.08
576.19
579.31
582.42
585.53
588.65
591.76
594.88
597.99
601.11
604.22
607.34
610.45
613.57
616.68
619.79
622.91
R. C Easby
PhD Thesis
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
619.4
619.8
620.3
620.7
621.2
621.6
622
622.4
622.9
623.3
623.7
624.1
624.5
624.9
625.3
625.6
626
626.4
626.8
627.1
627.5
627.9
628.2
628.6
628.9
629.3
629.6
630
630.3
630.6
631
631.3
631.6
632
632.3
632.6
632.9
633.2
633.5
633.8
634.1
634.4
634.7
635
635.3
635.6
635.9
636.2
636.4
636.7
637
637.3
637.5
571.1
571.7
572.3
572.9
573.4
574
574.5
575.1
575.6
576.2
576.7
577.2
577.7
578.3
578.8
579.3
579.8
580.3
580.7
581.2
581.7
582.2
582.7
583.1
583.6
584
584.5
584.9
585.4
585.8
586.2
586.7
587.1
587.5
587.9
588.4
588.8
589.2
589.6
590
590.4
590.8
591.2
591.5
591.9
592.3
592.7
593
593.4
593.8
594.1
594.5
594.9
523.9
524.6
525.3
526
526.7
527.4
528.1
528.7
529.4
530
530.7
531.3
532
532.6
533.2
533.8
534.4
535
535.6
536.2
536.8
537.4
538
538.5
539.1
539.6
540.2
540.7
541.3
541.8
542.3
542.9
543.4
543.9
544.4
544.9
545.4
545.9
546.4
546.9
547.4
547.9
548.4
548.8
549.3
549.8
550.2
550.7
551.1
551.6
552
552.5
552.9
323.8
324.2
324.5
324.9
325.2
325.6
325.9
326.3
326.6
327
327.3
327.7
328
328.3
328.7
329
329.4
329.7
330
330.4
330.7
331
331.4
331.7
332
332.4
332.7
333
333.3
333.7
334
334.3
334.6
335
335.3
335.6
335.9
336.2
336.6
336.9
337.2
337.5
337.8
338.1
338.5
338.8
339.1
339.4
339.7
340
340.3
340.6
340.9
301
301.3
301.6
301.9
302.2
302.6
302.9
303.2
303.5
303.8
304.1
304.4
304.7
305
305.3
305.6
305.9
306.2
306.5
306.8
307.1
307.4
307.6
307.9
308.2
308.5
308.8
309.1
309.4
309.6
309.9
310.2
310.5
310.8
311.1
311.3
311.6
311.9
312.2
312.4
312.7
313
313.3
313.5
313.8
314.1
314.3
314.6
314.9
315.2
315.4
315.7
315.9
156
284
284.3
284.6
284.9
285.1
285.4
285.7
286
286.3
286.5
286.8
287.1
287.3
287.6
287.9
288.2
288.4
288.7
289
289.2
289.5
289.7
290
290.3
290.5
290.8
291
291.3
291.5
291.8
292.1
292.3
292.6
292.8
293.1
293.3
293.6
293.8
294.1
294.3
294.6
294.8
295
295.3
295.5
295.8
296
296.3
296.5
296.7
297
297.2
297.4
267.3
267.5
267.8
268
268.3
268.5
268.8
269
269.3
269.5
269.8
270
270.2
270.5
270.7
271
271.2
271.4
271.7
271.9
272.1
272.4
272.6
272.8
273.1
273.3
273.5
273.7
274
274.2
274.4
274.6
274.9
275.1
275.3
275.5
275.8
276
276.2
276.4
276.6
276.9
277.1
277.3
277.5
277.7
277.9
278.2
278.4
278.6
278.8
279
279.2
626.02
629.14
632.25
635.37
638.48
641.6
644.71
647.83
650.94
654.05
657.17
660.28
663.4
666.51
669.63
672.74
675.86
678.97
682.09
685.2
688.31
691.43
694.54
697.66
700.77
703.89
707
710.12
713.23
716.35
719.46
722.57
725.69
728.8
731.92
735.03
738.15
741.26
744.38
747.49
750.61
753.72
756.83
759.95
763.06
766.18
769.29
772.41
775.52
778.64
&C Easby
PhD Thesis
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
708.5
637.8
638.1
638.3
638.6
638.8
639.1
639.3
639.6
639.8
640.1
640.3
640.6
640.8
641.1
641.3
641.5
641.8
642
642.2
642.5
642.7
642.9
643.1
643.3
643.6
643.8
644
644.2
644.4
644.6
644.8
645
645.2
645.4
645.6
645.8
646
646.2
646.4
646.6
646.8
647
647.2
647.4
647.6
647.8
647.9
648.1
648.3
648.5
595.2
595.6
595.9
596.2
596.6
596.9
597.3
597.6
597.9
598.2
598.6
598.9
599.2
599.5
599.8
600.1
600.4
600.8
601.1
601.4
601.7
602
602.2
602.5
602.8
603.1
603.4
603.7
604
604.2
604.5
604.8
605
605.3
605.6
605.9
606.1
606.4
606.6
606.9
607.1
607.4
607.6
607.9
608.1
608.4
608.6
608.9
609.1
609.3
553.3
553.8
554.2
554.6
555
555.4
555.8
556.2
556.7
557.1
557.4
557.8
558.2
558.6
559
559.4
559.8
560.1
560.5
560.9
561.2
561.6
562
562.3
562.7
563
563.4
563.7
564.1
564.4
564.8
565.1
565.4
565.8
566.1
566.4
566.7
567.1
567.4
567.7
568
568.3
568.6
568.9
569.2
569.5
569.8
570.1
570.4
570.7
341.2
341.5
341.8
342.2
342.5
342.8
343.1
343.4
343.7
344
344.3
344.6
344.9
345.2
345.5
345.8
346
346.3
346.6
346.9
347.2
347.5
347.8
348.1
348.4
348.7
349
349.2
349.5
349.8
350.1
350.4
350.7
351
351.2
351.5
351.8
352.1
352.4
352.6
352.9
353.2
353.5
353.8
354
354.3
354.6
354.9
355.1
355.4
316.2
316.5
316.7
317
317.3
317.5
317.8
318
318.3
318.6
318.8
319.1
319.3
319.6
319.8
320.1
320.3
320.6
320.8
321.1
321.3
321.6
321.8
322.1
322.3
322.6
322.8
323.1
323.3
323.5
323.8
324
324.3
324.5
324.7
325
325.2
325.5
325.7
325.9
326.2
326.4
326.6
326.9
327.1
327.3
327.6
327.8
328
328.3
157
297.7
297.9
298.1
298.4
298.6
298.8
299.1
299.3
299.5
299.8
300
300.2
300.5
300.7
300.9
301.1
301.4
301.6
301.8
302
302.3
302.5
302.7
302.9
303.1
303.4
303.6
303.8
304
304.2
304.4
304.6
304.9
305.1
305.3
305.5
305.7
305.9
306.1
306.4
306.6
306.8
307
307.2
307.4
307.6
307.8
308
308.2
308.4
279.4
279.6
279.8
280
280.2
280.4
280.6
280.9
281.1
281.3
281.5
281.7
281.9
282.1
282.3
282.5
282.7
282.9
283.1
283.3
283.4
283.6
283.8
284
284.2
284.4
284.6
284.8
285
285.2
285.4
285.6
285.7
285.9
286.1
286.3
286.5
286.7
286.9
287.1
287.2
287.4
287.6
287.8
288
288.2
288.3
288.5
288.7
288.9
158
Table B. Typical residual resin content output file for an 8mm polyester pultrusion. The table from
left to right reads;Time is seconds, followed by the residual resin content in % at nodes 1,3,5,40,
47,49 and 5 1.
(S)
RRC at ND= 1 ND= 3 ND= 5 ND=40 ND=47 ND=49 ND=51

100
100
0
100
100
100
100
100
100
100
3.1
100
100
100
100
100
100
100
6.2
100
100
100
100
100
100
100
100
9.3
100
100
100
100
99.999
100
12.5
100
100
100
100
100
99.992
100
15.6
100
100
100
100
100
18.7
99.966 99.999
100
100
100
100
100
99.89 99.993
21.8
100
100
100
100
100
99.709 99.979
24.9
100
100
100
100
100
99.347 99.943
100
28
100
100
100
100
98.712 99.871
100
31.1
100
100
100
100
97.727 99.743
100
100
34.3
100
100
100
96.344 99.538 99.996
37.4
100
100
100
100
94.557 99.243 99.992
40.5
100
100
100
100
92.386 98.844 99.983
43.6
100
100
100
100
89.871 98.338 99.971
100
46.7
100
100
100
87.056 97.72 99.952
49.8
100
100
100
100
83.986 96.991 99.926
100
52.9
100
100
100
80.703 96.151 99.892
56.1
100
100
100
100
95.2 99.847
77.245
59.2
100
100
100
100
73.647 94.141 99.792
100
62.3
100
100
loo
69.941 92.974 99.723
65.4
100
100
100
100
66.153 91.701 99.64
100
100
68.5
100
100
62.311 90.322 99.541
71.6
100
100
100
100
58.438 88.839 99.425
74.7
100
100
100
loo
54.557 87.251 99.291
77.9
100
100
100
100
81
50.69 85.558 99.135
100
100
loo
loo
46.857 83.762 98.958
84.1
100
loo
loo
100
43.077 81.861 98.756
87.2
100
100
100
100
39.371 79.855 98.528
100
90.3
100
100
100
35.758 77.743 98.272
93.4
100
100
100
100
32.256 75.527 97.985
96.6
100
100
100
loo
28.884 73.205 97.666
99.7
100
100
100
loo
25.661 70.777 97.311
100
102.8
100
loo
loo
22.606 68.244 96.917
100
105.9
100
100
100
100
19.735 65.608 96.483
100
109
100
100
17.064 62.869 96.004
100
112.1
100
100
100
14.611 60.031 95.478
100
115.2
100
loo
loo
12.386 57.099 94.902
118.4
100
100
loo
100
10.4 54.079 94.271
121.5
100
100
100
100
8.659 50.981 93.582
124.6
100
100
100
100
R. C Easby
PhD Thesis
127.7
130.8
133.9
137
140.2
143.3
146.4
149.5
152.6
155.7
158.8
162
165.1
168.2
171.3
174.4
177.5
180.6
183.8
186.9
190
193.1
196.2
199.3
202.4
205.6
208.7
211.8
214.9
218
221.1
224.2
227.4
230.5
233.6
236.7
239.8
242.9
246
249.2
252.3
255.4
258.5
261.6
264.7
267.9
271
274.1
277.2
RC Easby
PhD Thesis
7.163 47.817
5.908 44.602
4.882 41.357
4.067 38.104
3.44 34.869
2.972 31.681
2.636 28.568
2.402 25.559
2.246 22.681
2.145 19.956
2.082 17.406
2.045 15.049
2.023 12.896
2.012 10.957
2.006 9.239
2.002 7.742
2.001 6.463
2 5.393
2 4.519
2 3.824
2 3.285
2 2.881
2 2.586
2 2.379
2 2.237
2 2.143
2 2.084
2 2.047
2 2.026
2 2.013
2 2.007
2 2.003
2 2.001
2 2.001
22
22
22
22
22
22
22
22
229.899
228.638
227.523
226.548
225.704
224.983
224.373
92.833
92.019
91.137
90.185
89.16
88.06
86.883
85.628
84.294
82.879
81.383
79.804
78.141
76.392
74.557
72.634
70.623
68.526
66.343
64.077
61.732
59.314
56.829
54.284
51.689
49.051
46.38
43.687
40.983
38.281
35.592
32.933
30.317
27.763
25.286
22.903
20.63
18.48
16.466
14.597
12.878
11.312
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
99.996
99.993
99.99
99.987
99.984
99.98
99.977
99.974
99.971
99.967
99.964
99.961
99.955
99.949
99.942
99.936
99.93
99.923
99.917
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
99.998
99.996
99.994
99.993
99.991
99.989
99.987
99.985
99.983
99.982
99.98
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
loo
100
loo
100
100
100
100
100
100
100
loo
loo
loo
159
280.3
283.4
286.5
289.7
292.8
295.9
299
302.1
305.2
308.3
311.5
314.6
317.7
320.8
323.9
327
330.1
333.3
336.4
339.5
342.6
345.7
348.8
351.9
355.1
358.2
361.3
364.4
367.5
370.6
373.7
376.9
380
383.1
386.2
389.3
392.4
395.5
398.7
401.8
404.9
408
411.1
414.2
417.3
420.5
423.6
426.7
429.8
P, C Easby
PhD Thesis
223.865
223.446
223.106
222.834
222.62
222.454
222.327
222.232
222.161
222.11
222.074
222.049
222.032
222.02
222.013
222.008
222.005
222.003
222.002
222.001
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
99.91
99.904
99.894
99.884
99.875
99.865
99.855
99.846
99.836
99.825
99.812
99.799
99.786
99.774
99.761
99.748
99.734
99.718
99.701
99.685
99.669
99.653
99.637
99.618
99.599
99.58
99.561
99.541
99.522
99.501
99.478
99.456
99.433
99.411
99.388
99.363
99.337
99.312
99.286
99.26
99.233
99.204
99.175
99.146
99.117
99.087
99.055
99.023
98.99
99.978
99.976
99.974
99.973
99.971
99.969
99.967
99.963
99.959
99.956
99.952
99.949
99.945
99.941
99.938
99.934
99.931
99.927
99.923
99.92
99.916
99.911
99.905
99.9
99.895
99.889
99.884
99.878
99.873
99.867
99.862
99.857
99.851
99-845
99.838
99.831
99.824
99.817
99-809
99.802
99.795
99.788
99.78
99.773
99.766
99.757
99-748
99.739
99.73
100
100
100
100
100
99.998
99.996
99.994
99.993
99.991
99.989
99.987
99.985
99.984
99.982
99.98
99.978
99.976
99.975
99.973
99.971
99.969
99.967
99.966
99.964
99.962
99.96
99.958
99.956
99.955
99.953
99.951
99.949
99.947
99.946
99.944
99.942
99.94
99.938
99.937
99.935
99.933
99.931
99.927
99.923
99.92
99-916
99.913
99.909
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
99.999
99.997
99.996
99.994
99.992
99.99
99.988
160
432.9
436
439.2
442.3
445.4
448.5
451.6
454.7
457.8
461
464.1
467.2
470.3
473.4
476.5
479.6
482.8
485.9
489
492.1
495.2
498.3
501.4
504.6
507.7
510.8
513.9
517
520.1
523.2
526.4
529.5
532.6
535.7
538.8
541.9
545
548.2
551.3
554.4
557.5
560.6
563.7
566.8
570
573.1
576.2
579.3
582.4
P, C Easby
PhD Thesis
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
.222
222
222
222
222
222
222
222
222
222
222
98.958 99.721
98.924 99.712
98.888 99.703
98.853 99.694
98.818 99.685
98.781.99.676
98.743 99.665
98.704 99.654
98.666 99.643
98.627 99.632
98.585 99.621
98.543 99.611
98.501
99.6
98.459 99.589
98.414 99.577
98.369 99.565
98.324 99.552
98.277 99.54
98.229 99.527
98.181 99.514
98.133 99.502
98.082 99.489
98.031 99.475
97.979 99.461
97.926 99.447
97.872 99.432
97.817 99.418
97.761 99.403
97.704 99.389
97.646 99.374
97.588 99.358
97.527 99.341
97.466 99.325
97.404 99.309
97.34 99.292
97.276 99.276
97.211 99.259
97.143 99.24
97.076 99.222
97.007 99.204
96.937 99.186
96.866 99.168
96.793 99.149
96.719 99.129
96.645 99.109
96.568 99.089
96.491 99.069
96.412 99.05
96.332 99.028
99.905
99.902
99.898
99.894
69.891
99.887
99.884
99.88
99.876
99.873
99.869
99.866
99.862
99.858
99.855
99.851
99.847
99.844
99.84
99.837
99.833
99.827
99.822
99.816
99.811
99.806
99.8
99.795
99.789
99.784
99.778
99.773
99.768
99.762
99.757
99.751
99.746
99.741
99.735
99.73
99.724
99.717
99.71
99.703
99.696
99.689
99.681
99.674
99.667
99.986
99.985
99.983
99.981
99.979
99.977
99.976
99.974
99.972
99.97
99.968
99.967
99.965
99.963
99.961
99.959
99.958
99.956
99.954
99.952
99.95
99.949
99.947
99.945
99.943
99.941
99.939
99.938
99.936
99.934
99.932
99.93
99.929
99.927
99.925
99.923
99.921
99.92
99.918
99.916
99.914
99.912
99.911
99.909
99.907
99.905
99.903
99.902
99.9
161
Fire Behaviour ofPu Itruded Composites
585.5
588.6
591.8
594.9
598
601.1
604.2
607.3
610.5
613.6
616.7
619.8
622.9
626
629.1
632.3
635.4
638.5
641.6
644.7
647.8
650.9
654.1
657.2
660.3
663.4
666.5
669.6
672.7
675.9
679
682.1
685.2
688.3
691.4
694.5
697.7
700.8
703.9
707
710.1
713.2
716.3
719.5
722.6
725.7
728.8
731.9
735
R. C Easby
PhD Thesis
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
222
96.251 99.006 99.66 99.898

96.168 98.984 99.652 99.896
96.084 98.963 99.645 99.894
95.998 98.941 99.638 99.893
95.912 98.918 99.631 99.891
95.824 98.895 99.623 99.889
95.734 98.871 99.616 99.887
95.643 98.848 99.609 99.885
95.55 98.824 99.601 99.883
95.457 98.799 99.592 99.882
95.361 98.774 99.583 99.88
95.265 98.749 99.574 99.878
95.166 98.723 99.565 99.876
95.067 98.698 99.556 99.872
94.965 98.671 99.547 99.868
94.863 98.644 99.538 99.865
94.758 98.617 99.529 99.861
94.653 98.59 99.52 99.858
94.545 98.562 99.511 99.854
94.436 98.533 99.502 99.85
94.325 98.504 99.493 99.847
94.213 98.475 99.484 99.843
94.099 98.446 99.474 99.839
93.983 98.415 99.464 99.836
93.866 98.385 99.453 99.832
93.747 98.354 99.442 99.829
93.626 98.323 99.431 99.825
93.504 98.291 99.42 99.821
93.379 98.258 99.409 99.818
93.254 98.225 99.399 99.814
93.125 98.193 99.388 99.811
92.996 98.159 99.377 99.807
92.865 98.124 99.366 99.803
92.732 98.09 99.355
99.8
92.596 98.055 99.344 99.796
92.46 98-019 99.332 99.792
92.321 97.983 99.319 99.789
92.18 97.947 99.307 99.785
92.038 97.91 99.294 99.782
91.893 97.872 99.281 99.778
91.747 97.834 99.269 99.774
91.599 97.796 99.256 99.771
91.448 97.756 99.243' 99.767
91.296 97.716 99.231 99.764
91.142 97.677 99.218 99.76
90.986 97.636 99.205 99.756
90.828 97.594 99.191 99.753
90.667 97.553 99.177 99.749
90.505 97.511 99.162 99.745
162
738.1
741.3
744.4
747.5
750.6
753.7
756.8
759.9
763.1
766.2
769.3
772.4
775.5
778.6
R. C Easby
PhD Thesis
222
222
222
222
222
222
222
222
222
222
222
222
222
222
90.341
90.174
90.006
89.835
89.662
89.487
89.31
89.131
88.95
88.766
88.58
88.392
88.202
88.01
97.467
97.424
97.381
97.336
97.291
97.245
97.199
97.152
97.105
97.057
97.008
96.959
96.909
96.858
99.148
99.133
99.119
99.105
99.09
99.076
99.061
99.045
99.029
99.013
98.997
98.98
98.964
98.948
99.742
99.738
99.735
99.731
99.725
99.72
99.715
99.709
99.704
99.698
99.693
99.687
99.682
99.677
163
Fire Behaviour ofPultruded Composiles
164
Engineering Drawings
04
r_
.0i
-6
E
E
6
z
8
d
-0
42
C
C)
Ln
00'ZO
R. C Easby
PhD Thesis
165
.& a!
04
00'M
00'M
CD
ICE
R. C Easby
PhD Thesis
'Fire Behaviour ofPultruded Composites
a101
to
166
-T,
-_-
06
LE
00,09
Ln
00's
I
W
IIu
RC Easby
PhD Thesis
ca -- I<
167
CN
I's
jr
cq
OVOLL
Cl
(a
IUq
,01<-
RC Easby
PhD Thesis
Fire Behaviour ofPuhruded Composites
168
4
7
E
2
.
ET
tp
m
Alk
iE
8
43i
ISL
00,09
C)
C4
09,9
to
Ln
0(
00,00t
P, C Easby
PhD Thesis
00's L
169
co
tp
..
.% c
11
cn
---------------------Ct
kn
00'ZM
R C. Easby
PhD Thesis
'25
icl 11
170
09,9
CN
CD C
111
M
00'ZOL
de
le
os'u
OTE
nldHi CG'60
co
R. C. Easby
PhD Thesis
92
.cm
171
uj
CI4
00's'?
L
----------
8
9
cy
P)
4-
00,09
00'5
--
---------
--------- I
F-
C;
C3
IZ
In
8
cr
8
9
00,91F
Uu1
R. C Easby
PhD Thesis
00'ZS
ca
I<
172
cl
04
00'9 L
00's
C2
q
In
t
Ln
00'SZ
co, u
cc
R. C Easby
PhD Thesis
00 zs
173
19.
OOSV
L
CD
C:!
RC Easby
PhD 7besis
c6
174
00'Ot
OF
OC,
E
8
00
00'OG&
cs
00'k
tn
00,09
R C. Easby
PhD Thesis
175
-7 7
m
'A
1
00*00L
CL
2
1
01
.
00,00Z
04
C'4
14
Ul
M
C14
.
ID
00,001
x
!
M
00,009
00'0(;9
00,00Z
00,091F
&C Easby
PhD Thesis
176
TlCLJ
id
I
E2
Em
m
C4
1V
------------
LO
00,00Z
wl
to
P, C Easby
PhD Thesis
177
CL
00, L9
Rl
in
tn
blP
00'M
to
Li
XC Easby
PhD Thesis
41
1<
Fire Behaviour ofPultruded Composites,
(D
11
11
178
CF
C3
In
to
RC Easby
PhD Thesis
179
cli
cq
Ct
00'"
a-I
x
40 7E
kr
RC Easby
PhD Viesis
tD
cl
C4
- -0
to
180
.0
XC Easby
PhD Thesis
r4
181
00
CL
6
E<
E 3t
to m
00,09
00'9
L
F'=-,
en
it
I 'D At
8- -I - a
i-
I
, - -
,
8
00,00C
wi
09,661P
COVE
L
DO'09
-- - -----
00'M
-----69 ,CCE
C
0)
9L'LCZ
Lq
00
0
In
00'
00,00V
09'9%,
r
ca
RC Easby
PhD Thesis
to
"
182
00, Z
L
00's
go -
00,00C
oo-oos
C.
C
00 ,6zl
td
oo. 6zL
oo,
sz
00.0,;
ui
olr,6cc
00,00S
oo'Gzv
P, C Easby
PhD Thesis
'a
00,001F
183
f4
00,00C
Ct
0
C3
00,9
00,00V
Li
P, C Easby
PhD Thesis
184
00,00
'SZOO
co
R
CD
V)
C5
Q
Ct
Ul
rLa
In
40
tn
00,09
00,9Z
0
R
w
La
P, C Easby
PhD Thesis
e4
183
C.)
13
<
IQ
-
CN
oolo&
---
-------
A
-11
cq
rl
E,
II
IiI
111
11
Ct
Jj
wi
Of
L
e
cn
OO'SH
tn
4------
Ll
C)
q
a0
M
0-u
RC Easby
PhD Thesis
C3
ci
Ct
tn
li,
to
I<I
In
If
l
'a
I c:y -,
196
<
I
CID
7
I&
Is,b
M
Ct
co
CA
----------
In
00'9
C
R
to
C.4
40
F- C Easby
PhD Thesis
1;
CN
en
C3
Ln
Cq
00'Z
to
r4
I?
Ui
C5
R
Ln
19
0v
.
RC Easby
PhD Thesis
...
to
I<
187

Easby 07

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Easby 07

Uploaded by

Copyright:

Available Formats

NEWCASTLE UNIVERSITY

Fire Behaviour of Pultruded Composites

Thesis submitted in accordancewith the requirementsof the University of Newcastle

Centre for CompositeMaterials Engineering,

Fire Behaviour ofPultruded Composites

Fire Behaviour ofPultruded Composites

Fire Behaviour ofPultruded Composites

Introduction to Composite Materials

The 'Slow Burn-Through' Effect

Fire Behaviour ofPultruded Composites

Fire Behaviour under Load

Thermogravimetric Analysis (TGA) Parameters

Propane Burner Test

Rear Face Temperature Profile

Tensile Strength at Elevated Temperatures

Compressive Strength at Elevated Temperatures

Longitudinal (El) & Transverse (E2) Stiffness at Elevated Temperatures

FIRE TESTING UNDER LOAD

Tensile Tests in Fire

Compression Tests in Fire

Pool Fire Test under Load

, Furnace Test under Load

Columns under Compression in Fire

Fire Behaviour ofPultruded Composites

Columns under compression

MODELLING BEHAVIOUR UNDER LOAD

Input of Mechanical Properties vs. Temperature

Mechanical properties as a function of temperature

Fire testing under Load

Thermally Induced Deformations

Modelling of fire protective coatings

Fire Behaviour ofPultruded Composites

Visual Basic Code

Thermal Model Outputs

Fire Behaviour ofPultruded Composites

bending moment matrix

Fire Behaviour ofPultruded Composites

Fire Behaviour ofPultruded Composites

Introduction to Composite Materials

Composite materials have a high strength-to-weightratio when comparedto metals,

Table 1.1. Typical propertiesof E-glass [11].

Fire Behaviour of Pultruded Composites

Unidirectional (UD) reinforcement consists of fibres running in the same direction

Chopped Strand Mat/Needle Weave

Fire Behaviour ofPultruded Composites

Table 1.2. Typical propertiesof polyester and phenolic resins [11].

Fire Behaviour of Pultruded Composites

Often an enveloping layer of choppedstrand mat (CSM) needleweave mat is applied

Table 1.3. Advantagesand disadvantagesof the pultrusion process.

Very fast and economical

Limited to constant/nearconstantcross section

Resin content can be accuratelycontrolled

High die costs

High fibre volume fractions can be obtained

Fire Behaviour ofPultruded Composites

Fire Reaction & Fire Resistance

Fire reactionconcernsa material'sresponseto fire. Typically fire reactionexamines

Time-to-ignition (TTI) is a very important fire reaction property

since it defines the onsetof combustion.

Surface spread of flame. The speedof flame-spreadacrossthe surface of

Smoke and toxicity. Smoke reducesvisibility which is of great concern in a

Fire resistancedescribesthe ability of a material or structureto restrict the spreadof

Fire Behaviour ofPultruded Composites