Escolar Documentos
Profissional Documentos
Cultura Documentos
in Aqueous Environment
R. M. Guedes1, Alcides S2, Hugo Faria2
1
da Universidade do Porto, Rua Dr. Roberto Frias s/n, 4200-465 Porto, Portugal
2
Abstract
The aim of this work was to study the long-term failure of GRP (Glass Fiber Reinforced
Polymer) pipes under the influence of moisture absorption. These pipes are used in
water transportation which has an important effect on the mechanical properties of the
polymeric matrix. The GRP pipes are usually tested under ring deflection or internal
pressure conditions. This study presents and analyzes experimental creep-rupture data
obtained from standard test methods under ring deflection conditions. This loading
configuration simulates in laboratory the conditions verified in a sub-soil installation.
The creep testing was carried out under constant dead weight on unconditioned and
preconditioned samples in a submerged condition. The diametrical deflection of
samples was measured periodically and the time to failure of each sample was recorded.
The main purpose of this work was to determine the short and long-term rupture
energies of GRP pipes and assess the influence of moisture preconditioning on those
values. The observed failure mode was always the same. It was concluded that the
energy at failure decreases with time. The influence of the preconditioning on the creeprupture of GRP pipes was considered negligible. Different time-dependent failure
models were described and used for long-term extrapolation of the experimental data.
The maximum strain at failure decreased about 12% from 0.1 to 1000 hours of creep
testing. Furthermore data extrapolation to 50 years predicts a reduction of strength of
about 60%, founded on the most conservative time-dependent failure criterion.
1. Introduction
In recent years materials that possess high specific strength and specific modulus were
developed to fulfil the need for advanced lightweight structures. Fibre reinforced
plastics (FRP) have these characteristics, and are being used as primary as well as
secondary load carrying members in civil engineering structures. Consequently, the
development of testing procedures to predict the lifetimes of materials in extreme
service environments is becoming a high priority.
The problem tackle in this paper deals with durability issue which is usually associated
to different important fields of research. Those include viscoelastic effects, damage
initiation and propagation, environmental effects and glass-fiber stress corrosion. The
last case has been well studied and several models, based on fracture mechanics, have
been presented to model the propagation of a stress-corrosion crack in a glass fibre and
predict the time to failure in different environments [1-10].
High-performance polymer composites exhibit a time-dependent degradation in
modulus (creep or stress relaxation) and strength (creep rupture) as a consequence of the
viscoelasticity of the polymer matrix [11, 12]. One of the most important aspects of
long-term durability and dimensional stability of these materials is their long-term creep
behaviour. Prediction of long-term integrity of any polymeric composite structure
depends on the viscoelastic properties of these materials [11-13]. However, long-term
properties prediction of GRP pipes remains a difficult task because extensive long-term
mechanical tests on the components are required [14].
Yet the structural applications of composite materials in civil construction are becoming
more important. One major application is on rehabilitation in renewal of the structural
inventory, to repair or strengthen. The success of these applications has promoted the
development of new solutions based on FRP (Fiber-Reinforced Polymers). Although
these new products may promise a better mechanical performance on a long-term basis,
the lack of a historical record leads to overdesigned structures. The main reason is
because durability factors, usually determined after long-term experimental tests,
depend on each material system.
On the other hand the full comprehension of internal material changes from microscopic
scale up to the full structure length is far from being well known. The interaction
0
0
n
(1)
where D0, D1, n are material constants; and 0 represents the time unity (equal to
1second or 1hour or 1day, etc.).
The free stored energy, using Hunter [25] formulation, is given by
1
1 t t 2t 1 2 ( 1 ) ( 2 )
2
Ws ( t ) = ( t ) ( t ) D0 ( t ) D1
d 1d 2 .
2
2 0 0
0
2
1
n
(2)
( )
d .
(3)
Accordingly these time-dependent failure criteria [26] predict the lifetime under
constant load, as a function of the applied load 0 and the creep strength under
instantaneous condition R:
D0 2
R ,
2
t f 1 n D0 n 1
=
1
n
0 2 2 D1
Maximum Work Stress Criterion (MWS), states that Wt ( t )
1
n
(4)
D0 2
R ,
2
t f 1 D0 n 1
=
1
0 2 D1
1
n
(5)
t f D0 n 1
1
=
0 D1
1
n
(6)
1n
t f 1 n D0 n 1
=
1
.
n
2
D
1
0
(7)
where = 0 R .
2
Another possible approach is based on the kinetic rate theory of fracture. Based on this
Zurkov [27] used a simple relationship to predict lifetime of several different materials
(except for very small stresses) in terms of a constant stress level ,
t f = t0 exp (U 0 ) kT .
(8)
where t0 is a constant on the order of the molecular oscillation period of 10-13s, k is the
Boltzmann constant, T is the absolute temperature, U0 is a constant for each material
regardless its structure and treatment and depends on the previous treatments of the
material and varies over a wide range for different materials.
3. Materials
The GRP pipes used on the experimental program were produced by centrifugal casting.
The pipes are composed by unsaturated polyester reinforced with fibreglass filled with
sand. The mechanical properties provided by the manufacturer are described on Table I.
Accordingly with the manufacturer the pipe wall is made up of three different major
layers, each one possessing different properties according to its functions. Moreover the
pipe wall is composed by an inner layer which provides the chemical resistance, by a
mechanical resistant layer, and a top coat. The specimens used in creep tests are
depicted in Figure 1.
Burn-off tests were made according to the standard procedure NP 2216 1988 on 10
specimens removed from the pipes in order to analyse the components and the structure
of the specimens. The structure of the pipes is heterogeneous and it is constituted by
roving, random matt, polyester resin and sand according to the values displayed on
Table II. Ten plies were observed each one composed by a sub-layer of roving in the
hoop direction and another sub-layer of random matt as depicted in Figure 2. The matrix
is composed of a mixture of polyester resin with sand particles of small size distributed
uniformly.
The glass transition temperature (Tg) of the composite was measured, using a DMTA
machine of the Polymer Laboratories, as 106.5C (3.4C) for the unconditioned
specimens. The glass transition temperature was identified as the peak of loss modulus
measured at 1Hz. The measured glass transition temperature of the conditioned
specimens in water at 50C was 100.8C (8.7C).
4. Methods
Prior to testing, samples were stored at room temperature of 2025 C and 50% relative
humidity. Care was taken to prepare and maintain the pipe samples free of scratches. A
number of selected specimens were conditioned in a container filled with tap water. The
hoop edges were not protected to prevent water ingress.
The creep tests were carried out following to the European standard EN 1227:1997 [28].
This standard is used to determine the long-term ultimate relative ring deflection under
wet conditions on glass-reinforced thermosetting plastics (GRP) pipes. The
experimental creep tests were carried out applying a dead weight to the specimen
submerged in water at room temperature. The deflection was measured periodically and
the failure time recorded. The test set-up is depicted in Figure 3a and 3b.
Creep tests were preformed on 18 unconditioned specimens using several load levels
(10 to 14 KN) [28]. Creep tests were also performed on specimens preconditioned in
water at 50C. After the preconditioning the specimens were tested in the same
conditions, i.e. submerged in water at room temperature. In Table III are listed all the
specimens tested, with the load level and the applied preconditioning conditions.
The deflection was measured with a deflection dial gauge with a precision of 0.01mm.
Since the load was applied with a beam bar there were no local indentations to account
for.
In this particular case, a very flexible structure, large defections occur well before
failure takes place. Therefore the circumferential elastic modulus E, function of the
applied load P and the maximum deflection W of a pipe with an average radius R and
with a thickness h, must be calculated taking into account the geometrical nonlinearity
[29]:
E=
max
,
max
(9)
where
max =
max
1 12
+ 1 P ,
bR ( h R )
h
W
R
2
with (W )
( 8 + (W ) ) R
2
120 2 W
8) + 128
and = ,
9
which reduces to the linear relationship when = 0 . In Figure 5 is depicted the strain
calculated using the linear and the non-linear relationships against the strain measured,
using an electrical strain gauge. One should note that the strain gauge fail before the
pipe collapsed. It is clear that the non-linear relationship must be used, almost from the
beginning, to obtain the correct maximum strains from the maximum deflection.
The strain gauges were only used in static testes. They were placed in the tensile
stresses sides where the maximum strain was verified, i.e. along the loading line. This
was done to check the accuracy of the equation (9).Five specimens were tested with
strain gauges with a good repeatability.
During the creep test the load is prescribed as constant and the maximum deflection
W ( t ) becomes time dependent as well the modulus E ( t ) which can be obtained from
the previous elastic solution assuming a linear viscoelastic behavior and applying the
Elastic-Viscoelastic Correspondence Principle [30] as
E (t ) =
max
max ( t )
(10)
In fact this is not the real relaxation modulus but the reciprocal of creep compliance.
The relaxation modulus can be obtained using the simple approach suggested by Park et
al. [31]. However in this case the differences were negligible. Finally the creep
compliance is given by
t
1
D (t ) =
= D0 + D1 ,
E (t )
0
(11)
where D0, D1, n are material constants; and 0 represents the time unity (equal to
1second or 1hour or 1day, etc.). This expression is usually designated as the power law
and allows the creep compliance representation of many polymers and polymer based
composites materials in a time scale of practical interest [32].
6. Experimental results
In this section the creep and lifetimes results are presented. The objective is to assess the
long-term prediction methodologies and the hygrothermal effects due to the
preconditioning. Since the material presents initial stiffness variability, of 0.61 GPa for
unconditioned specimens and 0.75GPa for the conditioned specimens in water at 50C,
the creep compliance was normalized to the initial value D0 calculated at 0.01hours
which corresponds to initial creep deflection. This was done to compare more easily the
creep data.
Creep tests
Unconditioned specimens
The measured average initial modulus of the unconditioned specimens was 10.48
(0.61) GPa. The circumferential normalized creep compliance of some unconditioned
specimens against the power law equation is depicted in Figure 6.
Lifetime results
The creep tests were initiated using a loading rate as slow as necessary to obtain the
desired creep load in 3 min (0.5 min). Under creep loading conditions the short-term
(<0.1 h) failures take place at stress levels close to 165MPa.
Therefore the reference value of the creep strength was chosen to be165MPa. In Figures
8 and 9 are represented the maximum applied stress and the strain at failure,
respectively, against the lifetime. Three preconditioned specimens, indicated by arrows,
did not fail after 5000 hours when the test was stopped.
The Zurkov type relationship [27] fits reasonably the experimental data with a squared
correlation coefficient of 0.72. The low value of the correlation coefficient reflects the
scatter of the creep rupture data, which is typical of creep rupture in general. The data
extrapolation for 50 years gives a maximum stress limit of 103 MPa. This represents a
strength reduction of 37% compared against the reference strength (165MPa). Although
slightly different curves were determined for the unconditioned and preconditioned
specimens as depicted in Figure 8, the influence of the preconditioning on lifetime was
considered negligible. The main effect was a drop of 4% in the initial modulus. A
decrease on the maximum strain at failure was detected, 12% after 1000 hours and, by
extrapolation, 17% after 50 years.
7. Discussion of results
Both conditioned and unconditioned creep specimens were tested under the same load
and environment conditions. When comparing the instantaneous modulus (elastic) and
the creep compliance we can conclude that the preconditioning caused a 4% drop of the
elastic modulus but did not affect the viscoelastic behaviour. Nevertheless the reduced
number of preconditioned specimens tested as well the applied reduced stress range is
no sufficient to draw definitive conclusions.
The lifetime expressions previously summarized depended on the viscoelastic properties
and on the creep strength under instantaneous condition. Moreover these expressions
depend on the rate between the viscoelastic and elastic compliance parameters (D1/D0).
From the creep tests this ratio was determined as 0.030. However, for the lifetime
10
prediction, it was verified that the ratio D1/D0 should be larger, i.e. 0.08. One
explanation for this can be attributed to the linear viscoelastic assumption. Since the
stress state is not uniform, the stress influence over the creep compliance is averaged
when linear viscoelastic behaviour is presumed. Therefore one should expect higher
creep compliance values at the critical points, since the stress acts as an accelerator
factor on the viscoelastic behaviour [32]. The creep strength under instantaneous
condition was determined to be 165MPa. The viscoelastic parameters are summarized in
Table IV.
If we consider the normalized rupture free energy as
wf =
wf
D0 2
R
2
(12)
0 =
0
,
R
(13)
(14)
w f = 2n 0 + (1 2 n 1 ) .
2
(15)
(16)
(17)
If we plot the experimental data against the theoretical values, calculated using the
values in Table IV, it became quite clear that the R-W criterion is not appropriated for
the present case, as depicted in Figure 10.
In Figure 11 are depicted the theoretical lifetime predictions against the experimental
results. As it was expected from the previous analysis the R-W criterion is not well
suited for this case. All the other models follow quite well the experimental data. Table
V summarizes the lifetime predictions for each theoretical approach. The maximum
strain criterion is more conservative when compared against the other models.
11
In this case it seems appropriate to apply the MS criterion. The criterion can be
rewritten as follows
t f max D0 0
=
D1 0
0
1
n
(18)
Now using the MS criterion and considering a 50 year (438,000 hours) lifetime
prediction we obtain 79MPa and 68MPa for max=0.0160 and for max =0.0135,
respectively, see also Figure 12. These lifetime predictions imply a creep strength
reduction between 52 and 59 %, respectively.
Before closing this section something more has to be said about the preconditioned
effect. Although the preconditioning in water at 50C appeared to be irrelevant some
extra phenomena may occur. One phenomenon developed at high temperatures is the
post-curing process and the other is the acceleration of the ageing effect. In both cases
the outcome would be an increase of stiffness and strength. This in part would
compensate the stiffness decrease due to the water uptake. However neither the postcuring process nor the ageing effect were measured or taken into account in this study.
8. Conclusions
Creep tests were performed on GRP pipes in order to determine the long-term creep
compliance and creep-rupture. It was observed that preconditioning in water at 50C
had a small influence on the initial stiffness of GRP pipes, a decrease of about 4%.
Furthermore the creep and creep-rupture behaviour in aqueous environment of
preconditioned and unconditioned GRP pipes under lateral deflection was very similar.
Not quite surprisingly Zurkov [27] type relationship was able to fit the entire lifetime
experimental data with a reasonably correlation. Other lifetime models based on
different time-dependent failure criteria, which depend on the viscoelastic properties
and on the creep strength under instantaneous conditions, were briefly presented and
successfully applied, with the exception of Reiner-Weissenberg criterion. Moreover it
was verified that the maximum strain criterion was the most conservative criterion of
all. On that ground and considering a 50 year lifetime, a strength reduction of about
60% was predicted.
Acknowledgements
12
The research hereby presented was supported by Fundao para a Cincia e Tecnologia
(Ministrio da Cincia e do Ensino Superior) through project POCTI/EME/47734/2002.
The authors acknowledge the support of Dr. Joo Rodrigues (INEGI) who performed
and analyzed all the burn-off tests.
References
1. Sekine, H., Beaumont, P.W.R. A physically based micromechanical theory of
macroscopic stress-corrosion cracking in aligned continuous glass-fibre-reinforced
polymer laminates. Composites Science and Technology 1998; 58 (10):1659-1665
2. Sekine, H., Beaumont, P.W.R. On a simple power law for macroscopic crack
propagation rate due to stress-corrosion cracking in unidirectional GFRP
composites. Materials Science and Engineering 2000; A 285 (1-2): 298-302
3. Pauchard, V., Chateauminois, A., Grosjean, F., Odru, P. In situ analysis of delayed
fibre failure within water-aged GFRP under static fatigue conditions. International
Journal of Fatigue 2002; 24 (2-4):447-454
4. Pauchard, V., Boulharts-Campion, H., Grosjean, F., Odru, P., Chateauminois, A.
Development of a durability model applied to unidirectional composites beams
reinforced with glass fibers. Oil and Gas Science and Technology 2001;56 (6):581595
5. Khennane, A., Melchers, R.E. Durability of glass polymer composites subject to
stress corrosion. Journal of Composites for Construction 2003;7 (2):109-117
6. Barbero, E.J., Damiani, T.M. Phenomenological prediction of tensile strength of Eglass composites from available aging and stress corrosion data. Journal of
Reinforced Plastics and Composites 2003;22 (4):373-394
7. Ely, T., Armentrout, D., Kumosa, M. Evaluation of stress corrosion properties of
pultruded glass fiber/polymer composite materials. Journal of Composite Materials
2001;35 (9):751-773
8. Kumosa, L., Armentrout, D., Kumosa, M. An evaluation of the critical conditions
for the initiation of stress corrosion cracking in unidirectional E-glass/polymer
composites. Composites Science and Technology 2001;61 (4):615-623
9. Megel, M., Kumosa, L., Ely, T., Armentrout, D., Kumosa, M. Initiation of stresscorrosion cracking in unidirectional glass/polymer composite materials. Composites
Science and Technology 2001;61 (2):231-246
10. Ely, T., Kumosa, M. Stress corrosion experiments on an E-glass/epoxy
unidirectional composite. Journal of Composite Materials 2000;34 (10):841-878
11. Raghavan J.; Meshii M. Creep Rupture of Polymer Composites. Composites
Science and Technology 1997;57:375-388.
13
14
26. R.M. Guedes. Mathematical analysis of energies for viscoelastic materials and
energy based failure criteria for creep loading. Mechanics of Time-Dependent
Materials 2004;8(2):169-192.
27. Zhurkov, SN, Kinetic Concept of the Strength of Solids, International Journal of
Fracture Mechanics 1965;1(4):311.
28. Faria H. Failure Analysis of GRP Pipes Under Compressive Ring Loads. MSc
Thesis, Faculdade de Engenharia da Universidade do Porto; 2005.
29. Guedes RM. Stressstrain analysis of a cylindrical pipe subjected to a transverse
load and large deflections. Compos Struct (2008),
doi:10.1016/j.compstruct.2008.03.031.
30. Guedes RM, Marques AT, Cardon A. Creep or Relaxation Master Curves
Calculated from Experimental Dynamic Viscoelastic Function. Science and
Engineering of Composite Materials 1998;7 (3):259-267.
31. Park SW, Kim YR. Interconversion between relaxation modulus and creep
compliance for viscoelastic solids. Journal of Materials in Civil Engineering
1999;11(1): 76-82.
32. Findley WN, Lai JS and Onaran K. Creep and Relaxation of Nonlinear Viscoelastic
Materials. Dover Publications, New York;1989.
15
Figure Captions
16
List of Tables
17
ERadial (MPa)
ELongitudinal (MPa)
10547
3808
3808
18
Average values
Standard deviation
Polyester (matrix)
35.65
0.41
Roving
10.41
1.51
Matt
12.66
0.88
Sand
41.28
2.54
19
Table III Creep specimens with the correspondent load and preconditioning.
Specimen N
Preconditioning
Load (N)
C36
No
12196
C46
No
11831
C47
No
12087
C55
No
12400
C56
No
12400
C38
No
10643
C50
No
10380
C52
No
10431
C39a
No
12719
C48
No
12830
C49
No
12692
C51
No
13638
C57
No
13088
C39b
No
13970
C45
No
14420
C58
No
13988
C53
No
11411
C54
No
11139
C88
11096
C89
11096
C92
11353
C90
9865
C91
10567
C93
9084
C94
10059
C95
11358
20
D1/D0*
0.030
0.080
0.200
R
MPa
165
*
Value used for lifetime prediction
21
Maximum Stress
1000 h 1 year
146
138
130
116
129
117
125
110
50 years
119
86
93
79
22
List of Figures
23
R250mm
12mm
300mm
24
Figure 2 Detail of the ply reinforcement composed of roving in the hoop direction and
random matt.
25
(a)
(b)
Figure 3 Creep test apparatus; a) during load calibration and b) submerged in a tank
during creep tests.
26
27
300
Pipe failure
Stress (MPa)
250
200
150
100
Strain gauge
Linear calculation
50
Non-linear calculation
0
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
Strain
Figure 5 Comparison between the measured strain and the strain calculated using
linear and non-linear relationships.
28
1.20
D( t )
Dt
=1+ 1
D0
D0 0
1.10
Not Precond.
1.00
0.1
10
100
1000
10000
Time (hours)
Figure 6 Normalized creep compliance of GRP pipes under ring deflection, without
preconditioning.
29
1.20
D( t )
Dt
=1+ 1
D0
D0 0
Precond.
1.10
1.00
0.1
10
100
Time (hours)
1000
10000
30
tf
180
Stress (MPa)
0
140
100
=e
, 0 =1 hour, 2 = 0.72
Not Precond
Precond.
Zurkov (Not Precond.)
Zurkov (Precond.)
tf
60
0.01
156
4.09
0
0.1
10
153
4.87
=e
, 0 =1 hour, 2 = 0.72
100
1000
Figure 8 Lifetime GRP pipes under ring deflection versus the maximum applied
31
0.020
0.016
0.012
Precond
Not Precond
0.008
0.0108
max
0.004
0.000
0.01
0.1
t
= 0.0154
0
, 0 =1 hour, 2 = 0.257
100
10
1000
10000
100000
Time (hour)
Figure 9 Lifetime GRP pipes under ring deflection versus the maximum strain at
failure.
32
1.0
0.8
0.6
Experim.
R-W
MR-W
MSW
MS
0.4
0.2
0.0
0.0
0.2
0.4
0.6
0.8
1.0
Normalised Stress
Figure 10 Normalised free energy at failure versus normalised stress for the
experimental data and theoretical models.
33
Stress (MPa)
180
140
Not Precond
Precond.
R-W
MR-W
MSW
MS
100
60
0.01
0.1
10
100
1000
Figure 11 Experimental lifetime and theoretical predictions for GRP pipes under ring
deflection.
34
180
Stress (MPa)
max=0.0160
140
100
Not Precond
max=0.0135
Precond.
60
0.01
0.1
10
100
1000
Figure 12 Lifetime prediction of GRP pipes under ring deflection using the maximum
strain criterion, considering two extreme values.
35