Professional Documents
Culture Documents
Piping Applications
Design, Analysis and Optimization of
Subsea and Onshore Pipelines from FRP Materials
M ai n en t r y u n d e r t i t l e :
Composite Material s in Piping Ap plicat ions: Desig n, Analysis and Optimizatio n of Subsea
a n d On s h o r e P i p e l i n e s f r o m F R P M a t e r i a l s
A DE S t e c h P u b l i c a t i o n s b o o k
Bibliog rap hy: p .
In c lu d e s in d e x p . 3 9 5
D.G. Pavlou
xi
Introduction
The Minoans in 27th century BC were the first civilization known to use un-
derground clay pipes for sanitation, heating systems and water supply [1]. At
Knossos, on the island of Crete, Greece, pipes having a diameter of 4.0–6.0 inches
with perfect socket joints are the oldest ever discovered. The first known applica-
tion of metal for manufacture of pipes is “hydraulis,” a 3rd century Greek pipe
organ that operated by converting the dynamic energy of water into air pressure to
drive the pipes. Later, in the first century BC, the Romans used lead to fabricate
pipes with diameters from 0.5–22.0 inches for urban plumbing.
Fiber-reinforced pipes were first developed after World War II [2]. They con-
sisted of glass fiber cloth and resin applied over a mandrel by hand. An evolved
form of the handmade pipe is the filament-wound pipe consisting of tensioned
fibers properly oriented to bear the combination of hoop and axial forces.
Manufacturing of the first commercial FRP pipes started in the mid 1950s [2].
From 1960–1980 a continuous process for manufacturing of FRP pipes was de-
veloped and its efficiency improved to the point where large quantities of pipe
were produced for the chemical and oil industry. After the peak of 1981, a slight
decrease in the productivity rate occurred until 1986. At present, FRP pipes find
many applications around the world since they combine high resistance to corro-
sive fluids with a capacity for increased mechanical loads. FRP pipes are used on
offshore platforms in Alaska and the Persian Gulf, as well as in water-flood proj-
ects in Saudi Arabia and in a saltwater/crude oil/gas line in the deserts of South
Oman. In 1993 the largest FRP pipe, with a length of 390Km and a diameter of
350 mm, was installed in Algeria [2] and is used for oilfield applications.
Since the unit price of carbon steel is about 14 times cheaper than the unit
price of Glass Fiber Reinforced Polymers (GFRP), steel pipes are still the main
type of pipes uses for fluid transmission. However, the unit price is not the only
parameter controlling the material cost. The allowable tensile force for a mate-
rial sheet subjected to tension is P = ASy, where A is the cross section carrying
the load and Sy is the material’s yield stress. The above formula can also be
written as P = (V/L)Sy, where V is the volume and L is the length of the material
xiii
xiv Introduction
DENSITY Kg/m3
Cost index CI
References
[1] http://www.historywiz.com
[2] Oswald Kenneth, Thirty years of fiberglass pipe in oilfield applications: A
historical perspective, Materials Selection and Design, MP/May, 1996.
Preface
After World War II, offshore and continental oil, gas and water transmission
infrastructure, as well as chemical, sewage and irrigation installations, benefited
from the development of fiber reinforced polymeric (FRP) pipes. However, be-
cause the unit price of composite materials was historically expensive, steel pipes
remained in use for the transmission of liquid commodities. Today, high main-
tenance costs due to corrosion of aging steel pipelines, as well as the reduction
of the unit price of composite materials, have led to reconsidering the optimum
material cost for pipeline applications. Moreover, as will be shown in the intro-
duction, the final material cost of piping is strongly influenced by material density
and strength. Since FRP materials exhibit a much lower density and much higher
strength than carbon steel, the final cost of such materials is currently comparable
to the cost of carbon steel. Moreover, the lower maintenance costs of composite
pipelines, which results from their excellent resistance to corrosion and fatigue,
leads to the conclusion that the use of composite materials for pipeline applica-
tions has now become advantageous when compared to the use of carbon steel
pipelines.
Since FRPs are anisotropic materials, the methods and theoretical tools for
the mechanical design of composite pipelines are completely different from the
design procedures for steel pipelines. The existing design standards are rather
semi-empirical and cover simple loading cases.
The aim of the present book is to provide detailed analytic and numerical tools
for the analysis and design of FRP composite pipelines under pure and combined
loading conditions (e.g., bending, external pressure and axial tension). Failure
prediction in creep and fatigue conditions, design of joints and supports, esti-
mation of flow capacity for liquid gas and multi-phase fluids are major topics
investigated in the following chapters. A strong feature of the book is the devel-
opment of Mathematica-based computer algorithms corresponding to any load-
ing conditions, in order to facilitate direct design. Moreover, nomographs for a
wide range of loading cases, pure and combined, have been derived for multi-
layered filament-wound pipes made from the common composites E-glass/epoxy
ix
x Preface
and S-glass/epoxy for quick dimensioning. Since in the first chapter principles
of the mechanics of anisotropic elasticity are briefly explained, a reader unfamil-
iar with composite materials can appreciate and understand the design principles
presented.
The book is organized as follows:
Dimitrios G. Pavlou
Technical Institute of Chalkida
January 2013
Contents
Preface ix
Acknowledgments xi
Introduction xiii
10 Case Studies..................................................................299
Introduction 299
10.1 Axial Tension 299
10.1.1 Results of failure model for axial tension 299
10.2 Pure Bending 310
10.2.1 Results of failure model for pure bending 310
10.2.2 Results of buckling model for pure bending 316
viii Contents
Index 395
About the CD-ROM 399
Chapter 1
1
2 Mechanical Behavior of Fiber Reinforced Composite Materials
where
1
S11 = (1.4)
E1
v21
S12 = − (1.5)
E2
4 Mechanical Behavior of Fiber Reinforced Composite Materials
v31
S13 = − (1.6)
E3
v12
S 21 = − (1.7)
E1
1
S 22 = (1.8)
E2
v32
S 23 = − (1.9)
E3
v13
S31 = − (1.10)
E1
v23
S32 = − (1.11)
E2
1
S33 = (1.12)
E3
1
S55 = (1.13)
G13
1
S55 = (1.14)
G13
1
S66 = (1.15)
G12
Mechanical Behavior of Laminae 5
Combining equations (1.2) and (1.3), it can be concluded that Ci j can be written
in terms of the above engineering constants Si j . For shorthand notation, the equa-
tions (1.2) and (1.3) can be abbreviated as:
[C ] = [ S ]−1 (1.18)
v12 v21
= (1.19)
E1 E2
v13 v31
= (1.20)
E1 E3
v23 v32
= (1.21)
E2 E3
yielding
S31 = (1.23)
S13
S32 = (1.24)
S 23
Therefore, with the aid of eqs. (1.18) and (1.22–1.24), the members Cij of the
matrix [C ] are given by:
S 22 S33 − S 232
C11 = (1.25)
S
S13 S 23 − S12 S33
C12 = (1.26)
S
6 Mechanical Behavior of Fiber Reinforced Composite Materials
S12 S 23 − S13 S 22
C13 = (1.27)
S
S11 S 22 − S122
C33 = (1.33)
S
1
C44 = (1.34)
S 44
1
C55 = (1.35)
S55
1
C66 = (1.36)
S66
where
S = S11 S 22 S33 − S11 S 232 − S 22 S132 − S33 S122 + 2S12 S 23 S13 (1.37)
The matrices [C ] and [ S ] correlating the stress and strains are called the stiffness
matrix and the compliance matrix, respectively.
The mechanical strains used in the above equations are given by:
ε1mech ε1 − α1 ∆Τ
mech ε − α ∆Τ
ε2 2 2
ε3mech ε3 − α3 ∆Τ
mech = (1.40)
γ 23 γ 23
γ13
mech γ13
mech
γ12 γ12
For each of these three cases, the corresponding stress–strain relationships can be
written in the following form:
ε1 − α1 ∆Τ 0
ε − α ∆Τ 0
2 2
ε3 − α3 ∆Τ 0
= (1.42)
γ 23 0
γ13 0
γ12 0
The mechanical strains used in the above equations are given by:
ε1mech ε1 − β1 ∆Μ
mech ε − β ∆Μ
ε2 2 2
ε3mech ε3 − β3 ∆Μ
mech = (1.46)
γ 23 γ 23
γ13
mech γ13
mech
γ12 γ12
In the case of interaction between moisture strains and thermal strains, superposi-
tion of eqs. (1.38) and (1.44), as well as eqs. (1.39) and (1.45), yields:
10 Mechanical Behavior of Fiber Reinforced Composite Materials
γ13 = 0 (1.50)
Mechanical Behavior of Laminae 11
γ 23 = 0 (1.51)
Equations (1.50) and (1.51) show that planes x2 − x3 and x1 − x3 can be consid-
ered as free of shear strains.
Although the normal strain ε3 is not zero, the plane–stresses assumption
described above yields a simplified Hooke’s law involving only σ1 , σ 2 , τ12 and
ε1 , ε2 , γ12
ε1 s11 s12 0 σ1
σ
ε2 = s12 s22 0 2 (1.52)
γ 0 0 s66 τ12
12
where
1
S11 = (1.53)
E1
v12 v
S12 = − = − 21 (1.54)
E1 E2
1
S 22 = (1.55)
E2
1
S66 = (1.56)
G12
σ1 Q11 Q12 0 ε1
σ 2 = Q12 Q22 0 ε2 (1.57)
τ 0 0 Q66 γ12
12
where the parameters Qij are called reduced stiffnesses, and are given by the
following equations:
C132
Q11 = C11 − (1.58)
C33
12 Mechanical Behavior of Fiber Reinforced Composite Materials
C13C23
Q12 = C12 − (1.59)
C33
C232
Q22 = C22 − (1.60)
C33
The above equations, with the aid of eqs. (1.25)–(1.33), (1.36), (1.37) and eqs.
(1.4)–(1.15), yield:
E1
Q11 = (1.62)
1 − v12 v21
v12 E2 v E
Q12 = = 21 1 (1.63)
1 − v12 v21 1 − v12 v21
E2
Q22 = (1.64)
1 − v12 v21
Taking into account the plane–stresses assumption, equations (1.47) and (1.48),
describing the generalized Hooke’s law for the cases of the effects of free thermal
and moisture strains, can now be written as:
ε1 − α1 ∆Τ − β1 ∆Μ s11 s12 0 σ1
ε2 − α2 ∆Τ − β 2 ∆Μ = s12 s22 0 σ2 (1.66)
0 τ
γ12 0 s66 12
and
ε1mech ε1 − α1 ∆Τ − β1 ∆Μ
mech
ε2 = = ε2 − α2 ∆Τ − β 2 ∆Μ (1.68)
γ mech γ12
12
Figure 1.3 Definition of the principal coordinate system x1 − x2 − x3 with respect to the
global coordinate system x − y − z .
14 Mechanical Behavior of Fiber Reinforced Composite Materials
For the case of plane-stress, the above transformation is given by the following
matrix equation:
σ1 σ x
σ 2 = [T ] σ y (1.70)
τ
12 τ xy
where the transformation matrix [T ] is given by:
m2 n2 2mn
[T ] = n m −2mn
2 2
(1.71)
− mn mn m 2 − n 2
m2 n2 −2mn
2
[T ]−1
= n m 2
2mn (1.72)
mn − mn m 2 − n 2
The stresses σ x , σ y , τ xy can be expressed with respect to σ1 , σ 2 , τ12 :
σ x σ1
σ y = [T ]
−1
σ 2 (1.73)
τ
τ xy 12
ε1 εx
ε2 = [T ] ε y (1.74)
1 γ 1 γ
2 12 2 xy
With the aid of equations (1.70), (1.74), equation (1.52) yields
εx S 11 S 12 S 16 σ x
ε y = S 12 S 22 S 26 σ y (1.75)
γ
xy S 16 S 26 S 66 τ xy
where
(
S 12 = ( S11 + S 22 − S66 ) n 2 m 2 + S12 n 4 + m 4 ) (1.77)
(
S 66 = 2 ( 2 S11 + 2 S 22 − 4 S12 − S66 ) n 2 m 2 + S66 n 4 + m 4 ) (1.81)
The above parameters S i j are called transformed reduced compliances, and the
corresponding equation (1.75) is a fundamental equation for analysis of fiber-
reinforced layers. An expanded form of equation (1.75) and the corresponding
S i j parameters are given in the Appendix to Chapter I.
where
(
Q12 = (Q11 + Q22 − 4Q66 ) n 2 m 2 + Q12 n 4 + m 4 ) (1.84)
The parameters Q i j correlating the stress with respect to the strains in the global
coordinate system are called transformed reduced stiffnesses. Like the trans-
formed reduced compliances, the transformed reduced stiffnesses vary signifi-
cantly with fiber orientation ϑ .
σ x = Ex εx (1.89)
( )
For the above situation σ x ≠ 0 , σ y = 0 , τ xy = 0 equation (1.75) yields
εx = S 11 σ x (1.90)
Therefore, the combination of equations (1.89) and (1.90) provides the definition
of the elasticity modulus in the x direction.
Mechanical Behavior of Laminae 17
1
Ex = (1.91)
S 11
Taking into account equations (1.76), (1.52–1.56), the above equation can be writ-
ten as:
E1
Ex = (1.92)
E E
m 4 + 1 − 2v12 n 2 m 2 + 1 n 4
G12 E2
The Poisson’s ratio in the x − y direction is defined by the ratio of the contraction
strain ε y in the y direction over the extensional strain εx in the x direction or:
εy
vxy = − (1.93)
εx
ε y = S 12 σ x (1.94)
equation (1.93) with the aid of equations (1.90), (1.94) can be written as:
S 12
vxy = − (1.95)
S 11
Using the definitions of S 12 , S 11 and Sij, the above equation provides the follow-
ing formula:
E E
( )
v12 n 4 + m 4 − 1 + 1 − 1 n 2 m 2
E2 G12
vxy = (1.96)
E E
m + 1 − 2v12 n 2 m 2 + 1 n 4
4
G12 E2
For evaluating the modulus of elasticity in the y direction, the stress situation
σ x = 0, σ y ≠ 0 , τ xy = 0 will now be considered. In that case the modulus E y is
given by the following formula:
σy
Ey = (1.97)
εy
18 Mechanical Behavior of Fiber Reinforced Composite Materials
Since
ε y = S 22 σ y (1.98)
1
Ey = (1.99)
S 22
Using the definition of S 22 given by eq. (1.79) and the definitions of Sij given by
equations (1.53)-(1.56), equation (1.99) can now be written as:
E2
Ey = (1.100)
E E
m 4 + 1 − 2v12 n 2 m 2 + 2 n 4
G12 E1
εx
vy x = − (1.101)
εy
εx = S 12 σ y (1.102)
Therefore, with the aid of eqs. (1.98), (1.102), equation (1.101) provides
S 12
vy x = − (1.103)
S 22
Using the definitions of S 12 and S 22 the above equation can now be written as:
E E
( )
v21 n 4 + m 4 − 1 + 2 − 2 n 2 m 2
E1 G12
v yx = (1.104)
E E
m + 2 − 2v21 n 2 m 2 + 2 n 4
4
G12 E1
For the derivation of the formula providing the shear modulus Gxy , the stress situ-
ation σ x = 0 , σ y = 0 , τ xy ≠ 0 will be considered. In that case
Mechanical Behavior of Laminae 19
τ xy
γ xy = (1.105)
Gxy
γ xy = S 66 τ xy (1.106)
Therefore, from equations (1.105) and (1.106) the following can be derived:
1
Gxy = (1.107)
S 66
With the aid of the definition of S 66 the above equation yields:
G12
Gxy = (1.108)
G G
n + m + 2 2 12 (1 + 2v12 ) + 2 12 − 1 n 2 m 2
4 4
E1 E2
1.1.7 Free thermal and free moisture strains in the global coordinate system
(a) Transformation of thermal and moisture expansion coefficients.
Following the concept described in previous paragraphs, the relations of stress
and strain components in the x − y − z system, including the effects of free ther-
mal and moisture strains, can be derived. To this end, the thermal and moisture
expansion coefficients in the global coordinate system will be correlated initially
with the thermal and moisture expansion coefficients in the principal coordinate
system. Inverting equation (1.74) leads to the following formula:
20 Mechanical Behavior of Fiber Reinforced Composite Materials
Figure 1.4 Variation with fiber orientation ϑ of: (a) the elastic modulus Ex , (b) the elas-
tic modulus E y , (c) the shear modulus Gxy , (d) the Poisson’s ratios vxy and v yx .
Mechanical Behavior of Laminae 21
εx ε1
ε y = [T ]
−1
ε2 (1.109)
1 γ 1 γ
2 xy 2 12
Since in the case of free thermal strains,
ε1 = α1 ∆Τ (1.110)
ε2 = α2 ∆Τ (1.111)
γ12 = 0 (1.112)
εx α1
ε y = [T ]
−1
α2 ∆Τ (1.113)
1 γ 0
2 xy
Considering the following definitions of free thermal strains in the global coordi-
nate system
εx = α x ∆Τ (1.114)
ε y = α y ∆Τ (1.115)
γ xy = α xy ∆Τ (1.116)
αx α1
α y = [T ]
−1
α2 (1.117)
1 α 0
2 xy
With the aid of equation (1.72), the above equation provides the correlation of the
thermal expansion coefficients in the global coordinate system versus the thermal
expansion coefficients in the principal coordinate system:
Following the same concept, the coefficients of moisture expansion in the global
coordinate system are given by the following equations:
βx β1
β y = [T ]
−1
β2 (1.121)
1 β 0
2 xy
yielding
α1 αx
α2 = [T ] α y (1.125)
0 1 α
2 xy
and
β1 βx
β 2 = [T ] β y (1.126)
0 1 β
2 xy
the following matrix equation can be obtained:
ε1 αx β x S11 S12 0 σ1
2
ε = − [ ] y
T α ∆Τ − [ ] β y = S12
T S 22 0
σ 2 (1.127)
γ 1 α 1 β 0 τ
0 2 S 66
1
12 2 xy 2 xy 12
Taking into account equations (1.69), (1.74) the above equation yields
εx − α x ∆Τ − β x ∆M S 11 S 12 S 166 σ x
ε y − α y ∆Τ − β y ∆M = S 12 S 22 S 26 σ y (1.128)
γ x y − α x y ∆Τ − β x y ∆M S 16 S 26 S 66 τ xy
24 Mechanical Behavior of Fiber Reinforced Composite Materials
The inverse of the above equation provides the stress components σ x , σ y , τ xy ver-
sus the mechanical strains can be expressed as:
εmech
x = εx − α x ∆Τ − β x ∆Μ , εmech
y = ε y − α y ∆Τ − β y ∆Μ , γ mech
xy = γ x y − α x y ∆Τ − β x y ∆Μ
Table 1.1
Example of the stacking sequence of a laminate.
Layer’s location Fiber orientation θ
layer 1 0
layer 2 45°
layer 3 90°
layer 4 0
Mechanical Behavior of Laminates 25
Figure 1.7 Definition of the global coordinate system for a multilayered FRP laminate.
26 Mechanical Behavior of Fiber Reinforced Composite Materials
In cases where the stacking sequence for the side z ≥ 0 is a mirror image of
the stacking sequence of the side z ≤ 0 , the stacking notation can be abbrevi-
ated. For example, the symmetric six-layered laminate of Table 1.2 is denoted as
a [ 0 / 30 / 60]s laminate, where the subscript s means symmetric. In order for a
laminate to be characterized as symmetric, apart from the symmetric order of lay-
ers, the material properties, fiber orientation and thickness of the layer at a specific
location within the side z ≥ 0 should be identical to the material properties, fiber
orientation and thickness of the corresponding layer at the same location of the
side z ≤ 0 .
In cases where the symmetric laminate involves adjacent layers of opposite
orientation, the stacking notation can also be abbreviated. For example the six-
layered laminate of Table 1.3 can be defined using the notation [ ±30 / 0]s
Finally, when a group of layers is repeated within a symmetric laminate, fur-
ther shorthand notation is used. For the twelve-layered symmetric laminate de-
scribed in Table 1.4, the notation [ (±60 / 0) 2 ]s may be used.
Table 1.2
Example of stacking sequence of a symmetric six-layered laminate.
Layer’s location Fiber orientation θ
Layer 1 0
Layer 2 30°
Layer 3 60°
Layer 4 60°
Layer 5 30°
Layer 6 0
Table 1.3
Example of stacking sequence of a symmetric six-layered laminate
exhibiting adjacent layers of opposite orientation.
Layer’s location Fiber orientation θ
Layer 1 +30°
Layer 2 -30°
Layer 3 0
Layer 4 0
Layer 5 -30°
Layer 6 +30°
Mechanical Behavior of Laminates 27
Table 1.4
Example of a stacking sequence of a symmetric twelve-layered
laminate displaying repeated groups of layers.
In cases in the above notation where the subscript s is missing, the notation
[(±60 / 0) ]
2 represents the unsymmetrical laminate [ +60 / −60 / 0 / +60 / −60 / 0]
∂w0
u = u0 − z (1.130)
∂x
while its vertical movement is:
w = w0 (1.131)
28 Mechanical Behavior of Fiber Reinforced Composite Materials
∂w0
u = u0 − z (1.132)
∂y
where u 0 is the deflection of reference point O in the direction of the y-axis, and
∂wo / ∂x is the slope in the x-y plane of the y–axis at the point O after deformation.
∂u
εx = (1.133)
∂x
∂υ
εy = (1.134)
∂y
∂w (1.135)
εz =
∂z
∂υ ∂u
γ xy = + (1.136)
∂x ∂y
∂w ∂u
γ xz = + (1.137)
∂x ∂z
∂w ∂υ
γ yz = + (1.138)
∂y ∂z
yields:
εx = ε0x + zk x0 (1.139)
ε y = ε0y + zk y0 (1.140)
εz = 0 (1.141)
γ xz = 0 (1.143)
γ yz = 0 (1.144)
∂u
ε0x = (1.145)
∂x
∂υ
ε0y = (1.146)
∂y
∂υ0 ∂u
γ 0xy = + (1.147)
∂x ∂y
30 Mechanical Behavior of Fiber Reinforced Composite Materials
∂ 2 w0
k x0 = − (1.148)
∂x 2
∂ 2 w0
k y0 = − (1.149)
∂y 2
∂ 2 w0
k xy0 = −2 (1.150)
∂ x∂ y
0
The strains ε0x , ε y , γ xy are called reference surface extensional strain in the x
0
direction, reference surface extensional strain in the y direction and reference sur-
face in-plane shear strain respectively. The quantities k x0 and k y are the curva-
0
tures of the reference surface in the x and y directions respectively, while the
quantity k xy0 is the reference twisting curvature.
Since the strains and the reduced stiffnesses Q ij are functions of the location
z of each lamin, and since the material properties controlling the value of Q ij
are generally different for adjacent layers, the distribution of stresses through the
thickness of the laminate is expected to be incremental.
Figure 1.9 Nomenclature for (a) force resultants, and (b) moment resultants.
H
2
NX ≡ ∫H
σ x dz (1.152)
−
2
H
2
Ny ≡ ∫ H
σ y dz (1.153)
−
2
H
2
N xy ≡ ∫H
τ xy dz (1.154)
−
2
32 Mechanical Behavior of Fiber Reinforced Composite Materials
H
2
Μx ≡ ∫ H
σ x zdz (1.155)
−
2
H
2
Μy ≡ ∫ H
σ y zdz (1.156)
−
2
H
2
Μ xy ≡
H
∫ τ xy zdz (1.157)
−
2
The unit of the force resultants is force per unit length, while the unit of the mo-
ment resultants is moment per unit length. For example, the resultants N x , Μ χ
are
∧
Nx
Nx = (1.158)
AB
∧
Mx
Mx = (1.159)
AB
∧ ∧
where N x , M x are the force and moment acting on the segment AB in Figure 1.9,
and AB is the length of the segment AB.
With the aid of eq. (1.151), equations (1.152)–(1.154) can be written as:
H2 H2 H2 H2
0 0 0
N x = ∫ Q11dz εx + ∫ Q11 zdz k x + ∫ Q12 dz ε y + ∫ 12 k y0 +
Q zdz
− H − H − H − H
2 2 2 2
H2 H2
0 0
∫ Q16 dz γ xy + ∫ Q16 zdz k xy (1.160)
− H − H
2 2
Mechanical Behavior of Laminates 33
H2 H2 H H2
0 0 2 0
N y = ∫ Q12 dz εx + ∫ Q12 zdz k x + ∫ Q22 dz ε y + ∫ Q22 zdz k y0 +
− H − H − H − H
2 2 2 2
H2 H2
0 0
∫ Q26 dz γ xy + ∫ Q26 zdz k xy (1.161)
− H − H
2 2
H2 H2 H H2
0 0 2 0
N xy = ∫ Q16 dz εx + ∫ Q16 zdz k x + ∫ Q26 dz ε y + ∫ Q26 zdz k y0 +
− H − H − H − H
2 2 2 2
H2 H2
0 0
∫ Q66 dz γ xy + ∫ Q66 zdz k xy (1.162)
− H − H
2 2
H
+
2
For a laminate with N layers each integral ∫QH
ij dz has the form:
−
2
H
2
∫ Q dz
H
ij = Q ij1 ( z1 − z0 ) + Q ij 2 ( z2 − z1 ) + Q ij3 ( z3 − z2 ) + …. + Q ijk ( zk − zk −1 ) +
−
2
H
+
2
Aij = ∫ Q dz
H
ij (1.164)
−
2
34 Mechanical Behavior of Fiber Reinforced Composite Materials
Bij = ∫
−H 2
Q ij zdz (1.166)
yields
1 N
Bij = ∑ Q ( zk2 − zk2−1 )
2 k =1 ijk
(1.167)
Therefore, with the aid of equations (1.165), (1.167), the force resultants given by
equations (1.160)-(1.162) can be written in the following matrix form:
H2 H2 H2 H2
0 0 0
M x = ∫ Q11 zdz εx + ∫ Q11 z dz k x + ∫ Q12 zdz ε y + ∫ Q12 z dz k y0 +
2 2
− H − H − H − H
2 2 2 2
H2 H2
0 0
∫ Q16 zdz γ xy + ∫ Q16 z dz k xy
2
(1.169)
− H − H
2 2
H2 H2 H2 H2
0 0 0
M y = ∫ Q12 zdz εx + ∫ Q12 z dz k x + ∫ Q22 zdz ε y + ∫ Q22 z dz k y0 +
2 2
− H − H − H − H
2 2 2 2
H2 H2
0 0
∫ Q26 zdz γ xy + ∫ Q26 z dz k xy
2
(1.170)
− H − H
2 2
Mechanical Behavior of Laminates 35
H2 H2 H2 H2
0
M xy = ∫ Q16 zdz εx + ∫ Q16 z dz k x + ∫ Q26 zdz ε y + ∫ Q26 z dz k y0 +
2 0 0 2
− H − H − H − H
2 2 2 2
H2 H2
0 0
∫ Q66 zdz γ xy + ∫ Q66 z dz k xy
2
(1.171)
− H − H
2 2
Using the notation
H 2
Dij = ∫
−H 2
Q ij z 2 dz (1.172)
1 N
Dij = ∑ Qijk ( zk3 − zk3−1 )
3 k =1
(1.174)
In the above equation the 6 × 6 matrix consisting of the components Aij, Bij, Dij is
called the laminate stiffness matrix or ABD matrix. Inversion of the above equa-
tion provides the relation of the strains and curvatures with the force and moment
resultants :
36 Mechanical Behavior of Fiber Reinforced Composite Materials
a11 a12 a16 b11 b12 b16 Α11 Α12 Α16 B11 B12 B16
−1
a b26 Α12
12 a22 a26 b21 b22
Α 22 Α 26 B12 B22 B26
a16 a26 α66 b61 b62 b66 Α16 Α 26 Α66 B16 B26 B66
= =
b11 b12 b61 d11 d12 d16 B11 B12 B16 D11 D12 D16
b12 b22 b62 d12 d 22 d 26 B12 B22 B26 D12 D22 D26
b16 b26 b66 d16 d 26 d 66 B16 B26 B66 D16 D26 D66
(1.177)
and is called the laminate compliance matrix or abd matrix. Knowing the force
and moment resultants acting on the bounding surface of a laminate, the principal
stresses σ1, σ2, τ12 of each layer can now be calculated by following the algorithm
shown in Figure 1.10.
(a) Symmetric laminates are laminates where: (i) the stacking sequence of the
upper half side (with respect to the geometric mid-plane) is a mirror image
of the stacking sequence of the lower half side, and (ii) the material proper-
ties, fiber orientation and thickness of a pair of symmetric (with respect to
the geometric mid-plane) layers are identical.
(b) Symmetric Balanced laminates are laminates consisting of pairs of symmet-
ric (with respect to the geometric mid-plane) layers with identical material
properties and thickness but opposite fiber orientation.
(c) Cross-Ply laminates are ones where every layer has its fibers oriented at
either 0° or 90°.
(d) Symmetric Cross-Ply laminates meet the definitions of (a) and (c) above.
Mechanical Behavior of Laminates 37
Taking into account the above definitions, the ABD matrix can be simplified.
These simplifications yield the following formulations for equation (1.175):
Figure 1.10 Algorithm for calculation of the principal stresses on each layer of a laminate.
38 Mechanical Behavior of Fiber Reinforced Composite Materials
For the special case where the laminate is a single isotropic layer of thickness H,
the equation (1.175) yields
EH vEH
1 − v 2 0 0 0 0
1 − v2
vEH EH
0 0 0 0
Nx 1 − v 2 1 − v2 ε0x
N 0
EH
y 0 0 0 0 0 εy
N xy 2(1 + v) γ 0xy
= EH 3 vEH 3
0
Mx 0 0 0 0 kx
My 12(1 − v ) 12(1 − v 2 )
2 k y0
0
M
xy vEH 3 EH 3 k xy
0 0 0
12(1 − v 2 ) 12(1 − v 2 )
0
EH 3
0 0 0 0 0
24(1 + v)
(1.182)
In the above equation E, v are the modulus of elasticity and the Poisson’s ratio
respectively of the isotropic layer.
where
1 1
F1 = Τ + C (1.184)
σ1 σ1
1 1
F2 = Τ + C (1.185)
σ2 σ2
2
1
F66 = F (1.186)
τ12
1
F11 = − (1.187)
σ σ1C
Τ
1
1
F22 = − (1.188)
σ Τ2 σ C2
In Figure 1.11 the bounding surface including the allowable values of σ1 , σ 2 , τ12
in σ1 - σ 2 - τ12 Cartesian system is schematically represented, while Figure 1.12
shows the corresponding area of σ1 , σ 2 values for the case where τ12 = 0. In both
figures it can be shown that the corresponding 3-D and 2-D ellipsoids are very
long (in the direction of the σ1 axis), and very slender (in the direction of the σ2
and/or τ12 axis), indicating dependency on the high strength of the fibers.
Figure 1.12 Schematic 2-D graphical representation of Tsai–Wu criterion for the case of
τ12 = 0.
42 Mechanical Behavior of Fiber Reinforced Composite Materials
Appendix I
εx S 11 S 12 S 13 S 14 S 15 S 16 σ x
ε σ
y S 21 S 22 S 23 S 24 S 25 S 26 y
εz S 31 S 32 S 33 S 34 S 35 S 36
σ z
=
2 γ yz S 41 S 42 S 43 S 44 S 45 S 46 σ yz
2 γ σ
xz S 51 S 52 S 53 S 54 S 55 S 56 xz
2 γ xy S 61 S 62 S 63 S 64 S 65 S 66 σ xy
Compliance coefficients:
S 11 = S11 cos 4 θ − 2 S16 cos3 θ sin θ + ( 2 S12 + S66 ) cos 2 θ sin 2 θ − 2 S 26 cos θ sin 3 θ
+ S 22 sin 4 θ
S 12 = S12 cos 4 θ + ( S16 − S 26 ) cos3 θ sin θ + ( S11 + S 22 − S66 ) cos 2 θ sin 2 θ
+ ( S 26 − S16 ) cos θ sin 3 θ + S12 sin 4 θ
S 13 = S13 cos 2 θ − S36 cos θ sin θ + S 23 sin 2 θ
S 16 = S16 cos 4 θ + ( 2 S11 − 2 S12 − S66 ) cos3 θ sin θ + 3 ( S 26 − S16 ) cos 2 θ sin 2 θ
+ ( S66 − 2 S12 − 2 S 22 ) cos θ sin 3 θ − S 26 sin 4 θ
S 22 = S 22 cos 4 θ + 2 S 26 cos3 θ sin θ + ( 2 S12 + S66 ) cos 2 θ sin 2 θ
+ 2 S16 cos θ sin θ3 + S11 sin 4 θ
S 23 = S 23 cos 2 θ + S36 cos θ sin θ + S13 sin 2 θ
S 26 = S 26 cos 4 θ + ( 2 S12 − 2 S 22 + S66 ) cos3 θ sin θ + 3 ( S16 − S 26 ) cos 2 θ sin 2 θ
+ ( 2 S11 − 2 S12 + S66 ) cos θ sin 3 θ − S16 sin 4 θ
S 33 = S33
(
S 36 = 2 ( S13 − S 23 ) cos θ sin θ + S36 cos 2 θ − sin 2 θ )
( ) (
S 66 = S66 cos 2 θ − sin 2 θ + 4 ( S16 − S 26 ) cos 2 θ − sin 2 θ cos θ sin θ )
2
References
[1] Hyer M., Stress analysis of fiber reinforced composite materials, DEStech
Publications, 2009.
[2] Reddy J.N., Mechanics of laminated composite plates and shells, CRC
Press, 2004.
[3] Timoshenko S.P., and Goodier J.N., Theory of elasticity, McGraw-Hill,
1970.
[4] Tsai S.W., and Wu E.M., “A general theory of strength for anisotropic ma-
terials,” Journal of Composite Materials 5 (1971), pp. 58–80.
Chapter 2
45
46 Production Technology of FRP Materials
2.1.1 Thermosets
Due to their valuable properties, such as retention of mechanical behavior in
hot and moist conditions, good chemical resistance, good dimensional stability,
low processing temperatures, excellent fiber impregnation, low melt viscosity,
low cost etc., thermosets have become the most common type of resin used for
matrix fabrication for composites. Thermosets include the following types of
resins: (a) polyester resins, (b) vinyl ester resins, (c) bisphenol fumerate resins,
(d) liquid epoxy resins, (e) solid epoxy resins, (f) polyurethane resins, (g) furane
resins, (h) phenolic resins and (i) chlorenic resins. The main characteristics [1]
of the above types of resins used for thermosetting matrices are summarized in
Table 2.1, while in Table 2.2 typical values of the physical and mechanical proper-
ties of resins used in the filament winding process are presented.
Table 2.1
Main characteristics of thermosetting resins.
Types of resins used for Main characteristics
thermosets
Polyester resins (a) Orthophthalic polyester resins: Suitable for applications
requiring limited resistance to temperature or chemicals (e.g.,
water industry). It is the cheapest thermosetting resin.
(b) Isophthalic polyester and terephthalic resins: They offer
better all-around performance and molding properties com-
pared to orthophthalic resins.
Vinyl ester resins Suitable for mild to severe chemical applications. Offer very
high mechanical properties allied to an excellent corrosion
resistance.
Bisphenol fumerate resins Provide excellent corrosion resistance to strong acids and the
highest resistance to alkalis at elevated temperatures.
Liquid epoxy resins Suitable for manufacture of adhesives, laminates and coat-
ings. Widely used for manufacture of high-performance
pipe systems and pressure vessels for water treatment and
handling.
Solid epoxy resins Suitable for paints and powder coatings.
Polyurethane resins Offer high impact resistance, excellent bonding properties,
very high toughness and abrasion resistance. It is suitable for
coatings.
Furane resins Offer excellent resistance to alkaline solutions and acids
containing chlorinated solvents. However, furanes provide
poor mechanical properties.
Phenolic resins Provide excellent resistance to heat. Almost non-flammable,
phenolics also generate very limited smoke or toxic fumes.
Chlorenic resins Deliver reliable behavior in high operating temperatures and
highly oxidizing environments.
The Composite Matrix Material 47
Table 2.2
Typical properties of thermosetting resins used in filament winding process [1].
Thermosetting Specific gravity Tensile strength Tensile Flexural
resin (MPa) elongation (%) modulus (GPa)
Orthophthalic 1.10 70 2.5 3.8
polyester
Terephthalic 1.10 70 2.5 3.8
polyester
Isophthalic 1.10 70 2.5 3.9
polyester
Vinyl ester 1.11 80 5.0 3.2
Furane 1.10 36 1.0 4.1
Phenolic 1.25 40 2.0 4.0
Chlorenic 1.18 56 2.0 3.8
2.1.2 Thermoplastics
Thermoplastic resins can be amorphous or semi-crystalline. The semi-
crystalline type contains both amorphous and crystalline phases. Since thermo-
plastics are not cross-linked, they can be processed quickly, thereby reducing
their manufacturing cost. From a mechanical point of view, thermoplastics can
display and undergo large deformations before final fracture, thus offering higher
toughness compared to thermosets. Moreover, under graduated loading condi-
tions their deformation is time-dependent, due to creep. Since today new ther-
moplastics are obtained by modification or mixing of existing polymers, the full
range of these materials is extremely wide. The characteristics of a few main types
of thermoplastics, their abbreviations, and selected trade names are summarized
in Table 2.3, while Table 2.4 provides the mechanical and physical properties of
widely used thermoplastic polymers [2].
Table 2.3
Main characteristics of thermoplastics used for composite materials.
Type of thermoplastics Abbreviation Main characteristics
Poly-Benzimidazoles PBIs Offers good stability after aging. They are well
suited for use at temperatures up to 250°C but at
higher temperatures oxidate degradation occurs
accompanied with strength reduction.
Poly-Phenylene Sulfide PPS Offers excellent chemical and thermal stability.
Poly-Ether PEEK Excellent thermoplastics for engineering applica-
Ether Ketone tions due to good mechanical behavior. However,
their use for high-performance composites is limited
due to inadequate chemical resistance and adhesion.
48 Production Technology of FRP Materials
Table 2.4
Physical and mechanical properties of widely used thermoplastic polymers.
Thermoplastic Density Tensile Tensile Tensile
Polymers (Kg/m3) strength (MPa) Elongation (%) Modulus (GPa)
Poly-Phenylene 1340 70–75 3 3.3
Sulfide(PPS)
Poly-Ether-Ether- 1320 92–100 150 —
Ketone(PEEK)
Poly-Sulfone(PS) 1240 70–75 50–100 2.5
Poly-Propylene(PP) 900 25–38 300 1.0–1.4
Nylon 6,6 (NYLON) 1140 60–75 40–80 1.4–2.8
Poly-Carbonate (PC) 1060–1200 45–70 50–10 2.2–2.4
Poly-Ether Imide 1270 105 60 3
(Ultem)
Poly-Amide Imide 1400 95–185 12–18 5
(Torlon)
(i) A-Glass: It is a soda-lime glass, which was the first used and is still retained
for minor applications.
(ii) E-Glass: E-glass is a borosilicate glass, which exhibits very good corro-
sion resistance and is suitable for operating in water and mild chemical
environments.
(iii) C-Glass: C-glass is an improvement of E-Glass, providing better durability
when exposed to acids and alkalis.
(iv) S-Glass: S-glass exhibits increased strength and stiffness and is suitable for
high-performance applications.
Physical and mechanical properties of the main types of glass used for fiber
fabrication are presented in Table 2.5 [2].
Fiber Materials 49
Table 2.5
Properties of the main glass fibers.
Property E-Glass C-Glass S-Glass
Tensile strength (MPa) 3450 3160 4590
Tensile modulus (GPa) 72.4 68.9 85.5
Elongation (%) 1.8–3.2 4.8 5.7
3
Density (Kg/m ) 2541 2492 2492
Diameter (μm) 8–13 — 10
Table 2.6
Values of physical and mechanical properties of carbon fibers.
Property PAN Pitch Type-P
Table 2.7
Values of physical and mechanical properties of polymeric fibers.
Property Kevlar-29 Kevlar-49 Spectra 900
(Polyethylene)
Tensile strength 2760 2800–3792 2580
(MPa)
Tensile modulus 62 131 117
(GPa)
Elongation (%) 3–4 2.2–2.8 4–5
3
Density (Kg/ m ) 1440 1479 970
Diameter (μm) 12 12 38
Production Technologies for FRP Composite Pipes 51
This winding pattern is called helical and it is suitable for pipe production. The
main parameter controlling the mechanical behavior of the produced pipe is the
angle of the fibers with respect to the longitudinal direction of the pipe (winding
angle). It is obvious that winding angle θ = 90ο (hoop winding pattern) improves
the resistance in long pipes subjected to internal pressure, while winding angles
θ = 0 – 15o yield high-resistant pipes for axial tension or bending loading cases.
The prescribed winding angle can be achieved by controlling the rotational speed
of the mandrel and the longitudinal speed of the head that dispenses the tows
(payout head). Increasing the longitudinal speed of the payout head decreases the
winding angle. A hoop winding can be achieved by advancing the payout head
slowly along the mandrel axis so that the fiber tows are wound transversely to
the longitudinal axis. Generally, since the fibers tend to spread into bands due to
tension, a helical winding pattern does not put the tows in order on the mandrel’s
surface. Therefore, several circuits are used before the full surface of the mandrel
is covered.
Apart from the helical pattern, a polar one can be used for axisymmetric com-
posite shells. This pattern results when the mandrel does not rotate but the payout
head rotates about the longitudinal axis. This pattern is mainly used for axisym-
metric pressure vessels.
• The resin should not stick to the surface of the mandrel. Release agents need
to be applied.
• The mandrel must be extractable from the part after curing.
The separation of the final product from the mandrel is achieved by using ex-
tractable, collapsible, breakable or dissolvable mandrels.
References
[1] Derich Scott, Advanced materials for water handling: Composites and
Thermoplastics, Elsevier, 2000.
[2] Hyer M., Stress analysis of fiber reinforced composite materials, DEStech
Publications, 2009.
[3] Hoa S.V., Principles of the manufacturing of composite materials, DEStech
Publications, 2009.
53
54 Mechanical Design of Composite Pipelines
(a)
(b)
(c)
Figure 3.3 (a) Mid-depth tow, (b) Off-bottom tow, (c) Surface tow.
pressure and axial tension, vii) torsion (e.g., during the wave movement of the
installation ship), viii) combination of torsion, bending, axial tension and external
pressure. The loading cases above cause plane stress in the wall of composite
pipes. Therefore the stress state for every lamina has to be analyzed by a failure
criterion. Moreover, external pressure, bending, torsion and their combinations
can cause local buckling into the wall of the composite pipe. The loading cases
cited above are summarized in Table 3.1.
Table 3.1
Classification of loads acting on composite pipelines.
External pressure + +
Combination of bending and axial tension + +
Combination of external pressure and axial + +
tension
Combination of bending, external pressure and + +
axial tension
Torsion + +
Combination of torsion, bending axial tension + +
and external pressure
Constant internal fluid pressure + +
Fluctuation internal fluid pressure + +
Hydrodynamic forces due to internal axial flow + +
bending. In most bending cases, the determination of the allowable bending mo-
ment or the minimum radius of curvature of the deformed pipeline is usually
the main target of bending analysis. Since the plastic deformation of composite
materials is almost absent in their stress-strain behavior (compared to steel), the
stress and deflection analysis of composite pipelines will be based only on elastic-
ity equations.
58 Mechanical Design of Composite Pipelines
After the selection of the material, the data 1 (a)-(h) can easily be found in
existing data bases (e.g. Table 3.2) containing material properties. The input data
(2)-(4) are design parameters to be chosen by the designer taking into account,
e.g., manufacturing cost parameters, the fluid supply/demand scenario, the toler-
ance of the pipeline to further loading situations, etc.
Table 3.2
Material properties of widely used composites [4].
E-Glass/Epoxy S-Glass/Epoxy AS/3501 T300/5208
Carbon/Epoxy Carbon/Epoxy
E1 (GPa) 39 43 138 181
C
σ r = C1r n −1 − C2 r − n −1 + 3 + Agr sin ϕ (3.1)
r
C
σ ϕ = C1 ( n + 1) r n −1 + C2 ( n − 1) r − n −1 + 3 + 3 Agr sin ϕ (3.2)
r
1
σ z = Ar sin ϕ −
S33
( S13 σ r + S23 σ ϕ ) (3.3)
or
1
σz = − ( S13 + S 23 + nS 23 ) r n −1C1 + ( S13 + S 23 − nS 23 ) r − n −1C2 −
S33 (3.4)
( S13 + S23 ) r C3 + ( S33 − S13 g − 3S 23 g ) rA sin ϕ
−1
where [6]:
−1
Sij = Qij a ij Pij (3.6)
1 0 0 0 0 0
0 cos 2 θ sin 2 θ 0 cos θ sin θ 0
0 sin 2 θ cos 2 θ 0 cos θ sin θ 0
Qij =
0 0 0 cos θ 0 − sin θ
0 2 cos θ sin θ −2 cos θ sin θ 0 cos 2 θ − sin 2 θ 0
0 0 0 sin θ 0 cos θ
(3.7)
60 Mechanical Design of Composite Pipelines
1 0 0 0 0 0
0 cos 2
ϑ sin 2 ϑ 0 −2 cos ϑ sin ϑ 0
0 2
sin ϑ cos 2 ϑ 0 2 cos ϑ sin ϑ 0
[ Pij ] = (3.8)
0 0 0 cos ϑ 0 − sin ϑ
0 cos ϑ sin ϑ − cos ϑ sin ϑ 0 cos ϑ − sin 2 ϑ
2
0
0 0 0 sin ϑ 0 cos ϑ
1 ν 32 ν 21
E − − 0 0 0
E2 E1
2
ν 32 1 ν
− − 21 0 0 0
E2 E2 E1
ν 21 ν 21 1
− − 0 0 0
E E1 E1
[aij ] = 1
(3.9)
2(1 −ν 32 )
0 0 0 0 0
E2
1
0 0 0 0 0
G12
0 1
0 0 0 0
G12
S 23 − S13
g= (3.10)
β11 + 2 β12 + β 66 − 3β 22
β11 + 2 β12 + β 66
n = 1+ (3.11)
β 22
Si 3 S j 3
β ij = Sij − (3.12)
S33
According to [e.g., 7], the strains and stresses in directions z , ϕ , r (Figures 3.5
and 3.6) are correlated with the following relationships:
− − −
ε z = S 11 σ z + S 12 σ ϕ + S 13 σ r (3.13)
− − −
ε ϕ = S 21 σ z + S 22 σ ϕ + S 23 σ r (3.14)
− − −
ε r = S 31 σ z + S 32 σ ϕ + S 33 σ r (3.15)
Pure Bending 61
1−
γ rϕ = S 44 τ rϕ (3.16)
2
−
where S ij are given in the Appendix to Chapter 1 [7].
− 1
ε z = S 11 − ( S13 + S 23 + nS 23 ) r n −1C1 + ( S13 + S 23 − nS 23 ) r − n −1C2 −
S 33
{r }
−
S 13 n −1
C1 − r − n −1C2 + r −1C3 + grA sin ϕ
or
− 1
ε z = sin ϕ S 11 − ( S13 + S 23 + nS 23 ) r n −1C1 + ( S13 + S 23 − nS 23 ) r − n −1C2 −
S33
( S13 + S23 ) r −1C3 + ( S33 − S13 g − 3S23 g ) rA +
−
(3.18)
S 12 ( n + 1) r C1 + ( n − 1) r
n −1 − n −1
C2 + r C3 + 3 grA +
−1
−
S 13 r n −1C1 − r n −1C2 + r −1C3 + grA }
− 1
ε ϕ = sin ϕ S 21 − ( S13 + S 23 + nS 23 ) r n −1C1 + ( S13 + S 23 − nS 23 ) r − n −1C2 −
S 33
1−
γ rϕ = − S 44 cos ϕ r n −1C1 − r − n −1C2 + r −1C3 + grA (3.21)
2
∂ur
εr = (3.22)
∂r
ur 1 ∂uϕ
εϕ = + (3.23)
r r ∂ϕ
∂u z
εz = (3.24)
∂z
1 ∂ur ∂uϕ uϕ
γ rϕ = + − (3.25)
r ∂ϕ ∂r r
ur = ∫ ε r dr + F (ϕ ) (3.26)
or
1 − −
− − −
ur = r − n 2C2 − S13 S 31 + ( −1 + n ) S 23 S 31 + S33 S 32 − n S 32 + S 33 −
2nS33
− −
− − −
2C1r 2 n S13 S 31 + (1 + n ) S 23 S 31 − S33 S 32 + n S 32 + S 33 +
−
− − −
nr n Ar 2 S33 S 31 + g − S 31 ( S13 + 3S 23 ) + S33 3 S 32 + S 33 +
− − −
+2C3 − S 31 ( S13 + S 23 ) + S33 S 32 + S 33 ln ( r ) sin ϕ +
+ F (ϕ )
(3.27)
According to [5] the displacements should be single-valued functions.
Therefore:
C3 = 0 (3.28)
64 Mechanical Design of Composite Pipelines
r − n sin ϕ − −
− − −
ur = 2C2 − S13 S 31 + ( n − 1) S 23 S 31 + S33 S 32 − n S 32 + S 33 −
2nS33
− −
− − −
2C1r 2 n S13 S 31 + ( n + 1) S 23 S 31 − S33 S 32 + n S 32 + S 33 +
−
−
−
−
Anr n + 2 S33 S 31 + g − S 31 ( S13 + 3S 23 ) + S33 3 S 32 + S 33 +
+ F (ϕ )
(3.29)
uϕ = G (r ) + ∫ F ( ϕ )d ϕ +
r − n cos ϕ − −
−
2C1r 2 n n 2 S 23 S 21 − S33 S 22 − S 31 ( S13 + S 23 ) +
2nS33
− −
− − −
n S 21 ( S13 + S 23 ) − S 23 S 31 − S33 S 22 + S 23 + S 32 +
− − − − − −
Anr n+ 2 S33 S 33 2 S+21S33+g +( S13 + 3S 23 ) 2 S 21 − S 31 +
S 31 −S 32
− −6 S 22 − 2 S 23 + 3 S 32 + S 33
− − − − − −
2C2 n 2 S 23 S 21 − S33 S 22 S−33 S 31 ( 13 + S 23 ) +
− −
− −
− (3.30)
− S 21 ( Sof
where G(r) is an unknownn function r. S 23 ) + S 23 S 31 + S33 S 22 + S 23 − S 32 +
13 +
Taking into account the condition of symmetry
− −
S33 S 32 + S 33 +
uϕ (ϕ ) = −uϕ (π − ϕ ) (3.31)
G ( r ) + ∫ F (ϕ ) d ϕ = 0 (3.32)
Let
∫ F (ϕ ) d ϕ = Ε (ϕ ) (3.33)
Pure Bending 65
G ( r ) + E (ϕ ) = 0 (3.34)
yielding
G (r ) = 0 (3.35)
and
E (ϕ ) = 0 (3.36)
or
F (ϕ ) = 0 (3.37)
With the aid of eqs. (3.28), (3.35), (3.37) and using the following notations
r
−n
−
λ1 = −2 S13 S 31 + ( n + 1) S 23 S 31 − S33 S 32 + n S 32 + S 33 r 2 n
− − − −
(3.38)
2nS33
r
−n
λ2 = 2 − S13 S 31 + ( n − 1) S23 S 31 + S33 S 32 − n S 32 + S 33
− − − − −
(3.39)
2 nS 33
r
−n
−
λ3 = nr n + 2 S33 S 31 + g − S 31 ( S13 + 3S 23 ) + S33 3 S 32 + S 33
− − −
(3.40)
2nS33
λ4 = 2r 2 n n 2 S 23 S 21 − S33 S 22 − S 31 ( S13 + S 23 ) +
− − −
− −
− − −
n S 21 ( S13 + S 23 ) − S 23 S 31 − S33 S 22 + S 23 + S 32 + (3.41)
r
− n
− −
S33 S 32 + S 33
2nS33
66 Mechanical Design of Composite Pipelines
λ5 = 2 n 2 S23 S 21 − S33 S 22 − S 31 ( S13 + S 23 ) +
− − −
− −
− − −
n − S 21 ( S13 + S 23 ) + S 23 S 31 + S33 S 22 + S 23 − S 32 + (3.42)
r
− n
− −
S33 S 32 + S 33
2nS33
λ6 = nr n + 2 S33 S 31 − 2 S 21 + g ( S13 + 3S23 ) 2 S 21 − S 31 +
− − − −
r
−n
− − − −
S33 −6 S 22 − 2 S 23 + 3 S 32 + S 33
2 nS 33
(3.43)
the expressions (3.29), (3.30) can be written:
With the aid of eq. (3.28) and using the following notations
µ1 = r n −1 (3.46)
µ2 = −r − n −1 (3.47)
µ3 = gr (3.48)
µ4 = ( n + 1) r n −1 (3.49)
µ5 = ( n − 1) r − n −1 (3.50)
µ6 = − ( S13 + S 23 + nS 23 ) r n −1 (3.51)
µ7 = ( S13 + S 23 − nS 23 ) r − n −1 (3.52)
sin ϕ
σz = ( µ6C1 + µ7 C2 + µ8 A) (3.56)
S33
τ r1ϕ ( r2 ) = τ r2ϕ ( r2 )
τ ( r3 ) = τ ( r3 )
2
rϕ
3
rϕ
( N − 1) conditions (3.58)
τ rϕ ( rk +1 ) = τ rϕ ( rk +1 )
k k +1
τ rϕ ( rN ) = τ rϕ ( rN )
Ν −1 Ν
68 Mechanical Design of Composite Pipelines
u1r ( r2 ) = ur2 ( r2 )
u ( r3 ) = u ( r3 )
2
r
3
3
.....................
......................
.....................
k ( N − 1) conditions (3.59)
ur ( rk +1 ) = ur ( rk +1 )
k +1
.....................
.....................
......................
urN −1 ( rN ) = urN ( rN )
Pure Bending 69
uϕ1 ( r2 ) = uϕ2 ( r2 )
u ( r3 ) = u ( r3 )
2
ϕ
3
ϕ
.....................
......................
.....................
k ( N − 1) conditions (3.60)
uϕ ( rk +1 ) = uϕk +1 ( rk +1 )
.....................
.....................
......................
uϕN −1 ( rN ) = uϕN ( rN )
π rk +1
Μκ = 2∫ ∫ σ ( r,ϕ ) r
k 2
z sin ϕ dr dϕ .
0 rk
With the aid of the last equation the equilibrium between the bending moment M
acting at the ends of pipe is equal to the summation of the bending moments due
to N layers, i.e.
Ν
Μ = ∑ Μκ
κ =1
or
Ν π rk +1
Μ = 2∑ ∫ ∫ σ ( r,ϕ ) r
k
z
2
sin ϕ dr dϕ (3.62)
κ =1 0 rk
70 Mechanical Design of Composite Pipelines
With the aid of eq. (3.56) and taking into account eqs (3.51)-(3.53), the above
equation can be written as:
Ν
(
Μ = π∑ I 6 C1k + I 7 C2k + I 8 Ak ) (3.63)
k =1
where
rk +1
I6 = ∫
rk
µ6 r 2 dr (3.64)
rk +1
I7 = ∫
rk
µ7 r 2 dr (3.65)
rk +1
I8 = ∫
rk
µ8 r 2 dr (3.66)
S13 + S 23 + nS 23 n + 2 n + 2
I6 = − (r − r )
( n + 2 ) S33 k +1 k
(3.67)
S13 + S 23 − nS 23 − n + 2 − n + 2
I7 = (r − r )
( −n + 2 ) S33 k +1 k
(3.68)
S33 − S13 g − 3S 23 g 4
I8 =
4 S33
( rk +1 − rk4 ) (3.69)
For N-layered pipe the number of unknowns is 3N . On the other hand, the num-
ber of available conditions shown in eqs (3.58)–(3.62) is 3 ( N − 1) + 2 + 1 = 3 N .
Taking into account eq. (3.57) the conditions (3.58) for a multilayered pipe
with stacking sequence ( ±θ ) NP can be written:
Pure Bending 71
1 N
(3.70)
With the aid of eq. (3.44) the conditions (3.59) for a multi-layered pipe with stack-
ing sequence ( ±θ ) NP yield:
(3.71)
Similarly, with the aid of eq. (3.45), the conditions (3.60) can be written:
(3.72)
72 Mechanical Design of Composite Pipelines
The above equations (3.70)-(3.72) regarding the interfaces of the layers have to be
completed with the boundary conditions on exterior surfaces of the multi-layered
pipe.
Taking into account the eq. (3.57), the conditions (3.61) yield:
The equations (3.70)-(3.73), as well as the eq. (3.63), can be written in the follow-
ing matrix form:
[ M 1 ] [ M 2 ] [ M 3 ]
{C }
[ Λ1 ] [ Λ 2 ] [ Λ 3 ] 1
[ Λ 4 ] [ Λ 5 ] [ Λ 6 ] {C2 } = {L} (3.74)
[ 1 ] [ 2 ] [ 3 ] { A}
Β Β Β
[J ] [J ] [J ]
6 7 8
where:
{C1} = {C11 , C12 , C13 , , C1k , , C1N }
T
(3.75)
{ A} = { A1 , A2 , A3 , , Ak , , AN }
T
(3.77)
µi (2,1) − µi (2, 2) 0 0 0 0 0 0 0 0
0 µi (3, 2) − µi (3, 3) 0 0 0 0 0 0 0
0 0 µi (4, 3) − µi (4, 4) 0 0 0 0 0 0
[ M i ]( N −1) xN = 0 0 0 0 0 0 0 0 0
0 0 0 0 0 µi (k , k − 1) − µi (k , k ) 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 µ ( N , N − 1) − µ ( N , N )
i i
(3.78)
(3.79)
Pure Bending 73
µi (1,1) 0 0 0 0 0 0 0 0 0
[ Bi ]2 xN =
µ ( N , N − 1) 0 0 0 0 0 0 0 0 0
(3.80)
i
{ J i }1xN = {I i1 I i2 I i3 I ik −1 I ik I ik +1 I iN −1 I iN } (3.81)
T
M
{ L}3 Nx1 = 0 0 0 0 0 0 0 0 0 (3.82)
π
The solution of the matrix equation (3.74) yields the unknown constants C1k , C2k ,
Ak ( k = 1, 2, , NP ) for all layers. Therefore, using the values C1 , C2 , ANP
NP NP
of the critical exterior layer ( k = NP ) , its stress state can now be determined
by the eqs. (3.54)-(3.57). In that case, the determination of the principal stresses
σ 1 , σ 2 ,τ 12 of the exterior layer is possible by using eq. (1.70). Finally, the allow-
able bending moment can now be obtained by using a failure criterion (e.g. [8]).
Figure 3.8 (a) Buckling of narrow strip AB due to bending, (b) Buckling of narrow strip
AB due to axial load.
∧
We will approximate the critical value of M y for the case of Fig. 3.8a, taking
o
into account the critical strain ε ξ c derived from the model shown in Fig. 3.8b.
Because the maximum compressive stress causing buckling to the model of
Fig. 3.8b does act on the whole perimeter of the pipe’s cross-section, instead of
acting only on point A (as in the case of Fig. 3.8a), the practical assumption above
is expected to be conservative, and therefore, safe for designing purposes. This
assumption has been checked in [9] for the case of isotropic thin-walled tubes. In
that work it is seen that for pure bending the exact solution for the critical com-
pressive stress gives a value about 30% higher than that obtained by the above
assumption [10]. In regard to Figure 3.8b, the critical longitudinal strain at the
buckling state is:
N c = λcr N o (3.84)
Pure Bending 75
O L M 0 M n O L
T
Φ ⋅ [ J ] Φ 2 ⋅ [ J ]
det −λ 1 = 0
L O M n
(3.85)
M 0 L O Φ 2 ⋅ [ J ] Φ1 ⋅ [ J ]
The above equation, as well as the matrices [L], [O], [J] and the parameters Φ1,
Φ2, are given in ref. [11], i.e.:
−α 0 β 0 0 0
0
[ ]
O = −β α 0 0 0 (3.86)
0 2/ D 0 a + β 2 c22
2
β + a 2 c12
2
−2a β (1 + c1c2 )
β c2 0 −ac1 0 0 0
[ ] 0
L = ac1 − β c2 0 0 0
(3.87)
0 0 0 −2a β c2 −2a β c1 2(c1a 2 + c2 β 2 )
Α Β
[Μ n ] = − [M o ] (3.89)
Β D
0 0 0
[ J ] = 0 0 0
(3.90)
0 0 1
Φ1 = a 2 + β 2 c22 (3.91)
Φ 2 = −2a β c2 (3.92)
76 Mechanical Design of Composite Pipelines
In the above matrices the four constants a, b, c1, c2 characterize the wave pat-
tern of the buckling. For different combinations of the above constants, equation
(3.85) will result in different values of λcr. The lowest value of λcr is the appropri-
ate value to be used in equation (3.84).
ο
The local strain ε ξ at any point located on the mean diameter of a cross-
∧
section of a pipe subjected to pure bending by moment M y is [11]:
1
ε ξο = z (3.93)
ρy
where
∧
1 My
= (3.94)
ρ y EI yy
π
1 2 1 D
EI yy = 2 ∫ z + cos 2 θ dθ (3.95)
α
0 11
d11 2
D
z= cos θ (3.96)
2
ο
The local strain ε ξ of the strip AB shown in Fig. 3.8a occurs when θ = 0.
Therefore:
ο
My D
ε =
ξ (3.97)
EI yy 2
Combining eqs. (3.83), (3.84), (3.97), the critical bending moment that causes
local buckling into the strip AB can be approximated by the following equation:
∧ 2 EI yy a11λcr
M yc = (3.98)
D
External Pressure 77
In order to determine the external forces acting on the pipe’s wall, the equilib-
rium equation of the half pipe shown in Figure 3.11 should be used:
2N y L = pDL (3.99)
or
1
Ny = pD (3.100)
2
Because of the absence of external loads in directions x and xy, Ny is the only load
acting on the laminate.
Therefore:
Nx = 0 (3.101)
N xy = 0 (3.102)
Mx = 0 (3.103)
My = 0 (3.104)
M xy = 0 (3.105)
External Pressure 79
Taking into account the above equations as well as the inverse ABD matrix, it can
be written:
ε x0 = a12 N y (3.106)
ε y0 = a22 N y (3.107)
κ x0 = b21 N y (3.109)
κ y0 = b22 N y (3.110)
Therefore:
ε x = ε x0 + zκ x0 (3.112)
ε y = ε y0 + zκ y0 (3.113)
With the aid of equations (3.106)-(3.114), the stress-strain relation provides the
stresses σx, σy, τxy for every ply:
σ x ε x
σ y = Q ij (θ ) ε y (3.115)
τ xy γ xy
Using the matrix [T (θ )] , the principal stresses σ1, σ2, τ12 can be obtained by the
following well-known equation:
σ 1 σ x
σ 2 = [ Τ(θ ) ] σ y (3.116)
τ
12 τ xy
Using eq. (3.116), the Tsai-Wu failure criterion for every ply yields the allowable
value pa of external pressure. Since the Tsai-Wu is a second-order algebraic equa-
tion, we are going to obtain two values of pa for every ply. From the derived dif-
ferent values of pa, the minimum one should be adopted in order to cover all cases.
80 Mechanical Design of Composite Pipelines
2
Aani Dani − Bani
pcr = 3 (3.117)
Aani ( D / 2 ) + 2 Bani ( D / 2 ) + Dani ( D / 2)
3 2
In the above equation, the parameters Aani, Bani, Dani can be obtained by the follow-
ing matrix equation:
where
A12 B12
A B26
[ L1 ] = 26 (3.119)
B12 D12
B26 D26
h
α12 + b21
σ p
2
x
p
h 1
σ = Q (
y ij 22θ ) a + b22 pD (3.121)
p 2 2
τ xy h
a26 + 2 b26
1
σ xΜ S ( µ6 C1 + µ7 C2 + µ8 Α
Μ 33
σ y = µ 4 C1 + µ5C2 + 3µ3 Α (3.122)
Μ 0
τ xy
the following formulas for the principal stresses can be obtained where
(3.125)
It should be noticed that equation (3.122) provides the stresses of the exterior
lamina at the location φ = π/2 (location of the maximum bending stresses). In eq.
(3.121), h denotes the thickness of the lamina, D is the exterior diameter of the
pipe, and p is the external pressure.
82 Mechanical Design of Composite Pipelines
Since the principal stresses σ1, σ2, τ12 are known, the Tsai-Wu failure criterion
can be used for derivation of the diagrams pα = pα(M) providing the allowable
combinations of the external pressure pα and bending moment M.
2 ⋅ EI ⋅ a11
M cr = N x cr (3.126)
D
where
π
D 2 cos 2 θ 1 D
EI = 2 ∫ + cos 2 θ dθ (3.127)
0
4a11 d11 2
D 2 a11
M cr = π + N x cr (3.128)
4 d11
Therefore, the first step for the development of a buckling model for the combina-
tion of pure bending and external pressure is the solution for the case where the
compressive axial load Nx and the lateral pressure q are combined.
(d) the assumption that the moments are very small and we can neglect the
products of moments and derivatives of the displacements u, v, w,
∂N x ∂N xy
R + =0
∂x ∂θ
∂N y ∂N xy ∂ 2 v ∂M xy ∂M y
+R + RN x 2 + − =0 (3.129)
∂θ ∂x ∂x ∂x R∂θ
∂2w ∂2 M x
2 2
∂ M xy ∂ M y
RN x 2 + N y + R 2
−2 + = 0
∂x ∂x ∂x∂θ R∂θ 2
∂N x ∂N xy ∂ 2 v ∂w
R + + qR − =0
∂x ∂θ ∂x∂θ ∂x
∂N y ∂N xy ∂M y ∂M xy
+R − + =0 (3.130)
∂θ ∂x R∂θ ∂x
∂ 2 M xy 2
∂ 2
M ∂ 2
M
∂ Mx y xy 2
∂ w
− +R + − − q w + = 0
∂x∂θ ∂x 2 R∂θ 2 ∂x∂θ ∂θ 2
3.4.2.3 composite pipe under combined lateral external pressure and axial
compressive load
When both lateral external pressure q and axial compression Nx are simulta-
neously acting on a long composite pipe, the combination of eqs. (3.129) and
(3.130) yields:
∂N x ∂N xy ∂ 2 v ∂w
R + = − qR −
∂x ∂θ ∂x∂θ ∂x
∂N y ∂N xy 1 ∂M y ∂M xy ∂2v
+R − + = −Nx R 2 (3.131)
∂θ ∂x R ∂θ ∂x ∂x
2 2
∂2 M x 1 ∂ M y ∂ M xy ∂2 w ∂2 w
R + −2 = +q w + 2 − Nx R 2
∂x 2 R ∂θ 2 ∂ x∂ θ ∂θ ∂x
where [7]:
Combination of Bending and External Pressure 85
∂u
ε x0 =
∂x
∂v
ε y0 =
R∂θ
∂u ∂v
γ xy0 = +
R∂θ ∂x
(3.133)
∂2 w
k x0 = 2
∂x
2
1 ∂ w
k y0 = 2
R ∂θ 2
2 ∂2 w
k xy0 =
R ∂x∂θ
With the aid of eq. (3.133), eq. (3.132) can be written:
∂u A12 ∂v ∂2 w B ∂2 w
N x = A11 + + B11 2 + 122
∂x R ∂θ ∂x R ∂θ 2
∂u A22 ∂v ∂ w B ∂ w
2 2
N y = A12 + + B12 2 + 222
∂x R ∂θ ∂x R ∂θ 2
A ∂u ∂v 2 B66 ∂ 2 w
N xy = 66 + A66 +
R ∂θ ∂x R ∂x∂θ
(3.134)
∂u B12 ∂v ∂ 2 w D12 ∂ 2 w
M x = B11 + + D11 2 + 2
∂x R ∂θ ∂x R ∂θ 2
∂u B22 ∂v ∂2 w D ∂2 w
M y = B21 + + D12 2 + 222
∂x R ∂θ ∂x R ∂θ 2
B ∂u ∂v 2 D66 ∂ 2 w
M xy = 66 + B66 +
R ∂θ ∂x R ∂x∂θ
Taking into account the above equation (3.134), equation (3.131) yields:
∂ 2u A ∂ 2 v ∂ 3 w B ∂ 3 w A66 ∂ 2 u ∂v 2B ∂3 w
R A11 2 + 12 + B11 3 + 122 2
+ + A66 + 66 =
∂x R ∂x∂θ ∂x R ∂x∂θ R ∂θ 2
∂θ∂x R ∂x∂θ 2
∂ 2 v ∂w
= −qR − =0
∂x∂θ ∂x
(3.135)
86 Mechanical Design of Composite Pipelines
∂ 2 u A22 ∂ 2 v ∂ 3 w B22 ∂ 3 w A ∂ 2u ∂ 2v 2B ∂3 w
A12 + 2
+ B11 2
+ 2 3
+ R 66 + A66 2 + 66 2 −
∂θ∂x R ∂θ ∂θ∂x R ∂θ R ∂x∂θ ∂x R ∂x ∂θ
1 ∂u B ∂ v 2
∂ w D22 ∂ w B66 ∂ u
3 3 2
∂ v 2 D66 ∂ 3 w
2
− B12 + 22 2 + D12 + +
+ B66 + =
R ∂θ∂x R ∂θ ∂θ∂x 2 R ∂θ3 R ∂x∂θ ∂x 2 R ∂x 2 ∂θ
∂2v
= −Nx R 2 (3.136)
∂x
(3.137)
Assuming that the origin of the coordinates is located at the one end of the pipe
(see Fig. 3.13), the following general solutions for equations (3.135)-(3.137) will
be used:
(mπ x)
u = A ⋅ sin(nθ ) ⋅ cos
L
(mπ x)
v = B ⋅ cos( nθ ) ⋅ sin (3.138)
L
(mπ x)
w = C ⋅ sin(nθ ) ⋅ sin
L
where A, B, C, are unknown constants. The above equation assumes that during
buckling, the generators of the shell subdivide into m half-waves and the circumfer-
ence into 2n half-waves. Combining equations (3.138) and (3.135)–(3.137) yields:
The equation for calculating the critical value of the pairs of external pressure qcr
and axial compression N xcr has the following form:
In Figure 3.14 the curves N xcr vs qcr for several pairs (m,n) are schematically dis-
played. However, according to eq. (3.128), N x cr represents the quantity
( )
−1
π D / 4 + a11 / d11 M cr
2
Figure 3.14 Curves N xcr vs qcr for several pairs (m, n).
∧
N
Nx = (3.150)
πD
This is the only external load acting on the laminate constituting the wall of the
pipe and thus
Ny = 0 (3.151)
N xy = 0 (3.152)
Mx = 0 (3.153)
Axial Tension 89
My = 0 (3.154)
M xy = 0 (3.155)
∧
In order to estimate the allowable force N , a failure criterion, e.g., the Tsai-Wu
criterion, should be applied. To this scope, the principal stresses σ1, σ2, τ12 have to
be determined by following the general procedure shown in Fig. 3.16.
Taking into account this procedure, as well as equations (3.150)–(3.155) and
the formula for the inverse ABD matrix, one can formulate the following:
ε x0 = a11 N x (3.156)
ε y0 = a12 N x (3.157)
k x0 = b11 N x (3.159)
90 Mechanical Design of Composite Pipelines
İ [
1[1\1[\
$%' H F H \J [\
.LUFKKRII
İ \
0[0\0[\ 0DWUL[
N N N K\SRWKHVLV
[ \ [\
Ȗ[\
$OORZDEOHD[LDO 6WUHVV
)DLOXUH
FULWHULRQ ııIJ >7@ ı[ı\IJ[\ VWUDLQ
IRUFH 1 UHODWLRQ
∧
figure 3.16 Concept for estimation of allowable axial force N .
k y0 = b12 N x (3.160)
ε y = ε yo + zk y0 (3.163)
Thus, the stresses σx, σy, τxy for each lamina with fibers orientation θ can be ob-
tained by the well-known stress-strain relation:
σ y = Q 21 (θ ) ⋅ ε x + Q 22 (θ ) ⋅ ε y + Q 26 (θ ) ⋅ γ xy (3.166)
τ xy = Q16 (θ ) ⋅ ε x + Q 26 (θ ) ⋅ ε y + Q 66 (θ ) ⋅ γ xy (3.167)
With the aid of matrix [T(θ)], the principal stresses σ1, σ2, τ12 for every lamina can
be determined by the following matrix equation:
σ 1 σ x
σ 2 = [ Τ(θ ) ] σ y (3.168)
τ
12 τ xy
Combination of Bending and Axial Tension 91
By applying the Tsai-Wu failure criterion for the values σ1, σ2, τ12 of every lamina,
where:
1 1
F1 = ( + ) (3.170)
s 1Τ s 1C
1 1
F2 = ( + ) (3.171)
s 2Τ s 2C
1
F11 = − (3.172)
s 1Τs 1c
1
F2 = − (3.173)
s 2Τs 2c
1
F66 = ( F ) 2 (3.174)
τ 12
∧
the allowable axial tension Ν will be obtained. Since this procedure yields differ-
∧
ent values of Ν , we have to adopt the minimum one.
h
σ xN α11 + 2 b11
N �
h N X
σ y = Q ij (θ ) a12 + b12 (3.175)
N 2 π D
τ xy h
a16 + 2 b16
1
σ xΜ S ( µ6 C1 + µ7 C2 + µ8 Α
Μ 33
σ y = µ 4 C1 + µ5C2 + 3µ3 Α (3.176)
Μ 0
τ xy
Indicate that the following formulas for the principal stresses can be obtained:
τ 12 = (σ xΜ + σ xN )(− cos θ sin θ ) + (σ yΜ + σ yN ) cos θ sin θ + (τ xyΜ + τ xyN )(cos 2 θ − sin 2 θ )
(3.179)
It should be noticed that equation (3.176) provides the stresses of the exterior
lamina at the location φ = π/2 (location of the maximum bending stresses). In
equation (3.175), h denotes the thickness of the lamina, D is the exterior diameter
of the pipe, and N is the axial force.
Since the principal stresses σ1, σ2, τ12 are known, the Tsai-Wu failure criterion
can be used for derivation of the diagrams Nα = Να(M), providing the allowable
combinations of the axial force Να and bending moment M.
Figure 3.17 (a) Buckling of the narrow strip AB due to combined bending and tension,
(b) Buckling of the narrow strip AB due to axial load.
ο
M y D α11 ∧
ε =
ξ − Nx (3.180)
EI yy 2 π D
where [11]:
π
1 D2 1 D
EI yy = 2 ∫ cos 2 a + cos 2 a dθ (3.181)
α 4
0 11
d11 2
∧
In case ε ξο ≤ 0 , the tensile strain due to N x is greater than the compressive stain
∧
due to M y ; therefore, the check for local buckling is not needed. On the other
hand, we recall that the critical value of ε ξο , namely ε ξοc , is:
Therefore, combining equations (3.180) and (3.182), the boundaries of the allow-
∧ ∧
a
able values of M y and N xa in order to avoid local buckling, can be estimated by
the following relation:
94 Mechanical Design of Composite Pipelines
ο
M y D α11 ∧
ε =
ξ − N x = a11λcr (3.183)
EI yy 2 π D
∧
M y D a11 ∧
> Nx (3.184)
EI yy 2 πD
We recall that the value of λcr should be determined by the following equation
[11]:
O L M 0 M n O L
T
Φ ⋅ [ J ] Φ 2 ⋅ [ J ]
det −λ 1 = 0 (3.185)
L O M n
M 0 L O Φ 2 ⋅ [ J ] Φ1 ⋅ [ J ]
where the matrices [O], [L], [Mo], [Mn], [J] and the functions Φ1, Φ2 are given in
section 3.3.
It is important to remind that equation (3.185) contains four unknown con-
stants a, b, c1, c2. For different combinations of these constants, eq. (3.185) yields
different values for λ. The lowest value of λ is the appropriate parameter λcr to be
used in eq. (3.183).
∧
N x = N x / πD (3.186)
According to section 3.4, external pressure causes in-plane loading in the y-direction
(Fig. 3.18b) given by:
1
Ny = pD (3.187)
2
Combination of External Pressure and Axial Tension 95
ª D D D E E E − º 1[ ½ ½
«D ° ° ° °
« D D E E E − »» ° 1 \ ° ° °
« D D D E E E − » ° 1 [\ ° ° °
« » ° ° ° °
« E E E G G G − » ° 0 [ ° ° °
« E E E G G G − » ° 0 \ ° ° °
« » ° ° ° °
« E E E G G G −» ° 0 [\ ° ° °
« ⋅® ¾ = ® ¾
» ° ε [ ° ° 1Ö [ π ' °
« »
» ° ε \ ° ° S' °
«
« ° ° ° °
» ° γ [\ ° ° °
« »
» ° N[ ° ° °
«
« » °
° ° °
« » ° N\ ° ° °
«¬ »¼ °¯ N [\ ¿° °¯ °¿
(3.188)
96 Mechanical Design of Composite Pipelines
ε x0 , ε y0 , γ xy0 , k x0 , k y0 , k xy0
∧
Nx pD
ε x0 = a11 + a12 (3.189)
πD 2
∧
0 N pD
ε = a12 x + a22
y (3.190)
πD 2
∧
0 N pD
γ xy = a16 x + a26 (3.191)
πD 2
∧
0 N pD
k = b11 x + b12
x (3.192)
πD 2
∧
0 N pD
k = b12 x + b22
y (3.193)
πD 2
∧
0 N pD
k = b16 x + b26
xy (3.194)
πD 2
Therefore:
ε x = ε x0 + zk x0 (3.195)
ε y = ε y0 + zk y0 (3.196)
With the aid of equations (3.189)–(3.197), the stress components σx, σy, τxy, can be
obtained for every lamina using the well-known relation:
σ x ε x
σ y = Q ij (θ ) ε y (3.198)
τ xy γ xy
Using the matrix [ Τ(θ ) ] the principal stresses σ1, σ2, τ12 can be obtained by the
following equation:
Torsion 97
σ 1 σ x
σ 2 = T (θ ) σ y (3.199)
τ
12 τ xy
With the aid of eq. (3.199), the Tsai-Wu failure criterion yields the areas of the
∧
allowable combinations of the values of axial force N x and external pressure p
for avoiding failure for every ply.
The failure analysis described above is not sufficient for design purposes. The
derived geometry of the pipeline must be additionally checked by the buckling
model described in section 3.3.2. Acceptable pipe diameter (Dia), fiber orienta-
tion [±θ], number of plies (NP), thickness of lamina (h) are ones that satisfy both
the failure criterion and the buckling model.
3.8 Torsion
3.8.1 Failure analysis
When a torque M �
x (see Fig. 19a) is applied to a long composite pipe, a re-
sultant Nξ n is acting on the cross-section of the wall (see Fig. 19b). Taking into
account the equilibrium between Nξ n and M �
x it can be written as:
�
M
Nξ n = x
(3.200)
2π R 2
The corresponding strains ε ξo , ε no , γ ξon , kξo , kno , kξon can be obtained by the following
matrix equation:
98 Mechanical Design of Composite Pipelines
�
a16 M
ε ξo = x
(3.205)
2π R 2
a M �
ε no = 26 2x (3.206)
2π R
a M �
γ ξon = 66 2x (3.207)
2π R
Torsion 99
�
b61 M
kξo = x
(3.208)
2π R 2
�
b62 M
kno = x
(3.209)
2 πR 2
b M �
kξon = 66 2x (3.210)
2π R
Therefore
ε ξ ε ξo kξo
o o
εn = εn + ζ kn (3.211)
γ γ o k o
ξn ξn ξn
For symmetric lay-up of fibers, the maximum shear stresses τ ξ n take place in
the exterior layers of the pipe. Therefore, for ζ = h / 2 equations (3.205)–(3.211)
yield:
h
a16 + 2 b16
εξ
�
M h
x
εn = 2 a26 + b62 (3.212)
γ 2π R 2
ξn h
a66 + 2 b66
Taking into account the above equation, the stress-strain relation for the exterior
lamina with fiber orientation θ can be written as:
σ ξ εξ
σ n = Qij (θ ) ⋅ ε n (3.213)
τ γ
ξn ξn
Therefore, the principal stresses of the exterior lamina can be obtained by the fol-
lowing well-known formula:
σ 1 σ ξ
T
σ 2 = (θ ) ⋅ σ n (3.214)
τ τ
12 ξn
100 Mechanical Design of Composite Pipelines
� a can be obtained by
With the aid of the above equation, the allowable torque M x
the Tsai-Wu criterion:
where
(a) the geometric relations between the displacements u, v, w along the axes x,
y, z respectively;
(b) the equilibrium of the resultants acting in the wall, with torque Mx
(c) the equilibrium of exterior shear force per unit length T applied at the edges
of the pipe, with torque M� x , i.e. 2 πR 2T = M � x / 2 πR 2
� x or T = M
(d) the fact that approximation 1+ ( z / R) ≈ 1 yields M xy = M yx and N xy = N yx
[9] the following relations based on [9] can be written as:
∂N x ∂N xy M� x ∂ 2u
R2 +R − =0 (3.216)
∂x ∂θ πR ∂x∂θ
∂N y ∂N xy ∂M y ∂M xy � x ∂ 2 v ∂w
M
R + R2 − −R − + =0 (3.217)
∂θ ∂x ∂θ ∂x πR ∂x∂θ ∂x
∂2 M y ∂ 2 M xy ∂2 M x � x ∂v ∂ 2 w
M
2
+ 2R + 2
+ RN y − − =0 (3.218)
∂θ ∂x∂θ ∂x πR ∂x ∂x∂θ
Taking into account equations (3.134) as well as the following type of the
solutions:
λx
u = Α sin + mθ (3.219)
R
Torsion 101
λx
v = Βsin + mθ (3.220)
R
λx
w = C cos + mθ (3.221)
R
n πR
where λ = and n, m integers, the following system of equations can be
derived: L
where:
mλ �
G12 = − Α12 m λ − Α66 m λ + Mx (3.224)
πR 2
Β12 2 2Β Β
G13 = m λ + 66 m 2 λ + 11 λ3 (3.225)
R R R
102 Mechanical Design of Composite Pipelines
Β21 Β
G21 = − Α12 m λ − Α66 m λ + m λ + 66 m λ (3.226)
R R
Β 22 2 m λ � Β
G22 = − Α 22 m 2 + m + 2
M x − Α66 λ2 + 66 λ2 (3.227)
R πR R
D22 3 Β22 3 λ � D D
G23 = − 2
m + m − 2
M x − 122 m λ2 − 2 662 m λ2 +
R R πR R R (3.228)
Β12 2 Β
+ m λ2 + 66 m λ2
R R
Β21 2 2Β Β
G31 = Α12 λ − m λ − 66 m 2 λ − 113 λ3 (3.229)
R R R
Β22 3 λ � Β 2Β
G32 = Α 22 m − m − 2
M x − 123 m λ2 − 66 m λ2 (3.230)
R πR R R
D22 4 Β 22 2 m λ � D D
G33 = 2
m − m − 2
M x + 124 m 2 λ2 + 122 m 2 λ2 +
R R πR R R
(3.231)
4 D66 2 2 Β12 2 D11 4
+ 2 m λ − λ + 4 λ
R R R
The homogeneous matrix equation (3.222) can yield non-zero solutions for A, B,
C only if the following determinant is zero:
References
[1] Pavlou D.G., “Computational and experimental analysis of damaged mate-
rials,″ Transworld Research Network, 2007.
References 103
[2] Guedes R.M., Creep and fatigue in polymer matrix composites, Woodhead
publishing, 2011.
[3] Harris B., Fatigue in composites, CRC press, 2003.
[4] Berenberg B., http://composite.about.com
[5] Lekhnitskii S.G., Theory of elasticity of an anisotropic elastic body, Holden-
Day, 1963.
[6] Xia M., Takayanagi H., and Kemmochi K., “Bending behavior of filament-
wound fiber-reinforced sandwich pipes,” Composite Structures 56, 201–
210, 2002.
[7] Reddy J.N., Mechanics of laminated composite plates and shells, CRC
press, 2004.
[8] Tsai S.W., and Wu E.M., “A general theory of strength for anisotropic ma-
terials,″ Journal of Composite Materials 5, 58–80, 1971.
[9] Flügge W., Stresses in shells, Springer-Verlag, 1973.
[10] Timoshenko S.P., and Gere J.M., Theory of elastic stability, Dover publica-
tions 2009.
[11] Kollár L.P., and Springer G.S., Mechanics of composite structures,
Cambridge University Press, 2003.
[12] Rasheed H.A., and Karamanos S.A., “Stability of tubes and pipelines,” in:
Buckling and Postbuckling Structures, edited by Falzon B.G. and Aliabadi
M.H., Imperial College Press, 2008.
Chapter 4
The first known study of the dynamic analysis of pipes conveying liquids ap-
peared in 1950 [1]. It represented an attempt to explain vibrations observed in
the Trans-Arabian pipeline. Today the most complete analysis for flow-induced
vibrations of isotropic pipes (e.g., steel) can be found in the works of Païdoussis
[2,3]. In this chapter, analysis of dynamic stability will be performed for multilay-
ered filament-wound composite pipes.
105
106 Dynamic Stability of Composite Pipelines
Figure 4.1 (a) System of coordinates of a wall element of a multilayered composite pipe,
(b) System of coordinates of a multilayered composite pipe.
1
ε ξo = z (4.1)
ρy
1
κ ξo = cos a (4.2)
ρy
1 o
Nξ = εξ (4.5)
a11
1 o
Mξ = κξ (4.6)
d11
With the aid of equations (4.5) and (4.6), the above equation yields:
� y= 1 o 1 o
dM εξ ⋅ zdn + κ ξ cos a ⋅ dn (4.8)
a11 d11
Taking into account equations (4.1), (4.2), the above equation can now be written
as:
� y= 1 1 2 1 1
dM z dn + cos 2 adn (4.9)
ρ y a11 ρ y d11
� y= 1 1 1
M
ρy ∫ a
(s) 11
z2 +
d11
cos 2 a dn
(4.10)
108 Dynamic Stability of Composite Pipelines
D
z= cos a (4.11)
2
D
dn = da (4.12)
2
With the aid of eqs. (4.11), (4.12), eq. (4.10) yields:
� y = 1 D + 1 D cos 2 ada
π
ρ y 2a11 d11 ∫o
M (4.13)
or
πD D 1
+
� y= 2 2
11a d11
M (4.14)
ρy
� yy = M
EI � y⋅ρ (4.15)
y
Combining eqs. (4.14), (4.15), it can be concluded that the stiffness of a multilay-
ered composite pipe is given by:
� yy = πD D + 1
EI (4.16)
2 2a11 d11
It should be recalled that the parameters a11 , d11 can be derived by the inversion
of the stiffness matrix of the laminated wall, i.e.,
−1
a11 a12 a16 b11 b12 b16 A11 A12 A16 B11 B12 B16
a b26 A12
12 a22 a26 b21 b22 A22 A26 B12 B22 B26
a16 a26 a66 b61 b62 b66 A16 A26 A66 B16 B26 B66
=
b11 b21 b61 d11 d12 d16 B11 B12 B16 D11 D12 D16
b b b d12 d 22 d 26 B12 B22 B26 D12 D22 D26
12 22 62
16 26 b66
b b d16 d 26 d 66 B16 B26 B66 D16 D26 D66
(4.17)
Free Vibration of Composite Pipes 109
where
N
Aij = ∑ Q ijk ( zk − zk −1 )
k =1
1 N
Bij = ∑ Q ijk ( zk2 − zk2−1 ) (4.18)
2 k =1
1 N
Dij = ∑ Q ijk ( zk3 − zk3−1 )
3 k =1
In the above equation, zk , zk −1 are the distances of corresponding layers from the
middle plane of the cross-section of the laminate in Figure 4.2a.
figure 4.2 (a) cross-section and (b) plan form view of the laminate [5].
110 Dynamic Stability of Composite Pipelines
Q 66 = ( Q11 + Q22 − 2Q12 − 2Q66 ) n 2 m 2 + Q66 ( n 4 + m 4 )
where
n = sin θ (4.20)
m = cos θ (4.21)
E1
Q11 = (4.23)
1 −ν 12ν 21
v12 E2 v E
Q12 = = 21 1 (4.24)
1 − v12 v21 1 − v12 v21
E2
Q22 = (4.25)
1 − v12 v21
In the above equations (4.20)–(4.26), θ is the fiber direction of each ply, while
E1 , E2 , v12 , v21 , G12 are the modulus of elasticity (E), Poisson’s ratio (ν) and
shear modulus (G) in the principal directions (i.e., the directions along and normal
to the directions of fibers) of a ply. The above procedure for the derivation of the
stiffness of a composite pipe can be summarized in the following diagram:
Free Vibration of Composite Pipes 111
Figure 4.3 Schematic procedure for the derivation of the stiffness of multilayered com-
posite pipe.
Acting in the length direction of the deformed pipe are: (a) shear stresses q
due to the friction between the fluid and the interior cylindrical surface πDdx, and
(b) longitudinal tension T.
Acting in the direction normal to the deformed pipe are: (a) shear forces Q,
(b) reaction forces F between the fluid and the interior surface of the pipe, verti-
cal gravity forces mgdx due to the weight of the material of the corresponding
pipe element, and vertical dynamic forces mazp due to the vertical motion of the
elementary mass m of the pipe ( azp is the vertical acceleration). Acting in the y-
direction are the bending moments M � y . Taking into account that the pipe element
has a slope ∂w / ∂x (with respect to the x-axis) due to bending and applying the
equilibrium conditions of the forces projected on the axes x, z, y, the following
can be determined.
Free Vibration of Composite Pipes 113
∂T ∂w
+ πDq − F =0 (4.27)
∂x ∂x
∂Q ∂ ∂w ∂w
+ F + T + πDq + mg = mazp (4.28)
∂x ∂x ∂x ∂x
� y
∂M
= −Q (4.29)
∂x
Taking into account equation (4.15), the equation above can be written as:
� yy ∂ 1
EI
= Q (4.30)
∂x ρ y
1 ∂2 w
= (4.31)
ρ y ∂x 2
� yy ∂ w
3
Q = − EI (4.32)
∂x 3
In the direction along the axis of the deformed pipe, the forces acting on a fluid
element shown in Figure 4.6 are: (a) forces due to pressure p acting on the cross-
section area A = πD 2 / 4 ; (b) shear forces q due to friction of the fluid with the
interior cylindrical surface of the pipe; and (c) horizontal dynamic forces due to
the motion of the fluid in the x and z directions. In the direction normal to the
axis of deformed pipe, the forces acting on the fluid element are: (a) reaction
114 Dynamic Stability of Composite Pipelines
forces F between the fluid and the interior surface of the pipe; (b) the vertical
gravity forces Mgdx due to the weight of the fluid element; and (c) vertical
dynamic forces due to the motion of the fluid (in the x and z directions).
axp = 0 (4.35)
The definition of the acceleration of the fluid element is more complicated since
this motion takes place in two directions, i.e., in the direction along the pipe due
to its velocity U, and in the vertical direction z, which is due to the vertical motion
of the pipe (see Figure 4.7).
As the slope of the pipe ∂w / ∂x is very small, we can use the following
approximations:
Therefore, the projection of the velocity U along the axes x and z has the follow-
ing values:
Ux = U (4.39)
∂w
Uz = U (4.40)
∂x
Taking into account that the vertical velocity of the pipe is ∂w / ∂t , the vertical
velocity of the fluid element can be written as:
∂w ∂w
U zf = +U (4.41)
∂t ∂x
while the horizontal velocity of the fluid element is
U xf ≈ U (4.42)
rf ur f
Taking into account that a = ∂U / ∂t , the accelerations of the fluid element in
the vertical and horizontal directions can be obtained [2]:
∂2 w ∂2 w ∂ 2 w ∂U ∂w
azf = + 2U +U 2 2 + (4.43)
∂t 2
∂x∂t ∂x ∂t ∂x
and
∂U xf ∂U
axf = = (4.44)
∂t ∂t
∂w
F = πDq (4.45)
∂x
∂Q ∂w ∂2 w
+ F + πDq =m 2 (4.46)
∂x ∂x ∂t
� yy ∂ w
3
Q = − EI (4.47)
∂x 3
∂w
− q πD + F =0 (4.48)
∂x
∂w ∂2 w ∂2 w 2 ∂ w
2
− F − q πD =M 2 + 2U + U (4.49)
∂x ∂t ∂x∂t ∂x 2
Equation of Motion 117
∂w ∂Q ∂2 w
− F − πDq = −m 2 (4.50)
∂x ∂x ∂t
The combination of equations (4.49) and (4.50) yields :
∂Q ∂2 w ∂2 w ∂2 w ∂2 w
− m 2 = M 2 + 2 MU + MU 2 2 (4.51)
∂x ∂t ∂t ∂x∂t ∂x
Taking into account eq. (4.47), the above equation can now be written as:
4 2 2 2
� y y ∂ w + MU 2 ∂ w + 2 MU ∂ w + ( M + m) ∂ w = 0
EI (4.52)
∂x 4 ∂x 2 ∂x∂t ∂t 2
� yy is given by eq. (4.16) and can be derived by the procedure illustrated
where EI
in Figure 4.3.
The homogeneous partial differential equation above is the equation of motion
describing the free vibration of the composite pipe under consideration.
w( x, t ) t =0 = ws ( x)
dws ( x)
w '( x, t ) t =0 =
dx
d 2 ws ( x) (4.53)
w "( x, t ) t = 0 =
dx 2
3
d ws ( x)
w "'( x, t ) t = 0 =
dx 3
where ws ( x) is the static deflection curve because of the uniform self-weight
q = (M + m). Taking into account the expression for obtaining the stiffness of a
composite pipe, i.e., equations (4.16)–(4.26), the function ws ( x) for several types
of support can be approximated as follows:
118 Dynamic Stability of Composite Pipelines
( M + m) x 2 2
ws ( x) = ( x + 6 L2 − 4 Lx)
�
24 EI yy for cantilever pipe
2
( M + m) x
ws ( x) = ( L − x) 2 for fixed-fixed pipe
� yy
24 EI
(4.54)
( M + m) L L − x
4
( L − x )3 ( L − x) 4 for fixed-pinned
ws ( x) = − 3 + 2
48 EI L L3 L4 pipe
( M + m) x 3 for pinned-pinned
ws ( x) = ( L − 2 Lx 2 + x 3 )
� yy
24 EI pipe
Table 4.1
Boundary conditions for several types of support.
Type Support Equation
Pinned
w(0, t ) = 0 (4.55)
∂ 2 w(0, t )
=0 (4.56)
∂x 2
Fixed
w(0, t ) = 0 (4.57)
∂w(0, t )
=0 (4.58)
∂x
Free
∂ 2 w(0, t )
=0 (4.59)
∂x 2
∂ 3 w(0, t )
=0 (4.60)
∂x 3
Equation of Motion 119
Deflected
spring ∂ 2 w(0, t )
=0 (4.61)
∂x 2
� yy ∂ 3 w(0, t )
− EI = K D w(0, t ) (4.62)
∂x 3
Torsion
spring
w(0, t ) = 0 (4.63)
2
� yy ∂ w(0, t ) = K ∂w(0, t )
EI (4.64)
T
∂x 2 ∂x
Mass
∂ 2 w(0, t )
=0 (4.65)
∂x 2
� yy ∂ 3 w(0, t ) ∂ 2 w(0, t )
− EI 3
= me (4.66)
∂x ∂t 2
Dashpot
∂ 2 w(0, t )
=0 (4.67)
∂x 2
� yy ∂ 3 w(0, t ) ∂w(0, t )
− EI 3
=c (4.68)
∂x ∂t
For the boundary value problem under consideration, we can use solutions of
the form:
Substitution of the solution (4.69) into the partial differential equation (4.52)
yields:
4 2
� yy d f ( x) + MU 2 d f ( x) + 2 MU ωi df ( x) − ( M + m)ω 2 f ( x) = 0 (4.70)
EI
dx 4 dx 2 dx
For the above equation we can try the solution
f ( x) = Ceiβ x (4.71)
� yy β 4 − MU 2 β 2 − 2 MU ωβ − ( M + m)ω 2 = 0
EI (4.72)
β1 = −κ 7 − κ 8 (4.73)
β 2 = −κ 7 + κ 8 (4.74)
β 3 = +κ 7 − κ 8 (4.75)
β 4 = +κ 7 + κ 8 (4.76)
where:
1 4 MU ω
κ8 = 2κ 0 − κ 6 − (4.77)
2 2κ 7 EI y y
1
κ7 = κ0 + κ6 (4.78)
2
κ1 3 2 κ5
κ6 = + (4.79)
3EI y yκ 5 EI y y 3 54
κ 5 = (κ 2 + κ 4 )1/ 3 (4.80)
κ 4 = −4κ13 + κ 32 (4.81)
κ 3 = −2 M 3U 6 + 108 EI y y M 2U 2ω 2 − 72 EI y y M (m + M )U 2ω 2 (4.82)
Equation of Motion 121
κ 2 = −2 M 3U 6 − 72 EI y y mMU 2ω 2 + 36 EI y y M 2U 2ω 2 (4.83)
κ1 = M 2U 4 − 12 EI y y (m + M )ω 2 (4.84)
2 MU 2
κ0 = (4.85)
3EI y y
Application of the general solution equations (4.69), (4.71), and (4.73-4.85) into
the boundary conditions for several combinations of supports (Table 4.1) yields
the following:
C1 + C2 + C3 + C4 = 0 (4.86)
1 1 1 1
β β2 β3 β 4
det 2 1iβ1L =0 (4.90)
β1 e β 2 2 ei β 2 L β 3 2 ei β3 L β 4 2 ei β 4 L
3 iβ1L
β1 e β 2 3 ei β 2 L β 33 e i β 3 L β 4 3 ei β 4 L
Numerical solutions for the eigen-frequencies ωi can be obtained from eq. (4.90)
by taking into account eqs. (4.73)–(4.85).
By applying the concept above, the following equations for obtaining the eigen-
frequencies ωi can be derived for several types of supports:
122 Dynamic Stability of Composite Pipelines
1 1 1 1
β β2 β3 β 4
det iβ11L =0 (4.91)
e ei β 2 L ei β3 L ei β 4 L
iβ1L
β1e β 2 ei β 2 L β 3 ei β3 L β 4 ei β 4 L
1 1 1 1
β β2 β3 β 4
det iβ11L =0 (4.92)
e ei β 2 L ei β3 L ei β 4 L
2 iβ1L
β1 e β 2 2 ei β 2 L β 3 2 ei β3 L β 4 2 ei β 4 L
1 1 1 1
β2 β 22 β 32 β 42
det iβ11L =0 (4.93)
e ei β 2 L ei β3 L ei β 4 L
2 iβ1L
β1 e β 2 2 ei β 2 L β 3 2 ei β3 L β 4 2 ei β 4 L
As already mentioned, a pipe conveying fluid is stable when Im[ωi ] > 0 and un-
stable when Im[ωi ] < 0 . Therefore, the critical value of the fluid flow Ucr for
passing from stability to instability can be estimated by the following equation:
Im[ωi ] = 0 (4.94)
d 2w
• MU 2 represents the centrifugal force of the fluid flowing with constant
dx 2
speed for a curved portion of the pipe (because actually the derivative
d 2 w / dx 2 represents the value 1/ R , where R is the local radius of curvature).
Equation of Motion 123
∂2 w
• 2 MU represents the Coriolis force.
∂x∂t
∂2 w
• ( M + m) represents the inertia effects of the masses of fluid and pipe.
∂t 2
In the above four terms, only the second (centrifugal force) and third (Coriolis
force) are influenced by the flow velocity U. For small values of U, the dynamic
behavior of pipe is dominated by the Coriolis force that is proportional to U.
In that case, the system is subjected to flow-induced damping [3] because the
Coriolis forces in the fluid react to the pipe in a direction opposite the motion.
For higher values of U, the centrifugal force (which is proportional to U2) might
overcome the Coriolis damping effect, and the system can lose stability.
Instability can be introduced by either divergence (a static form of instability)
or flutter (a dynamic form of instability) [3]. Divergence may occur if both ends
of the pipe are supported (e.g., fixed-fixed, pinned-pinned, fixed-pinned). In the
case of cantilever pipe (where the one end is free), flutter may occur in the pipe.
For the special case that the direction of the flow is from the free end towards the
fixed one, the cantilever pipe will become unstable (due to flutter) for very small
values of U, and then be stabilized for larger values of U as first pointed out by
Païdoussis and Luu [6]. If we neglect the terms of Coriolis (causing damping in
low flow velocity) and inertial forces, eq. (4.52) yields:
4 2
� yy d w + MU 2 d w = 0
EI (4.95)
4
dx dx 2
From the above equation it is clear that the centrifugal force acts in the same
manner as a compressive axial force in a straight column. Therefore, when U
increases, the “compressive effect” of the centrifugal force can overcome the flex-
ural resistance of the pipe, which causes divergence. Since divergence is a static
rather than a dynamic form of instability, the critical flow U cr causing divergence
may be examined by considering only the time-independent terms of eq. (4.52),
so effectively eq. (4.95). From an engineering point of view, the solution of eq.
(4.52) can yield the critical values of U that cause instability for both cases of
pipes, i.e., cantilever or supported on both ends. However, the value of U that does
cause divergence instability in the second case, can be adequately obtained from
eq. (4.95). In cases where eq. (4.52) is used, obtaining the eigenvalues by solving
equations (4.90)–(4.93) is a very difficult task, even when advanced numerical
methods are implemented. On the other hand, eq. (4.95) has a very simple solu-
tion—but for a uniform pipe. However, for pipes where certain properties may
vary gradually along its length or for cases where long pipes are supported period-
ically by more than two supports, a direct solution is also quite difficult. For these
cases, approximate methods like the Galerkin or FEM can be effectively used for
124 Dynamic Stability of Composite Pipelines
obtaining the critical values of the velocity U cr that causes instability. In the au-
thor’s opinion, of existing methods for solving equations (4.52) or (4.95) that ap-
ply to cases involving additional supports or temperature gradients or additional
point masses or elastic foundation along the pipe’s length, the most advantageous
is the Transfer Matrices Method (TMM). The Transfer Matrices Method is es-
sentially an analytic method that reduces the higher-order differential equation
(here fourth-order) into a first-order matrix differential equation. The Transfer
Matrices Method provides simpler algebraic equations than eqs. (4.90)–(4.93),
which facilitates the derivation of the critical velocity U cr by using conventional
numerical methods.
n
dm
∑a
m=0
m
dx m
f ( x) = q ( x), an ≠ 0, x ∈ [ x1 , xn ] (4.96)
y = f ( x) (4.101)
Transfer Matrices Method (TMM) 125
and
the n known and the n unknown boundary conditions given by equations (4.97)–
(4.100) can be represented by the following matrix equations:
1 0 L 0
0 1 L 0
[ I ]nxn = (4.106)
M M O 0
0 0 L 1 nxn
and [X1]nx1, [XN]nx1 are the state vectors of the boundary points x1, xn respectively
while [EU]nx1, [UE]nx1 are the vectors containing the following values of the
(known and unknown) boundary conditions:
1 a a a a
yn = y 'n −1 = f ( n ) ( x) = q ( x) − n −1 yn −1 − n − 2 yn − 2 − − 1 y1 − 0 y (4.111)
an an an an an
Then the system of equations (4.101)–(4.103) and (4.111) can be written in the
following matrix form:
126 Dynamic Stability of Composite Pipelines
0 1 0 0 L 0 0
y y
y 0 0 1 0 L 0
y 0
1 0 0 0 1 L 0
1
0
d y2 y2
= 0 0 0 0 L 0 + 0
dx y3 y
M
3
M M M M O M
M M
a0 a1 a2 a a q( x)
yn −1 − − − − 3 L − n −1 yn −1
an an an an an an
(4.112)
where
[ X ] = [ y ( x), y3 ( x), yn −1 ( x) ]
T
y1 ( x), y2 ( x), (4.114)
T
q( x)
[ F ] = 0, 0, 0, 0, (4.115)
an
1 for j = i + 1
a j −1
[ A] = { Aij } = − for i = n and j = 1, 2,..., n (4.116)
an
0 for the other values i, j
For the end point x = xn the matrix equation (4.112) can be written as:
[ XN ]' = [ AN ][ XN ] + [ FN ] (4.117)
where [XN] is the state vector of the end point xn, given by eq. (4.108), and
T
q ( xn )
[ FN ] = 0, 0, 0, 0, (4.118)
an
Transfer Matrices Method (TMM) 127
1 for j = i + 1
a j −1
[ AN ] = { AN ij } = − for i = n and j = 1, 2,..., n (4.119)
an
0 for the other values i, j
Then, the solution of the matrix differential equation (4.117) can be obtained [e.g. 7]:
xn
[ XN ] = ∫ e[ AN ]( xn − s ) [ F ( s )]ds + e[ AN ]( xn − x1 ) [ X 1] (4.120)
x1
where [X1] is the state vector of the initial point x1, given by eq. (4.107), and
T
q( s)
[ F ( s )] = 0, 0, 0, 0, (4.121)
an
The eq. (4.120) is a matrix equation correlating the state vector [XN] of the end-
point xn with the state vector [X1] of the initial point x1. For the case that q(x) = 0
(e.g., for the case of free motion), the matrix e[ AN ]( xn − x1 ) correlating the state vec-
tors between two points i and n is called Transfer Matrix. Eq. (4.120) can be
considered as an algebraic linear system of n equations with 2n unknown val-
ues, which are incorporated in the boundary state vectors [X1] and [XN], i.e.,
[ y( x1 ), y1 ( x1 ), y2 ( x1 ), y3 ( x1 ), yn −1 ( x1 ), y( xn ), y1 ( xn ), yn −1 ( xn )].
To solve this (n × 2n) linear algebraic system, n additional linear equations are
required in order to yield a complete (2n × 2n) system. These n additional equa-
tions can be obtained by the n known boundary values e1, e2, …,ep, …,en which are
incorporated in matrix equations (4.104)–(4.105). Considering equations (4.104),
(4.105), (4.120) and taking into account the transformations given by equations
(4.101)–(4.103), the following linear system can be obtained:
[ A11 ] [ A12 ] [ X 1] [ B1 ]
[ A ] [ A ] ⋅ [ XN ] = [ B ] (4.122)
21 22 2
where the sub-matrices [A11], [A12], [A21], [A22], [B1], [B2] are given by the
following matrix equations:
[ I ] [ 0]
[ A21 ] = kxk
[0]
(4.125)
[0]
[0] [0]
[ A22 ] = (4.126)
[ 0 ] [ I ](n−k ) x(n−k )
xn
From the solution of the linear algebraic system (4.122), the unknown bound-
ary values [u1, …, un] of the boundary points x1 and xn (see equations (4.104),
(4.105)) can be obtained. Therefore, e.g., at the point x1, the value of the function
f(x1) = y(x1) and the values of its derivatives y1(x1), y2(x1), …,yk(x1), …,yn-1(x1), i.e.,
the vector [EU]nx1, are now known. Then, the solution
[ X Ξ] = [ y (ξ ), y3 (ξ ), yn −1 (ξ ) ]
T
y1 (ξ ), y2 (ξ ), (4.129)
where the sub-matrices [AA11], [AA12], [AA21], [AA22], [BB1], [BB2] are given by the
following matrix equations:
1 for j = i + 1
a j −1
[ AΞ] = { AΞ ij } = − for i = n and j = 1, 2,..., n (4.137)
an
0 for the other values i, j
a0 = −( M + m)ω 2 (4.138)
a1 = 2 MU ω i (4.139)
a2 = MU 2 (4.140)
a3 = 0 (4.141)
� yy
a4 = EI (4.142)
Therefore, according to eq. (4.119), the matrix [AN] can be written as:
0 1 0 0
0 0 1 0
[ ]
AN = 0 0 0 1 (4.143)
2
( M + m) ω −2 MU ω i − MU 2
0
EI � yy � EI yy � yy
EI
y ( 0) 0
y ( 0) 0
1
y 2 ( 0) 0
e[ AN ].L
[ − I 4 x 4 ] y3 (0) 0
= (4.144)
[ A21 ]
[ A22 ] y ( L) e1
y1 ( L) e2
y2 ( L) e3
y ( L ) e
3 4
In above matrix equation, the matrices [ A21 ] , [ A22 ] , {e1 , e2 , e3 , e4 } depend on the
T
boundary conditions at the end points x = 0 and x = L of the pipe. For the cases of
cantilever, fixed-fixed, pinned-pinned, and fixed-pinned pipes, the above matrices
have the following forms.
1 0 0 0
0 1 0 0
[ A21 ] = (4.145)
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
[ A22 ] = (4.146)
0 0 1 0
0 0 0 1
1 0 0 0
0 1 0 0
[ A21 ] = (4.147)
0 0 0 0
0 0 0 0
Estimation of Critical Velocity for Composite Pipes Conveying Fluid 131
0 0 0 0
0 0 0 0
[ A22 ] = (4.148)
1 0 0 0
0 1 0 0
1 0 0 0
0 0 1 0
[ A21 ] = (4.149)
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
[ A22 ] = (4.150)
1 0 0 0
0 0 1 0
1 0 0 0
0 1 0 0
[ A21 ] = (4.151)
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
[ A22 ] = (4.152)
1 0 0 0
0 0 1 0
For all above cases of support, the vector {e1 , e2 , e3 , e4 } has the form:
T
{e1 , e2 , e3 , e4 } = {0, 0, 0, 0}
T T
(4.153)
132 Dynamic Stability of Composite Pipelines
e[ AN ]⋅ L [ − I 4 x 4 ]
det =0 (4.154)
[ A21 ] [ A22 ]
L
∫ ε dx = 0
0
x
∂T
=0 (4.156)
∂x
From eq. (4.156) it can be concluded that the tension T has a constant value caus-
ing constant longitudinal stress. For a multi-layered composite pipe consisting of
NP plies with symmetric fiber orientation +θ or –θ, the longitudinal stress of each
ply has the value
Effect of Temperature (Thermal Load) 133
(T / NP )
σx = (4.157)
π Dh
where h is the thickness of each ply and D is the mean diameter of the pipe. Since
the pipe is subjected to internal pressure P, the stress s y for the multi-layered
long pipe is
D
s y = ( P / NP ) (4.158)
2h
We recall that for a lamina with fiber orientation θ, it can be written:
or
With the aid of the above equation, eq. (4.160) can now be written:
where:
1 1 1 2 v12
cos θ sin θ − ( cos θ + sin θ )
2 4 4
S12 = + − (4.164)
E
1 E 2 G12 E 1
Taking into account equations (4.157), (4.158) and (4.162)–(4.164), the condition
of no longitudinal movement at the pipe ends, which is
L
∫ ε x dx = 0 (4.165)
0
134 Dynamic Stability of Composite Pipelines
yields:
πD D ⋅ P ⋅ S12
T=−
S11 2
( )
+ NP ⋅ h ⋅ a1 cos 2 θ + a2 sin 2 θ ∆Τ (4.166)
If we do not neglect the tensioning from eq. (4.28), eq. (4.46) can be written:
∂Q ∂w ∂ ∂w ∂2 w
+ F + πDq + T = m (4.167)
∂x ∂x ∂x ∂x ∂t 2
Combining the above equation with eq. (4.49) yields:
∂Q ∂2 w ∂2 w ∂ 2 w ∂T ∂w ∂2 w ∂2 w
−M 2 + 2U +U 2 2 + + T 2 = m 2 (4.168)
∂x ∂t ∂x∂t ∂x ∂x ∂x ∂x ∂t
Since the axial force T has the same value for every point x, the derivative ∂T / ∂x
has the value
∂T
=0 (4.169)
∂x
With the aid of equations (4.47) and (4.169), eq. (4.168) yields:
4 2 2 2
� yy ∂ w + ( MU 2 − T ) ∂ w + 2 MU ∂ w + ( M + m ) ∂ w = 0
EI (4.170)
∂x 4 ∂x 2 ∂x∂t ∂t 2
Taking into account eq. (4.166), the above equation can now be written as:
(4.171)
π D D ⋅ P ⋅ S12
H (U , ∆Τ ) = MU 2 + + NP ⋅ h ( a1 cos 2 θ + a2 sin 2 θ ) ∆Τ
S11 2
(4.172)
Effect of Additional Mass 135
0 1 0 0
0 0 1 0
[ AN ] = 0 0 0 1 (4.173)
( M + m)ω − H (U , ∆Τ )
2
−2 MU ω i
0
� yy
EI � EI yy � yy
EI
The matrix above can be used in the Transfer Matrix Method (TMM) to determine
the critical flow velocity U cr for a multilayered composite pipe under the influ-
ence of temperature change.
(a) (b)
figure 4.8 (a) Installation of a composite pipe into the sea by using metallic collars, right
(b) Collar-pipe cross section.
136 Dynamic Stability of Composite Pipelines
0 1 0 0
0 0 1 0
[ AN c ] = 0 0 0 1 (4.174)
( M + m + mc ) ω
2 2
−2 MU ω i − MU
0
� c
EI � EI c � c
EI
where mc is the mass per unit length of the collar and EI � c is the stiffness of the
collar-pipe cross-section.
Therefore, the matrix equation correlating the state vectors { y, y1 , y2 y3 } of the
T
end points A and B of the pipe segment equipped with a collar is:
y y
y y
1 1
y
= [TM c ] ⋅ (4.175)
2 y2
y3 y3
B A
where [TM c ] is the transfer matrix of the collar given by:
[TM c ] = e[ AN ]⋅L
c c
(4.176)
In the above equation, Lc is the length of the collar. In order to calculate the
� c we shall consider the collar-pipe system as itself a composite pipe
stiffness EI
consisting of a multi-layered anisotropic pipe (with internal radius r1 and external
radius r2 = r1 + NP ⋅ h ), as well as an isotropic steel pipe (with internal radius r2
and external radius r3 ) perfectly bonded (Fig. 4.8b). In such a case, the elements
of the ABD matrix should be calculated by the following formulas:
c
Aijc = Aij + Q ij ( r3 − r2 ) (4.177)
In the above equation, Aij , Bij , Dij correspond to the multi-layered pipe, while
c
the parameters Q ij correspond to the steel collar. Since steel can be considered
as an isotropic material (n = 0, m = 1 because [T ] = I ) where E1 = E2 = E3 = E ,
c
ν 23 = ν 13 = ν 12 = ν , G23 = G13 = G12 = G , the parameters Q ij can be derived by
the following equations:
Effect of Additional Mass 137
c E
Q11 = (4.180)
1 −ν 2
c νE
Q12 = (4.181)
1 −ν 2
c
Q16 = 0 (4.182)
c E
Q 22 = (4.183)
1 −ν 2
c
Q 26 = 0 (4.184)
c
Q 66 = G (4.185)
where E, G, ν are, in order, the elasticity modulus, shear modulus and Poisson’s
ratio of steel.
With the aid of equations (4.177)–(4.179), the stiffness of the collar-pipe seg-
ment can be calculated by following the procedure outlined in Figure 4.3 and
using equation (4.16).
However, when the stiffness of a collar has a value that is much higher than
the value of the stiffness of the pipe, the inversion of the ABD matrix may yield
numerical errors. In such cases, since the stiffness of collar is dominant, the stiff-
ness of the pipe in the transfer matrix of the collar-pipe segment can be neglected.
For a better understanding of the above procedure, the critical velocity U cr will
be estimated for a cantilever pipe (Fig. 4.9) containing an additional mass located
at its end.
This model may represent the fallacious patent [8] for preventing torsional
buckling of drill-strings by the use of a floating drill-bit for oil exploration, rotat-
ing as a turbine under the action of the flow [2]. For this example the following
data will be used:
Orientation of fibers: ± ( π / 4)
Mechanical properties of ply:
E1 = 39 × 109 N / m 2
E2 = 8.6 × 109 N / m 2
ν 12 = 0.28
G12 = 3.8 × 109 N / m 2
Density of pipe: ρ p = 2.1× 103 Kg / m3
Density of fluid: ρ f = 103 Kg / m3
Inner diameter of pipe: 2r1 = 0.1 m
Exterior diameter of pipe: 2r2 = 2 ( r1 + NP ⋅ h ) = 0.103 m
Material of additional mass (collar): Steel
Mechanical properties of steel:
Es = 196 × 109 N / m 2
Gs = 73 × 109 N / m 2
ν s = 0.33
Density of steel: ρ s = 7800 Kg / m3
Exterior diameter of collar:
Length of the pipe: L12 = 1000 m
Length of the collar (in order to overcome the buoyancy force): L23 = 3.0 m
According to the procedure found in Fig. 4.3 and eq. (4.16), the calculation of the
value of the stiffness of the pipe (segment 1-2) yields:
� 12 =� EI yy = 39157 N ⋅ m 2
EI
Effect of Additional Mass 139
s
4
3 ( 2 )
� c = E π r 4 − r 4 � 2.31 × 109 N ⋅ m 2
EI
� yy /� EI c ≈ 0.00001 , along the segment 2-3 the stiffness of the collar dom-
Since EI
inates the bending behavior. Therefore, EI � 23 ≈� EI c .
0 1 0 0
0 0 1 0
[ AN ]12 =
0 0 0 1
−4
2.26 × 10 ⋅ ω
2
−4.01× 10 ⋅ U ω i −2.00 × 10−4 ⋅ U 2
−4
0
[ AN ]12 ⋅ L12
Therefore the transfer matrix for the segment 1-2 is [TM ]12 = e
2 2
( )
Mass per unit length of the collar: mc = π r3 − r2 ⋅ ρs = 11747 Kg / m .
0 1 0 0
0 0 1 0
[ AN ]23 =
0 0 0 1
−6
5.09 × 10 ⋅ ω
2
−6.8 × 10 ⋅ U ω i −3.4 × 10−9 ⋅ U 2
−9
0
[ AN ]23 ⋅ L23
and the corresponding transfer matrix is [TM ]23 = e
140 Dynamic Stability of Composite Pipelines
{ XN3 } = [TM 23 ]{ XN 2 }
{ XN 2 } = [TM 12 ]{ XN1}
where [ A21 ] and [ A22 ] are the transfer matrices incorporating the boundary condi-
tions for a cantilever pipe, given by equations (4.145), (4.146).
Because only a numerical solution for the above equation is possible, the soft-
ware Mathematica has been used to derive the value of ω for incremental values
of U. The critical value of U is the lowest value of U yielding a shift of Im {ω}
from a positive to a negative value. In the attached Mathematica code we started
by calculating the ω from a small initial value of U = 0.1 m/s. The value of ω is
calculated incrementally for the values U = 0.1, 0.11, 0.12, 0.13, 0.14, 0.15, 0.16,
0.17, 0.18, 0.19, 1.00. We found that Im {ω} moves from positive to negative
values where U cr = 0.9 m/s. This value corresponds to a critical fluid flow
Q = 3600 πr12U cr = 25 m3 / s .
Effects of an Elastic Foundation 141
4 2 2 2
� yy ∂ w + MU 2 ∂ w + 2 MU ∂ w + ( M + m) ∂ w + k w = 0
EI (4.186)
4 2
∂x ∂x ∂x∂t ∂t 2
Trying the solution given in eq. (4.69), the above equation yields:
� yy d f ( x ) + MU 2 d f ( x ) + 2 MU ω i df ( x ) + k − ( M + m ) ω 2 f ( x ) = 0
4 2
EI 4
dx dx 2 dx
(4.187)
Following the procedure for deriving the corresponding transfer matrix, the ma-
trix [ AN ] given in eq. (4.143) now takes the following form:
0 1 0 0
0 0 1 0
AN =
K
0 0 0 1 (4.188)
( M + m)ω − k
2 2
−2 MU ω i − MU
0
� yy
EI � EI yy � yy
EI
Therefore, the transfer matrix correlating the state vectors between two points i
and j of the pipe can now be written as:
TM ij = e ij
AN K ⋅ L
(4.189)
where Lij is the length between the nodes i and j. The equation providing the value
of ω versus U can now be written (see eq. (4.154)) as:
e AN K ⋅ Lij
[ − I 4 x 4 ]
det =0 (4.190)
[ A21 ] [ A22 ]
where the matrices [ A21 ] and [ A22 ] corresponding to the boundary conditions can
be obtained by eqs. (4.145)-(4.152). Here it should be recalled that critical flow
142 Dynamic Stability of Composite Pipelines
speed U cr is the lowest value of U yielding the shift of Im {ω} from a positive sign
to a negative one.
where L is the length of each span and [ AN ] is the matrix given by eq. (4.143).
Therefore:
y y
y y
1
y
= [ ]ij y1
TM j >i (4.192)
2 2
y3 y3
j i
y
y1
y2
y3 i
[TM ] [ − I 4 x 4 ] y = [0] (4.193)
ij
y1
y2
y
3 j
• •
Using the symbols ( L ) and ( R ) for the state variables (i.e., deflection w,
slope w’, bending moment M, shearing force Q) located just left and just right of
the node, the following compatibility equations and equilibrium conditions can
be written:
wL = 0 (4.194)
wR = 0 (4.195)
ML = MR (4.197)
Taking into account that M = EIw '' , the above equations can be written in the
following matrix form:
144 Dynamic Stability of Composite Pipelines
wL
w 'L
w ''L
1 0 0 0 0 0 0 0
0 0 0 0 1 0 0 0 L i
w '''
'
= [ 0] (4.198)
0 1 0 0 0 −1 0 0 wR
w 'R
0 0 1 0 0 0 −1 0
w ''R
w '''
R j
With the aid of equations (4.69) and (4.101)–(4.103), the above equation yields:
y
y1
1 0 0 0 0 0 0 0 y2
0
0 0 0 1 0 0 0 y3 L
= [ 0]
0 1 0 0 0 −1 0 0 (4.199)
y
0 0 1 0 0 0 −1 0 y1
y2
y
3 R
Taking into account the boundary conditions of the supports at the end points 1
and 5, the following matrix equations can be written:
y
1 0 0 0 y1
0 1 0 0 ⋅ y = 0 (4.200)
2
y3
1
y
1 0 0 0 y1
0 1 0 0 ⋅ y = 0 (4.201)
2
y3
5
Effect of Additional Supports 145
Applying equation (4.193) for all spans (i-j) namely (1-2), (2-3), (3-4), (4-5), and
equation (4.199) for all intermediate supports namely 2, 3, 4 and taking into ac-
count the boundary conditions (4.200), (4.201) of the end supports, the following
matrix equation can be derived:
where
{Y }1 = { y, y1 , y2 , y3 }1
T
(4.204)
{Y } j = { y R , y1R , y2R , y R T
}
R
3 j for j = 2, 3, 4 (4.206)
{Y }5 = { y, y1 , y3 , y4 }5 (4.207)
and
ϭ
ϭ
[70 ] [ − , × ]
ϭ
ϭ
ϭ Ͳϭ
ϭ Ͳϭ
[70 ] [ − , × ]
ϭ
[ × ]
* = ϭ
ϭ Ͳϭ
ϭ Ͳϭ
[70 ] [ − , × ]
ϭ
ϭ
ϭ Ͳϭ
ϭ Ͳϭ
[70 ] [ − , × ]
ϭ
ϭ
(4.208)
The above equation can be solved numerically yielding the values of ω versus
the values of U. It should be again noted that the value of U yielding the shift of
Im {ω} from a positive to a negative sign is critical.
4.9.1 Example
The fixed-fixed pipe containing three intermediate supports shown in Fig. 4.10
is modeled using the above procedure. The spans of this pipe are equidistant from
one another with length L12 = L23 = L34 = L45 = 10m . The interior diameter of the
pipe is 2r1 = 0.10m . The pipe wall is composed of Np = 50 layers of S-Glass/
Epoxy fiber reinforced composite. The thickness of each layer is h = 0.150 mm,
while the fiber orientation is θ = ±(π / 4) . The density of the composite material
9 3 3
is ρ p = 2.1× 10 kg / m , while the density of the fluid is ρ f = 1000kg / m . The
material properties in the principal directions of each ply are E1 = 39 × 10 N / m 2 ,
9
0 1 0 0
0 0 1 0
[ AN ] =
0 0 0 1
−5 2
6.524 × 10 ω −8.0051× 10 ωUi −4.0025 × 10−5 U 2
−5
0
Taking into account eq. (4.191), the transfer matrix [TM ]ij can now be derived
for every span. Using eq. (4.208) the matrix [G32×32 ] can be obtained yielding
eq. (4.209). Starting from a small initial value U = 1.0 m / s , which is changed
incrementally with step 1.0 m / s , the eigen-frequency ω has been calculat-
ed numerically by solving eq. (4.209) using Mathematica (see attached code).
The results indicated that for the values of U ≤ 70 m / s , Im {ω} ≈ 0 . For
U = 71 m / s the value of Im {ω} was positive (Im {ω} = +2.83007) , while for
the next increment of velocity (i.e. U = 72 m / s ) the Im {ω} became nega-
tive (Im {ω} = −4.02455) . Therefore the critical flow velocity of the sample
multi-supported pipe is U cr = 72 m / s . If the intermediate supports of the pipe
are removed, eq. (4.154) yields U cr = 28 m / s (see the attached computer code
in Mathematica). Therefore, the three intermediate supports have stabilized the
composite pipe by increasing the critical velocity almost 2.6 times.
Estimation of Critical Flow Velocity in Relation to Divergence 147
4 2
� yy d w + MU 2 d w = 0
EI (4.210)
dx 4 dx 2
The above equation can be modified in order to take into account thermal loads
(ΔΤ) or effect of elastic foundation, i.e.:
4 2
� yy d w + H (U , ∆Τ) d w = 0
EI (4.211)
4
dx dx 2
where
π ⋅ D ⋅ NP ⋅ h
H (U , ∆Τ) = MU 2 + (a1 cos 2 θ + a2 sin 2 θ )∆Τ (4.212)
S11
4 2
� yy d w + MU 2 d w + k w = 0
EI (4.213)
4
dx dx 2
where k is the elastic constant of the foundation.
In cases where the effects of thermal loads and the elastic foundation occur
simultaneously, the combination of eqs. (4.211) and (4.213) yields:
4 2
� yy d w + H (U , ∆Τ) d w + k w = 0
EI (4.214)
4
dx dx 2
148 Dynamic Stability of Composite Pipelines
Following the method of transfer matrices, the following form for the matrix [AN]
can be obtained for the case of divergence instability:
0 1 0 0
0 0 1 0
[ AN D ] = 0 0 0 1 (4.215)
−k 0
− H (U , ∆Τ)
0
EI
� �
EI
yy yy
Thus the transfer matrix correlating the state vectors {w,w’,w’’,w’’’}T of two suc-
cessive points i and j has the form:
TM ijD [ − I 4 x 4 ]
det = [ 0] (4.217)
[ A21 ] [ A22 ]
where [ I 4 x 4 ] is the 4 × 4 unit matrix and [ A21 ] , [ A22 ] are correlated with the
boundary conditions. For fixed-fixed, pinned-pinned and fixed-pinned pipe, the
matrices [ A21 ] , [ A22 ] are given by eqs. (4.146)–(4.152).
If the thermal load and elastic foundation effects are neglected, the solution of
eq. (4.217) for the cases of fixed-fixed, pinned-pinned and fixed-pinned pipes is
simple, i.e:
MU 2 2 κπ
= (4.218)
� yy
EI L
where k is an integer.
The lowest value of U corresponds to the lowest non-zero value of k, i.e.,
κ = 1. Therefore, for fixed-fixed pipe, the critical value of U is
Hydraulic Hammer 149
4π � yy
EI
U cr = (4.219)
L M
MU 2 κπ
= (4.220)
� yy
EI L
yielding
π � yy
EI
U cr = (4.221)
L M
MU 2 4.49
� (4.222)
� yy
EI L
yielding
� yy
4.49 EI
U cr = (4.223)
L M
The bulk modulus of elasticity E of the liquid can be derived by the relation
∆V
P=E (4.227)
Vbefore
where P is the pressure that causes the dilatation ΔV. With the aid of eqs. (4.224),
(4.225), the last equation yields
P(∆L + ∆L0 )
E= (4.228)
∆r
∆L − 2 ∆L0
r
Figure 4.12 Evolution of a liquid volume element after a sudden stoppage of flow.
Hydraulic Hammer 151
P ⋅ ∆L0
E= (4.229)
∆r
∆L − 2 ∆L0
r
F = m⋅a (4.230)
or
P(π r 2 ) = ρVa (4.231)
where ρ is the density of the liquid and V is the volume of the liquid element:
V = πr 2 ( ∆L + ∆L0 ) (4.232)
P
α= (4.233)
ρ (∆L + ∆L0 )
During the retardation, the center of gravity of the liquid element has traveled a
distance (ΔL/2). This distance is given by the well-known rule:
∆L U 2
= (4.234)
2 2a
where U is the liquid velocity. With the aid of eq. (4.232), the last equation yields:
U 2 ρ ( ∆L + ∆L0 )
∆L = (4.235)
P
Assuming that ∆L = U ∆t and ∆L + ∆L0 ≈ ∆L0 , the above equation can be written
as:
∆L0
P = ρU (4.236)
∆t
152 Dynamic Stability of Composite Pipelines
Taking into account the elastic properties of an anisotropic pipe’s wall, the pres-
sure P can be correlated with the dilatation ∆r / r . Considering the multi-layered
pipe of Fig. 4.13, the following known relations can be written:
1
Ny = PD (4.237)
2
ε y0 = a22 N y (4.238)
In the above equations, N y is the force per unit length acting in the y-direction
of the pipe’s wall, and ε y0 is the strain in the same direction in the middle of the
wall’s thickness. The parameter a 22 can be derived by the inversion of the ABD
matrix.
As the length of the perimeter of the pipe is 2πr , the strain ε y0 can also be
given by the formula:
∆ ( 2π r )
ε y0 = (4.239)
2π r
or
∆r
ε y0 = (4.2340)
r
∆r
= Pa22 r (4.241)
r
By writing eq. (4.228) in the form
∆r P∆L0
∆L − 2 ∆L0 = (4.242)
r E
and using eq. (4.241), the following formula can be written:
P∆L0
∆L − 2 Pa22 r ∆L0 = (4.243)
E
With the aid of eq. (4.236), the above equation yields:
∆L0 ρU ∆L
∆L − ρUa22 D∆L0 = ∆L0 0 (4.244)
∆t E ∆t
Dividing both parts of the above equation by ∆t the following is obtained:
2 2
∆L ∆L ρU ∆L0
− ρUa22 D 0 =
E ∆t
(4.245)
∆t ∆ t
∆L0 E/ρ
= (4.246)
∆t 1 + a22 DE
c 1
= (4.247)
c∗ 1 + a22 DE
It should be noted that the parameters c∗ and E are constants with known values
for most of the liquids (e.g., Table 4.2).
154 Dynamic Stability of Composite Pipelines
Table 4.2
Values of c∗ and E for widely used liquids [9].
Liquid c∗ ( m / s ) E ( N / m2 )
Water 1509 219 × 107
Benzene 1070 102 × 107
Crude oil 1402 175 × 107
Carbon tetrachloride 933 141× 107
Combining eqs. (4.236), (4.245), (4.246), the pressure change due to hydraulic
hammer can be estimated by using the following formula:
ρUc∗
P∗ = (4.248)
1 + a22 DE
Therefore, the total liquid pressure in the composite pipe will oscillate within the
range of:
Ptot = P0 ± P∗ (4.249)
The required time period T for the shock pressure to travel from, and back to, the
point where sudden stoppage of the flow occurred is
2L
T= (4.250)
c
where L is the length of the pipe to the point where a valve stopped the flow.
In cases where the time of closure of the valve is not zero but Tν , and in
specific case where Tv > T, the maximum overpressure can be determined by the
formula:
2L
P∗ = ρU (4.251)
Tν
When a main pipe 1 is branched into two branches, namely 2 and 3, the pressure
shock (due to hydraulic hammer) in each branch can be estimated by applying
the following formulas:
P3∗ = SP1∗
where
A1
2
C1
S= (4.253)
A1 A2 A3
+ +
C1 C2 C3
In the above equation, A1 , A2 , A3 are the areas of the cross-sections of the pipes 1,
2, 3, and C1 , C2 , C3 are the corresponding pressure speeds given by eq. (4.247).
∂N x
=0 (4.254)
∂x
∂
∂x
( RN xϕ + M xϕ ) = 0 (4.255)
Νϕ ∂2 M x ∂2 w
− − q z + ρ h =0 (4.256)
R ∂x 2 ∂t 2
156 Dynamic Stability of Composite Pipelines
Figure 4.14 Coordinate system for a composite pipe subjected to moving pressure shock
due to hydraulic hammer.
∂M x
Qx = (4.257)
∂x
∂M xϕ
Qϕ = (4.258)
∂x
In the above equations, ρ is the density of the composite material, and h is the
total thickness of the pipe’s wall.
The relationships between the displacements u , v, w on the directions
x, ϕ , z with the corresponding strains ε x0 , ε ϕ , γ xϕ and curvatures k x , kϕ , k xϕ in
0 0
∂u
ε x0 = (4.259)
∂x
w
ε ϕ0 = (4.260)
R
∂υ 0
γ x0ϕ = (4.261)
∂x
∂2 w
kx = − (4.262)
∂x 2
w
kϕ = − (4.263)
R2
2 ∂υ
k xϕ = − (4.264)
R ∂x
Wave Propagation Due to Hydraulic Hammer 157
We recall that the above strains and curvatures are correlated with the stress resul-
tants N x , Nϕ , N xϕ , M x , M ϕ , M xϕ by the well-known relation:
Β16 ε x
0
N x Α11 Α12 Α16 Β11 Β12
Nϕ Α12 Β 26 ε ϕ
0
Α 22 Α 26 Β12 Β22
N Α
xϕ 16 Α 26 Α66 Β16 Β 26 Β66 γ x0ϕ
= (4.265)
M
x 11 Β Β12 Β16 D11 D12 D16 k x
M ϕ Β12 Β 22 Β 26 D12 D22 D26 kϕ
M xϕ Β16 Β 26 Β66 D16 D26 D66 k
xϕ
where the members Aij , Bij , Dij , i = 1, 2, 6, j = 1, 2, 6 of the stiffness matrix are
given in Chapter 1.
Combining eqs. (4.254)–(4.265) and taking into account the symmetry
condition
υ =0 (4.266)
and the loading conditions
� x =0
N ( axial force ) (4.267)
T� = 0 ( torque ) (4.268)
∂4 w ∂2 w ∂2 w
f1 4
+ f 2 2 + f 3 w + ρ h 2 = qz (4.269)
∂x ∂x ∂t
f1 = H 22 (4.270)
1
f2 = ( H 21 + H12 ) (4.271)
R
H11
f3 = (4.272)
R2
where
H11 H12
= [ a1 ] − [ a4 ][ a3 ] [ a2 ]
−1
H
H 22
(4.273)
21
158 Dynamic Stability of Composite Pipelines
B22
A22 − R B12
[ a1 ] = (4.274)
B − D12 D11
12 R
B12
A12 − R B11
[ a2 ] = (4.275)
A − D26 D
B16 + 16
26 R 2 R
2 B16
A11 A16 −
R
[ a3 ] = (4.276)
A + B16 B 2D
A66 − 66 − 266
16 R R R
2 B26
A12 A26 −
R
[ a4 ] = (4.277)
B 2D
11 B16 − 16
R
The loading parameter qz represents the pressure acting on the wall’s surface.
Since the pressure shock due to hydraulic hammer is travelling with a constant
velocity c, its value in an arbitrary location x is given by the following equation:
qz = δ ( x − ct ) P∗ (4.278)
where δ is the Dirac delta function, t is time and P∗ is the value of the pressure
shock.
Taking into account eq. (4.278), eq. (4.269) can now be written:
∂4 w ∂2 w ∂2 w
f1 4
+ f 2 2 + f 3 w + ρ h 2 = δ ( x − ct ) P∗ (4.279)
∂x ∂x ∂t
Assuming that the ends of the shell are simply supported on their perimeters, the
above equation is associated with the following boundary conditions:
w ( 0, t ) = 0 (4.280)
Wave Propagation Due to Hydraulic Hammer 159
∂ 2 w ( x, t )
=0 (4.281)
∂x 2 x =0
w ( L, t ) = 0 (4.282)
∂ 2 w ( x, t )
=0 (4.283)
∂x 2 x=L
w ( x, 0 ) = 0 (4.284)
∂w ( x, t )
=0 (4.285)
∂t t =0
The partial differential equation (4.279) with the associated boundary and ini-
tial conditions given by eqs. (4.280)–(4.285) will be solved by using integral
transforms.
Taking into account the following relations of finite sine Fourier transform
[10]:
L
j πx
V ( j , t ) = ∫ w ( x, t ) sin dx j = 1, 2, 3,..... (4.286)
0
L
2 ∞ j πx
w ( x, t ) = ∑ V (i, t ) sin L (4.287)
L j =1
L
j πx j πct L
∫ δ ( x − ct ) P
∗
sin dx = P∗ sin for c〈 (4.288)
0
L L t
j 4π 4 j 2π 2 jπ ct
4
f1V ( j , t ) − 2
f 2V ( j , t ) + f 3V ( j , t ) + ρ hV ( j , t ) = P∗ sin (4.289)
L L L
160 Dynamic Stability of Composite Pipelines
j 4π 4 f1
ω(2j ) = (4.290)
L4 ρ h
ρ h f3
Ω(2 j ) = ω(2j ) − ω( j ) f 2 + (4.291)
f1 ρ h
πc
ω= (4.292)
L
eq. (4.289) can be written:
P∗
V ( j , t ) + Ω(2 j )V ( j , t ) = sin ( jωt ) (4.293)
ρh
Inserting the Laplace transform
V ∗ ( j , ξ ) = L {V ( j , t ) ; ξ } (4.294)
P∗ jω ξ 1
V ∗ ( j, ξ ) = (4.295)
(
ρ h ( ξ 2 + j 2ω 2 ) ξ 2 + Ω 2 j
( ) )
Taking the inverse Laplace transform [10] of the above equation, the following
solution can be derived:
jπ x
∞ sin
w ( x, t ) = w0 ∑ L j 2 ( j 2 − a 2 ) sin jωt − ja ( j 2 − a 2 ) sin Ω( j ) t (4.296)
2 2
j =1 j ( j −a )
2 2
and taking into account the period T(1) of the first free vibration
2π
T(1) = (4.299)
Ω(1)
as well as the time T for the shock pressure to travel over the pipe
L
T= (4.300)
c
the following relations can be obtained:
π
ω= (4.301)
T
2π
Ω(1) = (4.302)
T(1)
With the aid of eqs. (4.298)–(4.302), eq. (4.297) can be written in the following
form:
T(1) 2 π / Ω(1) c
a= = =
(Ω )
(4.303)
2T 2L / c
(1) L π
yielding
c
a= (4.304)
ccr
4.12.1 Example
For a multi-layered pipe made by E-glass/epoxy we shall derive the curves
w ( x, t ) for t = 0,1, 2,.....,15 sec.
Taking into account the mechanical properties of the composite material
E1 = 39 × 109 N m2
E2 = 8.6 × 109 N m2
ν 12 = 0.28
G12 = 3.8 × 109 N m 2
162 Dynamic Stability of Composite Pipelines
θ = π 12 (fiber orientation)
NP = 10 (number of plies)
h = 0.150 × 10−3 m (thickness of ply)
Taking into account the member a22 = 3.02085 × 10−7 of the above derived matrix
[ abd ] , and assuming the bulk modulus of elasticity∗ E = 219 × 107 N/m2 for water,
the pipe’s diameter D = 0.10 m and the celerity c = 1509 m s of the water, the
velocity of the pressure shock can be obtained by the equation:
c∗
c= = 184 m s
1 + a22 DE
For fluid velocity U = 3 m s , the pressure shock for instantaneous closure of the
valve is P∗ = ρUc = 1000 × 3 × 184 = 552000 Pa above the static pressure.
Wave Propagation Due to Hydraulic Hammer 163
With the aid of the results above, the following parameters of the governing equa-
tion of wave propagation (eq. 4.279) can be derived:
H11 H12
= [ a1 ] − [ a4 ][ a3 ] [ a2 ]
−1
H
H 22
21
or
f1 = H 22 = 15.5797
1
f2 = ( H 21 + H12 ) = 2060.8
R
H11
f3 = = 5.21754 × 109
R2
164 Dynamic Stability of Composite Pipelines
,
Wave Propagation Due to Hydraulic Hammer 165
,
166 Dynamic Stability of Composite Pipelines
,
Wave Propagation Due to Hydraulic Hammer 167
,
168 Dynamic Stability of Composite Pipelines
References
[1] Ashley H., and Haviland G., “Bending vibrations of a pipeline containing
flowing fluid,” Journal of Applied Mechanics 17, 229–232, 1950.
[2] Païdoussis M.P., Fluid-Structure Interactions, Slender Structures and Axial
Flow, Vol. 1, Academic Press, 1998.
[3] Païdoussis M.P., Price S.J., and de Langre E., Fluid-Structure Interactions,
Cross-Flow-Induced instabilities, Cambridge University Press, 2011.
[4] Kollár L.P., and Springer G.S., Mechanics of Composite Structures,
Cambridge University Press, 2009.
[5] Hyer M.W., Stress analysis of fiber-reinforced composite materials,
DEStech Publications, 2009.
[6] Païdoussis M.P., and Luu T.P., “Dynamic of a pipe aspirating fluid such
as might be used in ocean mining.” ASME Journal of Energy Resources
Technology 107, 250–255, 1985.
[7] Deif A.S., Advanced matrix theory for scientists and engineers, Abacus
Press, 1991.
[8] Den Hartog J.P., Mechanical Vibrations, 4th edition, Mc Graw Hill, N.Y.
1956.
[9] Simon A.L., Hydraulics, John Wiley & Sons, 1986.
[10] Frýba L., Vibration of solids and structures under moving loads,
ThomasTelford, 1999.
[11] Hoskins R.F., Delta Functions, Horwood Publishing, 1999.
Chapter 5
169
170 Connection and Supports of Composite Pipelines
Accurate determination of the stresses within the adhesive would require com-
plex elasto-plastic numerical modeling or experimental investigation. Since such
procedures are not efficient for engineering design [2], analytical models for es-
timating the allowable axial force and bending or torsional moment can be used
for design purposes.
The length of the initial non-deformed adhesive element ΑΒΓΔ is dx, while its
width is equal to the thickness ta of the adhesive layer. The axial force N acting
upon the strips 1 and 2 produces shear strain γ and normal mean strains ε1o and
ε 2o in the cross- section of each thin strip. Due to shear strain γ, the points Δ and Γ
will be moved to the new locations Δ1 and Γ1 respectively. Therefore, the element
ΑΒΓΔ will be transformed into the deformed element ΑΒΓ1Δ1 and the segment
Γ1∆1 will keep its length dx. Following the shear deformation, the normal strains
ε1o and ε 2o are assumed to act on the segments ∆1Γ1 and AB respectively, yield-
ing the corresponding elongation ε1dx and ε 2 dx . Therefore, the final location of
the points Γ and B will be Γ4B1 respectively.
If we draw the fictitious lines Β1Γ 2 and Β1Γ3 , the following geometric prop-
erties can be written:
Γ2Γ4
γ + dγ = (5.1)
Β1Γ 2
or
∆Γ 4 − ΑΒ1
γ + dγ = (5.2)
Β1Γ 2
Taking into account the definition of strain, the above equation can be written as:
(γ ta + dx + ε1dx ) − ( dx + ε 2 dx )
γ + dγ = (5.3)
ta
172 Connection and Supports of Composite Pipelines
or
ε1 − ε 2
dγ = dx (5.4)
ta
Taking the derivative with respect to x in both sides of the above equation, it can
be determined that:
d 2γ 1 d ε1 d ε 2
− − =0
dx 2 ta dx dx
(5.6)
Denoting by Ga the shear modulus of the adhesive, and using Hooke’s law
τ = Ga γ (5.7)
d 2τ Ga d ε1 d ε 2
− − =0
dx 2 ta dx dx
(5.8)
E1
Ex = (5.9)
E E
cos 4 θ + 1 − 2ν 12 sin 2 θ cos 2 θ + 1 sin 4 θ
G12 E2
where θ is the fiber orientation and E1 , E2 , G12 , ν 12 are the elasticity properties
along the principal axes, the stress in the x-direction (of each lamina) can be given
by Hooke’s law in a global coordinate system:
σ x = Ex εx (5.10)
σ x = NP Ex εx (5.11)
Therefore, the equilibrium equation of the strip element 1 (Fig. 5.5) can be writ-
ten as:
Ex ( ε1 + d ε1 ) NP s t − τ s dx − Ex ε1 NP s t = 0 (5.12)
which gives
d ε1 1
= τ (5.13)
dx Ex NP t
dε 2 1
=− τ (5.14)
dx Ex NP t
With the aid of equations (5.13) and (5.14), equation (5.8) can now be written:
d 2τ
− λ 2τ = 0 (5.15)
dx 2
174 Connection and Supports of Composite Pipelines
where
2 Ga
λ2 = (5.16)
Ex NP t ta
ε 2 ( L / 2) = 0 (5.18)
σx
ε 1 ( L / 2) = (5.19)
Ex NP
Since
∧
N
σx = (5.21)
π Dt
∧
where N is the axial force acting on a pipeline, eq. (5.19) can be written as:
∧
N
ε 1 ( L / 2) = (5.22)
π D t E X NP
dγ
=0 (5.23)
dx x = 0
Combining the above equation with equation (5.7) it can be determined that:
dτ
=0 (5.24)
dx x = 0
Joining of Composite Pipelines 175
Therefore, eq. (5.17) with the aid of the equation above yields:
A2 = 0 (5.25)
Thus, the general solution for shear stress distribution along the adhesive can now
be simplified:
τ ( x) = Α1 cosh(λ x) (5.26)
Combining equation (5.5) with equations (5.18) and (5.22) can be written as:
∧
dγ
=
1 N − 0 (5.27)
dx x=L/2 ta π D t Ex NP
The above equation with the aid of eq. (5.7) yields
dτ G
= a⋅ N (5.28)
dx x=L/2 ta π D t Ex NP
With the help of the above equation, the unknown constant A1 can now be deter-
mined by eq. (5.26):
∧
Ga N
A1 = (5.29)
ta π D t Ex NP λ sinh(λ L / 2)
Therefore, the distribution of shear stress along the length of the adhesive can be
approximated by the formula:
∧
Ga N
τ ( x) = cosh(λ x) (5.30)
ta π D t Ex NP λ sinh(λ L / 2)
τ max = τ ( L / 2) (5.31)
176 Connection and Supports of Composite Pipelines
∧
Ga N
τ max = (5.32)
ta π D t Ex NP λ tanh(λ L / 2)
∧
N
τm = (5.33)
π DL
Combining eqs. (5.32), (5.33), the stress concentration factor K h at the ends of
the adhesive layer is
τ max
Kh = (5.34)
τm
or
Ga L
Kh = (5.35)
ta t Ex NP λ tanh(λ L / 2)
Joining of Composite Pipelines 177
τ max = τ α (5.36)
where τ α is the allowable shear stress of the adhesive, the allowable axial force
∧
N a for the joint can be approximated by combining equations (5.32) and (5.36):
∧
τα
Na = ta π D t Ex NP λ tanh(λ L / 2) (5.37)
Ga
tan(λ L / 2) → 1 (5.38)
τα
Fmax → ta π D t Ex NP λ (5.39)
Ga
Results for the allowable axial force of butt joints for a wide range of multi-
layered filament wound pipes made from E-glass/epoxy and S-glass/epoxy are
presented in Chapter 10.
Figure 5.7 Capacity of a joint to bear axial force versus adhesive length.
178 Connection and Supports of Composite Pipelines
�
N M� D
α11 = (5.40)
π D EI
� 2
where
π
� = 2 D + 1 cos 2 θ D dθ
2
EI ∫0 4a11 d11 2
(5.41)
Therefore,
� = 4 πD 2 d11 �
N M (5.42)
d11 D 2 + 4a11
With the aid of the above equation, eq. (5.37) provides the approximate value of
the allowable bending moment M �
a :
Figure 5.8 Approximation of the maximum local strain caused by a bending moment M,
with the local strain caused by an axial force N.
Above-Ground Pipes 179
D a11 τ α
�
M a = + ta t Ex NP λ tanh ( λ L 2 ) (5.43)
4 d11 D Ga
Results for the allowable bending moment of butt joints for a wide range of mul-
tilayered filament-wound pipes made from E-glass/epoxy and S-glass/epoxy are
displayed in the nomographs in Chapter 10.
where λ is a coefficient given in Table 5.1, and w is the uniformly distributed load
per longitudinal unit. w depends on the specific gravities γ f , γ p and material
volumes V f , V p of the fluid and pipe respectively. Therefore,
π D2
w=π Dt γp + γf (5.45)
4
where t and D are the thickness of the pipe’s wall and the mean diameter,
respectively.
Combining equations (5.44) and (5.45), the maximum length between supports
can be estimated by the following rule:
M max
L≤ (5.46)
λπ D ( tγ p + 0.25 Dγ f )
In the above equation, the value of M max should be the minimum value between
the allowable bending moment M a (for avoiding failure) and the critical moment
M cr (for avoiding buckling) derived in Chapter 3.
According to Table 5.1 it can be shown that the value of λ tends to be stable for
more than six supports. The maximum value λ = 0.125 occurs in the span of a pipe
supported by two supports or over the intermediate support for a pipe resting on
three supports. In practice, it is prudent and safe to adopt this maximum value for
every case of pinned-supported pipeline. Taking into account equations (5.46),
(5.47) and the diagrams demonstrating the allowable and the critical value of the
bending moment for pipes under bending made of E-glass/epoxy and S-glass/
epoxy, results have been calculated that estimate the maximum spacing Lmax be-
tween supports for a wide range of multi-layered filament-wound pipelines. These
182 Connection and Supports of Composite Pipelines
Table 5.1
Values of coefficient λ for continuous pipeline over equal spans (Fig.5.12).
Number of supports Coefficient λ corresponding Coefficient λ corresponding
to maximum bending to maximum bending
moment over support moment in span
2 0 0.125
3 0.125 0.703
4 0.100 0.080
5 0.107 0.077
6 0.105 0.078
7 0.105 0.078
8 0.105 0.078
results are presented in Chapter 10. For the derivation of the above results, the
4 3
usual values γ p = 104 N / m3 and γ f = 1.3 x 10 N / m were adopted for the pipes’
material and the fluid, respectively.
�
M �
c ≈ 0.24QR (5.48)
�
and a compressive force N c
� ≈ 0.24Q
N � (5.49)
c
Figure 5.13 (a) Simulation of a pipe segment inside a hanger using a thin-walled ring,
�
(b) Bending moment M �
c and axial force N c due to Q.
Figure 5.14 shows the loading conditions of a pipe’s wall element located at
point C due to shear force Q � and bending moment M �
y .
Taking into account that the width of the element is denoted by B, the unit
forces M c , N c , Nξ can be obtained with the help of eqs. (5.48), (5.49), (5.42),
and (5.44):
�
0.24QR
Mc = (5.50)
B
�
0.24Q
Nc = (5.51)
B
4 Dd11 M� y
Nξ = 2
(5.52)
d11 D + 4a11 B
In the above equation the value of M � y can be substituted by the value obtained
�
by eq. (5.44). The shear force Q for every pipe-cross section, left or right of each
support in a continuous pipeline, as shown in Figure 5.12, is given by the formula:
� = µwL
Q (5.53)
184 Connection and Supports of Composite Pipelines
Figure 5.14 Loading conditions of a pipe’s wall element located at point C (see Figure 5.13).
0.145wLR
Mc = (5.54)
B
0.145wL
Nc = (5.55)
B
Table 5.2
Values of the coefficient µ for a continuous pipeline over
equal spans, as depicted in Figure 5.12.
Number of supports Coefficient µ corresponding to maximum shear force
on each side of supports L =left, R =right
L R
2 0 0.5
3 0.625 0.625
4 0.600 0.500
5 0.607 0.536
6 0.605 0.526
7 0.605 0.029
8 0.605 0.529
Above-Ground Pipes 185
With the aid of eq. (5.44) and taking into account the value λ = 0.125, eq. (5.52)
yields:
Combining the laminate compliance matrix (“abd” matrix) with eqs. (5.54),
(5.55), and (5.56) the following matrix equation can be obtained:
(5.57)
ε x0
0 b12 a12 a11
ε
y b a12
22 a22
0 M c
γ
xy b62 a26 a16
0 = − Nc (5.58)
k x d12 b21 b11 N
k 0 d b22 b12 ξ
y 22
k 0 d 26 b26 b16
xy
Therefore, on the exterior layers of the laminate, the strain components εx, εy, γxy
can be obtained from the following equation:
186 Connection and Supports of Composite Pipelines
ε x ε x0 k x0
0 0
ε y = ε y ± z k y (5.59)
0 0
γ xy γ xy k xy
σ x εx
σ y = Q ij (θ ) ε y (5.60)
τ γ
xy xy
Using the matrix [ Τ(θ ) ] , the principal stresses σ 1 , σ 2 ,τ 12 in the exterior layers
can be determined as follows:
σ 1 σ x
σ 2 = T (θ ) σ y (5.61)
τ τ
12 xy
With the aid of the above equation, the Tsai-Wu failure criterion can provide an
estimation of the minimum hanger width B. Results of the hanger width B for a
wide range of multi-layered filament wound pipes made of E-glass/epoxy and
S-glass/epoxy are presented in Chapter 10.
In the model shown in Figure 5.16, the left anchor (point A) and the point
located in the middle of the horizontal part of the expansion loop (point D) are
simulated by fixed supports. The point B joining the pipeline with the expansion
loop is guided horizontally. The aim of this analysis is to estimate the values of
the compressive axial force F in the pipe AB and the bending moment Μ Β at the
point B (critical point) due to a temperature increase of ΔΤ. These values will be
used for checking the tolerance of the structure to failure and buckling. In order
to estimate the axial force F, the L-shaped part of the extension loop (part BCD)
will be simulated by a linear spring (Fig. 5.17).
The equivalent spring constant k can be estimated by the ratio of the horizontal
force acting on point B of the expansion loop (Fig. 5.16) over its corresponding
horizontal movement, thus: ∆xB
F
k= (5.62)
∆xB
188 Connection and Supports of Composite Pipelines
Figure 5.17 Simulation of the expansion loop by a linear spring. (a) Structure when
∆Τ = 0 ; (b) Structure after temperature increase ∆Τ ≠ 0 ; (c) Deflection when the end
B is free.
According to [5], the horizontal movement ∆xB of the point B in the L-shaped
frame BCD (Fig. 5.16) can be obtained by the formula:
where VB , M B are the vertical reaction and bending moment acting in the support
B (Fig. 5.16). The parameters CHV , CHM , LFH are given by [5]:
ab 2
CHV = (5.64)
2 EI
a2 ab
CHM = + (5.65)
2 EI EI
LFH = FCHH (5.66)
where
a3 a 2b
CHH = + (5.67)
3EI EI
Above-Ground Pipes 189
3
D
I = π t (5.68)
2
where t, D are the thickness and the diameter of the pipe.
The reactions VB , M B can be obtained also from [5]:
where
ab 2
CVH = (5.73)
2 EI
a2 ab
CMH = + (5.74)
2 EI EI
a+b
CMM = (5.75)
EI
b2
CVM = (5.76)
2 EI
b3
CVV = (5.77)
3EI
Combining equations (5.69)–(5.72), one can arrive at:
With the aid of eqs. (5.66), (5.78), (5.79), eq. (5.63) yields:
Combining the above equation with eq. (5.62), the following formula, which pro-
vides the spring constant k, can be obtained:
According to Fig. 5.17c, for the case of free expansion of pipe due to a temperature
rise ΔΤ, the movement e1 + e2 of the end B can be approximated by the formula:
e1 + e2 = a x L ∆Τ (5.82)
where a x is the longitudinal (i.e., along the pipe’s axis) coefficient of thermal
deformation. a x can be approximated by the following formula:
e1 = ∆xB (5.84)
F = k ⋅ e1 (5.85)
where the constant k is given by the eq. (5.81). On the other hand, due to the axial
force F, the longitudinal deflection of the pipe (Fig. 17b) can be approximated by
the formula:
Ε⋅Α
F= e2 (5.86)
L
Above-Ground Pipes 191
where E can be approximately substituted with Ex, which is given by eq. (5.9), and
A is the pipe wall’s cross-section:
Α = πDt (5.87)
Lk
e2 = e1 (5.88)
EA
With the help of the above equation, eq. (5.82) can now be written:
E ⋅ A ⋅ α x ⋅ L ⋅ ∆Τ
e1 = (5.89)
E⋅ A+ k⋅L
Taking into account the above equation, the axial force F can now be estimated
by eq. (5.85)
E ⋅ A ⋅ α x ⋅ L ⋅ ∆Τ
F=k (5.90)
E⋅ A+ k⋅L
Combining the above equation with eq. (5.79), the value of bending moment MB
of the point B joining the pipe with the expansion loop can now be estimated as:
Using the values F, MB, the following conditions must be satisfied for an extension
loop to be acceptable:
F < Na (5.92)
MB < M a (5.94)
M B < M cr (5.95)
In the above equations, N a is the allowable axial force to avoid failure (Chapter 3),
λcr is the minimum eigenvalue (Chapter 3) in order to avoid buckling, M a is the
allowable bending moment to avoid failure (Chapter 3) and M cr is the critical
value of a bending moment to avoid buckling.
192 Connection and Supports of Composite Pipelines
3F
σz = (5.96)
2π z 2
According to Burland et al. [7], the above solution accurately describes the stress
s z for most soil conditions.
Assuming that the pipe segment influenced by the surface load F is a ring of
radius R and width b, the mechanical model shown in Fig. 5.19 can be safely
used for estimating the minimum installation depth zmin . In this model q is the
unit load given by:
q = bs z (5.97)
With the aid of eq. (5.96) the above equation can be written as:
3bF
q= (5.98)
2 πz 2
Figure 5.19 Mechanical model of the loading of an underground pipe embedded in instal-
lation depth z.
194 Connection and Supports of Composite Pipelines
� c = 1 q R2
M (5.99)
4
� c =q R
N (5.100)
3 R2
My = F (5.101)
8π z 2
3 R
Ny = F (5.102)
2π z 2
Combining the laminate compliance matrix with equations (5.101) and (5.102),
the following matrix equation can be obtained:
½
ª D D D E E E − º 1[ ½ ° °
«D » °1 ° ° °
D D E E E − » ° \° ° °
«
« D D D E E E − » ° 1 [\ ° °
°
« » ° ° ° °
« E E E G G G − » ° 0[ ° ° °
«E E E G G G − » °0\ ° °
°
« » ° ° ° °
« E E E G G G −» ° 0 [\ ° ° °
« »
® ¾= ® ¾
«
» ° ε[ ° ° 5 °
« » ° ε \ ° ° − π ] ) °
« ° ° ° °
» ° γ [\ ° ° °
« »
« » ° N [ ° ° °
« » ° ° ° °
« » ° N\ ° ° 5
°
° ° ° − ) °
«¬ ¼ » N
¯ [\ ¿ ° π ]
°
¯ ¿
(5.103)
Underground Pipelines 195
Figure 5.20 Bending moment and axial force acting on the longitudinal cross-section of
a composite pipe.
ε x0 b12 a12
0 b
ε y 22 a22 3 R2
γ xy0 b62 a26 F
8 z2
0 = − (5.104)
k
x d12 b21 3 R F
ky 0 d 22 b22 2π z 2
0
k xy d 26 b26
Using eq. (5.104) the strain components ε x , ε y , γ xy of the exterior layer of the
composite wall can be obtained by the following equation:
ε x ε y k x0
0
0 h 0
ε y = ε y ± k y (5.105)
0 2 0
γ xy γ xy k y
σ x ε x
σ y = Q ij (ϑ ) ε y (5.106)
τ xy γ xy
and the principal stresses σ 1 , σ 2 ,τ 12 can be obtained with the aid of matrix
Τ (ϑ )
σ 1 σ x
σ 2 = Τ (ϑ ) σ y (5.107)
τ
12 τ xy
Using the above equation, the Tsai-Wu failure criterion can provide an estimation of
the minimum installation depth zmin versus the wheel and the soil load F , the pipe’s
diameter Dia , the fiber orientation ϑ , and the number of plies NP . Results for
a wide range of multilayered filament-wound pipes made of E-glass/epoxy and
S-glass/epoxy are displayed in Chapter 10.
References
[1] Mallinson J.H. Corrosion-resistant plastic composites in chemical plant de-
sign, Marcel Dekker, 1988.
[2] Wang C.H., and Rose L.R.F., “Stress analysis and failure assessment of lap
joints” in: Recent Advances in Structural Joints and Repairs for Composite
Materials, edited by Tong L., and Soutis C., Kluwer Academic Publishers,
2010.
[3] Kollár L.P., and Springer G.S., Mechanics of Composite Structures,
Cambridge University Press, 2003.
[4] Vidosic J.P., “Mechanics of materials” in: Marks’ Standard Handbook for
Mechanical Engineers, 8th edition, edited by Baumeister Th. et al., McGraw-
Hill, 1979.
[5] Young W.C., ROARK’S Formulas for Stress & Strain, McGraw-Hill, 1989.
[6] Boussinesq J., Application des potentials_lY_tude de lY_quilibre et du
movement des solids elastiques, Gauthier-Villars, Paris, 1885.
[7] Burland J.B., Broms B.B., and de Mello V.F.B., “Behaviour of founda-
tions and structures,” Proc. 9th Int. Conf. Soil Mechanics and Foundation
Engineering, Tokyo, 1978.
Chapter 6
6.1 Introduction
The application of glass fiber-reinforced thermoset matrix composites as pri-
mary structures has been impeded by a number of problems. Ochoa and Salama
[1] pointed out five basic reasons, which include two very difficult challenges:
the need to have databases for long-term damage mechanisms for lifetime predic-
tion and the need for nondestructive evaluation (NDE) and in-service monitoring.
Both are interrelated in terms of simulating service loads, manufacturing, and
installation procedures. The proper environmental and loading conditions used to
accelerate long-term testing are not fully known in advance. To be representative,
they must generate damage evolution and failure modes similar those occurring in
real conditions over 20 to 30 years. The heterogeneity and anisotropy introduced
by the fiber and matrix make such an analysis very complicated. Moreover any
change in the manufacturing procedure, matrix or fiber, produces a different mate-
rial system, which in turn invalidates the use of previous experimental databases.
Pipelines, risers and piping systems are examples of primary applications of
filament-wound composites in the oil and gas industry. The thermosetting resins
commonly used are epoxy and polyester resin systems. The advantages are well
known: high stiffness-to-weight ratio and corrosion resistance.
The design of these structures is highly demanding, since they are expected
to remain in service for more than 20 years. During service, pipe structures may
197
198 Creep Design of Piping Applications Using Composite Materials
be subjected to both permanent and cycling loads. Although they do offer high
mechanical performance, over time their strength and stiffness may decay sig-
nificantly, a consequence of the viscoelastic nature of the matrix, damage accu-
mulation within the matrix and fiber breakage. One serious consequence of these
events is premature failure, usually catastrophic.
The lack of full understanding of the fundamental parameters controlling long-
term materials’ performance necessarily leads to over-design. In this context, the
lifetime prediction of these structures is an important issue not completely solved.
Standards for certification of glass reinforced plastic (GRP) pipes require at least
10,000 hours of testing and a high number of specimens. Even though these strin-
gent requirements may be seen as reasonable in terms of safety, they severely re-
strict the improvement and innovation of new products, which in turn may inhibit
greater commercial penetration in the markets.
The present chapter will present the following argument. The second sec-
tion reviews selected theoretical approaches for long-term failure criteria. Time-
dependent failure criteria are presented and developed in a manner useful for
practical engineering applications. The third section attempts to couple damage
with creep and fatigue effects on a long-term basis, based on safety factors. It
starts with a brief theoretical description of damage initiation and propagation
during static loading. Long-term sustained and cycling loads usually produce
damage similar to that originated under quasi-static loading. Experimental creep
failure data under sustained hydrostatic pressure are presented, followed by an
example of preliminary design for long-term creep. The fourth section is devoted
to fatigue failure. Experimental data published for filament-wound pipe failure
under cycling hydrostatic pressure are presented and discussed. The fifth section
describes and compares the standards for the design and qualification of pipes.
Finally, the sixth section presents a worked example with preliminary calculations
of qualification pressure for a filament-wound pipe, using known material proper-
ties, based on ISO 14692.
Failure
Region I
Time
figure 6.1 Typical creep strain evolution divided into three regions.
t d ( g 2σ (τ ) )
ε ( t ) = J 0 g 0σ ( t ) + g1 ∫ ∆J (ψ −ψ ′ ) dτ (6.1)
0 dτ
200 Creep Design of Piping Applications Using Composite Materials
tdτ ′ τ dτ ′
Ψ=∫ , Ψ′ = ∫ (6.2)
0 a 0 aσ
σ
ε ( t ) = ε ( 0 ) + g (σ ) t n (6.4)
t t − τ ∂σ (τ )
n
ε ( t ) = D0σ ( t ) + D1 ∫ dτ (6.5)
τ 0 ∂τ
0
where D0, D1, n are material constants and t0 represents the time unity (equal to
1 second or 1 hour or 1 day, etc.).
The free stored energy, using the Hunter [8] formulation, is given by:
1 t t 2t − τ 1 − τ 2 ∂σ (τ 1 ) ∂σ (τ 2 )
n
1
Ws ( t ) = ε ( t ) σ ( t ) − D0σ ( t ) − ∫ ∫ D1 dτ 1dτ 2
2
2 2 0 0 τ0 ∂τ 1 ∂τ 2
. (6.6)
t ∂ε (τ )
Wt ( t ) = ∫ σ (τ ) dτ . (6.7)
0 ∂τ
Accordingly, these time-dependent failure criteria [9] predict the lifetime under
constant load as a function of the applied load s0 and the strength under an instan-
taneous condition sR:
D
The Reiner-Weissenberg Criterion (R-W) states that Ws ( t ) ≤ 0 s R 2 ,
2
1 1
t f 1 n D0 n 1 1
= − 1
n
n
. (6.8)
τ 0 2 − 2 D1 γ
D0 2
The Maximum Work Stress Criterion (MWS) states that Wt ( t ) ≤ sR ,
2
1
tf 1 D0 n 1 1
= − 1
n
. (6.9)
τ0 2 D1 γ
1
t f D0 n 1 1
= − 1 n
. (6.10)
τ 0 D1 γ
202 Creep Design of Piping Applications Using Composite Materials
D
Wt ( t ) ≤ 0 s R s ( t )
2 ,
1 1
tf 1 n D0 n 1 1
= − 1 n
n
. (6.11)
τ0 2 − 2 D1 γ
where γ = σ 0 2 σ R 2 .
In short, these approaches establish a relationship between time to failure, vis-
coelastic properties and static stress failure throughout a stored elastic energy
limit concept. As an approximation, it is not difficult to conclude that we can take
t~σ−2/n for the R–W and the MSW criteria and t~σ−1/n for the MS and the MR-W
criteria. Similar results were obtained using fracture mechanics concepts [10]. In
fact, these concepts established a relationship between time-to-failure, viscoelas-
tic properties and strength properties [11-12], which are similar to the previous
approach. The main difference in these failure criteria is the interpretation of the
physical constants. According to Song et al. [13], there are three major phenom-
ena, which frequently occur simultaneously, responsible for the creep failure of
viscoelastic materials: (1) time-dependent constitutive equations; (2) time to the
formation of overstressed polymer chains in a localized plastic area, i.e., a fracture
initiation mechanism; and (3) the kinetics of molecular flow and bond rupture
of the overstressed polymer chains. The fracture mechanics approach assumes
the existence of defects from the start and develops a theory about the kinetic
crack growth, i.e., the fracture initiation process is neglected. In the previous ap-
proach, the stored energy in the material, i.e., the energy stored by all springs in
the viscoelastic model, can be compared with the energy necessary to stretch the
polymer chains and promote their bond rupture. In fact, it is possible to visualize
the polymer chains as linear springs acting as energy accumulators. Nevertheless,
these energy accumulators have a limited capacity above which bond rupture oc-
curs. Therefore the stored energy limit, called critical energy, can be related to the
energy involved in all microscale bond ruptures that lead to creep-rupture. Most
probably this critical energy depends on the internal state. In reality, there are ex-
perimental indications that the critical energy is temperature- and strain-rate- de-
pendent [14], at least for temperatures lower than the glass transition temperature
Tg (or for shorter times). This means that the R–W criterion is not universal. On
the other hand, there is experimental evidence, for polymers and composite poly-
mers, that change in the fracture mode is a result of change in critical energy with
temperature and strain rate [14]. Finally, it is not difficult to accept that creep-
rupture is strongly related to creep compliance or relaxation modulus. This rela-
tionship emerges naturally from theoretical approaches like fracture mechanics
Creep Damage Accumulation Mechanisms in Composite Materials 203
U
t f = t0 exp 0 − γσ , (6.13)
kT
known as the modified rate equation, for predicting the time to rupture of continu-
ous fiber-reinforced plastics with reasonable success.
The fracture mechanics analysis was extended to viscoelastic media to predict
the time-dependent growth of flaws or cracks. Several authors produced extensive
work in this area [11,12,17–20]. Schapery [19,20] developed a theory of crack
growth that was used to predict the crack speed and lifetime for an elastomer
under uniaxial and biaxial stress states. For a centrally cracked viscoelastic plate
with a creep compliance given by eq. (6.5) under constant load, Schapery [11]
deduced, after some simplifications, a simple relationship between stress and time
to failure:
tf −2 (1+1 n )
= ( Bσ 0 ) , (6.14)
τ0
where n is the exponent of the creep compliance power law and B a parameter
that depends on the geometry and properties of the material. Leon and Weitsman
[21] and Corum et al. [22] used this approach where B was considered an experi-
mental constant, to fit creep-rupture data with considerable success. Christensen
[23] developed a kinetic crack formulation to predict the creep rupture lifetime
for polymers. The lifetime was determined from the time required for an initial
204 Creep Design of Piping Applications Using Composite Materials
crack to grow sufficiently large to cause failure. The method assumed quasi-static
conditions and applies only to the central crack problem. The polymeric material
was taken to be in the glassy elastic state, as would be normal in most applica-
tions. For general stress, s ( t ) we have:
(ατ 0 )
1 − (σ ( t ) σ R ) (σ (τ ) σ )
1m tf 1 m +1
=∫ R dτ , (6.15)
0
tf α 1
= 1 m − 1 . (6.16)
τ0 γ γ
σ
σ = , (6.17)
1− ω
where 0 ≤ ω ≤ 1 is the damage variable. At rupture, no load-bearing area remains,
and the net stress tends to infinity when ω → 1.
Kachanov [24] assumes the following damage growth law:
ν
σ (t )
ω ( t ) = C , (6.18)
1 − ω (t )
where C and ν are material constants. This equation leads to a separable differ-
ential equation for ω ( t ) , assuming ω(0) = 0
(1 − ω ( t ) ) ω ( t ) = Cσ ν ( t ) ⇒ 1 − (1 − ω ( t ) )
ν 1+ν t
= C (1 + ν ) ∫ σ ν (τ ) dτ .
0
(6.19)
ω ( t ) = 1 − 1 − C (1 + ν ) ∫ σ ν (τ ) dτ . (6.20)
t 1+ν
0
Creep Damage Accumulation Mechanisms in Composite Materials 205
t
C (1 + ν ) ∫ σ ν (τ ) dτ = 1 . (6.21)
0
From the previous relationship, the time to failure for creep is readily obtained,
assuming s ( t ) = s 0 ,
1
tc = . (6.22)
C (1 + ν ) σ 0ν
Clearly, this result is equivalent to the one obtained previously by using the
Schapery theory [11]. Therefore, the creep lifetime expressions obtained for both
theoretical approaches are directly comparable and are, in fact, equivalent, even
though the parameters have distinct physical interpretations.
Damage evolution does depend strongly on a number of different factors acting
simultaneously, i.e., temperature, moisture, stress, viscoelasticity, viscoplasticity,
etc. Each of these factors is time-dependent. In practice, the influence of any one
on long-term failure is measured independently, i.e., under constant conditions.
Hence, a further methodology is needed to account for their combined effects.
One crucial question remaining to be solved completely is how to predict dam-
age accumulation, or the remaining strength, after a fatigue or creep cycle at mul-
tiple stress levels, based on the fatigue and creep master curves. Miner’s Rule
[26] is an example of a simple way to account for damage accumulation due to
different fatigue cycles. This damage fraction rule is also designated as the Linear
Cumulative Damage law (LCD). For fatigue, it states that failure occurs when the
following condition is satisfied:
N ∆ni (s i )
∑ n (s )
i
= 1 , (6.23)
f i
where n f is the number of cycles to failure at stress level s i and ∆ni is the num-
ber of cycles applied at each stress level s i of the fatigue cycle. Hence eq. (6.23)
provides a failure criterion for fatigue. The corresponding form for creep condi-
tions is given by:
N ∆ti (s i )
∑ t (s )
i
= 1 , (6.24)
c i
where tc is the creep rupture lifetime at stress level s i and ∆ti is the time ap-
plied at each stress level s i . Once more, equations (6.23) and (6.24) specify a
206 Creep Design of Piping Applications Using Composite Materials
criterion for a lifetime at multiple stress levels. Later, Bowman and Barker [27]
suggested a combination of both damage fraction rules to analyze experimental
data for thermoplastics tested until failure under a trapezoidal loading profile,
which combined fatigue with creep. Although the Miner’s rule can predict accu-
rately the failure of fiber-reinforced polymers under certain combined stress lev-
els, it proved to be inadequate in many other cases. However, due to its simplicity,
it is still used in engineering design.
A cumulative damage theory developed to address various applied problems in
which time, temperature, and cyclic loading are given explicitly, was developed
by Reifsnider et al. [28–30].
The basic form of the strength evolution equation calculates the remaining
strength Fr
z σ (τ ) j −1
Fr = 1 − ∫ 1 − Fa jτ dτ , (6.25)
X (τ )
0
E1 E2
1 − ν ν ν 12 0
1 − ν 12ν 21
12 21
E2
where Qij elastic = 0
1 −ν 12ν 21
Sym. G12
E1
2
E1 E2
β 2 ν 12 β 2 ν 12 0
1 −ν 12ν 21 1 −ν 12ν 21 1 −ν 12ν 21
2
damage E2
and Qij =ρ β2 0
1 −ν 12ν 21
Sym. β1G12 2
where E1 is the longitudinal modulus (in the fiber direction), E2 is the transverse
modulus (perpendicular to the fiber direction), ν 12 is the major in-plane Poisson’s
ratio (i.e., the Poisson’s ratio that corresponds to a contraction in the transverse
direction when an extension is applied in the longitudinal direction and is related
to the minor in-plane Poisson’s ratio by ν 21 E2 = ν 12 E1 ), and G12 is the in-plane
shear modulus. The dimensionless or normalized crack density in the ply is de-
fined by the ratio of ply thickness and crack spacing ρ = t s . The coefficient β1
relates to crack-face displacement in Mode III anti-plane strain, and the coefficient
β 2 relates to a Mode I crack opening. These two coefficients are expressed by:
2 π
β1 = ln cosh ρ
π G12 ρ 2
2
, (6.27)
π 1 −ν 12ν 21 10 an
β2 = ∑
2 E2 n =1 (1 + ρ )n
where the constants an are tabulated in [32]. For a ply with a free surface, β1 is
2.0 times, and β 2 2.51 times, their corresponding values [34].
The crack density effect on elastic constants is shown in Figure 6.2, with the
results normalized by its initial undamaged value, which is obtained from classical
laminated plate theory (CLPT) coupled with equation (22.6). The main conclu-
sion for angle-ply laminates is that the axial modulus decreases most rapidly with
increasing crack density. Frost [31] suggested an approximation using a linear
208 Creep Design of Piping Applications Using Composite Materials
1.20
Normalized change in elastic constant
0.60
Hoop modulus
0.40
Axial modulus
0.20
0.00
0.00 0.20 0.40 0.60 0.80 1.00 1.20
Normalized crack density
figure 6.2 Normalized change in elastic constants versus crack density for ±55º pipe
laminate, calculated from the model of Roberts at al. [32] (lines) and using the linear ap-
proximation given by Frost [31] (points).
relationship for elastic properties that decay with crack normalized density. The
simplified approach proposed by Frost [31] captures the initial changes in elastic
properties, as depicted in Figure 6.2. The observed differences arise from the fact
that the elastic material properties are not exactly the same. Hence, this simplified
approach does predict relevant decay in the pipe’s elastic properties with good
accuracy for the important crack density range prior to failure.
The relation between the crack density and the applied stress, for each ply, was
approximated as:
1+ k 2ρ 2 = f (σ 2 ,τ 12 ) [31], (6.28)
2 2
σ2 τ 12 σ2 τ 12
and f (σ 2 ,τ 12 ) = + −
σ
2, failure τ 12, failure σ τ
2 , failure 12 , failure
Short and Long-Term Static Failure of Composite Pipes 209
where s 2 and τ 12 represent the ply transverse and shear stresses and s 2, failure ,
and τ 12, failure the respective failure stresses. The function in (6.24) gives the nor-
malized crack density as a function of a second-order polynomial that depends
on both the shear and transverse stresses, which are the stresses contributing to
crack formation. Since the stiffness in (6.22) depends on crack density, a nonlin-
ear stress–strain relation is obtained. This was applied by Roberts et al. [32] to the
deformation of internally pressurized pipes with promising results.
The methodology is easy to implement, and some calculations were performed
using data for E-glass/epoxy ±55º filament-wound pipes provided by Soden et al.
[35–36]. In Figures 6.3 and 6.4, predictions and experimental data are compared
for stress-strain response under two different modes: hydrostatic pressure with
closed and open ends. The theory provides a good approximation to the stress-
strain experimental curves.
In general, the failure parameter is given as:
f (σ 2 ,τ 12 ) = C , (6.29)
LIVE GRAPH
Click here to view
300
250 Experimental
Axial strain
Hoop Stress (MPa9
Hoop strain
200 Calculated
150
100
σhoop/σaxial=2
50
0
0.000 0.005 0.010 0.015 0.020
Strain
figure 6.3 Comparison between experimental data and predictions of the stress-strain
behavior of an E-glass/epoxy ±55º filament-wound pipe under hydrostatic pressure with
closed ends.
210 Creep Design of Piping Applications Using Composite Materials
LIVE GRAPH
Click here to view
300
250
Hoop strain
Hoop Stress (MPa9
Axial strain
200
150
Experimental
100 Calculated
σaxial/σhoop=0
50
0
-0.050 -0.030 -0.010 0.010 0.030 0.050
Strain
figure 6.4 Comparison between experimental data and predictions of the stress-strain
behavior of an E-glass/epoxy ±55º filament-wound pipe under hydrostatic pressure with
open ends.
According to Frost [31], for long-term sustained loading the factor Acreep can be
set to 0.5 for 20-year conservative design. Using this assumption and the previous
theoretical approach, it is possible to predict the failure envelope for short-term
and long-term cases under different loading modes, as shown in Figure 6.5, for
an E-glass/epoxy ±55º filament-wound pipe. Failure was assumed to occur when
the normalized crack density reaches the value of 0.5 (2.2mm for a ply thickness
of 0.25mm).
LIVE GRAPH
Click here to view
180
160
140
Axial Stress (MPa)
120
100
80
60
40 Short-term prediction
20 Long-term prediction
0
0 50 100 150 200 250 300 350
Hoop Stress (MPa)
figure 6.5 Predicted failure envelope for short-term and long-term sustained loadings on
an E-glass/epoxy ±55º filament-wound pipe.
Short and Long-Term Static Failure of Composite Pipes 211
1
Acreep = , (6.31)
ta
where a is equal to the time exponent obtained by regression of the creep failure
curves.
For cyclic loads exceeding 7000 cycles over the design life, fatigue effects
must be considered. These effects are more severe than creep effects. The stan-
dard ISO 14692-3:2002 proposes the following factor:
N − 7000
Afatigue = R 2 + (1 − R 2 ) exp (1 − R 2 ) 1 −
1
, (6.32)
108
16
where R is the ratio between the minimum and maximum loads (or stresses) of
the load (or stress) cycle and N is the total number of cycles during service life.
For illustration purposes an example of static (creep) and cyclic pipe test re-
sults at 65ºC can be cited from [37]. In Figure 6.6, the creep failures curves are
shown, and in Figure 6.7 the experimental data and the predicted curves are rep-
resented. The fatigue curves were predicted from creep failure curves using the
factor Afatigue from ISO 14692-3:2002, with R = 0. The predictions are very close
to experimental data.
300
Water, 65ºC
250
Hoop Stress (MPa)
200
y = 445.7633x-0.0624
150 R2 = 0.9970
100 Experimental
Curve Fit
50
0
1.E+00 1.E+02 1.E+04 1.E+06 1.E+08 1.E+10
Time to Failure (s)
300
Water, 65ºC
250
Hoop Stress (MPa)
Experimental
200
150
ISO 14692
100
50
0
1.E+00 1.E+02 1.E+04 1.E+06 1.E+08 1.E+10
Cycles to Failure
figure 6.7 Experimental and predictions using creep data (ISO) for fatigue failure of
pipes at 65ºC.
\ [
5
+RRS6WUHVV03D
\ [ (SR[\
5 3RO\HVWHU
7LPHWR)DLOXUHKRXU
figure 6.8 Long-term weeping stresses in epoxy and polyester filament-wound pipes (af-
ter Mieras [40]).
LIVE GRAPH
Click here to view
214 Creep Design of Piping Applications Using Composite Materials
120
100
Hoop Stress (MPa)
80
60
after 1 min
40
after 1000 hour
20 after UEWS test
0
0 0.002 0.004 0.006 0.008 0.01
Hoop Strain
figure 6.9 Isochronous stress/strain curves for epoxy pipes compared with that of the pipe
after Ultimate Elastic Wall Stress (UEWS) tests.
LIVE GRAPH
Click here to view
in the first and tenth cycle of each step. The difference between these two strains
is used to verify whether elastic limits have been reached. Initially, the pipe speci-
mens are loaded using a hydrostatic pressure up to 10% of the expected UEWS
level with the hydrostatic pressure maintained for 1 minute. Then the hydrostatic
pressure is released, and the pipe remains unloaded for another minute. This load-
ing-unloading pattern is repeated 10 times with the strains being measured at the
end of the first and tenth one-minute load cycle. Subsequently, similar 10-cycle
series are successively repeated with the hydrostatic pressure in each step being
increased by 10% of the expected UEWS hydrostatic pressure. The test is sched-
uled to continue until a certain level of permanent deformation is reached (5%).
The UEWS test is a simple procedure that takes a few hours to perform. As
Gibson et al. [37] maintain, the UEWS can become an accepted alternative for
re-confirming pipe qualification whenever a minor product change is made. The
actual procedure for re-confirmation also mandates survival tests with samples
held under hydrostatic pressure for 1000 h. Survival in this test leads to a conclu-
sion: the samples tested are at least as good as those originally qualified.
Ghorbel et al. [3, 46] investigated the creep and damage of filament-wound
pipes reinforced with E-glass fibers wound at ±55º for two different resin systems:
polyester and vinylester. The response of the preconditioned pipes in water at
60ºC under sustained hydrostatic pressure with closed-ends was nonlinear visco-
elastic. It was observed that creep under hygrothermal conditions induces inter-
laminar cracking. The time-dependent failure under sustained loading followed a
trend similar to that depicted in Figures 6.10 and 6.11.
Short and Long-Term Static Failure of Composite Pipes 215
LIVE GRAPH
Click here to view
15.00
E glass/polyester
Water at 60ºC
Pressure (MPa)
10.00
5.00
y = 7.90603x-0.07181
R2 = 0.68232
0.00
0.01 0.1 1 10 100 1000 10000
Time to Failure (hour)
figure 6.10 Long-term creep failure of E glass/polymers pipes wound at ±55ºC immersed
in water at 60ºC [3].
20.00
ECR glass/vinylester
10.00
y = 16.40839x-0.04707
5.00 R2 = 0.99351
0.00
0.01 0.1 1 10 100 1000 10000
Time to Failure (hour)
figure 6.11 Long-term creep failure of E glass/vinylester pipes wound at ±55ºC immersed
in water at 60ºC [3].
LIVE GRAPH
Click here to view
It was concluded that the failure mechanisms and time-dependent failure de-
pend strongly on the resin, as can be observed in Figures 6.10–6.11. In contrast
to high pressure levels, at low pressure levels the effect of environment must be
assessed, since it significantly affects the time-dependent failure.
216 Creep Design of Piping Applications Using Composite Materials
Table 6.1
GFRP pipe specimens used in the experimental test series.
*)53SLSHW\SH
$ % & '
+\EULG±
ILODPHQW
ILODPHQW ILODPHQWZLQGLQJFHQWULIXJDO
&RQVWUXFWLRQ7\SH ZLQGLQJ
ZLQGLQJ83 PDW FDVWLQJ83
HSR[\
GHSRVLWLRQ83
1RPLQDO3UHVVXUHEDU
6SHFLPHQ/HQJWKPP
LIVE GRAPH
Click here to view
9.0
8.0 Pipe type A
Internal Pressure (MPa)
7.0
6.0
5.0
4.0
3.0 y = 5.786961x-0.006811
2.0 R² = 0.571338
1.0
0.0
0.1 1 10 100 1000 10000 100000
Time to Failure (hour)
Figure 6.13 Creep failure results for type A pipe.
LIVE GRAPH
Click here to view
9.0
8.0 Pipe type B
Internal Pressure (MPa)
7.0
6.0
5.0
4.0
3.0 y = 7.828656x-0.035688
2.0 R² = 0.741599
1.0
0.0
0.1 1 10 100 1000 10000 100000
Time to Failure (hour)
9.0
Internal Pressure (MPa) 8.0 Pipe type C
7.0
6.0
y = 3.785206x-0.059767
5.0 R² = 0.425245
4.0
3.0
2.0
1.0
0.0
0.1 1 10 100 1000 10000 100000
Time to Failure (hour)
LIVE GRAPH
Click here to view
9.0
8.0 Pipe type D
Internal Pressure (MPa)
7.0
6.0
y = 3.675653x-0.005417
5.0 R² = 0.125838
4.0
3.0
2.0
1.0
0.0
0.1 1 10 100 1000 10000 100000
Time to Failure (hour)
laminated plate theory (CLPT) to show the discrepancy between CLPT and the
netting theory and reveal significant factors for design. Verchery’s [52] approach
to the same problem proved to be more effective for deriving explicit design for-
mulas applicable to any state of stress and stacking sequence.
As an example of long-term preliminary analysis, let us suppose we have a
±55º glass/epoxy filament wound pipe with an internal diameter of 100mm and
a wall thickness of 5mm, sustaining a hydrostatic pressure of 0.6 MPa. The hoop
stress is
Pi D 0.6 ⋅100
s hoop = = = 6MPa , (6.33)
2t 2⋅5
1
sf = s hoop ≅ 9.0MPa , (6.34)
sin 2 ( 54.74º )
Now we can calculate the fiber strength, which remains after degradation on a
long-term basis [53], i.e., the allowable fiber design strength for the pipe as:
σ α = σ v Pt Psc Ps , (6.35)
where σ α is the virgin fiber strength, Pt is the thermal degradation factor, Psc is
the stress concentration factor and Ps is the factor for long-term static loading.
Let us assume that for the glass fiber we have σ α = 157MPa . The strength reduc-
tion due to thermal degradation is assumed as Pt = 0.8 . After the fiber is placed
in the strand, the strength tensile reduction is about 25%, i.e., Psc = 0.75 , due to
localized stress concentrations, fiber crossovers and residual stress, among others.
Since glass fibers are particularly susceptible to static fatigue effects (creep), in
many cases the strength reduction due to long-term static loading is very high,
i.e., Ps ∈ [ 0.1, 0.2] [53]. Assuming in this case Ps = 0.1 , the allowable fiber design
strength becomes:
This residual strength is slightly larger than the fiber stress calculated from the
netting analysis, which indicates a reasonable long-term preliminary design.
220 Creep Design of Piping Applications Using Composite Materials
t
log( P) = A − B log (6.37)
τ0
where P is the pressure (or hoop stress or axial stress), t the time, and τ 0 the
reference time. The curves are extrapolated for 20 years to define the long-term
hydrostatic pressure (LTHP). The resulting value is multiplied by a number of
factors (typically 0.5 [43]), resulting in a pressure known as the hydrostatic design
basis (HDB) [44].
Tarakcioglu et al. [55] tested ±55º glass-fiber/epoxy filament-wound pipes un-
der internal pressure. The pipes had four layers with 1.6 mm in thickness, 300mm
in length and an inner diameter of 72mm. Stress levels were 30% (121.5 MPa),
35% (141.7 MPa), 40% (162 MPa), 50% (202.5 MPa), 60% (243 MPa) and 70%
(283.5 MPa) of the static strength of the specimen (405 MPa). Sinusoidal stress
levels were applied at 0.42 Hz for R = 0.05 stress ratio (R = Maximum stress/
Minimum stress). Fatigue results were recorded for three different damage stages,
namely, whitening, leakage and final failure. The three stages of whitening (fiber/
matrix interface debonding and delamination), leakage initiation, and final failure
occurred sequentially. Micrographs from an SEM observation proved that there
is an analogy between the macro-damage stages and the micro-damage mecha-
nisms. It is possible to fit equation (6.37) to the three damage stages shown in
Figures 6.17–6.19 with a very good correlation. For all stress levels, ultimate
failure occurred almost immediately after leakage initiation.
Another study of E-glass/epoxy filament-wound pipes of four layers with a
±75º winding angle, using the same test conditions, led to similar conclusions
[56]. In both cases no evidence of a fatigue limit was found.
LIVE GRAPH
Click here to view
Lifetime of Composites Pipes Under Cyclic Loading 221
350
300
Hoop Stress (MPa)
250
200
150
100 y = 314.22x-0.1029
50 R2 = 0.9577
0
1 10 100 1000 10000 100000 1000000
Cycles to Whitening Initiation
figure 6.17 S–N curve for whitening initiation of ±55º glass-fiber/epoxy filament-wound
pipes.
LIVE GRAPH
Click here to view
300
250
Hoop Stress (MPa)
200
150
100
y = 620.51x-0.1555
50 R2 = 0.9694
0
1 10 100 1000 10000 100000 1000000
Cycles to Leakage Initiation
figure 6.18 S–N curve for leakage initiation of ±55º glass-fiber/epoxy filament-wound
pipes.
Fatigue under biaxial loading was obtained by Perreux and Joseph [57] for
E-glass/epoxy pipes with four layers with ±55º winding angle. The pipes had a
diameter of 60mm and a length of 2800mm and were cut into 350mm lengths.
The fatigue tests were done under three different modes: hydrostatic pressure with
closed and free ends and axial tensile loading.
222 Creep Design of Piping Applications Using Composite Materials
LIVE GRAPH
Click here to view
300
250
Hoop Stress (MPa)
200
150
100
y = 625.14x-0.1561
50 R2 = 0.9698
0
1 10 100 1000 10000 100000 1000000
Cycles to Failure
figure 6.19 S–N curve for final failure of ±55º glass-fiber/epoxy filament-wound pipes.
LIVE GRAPH
The results are summarized in Figures 6.20–6.22. Click here to view
700
600 Internal Pressure with free ends
Hoop Stress (MPa)
500
400
300
y = 585.58x-0.1622
200 R2 = 0.9402
100
0
0.1 1 10 100 1000 10000 100000 1000000
Cycles to Failure
figure 6.20 S–N curve for final failure of ±55º E-glass/epoxy filament-wound pipes under
hydrostatic pressure with free ends for a frequency of 0.2 Hz.
The previous results can be put in the form of an isonumber of cycles to failure,
depicted in Figure 6.23. This provides an approximate idea of the evolution of
failure envelopes with cycle fatigue. Hence, in this case, the failure envelope for
106 cycles can be obtained from the static failure envelope by applying a scaling
factor of 0.1.
Lifetime of Composites Pipes Under Cyclic Loading 223
700
600 Internal Pressure with closed ends
Hoop Stress (MPa)
500
400
300 y = 590.96x-0.1546
200 R2 = 0.9758
100
0
0.1 1 10 100 1000 10000 100000 1000000
Cycles to Failure
LIVE GRAPH
Click here to view
figure 6.21 S–N curve for final failure of ±55º E-glass/epoxy filament-wound pipes under
hydrostatic pressure with closed ends for a frequency of 0.2 Hz.
90
80 Tensile
70
Axial Stress (MPa)
60
50
40 y = 80.894x-0.0771
30 R2 = 0.888
20
10
0
0.1 1 10 100 1000 10000 100000
Cycles to Failure
LIVE GRAPH
Click here to view
figure 6.22 S–N curve for final failure of ±55º E-glass/epoxy filament-wound pipes under
axial tensile load for a frequency of 0.2 Hz.
The effect of frequency on fatigue lifetime was also studied by Perreux and
Joseph [57]. It was concluded that fatigue lifetime does increase with frequency
from 0.2 to 1Hz, i.e., 2~3 fold higher. Later, Perreux and Thiebaud [58] concluded
there are two concurrent phenomena that influence fatigue failure: (1) interaction
between creep and fatigue at low frequency, which increases the lifetime when the
224 Creep Design of Piping Applications Using Composite Materials
350
Scaling factor Static
300
Axial Stress (MPa)
250
200
150
102 cycles
100
50 104
106
0
0 100 200 300 400 500 600 700
Hoop Stress (MPa)
figure 6.23 Failure envelopes for ±55º E-glass/epoxy filament-wound pipes under biaxial
tensile load, for multiple fatigue cycles.
LIVE GRAPH
Click here to view
frequency is increased, and (2) a thermal effect due to viscoplastic dissipation at
a higher frequency, which reduces the lifetime when the frequency is increased.
Thus, all situations can be observed, depending on which phenomena are most
intense. It should be noted that there is no effect of the frequency on the lifetime if
the phenomena have similar intensity. However, this result shows that a material’s
lifetime as measured in fatigue can depend on the shape of the specimen as well
as on thermal dissipation.
Ellyin and Martens [59] investigated the multi-axial fatigue of pipes, showing
that there are two stages in failure: a functional one and final structure rupture.
Kaynak and Mat [60] studied tensile fatigue and showed that damage has three
stages: crack initiation, crack growth and concentration along the fibers’ direction,
and fiber failure. They also studied the effect of frequency on fatigue lifetime
and concluded, as Joseph and Perreux [57] had, that the tensile fatigue lifetime
increases with frequency.
In conclusion, the effect of cycling loading on damage initiation and propaga-
tion is more severe than sustained loading, i.e., creep loading. The UEWS tests,
previously mentioned and described, are strongly linked to fatigue tests as was
demonstrated by Gibson et al. [37]. A simple simulation of a UEWS test can be
done, using creep and fatigue failure curves from [37], to compute the cumulative
damage that originated separately from creep and fatigue. This was accounted for
using Miner’s law as suggested in [37]. The calculations are depicted in Figure
6.24. It is quite obvious that in the UEWS test the fatigue effect on damage is
much more pronounced.
Applicable Standards 225
1.0
UEWS
0.8
Damage factor
0.2
0.0
0 50 100 150 200 250 300
Hoop Stress (MPa)
figure 6.24 Computed Miner’s law sum, for creep and fatigue, at each hoop stress level
in the UEWS test.
These standards are cited because they form the current references for industry
and end-users worldwide. Each group of product standards (ASTM, AWWA, EN
and ISO) covers: pressure and non-pressure, aboveground and underground, off-
shore and onshore applications of GRP pipes. Although they are suitable for GRP
pipes manufactured by different processes (filament winding, centrifugal casting,
hand lay‑up etc.), and with different matrix materials (polyester, vinyl‑ester and
epoxy resins), each refers to a specific manufacturing process and material con-
figuration as the basis for the design and qualification methodology stated in it.
ASTM D3517, AWWA C950, EN 1796 and ISO 10639 are alternatively used
mainly for water supply piping systems with in‑service internal pressure loading
conditions. ASTM D3262, ASTM D3754, EN 14364 and ISO 10467 are alterna-
tively used mainly for sewage transportation piping systems where no internal
Table 6.2
Identification and comparison of GRP pipe standards.
Test/Parameter Standard
ASTM ASTM ASTM ASTM AWWA AWWA EN 1796 EN ISO ISO ISO
D3262 D3517 D3754 D2997 C950 M45 14364 10467 10639 14692
Initial circumferen- n/a ASTM ASTM ASTM ASTM ASTM EN 1394 EN 1394 ISO 8521 ISO 8521 ASTM
tial tensile strength D1599 D1599 D1599 D1599 D1599 D1599
(failure pressure)
Long-term circum- n/a ASTM ASTM ASTM ASTM ASTM EN 1447 EN 1447 ISO 7509 ISO 7509 ASTM
ferential tensile D2992 D2992 D2992 D2992 D2992 D2992
strength (failure
pressure)
Cyclic pressure n/a n/a n/a ASTM ASTM ASTM EN 1638 EN 1638 ISO ISO ASTM
strength D2143 D2143 D2143 15306 15306 D2143
Initial specific ring
227
ASTM ASTM ASTM ASTM ASTM ASTM EN 1228 EN 1228 ISO 7685 ISO 7685 ASTM
stiffness D2412 D2412 D2412 D2412 D2412 D2412 D2412
Long-term specific n/a n/a n/a n/a n/a n/a ISO ISO ISO ISO n/a
ring stiffness 10468 10468 10468 10468
Initial ring deflection ASTM ASTM ASTM n/a ASTM ASTM ISO ISO ISO ISO ASTM
D2412 D2412 D2412 D2412 D2412 10466 10466 10466 10466 D2412
Long-term ring n/a n/a n/a n/a n/a n/a ISO ISO ISO ISO n/a
deflection 10471 10471 10471 10471
Longitudinal tensile ASTM ASTM ASTM ASTM ASTM ASTM EN 1393 EN 1393 ISO 8513 ISO 8513 ASTM
strength D638 D638 D638 D2105 D2105 D2105 D2105
Methods for regres- n/a (ASTM (ASTM (ASTM (ASTM (ASTM ISO ISO ISO ISO (ISO
sion analysis of test D2992) D2992) D2992) D2992) D2992) 10928 10928 10928 10928 14692)
data
228 Creep Design of Piping Applications Using Composite Materials
pressure is considered. ISO 14692 is broadly used in the oil and gas industries,
mainly for off‑shore applications. Since it covers the design and qualification of
suspended pipe systems, AWWA M45 is often complimentarily used for under-
ground pipe systems within the same industries.
The main differences between ISO 14692 and the other ISO and CEN product
standards for pressure piping applications are that:
As can readily be observed in Table 6.2, the standard ISO 14692 adopted the
ASTM methodology and thus allows for design and qualification procedures
that are less expensive and time‑consuming than other ISO and CEN standards.
Nevertheless, the ISO 14692 procedure retains a conservative quality by: (1) re-
quiring higher test temperatures (thus aging the specimens during the tests) and
(2) using a lower confidence-limit level to determine the long-term hydrostatic
pressure (instead of directly extrapolating from the regression line).
In order to assess the regression line that permits the prediction of the long‑term
hydrostatic pressure or hoop stress level that leads to a lifetime of 20 years (ISO
14692) or 50 years (EN 1796), the critical procedure lies within the standards tests
ASTM D2992 or EN 1447, respectively. These standards establish the procedures
for obtaining the hydrostatic design basis (HDB) or the pressure-design basis
(PDB) for GRP piping products, by evaluating strength‑regression data derived
from pipe or fittings tests. The data obtained from these test methods is plotted as
hoop stress or internal pressure versus time‑to‑failure relationships at the selected
temperatures that simulate actual anticipated product end-use conditions. This
practice defines a hydrostatic design basis (HDB) for material in straight, hollow
Applicable Standards 229
Figure 6.25 Idealized envelopes for a single‑wound angle-ply GRP pipe with winding
angles ranging from 45º to 75º. (1) long‑term design envelope, (2) idealized long‑term
envelope, (3) idealized short‑term envelope, (4) schematic representation of the short‑term
failure envelope (after ISO 14692:2002).
cylindrical shapes where hoop stress can be easily calculated, and a pressure-
design basis (PDB) for fittings and joints where stresses are more complex.
ASTM D2992, in particular, includes two test procedures, accounting for two
typical in‑service loading cases (not limited to internal pressure):
responsible for long‑term failure are the same at different levels of load and from
short‑ to long‑term. Although this limitation is not explicitly addressed in the stan-
dards, the experimental evidence of the whole failure phenomenon and the lack
of adequate information were the main reason for extending the test periods to
over 10,000 hours. In a logarithmic time scale, 10,000 hours is only 1.2 and 1.6
decades distant from 20 years and 50 years, respectively. This makes the existing
test and prediction methods seem reasonable.
In the following section a practical case is presented where the regression
analysis procedure for long‑term hydrostatic pressure is directly applied and the
results are discussed.
• filament‑wound pipes are more resistant than the other construction types to
structural degradation when subjected to constant internal pressure. This is
observed from the smaller slope of the regression lines of pipes A and B;
• pipes of type A, made with epoxy resin, displayed very consistent behavior
for all test periods, suggesting that this resin system might be less vulner-
able to moisture absorption and softening by resin‑fiber de-cohesion and
general degradation;
• pipes of type C, manufactured by a hybrid combination of filament winding
(continuous fibers) and mat deposition (short fibers), with relevant inclu-
sions of silica, showed a very steep slope, thus displaying an overall degra-
dation rate much higher than the others. The lack of a continuous reinforce-
ment surely increases the creep factor of the structure, since only the resin
permanently sustains circumferential stress;
A
B
1,8 C
D
Pressure [log bar]
1,6
1,4
1,2
1
-2 -1 0 1 2 3 4 5
Time to Failure [log h]
figure 6.26 Results of the test series conducted accordingly to EN 1447:2009 – pressure
versus time‑to‑failure in log-log scale [59].
232 Creep Design of Piping Applications Using Composite Materials
• pipes A, C and D show very similar short‑term results, which agrees with
their equal nominal specifications, namely, initial stiffness and strength,
since they are classified for the same nominal service pressure and ring stiff-
ness. Specimens of type B seem clearly to be over‑designed for the specific
application of internal pressure.
Table 6.3
Main specifications of the GRE pipe specimens considered
in the practical design case [35].
Construction Type Filament Winding
fibre type Silenka E-glass 1200TEX
resin type MY750/HY917/DY063 epoxy
winding angle [º] ±55
number of ± plies 2
internal diameter (DN) [mm] 100
average wall thickness [mm] 1
fibre volume fraction [%] 60
post‑curing cycle 2 h @ 90ºC + 1.5 h @130ºC + 2 h @ 150ºC
Practical Design: A Case Study 233
Table 6.4
Mechanical properties of each single-angle (+55º or –55º) lamina of GRE
pipe specimens considered in the practical design case [35].
axial modulus [GPa] 39.4
circumferential modulus [GPa] 46.6
axial tensile strength [MPa] 767
axial compressive strength [MPa] 578
circumferential tensile strength [MPa] 1071
circumferential compressive strength [MPa] 739
In fact, type A pipe specimens are similar to the ones selected here, both in
terms of the applied materials (E‑glass fiber and epoxy resin) and manufactur-
ing technique (filament winding). Hence, the absolute value of the slope/gradi-
ent, G = 0.0069 , of the regression line observed in the tests mentioned above
(Fig. 6.26, etc.) can constitute a reference for the expected degradation of the
properties of the pipes under consideration. However, since the tests presented in
Figure 6.26 were conducted at room temperature (EN 1447 [59]) instead of at the
higher temperature specified in ISO 14692 [54], the gradient considered hereafter
is five (5) times higher than that one. Thus, the absolute value of the gradient of
the idealized regression line is conservatively given by:
2 t (s shu × 0.66 )
pLTHP = = 14.137 MPa = 141.37 bar (6.39)
D
where s shu is the short‑term hoop (circumferential) ultimate strength of the pipe
specimen according to the material property data provided in Table 6.4, and t and
D are the mean wall thickness and diameter of the pipe specimens, respectively.
For the sake of simplicity, the qualifying pressure, pq , is herewith assumed to
be equal to pLTHP .
The pressure rating to provide in the product literature is given by:
pNPR = f 2 f 3 pq (6.40)
234 Creep Design of Piping Applications Using Composite Materials
where f 2 is defined as the part load factor and f 3 is a factor accounting for the
limited axial load capability of GRP/GRE pipes. Default values for f 2 are given
in ISO 14692 [54] depending on the load-type assessment that is to be considered
for the application. These values are 0.67 for the case of sustained loads (pressure,
mass, etc.) excluding thermal effects and 0.89 for the combination of sustained
loads and occasional loads (water hammer, blast, etc.). Since our assessment is
based on sustained internal pressure (with eventual minor induced axial stress), a
value of f 2 = 0.67 is used.
For further assessments, let us assume that the biaxial stress ratio, circumferen-
tial stress/axial stress, of the application and installation for which this pipe is be-
ing designed is 1:1 and that the non‑pressure‑induced axial stress is s ab = 5 MPa .
For the calculation of f 3 , the biaxial strength ratio, r , must be assessed using
equation (6.40):
s sa ( 0:1)
r=2 (6.41)
s sh( 2:1)
where s sa ( 0:1) is the short‑term axial strength under axial loading only and s sh( 2:1)
is the short‑term hoop (circumferential) strength under 2:1 stress conditions.
Within the scope of the World Wide Failure Exercise, Soden et al. [36] presented
extensive results for the failure of GRP/GRE pipes under biaxial stress conditions.
Table 6.5 lists the failure data points for the ±55º GRE pipe considered in this
practical example. These pairs of hoop/axial stress at failure data points define the
short‑term failure envelope as shown later.
From these results one can easily calculate r by using the averaged values of
s sa ( 0:1) = 68 MPa and s sh( 2:1) = 736 MPa that these authors found for the specific
pipe under consideration. Hence, the value of the ratio r is
s sa ( 0:1) 68
r=2 =2 ≈ 0.185 << 1 (6.42)
s sh( 2:1) 736
2 s ab
f3 = 1 − (6.43)
r f 2 s fs
pqf D
s fs = (6.44)
2t
Table 6.5
Biaxial failure test results of thin ±55º GRE pipes subjected to combined internal pressure and axial loading [36].
Hoop Stress Axial Stress Stress Ratio Hoop Stress Axial Stress Stress Ratio Hoop Stress Axial Stress Stress Ratio
[MPa] [MPa] [hoop:axial] [MPa] [MPa] [hoop:axial] [MPa] [MPa] [hoop:axial]
0 74 0:1 741 370 2:1 761 138 11:2
0 62 0:1 717 358 2:1 676 67 10:1
107 143 3:4 750 375 2:1 516 0 1:0
198 198 1:1 835 334 5:2 594 0 1:0
235
pqf = A1 A2 A3 pq (6.45)
where A1 , A2 and A3 are, respectively, the partial factors for temperature, chemi-
cal resistance, and cyclic service. These additional factors account for specific ser-
vice conditions that eventually cannot be considered in the qualification program.
Since the operating temperature that one posits for the present example is less
than or equal to 65ºC, A1 = 1 . It is also assumed that the resin is chemically com-
patible with the operating fluid and therefore A2 = 1 . Given the nature of the pres-
ent study and the typical applications of such GRP pipes, it is supposed that the
predicted number of pressure or other cycles is less than 7,000 over the design life
(20 years), and so the service is considered static. Under these conditions, A3 = 1 .
The factored qualified pressure is equal to the non‑factored qualified pressure
and, therefore, the qualified stress, s fs , is determined as follows:
2 s ab 2×5
f3 = 1 − = 1− ≈ 0.886 (6.47)
r f 2 s fs 0.185 × 0.67 × 706.85
The typical nominal pressure rating of this specific GRE pipe ( D = 100 mm ,
t = 1 mm , ϕ = ±55º ) would thus be PN = 80 bar .
In order to establish the long‑term failure envelope for this specific GRE pipe,
which defines its domain of allowable operating conditions under combined biax-
ial stress states, one must first determine the idealized long‑term failure envelope.
This envelope is geometrically identical to the short‑term envelope (experimental
data in Table 6.5), with the three main data points scaled through f scale , which is
given by:
s qs s fs 706.85
f scale = = = ≈ 0.96 (6.49)
s sh( 2:1) s sh( 2:1) 736
Conclusions 237
6.7 Conclusions
The aim of the present chapter was to provide a broad description of the pres-
ent state-of-the-art concerning qualification and preliminary long-term design of
filament-wound pipes. The worked examples were mainly focused on angle-ply
stacking sequences, especially on the ±55º winding angle. Unfortunately, there is
a lack of published experimental data reporting long-term creep failure on fila-
ment-wound pipes. Moreover, since qualification standards only mandate one to
register the time or cycles to failure, experimental data on creep strain of filament-
wound pipes is even scarcer. For obvious reasons, companies are reluctant to pub-
licize important information derived from millions of creep and fatigue testing
Figure 6.27 Failure envelopes for the selected/designed GRE pipe, calculated in accor-
dance with ISO 14692. Dots are the short‑term experimental data. The solid lines repre-
sent idealized short‑term (thinnest lines), idealized long‑term (medium lines) and design
long‑term (thickest lines) failure envelopes.
238 Creep Design of Piping Applications Using Composite Materials
hours. Yet fatigue curves for various filament-wound pipes, available in the tech-
nical literature, are more substantial. It should be pointed out that damage under
cyclic loading is more severe than damage under sustained loading. The standards
used for design and qualification of filament-wound pipes, which cover different
types of applications, were presented and compared. In the specific case of pipes
for oil and gas transportation, one worldwide standard accepted by industry is ISO
14692. The present chapter closed with a worked example based on ISO 14692,
which discussed the safety factors used in durability design.
6.8 Acknowledgements
Some test series presented here were conducted within research supported
by the Growth Programme of the European Commission and the Portuguese
Foundation for Science and Technology through projects G6RD-CT2000-00259
and POCTI/EME/47734/2002, respectively. The authors also acknowledge Dr.
Catherine Hervé (CETIM) for the data compilation of hydrostatic pressure tests
from project G6RD-CT2000-00259 partners.
References
[1] Ochoa Ozden O., Salama Mamdouh M. “Offshore composites: Transition
barriers to an enabling technology,” Composites Science and Technology
65:2588–2596, 2005.
[2] Schapery, R. A., (1969), “On the Characterization of Nonlinear Viscolelastic
Materials,” Polymer Engineering and Science 9(4):295–310, 1969.
[3] Ghorbel I., “Durability of closed-end pressurized GRP filament-wound
pipes under hygrothermal aging conditions. Part II: creep tests.” Journal of
Composite Materials 30(14): 1581–1595, 1996.
[4] Scott David W., Lai James S. and Zureick Abdul-Hamid. “Creep Behavior
of Fiber-Reinforced Polymeric Composites: A Review of the Technical
Literature. “Journal of Reinforced Plastics and Composites 14: 588–617,
1995.
[5] Guedes R. M. “An energy criterion to predict delayed failure of multi-
directional polymer matrix composites based on a non-linear viscoelastic
model,” Composites Part A 35(5): 559–571, 2004.
[6] Guedes R.M. “Nonlinear viscoelastic analysis of thick-walled cylindri-
cal composite pipes.” International Journal of Mechanical Sciences 52
(8):1064–1073, 2010.
References 239
[22] Ren W. “Creep behavior of a continuous strand, swirl mat reinforced poly-
meric composite in simulated automotive environments for durability in-
vestigation; Part I: experimental development and creep-rupture,” Materials
Science and Engineering A 334(1–2):312–319, 2002.
[23] Christensen R.M., “An evaluation of linear cumulative damage (Miner’s
Law) using kinetic crack growth theory.” Mech Time-Dependent Mater
6(4):363–77, 2002.
[24] Kachanov, L. M., “On the Time to Failure Under Creep Conditions.” Izv.
Akad. Nauk SSR, Otd. Tekhn. Nauk 8:26–31, 1958.
[25] Stigh U. “Continuum Damage Mechanics and the Life-Fraction Rule.”
Journal of Applied Mechanics 73(4):702–704, 2006.
[26] Miner M.A. “Cumulative damage in fatigue.” J. Appl. Mech. 12:A159–
A164, 1945.
[27] Bowman J., Barker M.B.A. “Methodology for describing creep–fatigue in-
teractions in thermoplastic components.” Polym Eng Sci 26(22):1582–90,
1986.
[28] Reifsnider K.L., Stinchcomb W.W. “A critical element model of the residual
strength and life of fatigue-loaded composite coupons.” In: Hahn HT, editor.
Composite materials: fatigue and fracture (ASTM STP 907). Philadelphia
(PA): American Society for Testing and Materials; p. 298–313, 1986.
[29] Reifsnider K., Case S., Duthoi J. “The mechanics of composite strength
evolution.” Composite Sci Technol 60:2539–46, 2000.
[30] Reifsnider K., Case S. Damage tolerance and durability in material sys-
tems. Wiley-Interscience; 2002.
[31] Frost S.R. “Ageing of composite in oil and gas applications,” in Ageing of
composites Edited by R Martin, Woodhead Publishing Limited, Cambridge,
2008.
[32] Roberts S.J., Evans J.T., Gibson A.G., Frost S.R. “The effect of matrix mi-
crocracks on the stress-strain relationship in fiber composite tubes.” Journal
of Composite Materials 37 (17):1509–1523, 2003.
[33] Zang W. and Gudmundson P. “An Analytical Model for the Thermoelastic
Properties of Composite Laminates Containing Transverse Matrix Cracks,”
International Journal of Solids and Structures, 30: 3211–3231, 1993.
[34] Gudmundson Peter and Östlund Sören.“Numerical Verification of a
Procedure for Calculation of Elastic Constants in Microcracking Composite
Laminates.” Journal of Composite Materials 26(17):2480–2492, 1992.
References 241
[35] Soden P. D., Hinton M. J. and Kaddour A. S. “Lamina properties, lay-up con-
figurations and loading conditions for a range of fibre-reinforced composite
laminates.” Composites Science and Technology 58 (1998) 1011–1022.
[36] Soden P.D., Hinton M.J., Kaddour A.S. “Biaxial test results for strength
and deformation of a range of E-glass and carbon fibre reinforced compos-
ite laminates: failure exercise benchmark data.” Composites Science and
Technology 62:1489–1514, 2002.
[37] Gibson A.G., Abdul Majid M.S., Assaleh T.A., Hale J.M., Fahrer A.,
Rookus C.A.P., Hekman M. “Qualification and lifetime modelling of fibre-
glass pipe.” Plastics, Rubber and Composites 40 (2):80–85, 2011.
[38] Hale J.M., Gibson A.G., Speake S.D. “Biaxial failure envelope and creep
testing of fibre reinforced plastic pipes in high temperature aqueous envi-
ronments.” Journal of Composite Materials 36 (3):257–270, 2002.
[39] Soden P.D., Kaddour A.S., Hinton M.J. “Recommendations for designers
and researchers resulting from the world-wide failure exercise.” Composites
Science and Technology 64 (3–4):589–604, 2004.
[40] Mieras H.J.M.A. “Irreversible creep of filament-wound glass-reinforced
resin pipes,” Plastics & Polymers 41:84–88, 1973.
[41] Jones M. L. C., Hull D. “Microscopy of failure mechanisms in filament-
wound pipe.” Journal of Materials Science 14:165–174, 1979.
[42] Hull D., Legg M.J., Spencer B. “Failure of glass/polyester filament wound
pipe,” Composites 9 (1):17–24, 1978.
[43] Frost S. R. and Cervenka A. “Glass fiber-reinforced epoxy matrix filament-
wound pipes for use in the oil industry.” Compos. Manuf. 5 (2):73–81, 1994.
[44] Gibson A.G., Dodds N., Frost S.R., Sheldrake T. “Novel approach to quali-
fication of non-metallic pipe systems - As applied to reinforced thermoplas-
tic pipe.” Plastics, Rubber and Composites, 34 (7):301–304, 2005.
[45] Abdul Majid M.S., Assaleh T.A., Gibson A.G., Hale J.M., Fahrer A., Rookus
C.A.P., Hekman M. “Ultimate elastic wall stress (UEWS) test of glass fibre
reinforced epoxy (GRE) pipe.” Composites: Part A 42:1500–1508, 2011.
[46] Ghorbel D. Valentin, Yrieix M.-C., Spiteri P. “The effect of resin flexibil-
ity on the creep behaviour of filament-wound glass-reinforced resin pipes.”
Developments in the science and technology of composite materials: Fourth
European Conference on Composite Materials (ECCM4), September 25–
28, 1990, Stuttgart, F.R.G. J. Füller, European Economic Community.
[47] Faria H., Guedes R.M. “Long-term behaviour of GFRP pipes: Reducing the
prediction test duration.” Polymer Testing 29 (3):337–345, 2010.
242 Creep Design of Piping Applications Using Composite Materials
[48] Guedes R.M., Sá A., Faria H. “On the prediction of long-term creep-failure
of GRP pipes in aqueous environment,” Polymer Composites 31 (6):1047–
1055, 2010.
[49] Gdoutos E. E., Pilakoutas K., Rodopoulos C. A. Failure analysis of indus-
trial composite materials. McGraw-Hill Professional; first edition, 2000.
[50] Eckold G., “Mechanics of materials behaviour,” Chapter 3 in Design and
Manufacture of Composite Structures, Woodhead Publishing, Cambridge,
1994 pages 86–88. ISBN 1-85573-051-0.
[51] Evans J.T. and Gibson A.G. “Composite angle ply laminates and netting
analysis.” Proceedings of the Royal Society, Part A: Mathematical, Physical
and Engineering Sciences 458:3079-3088, 2002.
[52] Verchery G. “Design of laminates for in-plane loading.” ICCM International
Conferences on Composite Materials, 2009.
[53] Tew B.W. “Preliminary design of tubular composite structures using net-
ting theory and composite degradation factors.” Journal of Pressure Vessel
Technology, Transactions of the ASME 117 (4):390–394, 1995.
[54] ISO 14692:2002—Petroleum and natural gas industries Glass reinforced
plastics (GRP) piping, International Organization for Standardization,
Geneva, 2002.
[55] AWWA M45:2005—Fiberglass Pipe Design, 2nd edition, American Water
Works Association, Denver, 2005.
[56] Laney Patrick. Use of Composite Pipe Materials in the Transportation of
Natural Gas, Idaho National Engineering and Environmental Laboratory,
Idaho, 2002.
[57] HOBAS CC-GRP Pipes—Features, Tests and Benefits, Hobas Engineering
GmbH, Klagenfurt, 2010.
[58] ISO 10928:2009 Plastics piping systems—Glass-reinforced thermosetting
plastics (GRP) pipes and fittings—Methods for regression analysis and
their use, International Organization for Standardization, Geneva, 2009.
[59] EN 1447:2009 Plastics piping systems Glass reinforced thermosetting plas-
tics (GRP) pipes Determination of long term resistance to internal pressure,
European Committee for Standardization, Brussels, 2009.
Chapter 7
A u A
⋅ du + A ⋅ dP + sin a ⋅ dy + π Ddy ⋅τ = 0 (7.1)
gc υ u
where:
243
244 Flow Capacity of Composite Pipelines
Taking into account the following Fanning equation providing the energy losses
due to friction
2 fu 2
dF = dL (7.2)
gc D
where f is the friction factor and L the pipeline length, the Bernoulli equation
can be written in the following form:
1 u dP sin a 2 f u2
du + + dy + dL = 0 (7.3)
gc υ 2 υ υ2 gc D υ 2
In the above equation, the ratio u / υ expresses the mass flow rate m through a
cross-section A, i.e., u / υ = m / A , and can be considered as a constant c :
u
=c (7.4)
υ
In eq. (7.3) every member of the summation expresses energy. Table 7.1 summa-
rizes the meaning of each energy term:
Gas Transmission 245
Table 7.1
Physical meaning of the terms of Bernoulli equation.
Energy terms Physical meaning
Kinetic energy
1 u
du = K .E.
gc υ 2
dP / υ = PR.E. Pressure energy
Potential energy
sin a
dy = P.E.
υ2
Energy losses due to friction
2 f u2
dL = E.L.
gc D υ 2
Integrating each energy term of eq. (7.3) along a finite pipe element lying be-
tween the cross-sections 1 and 2 (see Fig. 7.1), it can be determined that:
Taking into account eq. (7.4), the above equation can be written as:
2
c2 du
K .E. =
gc ∫
1
u
(7.6)
or
c 2 u2
K .E. = ln (7.7)
g c u1
or
2
PR.E = ∫ ρ dP (7.9)
1
PV = nZRT (7.10)
m
n= (7.11)
M
m
ρ= (7.12)
V
where M is the average molecular weight of the gas, equation (7.9) can be writ-
ten as:
2
M ⋅P
PR.E = ∫ dP (7.13)
1
Z ⋅ R ⋅T
In the foregoing equations, Tci , Pci (i = 1, 2, ..., N ) are the critical tempera-
ture and pressure respectively of each component of the gas (Table 7.2), and
yi (i = 1, 2, ..., N ) is the corresponding mole fraction.
Gas Transmission 247
Table 7.2
Critical temperature and pressure of constituents of natural gas
(Campbell Petroleum Series).
Gas Molecular Tci Pci
Component i Weight (K) (MPa)
C1 16.043 191 4.60
C2 30.070 305 4.88
C3 44.097 370 4.25
iC4 58.124 408 3.65
nC4 58.124 425 3.80
iC5 72.151 460 3.39
nC5 72.151 470 3.37
nC6 86.178 507 3.01
nC7 100.205 540 2.74
N2 28.016 126 3.40
CO2 44.010 304 7.38
H 2S 34.076 373 8.96
O2 32.000 155 5.04
H2 2.016 33 1.30
H2O 18.015 647 22.06
Air 28.960 132 3.77
He 4.000 5 0.23
Using equations (7.16), (7.17), the pseudo-critical pressure Pr' and tempera-
ture Tr' can be obtained from the following equations:
Pm
Pr' = (7.18)
Pc'
Tm
Tr' = (7.19)
Tc'
∫ P dP
2
2 P1 ⋅ P2
Pm = 1
2
= P1 + P2 − (7.20)
3 P1 + P2
∫ PdP
1
248 Flow Capacity of Composite Pipelines
Having obtained Pr' and Tr' , the average compressibility factor can be estimated
by the ref. [1].
2 2
sin a
P.E. = ∫ 2
dy = ∫ ρ 2 sin a dy (7.21)
1 υ 1
2 2
PM
P.E. = ∫ sin a dy (7.22)
1
ZRT
2
∫ 1
sin ady = ∆Η (7.23)
where ΔH is the relative elevation of the cross section 2 with respect to cross sec-
tion 1, the potential energy term can be expressed by the following formula:
Pm2 M 2
P.E. = ∆Η (7.24)
Z m2 R 2Tm2
2 fC 2 2
E.L. =
gc D ∫ 1
dL (7.25)
or
2 fC 2
E.L. = L (7.26)
gc D
Gas Transmission 249
C 2 u2 M P22 − P12 P2 M 2 2 fC 2
ln + + 2m 2 2 ∆Η + L =0
g c u1 Z m RTm 2 Z RT g D (7.27)
1424 3 1442443 1m42m43 12 4c 4 3
Kinetic Pressure Potential Energy Losses
Energy Energy Energy Due to Friction
M P22 − P12 2 fC 2
+ L=0 (7.28)
Z m RTm 2 gc D
We recall that
•
m
C= (7.29)
A
where A is the cross-section area
πD 2
A= (7.30)
4
On the other hand, the gas law at standard conditions is
• •
Pb V b = nb Z b RTb (7.31)
•
where V b , Pb , Tb , Z b are the volumetric gas flow rate, pressure, temperature
and compressibility factor respectively, at standard conditions (Pb = 14.7 psia,
Tb = 520 °R, Zb ≈ 1).
Since
•
m •
n= (7.32)
M
equation (7.29), with the aid of eqs. (7.30)–(7.32), can be written as:
•
4 V b M Pb2
C= (7.33)
πRTb Z b D 2
250 Flow Capacity of Composite Pipelines
M
G= (7.34)
M air
can be
and combining equations (7.28), (7.33), the volumetric gas flow Vb
obtained:
• P12 − P22 1 52
Vb = A D (7.35)
Z mTm GL f
g c R Z bTb
A= π (7.36)
1.856 Pb
Equation (7.35) provides the gas flow rate at standard conditions of a pipe of di-
ameter D and length L conveying gas with inlet pressure P1 and exit pressure P2
for laminar, partially turbulent or fully turbulent flow and high pressure.
It is important to mention that the gas flow rate depends on the parameters
1 f , D 5 2 , 1 Tm .Therefore, the following conclusions can be determined:
7.1.1.1 friction factor for turbulent and partially turbulent flow regimes
Equation (7.35) requires a numerical value for the friction factor f. For rough
material surfaces and high values of fluid velocity (or flow rate), the flow is char-
acterized as turbulent, while for smooth material surfaces and low values of fluid
velocity, the flow is stable (or laminar).
A criterion for flow characterization is the value of the well-known Reynolds
dimensionless number Re:
Gas Transmission 251
ρ Du
Re = (7.37)
µ
where ρ is the fluid density and μ is the fluid viscosity. For Re > 2000 the flow is
turbulent, while for Re < 2000 the flow is stable or laminar.
For high-pressure gas pipelines, only fully and partially turbulent flow can be
observed. For partially turbulent flow, the friction coefficient can be correlated
with the Reynolds number by the following Prandtl-Von Karman equation:
1 Re
= 4 log10 − 0.6 (7.38)
f 1 f
In the case of fully turbulent flow, the friction factor f can be estimated by the
Nikuradse equation:
1 D
= 4 log10 3.7 (7.39)
f e
7.1.1.1.1 Example
Two pipelines with the same length and same diameter D = 1.0 m are fabricated
from different materials, with pipeline A constructed from glass fiber-reinforced
polymer and pipeline B from steel. Both pipelines carry the same gas with fully
turbulent flow under the same pressure and temperature conditions. The ratio of
the flow capacity V b of the two pipelines can be estimated by using eq. (7.35).
Table 7.3
Pipe roughness.
Material e (inches)
Glass-fiber reinforced pipe 0.0007869
Steel corroded 0.019688
Steel non-corroded 0.001966
252 Flow Capacity of Composite Pipelines
Dividing the flow capacities of the two pipelines, the following formula can be
obtained:
• A
Vb fB
= (7.40)
• B fA
Vb
where VbA , f A are the flow capacity and the friction factor of the glass-fiber
reinforced pipeline, while VbB , f B are the corresponding parameters of the steel
pipeline.
Taking into account equation (7.39), the ratio of the flow capacities can be
written as:
D
•
A log10 3.7 A
V e
b
= (7.41)
•
B D
V b log10 3.7 B
e
• •
VbA ≈ 1.60 VbB for corroded steel (7.42)
• •
VbA ≈ 1.12 VbB for non-corroded steel (7.43)
The above results show that after a period of years, when the steel pipeline of
D = 1.0 m becomes corroded, the pipeline made from glass-fiber reinforced plas-
tic can supply 60% more gas flow. However, even when both pipelines are new,
the gas flow capacity of the composite pipeline is 12% higher than that of the steel
pipeline.
ρ Lu 2
P1 − P2 = f − ∆P∆Η (7.44)
2D
where u is the mean velocity of liquid, ρ is the liquid density and ∆P∆H is the pres-
sure loss due to the elevation ΔΗ (Fig. 7.1).
Liquid Transmission 253
For laminar liquid flow (Re < 2000) the friction factor f is independent of sur-
face roughness and depends only on the Reynolds number:
64
f = (7.45)
Re
Taking into account the definition of the Reynolds number given by equation
(7.37), the combination of equation (7.44) and (7.45) yields:
u=
[ P1 − P2 + ∆P∆Η ] D 2 (7.46)
32 µL
Since the flow is given by:
πD 2
Q= u (7.47)
4
the above equation with the aid of eq. (7.46) can be written as:
π D 4 [ P1 − P2 + ∆P∆Η ]
Q= (7.48)
128µ L
When the difference of the elevation of the cross-sections 1 and 2 (see Fig. 7.1) is
ΔΗ, the parameter ΔPΔΗ is given by the following equation:
∆P∆Η = ws ∆Η (7.49)
π D 4 [ P1 − P2 + ws ∆H ]
Q= (7.50)
128µ L
From the above equation it can be seen that the flow capacity depends on the pa-
rameters D4 and 1/ µ .Therefore, an increase of diameter by a factor 2 yields an
increase of liquid flow rate of 16 times. On the other hand, transmission of a liquid
with double viscosity μ decreases the flow by 50%.
1 D
= 1.14 + 2 log10 (7.51)
f e
Combining the above equation with equations (7.44) and (7.49) yields:
π D 2 ( P1 − P2 + ws ∆Η ) 2.5
Q= 1.14 + 2 log D (7.52)
4 e ρL
It can be seen that the liquid flow rate in fully turbulent conditions depends on
D 2.5 . Therefore, an increase of diameter by a factor 2 will increase the liquid flow
rate by a factor of 5.66.
It should be noted that the estimation of f leads to uncertain results [9] in the tran-
sitional range when the flow changes from laminar to turbulent conditions. However,
since almost all designs concern flow in the fully turbulent range, this consequence
is rather immaterial. A complete plot of friction factor f versus Reynolds number
and roughness is given in Moody’s diagram (Fig. 7.2) for all flow regimes.
• Stratified smooth flow: In stratified smooth flow the gas flows on the top of
the liquid and their interface is smooth (Fig. 7.3a).This regime takes place
with low gas and liquid velocities.
• Stratified wavy flow: In stratified wavy flow, the gas again flows on the
top of the liquid, but when the gas and/or liquid velocity increases, their
interface becomes wavy (Fig. 7.3b).
• Slug flow: In slug flow, the waves of liquid flow are large enough to block
the gas flow (Fig. 7.3c). This regime occurs with higher values of gas flow.
The waves in such cases are called liquid slugs. Very often liquid slugs may
contain gas bubbles.
• Annular flow: With annular flow, the gas flows as a core in the center of the
pipeline and the liquid flow surrounds the gas flow (Fig. 7.3d). This regime
comes into play when gas velocity values are high. The gas core may con-
tain small liquid droplets.
• Dispersed bubble flow: Dispersed bubble flow is characterized by a con-
tinuous liquid phase containing a high density of discrete gas bubbles with
variable size and shape (Fig. 7.3e). This regime happens in conditions of
very low values of gas velocity and high values of liquid velocity.
Taking into account the superficial liquid velocity U sl , the superficial gas ve-
locity U sg , the pressure gradient for single-phase gas flow ( dP dx ) g , the
256 Flow Capacity of Composite Pipelines
Figure 7.3 Multiphase flow regimes: (a) stratified smooth flow, (b) stratified wavy flow,
(c) slug flow, (d) annular flow, (e) dispersed bubble flow.
pressure gradient for single-phase liquid flow ( dP dx )l , the densities of the liq-
uid and gas ρl , ρ g respectively, the liquid kinematic viscosity ν l , the pipeline
inclination angle θ, the inner diameter of the pipe D and the gravitational accelera-
2
tion g = 9.81 m / s , Taitel and Duckler [5] have developed a flow regime map
(Fig. 7.4a) based upon mechanistic models. The coordinate system for the curves
A, B, D is shown in Figs 7.4b and 7.4c.
Multiphase Flow 257
Figure 7.4 Multiphase flow regime map: (a) schematic representation of regimes [5];
(b) coordinate system for the curves A, B ; (c) coordinate system of the curve D.
258 Flow Capacity of Composite Pipelines
h
u g ≥ 1 − l
( ρ − ρ ) A g cos θ + u
l g g
(7.53)
D ( dA
l dhl ) ρg
w
where:
hl
≥ 0.35 (7.54)
D
Multiphase Flow 259
4vl ( ρl − ρ g ) g cos θ
ug ≥ (7.55)
Sc ρ g ul
where:
4 Ag g cos θ ρl − ρ g
ul ≥ (7.56)
Si fl ρl
where :
dp
p ⋅ Al − ( p + ∆x) Al + τ i si ∆x − τ l sl ∆x = 0 (7.57)
dx
260 Flow Capacity of Composite Pipelines
where p is the gas pressure, ∆x is the length of the fluid volume element, τ i and
τ l are the shear stresses at liquid-gas and liquid–pipe interfaces and si , sl are
the cross-sectional lengths at the gas-liquid interface and at the wetted periphery,
respectively (Fig. 7.6).
After several algebraic operations, equation (7.57) can be written as:
dp
− Al + τ i si − τ l sl = 0 (7.58)
dx
Figure 7.6 Geometry of the cross-section of a pipe in a stratified smooth flow regime
(gas-liquid flow).
Multiphase Flow 261
dp
− Ag + τ i si − τ g sg = 0 (7.59)
dx
where S g is the periphery of pipe in contact with gas (Fig. 7.7).
Combining equations (7.58) and (7.59) leads to:
sg sl 1 1
τg −τl + τ i si ( + ) = 0 (7.60)
Ag Al Ag Al
1
τ g = fg ρ g u g2 (7.61)
2
1
τ l = fl ρl ul2 (7.62)
2
1
τ i = fi ρ g (u g − ul ) 2 (7.63)
2
where f g , f l , f i are the corresponding friction factors.
We recall that for laminar and partially turbulent flow the friction factor de-
pends on the Reynolds number given by equation (7.37). Especially for gases, the
Reynolds number should preferably be expressed in terms of the gas flow rate. For
this reason, Kennedy [7] has proposed the following equation:
0.7105 ⋅ Pb ⋅ γ ⋅ Qg
Re = (7.64)
Tb ⋅ µ ⋅ D
Al
Dl = 4 (7.65)
sl
Ag
Dg = 4 (7.66)
sg + si
fi = f g (7.67)
Table 7.4
Friction factors for gas–liquid flow in a laminar,
partially turbulent, and fully turbulent regime.
GAS LIQUID
Laminar 360.31Tb µ g Ag 16 µl sl
Re < 2000 fg = fl =
Pbγ g Qg ( sg + si ) ρl Al ul
Partially
1 0.177 Pbγ g Qg ( sg + si ) f g 1 4ρ A u f
Turbulent = 4 log10 ( ) − 0.6 = 4 log10 ( l l l l ) − 0.6
2000 < Re < 3000 fg Tb µ g Ag fl µl sl
With the aid of the equations given in Table 7.4 and equations (7.61)–(7.63)
and (7.67), eq. (7.60) yields the following flow model for gas-liquid stratified flow
in laminar, partially turbulent and fully turbulent conditions:
f g ρ g u g2 sg 2
f l ρl ul2 sl f i ρ g (u g − ul ) si ( Al + Ag )
− + =0 (7.68)
2 Ag 2 Al 2 Al Ag
It should be noted that for the partially turbulent regime, the friction factor
f must be determined iteratively from the corresponding non-linear equation
(Table 7.4). The gas and liquid velocities are given by the ratios ug = Qg/Ag and
ul = Ql/Al , where Qg, Ql are the gas and liquid volumetric flow rates and Ag , Al
are the pipeline cross–sectional area occupied by gas and liquid, respectively.
Since the parameters Si, Sg, Sl, Ag, Al can be expressed in terms of liquid depth hl
(Fig. 7.6), the parameter hl can be obtained from equation (7.68). Once the liquid
depth hl and the correlated parameters Si, Sg, Sl, Ag, Al are known, the pressure drop
can be calculated using either eq. (7.58) or eq. (7.59).
si = 2 R 2 − (hl − R) 2 (7.69)
or
D2 D
si = 2 − (hl − ) 2 (7.70)
4 2
7.3.1.3.1.2 Evaluation of S g
Using the geometric definition of angle θ, one can determine:
sg = θ R (7.71)
Since
θ = π −ϕ (7.72)
and
hl − R
cos ϕ = (7.73)
si / 2
264 Flow Capacity of Composite Pipelines
With the aid of eq. (7.70), the above equation can now be written as:
hl − D / 2 D
sg = π − arc cos (7.75)
D 2 / 4 − (h − D / 2) 2 2
l
where D = 2R
7.3.1.3.1.3 Evaluation of S l
From Fig. 7.6 it can be shown that
sl + sg = πD (7.76)
D hl − D / 2
sl = π + arc cos (7.77)
2 D 2 / 4 − (h − D / 2) 2
l
7.3.1.3.1.4 Evaluation of Ag
Taking into account the symbols shown in Fig. 7.6, the following property of
the geometry can be written [e.g., 8]:
D − hl
Ag =
6 si
( 3( D − hl )2 + 4si2 ) (7.78)
D − hl
Ag =
2
12 D / 4 − (hl − D / 2) 2
(3( D − h )
l
2
+ 16 ( D 2 / 4 − (hl − D / 2) 2 ) )
(7.79)
Multiphase Flow 265
7.3.1.3.1.5 Evaluation of Al
Since
D2
Al + Ag = π (7.80)
4
a combination of the above equation with eq. (7.79) yields:
πD 2 D − hl
Al =
4
−
12 D / 4 − (hl − D / 2) 2
2
( (
3( D − hl ) 2 + 16 D 2 / 4 − (hl − D / 2) 2 ))
(7.81)
∆p ρl (uT − uF )(us − uF ) τ i si − τ F sF
= + − ρl g sin θ (7.82)
LF LF H lF A
where Δp is the pressure difference in the left and right boundaries of the liquid
film, A is the pipe’s cross-section area, uT is the slug translational velocity, us is
the slug velocity that equals the mixture velocity, uF is the liquid film velocity,
and H lF is the liquid holdup inside the liquid film given by H lF = Vl / V , where
Vl is the volume of the liquid film within the segment LF and V is the whole
266 Flow Capacity of Composite Pipelines
pipeline segment volume. Similarly, the equilibrium equation for the gas volume
lying above the liquid film can be written as:
∆p ρ g (uT − uc )(us − uc ) τ i si + τ c sc
= − − ρ g g sin θ (7.83)
LF LF (1 − H lF ) A
τ F sF LF
ρl (uT − uF )(us − uF ) − ρ g (uT − uc )(us − uc ) − +
H lF A
τ c sc LF 1
+ τ i si LF − ( ρl − ρ g ) g sin θ = 0 (7.84)
(1 − H lF ) A H lF (1 − H lF ) A
In the above equation, the shear stresses τ F , τ c , τ i can be given in the following
formulas [9]:
1
τF = f l ρl uF2 (7.85)
2
1
τc = f g ρ g uc2 (7.86)
2
1
τi = f g ρ g (uc − uF )(uc − uF ) (7.87)
2
where the friction factors can be obtained from Table 7.4.
The geometric parameters sF , si , sc are given by Zhang et al. [10] as:
sF = πDΘ l (7.88)
D2 sin(2 π Θ l )
sF π Θ l − − H lf A + H lF AD sin( π Θ l )
4 2
si = (7.89)
D2 sin(2 π Θ l )
4 π Θ l −
2
πD 2
sc = − sF (7.90)
4
Multiphase Flow 267
where [10]:
0.25 0.8
σ ρg 1 ρlU sl2 D U sg2
Θl = Θlo ( water )0.15 + (7.91)
σ ρl − ρ g cos θ σ 2
(1 − H lF ) gD
In the above equation, Θl , Θlo are the pipe wall fraction wetted by liquid with
curved and flat gas/liquid interfaces, respectively; s water and s are the water and
liquid surface tension. The symbols U sl , U sg represent the liquid and gas super-
ficial velocities given by the following relations:
Ql
U sl = (7.92)
A
Qg
U sg = (7.93)
A
where Ql , Qg are the liquid and gas volumetric flow rates.
With the aid of equations (7.85)–(7.93) and the equations of Table 7.4, equation
(7.84) can provide the value of the liquid holdup inside the liquid film H lF . Once
the H lF and the correlated parameters sF , si , sc are known, the pressure drop can
be calculated by either eq. (7.82) or eq. (7.83).
dp
− Af + τ i si − τ f s f − Af ρl g sin θ = 0 (7.94)
dx
and
dp
− Ac − τ i si − Ac ρ g g sin θ = 0 (7.95)
dx
where � A� f is the cross-sectional area of the liquid film, � Ac is the cross-sectional
area of the gas core, and si , s f are the perimeters of the gas–liquid and liquid–
pipe interfaces given by:
si = π ( D − 2δ ) (7.96)
s f = πD (7.97)
268 Flow Capacity of Composite Pipelines
πD 1 1
τf − τ iπ ( D − 2δ ) + + ( ρl − ρ g ) g sin θ = 0 (7.98)
Af A
f Ac
The cross-sectional areas Af , Ac can be correlated with the liquid film thickness δ:
Af = π ( D − δ )δ (7.99)
D
2
Ac = π − δ (7.100)
2
Moreover, the shear stresses τ i , τ f can be calculated by the following formulas:
1
τi = f g ρ g (u g − ul ) 2 (7.101)
2
1
τf = f l ρl ul 2 (7.102)
2
where the friction factors f g , f l can be obtained from Table 7.4 depending on
the flow regime. When the volumetric flow rates Qg � � and � Ql of the gas and liquid
respectively are known, the corresponding gas and liquid velocities can be esti-
mated by:
Qg
ug = (7.103)
Ac
Ql
ul = (7.104)
Af
Therefore, with the aid of equations (7.99)–(7.104), equation (7.98) can provide
the liquid film thickness δ. Consequently, when the thickness δ is known, the pa-
rameters si , s f , Ac , Af , τ i , τ f can be obtained by equations (7.96), (7.97), (7.99),
(7.100), and (7.101–7.104). Using the above results, the pressure drop for the
annular gas–liquid flow model can be estimated by either eq. (7.94) or eq. (7.95).
References
[1] Katz et al., Handbook of gas engineering, McGraw-Hill, 1959.
[2] Simon A.L., Hydraulics, John Wiley & Sons, 1986.
[3] Song S.H., Characterization and metering of multiphase mixtures from
deep subsea wells, Ph.D. dissertation, The University of Texas at Austin,
1994.
[4] Bergles A.E., Collier J.G., Delhaye J.M., Hewitt G.F., and Mayinger F.,
Two-phase flow and heat transfer in the power and process industries,
McGraw-Hill, 1981.
[5] Taitel Y., and Dukler A.E., “A model for predicting flow regime transitions
in horizontal and near horizontal gas-liquid flow,” AIChE Journal, 22, 1,
1976.
[6] Barnea D., Shoham O., and Taitel Y., “Flow pattern transition for vertical
downward inclined two-phase flow: Horizontal to Vertical,” Chem. Eng.
Sci., 37, 1982.
[7] Kennedy J.L., Oil and gas pipeline fundamentals, Pennwell, Tulsa, 1993.
[8] Gieck K., and Gieck R., Engineering formulas, McGraw-Hill, 2006.
[9] Guo B., Song S., Chacko J., and Ghalambor A., Offshore pipelines, Elsevier,
2005.
[10] Zhang H.Q., Wang Q., Sarica C., and Brill J.P., “Unified model for gas-
liquid pipe flow via slug dynamics-Part 1: Model Development,” Journal of
Energy Resources Technology, 125, 4, 2003.
Chapter 8
Therefore, for any specific material, only parameters (c), (d), (e) can control
the amount of the composite material needed for manufacturing a safe pipeline—
and, thus, eventually, the costs. Among these parameters, the diameter is cor-
related with the required flow capacity of the pipeline, and its estimation will be
based on a supply/demand forecast. On the other hand, the quantity of the material
is linearly dependent on the number of plies. Therefore, the present discussion
concerning the optimization of the material cost will focus on fiber orientation, θ.
271
272 Optimization of Material Cost
±30°, ±45°, ±60°, ±75°. From the foregoing values of θ, the values presented in
Table 8.1 maximize the corresponding allowable forces and, therefore, minimize
the material quantity.
When a combination of the pure loading cases is used, the optimum fiber ori-
entation will vary between the limiting values that correspond to the pure (uncom-
bined) loading cases. For a pipe with diameter Dia = 0.4 m consisting of NP =
50 plies that is subjected to combined bending moment and external pressure, the
allowable bending moment M (Nm) for four values of external pressure, namely p
= 100, 150, 200, 250 kPa, is shown below in Figure 8. 1.
From this figure it can be concluded that an increase in the external pressure
drives the value of the optimum fiber orientation (θ corresponding to maximum
M) towards the optimum fiber orientation of the pure pressure. Figure 8.2 cor-
relates the optimum fiber orientation θ shown in Fig. 8.1 with the value of the
external pressure.
For loading combinations, the value of the optimum fiber orientation can be
estimated with the aid of the values of the optimum fiber orientation of the pure
loads. Using linear interpolation, the following procedures for three types of load-
ing combinations can be derived.
8.1.1 Optimum fiber orientation for the combination of axial tension and
external pressure
Since the optimum fiber orientations for pure axial tension and pure external
pressure are 0° and 90° respectively, the optimum fiber orientation θop for any
combination of the above loading cases will have a value between 0° and 90°. It is
obvious that for high values of external pressure and low values of axial tension,
θop will have a value near 90°. For loading combinations where the axial tension
is predominant, θop will be a value near 0°. For any pipe diameter and a certain
number of plies that constitute the pipe’s wall, the allowable pure axial tension
Nα for a fiber orientation 0° and the allowable pure external pressure for a fiber
Table 8.1
Optimum fiber orientation θ (deg).
LOADING FAILURE BUCKLING
TYPE E-Glass/Epoxy S-Glass/Epoxy E-Glass/Epoxy S-Glass/Epoxy
Axial Tension 0° 0° — —
Extrnal Pressure 90° 90° 0° 0°
Bending 30° 30° 0° 0°
Torsion 45° 45° 60° 60°
Fiber Orientation and Loading Forces 273
Figure 8.1 Allowable bending moment versus fiber orientation of a pipe with diameter
Dia = 0.4 m and NP = 50 plies subjected to external pressure p = 100, 150, 200, 250 KPa.
orientation of 90° can create an envelope (Fig. 8.3), which indicates the optimum
fiber orientation for any combination (N, p). Assuming that the hypotenuse of the
triangle of Fig. 8.3 represents the variation of the optimum fiber orientation from
0° (optimum θ for pure axial tension) to 90° (optimum θ for pure external pres-
sure), then the intersection of the direction OC, corresponding to a combination
(N, p) with the hypotenuse, can provide an approximation of θ°P for the combined
loading of axial loading and external pressure (N, p).
In order to derive an analytic formula for the estimation of optimum fiber ori-
entation for a loading combination (N, p), the geometric properties of the triangle
OHD, shown in Fig. 8.4, will be used. According to this figure the following geo-
metric properties can be written as:
GB CA
= (8.1)
OB OA
and
Figure 8.2 Correlation of the optimum fiber orientation with the value of the external
pressure for the loading case of Figure 8.1.
Figure 8.3 Envelope for definition of the optimum fiber orientation for any combination
of axial tension and external pressure (N, p).
Fiber Orientation and Loading Forces 275
Figure 8.4 Geometric model for definition of optimum fiber orientation θop for any com-
bination of axial tension N with external pressure p.
Using the loading quantities corresponding to the sides of the triangles, the
above equations can be written as:
Ν1 Ν
= (8.3)
p1 p
Ν1 Ν a − Ν1
= (8.4)
pa − p1 p1
pa Ν a Ν
Ν1 = (8.5)
p a Ν + pN a
and
pa Ν a p
p1 = (8.6)
p a Ν + pN a
276 Optimization of Material Cost
With the aid of equations (8.5), (8.6), the segments α, β in Figure 8.4 can now be
determined:
β = ( Να − Ν1 ) 2 + p12 (8.7)
a = ( pa − p1 ) 2 + Ν12 (8.8)
Taking into account equations (8.5)–(8.8), the optimum fiber orientation for a
loading combination (N, p) can now be estimated:
β
θ op = 90 (8.9)
α +β
8.1.2 Optimum fiber orientation for the combination of bending and axial
tension
The envelope shown in Figure 8.5 corresponds to the combined bending mo-
ment M with an axial tensile force N.
Adhering to the procedure of the previous section, we obtain the following
formulas:
Μa Νa Ν
Ν1 = (8.10)
Μ a Ν + ΜN a
Μa Νa Μ
Μ1 = (8.11)
Μ a Ν + ΜN a
β = ( Να − Ν1 ) 2 + Μ12 (8.12)
β
θ op = 30 (8.14)
α +β
It should be noted that for the case of the (N, M) combination, θop has values
between 0° (optimum fiber orientation of the pure axial load) and 30° (optimum
fiber orientation of pure bending).
Fiber Orientation and Loading Forces 277
Figure 8.5 Definition of optimum fiber orientation θop for a combination of bending and
axial tension.
8.1.3 Optimum fiber orientation for the combination of bending and external
pressure
The corresponding envelope for the combination of bending moment M with
external pressure p is shown in Figure 8.6.
Figure 8.6 Definition of optimum fiber orientation θop for a combination of bending and
external pressure.
278 Optimization of Material Cost
Using the concepts explained in the previous sections, the following formulas
can be used for the estimation of θop for a combination of bending moment and
external pressure.
pa Μ a Μ
Μ1 = (8.15)
p a Μ + pΜ a
pa Μ a p
p1 = (8.16)
p a Μ + pΜ a
a = ( pa − p1 ) 2 + Μ12 (8.18)
β
θ op = 30 + (90 − 30 ) (8.19)
α +β
With reference to Figure 8.6, it should be mentioned that for the loading combina-
tion (M, p), θop has values between 30° (optimum fiber orientation of pure bend-
ing) and 90° (optimum fiber orientation of pure pressure).
Chapter 9
279
280 Quality Control of Composite Pipe Systems
to decades. Although creep testing can include any kind of loading, uniaxial or
bi-axial, the most common type is the tensile uniaxial creep test. To overcome the
disadvantage of the long-term testing requirements to complete a creep test, much
research is carried out with the objective of achieving accelerated characteriza-
tion of creep in composite materials. Since creep testing is the most common
testing for composites, detailed specifications (e.g., ASTM D 2990) are available
for comparison and validation of material properties. A representation of creep
testing results is shown in Fig. 9.1ab.
The configuration of straight-sided specimens used for creep testing is the same
as the configuration of the specimens for the static tensile test (ASTM D 3039).
However, for compressive creep specimens, special stiffening guides should be
placed on both flat sides of specimens in order to avoid buckling. The usual di-
mensions for uniaxial creep specimens are 150 mm for length, 13 mm for width
and 3 mm for thickness. Suitable support devices for holding the specimen’s ends
during creep testing are ones containing vise-like end tabs with inclined surfaces
to increase grip pressure and facilitate specimen alignment (Fig. 9.2).
Computerized servo-hydraulic machines are currently the best choice for per-
forming displacement-controlled or load-controlled creep tests. A typical creep-
testing machine is equipped with an environmental damper and heater for control-
ling the test conditions. The load capacity of creep machines varies from 1000 lbf
to 50000 lbf. The load acting on the specimen is measured by a pressure sensor
placed in a hydraulic network. The sensor is calibrated to provide a digital output
of the instant value of the load. Since conventional extensometers have the dis-
advantage of measuring a specimen’s deformation in only one direction, during
creep testing, strain is frequently measured by strain gages, which are a low-cost
solution and provide the capacity of multidirectional strain measurements.
Figure 9.1 Typical creep testing results for a certain temperature: (a) creep strain versus
time, (b) applied stress vs. time to failure.
282 Quality Control of Composite Pipe Systems
R(t ) = m γ (t ) + m g
(9.1)
U (t ) = ∫ γ (t )dt (9.2)
Figure 9.3 Schematic representation of impact testing results: (a) absorbed energy versus
time, (b) specimen resistance versus time.
dx
U (t ) = (9.4)
dt
eq. (9.3) yields:
dx
E (t ) = ∫ R(t ) dt (9.5)
dt
or
With the aid of equations (9.1), (9.2), the above equation yields:
R(t )
E (t ) = ∫ R(t ) − g dt (9.7)
m
The Charpy impact test is older than the drop-weight one. It was used for
the experimental determination of the fracture toughness of metals. Due to its
284 Quality Control of Composite Pipe Systems
simplicity, this method has been adopted for composites testing too. The Charpy
machine (Fig. 9.4) is equipped with a heavy pendulum that is raised to a known
height and released, impacting a prismatic specimen. The difference between the
initial and final heights H and h respectively is proportional to the amount of the
absorbed energy due to damage accumulation in the specimen. Therefore, the
total absorbed energy can be estimated by the formula:
Etot = G ( H − h) (9.8)
s max − s min
sa = (9.9)
2
s + s min
sm = (9.10)
2
In engineering practice, graduated variable amplitude loading histories (Fig. 9.6)
are often introduced.
Figure 9.5 Fatigue loading types: (a) harmonic, and (b) irregular.
286 Quality Control of Composite Pipe Systems
The test specimens (Fig. 9.7) may contain holes or notches or be smooth.
Standard dimensions for fatigue specimens are 195 mm for length and 12.5–
25 mm for effective width.
The aim of fatigue testing is the derivation of curves that describe the fatigue
life, i.e., the number of loading cycles up to failure versus the stress amplitude
of fatigue loading. These curves are called S-N or Woehler curves. A schematic
representation of the procedure for the derivation of an S-N curve is shown in
Fig. 9.8.
Figure 9.8 Derivation of an S-N curve from fatigue tests with differing stress amplitudes.
Table 9.1
Standards for composite pipes.
Code Subject
ASTM D 2996 Standard Specification for Filament-Wound Reinforced Thermosetting
Resin Pipe, 1–16 inches (25-400 mm) in diameter.
ASTM D 2517 Standard Specification for Reinforced Thermosetting Resin Pipe.
ASTM D 2997 Standard Specification for Centrifugally Cast, Reinforced Thermosetting
Resin Pipe.
ASTM D 3262 Standard Specification for Reinforced Plastic Mortar Sewer Pipe.
ASTM D 3517 Standard Specification for Fiberglass Pressure Pipe.
ASTM D 3754 Standard Specification for Fiberglass Sewer and Industrial Pressure Pipe,
8-144 inches in diameter (200–3600 mm).
ASTM D 4024 Standard Specification for Reinforced Thermosetting Resin Flanges.
ASTM D 4161 Standard Specification for Fiberglass Pipe Joints using Flexible Elastomeric
Seals.
ASTM D 5686 Standard Specification for ‘Fiberglass’ (Glass Fiber-Reinforced
Thermosetting Resin) Pipes and Fittings, Adhesive Bonded Joint Type
Epoxy Resin, for Condensate Return Lines.
ASTM D 3567 Standard Practice for Determining Dimensions of Reinforced
AMERICAN STANDARDS
Code Subject
ASTM D 2992 Obtaining Hydrostatic Design Basis for Reinforced Thermosetting Resin
Pipe and Fittings, Procedure A, Cyclic/Procedure B, Static.
ASTM D 2412 External Loading Characteristics of Plastic Pipe by Parallel Plate Loading.
AMERICAN STANDARDS (CONT’D)
Code Subject
BS 3974 Specification for Pipe Supports. Part 1: Pipe hangers, slider and roller-type sup-
ports. Part 2: Pipe clamps, cages, cantilevers and attachments to beams. Part 3:
Large-bore, high-temperature, marine and other applications.
BRITISH STANDARDS
BS 5350 Method of Test for Adhesives. Part C5: Determination of bond strength in longi-
tudinal shear.
BS 5480 Specification for Glass Fiber-Reinforced Plastics (FRP) Pipes and Fittings for Use
for Water Supply of Sewage. Part 1: Dimensions, materials and classifications.
Part 2: Design and performance requirements.
BS 6464 Specification for Reinforced Plastics Pipes, Fittings and Joints for Process Plants.
BS 7159 Design and Construction of Glass Reinforced Plastics (FRP) Piping Systems for
Individual Plants of Sites.
BS 8010 Code of Practice for Pipeline. Section 2.5. Glass reinforced thermosetting
plastics.
290 Quality Control of Composite Pipe Systems
Code Subject
DIN 16 867 Glass fiber-reinforced polyester resin (UP-GF) Pipes. Fittings and joints for
chemical pipelines. Technical delivery conditions.
DIN 16 868 Glass fiber-reinforced unsaturated polyester resin (UP-GF) Pipes. Part 1:
Wound, filled, dimensions. Part 2: Wound, filled. General quality.
DIN 16 869 Centrifugally cast filled fiber-reinforced unsaturated polyester resin
(UP-GF) Pipes. Part 1: Dimensions. Part 2: General quality requirements,
testing.
DIN 16 870-1 Wound glass fiber-reinforced epoxy pipes; dimensions.
DIN 16 871 Centrifugally cast glass fiber-reinforced epoxy pipe; dimensions.
DIN 16 964 Wound glass fiber-reinforced polyester resin (UP-GF) pipes, general quality
requirements. Testing.
DIN 16 965 Parts 1, 2, 4 and 5: Wound glass fiber-reinforced polyester resin pipes, types
GERMAN STANDARDS
A, B, D and E; dimensions.
DIN16 966 Part 1 – Glass fiber-reinforced polyester resin pipes. Fittings and joints.
General Quality requirements, testing.
Part 2- Elbows. Dimensions.
Part 4- Tees and nozzles. Dimensions.
Part 5 - Reducers. ngs and joints, bushings, flanges, flanged and butted
joints. General quality Requirements, testing.
Part 8 –Laminated joints. Dimensions.
DIN 19565-1 Centrifugally cast and filled polyester resin glass fiber-reinforced (UP-GF)
Pipes and fittings for buried drains and sewers, dimensions and technical
delivery conditions.
DIN 53 769 Part 1—Testing of glass fiber-reinforced plastic pipelines, determination of
the adhesive shear strength of type B pipeline components.
Part 2—Testing of glass fiber-reinforced plastic pipes; long-term hydro-
static pressure test.
Part 3—Testing of glass fiber-reinforced plastic pipes; short-term flattening
test and flattening endurance.
Part 6—Testing of glass fiber-reinforced plastic pipes; testing of pipes and
fittings under pulsating conditions.
DIN 54 815 Pipes of filled polyester resin molding materials. Part 1: Dimensions, mate-
rials, designation. Part 2: Requirements, testing.
International Standards for Composite Pipes 291
Code Subject
T57 200 Pipes and Fittings of composite glass thermosetting materials. General review.
Description. Classification. Characteristics.
T57 201 Pipes and Fittings in FRP. Test to determine the hoop rigidity.
T57 202 Reinforced plastic pipes. Sealing ring type joints for installation under pressure
FRENCH STANDARDS (AFNOR)
Code Subject
JIS K 7013 Fiber-reinforced plastic pipes.
JIS K 7014 Fittings and joints for fiber-reinforced plastic pipes.
JIS K 7020 Glass reinforced thermosetting plastics (GRP) pipes and fittings. Methods for
regression analysis and their use.
JIS K 7030 Pipes and fittings made of glass fiber reinforced plastics (GRP). Definitions
of terms relating pressure, including relationship between them, and terms for
installation and jointing.
JAPANESE STANDARDS
JIS K 7031 Plastics piping systems. Glass reinforced thermosetting plastics (GRP) pipe
and fittings. Test methods to prove the watertightness of the wall under short-
term internal pressure.
JIS K 7032 Plastic piping systems. Glass reinforced thermosetting plastics (GRP) pipes.
Determination of initial specific ring stiffness.
JIS K 7033 Plastic piping systems. Glass reinforced thermosetting plastics (GRP) pipes.
Determination of initial tensile properties.
JIS K 7034 Plastics piping systems. Pipes made of glass reinforced thermosetting plastic
(GRP). Determination of the resistance to chemical attack for the inside of a
section in a deflected condition.
JIS K 7035 Plastic piping systems. Glass reinforced thermosetting plastics (GRP) pipes.
Determination of the creep factor under wet conditions and calculation of the
long-term specific ring stiffness.
JIS K 7036 Plastics piping systems. Glass reinforced thermosetting plastics (GRP) pipes
and fittings. Test methods to prove the design of bolted flanged joints.
292 Quality Control of Composite Pipe Systems
Code Subject
JAPANSESE (CONT’D)
JIS K 7037 Plastics piping systems. Glass reinforced thermosetting plastics (GRP) pipes.
Determination of the apparent initial circumferential tensile strength.
JIS K 7038 Plastics piping systems. Glass reinforced thermosetting plastics (GRP) pipes.
Test method to prove the resistance to initial ring deflection.
JIS K 7039 Plastics piping systems. Glass reinforced thermosetting plastics (GRP) pipes.
Determination of the long-term ultimate bending strain and calculation of the
long-term ultimate relative ring deflection, both under wet conditions.
JIS K 7040 Plastics piping systems. Glass reinforced thermosetting plastics (GRP) pipes
and fittings. Test methods to prove the design of cemented or wrapped joints.
Code Subject
EN NF DIN 705 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)
pipes and fittings. Methods for regression analyses and their use.
EN NF DIN 761 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)
pipes. Determination of the creep factor under dry conditions.
EN NF DIN 1115 Plastic piping systems for underground drainage and sewerage under
pressure. Glass reinforced thermosetting plastics (GRP) based on poly-
ester resin.
Part 1: General.
Part 2: Pipes with flexible, reduced articulation or rigid joints.
Part 3: Fittings.
Part 4: Ancillary equipment.
EUROPEAN STANDARDS (EN)
Code Subject
EN NF DIN 1393 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)
pipes. Determination of initial longitudinal tensile properties.
EN NF DIN 1394 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)
pipes. Determination of the apparent initial circumferential tensile
strength.
EN NF DIN 1447 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)
pipes. Determination of a long-term resistance to internal pressure.
EN NF DIN 1448 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)
components. Test methods to prove the design of rigid locked socket and
spigot joints with elastomeric seals.
EUROPEAN STANDARDS (CONT’D)
EN NF DIN 1449 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)
components. Test methods to prove the design of cemented socket and
spigot joints.
EN NF DIN 1450 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)
components. Test methods to prove the design of bolted flanged joints.
EN NF DIN 1636 Plastic piping systems for non-pressure drainage sewerage. Glass rein-
forced thermosetting plastics (GRP).
Part 1: General.
Part 2: Pipes with flexible reduced articulation or rigid joints.
Part 4: Ancillary equipment.
EN NF DIN 1638 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)
pipes. Test methods for the effects of cyclic internal pressure.
EN NF DIN 1796 Plastic piping systems for water supply with or without pressure. Glass
reinforced thermosetting plastics (GRP) based on polyester resin (UP).
Part 1: General.
Part 2: Pipes with flexible, reduced articulation or rigid joints.
Part 4: Ancillary equipment.
Part 5: Fitness for purpose of the system.
Part 6: Recommended practice for installation.
EN NF DIN 1862 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)
pipes. Determination of the relative flexural creep factor following expo-
sure to a chemical environment.
Code Subject
ISO DIS 7370 Glass fiber-reinforced thermosetting plastics (GRP) pipes and fittings.
Nominal diameters, specified diameters and standard lengths.
INTERNATIONAL (ISO)
ISO DIS 7509 Glass reinforced thermosetting plastics (GRP) pipes. Determination of
time to failure under sustained internal pressure.
ISO 7510 Plastics piping systems. Glass reinforced thermosetting plastics (GRP)
pipes and fittings. Test methods to prove the watertightness of the wall
under short-term internal pressure.
ISO 7511 Glass fiber-reinforced thermosetting plastic (GRP) pipes and fittings. Test
methods to prove the watertightness of the wall under short-term internal
pressure.
294 Quality Control of Composite Pipe Systems
Code Subject
ISO 7684 Plastics piping systems. Glass reinforced thermosetting plastics (GRP)
pipes. Determination of the creep factor under dry conditions.
ISO 7685 Plastics piping systems. Glass reinforced thermosetting plastics (GRP).
Determination of initial specific ring stiffness.
ISO DIS 8483 Glass reinforced thermosetting plastics (GRP) pipes and fittings. Test
methods to prove the design of bolted flanged joints.
ISO DIS 8513 Glass reinforced thermosetting plastics (GRP) pipes. Determination of
initial longitudinal properties.
ISO 8521 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)
pipes. Determination of the apparent initial circumferential tensile
strength.
ISO DIS 8533 Glass reinforced thermosetting plastics (GRP) pipes and fittings. Test
method to prove the design of cemented wrapped joints.
ISO 8572 Pipes and fittings made of glass reinforced thermosetting plastics (GRP).
Definitions of terms relating to pressure, including relationships between
INTERNATIONAL STANDARDS (ISO) (CONT’D)
enters one end of a fiber optic cable, the light is completely reflected within the
cable, even in the presence of curvature. Changes in strain due to localized dam-
age affect the transmittance behavior of the optical fiber. Detection of the changes
in the intensity or frequency of the light signal provides information regarding the
type and severity of damage. EFPI (Extrinsic Fabry-Perot Interferometric) and
FBG (Fiber Bragg Grating)-type sensors are used for structural health monitoring.
The EFPI type is based on detection of the change of displacement between the
two ends of a cleaved optical fiber. The spacing of the two ends is on the order of
100μm. One fiber end transmits multiple frequency light signals, while the other
end acts as a receiver. The damage-induced strain changes the displacement be-
tween the two ends, resulting in phase differences between the reflected waves.
The FBG type sensor is based on strain-induced shifting in the light wavelengths
(corresponding to Bragg wavelength [5]). The shift in the Bragg wavelength is
detected by opto-electrical sensors and translated to strain.
References
[1] Adams D., Carlsson L., and Pipes B., Experimental characterization of ad-
vanced composite materials, CRC Press, 2003.
[2] Jenkins C.H., editor, Manual on experimental methods for mechanical test-
ing of composites, The Fairmont Press, Inc. Lilburn, GA, 1998.
[3] Kessler M., Advanced topics in characterization of composites, Trafford,
2004.
[4] Dallas G., “Thermal analysis,” ASM Handbook-Composites, ASM
International, 2001.
[5] Hare D., and Moore T.C., “Characteristics of extrinsic Fabry-Perot inter-
ferometric (EFPI) fiber-optic strain gages,” NASA Technical Publication,
NASA/TP-2000-210639, 2000.
Chapter 10
Case Studies
Introduction
Chapter 10 presents a collection of nomographs for the direct mechanical de-
sign of GFRP composite pipes under a wide range of pure and combined load-
ing conditions. These diagrams contain data that are based on model behavior
of composite materials, mainly E-glass/epoxy and S-glass/epoxy, given multiple
loading parameters. Data are also provided on parameters for the construction of
pipelines, hanger supports, and appropriate depths for underground pipe.
299
300 Case Studies
DĂƚĞƌŝĂů 'ĞŶĞƌŝĐ^Ͳ'ůĂƐƐͬƉŽdžLJh
dŚŝĐŬŶĞƐƐŽĨůĂLJĞƌ;ŵŵͿ Ϭ͘ϭϱϬ
&ŝďĞƌǀŽůƵŵĞĨƌĂĐƚŝŽŶ;йͿ Ϭ͘ϱϬ
>ĂŵŝŶĂƚĞĚĞŶƐŝƚLJ;ŐͬĐŵϯͿ Ϯ͘ϬϬ
KƌŝĞŶƚĂƚŝŽŶŽĨĨŝďĞƌƐ;ƌĂĚͿ ʋͬϭϮ
SD N3D
'LDP
332 Case Studies
External Pressure 333
334 Case Studies
10.4 Torsion
10.4.1 Results of failure model for torsion
Taking into account the model for torsion explained in Chapter 3, the allowable
torsion moment My has been estimated for pipes made from: (a) E-glass/epoxy
and (b) S-glass/epoxy. The following diagrams present the allowable values My
for pipes of diameter: Dia = 0.10–1.20 m constituted by plies of thickness 0.150
mm, fiber orientation θ = ±15°, ±30°, ±45°, ±60°, ±75° for a number of plies NP
= 10–50.
Number of Layers = 10
0.1 131868 0.1 102498 0.1 123046 0.1 131868 0.1 135402
0.2 275026 0.2 263637 0.2 273558 0.2 275026 0.2 275241
0.3 412909 0.3 410338 0.3 412773 0.3 412909 0.3 412917
0.4 550556 0.4 550052 0.4 550546 0.4 550556 0.4 550557
0.5 688196 0.5 688103 0.5 688195 0.5 688196 0.5 688196
0.6 825835 0.6 825819 0.6 825835 0.6 825835 0.6 825835
0.7 963474 0.7 963471 0.7 963474 0.7 963474 0.7 963474
0.9 1238750 0.9 1238750 0.9 1238750 0.9 1238750 0.9 1238750
1.0 1376390 1.0 1376390 1.0 1376390 1.0 1376390 1.0 1376390
1.1 1514030 1.1 1514030 1.1 1514030 1.1 1514030 1.1 1514030
1.2 1651670 1.2 1651670 1.2 1651670 1.2 1651670 1.2 1651670
llowable bending moment
Number of Layers = 10
0.1 4893.22 0.1 8159.01 0.1 9794.65 0.1 10496.9 0.1 10778.2
0.2 16314.4 0.2 20989 0.2 21770.7 0.2 21887.5 0.2 21904.7
0.3 29376.1 0.3 32654.7 0.3 32848.5 0.3 32859.4 0.3 32860
0.4 41975.6 0.4 43772.6 0.4 43811.9 0.4 43812.7 0.4 43812.7
0.5 53875.6 0.5 54758.2 0.5 54765.5 0.5 54765.5 0.5 54765.5
0.6 65307.8 0.6 65717.1 0.6 65718.4 0.6 65718.4 0.6 65718.4
0.7 76487.7 0.7 76671.1 0.7 76671.3 0.7 76671.3 0.7 76671.3
0.8 87543.9 0.8 87624.2 0.8 87624.2 0.8 87624.2 0.8 87624.2
0.9 98542.6 0.9 98577.1 0.9 98577.1 0.9 98577.1 0.9 98577.1
1.0 109515 1.0 109530 1.0 109530 1.0 109530 1.0 109530
1.1 120477 1.1 120483 1.1 120483 1.1 120483 1.1 120483
1.2 131433 1.2 131436 1.2 131436 1.2 131436 1.2 131436
llowable axial tension
Number of Layers = 20
0.1 64887.1 0.1 122943 0.1 169940 0.1 204997 0.1 229584
0.2 245886 0.2 409994 0.2 492185 0.2 527472 0.2 541610
0.3 509820 0.3 738277 0.3 804245 0.3 820675 0.3 824611
0.4 819988 0.4 1054940 0.4 1094230 0.4 110010 0.4 1100970
0.5 1147920 0.5 1354020 0.5 1374350 0.5 1376210 0.5 1376370
0.6 1476550 0.6 1641350 0.6 1651090 0.6 1651640 0.6 1651670
0.7 1797980 0.7 1922330 0.7 1926790 0.7 1926940 0.7 1926950
0.8 2109890 0.8 2200210 0.8 2202180 0.8 2202230 0.8 2202230
0.9 2412740 0.9 2476640 0.9 2477490 0.9 2477500 0.9 2477500
1.0 2708050 1.0 2752410 1.0 2752780 1.0 2752780 1.0 2752780
1.1 2997530 1.1 3027910 1.1 3028060 1.1 3028060 1.1 3028060
1.2 3282700 1.2 3303270 1.2 3303340 1.2 3303340 1.2 3303340
llowable bending moment
Number of Layers = 20
0.1 5169.74 0.1 9795.25 0.1 13539.6 0.1 16332.7 0.1 18291.6
0.2 19572.9 0.2 32636.1 0.2 39178.6 0.2 41987.4 0.2 43112.9
0.3 40575.6 0.3 58758.1 0.3 64008.3 0.3 65316 0.3 65629.2
0.4 65257.4 0.4 83956 0.4 87082.9 0.4 87550 0.4 87618.6
0.5 91353 0.5 107755 0.5 109373 0.5 109520 0.5 109534
0.6 117504 0.6 130619 0.6 131394 0.6 131438 0.6 131440
0.7 143082 0.7 152978 0.7 153333 0.7 153345 0.7 153345
0.8 167903 0.8 175090 0.8 175247 0.8 175251 0.8 175251
0.9 192002 0.9 197087 0.9 197156 0.9 197156 0.9 197156
1.0 215502 1.0 219033 1.0 219062 1.0 219062 1.0 219062
1.1 238538 1.1 240955 1.1 240968 1.1 240968 1.1 240968
1.2 261231 1.2 262868 1.2 262874 1.2 262874 1.2 262874
llowable axial tension
Number of Layers = 30
0.1 65570.9 0.1 127916 0.1 184415 0.1 233430 0.1 274370
0.2 255832 0.2 466861 0.2 614991 0.2 707587 0.2 761338
0.3 553245 0.3 922486 0.3 1107420 0.3 1186810 0.3 1218620
0.4 933722 0.4 1415170 0.4 1582410 0.4 1632140 0.4 1646220
0.5 1371850 0.5 1903340 0.5 2031040 0.5 2057780 0.5 2063210
0.6 1844970 0.6 2373620 0.6 2462030 0.6 2475230 0.6 2477170
0.7 2335300 0.7 2825920 0.7 2883500 0.7 2889690 0.7 2890340
0.8 2830350 0.8 3264280 0.8 3300310 0.8 3303110 0.8 3303320
0.9 3322250 0.9 3693040 0.9 3714950 0.9 3716180 0.9 3716250
1.0 3806690 1.0 4115560 1.0 4128620 1.0 4129150 1.0 4129170
1.1 4281800 1.1 4534190 1.1 4541860 1.1 4542090 1.1 4542090
1.2 4747240 1.2 4950470 1.2 4954910 1.2 4955010 1.2 4955010
llowable bending moment
Number of Layers = 30
0.1 5232.05 0.1 10206.7 0.1 14714.9 0.1 18262 0.1 21892.7
0.2 20372.2 0.2 37176.7 0.2 48972.4 0.2 56346 0.2 60626.2
0.3 44039 0.3 73431.1 0.3 88151.8 0.3 94471.8 0.3 97004
0.4 74315.7 0.4 112635 0.4 12594.8 0.4 129904 0.4 131024
0.5 109180 0.5 151480 0.5 161642 0.5 163771 0.5 164203
0.6 146829 0.6 188901 0.6 195964 0.6 196988 0.6 197142
0.7 185848 0.7 224892 0.7 229474 0.7 229967 0.7 230019
0.8 225241 0.8 259774 0.8 262642 0.8 262864 0.8 262881
0.9 264385 0.9 293892 0.9 295636 0.9 295734 0.9 295740
1.0 302935 1.0 327515 1.0 328554 1.0 328596 1.0 328598
1.1 340742 1.1 360828 1.1 361438 1.1 361456 1.1 361456
1.2 377781 1.2 393953 1.2 394307 1.2 394314 1.2 394315
llowable axial tension
Number of Layers = 40
0.1 65814.3 0.1 129774 0.1 190228 0.1 245886 0.1 295903
0.2 259548 0.2 491773 0.2 679761 0.2 819988 0.2 918337
0.3 570685 0.3 1019640 0.3 1310740 0.3 1476550 0.3 1564110
0.4 983546 0.4 1639980 0.4 1968740 0.4 2109890 0.4 2166440
0.5 1479510 0.5 2295840 0.5 2606850 0.5 2708050 0.5 2739250
0.6 2039280 0.6 2953110 0.6 3216980 0.6 3282700 0.6 3298440
0.7 2644690 0.7 3595960 0.7 3304560 0.7 3844670 0.7 3852180
0.8 3279950 0.8 4219770 0.8 4376640 0.8 4400420 0.8 4403860
0.9 3932220 0.9 4825470 0.9 4939930 0.9 4953270 0.9 4954810
1.0 4591680 1.0 5416100 1.0 5497410 1.0 5504830 1.0 5505500
1.1 5251270 1.1 5995060 1.1 6051760 1.1 6055810 1.1 6056100
1.2 5906220 1.2 6565400 1.2 6604360 1.2 6606550 1.2 6606670
llowable bending moment
Number of Layers = 40
0.1 5262.48 0.1 10376.7 0.1 15210.5 0.1 19660.9 0.1 23660.2
0.2 20679 0.2 39181 0.2 54158.5 0.2 65330.8 0.2 73166.6
0.3 4543.9 0.3 8113.7 0.3 104361 0.3 117563 0.3 124534
0.4 78291.6 0.4 130544 0.4 156714 0.4 167950 0.4 172451
0.5 117759 0.5 182732 0.5 207486 0.5 215541 0.5 218024
0.6 162303 0.6 235032 0.6 256033 0.6 261264 0.6 262517
0.7 210478 0.7 286186 0.7 302787 0.7 305979 0.7 356577
0.8 261030 0.8 335824 0.8 348332 0.8 350200 0.8 350474
0.9 312935 0.9 384022 0.9 393130 0.9 394192 0.9 394315
1.0 365412 1.0 431020 1.0 437431 1.0 438081 1.0 438135
1.1 417899 1.1 477091 1.1 481603 1.1 481925 1.1 481948
1.2 470018 1.2 522476 1.2 525576 1.2 525750 1.2 525760
llowable axial tension
Number of Layers = 50
0.1 65927.7 0.1 130656 0.1 193073 0.1 252222 0.1 307358
0.2 261313 0.2 504443 0.2 715940 0.2 889420 0.2 1024980
0.3 579218 0.3 1073910 0.3 1442610 0.3 1690450 0.3 1845690
0.4 1008890 0.4 1778840 0.4 2253930 0.4 2509700 0.4 2637360
0.5 1536790 0.5 2562460 0.5 3076160 0.5 3296700 0.5 3385060
0.6 2147820 0.6 3380890 0.6 3877360 0.6 4048000 0.6 4103380
0.7 2826440 0.7 4205270 0.7 4650080 0.7 4773230 0.7 4805830
0.8 3557680 0.8 5019400 0.8 5397330 0.8 5482120 0.8 5500520
0.9 4327830 0.9 5816040 0.9 6125050 0.9 6181520 0.9 6191590
1.0 5124920 1.0 6593390 1.0 6838960 1.0 6875650 1.0 6881030
1.1 5938940 1.1 7352570 1.1 7543540 1.1 7566940 1.1 7569760
1.2 6761790 1.2 8095990 1.2 8242030 1.2 8256720 1.2 8258190
Allowable bending moment
Number of Layers = 50
0.1 5285.71 0.1 10475.3 0.1 15479.5 0.1 20221.7 0.1 24642.2
0.2 20833.6 0.2 40217.6 0.2 57079.5 0.2 70910.5 0.2 81718.6
0.3 46131.1 0.3 85530.3 0.3 114895 0.3 134634 0.3 146998
0.4 80322.3 0.4 141622 0.4 179446 0.4 199809 0.4 209973
0.5 122331 0.5 203975 0.5 244866 0.5 262422 0.5 269455
0.6 170954 0.6 269099 0.6 308615 0.6 322196 0.6 326604
0.7 224956 0.7 334696 0.7 370098 0.7 379900 0.7 382495
0.8 283144 0.8 399478 0.8 429556 0.8 436305 0.8 437769
0.9 344429 0.9 462869 0.9 487461 0.9 491955 0.9 492757
1.0 407859 1.0 524725 1.0 544268 1.0 547188 1.0 547616
1.1 472635 1.1 585135 1.1 600333 1.1 602195 1.1 602420
1.2 538114 1.2 644292 1.2 655914 1.2 657083 1.2 657200
S-glass/epoxy material
llowable axial tension
Number of Layers = 10
0.1 61972.6 0.1 105114 0.1 128147 0.1 138728 0.1 143251
0.2 210228 0.2 277456 0.2 290251 0.2 292407 0.2 292762
0.3 384441 0.3 435376 0.3 438990 0.3 439231 0.3 439247
0.4 554911 0.4 584814 0.4 585642 0.4 585664 0.4 585664
0.5 716256 0.5 731906 0.5 732079 0.5 732081 0.5 732081
0.6 870752 0.6 878462 0.6 878497 0.6 878497 0.6 878497
0.7 1021240 0.7 1.024910 0.7 1024910 0.7 1.02491 0.7 1024910
0.8 1169630 0.8 1171330 0.8 1171330 0.8 1171330 0.8 1171330
0.9 1316970 0.9 1317740 0.9 1317750 0.9 1317750 0.9 1317750
1.0 1463810 1.0 1464160 1.0 1464160 1.0 1464160 1.0 1464160
1.1 1610420 1.1 1610580 1.1 1610580 1.1 1610580 1.1 1610580
1.2 1756920 1.2 1756990 1.2 1756990 1.2 1756990 1.2 1756990
llowable bending moment
Number of Layers = 10
0.1 4933.11 0.1 8367.21 0.1 10200.7 0.1 11042.9 0.1 11403
0.2 16730.7 0.2 22080.9 0.2 23099.1 0.2 23270.7 0.2 23299
0.3 30593.9 0.3 34647.3 0.3 34934.9 0.3 34954.1 0.3 34955.3
0.4 44159.3 0.4 46538.9 0.4 46604.7 0.4 46606.5 0.4 46606.6
0.5 56998.5 0.5 58243.9 0.5 58257.7 0.5 58257.8 0.5 58257.8
0.6 69292.8 0.6 69906.4 0.6 69909.1 0.6 69909.1 0.6 69909.1
0.7 81268.3 0.7 81559.9 0.7 81560.5 0.7 81560.5 0.7 81560.5
0.8 93076.4 0.8 93211.7 0.8 93211.8 0.8 93211.8 0.8 93211.8
0.9 104801 0.9 104863 0.9 104863 0.9 104863 0.9 104863
1.0 116487 1.0 116515 1.0 116515 1.0 116515 1.0 116515
1.1 128154 1.1 128166 1.1 128166 1.1 128166 1.1 128166
1.2 139812 1.2 139817 1.2 139817 1.2 139817 1.2 139817
llowable axial tension
Number of Layers = 20
0.1 65028.8 0.1 123945 0.1 172738 0.1 210228 0.1 237408
0.2 247891 0.2 420456 0.2 512588 0.2 554911 0.2 573005
0.3 518214 0.3 768882 0.3 848849 0.3 870752 0.3 876492
0.4 840911 0.4 1109820 0.4 1161000 0.4 1169630 0.4 1171050
0.5 1187040 0.5 1432510 0.5 1460820 0.5 1463810 0.5 1464120
0.6 1537760 0.6 1741500 0.6 1755960 0.6 1756920 0.6 1756990
0.7 1883230 0.7 2042480 0.7 2049510 0.7 2049810 0.7 2049830
0.8 2219640 0.8 2339250 0.8 2342570 0.8 2342660 0.8 2342660
0.9 2546550 0.9 2633940 0.9 2635460 0.9 2635490 0.9 2635490
1.0 2865020 1.0 2927620 1.0 2928310 1.0 2928320 1.0 2928320
1.1 3176660 1.1 3220840 1.1 3221150 1.1 3221150 1.1 3221150
1.2 3483010 1.2 3513850 1.2 3513990 1.2 3513990 1.2 3513990
llowable bending moment
Number of Layers = 20
0.1 5181.04 0.1 9875.09 0.1 13762.5 0.1 16749.5 0.1 18915
0.2 19732.4 0.2 33468.8 0.2 40802.7 0.2 44171.7 0.2 45612
0.3 41243.6 0.3 61193.8 0.3 67558.3 0.3 69301 0.3 69758.4
0.4 66922.6 0.4 88323.5 0.4 92396.6 0.4 93082.9 0.4 93196.1
0.5 94466.1 0.5 114001 0.5 116254 0.5 116492 0.5 116517
0.6 122375 0.6 138589 0.6 139739 0.6 139816 0.6 139821
0.7 149867 0.7 162540 0.7 163099 0.7 163123 0.7 163124
0.8 176637 0.8 186155 0.8 186419 0.8 186426 0.8 186426
0.9 202651 0.9 209605 0.9 209727 0.9 209729 0.9 209729
1.0 227994 1.0 232976 1.0 233031 1.0 233031 1.0 233031
1.1 252793 1.1 256309 1.1 256334 1.1 256334 1.1 256334
1.2 277171 1.2 279626 1.2 279636 1.2 279637 1.2 279637
llowable axial tension
Number of Layers = 30
0.1 65635.4 0.1 128404 0.1 185918 0.1 236590 0.1 279713
0.2 256807 0.2 473179 0.2 630683 0.2 733546 0.2 796045
0.3 557754 0.3 946025 0.3 1153320 0.3 1248550 0.3 1289260
0.4 946359 0.4 1467090 0.4 1664730 0.4 1728810 0.4 1748490
0.5 1398560 0.5 1990110 0.5 2148770 0.5 1754440 0.5 2193880
0.6 1892050 0.6 2497100 0.6 2612260 0.6 2195720 0.6 2634860
0.7 2408470 0.7 2985260 0.7 3063730 0.7 3074720 0.7 3074580
0.8 2934180 0.8 3457620 0.8 3508880 0.8 3513980 0.8 3513950
0.9 3459970 0.9 3918380 0.9 3950910 0.9 3.953230 0.9 3953220
1.0 3980220 1.0 4371240 1.0 4391440 1.0 4392480 1.0 4392480
1.1 4492010 1.1 4818920 1.1 4831260 1.1 4831730 1.1 4831730
1.2 4994200 1.2 5263320 1.2 5270770 1.2 5709800 1.2 5270980
llowable bending moment
Number of Layers = 30
0.1 5237.2 0.1 10245.6 0.1 14834.8 0.1 18878 0.1 22318.9
0.2 20449.9 0.2 37679.8 0.2 50222.1 0.2 58413.1 0.2 63390
0.3 44398 0.3 75304.9 0.3 91806.1 0.3 99386.3 0.3 102627
0.4 75321.6 0.4 116767 0.4 132498 0.4 137598 0.4 139164
0.5 111306 0.5 158385 0.5 171012 0.5 139638 0.5 174602
0.6 150576 0.6 198728 0.6 207892 0.6 174749 0.6 209691
0.7 191671 0.7 237573 0.7 243817 0.7 244692 0.7 244680
0.8 233505 0.8 275160 0.8 279240 0.8 279646 0.8 279643
0.9 275345 0.9 311826 0.9 314414 0.9 314599 0.9 314598
1.0 316745 1.0 347861 1.0 349469 1.0 349552 1.0 349552
1.1 357471 1.1 383486 1.1 384468 1.1 384506 1.1 384506
1.2 397433 1.2 418850 1.2 419443 1.2 419459 1.2 419459
llowable axial tension
Number of Layers = 40
0.1 65850.9 0.1 130058 0.1 191136 0.1 247891 0.1 299489
0.2 260115 0.2 495781 0.2 690951 0.2 840911 0.2 949631
0.3 573408 0.3 1036430 0.3 1349950 0.3 1537760 0.3 1642120
0.4 991562 0.4 1681820 0.4 2050350 0.4 2219640 0.4 2292020
0.5 1497440 0.5 2374080 0.5 2736870 0.5 2865020 0.5 2907710
0.6 2072850 0.6 3075530 0.6 3395400 0.6 3483010 0.6 3505970
0.7 2700190 0.7 3766470 0.7 4028790 0.7 4084970 0.7 4096630
0.8 3363640 0.8 4439290 0.8 4644010 0.8 4678510 0.8 4684200
0.9 4049850 0.9 5093090 0.9 5247330 0.9 5267880 0.9 5270570
1.0 4748150 1.0 5730050 1.0 5843280 1.0 5855250 1.0 5856500
1.1 5450480 1.1 6353310 1.1 6434840 1.1 6441690 1.1 6442260
1.2 6151060 1.2 6966020 1.2 7023830 1.2 7027700 1.2 7027960
llowable bending moment
Number of Layers = 40
0.1 5265.4 0.1 10399.3 0.1 15283.1 0.1 19821.2 0.1 23946.9
0.2 20724.1 0.2 39500.4 0.2 55050.1 0.2 66997.9 0.2 75659.9
0.3 45654.7 0.3 82520.2 0.3 107483 0.3 122437 0.3 130746
0.4 78929.7 0.4 133875 0.4 163211 0.4 176687 0.4 182448
0.5 119186 0.5 188959 0.5 217835 0.5 228035 0.5 231432
0.6 164974 0.6 244775 0.6 270233 0.6 277206 0.6 279033
0.7 214896 0.7 299755 0.7 320633 0.7 325103 0.7 326031
0.8 267690 0.8 353294 0.8 369586 0.8 372332 0.8 372785
0.9 322296 0.9 405320 0.9 417594 0.9 419229 0.9 419444
1.0 377864 1.0 456005 1.0 465016 1.0 465968 1.0 466068
1.1 433753 1.1 505601 1.1 512089 1.1 512633 1.1 512679
1.2 489502 1.2 554356 1.2 558958 1.2 559265 1.2 559286
llowable axial tension
Number of Layers = 50
0.1 65951.2 0.1 130841 0.1 193674 0.1 253581 0.1 309863
0.2 261681 0.2 507162 0.2 724021 0.2 905662 0.2 1051140
0.3 581021 0.3 1086030 0.3 1474200 0.3 1745290 0.3 1922210
0.4 1014320 0.4 1811320 0.4 2327060 0.4 2620150 0.4 2774560
0.5 1549320 0.5 2627850 0.5 3203680 0.5 3468200 0.5 3581280
0.6 2172060 0.6 3490590 0.6 4065190 0.6 4278980 0.6 4353760
0.7 2867970 0.7 4367880 0.7 4898970 0.7 5059850 0.7 5106210
0.8 3622650 0.8 5240310 0.8 5705310 0.8 5820620 0.8 5848140
0.9 4422590 0.9 6097780 0.9 6489170 0.9 6569020 0.9 6584840
1.0 5255690 1.0 6936390 1.0 7256200 1.0 7310170 1.0 7319060
1.1 6111470 1.1 7755900 1.1 8011520 1.1 8047200 1.1 8052110
1.2 6981180 1.2 8557960 1.2 8758690 1.2 8781960 1.2 8784620
Allowable bending moment
Number of Layers = 50
0.1 5287.59 0.1 10490.1 0.1 15527.6 0.1 20330.7 0.1 24843.1
0.2 20863 0.2 40434.4 0.2 57723.8 0.2 72205.4 0.2 83803.8
0.3 46274.7 0.3 86495.7 0.3 117411 0.3 139002 0.3 153092
0.4 80755.2 0.4 144208 0.4 185268 0.4 208603 0.4 220896
0.5 123328 0.5 209180 0.5 255017 0.5 276073 0.5 285075
0.6 172883 0.6 277830 0.6 323565 0.6 340581 0.6 346533
0.7 228261 0.7 347638 0.7 389907 0.7 402712 0.7 406401
0.8 288315 0.8 417059 0.8 454067 0.8 463245 0.8 465434
0.9 351972 0.9 485291 0.9 516439 0.9 522794 0.9 524054
1.0 418266 1.0 552022 1.0 577479 1.0 581768 1.0 582476
1.1 486365 1.1 617233 1.1 637576 1.1 640416 1.1 640806
1.2 555573 1.2 681056 1.2 697031 1.2 698882 1.2 699094
366 Case Studies
θ =π /12 L = 5
θ =π /12 L = 10
Hanger Width 367
θ =π /12 L = 15
θ =π /12 L = 5
368 Case Studies
θ =π /12 L = 10
θ =π /12 L = 15
Spaces Between Supports 369
NP = 50
NP = 40
L (m)
NP = 30
NP = 20
NP = 10
NP = 50
NP = 40
NP = 30
NP = 20
NP = 10
NP = 50
NP = 40
NP = 30
NP = 20
NP = 10
NP = 50
NP = 40
NP = 30
NP = 20
NP = 10
NP = 50
NP = 40
NP = 30
NP = 50
NP = 40
NP = 30
NP = 20
NP = 10
NP = 50
NP = 40
NP = 30
NP = 20
NP = 10
NP = 50
NP = 40
NP = 30
NP = 20
NP = 10
NP = 50
NP = 40
NP = 30
NP = 20
NP = 10
NP = 50
NP = 40
NP = 30
NP = 20
NP = 10
θ =π /6 F = 75 000 (N)
θ =π /4 F = 75 000 (N)
θ =π /3 F = 75 000 (N)
θ = π /6 F = 75 000 (N)
θ = π /4 F = 75 000 (N)
θ = π /3 F = 75 000 (N)
ABD matrix 35 36 37
acceleration 114 115
Acoustic emission 295
additional mass effect 135
adhesive joint 170
A-Glass 48
allowable shear stress 177
annular flow 255 256 258
267
axial load 84
E-Glass 48
Eigenfrequencies 117
elastic constant 141 147
elastic dilatation 150 152
elastic foundation 141 147
elevated temperature 56 57
energy losses 245 248
equation of motion 116 117
equibrium equations 67 143
expansion loops 179 186
extensional modulus of elasticity 173
external pressure 55 57 77
81 83 94
failure criteria 39
failure 56
Fanning equation 244
Fatigue testing 284 286
Fatigue 220
Fiber materials 48
fiber orientation 58 173 271
Filament winding 50 51
filament-wound 170 271
J-Lay 53 56
joining methods 169
Kevlar 45 50
kinetic energy 245
Kirchhoff assumption 27 37
Lamina 13
laminar flow 250 251 252
262
laminate nomenclature 24
laminate strains 28
laminate stresses 30
Laplace transform 160
law of real gases 246
Lekhnitskii formalism 59
lifetime prediction 197
Liquid epoxy resin 46
Liquid flow rate 253
liquid transmission 252
long-term hydrostatic pressure 220
Low-velocity impact 282
Mandrel 51
Mass 119
material characterization 279
operation loads 53 56
Optical fiber-based techniques 295
Optimization 271
optimum fiber orientation 272 273 276
277
Orthophlatic polyester 47
Overpressure 154
Polymer matrices 45
Poly-Phenylene Sulfide (PPS) 47 48
Poly-Propylene (PP) 48
Poly-Sulfone (PS) 48
Polyurethane resin 46
potential energy 245 248
Prandtl-Von Karman equation 251
pressure drop 255
pressure energy 245
pressure shock 149 153 155
158 161 162
principal coordinate system 1 13
principal directions 1 7
proportionality factor 243
pseudo-critical pressure 246 247
pseudo-critical temperature 246 247
pseudo-static loads 55
S-Glass 48
shear failure stress 40
shear modulus 172
shock wave 153
single-layered pipe 59
S-Lay 53 56
Slope 143
slug flow 255 256 258
265
socket adhesive joint 169 170
soil-pipe interaction 57
Solid epoxy resin 46
stiffness matrix 6 36 157
stratified smooth flow 255 256 259
stratified wavy flow 255 256 259
stress-strain relations 7 8
Structural health monitoring 295
Symmetric balanced laminates 36 38
Symmetric Cross-Ply laminates 36 38
Symmetric laminates 36 38
temperature gradients 56
velocity 115
vibration 56 105 155
Vinyl ester resin 46
Vinyl ester 47