7 views

Uploaded by samaneh

- CMS-16-QB
- Composites
- Benjamin_J_Phillips_Thesis_2009.pdf
- Composites
- Design of Anis Grid Composite Lattice Conical Shell Structures
- Carbon-Fiber-reinforced Polymer - Wikipedia, The Free Encyclopedia
- Experimental Determination of Dynamic Young's Modulus and Damping of an Aramid-fabricpolyester Composite Material
- Speech
- Construction Material 17209 Winter 14
- Cfrp Wrap Cfw-600
- noor1996
- congresso_sampe_sigmatex
- zaikako3
- Composites-1.pdf
- Serp 13 - Composite Prepregs 09-12-13
- AD061 Axial load capacity of a dented tubular member.pdf
- ebook_11600 (1)
- Sn003a - Elastic Critical Moment for LTB
- A13 Materials Selection in Design (1)
- BOF39 - Shave - FRP Composites in Bridge Design

You are on page 1of 417

Piping Applications

Design, Analysis and Optimization of

Subsea and Onshore Pipelines from FRP Materials

Technological Institute of Halkida (TEI-Halkida)

Mechanical Engineering Department

Greece

Composite Materials in Piping Applications

4 3 9 N o r t h Du k e S t r e e t

Lancaster, Pen nsyl van ia 1 7602 U.S.A.

A l l r i g h t s r es e r v e d

retrieval system, o r transmitt ed, in any fo rm or by an y means,

electro nic, mechan ical, photo cop ying, recording, or otherwi se,

without the p rior wri tten permi ssion of the publ ish er. The p ublisher

i s n o t l i a b l e f o r a n y e r r o r s r e s u l t i n g f r o m t h e u s e o f t h e C D- R O M .

10 9 8 7 6 5 4 3 2 1

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

ISB N N o . 9 7 8 -1 -6 0 5 9 5 -0 2 9 -7

BY PHONE: 877-500-4337 or 717-290-1660, 9AM–5PM Eastern Time

BY FAX: 717-509-6100

Order Department

BY MAIL:

439 North Duke Street

Lancaster, PA 17602, U.S.A.

BY CREDIT CARD: American Express, VISA, MasterCard, Discover

BY W W W SIT E: http://www.destechpub.com

Acknowledgments

developing this book. Dr. Rui Miranda Guedes of the Department of Mechanical

Engineering at the University of Porto and Dr. Hugo Faria of INEGI-Instituto de

Engenharia Mechânica e Gestão Industrial at Porto contributed the material in

Chapter 6 on creep design of piping applications using composites. I am indebted

to them for providing their expertise in this area. My former student (currently

Dipl. Mechanical Engineer) Stavros Lykakos was the tester of the specially de-

veloped software. His help was very valuable. Moreover, my students Michalis

Bazanos, Andronikos Miniatis, Giorgos Papastefanos, Panagiotis Bouyioukos,

Giorgos Macheras, Giorgos Roussos, Giorgos Vagalis, Tassos Tsitsakis, Apostolis

Kalaris provided considerable assistance with the figures, graphics and text typ-

ing. Particular thanks go to Asimina Kehagia and Giorgos Sioris for their valuable

linguistic comments.

Especially, the author wishes to express his enormous appreciation to Dr.

Joseph Eckenrode, publishing director of DEStech Publications for his kind in-

vitation to develop this book and for his encouragement during the efforts of the

last two years. Finally, of immense help in refining the book were the reviewers

who provided their helpful comments. For their time and comments I am deeply

grateful.

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

P = (Mass/ρL)Sy.

The material unit price is UP = Cost/Mass. Therefore, the allowable force can

now be written P = (Cost/UP) (1/ρL)Sy. Using this formula, a cost index demon-

strating the normalized cost, i.e., the cost for a unit length of the material sheet

per unit allowable force can be used for a material selection. With the aid of the

last formula providing P, the above cost index has the form CI = Cost/PL or CI

= UP ρ/Sy. From this formula it can be concluded that apart from the unit price

UP, the total material cost is strongly influenced by the material density ρ and the

yield stress Sy. The following figures illustrate the unit price ($/ton), density (Kg/

m3) and yield stress (MPa) for carbon steel, stainless steel, glass fiber-reinforced

polymers (GFRP) and carbon fiber reinforced-polymers (CFRP). Comparing for

example carbon steels and GFRPs, which are the main materials used for pipe

manufacturing, it can be shown that even though the GFRPs have a unit price that

is 11 times more expensive than that of steel, their density is about 4 times lower

and their yield stress is about 3 times higher. Taking into account the definition

of the cost index CI, it can be concluded that the lower density and the higher

strength of GFRPs reduce the total material cost significantly. Indeed, as shown in

the last figure that provides the cost index (CI) of four materials, the material cost

of GFRP is comparable to that of carbon steel.

Introduction xv

DENSITY Kg/m3

xvi Introduction

Cost index CI

corrosion resistance, it is apparent that the use of FRP pipes is advantageous com-

pared that of steel pipes. However, the mechanical design of FRP pipes is much

more complicated due to the fact that FRP materials are anisotropic. Apart from

the allowable service loads, the anisotropy also influences their dynamic behavior

and stability, fatigue and creep life, spaces between supports, joining etc. The sub-

sequent chapters are intended to address the complexities presented by anisotropy

and thus to provide composite materials designers with the analytical methods

and computer programs necessary for detailed, quantitative mechanical design of

filament-wound FRP pipelines.

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:

ite materials covering the anisotropic elasticity equations for laminae and

laminates, as well as the widely used Tsai-Wu failure criterion.

Chapter 2 explains the classification and properties of composite materials

and gives a brief description of the filament-winding method.

Chapter 3 develops theoretical tools for the mechanical design of pipelines

including pure bending, external pressure, axial tension, torsion and their

combinations in multilayered filament-wound pipelines by using failure and

buckling models.

Chapter 4 provides original models for the dynamic analysis of composite

pipes carrying fluids. Special emphasis is given to the estimation of critical

flow velocities that cause instability as well as the fluid hammer-induced

wave propagation.

In Chapter 5, methods are presented for designing joints as well as estimat-

ing the supporting spacing, hanger width and sizing of expansion loops.

Techniques for estimating the safe depth of underground pipelines situated

under in-service streets and railroads are also included.

Chapter 6 presents models for the life-time prediction of composite pipe-

lines under creep and fatigue conditions.

Chapter 7 explains flow models for estimating the flow capacity of compos-

ite pipelines that deliver liquid, gas or multi-phase fluids.

In Chapter 8, the parameters for optimum design of composite pipelines

with a view to reducing material costs are discussed.

Chapter 9 clarifies current quality control methods for the manufacture of

composite materials and composite pipelines.

Chapter 10 presents a collection of nomographs for direct mechanical

design of GFRP composite pipes under a wide range of pure and combined

loading conditions.

Dimitrios G. Pavlou

Technical Institute of Chalkida

January 2013

Contents

Preface ix

Acknowledgments xi

Introduction xiii

Composite Materials.........................................................1

1.1 Mechanical Behavior of Laminae 1

1.1.1 Generalized Hooke’s law 1

1.1.2 Effects of free thermal strains 6

1.1.3 Effects of free moisture strains 8

1.1.4 Plane stress constitutive relations 10

1.1.5 Coordinate transformation of stress and strain

components 13

1.1.6 Transformation of engineering properties 16

1.1.7 Free thermal and free moisture strains in the global

coordinate system 19

1.2 Mechanical Behavior of Laminates 24

1.2.1 Classical lamination theory 24

1.2.2 Laminate nomenclature 24

1.2.3 The Kirchhoff Assumption 27

1.2.4 Laminate strains 28

1.2.5 Laminate stresses 30

1.2.6 Laminate stiffness matrix 30

1.2.7 Classification of laminates 36

1.3 The Tsai–Wu Failure Criterion 39

References 43

iii

iv Contents

Technology of FRP Materials..........................................45

2.1 The Composite Matrix Material 45

2.1.1 Thermosets 46

2.1.2 Thermoplastics 47

2.2 Fiber Materials 48

2.2.1 Glasses 48

2.2.2 Carbon fibers 49

2.2.3 Synthetic fibers 50

2.3 Production Technologies for FRP Composite Pipes 50

2.3.1 Filament winding 50

2.3.2 Fiber placement process 52

References 52

3.1 Types of Loading Cases 53

3.1.1 Installation loads 53

3.1.2 Operation loads 56

3.2 Pure Bending 56

3.2.1 Failure analysis 56

3.2.2 Buckling model 73

3.3 External Pressure 77

3.3.1 Failure analysis 77

3.3.2 Buckling model 80

3.4 Combination of Bending and External Pressure 81

3.4.1 Failure analysis 81

3.4.2 Buckling model 82

3.5 Axial Tension 88

3.5.1 Failure analysis 88

3.6 Combination of Bending and Axial Tension 91

3.6.1 Failure analysis 91

3.6.2 Buckling model 92

3.7 Combination of External Pressure and Axial Tension 94

3.7.1 Failure analysis 94

3.8 Torsion 97

3.8.1 Failure analysis 97

Contents v

References 102

4.1 Free Vibration of Composite Pipes 105

4.1.1 Structural characteristics of composite pipes 105

4.1.2 Forces and bending moments acting on a composite

pipe element 111

4.2 Accelerations of the Fluid and Pipe Elements 114

4.3 Equation of Motion 116

4.3.1 Solution of equation of motion 117

4.3.2 Types of instability 122

4.4 Transfer Matrices Method (TMM) 124

4.5 Estimation of Critical Velocity for Composite Pipes

Conveying Fluid 129

4.5.1 Cantilever pipe 130

4.5.2 Fixed-fixed pipe 130

4.5.3 Pinned-pinned pipe 131

4.5.4 Fixed-pinned pipe 131

4.6 Effect of Temperature (Thermal Load) 132

4.7 Effect of Additional Mass 135

4.7.1 Transfer matrix of the segment 1-2 (pipe) 139

4.7.2 Transfer matrix of the segment 2-3 (collar-pipe) 139

4.7.3 Global transfer matrix 140

4.8 Effects of an Elastic Foundation 141

4.9 Effect of Additional Supports 142

4.9.1 Example 146

4.10 Estimation of Critical Flow Velocity in Relation to

Divergence 147

4.10.1 Elastic foundation effect 147

4.10.2 Thermal load and elastic foundation effects 147

4.10.3 Fixed-fixed pipe 148

4.10.4 Pinned-pinned pipe 149

4.10.5 Fixed-pinned pipe 149

4.11 Hydraulic Hammer 149

4.11.1 Shock pressure in a branched pipe 155

4.12 Wave Propagation Due to Hydraulic Hammer 155

vi Contents

References 168

5.1 Joining of Composite Pipelines 169

5.1.1 Approximate mechanical model for axial

loading 170

5.1.2 Approximate mechanical model for bending 178

5.2 Above-Ground Pipes 179

5.2.1 Maximum spacing between supports 180

5.2.2 Minimum hanger widths 182

5.2.3 Sizing of expansion loops 186

5.3 Underground Pipelines 192

References 196

Composite Materials.....................................................197

6.1 Introduction 197

6.2 Creep Damage Accumulation Mechanisms in Composite

Materials 198

6.3 Short and Long-Term Static Failure of Composite

Pipes 206

6.3.1 Damage modeling 206

6.3.2 Creep rupture 212

6.3.3 An example of preliminary design for the

long-term 218

6.4 Lifetime of Composites Pipes Under Cyclic Loading 220

6.5 Applicable Standards 225

6.5.1 Identification and comparison of main standards 225

6.5.2 Long-term qualification tests of four different types of

GRP 230

6.6 Practical Design: A Case Study 232

6.7 Conclusions 237

6.8 Acknowledgements 238

References 238

7.1 Gas Transmission 243

7.1.1 Estimation of gas flow rate 249

Contents vii

7.2.1 Flow capacity for laminar liquid flow 252

7.2.2 Flow capacity for turbulent flow 253

7.3 Multiphase Flow 255

7.3.1 Multiphase flow regimes for inclined pipelines 255

References 269

8.1 Fiber Orientation and Loading Forces 271

8.1.1 Optimum fiber orientation for the combination of

axial tension and external pressure 272

8.1.2 Optimum fiber orientation for the combination of

bending and axial tension 276

8.1.3 Optimum fiber orientation for the combination of

bending and external pressure 277

9.1 Test Methods and Material Characterization 279

9.1.1 Thermal analysis DSC (Differential Scanning

Calorimetry) 279

9.1.2 Measurement of residual stresses 280

9.1.3 Creep strain and creep rupture tests 280

9.1.4 Impact testing 282

9.1.5 Fatigue testing 284

9.2 International Standards for Composite Pipes 287

9.3 Detection of Defects and Structural Health

Monitoring 295

9.3.1 Piezoelectric techniques 295

9.3.2 Optical fiber-based techniques 295

9.3.3 Ultrasonic testing 296

References 297

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

10.3.1 Results of failure model for external pressure 326

10.3.2 Results of buckling model for external

pressure 334

10.4 Torsion 340

10.4.1 Results of failure model for torsion 340

10.5 Butt Joints of Multilayered Filament-Wound Pipes 345

10.5.1 E-glass epoxy material 346

10.5.2 S-glass/epoxy material 356

10.6 Hanger Width 366

10.6.1 E-glass/epoxy material 366

10.6.2 S-glass-epoxy material 367

10.7 Spaces Between Supports 369

10.7.1 E-glass/epoxy material 369

10.7.2 Material: S-glass/epoxy material 371

10.8 Installation Depth for Underground Pipelines vs. the

Vertical Load F 374

10.8.1 E-glass-epoxy materials 374

10.8.2 S-glass/epoxy material 384

Index 395

About the CD-ROM 399

Chapter 1

Composite Materials

1.1.1 Generalized Hooke’s law

Figure 1.1 shows a part of a fiber reinforced composite material. Since the ma-

terial properties of the fibers have different values from those of the matrix, it is

convenient to use an orthogonal coordinate system that has one axis aligned with

the fiber orientation. The axis x1 is aligned with the fiber direction, and perpen-

dicular to the fibers are the axes x2 (lying in the plane of the layer) and x3 (per-

pendicular to the plane of the layer). The orientations of the axes x2 and x3 are

called matrix directions. The coordinate system x1 , x2 , x3 is called the principal

coordinate system and the corresponding directions are called principal direc-

tions. From a macroscopic point of view, the mechanical behavior of each indi-

vidual fiber or matrix element is of no practical importance. For design purposes,

the two-material fiber-matrix system is treated as a homogeneous anisotropic ma-

terial. Obviously, this material does not have the same properties in all directions,

i.e., its stiffness and strength in the fiber’s direction are higher than in the matrix

directions. The study of the elastic response of the above material is equivalent

to determining the relations between the stresses applied to the bounding sur-

faces and the corresponding deformations of the material as a whole. Denoting by

E1 , E2 , E3 the modulus of elasticity in the directions x1 , x2 , x3 respectively, and by

νi j (i = 1, 2, 3 , j = 1, 2, 3) the Poisson’s ratios given by

εj

νi j = − ι = (1.1)

1, 2, 3 j = 1, 2, 3

ει

1

2 Mechanical Behavior of Fiber Reinforced Composite Materials

(Fig. 1.2) is given by the following generalized Hooke’s law:

σ C C22 C23 0 0 0 ε

2 21 2

σ 3 C31 C32 C33 0 0 0 ε3

= (1.2)

τ 23 0 0 0 C44 0 0 γ 23

τ13 0 0 0 0 C55 0 γ13

τ12 0 0 0 0 0 C66 γ12

The above equation can be written in terms of strains:

Mechanical Behavior of Laminae 3

ε S S 22 S 23 0 0 0 σ 2

2 21

ε3 S31 S32 S33 0 0 0 σ3

= (1.3)

γ 23 0 0 0 S 44 0 0 τ 23

γ13 0 0 0 0 S55 0 τ13

γ12 0 0 0 0 0 S66 τ12

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:

{ε1} = [ S ]{σ1} (1.17)

[C ] = [ S ]−1 (1.18)

among material properties can be obtained [e.g., 1, 2]:

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

C22 = (1.29)

S

C23 = (1.30)

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.

Temperature changes in a fiber reinforced composite material element can

cause significant stresses in the fiber and matrix. However, when the stresses are

integrated over the composite medium, the net result is zero. Taking into account

Mechanical Behavior of Laminae 7

anisotropic material, the simplification that free thermal strains of a material ele-

ment with no constraints on its bounding surfaces do not cause stresses has to be

reinterpreted. To this end, in order to maintain consistency with the definition of

stress and strain, the term “mechanical strains” will be used in the stress–strain

relations. According to this concept, mechanical strains are the total strains (i.e.,

the changes in length per unit length of a material element) minus the free thermal

strains. Therefore, the stress–strain relations can be written in the form:

ε − α ∆Τ S S 22 S 23 0 0 0 σ

2 2 12 2

ε3 − α3 ∆Τ S13 S 23 S33 0 0 0 σ3

= (1.38)

γ 23 0 0 0 S 44 0 0 τ 23

γ13 0 0 0 0 S55 0 τ13

γ12 0 0 0 0 0 S66 τ12

or

σ C C22 C23 0 0 0 ε − α ∆Τ

2 12 2 2

σ 3 C13 C23 C33 0 0 0 ε3 − α3 ∆Τ

= (1.39)

τ 23 0 0 0 C44 0 0 γ 23

τ13 0 0 0 0 C55 0 γ13

τ12 0 0 0 0 0 C66 γ12

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

and ΔΤ is the temperature change.

The above definitions take into account the cases of: (a) stress with no thermal

effects; (b) thermal effects but no stresses; and (c) thermal effects with stresses.

8 Mechanical Behavior of Fiber Reinforced Composite Materials

For each of these three cases, the corresponding stress–strain relationships can be

written in the following form:

ε S

2 12 S 22 S 23 0 0 0 σ

2

ε3 S S 23 S33 0 0 0 σ3

= 13 (1.41)

0 0 0 0 0

γ

23 S 44 τ 23

γ13 0 0 0 0 S55 0 τ13

γ

12 0 0 0 0 0 S66 τ12

Case (b) Thermal effects of a material element with no constraints on its

bounding surfaces:

ε1 − α1 ∆Τ 0

ε − α ∆Τ 0

2 2

ε3 − α3 ∆Τ 0

= (1.42)

γ 23 0

γ13 0

γ12 0

deformation) material element:

σ C − α ∆Τ

C22 C23 0 0 0 2

2 12

σ 3 C13 C23 C33 0 0 0 − α3 ∆Τ

= (1.43)

τ 23 0 0 0 C44 0 0 0

τ13 0 0 0 0 C55 0 0

τ12 0 0 0 0 0 C66 0

When a piece of composite material is exposed to a liquid, a certain amount

of that liquid is absorbed, yielding an increase in the composite’s weight and

expansion. Although the weight gain is negligible (usually less than 4%), the ex-

pansion can be important. The created free moisture strains can be considered to

Mechanical Behavior of Laminae 9

coefficient of thermal expansion, the term coefficient of moisture expansion can

be used in order to describe the free moisture expansion of a composite element.

Since the moisture expansion in the fiber direction is small (compared to the mois-

ture expansion of the polymer), the assumption of treating the fiber–matrix system

as a homogeneous anisotropic material and the concept of mechanical strains will

be again adopted. Using the notations β1 , β 2 , β3 for the moisture-expansion coef-

ficients in the principal directions and ΔΜ for the change of absorbed moisture,

the generalized Hooke’s law can be written in the form:

ε − β ∆Μ S S 22 S 23 0 0 0 σ

2 2 12 2

ε3 − β3 ∆Μ S13 S 23 S33 0 0 0 σ3

= (1.44)

γ 23 0 0 0 S 44 0 0 τ 23

γ13 0 0 0 0 S55 0 τ13

γ12 0 0 0 0 0 S66 τ12

or

σ C C22 C23 0 0 0 ε − β ∆Μ

2 12 2 2

σ 3 C13 C23 C33 0 0 0 ε3 − β3 ∆Μ

= (1.45)

τ 23 0 0 0 C44 0 0 γ 23

τ13 0 0 0 0 C55 0 γ13

τ12 0 0 0 0 0 C66 γ12

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

ε − α ∆Τ − β ∆Μ S S 22 S 23 0 0 0 σ

2 2 2 12 2

ε3 − α3 ∆Τ − β3 ∆Μ S13 S 23 S33 0 0 0 σ3

= (1.47)

γ 23 0 0 0 S 44 0 0 τ 23

γ13 0 0 0 0 S55 0 τ13

γ12 0 0 0 0 0 S66 τ12

and

σ C C22 C23 0 0 0 ε − α ∆Τ − β ∆Μ

2 12 2 2 2

σ 3 C13 C23 C33 0 0 0 ε3 − α3 ∆Τ − β3 ∆Μ

= (1.48)

τ 23 0 0 0 0 0

C44 γ 23

τ13 0 0 0 0 C55 0 γ13

τ12 0 0 0 0 0 C66 γ12

Structures made by composite materials are utilized in plates, beams, thin–

walled sheets, etc. The main characteristic of the above components is that the

value of at least one of their characteristic geometric dimensions is much smaller

than the other two dimensions. Therefore, three of the six components of stress are

much smaller than the other three. For plane structural components (e.g., plates

and thin-walled sheets), the in-plane stresses are much larger than the stresses

perpendicular to their plane. Usually, for design purposes, the stress components

perpendicular to the plane of these structures can be set to zero, which can simpli-

fy many problems. According to the above assumption, in a fiber-reinforced plate

with the principal directions shown in Figure 1.2, the stresses σ 3 , τ13 , τ 23 can be

set to zero, yielding the following simplification in the generalized Hooke’s law:

ε S σ

2 12 S 22 S 23 0 0 0 2

ε3 S S 23 S33 0 0 0 0

= 13 (1.49)

γ 23 0 0 0 S 44 0 0 0

γ13 0 0 0 0 S55 0 0

γ12 0 0 0 0 0 S66 τ12

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

eq. (1.2), the corresponding relation between stresses and strains can be written as:

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

σ 2 = Q12 Q22 0 ε2 − α2 ∆Τ − β 2 ∆Μ (1.67)

τ 0 0 Q66 γ12

12

Mechanical Behavior of Laminae 13

ε1mech ε1 − α1 ∆Τ − β1 ∆Μ

mech

ε2 = = ε2 − α2 ∆Τ − β 2 ∆Μ (1.68)

γ mech γ12

12

Composite laminates are made of multiple fiber-reinforced laminae, and each

lamina has its own specific fiber orientation. Therefore, in a multilayered, fiber-re-

inforced material, multiple x1 − x2 − x3 principal coordinate systems exist. On the

other hand, the loads acting on the multilayered laminate have a common direc-

tion for all layers composing the laminate. If we are dealing with an x − y − z or-

thogonal coordinate system (Fig. 1.3) to describe the directions of the loads acting

on the laminate, then the stresses and strains in each principal system x1 − x2 − x3

should be defined with respect to the stresses and strains to the x − y − z system

(global coordinate system).

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:

2 2

σ 2 = sin θ cos θ −2 sin θ cos θ σ y (1.69)

τ − sin θ cos θ sin θ cos θ cos 2 θ − sin 2 θ

12 τ xy

where ϑ is the angle defining the fiber orientation with respect to the x axis

(Figure 1.3).

Using the notations m = cos θ and n = sin θ , the above equation is usually

written in the following form:

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

−1

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

Concerning the strains involved in the plane-stress assumption, the following

transformation equation can be written as:

Mechanical Behavior of Laminae 15

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.

Taking into account equations (1.70), (1.74), equation (1.57) can now be writ-

ten as:

σ y = Q12 Q 22 Q 26 ε y (1.82)

τ xy Q16 Q 26 Q 66 γ xy

16 Mechanical Behavior of Fiber Reinforced Composite Materials

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 .

Apart from the engineering properties in the principal coordinate system, en-

gineering properties can also be defined within the x − y − z global coordinate

system.

Considering again a state of plane stress, when a stress σ x is acting on an FRP

plane element (Fig. 1.3), the element stretches in the x direction, contracts in the

y direction, and since the fibers are lying in a direction with orientation ϑ the

right corner angles do not remain as right angles. Denoting the modulus of elas-

ticity in the x direction by Ex , the Hooke’s law for this direction can be written

in the form:

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

properties in the global coordinate system is shown in Figure 1.4 (a)-(d). Portions

of this figure indicate that: (i) the modulus Ex becomes greatest when the fi-

bers are oriented in the x axis (i.e. ϑ = 0 ) and decreases rapidly with ϑ ; (ii) the

modulus E y has its smallest value at ϑ = 0 and increases rapidly as the fiber ori-

entation approaches the angle ± 90 ; (iii) the shear modulus Gxy takes its maxi-

mum value at ϑ = ± 45 and its minimum occurs at ϑ = 0 ; (iv) the maximum

value of vxy occurs in the area 0 < ϑ < 45 and the minimum at ϑ = 90 ; (v) the

Poisson’s ratio v yx has a minimum at ϑ = 0 while its maximum appears in the

area 45 < ϑ < 90 . The above conclusions correspond to the behavior of a typical

graphite-reinforced composite.

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:

22 Mechanical Behavior of Fiber Reinforced Composite Materials

orientation ϑ is shown in Figure 1.5.

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

the global coordinate system versus fiber orientation ϑ .

Mechanical Behavior of Laminae 23

yielding

principal coordinate system.

Using the concept of mechanical strains given by equation (1.66) and the in-

verted forms of equations (1.117), (1.121), i.e.,

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

σ y = Q12 Q 22 Q 26 ε y − α y ∆Τ − β y ∆M (1.129)

τ xy Q16 Q 26 Q 66 γ x y − α x y ∆Τ − β x y ∆M

1.2.1 Classical lamination theory

As was already stated, composite laminates are composed of multiple fiber-re-

inforced laminae (Fig. 1.6). Each lamina has a very small thickness and usually a

different fiber orientation. The stacking sequence of layers, their fiber orientation

and their number influence the mechanical behavior of the multilayered medium.

Since multiple combinations of stacking arrangements of layers are possible,

the nomenclature associated with the definition of the manner in which the lami-

nate is structured will be presented first.

A multilayered composite plate consisting of N layers (Figure 1.7) will be used

to describe the laminate nomenclature. The x axis of the global coordinate system

is located in the geometric midplane of the laminate, while the axes z and y are

aligned with the directions of the thickness and the width respectively (Fig. 1.7).

The specification of the fiber angles of the various layers starts with layer 1.

For example the four–layered laminate described in Table 1.1 is denoted as a

[0 / 45 / 90 / 0] laminate.

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.

Layer 1 +60° Repeated

Layer 2 -60° group

of layers

Layer 3 0

Layer 4 +60°

Layer 5 -60°

Layer 6 0

Layer 7 -60°

Layer 8 +60°

Layer 9 0

Layer 10 -60°

Layer 11 +60°

Layer 12 0

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]

The simplest laminated theory is an extension of the Kirchhoff isotropic plate

theory to laminated composite plates. According to this theory, the straight lines

normal to the xy plane (Fig. 1.7) before deformation remain straight and normal

to the mid-surface z = 0 after deformation. Therefore, on the basis of this as-

sumption, both transverse shear and transverse normal effects are neglected.

According to Figure 1.8 the movement of a point A in the x axis due to defor-

mation in the x-z plane is given by:

∂w0

u = u0 − z (1.130)

∂x

while its vertical movement is:

w = w0 (1.131)

28 Mechanical Behavior of Fiber Reinforced Composite Materials

point O located in geometric the mid-plane, w0 represents its vertical movement,

and ∂w0 / ∂x represents the slope in the x-z plane of the x-axis at the point O after

the deformation.

Following the same concept, the movement of the point A in the direction of

the y-axis (Fig. 1.7) can be given by:

∂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.

With the aid of eqs. (1.130)–(1.132), following the well-known [3] strain-

displacement relations:

Mechanical Behavior of Laminates 29

∂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.

If the plane stresses assumption is adopted for each lamina, the stresses of each

point of a lamina can be determined with the aid of equations (1.139), (1.140),

(1.142) by the already known stresses-strain relation in the global coordinate

system:

σx Q11 Q12 Q16 ε0 + zk 0

0x x

0

σ y = Q12 Q22 Q26 ε y + zk y (1.151)

τ γ 0 + zk 0

xy Q16 Q26 Q66 xy xy

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.

Input data for the design of composite structures are often the forces and mo-

ments acting on the boundary surfaces of laminates. The goal of the designer is

to calculate the stress field on each layer in order to apply a failure criterion. To

achieve this goal, formulae correlating the stress resultants, i.e., the loads and

moments required to produce the specified mid-plane deformations—with the

mid-plane strains and curvatures should be derived. Using the notations found

in Fig. 1.9 (a) (b), the force and moments resultants are given by the following

definitions:

Mechanical Behavior of Laminates 31

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

N

Aij = ∑ Q ijk ( zk − zk −1 ) (1.165)

k =1

+H 2

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:

0 0

N y = A12 A22 A26

ε y + B12 B22 B26 k y (1.168)

N A16 γ0 k 0

xy A26 A66 xy B16 B26 B66 xy

Combining equations (1.155) – (1.157) with equation (1.151), the following for-

mulas for the moment resultants can be obtained:

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)

0 0

M y = B12 B22 B26 ε y + D12 D22 D26 k y (1.173)

M γ0 k 0

xy B16 B26 B66 xy D16 D26 D66 xy

where

1 N

Dij = ∑ Qijk ( zk3 − zk3−1 )

3 k =1

(1.174)

N 0

y Α12 Α 22 Α 26 B12 B22 B26 εy

N xy Α16 Α 26 Α66 B16 B26 B66 γ 0xy

= 0 (1.175)

M x B11 B12 B16 D11 D12 D16 kx

M y B12 k y0

B22 B26 D12 D22 D26 0

M xy B16 B26 B66 D16 D26 D66 k xy

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

0 N

ε y a12 a22 a26 b21 b22 b26 y

γ 0xy a16 a26 α66 b61 b62 b66 N xy

0 = (1.176)

k x b11 b12 b61 d11 d12 d16 Mx

k y0 b b22 b62 d12 d 22 d 26 My

0 12

k xy b16 b26 b66 d16 d 26 d 66 M xy

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.

Depending on the specific stacking sequence of the layers composing a lami-

nate, the following types can be used in engineering practice:

(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

N 0

y Α12 Α 22 Α 26 0 0 0 εy

N xy Α16 Α 26 Α66 0 0 0 γ 0xy

= 0 (1.178)

Mx 0 0 0 D11 D12 D16 kx

My 0 k y0

0 0 D12 D22 D26 0

M xy 0 0 0 D16 D26 D66 k xy

N 0

y Α12 Α 22 0 0 0 0 εy

N xy 0 0 Α66 0 0 0 γ 0xy

= 0 (1.179)

Mx 0 0 0 D11 D12 D16 kx

My 0 k y0

0 0 D12 D22 D26 0

M xy 0 0 0 D16 D26 D66 k xy

N 0

y Α12 Α 22 Α 26 0 B22 0 εy

N xy 0 0 Α66 0 0 0 γ 0xy

= 0 (1.180)

M x B11 0 0 D11 D12 0 kx

My 0 k y0

B22 0 D12 D22 0 0

M xy 0 0 0 0 0 D66 k xy

N Α 0

y 12 Α 22 0 0 0 0 εy

N xy 0 0 Α66 0 0 0 γ 0xy

= 0 (1.181)

Mx 0 0 0 D11 D12 0 kx

My 0 k y0

0 0 D12 D22 0 0

M xy 0 0 0 0 0 D66 k xy

The Tsai–Wu Failure Criterion 39

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.

From a micromechanical point of view, the failure of fiber-reinforced mate-

rials is a complex and multi-parametric process. However, explaining the mi-

cromechanical approach to composites failure is beyond the scope of this book.

Macroscopically, models taking into account the applied stress components in

the principal coordinate system and the corresponding tensile and compressive

strengths, have been developed in order to estimate the limits of stresses that

cause failure.

The above models are called “failure criteria.”

Since “failure” is considered to be the loss of integrity of the material itself

[ e.g., 1 ] the stress situation yielding buckling is not covered by the failure criteria.

Generally speaking, the main target of a designer aiming to estimate the safety

factor of a composite structure is the determination (e.g., according to the algo-

rithm of Figure 1.10) of the principal stresses. In analogy with the failure criteria

for isotropic materials, a number of criteria for composites have been developed

in order to answer questions such as which principal stress component controls

failure.

40 Mechanical Behavior of Fiber Reinforced Composite Materials

failure?

Among existing failure criteria for composites, the Tsai-Wu [4] criterion seems

to be the most popular for design purposes and will be adopted for the case studies

in this book. Based on a concept similar to that followed for the derivation of the

von Mises criterion for isotropic materials, the Tsai-Wu criterion for composites,

assuming the plane–stresses assumption, is given by the following formula:

2

− F11 F22 σ1 σ 2 ≤ 1 (1.183)

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

σ1Τ : tensile failure stresses in the x1 direction

σ C2 : compressive failure stresses in the x2 direction

σ Τ2 : tensile failure stresses in the x2 direction

τ12F : shear failure stresses in the x1-x2 plane

The Tsai–Wu Failure Criterion 41

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

S 44 = S 44 cos 2 θ + 2 S 45 cos θ sin θ + S55 sin 2 θ

( )

S 45 = S45 cos 2 θ − sin 2 θ + ( S55 − S 44 ) cos θ sin θ

2 2

S 55 = S55 cos θ + S 44 sin θ − 2 S 45 cos θ sin θ

References 43

S 15 = S15 cos3 θ − ( S14 − S56 ) cos 2 θ sin θ + ( S25 − S46 ) cos θ sin 2 θ − S 24 sinn 3 θ

S 24 = S 24 cos3 θ + ( S 25 + S 46 ) cos 2 θ sin θ + ( S14 + S56 ) cos θ sin 2 θ + S15 sin 3 θ

S 25 = S 25 cos3 θ + ( − S 24 + S56 ) cos 2 θ sin θ + ( S15 − S 46 ) cos θ sin 2 θ − S 214 sin 3 θ

S 34 = S34 cos θ + S35 sin θ

S 35 = S35 cos θ − S34 sin θ

S 46 = ( 2 S14 − 2 S 24 + S56 ) cos 2 θ sin θ + ( 2 S15 − 2 S 25 + S 46 ) cos θ sin 2 θ

+ S 46 cos3 θ + S56 sin 3 θ

S 56 = ( 2 S15 − 2 S 25 + S 46 ) cos 2 θ sin θ + ( 2 S 24 − 2 S14 − S56 ) cos θ sin 2 θ

+ S56 cos3 θ + S 46 sin 3 θ

Si j = S ji

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

Technology of FRP Materials

fibers. Since the strength of the fibers is much higher than the strength of the ma-

trix, FRP materials have directionally dependent properties.

For the polymer matrices, two main classes of resins are used: thermosets and

thermoplastics. Thermosetting resins, such us epoxies, differ from thermoplas-

tics concerning their behavior during heating. At elevated temperatures, thermo-

sets transform their microstructure into an irreversible molecular chain (curing)

yielding an unchanged shape. Heating of thermosets after curing does not cause

melting. Actually, thermosets continue to retain their shape until their thermal de-

composition at very high temperatures. On the other hand, thermoplastics, such as

polyethylene, become malleable at high temperatures and solidify when cooled.

When reheated above a lower forming temperature, thermoplastics can be re-

shaped. This behavior of thermoplastics is important, since it allows them to be

repaired. For fabrication of fibers, three types of materials are in common use:

carbon (or graphite), glass and synthetics (e.g., Kevlar). The above materials are

characterized by high strength, thermal stability and low density.

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)

2.2.1 Glasses

Glass fibers are low-cost fibers suitable for piping applications (conveying cor-

rosive fluids), vessels, crafts, playground equipment etc. Their microstructure is

based on silica SiO2 which in the form of a polymer (SiO2)n does not melt but

softens progressively up to 2000oC. Although silica is a valuable material for en-

gineering applications, high temperatures are required to form the glass fibers.

Improvements in glass processing technology have led to the following types of

glasses suitable for composites manufacturing:

(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

The fabrication of carbon fibers starts from a precursor fiber, mainly made

from polyacrylonitril (PAN) or from pitch. Early precursor fibers were made from

rayon. Phenolics, polyvinylalcohols or polyimides may also be used but only for

limited cases. Although the final mechanical properties of carbon fibers are not

significantly affected by the type of precursor, their mechanical properties are

strongly influenced by processing techniques. Generally, by carbonizing organ-

ic precursor fibers, and then graphitizing them at very high temperatures, high

modulus carbon fibers can be produced. As provided by manufacturers, ranges

of the values of physical and mechanical properties of carbon fibers are given in

Table 2.6 [2].

Table 2.6

Values of physical and mechanical properties of carbon fibers.

Property PAN Pitch Type-P

modulus modulus

Tensile strength 2410–2930 2070–2900 1720 1720

(MPa)

(GPa)

Density (Kg/m3) 1780–1820 1670–1900 1860 2020

Diameter (μm) 8–9 7–10 7–10 10–11

50 Production Technology of FRP Materials

Synthetic (or polymeric) fibers have found increased use in the fabrication of

FRP composite materials due to their good chemical resistance and low density.

However, their maximum operating temperature is relatively low and varies be-

tween 100oC and 300oC. The most widely used synthetic fiber is Kevlar, devel-

oped in 1968 by the DuPont Co. It belongs to the family of aromatic polyamides

and its chemical name is poly (paraphenylene terephthalamide). Another impor-

tant polymer for fiber production is Spectra®, which has been developed by Allied

Signal. Spectra® is based on polyethylene containing oriented polymer chains.

Synthetic materials such as poly(benzobisoxazole) (PBO), aromatic copolyesters

and polyimides are also used for fiber fabrication with limited engineering use.

Physical and mechanical properties of widely used synthetic materials for fibers

are given in Table 2.7 [2].

Pipes

FRP composite pipes are fabricated by using the filament winding and fiber

placement methods. The matrix can be thermosetting or thermoplastic resins,

while the fibers can be glass, carbon or polymeric material. Both methods are

fully automated and yield lightweight and high-strength products.

2.3.1.1 Winding patterns

During filament winding, continuous reinforcements in the form of rovings or

monofilaments are wound over a rotating mandrel [3] (Fig.2.1).

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.

A critical parameter for successful filament wound products is the design of the

mandrel. The following design considerations should be taken into account [3]:

posed by the winding force.

52 Production Technology of FRP Materials

• 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.

The main concept of the fiber placement process is similar to the filament

winding method. However, in the fiber placement process, pre-impregnated tapes

instead of distinct fibers are placed on the mandrel’s surface (Fig.2.2). At the nip

point, a heat source is directed toward the fiber to melt the tape. A roller is used

to apply pressure at the nip point to spread the fiber and to apply compaction [3].

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.

Chapter 3

The loading conditions of a composite pipeline have fundamental importance

for its dimensioning. Underestimation of the initial sizing of a pipeline can cause

failure, with catastrophic environmental, economic and geopolitical consequenc-

es. On the other hand, as the length of a pipeline (especially for oil and gas trans-

mission) is extremely large, even minor overestimation in sizing (e.g., in the wall

thickness) can lead to critical consequences at the design competition stage of a

project. Therefore the type of loads to be considered is a key factor for a success-

ful design. The loads affecting the dimensions of a composite pipeline are classi-

fied into two main categories: 1. installation loads, and 2. operation loads.

Depending on the selected installation method, installation loads are very of-

ten more critical for pipeline dimensioning than operation loads. In the case of

offshore pipelines, the installation methods can be: (a) S-Lay (Fig.3.1), (b) J-Lay

(Fig.3.2), and (c) Towing (Fig.3.3 abc). In the case of continental pipelines, the

pipes are mostly embedded in soil (Fig.3.4), due to their sensitivity to ultraviolet

radiation. The installation of onshore pipelines thus requires movable cranes and

compaction machines.

53

54 Mechanical Design of Composite Pipelines

Types of Loading Cases 55

(a)

(b)

(c)

Figure 3.3 (a) Mid-depth tow, (b) Off-bottom tow, (c) Surface tow.

to pseudo-static loads of the following loading types: i) bending, ii) axial tension,

iii) external pressure, iv) combination of bending and axial-tension, v) combina-

tion of external pressure and axial-tension, vi) combination of bending, external

56 Mechanical Design of Composite Pipelines

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.

During service, composite pipelines can be subjected to static, pseudo-static,

or dynamic loading conditions at ordinary or elevated temperatures. During static

conditions, the loads continuously retain constant values, while during pseudo-

static conditions changes occur in them, assuming constant static conditions over

time. During static or pseudo-static loading, stresses can be produced by, for ex-

ample: a) internal pressure; b) deformation of the pipe due to its own weight

and the methods whereby it is supported; c) thermal stresses due to temperature

gradients; d) creep effects due to uniform (or variable) elevated temperatures;

e) moisture strain effects; and f) soil-pipe interactions, which affect underground

pipes. Dynamic loads can be caused by: a) vibrations resulting from hydrodynam-

ic forces due to internal axial flow or external cross-flow (fluid-pipe interactions);

b) impacts due to fluid hammer; or c) local impact due to foreign objects. During

variable loading conditions, the material mechanisms that can result in failure are

different from those taking place during static or pseudo-static loads. Therefore,

the failure criteria developed for static loads are inadequate for design purposes.

For variable loading cases, the designer has to use fatigue damage accumulation

rules in order to ensure the sustainability of the pipeline for the projected life.

Moreover, for long-term loading histories, especially at elevated temperatures,

the designer has to take into account deformation or possible rupture due to creep.

Current damage mechanics theory provides valuable theoretical tools [e.g., 1–3]

for an effective design that prevents fatigue or/and creep damage. Table 3.1 sum-

marizes the main composite pipeline operation loads, as well as required check-

points for sustainable design.

3.2.1 Failure analysis

Bending is the most common loading case in the installation of a composite

pipeline. During the procedure of, for example, S-Lay or J-Lay installation the

pipeline must be curved, which causes significant normal stresses in each lamina.

Moreover, while the pipeline is in operation, possible free-spans on the sea floor

of an offshore pipeline or soil settlement with underground piping can also cause

Pure Bending 57

Table 3.1

Classification of loads acting on composite pipelines.

Failure Buckling Fatigue Creep

criterion

Bending + +

Axial tension +

INSTALLATION LOADS

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 + +

OPERATION LOADS

Thermal stresses due to temperature gradients + + +

Moisture strain effects +

External pressure due to pipe-soil interaction + +

(forces due to soil weight and road or rail cross-

ings of underground pipelines)

Bending due to differential settlement of soil (of + +

underground pipelines)

Local impact by foreign objects +

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

pipeline under bending conditions include:

system, i.e.:

(a) The modulus of elasticity E1, E2

(b) The Poisson’s ratio ν12

(c) The shear modulus G12

(d) The longitudinal tensile strength s 1T

(e) The longitudinal compressive strength s 1C

(f) The transverse tensile strength s 2T

(g) The transverse compressive strength s 2C

(h) The in-plane shear strength τ 12F

2. the fiber orientation θ of each layer and the corresponding stacking se-

quence of the laminate;

3. the thickness h of each layer;

4. the internal diameter D of the pipe.

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

ν12 0.28 0.27 0.30 0.28

1080 1280 1447 1500

s 1T (MPa)

620 690 1447 1500

s 1C (MPa)

39 49 51.7 40

s 2T (MPa)

128 158 206 246

s 2C (MPa)

89 69 93 68

τ 12F (MPa)

Pure Bending 59

To derive a model for dimensioning a multi-layered filament wound pipe un-

der bending, the Lekhnitskii formalism for stress and displacements of a single-

layered pipe will be used, initially.

According to [5], the stress distribution on a single-layered pipe is given by the

following equations:

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].

62 Mechanical Design of Composite Pipelines

− 1

ε z = S 11 − ( S13 + S 23 + nS 23 ) r n −1C1 + ( S13 + S 23 − nS 23 ) r − n −1C2 −

S 33

(3.17)

{ }

−

S 12 ( n + 1) r C1 + ( n − 1) r

n −1

C2 + r C3 + 3 grA sin ϕ +

− n −1 −1

{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

−

(3.19)

S 22 ( n + 1) r n −1C1 + ( n − 1) C2 + r −1C3 + 3 grA +

− n −1

−

S 13 r n −1C1 − r − n −1C2 + r −1C3 + grA }

− 1

ε r = sin ϕ S 31 − ( S13 + S 23 + nS 23 ) r n −1C1 + ( S13 + S 23 − nS 23 ) r − n −1C2 −

S 33

−

(3.20)

S 32 ( n + 1) r n −1C1 + ( n − 1) r − n −1C2 + r −1C3 + 3 grA +

−

S 33 r n −1C1 − r − n −1C2 + r −1C3 + grA }

Pure Bending 63

1−

γ rϕ = − S 44 cos ϕ r n −1C1 − r − n −1C2 + r −1C3 + grA (3.21)

2

uϕ by the following relationships [e.g., 5]:

∂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)

into eq. (3.23) yields:

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)

Pure Bending 67

sin ϕ

σz = ( µ6C1 + µ7 C2 + µ8 A) (3.56)

S33

The equations (3.1)-(3.5) and (3.44), (3.45) express the stresses and deforma-

tions for a single-layered pipe. For the dimensioning of a multi-layered pipe under

bending (Fig. 3.7), we have to determine the stresses σr , σφ, σz, τrφ for each layer k.

Therefore, the next step is the determination of the unknown constants C1, C2, A

for all layers from k = 1 to k = N. To achieve this target, the following conditions

have to be satisfied:

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 )

The conditions on the exterior surfaces of the pipe located on r = r1 and r = rN+1

are:

τ r1ϕ ( r1 ) = 0

2 conditions (3.61)

τ rNϕ ( rN +1 ) = 0

Due to the action of stress σ zk (r , ϕ ) on an elementary area dA = r dϕ dr lo-

cated within an arbitrary layer k (Fig. 3.7), the corresponding elementary load is

dN z = σ zk ( r , ϕ ) .rdϕ .dr . This axial force yields an elementary bending moment

dM k = ydN , where y = r sin ϕ . Therefore, the bending moment due to the cross-

section of a layer k is

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

µ1 ( r3 , 2 ) C12 + µ2 ( r3 , 2 ) C22 + µ3 ( r3 , 2 ) A2 = µ1 ( r3 , 3) C13 + µ2 ( r3 , 3) C23 + µ3 ( r3 , 3) A3

...............................................................................................................................

...............................................................................................................................

...............................................................................................................................

k +1

µ1 ( rk +1 , k ) C1 + µ2 ( rk +1 , k ) C2 + µ3 ( rk +1 , k ) A = µ1 ( rk +1 , k + 1) C1 + µ2 ( rk +1 , k + 1) C2 + µ3 ( rk +1 , k + 1) A

k k k k +1 k +1

................................................................................................................................

.................................................................................................................................

................................................................................................................................

µ ( r , N − 1) C1 + µ2 ( rN , N − 1) C2 + µ3 ( rN , N − 1) A = µ1 ( rN , N ) C1 + µ2 ( rN , N ) C2 + µ3 ( rN , N ) A

N −1 N −1 N −1 N N N

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:

λ1 ( r3 , 2 ) C + λ2 ( r3 , 2 ) C + λ3 ( r3 , 2 ) A = λ1 ( r3 , 3) C + λ2 ( r3 , 3) C + λ3 ( r3 , 3) A

1

2 2

2

2 3

1

3

2

3

.............................................................................................................................

.............................................................................................................................

.............................................................................................................................

λ1 ( rk +1 , k ) C1k + λ2 ( rk +1 , k ) C2k + λ3 ( rk +1 , k ) Ak = λ1 ( rk +1 , k + 1) C1k +1 + λ2 ( rk +1 , k + 1) C2k +1 + λ3 ( rk +1 , k + 1) Ak +1

............................................................................................................................

.............................................................................................................................

.............................................................................................................................

λ1 ( rN , N − 1) C1N −1 + λ2 ( rN , N − 1) C2N −1 + λ3 ( rN , N − 1) AN −1 = λ1 ( rN , N ) C1N + λ2 ( rN , N ) C2N + λ3 ( rN , N ) AN

(3.71)

Similarly, with the aid of eq. (3.45), the conditions (3.60) can be written:

λ4 ( r3 , 2 ) C + λ5 ( r3 , 2 ) C + λ6 ( r3 ,1) A = λ4 ( r3 , 3) C + λ5 ( r3 , 3) C + λ6 ( r3 , 3) A

1

2 2

2

2 3

1

3

2

3

............................................................................................................................

............................................................................................................................

............................................................................................................................

k +1

λ4 ( rk +1 , k ) C1 + λ5 ( rk +1 , k ) C2 + λ6 ( rk +1 , k ) A = λ4 ( rk +1 , k + 1) C1 + λ5 ( rk +1 , k + 1) C2 + λ6 ( rk +1 , k + 1) A

k k k k +1 k +1

............................................................................................................................

............................................................................................................................

............................................................................................................................

λ4 ( rN , N −11) C1 + λ5 ( rN , N − 1) C2 + λ6 ( rN , N − 1) A = λ4 ( rN , N ) C1 + λ5 ( rN , N ) C2 + λ6 ( rN , N ) A

N −1 N −1 N −1 N N N

(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:

(3.73)

µ1 ( rN , N − 1) C1N + µ2 ( rN , N − 1) C2N + µ3 ( rN , N − 1) AN = 0

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)

T

(3.76)

{ 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)

0 λi (3, 2) −λi (3, 3) 0 0 0 0 0 0 0

0 0 λi (4, 3) −λi (4, 4) 0 0 0 0 0 0

[ Λi ]( N −1) xN = 0 0 0 0 0 0 0 0 0

0 0 0 0 0 λi (k , k − 1) −λi (k , k ) 0 0 0

0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 λi ( N , N − 1) −λi ( N , N )

(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

C2k is the coefficient C2 for the k-th layer

Ak is the coefficient A for the k-th layer

µi ( j , k ) is the coefficient µi for the k-th layer for r = rj

λi ( j , k ) is the coefficient λi for the k-th layer for r = rj

I ik is the coefficient I i for the k-th layer

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]).

When a bending moment M y is acting on a composite pipe (Fig. 3.8a), the

narrow strip AB is subjected to maximum compressive longitudinal strain ε ξ0 (0) . In

order to calculate the critical value of M y causing local buckling to the strip AB,

we have initially to estimate the critical value of εξο (0), namely εξcο, at the corre-

sponding buckling state. To this scope, we’ll assume that the above strip is a part

of the same pipe that has reached the critical longitudinal compressive strain εξcο

due to axial compression (Fig. 3.8b).

74 Mechanical Design of Composite Pipelines

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

the solution of the following equation:

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 )

A A22 0 B12 B22 0

12

0 0 A66 0 0 B66

[M o ] = B B12 0 D11 D12 0

(3.88)

11

B12 B22 0 D12 D22 0

0 0 B66 0 0 D66

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

3.3.1 Failure analysis

The long pipe under consideration (Fig. 3.9) with diameter D, thickness h,

stacking sequence [±θ NP ] , is subjected to external pressure pz. According to

Table 3.1, the first check to ensure its tolerance for this kind of loading should

be based on failure analysis. The maximum allowable external pressure satisfy-

ing the selected failure criterion can be determined by the procedure shown in

Figure 3.10.

78 Mechanical Design of Composite Pipelines

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

A long pipe with mean diameter D is made from a multi-layered laminate with

stacking sequence [±θ ] . According to ref. [12], the critical buckling pressure can

be estimated by the relation:

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:

− [ L1 ] L2 [ L1 ]

T

B =

Dani B22

(3.118)

ani D22

where

A12 B12

A B26

[ L1 ] = 26 (3.119)

B12 D12

B26 D26

Combination of Bending and External Pressure 81

A A66 B16 B66

[ L2 ] = B16 B16 D11 D16

(3.120)

11

B

16 B66 D16 D66

Pressure

3.4.1 Failure analysis

Superposition of the stresses σx, σy, τxy due to external pressure is given by:

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.

A laminated pipe of diameter D, made by a composite wall with stacking se-

quence [ ±Φ ] is subjected to combined pure bending and external compressive

pressure. The objective of this section is to determine the combination of the

critical values of the bending moment Mcr and the external pressure pcr that cause

buckling. We recall from section 3.3 that the critical bending moment Mcr can be

correlated (for design purposes) with the critical axial compression Nx cr through

the formula:

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.

Let u, v, w denote displacements along the axes x, y, z respectively. Taking into

account

(b) the equilibrium of resultant forces and compressive axial force Nx,

(c) the assumption that all resultant forces, except NX, are very small and we

can neglect the products of these forces with the derivatives of the displace-

ments u, v, w,

Combination of Bending and External Pressure 83

(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

Taking into account: (a) the geometric relations between the displacements

u, v, w; (b) the equilibrium of resultants and pressure q acting on a wall element;

(c) the assumption that all resultant forces, except Ny, are small and we can ne-

glect the terms containing the products of these resultants with the derivatives of

the displacements u, v, w; and (d) the assumption that the bending and twisting

moments are small and we can neglect the products of these moments with the

derivatives of the displacements u, v, w, the following relations based on [10] can

be written:

84 Mechanical Design of Composite Pipelines

∂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

For a general cross-ply laminated circular cylindrical shell, i.e., a shell with

stiffnesses A16 = A26 = A45 = B16 = B26 = D16 = D26 = 0, the resultant forces are given

[7] by:

N y = A12ε x0 + A22ε y0 + B12 k x0 + B22 k y0

N xy = A66γ xy0 + B66 k xy0

(3.132)

M x = B11ε x0 + B12ε y0 + D11k x0 + D12 k y0

M y = B21ε x0 + B22ε y0 + D12 k x0 + D22 k y0

M xy = B66γ xy0 + D66 k xy0

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

−2 2

− 2 B66 2 − 66 2 2 + R B11 3 + 12 2 + D11 4 + 122 +

R ∂x 2 ∂θ 2

R ∂ x∂ θ ∂x ∂ θ R ∂x ∂θ ∂x R ∂x ∂θ ∂x

1 ∂3u B ∂3 v ∂4 w D ∂4 w ∂2 w ∂2 w

+ B21 2 + 22 3 + D12 2 2 + 222 = +q w + 2 − N x R 2

4

R ∂θ ∂ x R ∂θ ∂θ ∂x R ∂θ ∂θ ∂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:

F21 F22 F23 B = 0 (3.139)

F F32 F33 C

31

where:

A11m 2 π 2 R 3

F11 = + A66 n 2 R + (3.140)

L2

A12 mn πR 2 A66 mn πR 2 mn πR 3

F12 = − + − q (3.141)

L L L

Combination of Bending and External Pressure 87

F13 = + + (3.142)

L L L3

B12 mn πR 2 B66 mn πR 2 A12 mn πR 3 A66 mn πR 3

F21 = − − − (3.143)

L L L L

B66 m 2 π 2 R 3 A66 m 2 π 2 R 4 m 2 π 2 R 4

F22 = B22 n 2 R − A22 n 2 R 2 − − − N x (3.144)

L2 L2 L2

B12 mn πR 2 B66 mn πR 2 A12 mn πR 3 A66 mn πR 3

F23 = − − − (3.145

L L L L

B12 mn 2 πR 2 2 B66 mn 2 πR 2 B11m3 π3 R 4

F31 = − + (3.146)

L L L3

B12 m 2 n π 2 R 3 2 B66 m 2 n π 2 R 3

F32 = B22 n3 R + − (3.147)

L2 L2

2 D12 m 2 n 2 π 2 R 2 4 D66 m 2 n 2 π 2 R 2

F33 = D22 n 4 + − + qR 3 −

L2 L2

(3.148)

2 3 m2 π2 R 4 D11m 4 π 4 R 4

n qR − Nx +

L2 L4

The equation for calculating the critical value of the pairs of external pressure qcr

and axial compression N xcr has the following form:

det F21 F22 F23 = 0 (3.149)

F31 F32 F33

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

curves are almost straight lines. The safe values N xcr , qcr are included within the

polygon OABCD.

Taking the point of intersection of the polygon with the horizontal axis (point

D), we obtain the critical value of q, providing that only lateral external pressure

alone is acting. On the other hand, the intersection of the polygon with the vertical

axis (point A) indicates the critical value of bending moment M when q = 0. Since

the procedure for deriving the lines M cr vs qcr for several values of (m,n) is

88 Mechanical Design of Composite Pipelines

Figure 3.14 Curves N xcr vs qcr for several pairs (m, n).

D and A and to draw the line AD. The points within the area of triangle OAD can

be used for engineering purposes and will deliver safe results.

3.5.1 Failure analysis

A multilayered pipe with a mean diameter D made by a wall composed by NP

layers with a stacking sequence [ ±θ ] (Fig.3.15) is considered.

∧

We shall assume that the load N is distributed uniformly around the circum-

ference of the cross-section. Therefore, the load per unit length Nx of laminate is:

∧

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)

Therefore:

ε x = ε x0 + zk x0 (3.162)

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.

3.6.1 Failure analysis

Superposition of the stresses σx, σy, τxy due to axial tension given by:

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

92 Mechanical Design of Composite Pipelines

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.

∧ ∧

When the combined bending moment M y and axial tension N x are acting

∧

on a pipe as in Figure 3.17a, both compressive (due to M y ) and tensile (due to

∧

N x ) stresses are placed upon the narrow strip AB. In cases where the value of the

∧

compressive stress due to M y is higher than the value of the tensile stress due to

∧

N x , a check for buckling has to be performed. To carry out such an inspection, the

concept followed for the approximation of a critical value of the bending moment

for the case of pure bending will again be adopted. However, the value of the local

strain ε ξο at the point A located at the center of the cross-section of the strip AB

∧ ∧

(Fig. 3.17a) is correlated with both bending moment M y and axial tension N x

through the formula:

Combination of Bending and Axial Tension 93

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).

Tension

3.7.1 Failure analysis

The composite pipe shown in Fig. 3.18(a) has a mean diameter D. Its wall is

composed of NP number of layers with the stacking sequence [±θ ] . This pipe is

∧

subjected to combined axial tension N x in the x-direction and external pressure

p. The axial load per unit length of the perimeter is:

∧

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

My = 0, Mxy = 0, the following matrix equation can be derived:

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

«D » °1 ° ° °

« D D E E E − » ° Q ° ° °

« D D D E E E − » ° 1ξ Q ° ° °

« » ° ° ° °

« E E E G G G − » ° 0ξ ° ° °

« E E E G G G − » ° 0 Q ° ° °

« » ° ° ° °

« E E E G G G 0

−» ° ξ Q ° ° °

« ⋅ ® ¾ = ® ¾ (3.204)

» ° εξ ° °

°

« »

» ° εQ ° ° °

«

« » ° ° ° °

° γ ξ Q ° ° 0Ö [ π 5 °

« »

» ° Nξ ° ° °

«

« » ° N ° ° °

« » ° Q ° ° °

«¬ »¼ °¯ Nξ Q °¿ °¯ °

¿

�

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 long laminated pipe of diameter D, made from a multilayered composite

wall with the stacking sequence [ ±φ ] is subjected to pure torsion. The target now

� x that causes elastic buckling.

is the estimation of the critical torque M

When one considers:

(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

G

G22 G23 Β = 0 (3.222)

21

31 G32

G G33 C

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:

det G21 G22 G23 = 0 (3.232)

G G33

31 G32

Therefore, the above equation yields an expression for Mx versus the parameters

m and λ. It should be noted that parameter m is an integer while λ is a real number.

For m = 1, it can be seen from equations (3.219)–(3.222) that a cross-section of

the pipe remains circular and moves, during buckling, only in its plane. For m = 1,

in order to calculate the critical value of Mx that causes elastic buckling, we have

to find the value of λ that minimizes the expression for Mx. Normally the minimi-

zation of the function Mx cannot be achieved by an analytical method. However,

for existing commercial software like Mathematica, numerical minimization of

Mx is an easy task.

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.

Since composite pipes are thin-walled, lightweight structures, they can be ana-

lyzed as slender beams. We consider a composite pipe of length L, flow area A

and mass per unit length m, conveying fluid of mass per unit length M with mean

axial flow velocity U. Due to axial flow, small lateral motions w(x, t) of a long

wavelength (compared to the diameter) will be observed. The aim of the present

section is to estimate the eigenfrequencies and eigenmodes, taking into account

the structural characteristics of the multilayered composite pipe and its interaction

with the fluid flow.

When lateral motions w(x, t) are induced due to bending, the main structural

parameter controlling vibration is the stiffness EI� yy (with respect to axis y) of the

composite pipe.

For a wall element such as that illustrated in Figure 4.1a located at (z, a) in the

cross-section of a multilayered pipe (Figure 4.1b), the local strain and curvatures

in its middle plane are:

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

On the other hand, the relationships between the strains and curvatures with

the forces and moments for a symmetrical laminate are given by the following

well-known [4] equations:

Free Vibration of Composite Pipes 107

o

ε n = a12 a22 a26 N n (4.3)

γ o a a26 a66 Nξ n

ξ n 16

o

κ n = d12 d 22 d 26 M n (4.4)

κ o d d 26 d 66 M ξ n

ξ n 16

Taking into account that for the wall element of Fig. 4.1a is N n = 0, Nξ n = 0, M n = 0,

M ξ n = 0, eqs (4.3), (4.4) above yield:

1 o

Nξ = εξ (4.5)

a11

1 o

Mξ = κξ (4.6)

d11

dM (4.7)

ξ ξ

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

Therefore, the bending moment M

pipe is:

� 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

It is known that the stiffness EI

the y-axis) is given by the following equation:

� 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

Q12 = ( Q11 + Q22 − 4Q66 ) n m + Q12 ( n + m )

2 2 4 4

Q16 = ( Q11 − Q12 − 2Q66 ) nm3 + ( Q12 − Q22 + 2Q66 ) n3 m

(4.19)

Q 22 = Q11n 4 + 2 ( Q12 + 2Q66 ) n 2 m 2 + Q22 m 4

3

Q 26 = ( Q11 − Q12 − 2Q66 ) n m + ( Q12 − Q22 + 2Q66 ) nm

3

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.

Each span of the composite pipe depicted in Figure 4.4 has length L, inter-

nal diameter D and is fixed by supports at both its ends. During fluid flow, the

forces and moments acting on an elementary section of pipe are demonstrated in

Figure 4.5.

112 Dynamic Stability of Composite Pipelines

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

From geometry it is well known that the curvature (1/ ρ y ) is associated with

the deflection w as expressed in the following relation:

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).

∂p ∂w

−A − q πD + F = Maxf (4.33)

∂x ∂x

In the above equation, axf is the horizontal acceleration of the fluid due to its

flowing motion in the x-direction, and in the z-direction because of the vertical

motion of the pipe.

∂ ∂w ∂w

−F − A p − q πD − Mg = Mazf (4.34)

∂x ∂x ∂x

where azf is the vertical acceleration of the fluid due to its motion in the x-direc-

tion due to flow, and in the z-direction due to the vertical motion of the pipe.

We consider that the motion of the pipe element takes place only in the vertical

direction. Therefore its acceleration in the horizontal direction is:

axp = 0 (4.35)

Accelerations of the Fluid and Pipe Elements 115

will be:

∂2 w

azp = (4.36)

∂t 2

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

116 Dynamic Stability of Composite Pipelines

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

Neglecting quantities associated with tensioning, pressurization effects and

gravity, which are not important for motion, and considering a constant velocity

U, the equilibrium equations of the pipe element (4.27), (4.28), (4.32) and the

fluid element (4.33), (4.34) can be simplified as follows:

∂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.

The aim of this section is to calculate the eigenfrequencies ω and critical values

Ucr of fluid velocity that cause instability of a composite pipe conveying fluid. The

corresponding boundary-value problem consists of the homogeneous fourth-order

partial differential equation (4.52) and the boundary and initial conditions of the

problem. Assuming that the motion w(x,t) starts at t > 0, the initial conditions at

t = 0 can be written as:

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

The boundary conditions depend on the type of supports. Table 4.1 summa-

rizes the boundary conditions for different types of support.

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:

mined. Generally the circular frequency ω can be a complex number. The system

will be stable when Im[ω ] > 0 and unstable when Im[ω ] < 0 . In cases where ω is

a real number, the system has neutral stability.

120 Dynamic Stability of Composite Pipelines

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:

For this type of pipe the ends are fixed-free. Therefore the boundary conditions

are given by equations (4.57), (4.58), and (4.59), (4.60). With the aid of equations

(4.69), (4.71), and (4.73–4.85) the boundary conditions above can be written as:

C1 + C2 + C3 + C4 = 0 (4.86)

(4.86)-(4.89), the corresponding determinant of the matrix of coefficients must

vanish:

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)

The equation of motion (4.52) contains the superposition of four terms. The

meaning of these terms is:

4

• EI

dx 4

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.

Problems in engineering are often formulated in terms of a boundary-value prob-

lem where the unknown function f(x) of the Ordinary Differential Equation (ODE)

n

dm

∑a

m=0

m

dx m

f ( x) = q ( x), an ≠ 0, x ∈ [ x1 , xn ] (4.96)

of the unknown function f(x) (or of some of its derivatives) are prescribed at two

distinct boundary points: x1 and xn.

The Ordinary Differential Equation given by eq. (4.96) is considered to be ac-

companied by the following boundary conditions at the boundary points x1 and xn

respectively:

known values. The proposed method presupposes the existence of the n unknown

boundary values u1,u2,…,un:

Assuming the transformations

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:

Differential Equation (ODE) (4.96) can be written:

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:

128 Dynamic Stability of Composite Pipelines

[ I ] [ 0]

[ A21 ] = kxk

[0]

(4.125)

[0]

[0] [0]

[ A22 ] = (4.126)

[ 0 ] [ I ](n−k ) x(n−k )

xn

x1

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)

algebraic system:

[ AA ] [ AA ] ⋅ [ X Ξ] = [ BB ] (4.130)

21 22 2

where the sub-matrices [AA11], [AA12], [AA21], [AA22], [BB1], [BB2] are given by the

following matrix equations:

x1

Estimation of Critical Velocity for Composite Pipes Conveying Fluid 129

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

Composite Pipes Conveying Fluid

(General case: solution of equation (4.52))

Taking into account the theory described in the previous section, eq. (4.122)

has to be formulated for the case of a composite pipe conveying fluid. Assuming

that the length of the pipe is L, the matrix [AN] should be derived by drawing on

the coefficients a0 , a1 , a2 , a3 of the differential equation (4.70). For n = 4, a com-

parison of eqs. (4.70) and (4.96) yields:

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

eq. (4.122) yields:

130 Dynamic Stability of Composite Pipelines

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 ]

with U. For any value of U, the value of ω is a complex number. As was already

mentioned, instability occurs for the value of U cr yielding Im [ω ] < 0 .

For several values of U, equation (4.154) can be solved numerically using a

standard commercial software, e.g., Mathematica. The obtained results Re [ω ]

and Im [ω ] can be presented in an Argand diagram ( Re [ω ] vs Im [ω ]) showing

the evolution of the system with increasing U, from stability (U < U cr ) to insta-

bility (U ≥ U cr ) .

In long pipes the temperature can vary over certain portions of their length. The

effect of thermal loads yields longitudinal strain εx. On the other hand, for pipes

where both ends are supported (e.g., fixed-fixed, pinned-pinned, fixed-pinned), no

longitudinal movement is allowed. Therefore, the condition

L

∫ ε dx = 0

0

x

(4.27) and (4.33) yields:

∂T ∂P

−A = maxf (4.155)

∂x ∂x

Assuming a constant flow velocity U, and assuming that the pressure is indepen-

dent from x, the above equation can be written as:

∂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

pansion in the x-direction. According to Chapter 1, the coefficient ax can be cor-

related with the coefficients of thermal expansion in the principal directions of a

lamina a1 , a2 by the equation:

With the aid of the above equation, eq. (4.160) can now be written:

where:

S11 = +− + cos θ sin θ + (4.163)

E1 E1 G12 E2

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 2

EI 4 1 2 2

∂x S11 2 ∂x

∂2 w ∂2 w

2 MU + ( M + m) 2 = 0

∂x∂t ∂t

(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.

Because the specific weight of composite materials is relatively low (almost

25% of the specific weight of steel) and their stiffness is very high, composite

pipes can be characterized as lightweight structures. During the installation of a

composite pipe of diameter D = 1.0 m and thickness 4.0 mm into the sea, a buoy-

ancy reaction of 0.78 tn/m acts on the pipe in a direction opposite to its weight.

Therefore, in order to install hollow composite pipe underwater, additional mass

must be added to offset the buoyancy forces. Such additional weights may be equal-

ly spaced metallic collars located, for example, on the joints of the pipe (Fig. 4.8a).

However, the mass and the additional stiffness of these collars alter the gravita-

tional forces of the pipe and therefore influence its dynamic behavior. In order to

estimate the critical velocity U cr for a composite pipe containing additional mass,

the local transfer matrix of each pipe segment equipped with a collar needs to be

derived. Taking into account equation (4.143), the matrix [ AN ] now has the form:

(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)

c

(4.178)

c

(4.179)

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.

138 Dynamic Stability of Composite Pipelines

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:

Thickness of ply: h = 0.15 mm

Number of plies: NP = 10

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

From the definition of a transfer matrix (equation (4.120)), the following can

be written:

{ XN3 } = [TM 23 ]{ XN 2 }

{ XN 2 } = [TM 12 ]{ XN1}

T

Therefore, the global transfer matrix [GTM ] of the system correlating the state

vectors of the end points 1 and 3 is always the product of the transfer matrices of

the segments, i.e., [GTM ] = [TM 23 ] ⋅ [TM 12 ] .

Since [ A] ⋅ [ B ] ≠ [ B ].[ A] , it is important to mention that the multiplication of

the transfer matrices should be started from the transfer matrix of the last segment.

We recall that the algebraic equation yielding ω versus the fluid flow U is

eq. (4.154). In this equation, the transfer matrix should now be replaced by the

[GTM ] derived above, i.e.,

[GTM ] [ − I 4 x 4 ] = 0

det

[ A22 ]

[ A21 ]

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

Very often, long composite pipes are embedded in soil or lie on the bottom of

the sea. The additional support of the soil yields a stiffer system. In order to simu-

late the dynamic behavior of a composite pipe resting on an elastic (Winkler-type)

foundation, the equation of motion (eq. 4.52) should be adapted [2] by taking into

account its elastic constant k, i.e.:

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.

Long composite pipes may be supported on more than two supports. In most

cases, the intermediate supports are simple, and the spans between them are

equally spaced as depicted in Figure 4.10.

The state vectors { y, y1 , y2 , y3 } of two successive nodes i, j are correlated by

T

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

Apart from the spans, a similar formulation can be derived for the cross-sections

just before (L) and just after (R), an intermediate support (Fig. 4.11).

Effect of Additional Supports 143

• •

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

L R L R L R T

(4.203)

{Y }1 = { y, y1 , y2 , y3 }1

T

(4.204)

T

{Y } j

L

j

{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)

146 Dynamic Stability of Composite Pipelines

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

of Fig. 4.3 and using eq. (4.16) the value EI � yy = 196224 N ⋅ m 2 has been obtained

for the stiffness of the pipe. Using the above data, the mass per unit length for

the pipe and the fluid have the values m = ρ p (2π r1 )( NP ⋅ h) = 4.948kg / m and

M = ρ f (π r12 ) = 7.854kg / m . Using the derived values EI � yy , m ,

M , the follow-

ing formula for the matrix [ AN ] has been obtained:

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

Relation to Divergence

As mentioned, divergence may occur if both ends of a pipe are supported (for

cantilevered pipes this instability is unlikely to happen). For sufficiently high-flow

velocities the bending due to divergence instability can be so extreme that the pipe

can fail. Therefore, the estimation of the critical flow velocity where divergence

instability can be encountered is an important consideration in the design of slen-

der pipelines that carry fluids. We recall that the eigenvalue problem associated

with divergence instability is described by the following differential equation:

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:

D

AN ]⋅ Lij

(4.216)

As it is already known, the eigenvalues ω versus the fluid speed U for a pipe

supported to its both ends can be obtained by the following equation:

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

A sudden change of discharge in a composite pipe conveying liquid may re-

sult in stresses of sufficient magnitude to exceed the design stresses. The reason

for this phenomenon is the created impulse force, which is a result of Newton’s

impulse-momentum equation. Actually the impulse force is a pressure shock that

travels with high velocity in the upstream and downstream directions. The cre-

ation of pressure shock in a fluid due to a sudden change of discharge is called

hydraulic hammer and must be taken into account as an important parameter for

pipelines designed with composite materials.

150 Dynamic Stability of Composite Pipelines

initially consider the evolution of the liquid volume element ABCD shown in

Fig. 4.12.

Before a sudden stoppage of flow, the diameter of the liquid element was r

and its length was (ΔL +ΔL0). After the sudden stoppage of flow, the length is

decreased by ΔL and the diameter increased by Δr, which causes elastic dilatation

of the pipe’s walls. The volume of the liquid element before and after this change

is given by the following known formulas:

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

of the movement of the center of gravity of the liquid element, which causes re-

tardation a. According to Newton’s law, the retardation a can be estimated by the

following formula:

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

Hydraulic Hammer 153

∆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

fluids of infinite extent, which is called “celerity,” while ∆L0 / ∆t is the celerity c

of the shock wave in the composite pipe. Therefore, eq. (4.246) can now be writ-

ten in the following form:

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

neous closure, and the shock pressure will reach its maximum value P∗ given by

eq. (4.248).

Wave Propagation Due to Hydraulic Hammer 155

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).

In this section, the effect of moving pressure shock on the deflection of a pipe’s

wall will be analyzed. The high velocity of pressure shock causes intensive vibra-

tion of the wall, which leads to dynamic stresses.

For the derivation of the dynamic model of the radial deflection w ( x, t ) cor-

responding to a point of the mid-plane of the pipe’s cross-section located at a

distance x where the equilibrium equations of a wall element (Fig. 4.14) will be

initially considered [e.g. 4]:

∂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

tions can be derived:

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

x = L / 2 of the pipe and

ω

a= (4.297)

Ω(1)

Ω(1) L

= ccr (4.298)

π

Wave Propagation Due to Hydraulic Hammer 161

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)

Qi , j ( i, j = 11, 12, 21, 22, 66 )

Qi , j ( i, j = 11,, 12, 16, 22, 26, 66 )

A A22 A26 B12 B22 B26

12

A A26 A66 B16 B26 B66

[ ABD ] = B16 B12 B16 D11 D12 D16

11

B12 B22 B26 D12 D22 D26

B16 B26 B66 D16 D26 D66

a a22 a26 b21 b22 b26

12

a a26 a66 b61 b62 b66

[ abd ] = b16 = [ ABD ]

−1

b12 b22 b62 d12 d 22 d 26

b16 b26 b66 d16 d 26 d 66

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

now be obtained (eqs. 4.274–4.277):

[ a1 ] =

20388 62.7832

[ a2 ] =

−217 993.966

[ a3 ] =

19300.3 7.89715 × 106

[ a4 ] =

62783.2 907.115

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

11 =

H 21 H 22 −103.248 15.5797

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

the normalized deformation w ( x, t ) / w0 for a pipe with length L = 1000 m and

a = 0.8 can be obtained from eq. (4.296).

The following diagrams display the results for t = 0, 0.1, 0.2, ...., 1.0 sec.

,

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

Numerous joining methods are employed in composite pipelines depending on

the manufacturer. However, the standard method for joining filament-wound FRP

pipes is the butt joint (Fig. 5.1). This conventional joint is made by wrapping lay-

ers of fiber impregnated with a catalyst resin over the butted joint. The butt joint is

economical, permanent and very satisfactory [1]. Another widely used connection

method based on joining pipe pieces by adhesive layer is the “socket adhesive

joint” (Fig.5.2). In order for the joint to be effective, the adhesive layer must be

able to carry the developed shear stresses.

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.

During axial loading of two filament-wound FRP pipes connected by an

adhesive joint (Fig. 5.1 or Fig. 5.2), the normal stresses in each pipe segment

are uniformly distributed along the periphery of the cross-section of each pipe.

Therefore, the mechanical model simulating the loading of the joint can be ap-

proximately represented by the configuration of two thin strips and an adhesive

layer, as depicted in Figure 5.3.

In this model, the center of coordinate x is located in the middle of L. The dis-

tribution of shear stresses along the length L of adhesive is not uniform. Therefore,

in order to estimate the maximum shear stress, the deformation of an element of

the adhesive layer will first be analyzed (Fig. 5.4).

Joining of Composite Pipelines 171

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

dγ 1

= ( ε1 − ε 2 ) (5.5)

dx 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)

Joining of Composite Pipelines 173

3] of a lamina given by:

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)

layered filament-wound pipe, the above Hooke’s law for a pipe’s laminated wall

consisting of NP layers can be written as:

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

x = L / 2 and x = 0

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)

this figure it can be concluded that the maximum values of τ ( x) occur at the ends

x = ± L / 2 of the adhesive layer. Therefore,

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

According to this figure it can be concluded that for large values of L, the capacity

of the joint to carry axial force is independent of the adhesive length.

Since for large values of L

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

For estimating the allowable bending moment for a butted joint, we shall again

use the value of an imaginary axial force N that causes the same maximum strain

ε ξo with the actual bending moment M (see Chapter 3).

According to this concept (Fig. 5.8), the axial force N yielding equivalent max-

imum strain with the bending moment M can be approximated from the following

formula:

�

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.

When a pipeline is installed above the ground, the types of supports shown in

Figures 5.9-5.11 are required.

Pipelines above ground can be suspended (Fig. 5.9), clamped on the ground

(Fig. 5.10), or placed in a shoe (Fig. 5.11).

Important parameters to be derived for pinned-supported pipes are: (a) the

maximum spacing between supports, (b) the minimum hanger widths, and (c) the

allowable deflection of expansion loops for anchored pipelines.

180 Connection and Supports of Composite Pipelines

In order to estimate the maximum space L between supports of a long pinned-

supported pipeline, the static model of a continuous beam (Fig. 5.12) will be ad-

opted to simulate the pipeline.

One criterion for estimation of L is the maximum bending moment M max given

by [4]:

Above-Ground Pipes 181

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.

In order to estimate the minimum hanger width B, the segment of pipe within

the hanger will be simulated using a thin-walled ring (Fig. 13a) of mean radius

R = D/2.

The contact force Q� transferred from the hanger to the pipe (point C) causes

[5] a bending moment M �

c to the lower area of the ring (Fig. 13b).

�

M �

c ≈ 0.24QR (5.48)

�

and a compressive force N c

� ≈ 0.24Q

N � (5.49)

c

ξ act in the x direction due to bending of the

�

pipeline by the bending moment M y (Fig. 5.13a). We recall that the above com-

pressive forces can be estimated by eq. (5.42).

Above-Ground Pipes 183

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).

According to Table 5.2 it can be shown that the value of µ tends to be sta-

ble for more than six supports. The maximum value µ = 0.605 can be adopted,

yielding:

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:

Nξ = (5.56)

(d 11 D 2 + 4a11 ) B

Combining the laminate compliance matrix (“abd” matrix) with eqs. (5.54),

(5.55), and (5.56) the following matrix equation can be obtained:

«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 [ ° ° − 1ξ °

« »

» ° ε \ ° °− 1 F °

«

« ° ° ° °

» ° γ [\ ° ° °

« »

» ° N [ ° ° °

«

« » °

° ° °

« » ° N \ ° °− 0 F °

¬« ¼» ¯° N [\ ¿° °¯ °¿

(5.57)

ε xo , ε yo , γ xyo , k xo , k yo , k xyo :

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

With the aid of eq. (5.59), the stress components σ x , σ y , τ xy can be obtained

for each lamina using the following well-known relation:

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.

Pipelines with free ends experience changes in dimensions as a result of tem-

perature variation. When the ends are anchored, the pipeline will be placed in

a condition of stress and will exert reactive forces and moments at its ends. To

avoid the catastrophic consequences of possible buckling due to the above condi-

tions, expansion loops (Fig. 5.15) are installed, especially for longer lines. The

basic problems in analyzing temperature change effects in a configuration such as

that seen in Fig. 5.15 are to estimate the magnitude of such stresses and to check

whether or not they are tolerable. For these problems, the mechanical model of

Fig. 5.16 will be used to simulate the system pipeline-expansion loop shown

in Fig. 5.15. Assuming that an expansion loop is located at the midpoint of the

distance between two anchors, we can analyze the half structure because of its

symmetry.

Above-Ground Pipes 187

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

modulus of elasticity Ex given by equation (5.9), and I is the moment of inertia

of a thin-walled pipe given by:

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]:

VB = (5.69)

CVV ⋅ CMM − CVM

2

MB = (5.70)

CVV ⋅ CMM − CVM

2

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:

VB = F (5.78)

CVV CMM − CVM

2

190 Connection and Supports of Composite Pipelines

ΜB = F (5.79)

CVV CMM − CVM2

With the aid of eqs. (5.66), (5.78), (5.79), eq. (5.63) yields:

∆xB = [CHV 2

+ CHM MH VV − CHH ]F (5.80)

CVV CMM − CVM CVV CMM − CV2M

Combining the above equation with eq. (5.62), the following formula, which pro-

vides the spring constant k, can be obtained:

k = [CHV 2

+ CHM MH VV 2

− CHH ]−1 (5.81)

CVV CMM − CVM CVV CMM − CVM

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:

coordinates 1 and 2, and θ is the fiber orientation in a filament wound pipe with

the stacking sequence [±θ ] .

Since the spring (Fig. 17b) is compressed by a deflection:

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:

MB = k ⋅ (5.91)

E A+k L CVV CMM − CVM2

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

Locating composite pipelines under roads and railroads is a common practice.

In these cases the pipeline should be installed at a great enough depth, which is the

main design parameter. For estimating the minimum value of installation depth,

the soil load and wheel load should be accounted for. Due to these loads, any

soil element located at a distance z below the surface is subjected to compressive

stresses s y , s z (see Fig. 5.18).

Since during maintenance the pipeline is empty, the above compressive stresses

cannot be equilibrated by internal pressure. Therefore, internal pressure should be

ignored and only s y , s z due to wheel and soil load should be taken into account

for dimensioning the installation of an underground pipeline. From Figure 5.18 it

can be shown that s y vanishes rapidly versus depth. Therefore, the estimation of

the installation depth by taking into account only s z should be a safe scenario.

The distribution of s z versus z below the contact surface can be obtained from

the following solution of Boussinesq [6] concerning semi-infinite elastic, homo-

geneous, isotropic soil:

3F

σz = (5.96)

2π z 2

Underground Pipelines 193

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 acting on the point C (conserva-

tive scenario) are given by the following formulas:

� c = 1 q R2

M (5.99)

4

� c =q R

N (5.100)

My = M� c b and the unit axial force N = N

� c b acting on the pipe’s cross-sec-

y

tion (Fig. 5.20) at location C.

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 , σ y ,τ xy of the exterior lamina are

196 Connection and Supports of Composite Pipelines

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

Composite Materials

1

Department of Mechanical Engineering (DEMec), Faculty of Engineering of the

University of Porto (FEUP), Rua Dr. Roberto Frias s/n, 4200-465 Porto, Portugal.

2

INEGI-Instituto de Engenharia Mecânica e Gestão Industrial, Rua Dr. Roberto Frias,

400, 4200-465 Porto, Portugal

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.

Composite Materials

In engineering, creep designates the gradual increase in strain that occurs in a

material when it is subjected to a sustained load over a certain period of time. Even

at room temperature polymers and polymer-based composites can undergo creep at

relatively low stress levels. Moreover, given enough time, creep will lead to rupture.

Usually the creep strain evolution may be divided into three distinct regions:

I) first-stage or primary creep, which starts at a rapid rate and slows with time; II)

second-stage (secondary) creep, where creep has a relatively uniform rate (mini-

mum gradient); III) third-stage (tertiary), where creep manifests an accelerating

creep rate and terminates with failure of the material. This is illustrated schemati-

cally in Figure 6.1.

Creep Damage Accumulation Mechanisms in Composite Materials 199

Failure

Region II

Region I

Time

figure 6.1 Typical creep strain evolution divided into three regions.

process. Since the stress-strain analysis is based on continuum mechanics, a dif-

ficulty presents itself when attempts are made to predict failure in general and

creep failure in particular of polymers and polymer matrix based composites.

Fracture mechanics and damage mechanics include the distribution of defects into

continuum models, which allow time-dependent failure prediction. Energy-based

failure criteria provide another possible approach. An example is establishing that

the energy accumulated in the springs of the viscoelastic mechanical model, des-

ignated as free energy, has a limit value. This limit can be or not be constant.

Earlier approaches to the prediction of time-dependent failures provided explicit

elementary equations to predict behavior over a lifetime. Following this line, a

global and homogeneous analysis was chosen because it is more convenient for

practical engineering applications.

In this context it becomes important to predict the effects of long-term load-

ing on deformation and failure behavior. A reliable prediction requires the use of

models that capture accurately the stress/strain time-dependent evolution of the

filament-wound cylindrical pipes. Most plastics, composites and other synthetic

materials exhibit strong hereditary type non-linear viscoelastic behaviors, which

render any stress-strain analysis of these materials a significant challenge. Single-

integral formulations, such as the Schapery nonlinear viscoelastic theory [2], have

proven to be very effective in representing the creep strain of filament-wound

pipes [3]:

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

∆J ( t ) is the time-dependent compliance. The compliance is usually expressed

by a power law [2–4]:

∆J ( t ) = J1t n (6.3)

Ghorbel [3] simplified the Schapery formulation since he did not perform all

the necessary creep and creep-recovery tests to obtain the full material stress-

dependent nonlinearizing parameters. Furthermore, it was found that g 0 = 1 . The

simplified expression for strain creep under sustained load was

ε ( t ) = ε ( 0 ) + g (σ ) t n (6.4)

m

wound cylinders is not usually done; only the time-to-failure is needed to qualify

pipes.

In the case of multi-directional filament-wound pipes, it is possible to perform

elastic-viscoelastic analyses using an approach based on classical laminated plate

theory (CLPT) as exemplified in [5]. That specific methodology includes a time-

dependent failure criterion. However for thick cylinders the CLPT approach does

not work. Hence a method to calculate the time-dependent stress-strain state in

nonlinear viscoelastic thick multilayered composite cylinders was proposed and

implemented [6]. This work demonstrated the importance of the viscoelasticity

effect over the time-dependent internal stress distribution through thick laminated

cylinders.

One of the first theoretical attempts to include time in a material strength for-

mulation for viscoelastic materials was developed by Reiner and Weissenberg [7].

The Reiner-Weissenberg criterion [7] states that the work done during the loading

by external forces on a viscoelastic material is converted into a stored portion

(potential energy) and a dissipated portion (loss energy). Hence the criterion as-

sumes that the instant of failure depends on a conjunction between distortional

free energy and dissipated energy, with the threshold value of the distortional

energy being the governing quantity.

Let us assume that the unidirectional strain response of a linear viscoelastic

material under arbitrary stress s(t), is given by the power law as:

Creep Damage Accumulation Mechanisms in Composite Materials 201

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

over time, measured experimentally, resemble one another in an extraordinary

manner. Most probably this signifies that a change in the relaxation modulus cor-

responds to a change in the strength.

The rate theory of fracture is based on a molecular approach, i.e., on the kinet-

ics of molecular flow and the bond rupture of the polymer chains. Drawing on

these approaches Zhurkov [15] presented one of the first models for predicting a

material’s lifetime tf (except for very small stresses) in terms of a constant stress

level s,

is the Boltzmann constant, T is the absolute temperature, U0 is a constant for each

material regardless its structure and treatment, and g depends on the previous

treatments of the material and varies over a wide range for different materials.

Griffith et al. [16] applied a modified version of the Zurkov equation:

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

is a parameter governed by the geometry and viscoelastic properties. For constant

stress, s = s 0 , the lifetime is given as:

tf α 1

= 1 m − 1 . (6.16)

τ0 γ γ

is provided by the method of continuum damage mechanics (CDM). Following

the original ideas of Kachanov [24], the net stress, defined as the remaining load-

bearing cross-section of the material is given [25] by:

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

process, Fa is the normalized failure function that applies to a specific controlling

failure mode, and j is a material parameter. This material parameter does influence

damage progression. If j < 1, the rate of degradation is greatest at the beginning,

but if j > 1, the rate of degradation increases as a function of time. However, if

j = 1, there is no explicit time dependence in the rate of degradation. The failure

criterion is given by Fr = Fa . This approach has been used successfully for more

than 20 years by Reifsnider et al. [28–30] to solve various applied problems where

time, temperature, and cyclic loading are explicit influences.

Composite Pipes

6.3.1 Damage modeling

The application of damage mechanics enables one to model the matrix degra-

dation resulting from cracking in filament-wound pipes [31]. This damage is quan-

tified in terms of crack density, i.e., the reciprocal of crack spacing. The model

used by Roberts et al. [32], based on the previous work of Zang and Gudmundson

[33], applies to microcracks with crack surfaces parallel to the fiber direction and

perpendicular to the lamina plane. In the end, the stiffness matrix of each ply is

obtained by subtracting the damage tensor from the ply stiffness as:

Short and Long-Term Static Failure of Composite Pipes 207

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)

with k2 = Click here to view

E1 E2

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)

the failure parameter becomes dependent on time, static or cyclic loading and

temperature:

C = Acreep ATemperature Afatigue , (6.30)

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

ament-wound pipes, does not change from short to long term. Based on this as-

sumption and using the Paris law, Frost [31] showed that the creep factor may be

given as:

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)

LIVE GRAPH

Click here to view

212 Creep Design of Piping Applications Using Composite Materials

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.

wound glass fiber-reinforced pipes that were subjected to prolonged exposure to

high- temperature water. Two materials systems in common use in offshore piping

applications were used, ±55º filament-wound E-glass/phenolic and ±55º filament-

wound E-glass/epoxy pipes. The loading modes were hydrostatic pressure with

closed and open ends and axial loading. The last mode was produced by applying

the same pressure to the interior and exterior of the pipe to eliminate the hoop

stress.

Biaxial failure envelopes were obtained for both types of pipe materials in

the water-saturated state at a range of temperatures from 20º to 160ºC. It was ob-

served that the strength of the epoxy pipes was significantly reduced as tempera-

ture increases, especially in the matrix-dominated modes close to pure hoop and

pure axial loading. In contrast, the phenolic pipes were unaffected by temperature

for all tested modes.

Concerning the application of failure theories to this type of filament-wound

pipe, we must note, as a consequence of the wwfe (world-wide-failure-exercise)

[39], it is recommended that special care be taken wherever large deformations

may be involved. This is especially relevant for ±55º laminates, since none of the

failure theories coped very well with these cases.

This section offers an overview of long-term creep failure data for internally

pressurized filament-wound pipes. Unfortunately, the available experimental data

on creep and creep failure of composite pipes is quite limited.

Short and Long-Term Static Failure of Composite Pipes 213

wound glass-reinforced epoxy and polyester pipes. It was shown how important

the resin is for the long-term behavior of the pipes. Creep occurred at all stress

levels down to 40 MPa. At higher stresses, creep leads to weeping. In Figure 6.8,

long-term weeping stresses are plotted.

The failure mechanism for weepage in filament-wound pipes was described in

a great detail by Jones and Hull [41] for two distinct modes, hydrostatic pressure

with open and closed ends. The interlaminar cracks associated with transverse

matrix cracking tend to grow and intersect, providing paths through the pipe wall

for weeping to occur.

Since creep loading at higher stress levels leads to weeping and eventually to

total failure, it must be concluded that creep does promote damage comparable to

the short-term cracking that was observed for the UEWS (Ultimate Elastic Wall

Stress) tests [40] as depicted in Figure 6.9.

Nevertheless, the results reported by Mieras [40] on hoop stress/hoop strain

curves were not observed by others, especially by Hull et al. [42]. For mode 2 (hy-

drostatic pressure with closed ends) the sharp bend in the hoop stress/hoop strain

curve was not observed. Hull et al. [42] advanced an explanation, suggesting the

phenomenon originated from the step-loading used by Mieras [40].

The UEWS test is used to identify a stress level beyond which permanent de-

formations are obtained and the material creep rate increases substantially. The

UEWS test applies increasing pressure levels based on groups of 10 one-minute

hydrostatic pressure cycles [40,43–45]. The procedure consists of performing a

series of loading cycles in successive increasingly higher loading steps (10 equal

cycles at each level), while measuring the maximum hoop or axial strain obtained

\ [

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

Pressure (MPa)

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

the standard EN1447, which is accomplished by imposing hydrostatic internal

pressure on the specimens using axial constrained-free-end sealing devices, was

described by Faria and Guedes [47–48]. Figure 6.12 depicts the testing apparatus.

GFRP pipes of three different construction types, from major European manu-

facturers, were used. The properties of the series of specimens used are displayed

in Table 6.1.

Figures 6.13 to 6.16 depict the creep failure curves for each pipe type. It is clear

that the pipes’ construction and material composition have a strong influence on their

long-term behavior, especially when comparing a matrix of epoxy resin and polyester

resin (UP). Despite considerable scatter in the tests results, which is usual, statistical

regression analysis determined no relevant differences between using complete data

(including tests up to 10,000 h) or using only data from shorter tests (up to 1000 h).

These long-term hydrostatic tests will be used later in this chapter in a discussion

of pipe qualification.

Short and Long-Term Static Failure of Composite Pipes 217

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

1RPLQDO6WLIIQHVV1P

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)

LIVE GRAPH

Click here to view

218 Creep Design of Piping Applications Using Composite Materials

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)

The netting analysis is useful for preliminary calculation of filament-wound

pipes under hydrostatic pressure. These structures are primarily loaded in mem-

brane, and that may be considered the simplest case for design. The netting analy-

sis assumes that the stresses induced in the structure are borne entirely by the fi-

bers, and the strength of the resin is neglected. Moreover, it assumes that the fibers

possess no bending or shearing stiffness, and carry only the axial tensile loads.

The netting analysis is described in detail in [49–50]. Evans and Gibson [51]

obtained analytical expressions for the stress–strain relations from classical

Short and Long-Term Static Failure of Composite Pipes 219

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

The netting analysis for this condition, with an optimal winding angle of

arctan 2 ≅ 54.74º , assumes that the stress normal to the fibers is null and the

stress in the fibers direction is calculated as [53]:

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

Loading

Polymers and polymer-based composites fail, given enough time, when sub-

mitted to cyclic loads at stresses well below their static failure loads. This phe-

nomenon is called fatigue. The typical approach to fatigue is to develop fatigue

curves, i.e., applied stress (S) against the number of cycles until failure (N). These

curves are usually designated as S-N curves and resemble a sigmoid function,

displaying a stress limit for large cycles, suggesting a fatigue limit. Moreover the

S-N curves for polymers and polymer based composites are extremely frequency

dependent.

As many research works have shown, cyclic loading can significantly degrade

the stiffness and strength of pipes. Frost [54] and Frost and Cervenka [43] ob-

tained the S-N (fatigue) curves of ±55º glass-fiber/epoxy filament wound pipes

with 100mm internal diameter (14 MPa rated pressure). The curves were fitted

using the power law expressed as:

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

Creep (Miner's Law)

0.4

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.

Click here to view

6.5.1 Identification and comparison of main standards

Among the increasing range of applications where glass-fiber-reinforced plas-

tic (GRP) pipes are used, oil/gas and sewage transportation are the most relevant.

For the former, internal pressure is considered to be the representative and gov-

erning loading condition. Therefore, the product and test standards specify de-

sign and qualification criteria based on this loading case. Sewage pipe are treated

mostly as a low-pressure application, and thus reduced qualification criteria may

be adopted. Also, whether the pipe is above‑ or underground determines the type

of external supports and the conditions to which the pipes are subjected during

their in‑service life time. This also can lead to changes in design and qualification

procedures.

Despite the fact that highly specific requirements may be needed for each in-

dividual piping application, the major standardization organizations worldwide

have developed product and test standards that strive to cover the most com-

mon applications. Namely, ASTM International (formerly known as the American

Society for Testing and Materials), AWWA—American Water Works Association,

CEN—the European Committee for Standardization and ISO—International

Organization for Standardization have all published multiple standards compris-

ing guidelines for both manufacturers and end-users of GRP piping systems.

During the last 25 years, such normative references have evolved from the initial,

empirically-based conservative approaches to a combination of experimental and

statistical analyses, which nonetheless are also empirical. Hence, these standards

226 Creep Design of Piping Applications Using Composite Materials

and operation of GRP piping systems (i.e., pipes, fittings and related accessories).

In Table 6.2, a comparison of the required/preferred procedures of the main

GRP pipe product standards is presented. The standards for the product-types

mentioned below [54–59] are compared:

Reinforced Thermosetting-Resin) Sewer Pipe.

• ASTM D3517—Standard Specification for “Fiberglass” (Glass-Fiber-

Reinforced Thermosetting-Resin) Pressure Pipe.

• ASTM D3754—Standard Specification for “Fiberglass” (Glass-Fiber-

Reinforced Thermosetting-Resin) Sewer and Industrial Pressure Pipe.

• ASTM D2997—Standard Specification for Centrifugally Cast “Fiberglass”

(Glass-Fiber-Reinforced Thermosetting-Resin) Pipe.

• AWWA C950—Standard for Fiberglass Pressure Pipe.

• AWWA M45—Fiberglass Pipe Design Manual.

• EN 1796—Plastics piping systems for water supply with or without pres-

sure—Glass‑reinforced thermosetting plastics (GRP) based on unsaturated

polyester resin (UP).

• EN 14364—Plastics piping systems for drainage and sewerage with or with-

out pressure—Glass‑reinforced thermosetting plastics (GRP) based on un-

saturated polyester resin (UP)—Specifications for pipes, fittings and joints.

• ISO 10467—Plastics piping systems for pressure and non‑pressure drain-

age and sewerage—Glass‑reinforced thermosetting plastics (GRP) systems

based on unsaturated polyester (UP) resin,

• ISO 10639—Plastics piping systems for pressure and non‑pressure water

supply—Glass‑reinforced thermosetting plastics (GRP) systems based on

unsaturated polyester (UP) resin.

• ISO 14692—Petroleum and natural gas industries—Glass-reinforced plas-

tics (GRP) piping.

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:

(approx. 175,400 hours), instead of 50 years (approx. 438,000 hours);

• the full regression qualification procedure requires only 18 specimens to be

tested, instead of 23 specimens;

• it provides default values for the slope of the regression line for preliminary

design calculations and/or determination of test conditions (in cases where

there is no data);

• it allows a breakdown of a product family into family representatives, prod-

uct sectors, component variants and product sector representatives in order

to reduce the overall quantity and duration of the qualification tests; in this

way, only the family representatives are required to pass through the full

qualification procedure while the other variants have to pass only a 1000 h

survival test (ASTM D1598);

• it establishes a normative procedure to determine the factored failure enve-

lope, considering a biaxial (hoop plus axial) stress state (Figure 6.25);

• the reference test temperature is 65°C, instead of room temperature;

• it establishes a limited (less stringent) qualification procedure for low‑pres-

sure water applications.

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):

under cyclic internal pressures at a cycle rate of 25 cycles per minute, at

several different pressures. The stress or pressure values for the test are

selected to obtain a proper distribution of failure points in the time decades

(in log scale). The cyclic long‑term hydrostatic strength (LTHS) of a pipe

is obtained by an extrapolation of a log-log plot of the linear regression line

for hoop stress versus cycles to failure.

• Procedure B—a minimum of 18 specimens of pipe (or fittings) are placed

under constant internal pressures at different pressure levels in a controlled

temperature environment. The time‑to‑failure for each pressure level is

recorded. The stress or pressure values for the test are selected to obtain a

proper distribution of failure points along the time decades (in log scale).

time objective (50 years instead of 20 years) and thereby requiring a greater num-

ber of specimens so that the extrapolation can be done with the same statistical

confidence level. Both standards methods implicitly assume that the mechanisms

230 Creep Design of Piping Applications Using Composite Materials

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.

Two typical in-service load conditions for GRP piping systems are internal pres-

sure and ring deflection. For these, empirical test methods have been developed

and are described in the standards ISO 7509:2000 (equivalent to EN 1447:2009)

and ISO 10471:2003 (replacing EN 1227:1997), respectively.

The lack of completely understood failure mechanisms and long‑term materi-

als performance knowledge necessarily leads to conservative over‑estimation and

consequently to over‑design of the pipe structures. The existence of various types

of GRP pipe construction, namely filament wound, centrifugal cast, and hybrid

ones, also inhibits any relaxation of test specifications.

Experimental results from standard test procedures, according to EN 1447:2009

and ISO 10928:2009 (which replaced EN 705:1994), conducted on GRP pipes of

four different types and respective data analysis are presented below.

The test procedure specified in EN1447 relates to the creep behavior of GRP

pipes and requires constant internal pressure loading conditions to be imposed at

different levels and during different periods of time on each specimen. The time

periods range from a few minutes up to 10,000 hours. Unlike other normative

procedures (such as ISO 14692), the test temperature is room temperature. The

long-term resistance to internal pressure is determined by imposing hydrostatic

internal pressure on the specimens using axial constrained-free end sealing devic-

es. Failure must occur within a determined valid failure zone (in the middle of the

specimen) in order to be validated. Time‑to‑failure is registered. In Figure 6.12

the testing apparatus used in the experimental tests for internal pressure property

analysis is depicted.

In this test series, GRP pipes of four different construction types, from four

main European manufacturers, were used. The properties of the series of speci-

mens used are indicated in Table 6.1, in section 6.3.

All tests showed considerable scatter in results, which is typical given the ma-

terial variability of these composite constructions. Different levels of admissible

loading and different trends in the results were also observed for GRP pipes that

varied in their construction.

Applicable Standards 231

that only data from campaigns A and B were suitable for extrapolation. However,

the experimental procedure showed the data to be applicable to all of the different

types, if a few more specimens had been tested in campaigns C and D. Graphic

representation of the data is shown in Figure 6.26, accomplished by plotting pres-

sure loading versus time‑to‑failure in log‑log scales. The regression trend lines

are also plotted.

From these results several conclusions may be drawn:

• 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.

In this section, a practical design case is addressed in accordance with ISO

14692:2002 [54]. In the example, material and experimental short‑term failure

test data are used to estimate the long‑term failure envelope of a specific GRP

pipe type. The nominal pressure rating of such a pipe system is then determined

by following the design procedure established in the referenced standard for a

design life of 20 years.

In order to correlate this practical example with real applications within the

oil & gas fields, as well as with the previous experience gathered from the ex-

perimental tests summarized in Figure 6.26, a glass‑fiber reinforced epoxy (GRE)

pipe system was selected. The properties of a filament‑wound GRE pipe with

two ±55º layers, characterized by Soden et al. [35], were considered. The type of

construction and geometry of the pipes are presented in Table 6.3. The mechanical

properties for each of the four single-angle laminae are summarized in Table 6.4.

Since no specific long‑term test data is available for the selected pipe speci-

mens, the test data obtained for the pipe specimens of type A in Figure 6.26 has

been used, in order to establish a reasonable long‑term hydrostatic pressure, pLTHP,

as a reference for the design procedure.

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:

proximately 34% during the 20 years of its useful life. The expected minimum

long‑term hydrostatic pressure, pLTHP , for the selected pipe is 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

which is related to the factored qualified pressure, pqf , by the following relation:

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

374 288 13:10 914 305 3:1 622 0 1:0

525 332 3:2 939 283 13:4 544 0 1:0

599 349 7:4 867 262 13:4 492 -27

723 365 2:1 921 263 7:2 256 -65

736 368 2:1 817 148 11:2

236 Creep Design of Piping Applications Using Composite Materials

follows:

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:

s fs = = = = 706.85 MPa (6.46)

2t 2t 2 ×1

2 s ab 2×5

f3 = 1 − = 1− ≈ 0.886 (6.47)

r f 2 s fs 0.185 × 0.67 × 706.85

as:

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

long‑term envelope multiplied by f 2 . In this particular case, the factored long‑term

design envelope would be equal to the non‑factored design envelope, since all the

partial factors ( A1 , A2 and A3 ) are equal to unity.

In Figure 6.27 the short‑term and long‑term envelopes are depicted, based on

the short‑term experimental data [36] and the factors just calculated above.

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

materials.” Rheol Leaflet 10:12–20, 1939.

[8] Hunter S. C. “Tentative Equations for the Propagation of Stress, Strain and

Temperature Fields in Viscoelastic Solids,” J Mech Phys Solids 9(1):39–51,

1961.

[9] Guedes R.M. “Mathematical analysis of energies for viscoelastic materials

and energy based failure criteria for creep loading.” Mechanics of Time-

Dependent Materials 8(2):169–192, 2004.

[10] Guedes RM. “Lifetime predictions of polymer matrix composites under

constant or monotonic load.” Composites Part A 37 (5): 703–715, 2006.

[11] Schapery R. A. “Theory of crack initiation and growth in viscoelastic me-

dia. 3. Analysis of continuous growth.” Internat. J. Fracture 11(4):549–

562, 1975.

[12] Christensen R. M. “Lifetime predictions for polymers and composites un-

der constant load.” J. Rheology 25:517–528, 1981.

[13] Song M.S., Hua G.X. and Hub L.J., “Prediction of long-term mechanical

behaviour and lifetime of polymeric materials.” Polymer Testing 17:311–

332, 1998.

[14] Raghavan J., Meshii M. “Creep Rupture of Polymer Composites.”

Composite Science and Technology 57:375–388, 1997.

[15] Zhurkov S.N. “Kinetic concept of the strength of solids.” Int J Fract Mech

1(4):311–323, 1965.

[16] Griffith W. I. Morris D.H. and Brinson H.F. “The accelerated character-

ization of viscoelastic composite materials.” Virginia Polytechnic Institute

Report VPI-E-80-15: Blacksburg, VA, 1980.

[17] Knauss W.G. “Delayed failure-the Griffith problem for linearly viscoelastic

materials.” Internat. J. Fracture 6:7–20, 1970.

[18] Kostrov B.V. and Nikitin L.V. “Some general problems of mechanics of

brittle fracture.” Arch. Mech. Stos. (English version) 22:749–776, 1970.

[19] Schapery R.A. “Theory of crack initiation and growth in viscoelastic media.

1. Theoretical development.” Int J Fract 11(1):141–59, 1975.

[20] Schapery R.A. “Theory of crack initiation and growth in viscoelastic media.

2. Approximate methods of analysis.” Int J Fract 11(3):369–88, 1975.

[21] Leon R., Weitsman YJ. “Time-to-failure of randomly reinforced glass

strand/urethane matrix composites: data, statistical analysis and theoretical

prediction.” Mechanics of Materials 33(3):127–137, 2001.

240 Creep Design of Piping Applications Using Composite Materials

[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

The influence of gas properties and pipeline structural parameters on flow effi-

ciency can be analyzed by using the flow equation. For a compressible fluid (e.g.,

natural gas) with density ρ and pressure P flowing with a mean velocity u through

an inclined pipe (Fig. 7.1) with cross-section area A, the flow is controlled by the

Bernoulli equation:

A u A

⋅ du + A ⋅ dP + sin a ⋅ dy + π Ddy ⋅τ = 0 (7.1)

gc υ u

where:

2

g c is a proportionality factor ( g c = 32.2 lbm × ft / lb f × sec )

a is the slope of the pipe (Fig. 7.1)

dy is an elementary length of the pipe

τ is the shear stress acting on the inner surface of the pipe.

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:

2

1 u

K .E. = ∫ du (7.5)

1

gc υ 2

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

2

dP

PR.E = ∫ (7.8)

1

υ

246 Flow Capacity of Composite Pipelines

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

T1 + T2

Tm = (7.14)

2

equation (7.13) yields

M P2 2 − P12

PR.E = (7.15)

Z m RTm 2

According to Kay’s rule [7], the compressibility factor Z m can be obtained by

using the values of the average pseudo-critical temperature Tc' and the average

pseudo-critical pressure Pc' given by the following equations:

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

the parameters Pm, Zm, Tm derived in the previous chapter.

Furthermore, since

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

According to Table 7.1 and eq. (7.4), the term of energy losses due to friction

can be written as follows:

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

With the aid of equations (7.7), (7.15), (7.24), (7.26), the integration of equa-

tion (7.3) yields:

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

kinetic and potential energy compared to the other terms is insignificant and can

be neglected. According to this assumption, equation (7.27) can be simplified,

yielding:

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:

pipeline to convey gas. Pipes with a smooth interior surface and small val-

ues of f can considerably increase the gas flow rate.

2. Since the exponent of D is 5/2, an increase of diameter by a factor 2 yields

an increase in the gas flow rate of 5.6 times.

3. The dependence of the gas flow rate on the parameter 1 Tm means that

lower temperature values will increase the flow capacity of the pipeline.

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

From eq. (7.39) it can be concluded that low values of surface roughness yield

low values of e, resulting in a higher flow capacity of the transmission line.

Since the interior surface of pipelines made from composite materials is

smoother and less corroded than the surface of steel pipes, their roughness e has

lower values (Table 7.3), which has advantages.

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.

7.2.1 Flow capacity for laminar liquid flow

For a pipe of length L (Fig. 7.1) conveying liquid, the Bernoulli equation yields:

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)

Therefore, the flow capacity of a pipe conveying liquid in laminar flow condi-

tions can be estimated by the following equation:

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%.

When Re > 2000 the liquid flow is turbulent. In that case, the friction factor f

of the eq. (7.44) cannot be obtained by eq. (7.45) because it depends on interior

surface roughness e. The dependence of f on e for turbulent flow of liquids can be

given by the Nikuradse equation:

254 Flow Capacity of Composite Pipelines

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.

Multiphase Flow 255

Pipelines often carry one or more fluids in both a gas and liquid phase (e.g.,

gas-condensate or gas-oil-water flow). These kinds of flow are called “multiphase

flows” and are controlled by the densities and viscosities of the liquid and gas

constituents as well as their velocities, volume fractions, and interfacial shear

stresses. Because of density differences among the fluids, the distribution of the

different phases within the pipeline creates different flow regimes, which are gov-

erned by different hydrodynamic equations. One of the most important data for

the mechanical design of a pipeline is pressure drop. However, in multiphase

flow, different flow regimes may induce different pressure drops. Therefore, for

successful pipeline design the flow regimes must be predicted and the relevant

flow model applied.

7.3.1.1 phenomenology of flow regimes

The classification of multiphase flow regimes is based on experimental ob-

servations. From a literature search [e.g. 3, 4], the following cases have been ac-

cepted as the most widely used definitions for multiphase 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

The elevation of the points along a pipeline often changes, causing alterations

in flow parameters (velocities, pressures etc.).The changes in flow parameters re-

sult in spatial variations in flow regimes. To predict the regime transitions, several

quantitative criteria have been published. These can assist the pipeline designer

in correctly modeling a multiphase flow. For gas-liquid flow the following are

widely used:

Criterion of Taitel and Duckler [5]:

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 is the surface liquid height (from the bottom of the pipe) at the area of

flat gas-liquid interface

Ag is the cross-sectional area occupied by gas at the area of flat gas-liquid

interface

Al is the cross-sectional area occupied by liquid at the area of flat gas-liquid

interface

D is the diameter of the pipe

θ is the slope of the pipe

ρl , ρ g are the densities of the liquid and gas respectively

g = 9.81 m / s 2 is the gravitational acceleration

Criterion of Barnea et al. [6]:

hl

≥ 0.35 (7.54)

D

Multiphase Flow 259

Criterion of Taitel and Duckler [5]

4vl ( ρl − ρ g ) g cos θ

ug ≥ (7.55)

Sc ρ g ul

where:

ul is the average liquid velocity

Sc ≈ 0.01 (sheltering coefficient [5])

Criterion of Taitel and Duckler [5]:

4 Ag g cos θ ρl − ρ g

ul ≥ (7.56)

Si fl ρl

where :

f l � is the liquid friction factor

Considering that all the flow parameters are independent of time, the pressure

drop in steady-state conditions may be estimated by the following models:

In a stratified flow regime, the multiphase flow of a gas–liquid fluid can be

considered as two separate flows. Taking into account the equilibrium conditions

of a fluid volume element (Fig. 7.5), the following model can be obtained:

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

can be evaluated by the following well-known equations:

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.

262 Flow Capacity of Composite Pipelines

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

gravity (dimensionless), μ is gas viscosity in cp, Qg is gas flow rate in f t 3 / day

and D is pipe diameter. In the case of a liquid-gas flow, instead of the pipe diam-

eter D, the corresponding hydraulic diameters should be used, which are given by:

Al

Dl = 4 (7.65)

sl

Ag

Dg = 4 (7.66)

sg + si

Taking into account: (a) the already mentioned equations for computing the fric-

tion factor for gas or liquid flow, and (b) eqs. (7.64)–(7.66), the formulas summarized

in Table 7.4 can be used for two-phase flow (gas-liquid) for the three flow regimes:

For the interface between the liquid and gas one can use:

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

1 1 14.8 Al

= 4 log10 = 1.14 + 2 log10

e( s + s )

Re > 3000

fg g i fl e ⋅ sl

Multiphase Flow 263

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).

7.3.1.3.1.1 Evaluation of Si

From the Pythagorean theorem one can write:

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

h − R

sg = π − arc cos l R (7.74)

si / 2

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)

with respect to liquid depth hl .Therefore, this equation should be solved numeri-

cally. The solution is more difficult for the case of partially turbulent flow, because

the friction factors (given in Table 7.4) must be derived iteratively.

As shown in Fig. 7.7, a slug flow is constituted by: (a) the region consisting

of liquid film at the bottom and the gas volume above it, and (b) the slug region

consisting of the slug body containing gas bubbles. During steady-state slug flow,

fluids picked up from the front equal the fluids being sloughed off from the back.

Therefore, the slug front velocity is higher than the mean value of the slug veloc-

ity. This slug front velocity, uT , is called “translational velocity.”

The film region, LF , together with the slug region, LS , is called the “slug unit”

[9]. Slug flow can be considered as a continuous (unit to unit) flow.

Considering both gas volume and liquid film within the segment LF as the

control volume, the equilibrium of pressure forces acting at the left and right

boundaries of the liquid film with the shear forces acting at the liquid–gas and

liquid–pipe interfaces yields:

∆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

Combining equations (7.82) and (7.83) leads to:

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).

Taking into account Fig. 7.3d, the equilibrium equation is applied to both the

liquid film and the gas core, which yields:

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

By combining equations (7.94)–(7.97), the following formula can be obtained:

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).

Assuming a uniform distribution of bubbles within the liquid phase, the dis-

persed bubble flow can be treated as a pseudo-single-flow case. Such a flow type

can be modeled by the theory of liquid transmission.

References 269

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

pipes, it can be determined that the parameters affecting the allowable loads, i.e.,

axial tensile force, external pressure, bending moment and torsional moment are:

(b) the strength of the material under tensile, compressive, and shear stress

conditions;

(c) the diameter of the pipe, D;

(d) the number of plies, NP, composing the pipe’s wall;

(e) the fiber orientation.

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.

In Chapter 10, one can find results of the allowable forces in axial tension,

external pressure, bending and torsion for multilayered filament-wound pipes

made from two widely used materials, E-glass/epoxy and S-glass/epoxy. The de-

rived diagrams correspond to pipes with the following dimensions: diameters,

D = 0.2–1.2m; number of plies, NP = 10–50; and fiber orientation θ = ±15°,

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

274 Optimization of Material Cost

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

Since composites are anisotropic materials, the conventional test methods for

characterizing homogeneous isotropic materials are not suitable. Therefore, com-

mittee D30 of ASTM and SACMA (Suppliers of Advanced Composite Materials

Association) have provided standards in order to compare, validate, and docu-

ment composite material properties. Since the field of composite materials’ char-

acterization is very wide, and therefore beyond the scope of the present book, only

a brief presentation of the main methods will be presented below. However, the

reader can find an excellent detailed overview in the refs. [1-3].

The DSC method is based on monitoring the difference in the heat flow be-

tween a sample and a reference material. From the results of this method, glass

transition and melting temperature, the degree of crystallinity for thermoplastics,

the residual heat of reaction for thermosets and the curing characteristics of ther-

mosets can be obtained, in order to evaluate the material for quality, process op-

timization and curing simulations. The DSC method is important where matrix-

dominated failures are likely to occur [4].

279

280 Quality Control of Composite Pipe Systems

As a result of the non-homogeneity of composite materials, the different mate-

rial properties (e.g., thermal expansion coefficients, modulus of elasticity etc.) of

the constituents yield residual stresses, i.e., left-over internal stresses existing in

the absence of external loads. Residual stresses can also be generated during pro-

cessing (e.g., curing). Since residual stresses may reduce the strength of a material

or change the shape of a structural part, determining what they are is very impor-

tant. The main techniques for measuring residual stress are: hole drilling, X-ray

diffraction, neutron diffraction, the compliance method, Raman spectroscopy etc.

The drilling method is based on residual stress relaxation that results from

drilling itself. Changes in the strains around the hole, measured by a special strain

gage rosette, permit one to calculate the principal residual stresses.

The X-ray diffraction method is based on irradiation of the composite mate-

rial with high-energy X-rays. The radiation penetrates the surface of the crystal

planes and diffracts some of the X-rays. Using diffractometer techniques, mea-

surement of the changes in the inner planar spacing allows the derived elastic

strains calculation.

The neutron diffraction method (NDM) is similar to that of X-ray diffraction.

However, since the penetration depth of neutrons is greater, the NDM method is

advantageous because it can provide large quantities of data over the entire sur-

face and through the depth of the sample [3]. This method is applied mainly to

metal matrix composites, since it requires crystalline samples.

The compliance method focuses on measurement of deformations after a slot

is incrementally machined into the sample. From the obtained strain versus slot

depth results, the required values of residual stresses are calculated. The measure-

ment of strains in the compliance method is carried out by using a strain gage

rosette mounted on the specimen’s surface.

Raman spectroscopy is based on the shift to lower wave length numbers of the

Raman band of the spectrum as fibers undergo tensile deformation. Therefore, this

method can provide a map of the strain along the fibers.

Since FRPs are composed of fibers (which exhibit elastic behavior) and a

polymer matrix (which is a viscoelastic material), they exhibit a combination of

the behavior of their constituents. Therefore, the deformation of FRPs is time-

dependent even under constant loading conditions. This behavior is called creep.

During creep, damage accumulation takes place within the material’s structure

and can lead to rupture. In order to determine the creep behavior of composites,

two types of tests are usually used: (i) creep compliance tests that aim to measure

strain as a function of time for a constant stress, and (ii) creep rupture tests that

aim to measure time to failure. The duration of creep tests varies from a few hours

Test Methods and Material Characterization 281

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

Composite structures in service often suffer from impact damage (e.g., a

pipe struck by foreign objects during transportation, installation or operation).

Therefore, impact resistance is an important mechanical parameter. The two main

types of testing are: (a) low-velocity impact, and (b) ballistic impact. Since ballis-

tic impact is important mainly for assessing projectiles striking composite armors,

and thus beyond the scope of this book, we shall focus on low-velocity impact.

Impact tests mainly measure energy absorption resulting from a specimen’s frac-

ture and the resistance of a specimen versus time. The two standard low-velocity

impact tests are: a) drop weight impact testing, and b) Charpy impact testing. The

drop weight impact test is the most common for composite materials. The test

machine is equipped with a weight raised to a prescribed height and released to

strike the specimen, which is a composite plate clamped along its edges according

to ASTM D 5628.

A schematic representation of the values of energy and resistance versus time,

obtained by an impact test, is demonstrated in Fig. 9.3.

Once the curve of resistance R = R(t) is measured, the corresponding accelera-

tion (or retardation) γ(t) can be obtained by Newton’s law:

R(t ) = m γ (t ) + m g

(9.1)

(g = 9.81 m/sec2).

Therefore, the velocity U(t) can be obtained by the following integration:

U (t ) = ∫ γ (t )dt (9.2)

Test Methods and Material Characterization 283

Figure 9.3 Schematic representation of impact testing results: (a) absorbed energy versus

time, (b) specimen resistance versus time.

Using the definition of the velocity:

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)

Modern Charpy machines are equipped with an electronic impactor that is able

to record the specimen’s resistance R(t) versus time. Typical dimensions for a

fiber- reinforced polymer Charpy specimen are: 126 mm length, 12.7 width and

3.0-12.7 mm thickness. Typical fiber orientation angles for the specimens are: 0o,

10o, 22.5o, 30o, 45o, 67.5o and 90o. To obtain reliable results it is important to test

specimens with different fiber directions. Sometimes, post-impact analysis using

ultrasound or other non-destructive test methods is needed to fully investigate the

failure type.

Variable loading conditions cause damage accumulation within a composite ma-

terials’ microstructure. From a micromechanical view point, such damage accumu-

lation is the result of fiber fractures, fiber-matrix debonding, matrix cracking etc.

Test Methods and Material Characterization 285

strength and the remaining life of the structural part, a phenomenon referred to as

fatigue. To obtain design data concerning the fatigue behavior of composites and

to compare them under standard conditions, fatigue test methods have been devel-

oped and standardized. Existing standards are related to coupon fatigue testing (e.g.,

ASTM D 3479, ASTM D 6115, EN ISO 13003) or to component fatigue testing

(e.g., ISO 14269 regarding GFRP pipes for use in the offshore petroleum industry,

EN 12245 for fiber wrapped gas cylinders, etc.). The fatigue loading can be har-

monic (i.e., consisting of well-distinguished loading cycles) or irregular (Fig. 5ab).

In the harmonic loading history shown in Fig. 9.5a, σa and σm are the values of

stress amplitude and mean stress respectively, while σmax and σmin are the maximum

and minimum stress levels. The relation of σα and σm to σmax and σmin is given by:

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.

International Standards for Composite Pipes 287

Figure 9.8 Derivation of an S-N curve from fatigue tests with differing stress amplitudes.

are used. These machines offer the ability to perform load- or strain-controlled

fatigue tests with harmonic, block or irregular loading histories. Since scatter in

the fatigue results is normal, statistical analysis is required to provide reliable data

for engineering design.

Due to the wide applications of composite pipes, their testing, design and in-

stallation are standardized. Currently, United States organizations covering the

composite pipes’ technology and applications are: ASTM (American Society for

Testing and Materials), ASME (American Society of Mechanical Engineers),

API (American Petroleum Institute) and AWWA (American Water Works

Association). In addition, international standards such as DIN (Deutsches Institut

fuer Normierung), BS (British Standards), AFNOR (L’ Association Francaise de

Normalisation), as well as Japanese standards JIS (Japanese Industrial Standards)

and International Standards ISO (International Standards Organization) are used

and offer important tools for developing composite pipeline networks for oil, gas

and water applications. Since the analysis of all standards is impossible, only a

sample of the main American, European, Japanese and International standards is

summarized in Table 9.1.

288 Quality Control of Composite Pipe Systems

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

ASTM D 2563 Standard Practice for Classifying Visual Defects in Glass Reinforced Plastic

Laminate Parts.

ASTM D 3839 Standard Practice for Underground Installation of Flexible Reinforced

Thermosetting Resin Pipe and Reinforced Plastic Mortar Pipe.

AWWA C 950 Appendix C-Installation.

API RP 15 L4 Recommended Practice for Care and Use of Reinforced Thermosetting

Resin Line Pipe.

API RP 15 A4 Recommended Practice for Care and Use of Reinforced Thermosetting

Resin Casing and Tubing.

ASTM D 638 Standard Test Methods for Tensile Properties of Plastics.

ASTM D 1599 Short-Term Hydraulic Failure Pressure of Plastic Pipe, Tubing and Fittings.

ASTM D 2290 Apparent Tensile Strength of Ring or Tubular Plastics and Reinforced

Plastic Pipe and Tube by Split Disk Method.

ASTM D 2105 Longitudinal Tensile Properties of Reinforced Thermosetting Plastic Pipe

and Tube.

ASTM D 695 Standard Test Methods for Compressive Properties of Rigid Plastics.

ASTM D 790 Standard Test Methods for Flexural Properties of Unreinforced and

Reinforced Plastics and Electrical Insulating Materials.

ASTM D Measuring Beam Deflection of Reinforced Thermosetting Plastic Pipe

12925 under Full Bore Flow.

ASTM D1598 Time to Failure of Plastic Pipe under Constant Internal Pressure.

ASTM D 2143 Cyclic Pressure Strength of Reinforced Thermosetting Plastic Pipe.

International Standards for Composite Pipes 289

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)

ASTM C 581 Standard Practice for Determining Chemical Resistance of Thermosetting

Resins used in Glass Reinforced Structures intended for Liquid Service.

ASTM D 3615 Chemical Resistance of Thermoset Molding compounds used in the

Manufacture of Molded Fittings.

ASTM D 3681 Chemical Resistance of Reinforced Thermosetting Resin Pipe in a

Deflected Condition.

AWWA C 950 Standard for Fiberglass Pressure Pipe for Water Service, 1–144 inches in

diameter (25–3600mm).

API 15 LR Specification for Low Pressure Fiberglass Line Pipe, 2-16 inches in diam-

eter, up to 1000 psi (cyclic)/25–300 mm, 70 bar.

API 15 HR Specification for High Pressure Fiberglass Line Pipe, 1-8 inches in diam-

eter, above 1000 psi (cyclic)/25–300 mm, 70 bar.

API 15 AR Specification for Fiberglass Tubing.

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)

T57 203 Glass fiber-reinforced plastic pipes. Dimensions.

T57 205 Glass fiber-reinforced plastic pipes. Test method for short-term resistance, under

pressure, to rupture.

T57-206 Glass fiber-reinforced plastic pipes. Glass epoxy resin pipes for the transport of

hot water under pressure. Characteristics and test methods.

T57-207 Glass fiber-reinforced plastic pipes. Collection of basic data for the dimensional

calculation of pipes and fittings under constant internal pressure. Test method.

T57-208 Fiber-reinforced plastic pipes. Design and dimensioning of cemented socket

assemblies.

T57-209 Fiber-reinforced plastic pipes. Underground installation of flexible pipelines

with or without pressure.

T57-213 Fiber-reinforced plastic pipes. Resistance determination under cyclic internal

test pressure. Test method.

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)

Part 6: Recommended practice for installation.

EN NF DIN 1119 Plastic piping systems. Joints for glass reinforced thermosetting plastics

(GRP) pipes and fittings. Test methods for watertightness and resistance

to damage of flexible and reduced articulation joints.

EN NF DIN 1120 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)

pipes and fittings. Determination of the resistance to chemical attack

from the inside of a section in a deflected condition.

EN NF DIN 1225 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)

pipes and fittings. Determination of the creep factor under wet conditions

and calculation of the long-term specific ring stiffness.

EN NF DIN 1226 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)

pipes. Test method to prove the resistance to initial ring deflection.

EN NF DIN 1227 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)

pipes. Determination of the long-term ultimate relative ring deflection

under wet conditions.

EN NF DIN 1228 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)

pipes. Determination of initial specific ring stiffness.

EN NF DIN 1229 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)

pipes. Test methods to prove the watertightness of the wall under shot-

term internal pressure.

International Standards for Composite Pipes 293

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)

ISO 8795 Plastic pipe systems for the conveyance of water intended for human con-

sumption. Migration assessment. Determination of migration values for

plastic pipes. (Note: valid for thermosetting and thermoplastic materials.)

ISO 10465 Underground installation of flexible glass reinforced thermosetting resin

(GRP) pipes.

Part 1: Installation.

Part 2: Comparison of static calculation methods.

Part 3: Installation parameters and application limits.

ISO 10466 Plastics piping systems. Glass reinforced thermosetting plastics (GRP)

pipes. Test methods to prove the resistance to initial ring deflection.

ISO DIS 10467 Plastic piping system for pressure and non-pressure drainage and sewer-

age. Glass reinforced thermosetting plastics (GRP) based on unsaturated

polyester (UP) resins.

ISO DIS 10468 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)

pipes. Determination of the long-term specific ring creep stiffness under

wet conditions and calculation of the wet creep factor.

ISO DIS 10471 Plastic piping systems. Glass reinforced thermosetting plastics (GRP)

pipes. Determination of the long-term ultimate bending strain and the

long-term ultimate relative ring deflection under wet conditions.

ISO DIS 10639 Plastic piping systems for water supply with or without pressure. Glass

reinforced thermosetting plastics (GRP) based on unsaturated polyester

(UP) resins.

ISO 10928 Plastic pipe systems. Glass reinforced thermosetting plastics (GRP) pipes

and fittings. Methods for regression analysis and their use.

ISO DIS 10952 Plastic pipe systems. Glass fiber-reinforced thermosetting plastics (GRP)

pipes and fittings. Determination of the resistance to chemical attack from

the inside of a section in a deflected condition.

ISO DIS 14828 Plastic pipe systems. Glass reinforced thermosetting plastics (GRP) pipes.

Determination of the long-term specific ring relaxation stiffness under wet

conditions and calculation of the wet relaxation factor.

Detection of Defects and Structural Health Monitoring 295

Monitoring

FRP composite pipelines usually do not exhibit visible fiber cracking, matrix

cracking, debonding or delamination prior to failure. Therefore, non-destructive

methods such as ultrasonography, infrared thermography, acoustic emission, elec-

tromagnetic infrared technique etc. have been applied to assess their structural

health. Acoustic emission is based on the detection (using piezoelectric sensors)

of transient elastic waves created by the rapid release of energy during loading-

induced damage accumulation. Acoustic emission can quantitatively detect the

location of defects; however, quantification of material damage is almost impos-

sible. Electromagnetic infrared thermography is predicated on the detection of

defects-induced local perturbations of an electromagnetic field, which are trans-

mitted through the composite material. However, the method is very expensive.

The most common method for detection of defects in composites is ultrasonog-

raphy. Due to the importance of this method, it will be explained in section 9.7.

Since pipe removal for testing entails downtime of the installation and very

high labor costs, in-situ structural health monitoring based on bonded or embed-

ded sensors within the composite material is now routinely used. Though numer-

ous methods, for example, electric impedance and electromagnetic response have

been developed, piezoelectric- and optical fiber-based systems are most common-

ly used for real-time monitoring of structural health.

Piezoelectric techniques utilize piezoelectric sensors made of ceramic or poly-

meric materials that produce electrical signals in response to strain vibrations. Strain

waves propagate through the wall of composite pipes and interact with the piezo-

electric device. Since the structural defects in the composite material change its

stiffness and since the stiffness changes yield changes in the wave frequency and

amplitude (e.g., in Lamb waves) special software is used to treat the changes in the

waveforms and thereby provide information regarding the existence of damage, its

location, type and severity. Piezoelectric sensors are low-cost devices and do not

need an external power source for their operation. They can either be bonded to, or

embedded within, the wall of a composite pipe and thus provide real-time monitor-

ing of the materials’ damage accumulation. Piezoelectrics are well suited for local-

ized damage detection, except in pipes conveying high-temperature fluids.

Optical fiber-based techniques were first used in 1979 by Langley Research

Center of NASA in order to monitor strain in composite panels operating at low

temperature. These techniques are based on the fact that when a beam of light

296 Quality Control of Composite Pipe Systems

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.

Ultrasonic testing is the most common non-destructive method for flaw detec-

tion, crack evaluation, material characterization, dimensional measurements etc.

It is based on the defect-induced reflection of an ultrasound wave, which is trans-

formed into an electrical signal that provides information concerning defect loca-

tion, orientation and size. The physical properties used for the description of how

the signal propagates through the anisotropic material are wave frequency and

acoustic impedance Z = ρc, where c is the acoustic velocity and ρ is the density of

the material. Two ultrasonic testing configurations can be used: pulse-echo (PE)

and through-transmission (TT). In the PE configuration the transducer for trans-

mission and reception of the signal is placed on the external side of the pipeline,

while in the TT configuration the transducer producing the signal is placed on the

one side and the transducer receiving the signal is located on the opposite side of

the composite wall. Since the TT configuration is impractical for testing pipelines,

only the PE technique is usually used for quality control.

Due to the high impedance mismatch of air, a liquid coupling medium, such as

water or an oil-based gel is needed between the transducer and the surface of the

composite material to be tested. However, because the use of liquid couplers is

impractical, recent advanced air-coupled ultrasonic transducers have been devel-

oped for pipeline testing. These highly sensitive transducers can be piezoelectric

or MEM (Micro-Electro-Mechanical). MEM transducers provide better acoustic

impedance than piezoelectric ones. Frequency and bandwidth are the main pa-

rameters to be considered for transducer data, to achieve the optimum sensitivity

and resolution of the system.

References 297

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.

10.1.1 Results of failure model for axial tension

Taking into account the derived model given in Chapter 3, the allowable ten-

∧

sile force Ν has been estimated for pipes made from the materials: (a) E-glass/

epoxy and (b) S-glass/epoxy. The following diagrams illustrate the allowable val-

∧

ues Ν for pipes of diameters Dia = 0.1–1.2 m constituted by plies of thickness

0.150 mm, fiber orientation θ = ±15°, ±30°, ±45°, ±60°, ±75° and with the number

of plies NP = 10, 20, 30, 40, 50.

299

300 Case Studies

force

Axial Tension 301

302 Case Studies

Axial Tension 303

304 Case Studies

Axial Tension 305

force

306 Case Studies

Axial Tension 307

308 Case Studies

Axial Tension 309

310 Case Studies

10.2.1 Results of failure model for pure bending

Taking into account the model described in Chapter 3, the allowable values

of bending moment M have been estimated for pipes made from the materials:

(a) E-glass/epoxy, and (b) S-glass/epoxy. The calculations were carried out by a

computer code titled “BENDING,” which was developed with the standard soft-

ware “Mathematica” (see enclosed CD-ROM). By using several values of M (last

command), the program can calculate the value of the Tsai-Wu expression. The

value of M that yields unit value for the Tsai-Wu expression should be adopted.

The diagrams below, derived from the software, present the allowable values M

for pipes of diameters Dia = 0.10 m, ,1.2 m consisting of plies with a thickness

0.150 mm and fiber orientation θ = ±15°, ±30°, …, ±75° for the number of plies

NP = 10, , 50.

bending

Pure Bending 311

312 Case Studies

Pure Bending 313

bending

314 Case Studies

Pure Bending 315

316 Case Studies

10.2.2.1 buckling model for multilayered filament-wound E-glass/epoxy pipes

subjected to pure bending

Pure Bending 317

Next Page

Previous Page

320 Case Studies

Pure Bending 321

subjected to pure bending

322 Case Studies

Pure Bending 323

324 Case Studies

Pure Bending 325

326 Case Studies

10.3.1 Results of failure model for external pressure

Taking into account the model for external pressure exhibited in Chapter 3,

the allowable external pressure Pa has been estimated for pipes made from the

materials: (a) E-glass/epoxy, (b) S-glass/epoxy. The following diagrams present

the allowable values Pa 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

number of plies NP = 10–50.

external pressure

External Pressure 327

328 Case Studies

External Pressure 329

330 Case Studies

External Pressure 331

external pressure

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

Using the buckling model for external pressure presented in Chapter 3, the

critical pressure pcr has been estimated for pipes made of: (a) E-glass/epoxy and

(b) S-glass/epoxy.

In the following curves, pcr is demonstrated for pipes of diameters Dia = 0.10

–1.20 m and made from plies with a thickness 0.150 mm, fiber orientation θ = ±30°,

±45°, ±60°, ±75° and number of plies NP = 10, 20, 30, 40, 50.

subjected to external pressure

External Pressure 335

336 Case Studies

External Pressure 337

subjected to external pressure

338 Case Studies

Next Page

Previous Page

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.

torsion

Torsion 341

342 Case Studies

Torsion 343

torsion

344 Case Studies

Butt Joints of Multilayered Filament-Wound Pipes 345

Pipes

Allowable axial tensions and bending moments are shown below for E-glass/

epoxy and S-glass epoxy materials with fiber orientation θ = ±45° and different

numbers of layers and ratios L/D (i.e. joint’s length L over its diameter D).

E-glass epoxy material

llowable axial tension

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

10.6.1 E-glass/epoxy material

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

10.7.1 E-glass/epoxy material

NP = 50

NP = 40

L (m)

NP = 30

NP = 20

NP = 10

NP = 50

NP = 40

NP = 30

NP = 20

NP = 10

370 Case Studies

NP = 50

NP = 40

NP = 30

NP = 20

NP = 10

NP = 50

NP = 40

NP = 30

NP = 20

NP = 10

Spaces Between Supports 371

NP = 50

NP = 40

NP = 30

NP = 50

NP = 40

NP = 30

NP = 20

NP = 10

372 Case Studies

NP = 50

NP = 40

NP = 30

NP = 20

NP = 10

NP = 50

NP = 40

NP = 30

NP = 20

NP = 10

Spaces Between Supports 373

NP = 50

NP = 40

NP = 30

NP = 20

NP = 10

NP = 50

NP = 40

NP = 30

NP = 20

NP = 10

374 Case Studies

Pipelines vs. the Vertical Load F

10.8.1 E-glass-epoxy materials

Installation Depth for Underground Pipelines vs. the Vertical Load F 375

376 Case Studies

θ =π /6 F = 75 000 (N)

Installation Depth for Underground Pipelines vs. the Vertical Load F 377

378 Case Studies

θ =π /4 F = 75 000 (N)

Installation Depth for Underground Pipelines vs. the Vertical Load F 379

380 Case Studies

θ =π /3 F = 75 000 (N)

Installation Depth for Underground Pipelines vs. the Vertical Load F 381

382 Case Studies

Installation Depth for Underground Pipelines vs. the Vertical Load F 383

384 Case Studies

Installation Depth for Underground Pipelines vs. the Vertical Load F 385

386 Case Studies

θ = π /6 F = 75 000 (N)

Installation Depth for Underground Pipelines vs. the Vertical Load F 387

388 Case Studies

θ = π /4 F = 75 000 (N)

Installation Depth for Underground Pipelines vs. the Vertical Load F 389

390 Case Studies

θ = π /3 F = 75 000 (N)

Installation Depth for Underground Pipelines vs. the Vertical Load F 391

392 Case Studies

Installation Depth for Underground Pipelines vs. the Vertical Load F 393

INDEX

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

bending 55 56 57

81 91 111

Bernoulli equation 243 244 252

Bisphenol fumerate resin 46

Buckling 57 73 80

82 92 100

bulk modulus of elasticity 150 162

buoyancy forces 135

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

130 137 147

carbon fibers 45 49

celerity 153 162

centrifugal force 122 123

C-Glass 48

Charpy machine 284

chlorenic resin 46

Chlorenic 47

circular frequency 119

classical lamination theory 24

compatibility equations 68 143

compliance coefficients 42

compliance matrix 6 36 42

185 194

Compliance method 280

compressibility factor 246 248 249

Compressive failure stress 40

compressive fluid 243

compressive load 84

connections 169

continental pipelines 53

coordinate system 13

coordinate transformation 13

coriolis force 123

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

Creep compliance test 280

Creep rapture test 280

critical temperature 246

critical velocity 117 129 146

147 148

Cross-Ply laminates 36 38

Curing 45

cyclic loading 220

damping 123

dashpot 119

deflected spring 119

deflection 143

Differential Scanning Calorimetry 279

Dirac delta function 159

dispersed bubble flow 255 256 268

divergence 123 147

Drilling method 280

drill-string 138

dynamic loads 56

dynamic model 155

dynamic stability 105

dynamic stresses 155

Index Terms Links

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

Index Terms Links

130 131 132

148

flexural force 122

floating drill-bit 138

flow capacity 243 252 253

flow equation 243

fluid hammer 6

fluid viscosity 251

flutter 123

Fourier transform 159

friction factor 244 250 251

252 253

Furane resins 46

Furane 47

gas constant 246

gas flow rate 249 250

gas gravity 250

gas specific volume 243

gas transmission 243

Glass fibers 48

global coordinate system 13

global transfer matrix 140

graphite 45

Index Terms Links

helical pattern 51

hole drilling 280

Hooke’s law 1 2 9

10 11 12

172 173

hydraulic hammer 149 150 154

155

hydrodynamic forces 56

hydrostatic design basis 220

impulse-momentum equation 149

inertia effects 123

Infrared thermography 295

Instability 122 123 124

installation loads 53

integral transform 159

International standards 287

J-Lay 53 56

joining methods 169

Index Terms Links

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

Index Terms Links

matrix directions 1

Matrix material 45

Maxwell-Betti Reciprocal Theorem 5

mechanical design 53

mechanical strains 6 7 9

13

metallic collars 135 136 138

modulus of elasticity 1 58

moisture expansion coefficients 9 19

moisture strains 8 9 12

19 23 56

57

molecular weight 246

Moody’s diagram 254

multi-layered pipe 67

multiphase flow 255

Neutron diffraction 280

Newton’s equation 149 151

Nikuradse equation 251 253

Nylon 48

Index Terms Links

operation loads 53 56

Optical fiber-based techniques 295

Optimization 271

optimum fiber orientation 272 273 276

277

Orthophlatic polyester 47

Overpressure 154

payout head 51

periodically supported pipe 143

Phenolic resin 46

Phenolic 47

Piezoelectric techniques 295

pinned-pinned pipe 118 122 123

130 131 132

148

plane stress 10 11 14

Poisson’s ratio 1 58

polar pattern 51

polyacrylonitril 49

Poly-Amide Imide 48

Poly-Benzimidazoles (PBls) 47

Poly-Carbonate (PC) 48

Polyester resin 46

Poly-Ether Imide 48

Poly-Ether-Ether Ketone (PEEK) 47 48

This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links

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

Rayon 49

reduced compliances 14 15

reduced stiffnesses 11 15 16

residual stresses 280

Index Terms Links

262

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

Index Terms Links

tension 55 57 88

91 94

Terephlatic polyester 47

thermal analysis 279

thermal expansion coefficient 133

thermal expansion coefficients 7 12 19

190

thermal load 132 147

thermal strains 6 7 9

12 19 23

thermal stresses 56 57

Thermoplastic resin 45 47 48

Thermosetting resin 45 46

Torsion 56 57 97

torsion spring 119

total strains 6

Towing 53

Transfer Matrices Method 124 135 148

Transfer Matrix 127 139

transformation matrix 14

transformation of engineering properties 16

translational velocity 265

Tsai-Wu Failure Criterion 39 40 41

turbulent flow 250 251 262

twisting curvature 30

Index Terms Links

Ultrasonography 295

underground pipe 56

underground pipelines 192

velocity 115

vibration 56 105 155

Vinyl ester resin 46

Vinyl ester 47

winding angle 51

Winkler-type foundation 141

- CMS-16-QBUploaded byGt Varadharajan
- CompositesUploaded byJohana Abarca Villalobos
- Benjamin_J_Phillips_Thesis_2009.pdfUploaded byAnonymous hprsT3WlP
- CompositesUploaded byسینا ایرانی
- Design of Anis Grid Composite Lattice Conical Shell StructuresUploaded bySEP-Publisher
- Carbon-Fiber-reinforced Polymer - Wikipedia, The Free EncyclopediaUploaded bysk211me1158
- Experimental Determination of Dynamic Young's Modulus and Damping of an Aramid-fabricpolyester Composite MaterialUploaded byAze
- SpeechUploaded byjaysonly90
- Construction Material 17209 Winter 14Uploaded bykiranpawar85
- Cfrp Wrap Cfw-600Uploaded byghazanfar
- noor1996Uploaded byGajaraj Gajapathi
- congresso_sampe_sigmatexUploaded byAlex
- zaikako3Uploaded bygopvij1
- Composites-1.pdfUploaded byPrithviDas
- Serp 13 - Composite Prepregs 09-12-13Uploaded byOoj Gen
- AD061 Axial load capacity of a dented tubular member.pdfUploaded byalvin_yee_17
- ebook_11600 (1)Uploaded byCésar Santana
- Sn003a - Elastic Critical Moment for LTBUploaded byDarren Thoon
- A13 Materials Selection in Design (1)Uploaded byAndres Carmona
- BOF39 - Shave - FRP Composites in Bridge DesignUploaded bydcqscribd
- Ada 221555Uploaded bymehdiq
- Expt 10.BucklingUploaded byVenkatesh Thota
- Drilling of CFRP by EDM (1)Uploaded bythasarathanr1993_939
- Brezier EffectUploaded byAscanio Barbosa
- Composolite - GEF IncUploaded byyohanes ong
- part of racket.docxUploaded byNurul Shahida Mohd Azilah
- ch11Uploaded byManjunatha Eikila
- seminar08_PlateBucklingUploaded byAndrei Cioroianu
- 1DA15ME002-1.docxUploaded byRahul Darshan Srt
- Pipe StressesUploaded bymark ignacio

- TrainUploaded by786908
- Custom Web AdiUploaded byarundvedi
- 8.Importanta Utilizarii Metodelor Interactive in Formarea La StudentiUploaded bycarmen
- CCH1500-5.pdfUploaded byIebe Seventynine
- 7 35Uploaded bypraveen
- R-intro.pdfUploaded bymaillatu
- 207115 SCMUploaded byMajamar Iguavita Guevara
- 3Ǥ Gothic Elements in Mary ShelleyUploaded byDipjyoti Bharali
- progression ii final essayUploaded byapi-340390725
- Power Cable Data SheetUploaded byFaty Bercasio
- ReadmeUploaded byaklreaxmer
- fpostUploaded byamdocm
- 8th_Biology_English.pdfUploaded byVadlakonda Sampathgoud
- ANALYSIS OF ELONGATION OF LASHING STRAPS.pdfUploaded byoopy Pp
- DX3340Uploaded bynaughtybigboy
- laplacian.pdfUploaded bygeo_bi
- Merits of c4 _coated Coconut Cover Crush_ Block Over Aggregate BlockUploaded byIAEME Publication
- Optimizer WSUploaded byvirgo_2k
- PRV2014-12 CEP Chemical-Engineering-Progress Dec-2014 - Choosing a Fine-Particle Filtration System US (1)Uploaded byreyorozcof
- Title Proposal Revised2Uploaded byRizza M. Mendoza
- meme rubricUploaded byapi-235058603
- Nozzle LoadsUploaded byKyaw Kyaw Aung
- Assignment 1Uploaded byChalermchai New Kawin
- Structured vs Unstructured DataUploaded byaakkii_chims
- mlk lesson plansUploaded byapi-348537046
- Awesome DbmsUploaded byrajesh kumar
- Simplified GSM Link BudgetUploaded bydoc_books
- german programUploaded byIshita Katyal