Experimental Study and Theoretical Modelling of Pipeline ... · Nomenclature T - Temperature ... V-groove preparatory joints as specified in (a) ... Heat loses and heat transfer in

i

Experimental Study and Theoretical Modelling of Pipeline Girth Welding

By

Krzysztof Borkowski

B. Eng.

A thesis submitted for the degree of Master of Philosophy at the

School of Mechanical Engineering

The University of Adelaide

Australia

Submitted: December 2014

ii

iii

Abstract

The thermal field induced by arc welding has been the subject of numerous

experimental, analytical and numerical studies in the past. However, few studies have

focused on the effects of the local geometry and pipeline welding procedure on the transient

thermal field at or near the vicinity of the weldline. The local geometry and welding

procedures are often simplified in computational or analytical studies and normally

disregarded in quantitative assessments. The objective of this thesis is to evaluate the

significance of these effects in order to understand their possible influence on the weld

quality, pipeline integrity and weldability. In this thesis, simplified analytical models are

developed, compared against outcomes from previous investigations, and validated with data

obtained from a full-scale experimental study completed by the candidate. The conducted

research indicates that the effects of the weld preparatory geometry (which is within the

industry acceptable variations) and pipeline welding procedures might have a significant

impact on the thermal history, specifically at low heat inputs and no preheats, which are

characteristic for pipeline girth welding. Therefore, the account of these effects is very

important for the adequate evaluation of the weld quality and, potentially, the pipe integrity.

The results presented in this thesis can be utilised in the quality control, advanced modelling

procedures and other activities directed towards the further improvement of pipeline

construction procedures.

iv

v

Thesis Declaration

I certify that this work contains no material which has been accepted for the award of

any other degree or diploma in my name, in any university or other tertiary institution and, to

the best of my knowledge and belief, contains no material previously published or written by

another person, except where due reference has been made in the text. In addition, I certify

that no part of this work will, in the future, be used in a submission in my name, for any other

degree or diploma in any university or other tertiary institution without the prior approval of

the University of Adelaide and where applicable, any partner institution responsible for the

joint-award of this degree. I give consent to this copy of my thesis, when deposited in the

University Library, being made available for loan and photocopying, subject to the provisions

of the Copyright Act 1968. I also give permission for the digital version of my thesis to be

made available on the web, via the University’s digital research repository, the Library

Search and also through web search engines, unless permission has been granted by the

University to restrict access for a period of time.

Krzysztof Borkowski,

9th

December 2014

vi

vii

Acknowledgements

I would like to thank A/Prof. Kotousov and A/Prof. Ghomashchi for their support and

guidance with this research project. Many thanks to Pascal Symons and Scott Letton from the

workshop for their fabrication and welding expertise as well as Alison-Jane Hunter for her

help with the editing of this thesis.

This research project was funded by the Energy Pipeline CRC and supported through

the Australian Government Cooperative Research Centre Program. The cash and In-kind

support from the APIA-RSC is gratefully acknowledged. I would also like to thank EPCRC

CEO, Prof. Valerie Linton and our industry advisors, Frank Barbaro, Leigh Fletcher, Chris

Jones, John Piper and Cameron Dinnis.

viii

ix

Nomenclature

T - Temperature [°C]

V - Voltage [V]

I - Current [A]

η - Arc efficiency

Q - Power Input [W]

v - Weld travel speed [m s-1

]

h - Plate thickness [m]

λ - Thermal conductivity of steel [W m-1

K-1

]

κ - Thermal diffusivity of steel [m2

s-1

]

cp - Specific heat [J kg-1

°K-1

]

ρ - Density [kg m-3

]

b - Dimensionless heat transfer factor

T0 - Initial temperature [°C]

T∞ - Ambient temperature [°C]

Tph - Preheat temperature [°C]

𝑥 and 𝑦 - Rectangular coordinates [m]

r - Radial coordinate in polar coordinate system, r = √x2 + y2 [m]

w - Coordinate along the weld direction, w = x − vt [m]

ξ - Moving radial coordinate in polar coordinate system, ξ = √w2 + y2 [m]

rz - Radial coordinate in cylindrical coordinate system, rz = √w2 + y2 + z2 [m]

rn - Function, rn = √w2 + y2 + (2nh − z)2 [m]

rn′ - Function, rn

′ = √w2 + y2 + (2nh + z)2 [m]

Uw - Heat transfer coefficient of weld surface [W m-2

K-1

]

x

Up - Heat transfer coefficient of plate surface [W m-2

K-1

]

d - Heat transfer/conductivity coefficient [m-1

]

dw - Heat transfer/conductivity coefficient of weld, dw = Uw λ⁄ [m-1

]

dp - Heat transfer/conductivity coefficient of plate, dp = Up λ⁄ [m-1

]

uwn - Eigenvalues satisfying characteristic equation,

tan(uwn) = 2hduwn (uwn2 + h2dw

2 )⁄

upn - Eigenvalues satisfying characteristic equation,

tan(upn) = 2hdupn (upn2 + h2dp

2)⁄

Awn - Coefficients of Fourier series, Awn = uwn2 (uwn

2 + h2dw2 + 2hdw)⁄

Apn - Coefficients of Fourier series, Apn = upn2 (upn

2 + h2dp2 + 2hdp)⁄

qp - Dimensionless plate heat reflection rate

R - Pipe radius [m]

t8/5 - Time it takes for the weld seam and adjacent heat-affected zone to cool from

800 °C to 500 °C

t100 - Time it takes for weld seam and adjacent heat-affected zone to reach 100 °C

List of Figures

xi

List of Figures

Figure 1: V-groove preparatory joints as specified in (a) AWS D10.11M.D10.11:2007 and (b)

AS2885.2-2007 standards. ......................................................................................................... 2

Figure 2: A simplified illustration of field pipeline girth welding procedure............................ 4

Figure 3: Pipe-line girth welding in field conditions (Miller Welding Equipment, 2014). ....... 5

Figure 4: SMA welding process. ............................................................................................... 8

Figure 5: Root, Hot, Filling and Capping passes in a pipe weld joint. ...................................... 9

Figure 6: Pipeline construction (The Joyce Road Neighbourhood, 2012)............................... 10

Figure 7: Pipeline construction procedure (Dunstone, 2004). ................................................. 11

Figure 8: Typical temperature history of a weld and characteristic cooling times .................. 13

Figure 9: Prediction of dominant microstructure from temperature histories (solid lines) using

a CCT diagram for X70 (Onsoien et al., 2009). TH 1 leads to a Martensite microstructure

with VH 340 and TH 2 facilitates a Bainitic microstructure, which is less brittle (VH 212). . 14

Figure 10: Diffusion constant of hydrogen in Ferritic steels versus temperature (Coe and

Chano, 1975). The Figure clearly demonstrates that there is a sharp drop in diffusivity of

hydrogen when the temperature drops below 100 °C. ............................................................. 15

Figure 11: Heat loses and heat transfer in SMAW. ................................................................. 17

Figure 12: An example of the numerical modelling of the transient thermal field of a welded

pipe (Feli et al., 2011). ............................................................................................................. 21

List of Figures

xii

Figure 13: An example of numerical modelling of stress field of a welded pipe (Feli et al.,

2011). ....................................................................................................................................... 22

Figure 14: Heat conduction and convective heat transfer from surface resulting from a

moving heat source on the plate surface. ................................................................................. 26

Figure 15: 1D Gaussian heat source. ....................................................................................... 30

Figure 16: 2D Gaussian heat source. ....................................................................................... 31

Figure 17: Goldak et al.’s 3D heat source................................................................................ 32

Figure 18: Application of the method of Mirror Images to the fundamental solution (7). ...... 36

Figure 19: Thermocouple diagram........................................................................................... 38

Figure 20: Various types of thermocouple enclosure options. ................................................ 40

Figure 21: K-type thermocouple setup to record the thermal history of the welded plate

(Attarha and Sattari-Far, 2011). ............................................................................................... 40

Figure 22: Example of a plunged thermocouple in a weld seam (Moore, 2003). .................... 41

Figure 23: Components of the temperature measurement and recording system. .................. 42

Figure 24: Signal “Hockey Puck” Transmitter, a) and Signal Isolators, b)

(Ocean Controls, 2014; RS Australia, 2014). .......................................................................... 43

Figure 25: Wavelength sections within the Electromagnetic Spectrum (Heaviside, 2011). ... 44

Figure 26: Microbolometer Pixel. ............................................................................................ 45

List of Figures

xiii

Figure 27: Typical examples of ZnSe (a) and Ge (b) optical windows (Knight Optical, 2014).

.................................................................................................................................................. 46

Figure 28: Typical transmissivity percentages of a variety of window materials against

wavelengths absorbed for SW and LW thermal cameras (Robinson, 2014). .......................... 47

Figure 29: Geometrical equivalence of the V groove and bead on plate welds with regard to

thermal distribution. ................................................................................................................. 52

Figure 30: Pipeline girth welding procedure. .......................................................................... 53

Figure 31: Schematic diagram to illustrate the mirror image method for pipes. ..................... 54

Figure 32: Representation of a pipe model (Equation (30)) which incorporates heat loss at the

free boundary surface. .............................................................................................................. 56

Figure 33: Lincoln Electric Invertec 415V, 3 Phase welding machine (WESS, 2014). .......... 60

Figure 34: Head Mount Signal “Hockey Puck” Transmitter (from PR Electronics 5331)

(RS Australia, 2014). ............................................................................................................... 60

Figure 35: Equipment setup for recording thermal history with thermocouples. .................... 61

Figure 36: Fitted ZnSe window to rubber manifold. ............................................................... 63

Figure 37: Transmissivity vs spectral range. ........................................................................... 63

Figure 38: Infrared Camera fitted with ZnSe window manifold. ............................................ 64

Figure 39: Infrared camera affixed to tripod............................................................................ 64

Figure 40: Plate test sample specifications. The R-type thermocouple is shown to illustrate

the temperature data acquisition technique. ............................................................................. 65

List of Figures

xiv

Figure 41: Top view of plate test sample with run on/run off tabs. ......................................... 66

Figure 42: Plate sample with tabs mounted on welding jig. .................................................... 67

Figure 43: Complete plate test setup with data acquisition equipment. .................................. 68

Figure 44: Local joint geometry specification of pipe test sample. ......................................... 71

Figure 45: Axial locations of K-type thermocouples. .............................................................. 71

Figure 46: Setup and data acquisition equipment for the pipe test. ......................................... 72

Figures 47(a) and (b): Experimental setup of the pipe test sample.......................................... 73

Figure 48: Infrared camera and pipe test sample setup. ........................................................... 74

Figure 49: Typical thermal histories acquired with the K and R-type thermocouples from the

plate test. .................................................................................................................................. 76

Figure 50: Thermal history of point B30°, see Fig. 45. The pipe is welded with the weld start

angle, ϕ30°, Tph = 25 °C and h = 6 mm. ................................................................................... 76

Figure 51: Typical thermal images captured during welding (left image) and cooling (right

image) of the pipe test sample. ................................................................................................ 77

Figure 52: Thermal history of thermal image sequence generated with IRBIS 3.0 of point

B90° on pipe welded with weld start angle, ϕ90°, Tph = 25 °C and h = 6 mm. .......................... 78

Figure 53: Example of weld metal thermal history. Symbols represent experimental

measurements and the solid line is the theoretical prediction utilising Equations (13) and (26).

.................................................................................................................................................. 80

List of Figures

xv

Figure 54: Calculated t8/5 cooling times with correction for local geometry (filled symbols)

and without (un-filled symbols) plotted against measured t8/5 cooling times of V groove

welding tests............................................................................................................................. 81

Figure 55: Calculated t8/5 cooling times with the equivalent thickness approach and variable

arc efficiency (filled symbols) and without (un-filled symbols) plotted against measured t8/5

cooling times of previous V groove welding tests performed with various root gaps. ........... 83

Figure 56: Comparison of thermocouple measurements and modelling predictions for 220

OD pipe welded with pipeline welding procedure ϕ30° and ϕ90° at B30° (a) and B90° (b)

respectively. Tph = 25 °C and h = 6 mm. ................................................................................ 86

Figure 57: Comparison of thermocouple readings and modelled predictions for 220 mm OD

pipe welded with pipeline welding procedure ϕ90°, h = 12.5 mm for Tph = 25 °C (a) 70 °C (b)

and 100 °C, respectively (c). .................................................................................................... 87

Figure 58: Comparison of thermocouple and infrared camera data thermal histories for pipe

welding procedure using weld start angle ϕ90° at B90°. Tph = 25 °C and h = 6 mm. ............... 89

Figure 59: Cooling time t100 along the pipe circumference for ϕ30° and h = 12.5 mm. .......... 91

Figure 60: Cooling time t100 along the pipe circumference for ϕ90° and h =12.5 mm. ........... 91

Figure 61: Cooling time t100 along the pipe circumference for ϕ30° and h = 6 mm. ............... 92

Figure 62: Cooling time t100 along the pipe circumference for ϕ90° and h = 6 mm. ............... 92

Figure 63: Cooling time t100 along the pipe circumference for ϕ30° and heat input of 0. 8 kJ

mm-1

. ........................................................................................................................................ 93

List of Figures

xvi

Figure 64: Cooling time t100 along the pipe circumference for ϕ90° and heat input of 0.4 kJ

mm-1

. ........................................................................................................................................ 93

List of Figures

xvii

List of Tables

Table 1: Arc efficiencies of various welding processes, η. ..................................................... 18

Table 2: Classification of Analytical Thermal Field Models................................................... 23

Table 3: Joint characteristics of the plate test samples. ........................................................... 69

Table 4: Welding parameters applied to each sample in the plate test. ................................... 69

Table 5: Dimensions of pipe test samples................................................................................ 74

Table 6: Welding parameters applied to the pipe test samples in Table 5. .............................. 75

Table 7: General high temperature region thermal properties of most steels. ......................... 79

Table 8: Geometry factors for various test piece thicknesses used in V groove welding tests.

.................................................................................................................................................. 82

Table 9: Determined weld arc efficiencies for V groove welds of various nominal thicknesses

and root gap.............................................................................................................................. 82

xviii

Table of Contents

xix

Table of Contents

Chapter 1: Introduction .......................................................................................................... 1

Chapter 2: Literature Review................................................................................................. 7

2.1 Welding ....................................................................................................................... 7

2.1.1 Pipeline construction ............................................................................................ 9

2.1.2 Effect of Temperature History on Weld Quality ............................................... 12

2.2 Thermal efficiency of welds ...................................................................................... 17

2.2.1 Comments on Numerical Approaches ............................................................... 20

2.3 Analytical Modelling of Thermal History ................................................................. 23

2.3.1 Point Heat Source Models ................................................................................. 24

2.3.2 Line heat source models..................................................................................... 28

2.3.3 Advanced heat source models ............................................................................ 29

2.4 Summary and Research Gap ..................................................................................... 33

Chapter 3: Research Methodology ...................................................................................... 35

3.1 Mathematical Modelling ........................................................................................... 35

3.2 Summary of Experimental Techniques ..................................................................... 37

3.2.2 Principles of thermal imaging ............................................................................ 44

Table of Contents

xx

Chapter 4: Development of Thermal Field Models for Pipeline Girth Welding ................. 49

4.1 Incorporation of the local preparatory joint geometry into a modelling approach ... 49

4.1.1 Thermal field model ........................................................................................... 50

4.1.2 Account for shape of V groove joint geometry: Equivalent thickness approach ..

............................................................................................................................ 52

4.2 Incorporation of pipeline girth welding procedure into modelling approach ........... 53

4.2.1 Development of thermal field model ................................................................. 54

4.3 Chapter Summary ...................................................................................................... 58

Chapter 5: Experimental studies .......................................................................................... 59

5.1 Experimental Equipment ........................................................................................... 59

5.1.1 Welding machine and consumables ................................................................... 59

5.1.2 Setup of temperature data recording equipment ................................................ 60

5.1.3 Software ............................................................................................................. 61

5.1.4 Thermocouple Calibration ................................................................................. 61

5.1.5 Temperature data acquisition with Infrared Camera ......................................... 62

5.2 Plate Tests ................................................................................................................. 65

5.3 Pipe Tests .................................................................................................................. 70

5.4 Selected examples of the recorded temperature history ............................................ 75

Table of Contents

xxi

5.5 Chapter Summary ...................................................................................................... 78

Chapter 6: Thermal Field Model for Pipeline Girth Welding ............................................. 79

6.1 Evaluation of Thermal Arc Efficiency during Pipeline Girth Welding .................... 79

Chapter 7: Effect of Welding Procedure on Thermal History ............................................. 85

7.1 Validation of pipeline welding procedure model with temperature data .................. 85

7.1.1 Comparison of thermal histories obtained with different data acquisition

techniques ......................................................................................................................... 88

7.2 Temperature Variation across the Pipe Circumference ............................................. 89

7.3 Chapter Summary ...................................................................................................... 94

Chapter 8: Overall Conclusion ............................................................................................ 95

8.1 Publications from current research ............................................................................ 97

References……………………………………………………………………………………………………………...…99

Table of Contents

xxii

Chapter 1: Introduction

1


There have been a large number of analytical, numerical and experimental studies

focussed on the investigation of the transient thermal field associated with the welding of

pipes (Alam et al., 1999; Nguyen, 2004; Deng and Murakawa, 2006; Akbari and Sattari-Far,

2009; Lee et al., 2013). One of the main objectives of these studies was the establishment of a

link between welding parameters (such as heat input and weld travel speed) and geometry

(plate or pipe wall thickness) from one side, and the generated transient thermal field from

the other side. This transient thermal field (or temperature history) is often described by

various cooling times, such as t8/5 or t100. The former is the time it takes for the weld seam

and adjacent heat-affected zone to cool from 800 °C to 500 °C and the latter is the time taken

to reach 100 °C. Both cooling times are widely accepted by the international and Australian

pipeline industries to analyse and characterise the weld quality as well as the susceptibility of

the weldment to hydrogen assisted cold cracking (HACC). These characteristic times for

many practical situations can be obtained from numerous simplified engineering procedures,

standards and codes available in the literature (Yurioka et al., 1986).

Despite significant progress made over the past century in predicting the thermal history

of weldments, the existing simplified analytical procedures, as well as very sophisticated

numerical approaches, usually disregard the actual local joint geometry, which, in accordance

with industrial standards can vary quite significantly from one weld to another, and the

particular way the weld is deposited. For example, the same pipeline weld run can be

completed with different start and stop positions, or completed by a different number of

welders.

There are various joint geometries used in pipeline welding such as bevel, square, single-

J, double-J, single-V, double-V, single-U and double-U groove joints (Lamit, 1981; Nayyar,


2

1992). However, single V-groove joints are most commonly used for pipe sections of small to

medium diameter (< 610 mm) pipes (Lamit, 1981) and are recommended by relevant

industrial standards (AWS, 2007; Standards Australia, 2007). For this reason this

investigation will solely focus on weld joints with V-groove geometry. Typical examples of

the industry acceptable preparatory geometries are presented in Fig. 1.

Figure 1: V-groove preparatory joints as specified in (a) AWS D10.11M.D10.11:2007 and (b)

AS2885.2-2007 standards.

Pipeline welding standards normally specify three controllable sizes of the local

geometry: root face, root gap (RG) and groove (or bevel) angle, (the maximum offset is a fit

up tolerance and not considered to be a joint design characteristic). In accordance with Fig. 1,

the actual size of the root gap is industry acceptable if it is (a) less or greater than the


3

diameter of the filler metal used for a particular welding method (AWS, 2007) or (b) it lies

within 1.4 ± 0.6 mm (Standards Australia, 2007). However, such a large variation of root gap

sizes in practice can significantly vary the arc efficiency, and, subsequently, the thermal

energy transferred to, or dissipated into, the different pipe joints welded with the same

welding parameters. With the wider gap, the arc efficiency normally decreases as more

energy escapes and dissipates in the environment through the wider gap. Other characteristic

dimensions (such as root face and bevel angle) are expected to have a much lesser influence

on the thermal losses; and this will be elaborated further in the literature review section of

this thesis.

The thermal energy dissipated in the joint generates transient temperature and stress

fields, and leads to radical changes of material properties and microstructure in the vicinity of

the weld. Therefore, it is important to know the actual arc efficiency of a welding process in

order to utilize the appropriate heat f1ow models, analytical or numerical techniques, and

provide a reliable assessment of the material properties, thermally-induced and residual

stresses in the weldment. It is well known that all these factors (various stresses and material

properties) have a significant impact on the integrity and durability of the welded structure.

One of the objectives of the current thesis is to investigate the effect of the root gap size on

arc efficiency within the typical geometry variations, which are tolerated by industrial

standards.

From the Australian pipeline industry perspective, thin walled (< 12 mm) and small

diameter (< 500 mm) pipes are usually considered for gas and oil transmission, in contrast

with Europe and America, where pipes are generally thicker and larger in diameter (Alam et

al., 1999). In the case of 400 OD pipe, normally two welders deposit the girth weld

simultaneously (Fletcher, 2011) to support a high rate of pipeline construction and ease the

stress conditions imposed by the pipe joining procedure (clamp release, lifting, etc.)


4

(McAlister, 1998; Dunstone, 2004). The typical welding procedure is shown schematically in

Fig. 2.

Figure 2: A simplified illustration of field pipeline girth welding procedure.

In accordance with this procedure, the first welder (Welder 1) starts to deposit the

weld seam from location A at the top of the pipe to location C at the bottom. At the same

time, the second welder (Welder 2) begins their weld run at location B, which is not clearly

specified in the welding procedure, and continues depositing the weld seam down to the

bottom location (C). Before the first welder reaches the bottom location (C), Welder 2 re-

starts welding at location A and completes the weld deposition of the right half-circle of the

pipe. Fig. 3 shows two welders completing the root pass in field conditions.

B

A

C

Pipe

Start location

End location

Welder 1

Welder 2

Start/End location Symbols


5

Figure 3: Pipe-line girth welding in field conditions (Miller Welding Equipment, 2014).

However, there exist many different pipe welding procedures that can vary from the

one described above and illustrated in Figures 2 and 3 (North et al., 1982; McAlister, 1998;

Sacks and Bohnart, 2005). These pipeline welding procedures depend on the pipe diameter

and may involve three or four welders working simultaneously (North et al., 1982; McAlister,

1998). In practice, specifically for larger diameters of pipes (> 220 mm), there may be several

interruptions in the continuous welding associated with the replacement of electrodes. For

example, in the described procedure, Fig. 2, the left side run (A-C) could not be completed

with a single electrode, so there may well be another stop/start location along this weld run.

Usually, these interruptions are disregarded in modelling investigations of pipeline welding.

However, these aspects of welding procedures can significantly affect the transient thermal

field, specifically in the close vicinity of stop/start locations. Therefore a different weld metal

and HAZ properties at these locations can be expected, affecting the weld quality, residual

stress profile as well as the susceptibility to cracking defects. In many practical situations,

manual welding is more preferable than the use of automatic welding machines as the manual


6

welding with two or more welders working simultaneously can support a much higher speed

of pipeline construction. For example, the girth weld for a 400 mm pipe can be completed in

approximately 90 seconds (Fletcher, 2011).

As stated above, the main objective of this thesis is to evaluate the significance and

effect of the actual geometry of the preparatory joint and specifics of the pipeline welding

procedure on the thermal history in order to understand their possible influence on the weld

quality, pipeline integrity and weldability. To address this objective, two simplified analytical

models will be developed and validated by comparing the theoretical predictions against

outcomes of previous investigations, and data obtained from full-scale experimental tests.

Chapter 2 will provide a broad introduction into the research area. In particular, various

analytical models for welding operations will be described in an historical context. Chapter 2

will also re-state the research gaps to be addressed in this thesis. Chapter 3 will present the

research methodology adopted in this project. In Chapter 4, the existing analytical models

will be evaluated critically and extended to simulate the previously stated aspects of pipeline

girth welding. These models will be further validated in Chapter 6 with the outcomes of

experimental studies including full-scale pipe tests described in Chapter 5. A case study will

be conducted in Chapter 7, based on the previously validated pipe model, in order to provide

answers to the pipeline welding issues posed earlier on in this chapter. The thesis will be

concluded with an overall summary, which will highlight the main outcomes of this project,

potential utilisation of these outcomes, as well as ideas for future research, which will address

some shortcomings of the present study.

Chapter 2: Literature Review

7


The literature review will first introduce the field of welding and, after that; it will

focus on several research topics, which are important for the current study. These topics

include the effect of the thermal history on the weld quality, evaluation of major factors

influencing the arc efficiency, experimental measurement techniques of transient thermal

field generated by welding and the development of theoretical modelling approaches for

evaluation of temperature history. A gap statement briefly described in the Introduction will

be further elaborated in this section of the thesis.

2.1 Welding

Welding is a fabrication process used to join materials securely together. This is usually

accomplished by melting the joining parts and adding a filler material or consumable to form

a molten weld pool that cools and solidifies to form a strong joint. Welding is widely used in

the manufacturing of airplanes, heavy machinery, general machinery parts and pipeline

construction. Various energy sources can be used for welding such as a gas flame, electric

arc, laser, electron beam, friction, and ultrasound. The most popular arc welding processes

which utilise an electric arc, including Shielded Metal Arc Welding (SMAW), Gas Tungsten

Arc Welding (GTAW), Gas Metal Arc Welding (GMAW), Flux-Cored Arc Welding

(FCAW) and Submerged Arc Welding (SAW). All these processes use different

consumables, methods of deposition and weld shielding techniques. However, SMAW has

been the most popular welding process used to construct pipelines in Australia for the past

several decades, and is still being widely used today (Fletcher and Piper, 2012). Therefore, in

this thesis the main focus will be on Shielded Metal Arc Welding or SMAW only.


8

SMAW is a manual arc welding process that uses a consumable electrode coated in

flux to lay the weld, see Fig. 4. An electric current, which can be either an alternating current

or a direct current supplied from a power source, generates an electric arc between the

electrode tip and the metal parts to be joined. The flux coating of the electrode melts down as

the weld is deposited, producing vapours that serve as a shielding gas and providing a layer

of slag, both of which protect the weld area from atmospheric contamination.

Figure 4: SMA welding process.

The process is very versatile, relatively simple and does not require sophisticated

equipment or highly trained personnel. Therefore, SMAW is one of the world's most popular

fabrication processes and joining methods, commonly used in construction industries,

maintenance procedure repertoires as well as the repair of structural components. The

SMAW process is primarily used to weld iron and steels (including stainless steel), however,

aluminium, nickel and copper alloys can also be welded with this process (Cary and Helzer,

2005). The SMAW process is illustrated schematically in Fig. 4.

Weld Pool Heat Affected Zone (HAZ)

Electrode

Weld Arc

Flux

Parent Metal


9

2.1.1 Pipeline construction

Time is a major constraint in pipeline construction. A faster rate of construction can

lead to a significant reduction in the overall cost of the pipeline, which can reach billions of

dollars. For example, in 2010 the Australian pipeline industry association (APIA) reported on

plans to improve the gas transmission infrastructure needed for the expansion of

Queensland’s coal seam gas (CSG) production industry. The APIA estimated that $10 billion

would be spent between 2010 and 2020 on the development and upgrading of the pipeline

infrastructure (APIA, 2010).

One particular factor significantly affecting the construction rate of the weldments is

the deposition speed of the first and second weld runs, known as root and hot passes,

respectively (see, Fig.5).

Figure 5: Root, Hot, Filling and Capping passes in a pipe weld joint.

The speed at which the root and hot passes are deposited by the front-end welding

team is termed the front-end speed. The filling and capping passes (see Fig. 5) can be

completed later by the second or multiple welding teams. Therefore the rate of productivity

of pipeline construction is largely limited by the front-end speed. To improve productivity,

construction management emphasises that the front-end welding team complete the root and

Filling passes

Capping passes

Root pass

Hot pass


10

hot passes as fast as possible by welding the pipe joint with extremely high rates of

deposition and subsequently low heat inputs. This managerial approach can challenge the

integrity and safety of the pipeline. Fig. 6 shows a photograph of the front-end pipeline

construction.

Figure 6: Pipeline construction (The Joyce Road Neighbourhood, 2012).

The typical procedure of pipeline construction has been outlined by Smart and Bilston (1995)

and is illustrated in Fig. 7. This procedure is highly optimised to avoid any delays and

interruptions.


11

Skids are used to support the pipeline above

the ground, so that the welding procedure can

be executed.

The line-up clamp is positioned at the front

edge of the leading pipe and clamped to that

leading edge.

A crane is used to carry and insert the next

pipe over the line up clamp. Wedges are used

to correct the root gap between the edges of

both pipes. The pipes are also restrained with

exterior clamps

The pipeline welding procedure is

commenced with the root pass welded

typically with cellulosic electrodes.

The front end of the pipeline is lifted by the

crane, while a support skid is placed beneath

the front end of the leading pipe.

The front-end team moves to the next weld.

Figure 7: Pipeline construction procedure (Dunstone, 2004).

Fig. 7 illustrates a typical pipeline construction procedure, which also demonstrates

that the speed of the pipeline assembling is largely determined by the production speed of the

front-end welding team. In addition to the increase of the welding speed to complete the root

and hot passes, Henderson et al. (1996) suggested that the removal of the line-up clamps and

earlier lowering of the pipe on the leading skid, (after only 50% of the root pass is

completed), can reduce the production time per pipe weld joint by 25%, or from 6 to 4.5

Extract line-up clamp

Conduct root welding

Insert pipe onto line up

clamp using crane

Crane lifts pipe

Crane lowers pipe onto skid


12

minutes. However, constructing pipelines with procedures based on Henderson et al.’s

suggestions could further compromise the integrity of the pipeline as the removal of the line-

up clamps and earlier lowering of the pipe on the leading skid can significantly aggravate the

stress conditions during welding and increase the risk of generation of critical welding

defects.

2.1.2 Effect of Temperature History on Weld Quality

The literature review indicates that the quality of a joint welded with the SMAW process

is influenced by (i) parent material and electrode composition; (ii) joint type; (iii) restraint

and (iv) environmental conditions as well as (v) temperature history (Radaj, 1992; Nguyen,

2004). As stated in the Introduction, one of the main goals of this research is to interpret the

transient thermal field generated by a weld in terms of welding parameters. Therefore, the

present study will focus on the last factor (v): the temperature history of the weld and HAZ as

a function of welding parameters, local geometry and welding procedures.

A typical temperature history of the weld metal is shown in Fig. 8. However, it is not

practical to calculate, evaluate or measure the entire temperature history for each individual

weld and welding conditions. Therefore, the transient thermal field (or temperature history) is

often characterised by cooling times, such as t8/5 or t100 (Yurioka et al., 1986; Kasuya and

Yurioka, 1993). In accordance with Fig. 8, t8/5 is the time it takes for the weld seam and

adjacent heat-affected zone to cool from 800 °C to 500 °C and t100 is the time to reach 100

°C from a liquid state (weld pool) (Terasaki et al., 1988; Radaj, 1992; Kasuya et al., 1995;

Nguyen, 2004).


13

Figure 8: Typical temperature history of a weld and characteristic cooling times

(t8/5 and t100).

The t8/5 is an important characteristic because the microstructure of the WM and the HAZ in

steel pipelines is largely determined by the cooling time from 800 to 500 °C, which is also

known as the transformation temperature range (Nguyen, 2004). For example, cooling times

longer than 7s (t8/5 > 7s) normally lead to the formation of a bainitic dominant

microstructure. While shorter cooling times of less than 3s, depending on the chemical

composition, (t8/5 < 3s) facilitate a martensitic microstructure, which is quite hard, brittle and

highly susceptible to cracking (Kasuya et al., 1995; Keehan et al., 2010).

The microstructure realisation in the WM and HAZ can be evaluated via a Continuous

Cooling Transformation (CCT) diagram (Karkhin et al., 2006; Onsoien et al., 2009), which is

shown in Fig. 9. It displays two temperature histories: Temperature History 1 (TH 1) has a

relatively short t8/5 cooling time, normally resulting in a hard and brittle martensitic

microstructure. TH 2 has a longer t8/5 cooling time, leading to more favourable bainitic

Welding Parameters:

Heat Input, HI = 1.7 kJ mm-1

Weld speed, v = 2.5 mm s-1

Plate thickness, h = 9 mm

0 200 400 600

500

1000

1500

Tem

per

ature

, [°

C]

time, [s]

t8/5

0

t100

800

100


14

microstructure, which is more ductile (Onsoien et al., 2009; Keehan et al., 2010). In welding,

it is normally imperative to avoid the formation of martensite in the weld metal and the HAZ

(Kasuya et al., 1995).

Figure 9: Prediction of dominant microstructure from temperature histories (solid lines) using

a CCT diagram for X70 (Onsoien et al., 2009). TH 1 leads to a Martensite microstructure

with VH 340 and TH 2 facilitates a Bainitic microstructure, which is less brittle (VH 212).

The t100 cooling time is another important characteristic because it determines the

amount of hydrogen, which can be defused from the weld during its cooling to ambient

temperature (Bailey et al., 1973; Yurioka and Suzuki, 1990; Nevasmaa, 2003). The residual

hydrogen, which stays trapped in the weld and HAZ, imposes a serious risk of hydrogen

assisted cold cracking (HACC). This phenomenon (HACC) can compromise the quality of

the weld and the overall integrity of the pipeline. Therefore, various methods are often

utilised in welding procedures to extend the t100 cooling time, such as preheating or wind

shields, in situations or in climate conditions where there is a risk of HACC formation. One

such situation is pipeline girth welding, which is completed at low heat inputs and with a high

weld travel speed. This situation is the focus of the current thesis.

0

Martensite 200

400

600

800

1000

Tem

per

ature

, [°

C]

10 100 1000 0

Time, [s]

t8/5

Austenite

Bainite

t8/5

Thermal History VH t8/5

TH1 340 1.7s

TH2 212 23.5s


15

It is well known from the literature that the diffusion rate of hydrogen in steel is a

function of temperature that decreases significantly when the temperature drops, see Fig. 10.

Figure 10: Diffusion constant of hydrogen in Ferritic steels versus temperature (Coe and

Chano, 1975). The Figure clearly demonstrates that there is a sharp drop in diffusivity of

hydrogen when the temperature drops below 100 °C.

The rate of hydrogen diffusion is very slow and almost negligible when the temperature

of the weld metal or HAZ drops below 100 °C (Bailey et al., 1973; Coe and Chano, 1975).

This explains the selection of 100 °C as the threshold temperature and the cooling time t100

as an important characteristic affecting the weld quality. However, the most representative

parameter influencing the weld quality is the residual hydrogen content. Nevertheless, direct

measuring of the hydrogen content is a complex and lengthy task that requires rather

sophisticated equipment (Yurioka and Suzuki, 1990). Therefore, many industrial standards

10 50 100 200 500

Temperature [°C]

10-8

10-7

10-6

10-5

10-4

10-9

Dif

fusi

on

co

nst

ant

[cm

2 s-1

]

Diffusion rate decreases

substantially when weld

temperature drops below

100 °C

10-3

Upper limit

Lower limit


16

and procedures utilise t100 cooling time, which can be measured with a simple temperature

probe, rather than the residual hydrogen content.

To facilitate the speed of pipeline construction in Australia, the root pass is often

deposited with cellulosic electrodes, which are covered by a cellulosic-enriched flux that can

absorb moisture from the environment (Dunstone, 2004). Cellulosic electrodes are used for

their superior penetration qualities and ability to support a high welding speed. However, the

use of cellulosic electrodes has a drawback: the weld metal becomes significantly

contaminated with hydrogen, which makes the weldment more susceptible to hydrogen

cracking (Yurioka et al., 1986; Yurioka and Suzuki, 1990). The phenomenon of hydrogen

assisted cold cracking (HACC) is largely impacted by microstructure, hydrogen content and

stresses. Therefore, both characteristic cooling times, t8/5 and t100, have a direct impact on

the quality and susceptibility of the weldment to HACC. The first characteristic, t8/5, largely

determines the microstructure realisation, as explained above, and the second one, t100,

governs the hydrogen diffusion (Bailey et al., 1973; Nevasmaa, 2003).

It is not surprising that there are many models and empirical relationships which predict

or relate the risk of the occurrence of HACC with these two characteristic cooling times

(Yurioka et al., 1986; Kasuya and Yurioka, 1993). Subsequently, many simplified models

and empirical formulas have been developed to predict t8/5 (Terasaki et al., 1988; Karkhin et

al., 2006) and t100 (Bailey et al., 1973; Nevasmaa, 2003) from the welding parameters and

the thickness of the weldment. However, all these models normally disregard the effects of

the pipeline welding procedures and local joint geometry on these characteristic cooling

times. Therefore, the objective of this thesis is to address this shortcoming.


17

2.2 Thermal efficiency of welds

The thermal efficiency of the weld is dependent on the efficiency of the weld arc (arc

efficiency). Arc efficiency, η, is a quantitative measure of the fraction of total energy, Etotal,

dissipated in the weldment, or:

η =Ew

Etotal , (1)

where Ew represents the total energy transferred to, or dissipated in, the work piece. The total

energy, Etotal, represents the welding arc energy generated at the electrode. The total energy

is normally distributed in two ways: a portion is lost to the environment (Etotal − Ew); and

the remainder, Ew, is transferred to the weldment (Fig. 11) (DuPont and Marder, 1995;

Nguyen, 2004).

Figure 11: Heat loses and heat transfer in SMAW.

The rate of energy (or power) generated by the arc is given simply by the product of the, arc

voltage, V, and the current, I. The heat input, q, is a more commonly used characteristic of the

welding process:

Losses (Etotal − Ew) Losses (Etotal − Ew)

Heat Affected

Zone (HAZ)

Electrode

Weld Arc

Etotal

Ew


18

q =ηVI

v. (2)

This heat input represents the quantity of energy generated by the arc per unit length of weld

(Rosenthal, 1946; Radaj, 1992). The appropriate values of the heat input are necessary for

analytical or numerical heat transfer models to predict the temperature history, mechanical

properties and stress field in the weldments with sufficient accuracy, as well as the overall

weld quality. The processes governing the arc efficiency are too complex to model and its

values are usually extracted from experimental data. For example, DuPont and Marder (1995)

conducted measurements of arc efficiency using a Seebeck arc welding calorimeter, which

were first described in Giedt et al. (1987). Yurioka et al. (1986) provided some typical values

of the arc efficiency for various types of welding processes. Table 1 summarises the typical

values for arc efficiency from various studies.

Table 1: Arc efficiencies of various welding processes, η.

Welding

Process

Sources of various arc heat efficiencies, η

Yurioka et

al. (1986)

Radaj (1992) Nguyen

(2004)

Christensen

et al. (1965)

Svensson

(1994)

SMAW

(Rutile)

0.9 0.8 0.66-0.85 0.66-0.85 0.85-2.50

SMAW (low

hydrogen)

0.8 0.8 0.66-0.85 0.66-0.85 0.85-2.50

Many analytical, numerical and experimental studies (North et al., 1982; Sawhill et

al., 1986; Noble and Pargeter, 1988; Alam et al., 1999) that focused on the investigation of

thermal, mechanical and micro-structural parameters of welds and heat-affected zones,

utilised empirical values of the arc efficiency. However, as is shown in Table 1, these values

vary considerably, which essentially negates the theoretical efforts (analytical or numerical)


19

to predict the transient thermal history of the weld and HAZ. Therefore, the scattering has to

be narrowed down in order to improve the predictive capabilities of all the previously

developed analytical, empirical and numerical models, which is one of the main objectives of

the current study.

It is known that the arc efficiency is only slightly affected by the welding parameters,

such as heat input and welding speed, for any given welding procedure (DuPont and Marder,

1995). In contrast, the actual local geometry of the preparatory joint can significantly affect

the portion of the total energy dissipated in the weldment (Nguyen, 2004). The local

geometry of pipeline girth welding can be characterised by three parameters: root height, root

gap and groove angle, as is shown in Fig. 1. The root height and groove angle do not

significantly influence the arc efficiency. For example, Terasaki et al. (1988) conducted a

study on the effect of the groove angle on the thermal history. The outcomes of this study

suggest that the weld arc efficiency has quite a low sensitivity to the changes of the groove

angle. A notable effect on the cooling time, t8/5 can only be observed when the variations of

the groove angle exceed 55°. The industry acceptable welds are unlikely to have such large

variations in the local geometry, see Fig.1 (AWS, 2007; Standards Australia, 2007).

The effect of root gap on arc efficiency is expected to be quite large as the heat losses

to the environment can substantially increase through the wider gap. This conclusion also

follows from experimental studies on the thermal field, which utilised different root gaps and

plate thicknesses (Noble and Pargeter, 1988; Alam et al., 1999; Suppiah, 1999). The current

thesis will utilise the outcomes of these studies, and will focus on the development of the

appropriate relationship between the root gap (as a main influential factor) and arc efficiency.

Another feature of the SMAW process is slag formation. The SMAW process

normally produces a layer of slag on top of the weld metal (Surhone et al., 2010). All

previous theoretical studies (Eagar and Tsai, 1983; Yurioka et al., 1986) and texts (Radaj,


20

1992; Nguyen, 2004) do not consider the effect of slag on the thermal field as the

temperature history near the weld is mainly affected by the heat dissipation into the

surrounding material due to a much higher conductivity of the steel in comparison with the

slag material. For this reason the effect of the slag layer will also not be considered in the

present study.

2.2.1 Comments on Numerical Approaches

Over the past twenty years, a significant effort has been made to develop a new area

in welding research called “Computational Welding Mechanics” (CWM) (Lindgren, 2007).

Very sophisticated models have been developed to simulate the transient thermal field:

thermally induced stresses, residual stresses, deformations and predictions of the

microstructure (Deng and Murakawa, 2006; Anca et al., 2011; Lee et al., 2013). The CWM

approach has many benefits in comparison with simplified analytical models, including

accounting for:

(i) Specific local geometry;

(ii) Temperature dependent material properties;

(iii) Nonlinear deformations;

(iv) Phase transformations in WM and HAZ during heating and cooling;

(v) Other effects.

In numerical approaches, the governing equations apply to finite elements, which are

assembled into a global matrix equation (Lindgren, 2006). This matrix equation is solved

using efficient computer programs. The thermal and mechanical analyses are normally

considered uncoupled. In other words, the thermal analysis serves as input data for the

solution of the mechanical problem.


21

The thermal analysis normally utilises a thermal energy conservation equation, which

can be written in a highly simplified form as:

MH = Text − Tint , (3)

where M representes the mass vector, Text is the thermal load vector and Tint is the internal

flux vector. H is the enthalpy matrix. Typical outcomes of the thermal analysis of welding are

shown in Fig. 12.

Figure 12: An example of the numerical modelling of the transient thermal field of a welded

pipe (Feli et al., 2011).

The mechanical analysis is based on Newton’s second law, which can be written as

MU = Fext − Fint , (4)

where Fext is a vector that represents the external loads and Fint is a vector that represents the

internal forces caused by stresses, both residual and thermally induced. The matrix U

represents the displacements, and MU the inertia term. In welding problems, MU ≈ 0 and is

usually disregarded for all practical purposes. A typical example of stress analysis is shown

in Fig. 13.

Temperature (°C)

2014

1535

1409

1283

1156

1030

651

399

146

20


22

Figure 13: An example of numerical modelling of stress field of a welded pipe (Feli et al.,

2011).

However, the numerical approaches have the same issue as the simplified analytical

models: all welding models have to assume a designated value of arc efficiency. The

calculation of arc efficiency is very complex as it has to utilise multi-physics modelling

(plasma, electric, gas, fluid, thermal, etc.), which incorporates coupled and highly non-linear

equations of Plasma Physics describing the behaviour of the weld arc (Lindgren, 2001a).

Despite the advantages of the numerical approaches, the accuracy can be greatly affected by

uncertainties associated with arc efficiency values. Therefore the current work is as

significant for analytical as well as numerical approaches. The current approach will utilise

the analytical modelling approach rather than the numerical approach because analytical

modelling offers a simplified and robust method of calculation to predict the thermal field,

whereas numerical modelling is complex, time consuming and impractical for a parametric

analysis (Lindgren 2001a). For this reason, this study will utilise the analytical modelling

approach to simulate the thermal field. Data obtained from a limited number of experimental

studies is used to validate the analytical models. Furthermore, these models will be utilised to

Axial Stress (MPa)

+418

+353

+222

+157

-39

-105

-170

-301

-366


23

obtain an empirical relationship between arc efficiency and the root gap, as well as to

investigate the effect of welding procedures on the transient thermal history of the weldment.

2.3 Analytical Modelling of Thermal History

Analytical models are simpler in form and less cumbersome to apply than numerical

models, as highlighted earlier. Analytical models can be classified in terms of the:

(i) Dimensionality of the governing equations,

(ii) Representation of the heat source and,

(iii) Incorporation of different boundary conditions.

A summary of past and current most popular analytical models of welding processes is

presented in Table 2.

Table 2: Classification of Analytical Thermal Field Models.

Number of dimensions of the undersigning heat equation

Heat Source 1D + t 2D + t 3D + t

Point heat source Rosenthal (1946), Rykalin

(1957)

Rykalin (1957), Nunes (1983)

Radaj (1992), Yurioka et al.

(1986), Terasaki et al. (1988)*,

Kasuya and Yurioka (1993)*

Line heat source, Nguyen

(2004)

Rosenthal (1946), Radaj

(1992)*, Zhang (1989)*,

Nguyen (2004)

Heat intensity

uniformly distributed

over an area

Radaj (1992) Carslaw and Jaeger (1947) Carslaw and Jaeger (1947),

Darmadi et al. (2011)

Gaussian heat

distribution

Fassani and Trevisan

(2003)

Eagar and Tsai (1983), Boo and

Cho (1990)

Double ellipsoid

heat distribution

Nguyen et al. (2004), Winczek

(2010)

* Taking into account convection at free boundary surfaces.


24

All analytical models are based on the general governing heat conduction equation

(5), representing the Fourier law (Rosenthal, 1946; Carslaw and Jaeger, 1947):

D1

∂2T

∂x2+ D2

∂2T

∂y2+ D3

∂2T

∂z2=

1

κ

∂T

∂t, (5)

where Di = 1 or 0, depending on the dimensionality of the model; κ represents the thermal

diffusivity of the material and κ = λ ρcp⁄ , where λ is the thermal conductivity, ρ is the

density and cp is the specific heat of the material. In analytical models, all material properties

are normally considered as constants.

2.3.1 Point Heat Source Models

As mentioned above, analytical models can utilise various representations of the heat

source or heat flux distribution. The heat flux associated with welding is often written as:

Q = ηVI , (6)

where V and I are the welding voltage and current, and η is the arc efficiency, which is one of

the main foci of this research.

All 2D thermal field models in Table 2 utilise the fundamental 2D solution for

instantaneous heat source release:

T(x, y, t) − T0 =Q

4πλtexp (−

r2

4κt), (7)

where T0 is the initial temperature of the plate, r is the distance from the source to the

observation point (r = √x2 + y2), and t is time calculated from the instance of the heat

release.


25

The thermal field caused by moving heat sources can be found as a superposition of

the instantaneous heat sources. In the coordinate system, located at the heat source moving

with the constant speed, v, the temperature distribution is given by Rosenthal (1946) as:

T(x, y, t) − T0 = ∫Q

4πλtexp (−

(x − vt)2 + y2

4κt) dt

0

−∞

=q

2πλexp (−

vx

2κ) K0 (

vr

2κ), (8)

where q = Q v⁄ and K0 is the modified zero order Bessel function of the second kind, x is the

distance along the weldline. Studies by Eagar and Tsai (1983), Fassani and Trevisan (2003)

and many others have demonstrated that equation (8) satisfactorily describes the thermal field

away from the heat source. However, this model produces large errors in temperature

estimates in the vicinity of the heat source, where the temperature is predicted to be infinite in

accordance with this idealisation.

Several researchers have developed point heat source based models that incorporate

convection effects at free surfaces in order to improve the predictive capabilities of

Rosental’s solution (Yurioka et al., 1986; Terasaki et al., 1988; Kasuya and Yurioka, 1993).

The models are still based on the governing heat conduction equation (5), but also utilise

convection boundary conditions at free surfaces, which are normally based on Newton’s Law

of Cooling (Kasuya and Yurioka, 1993):

λ∂T

∂n= −U(T − T∞), (9)

where ∂T ∂n⁄ is the gradient of temperature normal to the free surface, T∞ is the ambient

temperature. The heat transfer and convection coefficients, U and λ, respectively are

considered to be constants in analytical models. One such model was developed by Yurioka

et al. (1986). The solution is obtained through a Fourier series and can be written as:


26

T(x, y, z, t) =

T∞ +Q

πλhexp (−

vw

2κ) ∑ Awn (cos

uwnz

h+

hdw

uwnsin

uwnz

h) ×

∞

n=0

K0 (rz

h√uwn

2 + (vh

2κ)

2

)

+ 2(Tph − T∞) ∑ Apn (cosupnz

h+

hdp

upnsin

upnz

h)

∞

𝑛=0

exp (−upn

2

h2t)

× (sin(upn)

upn−

hdP(cos(upn) − 1)

upn2

),

where w = x − vt, is the moving co-ordinate in the welding direction, rz = √w2 + y2 + z2,

which is the distance from the moving heat source in Fig. 14.

Figure 14: Heat conduction and convective heat transfer from surface resulting from a

moving heat source on the plate surface.

The model takes into account the convective heat transfer at the surface of the weld,

Uw, and plate, Up. The heat transfer/conduction coefficients at the weld and plate are

dw = Uw λ⁄ and dp = Up λ⁄ , respectively.

Eigenvalues uwnand uPn satisfy the following characteristic equations:

tan(uwn) = 2hduwn (uwn2 + h2dw

2 )⁄ , (11a)

(10)

z y

x

v

h

T∞

Tph Up

Uw Up

λ, κ


27

and

tan(upn) = 2hdupn (upn2 + h2dp

2)⁄ . (11b)

The Fourier coefficients are:

Awn = uwn2 (uwn

2 + h2dw2 + 2hdw)⁄ , (12a)

and

Apn = upn2 (upn

2 + h2dp2 + 2hdp)⁄ . (12b)

The described above thermal model is a superposition of the thermal field, generated

by the point source with intensity, Q, and the thermal field induced by preheat temperature,

Tph. This model was used to generate the characteristic cooling times, t8/5 and t100 for a

wide range of welding parameters, Yurioka et al. (1986). This model is often utilised in the

welding industry for prediction of the thermal history or in the evaluation of the susceptibility

of a weldment to HACC.

Later, Yurioka and Kasuya (2004) presented an interesting model, which, strictly

speaking, is mathematically incorrect since the heat reflection rate, qP, is contained. The

empirical heat reflection rate was introduced to improve the correlation of the theoretical

predictions with measured values of the temperature field. The model was obtained using a

method of mirror images and the superposition principle. It can be written in the following

form,

T(x, y, z, t) = T∞ + Tw(x, y, z, t) × exp (−2tUw

ρcph) + (Tph − T∞) × exp (−

2tUp

ρcph),

(13)

where:


28

Tw(x, y, z, t) = Q

2πλexp (−

vw

2κ) (

exp (−vrz

2κ )

rz+ ∑ q

Pn (

exp (−vrn

2κ )

rn+

exp (−vrn

′

2κ)

rn′

)

∞

n=1

).

The only parameters, which are not described above are rn and rn′ , which can be

found as:

rn = √w2 + y2 + (2nh − z)2, (14a)

and

rn′ = √w2 + y2 + (2nh + z)2. (14b)

The empirical heat reflection rate, qP, is different for various welding processes and for

SMAW, qP = 0.8. This model was also extensively validated in various experimental studies

by a number of researchers (Terasaki et al., 1988; Yurioka and Kojima, 2004) and it was also

applied to predict various temperature characteristics of welding (Yurioka, 2004; IQSim,

2010).

2.3.2 Line heat source models

At low welding speeds, the integration of the point heat source solution over time can

provide an adequate estimate of the thermal field or temperature history; see Equations (10)

to (14). Conversely, when the speed of the weld (heat) source is sufficiently large, another

representation of the heat source could be adequate to model the welding process. This

representation utilises a continuous line heat source. This representation can lead to the

generation of both 1D as well as 2D models, as is highlighted in Table 2. An example of 2D

modelling was first introduced by Rosenthal (1946), and can be written as:


29

T(x, y, t) − T0 =q

2πλh√vξexp (−

v(w + ξ)

2κ) √πκ , (15)

where:

w = x − vt, and ξ = √w2 + y2.

Zhang (1989) modified Equation (8) to represent thermal flow as constant through the

finite thickness, h, of the thin plate. However, the model was also modified to represent the

thermal field for a finite time of welding, t, and includes heat losses due to convection and

radiation. The modified equation can now be written as,

T(x, y, t) − T0 = ∫Q

4πλh(t − t′)exp (−

(x − vt′)2 + (y − y′)2

4κ(t − t′)− bt′) dt′

t

0

, (16)

where:

b =2(αc + αr)

cpρh.

Parameters αc and αr represent the respective heat transfer coefficients for the convection and

radiation mechanisms of cooling. As with point source models, the line heat source models

also predict infinite temperature at the heat source.

2.3.3 Advanced heat source models

As mentioned above, the point and line heat source models can reliably predict the

thermal history at some distances away from the source. However, the accuracy of analytical

models can be significantly improved, specifically, in the high temperature region, with the

representation of the welding heat input with Gaussian 1D, 2D or 3D volumetric distribution.

The power of the Gaussian heat source can be written as:


30

Q = ∫ q(y)dy

∞

−∞

, (17)

where, Q is the total power of the heat and q(y) is the power per unit length (Fassani and

Trevisan, 2003). In the 1D case, the Gaussian heat distribution is applied equally along the y

direction from the heat source centre, 0, see Fig. 15 (Radaj, 1992).

Figure 15: 1D Gaussian heat source.

The power per unit length q(y) can be written as:

q(y) = Qmaxexp(−Ay2), (18)

where Qmax is the maximum value of q(y), and A is the coefficient of arc concentration.

Coefficient A is determined with regard to the distance, yb, which corresponds to the distance

from the distribution centre C, where the power is reduced to 5% of its maximum value.

Coefficient A is written as:

A =3

yb2 . (19)

As yb increases, coefficient A and therefore q(y) decreases. As yb → 0, the Gaussian

distribution essentially reduces to a point heat idealisation.

q(y)

−y𝑏

Qmax

y y𝑏

0

2.5% 2.5%


31

The 2D Gaussian heat source was first introduced by Pavelic et al. (1969). The 2D

distribution is presented in Fig. 16 and can be written as (20):

Figure 16: 2D Gaussian heat source.

q(x, y) = Qmax exp(−kr2) =Qk

πexp(−kr2), (20)

where k (noted as 1 2σ2⁄ ) is the coefficient of arc concentration, also known as the

distribution parameter (higher values of k narrow the distribution and make it approach to a

point source idealisation). At r = 0, the heat density is equal to Qmax, therefore the Gaussian

heat source representation will predict a finite temperature in the vicinity of the heat source,

which better corresponds to the modelling expectations.

Goldak et al. (1984) further updated the Gaussian heat source by developing the most

seemingly appropriate heat source to represent a weld on a plate. Unlike the point, line and

Gaussian heat sources, Goldak et al.’s heat source considers the depth of the weld bead,

which represents the penetrative qualities of the weld bead itself (Fig. 17).

Qmaxexp (−1

2)

Qmax

σ

r

q(x, y)

x

y


32

Figure 17: Goldak et al.’s 3D heat source.

Goldak et al.’s heat source consists of two different elliptical parts, each representing the

front and back of the weld bead (Equations (21a) and (21b)).

q(x, y, z) =6√3ffQ

abcfπ√πexp (−

3x2

a2−

3y2

b2−

3z2

cf2

),

q(x, y, z) =6√3fbQ

abcbπ√πexp (−

3x2

a2−

3y2

b2−

3z2

cb2

),

(21a)

(21b)

where:

ff + fb = 2. (21c)

The resultant temperature fields for T(x, y, t), as well as T(x, y, z, t), due to the

respective instantaneous distributed temperature heat sources or moving heat sources q(y),

q(x, y) (Gaussian) and q(x, y, z) (Goldak et al.), can be found by using the principle of

superposition or, technically, by integration over the heat source area. The final equations are

quite lengthy and cumbersome (Eagar and Tsai, 1983; Nguyen, 2004; Nguyen et al., 2004),

and are not given in this thesis as they can be found elsewhere. As with the point and line

heat source thermal field models, the resultant thermal fields, based on Gaussian and Goldak

y

x

z cb

cf

b

a

q(x, y, z)


33

et al.’s heat sources, have also been validated by numerous studies (Eagar and Tsai, 1983;

Boo and Cho, 1990; Radaj, 1992; Fassani and Trevisan, 2003; Nguyen, 2004).

The point heat source models are much simpler than the relevant line or Gaussian heat

source representations of welding. The latter models require extensive experimental studies in

order to specify all parameters of these models. The utilisation of the simple point heat source

model in conjunction with experimental data from the relevant and limited experimental

studies can provide a very accurate evaluation of the thermal field (Eagar and Tsai, 1983;

Radaj, 1992). Therefore, it will be utilised in the current study to investigate the effects (i) the

V-groove local geometry and (ii) the pipeline welding procedure on the transient history field

of weld.

2.4 Summary and Research Gap

From the literature review provided above, it is demonstrated that there are a large

number of studies (analytical, numerical and experimental) that have investigated the

temperature field during welding. These investigations address the industry’s need for a

reliable evaluation of microstructure, thermally induced stresses and deformations, as well as

weld quality. However, there is little research on the effect of the local geometry of the weld

joint and procedures in pipeline welding on the thermal history of WM and HAZ. Moreover,

many sophisticated analytical and numerical thermal weld models utilise empirical values of

arc heat efficiency (Table 1), which can vary in a wide diapason. Therefore, the accuracy of

these models is significantly affected by the adopted values of the arc efficiency. However,

the selection of the appropriate values is totally at the discretion of the researcher or welding

engineer. The current work is directed to narrow this diapason and establish a link between

the arc efficiency and the most influential geometrical features of the local joint geometry

utilised in pipeline girth welding.


34

Another important aspect of modelling is the effect of the welding procedures adopted

in the pipeline industry on the thermal field in WM and HAZ. There are few (if any) research

papers, which have focused on the evaluation of the thermal field in actual pipeline welding.

Previous papers have usually disregarded the method by which the pipe is welded. The

method is expected to disturb the transient thermal field specifically at the location associated

with the start and end of the weld runs. There exist many different pipe welding procedures

(McAlister, 1998; Sacks and Bohnart, 2005) that can differ from the one described above and

they are illustrated in Fig. 2 (Fletcher and Yurioka, 1999; Fletcher and Piper, 2012). In many

practical situations, particularly for larger diameter pipes, the welding electrode is expected to

be consumed before the completion of the corresponding weld segments. This will interrupt

the continuous welding process and will influence the transient temperature field. The

previous studies largely ignored this aspect of welding practice. In the current work, a

mathematical model incorporating the realistic features of pipeline girth welding procedures

will be developed and the influence of these procedures on the transient thermal field will be

investigated. The current work will address this gap by evaluating the possible deviations in

the thermal history by using simplified analytical modelling. The analysis and conclusions

from the simulations can be used to improve quality of welding and to identify the critical

and high-risk locations for non-destructive defect evaluation inspections.

Chapter 3: Research Methodology

35


This research will address the gaps described in the previous chapter, namely the effect

of local geometry and pipeline welding procedures on the transient thermal field by

conducting theoretical and experimental studies. The underlying concepts of methodology

and research approach are summarised below.

3.1 Mathematical Modelling

A review of the mathematical models for simulating the transient thermal field was

presented in Section 2.3. Two main parameters of temperature history affecting the weld

quality, t8/5 and t100 cooling times, were identified in Section 2.1.2. In particular, t8/5 is the

time it takes for the weld seam and adjacent heat-affected zone to cool from 800 °C to 500 °C

and t100 is the time to reach 100 °C from a liquid state (weld pool). For the purpose of

predicting these cooling time characteristics, a simplified analytical modelling approach was

selected. Additionally, since the accurate prediction of the temperature field near the weld

pool is not a focus of the current thesis; the point heat source thermal models were selected

for the analysis of the effect of the local geometry and welding procedure on the transient

thermal field.

A 2D non-steady state point heat source analytical model is utilised to investigate the

effect that varying welding procedures has on the transient thermal field and characterise

cooling times in pipes. In order to model the thermal field in pipes, mirror images will be

utilised and are described below.

The method of mirror images is often applied to problems with finite geometries.

Examples of the application of this method can be found in Boo and Cho (1990), Nguyen


36

(2004) and Yurioka and Kojima (2004). In this study, the method of mirror images will be

used to develop the transient thermal field model for pipes, using the fundamental heat source

solution for an infinite plate (7). As an example, Fig. 18 illustrates the application of this

method and the fundamental solution for an infinite plate to derive the thermal field in a half-

plate.

Figure 18: Application of the method of Mirror Images to the fundamental solution (7).

The summation of the thermal field from the real and imaginary heat sources leads to

zero flux through the boundary. Equation (7) represents the thermal field due to an

instantaneous release of heat for the fundamental solution (which is given below, for the

reader’s convenience):

T(x, y, t) − To =Q

4πλtexp (−

r2

4κt). (7)

= Isolated boundary (no heat flux)

y

x

d

Real source Q at (0,+d)

r Observation Point

(xP, yP)

Original Problem

Equivalent Representation of the

Original Problem

y - mirror plane

y

x

d

d

Real source +Q at (0,+d)

Imaginary source −Q at (0,−d)

r′

Observation Point

(xP, yP)

r2

r1

No heat flux due to symmetry


37

The application of the method of images to the fundamental solution (7) and can be written

as:

T(x, y, t) − To =Q

4πλtexp (−

r12

4κt) +

Q

4πλtexp (−

r22

4κt), (22)

where:

r1 = √xP2 + (yP − d)2,

and,

r2 = √xP2 + (yP + d)2.

The thermal field in pipes can be modelled as a superposition of an infinite number of

imaginary heat sources producing no heat flow through the line of symmetry corresponding

to the current position of the heat source. This procedure will be described in Section 4.2.

3.2 Summary of Experimental Techniques

In this thesis, the experimental studies will be focusing on the investigation of the arc

efficiency and pipeline girth welding procedure. The experimental studies will generate data

to validate the selected, developed thermal field models for plates and pipes. In this study the

temperature data will be acquired using:

K-type thermocouples, an

R-type thermocouple and an

Infrared thermal camera.

The physical principles of temperature measurements and associated measurement techniques

will be described briefly in the following sections.


38

The majority of experimental studies in the area of welding have utilised

thermocouples for the evaluation of temperature history (Alam et al., 1999; Nguyen, 2004;

Attarha and Sattari-Far, 2011). Several studies have also applied infrared cameras to

investigate the thermal field during welding (Al Karawi and Schmidt, 2002; Camilleri et al.,

2004). A description of the operating principles of thermocouples will be presented next,

followed by a description of thermal imaging principles.

3.2.1.1 Operating Principles of Thermocouples

Large varieties of thermocouples are currently on the market and utilised for

measuring the temperature of solid and molten substances. Thermocouples usually consist of

a pair of dissimilar metal alloy conductors, housed in a ceramic insulator and fused at the tip,

as illustrated in Fig. 19.

Figure 19: Thermocouple diagram.

The principle of operation of thermocouples utilises a physical phenomenon known as

the Seebeck effect. The Seebeck effect relates to the following situation: when metal is

+ ve

- ve

Ceramic insulator Thermocouple tip

(junction)

Thermocouple tip

(junction) Ceramic insulator

+ ve

- ve

Dissimilar metal

alloy conductors


39

subjected to a thermal gradient, ∆T, there will be a voltage reading, ∆V, which can be written

as,

∆V = −S(T)∆T, (23)

where S(T) is a material property of the conducting metal, which, in general, is temperature

dependent.

The use of a pair of conductors with the same electro-thermal properties (S+ve(T) =

S−ve(T)) will produce an equal voltage difference of opposite polarity, which cancels out the

reading. For this reason, dissimilar conductors are used in thermocouples (S(T)+ve ≠

S(T)−ve) such that a non-zero voltage reading can be obtained. The induced voltage due to

the temperature gradient between the tip of the thermocouple and a reference location can be

written as:

V = ∫ (S+ve(T) − S−ve(T))dT

Ttip

Tref

, (24)

where Ttip is the temperature at the thermocouple tip and Tref is the temperature at the

reference location of both conductors. The thermocouple tip must have firm contact with a

solid or molten substance in order to measure its temperature. The thermocouple must also be

calibrated using a special procedure to convert the non-zero voltage reading to a temperature

reading.

The ceramic insulator of the thermocouple is usually covered in a stainless steel

sheath for a variety of enclosure options, as illustrated in Fig. 20. These different enclosures

are used for measurements of the temperature of the solid surface or liquids, including molten

metals.


40

Figure 20: Various types of thermocouple enclosure options.

3.2.1.2 Utilisation of thermocouples for the measurement of the temperature field during

welding

Figure 21: K-type thermocouple setup to record the thermal history of the welded plate

(Attarha and Sattari-Far, 2011).

As mentioned above, many previous experimental studies have utilised thermocouples

to measure the thermal field generated by welding (Alam et al., 1999; Nguyen, 2004; Attarha

and Sattari-Far, 2011). For temperature measurements, small diameter holes (typically less

than 1.5mm, and to a desired depth) are drilled at specified distances from the weld-line. The

Exposed tip Grounded tip Ungrounded tip

Ceramic

insulator

Stainless

steel sheath

Thermocouples

Weld


41

thermocouples are then inserted into these holes, making firm contact with the surface of the

bottom of the drilled hole. Thus, the temperature of the bottom of the hole is actually

measured during welding. Fig. 21 presents a typical setup of thermocouples used to record

temperature data.

Certain types of thermocouples, specifically R-types, can be used to measure the

temperature history of the weld metal from its molten stage. This is normally accomplished

by plunging the thermocouple behind the weld arc into the molten weld pool. The molten

pool solidifies around the thermocouple as it cools, see Fig.22. This technique is called

“harpooning” (Smith, 1974; Moore, 2003) and, in particular, allows for the measurement of

t8/5 cooling times. As described in Chapter 1, t8/5 cooling times are linked to the weld

microstructure and impact the quality and susceptibility to HACC of the weldment. This

technique requires a well-trained technician to perform the task. The success rate of plunging

is strongly dependent on the skills and experience of the technician or researcher.

Figure 22: Example of a plunged thermocouple in a weld seam (Moore, 2003).

In this research, both K and R-type thermocouples are used to record the thermal

history as a result of welding. The first type represents an inexpensive and versatile option.

K-type thermocouples consist of a pair of chromel and alumel conductors and can be used in


42

temperature ranges of -200 °C to 1350 °C. However, these types of thermocouples cannot be

used near the molten pool, where the temperature reaches 1500 °C to 1600 °C. R-type

thermocouples are normally manufactured from a platinum-rhodium alloy with a platinum

conductor pair. R-type thermocouples are quite expensive; however this type of thermocouple

can provide temperature measurements near or in the welded pool and withstand temperature

up to 1600 °C.

3.2.1.3 Isolation

Experimental equipment, which is necessary to measure and record temperature,

generally consists of the following components: a thermocouple, DAQ, PC and monitor, as

illustrated in Fig. 23.

Figure 23: Components of the temperature measurement and recording system.

The use of thermocouples has the potential to damage the recording equipment

permanently by exceeding the maximum allowable voltage limit of the DAQ. In welding

applications, the thermocouple (inserted in the plate, see Fig. 21) would also be exposed to

the welding current and voltage, which can result in a short circuit. To prevent electrical

damage to the temperature measurement system due to these mechanisms, the thermocouple

voltage has to be conditioned. Signal conditioning can be performed with a signal transmitter

or signal isolator. Typical examples of both devices are shown in Fig.24.

Thermocouple

DAQ

PC Monitor


43

Figure 24: Signal “Hockey Puck” Transmitter, a) and Signal Isolators, b)

(Ocean Controls, 2014; RS Australia, 2014).

Signal transmitters and isolators are generally affixed within some type of physical

housing (i.e. an “Isolation box”). Customised “Data Loggers” have also been developed

specifically to acquire temperatures for welding applications safely (CWT Inc., 2014; Hürner,

2014). These “Data Loggers” normally incorporate a DAQ, which is protected from any

excessive voltage.

Thus, there exist two options that can be applied to record the temperature history during

welding safely. These options are as follows:

1) Thermocouple → ”Isolation Box” → DAQ → PC → Monitor

2) Thermocouple → “Data Logger” → PC → Monitor

The second option is relatively expensive. The experimental study conducted as a part

of this thesis utilised the first option or “Isolation Box” to protect the DAQ from the

excessive voltage and short circuit damage.

a) b)


44

3.2.2 Principles of thermal imaging

Another method widely utilized for recording temperature history is thermal imaging,

which will be briefly discussed in this section. The electromagnetic spectrum includes

radiation from gamma rays, x-rays, ultra-violet (UV) light, a thin region of visible light,

infrared (IR) light, terahertz waves, microwaves and radio waves. All of these waves and rays

are characterised by the length of their wave (or wavelength), as illustrated in Fig. 25.

Figure 25: Wavelength sections within the Electromagnetic Spectrum (Heaviside, 2011).

Infrared waves reside in a small section within the electromagnetic spectrum (see Fig.

25). These waves carry the energy of the infrared spectrum, which is emitted by the object

under investigation through radiation. All objects with non-zero temperature (K) emit

radiation in the infra-red spectrum. The power of this radiation can be described by Stefan-

Boltzmann’s law (Camilleri et al., 2004), which can be written as,

P = εσ0A(T4 − T∞4 ), (25)

where σ0 = 5.6703 × 10−8 [W m-2

K-4

] is Stefan Boltzmann’s constant, P is the power

radiating from the area A. The percentage of radiation is termed as the emissivity ε and is


45

expressed as a fraction between 0 (for radiation emitted from an ideal white body) and 1 (for

radiation emitted from an ideal black body). Therefore, according to Equation (25), the ideal

black body emits the greatest amount of infrared radiation at a defined temperature. This

radiation can be detected with an infrared camera, similar to the way in which visible light is

detected with a photographic camera. The infrared camera can even work in pitch black

darkness as its operation is not dependent on ambient light, making the camera particularly

useful for night time rescue and underground operations.

3.2.2.1 Infrared radiation detection method

The intensity of the IR radiation is detected by an uncooled thermal sensor called a

microbolometer, which consists of an array of pixels (Orzanowski and Madura, 2010). A

single microbolometer pixel is presented in Fig. 26.

Figure 26: Microbolometer Pixel.

The surface of the pixel can be formed from a number of different absorbing

materials. The base of the pixel is usually made out of a silicon substrate and a readout

integrated circuit (ROIC), which can readily transmit electrical signals (Orzanowski and

Madura, 2010).

IR Wave IR absorbing

material

Reflective

layer

Electrode

leg

Read out circuit /

substrate

25μm

0.5μm 2μm


46

There are two types of thermal cameras; short wave (SW) and long wave (LW),

which absorb IR waves at the short and long wavelength section of the infrared region of the

electromagnetic spectrum, respectively. Typically, SW cameras absorb IR waves measuring

2.5 to 5×10-6

m, while LW cameras absorb IR waves measuring 7 to 14×10-6

m.

3.2.2.2 Safe image acquisition with windows (radiation filters)

The weld arc emits ultra-violet (UV), visible light and IR rays of length waves

between 0.26×10-6

m to 1×10-6

m (Extruflex, 2014), which can potentially damage the thin

absorbing material in the microbolometer. Therefore, to prevent damage to the

microbolometer, the thermal camera should only be operated behind an optical window (or

filter) with an anti-reflective (AR) coating (Al Karawi and Schmidt, 2002).

The optical window is usually fabricated from various types of elemental compounds

which absorb the smaller wavelengths emitted by the intensely bright arc flashes during

welding, thereby protecting the microbolometer from damage (Robinson, 2014). Windows

used for the protection of thermal cameras are normally made from Zinc Selenide (ZnSe)

(Fig. 27(a)), Germanium (Ge) (Fig. 27(b)), Potassium Bromide (KBr), Calcium Fluoride

(CaF2), Aluminium Oxide (Al203) and Barium Fluoride (BaF2).

Figure 27: Typical examples of ZnSe (a) and Ge (b) optical windows (Knight Optical, 2014).

(a) (b)


47

Fig. 28 presents transmissivity percentages of a variety of window materials against

wavelengths absorbed for SW and LW thermal cameras.

Figure 28: Typical transmissivity percentages of a variety of window materials against

wavelengths absorbed for SW and LW thermal cameras (Robinson, 2014).

According to Fig. 28, CaF2, BaF2 and Al2O3 windows are the best for SW thermal cameras,

whereas ZnSe and Ge windows are most appropriate for LW thermal cameras.

3.2.2.3 Thermal camera calibration technique

To ensure accurate data acquisition, calibration of the thermal camera is required

(Camilleri et al., 2004). Calibration is generally performed by comparing the thermal history

of a particular point (or points) obtained with a thermocouple(s) against the thermal history of

the same point(s) obtained with the infrared camera.

100

90

80

70

60

50

40

30

20

10

0

1 2 3 5 6 7 8 9 10 12 15 20

CaF2

Al2O3

Ge

ZnSe

BaF2 % T

ran

smis

sivit

y

Wavelength ×10-6

[m]

SW LW


48

Chapter 4: Development of Thermal Field Models for Pipeline Girth Welding

49

Chapter 4: Development of Thermal Field Models for Pipeline Girth

Welding

In this thesis the development of thermal field models for pipeline girth welding is

focused on two aspects of the problem: the effects of (1) local joint geometry and (2) pipeline

welding procedures on the transient thermal field. In this Chapter, two approaches are utilised

to develop the simplified predictive models, addressing each aspect, respectively, the:

I. Equivalent thickness approach, first introduced by Radaj, (1992) and further

developed by Yurioka (2004) and IQSim (2010), and

II. Mirror image method, described in the previous Chapter, which will be applied to

extend the known point heat source fundamental solutions for infinite plates to

pipe geometry as well as to different welding procedures.

This Chapter will briefly outline the above approaches and will present the development

of two analytical thermal field models. These models will be further utilised (1) to analyse the

transient thermal field during pipeline girth welding, (2) determine the arc efficiency for

various configurations of preparatory joints and (3) effect of welding procedures on the

transient thermal field during pipeline girth welding.

4.1 Incorporation of the local preparatory joint geometry into a modelling

approach

As described in Chapter 2 (Literature Review), the pipeline welding standards specify

the local geometry of the preparatory joint, which typically represents a V-groove shape for a

pipeline welded with the SMAW process, see Fig. 1. Due to cost-efficiency reasons, the

characteristic dimensions of the joint normally have a relatively wide range of tolerances to


50

avoid any delays associated with the preparation of pipe joints for welding and support the

high rate of pipeline construction. However, these tolerances can significantly affect the

welding conditions, in particular the arc efficiency, and, therefore, the transient thermal field.

In this chapter, analytical thermal models will be developed to assist with investigations of

the effects of arc efficiency and welding procedures on the temperature history, which are the

main objectives of the current thesis.

In the beginning, an effective thickness approach (Radaj, 1992; Yurioka, 2004) is

applied to incorporate the effect of the local geometry (V-shaped groove) on the transient

temperature field. The effective thickness for particular geometry dimensions will be derived

from the comparison of the experimental and analytical results, in particular, the cooling rate

from 800 °C to 500 °C (or t8/5). This is because the t8/5 parameter is widely adopted by the

industry to characterise the temperature history, as described in the literature review

(Terasaki et al., 1988; Radaj, 1992; British Standards, 2001; Nguyen, 2004; Yurioka, 2004).

In addition, this parameter (t8/5) is not significantly affected by other factors, such as weather

conditions and surface heat conductance, which are often unknown or unspecified in welding

tests (Kasuya and Yurioka, 1993). Furthermore, the same model will be applied to evaluate

the arc efficiency from the temperature history reported in the literature. Finally, an empirical

equation will be suggested to link the arc efficiency with the root gap size, which is

considered to have the largest effect on the arc efficiency.

4.1.1 Thermal field model

A number of simplified analytical models have been developed in the past to predict

the transient thermal field during various welding operations (Yurioka et al., 1986; Terasaki

et al., 1988; Kasuya and Yurioka, 1993; Trevisan and Fals, 1999; Yurioka and Kojima,

2004). Some of these models were briefly summarised in Chapter 2. Analytical thermal


51

models of welding processes are normally based on various assumptions and simplifications,

which are needed in order to derive a closed-form solution suitable for practical applications.

For the purpose of this investigation, Yurioka and Kojima’s thermal field model (Yurioka and

Kojima, 2004) (Equation (13), repeated for easy reference) has been selected to correlate the

experimental results and theoretical approach predictions. This model is based on the exact

solution to the problem of a point heat source moving with a constant velocity on the surface

of a plate of finite thickness. Strictly speaking, the model can only be applied to bead-on-

plate welding problems. However, by implementing the equivalent thickness approach

suggested by a number of authors (Radaj, 1992; Yurioka, 2004), and to be described later in

this section, this model can be extended to describe the temperature history for more

complicated geometries and welding conditions.

T(x, y, z, t) = T∞ + Tw(x, y, z, t) × exp (−2tUw

ρcph) + (Tph − T∞) exp (−

2tUP

ρcph), (13)

where:

Tw(x, y, z, t) = Q

2πλexp (−

vw

2κ)

× [exp (−

vrz

2κ )

rz+ ∑ rn (

exp (−vrn

2κ )

rn+

exp (−vrn

′

2κ)

rn′

)

∞

n=1

].

The selected analytical model was extensively validated by many researchers and is

often applied to predict various transient temperature characteristics of welding processes,

such as t8/5 cooling time (Kasuya and Yurioka, 1993; Yurioka, 2004; Yurioka and Kojima,

2004). This thermal characteristic is often used to evaluate the microstructure, hardness of the

weld metal and HAZ (Karkhin et al., 2006). The model (Eq. 13) represents the thermal field

for a plate. However, it can still be applied to pipe geometry in the high temperature range to


52

evaluate, for example, t8/5. This is due to the fact that the curvature of the pipe has a minimal

effect on the t8/5 cooling time of the weld metal. In other words, the radius of the curvature

is normally much larger than the characteristic size of the temperature affected zone in the

interval between 800 and 500 °C, which is typically limited by several millimetres from the

weldline.

4.1.2 Account for shape of V groove joint geometry: Equivalent thickness approach

As mentioned in Chapter 2, the relatively complex geometry of the V-groove joint

does not allow modelling of the thermal field with analytical approaches. The heat transfer in

V-grooves and flat plates are essentially different. Radaj (1992) has suggested, the effective

thickness approach, which can be applied to avoid this difference, and simplified analytical

models, such as (13), can still be used to predict the thermal field. The equivalent thickness in

this approach is defined as:

heq = h × f, (26)

where f is the geometry factor and heq is the effective thickness. Fig. 29 provides a

simplified illustration of the equivalent thickness approach.

Figure 29: Geometrical equivalence of the V groove and bead on plate welds with regard to

thermal distribution.

Essentially, the employment of the equivalent thickness, heq, in the analytical model is

intended to replicate the thermal conditions in the vicinity of the weld root for the real V-

heq

Q Q h Q Q


53

groove geometry at the same welding parameters such as heat input and welding speed as

would be performed on a bead on plate weld. The geometry factor, f, is different for different

V-groove geometries and has to be identified from a correlation with either the experimental

results or 3D numerical simulations (Radaj, 1992). The latter is a less reliable approach as it

attempts to theoretically model extremely complex non-linear and coupled phenomena

accompanying welding operations. Many of these phenomena are very difficult to formalize

and describe, even with contemporary numerical approaches and theoretical models.

Moreover, these approaches always require many material constants and functions, only

obtainable from experiments, which largely devalue the theoretical efforts.

4.2 Incorporation of pipeline girth welding procedure into modelling approach

A typical pipeline girth welding procedure was presented and generally described in the

Introduction (Ch. 1, Fig. 2) and shown in Fig. 30 for the reader’s convenience.

Figure 30: Pipeline girth welding procedure.

B

A

C

Pipe

Start location

End location

Welder 1

Welder 2

ϕ

Run 1 (A-C)

Run 2a (B-C)

Run 2b (A-B)


Runs in welding procedure


54

In Fig. 30, ϕ is the angle corresponding to the start/stop point B. The development of the

thermal model describing the transient temperature history corresponding to the specific

welding procedure will be presented in the following section.

4.2.1 Development of thermal field model

Figure 31: Schematic diagram to illustrate the mirror image method for pipes.

Most of the analytical models reviewed in Chapter 2 rely on the two-dimensional

simplification (2D + t) of the actual three-dimensional transient temperature field (3D + t)

(Eagar and Tsai, 1983; Boo and Cho, 1990; Fassani and Trevisan, 2003). However, some

researchers advocated the need for full three-dimensional considerations (Goldak et al., 1984;

Nguyen et al., 2004; Dong and Wei, 2006). Since the welding process involves very complex

physical phenomena, it is impossible to establish a single and generally accepted model or

analytical modelling approach. It is the intention of this investigation to develop and utilise a

model that represents a compromise between simplicity and accuracy. The accuracy of the

developed model can be evaluated by comparing the theoretical predictions with

y

x v

2πR 2R

2πR

2πR Imaginary source

Imaginary source

r y

x

Actual pipe geometry

Equivalent plate problem


55

experimental data. This model will be applied to evaluate the characteristics of the

temperature history at low temperatures or longer times, such as t100. At these conditions, the

effect of the local geometry is not significant and can be disregarded.

To take into account the pipe geometry, the thermal field due to a point heat source

can be obtained using the mirror image method; the fundamental point heat source solution

for infinite plate geometry. This is illustrated in Fig. 31, and represented mathematically in

Equation (27):

T(x, y, t) − T∞ =Q

4πλht∑ exp (−

x2 + (y + n2πR)2

4κt)

∞

n=−∞

, (27)

where R is the averaged pipe radius and n is the number of the imaginary heat source. The

transient temperature field due to a propagating heat source in a circumferential direction at

constant speed v, is given by the following equation:

T(x, y, t) − T∞ = ∫Q

4πλh(t − t′)∑ exp (−

x2 + (y − vt′ + n2πR)2

4κ(t − t′))

∞

n=−∞

dt′

t

0

. (28)

Equation (28) does not consider the possible pre-heat or heat losses through the pipe surface,

see Fig. 32.

The pipe surface temperature due to pre-heat, Tph, applied to a sufficiently large area

of the pipe can be expressed as (29):

T(t) = T∞ + (Tph − T∞) exp(−βpt). (29)

The pre-heat contribution to the pipe temperature history (29) and the heat losses through the

pipe surface can be incorporated into Equation (28), similar to Yurioka et al (1986) and

Yurioka and Kojima’s (2004) solutions. The combined transient temperature model for pipe


56

welding incorporating the pre-heat temperature, Tph, and heat convection at free surfaces can

be written as:

T(x, y, t) = T∞ + ∫Q

4πλh(t − t′)

t

0

∑ exp (−x2 + (y − vt′ + n2πR)2

4κ(t − t′)− βwt) dt′

∞

n=−∞

+ (Tph − T∞) exp(−βpt), (30)

where the constants βp and βw are the convection coefficients for the pipe and the weld, and

can be represented (Yurioka and Kojima, 2004) as:

βp =2Up

ρcph, (31a)

βw =2Uw

ρcph. (31b)

The developed Equation (30) is represented in Fig. 32. In particular cases, Equation (30) can

be reduced to the earlier derived temperature history models, such as Zhang (1989), Yurioka

et al (1986) and Yurioka and Kojima (2004).

Figure 32: Representation of a pipe model (Equation (30)) which incorporates heat loss at the

free boundary surface.

h

(0,0)

y

x

T0 = Tph

Up

T∞ Uw

v

R


57

In order to incorporate the effect of the welding procedure (see Fig. 30) on the

transient temperature field, the thermal fields from two heat sources Q1 and Q2 (two welders

each welding at respective speeds v1 and v2) are superimposed. Also, during the field pipe

welding, Welder 2 often experiences a short time delay between the completion of the first

run 2a to point C and the striking of the second run 2b at point A (see Fig. 30). This delay is

represented in the model by the delay time, td. The final transient temperature distribution,

taking into account the actual welding procedure, is presented in Equation (32).

T(x, y, t)

= ∫Q1


x2 + (y − v1t′ + n2πR)2

4κ(t − t′)− βwt)

∞

n=−∞

dt′

t

0

+ ∫Q2


x2 + (2πR − ϕR − y − v2t′ + n2πR)2

4κ(t − t′)− βwt)

∞

n=−∞

dt′

t

0

+ θ (t −πR − ϕR

v2− td) ∫

Q2

4πλh(t − t′)

t−td

πR−ϕR v2

× ∑ exp(

−x2 + (2πR − y − v2 (t′ −

πR − ϕR v2

) + n2πR )2

4κ(t − t′)− βwt

)dt′

∞

n=−∞

+ T∞ + (Tph − T∞) exp(−βpt),

(32)

where ϕ is the initial location of the weld deposition (point B) by Welder 2 (see Fig. 30).

Symbol θ represents the standard Heaviside step function. The circumferential linear

coordinate is, y, whereas, x is the distance from the weld centre line in an axial direction. It is

clear that the temperature field is symmetric with respect to y.


58

4.3 Chapter Summary

Two predictive models have been developed and presented in Sections 4.2 and 4.3 to

simulate the transient temperature field due to pipeline welding. The first model incorporates

the local geometry of the preparatory joint by utilising the equivalent thickness approach

suggested by Radaj (1992). This model is based on the solution presented by Yurioka and

Kojima (2004). The model will be applied to evaluate the effect of the geometry of the

preparatory joint on the arc efficiency. This will be accomplished by correlating the

modelling predictions with the experimental data obtained for various root gaps.

The second model is an extension of Zhang’s (1989) temperature model as well as,

Yurioka et al (1986) and Yurioka and Kojima’s (2004) approaches to incorporate the pre-heat

temperature and heat losses at the pipe’s free surface. The new model was obtained by using

the standard method of mirror images. This model accounts for a specific welding procedure

and will be applied later on to find the effect of the welding procedure on long cooling times,

such as t100. These cooling times are largely responsible for the weld quality and, therefore,

it is important to identify the possible critical locations with shorter cooling times, which

might compromise the integrity of the pipeline.

Chapter 5: Experimental studies

59


The experimental techniques used in the current study have been presented in Chapter 3

of this thesis. The current tests have been conducted to:

a) Validate the developed models,

b) Identify the effective thickness, heq, for V-groove joints (see Section 4.1.2) and

develop an empirical relationship between the root gap and arc efficiency.

This chapter will present the details of the experimental studies, selected examples of the

temperature history and will provide a summary of the main outcomes.

5.1 Experimental Equipment

The following sections describe the details of the equipment utilised in the

experimental study.

5.1.1 Welding machine and consumables

The welding of the plates and pipes was conducted with a Lincoln Electric Invertec

415V, 3 Phase welding machine (Fig. 33) and Lincoln Electric 5P+ Pipeliner E6010

cellulosic electrodes.


60

Figure 33: Lincoln Electric Invertec 415V, 3 Phase welding machine (WESS, 2014).

5.1.2 Setup of temperature data recording equipment

The “Insulation box” and DAQ setup (see Section 3.2.1.3) was utilised to acquire

temperature data through the use of thermocouples. The “Insulation box” housed 8 × Head

Mount Signal “Hockey Puck” Transmitters (PR Electronics 5331) (Fig. 34). The

thermocouples were wired to the “Isolation box”, which was connected to the NDC cable box

and integrated into the DAQ and PC, see Fig. 35.

Figure 34: Head Mount Signal “Hockey Puck” Transmitter (from PR Electronics 5331)

(RS Australia, 2014).


61

Figure 35: Equipment setup for recording thermal history with thermocouples.

The DAQ used in the experimental studies was a National Industries (NI) cDAQ 9188, which

housed 1 × analog input voltage module (NI 9215), with 4 channel inputs and 2 × analog

input universal modules (NI 9219), with 8 channel inputs (combined).

5.1.3 Software

To process and interpret the temperature data, a computer program was developed

using a LabVIEW v.10 environment for measuring and recording the temperature data. The

current experimental study also utilised the capabilities of LabVIEW v.10 to calibrate the K

and R-type thermocouples used in the tests.

5.1.4 Thermocouple Calibration

To ensure correct recording of the temperature history, all the thermocouples were

calibrated prior to use. During the calibration process, the thermocouples were exposed to a

range of temperature environments (within their specified operating range). A heated oven,

ambient air and a cup of cold water were used to create these temperature environments. The

Thermocouple “Isolation Box”

NDC Cable Box

DAQ

PC Monitor


62

Voltage (V) versus Temperature (T) relationship was recorded in each of these temperature

environments. The calibration demonstrated that this relationship is linear. The voltage-

temperature relationships for K and R-type thermocouples were:

For K-type thermocouples:

T(°C) = 298V + 298. (33a)

For R-type thermocouples:

T(°C) = 439V + 425. (33b)

Effectively, this calibration procedure has allowed for the transferral of the voltage

signal from the thermocouples into the temperature readings close to the thermocouple tip.

The sampling frequency for the K and R-types was set at 10 Hz.

5.1.5 Temperature data acquisition with Infrared Camera

The thermal imaging technique was used to collect the temperature history in the pipe

tests independently (see Section 3.2.2.3). Important aspects concerning the accuracy and safe

performance of the infrared camera were:

(a) Specifications and manual settings,

(b) Window (filter) selection,

(c) Transmissivity of the selected window,

(d) General camera setup.

The InfraTec VarioCAM hr has a sampling frequency of 1 Hz (InfraTec, 2010),

which means that the camera is able to capture thermal image sequences at 1 second

intervals. These thermal image sequences were processed later, using the camera software to


63

generate the thermal history of the weld metal and surrounding area. The camera has few

temperature range settings. In all tests it was set to capture thermal images in the 600 °C to

100 °C temperature range, in order to have an opportunity to evaluate two important

characteristics of the weld thermal history, t8/5 and t100 independently from the

thermocouples.

A Zinc Selenide (ZnSe) window (Fig. 36) with an Anti-Reflective (AR) coating was

selected to protect the thermal camera from the intense radiation exposure due to welding.

Figure 36: Fitted ZnSe window to rubber

manifold.

Figure 37: Transmissivity vs spectral range.

The AR coated ZnSe window (Fig. 36) was supplied by the manufacturer, Crystran

Ltd. UK. The window is designed to transmit radiation in a particular spectral range (7.5 to

14 μm). All other high intensity damaging IR and UV emissions that the weld arc produces

are not transmitted through this window. The average transmissivity versus the spectral range

of the AR coated ZnSe window was provided by the manufacturer and presented in Fig. 37.

The ZnSe window manifold was attached to the camera lens housing to protect the

camera’s IR detector (microbolometer) from the intense light and high temperature of the

10

9

8

7

6

5

4

3

2

1

0

7 14 Wavelength [μm]

Tra

nsm

issi

vit

y %

Average Transmissivity > 93%


64

weld arc (Fig. 38). The thermal camera was then affixed to the pivoting head of a tripod

stand, see Fig. 39.

Figure 38: Infrared Camera fitted with ZnSe window manifold.

Figure 39: Infrared camera affixed to tripod.


65

5.2 Plate Tests

The purpose of the plate test was to obtain temperature history, specifically, t8/5,

during the welding of a V-groove joint. The temperature data was acquired with K and R type

thermocouples.

The overall geometry of the test sample is shown in Fig. 40.

A mild steel plate is used to make the plate test sample (Fig. 40). The plate test sample has

the following details: (i) a bevelled groove angle of 60 degrees, commonly used in previous

studies (Sawhill et al., 1986; Noble and Pargeter, 1988; Alam et al., 1999), (ii) a root gap and

root height of 0.8mm, which corresponds to the minimum extreme in AS2885.2-2007. It is

known that smaller root gaps and root heights facilitate a faster weld travel speed, which is

important for pipeline construction cost. (iii) Diagonally drilled holes, parallel with the bevel

Figure 40: Plate test sample specifications. The R-type thermocouple is shown to illustrate

the temperature data acquisition technique.

Root Gap:

0.8 mm

Root Height:

0.8 mm

Nominal

thickness: h

Groove angle: 60°

Thermocouple hole

depth: h-1 mm

K-type Thermocouple

hole diameter: 1.2 mm

Thermocouple distance

from weldline

6 mm 3 mm 3 mm

R-type Thermocouple

30°


66

angle, at selected lengths of 6 mm, 9 mm and 12 mm from the centre line, are drilled in the

test samples to house K-type thermocouples.

During preliminary tests it was found to be very challenging for manual welding to

ensure a constant weld travel speed with very short lengths (the specimen’s width is only 50

mm). To achieve this speed, the effective length of the weld run was increased by tacking

mild steel run on and run off tabs of 80 mm each onto the weld sample, as illustrated in Fig.

41. After the addition of the run on and run off tabs, the total length of the plate test sample

was increased from 50 mm to 210 mm.

Figure 41: Top view of plate test sample with run on/run off tabs.

A vertical welding angle closely resembles the actual welding angles encountered in

pipeline welding (Coniglio et al., 2010). For this reason, the plate sample with tabs (Fig. 41)

was welded vertically. The plate sample with tabs was tacked onto a 20 mm thick mild steel

strongback (T joint). The strongback was placed onto a mount (100 mm mild steel angled

section), which was bolted to the jig. The strongback is held to the mount with a G-clamp.

Sample weld

length: 50 mm

Run on/Run off

tab length: 80 mm

Sample width: 150 mm Root Gap: 0.8 mm

Array of Thermocouple holes

Plate sample

(from Fig. 40)

Run off tab

Run on tab

Tack


67

The jig can move the plate sample up and down on a vertical axis at a constant speed v. The

setup of the plate test sample on the jig is presented in Fig. 42.

Figure 42: Plate sample with tabs mounted on welding jig.

The K-type thermocouples were inserted into the holes of the plate test sample and

connected to the “Isolation box”, which is part of the data acquisition equipment, as described

above. The R-type thermocouple was also connected to the “Isolation Box”. The schematic

picture of the plate test sample mounted on the jig (Fig. 42) is presented in Fig. 43.

G-clamp

Mount

Bolt

Jig

Strongback

Tack

Plate Sample with tabs

(from Fig. 41)


68

Figure 43: Complete plate test setup with data acquisition equipment.

In order to identify the geometry factor, f, and evaluate the equivalent thickness, heq,

a number of weld tests were conducted with V-shaped joints of varying nominal thicknesses,

h. Weld tests were conducted on the plate test sample setup, schematically shown in Fig. 43.

Also, as stated in Section 2.2, the root height and groove angle are unlikely to have an effect

on the thermal field history and therefore, these parameters were the same in all plate tests.

The characteristic dimensions of each plate test sample are given in Table 3.

“Isolation Box”

NDC Cable Box

DAQ

PC Monitor

K-type Thermocouples

R-type

Thermocouple


69

Table 3: Joint characteristics of the plate test samples.

Sample #1 Sample #2 Sample #3

Root Gap [mm] 0.8 0.8 0.8

Root Height [mm] 0.8 0.8 0.8

Nominal thickness, h [mm] 10 16 20

Root Angle 60° 60° 60°

The plate test samples in Table 3 were welded with the welding parameters presented in

Table 4.

Table 4: Welding parameters applied to each sample in the plate test.

Sample # (h [mm]) Weld speed, v [mm s-1

] Heat Input, HI [kJ mm-1

]

#1 (10 mm) 5.86 0.52

#1 (10 mm) 6.22 0.47

#1 (10 mm) 5.82 0.77

#1 (10 mm) 5.93 0.76

#2 (16 mm) 7.37 0.63

#2 (16 mm) 7.13 0.42

#2 (16 mm) 9.93 0.47

#2 (16 mm) 7.51 0.39

#3 (20 mm) 7.10 0.59

#3 (20 mm) 6.83 0.61

#3 (20 mm) 6.05 0.43

#3 (20 mm) 5.41 0.51

#3 (20 mm) 5.63 0.59

#3 (20 mm) 6.05 0.66

#3 (20 mm) 4.46 1.05

#3 (20 mm) 5.50 0.49

#3 (20 mm) 6.81 0.69

#3 (20 mm) 6.48 0.64

#3 (20 mm) 5.09 0.57


70

5.3 Pipe Tests

One of the key aims of the pipe tests was to experimentally evaluate the effect of the

welding procedure on the transient temperature history, as well as to validate the theoretical

predictions. For this purpose, the start angle, ϕ (see Fig. 30) pipe wall thickness, h, and pre-

heat temperature, Tph, were varied in these tests.

Figure 30: Pipeline girth welding procedure for model development (reproduced for easy

reference).

The temperature measurements in the pipe tests were collected at the critical locations

(the stop/start weld run locations at points A, B and C, see Fig. 30).

The welding electrodes used in the pipe tests were Lincoln Electric Pipeliner 5P+.

These electrodes are 350 mm in length and generally deposit a weld seam of approximately

350 mm in length. The pipe test samples were selected to have an outer diameter (OD) of

approximately 220 mm with circumferential length of π × OD = 691 mm, which is twice

the length of the weld seam deposited with a single electrode. Sufficiently long specimens,

each of 220 mm in length, were cut from bulk pipes with three different wall thicknesses of 6,

B

A

C

Pipe

Start location

End location

Welder 1

Welder 2

ϕ

Run 1 (A-C)

Run 2a (B-C)

Run 2b (A-B)


Runs in welding procedure


71

8.6 and 12.5 mm. The selected length of the specimens (220 mm) avoids the effect of the

final geometry of the specimen on the transient temperature history.

Figure 44: Local joint geometry specification of pipe test sample.

Figure 45: Axial locations of K-type thermocouples.

The weld run starts at point B and the thermocouples are located at A, B, C and D, as

specified in Fig. 45.

Thermocouple locations at

A - 0,

C - πR/2,

D - 3πR 2⁄

Thermocouple locations and

weld run start at

B30° - πR 6⁄ ,

B90° - πR 2⁄ ,

B90°

B30°

A

C

ϕ30°

ϕ90° D

y

R

Thermocouple

Root Gap: 0.8 mm

Root Height: 0.8 mm

Pipe wall thickness: h

Groove angle: 60°

Thermocouple hole depth: h-1 mm

Thermocouple hole diameter: 1.2 mm

Thermocouple distance from

weldline: 8 mm


72

Figure 46: Setup and data acquisition equipment for the pipe test.

“Isolation Box”

NDC Cable Box

DAQ

PC Monitor

K-type Thermocouples

Table stand Pipe test sample

Tripod

positioning

angle

Tripod

Infrared camera


73

The preparatory joint geometry of the pipe test samples was fabricated in accordance

with AS2885.2-2007 and had the following parameters: a groove angle of 60 degrees and

root gap and root height of 0.8 mm (as in the plate tests). Holes of 1.2 mm to house K-type

thermocouples were drilled straight down to a depth of 1 mm from the inner wall surface, 8

mm away from the weldline. A schematic picture of the local joint geometry of the pipe tests

sample is presented in Fig. 44. The locations of the drilled holes and thermocouples are

summarised in Fig. 45. The schematic diagram of the pipe tests is shown in Fig. 46.

Figures 47a and b show the experimental samples equipped with thermocouples and

Fig. 48 shows the overall view of the test samples and the location of the infrared camera.

Figures 47(a) and (b): Experimental setup of the pipe test sample.

(a)

(b)


74

Figure 48: Infrared camera and pipe test sample setup.

The dimensions of pipe test samples used in the pipe test are presented in Table 5.

Table 5: Dimensions of pipe test samples.

Sample #1P Sample #2P Sample #3P

Root Gap [mm] 0.8 0.8 0.8

Root Height [mm] 0.8 0.8 0.8

Nominal thickness, h [mm] 6 8.6 12.5

Root Angle 60° 60° 60°

Pipe outer diameter, OD [mm] 220 220 220

The pipe test samples were welded according to the pipeline welding procedure

(Chapter 1) for a variety of start/stop locations B (see Fig. 45), pre heat temperatures, Tph and


75

welding parameter combinations. The welding parameters applied to each pipe test sample

are presented in Table 6.

Table 6: Welding parameters applied to the pipe test samples in Table 5.

Pre-heat

Temperature

Weld speed, v of

welders 1 and 2

Heat Input, HI provided

by welders 1 and 2

Sample #

(h [mm]) B Tph [°C]

v1

[mm s-1

]

v2

[mm s-1

]

HI1

[kJ mm-1

]

HI2

[kJ mm-1

]

#1P (6 mm) B90° 25 6.4 7.02 0.56 0.46

#1P (6 mm) B90° 70 4.94 5.68 0.74 0.51

#1P (6 mm) B90° 100 5.37 6.91 0.6 0.48

#1P (6 mm) B30° 25 7.66 7.29 0.49 0.48

#2P (8.6 mm) B90° 25 5.28 6.65 0.69 0.5

#2P (8.6 mm) B90° 70 6.22 7.07 0.54 0.52

#2P (8.6 mm) B90° 100 6.03 6.83 0.69 0.54

#2P (8.6 mm) B30° 25 7.5 7.77 0.49 0.49

#3P (12.5 mm) B90° 25 6.78 7.82 0.61 0.49

#3P (12.5 mm) B90° 70 5.92 6.72 0.62 0.56

#3P (12.5 mm) B90° 100 4.85 6.76 0.73 0.56

#3P (12.5 mm) B30° 25 7.56 7.55 0.43 0.49

5.4 Selected examples of the recorded temperature history

In the plate test (Section 5.2), the temperature data history was collected with K-type

and R-type thermocouples. Fig. 49 shows the typical thermal history of the plate at locations

6mm and 9mm away from the weld centreline. It is seen from Fig. 50 that the thermal

histories of the K-type thermocouples in the near vicinity of the weld (6 mm and 9 mm away


76

from the weld centreline) correlate very well with the temperature data collected with the R-

type thermocouple plunged into the weld pool, especially below 300 °C.

Figure 49: Typical thermal histories acquired with the K and R-type thermocouples from the

plate test.

Figure 50: Thermal history of point B30°, see Fig. 45. The pipe is welded with the weld start

angle, ϕ30°, Tph = 25 °C and h = 6 mm.

0

300

600

900

1200

1500

0 50 100 150 200 250 300 350

Tem

per

atu

re [

°C]

time [s]

R-Type

K-Type (6mm away)

K-Type (9mm away)

0

100

200

300

400

0 100 200 300 400

Tem

per

ature

[°C

]

time [s]

Welding Parameters:



Plate thickness, h =6 mm

Welding Parameters:




R-type

K-type, (6 mm away)

K-type, (9 mm away)


77

This provides extra confidence in the experimental approach and indicates, for example, that

only the K-type thermocouples need to be used to obtain the temperature characteristics in

low temperature ranges, such as the t100 cooling time. Fig. 50 shows a typical temperature

history for the pipe sample.

The infrared camera was used to capture thermal images of the pipe during welding

and cooling for the purpose of generating independent measurements of temperature history,

which were used for the validation/support of data obtained with the thermocouples. These

images were processed with IRBIS 3.0 software and are presented in Fig. 51, for welding

(left image) and cooling (right image) stages of the pipe test, respectively.

Figure 51: Typical thermal images captured during welding (left image) and cooling (right

image) of the pipe test sample.

The thermal images captured during the welding and cooling of the pipe test sample

were compiled to create a thermal image sequence. The thermal image sequence was also

processed with IRBIS 3.0 software to generate the thermal history of a particular point of

interest. For example, Fig. 52 shows the thermal history at point B90. A careful comparison

between thermal histories obtained with thermal image sequences and K-type thermocouples

will be presented in Chapter 6.


78

Figure 52: Thermal history of thermal image sequence generated with IRBIS 3.0 of point

B90° on pipe welded with weld start angle, ϕ90°, Tph = 25 °C and h = 6 mm.

5.5 Chapter Summary

Experimental equipment, specimens and techniques used in the experimental study were

described in this chapter. Two types of specimens were successfully tested: plate and pipe

specimens. The objectives of the experimental study were (1) the validation of temperature

models developed in Chapter 4 as well as (2) an investigation of the size of the root gap of

the preparatory joint and (3) the welding procedure on the transient temperature field due to

pipeline girth welding. The welding parameters and the specimen geometries were selected to

reflect the actual conditions and geometries typically adopted in field pipeline welding. The

thermal histories for different specimens were recorded with K- and R-type thermocouples as

well as with an infrared camera. The recorded temperature histories with different techniques

demonstrated a high level of consistency, which provided confidence in the experimental

measurements. The outcomes of the experimental study will be utilised in the next two

chapters.

0

100

200

300

400

0 100 200 300 400

Tem

per

atu

re [

°C]

time [s]

Welding Parameters:




Chapter 6: Thermal Field Model for Pipeline Girth Welding

79


In this chapter the outcomes of this experimental study, as well as previous studies, will

be applied to investigate the effect of the variations in the sizes of the preparatory joint on arc

efficiency. This will be accomplished by correlating the temperature history as obtained from

mathematical modelling with the corresponding experimental data generated at various root

gaps in a high temperature range.

6.1 Evaluation of Thermal Arc Efficiency during Pipeline Girth Welding

The present approach to the evaluation of the arc efficiency is based on the analysis of

the experimental results at elevated temperatures, therefore, the averaged thermal parameters

corresponding to this temperature diapason are utilised in the theoretical modelling and the

thermal model described above. Table 7 specifies these parameters, which also have been

widely utilised in previous studies (Yurioka et al., 1986).

Table 7: General high temperature region thermal properties of most steels.

Thermal Properties

Thermal conductivity, λ [W m-1

K-1

] 25.13

Thermal Diffusivity, κ [m2

s-1

] 4.2 × 10-6

Density × Specific Heat, ρcp [J m-3

K-1

] 5.976 × 106

Overall heat transfer coefficients, Uw

and UP [W m-2

K-1

]

41.8

The thermal history as well as t8/5 cooling times obtained from the plate test (Section

5.2) were compared with Yurioka and Kojima’s modelled prediction (Equation (13) using


80

thermal properties in Table 7) for corresponding welding parameters, to identify the geometry

factors, f, for a range of specimen thicknesses detailed in Table 3. The best fit method for the

identification of the geometry factor, f, in accordance with the equivalent thickness approach,

was implemented with MathCAD software. The nominal thermal efficiency (η) for the

minimum root gap of 0.8 mm was set at 0.8, which corresponds to the averaged values of arc

efficiency for SMAW used in the previous studies (Nguyen, 2004).

Figure 53: Example of weld metal thermal history. Symbols represent experimental

measurements and the solid line is the theoretical prediction utilising Equations (13) and (26).

Fig. 53 shows a typical outcome of the utilisation of Equations (13) and (26)

(equivalent thickness approach) to best fit the experimental data. The summary of the

theoretical analysis is presented in Fig. 54, where the calculated values of t8/5 cooling times

are plotted against those experimentally measured with R-type thermocouples.


81

Figure 54: Calculated t8/5 cooling times with correction for local geometry (filled symbols)

and without (un-filled symbols) plotted against measured t8/5 cooling times of V groove

welding tests.

Fig. 54 shows two sets of results: the theoretical predictions (1) based on the equivalent

thickness approach (filled symbols) and (2) the predictions without taking into account the

local geometry (un-filled symbols). For the former approach, the correlation is much closer

and can be considered as very good, as the range of scattering is quite typical for this type of

experimental study (Chen and Wang, 2008).

The geometry factors, f and equivalent thicknesses, heq were determined for each

thickness, h. The results are presented in Table 8. Simple empirical relationships between the

equivalent and nominal thickness and the geometry factor are given in Equations (34) and

(35). It is important to highlight that these equations are only valid in the specified range of

nominal thicknesses, i.e. from 6.3 to 20 mm.

0

2

4

6

8

10

0 2 4 6 8 10

t 8/5

(ca

lcula

ted)

[s]

t8/5 (measured) [s]

Equivalent thickness approach

without Equivalent thickness approach

Correction for local geometry

Disregarding local geometry


82

Table 8: Geometry factors for various test piece thicknesses used in V groove welding tests.

Test piece thickness, h [mm] Geometry factor, f heq [mm]

20 0.35 7.00

16 0.42 6.72

10 0.65 6.50

heq = 0.048 × exp(0.132h) + 6.318, for 6.3 mm ≤ h ≤ 20 mm, (34)

f =0.048 × exp(0.132h) + 6.318

h,

(35)

where all dimensions in equations (34) and (35) are in millimetres.

To further validate the approach and empirical Equations (34) and (35), the outcomes

of a number of previous experimental studies, including Noble & Pargeter (1988) and Alam

et al. (1999), were analysed. Table 9 summarises the selected experiments conducted for

various nominal thicknesses and root gaps in the past.

Table 9: Determined weld arc efficiencies for V groove welds of various nominal thicknesses

and root gap.

Test Nominal thickness

h [mm]

Root gap RG [mm] Arc efficiency

η

Selected datum N/A 0.8 0.80

Noble & Pargeter (1988) 9.5 1.7 0.60

Noble & Pargeter (1988) 15.5 2.0 0.51

Alam et al. (1999) 8.6 1.5 0.68


83

Fig. 55 shows two sets of data: the theoretical predictions (1) based on the equivalent

thickness approach (filled symbols) and (2) the predictions without taking into account the

local geometry (un-filled symbols). Again, a much better correlation between the theoretical

and experimental results can be obtained if one implements the geometry factor, f, for

calculation of the temperature field in the high temperature range, which is significantly

affected by the local geometry.

Figure 55: Calculated t8/5 cooling times with the equivalent thickness approach and variable

arc efficiency (filled symbols) and without (un-filled symbols) plotted against measured t8/5

cooling times of previous V groove welding tests performed with various root gaps.

To achieve this correlation, the arc efficiency was evaluated using the best-fit method

for different root gap sizes. The empirical equation linking the size of the root gap, RG, and

arc efficiency, η, was found to be,

0

2

4

6

8

10

12

14

0 2 4 6 8 10 12 14

t 8/5

(ca

lcula

ted)

[s]

t8/5 (measured) [s]

0.8

1.5

1.7

2

0.8

1.5

1.7

2

RG [mm]


84

η = −(0.066)RG2.332 + 0.841, for 0.8 mm ≤ RG ≤ 2.0 mm. (36)

Thus, the equivalent thickness approach proposed by Radaj (1992) and Yurioka (2004)

was utilised to take into account the local V-shaped geometry of the preparatory joint. This

approach allowed for the investigation of the effect of the root gap on the arc efficiency, see

Equation (36). It can be seen from Equation (36) that the arc efficiency decreases as the root

gap length increases because a greater amount of heat from the arc is lost through a larger

root gap. The developed Equations (34) – (36) can be used to evaluate the t8/5 cooling time

using the analytical approach, which is much more versatile and much less expensive than the

experimental or numerical approaches. The steps in this analytical approach are:

Step 1: Calculation of the equivalent thickness, heq, based on the nominal plate thickness –

Equation (34)

Step 2: Calculation of the arc efficiency, η, based on the root gap size – Equation (36)

Step 3: Analytical analysis of thermal history using the calculated heq and η, using, Equation

(13).

It is important to highlight that the empirical relationships cannot be applied beyond the

specified limits. In this particular study the limits are: nominal plate thickness range 6.3 – 20

mm and root gap size variations from 0.8 – 2.0 mm.

Chapter 7: Effect of Welding Procedure on Thermal History

85


In this chapter the effect of the welding procedure on the transient thermal field will be

investigated. A particular focus will be placed on t100 cooling time, which is an important

factor in the HACC mechanism, as explained in Chapter 2. The calculations will be based on

the model developed in Chapter 4, which will first be validated against the results of

experimental study (Chapter 5).

7.1 Validation of pipeline welding procedure model with temperature data

Figure 45: Axial locations of K-type thermocouples (reproduced for convenience).

The modelling predictions for pipes utilised the equivalent thickness approach

(described in Section 6.1) to take into account the local weld preparatory geometry. Then, the

thermal pipe model (Equation (32) was used to generate the thermal history for selected

points on each pipe’s circumference. The modelled predictions were compared with the

B90°

B30°

A

C

ϕ30°

ϕ90° D

y

R

Thermocouple hole

Fixed Thermocouple locations

A - 0,

C - πR/2,

D - 3Rπ 2⁄

Thermocouple locations and

weld run start angle, ϕ

B30° - πR 6⁄ ,

B90° - πR 2⁄ ,

where,

ϕ30° - π 6⁄ ,

ϕ90° - π 2⁄


86

temperature data obtained from the pipe test (Section 5.3). The modelling predictions were

conducted for two weld start locations, ϕ30° or ϕ90° (see Fig. 45).

Fig. 56 shows the typical comparisons between the modelled predictions and the actual

experimental data from the pipe tests, described in Chapter 5, when the start location, ϕ is

changed from 30° to 90°.

Figure 56: Comparison of thermocouple measurements and modelling predictions for 220 OD

pipe welded with pipeline welding procedure ϕ30° and ϕ90° at B30° (a) and B90° (b)

respectively. Tph = 25 °C and h = 6 mm.

Fig. 57 presents a comparison of the modelling predictions and experimental results with

the weld start location at ϕ90° and the pipe wall thickness h = 12.5 mm. In this series of

predictions, the pre-heat, Tph is changed from 25 °C (Fig. 57(a)) to 70 °C (Fig. 57(b)) and

100 °C (Fig. 57(c)), respectively.

0

100

200

300

400

0 100 200 300 400

Tem

per

atu

re [

°C]

time [s]

Thermocouple data

Modelled prediction

0

100

200

300

400

0 100 200 300 400

Tem

per

atu

re [

°C]

time [s]

Thermocouple data

Modelled prediction

(a) (b)


87

Figure 57: Comparison of thermocouple readings and modelled predictions for 220 mm OD

pipe welded with pipeline welding procedure ϕ90°, h = 12.5 mm for Tph = 25 °C (a) 70 °C (b)

and 100 °C, respectively (c).

Figures 56 and 57 show a good agreement with the thermocouple data in the low

temperature range, i.e. below 200 °C. However, there are some discrepancies between the

theoretical predictions and measurements of the peak temperature near the weld-line. This

can be explained by various inconsistences in the welding speed and by difficulties of

depositing the weld along the centre line. In other words, in practice, there are always some

deviations of the weld deposition from the straight line, which could significantly affect the

temperature history in a high temperature range or at comparatively short times.

Another reason is that the two-dimensional thermal model utilised in this study is not

capable of capturing the temperature variations across the thickness, which can be significant

0

100

200

300

400

0 100 200 300 400

Tem

per

atu

re [

°C]

time [s]

Thermocouple data

Modelled prediction

0

100

200

300

400

0 100 200 300 400

Tem

per

atu

re [

°C]

time [s]

Thermocouple data

Modelled prediction

0

100

200

300

400

0 100 200 300 400

Tem

per

atu

re [

°C]

time [s]

Thermocouple data

Modelled prediction

(a) (b)

(c)


88

in short time intervals straight after the deposition. However, many previous experimental

and numerical results have demonstrated that these variations disappear at relatively low

temperatures due to conduction mechanisms.

The equivalent thickness is identified with the best fit method, as described in Chapter

6, and for pipe welding is described by the following equation:

heqII = 2.043 ln(h) + 2.362, 6 mm ≤ h ≤ 12.5 mm, 25 °C ≤ Tph ≤ 100 °C,

RG = 0.8 mm, D = 2R = 220 mm.

(37)

The calculation procedure of the transient temperature field for pipes is the same as for plates,

and it is outlined in Section 6.1.

7.1.1 Comparison of thermal histories obtained with different data acquisition techniques

The infrared camera was also used to capture thermal images of the pipe temperature

during welding and cooling with the purpose of providing (1) independent temperature

measurements, and (2) validation/support of the thermocouple readings. The thermal images

recorded every second were processed with IRBIS 3.0 software and are presented in Fig. 51,

for the welding and cooling stages of the pipe test, respectively.

The same software IRBIS 3.0 was applied to extract the temperature profile versus

time at the locations of interest (see Fig. 52). Fig. 58 presents a comparison of the thermal

history obtained with thermocouples and an infrared camera.


89

Figure 58: Comparison of thermocouple and infrared camera data thermal histories for pipe

welding procedure using weld start angle ϕ90° at B90°. Tph = 25 °C and h = 6 mm.

It can be seen from Fig. 58 that the infrared camera results agree quite well with the

thermocouple measurements, particularly in the low temperature region. However, there are

some differences at high temperatures. The reasons behind the inconsistencies can be

attributed to the temperature variations across the pipe wall thickness as the thermocouples

measure the temperature inside the pipe and the infrared camera captures the temperature

images at the surface of the pipe. At low temperatures, the through-the-thickness temperature

variations become negligible and both methods produce the same values for the temperature.

7.2 Temperature Variation across the Pipe Circumference

In this section, the variation of t100, which represents one of the important characteristics of

the temperature history, is investigated with the developed pipe model. Pre heat was not

applied to the pipe (i.e. Tph = 25 °C). Figures 59 and 60 show the t100 cooling time for a

range of heat inputs corresponding to pipeline girth welding specifications but with two

different weld run start points: at 30° and 90° (see Fig. 30). The nominal pipe thickness in

both simulations was set at 12.5 mm. The part of the pipe affected by the welding procedure

is greater (from A to C) when the weld starts at 90° (see Fig. 60), however, the intensity of

0

100

200

300

400

0 100 200 300 400

Tem

per

atu

re [

°C]

time [s]

Thermocouple data

Thermal camera data


90

these variations is slightly lesser for the case of 30°. In both cases, the variations of the

cooling times are quite significant and cannot be disregarded. These two graphs also identify

the higher risk locations with lower t100 along the pipe circumference, which can potentially

lead to structural defects, such as HACC.

In Figures 61 and 62 the same parameters were used to generate the results for 6 mm

pipes. The cooling time for 6 mm pipes is much longer and the relative variations are less

pronounced. This is consistent with HACC observations and empirical criteria, which

indicate that thicker pipes are much more susceptible to HACC than thin pipes welded with

the same welding parameters.

The effect of the thickness on the cooling time, t100, is summarised in Figures 63 and

64. Heat input is another parameter which significantly affects the cooling time. At high heat

inputs (0.8 kJ mm-1

), the relative variations in the temperature history are small; with a

decrease of heat input to lower values (0.4 kJ mm-1

) the relative variations of the t100

increase significantly (up to 400%), which can lead to a high level of quality variation of the

circumferential weld seam. The minimum cooling time in all the simulations was found

between points A and B, see Fig. 45. This part of the pipe can be considered as being

subjected to a higher risk of the presence of structural defects.


91

Figure 59: Cooling time t100 along the pipe circumference for ϕ30° and h = 12.5 mm.

Figure 60: Cooling time t100 along the pipe circumference for ϕ90° and h =12.5 mm.

0

100

200

300

400

500

0 90 180 270 360

0

100

200

300

400

500

0 90 180 270 360

Heat input

increasing

t100[sec]

ϕ30°

A C B A

Φ

Heat input

increasing

t100[sec]

ϕ90°

A C B A

Φ

Heat Input [kJ mm-1

]

0.4

0.6

0.8

Heat Input [kJ mm-1

]

0.4

0.6

0.8


92

Figure 61: Cooling time t100 along the pipe circumference for ϕ30° and h = 6 mm.

Figure 62: Cooling time t100 along the pipe circumference for ϕ90° and h = 6 mm.

0

100

200

300

400

500

600

700

0 90 180 270 360

0

100

200

300

400

500

600

700

0 90 180 270 360

Heat input

increasing

t100[sec]

ϕ30°

A C B A

Φ

Heat input

increasing

t100[sec]

ϕ90°

A C B A

Φ

Heat Input [kJ mm-1

]

0.4

0.6

0.8

Heat Input [kJ mm-1

]

0.4

0.6

0.8


93

Figure 63: Cooling time t100 along the pipe circumference for ϕ30° and heat input of 0. 8 kJ

mm-1

.

Figure 64: Cooling time t100 along the pipe circumference for ϕ90° and heat input of 0.4 kJ

mm-1

.

0

100

200

300

400

500

600

700

0 90 180 270 360

0

100

200

0 90 180 270 360

Pipe wall

thickness

increasing

t100[sec]

ϕ30°

A C B A

Φ

Pipe wall

thickness

increasing

t100[sec]

ϕ90°

A C B A

Φ

Pipe wall thickness [mm]

6

8.6

12.5

Pipe wall thickness [mm]

6

8.6

12.5


94

7.3 Chapter Summary

To investigate the effect of welding procedures on the transient temperature field, the

analytical model developed in Chapter 4 was extensively validated using the outcomes of the

experimental study outlined in Chapter 5. Despite the fact that the developed prediction

model for pipes produces some discrepancies in the high temperature range, it quite

accurately describes the variation of the temperature at low temperatures i.e. below 200 °C.

This affirmed that the analytical predictions will reflect the actual temperature variations

which often challenge pipeline integrity in pipes during girth pipeline welding,

A range of results were generated with the developed simplified analytical model for

typical welding parameters corresponding to the pipeline girth welding procedure: HI = 0.4 –

0.8 kJ mm-1

; Tph = 25 °C – 100 °C and pipe wall thickness from 6 to 12.5 mm. All modelling

results are consistent with the experimentally obtained tendencies. For example, the higher

pipe thickness, lower preheat temperatures and lower heat inputs significantly shorten the

t100 cooling time. The new results have indicated that the welding procedure has a significant

impact on relative variations of this characteristic time, specifically at low heat inputs, high

plate thicknesses and low pre-heat temperatures. The modelling results demonstrated that

there is a section of a pipe (specifically between points A and B), which is subjected to a

higher risk of the presence of structural defects due to HACC mechanisms. The practical

implementation of this particular result can result in directing Non-Destructive Testing

Technologies (NDTT) to focus on this pipe section at the specified parameters of the welding

procedure as well as pipe geometry. The specifics of these parameters and geometry could be

the subject of further investigations, which can be coupled with procedures and criteria for

the evaluation of the susceptibility to HACC.

Chapter 8: Overall Conclusion

95


The research presented in this thesis was focused on the investigations and prediction of

the effects of the local preparatory geometry and welding procedures on the transient thermal

field. The temperature history of the weldment has a significant impact on defect formation,

integrity and durability of the whole structure. The current research utilised both

experimental and theoretical methods to investigate these effects.

The careful literature review conducted in Chapter 2 allowed for an appropriate selection

of simplified analytical temperature models to address the purposes of the current study.

These models were extended to incorporate the effects of the V-shaped geometry of the

preparatory joint as well as the welding procedure on the transient thermal field due to girth

welding.

The first model was utilised to evaluate the effect of the root gap on the arc efficiency.

The outcomes of the modelling investigation were correlated with experimental data obtained

with flat plates as well as with the outcomes of the previous studies. The best-fit technique

provided an empirical relationship between the root gap and the weld arc efficiency. This

relationship can be used in analytical and numerical techniques for the theoretical evaluation

of the temperature history or thermal stresses. It was also shown that the variations of the arc

efficiency within the industry acceptable range of root gaps is significant and can be

responsible for the variations in the weld quality. It is recommended that large variations in

the root gap between the joints be avoided to achieve consistency and equally strong weld

joints.

The second model was used to evaluate the impact of the welding procedures on the

transient temperature field in pipes during pipeline girth welding. The temperature variations


96

are quite small for high heat inputs, relatively thin pipes and in the case of pre-heat, and,

probably, can be disregarded in most cases. However, in pipeline girth welding at no-preheat

and heat inputs of ~ 0.5 kJ mm-1

(which are typical for Australian conditions) the welding

procedure can have a significant impact on the temperature history. It was demonstrated in

the current thesis that for 12.5 mm pipes, the variation in the cooling time, t100, along the

pipe circumference could reach up to 30%. This large difference in cooling times, t100, is

caused by a much faster heat dissipation rate in thick pipes in comparison with thin pipes.

Essentially, the pipe thickness changes the geometry of the heat flow from 2D (relatively thin

pipes) to 3D (thick pipes), see Fig. 29. In addition, there is a pipe segment (section A to B)

which has a consistently shorter cooling time than the rest of the pipe. This implicates that the

non-destructive procedures should focus on this area, and consider this segment as having a

greater risk of containing critical structural defects. The analytical simulations were further

compared with full scale tests, specifically designed to investigate this problem

experimentally. Some discrepancies at high temperatures were found, which were attributed

to various reasons, such as the temperature gradient through the thickness, small deflections

of the weld from a straight line, travel speed variations, etc. The temperature field in pipes

was investigated with thermocouples and an infrared camera, which supported the

thermocouple data and provided greater confidence in the outcomes of the experimental

investigations.

From a practical point of view, it is recommended that the measurements of the

temperature history have to be made in the critical segment of the pipe. Some additional

measures could be undertaken to extend the cooling time in this segment such as shielding

from the wind, thermal isolation, etc.

Future work can incorporate the developed temperature models with various empirical

criteria or theoretical techniques, which have been developed previously for the evaluation of


97

weldability and the quality of the weld. The range of the pipe diameters and plate thicknesses

can be also extended to derive more comprehensive relationships between the local

preparatory geometry and arc efficiency. Currently, these relationships include only the size

of the root gap. It is argued that the development of such relationships will significantly

improve the predictive capabilities of analytical and numerical models.

One important step, which has not been completed in this thesis and could be a subject

for future research, is the validation of the theoretical predictions with the field measurements

obtained during pipeline construction. Unfortunately, there were no opportunities to complete

such measurements in the field. Any adaptation of the theoretical conclusions obtained with

the developed models by the pipeline industry has the potential to produce significant

outcomes in the future. It is believed that the outcomes of the current study will stimulate

further research into the evaluation of various effects and parameters, which are often

disregarded or ignored in theoretical calculations and modelling approaches. This includes,

but is not limited to, an investigation into the effects of the preparatory geometry and the way

the welding run is completed on the transient temperature field.

8.1 Publications from current research

The research approaches and findings presented in this thesis were also used to produce

the following publications;

1) Kotousov, A., Borkowski, K., Fletcher, L. and Ghomashchi, R. (2012). A model

of hydrogen assisted cold cracking in weld metal, 9th

International Pipeline

Conference, 24-28 September 2012, Calgary, Alberta, Canada.

2) Borkowski, K., Kotousov, A., Kurji, R. and Ghomashchi, R. Modelling the arc

efficiency of pipeline girth welding, (awaiting journal selection).

Overall Conclusion

98

References

99

References

Akbari, D. and Sattari-Far, I. (2009) Effect of the welding heat input on residual stresses in

butt-welds of dissimilar pipe joints, Int. J. Pressure Vessels and Piping, vol.86, no.11, pp.769-

776.

Al Karawi, J. and Schmidt, J. (2002) Application of Infrared thermography to the Analysis of

Welding processes, Institute of Fluid Dynamics and Thermodynamics.

Alam, N., Dunne, D. and Barbaro, F. (1999) Weld metal crack testing for high strength

cellulosic electrodes, First International Conference on Weld metal Hydrogen Cracking in

Pipeline Girth Welds. WTIA. Wollongong, Australia.

Anca, A., Cardona, A., Risso, J. and Fachinotti, V. D. (2011) Finite Element modelling of

welding processes, Applied Mathematical Modelling, vol.35, no.2, pp.688-707.

APIA (2010) Pipeline industry set for new multi-billion dollar Australian developments,

viewed 15 September 2014, from <http://www.apia.net.au/blog/2010/09/12/pipeline-

industry-set-for-new-multi-billion-dollar-australian-developments-apia-media-release-12-

september-2010/ >.

Attarha, M. J. and Sattari-Far, I. (2011) Study on welding temperature distribution in thin

welded plates through experimental measurements and finite element simulation, Journal of

Materials Processing Technology, vol.211, no.4, pp.688-694.

AWS (2007) AWS D10.11M/D10.11:2007: Guide for Root Pass Welding of Pipe Without

Backing.

http://www.apia.net.au/blog/2010/09/12/pipeline-industry-set-for-new-multi-billion-dollar-australian-developments-apia-media-release-12-september-2010/



References

100

Bailey, N., Coe, F. R., Gooch, T. G., Hart, P. H. M., Jenkins, N. and Pargeter, R. J. (1973)

Welding steels without hydrogen cracking, Abington Publishing.

Boo, K. S. and Cho, H. S. (1990) Transient temperature distribution in arc welding of finite

thickness plates, Proceedings of the Institution of Mechanical Engineers, vol.204, no.B3,

pp.175-183.

British Standards (2001) Welding - Recommendations for welding of metallic materials, BS

EN1011-2:2001, Part 2: Arc welding of ferritic steels.

Camilleri, D., Gray, T. and Comlekci, T. (2004) Use of thermography to calibrate fusion

welding procedures in virtual fabrication applications, Proceedings of Inframation 2004

Conference, pp.121-131.

Carslaw, H. S. and Jaeger, J. C. (1947) Conduction of Heat in Solids, Oxford, Clarendon

Press.

Cary, H. B. and Helzer, S. C. (2005) Modern Welding Technology, Upper Saddle River, New

Jersey, Pearson Education.

Chen, Y. and Wang, Y. Y. (2008) Thermal modeling of gas metal arc girth weld with narrow

groove, 8th International Conference on Trends in Welding Research. Pine Mountain,

Georgia, USA.

Christensen, N., Davies, V. and Gjermundsen, K. (1965) The distribution of temperature in

arc welding, British Welding Journal, vol.12, no.2, pp.54-75.

Coe, F. R. and Chano, Z. (1975), Welding Research International, vol.5, no.1, pp.33-90.

References

101

Coniglio, N., Linton, V. and Gamboa, E. (2010) Coating composition, weld parameter and

consumable conditioning effects on weld metal composition in shielded metal arc welding,

Science and Technology of Welding and Joining, vol.15, no.5, pp.361-368.

CWT Inc. (2014) WDL II Weld Data Logger, viewed 10 October 2014, from

<http://www.cweldtech.com/product-WDL2.html >.

Darmadi, D. B., Norrish, J. and Teiu, A. K. (2011) Analytic and finite element solutions for

temperature profiles in welding using varied heat source models, World Academy of

Science, Engineering and Technology, vol.81, no., pp.154-162.

Deng, D. and Murakawa, H. (2006) Numerical simulation of temperature field and residual

stress in multi-pass welds in stainless steel pipe and comparison with experimental

measurements, Computational Materials Science, vol.37, no.3, pp.269-277.

Dong, Z. B. and Wei, Y. H. (2006) Three dimensional modelling weld solidification cracks in

multipass welding, Theoretical and Applied Fracture Mechanics, vol.46, no.2, pp.156-165.

Dunstone, A. (2004) Numerical Modelling of Pipeline Construction, PhD, University of

Adelaide.

DuPont, J. N. and Marder, A. R. (1995) Thermal efficiency of arc welding processes,

Welding Journal, vol.74, no.12, pp.406s-416s.

Eagar, T. W. and Tsai, N. S. (1983) Temperature Fields produced by traveling distributed

heat sources, Welding Journal, vol.62, no.12, pp.346s-354s.

Extruflex (2014) Screenflex eye protection, viewed 28 September 2014, from

<http://www.extruflex.com/en/node/329 >.

http://www.cweldtech.com/product-WDL2.html

http://www.extruflex.com/en/node/329

References

102

Fassani, R. N. S. and Trevisan, O. V. (2003) Analytical modeling of multipass welding

process with distributed heat source, Journal of the Brazilian Society of Mechanical Sciences

and Engineering, vol.25, no.3, pp.302-305.

Feli, S., Aalami Aaleagha, M. E., Foroutan, M. and Borzabadi Farahani, E. (2011) Finite

Element Simulation of Welding Sequences Effect on Residual Stresses in Multipass Butt-

Welded Stainless Steel Pipes, Journal of Pressure Vessel Technology, vol.134, no.1,

pp.Article Number: 011209.

Fletcher, L. (2011) Stovepipe welding procedure (video clip), Wollongong, Australia, EP-

CRC, running time: 2 minutes.

Fletcher, L. and Piper, J. (2012) Resulting from discussion with authors at the Energy

Pipelines-CRC Steering Commitee Meeting 5th June 2012. University of Wollongong,

Wollongong, Australia.

Fletcher, L. and Yurioka, N. (1999) A holistic model of hydrogen cracking in pipeline girth

welding, First International Conference on Weld metal hydrogen cracking in pipeline girth

welds. WTIA. Wollongong, Australia.

Giedt, W. H., Fuerschbach, P. W. and Tallerico, L. N. (1987) GTA welding efficiency -

Calorimetric and temperature field measurements, 68th American Welding Society (AWS)

Annual Meeting.

Goldak, J., Chakravarti, A. and Bibby, M. (1984) A new finite element model for welding heat

sources, Metallurgical and Materials Transactions, vol.15, no.2, pp.299-305.

Heaviside, O. (2011) Electromagnetic Waves, Cambridge, UK, Cambridge University Press.

References

103

Henderson, I. D., Krishan, K. N. and Cantin, D. (1996) Investigation of field welding precise

limits for girth welding of thin-walled, high strength pipes, Final Report on CRC Project

96.32.

Hürner (2014) Hürner Whiteline SPG 2.0, viewed 10 October 2014, from

<http://huerner.de/en/portfolio/butt-welding/data-logging-systems/ >.

InfraTec (2010) VarioCAM high resolution User Manual.

IQSim. v2.0 (2010), Sør Trøndelag University College, Faculty of Technology, Norway.

Karkhin, V. A., Khomich, P. N. and Michailov, V. G. (2006) Prediction of Microstructure

and mechanical properties of weld metal with consideration for real weld geometry, 3rd

International Conference on Mathematical Modelling and Information Technologies in

Welding and Related Processes. E.O. Paton Electric Welding Institute of NASU.

Kasuya, T., Okumura, M. and Yurioka, N. (1995) Methods for predicting maximum hardness

of heat affected zone and selecting necessary preheat temperature for steel welding, Nippon

Steel Technical Report,no.65, pp.7-13.

Kasuya, T. and Yurioka, N. (1993) Prediction of welding thermal history by a comprehensive

solution, Welding Journal, vol.72, no.3, pp.107s-115s.

Keehan, E., Zachrisson, J. and Karlsson, L. (2010) Influence of cooling rate on

microstructure and properties of high strength steel weld metal, Science and Technology of

Welding and Joining, vol.15, no.3, pp.233-238.

Knight Optical (2014) Infrared Optics, viewed 15 September 2014, from

<https://www.knightoptical.com/stock/optical-components/infrared-optics/ >.

http://huerner.de/en/portfolio/butt-welding/data-logging-systems/

http://www.knightoptical.com/stock/optical-components/infrared-optics/

References

104

Lamit, L. G. (1981) Piping Systems: Drafting and Design, Prentice Hall Inc.

Lee, C. H., Chang, K. H. and Park, J. U. (2013) Three-dimensional finite element analysis of

residual stresses in dissimilar steel pipe welds, Nuclear Engineering and Design, vol.256,

no.3, pp.160-168.

Lindgren, L. E. (2001a) Finite element modelling and simulation of welding. Part 1.

Increased complexity, Journal of Thermal Stresses, vol.24, no., pp.141 - 192.

Lindgren, L. E. (2006) Computational Modelling of welding, Computer Methods in Applied

Mechanics and Engineering vol.195, no.48-49, pp.6710-6736.

Lindgren, L. E. (2007) Computational Modelling of Welding, Woodhead Publishing.

McAlister, E. W. (1998) Pipeline rules of thumb Handbook, Houston, Texas, USA, Gulf

Publishing Company.

Miller Welding Equipment (2014), viewed 15 November 2014, from

<http://www.millerwelds.com >.

Moore, P. L. (2003) Novel method of recording cooling curves during laser & laser/arc

hybrid welding, International Journal for the Joining of Materials, vol.15, no.4, pp.14-20.

Nayyar, M. L. (1992) Piping Handbook, McGraw-Hill Inc.

Nevasmaa, P. (2003) Predictive model for the prevention of weld metal cracking in high

strength multipass welds. University of Oulu, Oulu, Department of Mechanical Engineering.

Nguyen, N. T. (2004) Thermal Analysis of Welds, WITPress.

http://www.millerwelds.com/

References

105

Nguyen, N. T., Mai, Y. W., Simpson, S. and Ohta, A. (2004) Analytical Approximate

Solution for Double Ellipsoidal Heat Source in Finite Thick Plate, Welding Journal, vol.83,

no.3, pp.82s-93s.

Noble, D. N. and Pargeter, R. J. (1988) Field weldability of high strength pipeline steels.

Columbus, Ohio, Edison Welding Institute.

North, T. H., Rothwell, A. B., Glover, A. G. and Pick, R. J. (1982) Weldability of high

strength line pipe steels, Welding Journal, vol.61, no.8, pp.243s-257s.

Nunes, A. C. J. (1983) An extended Rosenthal weld model, Welding Journal, vol.62, no.6,

pp.165s-170s.

Ocean Controls (2014) Single Loop Isolator, viewed 13 October 2014, from

<http://oceancontrols.com.au/SIG-201.html >.

Onsoien, M. I., M'Hamdi, M. and Mo, A. (2009) A CCT diagram for an offshore pipeline

steel of X70 type, Welding Journal, vol.88, no.1, pp.1s-6s.

Orzanowski, T. and Madura, H. (2010) Test and evaluation of reference-based nonuniformity

correction methods for microbolometer infrared detectors, Opto-electric Review, vol.18,

no.1, pp.91-94.

Pavelic, V., Tanbakuchi, R., Uyehara, O. A. and Myers, P. S. (1969) Experimental and

computed temperature histories in Gas Tungsten arc welding of thin plates, Welding Journal,

vol.48, no.7, pp.295s-305s.

Radaj, D. (1992) Heat Effects of Welding, Springer Verlag.

http://oceancontrols.com.au/SIG-201.html

References

106

Robinson, M. (2014) A thermographer's guide to infrared windows, viewed 28 September

2014, from < http://www.irinfo.org/articles/article_7_1_2006_robinson.html >.

Rosenthal, D. (1946) The Theory of moving Heat Sources and its Application to Metal

Treatments, Trans. Am. Soc. Mech. Engrs. 68., vol.68, no., pp.849-866.

RS Australia (2014) 2-wire Programmable Transmitter 5331A3B, viewed 22 September

2014, from < http://au.rs-online.com/web/p/temperature-transmitters/7966900/ >.

Rykalin, N. N. (1957) Berechnung der Warmevorgange beim Schweissen, Berlin, VEB

Verlag Technik (Original: Raschety teplovykh protsessov pri svarke. Moscow: Mashgiz

1951).

Sacks, R. J. and Bohnart, E. R. (2005) Welding Principles and Practices, MacGraw Hill.

Sawhill, J. M., Baker, J. C. and Howe, P. (1986) Hydrogen-Assisted Cracking in High-

Strength Pipeline Steels, Welding Journal, vol.65, no.7, pp.175s-183s.

Smart, R. and Bilston, K. (1995) Stress upon partially completed girth welds during line-up

clamp removal and lowering off, WTIA/APIA Panel 7 Research Seminar, Welding high

strength thin walled pipelines, Wollongong, Australia.

Smith, F. J. (1974) Weld metal temperature measurement - Harpoon technique with

Rhodium-Platinum thermocouples, Platinum Metals Review, vol.18, no.2, pp.73.

Standards Australia (2007) AS2885.2-2007: Pipelines - Gas and Liquid Petroleum, Part 2:

Welding.

Suppiah, M. (1999) Weld metal cracking in cellulosic welds of X80 steel, Master of

Engineering (Hons), University of Wollongong.

http://www.irinfo.org/articles/article_7_1_2006_robinson.html

http://au.rs-online.com/web/p/temperature-transmitters/7966900/

References

107

Surhone, L. M., Timpledon, M. T. and Marseken, S. F. (2010) Shielded Metal Arc Welding,

VDM Publishing.

Svensson, L. E. (1994) Control of Microstructures and properties in steel arc welds, CRC

Press.

Terasaki, T., Akiyama, K., Ishimoto, K. and Yasuhiko, M. (1988) Proposal of Equations to

estimate the cooling time t8/5 from 800C to 500C, Quarterly Journal of the Japan Welding

Society, vol.6, no.2, pp.105-109.

The Joyce Road Neighbourhood (2012), viewed 18 September 2014, from <http://joyce-

road.blogspot.com.au/2012/02/pipeline-construction_11.html >.

Trevisan, R. E. and Fals, H. C. (1999) Fracture Modes and Acoustic Emission

Characteristics of Hydrogen-Assisted Cracking in High-Strength Low-Alloy Steel Weldment,

Journal of the Brazilian Society of Mechanical Science and Engineering, vol.21, no.4,

pp.675-682.

WESS (2014) VRD Safety Units viewed 22 September 2014, from

<http://www.wess.com.au/for-hire/browse/vrd-safety-units-(in-built-welders) >.

Winczek, J. (2010) Analytical solution to transient temperature field in a half-infinite body

caused by moving volumetric heat source, International Journal of Heat and Mass Transfer,

vol.53, no.25-26, pp.5774-5781.

Yurioka, N. (2004) The Japanese Welding Engineering Society, viewed 2012, from

<http://www-it.jwes.or.jp/weld_simulator/en/cal2.jsp >.

Yurioka, N. and Kojima, K. (2004) A predictive formula of weld metal tensile strength,

Quarterly Journal of the Japanese Welding Society, vol.22, no.1, pp.53-60 (Japanese).

http://joyce-road.blogspot.com.au/2012/02/pipeline-construction_11.html

http://joyce-road.blogspot.com.au/2012/02/pipeline-construction_11.html

http://www.wess.com.au/for-hire/browse/vrd-safety-units-(in-built-welders

http://www-it.jwes.or.jp/weld_simulator/en/cal2.jsp

References

108

Yurioka, N., Okumura, M., Kasuya, T. and Ohshita, S. (1986) Welding Note - Second

Edition. Kanagawa, Japan, R&D Laboratories-II, Nippon Steel.

Yurioka, N. and Suzuki, H. (1990) Hydrogen assisted cracking in C-Mn and low alloy steel

weldments, International Materials Reviews, vol.35 no.4, pp.217-248.

Zhang, W. Y. (1989) Heat Transfer in Welding, Beijing, China Machine Press (Chinese).

Documents

Experimental Study and Theoretical Modelling of Pipeline ... · Nomenclature T - Temperature ... V-groove preparatory joints as specified in (a) ... Heat loses and heat transfer in