Hybrid Laser Welding in API X65 and X70 Steels

THESIS

Presented in Partial Fulfillment of the Requirements for the Degree Master of Science in the Graduate School of The Ohio State University

By

Andrês Fabrício Fischdick Acuña

Graduate Program in Welding Engineering

The Ohio State University

2016

Master's Examination Committee:

Antonio J. Ramirez “Advisor”

John C. Lippold

Copyright by

Andrês Fabrício Fischdick Acuña

2016

Abstract

Hybrid laser welding presents an important advance in productivity due to high welding speeds. However, fast cooling rates are inherent to the process, affecting the resultant microstructures and joint performance. In this research, three API steels were welded using hybrid laser welding with three distinct preheating conditions. The specimens, which were obtained using one hybrid laser root pass and two other GMAW filling passes, were subjected to microstructural characterization and performance evaluation using hardness and toughness measurements. Incomplete joints with only the hybrid root pass and completed joints (root and filling passes) were evaluated. Hardness mapping revealed as the critical area the top portion of hybrid laser fusion zone, which was subsequently reheated by the GMAW filling pass. Optical and scanning electron microscopy revealed a bainitic-martensitic microstructure with the proportion of those two phases varying as a function of the preheating. Miniaturized Charpy V-notch testing was used to evaluate the local toughness and ductile-to-brittle transition of several regions within the joint.

Fractographic analysis confirmed the abrupt transition from ductile-to-brittle behavior. The localized fracture toughness testing showed an adequate joint performance for all tested conditions. Nevertheless, the hardness values meet the requirements only for higher preheating temperature conditions.

ii

Dedication

This document is dedicated to my family, my father and mother which taught me to always seek for knowledge, my beloved wife Raquel, my daughter Nicole and my son

Lucca which always supported me in everything.

iii

Acknowledgments

I want to acknowledge everyone who directly indirectly contributed to the development of this research and in special:

To God because everything comes from Him. “ You shall remember the Lord your God, for it is He who gives you power to get wealth, that he may confirm his covenant that he swore to your fathers, as it is this day.” Deuteronomy 8:18 ESV

To my wife Raquel, my daughter Nicole and my son Lucca, which supported me every day and embraced all the changes and challenges in their lives so we could start this master project abroad.

To my Mom and Dad for all their support and comprehension.

To the professor Antonio Ramirez for the guidance, support and help as my advisor.

To all the faculty and staff from the welding engineering department that made this work possible.

To all the graduate and undergraduate students, which shared so many ideas, advice tips and time to discuss the development of the work.

To Petrobras for funding all the tuition cost and support me in a foreign country, especially to Leonardo Paixão for all the time shared in the discussions and meetings with SENAI, and Gilmar Zacca for the help reviewing and discussing the thesis.

To EWI that not only allowed me to analyze some previous welds but also produced new

HLAW welds in the material and parameters required

To SENAI CTS Solda, which prepared part of my welds and specimens in Brazil only, guided by our digital meetings. iv

To NIST, especially to Jeffrey Sowards and Enrico Lucon for support and for the impact testing and data analysis.

To LAMEF from Universidade Federal do Rio Grande do Sul, for their support for the

SEM analyses.

v

Vita

2001...... Colégio Adventista de Esteio

2004...... Electronic Technician, Colégio Cristo

Redentor

2008...... B.S. Mechanic Engineering, Universidade

Federal do Rio Grande do Sul

2008...... Equipment Engineer, Petrobras

2009...... Terminals and Pipeline Engineering,

Petrobras University

2010...... Welding Engineering, Petrobras University

2014 to present ...... Graduate Student, Department of Materials

Science, The Ohio State University

Publications

ZACCA, G. ; MENEZES, M. S. ; ACUNA, A. F. F. ; SCHNEIDER, E. ; SILVEIRA, T.

L. ; ARAUJO, C. R. Welding and Heat Treatment of Heavy Wall API 5l X65 Sour

Service Pipes. 2014.

ZACCA, G. ; ACUNA, A. F. F. ; MENEZES, M. S. ; SCHNEIDER, E. ; SILVEIRA, T.

vi

L. Welding and heat treatment of heavy wall API 5L X65 sour service pipes. 2013.

MAGALHAES, V. A. N. ; VILARINHO, L. O. ; ACUNA, A. F. F. ; CARVALHO, L. P.

; FREITAS, J. C. EVALUATION OF CONVENTIONAL AND CONTROLLED

SHORT-CIRCUIT GMAW PROCESSES FOR ROOT PASS IN PIPE WELDING. 2013.

ACUNA, A. F. F. ; AWRUCH, A. M. Um modelo computacional de silos em estruturas metálicas. 2008.

ACUNA, A. F. F. ; AWRUCH, A. M. . Pontes ferroviárias metálicas: determinação do número de ciclos de carga significativos para análise de fadiga. 2007.

Fields of Study

Major Field: Welding Engineering

vii

Table of Contents

Abstract ...... ii

Dedication ...... iii

Acknowledgments...... iv

Vita ...... vi

Publications ...... vi

Fields of Study ...... vii

Table of Contents ...... viii

List of Tables ...... xii

List of Figures ...... xiii

Chapter 1: Introduction and Objectives ...... 1

1.1. Introduction ...... 1

1.2. Objectives ...... 3

Chapter 2: Literature review ...... 4

2.1. The Laser Welding for the Oil industry ...... 4

2.2. Laser Beam Welding (LBW) ...... 4

2.3. Welding Modes ...... 7 viii

2.3.1. Conduction Mode...... 8

2.3.2. Keyhole Mode Welding ...... 9

2.4. Hybrid Laser Welding ...... 13

2.5. The requirement for gap-bridging ...... 16

2.5.1. The API 5L ...... 17

2.5.2. The API 1104 ...... 19

2.6. Geometrical Characteristics of Hybrid Laser Weld ...... 20

2.7. Welding Parameters ...... 23

2.8. Typical HLAW Welded Microstructure on Steels ...... 26

2.9. The Hybrid Laser Welding on API steel JIP ...... 30

3.1. Materials used ...... 34

3.1.1. Base Materials ...... 34

3.1.2. Filler Metals ...... 35

3.2. Chemical Composition ...... 35

3.3. Welding Pre Heat ...... 35

3.4. Welding Matrix ...... 37

3.5. Welding System ...... 38

3.5.1. Material A (X65 Plate) Welding Parameters ...... 39

3.5.2. Material B (X65 Seamless Pipe) Welding Parameters ...... 40

ix

3.5.3. Material C (X70 UOE Pipe) Welding Parameters ...... 41

3.6. Thermal history ...... 41

3.7. Metallographic Analysis ...... 43

3.7.1. Optical Microscopy ...... 43

3.7.2. Scanning Electron Microscopy ...... 44

3.8. Hardness Mapping...... 44

3.9. Impact Testing ...... 45

3.10. Thermal simulation ...... 48

Chapter 4: Results and Discussion ...... 50

4.1. Development of the welded Joints ...... 50

4.2. Thermal History ...... 50

4.3. Chemical composition ...... 52

4.3.1. Dilution ...... 54

4.3.2. Continuous Cooling Transformation – CCT diagrams ...... 57

4.4. Hardness Mapping...... 62

4.5. Microscopy ...... 92

4.5.1. Base Material ...... 92

4.5.2. Welded Joints ...... 98

4.6. Impact Test ...... 125

x

4.6.1. MCVN specimen Fractography ...... 132

4.7. Computational Model ...... 137

Chapter 5: Feasibility ...... 140

5.1. Time per weld...... 141

5.2. Welds per day ...... 142

5.3. Overall cost per day...... 143

5.4. The SMAW Pipeline construction in Brazil ...... 143

5.5. Defects rate and rework...... 144

5.6. The GMAW Pipeline construction in Brazil ...... 145

Chapter 6: Conclusions ...... 146

Chapter 7: Future Work ...... 149

References ...... 150

xi

List of Tables

Table 1 - Weld Matrix showing condition, pass order, preheating temperature, and adopted nomenclature for specimens ...... 37

Table 2 - Measured chemical composition of the base metals (material A X65 Plate, material B X65 Seamless Pipe and material C X70 UOE Pipe) and the filler metals used,

(ER80S-Ni1 for materials A and B and ER80S used for the material C) ...... 53

Table 3 - Through-thickness chemical composition measurements of welds A3-FJ-300C-

300C, B3-FJ-300C-300C, and C3-FJ-300C-300C at locations J, K, and L...... 55

Table 4 - Base materials hardness average and standard error ...... 63

Table 5 - Hardness results summary in the 190-290HV0.2 scale ...... 64

Table 6 - Hardness results summary in the 190-380HV0.2 scale ...... 65

Table 7 - HLAW WCL Microstructure within the subcritical GMAW HAZ ...... 124

Table 8 - SEM images of KLST specimen fracture surfaces used on DBTT curves development ...... 133

Table 9 - MCVN DBTT and USE values put in perspective to full-size CVN through correlation reported by Lucon [49] ...... 136

xii

List of Figures

Figure 1 - Schematic of laser profile in both orientation (a) transverse and (b) longitudinal considering welding speed to the left, for a diffuse Gaussian intensity distribution (q) on the figure, presenting the vapor phase (v), molten material (l) and solid (s). From

Martukanitz [4]...... 9

Figure 2 - Schematic of laser profile in both orientation (a) transverse and (b) longitudinal considering welding speed to the left, for a concentrated Gaussian intensity distribution

(q) on the figure, presenting the vapor phase (v), molten material (l) and solid (s). From

Martukanitz [4] ...... 10

Figure 3 -Schematic view of Keyhole welding. (a) Simplified set, (b) real representation of keyhole shape by thermal distribution – From Principle of lasers materials processing

[5] ...... 13

Figure 4 - Effect of Laser and arc combination on current and voltage for (a) stable arc and (b) non-stable arc. From Steen [12] ...... 14

Figure 5 - HLAW pass cross-section, A - laser governed region, and B - Laser-ARC interaction zone, from Lei [20] ...... 21

Figure 6 - EWI's electrode mixing example, stainless steel electrode in carbon steel base material. From [2] ...... 23

Figure 7 - J Bevel of narrow groove joint configuration used on the welds. From [2] .... 31

Figure 8 - Through-thickness macro hardness indents, HV10. EWI [2] ...... 33 xiii

Figure 9 - Thermocouple disposition on the pipe inner surface on the material B weld. . 42

Figure 10 - Kleinstprobe KLST miniaturized Charpy V-notch specimen used on the

DBTT curves development from ISO 14556 [35] ...... 47

Figure 11 - Finite Element Method mesh used to model the welding steps for the filling passes on the software Sysweld®...... 49

Figure 12 - Acquired thermal cycles for the GMAW first filling pass on the B1-FJ-100C-

100C specimen...... 51

Figure 13 - Acquired thermal cycles of the GMAW second filling pass of the B1-FJ-

100C-100C specimen ...... 52

Figure 14 - Locations of chemical composition measurements, weld center line face. ... 55

Figure 15- Carbon measured content at locations J, K, L, and base material ...... 56

Figure 16 - Vanadium measured content at locations J, K, L, and base material ...... 57

Figure 17 - CCT diagram for material A, API X65 TMCP accelerated cooling plate ..... 58

Figure 18 - CCT diagram for material B, API X65 Seamless pipe ...... 59

Figure 19 - CCT diagram for material C, API X70 UOE pipe ...... 60

Figure 20 - CCT diagram for the ER80S-Ni1 filler metal used on welds of materials A and B...... 61

Figure 21 - CCT diagram for the ER80S filler metal used on weld of material C ...... 62

Figure 22 - Hardness map on the HLAW root pass of the material A using 300 °C preheating – A3-RP-300C – HV 0.2 ...... 66

xiv

Figure 23 - Hardness map on full joint of the material A (X65 rolled plate with accelerated cooling) using 300°C preheating for the root pass and the GMAW filling pass, - A3-FJ-300C-300C – HV 0.2 ...... 68

Figure 24 – Distance measurements, detail of hardness map A3-FJ-300C-300C specimen

– HV 0.2 ...... 70

Figure 25 - Hardness map on the full joint of the material B (X65 seamless pipe) using

100°C preheating for the HLAW root pass, - B1-RP-100C – HV 0.2 ...... 71

Figure 26 - Hardness map on full joint weld of the material B (X65 seamless pipe) using

100°C preheating for the HLAW root pass and for the GMAW filling passes, - B1-FJ-

100C-100C – HV 0.2 ...... 73

Figure 27 – Hardness map on full joint weld of the material B (X65 seamless pipe) using

300°C preheating for the HLAW root pass and no preheating on the GMAW filling passes, - B2-FJ-300C-No – HV 0.2 ...... 76

Figure 28 - Hardness map on root pass weld of the material B (X65 seamless pipe) using

300°C preheating for the HLAW root, - B3-RP-300C– HV 0.2 ...... 78

Figure 29 - Hardness map on full joint weld of the material B (X65 seamless pipe) using

300C preheating for the HLAW root and 300C for the GMAW filling passes, - B3-FJ-

300C-300C – HV 0.2 ...... 80

Figure 30 - Hardness map on full joint weld of the material C (X70 UOE pipe) using

100C preheating for the HLAW root pass, - C1-RP-100C – HV 0.2 ...... 82

xv

Figure 31 - Hardness map on full joint weld of the material C (X70 UOE pipe) using

100C preheating for the HLAW root and 100C for the GMAW filling passes, - C1-FJ-

100C-100C – HV 0.2 ...... 84

Figure 32 - Hardness map on full joint weld of the material C (X70 UOE pipe) using

300C preheating for the HLAW root and no preheating for the GMAW filling passes, -

C2-FJ-300C-No – HV 0.2 ...... 86

Figure 33 - Hardness map on full joint weld of the material C (X70 UOE pipe) using

300C preheating for the HLAW root, - C3-RP-300C – HV 0.2 ...... 88

Figure 34 - Hardness map on full joint weld of the material C (X70 UOE pipe) using

300C preheating for the HLAW root and for the GMAW filling passes, - C3-FJ-300C-

300C – HV 0.2 ...... 90

Figure 35 - Distance measurements, detail of hardness map C3-FJ-300C-300C specimen

– HV 0.2 ...... 91

Figure 36 - Material A microstructure, elongated granular bainitic structure ...... 93

Figure 37 - Material B microstructure, elongated granular bainitic structure ...... 94

Figure 38 - Material C microstructure, banded ferrite and perlite structure elongated in the rolling direction ...... 95

Figure 39 - Schematic representation of upper and lower bainite formation from

Bhadeshia [44] ...... 98

Figure 40 – A3-RP-100C Microstructure detail on the hardness map. Specimen etched with Nita 2%...... 100

xvi

Figure 41 – A3-FJ-300C-300C Microstructure detail on the hardness map. (a), (c) and (f) are from the HLAW fusion zone weld center line reheated by the GMAW filling passes,

(b) GMAW fusion zone, (d) is a half of HLAW root pass penetration region and (e) is the weld center line WCL on bottom of the root pass, specimen etched with Nital 2% ...... 101

Figure 42 - B1-RP-100C - Microstructure detail on the hardness map. Details (a), (b) and

(c) on the top of the HLAW root pass within the fusion zone at the weld center line, specimen etched with Nital 2%...... 103

Figure 43 - Chart extracted from Bain and Paxton, Alloying in Steels, ASM, 1961[48] 104

Figure 44 - SEM combination images of B1-RP-100C microstructure bainite and martensite detail 1 on bainitic grain and detail 2 in carbide aggregate between sheaves, specimen etched with Nital 2% ...... 106

Figure 45 - B1-FJ-100C-100C Microstructure detail on the hardness map. (a) HLAW fusion zone weld center line reheated by the GMAW filling passes, (b) HLAW fusion

Line, (c) Bainitic formation at WCL, (d) FGHAZ, (e) Fusion zone to CGHAZ transition and (f) is the weld center line WCL on bottom of the root pass, specimen etched with

Nital 2% are from the, ...... 107

Figure 46 - weld center line WCL on the bottom of the root pass, specimen etched with

Nital 2%...... 109

Figure 47 - B2-FJ-300C-No Microstructure detail on the hardness map. In (a) the GMAW

HAZ below A 1, (b), (c) and (e) presents the microstructure on WCL of HLAW root pass at distinct penetration the latter also presents martensite, (d) is the microstructure of the bottom of the root and (f) presents the transition from the fusion zone to the FGHAZ . 110

xvii

Figure 48 - B3-RP-100C Microstructure detail on the hardness map. (a) and (b) show the microstructure that would be heat treated in a subcritical regime by the GMAW filling pass, (c) presents the microstructure on the root bottom and (d) macro of the HLAW pass

...... 112

Figure 49 - B3-FJ-300C-300C Microstructure detail on the hardness map. In (a), (b) and

(c) bainitic structure on GMAW subcritical HAZ over HLAW root pass, (a) is presents using a polarizer filter, (d) and (e) presents the microstructure on the bottom of the and (f) is the transition from fusion zone to HAZ...... 114

Figure 50 - SEM combination images of B3-FJ-300C-300C Bainite microstructure, detail of a three-grain boundary encounter, specimen etched with Nital 2%...... 116

Figure 51 - C1-RP-100C Microstructure detail on the hardness map Bainite grain surrounded by Martensite...... 117

Figure 52 – C1-FJ-100C-100C Microstructure detail on the hardness map. In (a), (b) and

(c) structure on GMAW subcritical HAZ over HLAW root pass increasing magnification,

(d) macro of the weld cross section, (e) untempered martensite and bainite structure on the bottom of the root pass and (f) CGHAZ to FGHAZ transition...... 118

Figure 53 – C2-FJ-300C-No Microstructure detail on the hardness map. In (a), (b) and (c) bainitic structure on GMAW subcritical HAZ over HLAW root pass, (a) low magnification of the HLAW fusion zone, (d) macro of the weld cross section is and (e) presents the microstructure on the bottom of the and (f) is transition from fusion zone to

HAZ...... 120

xviii

Figure 54 - C3-RP-300C Microstructure detail on the hardness map, Bainitic structure in the center of the weld...... 121

Figure 55 - C3-FJ-300C-300C Microstructure detail on the hardness map. In (a) and (b)

Bainite on HLAW weld center line and (c) fusion line to HAZ transition (d) martensite formation on droplet attached to root, (e) bainite and Widmanstätten ferrite on root bottom part, (f) weld Macro cross-section...... 122

Figure 56 - DBTT KV values in degrees C, calculated by the impact energy...... 126

Figure 57 - DBTT LE values in degrees C, calculated by lateral expansion ...... 127

Figure 58 - USE value in Joules measured on miniaturized Charpy V-notch test using the

KLST specimen ...... 129

Figure 59 - Base Materials DBTT curves developed for the MCVN KLST specimen .. 130

Figure 60 - Full joint welds with 300C pre-heat condition DBTT curves developed for the

MCVN KLST specimen ...... 131

Figure 61 - Inclusion on fracture surface of B1-RP-100C specimen broken at -25C..... 134

Figure 62 - Detail dimples on A3-FJ-300C-300C specimen fracture surface, -48 °C. ... 135

Figure 63 - Numerically calculated thermal history plot on the nodes of 2, 3 and 4mm 138

Figure 64 - Numerical simulation of temperature fields applied by the filling passes ... 139

xix

Chapter 1: Introduction and Objectives

1.1. Introduction

Several industries such as oil and gas and shipbuilding present a high construction cost per day. Thus, an increase in the construction speed, productivity, is always the target.

Since welding is a key feature in industrial construction, the increase in welding speed is a constant demand in the search for productivity and cost reduction. This has pushed the development of new welding processes and technologies. In this scenario, laser welding outstands due to one of its more relevant advantages, high welding speed.

However, there are also challenges. The high energy density source and elevated welding speed produce rapid solidification and subsequent cooling, which can produce microstructures with lower toughness on some alloys. In addition, the laser welding requires precise fit-up, which raises the cost and might be not feasible within heavy industries. Therefore, a solution involving laser requires addressing the toughness fit-up issues. The Hybrid Laser Arc Welding (HLAW) process, which combines the advantages of the laser beam welding (LBW) and gas metal arc welding (GMAW) overcomes the presented challenges.

The use of HLAW provides high welding speed and robustness regarding joint fit- up while providing a joint with better toughness and overall good quality. However, the standards requirements for the oil and gas industry are challenging due to the critical nature of its products, gas, oil, fuels and corrosive products. In addition, the literature available 1 on HLAW has plenty of information on process modeling [1] , some of properties and microstructure. However, there are few studies on the use of process parameters to develop a more suitable microstructure.

There is a demand for research seeking to assess this welding process and its results in properties and microstructure for industrial applications. It is essential to consider standards requirements and the critical results as toughness and hardness. Since HLAW process bases its advantage on high welding speed it is important to account the influence of preheating as a parameter to final microstructure and, consequently, to toughness and hardness.

Results and joints from a previous study [2] on HLAW on API X65 pipeline steel plates were analyzed and used as start point for this study on similar materials, X65 and

X70 pipeline steel. The preheating was a parameter for root and filling passes. Root and filling passes were welded in combinations of room temperature, 100C° and 300C° preheat.

Microhardness maps were used to identify the critically elevated hardness regions.

Numerical simulation was used to evaluate the thermal history associated with the welding processes. Those critical regions microstructure was evaluated using optical and scanning electron microscopy. Miniaturized Charpy V-notch testing and fractographic analyses were used to evaluate the fracture toughness and ductile-to-brittle transition temperature of specific regions of such joints.

The results have shown predominant bainitic microstructure of HLAW fusion zone where hardness is the limiting property. The 300°C preheating produced hardness results meeting the NACE requirement. Small-scale impact test produced the DBTT curves which

2 have shown good toughness on all preheat conditions. The (DBTT KV ) values for the base materials were below -125°C and below -85°C for the welds. The fracture surfaces have shown a mixture of larger amount brittle mechanism and a small amount of ductile behavior at low impact energies while the pure ductile behavior was seen at high impact energies.

1.2. Objectives

This research had the following objectives:

− To study the use of hybrid laser arc welding process on currently used API pipeline

carbon steels.

− To evaluate the influence of preheating temperature on the final microstructure and

properties of the welded joints.

− To identify the most suitable tests or methodologies in order to better characterize and

evaluate welded joints produced by HLAW.

− To analyze the performance of the welded joints regarding standard requirements of

hardness and toughness.

3

Chapter 2: Literature review

2.1. The Laser Welding for the Oil industry

Welding is a key step for most industries. The Oil and Gas is one that is most based on welding. Pipelines, tanks, refinery structures, platforms, drilling system and others steps of its chain production are all manufactured with or by welding.

For a key step in its construction, the demand for improvement in welding productivity is endless. Therefore, there is a major interest on high productivity process, such as Laser Welding, Hybrid Laser Arc Welding, Tandem Gas Metal Arc Welding,

Friction Stir Welding, Radial Friction Welding, etc.

The Brazilian National Oil and Gas Company – Petrobras, and others [3] have classified the laser welding as one of the trends in high productivity welding processes for the future. Therefore, it has been mapping, sponsoring and developing research on laser welding area.

2.2. Laser Beam Welding (LBW)

The Laser beam welding – LBW – is a high energy density process using a laser generating system as a power source to provide heat in order to join materials by melting and coalescing it. It was developed following the establishment of the laser cutting in the end of sixties decade. Its key characteristic is the high energy density – in the order of 10 6

4

W/cm 2 intensities [4]. The energy density is obtained by focusing the beam in the desired area of the workpiece.

The Laser is generated in a resonator cavity made of two mirrors and a laser medium. It uses mirror reflection the laser medium to increase the energy beam while reflecting inside the cavity. The energy increase is due to continuous stimulation from a pumping system – external power source providing specific wavelength stimuli. The generated laser beam is optically transported and focused in the material to achieve the desired energy density. This concentrated heat is intense enough to produce melting and even vaporization. Once all those steps are fulfilled, the beam can move as a welding source

[5].

The result of concentrating this energy beam with high spatial and temporal coherence in a small spot produces intense heating. The vibration of the material electrons due to the photon oscillation electric field, this promotes local coalescence and finally causing interatomic bonding of the materials [6].

The beam projected area is defined as the spot size. It is an important parameter for the energy density since it defines the amount of energy and the interaction area of this energy. Therefore, the laser power and/or the spot size area can manipulate the energy density. For a defined beam power, the larger the spot size the smaller the energy density.

Due to its high energy density, the LBW is able to deliver much faster welding speeds combined with high penetration if compared to arc welding processes. With the latest developments in the laser generating systems, the laser sources become more

5 compact and have achieved high powers in the last fifteen years, with significant cost reduction. Nowadays 16kW and 20kW are commercially available for the industry.

The main advantages of laser welding when compared to conventional processes are:

• high penetration-width ratio,

• low heat input and highly localized heat affected zone,

• the non-contact process,

• the beam is not affected by magnetic fields and airflow,

• easy access to the welding region,

• ability in dissimilar welds,

• good control of power density by focusing,

• high welding speeds,

• high strength welds

However, the process also has limitations. The main ones are inherent consequences of some advantages.

− The small spot size, leads to the small-impinged area and even small

geometric variations of the joint could severely change its energy

distribution, hence, the LBW requires a high precision finishing and fit-up.

− It is an autogenous process; its fusion zone is composed only of the melted

and re-solidified base metal. In addition, it has poor gap bridging ability,

thus, volumetric changes such as gaps, misalignments, and wall thickness

variations will result in a lack of material to properly fill the joint. 6

− Even though the high welding speed is usually seen as an advantage, due to

the productivity increase. There is also an inherent critical characteristic

regarding its heat input since the heat source moves at high welding speed.

The heat extraction from conduction and convection promotes fast cooling

rates. The cooling rates are extremely important for materials which present

time-temperature dependent resulting microstructures. The thermal cycle

resulting from the laser welding is very challenging for the metallurgical

and mechanical results. Properties such as hardness and toughness might be

severely affected.

However, the biggest disadvantage for industrial application still is the high equipment cost. The growing demand for equipment has pushed the production and reduced the cost.

2.3. Welding Modes

The high energy density is the key characteristic of the laser welding, the reduced area receiving energy allows a localized heating and melting. That means that just a small section of the base metal is effectively affected by the laser thermal cycle (narrow heat affected zone).

There are two main Laser welding modes, based on heat flow ruled by an energy density and welding speed; conduction mode and keyhole mode. The conduction-welding mode occurs with lower density, there is not energy enough to produce metal vaporization.

The resulting welds have less penetration and a shallower energy distribution. However,

7 there is a gradual transition between the conditions for conduction or keyhole mode, not only the energy density and source wavelength but also material’s physical, thermal and optical properties [4], [7].

If the energy density at a given welding speed is enough to vaporize the metal, a cavity will form, due to mainly the recoil force of vapor. This cavity is capable of trapping the radiation and effectively distributing along the material thickness. Thus, it increases the energy effectively absorbed by the material. This cavity is called a keyhole and it gives the name to the mode, keyhole mode welding, known by the high deep-width ratio and for allowing extreme welding speeds [5].

2.3.1. Conduction Mode

As the name indicates this mode has its heat flow, on working piece, governed by thermal conduction. This is typical when using small energy densities in the order of kW/mm 2. If the energy amount is not enough to vaporize the material the keyhole cannot be established. Hence, reduced amount of energy is transferred for the workpiece. The melting will only start under the highest intensity of the laser profile, (as presented in Figure

1) [4] and the main heat extraction mode is conduction.

8

Figure 1 - Schematic of laser profile in both orientation (a) transverse and (b) longitudinal considering welding speed to the left, for a diffuse Gaussian intensity distribution (q) on the figure, presenting the vapor phase (v), molten material (l) and solid (s). From Martukanitz [4].

Trautmann [8] in his work in laser welding related the heat flow to the Peclet number, Pe. This is a way to measure the heat flow in the welding by relating the magnitude of the travel speed to the rate of thermal diffusion. In this mode, the Marangoni effect establishes a driving force that produces an outwards flow on the melting pool, in the direction of the cooler regions near the fusion boundaries [9].

The conduction mode offers the advantage of easily define processing conditions without defects. Therefore, it is extensively used in industries working with thin materials such as automotive, electronics and aerospace [4].

Since the conduction does not present the same energy density and therefore does not allow deep penetration, it is often used for thinner plates which are not the subject studied in this research.

2.3.2. Keyhole Mode Welding

9

The keyhole mode welding occurs for high energy densities processes, which are able to establish a cavity through the material due to its gases evaporation. The name, keyhole is due to its shape, since it is a deformed “cylinder” across the material thickness.

Hence, the energy is not just distributed on the material surface; rather the energy is delivered in a distribution across the material thickness with its peak in the beam focal region [5].

Martukanitz [4] in its review presented a schematic of the keyhole, reproduced in figure 2, which shows the geometric distribution of the phases. It is possible to compare it with the conduction mode. In Figure 2, it can be seen the differences in the amount of the vapor phase creating the cavity, the keyhole.

Figure 2 - Schematic of laser profile in both orientation (a) transverse and (b) longitudinal considering welding speed to the left, for a concentrated Gaussian intensity distribution (q) on the figure, presenting the vapor phase (v), molten material (l) and solid (s). From Martukanitz [4]

10

Since the laser is a light energy beam and metals have high reflectivity. A large portion of the incident beam power is reflected, metals also have high thermal conductivity.

As a consequence, the small amount of absorbed energy is rapidly extracted by conduction.

Therefore, in order to weld, the small-absorbed portion must be able to almost instantaneously melt the surface. This is achieved by focusing the power in a small spot size, and hence, the energy density needs to be increased to the order of MW/mm 2.

Kaplan [10, 11] have demonstrated that once the surface melts, there is a change in the reflection absorption balance. The absorption becomes dominant since the molten metal presents higher absorptivity than its solid state. The new liquid surface absorbs much more of the light energy [12], converting it into heat which increases the metal temperature above the boiling point, vaporizing more volatile elements such as Mg [13]. The vaporization causes a recoil force on the weld pool, which further depresses it forming a “hole” or

“cavity”. The Keyhole stability is based on a force balance between the vaporization pressure and the surface tension on the molten walls of the cavity [14].

The energy absorption on Laser welding is a complex mechanism composed of multiple effects.

A part that means gain for the heat input:

− energy absorption at the keyhole wall directly from the beam,

− absorption from the multi-reflected and scattered light,

− workpiece impinged surface absorption;

Another part, the losses from the energy beam:

− plasma absorption,

11

− beam reflection from workpiece surface,

− reflection and transmission through the keyhole [11].

Dowden [15] explained that a fully developed keyhole could capture almost all the incident laser. It acts as an optical black body where the radiation is captured in a sequence of multiple reflections within the cavity. This enhances its conversion from light into thermal energy. The vapor plume dissipates a portion of energy, but the biggest portion is still reflected within the channel, depositing energy in each reflection. The keyhole allows the narrow and deep weld due to the energy distribution along the channel. The high deep- width ratio characteristic for LBW is obtained due to the heat distribution through the keyhole cavity, which leads to a trapezoidal cross section generally with narrower root and a broader cap.

The keyhole is a deformed cylinder through the thickness. Since the vapor pressure varies through the cross section of the material and along beam radius. This effect combined with the advancing movement of the beam gives the weld pool an elliptical drop- like shape and the cavity has the shape of a keyhole. In addition, with this advancing welding speed and the heat distribution from the top to the bottom, the axis of the keyhole is bent in J shape, Figure 3(b).

12

Figure 3 -Schematic view of Keyhole welding. (a) Simplified set, (b) real representation of keyhole shape by thermal distribution – From Principle of lasers materials processing [5]

2.4. Hybrid Laser Welding

The Hybrid laser welding i.e. laser and arc welding combination were first proposed by Steen [12] in the final 70’s. With the objective to enhance the laser welding properties with a reduced cost. The combination proposed in his experiments were CO 2 Laser, which was the leading laser technology available at that time, and Gas Tungsten Arc welding power sources.

The initial motivation of hybrid laser welding was cost reduction since the early lasers were extremely expensive and presented low power. Therefore, the combination with an arc source targeted to add power without a huge increase in cost. However, the experiments demonstrated a synergic interaction between the arc and the laser. Allowing not only to combine the advantages but also to enhance its properties and to reduce the disadvantages.

The Figure 4 from Steen’s [12] work presents the result obtained by the interaction between the laser and the arc.

13

Figure 4 - Effect of Laser and arc combination on current and voltage for (a) stable arc and (b) non-stable arc. From Steen [12]

Steen [12] have shown that for a non-stable arc, the laser-arc interaction helps to stabilize it. On the other hand, for a stable arc, the arc resistance is reduced when combining arc and laser. In general, the welding speed and penetration increased if compared with the

LBW or the arc separately.

Matsuda [16] in 1988 tested Laser + GTAW and the Laser + GMAW combination determining parameters effects as arc-beam distance, focal position and laser beam power.

The Laser-GTAW combination produced less steep cooling rates than the Laser itself and achieved thick plates (maximum 8mm) welding. In addition, he verified that the distance between the electrode and the beam influences defects occurrence, (undercut and humping) which, the GTAW by itself would produce at high speeds. In order to achieve higher

14 penetration and avoid consuming the tungsten electrode, 3mm was the optimal spacing used. Ultimately, the Laser + GTAW achieved penetration 1.3-2.0 times greater than the

Laser alone. However, by using two autogenous processes the weakness of facing gaps and misalignments continues.

When using the GMAW and Laser combination 12mm thick plates were welded in a single pass. Which was not possible if welded by either the LBW or the GMAW itself.

In addition, Matsuda showed that in order to obtain the higher penetration the beam must be incident on the deepest region of the arc molten pool while the focal position should coincide the surface depression in the weld pool.

The target of developing a hybrid welding process is to enhance advantages of each process and overcome some of the disadvantages. Therefore, an effective process combination should present benefits from the interaction between the arc and the laser, maintain or enhance the advantages such as high welding speeds, reduced HAZ, and at the same time reduce the challenges such as gap bridging and misalignments.

In the early 90’s the GMAW was a process well established and widely used on industry, this reduced the uncertainties when making combinations with a distinct process.

In addition, the GMAW has the ability of continuously provide a clean filler metal, no flux is required and there is no slag formation. Hence, it became the most used arc process in combination with Laser Welding. Nowadays the term hybrid laser arc welding - HLAW refers directly to the laser beam and gas metal arc welding combination.

There is a specific interaction between the arc cathode and on the weld pool with the laser. The keyhole stabilizes the cathode position on the molten metal. Once the arc is

15 igniting it starts from the keyhole to the electrode [17]. There is a synergic effect when using the HLAW process in thick sections. Due to the filler metal deposit efficiency of the

GMAW, the use of narrow groove joints is possible with the HLAW.

Martukanitz [4] stated that the advantage of the HLAW is the ability to achieve high processing speeds and penetration from the laser beam welding, combined with the ability to accommodate gaps and fulfill misalignments with the filler metal addition which is related with the GMAW process.

The penetration of the HLAW is directly related to the keyhole and the welds are governed by the Laser Beam Welding parameters [17]. Hence, the HLAW welds present higher influence and advantages from the LBW.

2.5. The requirement for gap-bridging

As the LBW process relies on the high energy densities, this means a heat source focused in a small area. The fact of operating in narrow regions present challenges when welding large components. [4]

Precise fit-up and small tolerances are typically challenging for heavy industries, such as oil and gas. Large and bulky pieces present a fit-up challenge for many reasons, some of them are inherent to the size and weight of the pieces. Pipelines for example, due its aspect ratio, is highly susceptible to measurements variations that may affect the joint fit-up. In addition, the pipeline manufacturing process can severely interfere in its final dimensions and tolerances.

16

The challenge on fit-up is not noticeable in regular arc welding as it is for high energy density processes, due to the gap bridging ability and the broad heat distribution envelope of the arc. In contrast, the LBW is focused on a spot size smaller than one millimeter. In addition, for an autogenous welding process as the Laser, even small geometrical variations, (fit-up gaps, misalignments or not parallel root faces means volumetric defects) can cause the laser beam completely miss the joint, distribute the power unevenly or not being able to fill the joint volume.

In order to enhance the welding results, it is necessary to provide some gap bridging ability and/or ensure better geometric finishing tolerances for the pieces which will be welded.

Considering the pipeline construction sector, inside the Oil and gas industry, increase the quality of finishing is challenging since it implies deep changes in the standards. The API - American Petroleum Institute is the entity which regulates most of the standards for the oil industry, including the pipeline manufacturing standard – API 5L

[18], and the welding requirements API 1104 [19].

2.5.1. The API 5L

The API 5L is the standard that specifies pipeline manufacturing. Its requirements to classify the steel are more related to mechanical properties. Its classification is broad on chemical composition and there is no restriction of the microstructure. This standard also defines manufacturing requirements such as size, geometry, and tolerances for the pipes.

Part of the gaps and misalignments faced during welding have its origin on these tolerances.

17

A combination of the geometric factors, (out of roundness, thickness variation and out of squareness), on the manufactured pipe may lead to a poor joint fit-up.

The standard presents those geometric factors on chapter 9 “Acceptance Criteria” in its sub-chapter 9.11 “Dimensions, mass and tolerances”. The main source of the misalignments is related with wall thickness variations and out of roundness while the main source of gaps is related to out of squareness.

The permissible wall thickness range and its tolerances are specified as a function of diameter. However, there are manufacture steps such as, plate lamination, plate bending, pipe expansion for seamed pipes or the extrusion for seamless pipes, which stretches the material, and are sources of thickness variations. In addition, during the pipeline installation there is also bending and expansion steps those could severely alter the thickness and even the mechanical properties as ultimate tensile strength and yield stress.

Another issue due to manufacturing is the out of roundness, no pipe is perfectly round, and the manufacturing process plays a big role on pipe final shape. Once scanned with high resolution it reveals distinct shapes depending on manufacturing process, seamless pipe, longitudinal seam or helicoidally seamed. Therefore, each pipe has its own ellipsoidal cross section. In order to limit these variations, the API standard establishes an out-of-roundness tolerance range as a function of the pipe outer diameter. However, to achieve a proper joint using two randomly ellipsoidal sections is challenging and may compromise the welding process performance.

The out of squareness defines how the perpendicular is pipe cross-section plane regarding its longitudinal axle. This is critical for forming gaps during the fit-up and may

18 be incremented due the weld distortion on the previous sections or due the longitudinal weld.

It must be clear that this is a manufacturing standard. Hence, its tolerances are not made based on welding process requirements. Rather it is a sort of balance between the minimum production (machining and forming) cost and the generally required tolerances for a general welding process. That ultimately means that within this specification there always will be gaps and misalignments, therefore, in order to produce sound welds the welding process selected must be able to overcome those gaps and misalignments.

2.5.2. The API 1104

The welding joints requirements are specified in another standard, the API 1104

“Welding of Pipelines and related Facilities” [19]. This standard presents the requirements for mechanical properties, procedure qualification, and non-destructive evaluation.

It is important to consider the form of the base material, there is metallurgical, mechanical and even welding differences from welding plates and pipes. Pipe presents more accumulated strain due its manufacturing process, and therefore mechanical properties such toughness and yield strength may be distinct, also, the recrystallization is more likely to occur due to the stored energy. In addition, welding parameters developed for plates are not interchangeable with pipes the geometry and the gravity play a role in the weld pool stability, which will require some welding parameters correction. It is important to notice that for some research sources is much easier and less expensive to obtain plates

19 than pipelines. However, in other cases pipelines are available while the same base material in plate form is not.

Regarding the fit-up misalignment the API 1104, at chapter 7, “Design and

Preparation of a Joint for Production Welding” on item 7.2 defines a maximum value of

3mm. It is possible to have a non-specified joint (by the API 1104 criteria) even using two pipes that meet all the manufacturing requirements the (API 5L) requirements, due to the tolerances combination. Besides, there is the joint preparation process which in the field is really challenging to control it on a millimeter scale.

Due the pipe manufacturing process and the fit-up procedure in the pipeline field, is virtually impossible to completely eliminate the gaps and misalignments of the joints.

Therefore, the welding process applied has to be able to tolerate some gap. This practical requirement almost eliminates the use of autogenous process as the laser welding, since some extra material is required to fill the joint.

Using hybrid laser welding it is possible to get the benefit of high welding speeds and the high deep-width ratio relying on the GMAW to provide the filler metal. However, there are still fit-up requirements but not as tight as for autogenous welding. The tradeoff is to achieve welding speeds five times faster as the expense of a more precise fit-up.

2.6. Geometrical Characteristics of Hybrid Laser Weld

In order to have a better characterization of the weld is easier to define geometric regions of the weld bead. The hybrid configuration presents a specific heat distribution due its combination of the laser as a much-localized heat source, which deep penetrates in the

20 material due that keyhole ability to trap the energy and arc with a broader and shallower heat distribution.

Lei et al. [20] proposed the distinction of the weld bead in two regions as presented on the Figure 5, (A) with governed mainly by the laser source, defined as laser zone and a second one (B) where the interaction between laser and arc heat sources broadens the top of weld pool generating a conic region, defined as arc zone. He defined geometric parameters of the weld bead where W1 and H1 are the maximum weld width and depth, respectively, of the top part of the weld (A) and the W2 and H2 are the maximum weld width and depth, respectively, of the bottom part of the weld (B).

B

A

Figure 5 - HLAW pass cross-section, A - laser governed region, and B - Laser-ARC interaction zone, from Lei [20]

Lei [20] demonstrated that those geometric parameters H and W may change as a function of heat distribution. Therefore, variables such source power, welding speed, joint

21 preparation and surface tension will affect the resulting shape and microstructure of the weld bead. In his study, a pipeline was welded and the weld bead profile and properties were studied as a function of the welding positions.

There is gradient on the chemical composition through the material thickness. Since the heat source is not horizontally symmetric and the filler metal deposition on the bottom of the root is not as effective as on weld top. This leads to a quasi-autogenous region near the bottom of the root and a more diluted region on the top.

The EWI in previous research developed a dissimilar weld using stainless steel filler metal on welding of carbon steel base material in order to show the effect of electrode mixing [2]. Figure 6present the etched cross-section of this dissimilar weld, due the distinct response of the etchant is easy to notice the gradient of the electrode mixing through the thickness of the plate, that variation is critical when properties are expected in specific regions, such as on the bottom of the root.

22

Figure 6 - EWI's electrode mixing example, stainless steel electrode in carbon steel base material. From [2]

The Figure 6 shows that there is a quasi-autogenous region on the bottom (A, laser region defined by Lei), the filler metal provided by the arc process almost do not reach the bottom so the material in this region is basically the same base metal, a dilution gradient can be found increasing in the direction of the top.

On the other hand, at the top of the root pass, (B – arc zone defined in Lei’s work), the influence of the arc energy and the filler metal is more intense. There is a mixed zone since the dilution increases as well the width of the fusion zone since the heat source is acting in a larger area.

2.7. Welding Parameters

23

Spot size: Spot size is the diameter of the beam within the focusing region, this parameter combined with the beam power will provide the energy density.

Laser Power: power is the parameter that controls most of the weld energy. It is highly dependent on the laser power source. The development of laser power sources on the last 15 years, increasing power, efficiency, and robustness is the main reason for the type of applications available nowadays [3].

Laser Heat Input: Since the energy is concentrated in a small area and the travel speeds are high, the applied energy per length on the welding piece is low. The direct consequence of the low heat input is the reduced distortion and a narrow heat affected zone.

The reduced distortion plays a big role for industries as shipbuilding and automotive where precise alignment is critical. However, the narrow HAZ presents positive and negative sides. While the HAZ zones are reduced in size which reduces the extension of critical regions on another hand the properties transition from the weld metal to the HAZ and finally to the base metal is abrupt which can compromise the joint performance in some mechanical properties as toughness and tensile stress.

Welding speed; one of the advantageous characteristics of the laser beam welding are the high welding speeds achieved, which is also related to the high energy density feature. The energy amount in the spot area is able to vaporize quasi-instantaneously the metal establishing the keyhole. Once the keyhole is stable it traps and absorbs most of the energy of the beam, which increases the energy available to melt and vaporizes the material. Therefore, the welding travel speed can be increased, as long the keyhole remains stable. Welding speeds up to 8m/min in API pipeline materials have been reported on

24 research from Cranfield University [21-23]. However, extremely high welding travel speeds may lead to weldability problems such as lack of fusion, solidification cracking, and humping. Most research for industrial applications use the welding speed parameters in a range from 1m/min up to 2.5m/min Gruenenwald [24], which still is remarkable if compared with the current travel speed of manual arc processes such 0.3m/min or even the automated systems, which are around 1m/min. In practical evaluation high welding speeds directly related with productivity, this is one of the main parameters for industrial production, weld faster means manufacture time reduction and cost reduction.

Arc Current: Effectively controls the melting rate of the wire which is deposited on the joint, affects the arc and its metallic transfer mode of the electrode.

Arc Voltage: Controls the arc stability and once combined with the current determines the power applied by the arc on the weld.

Arc Heat Input: Analogous to the Laser Heat input defines the energy per length unit imposed during the welding.

The distance between beam and electrode: This is not defined as a parameter.

However, Wei et. al. [25] have demonstrated trough simulation and experimental procedure how the distance between the beam and the electrode can significantly affect the results. In their study the laser arc separation was changed from 1 mm to 5 mm with all other parameters kept constant, they have shown that even for the same heat input, the heat distribution is distinct with the higher spacing configuration presenting a larger welding pool and as consequence of its hotter condition a reduced cooling rates which ultimately

25 affects the resulting microstructure to the point that with 1mm a 64%volume fraction of martensite was measured and with 5mm spacing no martensitic content was found.

2.8. Typical HLAW Welded Microstructure on Steels

Steels present a resulting weld microstructure very dependent on the chemical composition and on the cooling cycle. In addition, the lower the cooling heat input the faster is the cooling rate, since there is less energy for heat extraction therefore it is expected that a process with low heat input such as laser welding will produce high cooling rates.

Since it is a fusion welding process on steel the austenite-ferrite transformation will eventually occur, due its steep cooling rates those microstructures that are promptly formed on fast cooling will predominate, typically bainitic ferrite, martensite, and some polygonal ferrite.

Miranda [21-23], compared the autogenous LBW with Gas Tungsten Arc Welding

GTAW at distinct heat inputs, low 160J/mm and high 960J/mm for both processes, welding an API - American Petroleum Institute X100 grade steel. For the low heat input, both processes formed mainly martensite and bainite; however, the laser weld produced a thinner microstructure that improves the toughness. At high heat input, the GTAW does produces a softer (325HV 10 ) microstructure mainly constituted of ferrite and perlite while the Laser also is claimed to present the same microstructures but in a narrower area and a maximum hardness level of 375HV 10 on the fusion zone.

26

Moore [26] et al. studied specifically microstructure and properties for autogenous and hybrid laser welds in six distinct pipeline steels. The hybrid laser used metal cored filler metal in order to promote acicular ferrite nucleation since the powder particles carry a higher oxygen content in oxide form to nucleate the acicular ferrite AF phase and therefore increase its toughness properties.

On the autogenous welds shows the influence only from the base material composition. In this group only one material presented primarily acicular ferrite which was only the on with a higher combination of oxides particles promoters as O and Ti. All other five materials resulted in predominant ferrite with aligned phases FS or AC, mainly bainite and Widmanstätten ferrite.

In another hand, the hybrid laser welds were executed with lower heat input which also affects the cooling rates, however, the greater difference on weld microstructure came from the addition of an arc heat source and filler metal. As expected, the oxide layer on the powder particles promoted a greater number of oxides inclusions which acts as nuclei sites for intragranular solidification of acicular ferrite. However, the predominant microstructure was ferrite with aligned phases and the results on toughness did not present significant differences even on those materials with acicular ferrite presence.

Moore’s most important conclusion is regarding hardness and toughness discrepancy since almost all materials failed to meet the requirement of weld metal hardness, even when the toughness requirements were meet. The hardness limits for non- sour were defined regarding hydrogen induced cracking for arc welding, even more critical for cellulosic electrodes used on SMAW. However, for laser/arc hybrid welding the

27 hydrogen content is controlled and the hardness standard limits are extremely challenging to meet for high energy density processes due to its inherent rapid cooling rates.

The microstructure found by Lei [20] was classified in the two zones: region A - laser zone and region B - arc zone, Figure 5 defined in his work. The laser zone was more complex since its compressed in a narrow area and it has received the even faster cooling rate than the arc zone. The structure was classified as upper bainite nucleating from the prior austenitic grain boundaries, acicular ferrite, and granular bainite. The arc zone consisted of columnar polygonal ferrite grains growing from the fusion line, acicular ferrite and some polygonal ferrite on the prior austenite grains. Liu [27] et al. used the same classification of arc and laser zone to compare the resulting microstructure and hardness on sintered powder metallurgy steels successfully achieving 10 mm penetration.

A more technological approach was developed by Wei [25] which developed a complete simulation to address and predict the cooling rates, weld bead profile, and microstructure. This simulation was then compared with the real welds on DH-36 steel presenting and good agreement between the simulation predictions and the welds regarding the weld profile and the microstructure phase proportion.

An important point addressed in his research was the influence of the beam to arc separation distance were the same nominal heat input and welding parameters might form distinct heat distribution. Increasing the separation distance while in the same weld pool,

(otherwise, it becomes a tandem process rather than a hybrid), generates a longer and wider molten pool which spread the heat distribution and makes the cooling rates slower effectively influencing the microstructure.

28

Gruenenwald [24] et al. has welded API X65 with autogenous laser and API X70 with HLAW, materials commonly applied in the industry and metal core filler metals which should promote better quality results. Since the materials employed were less challenging in terms of weldability he obtained promising results, a maximum hardness of

256HV 10 for the X65 material in 9.5 mm thick plates, and for X70 steel, the results were near 350HV 10 which would be enough for onshore applications.

Rethmeier [28] et al. tested single pass and multi-pass conditions up to 32mm thick joint also varying misalignment and gaps for X65 butt welds in a plate shape. The results showed 1mm vertical misalignment without parameters change and up to 2mm misalignment and 0.3 mm gap through parameters correction. The fast 2.5m/min welding speeds lead to steep cooling rates which resulted in 280HV 0.5 for the lower carbon material

(0.04%wt) and 410HV 0.5 for the high carbon material (0.08%wt), in order to reduce this value a 160°C preheating condition was used and achieved 330HV 0.5 .

A continuation of Rethmeier’s work was done by Gook [29] et al were API X65

36in diameter and 16mm thick pipeline were welded varying the preheat condition.

Parameters were varied as a function of the orbital welding position on the pipeline. His results on hardness reduced from 380HV 0.5 with no preheat condition to 280HV 0.5 at 200°C and finally 240HV 0.5 at 300°C preheat condition. However, the toughness values did not exceed 40J at 0°C and its microstructure was martensitic with some bainite formation.

Later Gook [30] tested four different filler metals on new pipeline API X80 and

API X120 grade steels for seam welds. The X80 presented a ferritic-bainitic microstructure

29 and the X120 a fully bainitic. It was interesting that the toughness produced with the metal cored wires were consistently superior to the values achieved with solid wire filler metals.

Roepke [31] studied the HLAW on HY-80 bainitic/martensite base material which counted on the filler metal influence to for acicular ferrite. A comparison between cooling rates through the ∆t8-5 and its correspondent microstructure for the LBW, HLAW and

GMAW as 1.9s, 6s, and 5.8s respectively. A large combination of preheating and heat input was tested and achieved 95% of acicular ferrite.

2.9. The Hybrid Laser Welding on API steel JIP

One of the motivations and references for this research was a previous development which PETROBRAS, along with other sponsors companies have funded. A Joint Industry

Program (JIP) research developed by the Edison Welding Institute – EWI on hybrid laser welding for API carbon steels focused on the HLAW root pass [2] since this is most the critical weld step for pipelines applications.

The JIP started in 2012 with the specific objectives, develop HLAW root pass that result in acceptable NACE hardness levels (248HV max) and acceptable toughness (36.6J

(27 ft-lb) at -30°C (-22°F)) on X65 material with the thickest joint face possible, also checking it regarding the common oil and gas industry standards.

The 15KW IPG fiber laser available allowed to target a 15 mm deep root pass on a

19mm API X65 bainitic steel plates. This material and welds were used later in this dissertation study defined as Material A in chapters 3 and 4. In addition, the welding parameters used were as a reference since the welds in other steels were supposed to

30 reproduce the heat input used in this JIP welds. The joint used a narrow groove preparation in a J bevel with 15 mm root face, as presented in Figure 7.

Figure 7 - J Bevel of narrow groove joint configuration used on the welds. From [2]

The weld was composed of three passes, an HLAW, and two GMAW filling passes.

All the welding passes root and filling ones, the filler metal Lincoln Electric LA-75

(ER80S-Ni1) electrode, 1.2 mm (0.045-in.) diameter.

In order to evaluate the weld mechanical properties, the American Petroleum

Institute - API 1104 standard requirements were used. Hence, the API tests were combined with some additional tests regarding toughness due the research objectives, the tests applied on the JIP are listed below:

• Transverse tensile test; 31

• Bend test; • Nick-break test; • Hardness testing; • Charpy V-Notch (CVN) testing at weld metal (WM) and heat affected zone (HAZ), full size and sub size (three levels through the thickness); • Crack Tip Opening Displacement (CTOD) testing at weld metal (WM) and heat affected zone (HAZ);

Initially, a short pre-test only for hardness and the CVN tests were applied in three root passes penetrations 6 mm, 12 mm, and 15 mm, since one of the objectives was to obtain approved welds with the deepest pass possible and those properties were equivalent in all welds, only the 15 mm penetration was used on the rest of the study.

Since it is known that the thermal cycle of HLAW presents high cooling rates and to obtain low hardness is a challenge, in order to meet the maximum hardness of 248HV 10 a first test using 100ºC preheat was applied. However, its hardness exceeded the defined threshold, hence, 300ºC was used as preheat. Using the higher preheat temperature, it was possible to meet the 248HV 10 criterion for the standard linear hardness measurements, indents made at 2 mm above the root surface.

In this development, the CVN results presented low response, which is not typical for this material and not typical for the hardness levels measured. Therefore, in order to better characterize the weld a vertical line of hardness indents was done through-thickness on the root pass only and on the full joint condition, Figure 8. While the root pass did not present large variations, the full joint condition revealed a peak of hardness near the transition from the GMAW filling pass to the HLAW root pass.

32

Figure 8 - Through-thickness macro hardness indents, HV10. EWI [2]

A multi-level sub-size Charpy V-notch, 7mm X10mm, was used to evaluate three discrete regions; top (GMAW), center (HLAW transition to GMAW, and bottom (HLAW).

Those results were low near the transition region.

Since a more detail study was beyond the scope, funds and time, the JIP did not address the reasons, details or mechanism of this toughness behavior.

This unexpected behavior and the fact of most of the HLAW articles did not explain the microstructure formation or its mechanical results, evidenced room for this research. 33

Chapter 3: Experimental Procedure

3.1. Materials used

3.1.1. Base Materials

Three base materials were used in this research. All were carbon steel specified by the API 5L standard [18] with 19mm thickness. The manufacturing process, however, was distinct from each material. Therefore, for organization the set of welds were classified as

A, B and C based on the different on material condition, as detailed below:

• Material A: 19 mm thick carbon steel plate, API X65 grade, which was used on

previous work [2]. This steel was produced using Thermomechanically Controlled

Processing followed by accelerated cooling (TMCP-AC). The weld joint was

composed by two plates of 6 inches wide plates by 18 inches long.

• Material B: 16 inches diameter seamless pipe with 19 mm wall thickness, API X65

grade steel, produced by extrusion. This manufacturing method presents a different

thermal-mechanical history when compared with the conventional rolling

processes. The pipe sections of were welded using parameters adapted for such

shape. The joint was composed of two pipe sections of 6 inches wide.

• Material C: 38 inches diameter pipe with 19 mm wall thickness, API X70 grade

steel. The manufacturing process used plates produced by TMCP without

accelerated cooling, and later bent and seam welded to form a pipe by the UOE

process. The UOE pipe manufacturing process consists of 1. bending the plate in U

shape; 2. transforming the U shape in an O shape pipe; 3. radial expansion of the

34

pipe to assure roundness, followed by a stress relieving heat treatment. The joint

was composed of two parts of pipe sections of 6 inches wide by 18 inches long.

3.1.2. Filler Metals

Both the HLAW and the GMAW processes used the same filler material. The welds on materials A and B used the filler metal ER80S-Ni1 produced by Lincoln Electric while the filler metal for the welds on material C had the same AWS specification (ER80S) manufactured by Hyundai. Even though, it did not have 1% nickel maximum content limitation. The decision for avoiding NACE 1% limitation was due to cost and material availability

3.2. Chemical Composition

All materials were specified by the API Spec 5L [18]. However, this standard controls mainly the yield strength and accepts a wide range of chemical composition.

Therefore, even though all materials were API specified and only presented 5 ksi strength difference, the chemical composition could present more differences.

Optical emission spectrometer was used to measure the chemical composition of each material. All presented values are the average of three measurements.

3.3. Welding Pre Heat

Previous work Purslow [2] have revealed that 100 ºC preheat was not enough to keep hardness below 248HV or 22HRC, which is a NACE MR 0175 [32] requirement. On

35 the other hand, 300 ºC preheat produced hardness levels as required but such standard.

Therefore, it was decided, due to time, material and cost optimization, to only use the 300

ºC preheating condition in this study. Thus, the material A welded joints were all produced using 300 ºC preheating temperature for both, HLAW root and GMAW filling passes.

For the welded joints using the materials B and C, 100 ºC preheating temperature condition was sued for multiple reasons:

• Such condition is less a challenging for field application;

• 100 ºC preheat it is already required in order to eliminate moisture from the

base material;

• The hardness values found during previous work [2] exceed the NACE [32]

MR 0175 requirements but would be meet the BS 4215-1[33] standard

criteria.

Materials B and C were welded using three different preheat conditions, in order to evaluate their effect on the hardness, toughness and resulting microstructure. such preheating conditions were:

• Welding condition 1– root pass (HLAW) using 100ºC preheating and filling

passes (GMAW) using 100ºC preheating condition.

• Weld condition 2 – root pass (HLAW) at 300ºC preheating and filling

passes (GMAW) with no preheat condition.

• Weld condition 3 – root pass (HLAW) at 300ºC preheating and filling

passes (GMAW) at 300ºC preheating condition.

36

3.4. Welding Matrix

The amount of base materials and the preheating conditions generated a combination of conditions. In addition, the insight from previous work showed a hardness increase on the HLAW fusion zone after executing the filling passes. Filling passes impose an extra thermal cycle on the as welded microstructure. Therefore, for every preheat condition incomplete joints which had only the root pass and complete joints with the root pass and filing passes were evaluated.

The welding matrix was developed by defining the base materials, the preheat temperature and the condition (root pass only or full joint), as presented in Table 1.

Material Condition Pass Pre-heat Temp Name Root Pass 300 ºC A3-RP-300C A (X65 Plate) 3 Full Joint 300 ºC A3-FJ-300C-300C Root Pass 100 ºC B1-RP-100C 1 Full Joint 100 ºC B1-FJ-100C-100C Root Pass 300 ºC B2-RP-300C B (X65 Seamless Pipe) 2 Full Joint No preheat B2-RP-300C-No Root Pass 300 ºC B3-RP-300C 3 Full Joint 300 ºC B3-FJ-300C-300C Root Pass 100 ºC C1-RP-100C 1 Full Joint 100 ºC C1-FJ-100C-100C Root Pass 300 ºC C2-RP-300C C (X70 UOE Pipe) 2 Full Joint No preheat C2-RP-300C-No Root Pass 300 ºC C3-RP-300C 3 Full Joint 300 ºC C3-FJ-300C-300C Table 1 - Weld Matrix showing condition, pass order, preheating temperature, and adopted nomenclature for specimens

37

Table 1 presents the nomenclature used on this work. A name was assigned to each sample/welding condition in order to provide complete information on the evaluated sample. The name structure is AB-CC-DDD-EEE. Where A corresponds to the base metal material (material A, material B, and material C). B is the weld condition (1, 2 and 3). CC is the joint configuration (Root pass-RP and Full Joint-FJ). DDD is the root pass preheating temperature (100ºC or 300ºC). EEE is the filling passes preheating temperature (NO for room temperature, 100ºC or 300ºC).

In total, fourteen welding conditions were used. However, conditions 3 and 2 for materials B and C present equivalent pre-heating on the root pass. The specimens B2-RP-

300C is equivalent to B3-RP-300C, as C2-RP-300C is equivalent to C3-RP-300C.

Therefore, condition 2 for the root pass was not evaluated.

3.5. Welding System

The welds on material A and B were performed at the same laboratory and using the same laser system. It consisted of:

• Root pass: IPG YLR 20000 power source Laser and a GMAW Lincoln

power wave 455 mounted on Kuka arm robot with a specific laser head for

the hybrid configuration.

• Filling passes: Multi-process Miller power source set for GMAW process.

Material C welding was performed in a different laboratory in Brazil. The welding system used was also different, which consisted of:

38

• Root Pass: Trumpf Disk 16kW Laser combined with Fronius GMAW power

source.

• Filling passes: Fanuc arm robot using a Lincoln 455 power source.

3.5.1. Material A (X65 Plate) Welding Parameters

All welds from steel A had the workpiece plate fixed and the torch moved linearly.

Furthermore, the weld was done in three passes, 15mm depth HLAW root pass and two

5mm depth GMAW filling passes.

The HLAW root pass used parameters were:

All welds from steel A had the workpiece plate fixed and the power sources moved linearly. Furthermore, the weld was done in three passes, 15 mm depth HLAW root pass and two 5 mm depth GMAW filling passes.

The HLAW root pass used parameters were:

Laser Parameters; laser leading configuration, travel speed of 2 m/min (80 ipm), laser angle 0º, laser power 15.7 kW, beam to arc spacing of 2 mm.

GMAW on the hybrid root pass parameters; wire feed speed 9.38 m/min (375ipm), torch angle of 25º pushing, contact tip-to-work distance of 17 mm and shielding gas mixture using 90% Argon and 10% CO 2.

Parameters for the GMAW filling passes; wire feed speed of 8.1m.min (325ipm),

Voltage 29.5 V, Current of 255 A, Travel speed 0.23 m/min (9.3 ipm) and shielding gas mixture composed of 90% Argon and 10% CO 2.

39

3.5.2. Material B (X65 Seamless Pipe) Welding Parameters

Since this was the circumferential girth weld on the pipe. In order to avoid positional welding effects, common for pipe welds, it was chosen to rotate the pipe with a positioner and keep the power sources constant on the top (1C) position. The weld was done in three passes, 15 mm depth HLAW root pass and two 5 mm depth GMAW filling passes.

The ideal methodology would be to use the same parameters from A material welds.

However, the material B was a pipe section and the previous parameters developed from material A plates required adjustments. Otherwise, welding without defects was not possible, due to humping and uneven deposition. The biggest change was the laser-arc spacing, which was increased from 2 to 4 mm. This change ensures a longer weld pool as demonstrated by Wei et al. [25]. Since the pipe was rotated at constant speed, the longer the weld pool the thinner is the layer of molten metal avoiding this molten material slide back over the weld bead.

Laser Parameters; laser leading configuration, travel speed of 2.1 m/min (80ipm), laser angle 0º, laser power 15.7 kW, beam to arc spacing of 4 mm.

GMAW for the hybrid root pass parameters; wire feed speed 9.38 m/min (375 ipm), torch angle of 25º pushing, contact tip-to-work distance of 17 mm and shielding gas mixture using 90% Argon and 10% CO 2.

Parameters for the GMAW filling passes; wire feed speed of 8.1 m/min (325 ipm),

Voltage 28 V, Current of 280 A, Travel speed 0.46 m/min (18 ipm) and shielding gas mixture composed of 90% Argon and 10% CO 2.

40

3.5.3. Material C (X70 UOE Pipe) Welding Parameters

Even though this material was in a pipe shape, its 38 inches outer diameter do not allow the pipe to be rotated during the weld on the available system. The alternative used was to cut the pipe perimeter in five sections maintaining the same width used for the welds on Material A and B (12 in). Hence, the workpiece was kept stationary and the torch moved in a semi-circular trajectory in order to follow the pipe section curvature.

It is important to clarify that due to power restriction of the hybrid torch system of

10kW the penetration depth was limited to 10 mm. In addition, two extra filling pass were used fill the same 19mm thick joint. Therefore, the whole weld was done in five passes, 10 mm depth HLAW root pass and four 4 mm depth GMAW filling passes

Laser Parameters; laser leading configuration, travel speed of 2 m/min (80 ipm), laser angle 0º, laser power 8.5 kW, beam to arc spacing of 2 mm.

GMAW for the hybrid root pass parameters; wire feed speed 12 m/min (472.5 ipm), torch angle of 20º pushing, contact tip-to-work distance of 17 mm and shielding gas mixture using 90% Argon and 10% CO 2.

Parameters for the GMAW filling passes; wire feed speed of 11 m/min (433 ipm),

Voltage 18.3 V, Current of 127.7 A, Travel speed 0.24 m/min (9.4 ipm) and shielding gas mixture composed of 90% Argon and 10% CO 2.

3.6. Thermal history

41

Type K thermocouples were used to acquire the joint thermal history during welding for material B. The thermocouples registered the temperature-time data for all the three passes separately.

Three thermocouples were installed on the internal surface of the pipe (bottom part of the weld) in distances of 2, 3 and 4 mm from the joint bevel (figure 7). The thermocouples were separated by an angle of 60 degrees, therefore, only one-half of the pipe had thermocouples installed, as presented in Figure 9.

Figure 9 - Thermocouple disposition on the pipe inner surface on the material B weld.

42

The acquiring time interval set was 10 ms, which proved to have resolution enough to acquire the GMAW cycle. However, due to its rapid cooling, it was not enough for the

HLAW. Even though, the acquiring frequency was not enough for the HLAW cooling the acquired data is still a reference of the HLAW thermo-cycle, if needed curve fitting could be used to reproduce its behavior.

The objective of the thermal history measurement was to calibrate the numerical model.

3.7. Metallographic Analysis

The welded samples on the material A they were extracted from the welded plate with the weld bead parallel to the rolling direction. For the other materials, since the welds were pipe circumferential butt joints the weld bead was perpendicular to the rolling direction of material B and extrusion direction for material C.

The samples were all prepared by sand grinding in sequence, reducing the grinding grain, 240, 320, 400, 600 and 800 to finally polish using diamond paste in two steps, 6 microns, and 1 micron.

3.7.1. Optical Microscopy

All the optical microscopy analysis used an Olympus G51X microscope. In order to evaluate the microstructure by optical microscopy, the samples were chemically etched

43 with Nital reagent in a 2% solution (2 ml of nitric acid (HNO 3) and 98 ml of ethanol). The etching average time was 25 seconds for the material A, 35 seconds for weld set B and 30 seconds for the material C.

3.7.2. Scanning Electron Microscopy

The SEM images of materials A and B were obtained using a Quanta 200 and a

Sirion 20kV microscopes. For Material C a Shimadzu SSX-550 microscope.

3.8. Hardness Mapping

Results previous study [2] demonstrated distinct behavior in three measurements at different thickness locations, (bottom of the root, below the first GMAW HAZ and across the second GMAW pass). In addition, a vertical hardness line profile along the weld centerline showed that the hardness was not evenly distributed across the thickness, as shown in Figure 8. The heat source causes a thermal gradient through the thickness in every pass, shown as molten material flow, Figure 2.

Therefore, it is possible to infer that the standard linear hardness measurements

[32] at 2 mm from the top and bottom surfaces does not adequately evaluates the joint hardness and misses the critical regions. Even a several line profiles across the joint at different depths do not elucidate the joint properties and possible performance. Hence, hardness mapping is very important to reveal with proper resolution the distinct joint regions. 44

The Vickers microhardness maps were obtained using a Leco AMH43 equipment, with a load of 200 g applied during 13 seconds (HV 0.2 ). The maps comprised the whole fusion zone, the HAZ and, at least, part of the base material in order to fully represent the welded joint.

Filling passes impose an additional thermal cycle to the joint. This heat is capable of melting and heat treating part of the root pass. Thus, it is important to map the incomplete joint with only the root pass and the complete joint. This will allow verifying the effect of the filling passes on the microstructure and hardness of the joint. Therefore, for each weld two hardness maps were developed root pass, (hybrid laser), and complete joint, (HLAW pass and GMAW filling passes).

For Materials A and B, all the maps were executed with 170 microns spacing between indents in both directions, vertical and horizontal.

For material C, 190 microns of spacing was used in both directions.

3.9. Impact Testing

One of motivation for this study was the unexpected low results on impact energy found on the JIP development. Therefore, the toughness-hardness evaluation is one major aspect of this study.

Due to the specimen producing low cost, and the fast and simple results evaluation the Charpy V-notch (CVN) is the most used test to address impact toughness. Even though it does not produce quantitative toughness results, as the crack tip opening displacement

45

(CTOD). It does presents the same behavior for ductile-to-brittle transition curves.

However, CTOD is more time and material consuming, presenting an elevated cost.

In order to have a general performance information of the material impact toughness the ductile-to-brittle transition temperature (DBTT) curves were developed. This approach not only obtains the DBTT values but also presents the materials fully ductile behavior on the upper shelf energy - USE regime and lower shelf energy - LSE regime,

(completely fragile).

The indication of from previous work [2] through-thickness hardness combined with the results of the developed hardness maps showed a hard region. This region is located in the fusion zone of HLAW root pass, immediately below HAZ of the first GMAW filling pass, therefore, expected in the subcritical heated affected zone (SCHAZ).

Full-size CVN specimen might not reveal in detail properties of that small region,

(less than 1 mm extension), rather it will present an average value of the 10 mm specimen section of the weld. Alternatively, the miniaturized Charpy V-notch (MCVN) test was used evaluate this region fracture toughness. Two standards address the MCVN specimens, the

ASTM E2248-15 [34] for the reduced half-size RHS specimen with dimensions of 26.6 x

4.8 x 4.8 mm and the ISO 14556 Annex D standard [35] for the Kleinstprobe KLST specimen with 27 x 4 x 3 mm dimensions, Figure 10. The small dimensions of the specimen required accurate machining with narrow tolerances range. Lucon [36] have found no deleterious effect on using Electrical Discharge Machining (EDM) on the specimen notches. KLST specimens have been used for local analysis of fusion welded joints Xue et

46 al . [37], friction stir welds Avila et al . [38] and nuclear power plants pressure vessels Lucon

[39].

Figure 10 - Kleinstprobe KLST miniaturized Charpy V-notch specimen used on the DBTT curves development from ISO 14556 [35]

The KLST specimen, presented in Figure 10, was used to develop base material and welds DBTT curves. The specimens were specifically extracted from the hard region identified by hardness mapping within the HLAW fusion zone.

Fractography was used to identify the fracture mode. The specimen fracture surfaces were preserved with desiccant to avoid surface oxidation. In order to analyze the

47 fracture surface on the SEM, the bent and unbroken KLST specimens were broken or separated after immersion in liquid nitrogen. All the SEM fractography images were obtained on the Shimadzu SSX-550.

3.10. Thermal simulation

A numeric model was developed in order to simulate the thermal field through the cross-section of the weld modeling the heat source, using the acquired data from the thermocouples this model was calibrated in order to better represent the real welding conditions.

The model started from digitizing a macro section using the software Engauge

Digitizer®, this tool allowed to acquire the weld bead profile from a macro section of the weld. This data was then used to create a three-dimension model of the welded plate of material A in Abaqus® which use volume partitioning to separate the weld beads to then create the mesh and finally export the model to Sysweld®.

The target of the modeling is to reproduce accurately the thermal cycle imposed by the GMAW heat source. Once the model is calibrated with the thermal history acquired by the thermocouples and properly represents those cycles on the real distances it is possible to extrapolate the temperature field with respect to time on other positions of the weld.

On VisualWeld® software, the modeling of the heat source was created using a

Goldak’s double ellipsoidal model to represent each weld pass of the GMAW arc welding, the HLAW pass was not simulated since the objective of the simulation was to impose the heat from the arc source over the HLAW weld bead and the base material.

48

The Figure 11, presents the mesh used on the model, even though the mesh was originally created on Abaqus® it was reseeded in order to optimize computational time and avoid instabilities due the model exportation.

A numeric model was developed in order to simulate the thermal field through the cross-section of the weld modeling the heat source, using the acquired data from the thermocouples this model was calibrated in order to properly represent the real weld.

Figure 11 - Finite Element Method mesh used to model the welding steps for the filling passes on the software Sysweld®.

49

Chapter 4: Results and Discussion

In this chapter, the results of the analysis on the three distinct API steel base metals,

(material A X65 plate, material B X65 seamless pipe and material C X70 UOE pipe), and on the correspondent welded joints using different pre-heating conditions are presented.

4.1. Development of the welded Joints

Welded joints were prepared transversally to rolling direction, even for material A in a plate shape, in order to represent the pipeline girth joint. All root passes were carried out by hybrid laser arc welding which in this research is considered a combination of laser beam and gas metal arc welding GMAW with the laser in leading configuration.

The nominal thickness of 19 mm was used for all materials. All welds were executed in full penetration combining an HLAW root pass and GMAW filling passes. For all the welds and preheating conditions, the root pass and full joint configuration were evaluated.

4.2. Thermal History

The sequence of figures presents the acquired temperature-time data plots for the

B1-FJ-100C-100C welds. This data was acquired by thermocouples installed in the inner surface of the pipe at 2, 3 and 4 mm distance, Figure 9. Each figure represents one step of the filling procedure using exactly the same welding parameters. Figure 12 shows the first

GMAW filling pass. Figure 13 shows the second GMAW filling pass.

50

The data presents a very similar behavior and peak values for all thermocouples. It is important to notice that during mount procedure, the 3 mm and 2 mm thermocouples had its position inverted and therefore the plot on time basis shows first the cycle for the 3mm and later the 2mm and 4 mm thermocouples. It is also noticeable that the peak registered values for the 3 mm and 4 mm position are very similar, this has two possible reasons, low accuracy in the distance measurement during the installation, the low influence of the heat conduction to be detectable within a short distance.

B1-FJ-100C-100C - GMAW 1 - Thermal History 500 450 400 C]

o 350 300 250 3mm 200 2mm 150 Temperature[ 100 4mm 50 0 0 50000 100000 150000 200000 250000 Time [µs]

Figure 12 - Acquired thermal cycles for the GMAW first filling pass on the B1-FJ-100C- 100C specimen.

51

Figure 13 presents the same trend, as expected, for the second GMAW filling pass.

However, the peak temperature values are lower since the heat source is further from the thermocouple. It is possible to notice in both plots that the thermocouple at 4 mm presents a higher cooling rate due to transient effect due to the arc extinction only one inch after it passed by such thermocouple position.

B1-FJ-100C-100C - GMAW 2 - Thermal History 400 350 300 C] o 250 200 3mm 150 2mm

Temperature[ 100 4mm 50 0 0 50000 100000 150000 200000 250000 300000 Time [ µs]

Figure 13 - Acquired thermal cycles of the GMAW second filling pass of the B1-FJ-100C- 100C specimen

4.3. Chemical composition

52

The chemical composition of the base metal and the filler metals was measured by optical emission spectrometry. Here is presented the average of three consecutive measurements for every material, A, B and C and the correspondent filler metals used.

Table 2 presents the measured chemical composition for the materials A, B and C and also for the filler metals used on each weld.

Chemical %C %Si %Mn %P %S %Mo %Ni %Nb *CE *CE Composition pcm IIW A - X65 Plate 0.052 0.247 1.267 0.003 0.001 0.074 0.095 0.040 0.139 0.261 ER80S-Ni1 0.097 0.653 0.987 0.008 0.004 0.01 0.854 0.000 0.185 0.241 B - X65 0.12 0.190 1.403 0.006 0.0003 0.143 0.15 0.025 0.212 0.299 Seamless Pipe ER80S-Ni1 0.097 0.653 0.987 0.008 0.004 0.01 0.854 0.000 0.185 0.241 C - X70 UOE 0.116 0.198 1.65 0.0045 0.0025 0.0005 0.011 0.044 0.210 0.288 Pipe ER80S 0.691 0.543 0.933 0.0105 0.0132 0.0102 0.495 0.000 0.162 0.274 Table 2 - Measured chemical composition of the base metals (material A X65 Plate, material B X65 Seamless Pipe and material C X70 UOE Pipe) and the filler metals used, (ER80S-Ni1 for materials A and B and ER80S used for the material C) ( + + ) = + + + + + + 5 60 30 20 15 10

( + ) ( + + ) = + + + 15 6 5

Table 2 also presents the carbon equivalent content calculated by two equations the

CE pcm equation and the CE IIW equation. The API 5L [18] defines for these materials the

CE pcm as an adequate carbon equivalent, due their low alloy content. However, if the carbon content is higher than 0.12% the CE IIW equation should be used for the carbon equivalent calculation. Materials B and C are on the limit for the calculation, therefore, both values are presented on the table.

53

It is interesting to highlight that material A presents a much leaner alloy content if compared with the other two materials, which is evidenced by both, the carbon content and carbon equivalent. This is expected due to the TMCP-AC condition for material A, which provides equivalent YS and UTS for lower alloy content.

In addition is possible to notice that the CE pcm (PCM, critical metal parameter) for the presented compositions, is significantly affected by the alloy content while the CE IIW presents similar values for the three materials.

4.3.1. Dilution

It is understood that the dilution (mixing of the base material on the filler metal composition) varies though the thickness on the HLAW root pass. A lower base metal dilution is expected on the top of the weld pool while a very high dilution is expected on the bottom, as shown on the on Figure 6. While at the top the fusion zone has strong influence of the GMAW process, at the bottom is governed mainly by the LBW keyhole and therefore, low or none filler metal mixing is expected

Chemical compositions measurements were taken in three locations through the thickness of the hybrid pass, J, K and L as shown in Figure 14. All three locations were on the HLAW root pass at weld center line. Those local measurements allow inferring a trend of how the chemical elements are mixed through the root pass.

54

Figure 14 - Locations of chemical composition measurements, weld center line face.

The measured values for all three locations and the base material reference are presented in Table 3. The specimen measured presents the measured values in those locations and the base material references.

%C %Si %Mn %Mo %Ni %Co %Nb %V %Fe

Location J 0,064 0,3 1,18 0,057 0,32 0,002 0,03 0,033 97,8 Location K 0,056 0,26 1,25 0,067 0,2 0,002 0,034 0,038 98 300C 300C- A3-FJ- Location L 0,056 0,25 1,25 0,071 0,16 0,002 0,037 0,041 98

Material A 0,052 0,2467 1,267 0,0743 0,09467 0,00267 0,03967 0,043 98

Location J 0.093 0.458 1 0.0514 0.697 0.0031 0.0036 0.0011 96.9 Location K 0.0943 0.24 1.32 0.154 0.153 0.0063 0.0264 0.0006 96.9 300C 300C- B3-FJ- Location L 0.0997 0.245 1.27 0.153 0.141 0.0046 0.0268 0.0005 97.2

Material B 0,12 0,19 1,403 0,1433 0,15 0,001867 0,02467 0,00367 97,7

Location J 0,0894 0,599 1,14 0,0058 0,282 0,0053 0,0144 0,0191 97 Location K 0,0937 0,383 1,38 0,002 0,128 0,0042 0,0321 0,0332 97,5 300C 300C- C3-FJ- Location L 0,114 0,326 1,57 0,0005 0,0097 0,0014 0,0457 0,0419 97,7 Material C 0,1160 0,198 1,650 0,0005 0,0111 0,0025 0,0439 0,0368 97,8 ER80S-Ni1 0,0968 0,653 0,987 0,0100 0,8540 0,0050 0,0005 0,0013 96,8 ER80S 0,0691 0,543 0,933 0,0101 0,4950 0,0069 0,0005 0,0033 96,7 Table 3 - Through-thickness chemical composition measurements of welds A3-FJ-300C- 300C, B3-FJ-300C-300C, and C3-FJ-300C-300C at locations J, K, and L.

55

The Figure 15 and Figure 16 show the weight percent concentration of carbon and vanadium respectively. It is possible to verify a gradient across the thickness as a function of the filler and base metals mixing.

.

C Content 0,14 0,12 0,1 0,08

WT% 0,06 0,04 0,02 0 Location J Location K Location L Base Material

Material A Material B Material C

Figure 15- Carbon measured content at locations J, K, L, and base material

56

V Content 0,05

0,04

0,03

Wt% 0,02

0,01

0 Location J Location K Location L Base Material

Material A Material B Material C

Figure 16 - Vanadium measured content at locations J, K, L, and base material

It is possible to verify that the mixing of base material and filler metal is distinct as a function of location. In location J (top part of the root pass) for materials A and C a greater influence of the filler metal can be seen while in the other locations the chemical content gets closer to the base material reference, for both elements carbon and. The measured values show that material A and C presents a higher content of vanadium on weld metal and therefore are more susceptible to secondary hardening due to precipitation.

It is important to notice that even though some vanadium carbides are expected as secondary precipitation, the amount of C, V and Nb are relatively small and therefore, even

SEM images may not reveal the precipitates.

4.3.2. Continuous Cooling Transformation – CCT diagrams

57

Since the base materials are steels, and those are thermal history dependent materials, the use of CCT diagrams helps to estimate the resulting microstructure as a function of the thermo-cycle experienced. The base materials and the filler metals CCT diagrams were calculated on JmatPro® software, based on chemical composition.

Figure 17 presents the CCT diagram for the material A based on its measured chemical composition.

Figure 17 - CCT diagram for material A, API X65 TMCP accelerated cooling plate

58

Figure 18 presents the CCT diagram for the material B based on the measured chemical composition.

Figure 18 - CCT diagram for material B, API X65 Seamless pipe

Figure 19 presents the CCT diagram for the material C based on the measured chemical composition.

59

Figure 19 - CCT diagram for material C, API X70 UOE pipe

Figure 20 presents the CCT diagram for filler metal ER80S-Ni1 used on welds for material A and B, based on the measured chemical composition. This material meets the

NACE [32] maximum nickel content.

60

Figure 20 - CCT diagram for the ER80S-Ni1 filler metal used on welds of materials A and B.

Figure 21 presents the CCT curve for filler metal ER80S used on welds for material

A and B, based on the measured chemical composition.

It is interesting to notice that the HLAW in this research always produced cooling rates faster than 10 oC/s while the modeling of the GMAW filling passes presented calculated cooling rates of 15 oC/s. Therefore, ferrite, bainite, and martensite are the expected microstructures.

61

Figure 21 - CCT diagram for the ER80S filler metal used on weld of material C

4.4. Hardness Mapping

The hardness maps were used as a way to identify regions-of-interest on the weld cross-sections, and as a method to evaluate the differences between the incomplete joint with only the root pass and the completed joint with the filler passes.

Hardness maps were developed for the base materials with an average of 400 indents in order to verify the homogeneity of the property through the thickness. Even

62 though some of the base material hardness can be verified on same weld specimen, all the base material measurements were taken away (200mm) from the welded joint.

TTable 4 presents the results of the base materials hardness measurements, the values are presented as an average and standard error. Materials A and B presents equivalent hardness, and microstructures. However, material C even with carbon content equivalent to material B presents a higher average hardness.

Average Hardness Standard Error

[HV 0.2 ] [HV 0.2 ] Material A (X65 Plate) 203 ± 8.95 Material B (X65 Seamless 200 ± 4.59 Pipe) Material C (X65 Plate) 228 ± 8.68 Table 4 - Base materials hardness average and standard error

Due to the effect of fast cooling, the welds preheated with 100 ºC presented higher hardness levels that required using a larger scale range 190-380HV. In order to have contrast resolution Table 5 presents a summary of the hardness maps in the scale 190HV 0.2 -

290HV 0.2 , and Table 6 presents the same maps in the scale of 190-380HV.

63

A3-RP-300C A3-FJ-300C-300C B1-RP-100C B1-FJ-100C-100C

B2-FJ-300C-No B3-RP-300C B3-FJ-300C-300C C1-RP-100C

C1-FJ-100C C2-FJ-300C-No C3-RP-300C C3-FJ-300C-300C

Table 5 - Hardness results summary in the 190-290HV0.2 scale

64

A3-RP-300C A3-FJ-300C-300C B1-RP-100C B1-FJ-100C-100C

B2-FJ-300C-No B3-RP-300C B3-FJ-300C-300C C1-RP-100C

C1-FJ-100C C2-FJ-300C-No C3-RP-300C C3-FJ-300C-300C

Table 6 - Hardness results summary in the 190-380HV0.2 scale

• A3-RP-300C

65

Figure 22 shows the hardness map in the 190-290HV 0.2 scale developed on the root pass weld for material A, welded with 300 ºC pre-heating.

Figure 22 - Hardness map on the HLAW root pass of the material A using 300 °C preheating – A3-RP-300C – HV0.2

66

From the Figure 22, it is possible to confirm the 203HV 0.2 hardness of the base metal, even though nearby the weld. It also shows the low values of hardness on both HAZ and FZ. The hardness map shows a small hardness gradient from the top part of the hybrid pass (about 230HV0.2), where the influence of the GMAW heat and filler metal is higher than on the bottom part of the root, where almost pure autogenous weld is produced by the laser keyhole (215HV 0.2).

The hardness values do not exceed 240HV 0.2, which is within the initial NACE [32] requirement. In addition, it is possible to identify an initial reduction in hardness on the

HAZ, before to reach back the base material average.

• A3-FJ-300C-300C

Figure 23 presents the hardness map developed on the material full joint configuration using 300 ºC preheating for both root and filling passes (A3-FJ-300C-300C).

67

Figure 23 - Hardness map on full joint of the material A (X65 rolled plate with accelerated cooling) using 300°C preheating for the root pass and the GMAW filling pass, - A3-FJ- 300C-300C – HV 0.2

The map in Figure 23 presents a gradient across the fusion zone of the HLAW similarly to the root pass only specimen. However, it does presents a very specific region with high hardness on the top of the root pass. This region is the top part of the HLAW

68 fusion zone, which was reheated by the first GMAW first filling pass. as the hardness map shows it is immediately visible HAZ limit, A 1 isothermal.

It is important to remember the previous work [2], which indicated low toughness and a hardness peak within this area. Therefore, the analysis was concentrated within this high hardness region.

ImageJ® software was used to measure the extension and the location of the region of interest. Once the location is defined, a numerical model will allow determining the thermal history experienced by this region and its influence on the formed microstructure.

The distance from the bottom surface was measured as 8.9 mm also, the distance from the

GMAW fusion line to the center of the high hardness region was measured as 1.6 mm and its vertical extension was 0.6 mm, as detailed in Figure 24.

Nevertheless, a comparison of the microstructure and hardness in the condition prior and after the filling passes can be established. The incomplete joint with only the root pass, figure 26, did not present hardness values above 240 HV 0.2 . However, after the filling passes the same region presented 280 HV 0.2 . It is possible to infer that the thermal cycle provided from the GMAW passes caused this change.

Some high hardness regions can be seen on the HAZ of the filling passes, however, the subject of the study was focused on the HLAW. In addition, the hardness levels of

GMAW filling passes for those materials could be corrected with process parameter optimization which is a typical step during weld procedure qualification, especially for the last filling pass.

69

Figure 24 – Distance measurements, detail of hardness map A3-FJ-300C-300C specimen – HV 0.2

The measurements shown in Figure 24 were used as a reference to evaluate the microstructure at similar locations on the other welds. It was also essential in order to define the region from where the KLST specimens were extracted.

• B1-RP-100C

70

The hardness map of the specimen made of the HLAW root pass only, on the material B, using 100C preheating, is presented in Figure 25. It is important to notice that the scale used in this picture is on the 190-380HV 0.2 since the reduced preheating allowed higher hardness.

Figure 25 - Hardness map on the full joint of the material B (X65 seamless pipe) using 100°C preheating for the HLAW root pass, - B1-RP-100C – HV 0.2 .

71

Figure 25 shows a steep hardness gradient on the fusion zone from the top part of the root pass 360HV 0.2 to the bottom 290HV 0.2. Even though the top part of the root pass was re-melted by the first GMAW filling pass, there are also some hardness areas above

320HV 0.2 along the rest of the root.

It is also possible to note an isolated low hardness point at the weld center line. This was due to solidification cracking. Verified by optical and scanning electron microscopy.

Solidification cracking is common for high welding travel speeds and high solidification rates as the ones observed during HLAW process. This can be corrected controlling chemical composition, welding parameters, increasing heat input or reducing weld constraint. It is important to notice that the heat input should not be changed to keep a similar thermal effect . Restraint, however, was changed due to the shape of the coupon, plate in material A to pipeline in material B and pipe section in material C.

• B1-FJ-100C-100C

The hardness map presented on Figure 26 corresponds to the full joint configuration of the welds made on the material B using 100°C preheating for the HLAW root pass and

GMAW filling passes.

72

Figure 26 - Hardness map on full joint weld of the material B (X65 seamless pipe) using 100°C preheating for the HLAW root pass and for the GMAW filling passes, - B1-FJ- 100C-100C – HV 0.2

The Figure 26 is also presented on the 190-380HV 0.2 scale due to the same reasons, reduced heat input resulting in higher hardness. The results confirmed the behavior seen

73 on the previous map (B1-RP-100C) of hardness values above 320HV 0.2 even in the bottom of root pass. Therefore, a high hardness microstructure was expected to be seen in this region.

On the top part, the same region where the peak values were registered in Figure

23 on material A, have shown local heat-treating from the filling passes. This which reduced the hardness levels to the order of 270HV 0.2. It is important to notice that this heat treatment is local and does not extend through the whole root pass.

It is interesting that none of the welds on base material B presented high hardness on the HLAW fusion zone after the filling passes. This is due to the low content of elements alloying elements in this steel. The macro cross section and the hardness maps indicate that the location of measurement J, Figure 14, is reheated during the GMAW passes at a temperature below A 1 temperature. Therefore, no phase transformation occurs at that temperature but is local precipitation of carbides formers as vanadium and niobium should occur.

The Figure 15 and Figure 16 presented the weight percentage of carbon and vanadium on weld center line across the HLAW root pass. From Table 3 and Figure 15 is clear that on material B, the vanadium and niobium amount are very low in the first measurement, location J. Therefore, those elements are not available to form carbides precipitated and increase the hardness. These alloys stay in solution on the base material and during the weld of filling passes the thermal cycle precipitation can occur. Lagneborg

74 et al. [40] shown modeling that general precipitation occurs from 700 °C and below and the particle diameter reduces as a function of nitrogen content.

• B2-FJ-300C-No

The Figure 27 shows the hardness map of the full joint configuration specimen welded on material B (Seamless Pipe X65 steel) with 300°C preheating on all passes,

HLAW root pass and the two GMAW filling passes.

The map on Figure 27 clearly presents a difference in hardness values from the 100

°C condition. With the increased preheating to 300°C, an average hardness of 255HV 0.2 was found on the HLAW fusion zone. In addition, the results were more evenly distributed across the whole root pass. However, on the top part of the HLAW pass, there were some regions with higher hardness presenting values around 265HV 0.2. Which are still low if compared with the results obtained from the welds using 100°C preheating on the same material.

The GMAW passes in this condition, did not use preheating. Therefore, the resulting hardness for its weld metal and HAZ presented a higher average than what obtained with 100 °C preheating.

75

Figure 27 – Hardness map on full joint weld of the material B (X65 seamless pipe) using 300°C preheating for the HLAW root pass and no preheating on the GMAW filling passes, - B2-FJ-300C-No – HV 0.2

• B3-RP-300C

76

The map presented in Figure 28 corresponds to the weld made with 300 °C preheating condition on the HLAW root pass on the X65 seamless pipe, material B. This condition was made to be somewhat comparable to the weld made on the material A (A3-

RP-300C). However, there were differences since the welds were in a new material with a distinct chemical composition and a small adjust of the welding parameters due to the shape change from a plate to a pipe.

77

Figure 28 - Hardness map on root pass weld of the material B (X65 seamless pipe) using 300°C preheating for the HLAW root, - B3-RP-300C– HV 0.2 .

Even though a high hardness 280 HV 0.2 region can be seen on the very bottom of the root pass, the map shows a good answer regarding hardness along the weldment with

78 an average under 248 HV 0.2 . The higher hardness levels at the bottom of the root pass are due to two main reasons, higher heat extraction on that region and expulsion of some molten material from the keyhole, this molten drop solidifies faster than the rest of the weld and may present higher hardness. In addition, is important to consider that this a very localized effect and it depends on the stability of the keyhole, the root pass of the B3-FJ-

300C-300C is the same root weld and it does not exhibit this hardness on the bottom.

The map shown a similar behavior of the hardness fields as seen on the A3-RP-300

°C, there is a small gradient from top to bottom, however, there is no major distinction on hardness through the HLAW fusion zone. Again is possible to notice that on the top where the GMAW heat source and filler metal interacts more with the weld the resulting hardness is higher. The HLAW HAZ just adjacent to the fusion line presents an average hardness level of 210 HV 0.2 and hardness but there is a deep to 190 HV 0.2 1mm from it fusion line.

• B3-FJ-300C-300C

The hardness map presented on Figure 29 shows the specimen B3-FJ-300C-300C, full joint on material B using 300°C preheating for the HLAW root pass and both GMAW filling passes. This specimen is most comparable to the A3-FJ-300C-300C and it is clear to notice that the peak of hardness is smaller and less concentrated. Almost the whole weldment is under the 248 HV 0.2 , which proved to be an effective result of the preheating.

79

Figure 29 - Hardness map on full joint weld of the material B (X65 seamless pipe) using 300C preheating for the HLAW root and 300C for the GMAW filling passes, - B3-FJ- 300C-300C – HV 0.2

Figure 29 presents the B3-FJ-300C-300C specimen in the 190-290HV 0.2 scale, the map shows an area of higher hardness constricted within HLAW fusion zone, and again there is an increase of hardness towards the top of root pass. It is interesting that the first

80

GMAW pass caused a deep hardness reduction on its HAZ over the base material and over the HLAW fusion zone. Similarly, it is possible to see, parallel to the root pass fusion line, a decrease of the hardness levels. This effect is less noticeable on the bottom part of the root, due to the smaller influence of the arc heat source.

There were high hardness measurements on the HAZ of the GMAW filling passes.

However, the GMAW HAZ on the base metal is not the subject of this.

• C1-RP-100C

The root pass on all material C had only 10 mm penetration due to power limitation of the hybrid laser welding head. Even though the power source had 16 kW power the welds used only 8.5kW for full penetration root pass.

Figure 30 presents the hardness map, in the 190-380 HV 0.2 scale, due to the 100 °C preheating. In this weld, the peak of hardness was at the HLAW HAZ. However, the root pass fusion zone also presented high hardness.

81

Figure 30 - Hardness map on full joint weld of the material C (X70 UOE pipe) using 100C preheating for the HLAW root pass, - C1-RP-100C – HV 0.2

• C1-FJ-100C-100C

The specimen C1-FJ-100C-100C hardness map is presented in Figure 31. It was mapped on the material C (API X70 UOE pipe) with the lowest preheating configuration,

100°C for the HLAW root pass and the same for the GMAW filling passes. This specimen presented hardness reaching a peak of 410 HV 0.2 on the HLAW fusion zone. However, it 82 is plotted on the 190-380 HV 0.2 scale for comparison simplicity, otherwise, the other regions would present poor contrast on hardness.

The high hardness is due to a combination of conditions, this material C present higher alloy content, similar to material B, but the root pass had only 10mm penetration, therefore, there was more mass for heat extraction.

It is also possible to notice high hardness on the HAZ of the last GMAW filling pass. Since this could be fixed by adjusting the welding parameters and the focus of the study is the hybrid laser welding, this region is not analyzed.

The effect of the filling pass promoting a heat treatment on the previous weld bead can be seen on the top of the root pass were the hardness levels presented a decrease of 40

HV 0.2 .

83

Figure 31 - Hardness map on full joint weld of the material C (X70 UOE pipe) using 100C preheating for the HLAW root and 100C for the GMAW filling passes, - C1-FJ-100C- 100C – HV 0.2 .

C2-FJ-300C-No

The Figure 32 shows the hardness map in the 190-290 HV 0.2 scale developed on the material C (API X70 UOE pipe) using the 300°C only on the HLAW root pass and no preheating for the GMAW filling passes. This pre-heat condition was considered due to the difficulty to execute and control preheating on the field, therefore, if only the root pass required preheating it would reduce one extra step saving time and money. 84

It is important to notice the base material by itself already presents higher hardness than the other tested materials, as presented in Table 4. However, the HLAW heat affected zone presented on one side a high hardness field with 310 HV 0.2 .

The asymmetry in the hardness fields is highlighted due to misalignment, a slight curvature can be seen on the right side and the bottom surface reference also indicates a misalignment. Therefore, uneven beam exposure combined with the keyhole fluctuations while moving could cause rapid cooling in one of the walls.

85

Figure 32 - Hardness map on full joint weld of the material C (X70 UOE pipe) using 300C preheating for the HLAW root and no preheating for the GMAW filling passes, - C2-FJ- 300C-No – HV 0.2 .

• C3-RP-300C

86

Figure 33 presents the map developed for the specimen C3-RP-300C, this represents the root pass weld made on the material C (API X70 UOE pipe) using 300C preheating.

As on the other welds made on material C is possible to notice the higher base material hardness. A large portion of the fusion zone presented hardness levels around

230HV 0.2, however, is still possible to verify a hardness increase on the top part of the root pass, as seen all other welds for all materials. This is likely due to the higher influence of the arc welding on that region. That means filler metal addition which in general increases the hardenability combined with the broader area of the molten pool, increase on convection area.

87

Figure 33 - Hardness map on full joint weld of the material C (X70 UOE pipe) using 300C preheating for the HLAW root, - C3-RP-300C – HV 0.2 .

Even though there was an increase of hardness on the top part the condition of multi-pass weld could alter those hardness fields, usually a tempering heat treatment effect is expected, however, the heat of subsequent pass can also start other mechanisms that will not necessarily promote tempering.

88

In this weld is possible to see the edge of the fusion zone with hardness a slightly higher (255 HV 0.2 ) this is likely to be the coarse grain heated affected zone CGHAZ.

• C3-FJ-300C-300C

Figure 34 presents the hardness map developed on the material C, filled joint using

300°C preheating on root and filling passes, specimen named as C3-FJ-300C-300C.

The figure shows two unexpected high hardness regions at the sides of the root, this is not typical and is likely to be molten material ejection from the bottom part of the keyhole, which did not detach completely and impinged on the side of the root. The microscopy will also look to those regions in order to confirm its microstructure.

As mentioned before, the material C was welded in Brazil and a small amount of material was sent to OSU to produce this characterization. Since the amount of material consisted in 12 mm of extension, all the material available presented the same feature. In another hand, this material was also characterized and mapped in the lab in Brazil were the weld was produced. Those specimens did not present this feature either on hardness map nor in microscopy. Therefore, it is believed that is just a local effect related to the keyhole stability or molten material ejection. If there was more material available, the map section should not present this specific high hardness fields.

Other than the top part, which was heat treated by the filling pass, the HLAW fusion zone presented a homogenously distributed low hardness. The HLAW HAZ presented values smaller than the base metal itself while the GMAW HAZ showed some elevated hardness which could fix with better welding parameters set up.

89

Figure 34 - Hardness map on full joint weld of the material C (X70 UOE pipe) using 300C preheating for the HLAW root and for the GMAW filling passes, - C3-FJ-300C-300C – HV 0.2

This map indicated, as on the A3-FJ-300C-300C map, a high hardness region near the top of the HLAW pass, just below the visible GMAW HAZ on the HLAW fusion zone.

90

This high hardness region presents a local increase from the 230HV 0.2 to 275HV 0.2, the location, and extension of this region is a reference to focus on microstructural analysis.

Using the software ImageJ® the distance from the bottom surface until the center of the high hardness region was measured as 7.1 mm, also, the distance from the GMAW fusion line to the center of the high hardness region was measured as 2.9 mm and its extension was 0.9 mm, as detailed in Figure 35.

Figure 35 - Distance measurements, detail of hardness map C3-FJ-300C-300C specimen – HV 0.2 91

As mentioned before, there were differences in the welding parameters. Therefore, the welds are only completely comparable within the same material group, those used the same parameters and only the preheating condition varied. However, even if the welding parameters were slightly different, the A3-FJ-300C-300C, the B3-FJ-300C-300C, and the

C3-FJ-300C-300C, could be compared since they presented almost the same classification of base materials and filler metals, same preheating and same welding speeds.

4.5. Microscopy

4.5.1. Base Material

Since the three materials were produced by different manufactures its composition and processing were completely different, therefore, the microstructures were also distinct.

As mentioned earlier, the material A is an API X65 carbon steel rolled on TMCP process with the accelerated cooling process in a plate shape. The resultant microstructure by this process for the measured chemical composition presented a mainly bainitic microstructure. Figure 36 presents fine grains of granular bainite elongated in the rolling direction, which is not typical for API X65 steels since its strength limit is not too high and can easily be reached just by the conventional rolling process. However, this microstructure is obtained when used TMCP rolling combined with accelerated cooling.

92

Figure 36 - Material A microstructure, elongated granular bainitic structure

It is noticeable that even though it is a rolled plate specimen the microstructure is not banded with layers of ferrite and perlite as it will be shown on the material C specimen.

This bainitic microstructure is typical for higher resistance grades as X80 or X100 that in order to achieve strength and toughness without severely affect weldability use accelerated cooling during with the TMCP. However, depending on the chemical composition and the cooling procedure, even lower resistance grades as X65 may present predominant bainitic structure.

93

Material B is also an API X65 carbon steel but it was processed already in pipe shape by extrusion. Even with the distinct manufacturing process and chemical composition, it also presented mainly bainitic structure, Figure 37.

Figure 37 - Material B microstructure, elongated granular bainitic structure

Also in Pipe shape, the material C is an API X70 carbon steel, which was rolled, wrapped and seam welded to form a pipe in a process called UOE. This material, however,

94 presented a banded microstructure, typical for rolled plates, mainly formed of polygonal ferrite (PF) with bands of pearlite (FC(P)) elongated in the rolling direction as presented in

Figure 38.

Figure 38 - Material C microstructure, banded ferrite and perlite structure elongated in the rolling direction

The bainite is one of the resultants microstructure on the displacive transformation regime, shear dominant and with high reaction rates [41]. A displacive model for the

95 bainitic transformation is generally accepted [42], however, there is not a consensus. On this model, the bainitic ferrite nucleation occurs by diffusion mechanism while its growth is then controlled by a displacive mechanism.

This transformation occurs between the austenite to martensite transformation temperature and the austenite to pearlite transformation, either through continuous cooling or isothermal transformation. Davenport and Bain [43] showed in his work that the microstructure formed in those intermediate temperature was, microscopically distinct of pearlite and martensite. However, this bainitic structure also presents similarities with either perlite or the martensite. Similar to the perlite it is a two-phase aggregate of ferrite and carbides, however in a non-lamellar form since the phase forms consecutively and does not alternate as in the perlite form. It is also similar to martensite since bainitic ferrites may be presented as plates or sheaves.

Even though the displacive mechanism has been proposed and defended by part of science community there is still no consensus in the bainite mechanism of transformation and growth. There is also more than one system for the classification its morphology and microstructural classification. Classically, two distinct morphology categories were used to describe the bainite, upper, (Bu), and lower, (Bl), depending on the transformation temperature. Both upper and lower bainite consists in laths or platelets of bainitic ferrite which are separate by a residual phase, (residual due to the carbon partition in the austenite- ferrite interface), such as cementite, martensite or even untransformed or retained austenite

[44].

96

The upper bainite forms at temperatures slightly below the pearlite transformation temperature, (higher or upper temperatures). The bainitic ferrite in its upper form does not have carbides itself; rather the cementite carbides are formed as a consequence of enrichment in carbon of the residual austenite trapped between the platelets of upper bainitic ferrite. [42] Therefore, it tends to present in parallel ferrite laths separated by its carbides with the same orientation. There may be also cementite precipitation at austenite grain boundaries, which is very detrimental for toughness. [44]

The lower bainite forms at a temperature slightly above the martensite start formation (Ms) temperature and it is also composed of ferrite and carbide aggregate in a non-lamellar form, also with parallel platelets of ferrite containing fine carbides in it.

However, there are two known forms of carbide precipitation. A fine disperse carbide ( ε- carbide) precipitation within the ferrite plate. And the mechanism of carbide precipitation from carbon enriched austenite between plates of ferritic bainite [45], those, however, are thinner with a smaller volume fraction, since the part of the carbon is already trapped on carbide precipitation within the ferrite [46] as shown in Figure 39.

97

Figure 39 - Schematic representation of upper and lower bainite formation from Bhadeshia [44]

4.5.2. Welded Joints

All welds were characterized in the root pass and the full joint condition. The root pass presents a microstructure purely resulting from dilution and welding thermal cycle, whereas the completed joint presents an HLAW weld metal and an HAZ that were thermally treated by the filling passes. The hardness maps presented earlier provided references of regions of interest that should be analyzed in more detail.

98

From the hardness maps presented on item 4.4 It was shown that there are differences on hardness levels on the conditions before and after the filling passes, therefore, a change in microstructure might also be expected.

• A3-RP-300C

The microstructure found on the whole root pass is mainly bainite or bainitic ferrite with second phase AC as described on IIW scheme [47] or (FS(B)) as defined by Thewlis

[41] and some bainitic ferrite with non-aligned phase FS(NA) and some Widmanstätten ferrite also defined as AC in the IIW scheme. Macroscopic large columnar grains growing from the fusion line towards the weld center line.

Figure 40 presents the microstructure for the regions of interest over the hardness map on the A3-RP-300C. It is interesting to notice that the characterized microstructure is in accordance with the hardness values registered. Since it was verified on the hardness maps of the welds on material A presented higher differences in the weld metal just below the first GMAW HAZ over the fusion zone of the HLAW pass the figure presents the details of this region microstructure.

99

FS(B) FS FS(B) FS(NA)

FS(NA) FS(B)

Figure 40 – A3-RP-100C Microstructure detail on the hardness map. Specimen etched with Nita 2%.

In the detail of the map two distinct regions vertically apart by 1mm, this is the region which after the filling passes presented higher hardness. The microstructure shown is composed mainly of ferrite with second phase which is shown in the form of Bainite

FS(B), granular Bainite FS(NA) and some fraction of Widmanstätten ferrite.

100

• A3-FJ-300C-300C

On the A3-FJ-300C-300C specimen, there were a couple of interesting regions which its microstructure is presented as details of the hardness maps in Figure 41.

Figure 41 – A3-FJ-300C-300C Microstructure detail on the hardness map. (a), (c) and (f) are from the HLAW fusion zone weld center line reheated by the GMAW filling passes, (b) GMAW fusion zone, (d) is a half of HLAW root pass penetration region and (e) is the weld center line WCL on bottom of the root pass, specimen etched with Nital 2%

101

The region showing the peak of hardness was extensively analyzed, its microstructure is shown in the figure in (a), (c) and (f). In (a) a large granular bainitic grain

FS(NA) can be seen on the up right corner with large non-oriented carbide precipitation intragranular, this large grain which was surrounded by smaller grains of ferrite with second phase FS(B), some ferrite with side plates some polygonal ferrite PF. In (c) more

FS(B) bainitic structures can be seen also with some grain boundary ferrite. While in (f) large bainitic grains FS(B) are presented as upper bainite.

This structures formation of upper bainite, granular bainite, grain boundary ferrite and Widmanstätten ferrite are detrimental for toughness. Therefore, reduced values of toughness might be locally expected.

The microstructures of the remaining HLAW fusion zone regions can be seen in the middle (d) of the root pass and in (e) bottom part of the root. Even though the scale is different a grain size reduction can be seen and it was expected since the heat source is more intense on the top. Predominantly bainite structure dominates this region with a higher presence of FS(B) and less amount of granular bainite FS(NA) and polygonal ferrite

PF.

The GMAW microstructure is not the subject of this study but it can be seen in (b) long grain boundary ferrite chains with a strong presence of fine acicular ferrite AF structure.

• B1-RP-100C

Figure 42 presents a detail of the microstructure over the same hardness maps presented earlier, note that for the B1-RP-100C specimen the scale range was increased to

102

190-380HV 0.2. The figures (a) presents bainite grains surrounded by martensite, needle- like structures disposed of approximately 120 degrees spacing between its sheaves. The figure (b) presents lower magnification and allows a more general view of the bainite grains and martensite dispersion in between while in figure (c) a closer detail on the needle martensitic structure is shown.

Figure 42 - B1-RP-100C - Microstructure detail on the hardness map. Details (a), (b) and (c) on the top of the HLAW root pass within the fusion zone at the weld center line, specimen etched with Nital 2%.

103

Combining the information of hardness, the carbon content of this material and the microstructure verified it is possible to define the presence of martensite on it. It is also nice to check the hardness values related to the carbon content, the Figure 43 extracted from Bain and Paxton’s [48] book presents a chart with the relationship of martensite hardness as a function of carbon content. The figure was modified to show the carbon content of materials B and C, 0.12% and 0.116% respectively, and the expected martensite hardness on the welds.

Figure 43 - Chart extracted from Bain and Paxton, Alloying in Steels, ASM, 1961[48]

104

Figure 44 shows a sequence of amplifications of the specimen B1-RP-100C. In the larger scale is possible to verify a large amount of martensite M in the top and bottom of the figure. The first close up is on a bainite FS(B) grain were an encounter of two grains directions can be seen with the change in the sheaves growth direction, in the bottom right.

Finally, the third amplification is a detail of the aligned carbides separating the sheaves.

105

Figure 44 - SEM combination images of B1-RP-100C microstructure bainite and martensite detail 1 on bainitic grain and detail 2 in carbide aggregate between sheaves, specimen etched with Nital 2%

106

• B1-FJ-100C-100C

The specimen B1-FJ-100C-100C microstructure is presented in Figure 45. In the detail (a) the GMAW intercritical and subcritical HAZ over HLAW fusion zone is presented, there is possible to see a refined grain formed over the former HLAW root pass until the A 1 temperature limit, (darker in the picture) after it the structure become coarser.

Figure 45 - B1-FJ-100C-100C Microstructure detail on the hardness map. (a) HLAW fusion zone weld center line reheated by the GMAW filling passes, (b) HLAW fusion Line, (c) Bainitic formation at WCL, (d) FGHAZ, (e) Fusion zone to CGHAZ transition and (f) is the weld center line WCL on bottom of the root pass, specimen etched with Nital 2% are from the,

107

In the detail (b) an image of the fusion line presents mainly bainite microstructure and at the bottom, there are some martensite needles starting to appear. In figure (c) the

WCL of the HLAW root is shown in which a large amount of bainite FS(B) is present together with some grain boundary ferrite. In figure (d) the fine grain heated affected zone

FGHAZ of the HLAW root pass. In figure (e) the transition from the bainitic fusion zone to coarse grain heated affected zone CGHAZ is presented. Finally, in the figure detail (f) the bottom part of the root pass microstructure is presented with large FS(B) bainite grains and large colonies of martensite M grains, which agrees with the increase in hardness.

Figure 46 presents an SEM image of the WCL of the HLAW pass where, as measured on the hardness map of A3-FJ-300C-300C, a higher hardness was expected.

However, as the hardness maps have shown for this material there is not an increase of hardness. The image does show bainite grain with two close-ups on its boundary showing the end of a carbide aggregate. It is also possible to notice that spacing between the sheaves and the carbide thickness can vary, in the higher magnification the carbide thin is on the order of a fifth of a micron.

108

Figure 46 - weld center line WCL on the bottom of the root pass, specimen etched with Nital 2%.

• B2-FJ-300C-No

The Figure 47 presents the microstructure found for the specimen B2-FJ-300C-No, in that is easy to notice that the hardness levels are reduced and the microstructure presents much less martensite than the specimen welded using only 100°C preheating. The 300°C on root pass effectively changed the thermal cycle and therefore the cooling rate promoting

109 a less hard microstructure. However, it is still possible to see some martensitic structure in figure (e) it is not the predominant one but is present in the region with higher hardness.

Figure 47 - B2-FJ-300C-No Microstructure detail on the hardness map. In (a) the GMAW HAZ below A 1, (b), (c) and (e) presents the microstructure on WCL of HLAW root pass at distinct penetration the latter also presents martensite, (d) is the microstructure of the bottom of the root and (f) presents the transition from the fusion zone to the FGHAZ

110

Similarly to what was shown for the other welds the fusion zone of the HLAW is predominantly bainitic changing the spacing between sheaves, grain size, carbide size as a function of the chemical composition and thermal cycle received.

• B3-RP-300C

The condition of 300 °C preheating used on the HLAW promoted slower cooling rates which reduced the formation of martensite and hardness.

111

Figure 48 - B3-RP-100C Microstructure detail on the hardness map. (a) and (b) show the microstructure that would be heat treated in a subcritical regime by the GMAW filling pass, (c) presents the microstructure on the root bottom and (d) macro of the HLAW pass

In all regions of Figure 48, figure (a), (b) and (c) some martensite can be seen while for the material A did not present it when subjected on the same weld and preheat condition, however, the material B has the double of the carbon content which increases the

112 hardenability. However, bainite FS(B) is the predominant microstructure, there is also some grain boundary ferrite that forms some long veins in the microstructure.

Even with a higher hardness region on the bottom of the root it is not critical to exhibit too much martensite. In addition, the hardness levels locally exceed the 248HV 0.2 but does not exceed the 300HV 0.2 requirement of the BS 4515-1 [33]. For a research study, the hardness average under the 250 HV 0.2 and only locally reaching 280 HV 0.2 is a reasonable achievement.

• B3-FJ-300C-300C

Figure 49 presents the microstructure of the material B heat treated with 300°C in all passes. As it was seen on the other specimens, using 300°C preheat inhibits the martensite formation and therefore the martensite content is small. In (a) a polarizer filter was used to reveal more texture details of the topography, using the bainite sheaves become very clear, in addition, is possible some very flat grains with little or no carbide precipitation such as granular bainite.

113

Figure 49 - B3-FJ-300C-300C Microstructure detail on the hardness map. In (a), (b) and (c) bainitic structure on GMAW subcritical HAZ over HLAW root pass, (a) is presents using a polarizer filter, (d) and (e) presents the microstructure on the bottom of the and (f) is the transition from fusion zone to HAZ.

In (b) bainite can be seen again, in large and long grains with the presence of some ferrite with side plates or Widmanstätten ferrite FS formation. Still on the HLAW fusion zone figure (c) shows the columnar growth in the direction of the WCL, the structure is

114 composed of large columnar bainite grains which are surrounded by other small bainitic grains and some grain boundary ferrite.

In the detail (d) and (e) the root bottom microstructure shows almost only bainite on its structure. The detail (f) present the transition from fusion zone to HAZ were granular bainite predominates in the structure. It is clear that the preheating acts in order to reduce the hardness levels but the toughness will be verified with the impact testing.

SEM images in order to provide a more details in a closer assessment of the surface weld cross-section surface. Figure 50 presents a combination of images using 1000X and

4000X magnification it does not exhibit martensite or a large amount of grain boundary ferrite. The strong bainitic formation shows distinct carbide precipitation some thick carbide formation between the sheaves, which could be classified as upper bainite, and some thin carbide presenting additional fine precipitation inside the sheaves, which are classified as lower bainite. On close-up detail, it is possible to see a three-grain boundary encounter and some small carbide particles within the sheaves.

115

Figure 50 - SEM combination images of B3-FJ-300C-300C Bainite microstructure, detail of a three-grain boundary encounter, specimen etched with Nital 2%.

• C1-RP-100C

The microstructure presented in Figure 51 corresponds to the material C using

100°C condition as seen for the material B, which has a very similar carbon content, it forms a high amount of martensite, the detail shows large colonies of martensite M around some grains of bainite.

116

Figure 51 - C1-RP-100C Microstructure detail on the hardness map Bainite grain surrounded by Martensite.

• C1-FJ-100C-100C

Due to the low heat input combined with the only 100ºC preheating temperature this specimen shown predominant martensite formation on the HLAW fusion zone. In

Figure 52, (a), (b) and (c) presents distinct magnifications of the region heat treated by the

GMAW filling pass. In figure (c) a lower bainite FS(Bl) grain is surrounded by martensite and grain boundary ferrite GF. It is important to notice that the region represented in on the top has tempered martensite due to the heat treat promoted during the filling pass.

Figure (d) is a macro of the welded joint, figure (e) shows a mainly martensitic structure, bainite (mostly upper bainite FS(Bu)) and small amount of grain boundary ferrite

117

GF. The hardness map shows that the martensite presented in the middle and on the bottom of the root is on the untempered condition, therefore, brittle and harder.

In figure (f) the HAZ transition from coarse to fine bainitic grain is shown.

Figure 52 – C1-FJ-100C-100C Microstructure detail on the hardness map. In (a), (b) and (c) structure on GMAW subcritical HAZ over HLAW root pass increasing magnification, (d) macro of the weld cross section, (e) untempered martensite and bainite structure on the bottom of the root pass and (f) CGHAZ to FGHAZ transition.

118

• C2-FJ-300C-No

The microstructure encountered in the specimen of material C welded using 300°C for the root pass and no preheating on the root has presented high content of bainite as can be seen in the details of Figure 53. The figure (a) present a low magnification of the fusion zone were the previous columnar grains can be seen, figure (b) and (c) presents higher amplifications of the same region were intragranular bainite and Widmanstätten are predominant with some veins of grain boundary ferrite. The figures (e) and (f) shows the bainitic microstructure of the bottom of the root and (d) is the macro cross-section of the weld.

119

Figure 53 – C2-FJ-300C-No Microstructure detail on the hardness map. In (a), (b) and (c) bainitic structure on GMAW subcritical HAZ over HLAW root pass, (a) low magnification of the HLAW fusion zone, (d) macro of the weld cross section is and (e) presents the microstructure on the bottom of the and (f) is transition from fusion zone to HAZ.

• C3-RP-300C

120

The microstructure of the material C specimen with 300 °C preheating is shown on

Figure 54 showing bainite formation and a very refined ferritic grain and also some regions of martensite. However, this microstructure did not present high hardness.

Figure 54 - C3-RP-300C Microstructure detail on the hardness map, Bainitic structure in the center of the weld.

• C3-FJ-300C-300C

The full joint of the 300 °C configuration used an equivalent weld procedure of the other two materials on the 300 °C configuration. As in the example of the material B the preheating effectively slowed the martensite formation by that increasing the bainite content. It must be noticed the formation of higher hardness field on the HLAW fusion zone under the intercritical HAZ of the GMAW, similar to what was seen for the A3-FJ-

121

300C-300C specimen martensite between the bainite grains can be seen in Figure 55 (a) and (b). This region presents the higher content of martensite than the rest of the root pass but the main presence is still bainite. Figure (c) shows the transition from bainitic and martensite grains on the fusion zone to a fully bainitic CGHAZ, where its hardness did not exceed 230HV 0.2, close to the base material measured earlier.

Figure 55 - C3-FJ-300C-300C Microstructure detail on the hardness map. In (a) and (b) Bainite on HLAW weld center line and (c) fusion line to HAZ transition (d) martensite formation on droplet attached to root, (e) bainite and Widmanstätten ferrite on root bottom part, (f) weld Macro cross-section.

122

A non-expected effect was the high hardness fields on the bottom part, this discussion was mentioned in the hardness results, and however the microstructure was also assessed. It is believed that the local higher hardness field is due to secondary precipitation hardening. This precipitates form at temperatures below A 1, agreeing with the macro image and hardness maps. The macro image on figure (f) shows a microstructure direction solidification from the top, base material, to the bottom molten metal. Since there was no heat source on the bottom and the HLAW produces a constricted root pass, this is likely to be expulsed molten droplets that did not detach properly and re-solidified once it touched the base metal surface. That reached local levels of 315HV 0.2 and high presence of martensite microstructure as seen in figure (d).

In order to allow an easy comparison on the microstructure of all specimen in the same region, GMAW subcritical HAZ over the HLAW pass on weld center line, Table 7 was developed. The measured distance from the bottom surface was used, as presented on hardness maps, to make sure that same region was selected.

123

A3-RP-300C A3-FJ-300C-300C B1-RP-100C B1-FJ-100C-100C

B2-FJ-300C-No B3-RP-300C B3-FJ-300C-300C C1-RP-100C

C1-FJ-100C-100C C2-FJ-300C-No C3-RP-300C C3-FJ-300C-300C Table 7 - HLAW WCL Microstructure within the subcritical GMAW HAZ

In the pictures shown in Table 7, it is important to make it clear that the chemical compositions are distinct. However, the carbon equivalent is close and the welding parameters were adapted from the weld on material A (plate) to the other materials (pipes) thus a direct comparison is not the best approach. On another hand, all three materials meet the API X65 strength definition (even the X70) and all the filler metal used also had the same specification. Therefore, there are reasonable similarities to apply the same welding procedure and the same properties requirements. Hence, the comparison is in order to address the resulting properties for each material once subjected to a novel welding process and independently of materials performance.

124

It is clear that the grain size is smaller for the specimens that received only 100 °C preheating. In addition, those specimens formed the higher amount of martensite, the lower the heat input the faster the cooling rates. It is also important to notice that materials B and

C have the double of the material A carbon content, which is also visible through the carbides amount and size from material A to B and C under the same magnification.

Even though martensite was visible in all specimen, the amount of bainite and martensite varied, those with lower heat input formed predominantly martensite while those using higher preheat formed predominantly bainite with some disperse martensite.

4.6. Impact Test

The DBTT curves were developed for each material set and were grouped in the charts by material in order to have a clean plot to analyze its data.

The analyzed DBTT temperatures calculated by the energy impact measured is presented in the chart of Figure 56.

125

-160 -149,0 -147,7

-140 -125,1 -121,7 -120 -104,6 -104,9 -101,5 -104,3 -100 -89,4 -87,3 -85,1 -87,4 -80 DBTTKV DBTTKV (˚C) -60 -56,5

-40

-20

0

Figure 56 - DBTT KV values in degrees C, calculated by the impact energy.

The DBTT based on the encoder measurement of the pendulum energy (KV) have shown a really good agreement with the lateral expansion LE method, which is presented in Figure 57 below.

126

-160 -150,4 -144,2 -140

-121,7 -120,5 -120 -105,7 -102,5 -101,8 -99,2 -100 -92,4 -87,8 -87,0 -84,1 -80

DBTTLE (˚C) DBTTLE -59,3 -60

-40

-20

0

Figure 57 - DBTT LE values in degrees C, calculated by lateral expansion

The values in both charts are presented in Celsius, with an excellent agreement. In addition, the behavior was as expected the base materials presented a lower DBTT than the welds since the presence of defects, stress concentrators and segregation is reduced if compared to the welds.

127

Within the welds, an interesting result is that the welds with only 100 °C preheat

(B1-RP-100C and C1-RP-100C) presented the highest DBTT values, which was expected since those welds had the fastest cooling rates and presented a significant amount of martensite in its microstructure. Oddly the full joint welds with 100 °C (B1-FJ-100C-100C and C1-FJ-100C-100C), were the ones that with the lowest DBTT for the same material group. The reason for this is the effective heat treatment of the GMAW filling passes.

However, that only happens because the impact specimen has small dimensions since the

KLST has only 3 mm in this direction, the microstructure locally changed from untempered martensite on the root pass (B1-RP-100C and C1-RP-100C) to tempered martensite on the full joint configuration (B1-FJ-100C-100C and C1-FJ-100C-100C). The rest of root pass still presented untempered martensite and is expected that a full-size Charpy specimen would register an average value.

The upper shelf energy chart is presented in Figure 58.

128

10,7 10,1 10 9,4 9,1 8,9 8,9 8,7 8,6 8,5 8,3 8 7,1 7,3

6 5,2

4 Upper Shelf Shelf EnergyUpper (J)

2

0

Figure 58 - USE value in Joules measured on miniaturized Charpy V-notch test using the KLST specimen

The USE limit consider only the completely ductile regime, Figure 58 presents the

USE values for all the conditions tested. It is interesting that Materials A and B presented a short variation from the welded condition to the base material while material C presented a large energy variation.

The base material DBTT curves constructed with a miniaturized Charpy V-notch specimen are plotted in Figure 59. In this chart the values presented earlier were materials

129

A and B (both bainitic microstructure) have a very similar DBTT value, -149°C and -

147.7°C respectively, and USE values, 9.1J and 8.7J respectively. Nevertheless, Material

C, ferritic-pearlitic microstructure, presented higher DBTT -125°C and higher USE 10.1J.

The DBTT curves were developed using hyperbolic tangent curve - HTF.

Figure 59 - Base Materials DBTT curves developed for the MCVN KLST specimen

The DBTT results are consistent with the microstructure, both bainitic had a similar behavior with a higher value while the banded ferritic-pearlitic material had a lower value.

130

The bands are stress concentrators and might act as a void nucleator similarly as an inclusion on the matrix reducing its toughness performance.

Notice that the materials A and B did not present a smooth fracture transition it presented a bimodal behavior. In addition, the material C presented a more gradual transition.

Figure 60 - Full joint welds with 300C pre-heat condition DBTT curves developed for the MCVN KLST specimen

The Figure 60 presents the DBTT developed for the full joint welds using 300°C preheating. Even though, in this condition of 300°C preheating all the DBTT results were 131 very similar, -87.3°C for material A, -87.4°C for material B and -89.4°C for material C, the behavior of the curves was distinct. The materials A and B again presented a very close behavior and both curves had a bimodal fashion, in some cases the transition from fully ductile to fully brittle occurred in less than five degrees Celsius

As expected, for all materials and conditions, there was a clear shift increasing the

DBTT from the base material to the welded condition. The material C was the more predictable presenting a smooth transition on failure mode, however, its USE was significantly reduced.

4.6.1. MCVN specimen Fractography

Fractography analysis was used in order to access the fracture surfaces of the

MCVN specimen. For each DBTT curve at least three specimens were evaluated, one on the USE, one at the lower shelf energy - LSE and one within the transition.

Table 8 is a summary of specimen fracture surfaces. In it, this table the images correspond to one broken specimen of each DBTT curves, mainly in the ductile fracture mode.

132

A BM @-138C A3-FJ-300C-300C @-48C B BM @-132C

B1-RP-100C @-25C B1-FJ-100C-100C @-26C B3-RP-300C @-85C

B3-FJ-300C-300C @-86C C BM @-147C C3-FJ-300C-300C @-92C Table 8 - SEM images of KLST specimen fracture surfaces used on DBTT curves development

It was interesting that the bainitic materials presented a lot of inclusions, either in the base metals or in the welded specimen, which nucleated small and large voids.

Figure 61 is a detail of B1-RP-100C broken at -25 °C in the ductile regime were a large void and its nucleating inclusion can be seen. Notice that the inclusion void is surrounded by small dimples or microvoids. In addition, the inclusion can be seen on the bottom and slip lines can be seen along the walls of the cave. Even though this was on

133 ductile failure mode the inclusion reduces its toughness, also this was the specimen which presented the highest martensite content.

Figure 61 - Inclusion on fracture surface of B1-RP-100C specimen broken at -25C

It was interesting that the A3-FJ-300C-300C specimen at -48 °C which is in the

USE values, hence fully ductile failure mode, presented a lot of micro-cracks and even inside those cracks the failure was ductile, shown in the detail in Figure 62.

134

Figure 62 - Detail dimples on A3-FJ-300C-300C specimen fracture surface, -48 °C.

The MCVN values are not comparable with the full-size specimen, and it should not be comparable since the reason to use MCVN is for small sections, components or local effects were the standard specimen does not fit. In addition, the specimen size reduction will cause a loss in constraint hence a decrease in the DBTT.

Lucon et al. [49] in his study validated correlations between the MCVN and CVN where the DBTT shift correlation is calculated based on other two authors Towers [50] and

135

Wallin [51] and the USE correlation for the KLST specimen is best described in the work other work of Lucon [52].

Table 9 presents the values calculated based on the correlations proposed from the

MCVN specimen to the CVN. These correlations values are only to show the results in perspective, showing what would be expected if a whole full-size (CVN) specimen could evaluate the local effect.

Measured Measured Measured Towers's Towers's Wallin's Wallin's Lucon's Values Values Values correlation correlation correlation correlation Correlation DBTT DBTT USE DBTT DBTT DBTT DBTT USE Lucon Weld KV LE KV LE KV LE (˚C) (˚C) (J) (˚C) (˚C) (˚C) (˚C) (J) BM -149.0 -150.4 9.12 -114.7 -116.1 -111.3 -112.7 257.33 A3-FJ-100C-100C -87.3 -87.0 8.86 -53.0 -52.7 -49.6 -49.3 242.08 BM -147.7 -144.2 8.68 -113.4 -109.9 -110.0 -106.5 232.25 B1-RP-100C -56.5 -59.3 8.58 -22.2 -25.0 -18.7 -21.6 226.41 B1-FJ-100C-100C -101.5 -102.5 9.39 -67.2 -68.2 -63.8 -64.8 274.63 B3-RP-300C -85.1 -84.1 8.53 -50.8 -49.8 -47.4 -46.4 223.91 B3-FJ-300C-300C -87.4 -87.8 8.87 -53.1 -53.5 -49.7 -50.1 242.83 BM -125.1 -121.7 10.08 -90.8 -87.4 -87.4 -84.0 323.46 C1-RP-100C -104.6 -101.8 8.29 -70.3 -67.5 -66.9 -64.1 211.49 C1-FJ-100C-100C -121.7 -120.5 7.11 -87.4 -86.2 -84.0 -82.8 159.87 C2-FJ-300C-No -104.3 -105.7 5.24 -70.0 -71.4 -66.6 -68.0 102.40 C3-RP-300C -104.9 -99.2 10.75 -70.6 -64.9 -67.2 -61.5 379.17 C3-FJ-300C-300C -89.4 -92.4 7.32 -55.1 -58.1 -51.7 -54.7 167.73

Table 9 - MCVN DBTT and USE values put in perspective to full-size CVN through correlation reported by Lucon [49]

What is very interesting is that the A3-FJ-300C-300C specimen which on previous work [2] did not achieve the 36.6J at -30°C requirements either on the full-size or sub-size

136 specimen. In this research, even using the correlation to have a full-size specimen perspective, the highest DBTT value for material A is 49.3°C. In addition, its DBTT curve shows a 10°C range transition. Thus, at -30°C the fracture mode should be at the USE which is fully ductile. From the correlation proposed by Lucon, shown in table 8, the USE value for this specimen was 242 J which is much higher than the 36.6 J requirement.

A hypothesis to explain the values discrepancy is the number of specimens tested.

The full size and the sub-size earlier tests had only three specimens each, whereas the

DBTT curve developed used thirteen specimens in which 76% percent were under -30 °C.

This condition is equivalent or more challenging than tested previously, as it can be verified in the previous Figure 60. In addition, fractography was not used to validate the measured values in the previous work [2], therefore, no weld defect on fracture face was considered.

In a larger volume, as in the full-size specimen, the chances of encounter weld defects are higher than in the KLST specimen. Besides that, all MCVN specimens used on the DBTT curve had its fracture surfaces assessed in order to identify weld defects.

The table also shows that the worst DBTT results were for the 100C preheat conditions and the root pass only B1-RP-100C and C1-RP-100C, those are ones that formed a large amount of martensite and did not have a subsequent welding pass to temper its microstructure. Considering the USE values, as long the specimens are in the ductile fracture mode all the values are much higher than the limit of 36.6J.

4.7. Computational Model

137

The model was used to describe the thermal history of the two GMAW filling passes over the HLAW weld bead and the base material plate. In order to calibrate the model, the thermal history acquired with type k thermocouples were compared with the predicted by the modeling in the three spacing distances, 2, 3 and 4mm. Figure 63 presents the model data and thermal history plot of the nodes on the defined spacing distances.

Figure 63 - Numerically calculated thermal history plot on the nodes of 2, 3 and 4mm

The thermal history simulated presented a similar cooling behavior, however, the peak temperatures reached were higher than the measured ones. The welding parameters set in the simulation were the real ones defined on the welds. Still there is some source of

138 uncertainty since the heat source modeling is also based on the molten pool geometry.

Without using high-speed camera is hard to measure precisely this data.

The simulation allowed visualizing temperature fields on the cross-section once the heat source is crossing a defined plane as presented in Figure 64, this data is used to predict where the locations of the A 3, A 1 regions on the cross-section are.

Figure 64 - Numerical simulation of temperature fields applied by the filling passes

The temperature fields presented shown that there is a large region of the HLAW pass that is heated above the A 1 temperature. This means re-austenitization of part of the bainite which was cooled down and might form brittle and harder microstructures.

139

Chapter 5: Feasibility

There are some situations that a switch for a new process can be justified if detailed comparison established. The hybrid laser process could improve the productivity in four or six times. The biggest advantage presented is welding speed since the HLAW travel speed is at least six times faster than SMAW and three times faster than GMAW, more welds could be done per day, so the total time length of the work could be done sooner which means it saves money.

The Hybrid laser configuration is able to weld in very high speeds 2.5-3.5 m/min but is important to understand that the window process could be severely reduced if any parameter is changed, in order to reduce this dependence and stay in a more robust configuration the first target is to reach speeds within 1.5-2 m/min. As it can be noticed this travel speed is at least seven times more than a regular SMAW welding (weld speed approximately 0.15-0.2 m/min) and at least the triple of the GMAW (0.4 – 0.5 m/min).

The critical procedure is the root and the second pass because all others sequence of welding and sections needs to wait for these steps. Hence, if at least the first two passes were made by HLAW it would mean achieve almost the double of welds per day. Thus, severally decreasing the work schedule. Which means less cost and start the facility earlier, which is also increase profit. So changing the welding process is a direct way to increase gains for the company.

In a practical way the benchmark today is 50 welds per day, if welded faster a rate of 90 welds per day is achievable that is already a strong improvement but also the

140 defect/repair rate should be reduced, in a more controlled process, to a maximum of 1% this would short even more the work schedule.

As detailed below each work day costs around 350000 USD. So even expensive equipment, (3000000 USD), and the initial research cost could be covered if, only in one construction, 14 days could be shortened in 6 months’ time frame. However, it is necessary to consider that the cost of work will increase but the time reduction will impact in all others constructions in the future, also, it is supposed to reduce much more than just 2 weeks so the savings could be much more.

A simple estimative of savings due to time reduction is demonstrated in sequence.

5.1. Time per weld

The time to perform a butt joint in pipelines is dependent on the diameter and thickness, but it also highly dependent on the welding process, travel speed, and deposit rate. However, the weld time should not be considered only as the open arc time, or considering the ratio between open arc time and cleaning. Weld time in a practical way considers joint preparation, not bevel machining, but cleaning, preheating time, mounting or not equipment etc. All these steps take place when more than one welding process or method is compared.

A regular gas pipeline joint in Brazil would be 28 in outer diameter per ½ in thick to perform a weld on this pipeline if the procedure is made linearly it would take 3 hours at least. As an alternative the welding work is done in a sequence of parallel passes, one team does the root and second pass and then move to another joint, other team comes for

141 the next two passes and finally other team finishes the next two or four passes. By doing this, the time spent to perform a whole weld is reduced to approximately of 1 hour.

5.2. Welds per day

One way to control the production rate considering all that previous factors presented on the previous topic is the amount of weld made per day. Considering this rate all the side procedures regarding weld are already included in this number.

Even on offshore pipeline construction this rate is very important since contractors and equipment receive by workday.

Also in onshore pipelines constructions, the same evaluation can be applied. By specification of API SPEC 5L [18],each pipe has 12 meters nominal length or one weld at each 12 meters, roughly 84 welds per kilometer.

On the very beginning of pipeline construction in Brazil, the onshore construction used to assembly 700 meters per day while a lay barge would be able to deliver 300 meters per day.

Nowadays the cost of offshore infrastructure required a huge improvement on automation and control of process reaching rate of 1.5-2 km per day. In the opposite way, onshore production has been reduced to 0.5km per day. Since the cost of staff, equipment, and infrastructure is severally less than the offshore the same development or investment in equipment, welding process and techniques were not required. Consequently, less automation was developed which means that the process applied are still highly manual than with the improvement of safety procedures the work rate per day was downgraded.

142

5.3. Overall cost per day

There is still a lot of discussion regarding how to evaluate the weld cost, most of the time only equipment, consumables, and staffs are measured, however other variables as repair rate, efficiency, and production rate are not measured.

So in order to establish a comparison, a mean cost per day can be defined, even though, it will vary with project requirements and locations since some places are easier to build and others not, so a mean from the four long pipelines constructed was taken. An overall cost was developed simply dividing the total cost of the project by the length in days required to finish it. Of course, there are a lot of things that are together, not only the weld. However, the weld is a critical part and generally is the longest during the work so this approach gives an idea that how much a faster or more productive welding process can impact in cost.

The mean cost per day based on the last four pipelines was approximately nine hundred thousand Reais (Brazilian currency - BRL) per day some closer to a 350 thousand

USD per day. So now it is possible to consider that even a more expensive welding process could save 350.000 USD at each day shorted at the schedule.

5.4. The SMAW Pipeline construction in Brazil

In Brazil, the most widely welding process used the shielded metal arc welding

(SMAW) which is one of the most versatile welding processes and theoretically a

143 dominated technique which would not require highly skilled welders. However, this does not justify the process in every, also, there is room for a lot improvement if a proper welding process is selected for every case.

The most reliable advantages of the process are the versatility and that is non- expensive welding process. However, this “non-expensive” title is obtained only considering the cost of equipment, consumables, and staff and is not considering, the low efficiency of the process, or the extremely low productivity.

If the same weld could be done in a half of the time with a welding process that cost the double, the overall cost is still the same but the facility is obtained sooner which means cost optimization.

5.5. Defects rate and rework

As a noun automated process and highly dependent of welders the repair rate still presents an inconvenient rate of 6% of repair rate due defects detected on NDE inspection this a terrible number and should be counted when the welding process cost is assessed.

It is important to take into account that the repair rate is not only the sum an extra

6% of work since it is now necessary properly remove the whole weld at the defect region and do it again. In addition, the probability of having another defect once the weld is open is high. Another important factor is that this weld is not produced as the construction series, so it highly increases the time expended on those welds the are required to be repaired.

144

5.6. The GMAW Pipeline construction in Brazil

An alternative to getting more productivity in the root pass and in order to reduce the defect/repair rate is the use of GMAW. Which has been required for the new pipelines projects. But it can be applied as a manual process, so again highly dependent on welder’s skills, or automated process which should be more reliable. Even though this process presents an improvement from SMAW, the welding speeds are roughly 0.4m/min which is far less the LASER could achieve besides the repair rate are still in 1.5-2% which is good but a fully automated process if properly calibrated should reduce this rate to less than 1%.

145

Chapter 6: Conclusions

Based on the results obtained and with the objectives proposed is possible to conclude that:

− The evolution of the laser welding over the decades has developed a hybrid

combination, which is capable of being applied in heavy industry such as Oil and gas.

− It is possible to produce sound welds by hybrid laser welding on both API steel grades

studied X65 and X70 meeting the requirements of toughness and hardness proposed.

− 15mm and 10mm penetration HLAW root pass were achieved at 2m/s welding speeds.

− The current standard hardness linear measurement does not have the resolution to

reproduce the weld. Hardness maps are required in order to represent the whole joint.

− The preheating conditions considered severely changed the resulting microstructure

and therefore the resulting hardness and toughness.

− 100 °C preheating was not enough to inhibit the high hardness resulting from the

HLAW thermal cycle. Even though 100 °C is a typical heating temperature in order to

remove moisture of the pipe and therefore it would not be an extra step or challenge to

apply, its resulting microstructure had a strong presence of martensite and the hardness

levels exceeded 310HV 0.2 .

− 300 °C preheating was enough to change the thermal cycle and effectively reduce the

martensitic content and reduce the resulting hardness below the 248HV 0.2 [32]

requirement proposed on previous work [2].

146

− The hardness requirement [32] may not be a proper target for an initial research. In

addition, the BS 4215 [33] only presents 300HV 0.2 requirements and even higher values

could be accepted if hydrogen induced cracking - HIC is not a concern. Most of the

hardness requirements were established for arc welding and should not necessarily

applicable to a novel welding process. In 2004 Moore [26] also stated that the hardness

requirements should not be applied on laser welding as long the welds were approved

on the other requirements and is not an intrinsically HIC exposed process.

− Keyhole instabilities may cause weld defects or undesired properties such lack of

fusion and porosity. In addition, the incomplete detachment of molten material on the

bottom may result in undesired fusion zone and HAZ.

− The microstructure formed by the thermal cycle imposed by the hybrid laser will form

mainly bainite, martensite, and some small amount of grain boundary and

Widmanstätten ferrite. However, the chemical composition and heat input will vary the

proportion of those phases and may promote other forms of ferrite.

− Since the predominant microstructure was bainite, but the resulting properties of upper

and lower bainite are distinct, and both are formed. A more detailed examination could

clarify the proportion of each bainitic formation and its relation with the final hardness

and toughness.

− The GMAW filling passes provide a thermal cycle, which partially heat treats the

HLAW root pass. However, this not assures only tempering. In fact, materials A and C

did show an increase of hardness after filling pass thermal cycle, which is due to

precipitation hardening of small vanadium carbides.

147

− There is a localized tempering effect over the HLAW fusion zone. However, it does

not heat treats the entire pass. Thus, part of the HLAW root pass will always be in the

as-welded condition.

− The use of miniaturized Charpy V-notch specimen proved effective in order to evaluate

local properties and do not address an average behavior for the whole joint or more

than one welding pass. That was clear with the DBTT values of the full joints with 100

°C conditions, which presented the lowest DBTT values for the welds when there was

a lot of martensite on the bottom of the root pass. Only the local tempered region was

tested within the 3mm thick KLST specimen.

− The DBTT curves for base material shown an almost bimodal transition curve for the

bainitic base materials whereas the ferritic/pearlitic material presented a smoother

transition, however, at a higher temperature.

− The developed DBTT curves have shown a clear shift in transition temperature from

the base material to the welded condition

− The SEM fractography used to access fracture surfaces allowed to clearly identify the

transition on failure modes and also added reliability to the tests used to develop the

DBTT curves.

148

Chapter 7: Future Work

This study was focused on toughness and hardness study due to the reduced amount of material available. With the availability of material and funds a broader properties investigation, such crack tip opening displacement - CTOD, lateral bending and tensile should be done.

In addition, Physical simulation of the filling passes over the laser welds varying the heat input should provide a better understanding of the thermal cycle influence on this material.

Since there is a dilution gradient across the HLAW root pass the MCVN approach to developing DBTT curves should be applied to compare the microstructure at the very bottom of the root pass, at the top (under the GMAW HAZ as it was done in this research) and the correspondent HAZ of the HLAW pass.

Moore [26] et al. suggested a large influence on toughness and microstructure from the filler metal, therefore, a comparison with different filler metals for the hybrid pass will add knowledge on the expected results for the HLAW. In fact, there is a new Petrobras research development considering the use of metal core filler metals on the HLAW pass and distinct welding process on the filling passes.

149

References

[1] A. P. Mackwood and R. C. Crafer, "Thermal modeling of laser welding and related processes: a literature review," Optics and Laser Technology, vol. 37, pp. 99-115, Mar 2005. [2] M. Purslow and E. W. I.-. EWI, "Development of Mechanical Operating Envelope for HLAW – Final Report," EWI, Ohio2014. [3] D. Yapp and S. A. Blackman, "Recent developments in high productivity pipeline welding," Journal of the Brazilian Society of Mechanical Sciences and Engineering, vol. 26, pp. 89-97, 2004-03 2004. [4] R. P. Martukanitz, "A critical review of laser beam welding," in Conference on Critical Review , San Jose, CA, 2005, pp. 11-24. [5] E. Kannatey-Asibu, Principles of Laser Materials Processing vol. 4: John Wiley & Sons, 2009. [6] W. M. Steen and J. Mazumder, Laser material processing , 4th ed. London: Springer, 2010. [7] K. H. Leong, H. K. Geyer, K. R. Sabo, and P. G. Sanders, "Threshold laser beam irradiances for melting and welding," Journal of Laser Applications, vol. 9, pp. 227-231, Oct 1997. [8] T. Andreas, "Bifocal Hybrid Laser Welding: A Technology for Welding of Aluminium and Zinc-coated Steels," Doctoral, Institut für Werkzeugmaschinen und Betriebswissenschaften, Technischen Universität München , Munich, 2009. [9] J. Achebo and O. Oviemuno, "Numerical Computation of Melting Efficiency of Aluminum Alloy 5083 During CO2 Laser Welding Process," in Materials with Complex Behaviour II . vol. 16, A. Öchsner, Ed., ed Berlin Heidelberg: Springer-Verlag, 2012, pp. 601-617. [10] A. F. H. Kaplan, "Model of the absorption variation during pulsed laser heating applied to welding of electronic Au/Ni-coated Cu-leadframes," Applied Surface Science, vol. 241, pp. 362-370, Mar 15 2005. [11] A. Kaplan, "Keyhole Welding: The Solid and Liquid Phases," Theory of Laser Materials Processing: Heat and Mass Transfer in Modern Technology, vol. 119, pp. 71-93, 2009 2009. [12] Steen and William, "Arc augmented laser processing of materials," Journal of Applied Physics, vol. 51, 1980. [13] J. Weston, "Laser welding of aluminum alloys," PhD, Department of Material Science and Metallic Engineering, University of Cambridge, 1999. [14] H. Al-Kazzaz, M. Medraj, X. Cao, and M. JahaZi, "Nd : YAG laser welding of aerospace grade ZE41A magnesium alloy: Modeling and experimental investigations," Materials Chemistry and Physics, vol. 109, pp. 61-76, May 15 2008. [15] J. Dowden, "Laser Keyhole Welding: The Vapour Phase," Theory of Laser Materials Processing: Heat and Mass Transfer in Modern Technology, vol. 119, pp. 95-128, 2009 2009. [16] Matsuda, Utsumi, Katsumura, and Hamasaki, "TIG or MIG Arc Augmented Laser Welding of Thick Mild Steel Plate," Joining and Materials, vol. 1, pp. 31-34, 1988.

150

[17] Y. P. Kim, N. Alam, and H. S. Bang, "Observation of hybrid (cw Nd : YAG laser plus MIG) welding phenomenon in AA 5083 butt joints with different gap condition," Science and Technology of Welding and Joining, vol. 11, pp. 295-307, May 2006. [18] A. P. I.-. API, "Specification for Line Pipe," vol. 5L, ed. Washington DC: API Publishing Services, 2013, p. 192. [19] A. P. I.-. API, "Welding of Pipelines and Related Facilities," vol. 1104, ed: API Publishing Services, 2013, p. 118. [20] Z. Lei, C. Tan, Y. Chen, and Z. Sun, "Microstructure and Mechanical Properties of Fiber Laser-Metal Active Gas Hybrid Weld of X80 Pipeline Steel," Journal of Pressure Vessel Technology-Transactions of the ASME, vol. 135, Feb 2013. [21] R. Miranda, A. Costa, L. Quintino, D. Yapp, and D. Iordachescu, "Characterization of fiber laser welds in X100 pipeline steel," Materials & Design, vol. 30, pp. 2701-2707, Aug 2009. [22] R. Miranda, L. Quintino, S. Williams, and D. Yapp, "Welding with High Power Fiber Laser API5L-X100 Pipeline Steel," 1-2, 2010. [23] L. Quintino, A. Costa, R. Miranda, D. Yapp, V. Kumar, and C. J. Kong, "Welding with high power fiber lasers - A preliminary study," Materials & Design, vol. 28, pp. 1231-1237, 2007. [24] S. Gruenenwald, T. Seefeld, F. Vollertsen, and M. Kocak, "Solutions for joining pipe steels using laser-GMA-hybrid welding processes," Laser Assisted Net Shape Engineering 6, Proceedings of the Lane 2010, Part 2, vol. 5, pp. 77-87, 2010 2010. [25] H. L. Wei, J. J. Blecher, T. A. Palmer, and T. Debroy, "Fusion Zone Microstructure and Geometry in Complete-Joint-Penetration Laser-Arc Hybrid Welding of Low-Alloy Steel," Welding Journal, vol. 94, pp. 135S-144S, Apr 2015. [26] P. L. Moore, D. S. Howse, and E. R. Wallach, "Microstructures and properties of laser/arc hybrid welds and autogenous laser welds in pipeline steels," Science and Technology of Welding and Joining, vol. 9, pp. 314-322, Aug 2004. [27] S. Liu, H. Zhang, J. Hu, and Y. Shi, "Microstructure of Laser-MAG Hybrid Welds of Sintered P/M Steel," Journal of Materials Engineering and Performance, vol. 22, pp. 251- 257, Jan 2013. [28] M. Rethmeier, S. Gook, M. Lammers, and A. Gumenyuk, "Laser-Hybrid Welding of Thick Plates up to 32 mm Using a 20 kW Fibre Laser," QUARTERLY JOURNAL OF THE JAPAN WELDING SOCIETY, vol. 27, pp. 74s-79s, 2009. [29] G. Sergej, G. Andrey, and RethmeierMichael, "Weld Seam Formation and Mechanical Properties of Girth Welds Performed with Laser-GMA-Hybrid Process on Pipes of Grade X65," 2010. [30] S. Gook, A. Gumenyuk, and M. Rethmeier, "Hybrid laser arc welding of X80 and X120 steel grade," Science and Technology of Welding and Joining, vol. 19, pp. 15-24, Jan 2014. [31] C. Roepke and S. Liu, "Hybrid Laser Arc Welding of HY-80 Steel," Welding Journal, vol. 88, pp. 159S-167S, Aug 2009. [32] A. N. /ISO, "Petroleum, petrochemical, and natural gas industries — Materials for use in H2S - containing environments in oil and gas production

151

Part 1: General principles for selection of cracking - resistant materials," in Part 1: General principles for selection of cracking - resistant materials vol. MR 0175, ed. Houston, TX: NACE Internationa, 2015, p. 164. [33] B. S. Institution, "Specification for welding of steel pipelines on land and offshore Part 1: Carbon and carbon manganese steel pipelines," vol. BS 4215-1, ed: British Standards, 2009, p. 80. [34] A. S. f. T. a. M.-. ATSM, "E2248 Impact Testing of Miniaturized Charpy V-Notch Specimens," vol. E2248, ed. United States of America: ASTM, 2015, p. 7. [35] B. S. I.-. BSI, "ISO 14556 Metallic materials - Charpy V-notch pendulum impact test - Instrumented test method," vol. 14556, ed. United Kingdom, 2015, p. 30. [36] E. Lucon, "Effect of Electrical Discharge Machining (EDM) on Charpy Test Results from Miniaturized Steel Specimens," Journal of Testing and Evaluation, vol. 41, pp. 1-9, Jan 2013. [37] P. Xue, Y. Komizo, R. Ueji, and H. Fujii, "Enhanced mechanical properties in friction stir welded low alloy steel joints via structure refining," Materials Science and Engineering a-Structural Materials Properties Microstructure and Processing, vol. 606, pp. 322-329, Jun 12 2014. [38] J. A. Avila, E. Lucon, J. Sowards, P. R. Mei, and A. J. Ramirez, "Assessment of Ductile-to- Brittle Transition Behavior of Localized Microstructural Regions in a Friction-Stir Welded X80 Pipeline Steel with Miniaturized Charpy V-Notch Testing," Metallurgical and Materials Transactions A, pp. 1-11, 2016. [39] E. Lucon, "Testing of Small-Sized Specimens," Comprehensive Materials Processing, Vol 1: Assessing Properties of Conventional and Specialized Materials, pp. 135-163, 2014. [40] R. Lagneborg, T. Siwecki, S. Zajac, and B. Hutchinson, "The role of vanadium in microalloyed steels," Scandinavian Journal of Metallurgy, vol. 28, pp. 186-241, Oct 1999. [41] G. Thewlis, "Materials perspective - Classification and quantification of microstructures in steels," Materials Science and Technology, vol. 20, pp. 143-160, Feb 2004. [42] H. K. D. H. Bhadeshia, R. W. K. Honeycombe, and ebrary Inc. (2006). Steels microstructure and properties (3rd ed.) . [43] E. S. Davenport and E. C. Bain, "Transformation of austenite at constant subcritical temperatures," Transactions of the American Institute of Mining and Metallurgical Engineers, vol. 90, pp. 117-154, 1930. [44] H. Bhadeshia and J. W. Christian, "BAINITE IN STEELS," Metallurgical Transactions a- Physical Metallurgy and Materials Science, vol. 21, pp. 767-797, Apr 1990. [45] H. Bhadeshia, "THE LOWER BAINITE TRANSFORMATION AND THE SIGNIFICANCE OF CARBIDE PRECIPITATION," Acta Metallurgica, vol. 28, pp. 1103-1114, 1980 1980. [46] H. Bhadeshia and D. V. Edmonds, "THE MECHANISM OF BAINITE FORMATION IN STEELS," Acta Metallurgica, vol. 28, pp. 1265-1273, 1980 1980. [47] W. International Institute of and I. Welding, Compendium of weld metal microstructures and properties : submerged-arc welds in ferritic steel = Collection de microstructures du metal fondu et proprietes mecaniques correspondantes : soudures à l'arc sous flux solide sur acier ferritique . Cambridge, UK: Published for the International Institute of Welding by the Welding Institute, 1985. [48] E. C. Bain and H. W. Paxton, Alloying elements in steel , 2nd ed. Metals Park, Ohio: American Society for Metals, 1961. 152

[49] L. E., M. C.N., and S. R.L. (2015, Overview of NIST activities on sub-size and miniaturized charpy specimens: Correlations with full-size specimens and verification specimens for small-scale pendulum machines. American Society of Mechanical Engineers, Pressure Vessels and Piping Division [50] O. L. Towers, Charpy V-notch tests : influences of striker geometry and of specimen thickness . Cambridge [England]: The Welding Institute, 1983. [51] K. Wallin, New improved methodology for selecting charpy toughness criteria for thin high strength steels . [Espoo: VTT, 1994. [52] E. Lucon, R. Chaouadi, A. Fabry, J. L. Puzzolante, and E. Van Walle, "Characterizing material properties by the use of full-size and sub-size Charpy tests: An overview of different correlation procedures," Pendulum Impact Testing: a Century of Progress, pp. 146-163, 2000.

153