Porosity Reduction and Elimination in
Laser Welding of AA6014 Aluminium
Alloys for Automotive Components
Manufacture and Industrial Applications
A thesis submitted to The University of Manchester for the degree of
Doctor of Philosophy (PhD)
in the Faculty of Science and Engineering
2016
Ahmad Wael Al Shaer
School of Mechanical, Aerospace and Civil Engineering
TABLE OF CONTENTS
Table of Contents ........................................................................................ 2
List of Figures .............................................................................................. 8
List of Tables ............................................................................................. 15
Nomenclature............................................................................................. 17
List of Operators ....................................................................................... 20
List of Abbreviations................................................................................. 21
Abstract ...................................................................................................... 22
Declaration ................................................................................................. 23
Copyright Statement ................................................................................. 24
Acknowledgements.................................................................................... 25
Publications ................................................................................................ 26
Chapter 1 Introduction ............................................................................ 27
1.1
Overview .......................................................................................................... 27
1.2
Research motivation ......................................................................................... 28
1.3
Aim and objectives ........................................................................................... 30
1.4
Thesis outline ................................................................................................... 31
Chapter 2 A Literature review Part I: Laser welding technology ..... 33
2.1
Introduction ...................................................................................................... 33
2.2
Fusion Welding ................................................................................................ 35
2.3
Laser Welding Mechanism ............................................................................... 37
2.3.1
Conduction Mode ...................................................................................... 38
2.3.2
Keyhole Mode ........................................................................................... 39
2.4
Aluminium Alloys ............................................................................................ 43
2
2.4.1
Aluminium-Mn alloys (3000 series) ......................................................... 47
2.4.2
Aluminium-Si alloys (4000 series) ........................................................... 47
2.4.3
Aluminium-Mg alloys (5000 series) ......................................................... 48
2.4.4
Aluminium-Mg-Si alloys (6000 series) .................................................... 48
2.5
Porosity in welded Al alloys ............................................................................ 49
2.6
Porosity formation mechanism ......................................................................... 51
2.6.1
Pores shape and size .................................................................................. 52
2.6.2
Effect of process parameters on porosity formation ................................. 53
2.6.3
Hydrogen solubility................................................................................... 59
2.6.4
Solubility of Oxygen ................................................................................. 62
2.6.5
Nitrogen solubility .................................................................................... 62
2.6.6
Keyhole Collapse ...................................................................................... 63
2.7
Porosity reduction............................................................................................. 66
2.7.1
Effects of magnetic field ........................................................................... 66
2.7.2
Effects of the shielding gases .................................................................... 68
2.7.3
Effects of surface preparation ................................................................... 71
2.7.4
Effects of laser beam characteristics ......................................................... 74
2.8
Summary .......................................................................................................... 76
Chapter
3
A Literature review Part II: Smoothed Particle
hydrodynamics (SPH) modelling of laser processing ............................ 78
3.1
Introduction ...................................................................................................... 78
3.2
Laser cleaning/ablation modelling ................................................................... 79
3.2.1
Mesh-based modelling of material processing .......................................... 81
3.2.2
Mesh-less based modelling of material processing................................... 83
3.3
SPH interpolation and summation .................................................................... 87
3.3.1
SPH modelling of material processing ...................................................... 90
3
3.4
Summary .......................................................................................................... 93
Chapter 4 Materials & Experimental set-up ......................................... 94
4.1
Materials and experimental set-up ................................................................... 94
4.2
Laser welding process ...................................................................................... 95
4.3
Laser cleaning process ...................................................................................... 96
4.4
Samples preparation ......................................................................................... 97
4.5
Scanning electron microscope (SEM): ............................................................. 97
4.6
X-ray Diffraction technique: ............................................................................ 97
4.7
Laser-Induced Breakdown Spectroscopy (LIBS): ........................................... 98
4.8
UV-VIS-IR Spectrometer ................................................................................. 98
4.9
White Light Interferometer .............................................................................. 99
4.10
Summary .......................................................................................................... 99
Chapter 5 Understanding the effect of heat input and sheet gap on
porosity formation in laser fillet edge and flange couch welding of AC170PX aluminium alloy for automotive component manufacture ..... 103
5.1
Abstract .......................................................................................................... 103
5.2
Introduction .................................................................................................... 103
5.3
Materials and experimental procedure ........................................................... 107
5.4
Results and Discussion ................................................................................... 111
5.4.1
Macrostructure and microstructure characteristics ................................. 111
5.4.2
Porosity characteristics............................................................................ 115
5.4.3
Weld penetration and fusion zone geometry characteristics ................... 122
5.4.4
Tensile strengths and micro-hardness ..................................................... 125
5.4.5
Microhardness ......................................................................................... 126
5.5
Summary ........................................................................................................ 129
4
Chapter 6 Effect of filler wire properties on porosity formation in
laser welding of AC-170PX aluminium alloy for automotive component
manufacture ............................................................................................. 132
6.1
Abstract .......................................................................................................... 132
6.2
Introduction .................................................................................................... 132
6.3
Materials and experimental procedure ........................................................... 135
6.4
Experimental results ....................................................................................... 139
6.4.1
Fillet edge joints ...................................................................................... 139
6.4.2
Flange couch joints ................................................................................. 146
6.5
Discussion ...................................................................................................... 151
6.6
Summary ........................................................................................................ 153
Chapter 7 The effects of short pulse laser surface cleaning on porosity
formation and reduction in laser welding of aluminium alloy for
automotive component manufacture..................................................... 156
7.1
Abstract .......................................................................................................... 156
7.2
Introduction .................................................................................................... 157
7.3
Materials and experimental procedure ........................................................... 160
7.4
Experimental Results ...................................................................................... 164
7.4.1
Effect of laser cleaning on the surface characteristics ............................ 164
7.4.2
Porosity characteristics after laser welding with and without laser pre-
cleaning …………………………………………………………………………168
7.4.3
Weld dimensions ..................................................................................... 171
7.5
Discussion ...................................................................................................... 173
7.6
Summary ........................................................................................................ 175
5
Chapter 8
Smoothed Particle Hydrodynamics (SPH) modelling of
transient heat transfer in pulsed laser ablation of Al and associated
free-surface problems ............................................................................. 179
8.1
Abstract .......................................................................................................... 179
8.2
Introduction .................................................................................................... 180
8.3
Model Description .......................................................................................... 183
8.3.1
8.4
Physical Phenomena................................................................................ 183
SPH Methodology .......................................................................................... 186
8.4.1
SPH Interpolation .................................................................................... 186
8.4.2
Kernel gradient correction....................................................................... 188
8.4.3
Governing Equations and SPH Discretisation ........................................ 189
8.4.4
Time-step scheme and CFL number ....................................................... 192
8.4.5
Boundary Conditions .............................................................................. 193
8.5
Results and Discussion ................................................................................... 193
8.5.1
Kernel Gradient Correction (KGC) ......................................................... 194
8.5.2
Laplacian Operator Correction (Schwaiger correction) .......................... 197
8.5.3
Transient Heat Transfer Test Cases ........................................................ 201
8.5.4
Laser Ablation Model ............................................................................. 211
8.6
Summary ........................................................................................................ 224
Chapter 9 Conclusions and future work ............................................... 227
9.1
Conclusions .................................................................................................... 227
9.1.1
Experimental work .................................................................................. 227
9.1.2
SPH modelling ........................................................................................ 230
9.2
Limitations of the SPH model ........................................................................ 231
9.3
Future work recommendations ....................................................................... 232
9.3.1
Experimental work .................................................................................. 233
6
9.3.2
SPH modelling ........................................................................................ 233
Appendix A .............................................................................................. 258
Appendix B .............................................................................................. 277
Total word count: 56,014
7
LIST OF FIGURES
Figure 2.1 Master Chart of various welding processes (Khan, 2007) ............................. 34
Figure 2.2 Common types of welding joints (Minnick, 2008)........................................ 36
Figure 2.3 Schematic of the welding area (Black et al., 2008) ....................................... 37
Figure 2.4 (a) Conduction mode (Dahotre and Harimkar, 2008) (b) Keyhole mode
(Ion, 2005a) ..................................................................................................................... 38
Figure 2.5 Shape of the weld with a keyhole at different welding speeds (Dahotre and
Harimkar, 2008) .............................................................................................................. 40
Figure 2.6 Keyhole in deep penetration welding (Schubert et al., 2001) ....................... 41
Figure 2.7 Plasma formation in laser welding (reformatted for the sake of clarity) (Steen
and Mazumder, 2010) ..................................................................................................... 41
Figure 2.8 The global consumption of Al (a) in 2010 (Melik and Kouzmenkov, 2010)
(b) in 2014 (Klein et al., 1994) ....................................................................................... 44
Figure 2.9 Scanning electron micrograph of a small pore in a titanium alloy (Zhou and
Tsai, 2007a) ..................................................................................................................... 52
Figure 2.10 Different types of porosity distribution (Gupta, 2015) ................................ 53
Figure 2.11 SEM picture of various pores shapes (a) spherical pores due to gases
entrapment (also called metallurgical pores) (b) irregular pores due to the keyhole
collapse (Yao and Gong, 2011) ....................................................................................... 53
Figure 2.12 Relationship between welding velocity and porosity formation (Katayama
and Kawahito, 2009) ....................................................................................................... 54
Figure 2.13 Main flow directions during ice water welding using a CO 2 laser, laser
power: 500W, welding velocity: 2m/min (Berger et al., 2011) ...................................... 55
Figure 2.14 Pore morphologies in (a) partial penetration and (b) full penetration laser
welds in AA5083 (Chang et al., 2013) ........................................................................... 56
Figure 2.15 Porosity inside the Al weld under various settings (weld length: 200mm)
(Tao et al., 2013) ............................................................................................................. 57
Figure 2.16 (a) Variation of solubility of H2 in Al with temperature (ASM International,
1993), (b) solubility of H2 in Al and iron (Howden, 1971)............................................. 60
Figure 2.17 Effect of alloying elements on the solubility of hydrogen in liquid Al at 973
K and 1 atm partial pressure of H2 (Anyalebechi, 1995) ................................................ 61
8
Figure 2.18 (a) Schematic diagram showing the formation of the void at weld root due
to the imperfect collapse of Keyhole (left) (Pastor, 1998) (b) Liquid projection
formation at a level where the vapour pressure force approximately balances the surface
tension force (right) (Schauer and Giedt, 1978).............................................................. 65
Figure 2.19 Macro sections of two welds (a) a reference case without a magnetic field
(b) the case with a magnetic field with different polarity (Bachmann et al., 2013) ....... 67
Figure 2.20 Effect of shielding gas type on porosity formation in 5083 Al Co2 laser
welding (Seto et al., 2000) .............................................................................................. 68
Figure 2.21 Temporal behaviour of metallic plasma and keyhole in CO2 laser welding
using nitrogen shield (intermittent generation of Nitrogen plasma) (Matsunawa et al.,
2003) ............................................................................................................................... 69
Figure 2.22 The effect of nozzle position on the molten pool and the gas/material
interaction in stainless steel laser welding. The gas flow is directed towards (a) the
molten pool surface (b) the keyhole opening and (c) the front of the keyhole
(Kamimuki et al., 2002) .................................................................................................. 70
Figure 2.23 X-ray inspections followed by image analysis of laser-welded beads
showing the influence of surface preparation in AA5083 (Haboudou et al., 2003b). .... 73
Figure 2.24 Effect of pulse modulation on the reduction of porosity in AA5182 laser
welding (Matsunawa et al., 2003) ................................................................................... 75
Figure 2.25 (a) Dual spot laser welding (an inline beam and a cross beam) (b) effect of
dual spot welding on porosity in AA5083 and AA356 alloys (Yao and Gong, 2011) ... 76
Figure 3.1 Flow chart of a typical SPH thermal simulation approach ............................ 90
Figure 4.1 (a) Fillet edge joint (b) Flange couch joint with offset .................................. 95
Figure 4.2 Incident beam and its inclination angles of the normal plane ....................... 96
Figure 4.3 Hitachi High Technologies S-3400N SEM ................................................... 97
Figure 4.4 Philips X'pert -1 Copper XRD machine (University of Manchester, 2014) .. 98
Figure 4.5 LIBSCAN 0165 laser-induced breakdown spectroscopy .............................. 98
Figure 4.6 Analytic Jena-SPECORD 250 Spectrometer ................................................. 99
Figure 4.7 Wyko NT1100 (Analytik Jena Specord 250) Interferometer ........................ 99
Figure 5.1 (a) Fillet edge joint (b) Flange couch joint with offset ................................ 107
Figure 5.2 Schematic of the Scansonic ALO3 laser welding head ............................... 109
Figure 5.3 Incident beam and its inclination angles of the normal plane ..................... 109
9
Figure 5.4 Macro-sections of the fillet edge joints at various welding parameters where
the white spots are porosity ........................................................................................... 112
Figure 5.5 Macro-sections of the flange couch joints at various welding parameters .. 113
Figure 5.6 Microstructure of the weld-HAZ-Base material of a fillet edge joint welded
using 2000W and 20 mm/s............................................................................................ 114
Figure 5.7 Microstructure of the weld-HAZ-Base material of a flange couch joint
welded using 3800 W and 35 mm/s .............................................................................. 114
Figure 5.8 X-ray diffraction test results for the AC170-PX (AA6014) joints welded
using 4043 filler wire (phases detected are Al-Si , Mg2Si , SiO2) ................................ 115
Figure 5.9 Porosity percentages for fillet edge joints (a) without gaps and (b) with a
0.2mm gap ..................................................................................................................... 116
Figure 5.10 Porosity percentages for flange couch joints (a) without gaps and (b) with a
0.2 mm gap .................................................................................................................... 117
Figure 5.11 Porosity distribution curve for a fillet edge joint welded at 5300 W and 50
mm/s .............................................................................................................................. 119
Figure 5.12 Porosity distribution curve for a flange couch joint welded at 2200 W and
20 mm/s ......................................................................................................................... 119
Figure 5.13 Fillet edge weld dimensions (t1 and t2 sheets’ thicknesses, s the gap size, Pw
weld width, and Pd penetration depth ........................................................................... 122
Figure 5.14 Flange couch weld dimensions (t1 and t2 sheets’ thicknesses, s the gap size,
Pw weld width, and Pd penetration depth) ..................................................................... 123
Figure 5.15 Penetration depths and weld width values for the fillet edge joints .......... 123
Figure 5.16 Penetration depths and weld width values for the flange couch joints ...... 124
Figure 5.17 (a) Ultimate tensile strength values for fillet edge joints (b) UTS values as a
percentage of the base materials strength 195,000 kN/m2 ............................................ 125
Figure 5.18 (a) Ultimate tensile strength values for Flange couch joints (b) UTS values
as a percentage of the base materials strength 195,000 kN/m2 ..................................... 126
Figure 5.19 Microhardness profile for a fillet edge joint welded at 3500W and 35 mm/s
....................................................................................................................................... 127
Figure 5.20 Microhardness profile for a flange couch joint welded at 3800W and
35mm/s .......................................................................................................................... 128
Figure 6.1 Two types of joint configurations used in this investigation (a) Fillet edge
and (b) flange couch joints configuration ..................................................................... 134
10
Figure 6.2 Scansonic ALO3 Laser head ....................................................................... 137
Figure 6.3 Incident beam and its inclination angles of the normal plane ..................... 137
Figure 6.4 Microsections of fillet edge joints welded using different filler wires ........ 140
Figure 6.5 Microstructure of the weld-HAZ-base material of a fillet edge joint welded
at 5300W laser power and 50mm/s welding speed (AlSi5 filler wire) ......................... 141
Figure 6.6 Microstructure for a fillet edge joint welded using AlSi12 filler wire ........ 141
Figure 6.7 A Microstructure comparison between fillet edge joints welded using (a)
AlSi12 filler wire and (b) AlSi5 filler wire ................................................................... 142
Figure 6.8 Microstructure of the weld-HAZ-Base material of a fillet edge joint welded
at 5.3KW and 50mm/s (AlMgMn filler wire) ............................................................... 143
Figure 6.9 Weld areas used in porosity proportions determination for (a) Fillet edge and
(b) Flange couch joints. ................................................................................................. 143
Figure 6.10 Porosity content in fillet edge joints welded with different three filler wire
at 5.3KW and 50mm/s................................................................................................... 144
Figure 6.11 Fillet edge weld dimensions (t1 and t2 sheets’ thicknesses, s the gap size, Pw
weld width, and Pd penetration depth) .......................................................................... 145
Figure 6.12 Weld dimensions for fillet edge joints using various filler wires at 5.3KW
and 50mm/s ................................................................................................................... 145
Figure 6.13 Tensile strength for various filler wires welds at 5.3 KW and 50 mm/s as a
proportion of the base material UTS (195 MPa) ........................................................... 146
Figure 6.14 Microsections of flange edge joints welded using different filler wires ... 147
Figure 6.15 Microstructure of the weld-HAZ-Base material of a flange couch joint
welded at 3500W and 35mm/s (AlMg4,5Mn filler wire) ............................................. 148
Figure 6.16 Porosity content in flange couch joints welded with different three filler
wire at 3.5KW and 35mm/s .......................................................................................... 149
Figure 6.17 Flange couch weld dimensions (t1 and t2 sheets’ thicknesses, s the gap size,
Pw weld width, and Pd penetration depth) ..................................................................... 149
Figure 6.18 Weld dimensions for flange couch joints using various filler wires at
3.5KW and 35mm/s ...................................................................................................... 150
Figure 6.19 Tensile strength for various filler wires welds at 3.5KW and 35mm/s as a
proportion of the base material UTS (195 MPa) for the flange couch joints ................ 151
Figure 7.1 (a) Fillet edge joint (b) Flange couch joint with offset ................................ 162
Figure 7.2 Incident beam and its inclination angles of the normal plane ..................... 164
11
Figure 7.3 A cross-section view of the laser cleaned/uncleaned interface showing the
material removal depth by laser cleaning...................................................................... 164
Figure 7.4 Tomography of (a) a reference surface for which the roughness is Ra= 982
nm and of (b) cleaned surface with a roughness Ra= 882 nm ....................................... 165
Figure 7.5 Cross sections of (a) not cleaned surface and (b) a laser cleaned surface. .. 165
Figure 7.6 SEM image of surface characteristics in which the bottom part is laser
cleaned while the top portion is kept as a reference...................................................... 166
Figure 7.7 SEM images of surface characteristics (a) not cleaned surface of the treated
AC170-PX and (b) the same surface after laser cleaning ............................................. 167
Figure 7.8 Elemental analysis of as-received AA6014 surface .................................... 168
Figure 7.9 Elemental analysis of laser-cleaned AA6014 surface ................................. 168
Figure 7.10 Comparison in porosity level between laser fillet edge welded joints (a)
without cleaning and (b) with laser cleaning ................................................................ 169
Figure 7.11 Comparison in porosity level in laser flange couch welded joints (a) without
laser cleaning (b) with laser cleaning ............................................................................ 170
Figure 7.12 Porosity percentages in laser welded joints with and without laser cleaning
(a) for fillet edge joints, (b) flange couch joints............................................................ 171
Figure 7.13 Comparisons in weld width and penetration depth between a laser-cleaned
fillet edge joint and welds without laser cleaning ......................................................... 172
Figure 7.14 Comparisons in weld width and penetration depth between laser-cleaned
flange couch joints with and without laser cleaning ..................................................... 172
Figure 7.15 Reflectance of (a) the un-cleaned AC-170PX surface and (b) cleaned
surface using the Nd:YAG laser (1.06 µm) .................................................................. 173
Figure 8.1 Simplified schematic laser ablation of a metallic sample ............................ 184
Figure 8.2 Particles i and j within the smoothing kernel support ................................. 188
Figure 8.3 Functions used to validate the kernel truncation correction at the free surface
in SPH ........................................................................................................................... 194
Figure 8.4 Kernel gradient Correction (KGC) effect on SPH results at the free ends of
the 1-D domain.............................................................................................................. 196
Figure 8.5 L2 Norm Errors in SPH results for the gradient of different functions
with/without Kernel Gradient Correction ..................................................................... 197
Figure 8.6 Effect of Schwaiger operator without KGC on SPH results at the boundaries
of the 1-D domain ......................................................................................................... 198
12
Figure 8.7 Effect of Schwaiger operator with KGC on SPH results at boundaries of the
1-D domain ................................................................................................................... 200
Figure 8.8 L2 Norm Errors in SPH results for the second derivative of different
functions with/without correction ................................................................................. 200
Figure 8.9 SPH computational domain (not to scale) ................................................... 202
Figure 8.10 Surface temperature using different particles spacing for single Laser pulse
tp=100 ns ....................................................................................................................... 205
Figure 8.11 L2 Norm Error for SPH results at different resolutions ............................ 206
Figure 8.12 Effect of time step in SPH modelling using the step-predictor corrector time
scheme ........................................................................................................................... 207
Figure 8.13 Single Laser pulse of 100 ns and 75% surface reflectivity on the Al target
using the 3-D SPH code ................................................................................................ 208
Figure 8.14 SPH modelling of three consecutive laser pulses of 100 ns pulse duration
and 100 ns relaxation time (a) 95% surface reflectivity (b) 75% surface reflectivity .. 209
Figure 8.15 Temperature history during laser heating (a) Temperature variation over
time (b) Temperature profile across the sample ............................................................ 210
Figure 8.16 Heating/Cooling rates at the surface during three consecutive laser pulses
....................................................................................................................................... 210
Figure 8.17 Spatial distribution of the Laser intensity as a function of time ................ 212
Figure 8.18 3-D view of the ablated surface showing the temperature evolution and
phase change with time. Particles are ejected within 1.5 ns time (single shot at 10 J/cm 2
and 10ns pulse duration) ............................................................................................... 214
Figure 8.19 Cross section of the Al sample showing the ablation with taper effect (using
single shot at 10 J/cm2 and 10 ns pulse duration) ......................................................... 215
Figure 8.20 Temperature change with time for a surface particle obtained by SPH model
(Region I: conduction heating, Region II: further heating, Region III: partial cooling
after ejection) ................................................................................................................ 217
Figure 8.21 SPH values of vapour pressure for surface particles a function of
temperature.................................................................................................................... 218
Figure 8.22 Vapour velocity vectors at the workpiece surface (a) 3-D view at 2 ns (b)
half-section to be considered in the following subfigures (c) cross section B-B at 1 ns
(d) cross section B-B at 3.0 ns ...................................................................................... 219
13
Figure 8.23 SPH results for the ejected particles’ velocity with time (Region I:
stationary state, Region II: instant ejection, Region III: velocity change with
temperature, Region IV: stable speed) .......................................................................... 220
Figure 8.24 Ablation depths at different fluences 4-23 J/cm2 using 10 ns pulse with 30
kHz repetition rate. SPH results are compared with experimental and numerical values
reported in (Lutey et al. 2013)....................................................................................... 222
Figure 8.25 SPH results of Ablation depth compared to literature experimental and
numerical data at different fluences 4-23 J/cm2 using 6 ns pulse. ................................ 223
14
LIST OF TABLES
Table 2.1 Comparison of various joining processes (Gillner et al., 2000) (reformatted)35
Table 2.2 Ionisation potential and other properties of common shielding gases (Ion,
2005a).............................................................................................................................. 42
Table 2.3 Standard forms and applications of wrought Al alloys (Mathers, 2002) ........ 46
Table 2.4 Chemical composition of three Al alloys used in automotive industry, (Note:
single-numbers refer to the maximum limits) (Kamat et al., 2002)................................ 49
Table 3.1 Selected published work on heat transfer modelling using some common
meshless methods ............................................................................................................ 84
Table 3.2 A number of published works on metal processing using some common
meshless methods ............................................................................................................ 86
Table 4.1 Chemical composition of the Al alloys (wt%) (Novelis Deutschland GmbH,
2011) ............................................................................................................................... 94
Table 4.2 Physical and mechanical properties of AC-170PX after tempering (Novelis
Deutschland GmbH, 2011).............................................................................................. 95
Table 4.3 Laser cleaning optical and process parameters (Sun and Karppi, 1996) ........ 96
Table 5.1 Chemical composition of the Al alloys (wt%) (Novelis Deutschland GmbH,
2011; Anon, 2011b) ...................................................................................................... 107
Table 5.2 Physical and mechanical properties of AC-170PX after tempering (Anon,
2011b; Novelis Deutschland GmbH, 2011) .................................................................. 107
Table 5.3 TRUMPF TruDisk 5302 laser unit technical Properties (Trumpf-Lasers,
2013) ............................................................................................................................. 108
Table 5.4 Welding parameters and line energy (J/mm) for fillet edge and flange couch
joints .............................................................................................................................. 110
Table 6.1 Chemical composition of the Al alloys (wt%) (Anon, 2011a; Kutsuna et al.,
2006; Efunda, 2016) ...................................................................................................... 135
Table 6.2 Physical and mechanical properties of AC-170PX (after tempering) as well as
AA4043 filler wire (Efunda, 2016; Kutsuna et al., 2006; Anon, 2011a) ...................... 136
Table 6.3 TruDisk 5302 laser unit technical properties (Trumpf-Lasers, 2013) .......... 138
Table 6.4 Welding parameters for fillet edge joints ...................................................... 139
Table 6.5 Welding parameters for flange couch joints ................................................. 146
Table 7.1 Chemical composition of the Al alloys (wt-%) (Anon 2011a, 2011b) ......... 161
15
Table 7.2 Physical and mechanical properties of AC-170PX and AA4043 filler wire
(Anon 2011a, 2011b) .................................................................................................... 161
Table 7.3 cleanLASER CL600 laser beam properties (CleanLaser, 2013) .................. 162
Table 7.4 TruDisk 5302 laser unit technical properties (Trumpf-Lasers, 2013) .......... 163
Table 8.1 Aluminium alloy AA6014 thermo-physical properties (AlShaer et al., 2015a)
....................................................................................................................................... 202
Table 8.2 Material properties and process parameters used in the SPH model ............ 212
16
NOMENCLATURE
Nomenclature used in the experimental work
Symbol
Definition
σ
Stress
L
Solubility of hydrogen in aluminium
P
Laser power
Pd
Penetration depth
Pw
Weld width
s
Gap size between sheets
tn
Sheet thickness
v
Traverse speed
Nomenclatures used in fluid dynamics analyses
Symbol
A
CFL
Definition
An arbitrary variable
CFL number
co
Numerical speed of sound
F
Lagrangian marker force density
fs
Surface tension force
g
External body force vector per unit mass
17
h
Smoothing length
L
Corrective tensor used to evaluate corrected kernel derivative
L2
L2 error norm
m
Mass
N
Total number of fluid particles
n
Unit vector normal to surface
Nb
Number of boundary particles
P
Pressure
r
Position vector
t
Time
u
A component of velocity vector u
u
Fluid velocity vector
V
Volume
W
Smoothing kernel function
t
Time integration step
tcv
Maximum allowable time step based on Courant time step control
tf
Maximum allowable time step based on force conditions
tst
Maximum allowable time step based on surface tension conditions
tv
Maximum allowable time step based on viscosity conditions
x
Particle spacing
SPH
SPH volume integral approximation of a variable or property
18

Boundary surface

Viscous term in SPH discrete form of Navier-Stokes momentum equation

Interpolation region

Free parameters in artificial viscosity model

Dirac delta function

Viscosity

Kinematic viscosity

Fluid density
19
LIST OF OPERATORS
Symbol
⊗
∙
∂a/∂b
Definition
Vector tensor product
Vector dot product
Partial derivative of a with respect to b
∇
Gradient
∇∙
Divergence
∇2
Laplacian
da/db
Mass derivative of a with respect to b
max
Maximum value of
min
Minimum value of

As a superscript denotes transpose of a matrix
20
LIST OF ABBREVIATIONS
CW
Continuous Wave
PW
Pulsed Wave
CFD
Computational Fluid Dynamics
CFL
Courant Fridrich Lewy
CPU
Central Processing Unit
DB
Dynamic Boundary
FDM
Finite Difference Method
FEM
Finite Element Model
FVM
Finite Volume Model
GMA
Gas Metal Arc
ISO
International Standards Organisation
LIBS
Laser Induced Breakdown Spectroscopy
MIG
Metal Inert Gas
MLS
Moving Least Square
SEM
Scanning Electron Microscope
SPH
Smoothed Particle Hydrodynamics
UTS
Ultimate Tensile Strength
WCSPH
Weakly Compressible Smoothed Particle Hydrodynamics
XRD
X-ray Diffreaction
YAG
Yttrium-Aluminium-Garnet
21
ABSTRACT
Porosity reduction and elimination in laser welding of AA6014 aluminium alloy for
automotive component manufacture and industrial applications
Ahmad W. AlShaer
Thesis submitted for the degree of Doctor of Philosophy
The University of Manchester
2016
Automotive and aerospace industries consume a significant amount of Al alloys in structures
and framing. There is, however, a significant challenge to join the alloy components by laser
welding. A key problem is the presence of large amount of porosity in the welds. This research
work aimed to understand factors affecting porosity formation in laser welding of AA6014 Al
alloy and identification and verification of a suitable method for the porosity reduction and
elimination.
AC-170PX (AA6014) Al alloy was welded, for the first time, using a 5 kW disk laser in two
different configurations: fillet edge and flange couch joints using a number of different filler
wires. The experimental results showed that laser power (2-5 kW) and welding speed (20-50
mm/s) had a significant influence on porosity generation. Also, the introduction of a 0.2 mm gap
between the sheets significantly reduced porosity for the fillet edge joint while it had a marginal
effect for the flange couch joint.
The effect of the chemical composition of the filler wire on the AA6014 weld quality was also
evaluated for the first time by using different filler wires (AA3004, AA4043, AA4047 and
AA5083) over a range of laser powers and welding speeds. The increase in Mg and Mn content
in the filler wire’s composition was found to reduce porosity in comparison with high silicon
content filler wires.
Nanosecond pulsed Nd:YAG laser cleaning was investigated as a surface preparation method
for laser welding for AA6061, and its effect on porosity at various welding parameters was
examined. The effect of laser cleaning on porosity reduction during laser welding using a filler
wire has not been reported before. The surface characteristics before and after laser cleaning
were analysed. The results showed that laser cleaning played an essential role in significantly
reducing porosity in both the fillet edge and flange couch joints at different levels of power and
laser welding speed. The developed surface preparation technology as a method for porosity
reduction in laser welding has been successfully implemented in one of the largest
UK/international car manufacturers.
To study the laser cleaning process, a novel Smoothed particle hydrodynamics (SPH) meshless
model has been implemented using a new 3-D multi-phase transient model. For the first time, a
study was conducted to validate the temperature field distribution predicted in SPH method
under nanosecond pulsed laser heating. The need for special surface treatment of the kernel
truncation was also investigated. The proposed model accurately predicted the laser ablation
depth and the crater shape and was validated using a significant number of experimental and
numerical data reported in the literature. Moreover, a primitive laser welding model has been
created to predict the material flow inside the welding pool.
The research work has resulted in four publications in peer-reviewed journals. The research
highlighted that future work should include the development of a more advanced SPH model for
the prediction of porosity in laser welding and to fully describe the relationship between laser
cleaning and porosity reduction in laser welding.
22
DECLARATION
No portion of the work referred to in the thesis has been submitted in support of an
application for another degree or qualification of this or any other university or another
institute of learning.
23
COPYRIGHT STATEMENT
The author of this thesis (including any appendices and/or schedules to this thesis) owns
the certain copyright or related rights in it (the “Copyright”) and s/he has given The
University of Manchester certain rights to use such Copyright, including for
administrative purposes.
Copies of this thesis, either in full or in extracts and whether in hard or electronic copy,
may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as
amended) and regulations issued under it or, where appropriate, in accordance with
licensing agreements which the University has from time to time. This page must form
part of any such copies made.
The ownership of certain Copyright, patents, designs, trademarks and other intellectual
property (the “Intellectual Property”) and any reproductions of copyright works in the
thesis, for example graphs and tables (“Reproductions”), which may be described in this
thesis, may not be owned by the author and may be owned by third parties. Such
Intellectual Property and Reproductions cannot and must not be made available for use
without the prior written permission of the owner(s) of the relevant Intellectual Property
and/or Reproductions.
Further information on the conditions under which disclosure, publication and
commercialisation of this thesis, the Copyright and any Intellectual Property University
IP Policy (see http://documents.manchester.ac.uk/display.aspx?DocID=24420), in any
relevant Thesis restriction declarations deposited in the University Library, The
University Library’s regulations (see
http://www.library.manchester.ac.uk/about/regulations/) and in The University’s policy
on Presentation of Theses
24
ACKNOWLEDGEMENTS
‫ال حمد هلل رب ال عال م ين‬
I have been very honoured to have two supervisors who are shiny stars in their fields. I
would like to express my endless gratitude to Prof. Lin Li for his continuous guidance
and support throughout my PhD and for sharing his expertise with me during all our
meetings. I would like also to deeply thank Dr Benedict Rogers for his endless
guidance and advice which helped me to bring this study into success. I am grateful to
him for sharing his experience with me, on the professional and personal levels.
Many thanks to Mr Anil Mistry, Mr Gary Watson and Mr Earl Russel for the supply
of the materials and for their support and the industrial discussions throughout this
project, and for the great time I had while working together as a team.
I would like also to sincerely acknowledge Dr John Francis and Dr Wei Guo for the
constructive discussions we had, which helped me to further understand the
experimental aspects of my work. I am also grateful to the SPH group, Prof. Peter
Stansby, Dr Steven Lind, Dr Athanasios Mokos, Dr Abuzeid Nasar and Dr
Georgios Fourtakas who shared their knowledge in SPH with me and helped me in
completing my simulation work.
Thanks to my friends and colleagues in the LPRC, Dr Ahmed Almasalmey, Fatima
Rajab (PhD) and Dr Abobaker Hamad for the great time we had and for helping me
on the lab equipment.
I would like to thank, with all my love and gratitude, my beloved mother, father and
siblings who supported and believed in me and were always there for me in the most
stressful moments. Also, I am truly thankful to my soulmate, my only love and wife
who has been my inspiration and motivation for expanding my knowledge and move
my career forward. I am grateful to her for her continuous support and encouragement
during all challenges in my study and life.
25
PUBLICATIONS

A.W. AlShaer, B. D. Rogers and L. Li (2017). “Smoothed Particle
Hydrodynamics (SPH) modelling of transient heat transfer in pulsed laser
ablation of Al and associated free-surface problems”. Computational Materials
Science, 127, pp. 161-179. http://dx.doi.org/10.1016/j.commatsci.2016.09.004

A.W. AlShaer, B. D. Rogers and L. Li (2016). “Smoothed Particle
Hydrodynamics Modelling of Pulsed Laser Ablation of Aluminium and
Associated Free-Surface Problems”. In Proc. 11th International SPHERIC
Workshop, June 13-16.

A.W. AlShaer, L. Li and A. Mistry (2015) “Understanding the Effect of Heat
Input and Sheet Gap on Porosity Formation in Fillet Edge and Flange Couch
Laser Welding of AC-170PX Aluminium Alloy for Automotive Component
Manufacture” ASME J. Manuf. Sci. Eng. 137(2), 021011.

A.W. AlShaer, L. Li and A. Mistry (2015) “Effect of filler wire properties on
porosity formation in laser welding of AC-170PX aluminium alloy for
lightweight automotive component manufacture”. Proceedings of the Institution
of Mechanical Engineers, Part B: Journal of Engineering Manufacture 1–13,
doi: 10.1177/0954405415578584.

A.W. AlShaer, L. Li and A. Mistry (2014) “The effects of short pulse laser
surface cleaning on porosity formation and reduction in laser welding of
aluminium alloy for automotive component manufacture”. Optics & Laser
Technology, 64, pp. 162-171.
26
CHAPTER 1
INTRODUCTION
1.1 Overview
Over the last 40 years, laser welding has become one of many technologies developed
to join metallic materials. Laser welding has attracted the interest of researchers and
industry through the wide range of applications that were enabled by the technology.
Such applications can be found in the aerospace, medical, automotive, defence,
electronics industries, etc. This success can be attributed to the specific properties of the
laser beam welding process including low energy input, producing small heat affected
zones, high welding speeds, high productivity and reliability (Chen, 2010).
With the growing production of new materials, more interest is shown in joining
materials with the highest quality and lowest cost (Khan et al., 2012). In addition to
that, production lines are believed to be more flexible and less demanding when
conventional welding processes are replaced with laser welding (Sun and Ion, 1995). A
broad spectrum of developments were made in laser welding, and the research has
covered many areas of interests: such as, the influence of different process parameters
on the quality of welds (Katayama et al., 2010; Sabbaghzadeh et al., 2008); welding
metals, plastics and even ceramics to metallic alloys (Haberstroh and Hoffmann, 2008),
and enhancing the effectiveness of conventional welding processes using hybrid
welding systems which utilise lasers (Liu and Zhao, 2008).
The automotive sector is looking for new lightweight materials and their fabrication
technologies to reduce the structure's weight and increase their strength simultaneously.
That need has coincided with environment protection measures and the need for energy
reduction (Chen, 2010). Weight reduction can be achieved by introducing Aluminium
(Al) alloys instead of using heavy steel parts to build the entire automobile bodies. Al
27
alloys are commonly welded using a filler material of the same or different composition.
The main reason behind the use of filler wires with different chemical compositions in
laser welding is the combination of the excellent mechanical properties of one material
with the thermal properties and good weldability of the other (Taban et al., 2010).
As indicated by some researchers, porosity and hot cracking are the most predominant
defects in laser welded seams in Al alloys. Various investigations have been carried out
to understand the formation of such defects further. For instance, the application of Xray radiation to the welding pool to show formation of pores during welding (Seto et al.,
2000), welding ice water using a CO2 laser to examine the fluid flow characteristics
(Berger et al., 2011), monitoring the plume during welding and the use of acoustic
signals to evaluate the weld quality (Hongping and Duley, 1996), etc. However, all
methods were limited to identifying porosity location and its movement rather than
exploring the methods for its elimination or reduction. Hence, this thesis presents an
experimental study supported by numerical simulations to reduce porosity in Al alloy
seams used in the assembly of the outer structural car body.
1.2 Research motivation
Focusing on automotive applications, various separate structural panels are used in
different parts of the car body, which limits the methods that can be utilised for joining
them with the adjacent parts.
Over the last 20 years, car companies have commonly used mechanical fasteners,
adhesive bonding, resistance and arc welding for joining structural Al alloys parts.
However, chemical processes such as adhesive bonding are inherently hazardous and
detrimental to the operators’ health due to the chemicals used in the process. Also, this
method embraces an additional cost due to the need for specialised storage areas and
special procedures for their handling and disposal. This problem continues when
automation takes place due to the need for efficient ventilation of the areas where those
chemicals are applied. Besides the additional cost, mechanical fastening and welding
remain superior when considering joint strength to adhesion (Hayat, 2011).
Mechanical fastening, such as bolts, pins and rivets, produces an acceptable strength for
metallic and non-metallic joints but they cannot be used for all structural panels in all
28
locations, especially for the outer panels which are meant to hide all structural frames
and hardware inside the car body.
Although resistance welding has still been applied in the automotive sector for joining
structural parts, it has a number of disadvantages that pushed companies to look for
alternatives. These drawbacks can be: the electrodes wear over time, the significant
dependence of the weld’s dimensions on the electrodes condition, the need to access
both sides of the joints, high electrical energy consumption and the process limitation to
small thicknesses for certain types of electrodes (Rajput, 2007). More importantly,
resistance welding produces stitched seams rather than continuous seams, which may
not be acceptable for sealing purposes.
Laser welding has been employed for joining steels, alloy steels and dissimilar materials
for automotive applications. However, joining lightweight alloys such as Al and Mg
alloys, without any defects, is a challenge. These defects can be a serious issue for
panels that form the external shape of the car body and in which dents and surface pores
are not allowed. Additionally, pores and cracks adversely affect the tensile strength and
corrosion resistance of the welds that are in contact with various corrosive environments
during their service. Car body panels, which are of the author’s interest, have a thin
thickness of 1.1 mm that requires a focused heat source during welding to minimise the
heat losses and to avoid burn-through defects and minimise microstructural changes of
the base material.
It is essential in a production line that produces a car every circa 80 seconds to
guarantee the quality of the welded parts to eliminate any delays or stoppages in the
production line. Defective parts not only cause delays in the process due to the need for
human intervention but also raise the need for repair which embraces additional cost to
the process. The welding process, therefore, should be completed as quickly as possible
with the minimum number of scraps or the need for repair parts. Hence, the welding
process parameters such as the laser power and traverse speed should be modulated to
weld the parts with the highest quality at the maximum speed possible.
The final assembly strength significantly depends on the weld strength which is
determined by the process parameters and the base and filler materials’ chemical
compositions and microstructures. Hence, it is important to investigate the most suitable
29
filler wire (or set of wires) which produce(s) the highest mechanical properties
accompanied with the acceptable external appearance.
Over the years, surface preparation has been carried out using chemical etchant
(Mathers, 2002), sandblasting and mechanical scraping (Vargel, 2004), for reducing
porosity that originates from the hydro-oxide layers on the top of metallic alloys. In
addition to the environmental and operational disadvantages of chemical cleaning, this
process may have detrimental effects on the base material depending on its chemical
composition and surface treatment. It has been reported in (Joshi et al., 2011) that
alkaline cleaning of AA7075-T6 may remove the outer Mg-rich layer while the Al-rich
oxide layer remains intact. Also, acids can actively react with Al and can damage the
oxide film, producing Al hydroxides and atomic hydrogen which will lead to porosity
after welding. Mechanical scraping and sandblasting create uneven cleaning and
severely damage the welding surface, and may subsequently cause large rejection rates
of the final products, especially when precision is required during production
(Haboudou et al., 2004). Moreover, with many industries increasing the use of
automation, this emphasises the need for new methods which eliminate the human
factor in the process and can be fully integrated into the modern production lines.
Therefore, dry laser cleaning is necessary to be investigated for its feasibility to
eliminate all other methods’ disadvantages and offer a better and a faster solution to
porosity in laser welding. The application of laser cleaning to the Al alloys as a surface
preparation was a new topic to investigate at the beginning of this PhD project.
1.3 Aim and objectives
The aim of this project was to study the feasibility and the output of three different
experimental and numerical methods for eliminating/reducing porosity in laser welded
AC-170PX Al alloy. This PhD work is industrially oriented, and the method developed
for porosity reduction in laser welding of Al alloys has been successfully implemented
and embedded in one of the largest UK/International car manufacturers.
The objectives of this work can be summarised as follows:
30

To investigate the effect of the laser welding process parameters on the porosity
formation and the mechanical properties of the welds. Additionally, the
influence of a 0.2 mm gap on porosity reduction was investigated for two
different welding configurations: fillet edge and flange couch joints.

To study the effect of changing the filler wire chemical composition on porosity
content, mechanical properties and the microstructure of the weld seam. During
this investigation, the weld appearance was evaluated to the industrial standards.

To investigate the role of laser cleaning before laser welding on the welding
quality over a range of a given process parameters.

To further understand the process characteristics of laser cleaning (ablation) of
Al alloys by developing a multi-phase transient model using a novel smoothed
particle hydrodynamics (SPH) numerical model for nanosecond laser pulses.
This model was the first of its kind and has been validated using analytical and
experimental data.
 To formulate a new SPH model of the molten material flow during laser welding
using SPH (included in Appendix A).
1.4 Thesis outline
The thesis is structured according to the “Alternative” format instead of the “Standard”
format. Thus, the key chapters are given in the form of “published” journal articles.
Between every two consecutive chapters published as papers, a brief introduction is
given to emphasise the relation between these chapters followed by a page showing the
status of the article. It is worth mentioning that the published manuscripts are
reformatted according to the thesis presentation policy set by The University of
Manchester. Also, the references used in all chapters are compiled and grouped in the
“References” section at the end of this thesis.
The thesis is structured as follows:
Chapter 2 reviews the reported literature on laser welding and cleaning of Al alloys and
porosity formation mechanism and the methods used to create the pores’ trajectories
within the welding pool. Additionally, the methods employed for porosity reduction
during laser welding of different materials are reviewed in some detail.
31
Chapter 3 navigates through the reported literature on mesh-based, meshless and
specifically SPH modelling of laser materials processing, highlighting the novelty of the
multi-phase 3-D SPH model created in this project on laser ablation/cleaning.
Chapter 4 gives a brief description of the experimental set-up and some of the
instruments used for the welding characterisation.
Chapter 5 investigates the significance of the heat input and a 0.2 mm gap between the
welded sheets on porosity formation during disk laser welding of AC-170PX Al alloy
for automotive applications. The welds were characterised after welding with and
without the 0.2 mm gap between the sheets for two different welding configurations:
fillet edge and flange couch joints. (Alshaer et al., 2015b)
Chapter 6 studies the effect of the filler wire’s chemical composition on porosity
generation in the welds during laser welding of AC-170PX alloy. In addition to the
mechanical and the appearance characteristics of the welds, the effect of the wire’s
chemical composition on the microstructure was also analysed since different heat
treatable and non-heat treatable wires were used. (Alshaer et al., 2015a)
Chapter 7 investigates the effects of nanosecond-pulsed laser surface cleaning on
porosity formation and reduction in laser welding of AC-170PX alloy for the
automotive component manufacturer. The effect of laser cleaning on the sample surface
regarding melting, surface roughness and reflectivity were evaluated. Additionally, the
weld dimensions and the porosity content were also examined. (Alshaer et al., 2014)
Chapter 8 presents the development of smoothed particle hydrodynamics (SPH)
modelling of transient heat transfer in pulsed laser ablation of Al alloys and the
associated free-surface problems. The open-source code SPHysics_3D was used to
develop the SPH model for the laser processing phenomena. This chapter demonstrates
the lagrangian characteristics of the SPH method and its ability to model nanosecond
laser processing. The model predicts not only the ablation crater's dimensions and shape
but also the thermo-physical characteristics of the ejected material (Alshaer et al., 2017)
Chapter 9 states the conclusions of this work and the recommendations for future work.
32
2 CHAPTER 2
A LITERATURE REVIEW PART I: LASER
WELDING TECHNOLOGY
2.1 Introduction
Welding is a process in which two materials are joined permanently together through
localised cohesion resulting from the correct selection of pressure, temperature, and
metallurgical conditions (Khan, 2007). Welding was known for a very long time in
joining copper-gold in the jewellery industry. By the time electricity was readily
provided in the 19th century, welding had started to develop rapidly and was used to join
metals. The welding term can be applied to autogenous metals joining, brazing and
soldering. In fact, welding embraces a broad spectrum of processes which can be
classified into four main categories: fusion welding, soldering/brazing, and solid state
welding. Fusion welding includes melting of the base materials while soldering/brazing
includes melting of a filler material which in turn joins the base materials together.
However, solid state welding combines the base materials without melting them but by
softening using heating.
Figure 2.1 and Table 2.1 show a master chart of welding and adhered processes and a
comparison of the different methods respectively. Each welding/joining process has
advantages and disadvantages which drive its use in certain applications. For instance:
adhesive bonding is used where the product is operating at low temperatures (less than
100oC); resistance welding is widely applied in automotive industries; soldering/brazing
is used in electronics and jewellery industry, and fusion welding is employed to join
33
metals of relatively high melting points. The laser beam and electron beam welding
techniques have unique properties which allow them to be used in either macro-scale
welding to join small and large parts, or in micro-scale welding that is used for MEMS
or medical applications (Chen, 2010; David and DebRoy, 1992; Gale and Totemeier,
2004).
In this chapter, the principle behind the laser welding process will be briefly introduced,
and later different types of Al alloys, as well as porosity generation mechanisms, will be
discussed in detail.
Figure 2.1 Master Chart of various welding processes (Khan, 2007)
34
Table 2.1 Comparison of various joining processes (Gillner et al., 2000) (reformatted)
Technology
Adhesive bonding
Temperature
Range
o
< 100 C
Advantages/
Disadvantages
Application
+ low cost
-selective process
 Low temperature
sealing
 Optoelectronics
 Electronic
interconnection
 Chip
interconnection
Soldering
< 200 C
+ low cost
+ large quantity
-long-time instability
Brazing
< 600 oC
+ fast
-Restricted material choice
-selective
Resistance welding
< 1500 oC
+fast
+all materials
-selective
 Connector package
Electron beam
welding
> 1500 oC
+fast
+all materials
-vacuum environment
 Joining different
material
 Medical device
Laser welding
> 1500 oC
+fast
+all materials
 Packaging
 Join different
material
o
2.2 Fusion Welding
The fundamental principle of fusion welding involves concentrating the heat source (gas
flame, electric arc or a high energy laser beam) in small areas resulting in melting and
mixing a part of the base material to produce the weld. Fusion welding includes heating
and cooling cycles that may induce undesirable phase transformations to take place
inside the fusion zone depending on the materials properties, physical dimensions and
the process parameters including the heat input and the number of passes (Messler,
1999; Black et al., 2008; Kou, 2002). Irrespective of the heat source being used,
different joint types can be welded using fusion welding as shown in Figure 2.2.
35
Figure 2.2 Common types of welding joints (Minnick, 2008)
It is pertinent to mention that fusion welding can be carried out with or without filler
materials, depending on the type of joint configuration, material composition and the
number of passes. As the power density of the heat source increases, the heat input,
which is the ratio of the laser density to the traverse speed, contributing to welding
increases. Fusion welding conventionally consists of two stages: substrate heating
(melting) and subsequent cooling (solidification). The cooling rates, which depend on
the workpiece’s material and dimensions, can range from less than one hundred to a
number of millions of degrees per second dividing the welding zones three different
regions: the fusion zone, the heat-affected zone (HAZ), the parent material which has
not been affected during the process and finally the so-called “partially melted zone
(PMZ)” that can be recognized during welding of alloys using filler wire of a different
chemical composition (David and DebRoy, 1992; Kou, 2002). Therefore, mechanical
properties and even the chemical composition of the weld may vary from that of the
base materials according to the added filler (Kwok et al., 2006). Figure 2.3 shows a
schematic of the different parts of the welding zone.
36
Figure 2.3 Schematic of the welding area (Black et al., 2008)
It is worth mentioning that fusion welding can be classified into three categories
depending on the use of a filler material and the chemical composition of that filler.
These three types are briefly introduced as follows (Messler, 1999):

Autogenous welding: Fusion welding is conducted without the use of filler
material.

Homogeneous welding: Both base and filler materials have identical chemical
composition.

Heterogeneous welding: substrate and the filler material have distinct chemical
composition
2.3 Laser Welding Mechanism
In laser welding, the increase in the substrate’s temperature is achieved by transferring
the photons’ energy to the electrons cloud that occupies the surface of the metal
(Dahotre and Harimkar, 2008). The laser beam can show two different behaviours
during laser welding depending on the laser-material interaction characteristics:
“conduction mode” and “keyhole mode” (see Figure 2.4). The difference between the
two modes can be based on the aspect ratio (ratio of the weld depth to the width) that
describes the penetration depth or on the power density (Klages et al., 2003).
37
(a)
(b)
Figure 2.4 (a) Conduction mode (Dahotre and Harimkar, 2008) (b) Keyhole mode
(Ion, 2005a)
2.3.1
Conduction Mode
In conduction mode, the laser power density does not typically exceed 103 W/mm2, and
this power is used to melt the material without any vaporisation (Ion, 2005a). Using the
aspect ratio concept, welding is considered to be carried out under conduction mode if
the weld aspect ratio is around 0.5 or less (Martukanitz, 2005). There are two types of
conduction welding: direct heating in which the welded material absorbs the beam
energy and converts it into heat (as typically occurs in metals and alloys), and the
energy transmission in which the base material transmits the laser beam without
absorbing it (as typically occurs in transparent polymers). Therefore in energy
transmission, welding is often conducted using a novel interfacial ink that is applied at
the joint interface to enhance the beam absorption (Ion, 2005a).
Conduction welding utilises a defocused laser beam with a low power density which
results in a semi-spherical weld pool shape that remains intact. Heat is conducted from
the pool surface to the rest of the bulk material (Miyamoto et al., 2004; Dahotre and
Harimkar, 2008) and achieves a shallow penetration, welding with high energy waste
and low efficiency. The intact and steady weld pool and the shallow penetration make
conduction mode welding suitable for joining thin sheets and materials of low boiling
temperatures and it can be used in butt or lap joints which are applied in different
industrial sectors including automotive, electronics and medicine (Martukanitz, 2005).
Although this type of welding produces a smooth weld surface and a homogeneous
cross section, a significant amount of the energy is wasted by conduction to the bulk
38
material without contributing to the welding process, which may cause undesirable
metallurgical changes in the adjacent regions to the weld. This welding mode in effect
will reduce the process efficiency and produces weld beads with limited penetration
(Quintino et al., 2007).
2.3.2
Keyhole Mode
Keyhole mode is differentiated from conduction mode by the aspect ratio of the weld
depth to the weld width, which can be greater than 0.5, while the power density can
range from 103 to 105 W/mm2 (Martukanitz, 2005; Baba and Watanabe, 2005). Parts of
the material, which receive such high power density, melt and evaporate instantly
creating a hole within the weld pool. The laser beam then enters the keyhole and gets
reflected at the side walls of the keyhole and increases the efficiency of the process due
to sequential Fresnel effects (absorption from the beam reflection on a surface). As soon
as the keyhole is established in the weld pool, the absorptivity of the material increases
sharply from ~3% to ~98% and therefore care should be taken to prevent the structure
failure (Ion, 2005a; Steen and Mazumder, 2010).
The majority of existing applications use keyhole welding due to its beneficial
properties such as high power density, low heat input into the workpiece, deep
penetration, accuracy and high traverse speed (Cho et al., 2004). The weld in keyhole
mode usually consists of two regions: one with a semi-sphere shape at the top of the
weld and a bottom region that has near parallel sides (see Figure 2.5) (Steen et al.,
1988). In real life application, the shape of the weld may differ according to the
materials properties, the joint configuration and the process parameters.
39
Figure 2.5 Shape of the weld with a keyhole at different welding speeds (Dahotre and
Harimkar, 2008)
When the beam is moving, the welding speed becomes dependent on the stability of the
keyhole, and the geometry of the keyhole becomes a function of the welding speed.
Figure 2.6 demonstrates the weld pool and the keyhole shape. At low speeds, the
keyhole takes a rotationally symmetric shape, while at high speeds, a remarkable
difference can be seen between the front and the rear walls (Kaplan, 1994; Dahotre and
Harimkar, 2008). It should be emphasised that these predictions are based on
mathematical models only, and not on experiments.
40
Figure 2.6 Keyhole in deep penetration welding (Schubert et al., 2001)
In keyhole welding, the material’s temperature rises above the boiling point and plasma
that consists of electrons, ionised metal vapour and gases accumulates at the weld top
surface. Plasma temperature can reach 10,000 to 30,000oC in some cases with a very
high heat input (Steen and Mazumder, 2010). Figure 2.7 illustrates the process of
plasma formation as the power density increases.
Heats
Heats
103 Wmm-2
104 Wmm-2
Heats
Plasma
105 Wmm-2
formation
Figure 2.7 Plasma formation in laser welding (reformatted for the sake of clarity)
(Steen and Mazumder, 2010)
In fact, the amount of plasma that is generated from the ionised gas and metal may
increase in continuous wave (CW) laser welding in comparison with pulsed laser
welding (Chen, 2010). Although the high temperature of the plasma helps to form the
keyhole, plasma absorbs a portion of the incident beam and elevates the focal point
changing the original spot size and the pre-selected power density. This elevation in the
focus negatively affects the process and causes a disturbance in the keyhole opening
41
which in turn reduces the efficiency of the welding process (Chen, 2010; Dahotre and
Harimkar, 2008). The following equation calculates the fraction of the beam absorbed
by the plasma, and it is known as Beer–Lambert law (Dahotre and Harimkar, 2008):
a pl  1  exp(aiB hpl )
( 2-1 )
where αpl is the portion of the beam absorbed by plasma, αiB is plasma absorption
coefficient and hpl is the height of the plasma plume over the workpiece.
Table 2.2 Ionisation potential and other properties of common shielding gases (Ion,
2005a)
To reduce plasma effects, it is recommended to use shield gases which have high
ionisation potential such as He and Ar. Shielding gases not only contribute to plasma
suppression during welding but also enhance the heat transfer to the welding pool when
gases or a mixture with high thermal conductivity are used. It is evident from Table 2.1
that He has the greatest potential among all gases and followed by Ar; therefore, they
are the most common shielding gases used in industry. Although helium has better
shielding properties, i.e. ionisation potential, Ar is cheaper and often used for most
industrial applications (Weman, 2003). In fact, the type of gas is strongly dependent on
the material being welded. Non-ferrous metals, which are widely used in aerospace and
automotive industries, show a high tendency to react with surrounding gases when they
are at elevated temperatures (Gouret et al., 2004). Greses et al. (2001) showed that the
plume’s temperature produced in Nd:YAG laser welding is about 2000oC, whereas the
temperature in CO2 laser welding is about 6,000–10,000oC. This behaviour was
explained by Hussein et al. (2013) who concluded that the inverse bremsstrahlung
absorption of the laser radiation by the plasma becomes dominant for longer
wavelengths (1 m and 10 m) than for shorter wavelengths (532 nm and 266 nm). This
42
implies that plasma absorption for the 10 m radiation is more significant than the
absorption of the 1 m wavelength; therefore, a cheaper shielding gas can be used in
Nd:YAG welding in comparison with CO2 welding (Steen and Mazumder, 2010).
Another research conducted by Tani et al. (2007) showed that the gas flow rate was
probably not the principle factor affecting the weld quality when high-density shielding
gases are used. Moreover, the weld penetration can be improved using a mixture of 90%
argon and 10% oxygen instead of pure argon (Jorgensen, 1980). If the plasma formed
above the workpiece surface is thin and close to the surface, Ar is recommended to be
used. Conversely, if plasma is becoming thicker and is leaving the surface, its blocking
effects become considerable, and helium should be supplied with a side-blown jet
despite the high cost (Steen and Mazumder, 2010).
In the following section, an introduction is given on the Al alloys used in this project in
addition to the effect of the alloying elements on the mechanical properties of the alloy
and the weld bead.
2.4 Aluminium Alloys
Nowadays, a full spectrum of industrial applications is increasingly using Al and its
alloys due to their beneficial properties such as high strength-to-weight ratio, good
weldability and formability, and excellent corrosion resistance (Ion, 2005a). It is
essential to know the main properties of Al before introducing the different alloys
(Mathers, 2002):

Aluminium oxide exceeds Al melting point (~660oC) by approximately 1400oC
reaching up to 2060oC. This high melting temperature has a remarkable
importance in Al welding since the oxide film may hinder the welding process,
and therefore it is intrinsic to reduce/eliminate this film before and during
welding to produce good quality welds.

This oxide film is self-healing and has excellent corrosion resistance which
makes Al alloys very suitable to be used in exposed environments without any
extra protection.

Aluminium has a thermal conductivity coefficient which is six times higher than
that of steel and has a specific heat twice as that of steel. These characteristics
43
make Al difficult to be welded since a significant amount of heat is carried away
from welding zone, and hence higher power density is required to achieve the
desired penetration.

The colour of Al does not change as temperature increases. In the case of
manual arc welding, which can make it difficult for the welder to determine
when melting is going to start. However, since laser welding is an automated
process, this stability in colour is not a serious concern.

Unlike steel which suffers from crystal transformation at certain temperatures,
pure Al is not prone to such transformation during heating or cooling. This
generally means that the rapid cooling has a negligible or no effect on Al
hardening. Despite that, pure Al can be hardened by a process called
“precipitation hardening”.
Figure 2.8 gives an idea of the global consumption of Al in 2010 and 2014 in which it
can be seen that the transport sector consumes the highest portion of Al in comparison
with other sectors.
(a)
(b)
Figure 2.8 The global consumption of Al (a) in 2010 (Melik and Kouzmenkov, 2010)
(b) in 2014 (Klein et al., 1994)
Most of the commercial Al alloys are based on ternary or quaternary systems in which
alloying elements participate in enhancing the mechanical properties of the alloys by
precipitation (Davis, 1993). Al alloys are applied in many sectors such as rocket covers,
liquid nitrogen containers (at -200oC) and automotive industry (up to 150oC) (Ashby et
al., 2010). Al alloys can be classified into two categories, wrought and cast alloys.
44
Wrought Al alloys differ from each other by the alloying elements added to them. Table
2.3 shows the designation of various alloys according to the European system and their
typical applications (Mathers, 2002).
It is pertinent here to mention the effect of every added element on mechanical
properties of the Al alloy (Mathers, 2002; Ion, 2005a):

Copper (Cu): improves strength and precipitation hardening; reduces
weldability, corrosion resistance and ductility.

Manganese (Mn): improves ductility and strength through the formation of a
solid precipitate.

Silicon (Si): increases strength, ductility and fluidity of the alloy; it produces
with magnesium a precipitate that results in precipitation hardening.

Magnesium (Mg): promotes strength and work hardening characteristics.

Zinc (Zn): ultimately increases strength; allows precipitation hardening; can
result in stress corrosion.

Iron (Fe): enhances strength; can reduce solidification cracking but it is usually
a residual element.
45
Table 2.3 Standard forms and applications of wrought Al alloys (Mathers, 2002)
The 1000, 3000, 4000, and 5000 series of Al alloys are not heat-treatable while the rest
are heat-treatable. The mechanical properties of Al alloys can be controlled using
different procedures such as strain hardening, annealing, tempering and age hardening
(Mathers, 2002). The European Committee for Standardisation (CEN) has provided a
different designation system to distinguish the different heat treatment procedures. For
example:

F- as fabricated. To indicate that there is no control over the strength and the
mechanical properties of the alloys as this is a function of the manufacturing
process.

O- annealed. This is used to indicate the lowest strength for specific alloys.

H- strain hardened. This is used to refer to the cold work which has been exerted
on the alloy to achieve the required mechanical properties.

T- tempered. This is to identify the degree of ageing and heat treatment that have
been applied to the alloy.
46
It is pertinent to mention that a number of digits usually follow the previous letters to
identify the exact processes followed to achieve the exact desirable properties (Mathers,
2002).
The following subsections will introduce the characteristics of some Al alloys that are
most used in vehicles since they are the primary interest in this project. Al 5000 series
alloys, especially 5754-O alloys, have high formability and are widely employed in the
production of car body panels (Bolt et al., 2001; Story et al., 1993; Jain et al., 1998;
Zhang et al., 2015). In comparison with other types of Al alloys, 2000 and 6000 series
alloys are commonly applied in car panels manufacture due to their excellent properties
such as high corrosion resistance, superior surface finish and good formability (Kamat
et al., 2002; Wang and Kassner, 2002; Murtha, 2000). Moreover, Al alloys selected to
form the exterior of the car body should be age hardened during the paint bake process
to gain the required dent resistance (Cole and Sherman, 1995).
2.4.1
Aluminium-Mn alloys (3000 series)
This type of Al alloys contains Mn as a major alloying element beside other
complementary elements such as Fe and Mg. It is a common practice to add other
elements such as Cu and Mn to improve weldability and strength of these alloys. Mn
solubility in 3000 significantly depends on the Fe percentage in the alloy; therefore,
manufacturers tend to reduce Fe content to its minimum (Mathers, 2002).
2.4.2
Aluminium-Si alloys (4000 series)
Al-Si alloys usually form a eutectic at 11.7% Si without forming intermetallic
compounds. Molten 4000 series have a high fluidity and are often alloyed with Mg to
enhance precipitation hardening to make it ideal for casting (Mathers, 2002). Because
Al-Si alloys have low viscosity, non-heat treatable, and less susceptible to cracking
above 5% Si, they became an excellent choice as a filler wire for joining 6000 series
with minimum cracks especially when welding thin sheets such as car body panels.
Also, Si content lowers the melting point for the alloys and promotes it to solidify over
a small temperature range preventing potential cracks nucleation (Ion, 2005a).
47
2.4.3
Aluminium-Mg alloys (5000 series)
The maximum percentage of Mg in 5000 series alloys is limited to 5%. Mg forms a
solid solution with Al which is responsible for the increase in strength. The more Mg
solutes in Al, the more strength is gained. It is important to note that the maximum Mg
amount that can dissolve in Al at room temperature is only 1.4%, meaning that Mg will
segregate if Mg-rich Al alloys are heated and cooled slowly (Mathers, 2002). Either or
both Mn (0.1-1 %) and Cr (0.1-0.25%) can be added to increase the strength further
(Ion, 2005a). Al-Mg alloys which contain Mg between 1% and 2.5% may be more
prone to hot cracking if welded autogenously or with a filler wire of the same
composition. To overcome this problem, a high alloyed filler material that contains
more than 3.5% Mg should be used during welding (Mathers, 2002).
2.4.4
Aluminium-Mg-Si alloys (6000 series)
Because 6000 series alloys are heat treatable alloys, they are well known for their ability
to be formed in a solution-treated state and then age hardened during the paint bake
stage (Engler and Hirsch, 2002; Birol, 2005; Song et al., 2007). These alloys contain
about 1% of each of the main elements (Si and Mg). However, they can include other
elements such as Cr, Zn, Cu and Mn of equal percentages. The principle hardening in
theses alloys is gained by the presence of a metastable magnesium silicide (Mg2Si)
known as β`` phase. In fact, 6000 alloys are very susceptible to hot cracking especially
when a great proportion of Al exists in welding zones. To overcome this, filler wire of
AA4043, which contains high silicon content, can be used to reduce cracks, unlike
AA5356 filler materials. Accordingly, it is rare to weld 6000 series alloys autogenously
without a filler wire (Mathers, 2002; Ion, 2005a). Table 2.4 gives the chemical
composition of 6000 series alloys that are most used as Al sheets in the car industry.
48
Table 2.4 Chemical composition of three Al alloys used in automotive industry, (Note:
single-numbers refer to the maximum limits) (Kamat et al., 2002)
Alloy
Si
Fe
Cu
Mn
Mg
Cr
Zn
Ti
6022
0.8-1.5
0.05-0.2
0.01-0.11
0.02-0.1
0.45-0.7
0.1
0.25
0.15
6111
0.6-1.1
0.4
0.5-0.9
0.1-0.45
0.5-1.0
0.1
0.15
0.1
6016
1.0-1. 5
0.5
0.2
0.2
0.25-0.6
0.1
0.2
0.15
2.5 Porosity in welded Al alloys
A significant number of studies have been reported in the literature on porosity
formation in Al alloys since these alloys have a broad range of critical applications such
as aeroplanes’ and car bodies’ structures.
Alfieri et al. (2011) discussed porosity content in AA2024 bead-on-plate (BOP) welds
during disk laser welding. They concluded that the use of high line energy for this type
of joints and materials led to a significant reduction in porosity because it created a
more stable keyhole and achieved full penetration in butt and BOP welds.
Two AA6013 sheets were laser welded using a fibre laser of 1.0 kW power by Vilar et
al. (2008) and the optimum welding parameters were stated as 1.0 kW and 5 m/min.
Although no cracks were observed in the produced welds, pores and holes were present
in the longitudinal sections of the welds and the proposed methodology was not able to
eliminate all defects. Reinhold et al. (2002) welded the same alloy using a YAG-laser
with different filler powders such as AlSi12, AlSi12Mg5, AlSi10Mg, AlMgSi and
AlMg4,5Mn. They measured porosity levels in these seams and related the increase in
porosity to the rise in Mg content in the filler powder. According to the authors, Mg
makes it easier for H2 to solute in Al, increasing the pores percentage inside the weld.
However, porosity was not eliminated, and the authors did not discuss the effect of
feeding separate powder particles on porosity compared with feeding a continuous
medium such as a filler wire. Another investigation (Pakdil et al., 2011) on 6000 series
focused on fatigue properties of AA6056 welds when it was welded using a CW CO2
laser with AlSi12 filler. The authors concluded that pores not only hindered crack
49
propagation but also reduced the propagation rate in relation with the stress ratio R.
Although pores acted to some extent in favour of increasing fatigue resistance, the
authors were not able to eliminate porosity and produce defect free-welds.
Researchers have also drawn a remarkable attention to 5000 Al alloys due to their high
strength and corrosion resistance which makes them ideal to be used for aeroplanes’
structures. However, hydrogen porosity is still the main issue faced during welding
AA5182 and AA5754 alloys. In addition to hydrogen pores, porosity in these alloys is
also generated due to the instability of the keyhole. Pastor et al. (1999) concluded that
focussing the laser beam at a point above the surface created a more stable keyhole, but
the problem of hydrogen porosity insisted.
Yu et al. (2010) showed that the use of filler wire during laser welding of 5A06 Al alloy
increases pores generation in the weld because the molten wire droplets can disturb the
stability of the keyhole and causes its fluctuation. To avoid this, they introduced a prefabricated gap between the sheets in butt welding and found that porosity can
significantly be reduced because the pre-existing gap joins the keyhole and makes it
larger and steadier (Yu et al., 2010). It is pertinent to mention that porosity was also
observed in YAG laser welding of Al-Li alloys such as 5A90 in which metallurgical and
unstable keyhole porosities were detected. Post-treatment of the weld such as surface
remelting or hot isostatic pressing can be used as methods for eliminating porosity and
shrinkage in the welds (Chen and Gong, 2011).
Hybrid laser welding was investigated as a method for porosity reduction in Al welding
after porosity problem had been observed in autogenous laser welding (Liu et al., 2006;
Liu et al., 2005a; Liu et al., 2005b; Katayama et al., 2007b). The decrease in porosity
percentage in hybrid laser welding is attributed to the lower cooling rate compared to
laser welding due to the extra heat generated by the arc (Naito et al., 2002; Reutzel et
al., 2006; Ribic et al., 2009). Autogenous and hybrid laser/MIG welding of a 7000 Al
alloy was also investigated and full penetration in a 12.7 mm thick plate was achieved in
both autogenous and hybrid laser/MIG welding. Porosity levels did not exceed 0.3% in
both types of welds (Allen et al., 2006).
AA7075-T6 alloys have been hybrid laser welded, but a significant loss in Zn content
has been observed. Wu et al. (2013) showed that this reduction in Zn caused a
50
significant segregation of Cu in the heat affected zone and prevented the uniform
distribution of alloying elements in the welds. Hybrid laser/MIG welding of AA5052
alloys using AA5356 filler wire was found to suppress porosity in the welds, and that
was confirmed by X-ray transmission observation (Katayama et al., 2006). Ascari et al.
(2012) studied hybrid GMA/laser welding of AA6082 Al alloy and used analysis of
variance (ANOVA) software to analyse the results. They concluded that porosity was
significantly reduced using weak current in short-arc transfer only. However, the effect
of the current was significant on porosity size in both short-arc and pulsed-arc transfer
modes.
Laser welding was not only evaluated under atmospheric pressure but also in a vacuum
and under microgravity. This evaluation was conducted due to the need for space crafts’
repair in orbits, where the gravity force is negligible. Microgravity- and vacuumwelding of Al alloys, titanium, medium carbon steel and stainless steel were
investigated by Katayama et al. (2000). Welding was performed in a chamber in which
microgravity was applied on two types of welds: flat position and overhead welds.
Additionally, vacuum welding was also investigated to observe its influence on welds
characteristics such as penetration, weld geometry and porosity. Their results clarified
that porosity percentages values were approximate under all circumstances and that a
vacuum did not have any effect on porosity. Vacuum and microgravity did not have any
exceptional influence on porosity reduction because voids were generated due to
recurrent keyhole collapse, which in turn caused gases entrapment in the weld. The
same trend was also spotted in welding of AA1050, AA5083 Al alloys, pure titanium,
carbon steel and SUS 403 (Katayama et al., 2000).
2.6 Porosity formation mechanism
To solve the problem of porosity in Al alloys, it is essential to understand its
characteristics, generation mechanism and the factors influencing it.
2.6.1
Pores shape and size
There are two main types of pores that are generated inside Al welds: gas porosity
which is also known as hydrogen porosity (or metallurgic porosity) and shrinkage
porosity. Gas pores are produced due to the reduction in hydrogen’s and other gases’
51
solubility in the solid phase after solidification from the liquid. However, shrinkage
porosity occurs when the metal is transformed from a liquid state to solid increasing its
density and leaving an empty volume at the boundaries. For Al alloys, this shrinkage
ratio takes a significant value from 3.5% to 8.5% depending on the alloying elements,
solidification temperature range and the volume fraction of the solid and liquid phases.
This change in volume results in shrinkage pores which may develop into cracks
especially for alloys with high susceptibility to cracks (ASM International, 1993).
According to another classification, pores can be divided into two categories by size:
fine porosity with an average diameter of less than 0.5mm, and coarse porosity with a
diameter greater than 0.5mm (Ion, 2005b).
Fine porosity tends to appear regularly distributed in a line of spherical voids. They
usually exist in lines parallel to the fusion line and are formed mainly from dissolved
gases such as hydrogen, oxygen, and nitrogen which have lower solubility in solid than
in liquid metals. The primary source of these gases and porosity can be the base
material, filler material, shielding gases and surface contamination (Ion, 2005b). Figure
2.9 shows the shape of the inner surface of a small pore in a titanium alloy.
Figure 2.9 Scanning electron micrograph of a small pore in a titanium alloy (Zhou and
Tsai, 2007a)
Coarse porosity appears irregularly distributed throughout the entire weld bead (Ion,
2005b). However, they appear commonly in larger number in the bottom section of the
bead than in the upper part. According to some researchers, this type of pore may form
due to keyhole collapse which is induced by the long interaction time between the laser
beam and the molten materials (Katayama et al., 2003). Different types of porosity
distribution inside the weld are depicted schematically in Figure 2.10 followed by two
original pore shapes shown in Figure 2.11 (Yao and Gong, 2011).
52
Figure 2.10 Different types of porosity distribution (Gupta, 2015)
Figure 2.11 SEM picture of various pores shapes (a) spherical pores due to gases
entrapment (also called metallurgical pores) (b) irregular pores due to the keyhole
collapse (Yao and Gong, 2011)
According to Ion (2005b) and Zhou and Tsai (2007b), coarse porosity is formed due to
different factors: high metal vapour pressure, disturbance at the back edge of the
keyhole, fluid dynamics in the surrounding molten metal, plasma formation at the top of
the keyhole, gases entrapment, small temporal and spatial fluctuations in beam intensity
and mode and solidification process. Research conducted by Zhou and Tsai indicated
that porosity exists in higher percentages in keyhole mode welding than in conduction
mode welding when the weld width is kept constant. Besides, it has been found that
welding speed had a significant influence in affecting porosity formation in laser
welding.
2.6.2
Effect of process parameters on porosity formation
Katayama et al. (2003) explained the mechanism of porosity generation at different
welding speeds during autogenous diode and YAG welding of stainless steel SUS304.
At lower travel speeds (less than 5 m/min), bubbles are created due to the collapse of
53
the front wall of the keyhole, and they can be easily observed at the tip of the keyhole.
The dynamics inside the pool, in this case, help the bubbles to remain close to the
solidification bottom surface, and porosity distributes randomly following the shape of
the solidification front. Conversely, the flow of the molten material at high speeds
(above 6 m/min) becomes slower at the bottom of the weld pool accelerating
solidification and preventing the collapse of the keyhole such that porosity is rarely seen
in this case. Figure 2.12 shows the effect of different speeds and laser powers on the
porosity and its position.
Figure 2.12 Relationship between welding velocity and porosity formation (Katayama
and Kawahito, 2009)
The flow inside the weld pool was also observed in other research by tracing tungsten
particles of 0.1 - 0.4 mm in diameter. The particles were embedded between AA5083
plates prepared for welding. During welding, the particles moved directly to the bottom
of the keyhole and then moved along an eddy backwards to the rear wall of the pool.
The speed of the flow reached between 0.25 and 0.35 m/s which considerably exceeds
the natural convection velocity in similar flows (Matsunawa et al., 1998).
54
In the same context, Arata (1980) and Antonov et al. (1977) investigated welding glass
and water as a new technique to investigate porosity formation and movement during
welding of transparent materials. Following their research, Berger et al. (2011)
conducted some experiments to visualise porosity formation in deep penetration keyhole
welding but using a cube of ice water heated by a CO2 laser beam. Although there is a
significant difference between ice water and metals regarding viscosity, thermal
expansion and conductivity, water was used to allow the porosity observation using a
high-speed camera. Their results showed that bubbles generated in water were filled
either by water vapour, the ambient air or by a mixture of both. Some of these bubbles
vanished within 1 ms, and some lasted for a relatively long time. The authors concluded
that the majority of pores aggregated at the root of the capillary and caused by the
accumulation of the water vapour inside the keyhole before it collapses. Moreover,
porosity generation at the capillary tip is very complicated to explain, and welding
velocity has a significant influence on porosity generation (Berger et al., 2011). Figure
2.13 gives a picture of the main flow of the pores during CO2 laser welding of ice water.
It should be noted that although transparent materials can be used only to show pores
motion within the pool, metals’ thermo-physical properties and their dependence on
temperature play a major role in generating the currents that drive the molten material
and pores, and this is quite different from water behaviour and its properties.
Figure 2.13 Main flow directions during ice water welding using a CO2 laser, laser
power: 500W, welding velocity: 2m/min (Berger et al., 2011)
It has been indicated by many researchers that porosity levels were lower in the
conduction mode welding than in keyhole mode welding. In keyhole mode, gases have
not got adequate time to escape from the welding pool due to rapid solidification of the
55
pool. In contrast, the stability of the welding pool in conduction mode allows gases to
be released from the bead due to the longer interaction time between the laser beam and
the material (Okon et al., 2002; Kuo and Lin, 2006c; Sánchez-Amaya et al., 2009b).
Also, Chang et al. (2013) showed using a computational fluid dynamics (CFD) model
that less porosity was observed in partial penetration in keyhole mode than in full
penetration keyhole mode welding. The simulation results indicated that the liquid
velocity reached a maximum value of 0.87 m/s at the top surface of the weld in the
partial penetration, while it peaked in full penetration welding at both the top and the
bottom surface. The validity of these approaches is discussed in detail in Chapter 3.
According to the authors, gases in partial penetration may be trapped inside the pool at
the lower surface due to the fluid movement and large pores were observed in this
situation. On the other hand, gas bubbles had the opportunity in full penetration welding
to be released from both the top and bottom surfaces due to the presence of the keyhole
(Chang et al., 2013). Figure 2.14 shows a comparison in porosity level between partial
and full penetration welding.
(a)
(b)
Figure 2.14 Pore morphologies in (a) partial penetration and (b) full penetration laser
welds in AA5083 (Chang et al., 2013)
In addition to the factors above, the effects of the filler wire melting behaviour and
process stability on porosity in the Al welding were studied by Tao et al. (2013). Two
different alloys, AA6156-T6 and AA6056-T4 of thickness 1.8 mm were laser welded
using an AA4047 filler wire. The authors investigated the influence of overlapping
between the laser beam and the filler wire tip as well as the leading and trailing angles
on the number of pores generated in the bead. Figure 2.15 shows porosity in the welds
at different parameters.
56
Figure 2.15 Porosity inside the Al weld under various settings (weld length: 200mm)
(Tao et al., 2013)
According to Figure 2.15 and Tao’s conclusions, porosity was found to be less when the
filler wire is partially overlapped with the laser beam in comparison with complete
overlapping. Complete overlapping directs the entire melting energy to the filler wire
without sufficiently melting the base material. This periodic and intermittent melting of
the wire results in spherical growing droplets of the molten wire that drop into the
melting pool and disturb the stable flow of the molten material. This procedure
deteriorates the weld appearance and makes it easy for the molten metal droplets to push
the surrounding gases into the weld pool. Also, feeding the wire from the leading
direction is more advantageous because the wire helps to fill the space around the
keyhole with the material faster than it does if it is fed from the following direction. It
should be noted that the wire feeding angle plays an important role in porosity
generation and small feeding angels are preferable for porosity reduction than large
angles; this is because small angles enhance keyhole stability and allow a more steady
flow of the material inside the pool (Tao et al., 2013). It is important to mention all the
samples used in the authors’ study were chemically cleaned and the porosity in noncleaned samples was never given.
57
In a related research (Park et al., 2010), the authors showed that the filler wire feed rate
does not affect porosity percentage in AA2024 welds, but it may, along with other
feeding parameters, reduce underfill and undercuts.
Understanding pores development was not limited to experiments and practical work.
Ol'shanskii and Morozov (2004) highlighted the effects of the shielding gas pressure on
the nucleation and development of porosity during Al welding using a mathematical
model. They developed their model upon previous works until they introduced Equation
( 2-2 ) which gives the minimum gas pressure under which full suppression of porosity
can be achieved:
Pcr 
([ H ]0  [ H ]d )2
[ H ]2P
( 2-2 )
where Pcr is the external pressure resulting in complete pores suppression, [H]0 is the
hydrogen concentration in the parent metal, [H]d is the additional concentration of the
hydrogen transferred into the pool from the surface of the metal and [H]p is the
equilibrium concentration of H2 at the solidification temperature of the metal
(Ol'shanskii and Morozov, 2004).
Since high shielding gas pressures have a direct influence on pore generation, Katayama
et al. (2003) studied the content of pores by drilling small holes in the welds in a
vacuum chamber to reach the gases trapped inside. A Q-mass-spectroscopic analysis
was then carried out, and they confirmed that He and H2 were found to be the dominant
gases inside the weld when He was used as a shielding gas. Moreover, the H2
percentage increased with time during diffusion. Zhao and DebRoy (2011) concluded
that porosity may be generated due to the pores existed in the parent material during
manufacturing (such as casting); however, these pores are less pronounced in wrought
Al alloys and hence have less influence on the ultimate porosity after welding. In this
case, porosity was found to increase in the fusion zone due to the pores expansion and
the reduction in the gas pressure inside the pores; this occurs due to the heat generated
in the fusion zone.
58
2.6.3
Hydrogen solubility
As mentioned before, pores may be formed when gases are entrapped inside the weld
pool. In this section, the origin of these gases will be discussed, and more attention will
be drawn to the effect of H2 since it is considered to be the most common source for
metallurgical porosity in the welds.
In fact, there are various sources of H2 in Al welds. Hydrogen can be released from
moisture at the molten surface of the pool and dissolves in the liquid metal forming
pores; or it can be generated from contaminants on the surface, lubrication, parent
alloys, oxide layers, coated surfaces or water vapour condensed on the inner walls of the
inert gas cylinders and tubes (ASM International, 1993). Moisture is a major factor
responsible for increasing H2 concentration in the weld zone because molten Al can
easily react with water vapour and release hydrogen gas according to the following
formula:
2 Al  3H 2O  Al2O3  3H 2
( 2-3 )
In general, the solubility of H2 in Al and its alloys increases with temperature. Figure
2.16(a) illustrates the variation of H2 solubility in Al at different temperatures. It is
evident from the diagram that the solubility climbs sharply when Al is transformed from
solid into a liquid state at the temperature of around 660oC for pure Al. Additionally, H2
solubility in liquid Al increases dramatically and approximately doubles its value for
every 100 oC increase in temperature (ASM International, 1993). The difference in H2
solubility between Al and iron is given clearly in Figure 2.16(b). The ratio between the
maximum solubility of H2 to the solubility at the melting temperature can reach “70” for
Al compared to only “1.6” for iron. This may explain the ease of H2 porosity generation
in Al welds in comparison with steel welds (Mazur, 1992).
59
(a)
(b)
Figure 2.16 (a) Variation of solubility of H2 in Al with temperature (ASM
International, 1993), (b) solubility of H2 in Al and iron (Howden, 1971)
Different investigations were made to measure H2 solubility in pure Al. Ransley and
Neufeld (1989) formulated two equations to calculate H2 solubility in Al as a function
of temperature and H2 partial pressure (see later in Chapter 5). The calculated values at
certain temperatures indicated that H2 is 20 times more soluble in molten Al than in
solid Al. Due to the high temperature of plasma above the weld pool, H2 solubility can
be promoted due to the generation of atomic hydrogen (Ransley and Neufeld, 1989;
Gingell and Gooch, 1997; Sekhar).
It can be seen from the previous discussions that different references gave various
values for the ratio of H2 solubility in solid and liquid Al; however, it can generally be
concluded that the H2 solubility climbs sharply once the phase transition occurs.
Anyalebechi (1995) studied the impact of alloying elements on the hydrogen solubility
in Al alloys. He showed that the increase in Ti, Li and Mg content promotes H2
solubility whereas the presence of Zn, Si, Cu and Fe reduces it. The author attributed
this trend to the strong attraction between Ti, Li, Mg and H2 which pulls H2 atoms away
from Al, and to the strong bonding between Al and Fr, Zn, Cu and Si which leaves very
few Al atoms to bond with H2. In addition, the introduction of some alloying elements
was found to affect the viscosity of Al and the internal energy by acting as surfactants.
The addition of surface active elements such as Li, Mg and Bi lowers the interfacial
energy and enhances H2 pores formation (Zhao et al., 1999b; Fang and Anyalebechi,
1988). Moreover, the author stated that high Mg content in Al alloys increases Mg
vaporisation and its content loss at high energy input levels and contributes to porosity
60
formation. This was pronounced in AA5754 and AA5182 laser welding when high
energy input was desired to achieve deeper penetration as indicated by Punkari et al.
(2003). Figure 2.17 shows the variation in hydrogen solubility in Al as a function of
alloying elements contents. The effect of the alloying elements will be examined in
Chapter 6 when the effect of the filler wire’s composition on porosity is studied in laser
welding.
Figure 2.17 Effect of alloying elements on the solubility of hydrogen in liquid Al at 973
K and 1 atm partial pressure of H2 (Anyalebechi, 1995)
It should be noted that H2 micro-porosity appears at high cooling rates and high
solubility while macro-porosity usually becomes visible by a different mechanism,
namely the keyhole instability (Zhao et al., 1999b).
The increase in the weld pool temperature has a direct influence on the growth of H2
pores due to temperature-dependent gas expansion. Moreover, spherical bubbles tend to
move up to the surface by buoyancy and the rising speed u depends on, according to
Stoke’s law, both the viscosity of the liquid and the radius of the pore as shown in the
following equation (Zhao et al., 1999b):
61
u
2r 2 . .g
9
( 2-4 )
where r is the radius of the pore, Δρ is the density difference between the liquid phase
and the gas inside the pore, g is the gravitational acceleration, and µ is the dynamic
viscosity of the fluid. For a schematic view of the welding pool, see Figure 2.12.
Cooling rate and the heat input also control the growth of the H2 pores inside the welds.
High cooling rates do not give sufficient time for H2 to diffuse and lead to porosity
formation and growth with time. Therefore, welding processes such as laser welding at
high welding speeds reduce porosity formation or form porosity with small diameters
because it increases the cooling rates at a fixed laser power (Zhao et al., 1999b). Many
investigations (Thomas and Gruzleski, 1978; Vaccari, 1991; Metcalfe, 1945) showed
that rapid cooling (high cooling rates) decreases the volumetric fraction of H2 porosity
and Fang and Anyalebechi (1988) indicated that this could reduce the diameter of H2
pores.
2.6.4
Solubility of Oxygen
Oxides of Al, Mg, Mn and Si are very stable at the welding temperature and can
withstand high temperatures. Also, Al oxide has a lower density than Al and its
solubility in molten Al is very limited and difficult to estimate. Accordingly, molecular
oxygen is rarely noticed in the pores generated in the Al welds since it interacts
immediately with molten Al (Zhao et al., 1999b).
2.6.5
Nitrogen solubility
Nitrogen solubility in molten Al was estimated to be lower than 10-11 at% at about
660oC. N2 gas can react with Al at welding temperature and forms Al nitride according
to the following reaction (Massalski, 1990):
1
Al  N 2  AlN
2
( 2-5 )
Aluminium nitride AlN usually appears as a solid film floating at the weld pool surface
reducing the wavy motion of the pool surface and stabilising the formed keyhole, which
in turn, reduces keyhole-collapse bubbles. Moreover, the use of nitrogen as a shielding
62
gas eliminates the gas originated pores in the pool since the trapped N2 reacts with Al
and does not stay segregated for a long time (Matsunawa et al., 1998).
2.6.6
Keyhole Collapse
Compared to Mg alloys, Al alloys have higher surface tension, higher boiling
temperature and lower equilibrium vapour pressure. Therefore, Al alloys suffer from the
instability of the keyhole due to evaporation of light alloying elements such as Mg and
Zn (Zhao and DebRoy, 2011; Beyer et al., 2012; Caiazzo et al., 2012). This arbitrary
evaporation causes the vapour pressure inside the keyhole to decrease and promotes the
collapse of the keyhole during welding (Zhao and DebRoy, 2011).
In a different aspect, some theoretical models concerning the transient state of the forces
acting on the keyhole indicated that even small fluctuations in the incident laser beam
power for a certain keyhole radius may lead to the keyhole collapse (Kroos et al., 1993).
Klein et al. (1994) identified the lower and upper limits of the absorbed power at which
the keyhole becomes stable and unstable respectively. In later research (Klein et al.,
1996), they stated that fluctuations greater than 1% in the incident power could be
significantly amplified to cause the keyhole fluctuations if they meet the resonance
frequency of the keyhole collapse. For instance, the resonance frequencies of the
keyhole during laser welding of 1 mm Fe plate thick lay between 500 and 3500 Hz
depending on the process parameters.
Most of the researchers in the field investigated the keyhole collapse phenomena from a
hydrodynamic point of view since the fluid-flow-related properties of the molten
material play an active role in the process. Kroos et al. (1993) created a theoretical
model for the collapse of the keyhole assuming that the keyhole takes a cylindrical
geometry. In keyhole welding, in order to maintain the keyhole, the pressure of the
metal vapour inside the keyhole should balance the hydrostatic pressure and the
pressure generated by the flow of the metal around the keyhole. However, the dynamic
pressure and the hydrostatic pressure can be neglected due to the small dimensions of
the weld pool, and the keyhole diameter was assumed to be 1.7 times larger than the
laser beam diameter. In addition, they estimated the time of the keyhole collapse in Fe,
Cu and Al welding to be of the order 0.1 ms. Their results concluded that in order to
63
obtain a continuous keyhole during pulsed laser welding, the time interval between the
pulses should be less than 0.1 ms (Kroos et al., 1993). Kim et al. (2004) have
determined the time of the keyhole collapse in AA5083 laser welding as 0.6 µs while
the solidification time was much larger than 8 µs. The same phenomenon was observed
by Matsunawa et al. (1992) in pulsed laser welding of Al alloys, and they concluded
that this type of porosity could be avoided by changing the laser pulse shape.
Another research attributed the keyhole collapse to the surface tension that the keyhole
walls cannot balance by the vapour pressure on the inside. Their simulation results show
that the keyhole collapse occurs faster at the middle part of the keyhole wall than on the
top or the bottom, which is believed to be the cause for gas entrapment at the bottom of
the keyhole (Ducharme et al., 1993).
Schauer and Giedt (1978) studied the keyhole stability in electron beam welding (EBW)
of AA1100 alloy. The keyhole in EBW is formed by a similar mechanism to laser beam
welding, and its stability is a common problem in EBW as well. They calculated the
surface tension force and the vapour pressure force acting on the inside wall of the
keyhole using the temperatures at different heights of the keyhole’s walls. Their results
show that the surface tension in the upper region of the keyhole is greater than the
pressure force; however, the two forces balance each other at a certain level and then the
pressure force begins to increase and becomes larger than the surface tension force at
the middle of the keyhole. This indicates that the liquid will flow downwards from the
upper region of the keyhole due to the low-pressure force while the space at the keyhole
bottom will be maintained by the high vapour pressure. This, in turn, will cause the
molten material to aggregate in the middle of the keyhole and cause the keyhole to
collapse entrapping the vapour and other gases at the bottom of the keyhole as shown in
Figure 2.18 (Schauer and Giedt, 1978).
64
(a)
(b)
Figure 2.18 (a) Schematic diagram showing the formation of the void at weld root due
to the imperfect collapse of Keyhole (left) (Pastor, 1998) (b) Liquid projection
formation at a level where the vapour pressure force approximately balances the surface
tension force (right) (Schauer and Giedt, 1978)
It should be noted that the flow dynamics in keyhole welding at low Péclet numbers Pe,
which compares the advection rate of a physical quantity to the diffusion rate of that
quantity, are significantly different in comparison with welding at high numbers. Pe
number can be described in laser keyhole welding context using the thermo-physical
properties of the molten material and the process parameters as follows (Steen, 1998):
Pe 
VRk
2D
( 2-6 )
where V is the welding velocity, Rk is the keyhole radius and D 
k
is the thermal
cp
diffusivity.
The differences induced by the transition from low to high Pe numbers can be attributed
to the change in the weld pool’s dimensions and shape, molten material flow inside the
pool and the cooling rates that control solidification (Dowden et al., 1987).
65
2.7 Porosity reduction
In the previous sections, the mechanism and sources of porosity were explained. In
general, porosity can be eliminated when the reasons behind its formation are excluded.
In this section, the review will focus on some of the most common methods used to
reduce porosity in Al laser welding.
2.7.1
Effects of magnetic field
The application of a magnetic field perpendicular to the welding direction was one of
the common methods investigated by many researchers. Zhou and Tsai (2007b)
developed a mathematical model to simulate the effects of the static magnetic field on
the quality of the weld in pulsed laser welding of steel. They concluded that the
electromagnetic force for medium aspect-ratio pools has the ability to control the
backfilling speed of the molten material. Although a high value of the electromagnetic
force can remove porosity from the weld root, it may create other defects such as a
protrusion in the centre accompanied with undercuts at the edge.
Bachmann et al. (2013) ran a CFD simulation for laser welding of a thick Al alloy plate
under the application of around a 500 mT magnetic field. The movement of the molten
Al in the magnetic field induces Lorentz forces that act in the opposite direction of the
flow if the magnets were located perpendicular to the welding direction. The Hartmann
number Ha2 is a dimensionless number that can be used to evaluate the strength of the
electromagnetically induced frictional (viscous) forces and can be expressed as follows:
Ha
2
 B L

2

( 2-7 )

where B is the magnetic flux density, L is a characteristic length (can be the half-width
of the weld pool), σ is the electric conductivity and η is the dynamic viscosity.
Using a Hartmann number of about 104, the simulation results showed that eddy
currents induced by the magnetic fields not only slows the flow movement inside the
pool but also result in considerable porosity reduction in the welds. The simulations
output were of an acceptable agreement with the experiments conducted using a 16 kW
66
disk laser to weld an AlMg3 alloy (Bachmann et al., 2013). Figure 2.19 illustrates the
effect of the electro-magnetic field (EMF) on an AlMg3 Al weld. Despite the fact that
the EMFs reduced the weld defects and improved the weld shape, large pores were still
observed in the upper and lower cross sections of the weld and the welding was only
performed on thick (25-30 mm) bead on plates (BOP).
(a)
(b)
Figure 2.19 Macro sections of two welds (a) a reference case without a magnetic field
(b) the case with a magnetic field with different polarity (Bachmann et al., 2013)
In addition to the aforementioned observations, Schneider et al. (2013) noted that an
oscillating EMF could enhance the outgassing through the top oxide layer in laser
welding of Al alloys, which results in a more uniform weld shape and less porosity.
However, this is actually dependent on the relative location of the laser point to the
magnetic poles location. In the same context, Vollertsen and Thomy (2006) studied the
use of the alternating EMF to achieve an even distribution of the filler wire across the
weld bead of 3 mm thickness. They applied a coaxially EMF with intensity up to 60 mT
and frequency up to 20 Hz, and they used a Cu foil in between the Al plates to detect the
effect of magnets on the melt flow since Cu has a high electric conductivity. Their
experiments showed that dilution was improved using this level of intensity since the
EMF affected the melt flow positively and it did not cause any change in the weld bead
shape. However, the authors stated that the use of the Cu foil has an effect on the melt
flow and this drawback cannot be avoided in their experiments.
67
Although the previous research showed that EMF could be used as a technique to
reduce weld defects, authors have limited their experiments only to relatively thick
sheets, in which a large mass is being circulated in weld pool, and only to butt or
overlap joints without investigating the effect of EMF on thin sheets, in which the flow
is restricted to small volumes or the effects on different joint types such as T- or edge
joints. Moreover, the use of EMF requires more equipment to be integrated with the
laser heads and more safety requirements to be met either by hardware or personnel,
which causes an increase in the production cost.
2.7.2
Effects of the shielding gases
Matsunawa et al. (1998) investigated the influence of the shielding gas type on porosity
formation in CO2 laser welding. The author compared the quality of laser welded plates
using two different types of shielding gases such as helium and pure nitrogen. Their
experiments showed that the use of nitrogen as a shielding gas produces Al welds with
less porosity and deeper penetration. The authors related that to the chemical reaction
that takes place between Al and nitrogen which results in the AlN compound which is
believed to be responsible for a better keyhole stability as discussed in section 2.6.5
(Matsunawa et al., 1998). Seto et al. (2000) used an X-ray transmission high-speed
camera to observe the voids in the welds shielded with He and N2 emphasising
Matsunawa et al.’s (1998) findings through the images shown in Figure 2.20.
Figure 2.20 Effect of shielding gas type on porosity formation in 5083 Al Co2 laser
welding (Seto et al., 2000)
68
Figure 2.21 Temporal behaviour of metallic plasma and keyhole in CO2 laser welding
using nitrogen shield (intermittent generation of Nitrogen plasma) (Matsunawa et al.,
2003)
In further research, Matsunawa et al. (2003) studied the stability of the keyhole and the
impact of plasma generated from shielding gases on porosity content in the welds.
Although the shielding gas plasma was absent due to the high ionisation potential of
helium, the metallic plasma formed a continuous keyhole during all times. The
continuous and open keyhole helped the metallic plasma to grow and to build up in the
open volume of the keyhole. This, in turn, caused the metallic plasma to be trapped at
the tip of the keyhole and to form porosity at the solidification front when the keyhole
collapsed under the hydrodynamic forces effects.
On the other hand, nitrogen plasma was noted to cyclically form and extinct during the
course of welding. Figure 2.21 depicts the formation/suppression cycle of the nitrogen
plasma during CO2 laser welding of stainless steel. The keyhole, therefore, follows a
cycle of formation/suppression reciprocally over time when the metallic plasma escapes
with the gas plasma reducing the keyhole depth. Due to the fact that the cycle time of
the metallic and N2 plasma formation is shorter than that of the keyhole instability and
the fact that nitrogen reacts with Al to form a chemical compound, porosity is
eliminated when nitrogen is used as a shroud (Matsunawa et al., 2003).
69
It was observed that there was no fluctuation in the angle of the plume generated in
keyhole laser welding at high speeds more than 100 mm/s (Matsunawa et al., 1998).
This indicates that the semi-stationary state of the keyhole, which can be obtained at
high welding speeds, was in a good agreement with Matsunawa and Semak’s (1971)
mathematical model that studied the keyhole dynamics.
Kamimuki et al. (2002) and Blackburn et al. (2009) discussed a different issue in laser
welding, namely the position of the shielding nozzle and its impact on keyhole stability
and porosity. Figure 2.22(a) shows that when the nozzle is shifted from the optimum
position, the gas drives the material into the keyhole and makes it narrower and
unstable. This causes a frequent collapse of the keyhole and hence generates more
porosity. However, when the nozzle is shifted away from the welding zone as in Figure
3.18(c), this will eliminate the protection of the fusion zone that shielding gases usually
provide. The authors proposed the optimum position of the nozzle to be as shown in
Figure 2.22(b) in which the gas helps to widen the keyhole and makes it easier for the
gases to escape from the welding pool during translation. The use of side gas jet was
found to decrease porosity and spatter and improve penetration during Nd:YAG laser
welding of stainless steel (Kamimuki et al., 2002).
Welding direction
(a)
(b)
(c)
Figure 2.22 The effect of nozzle position on the molten pool and the gas/material
interaction in stainless steel laser welding. The gas flow is directed towards (a) the
molten pool surface (b) the keyhole opening and (c) the front of the keyhole
(Kamimuki et al., 2002)
70
2.7.3
Effects of surface preparation
Some methods, such as scraping, chemical etching, sandblasting and laser cleaning,
were used as a surface pre-treatment before welding to prevent or reduce voids
generation.
2.7.3.1 Chemical and mechanical cleaning
Chemical etching and mechanical cleaning such as scraping, grinding, and sandblasting
have been in service for many decades. Chemical cleaning was preferable because it is
easy to apply by wiping or brushing and it can clean large areas in a short time.
Some researchers utilised chemical and steel brush cleaning to reduce the bubbles
generated during welding Al-Li alloys (Yang et al., 2008; Wang et al., 2008). Surface
preparation with optimised shielding gas parameters was also found to have positive
effects on porosity reduction during CO2 laser welding of AA6061, AA5052 and
AA5083 alloys (El-Batahgy and Kutsuna, 2009). The surfaces of these alloys were
cleaned using a stainless steel brush and nitric acid solution to remove the oxide layer
on the top surface. Although it was mentioned that chemical cleaning improves the
welding quality, the authors did not present any comparison between the weld quality
before and after cleaning. Moreover, the authors stated that the improvement in the weld
quality and the CO2 welding efficiency is attributed to the increased surface roughness
of the cleaned surfaces, which promoted the laser-material interaction (El-Batahgy and
Kutsuna, 2009).
Similar to the previous work, Verhaeghe and Hilton (2004), who used helium gas with a
low dew-point during welding of a cleaned AA2024 alloy, indicated that scrapping and
chemical etching could hinder bubble generation in the weld, provided welding occurs
directly after cleaning to prevent atmospheric moisture pick up. The contaminants,
lubricants and dirt were removed in this work by rinsing the samples in nitric and HF
solution for about 12 min followed by acetone wiping (Verhaeghe and Hilton, 2004).
The cleaning procedure presented in this work is very demanding and hazardous due to
the manual labour work required for scraping and grinding, and the application of
hazardous chemical substances.
71
The main drawback of using mechanical cleaning such as scraping and the use of a steel
brush is that these methods severely damage the parent material surface and produce
evident grooves and scratches that may affect the final appearance of the weld bead.
Moreover, these techniques are difficult to control and are entirely operator-dependent
regarding cleaning evaluation and repeatability.
In the same context and despite the quick and easy application of chemical cleaning,
chemicals may have detrimental effects on the base material as acids can interact with
the Al oxide film and may generate hydroxides, which are the primary source for
hydrogen pores in welds. Moreover, some chemicals such as Alkalis may attack the
magnesium in the AA7075 Al alloys causing a significant reduction in the alloying
content of the superficial layer. The detrimental effect of chemicals on the parent
material is not the only disadvantage of chemical cleaning, environmental problems
associated with the storage, disposal and the use of these chemicals become a challenge
and a costly process in the industry.
2.7.3.2 Sandblasting
Sandblasting involves projecting a high-velocity jet of sand or metallic powder on
surfaces to be cleaned. It has been indicated by some researchers that sandblasting is
necessary to reduce porosity in laser welding of Al (Haboudou et al., 2003b) and
stainless steel (Guo et al., 2015). Sandblasting before A5083 laser welding, as shown in
Figure 2.23, reduced porosity by only 3% in comparison to the as-degreased welds.
Despite the advantage of being a dry cleaning method in which no acids or liquids are
used, sandblasting removes a variable depth of the substrate and cannot be repeated to
the same condition by all workers. Automated systems can be used in a controlled
environment for production of small products; however, they are difficult to use for
large workpieces such as boats, aeroplanes and locomotive vehicles which require decoating or paint stripping since intensive programming is required for identifying the
origin for the cleaning paths. Moreover, sandblasting promotes environmental pollution
due to the types of powders used and may cause health issues to the operators (Vargel,
2004).
72
2.7.3.3 Laser cleaning
Laser cleaning was first developed for cleaning semiconductors used in board-based
industries in 1991 (Imen et al., 1991). Laser cleaning was then utilised for other
applications including TIG welding of 2-3 mm Al alloys using pulses duration of 10 1000 ns and a pulse frequency of 1-30 kHz (Lima et al., 2002; Qiang et al., 2016).
Focusing on a different material, Pashby et al. (1997) removed the zinc coating layer
from the steel substrate using a pulsed laser as a preparation procedure for spot welding.
Haboudou et al. (2003b) focused on the surface preparation of the parent material prior
to welding and its influence on bubble suppression. They compared different techniques
for surface cleaning and observed their effect on porosity when welding A356 and
AA5083 Al alloys. A number of samples were laser cleaned (pulsed Nd:YAG laser, 1.5
J/cm2, 20 Hz, 20 mm/s).
Figure 2.23 X-ray inspections followed by image analysis of laser-welded beads
showing the influence of surface preparation in AA5083 (Haboudou et al., 2003b).
The results showed that surface preparation had reduced porosity by different
percentages depending on the cleaning method used. Figure 2.23 illustrates the
percentages by which porosity was reduced for each cleaning process during AA5083
laser welding. Although Haboudou et al. (2003b) examined laser cleaning on porosity
generation, their work was limited to autogenous welding of thick Al plates (4 mm in
73
thickness) without the use of a filler wire. The authors did not support their findings
with surface analysis after laser cleaning and only analysed the surface after
sandblasting. Also, they attributed the porosity reduction to the use of low welding
speeds rather than high speeds as reported in other literature. The Al alloys were welded
using only one level of power at two levels of welding speeds, representing a narrow
processing window within which laser cleaning can effectively reduce porosity. Also,
the welding area was shielded with helium of 20 l/min flow as well as a cross jet above
the sample for preventing optic damage. It should be noted that this shielding set-up
may not be industrially feasible due to the He high cost.
Only a couple of studies have been carried out on laser cleaning prior to welding and the
majority aimed to weld Al alloys using TIG, spot resistance welding or using
autogenous laser welding of two alloys. To the best of the author’s knowledge, no work
has been reported on laser cleaning of Al alloys before laser welding with filler wire.
Therefore, AlShaer et al. (2014) presented for the first time the laser cleaning effect on
porosity during heterogeneous laser welding of AC-170PX sheets for the automotive
industry. More details on this work are presented in Chapter 7.
2.7.4
Effects of laser beam characteristics
Most of the aforementioned work done on porosity reduction concentrated on external
factors related to parent materials, surface preparation, shielding gases. In contrast,
some researchers including Matsunawa et al. (2003) investigated the difference that
pulsed laser beam may have on porosity reduction compared to continuous wave (CW)
laser welding. The authors built their experiments on the phenomenon that the keyhole
should be rapidly closed during a pulsed laser ON/OFF cycle before the voids can even
form. Many experiments were conducted to test various pulse duty cycles from 50% to
100%. Their results shown in Figure 2.24 clearly concluded that reducing the pulse duty
cycle up to 50% can be significantly effective for porosity suppression.
74
Figure 2.24 Effect of pulse modulation on the reduction of porosity in AA5182 laser
welding (Matsunawa et al., 2003)
Twin spot laser welding, as an alternative technique, was also found to reduce porosity
in Al welds that can be used in car body structures and other applications (Yao and
Gong, 2011; Shibata et al., 2003; Haboudou et al., 2003a; Hayashi et al., 2003; Gref et
al., 2003). Haboudou et al. (2003b) investigated the use of two laser spots instead of a
single spot Nd: YAG laser during Al laser welding (see Figure 2.25). A 4 kW Nd: YAG
laser was used in welding with a spot diameter of 600µm and an optical prism was used
to split the beam into two spots. It was observed from the experiments that dual spot
welding stabilises the flow inside the pool (observed using high-speed camera analysis)
and reduces porosity to less than 2% for AA5083 alloy.
In a relevant research, Yao and Gong (2011) explained that the two laser spots create
two adjacent keyholes in the welding pool which in turn join together instantly and form
a bigger keyhole. The large keyhole is more stable than the small keyhole and does not
suffer from fluctuation at its walls. Moreover, this type of keyhole reduces the flow to
the keyhole back edge when vapour plasma reaction takes place. This mechanism
promotes the suppression of detrimental bubbles in Al laser welding.
Although the double spots technique may reduce porosity, special optics are required
for dividing the beam into two regions and for offsetting them at a given distance.
75
Moreover, it is usually needed to reduce the welding speed in order to achieve similar
penetration depth as in the single-spot mode (Haboudou et al., 2003a). These
drawbacks, however, can be avoided in conventional laser welding using surface
preparation, which is independent of the laser welding process.
(a)
(b)
Figure 2.25 (a) Dual spot laser welding (an inline beam and a cross beam) (b) effect of
dual spot welding on porosity in AA5083 and AA356 alloys (Yao and Gong, 2011)
2.8 Summary
In this chapter, the deteriorating effects of porosity, as well as the formation
mechanism have been highlighted in some detail. Various methods were introduced
to reduce porosity in Al welds such as the use of a magnetic field perpendicular to
the welding direction; the use of dual spot laser beams instead of a single spot;
utilising pulse modulation; surface preparation of the parent material and the use of
He and N2 as shielding gases during welding.
Although major efforts have been exerted to understand porosity formation
mechanism, there is still no consensus as to what the main cause of porosity is. In
order to comprehensively understand the origin of pores inside Al welds, further
research is needed through experiments and modelling. It should be noted that for
automotive applications, porosity content should be less than 10% of the weld
cross-sectional area in all types of joints. For weld seams used for sealing purposes,
76
the ultimate tensile strength is not subject to a specific criterion since the weld is
used only for sealing and not for carrying large structural loads. However, porosity
content and the weld appearance are of a great importance since the structural welds
will be treated and painted immediately after welding without any further
processing. Therefore, any superficial pores will not be allowed under any
circumstances and the weld appearance and colour should be controlled carefully
by choosing the correct filler wire. More details on the target welding specifications
required in this project will be provided in Chapter 4.
Since a part of this project involved SPH modelling of the laser cleaning/ablation
process, a literature review on laser processing modelling is given in the next
chapter.
77
3
CHAPTER 3
A LITERATURE REVIEW PART II:
SMOOTHED PARTICLE
HYDRODYNAMICS (SPH) MODELLING OF
LASER PROCESSING
3.1 Introduction
Numerical modelling is established as an efficient tool for understanding complex
problems and phenomena in engineering. Numerical simulation of engineering
processes relies on expressing physical behaviour as mathematical relations that can be
analysed and solved by computers in order to model the specific problem. Modelling is
usually applied whenever experimental or analytical procedures fail to give a clear
understanding of the studied problem or can be used in case of time-consuming,
expensive and dangerous laboratory work (Liu and Liu, 2003).
Laser material processing is one of many complex phenomena in materials processing
and, as highlighted in the previous chapter, is very difficult to be fully understood using
experiments only. The knowledge of the temperature gradient in laser cleaning or
welding, for example, is a challenge due to the small dimensions of the molten zone.
Moreover, it is challenging (if not impossible) to attach thermocouples around such
small fusion zones or even to use a thermographic camera due to the plasma presence.
Once the model is created and validated with experiments, it reduces the need for
further experimentation (Rai, 2009). Additionally, it is very difficult to observe the
motion of the fluid inside the melting pool during welding and to obtain detailed
information about the velocity field without the help of numerical simulations.
78
Of the techniques reviewed in this chapter, SPH emerges as a viable technique that can
simulate problems with highly non-linear deformation such as the motion of multiphase fluids, movement, etc. SPH is one of many meshless (mesh-free) techniques
which are based on a Lagrangian description of motion (explained herein). SPH has
demonstrated its ability to model various physical phenomena such as fluid flow, heat
and mass transfer and elastic-plastic deformation in a range of applications (Liu and
Liu, 2003; Dowden et al., 1987).
This chapter will navigate through the reported literature on modelling of laser
processing and will review the key publications in the field. The main motivation of
using SPH as a tool for modelling will be explained by highlighting the drawbacks and
limitations of the most common modelling techniques used.
3.2 Laser cleaning/ablation modelling
The laser cleaning process can remove surface contaminants either thermally by
evaporation, or mechanically by developing laser-induced shear stress at the interface
between the base material and contaminants. Lasers can also ablate the base material
(which does not contain contaminants) using either mechanism; therefore, herein, the
focus will be directed towards laser ablation due to the increasing body of experimental
work reported in the literature which can be used in validating the proposed model in
this thesis.
Laser-material interaction is a complicated process due to its dependence on the various
material- and process-related factors. The mechanism of laser ablation has not been
fully understood, and different explanations have been proposed to explain the
phenomenon (Bityurin, 2005; Russo et al., 2002). Modelling is one of the methods used
to understand this process and started in the early 1970s when the laser drilling process
began to be employed in industrial applications.
Different analytical solutions were developed to describe the effect of the process
parameters on the process output and to predict the behaviour of the material heated by
the laser. One of the early publications in this field, Allmen (1976), proposed a onedimensional analytical solution for predicting the surface temperature, which was
subsequently used to determine the rate of vaporisation for the drilled holes with an
79
aspect ratio less than “7”. The author’s model, which was created to simulate relatively
long pulses on the order of micro-seconds and used copper as a base material heated
with step-shaped pulses, succeeded in predicting the drilling depth and velocity within a
certain range of laser intensities. However, the one-dimensional model was unable to
account for the radial heat losses to the surrounding boundaries, and the model results
showed a poor agreement with the experimental results within the high-intensity regime.
This model was further developed by Zweig (1991) who created a thermo-mechanical
model to account for the thermal evaporation of heated gelatin and the radial ejection of
the molten material. The drilling depth, which was formed due to vaporisation and the
advancement of the molten material push-front, was predicted to observe the effect of
the optical penetration during microsecond pulsed laser drilling. Although the model
accounts for two mechanisms in one model, it was created to simulate laser drilling of
soft materials which behave quite differently from solid materials such as metals.
Moreover, the laser pulses used in this model were on the order of microseconds which
are much longer than the laser pulses used now in modern laser ablation, namely nanoand picosecond pulses.
Following this initial work, more analytical models were developed to predict the
ablation depth, the shape of the crater, short-pulse effects and the associated phenomena
such as plasma formation and shockwaves. Nevertheless, some of the more recent
formulation work (Shusser, 2009; Zheng et al., 2001; Song et al., 1996; Lutey, 2013a;
Gusarov and Smurov, 2005; Jeong et al., 1998; Bardy et al., 2016) still depended on a
one-dimensional model to predict the surface temperature and the ablation depth of the
three-dimensional metallic solids. Moreover, some models (Lutey, 2013a) used a
“shielding factor” to align the model results with the experimental data while others
neglected the effect of important physical properties in the process such as viscosity and
thermal conductivity (Shusser, 2009), radiated heat losses and the vapour-substrate
interaction in a superheated substrate (Oderji et al., 2016).
Tangwarodomnukun (2016) analytically modelled the cavity formation created by laser
ablation of an Al target immersed in a thin film of water. While the cavity profiles and
depths were well predicted at different parameters using still and circulated water, the
model failed to predict the correct surface roughness which was one of the main
motivations for developing the model. Moreover, the model was only capable of
80
predicting the cavity depth under the presence of water, and no results were shown for
dry laser ablation.
Some researchers (Ait Oumeziane et al., 2016; Cheng et al., 2016; Bulgakova et al.,
2016; Amoruso, 1999) have considered the effect of the photoelectric breakdown of the
base material due to the ultrashort laser pulse ablation, while others (Lunney and
Jordan, 1998; Bardy et al., 2016; Wu et al., 2007; Morel et al., 2008) focused on the
plasma interaction with the substrate and the shock wave role in creating the cavity.
However, these topics were out of the scope of this thesis since the model created in this
work concerns nanosecond laser ablation.
3.2.1
Mesh-based modelling of material processing
In order to overcome the limitations of using a one-dimensional model to predict the
surface temperature that fed into the laser-material interaction model, a significant
amount of research has been calculated using numerical techniques to simulate laser
processing. Finite element (FE) and finite volume (FV) methods have been the
dominant methods for a long time to simulate laser processing applications using
semiconductors, polymers and metals.
Although FE and FV methods have been used to simulate various laser processes such
as surface hardening (Jerniti et al., 2016; He et al., 2014), laser surface modification
(Tran et al., 2016; Picasso and Hoadley, 1994) and laser welding (Casalino et al., 2015;
Hartel et al., 2016), very limited publications in the area of laser ablation have been
generated and discussed only limited aspects of the process. Dry laser ablation was
modelled using FE (Liyang et al., 2012; Marimuthu et al., 2013; Lutey, 2013a; Singh
and Sharma, 2016; Lutey, 2013b; Arronte et al., 2003) and FV (Zhao and Shin, 2013;
Marla et al., 2013) methods to predict the ablation depth and the shape of the crater
generated by the consecutive pulses. As formulated, the FE method relies on the meshed
connection between the nodes of the elements that carry the mechanical or thermal
loads to the adjacent elements and so on. This irreplaceable connection between the
elements makes it difficult for the method to generate holes or depressions on flat
surfaces without depending on cracks that separate the mesh into two parts. However in
laser material ablation, the surface material abandons the substrate when its temperature
81
exceeds the boiling point without initiating cracks inside the base material. In order to
overcome this problem, the elements whose temperatures exceed the boiling point were
simply deleted and removed from the computation to simulate the protrusion generated
by laser processing. This procedure prevents mesh-based methods from simulating the
ejected material or providing any information on the interaction between the different
phases at the surface. In addition to the previous issue, the mesh-based methods require
mesh optimisation prior to simulation while the selection of the mesh resolution,
element type and degrees of freedom can be critical and time-consuming. Moreover,
small physical domains of micro- or nanometers scale can still represent a
computational challenge when the element size must be smaller than the ejected
material or the liquid droplet by several times (Gross, 2008a).
In addition to modelling the solid materials with the FE and FV methods, CFD was also
implemented in the modelling of laser processing in order to include the hydrodynamic
behaviour of the multi-phase flows. A significant number of publications were reported
in the literature concerning the plasma/plume behaviour during laser ablation at very
high fluences greater than 10 J/cm2, in which phase explosion is clearly pronounced.
These publications were produced using either one-dimensional (Ding et al., 2012;
Vertes et al., 1989; Autrique et al., 2013) or a two-dimensional hydrodynamic models
(Rajendiran et al., 2014; Zhao and Shin, 2013; Ohnishi et al., 2015; Povarnitsyn et al.,
2015; Colombier et al., 2006) in which the ions and plasma flows were simulated under
different surrounding conditions. Singh and Sharma (2016) and Ohnishi et al. (2015)
studied the effect of a magnetic field on the plasma-induced shock wave and its
relationship with the ablation depth, while Yu et al. (2015) investigated the electron
collision frequency influence on the laser ablation process on gold.
As previously discussed, the hydrodynamic models were created to study the plasma
behaviour without drawing attention to the main objectives of the modelling which are
the ablation depth, crater shape and the process parameters effect on ablated surface.
Although the proposed hydrodynamic models contributed to further understanding the
issues associated with plasma generation during laser ablation, some simulations were
not validated (Autrique et al., 2013; Rajendiran et al., 2014) or were poorly validated
against available experimental data (Vertes et al., 1989). Moreover, the majority of the
models as previously mentioned were limited to 1-D or 2-D computational domains
82
without being extended to three-dimensional spaces, and they were all mesh-based
models which are inherently unable to provide a full picture of both the ablated and the
base materials. Additionally, such models discussed only the high fluences (f) regime (f
>10 J/cm2) used for generating the plume without utilising their advantages in
investigating the low (f <1 J/cm2) and medium fluences regime (1 J/cm2< f <10 J/cm2)
within which most industrial applications operate.
3.2.2
Mesh-less based modelling of material processing
3.2.2.1 Transient heat problem modelling
Modelling of transient heat problems and high deformations in mesh-based methods
requires a fine mesh in order to avoid singularities and to obtain a reliable prediction of
the physical quantities. Computationally expensive fine meshes can be replaced by an
adaptive mesh which changes accordingly to the distorted shape every single time step.
This procedure can reduce the computation time for relatively simple problems;
however, it will, according to Pham et al. (2012), increase the computation time when
the mesh experiences large non-linear deformations in a very short period (fraction of a
second), and re-meshing is required every time step in order to follow the shape of the
deformed mesh. Moreover, adaptive meshes are complex to produce on parallel codes
and do not necessarily conserve fundamental quantities such as mass and momentum
(Nijhuis, 2011). Due to the limitations of the mesh-based techniques, the alternative of
mesh-free methods have been developed which includes smoothed particle
hydrodynamics (to be discussed in detail in the next section), Meshless local PetrovGalerkin Method (Shibahara and Atluri, 2011), Meshless Finite-Difference Method
(Varanasi et al., 2012), the meshless element free Galerkin (MEFG) method
(Belytschko et al., 1994), a space-time meshless method (Sophy et al., 2012), VoronoiBased Discrete Least Squares Meshless Method (Labibzadeh, 2016).
A large number of publications summarised in Table 3.1 treated the moving or the static
heat source as a 2-D problem using the moving least square (MLS) approximation
without the expansion to 3-D of a real application (Das et al., 2011; Labibzadeh, 2016).
Although some models were created in three-dimensional space, most models were only
limited to heat transfer modelling and temperature fields prediction without showing the
83
possibility of coupling the thermal problems with other physical phenomena within the
same model.
Table 3.1 Selected published work on heat transfer modelling using some common
meshless methods
Problem
Method
Reference
(Das et al., 2011; Singh et
Welding heat transfer
Element Free Galerkin
al.,
2007;
Singh,
2004;
analysis
Method
Pham, 2013; Zhang et al.,
2013)
Heat conduction problems
Weighted Least-Squares
(Dai et al., 2011; Fang et al.,
in irregular domains
Collocation
2009)
Voronoi-Based Discrete
(Labibzadeh, 2016)
Transient heat conduction
Least Squares Meshless
Method
Inverse heat conduction
Meshless local Petrov–
(Sladek et al., 2006)
Galerkin
(Shibahara and Atluri, 2011;
Transient heat conduction
Meshless local PetrovGalerkin
Wu et al., 2009; Thakur et
al., 2010; Tian and Rao,
2012; Mirzaei and Schaback,
2014; Liu, 2006)
Pham (2013) used the meshless element free Galerkin (MEFG) method to simulate a
two-dimensional temperature field induced by a moving heat source on a steel plate,
such as what occurs in the welding process. Their model generated very primitive
results in terms of the temperature fields and did not present the method capability to
simulate the welding process and the heat-induced material flow inside the weld to
show the benefit of using such meshless methods. Moreover, the selection of some
method-related constants was carried out using the comparison with the FEM results for
the heat transfer problems, which questions the ability to pre-define the model’s
constant for problems that may not have an FE solution. In the same context, Shibahara
and Atluri (2011) simulated the same problem using meshless local Petrov-Galerkin
84
method (MLPG) in order to obtain the heat-induced residual stresses in the welded
plates. Although the authors accounted for the effect of temperature-dependent
parameters on the process and investigated a regular and an adaptive nodal distribution
for the methods, the supporting size for all cases was always larger than the test function
size by a factor of “2” or “3” and no other cases were presented to show the method
capabilities beyond these limits. Also, similar to the aforementioned research, the heat
transfer model was limited to 2-D cases and no structural analysis was presented.
As previously mentioned, mesh-free methods were developed in order to simulate large
deformations which accompany many physical problems such as fluid flows, explosions
and structural deformations. Over the past decade, researchers have been able to obtain
a solution through these methods to other industrial problems such as forming, welding
and cutting by combining the fluid flow formulations with the formulation of the
problem of interest. Table 3.2 shows selected publications on different engineering
problems in the field.
Although the published work given in Table 3.2 used different meshless methods in
order to simulate thermo-physical problems, some methods were not able to show the
flow of the material inside the studied area (Xiao et al., 2017); however, the results were
limited only to the temperature field distribution in the sample. While others (Alfaro et
al., 2009) were not able to show the mixed material final geometry in friction stir
welding, and, although the resolution was very poor for the application, Cueto and
Chinesta (2015) stated that most of the available meshless methods suffer from
unphysical representation of the process in the regions away from the boundaries.
Despite the fact that more developments are currently underway in these methods, they
are still facing issues in the true representation of the process and the computational
time especially the NEM method which depends on Voronoi representation of the
domain which must be created for each time step (Alfaro et al., 2006b).
85
Table 3.2 A number of published works on metal processing using some common
meshless methods
Application
Method
References
(Alfaro et al., 2006b; Cueto and Chinesta,
Metal Forming
-
2015; Chen et al., 1998a; Chen et al., 1998b;
Bonet and Kulasegaram, 2000)
Friction Stir welding
-
RKPM*
Forging,
Indirect Extrusion
(Alfaro et al., 2007; Alfaro et al., 2009; Xiao
et al., 2017; Pan et al., 2013)
(Wang et al., 2007; Xiong et al., 2005;
Xiong et al., 2007; Shangwu et al., 2003).
(Guedes and César de Sá, 2007; Li and
EFG*
Belytschko, 2001; Guo and Nakanishi,
2003)
Upsetting
Extrusion
Metal Arc Welding
EFG*
SPH*
Metal machining and
(Kwon and Youn, 2006)
(Ito et al., 2012)
(Cueto et al., 2003)
Biomechanics
Resin Moulding
(Kwon et al., 2005)
NEM*
Extrusion
(García et al., 2007)
(Alfaro et al., 2006a; Filice et al., 2009)
*RKPM: Reproducing Kernel Particle Method, EFG: Element Free Galerkin, SPH: Smoothed Particle
Hydrodynamics, NEM: Natural Element Method
3.2.2.2 Laser processing modelling
In another area of interest, some of the published work (for example Kim, 2011)
exploited the mesh-free isoparametric point interpolation method (IPIM) to simulate the
laser drilling process. The proposed model showed a significant sensitivity of the results
to the number of iterations for a given resolution. The shape of the laser drilled groove
at low resolutions was far from the real shape obtained experimentally, while the
groove’s right and left sides at high resolutions suffered from numerical oscillations and
no accurate representation of the groove could be generated. Moreover, the authors
assumed that the points which have a temperature higher than the vaporisation
temperature are moved away from the laser only in the vertical direction, and not in a
86
normal direction as occurs in the real application. More importantly, it was concluded
that the 2-D results cannot be taken as a true representation of the laser drilling process
since the 2-D numerical methods usually over-predicts the groove depth by three times
more than the 3-D models. Although the author further developed his model (Kim,
2012) and reduced the sensitivity of the model to the mesh resolution, the same
conclusion was reached, and the same limitations still exist.
Mesh-free collocation was employed by Xu et al. (2010) to model the use of laser
radiation in healing tumours for medical applications. The radial basis function
collocation technique was used to predict the laser fluence (laser power divided by the
area) distribution inside the tissues as a means for monitoring its concentration of the
thermal effect inside the tissues. The main motivation of using a mesh-free method for
such applications was the ability to deal with complex geometries and singularities in
the fluence values that other mesh-based methods may not be able to predict correctly.
It should be noted that there has been limited research focusing on developing laser
processing modelling using mesh-free methods whereas a large number concerned the
heat transfer and other engineering problems as a proof of concept. Many of these
methods are not straightforward to apply. In the meantime, only a limited number of
research works using SPH for laser applications have taken advantage of this method’s
potential flexibility for simulating molten and vaporised material flows which are the
area of interest for this project.
3.3 SPH interpolation and summation
Unlike many of the mesh-free methods in Table 3.2, SPH relies on discrete
computational points to evaluate different variables such as density, pressure,
temperature, where the studied domain is divided into arbitrarily distributed points that
are called particles. Each particle has unique properties such as mass mi, volume ωi,
pressure Pi, and velocity vi, where the subscript denotes the ith particle (Muhammad,
2012). Here, the basic SPH equations are briefly introduced for clarity of understanding
material presented later and the discussion.
87
The value of a function A(r) at a particular particle can be found by a local interpolation
for a set of surrounding particles at position r. The interpolated value of the function, A,
can be expressed by the following equation in continuous form (Gingold and J.J., 1977):
A(r )    (r  r ') A(r ')d 
( 3-1 )

where A(r) is the function value at a distance r, A(r' ) is the function value at a distance
r' , δ is the Dirac delta function, r  r' is the distance between particles.
Although the delta function provides an exact value for the interpolated function, it is
not used because it is infinitesimally narrow such that the interpolation region around
each particle does not overlap with other particles in the studied domain. Alternatively,
a weighting function or a smoothing kernel, W, is introduced to replace the delta
function. Using the smoothing kernel function, Equation ( 3-1 ) becomes (Gingold and
J.J., 1977):
A  ri    A  rj W  ri  rj , h  d 
( 3-2 )

where W is the smoothing kernel, h is the smoothing length (defines the width of the
kernel). This equation is a continuous approximation of Equation ( 3-1 )
If the infinitesimal length dΩ seen in Equation ( 3-2 ) is replaced with the particle
volume Vj, the mass of the particle can be calculated as follows:
m j  V j . j
( 3-3 )
By discretising Equation ( 3-2 ) and using Equation ( 3-3 ) (Violeau, 2012), the
approximate interpolated value of A at location ri is obtained:
A(ri )   A(rj ) W  ri  rj , h V j
( 3-4 )
j
Using a similar approach, the gradient of the function A(r) becomes the summation of
the smoothing kernel gradient over the neighbouring particles:
88
A  ri   
j
mj
j
A  rj  iW  ri  rj , h 
( 3-5 )
For the sake of simplicity, the value A(ri) will be written as Ai, similarly
W  ri  rj , h   Wij and iW  ri  rj , h   iWij .
Therefore, the previous equation yields:
Ai  
j
mj
j
where iWij 
Aj iWij
ri  rj Wij
rij
rij
( 3-6 )
.
In practice, different versions of gradient approximation are used depending on
accuracy and conserving fundamental quantities such as momentum (Violeau, 2012).
The smoothing kernel can take different forms: Gaussian, Quadratic, cubic spline (Bspline), or higher order kernels such as 4th or 5th order. Regardless of the kernel type,
the weighting function should be continuous over the smoothing length and its integral
should equal to unity over the support domain (Violeau, 2012). Figure 3.1 illustrates a
typical flow chart of SPH simulation for thermal problems. More details on the SPH
formulation and the smoothing kernel properties used in this project will be given in the
relevant chapters as well as any other related formulations.
89
Material’s thermal
properties dependent on
Temperature
Boundary conditions
Heat source properties
(Laser power distribution,
moving heat source)
SPH formulation of
conservation equations
Thermal Analysis
Simulation Output
Temperature
distribution
Ablation/weld/kerf
dimensions
Material flow and
its characteristics
Figure 3.1 Flow chart of a typical SPH thermal simulation approach
3.3.1
SPH modelling of material processing
SPH can simulate problems with highly non-linear deformation such as complex
movement of multi-phase fluids and topology changes. SPH is one of many meshless
techniques that are based on the Lagrangian description of motion which have proven
their ability to model various physical phenomena such as fluid flows, heat and mass
transfers, and elastic-plastic deformation (Violeau, 2012; Liu and Liu, 2003). SPH was
initially developed in 1977 by Lucy (1977) and Gingold and Monaghan (1977) to
capture astrophysical phenomena in which boundary conditions are neglected when an
open domain is under consideration. Later, by introducing boundary conditions, the
SPH method has enabled the simulation of engineering problems such as the flow of
water waves (Dalrymple and Rogers, 2006; Violeau and Issa, 2007), metal forming
(Cleary et al., 2006) and phase change problems (Zhang et al., 2007; Xiong and Zhu,
2010). SPH, therefore, appears appropriate for modelling fusion processes in which
liquid or vaporised metals can be handled easily.
90
Despite its obvious attraction, SPH modelling of laser processing is still in its early
stage and needs significant development since very few publications have been
conducted in this field. Demuth et al. (2012) simulated laser interference patterning of
metallic surfaces using SPH, following Tong and Browne (2011) who modelled laser
spot welding of Al with a very primitive 2-D model and a limited resolution without
providing any validation for the model. Moreover, the proposed model considered the
molten metal as an incompressible fluid and ignored the density change with
temperature which can be physically unrealistic.
Hu et al. (2016) and Hu and Eberhard (2016) modelled the laser welding process using
a 3-D SPH model coupled with a ray tracing program to simulate the laser absorption in
the keyhole. The authors identified the deep penetration in welding when the aspect
ratio of the weld depth to the width equals to “0.5” and presented the deep penetration
and vaporisation thresholds as a function of the feed rate (traverse speed). In addition to
the fact that these results were obtained with a very coarse resolution (~0.1 mm particle
size), they were compared to the experimental data and significant deviations of about
10% and 60% were recorded for the two studied thresholds respectively. The model did
not give any validation for the temperature field obtained by laser heating and did not
provide an estimation of the thermal stresses and strains. Moreover, the authors plotted
the Marangoni force in the x, y and z directions but did not present the particles
movement or the velocity field induced by these forces despite Marangoni forces
playing a further role in thermodynamic convection during laser welding. Therefore and
according to the authors, a significant development is needed in order for the model to
become an established and validated model.
In another approach, Gross’s (2008a) model on laser cutting of metals was further
developed by Muhammad et al. (2013) who simulated both dry and wet cutting of
stainless steel stents used for medical applications. The SPH model in (Muhammad et
al., 2013) considered the assist gas pressure to drive the molten material away from the
hole according to Bernoulli’s equilibrium equation. However, the velocity components
derived from Bernoulli’s equation was applied to all fluid particles whether they are
liquid or vapour, and this may jeopardise the simulation’s accuracy. While Yan et al.
(2011) modelled CO2 laser underwater machining of alumina ceramics, Chen and
Beraun (2001) presented their preliminary work on SPH modelling of ultrashort laser
91
pulse interactions with a metal film, in which electron heating and cooling phenomena
had to be considered. Cao and Shin (2015) simulated the phase explosion phenomena in
laser ablation of Al and Cu at very high laser fluences of up to 36 J/cm2. The authors
presented the ejected material’s size distribution during the ablation, but their model did
not predict the ablation depth for Al at different fluences and did not show the
simulation progression to the end of the ablation process, although the material’s
temperature at that time was still above the boiling point and more material was still to
be ejected. Hence, the authors did not show the final stage of the ablation process at
which the final ablation depth must be measured and compared with the experiments.
Moreover, the proposed SPH model was not able to predict the entire process but relied
on molecular dynamics and hydrodynamics models to calculate the initial position of
the ejected material, such that the SPH model could not simulate the interaction between
the vaporised material and the solid substrate.
None of the previous investigations on SPH modelling of laser heating evaluated the
accuracy of temperature values which were mainly used for calculating the melt ejection
velocity, kerf width and depth. Brookshaw (1986) firstly introduced the heat conduction
equations into the SPH formulation for cases where the thermal conductivity is constant
or changing slowly. The model created was robust and can be extended to embrace
more complicated cases. Due to its limitations, however, this model cannot be applied in
its form to applications such as phase transformation in which the thermal conductivity
and other thermal properties may vary rapidly with temperature or pressure. Also, the
model was not able to simulate multi-phase flows such as porosity in metallic welding
or sand casting in which discontinuities in the thermal properties can be clearly
observed at the interface between the different phases. Cleary and Monaghan (1999)
created an SPH model that takes into account the change in thermal conductivity and
behaves in a stable manner even if a discontinuity in the thermal conductivity exists.
The two-dimensional model was successfully capable of dealing with 1000:1 ratio in
thermal conductivity. Also, the boundary conditions were selected accurately to show
that the heat flux is conserved at the boundaries when complex geometries such as
curved boundaries are required or when boundaries are not aligned with the particles.
Jeong et al. (2003) modelled heat transfer with SPH using a different approach in which
the second-order partial differential equations (PDE) were decomposed into two first92
order PDEs: one is what was called a “balance equation” which is exact, and the other is
approximate. The main purpose of this separation is to make it easier to deal with multiphase flows in which only the approximate equations vary from one phase to another
while the balance equation remains the same.
Although the authors investigated different geometries and initial temperature profiles,
they only studied steady-state problems and did not include any heat source terms in
their formulation. Additionally, all geometries studied were in either 1-D or 2-D closed
domains with no treatment of the free surface where the adiabatic condition applies as in
laser heating of metal surfaces.
Hence for the state-of-the-art laser processing, there is a need to investigate the accuracy
of transient solutions of problems with rapid heating occurring with nanosecond pulses.
3.4 Summary
This chapter has reviewed the state-of-the-art for modelling laser processing and
welding. Mesh-based models failed to give all details on the characteristics of the
ejected and the remaining base materials within the same model. Therefore, new
methodologies are required that can progress beyond the limitations of the mesh-based
approaches. The meshless method SPH has the capability to simulate not only the
ejected material but also the interaction between the multiple phases in order to obtain
the closest representation of the real life application.
To date, no previous work has been reported on modelling the laser cleaning using a
meshless method only, which makes the SPH ablation model in this project to be the
first in this area. The new SPH work on laser welding is presented in Chapter 9.
The next chapter describes the materials and experimental equipment used in laser
welding and cleaning, as well as the materials characterisation procedure and
equipment.
93
4 CHAPTER 4
MATERIALS & EXPERIMENTAL SET-UP
This chapter presents a description of the experimental equipment and the materials
employed in the investigations conducted at the University of Manchester.
4.1 Materials and experimental set-up
AC170-PX (AA6014) Al sheets of 1.1 mm thickness were laser welded using different
filler wires with a diameter of 1.2 mm. The chemical compositions and physical
properties of these materials are given in Table 4.1 and Table 4.2 respectively. The
parent material has been tempered to T4, and its physical and mechanical properties
were measured according to EW 10002 in the transverse direction to rolling direction.
The base material was coated with titanium and zirconium (4 mg/m2) and lubricated
using a dry lubricant AlO70 (1.5 g/m2) to enhance formability and other performance
characteristics. The total coating thickness was calculated to be 0.503 μm since Ti and
Zr coatings together were about 3 nm in thickness and AlO70 lubricant has a thickness
of 0.5 μm.
Table 4.1 Chemical composition of the Al alloys (wt%) (Novelis Deutschland GmbH,
2011)
Si
Fe
max.
Cu
max.
Mn
Mg
Zn
max.
Ti
max.
Cr
max.
Others
each
max.
Others
Total max.
AC-170PX
(T4)
0.5-0.7
0.35
0.2
0.05-0.2
0.4-0.7
0.1
0.1
0.1
0.05
0.15
(AA4043)
5.2
0.5
0.1
0.05
0.05
0.1
0.1
-
0.05
0.15
Element
94
Table 4.2 Physical and mechanical properties of AC-170PX after tempering (Novelis
Deutschland GmbH, 2011)
Alloy
Density
[103 kg/m3]
Yield
strength
(0.2%)
[MPa]
Thermal
Conductivity
[W/m.K]
Electrical
Conductivity
[m/Ωmm2]
Coefficient of
Thermal Expansion
[10-6 K-1]
Rm
[MPa]
AC-170PX
(T4)
2.7
90
160-190
26-30
23.4
195
(AA4043)
2.68
40
170
24-32
22.1
160
4.2
Laser welding process
Two different types of joint configurations were evaluated in this work: fillet edge and
flange couch joints (shown in Figure 4.1). The joints were welded using a Trumpf disk
laser TruDisk 5302 (maximum output power of 5300W) with the beam properties given
in Table 4.3.
The laser welding was performed using a laser beam with a focal point of 8 mm below
the workpiece and shielded by Ar gas (99.9% purity) at a flow rate of 15 l/min. A spot
of 600 µm in diameter (on the workpiece surface) was used in the welding, and the
laser beam was focused using a lens with a focal length of 176 mm at 8 mm below the
workpiece surface. A 1.2 mm filler wire of different chemical composition (detailed in
the relevant chapters) was used with a 10 mm stick-out distance. The laser beam
properties are also presented in detail in the experimental results chapters.
For both types of joints, a 290 mm length seam was welded with a drag angle of 10 deg.
The lateral angles for the fillet edge and flange couch joints were 45 deg and 20 deg
respectively. Figure 4.2 illustrates the drag and the lateral angles relative to the normal
plane to the weld.
Weld
(a)
(b)
Figure 4.1 (a) Fillet edge joint (b) Flange couch joint with offset
The samples were welded at different parameters such as power, welding speed, wire
feed and they are shown in detail in each relevant chapter.
95
Drag
angle
Lateral
angle
Welding
Head
Welding
Head
Laser
Lateral
angle
Filler
45
 Welding Direction
Welding Direction
 Welding Direction
Figure 4.2 Incident beam and its inclination angles of the normal plane
4.3
Laser cleaning process
Before laser welding, some samples were cleaned using a Q-switched Nd:YAG laser
(cleanLASER CL600) and operated at the parameters shown in Table 4.3.
Table 4.3 Laser cleaning optical and process parameters (Sun and Karppi, 1996)
Nd:YAG CL600
Unit
Value
Wavelength
[nm]
1064
Operating power
[W]
600
Min. diameter laser light cable
[µm]
310
Spot diameter
[µm]
780
Pulse frequency
[kHz]
20
Pulse duration
[ns]
100
Scan frequency
[Hz]
180
Scan width
[mm]
20
Scanning speed
[mm/s]
95
Overlap in scan direction
[%]
53.8
Laser cleaning aimed to remove the contaminants and the oxide layer on the surface of
the sheets in order to reduce porosity in the welds.
96
4.4
Samples preparation
All welded samples were cut perpendicular to the welding direction, mounted in the
resin and then ground using silicon carbide abrasive papers at five stages of 80, 300,
600, 1200, 4000 grits using a rotating disk (Mecatech 334). After that, the specimens
were polished using 3 µm and 1 µm diamond pastes and then etched by immersing them
in sodium hydroxide solution (1g NaOH + 100 ml H2O) as an etchant for 45 seconds.
The macrostructure and microstructure of the samples were examined using a digital
microscope KEYNCE VHX-500F.
4.5
Scanning electron microscope (SEM):
SEM was used to analyse the laser cleaned surfaces in comparison with a reference
surface in order to determine any phase change at the cleaned surfaces. Hitachi High
Technologies, S-3400N Type I, 0-30 kV scanning electron microscope was used in this
project as shown in Figure 4.3.
Figure 4.3 Hitachi High Technologies S-3400N SEM
4.6
X-ray diffraction technique:
Philips X'pert -1 Cobalt and Copper XRD machines were used to analyse and determine
the phases exist in the weld and the parent material.
97
Figure 4.4 Philips X'pert -1 Copper XRD machine (University of Manchester, 2014)
4.7
Laser-induced breakdown spectroscopy (LIBS):
LIBSCAN 0165 was used for the surface chemical analysis before and after laser
cleaning. A 1064 nm laser with an average power of 2 W and peak power of 35 MW
was used in a pulsed mode with 3-10 ns pulses. Figure 4.5 shows the LIBS in the
Photon Science Institute at the University of Manchester.
Figure 4.5 LIBSCAN 0165 laser-induced breakdown spectroscopy
4.8
UV-VIS-IR spectrometer
The UV-VIS-IR spectrometer (Analytic Jena, SPECORD 250, Dual beam) was used to
determine the reflectivity of the Al alloys before and after laser cleaning. This piece of
equipment allows measuring reflectivity and absorptivity over the entire light spectrum.
98
Figure 4.6 Analytic Jena-SPECORD 250 Spectrometer
4.9
White light interferometer
A Wyko NT1100 (Analytik Jena Specord 250), white light Interferometer with a
wavelength range of 200 to 1100nm with less than 1 nm resolution, was used for surface
topography and for the surface roughness determination before and after cleaning.
Figure 4.7 Wyko NT1100 (Analytik Jena Specord 250) Interferometer
4.10 Summary
This chapter described the equipment and the experimental procedure utilised in the
experimental part of this project. Also, pictures with a full description of the materials
characterisation equipment were provided since they are not fully described in the
following chapters. The next chapter presents the outcome of an experimental study on
99
the effect of heat input and sheet gap on porosity generation in laser welding of AC170PX alloy with two weld configurations.
100
 Brief Introduction
After reviewing the state-of-the-art in laser welding of Al alloys, it has been noted, to
the best of the author’s knowledge that no previous studies have been carried out on
porosity reduction in laser welding of AC-170PX (AA6014) alloys for the fillet edge
and flange couch joints. Therefore, a number of experiments were conducted to study
the effect of the heat input on porosity during laser welding of such alloy using the
recommended and commercially available filler wire AA4043.
Moreover, the research was taken one step further by exploring the response of porosity
content to the introduction of a 0.2 mm gap between the sheets at different levels of
laser powers and welding speeds. The selection of the gap size was based on the
industrial practice and the available literature on laser welding of zinc coated steels such
as (Gualini, 2001). The gap can be achieved using accurately machined fixtures to
maintain a fixed separating distance between the welded sheets. Although it can be a
challenge to achieve this tight fit-up tolerance for large and 3-D complex panels, it is
still important to investigate the effect of the gap on porosity reduction.
The study presented in the next chapter fills in the knowledge gap identified in the
literature review and identifies the baseline for the following investigations presented in
Chapter 6, 7 and 8 of this thesis.
101
Publication Title:
Understanding the Effect of Heat Input and Sheet Gap on Porosity
Formation in Fillet Edge and Flange Couch Laser Welding of
AC-170PX Aluminium Alloy for Automotive Component
Manufacture
Authors:
A.W. AlShaer, L. Li and A. Mistry
Journal:
ASME Journal of Manufacturing Science and Engineering
Year, Volume, Issue and Pages: 2015, 137 (2), 1-13
Publication: No. 021011-1
DOI: 10.1115/1.4028900
STATUS: PUBLISHED
102
5 CHAPTER 5
UNDERSTANDING THE EFFECT OF HEAT
INPUT AND SHEET GAP ON POROSITY
FORMATION IN LASER FILLET EDGE
AND FLANGE COUCH WELDING OF AC170PX ALUMINIUM ALLOY FOR
AUTOMOTIVE COMPONENT
MANUFACTURE
5.1 Abstract
An investigation is reported on the characteristics of porosity formation in high power
disk laser welding of AC-170PX (AA6014) alloy sheets (coated with titanium and
zirconium) in two weld joint configurations: fillet edge and flange couch with AA4043
filler wire for potential automotive manufacturing applications. Porosity, macro- and
microstructure characteristics, tensile strengths, microhardness, and joint geometry were
investigated. It has been found that an increase in heat input and welding speed
generates more porosity in both types of joints. The introduction of a 0.2 mm gap
reduces porosity significantly in the fillet edge joints, but it does not have a noticeable
effect on the flange couch joints. The mechanism of the porosity formation is discussed.
Keywords: Laser welding, fillet edge joint, flange couch joint, aluminium alloy,
porosity, joint geometry, car, automotive.
103
5.2
Introduction
In the past few decades, the use of lightweight alloys has been growing at an
accelerating rate in many industrial sectors such as aerospace, rail vehicles, and
automotive industries. Al alloys allow the manufacture of high strength and lightweight
structures that can reduce the weight of the vehicles by up to 50% compared with
conventional steel structures in automotive manufacture (Rapp et al., 1995). Although
the use of Al alloys in design allows manufacturers to reduce the weight of the
structures, welding Al alloys is a real challenge since a significant amount of porosity is
usually observed during the welding process. Porosity not only forms a serious defect in
the weld joints but also stimulates other defects such as corrosion and fatigue failure.
For instance, the aggregation of close pores may develop to form micro-cracks in the
weld joints. Moreover, hydrogen trapped inside the voids increases the susceptibility to
the cold crack formation. Porosity that is located in the weld root decreases bending
strength of the weld because it acts as nucleation centres for cracks formation and
propagation (Ion, 2005b). In addition to porosity formation and cold cracking that is
associated with Al welding, hot cracks are also a very common defect that has been
observed in Al alloys welding (Katayama et al., 1998).
Porosity formation mechanism has been disputable due to the fact that different
researchers investigated different materials and different processing conditions with
different porosity formation origins. Katayama et al. (2007a) explained porosity
formation mechanism at various welding speeds during fibre laser welding of 304
stainless steel. They concluded that pores are generated at lower speeds due to the
keyhole collapse, and they are trapped in the weld due to the turbulent flow of the
material at the bottom of the welding pool. However, porosity is suppressed at higher
speeds due to the slow movement of the molten material in the weld pool. Other
investigations focused on observation of porosity and its movement inside the weld
pool. Arata (1980) and Antonov et al. (1977) observed bubble generation during glass
and transparent medium welding respectively. Similar to their research, Berger et al.
(2011) used ice water as a transparent material to observe porosity during CO2 laser
welding. Their results showed that pores generated in ice water were filled with water
vapour or air or a mixture of them. The authors concluded that porosity generation at the
104
capillary tip is very difficult to explain and welding speed has a significant impact on
pores formation. Despite the fact that the use of transparent materials enabled the
authors to visualize porosity formation in the welds, ice water and glass behave
differently from metals (due to their different physical, chemical, and mechanical
properties) during laser welding process; therefore, more research on porosity formation
in laser welding metal is required since the formation mechanism is still not fully
understood.
Alfieri et al. (2011) studied porosity content in AA2024 bead-on-plate (BOP) Al welds
during disk laser welding. They concluded that the use of high power to speed ratio for
this type of joints and materials reduces porosity significantly since it forms more stable
keyhole and achieves full penetration in butt and BOP welds.
AA6013 alloy welding was investigated to determine the optimum welding parameters
for 1 kW fibre laser welding. After conducting different experiments, the optimum laser
power and welding speed at which sound weld can be achieved were found to be 1 KW
and 5 m/min (Vilar et al., 2008). The AA6013 alloy was also welded using a YAG laser
and using different filler materials namely: AlSi12, AlSi12Mg5, AlSi10Mg, AlMgSi1
and AlMg4,5Mn. Reinhold et al. (2002) measured porosity percentages in these welds
and attributed the increase in porosity to the increase in Mg content in the filler material
since magnesium eases hydrogen solution in Al and that increases porosity content
inside the welds. An investigation on AA6056 by Pakdil et al. (2011) focused on fatigue
properties when it is laser welded using AA4047 filler wire. Pores were found to slow
cracks propagation in relation with the value R (stress ratio). Moreover, they found that
crack propagation in the direction perpendicular to rolling direction was slower than that
correspondent to the rolling direction.
Yu et al. (2010) showed that pore generation increased due to the use of filler wire
during laser welding of 5A06 Al alloy. This is because the liquid metal from the wire
disturbs the stability of the keyhole and promotes its oscillation. To prevent this, they
introduced a gap between the sheets in butt joints and found that porosity can be
reduced significantly because the pre-existing gap joins the keyhole and makes it larger
and more stable, and it provides an escape for the gases from the fusion zone (FZ) (Yu
et al., 2010). It is worth mentioning that porosity was also detected in YAG laser
105
welding of 5A90 alloy (an Al–Li alloy) in which hydrogen and keyhole porosities were
observed. To reduce porosity, pre-treatment techniques such as re-melting the surface or
hot isostatic pressing can be used for this type of the welds (Chen and Gong, 2011).
A number of investigations were reported on porosity elimination in laser welding using
various methods such as application of a magnetic field on the weld pool (Schneider et
al., 2013; Bachmann et al., 2013; Zhou and Tsai, 2007a), the use of a modulated beam
(Matsunawa et al., 2003), surface preparation (Katayama et al., 1998), the use of hybrid
laser-MIG or laser-TIG welding techniques(Yan et al., 2009), and the application of
dual spots on the surface (Blackburn et al., 2009, 2010b; Blackburn et al., 2012). Only a
few investigations drew attention to porosity reduction by introducing a gap between the
welded sheets. Some experiments were conducted to determine the optimum gap size at
which porosity is eliminated during hybrid Nd-YAG laser-MIG welding of a similar
AA5083 Al joint (Andersen and Jensen, 2001), and a dissimilar AA6016-Steel joint
(Thomy and Vollertsen, 2009). Both experiments focused only on the butt joints that
can be used in automotive or in aerospace industries. The studies on the effect of the
gap size on porosity formation and weldability during laser welding covered a number
of different materials such as: titanium (Kawahito et al., 2007), AA5754 Al alloys (Kuo
and Lin, 2006b), and dissimilar Al/steel welds (Chen et al., 2011).
So far, there has been no reported work by other researchers on laser welding of AC170PX (Al-Mg-Si) alloy with a filler wire, despite the fact that this material is widely
used in the automotive industry due to its unique properties such as good formability
and corrosion resistance.
In this work, an investigation is reported on the effect of laser welding parameters and
joint gap on porosity in the weld joints in laser welding of AC-170PX (AA6014) Al
alloy with an AlSi5 (AA4043) filler wire. Two different joint geometries were
evaluated: fillet edge and flange couch joints as shown in Figure 5.1.
106
Weld
Figure 5.1 (a) Fillet edge joint (b) Flange couch joint with offset
5.3
Materials and experimental procedure
The chemical compositions of the parent material AC-170 PX (AA6014-T4) and the
filler wire AA4043 are given in Table 5.1. The parent alloy was tempered to T4, and its
properties were measured according to EW 10002 transversely to the rolling direction.
Table 5.2 gives the physical and mechanical properties of AC-170PX after tempering.
The parent material is coated with titanium and zirconium (4 mg/m2) to enhance
weldability and other performance characteristics and was lubricated using a dry
lubricant AlO70 (1.5 g/m2). Two sheets of 1.1 mm in thickness were welded using the
filler wire with a diameter of 1.2 mm.
Table 5.1 Chemical composition of the Al alloys (wt%) (Novelis Deutschland GmbH,
2011; Anon, 2011b)
Element
Si
Fe
max.
Cu
max.
Mn
Mg
Zn
max.
Ti
max.
Cr
max.
Others
each
max.
Others
Total
max.
AC-170PX
(T4)
0.5-0.7
0.35
0.2
0.05-0.2
0.4-0.7
0.1
0.1
0.1
0.05
0.15
(AA4043)
5.2
0.5
0.1
0.05
0.05
0.1
0.1
-
0.05
0.15
Table 5.2 Physical and mechanical properties of AC-170PX after tempering (Anon,
2011b; Novelis Deutschland GmbH, 2011)
Alloy
Density
[103 kg/m3]
Yield
strength
(0.2%)
[MPa]
Thermal
Conductivity
[W/m.K]
Electrical
Conductivity
[m/Ωmm2]
Coefficient of
Thermal
Expansion
[10-6 K-1]
Rm
[MPa]
AC-170PX
(T4)
2.7
90
160-190
26-30
23.4
195
(AA4043)
2.68
40
170
24-32
22.1
160
107
The materials were welded using a TRUMPF disk laser TruDisk 5302 (max power of
5300 W) with beam properties given in Table 5.3 and using a Scansonic ALO3 laser
welding head (see Figure 5.2). Welding was performed using a lens with a focal length
of 176 mm that focuses the laser beam at 8 mm below the target surface, producing a
600 µm spot at the specimen surface. The weld zone was shielded by Argon gas with a
flow rate of 15 l/min. For both types of joints, a length of 290 mm was welded with a
drag angle of 10 deg. The lateral angles for fillet edge and flange couch joints were
45 deg and 22 deg, respectively. Such angles were used to prevent the beam interaction
with plasma (Narikiyo et al., 1996). Moreover, the inclined beam should be used during
laser welding of Al to prevent the beam reflection and damage to the optics. Figure 5.3
illustrates the inclination angles of the laser beam during welding. A KUKA 6-axis
robot was used to control the motion of the laser head.
Table 5.3 TRUMPF TruDisk 5302 laser unit technical Properties (Trumpf-Lasers,
2013)
TruDisk 5302
Unit
Value
Wavelength
[nm]
1030
Maximum Laser power
[W]
5300
[mm.mrad]
8
Min. diameter laser light cable
[µm]
200
Power stability at nominal power
[%]
±1
Cooling water temperature range
[oC]
5 - 20
Beam quality
108
Figure 5.2 Schematic of the Scansonic ALO3 laser welding head
Drag
angle
Welding
Head
Welding
Head
Lateral
angle
Laser beam
Lateral
angle

Welding Direction
 Welding Direction
Filler wire
45
Welding Direction
Figure 5.3 Incident beam and its inclination angles of the normal plane
The welding parameters, as well as the line energy input for each type of configuration,
are presented in Table 5.4. The line energy input was calculated as the ratio of the laser
power to welding speed; therefore, it can be seen that the line energies varied according
to the selected power and speed for each type of joint.
109
Table 5.4 Welding parameters and line energy (J/mm) for fillet edge and flange couch
joints
Fillet Edge Joint
Welding
Wire
speed
feed
[mm/s]
[m/min]
20
1.2
35
2.1
50
3.5
Flange Couch Joint
Laser Power [W]
2000
3500
5300
Welding
Wire
speed
feed
[mm/s]
[m/min]
20
1.9
35
3.2
50
4.2
100
100
106
Laser Power [W]
2200
3800
4400
110
108.5
88
Two different gap sizes (zero-gap and 0.2 mm gap) were applied between the sheets, in
order to observe their effects on the weld properties besides the influence of the heat
input on the weld characteristics.
The size of the gap used in this study was chosen according to the industrial
requirements and due to joint dimensional limitations. Moreover, this value (i.e.
0.2 mm) was not far from the value selected by other researchers such as (Harooni et al.,
2012) who investigated the effect of the root gap on the shape of ZEK100 Magnesium
alloy sheets in lap joint configuration. They studied different gap sizes ranging from
0.0 mm to 0.2 mm with 0.05 mm intervals. They concluded that increasing the gap size
produces a “bulged-shape” weld as the molten material tends to fill in the gap at the
interface. Meng et al. (2013) concluded that the use of a gap less than 0.4 mm helps to
control the weld quality during CO2 laser welding of HSLA steel with 4 mm in
thickness. Additionally, the value 0.2 mm falls within the limits specified by the
standards EN 30042:1994 and BS EN ISO 13919−2:2001 which specify a maximum
value for the gap to be 0.57 mm (calculated) for 1.1 mm sheets.
After the welding had been completed, the samples were sectioned perpendicular to the
welding direction, and then ground using silicon carbide abrasive papers. Grinding was
performed in five stages using 80, 300, 600, 1200, and 4000 grit papers using a
polishing disk (Mecatech 334). After that, the samples were polished using 3 μm and
1 μm diamond pastes and then immersed in sodium hydroxide solution (1g
110
NaOH + 100 ml H2O) as an etchant for 45 sec (ASM International, 1993) in order to
reveal microstructures. The microstructure, porosity, and the FZ geometry of the welded
samples were examined using a digital optical microscope KEYNCE VHX-500F. Three
samples were used for generating the error values for each set of experiment and the
measurements of porosity level, tensile strength, weld dimensions, and microhardness
were repeated three times to guarantee reliable data and to evaluate the error.
Microhardness profiles were examined using 0.3 kgf (2.942 N) load and a test duration
of 10 sec. Tensile tests were performed using Galbadini Quasar 100 kN machine and the
ultimate tensile strength (UTS) was measured for all samples (prepared to ISO 68921:2009). X-ray diffraction (XRD) analysis was carried out to identify the phases in the
joints.
5.4
5.4.1
Results and Discussion
Macrostructure and microstructure characteristics
The weld cross section macrostructures were examined using low magnification (×50)
optical microscopy, in which larger pores and cracks can be clearly observed. Figure 5.4
and Figure 5.5 give typical weld joint cross-section characteristics of the fillet edge and
flange couch samples at different welding parameters.
From Figure 5.4, it can be seen that for the fillet edge joints, as the laser power increases
while increasing the welding speed, the amount of porosity increases and the welding
penetration increases despite the fact that the line energy was increased very slightly. In
addition, microcracks had developed in certain welds using certain welding parameter
range. At low laser power and welding speeds, the welds suffered from lack of fusion in
the root of the weld. Cracks and some large pores can be seen located mostly at the
welds root.
For the flange couch joints, on the other hand, porosity is significantly less and
microcracks were not found in most welds (Figure 5.5). Only at the highest laser power
and welding speed, very small amount of porosity was found. However, penetration
depth decreased gradually due to the decline in line energy as shown in Table 4.4.
111
Fillet Edge
Joints
No Gap
Welding
Speed
0.2mm Gap
20
2
35
3.5
Power
[kW]
[mm/s]
50
5.3
Figure 5.4 Macro-sections of the fillet edge joints at various welding parameters where
the white spots are porosity
Using a higher magnification (×400), microstructures can be clearly observed. Figure
5.6 shows the microstructure in the fillet edge weld FZ, heat affected zone (HAZ), and
the parent material at 2000 W and 20 mm/s. In the parent AC-170PX material, dark
spots were found distributed over the entire area in the parent material. The plane area is
mainly the Al. By moving to the left of this region, a new area covered with dendritic
Al-Mg2Si begins. This area is the HAZ which is full of elongated Al-Mg2Si dendrites
that are distributed on the grains boundaries.
112
Flange couch
joints
No Gap
0.2mm Gap
20
2.2
35
3.8
50
4.4
Welding
Speed
Power
[mm/s]
Figure 5.5 Macro-sections of the flange couch joints at various welding parameters
The AC-170PX alloy contains many alloying elements that can form precipitates.
However, due to the high content of Mg and Si, magnesium silicide forms and
precipitates in both the parent material and the HAZ. This compound improves strength
and corrosion resistance of the material, and this is the main reason for selecting such an
alloy in the automotive industry. The third region shown in Figure 5.6 is the FZ, in this
area a eutectic of Al and silicon forms at the Al grain boundaries. Al-Si eutectic
becomes the dominant phase in this region due to the very high silicon content that is
originated from the AlSi5 filler wire and from the silicon content in the parent material.
The magnesium content in the filler wire (as shown in Table 5.1) is very low. Therefore,
Mg2Si does not have the ability to form as a precipitate in the FZ.
Figure 5.7 shows the microstructure of the flange couch joint and similar phases to that
observed in the fillet edge joint was detected in the flange couch as well. This can be
113
[kW]
understood since the same base material, and filler wires were used to create these
joints. As previously indicated, the eutectic Al-Si tends to form on the grains boundaries
in the FZ while Al-Mg2Si dendrites are formed in a uniform direction in the HAZ as
shown in Figure 5.7.
Elongated
grains
Figure 5.6 Microstructure of the weld-HAZ-Base material of a fillet edge joint welded
using 2000W and 20 mm/s
Phases that appear in Figure 5.6 and Figure 5.7 were confirmed using the XRD analysis
given in Figure 5.8, which have identified Al-Si in the weld zone and Al-Mg2-Si in the
HAZs.
Elongated
grains
Figure 5.7 Microstructure of the weld-HAZ-Base material of a flange couch joint
welded using 3800 W and 35 mm/s
114
Al-
AlAl-
Mg2S
SiO2
Figure 5.8 X-ray diffraction test results for the AC170-PX (AA6014) joints welded
using 4043 filler wire (phases detected are Al-Si , Mg2Si , SiO2)
5.4.2
Porosity characteristics
For both types of joints, welding parameters (laser power and welding speed) were
changed simultaneously in order to maintain an approximate range of line energy
values. According to Figure 5.9 and Figure 5.10, all samples suffered from porosity
formation but with different proportions. Generally speaking, pores may form due to
many factors (Couso and Gómez, 2012):
1. Hydrogen that can be dissolved in Al, since Al can dissolve about 70 times more
hydrogen at melting temperature than it does at low temperatures.
2. The lubricant film that is applied to the parent material, oxide films, hydrogen
solved in filler wires, contaminations, the moisture in the environment, or in the
shielding gas.
3. The collapse of the unstable keyhole.
4. Vaporisation of some of the alloying materials especially those which have
lower melting temperature than Al.
115
Porosity characteristics were determined according to the ratio of the area that the pores
occupy in the weld zones to the area of the weld as well as the number of pores and
their diameters (Figure 5.9 to Figure 5.12). The gap configurations for the two types of
joints are illustrated in Figure 5.13 and Figure 5.14.
From the results, it can be seen that for the fillet edge joints, increasing laser power and
welding speed increased the porosity significantly and the presence of a gap would
reduce porosity. At the 5 kW laser power without a gap, the porosity level can reach up
to 85% from less than 20% at lower powers despite that the line energy is slightly
changed from the case with lower laser power. At the 2 kW laser power with a 0.2 mm
gap, porosity can be reduced to less than 5%. However, even with a gap of 0.2 mm, at
the 5 kW laser power and welding at high speed, the porosity can reach up to 40%. The
pore sizes are generally between 20 μm and 140 μm with the majority of the pores
between 20 and 50 μm.
(a)
(b)
Figure 5.9 Porosity percentages for fillet edge joints (a) without gaps and (b) with a
0.2mm gap
116
(a)
(b)
Figure 5.10 Porosity percentages for flange couch joints (a) without gaps and (b) with a
0.2 mm gap
Despite the fact that the line energy values for the three parameter sets were
approximately the same as shown in Table 5.4, porosity levels were significantly
different in each case. This can be attributed to the increased welding pool turbulence
and shorter melting and solidification time of the molten alloys as welding speed is
increased. Zhou and Tsai (2007b) created a mathematical model to simulate laser
welding of 304S steel. Their results show that porosity can be formed due to the
collapse of the keyhole and the slow backfilling speed of the material and the high
solidification rate. As welding speed increases, the material will not have enough time
to backfill the keyhole during welding. Moreover, increasing welding speed stimulates
the turbulent flow of the material inside the welding pool. Taking into account the low
viscosity of Al in our case, turbulence will be greater than that shown in Zhou and Tsai's
model for 304S steel (Mäkikangas et al., 2007). According to that, increasing welding
speed will enhance porosity formation in laser welding of the Al alloy if no other
measurements were taken to avoid this trend.
Many investigations indicated that the application of a 0.1 mm to 0.2 mm gap during
laser welding of zinc-coated steel could eliminate porosity inside the weld. Researchers
related this to the fact that zinc vapour which originates from the coating can escape
easily through the gap applied between the sheets, reducing porosity and eliminating
other weld defects such as undercuts and lack of material (Mäkikangas et al., 2007).
According to Figure 5.9, fillet edge samples with a 0.2 mm gap show a significant
reduction in porosity proportion in comparison with fillet edge welds without gaps. This
117
decrease in porosity can be explained by the presence of the gap that acts as a canal that
helps gases to escape.
For the flange couch welds (shown in Figure 5.10), the porosity levels are much reduced
to generally less than 5% and the effect of laser power is negligible without a gap.
Increasing the laser power has been found to increase the porosity when a 0.2 mm gap
was introduced. The pore sizes are also generally much smaller, typically in the range of
10–30 μm in the flange couch welds compared with those of fillet edge welds.
For the flange couch joints with a gap, the increase in power increases porosity in
comparison with the negligible effect of power over pore generation when there is no
gap. This can be explained since deeper penetration and narrower weld bead was
generated when the gap was introduced. This means that higher depth-to-width ratio has
been achieved and it has been found that porosity is dependent on the depth-to-width
ratio; higher ratio would lead to higher porosity in the weld (Zhou and Tsai, 2007b).
According to this, increasing the laser power and the welding speed in the weld leads to
a higher aspect ratio that promotes porosity formation. However, pores find an easy way
to escape from welds with low aspect ratio especially with the existence of the small
room under the flange weld due to the weld configuration.
As aforementioned, pores are formed when gases are entrapped inside the welding pool
or due to the keyhole collapse. Hydrogen usually originates from moisture at the weld
surface and dissolves in the molten Al, forming porosity. Moreover, H2 can be
generated from contaminants on the surface, the dry lubrication used on the sheets, or
from water vapour condensed inside the shielding gas delivery system (the inner walls
of the inert gas cylinders and tubes).
118
Figure 5.11 Porosity distribution curve for a fillet edge joint welded at 5300 W and 50
mm/s
Figure 5.12 Porosity distribution curve for a flange couch joint welded at 2200 W and
20 mm/s
Moisture is found to be a major cause for increasing hydrogen content in the weld zone
because molten Al reacts easily with water vapour and releases hydrogen gas according
to the following reaction (ASM International, 1993):
2 Al  3H 2O  Al2O3  3H 2
( 5-1 )
In addition, many investigations were conducted to evaluate hydrogen solubility in pure
Al. Ransley and Neufeld (1989) formulated two equations to calculate hydrogen
solubility in Al:
For liquid Al:
119
 2760  1
log L   
  log P  1.356
 T  2
( 5-2 )
For solid Al:
 2080  1
log L   
  log P  0.652
 T  2
( 5-3 )
where L is the solubility of hydrogen (mL/100 g Al) measured at 273 K and 760 Torr, T
is the temperature (K), and P is the partial pressure of hydrogen (1 Torr =0.133 mbar).
The calculated values of hydrogen solubility at different temperatures indicated that
hydrogen solubility increases by 20 times in Al as the material transferred from solid to
liquid state. Additionally, hydrogen solubility can be promoted due to the high
temperature of plasma above the weld pool that generates atomic hydrogen (Ransley
and Neufeld, 1989). Atomic hydrogen reacts much easier with Al than molecular
hydrogen.
As aforementioned, porosity flows inside the weld due to the material flow and tends to
move up to the weld surface by buoyancy. The rising speed u depends on, according to
Stokes’ law, both viscosity of the liquid and the radius of the pore as shown in (Zhao et
al., 1999b):
u
2r 2 . .g
9
( 5-4 )
where r is the radius of the pore, Δρ is the difference in density between the liquid and
the gas inside the pore, g is gravitational acceleration, and µ is the dynamic viscosity of
the liquid.
The density of the liquid 6000 alloy is not only different from solid Al but also changes
with temperature. This change can be calculated from the following equation (Mills,
2002):
l (kg.m3 )  2415  0.28(T  624o C)
( 5-5 )
120
l
at 700o C
 2415  0.28(700  642)  2398.7kg.m3
( 5-6 )
The density of hydrogen is 0.089 kg/m3 (Wiberg and Wiberg, 2001), the dynamic
viscosity of Al at 700 oC is 2.96 mPa.s (Brandes and Brook, 1992), g is 9.81 m/s2.
By applying the above equations for liquid Al at temperature 700 °C and assuming a
pore radius of 50 μm, the rising velocity is found to be approximately 4.4 mm/s. For
bigger pores of radius 100 μm, velocity value becomes 17.6 mm/s. From these
calculated values, it can be seen that pores are moving up to the surface at a very slow
speed relative to the solidification rate associated with laser welding (solidification time
for Al laser welding is about 8 μs (Kim et al., 2004)). Therefore, pores will not have
enough time to escape from the weld pool before solidification takes place. This case is
more pronounced in the fillet edge joints since bubbles have a very long path to the
surface. However, some pores generated in the flange couch joints do not have to travel
a long distance as they can escape from the space associated with the joint
configuration. Moreover, this may explain the reduction in porosity when applying a
gap between the sheets for both types of joints.
It is worth mentioning that the collapse of the keyhole and its instability is the other
major factor behind porosity formation. The weld pool during Al welding is usually
deeper than that in steel due to lower Prandtl and Stefan numbers (Wei et al., 2012).
Thus, the pores have to travel a longer distance to exit the Al weld pool if compared to
other metals. In comparison with magnesium alloys, Al alloys have greater surface
tension, higher boiling temperature, and lower equilibrium vapour pressure. As a result,
these alloys experience keyhole instability due to evaporation of alloying elements such
as magnesium and zinc. This evaporation causes a drop in the vapour pressure inside the
keyhole and stimulates the collapse of the keyhole during welding (Zhao and DebRoy,
2011). Many investigations were carried out to estimate the keyhole collapse time. Kim
et al. (2004) estimated the time of the keyhole collapse in AA5083 laser welding to be
0.6 µs, while the solidification time was much larger 8 µs. This means that keyhole may
collapse many times before solidification occurs.
121
Ducharme et al. (1993) attribute the keyhole collapse to the fact that that the surface
tension at keyhole walls fails to balance the vapour pressure inside the keyhole. This
causes the material to flow to the inside of the keyhole. Simulation results show that the
keyhole collapse occurs faster at the middle part of the keyhole wall than on the top and
the bottom, which is believed to be the cause for gas entrapment at the lower part of the
keyhole (Ducharme et al., 1993).
Schauer and Giedt (1978) discussed the instability of the keyhole in electron beam
welding (EBW). Although this method of welding is different from laser welding,
keyholes form in a similar manner, and their collapse is a common problem in EBW.
The authors calculated the surface tension and vapour pressure forces at different
temperatures and heights at the keyhole walls. Their results show that the vapour
pressure force is less than the surface tension force at the upper region of the keyhole.
Conversely, the pressure force becomes greater than the surface tension at the bottom of
the keyhole while they balance each other at the middle section. Therefore, molten
material starts to flow from the top of the keyhole and aggregate at the middle section.
However, molten metal cannot flow to the bottom due to the high vapour pressure, and
this causes the vapour to be trapped in the inside and porosity forms at the bottom of the
keyhole.
5.4.3
Weld penetration and fusion zone geometry characteristics
The weld penetration depth and weld width were measured for both types of joints and
are presented in Figure 5.15 and Figure 5.16 for the fillet edge joints and flange couch
joints, respectively. Figure 5.13 and Figure 5.14 illustrate these dimensions on the
samples configurations.
(a)
(b)
Figure 5.13 Fillet edge weld dimensions (t1 and t2 sheets’ thicknesses, s the gap size, Pw
weld width, and Pd penetration depth
It is known that increasing welding speed in laser welding leads to a reduction in the
heat input and an increase in the cooling rate. This results in shallower penetration and
122
narrower weld beads (Kutsuna et al., 2006). Consequently, in order for welding to
achieve deeper penetration and wider welds, an increase in laser power is expected to
maintain constant line energy.
(a)
(b)
Figure 5.14 Flange couch weld dimensions (t1 and t2 sheets’ thicknesses, s the gap size,
Pw weld width, and Pd penetration depth)
In Figure 5.15, the dimensions of fillet edge joints are depicted for welds with and
without a gap. From this figure, it can be seen that as the laser power increases (the
welding speed increases as well) the weld penetration depth and width generally
increases.
However, it should be noted that penetration depth and weld width have lower values
for fillet joints with 0.2 mm gaps than the welds without gaps. This is due to the fact
that the molten material tends to fill in the space between the sheets since it is easier to
flow through the gap.
Figure 5.15 Penetration depths and weld width values for the fillet edge joints
123
Figure 5.16 Penetration depths and weld width values for the flange couch joints
From Figure 5.16, the decrease in line energy for the flange couch joints generally led to
a gradual decrease in the weld dimensions for all joints. This trend is not strictly true for
the fillet edge joints. This can be attributed to the space already existed between the
sheets according to the weld configuration that restricted the heat flow.
It is worth mentioning that weld dimensions in the couch joints with 0.2 mm gap
followed a similar trend to the joints without a gap, this is justifiable since the line
energy was reduced.
It can be concluded that weld dimensions have a proportional relation with the line
energy. When the line energy increases, weld dimensions increase and vice versa. This
trend was described in (Steen and Mazumder, 2010):
0.483P(1  rf )  v.w.g..c pTm
( 5-7 )
where P is the laser power, rf is the reflectivity of the surface, v is the welding
speed, w is the weld width, g is the penetration depth, ρ is density, cp the heat capacity
of the material, and Tm is melting temperature of the material. This equation has not
considered the effect of thermal conduction and convention. The experimental findings
reported in this paper show that higher laser power and higher welding speed would
increase the weld FZ volume and penetration depth (Figure 5.15) for the fillet edge
welds.
124
5.4.4
Tensile strengths and micro-hardness
The UTS for both types of joints is given in Figure 5.17 and Figure 5.18. Since the
tensile strength of the weld is usually compared with the UTS of the parent material, the
UTS values of the weld can be given as a proportion of the UTS of the parent material.
These are shown in Figure 5.17(b) and Figure 5.18(b).
From Figure 5.17, it can be seen that, for the fillet edge welds, tensile strengths reduce
despite the increase in the heat input. The tensile strengths are much lower than the
parent material at typically 38–55% of the parent material UTS. The introduction of a
gap increases the tensile strength slightly. For the flange couch joints, UTS values
decreased from 60% of the parent material's UTS at the highest line energies to about
50% when the line energy was decreased from 110 J/mm to 88 J/mm. This reduction in
the strength is mainly caused by the increase in porosity, cracks and other defects in the
weld.
(a)
(b)
Figure 5.17 (a) Ultimate tensile strength values for fillet edge joints (b) UTS values as
a percentage of the base materials strength 195,000 kN/m2
Although a reduction in the strength was observed in both type of joints compared with
that of the parent material, flange couch joints achieved higher strength (about 62% of
the parent material strength) than the fillet edge joints (only 55% of the parent material
strength) when the higher line energy was used 110 J/mm in comparison with value
used for the fillet edge joint 100 J/mm. This can be attributed to the large proportion of
125
porosity in the fillet edge joints compared to the slight difference in porosity proportions
in the flange couch joints.
(a)
(b)
Figure 5.18 (a) Ultimate tensile strength values for Flange couch joints (b) UTS values
as a percentage of the base materials strength 195,000 kN/m2
5.4.5
Microhardness
The microhardness profile of the fillet edge profile is shown in Figure 5.19, from which
it can be seen that the microhardness increases slightly (by up to 20%) in the weld FZs
in comparison with the parent material. However, softer zones were observed in the
HAZs.
The AlSi5 (A4043) filler wire was used because it contains high levels of silicon to
reduce the susceptibility of the weld to hot cracking. Welding a 6000 series Al alloy
with 4043 filler (contains about 5% silicon) produces a weld with hypo eutectic Al–
silicon alloy with a dendritic magnesium silicide Mg2Si surrounding the grains (ASM
International, 1993; Pinto et al., 2010).
From Figure 5.19, it can be seen that the microhardness starts to increase when leaving
the base material and entering the HAZ. This can be explained by the concentrated
presence of the precipitate Mg2Si which takes a dendritic form in this region of the
weld. Microhardness starts to drop by about 10–15 HV0.3 when moving from the HAZ
to the centre of the weld. This decrease in hardness is due to the lower Mg content in the
FZ that is filled mainly by the 4043 filler wire. FZ contains more solid solution Al–Si
126
and much less magnesium silicide than the HAZ; this can explain the drop in hardness
in FZ in comparison with the HAZ. It can be seen from Figure 5.19 that there is some
fluctuation in hardness value in the FZ. This can be attributed to two reasons: first, FZ is
formed from a mixture of the two 6000 and 4000 alloys, and it is difficult to achieve a
homogeneous medium that can obtain approximate microhardness values; second, FZ
contains some micropores that may cause the microhardness to drop sharply at certain
points.
Figure 5.19 Microhardness profile for a fillet edge joint welded at 3500W and 35 mm/s
Figure 5.20 shows the microhardness profile for the flange couch joint. The joint was
observed under the microscope with lower magnification, and it was rotated in 45 deg
so that the entire profile can be seen in the picture. The indentations were numbered,
and the value of the hardness for each point is depicted in the same figure. It can be seen
from the microhardness profile that the microhardness values rises from about 64 HV in
the parent material up to 74.5 HV in the HAZ. The hardness then starts to drop to reach
58.7 HV in the FZ of the weld. The microhardness then increases again in the HAZ and
then stabilises in the parent material at approximately 65 HV.
127
This trend in the microhardness change is similar to that observed for the fillet edge
joint. However, it can be noted that the change in the hardness values is smaller in the
flange couch joint to that in the fillet edge. This may be attributed to the existence of the
small gap behind the weld (which is due to the configuration of the flange joint) that
maintains some of the heat during welding. The heat maintained in this space relieves
some of the internal stresses between the grains and hence reduces the microhardness.
80
5
75
70
65
HV 0.3
6
4
1 2
3
8 9
7
11 12 13
10
18
15
14
16 17
23
19 20
60
24 25 26 27 28
21 22
55
50
45
40
15 14 13 12
21
22
17
18
19
20
16
11
10 9
8
7 6
5
4
3
24 23
2
25
26
28
1
27
Figure 5.20 Microhardness profile for a flange couch joint welded at 3800W and
35mm/s
128
5.5
Summary
The work has shown the effect of laser welding parameters and joint gap on the porosity
and weld FZ geometry formation. The following conclusions can be drawn:
1. Porosity in fillet edge joints increased from about 20% to about 85% when the
laser power and welding speed were increased simultaneously from 2 kW to
5.3 kW and from 20 mm/s to 50 mm/s, respectively (line energy was increased
from 100 J/mm to 106 J/mm). However, porosity is far less affected by the laser
power and speed changes for flange couch joints. Porosity is increased by only
4% when the laser power and speed were increased. The application of a 0.2 mm
gap between the welded sheets led to a significant drop in porosity for the fillet
welds. In contrast, the gap did not show any effects on porosity for flange couch
joints. It has been found that the application of a 0.2 mm gap plays an important
role in porosity reduction in fillet edge joint because it acts as a canal for gases
to escape from the weld.
2. Increasing welding speed has been found to have a significant effect on porosity
formation because it stimulates the turbulent flow of the molten material inside
the welding pool. This has been observed despite the slight increase in line
energy for the fillet edge joint and despite the decrease in line energy for the
flange couch joint.
3. Weld joint dimensions such as weld width and penetration depth for the two
types of joints change proportionally with the line energy. Moreover, the 0.2 mm
gap led to a drop in weld dimensions in fillet welds in comparison with fillet
welds with zero gap. On the other hand, the application of a gap did not have a
significant influence on weld dimensions for the flange couch joints.
4. Tensile strength (UTS) of the weld reduced in fillet joints by about 34% despite
the increase in line energy from 100 J/mm to 106 J/mm due to the high increase
in porosity percentage. The same trend was observed for flange couch joints as
UTS was reduced by about 10–15% as line energy was reduced from 110 J/mm
to 88 J/mm. For both joints, 0.2 mm gap caused a slight difference in UTS
values, and that is due to the gap effects as concluded previously.
129
 Brief Introduction
In the previous chapter, the influence of the process parameters and the gap between the
sheets on pores was studied.
Without the introduction of any spaces between the sheets, the results showed that the
slower travel speed during welding generates less porosity in the solidified seam. This
implies, from the production point of view, that the welding process should be restricted
to low speeds to improve the welding quality causing a significant increase in the lead
time and may make the welding process unable to match the existing production line
speed.
The introduction of a 0.2 mm gap only improved the quality of the fillet edge welds
with a marginal reduction in porosity for the flange couch joints. Despite that, this
method did not eliminate porosity entrapment and did not remarkably widen the
operating window. Therefore, a question was raised as to whether the change of the wire
chemical composition may improve the weld quality and allow for higher speeds to be
applied in the process. The use of different wires involves the formation of different
microstructures and intermetallic phases which may affect the mechanical properties
and corrosion resistance of the weld.
The next chapter presents the outcome of investigating the change in the wire’s
chemical composition on porosity and the weld mechanical properties.
130
Publication Title:
Effect of Filler Wire Properties on Porosity Formation in Laser
Welding of AC-170PX Aluminium Alloy for Lightweight
Automotive Component Manufacture
Authors:
A.W. AlShaer, L. Li and A. Mistry
Journal:
Proceedings of the Institution of Mechanical Engineers, Part B:
Journal of Engineering Manufacture
Volume and Pages: (2015), 1-13
Publication No.: DOI. 10.1177/0954405415578584
STATUS: PUBLISHED
131
6 CHAPTER 6
EFFECT OF FILLER WIRE PROPERTIES
ON POROSITY FORMATION IN LASER
WELDING OF AC-170PX ALUMINIUM
ALLOY FOR AUTOMOTIVE COMPONENT
MANUFACTURE
6.1 Abstract
AC-170PX (AA6014) alloys are typically used in lightweight automobile vehicles.
Laser welding can be a viable tool for the assembly of components. However, porosity
is often generated during Al welding. In this article, an investigation is reported on the
characteristics of porosity formation in high-power disc laser welding of AC-170PX Al
alloy sheets in two weld joint configurations: fillet edge and flange couch with three
different filler wires of 4000, 3000 and 5000 Al series for each joint. Porosity,
microstructures, tensile strengths and joint geometry were investigated. It has been
found that the use of filler wires with higher Mg and Mn content such as AA5083 and
AA3004 leads to a significant reduction in porosity to less than 1.5% in both types of
joints compared with up to 80% porosity with the silicon-rich AA4043 wire. The
mechanism that led to this improvement is discussed.
Keywords: Laser welding, fillet edge joint, flange couch joint, aluminium alloy,
porosity, joint geometry, car, automotive.
6.2
Introduction
Al alloys are becoming commonly used in rail vehicles, electronics, automotive and
aerospace industries due to their high strength-to-weight ratio. Different welding
132
techniques such as friction stir welding and arc welding have been successfully applied
for joining components (Palmeri et al., 2007; Campbell et al., 2012; Bahemmat et al.,
2010). Laser welding would enable several advantages over the conventional welding
techniques. These advantages include higher welding speed, higher precision and more
flexibility (Spalding, 1987). However, laser welding of Al alloys can be challenging due
to defects associated with melting, solidification, shrinkage, thermal expansion,
hydrogen trapping and other gases’ solubility (Matrukanitz, 1990; Pinto et al., 2010),
etc. The type and amount of defects depend on the chemical compositions of the parent
materials and the filler wires as well as welding parameters. Porosity and cracks are the
most common defects that are observed during Al alloys welding. Some investigations
reported crack initiations during welding of Al alloys (6000 series) that have a wide
solidification range between the liquidus and the solidus, while others reported porosity
generation in Al alloys with a high magnesium content. This was attributed to the lower
melting temperature of Mg compared to Al (Zhao et al., 1999a; Ramasamy and
Albright, 2000). Haboudou et al. (2003b) used different methods for porosity reduction
such as surface preparation including sandblasting, polishing and laser cleaning, as well
as dual spot laser welding. Their experiments showed that surface cleaning reduced
porosity in the weld and laser cleaning had the most significant effect on other cleaning
techniques. Moreover, the increase in the interaction time between the laser beam and
the metal using dual spot beams led to a reduction in porosity by about 8% for AA5083
and A356 alloys. A similar trend has been observed by (Blackburn et al., 2010a, 2012),
who not only investigated the effect of twin spot laser beams on pore reduction in
titanium alloys but also optimised the directed assist gas parameters to achieve the same
goal. The directed argon gas jet led to less porosity, deeper penetration and eliminated
other weld defects such as undercuts and spatter. AlShaer et al. (2014) conducted an
investigation on porosity reduction using laser cleaning as a surface preparation method
prior to laser welding of Al alloys for automotive applications. They used an Nd:YAG
nanosecond pulsed laser to clean the alloys before being welded using a 5 kW disk
laser. Their findings indicated that laser cleaning has eliminated all contaminants and
dry lubrication from the surface of AA6014 alloys and led to a significant reduction in
porosity to less than 1% for fillet edge and flange couch joints.
133
Although much attention has been drawn to porosity formation in laser welding of Al
alloys, few investigations concentrated on the effect of the filler wire properties on weld
defects. The influence of shielding gases and their delivery in laser welding of Al alloys
were reported by Ancona et al. (2005), Katayama and Kawahito (2009) and Katayama
et al. (2010) who indicated that the use of helium gas achieved deeper penetration than
both Ar and nitrogen gasses due to the reduction of plasma. The influence of different
filler wires on Nd:YAG laser welding of dissimilar materials was reported by Pinto et
al. (2010), including welding of AA6061-to-LM25, AA6063-to-AA6083 and AA6060to-AA5754 using 4000 and 5000 series filler wires. Their findings concluded that
AA4047 filler wire is the most suitable for welding 5000-to-6000 Al alloys due to its
high silicon content if a suitable shielding gas is used. A 5000 series filler wire causes
hot cracking in dissimilar 5000-to-6000 laser welding. Welding different materials with
different chemical composition using filler wires that have another chemical
composition makes it difficult to observe the effect of the filler wire composition on
porosity formation in the welds. To the authors’ knowledge, no investigation has been
reported on the effect of filler wires on porosity formation in laser welding of AC170PX (AA6014) alloy.
In the present investigation, AC-170PX (AA6014) Al alloy sheets were laser welded
using three different filler wire types to examine their influence on porosity formation.
AC-170PX Al alloys are typically used for automotive applications owing to its high
strength to weight ratio and corrosion resistance. Two types of joints were welded using
a high power disk laser and the quality of the welds was evaluated for different filler
wires. Figure 6.1 illustrates the joint configurations.
(a)
(b)
Figure 6.1 Two types of joint configurations used in this investigation (a) Fillet edge
and (b) flange couch joints configuration
134
6.3
Materials and experimental procedure
Chemical compositions of the parent material and the filler wires are given in Table 6.1.
Sheets of 1.1 mm thickness were used as the parent material. In this investigation, the
Al alloy sheets were welded using different filler wires of 1.2 mm diameter: AlSi5
(AA4043), AlSi12 (AA4047), AlMgMn (AA3004) and AlMg4,5Mn (AA5083). In each
type of welds (fillet edge and flange couch) the three types of filler wires were
compared. The base material AC-170PX (AA6014) is tempered to T4 and coated with
titanium and zirconium (4 mg/m2) and lubricated using dry lubricant AlO70 (1.5 g/m2)
as part of the material forming process. The mechanical, physical and microstructural
properties after laser welding were characterised in the transverse direction of rolling
according to EW 10002. The physical and mechanical properties of the alloys are given
in Table 6.2.
Table 6.1 Chemical composition of the Al alloys (wt%) (Anon, 2011a; Kutsuna et al.,
2006; Efunda, 2016)
Cu
max.
Mn
Mg
Zn
max.
Ti
max.
Cr
max.
Others
each
max.
Others
Total
max.
Element
Si
Fe
max.
AC-170PX
(T4)
0.50.7
0.35
0.2
0.05-0.2
0.4-0.7
0.1
0.1
0.1
0.05
0.15
(AA4043)
5.2
0.5
0.1
0.05
0.05
0.1
0.1
-
0.05
0.15
(AA4047)
12.0
0.5
0.1
0.1
-
0.1
0.1
-
0.05
0.15
AlMgMn
(AA3004)
0.3
0.7
0.25
1.0-1.5
0.8-1.3
0.25
-
-
0.05
0.15
AlMg4.5Mn
(AA5083)
0.4
0.4
0.1
0.4-1.0
4-4.9
0.25
0.15
0.05-0.25
0.05
0.15
135
Table 6.2 Physical and mechanical properties of AC-170PX (after tempering) as well as
AA4043 filler wire (Efunda, 2016; Kutsuna et al., 2006; Anon, 2011a)
Alloy
Density
[103
kg/m3]
Yield
Strength
(0.2%)
[MPa]
Thermal
Conductivity
[W/m.K]
Electrical
Conductivity
[m/Ωmm2]
Coefficient of
Thermal Expansion
[10-6 K-1]
Rm
[MPa]
≥175
AC-170PX
(T4)
2.7
90
160-190
26-30
23.4
195
AA4043
2.68
40
170
24-32
22.1
160
AA4047
2.65
80
150-170
17-27
20
180
AlMgMn
(AA3004)
2.72
170
162
-
23.2
215
AlMg4.5Mn
(AA5083)
2.66
140
110-120
16-19
23.7
280
A TruDisk 5302 (max power 5300 W) disk laser with a Scansonic ALO3 head was used
for welding the sheets. Table 6.3 and Figure 6.2 give the laser technical specifications
and the schematic of the laser head respectively. A lens with focal length of 176 mm
was used to focus the beam at 10 mm below the workpiece surface producing a spot size
of 600 µm at the surface. The use of a defocused laser beam for the welding process is
to compensate for the effect of thermal lensing (that would effectively shorten the focal
length to move the focal position closer to the lens) and to provide sufficient heating
area for the weld zone formation. Argon gas was used for shielding the weld zones with
a flow rate of 15 l/min. Argon is widely used in industry as a shielding gas for the weld
zone protection from oxidation since it is cheaper and heavier than helium (Olabode et
al., 2013). A length of 300 mm weld seam was produced for each experimental
combination using a drag angle of 10 deg and a lateral angle of 45 deg. An inclined
beam was used in laser welding of Al to prevent beam interaction with plasma and to
avoid back reflection and the optics damage. Inclined gas shroud removes the plume
and prevents its interaction with the beam. Figure 6.3 demonstrates these angles in the
weld configuration. The ultimate tensile strength (UTS) was measured for all samples
using Galbadini Quasar 100 kN test machine (Samples were prepared according to ISO
6892-1 2009).
136
Figure 6.2 Scansonic ALO3 Laser head
Drag angle
Welding
Head
Welding
Head
Lateral
angle
Laser beam
Lateral
angle
Filler wire
45
Welding Direction
Welding Direction
Welding
Figure 6.3 Incident beam and its inclination angles of the normal plane
The wire feeding rates were selected in the range of 3.1 - 5.7 m/min to provide
sufficient but not excessive filler material in the weld zone. This allows welding under
different laser line energy inputs that would bond the two sheets to meet the industrial
joint geometry specification (e.g. weld depth and width).
On one hand, the increase in silicon content in Al alloys increases the fluidity since the
heat of fusion of primary silicon is higher by 4.5 times than that of pure Al. Hence,
137
liquid AlSi12 alloys are less viscous than molten AlSi5 and Al-Mg alloys (Committee,
1979; Sabatino, 2005).
Table 6.3 TruDisk 5302 laser unit technical properties (Trumpf-Lasers, 2013)
TruDisk 5302
Unit
Value
Wavelength
[nm]
1030
Maximum Laser power
[W]
5300
[mm.mrad]
8
Min. diameter laser light cable
[µm]
200
Power stability at nominal power
[%]
±1
Beam quality
o
Cooling water temperature range
[ C]
5 - 20
Dimensions
[mm]
1990  1550  1200
On the other hand, Al-Mg wires produce more spatter and “smut” (black deposit of the
metal oxide on the weld surface) than Al-Si wires that is very detrimental to the
appearance of the welds in the structural components (ESAB Knowledge Center, 2014).
Accordingly, limits for wire feeding rates apply when Al-Mg fillers are used for such
applications. Therefore, identical feeding rates cannot be utilised for all cases regardless
of the wire composition. For each wire, the feed rates were selected to achieve optimum
welds.
To compare the effect of filler wires, the laser welding parameters were fixed in each set
of welding experiments at 35 to 50 mm/s welding speed at a laser power of 3500 W to
5300 W. The selection of laser powers and welding speeds was to provide the required
line energy (approximately 100 J/mm) to melt the wire and some surrounding material.
On the other hand, welding parameters were selected to meet the industrial requirements
requested and to be in the range that is utilised and feasible for automotive industry.
After welding had been completed, all samples were sectioned in the perpendicular
direction to the weld axis, and then ground using abrasive papers that contain silicon
carbide particles. Grinding was performed in five stages using 80, 300, 600, 1200, 4000
grit papers using a rotating disk (Mecatech 334). Following that, the samples were
polished using 3 μm and 1 μm diamond pastes on a rotating cloth. The samples were
then etched using Sodium Hydroxide solution (1g NaOH + 100ml H2O) in which they
138
were immersed for 45 seconds. The microstructure and the dimensions of the welds
were then examined using a digital microscope KEYNCE VHX-500F. It is worth
mentioning that porosity levels were determined using cross sections along the weld
bead and all measurements were repeated multiple times. Three cross sections were
made of the weld bead, and porosity content was calculated as a ratio of the pores
projected area to the weld cross section area. The measurements of porosity content for
each section were then repeated three times. The errors shown in Figure 6.10
demonstrate the level of deviation in these measurements. The same procedure was also
followed for flange couch joints and the parameters used in welding the fillet welds are
shown in Table 6.4.
Table 6.4 Welding parameters for fillet edge joints
Fillet Edge Joint
Welding
speed
[mm/s]
50
Laser
Power
[W]
5300
Wire Speed
[m/min]
Filler Wire Type
3.4
AlSi5 (AA4043)
5.7
AlSi12 (AA4047)
4.4
AlMgMn (AA3004)
6.4 Experimental results
6.4.1
Fillet edge joints
6.4.1.1 Microstructure and porosity characteristics
Micro-sections were examined initially using low magnification optical microscopy
(x100) and porosity levels and other defects were examined. Figure 6.4 shows the
micro-sections of the fillet edge samples welded using different filler wires.
From Figure 6.4, it can be seen that a large quantity of porosity is observed during
welding using the AlSi5 filler at the specified parameters. However, less porosity was
observed when welding with AlSi12 filler and it was further reduced with the AA3004
filler. Some cracks were also observed with the AA4043 (AlSi5) filler wire welds
especially at the root of the joints but no macro- or micro-cracks were found in the weld
joints welded with AlSi12 and AA3004 wires.
139
Moreover, both AlSi5 and AlSi12 welds suffered from the lack of fusion at the weld
root while no such defect exists in the AlMgMn (AA3004) welds.
Laser Power
[ 5300 W ]
Filler Wire type
AlSi5 (AA4043)
Welding
Speed
AlSi12 (AA4047)
[50 mm/s]
AlMgMn
(AA3004)
Figure 6.4 Microsections of fillet edge joints welded using different filler wires
Using higher magnification (x400), the welds microstructure can be clearly examined.
Figure 6.5 shows the microstructure in the fusion zone (FZ), heat affected zone (HAZ)
and the parent material after welding using the AlSi5 filler wire. The microstructure of
the base material consists of elongated Al grains whose boundaries do not appear
clearly due to the large elongation occurred during the rolling process. Black dots of
magnesium silicide Mg2Si particles can be seen distributed randomly within the grains.
In the HAZ, Mg2Si is rearranged to form a eutectic phase and precipitates at the grain
boundaries in a dendritic form.
140
In the FZ, Mg content in the filler wire is very low in comparison with the silicon
content. Therefore, Al forms an (Al-Si) eutectic with silicon and settles at the grain
boundaries. The magnesium silicide is not clearly observed in this region due to the low
Mg content in comparison with the parent material.
FZ
Figure 6.5 Microstructure of the weld-HAZ-base material of a fillet edge joint welded
at 5300W laser power and 50mm/s welding speed (AlSi5 filler wire)
Elongated
grains
Figure 6.6 Microstructure for a fillet edge joint welded using AlSi12 filler wire
141
Figure 6.6 shows the microstructure of fillet joints welded using the AlSi12 filler wire.
The microstructure in the AlSi12 weld is similar to that of AlSi5 in terms of the phases.
However, due to the high silicon content in the AlSi12 (4047) filler wire, dendritic Al-Si
eutectic becomes coarser and more visible in between the Al grains and it looks like a
crack under the microscope using low magnifications. Pores of different sizes appear
clearly in Figure 6.6 in the fusion zone.
Figure 6.7 shows the difference in Al-Si amount in the fusion zone for both AlSi5 and
AlSi12 filler wire. It can be seen that the grain size in the case of AlSi12 ranges between
5 to 7 μm, compared with 10 to 15 μm in the AlSi5 welds.
(a)
(b)
Figure 6.7 A Microstructure comparison between fillet edge joints welded using (a)
AlSi12 filler wire and (b) AlSi5 filler wire
142
Figure 6.8 Microstructure of the weld-HAZ-Base material of a fillet edge joint welded
at 5.3KW and 50mm/s (AlMgMn filler wire)
The microstructure of weld with the AlMgMn filler wire is shown in Figure 6.8. The
HAZ takes a different appearance in which Al solid solution appears in the light yellow
zone containing some coarse and fine particles. Since the filler wire and the base
materials are of different chemical composition mixed together, the black particles
which appear in between the grains boundaries in the HAZ can be Mg2Si which
originates from the base material and the eutectic Al6Mn formed due to the high
Manganese content in this filler wire in comparison with AlSi5 and AlSi12 fillers. In the
FZ, randomly-distributed dark particles of FeAl3 eutectic and Al6Mn are observed
within the Al solid solution with a light colour. FeAl3 becomes more pronounced in the
FZ because AA3004 filler wire contains two times more iron than the base material
(Mondolfo, 1976).
Weld
Area
Weld
Area
(a)
(b)
Figure 6.9 Weld areas used in porosity proportions determination for (a) Fillet edge and
(b) Flange couch joints.
143
It is obvious from Figure 6.4 that porosity reached significant proportions for the welds
with silicon-rich filler wires, while very little pores were detected for welds with Mgrich filler wires. Porosity was estimated as the ratio of the pores’ projected area to the
weld cross-section area averaged in multiple cross sections. Pores and weld areas were
determined directly using the microscope image processing software.
Figure 6.10 Porosity content in fillet edge joints welded with different three filler wire
at 5.3KW and 50mm/s
Figure 6.9 and Figure 6.10 illustrate the weld areas for both types of joints where
porosity was characterised, and porosity percentages for the three different filler wires
respectively. Porosity content in the AlSi5 welds has reached up to 80% at the specified
weld parameters. It was reduced to less than 10% for the AlSi12 to reach only 1.4 % for
AlMgMn welds. The maximum pore diameter observed in the fillet edge joints reached
up to 690 μm for the AlSi5, 90 μm for AlSi12, and 56 μm for AlMgMn welds.
6.4.1.2 Weld dimensions and tensile strength
The weld penetration depth and weld width definition for fillet edge joints are illustrated
in Figure 6.11.
144
Gap
(a)
Weld
Weld
(b)
Figure 6.11 Fillet edge weld dimensions (t1 and t2 sheets’ thicknesses, s the gap size, Pw
weld width, and Pd penetration depth)
The weld dimensions were measured using the optical microscope for all the joints as
shown in Figure 6.12. All weld dimensions were found to be maximum when AlMgMn
was used as a filler since the penetration depth and the weld width reached up to 1000
μm and 2500 μm respectively. Penetration depth for the welds using either AlSi5 or
AlSi12 wires were approximately half that for AlMgMn. This difference in weld
dimensions, although welding parameters were the same, can be attributed to the root
cracks which caused the lack of fusion for both AlSi5 and AlSi12 fillers.
Figure 6.12 Weld dimensions for fillet edge joints using various filler wires at 5.3KW
and 50mm/s
The difference in the weld dimensions and porosity content besides the difference in the
chemical composition of the filler wires led to a difference in the ultimate tensile
strengths. As shown in Figure 6.13, the UTS has increased from its minimum value for
AlSi5 (38%) to the maximum value for the AlMgMn (55%) weld. Although AlSi12
weld dimensions were less than those obtained for the AlSi5 welds, porosity caused a
significant reduction in the UTS of the AlSi5 welds since it occupied 80% of the weld
cross section area. The use of AlMgMn filler achieved the maximum UTS due to three
reasons: the low porosity content in the weld, the large fusion zone volume and the high
145
Mg and Mn content that is responsible for increasing the strength (ASM International,
2004a).
Figure 6.13 Tensile strength for various filler wires welds at 5.3 KW and 50 mm/s as a
proportion of the base material UTS (195 MPa)
6.4.2
Flange couch joints
6.4.2.1 Microstructure and Porosity characteristics
The same analytical procedures were followed for the flange couch joints and the
welding parameters are given in Table 6.5. Figure 6.14 shows the micro-sections of
flange couch joints welded using AlSi5, AlSi12 and AlMg4,5Mn filler wires at 3500 W
laser power and 35 mm/s welding speed.
Table 6.5 Welding parameters for flange couch joints
Flange Couch Joint
Welding
speed
[mm/s]
35
Laser
Power [W]
3500
Wire Speed
[m/min]
Filler Wire Type
3.1
AlSi5 (AA4043)
3.8
AlSi12 (AA4047)
4.1
AlMg4,5Mn (AA5083)
146
Laser Power
[ 3500 W ]
Filler Wire type
AlSi5 (AA4043)
Welding
AlSi12 (AA4047)
Speed
[35 mm/s]
Al Mg4,5Mn
(AA5083)
Figure 6.14 Microsections of flange edge joints welded using different filler wires
Figure 6.14 shows that the use of AlSi12 (AA4047) wire increases porosity in
comparison with AlSi5 (AA4043) filler at the same operating parameters. AlMg4,5Mn
(AA5083) reduced porosity in the flange welds significantly compared to other fillers.
Micro-cracks were not observed in any of the joints and welds did not suffer from other
defects such as lack of fusion, or undercuts, etc.
The microstructures of the flange couch welds were found to be identical to that shown
in Figure 6.5 and Figure 6.6 when AlSi5 and AlSi12 wires were used respectively. This
can be related to the fact that the same filler wires were used to weld the same type of
the parent material. Therefore, the joint configuration did not have any effect on the
type of phases exist in the flange welds. A different microstructure was observed in
Figure 6.15 due to the use of AA5083 filler wire. In the heat affected zone (HAZ), black
147
dendrites of magnesium silicide that originates from the base material are pronounced at
the grains boundaries since this region contains relatively high silicon content. Beside
this phase, FeAl3 phase is observed as dark rounded dots which are distributed randomly
over the heat affected zone. In the fusion zone, FeAl3 becomes more evident since
higher iron content exists in the fusion zone than in the HAZ. This phase is distributed
within a light grey network of Mg2Al3 eutectic which is formed from the very high
magnesium content in the AA5083 alloy (Mondolfo, 1976; Punkari et al., 2003). Small
pores can also be seen within different zones.
Figure 6.15 Microstructure of the weld-HAZ-Base material of a flange couch joint
welded at 3500W and 35mm/s (AlMg4,5Mn filler wire)
After calculating porosity percentages, Figure 6.16 was created to compare the different
flange welds. It can be seen from this figure that porosity recorded the highest value for
AlSi12 (AA4047) welds while the minimum value was achieved in AlMg4,5Mn
(AA5083) welds. At this level of power and welding speed, although the joint
configuration helps the gases to escape from the welding pool, more pores were
generated in AA4047 flange joints. The high silicon content (about 12%) of the
AA4047 wire increases fluidity and hence the speed of the flow inside welding pool,
which in turn promotes the interaction between liquid Al and ambient gases especially
shielding gases which are blown directly at a high flow rate (15 l/min) onto the welding
pool. Another source for porosity during laser welding of Al alloys can be the hydrogen
148
gas. Some works reported that hydrogen was the main reason for porosity formation and
even 2 ppm of hydrogen can still lead to porosity in Al welds (Askeland, 1985).
Figure 6.16 Porosity content in flange couch joints welded with different three filler
wire at 3.5KW and 35mm/s
Moreover, some pores can be formed due to the keyhole instability which is dependent
on how stable the flow inside the welding pool is. Higher fluidity may increase the
chance of keyhole collapse and leads to more porosity (Bachmann et al., 2013; Zhao
and DebRoy, 2011; Zhao et al., 1999b). The maximum pore size observed for the flange
couch joints in this investigation was about 65 μm in diameter for both AlSi5 and
AlSi12 welds and 22 μm for AlMg4,5Mn welds.
6.4.2.2 Weld dimensions and tensile strength
The weld penetration depth and weld width for flange couch joints are illustrated in
Figure 6.17.
Gap
Weld
(a)
Weld
(b)
Figure 6.17 Flange couch weld dimensions (t1 and t2 sheets’ thicknesses, s the gap size,
Pw weld width, and Pd penetration depth)
149
The weld dimensions values are depicted in Figure 6.18. It is clear from Figure 6.18,
which shows the weld dimensions for the flange couch weld, that there is no significant
change in both penetration depth and weld width for the various welds. This can be
related to two reasons; same welding parameters and the same weld configuration were
used for all filler wires, and more importantly no defects such as lack of fusion or root
cracks were observed in these joints.
Figure 6.18 Weld dimensions for flange couch joints using various filler wires at
3.5KW and 35mm/s
Tensile strengths of the joints were also evaluated to the ISO 6892-1 2009 standard, and
their values are given in Figure 6.19. It can be seen that AlSi12 (AA4047) welds
showed the least tensile strength among other joints due to its containment of the
highest porosity percentage. Porosity affects UTS directly since it reduces the cross
section area that carries the tensile load. The highest UTS value of more than 80% was
achieved using AlMg4,5Mn (AA5083) filler wire which obtained its strength from the
high magnesium content (more than 4.5%) and other alloying elements which form high
strength precipitates.
150
.
Figure 6.19 Tensile strength for various filler wires welds at 3.5 KW and 35mm/s as a
proportion of the base material UTS (195 MPa) for the flange couch joints
6.5 Discussion
This work shows clearly that the addition of Mg and Mn to the filler wires can
significantly reduce porosity and cracks compared to Al-Si filler wires in laser welding
of the Al alloys. This is opposite to some of the previous findings (Ramasamy and
Albright, 2000; Zhao et al., 1999b) when filler wires were not used.
This may be attributed to the fact that Si element in the filler wires increases the fluidity
of Al alloy by decreasing the viscosity to less than 1cP at 700oC for alloys at 12%
silicon (Zhao et al., 1999b). This increases the disturbance in the weld pool especially at
high welding speeds, which in turn enhances the entrapment of the gases inside the
weld. Although the increase in Mg content increases hydrogen solubility in Al, the
AA3004 filler wire contains more Fe and Cu than the base material and the other two
filler wires, which in turn reduces hydrogen solubility to a certain level (Zhao et al.,
1999b). More importantly, the microstructure of the dendritic Al-Mg eutectics that are
produced in the AlSi5 and AlSi12 welds makes magnesium evaporation much easier to
occur than in the refined and rounded phases in the AlMgMn welds. In a previous study,
it was found that the refinement of the microstructure in AA3004 welds reduces
alloying elements evaporation and subsequently reduces porosity generation (Kuo and
Lin, 2006a).
151
Keyhole stability is dependent on the vapour pressure inside the keyhole as well as the
surface tension of the liquid. When the surface tension becomes greater than the vapour
pressure, the keyhole collapses and results in voids inside the weld (Schauer and Giedt,
1978). It is pertinent to mention that unlike Cu and Si, Magnesium has lower boiling
and melting points, and has a very high vapour pressure that leads to a reduction in the
surface tension of molten Al (Szekely, 1979). However, it may be possible that Mg
helps to stabilise the keyhole and results in less porosity in welds with high Mg content.
It is worth mentioning that less porosity was observed in the flange couch joints in
comparison with the filler edge ones although the same line energy (approximately 100
J/mm) was used. This can be attributed to the gap that exists at the root of the joint due
to the joint’s configuration. This gap allows gases to escape from the weld while they
are trapped inside the welding pool in fillet edge joints.
It should be noted that no cracks were observed in the welds produced in this work
when Mg filler wires were used. Many researchers reported that some of the Al-Mg
alloys can be used to produce welds which are free from solidification cracking
(Punkari et al., 2003; ASM International, 2004a; Kutnusa et al., 1993). This can be
attributed to the high Mg content (4.5%) which alters the FZ composition and reduces
the weld crack susceptibility, as well as to its role in the formation of equiaxed grains in
the FZ instead of the elongated grains which are associated with autogenous laser
welding of 6000 alloys. The fine and equiaxed grains hinder cracks initiation and
propagation in the fusion zone which is also observed when Al-Si filler wires are used
(Zhao et al., 1999b; Punkari et al., 2003).
Some previous work reported that AA6013 alloy could be welded without filler wire
using a fibre laser. However, the process window within which sound welds are
produced was found to be very narrow (0.8 to 1.5 kW and 70 to 100 mm/s), and many
defective welds were produced outside that range (Vilar et al., 2008). The work reported
in this paper shows that the use of filler wire broadens the process window within which
low porosity welds can be achieved.
152
6.6 Summary
This work reported the effect of filler wires composition on the joint characteristics such
as porosity, tensile strength, and weld dimensions during laser welding. The work can
be concluded as follows:
1. Porosity in fillet edge joints decreases from 80% (where AlSi5 filler wire is
used) to less than 1.5% when AlMgMn filler wire is used. For flange couch
joints, the maximum porosity proportion is observed in the AlSi12 joints (3%),
while it is reduced to about 1% using AlMg4,5Mn filler wires. In both types of
joints, filler wires with higher Mg and Mn content lead to a significant reduction
in porosity.
2. Weld dimensions such as weld width and penetration depth for the fillet edge
joints vary due to various weld defects such as lack of fusion and root cracks.
However, no significant difference is observed in the weld dimensions in the
flange couch joints due to the joint configuration.
3. The tensile strength (UTS) of the weld is maximum in the fillet and flange joints
when AlMgMn and AlMg4,5Mn fillers are used respectively. This can be
attributed to the high Mg and Mn content in these wires compared to AlSi5 and
AlSi12 fillers.
Further investigations may include an evaluation of the corrosion resistance as well as
other mechanical properties such as impact and fatigue strength for the studied wires.
The weld appearance is of a great interest for structural applications in which the outer
skin should be sustainable, smooth and defects-free. Therefore, corrosion resistance can
be explored in more detail. Moreover, a systematic Design of Experiment can be created
to examine a wider process window using two or three filler wires.
153
 Brief Introduction
In the previous chapter, the influence of the wire’s chemical composition on porosity
was studied.
Generally speaking, during welding of different materials, various defects including
pores, burn through, smut and spatter appear on the weld surface. For certain
applications, these defects can be a minor issue or can be easily removed using a steel
brush or other tools. However, for the outer car body sheets, these imperfections are
strictly taken into consideration and must be eliminated since these sheets are directly
taken to the paint shop after welding without any further post-treatment. Using a steel
brush to remove spatter or the use of chemicals to remove the weld dark colour (smut)
adds to the process lead time and requires additional workforce and tools.
The use of filler wires with high magnesium and manganese content has improved the
weld mechanical properties but produced welds with undesirable imperfections
including spatter, rough surfaces and a darker surface colour. Removing these defects
not only adds to the process time but also may damage the sheets’ surfaces causing
scratches or dents, which increase the part-rejection percentage during production.
Therefore, another technique, namely laser cleaning, was discussed with the industrial
partner as a surface preparation method to reduce porosity and to produce a better
appearance during welding with silicon filler wires.
The next chapter presents the outcome of investigating laser cleaning effect on porosity
and the weld appearance.
154
Publication Title:
The effects of short pulse laser surface cleaning on porosity
formation and reduction in laser welding of aluminium alloy for
automotive component manufacture
Authors:
A.W. AlShaer, L. Li and A. Mistry
Journal:
Optics and Laser Technology
Volume, Issue and pages: 64 (2014) 162–171
Publication No.:
DOI:10.1016/j.optlastec.2014.05.010
STATUS: PUBLISHED
155
7 CHAPTER 7
THE EFFECTS OF SHORT PULSE LASER
SURFACE CLEANING ON POROSITY
FORMATION AND REDUCTION IN LASER
WELDING OF ALUMINIUM ALLOY FOR
AUTOMOTIVE COMPONENT
MANUFACTURE
7.1 Abstract
Laser welding of Al alloys typically results in porosity in the fusion zones, leading to
poor mechanical and corrosion performances. Mechanical and chemical cleaning of
surfaces has been used previously to remove contaminants for weld joint preparations.
However, these methods are slow, ineffective (e.g. due to hydrogen trapping) or lead to
environmental hazards. This paper reports the effects of short pulsed laser surface
cleaning on porosity formation and reduction in laser welding of AC-170PX (AA6014)
Al sheets (coated with Ti/Zr and lubricated using a dry lubricant AlO70) with two types
of joints: fillet edge and flange couch, using an AA4043 filler wire for automotive
component assembly. The effect of laser cleaning on porosity reduction during laser
welding using a filler wire has not been reported before. In this work, porosity and weld
fusion zone geometry were examined prior to and after laser cleaning. The nanosecond
pulsed Nd:YAG laser cleaning was found to reduce porosity significantly in the weld
fusion zones. For the fillet edge welds, porosity was reduced to less than 0.5%
compared with 10–80% without laser cleaning. For flange couch welds, porosity was
reduced to 0.23–0.8% with laser cleaning from 0.7% to 4.3% without laser cleaning.
156
This has been found to be due to the elimination of contaminations and oxide layers that
contribute to the porosity formation. The laser cleaning is based on thermal ablation.
Keywords: Laser welding, porosity, Al alloy.
7.2 Introduction
Lightweight alloys such as Al alloys usually have a layer of oxides or hydroxides on
their surfaces. These types of surface layers may affect fusion joining processes due to
hydrogen entrapment in the fusion zones (Feliu et al., 2009). Various approaches have
been reported on the elimination of surface oxide layers before welding. The use of
chemical substances is the main method to remove the oxide layer (Cherepy et al.,
2005). Sandblasting, silicon carbide polishing and plasma etching were also reported as
weld surface preparation methods (Kersten et al., 1996; Haboudou et al., 2003b). Most
cleaning methods used in the past depend on chemical substances that react with the
metal on the surface. These methods were referred to as wet cleaning. Although they
can remove the undesirable layer, the use of these substances and their storage may
harm the workers‫ ׳‬health and can raise environmental and safety risks (Dimogerontakis
et al., 2005). Furthermore, due to the use of fluid in cleaning, hydrogen trapping was
difficult to avoid, which would cause porosity in welds.
Lasers have been used in many industrial applications such as drilling, cutting,
machining and recently for surface cleaning (Watkins et al., 2003). Laser cleaning is
usually considered to be one of the dry cleaning methods. It should be noted that not all
lasers are suitable for cleaning applications. Laser dry cleaning leads to particle or film
ejection from the surface using high power laser pulses. Ejection of contaminants from
the surface is achieved due to the sudden expansion of particles or the surface or by
evaporation. According to the laser wavelength and the type of the substrate material,
the irradiation energy is absorbed by the surface, the particle or the contaminated film
(Yong-Feng et al., 1998). Tam et al. (1992) found that the ejection acceleration
(~1010 cm/s) given to a particle by the laser radiation is much greater than that of
gravitational acceleration. This results in particles ejection without melting the surface.
Their study shows that higher laser power intensities, shorter pulse duration and shorter
wavelength lead to a high cleaning efficiency. Contaminant particles usually adhere to
157
the metallic surface based on the Van der Waals or electrostatic forces (Bäuerle, 2000).
Laser cleaning can remove the contaminating particles according to two different
mechanisms, either by direct ejection of particles which is achieved by vaporisation or
hydrodynamic sputtering or by the thermoplastic expansion of the underneath parent
material (Psyllaki and Oltra, 2000). This means that laser cleaning is achieved when the
energy is absorbed either by the contaminant particles or by the substrate. Sometimes a
combination of the two mechanisms can occur. For the removal of contaminants in the
form of thin films (such as oxides, grease machining coolant residuals and oil), the
mechanisms of material removal are strongly dependent on the wavelengths of the laser.
For example, excimer laser at 248 nm wavelength would cause photochemical ablation
of certain organic contaminants, and Nd:YAG and CO2 lasers would be mainly based on
thermal ablation processes with Nd;YAG laser at 1064 nm wavelength interacting with
materials both on the surfaces of the contaminants and directly with the substrates while
the CO2 lasers (10.6 μm) would only interact with materials on the surfaces (Turner et
al., 2007).
The energy absorbed by the contaminant can be calculated from (Turner et al., 2006b):
E 
I0 A
 BL
d
e
 z /  BL
dz
( 7-1 )
0
where I0 is the intensity, A is the laser spot area, δBL is Beer–Lambert absorption depth
and d is the thickness of the contaminant layer.
Laser cleaning offers many technical advantages such as remote application, localised
cleaning, cleaning of components with complex 3-D shapes and elimination of the use
of hazardous chemical solutions. Since lasers that are used in surface preparation should
only remove a thin layer without affecting the bulk material, the selection of the process
parameters becomes very critical. These parameters can include wavelength, pulse
duration, pulse energy, repetition rate, scanning speed, laser fluence,
etc
Dimogerontakis et al., 2005).
It should be noted that during the interaction between a laser beam and metallic alloys,
photon–electron harmonic interactions may be observed that are closely related to the
non-linear optical effects. (Turner et al., 2005, 2006a) investigated and modelled
158
Nd:YAG and CO2 laser cleaning of the Ti6Al4V alloy. They concluded that the
prevailing effect, in the case of YAG laser cleaning, is the heat conducted to the
contaminants from the substrate surface. However, contaminants were removed by
direct heating if a CO2 laser is used. In a further investigation (Turner et al., 2007), they
compared the use of CW CO2, pulsed YAG and excimer lasers for titanium alloys
cleaning. Their results showed that photochemical ablation takes place only when the
excimer laser is used due to its short wavelength, while thermal vaporisation becomes
dominant in the YAG and CO2 laser cleaning at the selected parameters. In addition,
oxidation was noticed to be less after pulsed laser cleaning than CW laser cleaning
(Rechner et al., 2010).
Yue et al. (2012) investigated the removal of the alpha case in Ti alloys while
(Marimuthu et al., 2010, 2013) monitored the TiN removal from WC micro-tools and
modelled excimer laser cleaning mechanism of film and particle contamination. They
concluded that the peak power in excimer laser cleaning had more significant influence
than the number of pulses. Some models were created to simulate the laser bending of
Ti sheets while others focused on the estimation of the optimal solution for minimising
the surface roughness in end milling of Ti alloys (Zhang et al., 2014).
Other materials were laser cleaned using different lasers such as: AA5083 Al alloy
(using a YAG laser) (Haboudou et al., 2003b), stainless steel (using a YAG laser)
(Psyllaki and Oltra, 2000), stainless steel 316L and Inconel (using an excimer laser)
(Delaporte et al., 2003), Silicon (a CO2 laser) (Allen et al., 1998), and Copper (using a
femtosecond Ti-Saphire laser) (Hallo et al., 2006).
Haboudou et al. (2003b) studied porosity reduction using laser cleaning and other
methods for removing the surface oxide layer on AA5083 and A356 Al alloys. They
used a pulsed YAG laser for cleaning with a fluence of 1.5 J/cm2, a scanning speed of
20 mm/s, a frequency of 20 Hz and a 100 nm pulse length. Their results show that laser
cleaning of the surface reduced porosity in the welds to less than 2% and 7% for A356
and AA5083, respectively. They attribute the difference in porosity generated in the
different materials to Mg content in these alloys since Mg vaporisation increases as Mg
content increases. Laser cleaning was found to be more efficient than mechanical
polishing or sand blasting for this type of material. Meja et al. (1999) investigated laser
159
cleaning of anodised Al (black anodised) with an oxide layer thickness of 20 µm using a
laser at 248 nm, 532 nm and 1064 nm wavelengths in order to determine the most
efficient radiation for the oxide removal. They concluded that there was a shift in the
energy density threshold at which cleaning took place between 532 nm and 1064 nm
wavelengths. They attributed this to the higher absorption at a shorter wavelength.
Rechner et al. (2010) studied Nd:YAG laser cleaning of an AA6016 Al alloy which was
coated with a TiZr oxide layer. They used XPS analysis to confirm the reduction in
oxygen and carbon content on the rolled alloy surface, and it showed that laser cleaning
had removed a significant amount of Fe, C, Si, and O2 from the surface. The sheets were
subsequently adhered together using epoxy adhesive and surface cleaning was found to
improve the tensile strength joint for this type of material.
As seen from the literature, research on laser cleaning focused so far mainly on cleaning
of steels, titanium, and very few types of Al alloys. Therefore, further research is
required on other Al alloys in order to understand the effect of laser cleaning on weld
defects such as porosity since it is one of the most common problems faced in Al
welding using high energy beams such as laser and electron beams. In the previous
work, some researchers discussed the effect of mechanical and chemical cleaning of the
surface on laser welding using a filler wire (Vilar et al., 2008; El-Batahgy and Kutsuna,
2009; Sánchez-Amaya et al., 2009a), while others discussed laser cleaning effect on
autogenous Al welding (Haboudou et al., 2003b). However, the effect of laser cleaning
on porosity reduction during laser welding using a filler wire has not been reported
before, and laser cleaning of AA6014 Al alloy has not been reported before.
In this study, the effect of laser cleaning as a surface preparation method for laser
welding of AC-170PX (AA6014) is reported. The Al sheets were laser cleaned and
welded. The porosity, microstructure, and weld dimensions were compared with and
without laser cleaning.
7.3 Materials and experimental procedure
AC170-PX Al sheets of 1.1 mm thickness were laser welded using AA4043 filler wire
with a diameter of 1.2 mm. The chemical compositions of these materials are given
in Table 7.1. The parent material has been tempered to T4 (solution heat treated and
160
naturally aged) and its physical and mechanical properties, which are given in Table 7.2,
are measured in the transverse direction to rolling direction according to EW 10002 (a
tensile test in which the sample is strained to fracture in order to determine the
mechanical properties) (British Standards Institution, 2001). The base material was
coated with titanium and zirconium (4 mg/m2) and lubricated using a dry lubricant
AlO70 (1.5 g/m2) as part of their sheet metal forming process. The total coating
thickness was calculated to be 0.503 μm since Ti and Zr coatings together were about
3 nm in thickness and the AlO70 lubricant has a thickness of 0.5 μm.
Table 7.1 Chemical composition of the Al alloys (wt-%) (Anon 2011a, 2011b)
Element
Si
Fe
max.
Cu
ma
x.
Mn
Mg
Zn
max.
Ti
max.
Cr
max.
Others
each
max.
Others
Total
max.
AC-170PX
(T4)
0.50.7
0.35
0.2
0.050.2
0.4-0.7
0.1
0.1
0.1
0.05
0.15
AA4043
5.2
0.5
0.1
0.05
0.05
0.1
0.1
0.05
0.15
Table 7.2 Physical and mechanical properties of AC-170PX and AA4043 filler wire
(Anon 2011a, 2011b)
Electrical
Conductivity
[m/Ωmm2]
Coefficient
of
Thermal
Expansion
[10-6 K-1]
Rm
[MPa]
≥175
Alloy
Density
[103 kg/m3]
Yield
Strength
(0.2%)
[MPa]
AC-170PX
(T4)
2.7
90
160-190
26-30
23.4
195
AA4043
2.68
40
170
24-32
22.1
160
Thermal
Conductivity
[W/m.K]
Two different joint configurations were examined in this work: fillet edge and flange
couch joints (Figure 7.1). Before laser welding, samples are cleaned using a Q-switched
Nd:YAG laser (CleanLASER CL600) and operated at the parameters given in Table
7.3. After the sheets are cleaned, they are welded using a Trumpf disk laser TruDisk
5302 (maximum output power of 5300 W) with the beam properties given in Table 7.4.
It should be noted that both types of lasers used a Gaussian beam shape. The laser
welding was performed with a defocused laser beam with the focal point at of 8 mm
below the workpiece and shielded by argon gas at a flow rate of 10 L/min. A 600 µm
beam spot (on the workpiece surface) was used in welding, and the beam was focused
using a lens with a focal length of 176 mm. For both types of joints, a 290 mm length
161
seam was welded with a drag angle of 10 deg. The lateral angles for fillet edge and
flange couch joints were 45 deg and 20 deg, respectively. Such angles were used to
prevent the beam interaction with plasma (Narikiyo et al., 1996). This also prevents
beam back reflection to the laser optics. Figure 7.2 illustrates the drag and lateral angles
relative to the normal plane to the weld.
Weld
Weld
(a)
(b)
Figure 7.1 (a) Fillet edge joint (b) Flange couch joint with offset
Table 7.3 cleanLASER CL600 laser beam properties (CleanLaser, 2013)
Nd:YAG CL600
Unit
Value
Wavelength
[nm]
1064
Operating power
[W]
600
Min. diameter laser light cable
[µm]
310
Spot diameter
[µm]
780
Pulse frequency
[kHz]
20
Pulse duration
[ns]
100
Scan frequency
[Hz]
180
Scan width
[mm]
20
[mm/s]
95
[%]
53.8
Scanning speed
Overlap in scan direction
162
Table 7.4 TruDisk 5302 laser unit technical properties (Trumpf-Lasers, 2013)
TruDisk 5302
Unit
Value
Wavelength
[nm]
1030
Maximum Laser power
[W]
5300
[mm.mrad]
8
Min. diameter laser light cable
[µm]
200
Power stability at nominal power
[%]
±1
Cooling water temperature range
[oC]
5 - 20
Beam quality
The laser cleaned samples were welded at different laser powers (from 2 kW to 5 kW)
and welding speeds (from 20 mm/s to 50 mm/s). Both cleaned and uncleaned samples
were then cut perpendicular to the welding direction and then ground using silicon
carbide abrasive papers at four stages of 80, 300, 600, and 1200 grits using a rotating
disk (Mecatech 334). After that, the specimens were polished using 3 μm and 1 μm
diamond pastes, respectively and then etched by immersing the specimens in Sodium
Hydroxide solution (1 g NaOH+100 ml H2O) for 45 s. The macrostructure and
microstructure of the samples were examined using a digital microscope KEYNCE
VHX-500F. Wyko white light interferometry was used to examine the surface
topography. Scanning electron microscopy (Hitachi S-3400N Type I) was used to
examine the damage to the substrate. In addition, both penetration depth and the weld
width were measured for the laser cleaned sample and compared with the uncleaned
one. Laser-induced breakdown spectroscopy (LIBS) LIBSCAN 0165 was utilised for
the surface chemical analysis before and after laser cleaning.
163
Figure 7.2 Incident beam and its inclination angles of the normal plane
7.4 Experimental Results
7.4.1
Effect of laser cleaning on the surface characteristics
In addition to the microstructures that have been examined for both samples, the surface
topography was investigated for the cleaned surface in order to visualise the effect of
laser cleaning on the surface characteristics. Figure 7.3 shows a cross section view of
the interface between laser cleaned and uncleaned zones. A surface layer of around
19 µm was removed, and the laser cleaned area appeared to be smoother. Figure
7.4 shows the surface morphology before and after laser cleaning. A smoother surface
(Ra=882 nm) was observed compared with that of the un-cleaned surface (Ra=982 nm).
Laser cleaned
As-received
Surface
surface
Figure 7.3 A cross-section view of the laser cleaned/uncleaned interface showing the
material removal depth by laser cleaning
164
(b)
(a)
Figure 7.4 Tomography of (a) a reference surface for which the roughness is Ra= 982
nm and of (b) cleaned surface with a roughness Ra= 882 nm
Laser cleaning was aimed to remove surface contaminants, the surface coating and the
oxide layer. The presence of lubricants and contaminants usually makes the surface
microscopically uneven. The removal of these layers reveals the plane surface obtained
after rolling processes. Such effect is also demonstrated in Figure 7.5 where the cross
section of laser cleaned surface is smoother and more uniform compared with that of the
uncleaned surface.
(a)
(b)
Figure 7.5 Cross sections of (a) not cleaned surface and (b) a laser cleaned surface.
In addition to the reduction in roughness, laser cleaning leads to a small change in the
microstructure of the interfacial layer as shown in Figure 7.6. It can be noted from this
figure that magnesium silicide precipitants that exist in the surface layer are much
smaller in size than the precipitants located deeply in the bulk material. This can be
attributed to the rapid heating and cooling induced by the laser pulses which do not
allow enough time for the precipitates to grow in comparison with those inside the bulk
material.
The prepared surface was also examined using scanning electron microscopy (SEM) to
understand whether any part of the surface was melted. This is shown in Figure
165
7.6 and Figure 7.7 which highlight the difference between cleaned and uncleaned
surfaces.
Figure 7.6 SEM image of surface characteristics in which the bottom part is laser
cleaned while the top portion is kept as a reference
According to the SEM pictures of the as-received and laser cleaned surfaces in Figure
7.6 and Figure 7.7, it is observed that the surface layer is melted during the course of
laser cleaning process. Before laser cleaning, lubricant was distributed on the surface in
the same direction to the rolling direction. However, laser cleaning as shown in Figure
7.7 caused the surface to be melted, and it formed a solidified material on the surface.
Dimogerontakis et al. (2005), who studied Nd:YAG laser cleaning of anodised Al-Mg
alloy using nanosecond pulses, specified the threshold at which laser induced surface
oxidation could occur. Their experiments showed that the fluence at which oxidation
might take place on the surface was in the range 0.6–1.4 J/cm2. The fluence according to
the parameters listed in Table 7.3 can be calculated from:
F
Pave
600
J

 6.27 2
4 2
f pulse  A 20000    (390 10 )
cm
( 7-2 )
where F is the fluence (energy density), Pave is the average power, fpulse is pulse
frequency, and A is laser spot area.
166
(a)
(b)
Figure 7.7 SEM images of surface characteristics (a) not cleaned surface of the treated
AC170-PX and (b) the same surface after laser cleaning
To observe the variation in surface chemistry before and after cleaning, laser-induced
breakdown spectroscopy (LIBS) technique was used. Figure 7.8 and Figure 7.9 show
the surface chemistry before and after cleaning, respectively. From Figure 7.8, it is clear
that as-received surface contains Al, Zr, and Ti as it is coated with Zr and Ti. Figure
7.9 shows that Zr peak as some of Ti peaks has disappeared in comparison with Figure
7.8. The high fluence of the pulsed laser is able to remove about 19 μm of the surface of
the Al alloy. This causes the Ti/Zr coating that is adhered to the top of the surface layer
to be removed during laser cleaning since the Ti/Zr coating is only 3 nm in thickness,
together with the lubricants. However, Figure 7.8 indicates that some titanium particles
still exist on the laser-cleaned surface although other Ti peaks in the range 400–600 nm
have disappeared. It appears that laser-cleaning has removed most of the coating and
contaminants.
167
Figure 7.8 Elemental analysis of as-received AA6014 surface
Figure 7.9 Elemental analysis of laser-cleaned AA6014 surface
7.4.2
Porosity characteristics after laser welding with and without laser precleaning
Pores after laser welding can be generated from various sources such as the hydrogen
gas, contaminants and lubricants on the surface, oxidation layers and the shrinkage of
the molten material during solidification. Therefore, elimination of porosity formation
168
sources such as contaminants, lubricant and oxidation can reduce gaseous porosity to
very low levels.
Figure 7.10 and Figure 7.11 show cross sections of the fillet edge and flange couch
joints with and without laser cleaning, respectively. It is noted that the flange weld
joints have less porosity compared to those of fillet edge joints due to the presence of
the gap at the root of the weld. The gap that exists due to the flange joint configuration
helps the gases generated during welding to escape from the welding pool (Pan and
Richardson, 2007; Mäkikangas et al., 2007; Gualini, 2001).
Fillet Edge Joints
(a)
(b)
20
2
35
3.5
50
5
Welding
Speed
[mm/s]
Figure 7.10 Comparison in porosity level between laser fillet edge welded joints (a)
without cleaning and (b) with laser cleaning
169
Power
[kW]
Flange couch joints
(a)
Welding
Speed
(b)
20
2.4
35
3.5
Power
[kW]
[mm/s]
50
5
Figure 7.11 Comparison in porosity level in laser flange couch welded joints (a)
without laser cleaning (b) with laser cleaning
By comparing the cross sections of the as-received and cleaned samples (Figure
7.10 and Figure 7.11), it becomes evident that laser cleaning reduced porosity
significantly in the weld zones. Porosity percentages were specified as a ratio of the area
occupied by the pores to the total cross-sectional area of the weld fusion zone.
According to graphs shown in Figure 7.12, porosity is reduced after cleaning by about
95–98% for the fillet edge welds depending on welding parameters used in the process.
Before cleaning, porosity reached values between 10% and 80% of the cross section of
170
the weld area, while after cleaning it was reduced to less than 1% regardless of welding
parameters. For flange couch joints, porosity was also decreased from 56% to 75% to
reach values of 0.23–0.8% after cleaning compared with 0.7–4.3% before laser cleaning
depending on welding parameters.
Nanosecond Nd:YAG laser cleaning used in this work has removed most of the
coatings, lubricants and contamination layers from the substrate surface. This reduces
the sources of harmful gases that cause porosity in the welds.
(a)
(b)
Figure 7.12 Porosity percentages in laser welded joints with and without laser cleaning
(a) for fillet edge joints, (b) flange couch joints
7.4.3
Weld dimensions
The weld's geometry was characterised by the weld width and penetration depth. These
two parameters are measured and presented for the fillet edge and flange couch joints as
shown in Figure 7.13 and Figure 7.14, respectively. It can be seen from these figures
that weld fusion zone width and penetration depth are greater prior to laser cleaning
than after cleaning.
For the fillet edge joints, weld fusion zone width and the penetration depth were
reduced by about 30–65% and 8–50%, respectively. However, the change in weld width
and penetration in flange couch joints was only about 1–3% and up to 25%,
respectively. This can be attributed to two reasons: less porosity is generated after
cleaning which in turn reduces the volume of the bead and its dimensions; and the
increase in reflectivity of the cleaned surface after cleaning.
171
Figure 7.13 Comparisons in weld width and penetration depth between a laser-cleaned
fillet edge joint and welds without laser cleaning
Figure 7.14 Comparisons in weld width and penetration depth between laser-cleaned
flange couch joints with and without laser cleaning
Laser cleaning removes the lubricant and the oxide layer and exposes the base material.
The cleaned Al surface becomes brighter than the as-received surface and hence would
reflect more laser beam energy. To prove this, the reflectance of the as-received and
cleaned surfaces was measured using an SPECORD 250 spectrometer. The reflectance
is measured for a wavelength range between 500 nm and 1100 nm as shown in Figure
7.15. The reflectance value of the as-received surface at the welding laser wavelength
(1064 nm) reached approximately 70%. On the other hand, the reflectance of the
cleaned surface at the same wavelength reached up to 85%. Accordingly, the reflectance
of the surface has increased by about 20% after laser cleaning
Therefore, a larger portion of the incident laser beam will be reflected by the cleaned
surface. This, in turn, means that less energy would be delivered to the base material
172
after laser cleaning during laser welding and hence producing welds with smaller
dimensions.
(a)
(b)
Figure 7.15 Reflectance of (a) the un-cleaned AC-170PX surface and (b) cleaned
surface using the Nd:YAG laser (1.06 µm)
7.5
Discussion
It is useful to understand the thermal penetration depth for each laser pulse delivered.
This characteristic thermal penetration depth is given by the following approximate
equation (Steen and Mazumder, 2010):
z  2 D.t
( 7-3 )
Where D [cm2 s-1] is thermal diffusivity (0.85 cm2 s-1 for 6000 alloys), and t [s] is the
pulse duration (100 ns in this work).
z  2 0.85 10-7  5.8 10-4 cm  5.8 m
( 7-4 )
Using simple maths, the number of pulses irradiated per spot can be easily found using
the pulse frequency, scanning speed and the spot size values in Table 7.3 as follows:
N pusle 
spot diameter
0.78
 f pulses 
 20000 164 pulse/spot
scanning speed
95
173
( 7-5 )
It should be noted that the above estimation is for 6000 series Al alloys. In reality, the
laser beam first interacts with the contaminants.
The change in the surface temperature, assuming a uniform laser beam, can be
estimated using (Steen and Mazumder, 2010):
T 
2I0 
k
D.t
( 7-6 )

Where I0 is the power intensity, β =(1-R) and R is the surface reflectivity (70% as shown
in Figure 7.15(a)), k is thermal conductivity of AA6016 alloy (175 W/m.K as given in
Table 7.3). For one pulse and taking the reflectivity value from Figure 7.15, equation (
7-6 ) yields:
2  6.27 1011  0.3 0.85 104 107
T 
 3236 C
175

( 7-7 )
If the room temperature is assumed to be 25 °C, the surface temperature will be 3261°C
which is greater than the alloy boiling temperature (2520°C). It should be noted that this
heat is generated only by one pulse of 20000 pulses in a second. The heat accumulated
due to the sequent pulses generates a very high temperature that results in vaporisation
of the surface material.
From the above calculation, it can be seen that the surface temperature has reached
more than the boiling point of the Al alloy. Thus the material removal mechanism
would be based on vaporisation.
Hydrogen solubility in Al differs significantly when Al is transformed from solid to
liquid. It was found that H2 solubility at melting temperature can reach up to 70 times
greater than the solid solubility for Al, while it is only 1.6 times higher than solid
solubility for iron (ASM International, 2004b).
Hydrogen can be generated from the alloys material itself, moisture in the atmosphere
and the shielding gas, contaminants on the metal surface, moisture in shielding gas
174
tubes, and oxide layers. The following reaction explains how hydrogen can be generated
from moisture (ASM International, 2004b):
2 Al  3H 2O  Al2O3  3H 2
( 7-8 )
Although laser cleaning helped to significantly reduce porosity in the studied Al alloys,
there is still a small percentage of porosity in both types of welds after cleaning. This
can be attributed to many sources (Dimogerontakis et al., 2005; Couso and Gómez,
2012):
1. A tiny layer of oxide that might have been formed after laser cleaning at fluence
(6.27J/cm2). This thin layer may generate a small percentage of porosity.
2. Ti particles that may still exist at the surface.
3. Hydrogen that already exists in the alloy composition.
4. A small amount of the shielding gas applied during the welding process.
5. The keyhole collapse and instability during the welding process.
It is worth mentioning that the automotive manufacturers set the maximum porosity
level allowed in Al welding to 10% of the weld cross section area (in accordance with
BS EN ISO 13919-2:2001 standards). In addition, they require a minimum value of
25% of the sheet thickness for the penetration depth, as well as a value of the weld
width equal to the sheet thickness. According to Figure 7.12, Figure 7.13 and Figure
7.14, laser cleaning allowed the authors to meet the industrial requirements in terms of
porosity level and weld dimensions values and hence enabled Al welding to be applied
for car body manufacture
7.6
Summary
According to the results reported in this work, the following conclusions can be drawn:
1. Laser cleaning was found to remove most of the lubricant and contaminants
placed on the surface of AC-170PX Al alloys. Moreover, laser cleaning removed
about 19 μm from the substrate at the used cleaning parameters.
175
2. Depending on welding parameters, laser cleaning was found to achieve a
significant improvement in the weld quality by reducing porosity level to less
than 1% for both types of joints.
3. It is believed from this work that laser cleaning reduced porosity during the
welding of coated AC170 PX Al alloy due to the elimination of hydrogen and
other gases produced from the coating, lubricant and the contaminants on the
surface.
4. Weld dimensions were reduced when AC-170PX sheets were cleaned due to the
increase in reflectance of the surface after cleaning.
176
 Brief Introduction
After the successful reduction in porosity had been achieved using laser cleaning prior
to welding, the author identified the need to investigate the mechanism of laser cleaning
which eliminates porosity generation during Al alloys laser welding.
Due to the limitations of the experimental procedures that can be used to monitor the
cleaning and welding processes, numerical modelling was chosen as a study tool for
understanding the mechanism of laser cleaning. SPH method was utilised for this
purpose due to its lagrangian nature which allows the user to track each computational
element (particle) independently. This useful characteristic can be used to understand
the ejected material behaviour during laser cleaning.
The next chapter introduces a novel 3-D multi-phase SPH model for laser
cleaning/ablation and presents a systematic methodology for validating the model
results.
177
Publication Title:
Smoothed Particle Hydrodynamics (SPH) modelling of transient
heat transfer in pulsed laser ablation of Al and associated freesurface problems
Authors:
A.W. AlShaer, B. D. Rogers and L. Li
Journal:
Computational Materials Science
Volume, Issue and Pages: 127 (2017) 161–179
Publication No.:
http://dx.doi.org/10.1016/j.commatsci.2016.09.004
STATUS: PUBLISHED
178
8 CHAPTER 8
SMOOTHED PARTICLE
HYDRODYNAMICS (SPH) MODELLING OF
TRANSIENT HEAT TRANSFER IN PULSED
LASER ABLATION OF AL AND
ASSOCIATED FREE-SURFACE PROBLEMS
8.1 Abstract
A smoothed particle hydrodynamics numerical model is developed to simulate pulsedlaser ablation processes for micro-machining. Heat diffusion behaviour of a specimen
under the action of nanosecond pulsed lasers can be described analytically by using
complementary error function solutions of second-order differential equations.
However, their application is limited to cases without loss of material at the surface.
Compared to conventional mesh-based techniques, as a novel meshless simulation
method, SPH is ideally suited to applications with highly nonlinear and explosive
behaviour in laser ablation. However, little is known about the suitability of using SPH
for the modelling of laser-material interactions with multiple phases at the micro scale.
The present work investigates SPH modelling of pulsed-laser ablation of Al where the
laser is applied directly to the free-surface boundary of the specimen. Having first
assessed the performance of standard SPH surface treatments for functions commonly
used to describe laser heating, the heat conduction behaviour of a new SPH
methodology is then evaluated through a number of test cases for single- and multiplepulse laser heating of Al showing excellent agreement when compared with an
analytical solution. Simulation of real ablation processes, however, requires the model
to capture the removal of material from the surface and its subsequent effects on the
179
laser heating process. Hence, the SPH model for describing the transient behaviour of
nanosecond laser ablation is validated with a number of experimental and reference
results reported in the literature. The SPH model successfully predicts the material
ablation depth profiles over a wide range of laser fluences 4-23 J/cm2 and pulse
durations 6-10 ns, and also predicts the transient behaviour of the ejected material
during the laser ablation process. Unlike conventional mesh-based methods, the SPH
model was not only able to provide the thermo-physical properties of the ejected
particles, but also the effect of the interaction between them as well as the direction and
the pattern of the ejection.
Keywords: Smoothed Particles Hydrodynamics (SPH), Heat Conduction, Kernel
Correction, Free surface, Laser Ablation, Aluminium.
8.2
Introduction
Lightweight alloys such as Al alloys usually have a layer of oxide or hydroxide on their
surfaces. These surface layers may deteriorate the properties of the joints after fusion
joining due to hydrogen entrapment in the fusion zones (Feliu et al., 2009). Laser
ablation is usually considered to be one of the dry cleaning methods that lead to
particles or film ejection from the surface using high-power laser pulses. Ejection of the
material from the surface is achieved due to the sudden expansion of the surface
particles or by evaporation (Yong-Feng et al., 1998; AlShaer et al., 2014). Laser
ablation/cleaning was used as one of the successful applications for weld joint surface
preparation for Al6014 alloys and Ti6Al4V alloys which have been used for automotive
and aero-engine manufacture respectively (AlShaer et al., 2014; Turner et al., 2007).
In order to understand and further improve the process parameters, different modelling
approaches have been developed either analytically (Zheng et al., 2001; Lu et al., 2001;
Song et al., 1996; Oltra et al., 1996; Bloisi et al., 2006; Marczak et al., 2005; Kelly,
1990; Kudryashov and Allen, 2006; Luk‘yanchuk et al., 2003; Arnold, 2003; Arnold et
al., 2001), or numerically (Luk‘yanchuk et al., 2003; Colina et al., 2011; Autrique et al.,
2013; Bityurin et al., 2003; Arronte et al., 2003; Liyang et al., 2012; Lutey, 2013c).
Conventional numerical modelling methods such as finite elements (FE), finite
difference (FD) and finite volume (FV) may not be able to predict complex processes
180
when multiple materials and phases are interacting due to the connected mesh that
describes the computational domain. Due to the fact that all the elements in the mesh are
connected to each other and cannot leave the mesh, the splashed particles in laser
cutting, drilling and ablation cannot be observed in such methods. Some authors (Lutey,
2013c; Marimuthu et al., 2013; Shalahim et al., 2010) simulated the protrusion formed
by the laser ablation/machining by deleting the elements that have their temperatures
greater than the boiling temperature. However, such an approach fails to provide
information on the behaviour of ejected material. Moreover, FE methods may not be as
accurate as required when the studied domain is of micro- or nanometers scale where
the element size should be less than the droplet or the ejected material size by several
times (Gross, 2008a).
Smoothed particle hydrodynamics (SPH) can simulate problems with highly non-linear
deformation such as complex movement of multi-phase fluids and topology changes.
SPH is one of many meshless (mesh-free) techniques that are based on the Lagrangian
description of motion. SPH has proven its ability to model various physical phenomena
such as fluid flows, heat and mass transfers, and elastic-plastic deformation (Liu and
Liu, 2003; Monaghan J.J., 2005). In SPH, the computational domain is divided into
arbitrarily distributed points called particles, which move independently from each other
eliminating the time and resource-consuming calculations of spatial derivatives
associated with mesh-based methods (Muhammad et al., 2013). SPH was initially
developed in 1977 by Lucy (1977) and Gingold and Monaghan (1977) to capture
astrophysical phenomena in which boundary conditions are neglected when an open
domain is under consideration. Later, by introducing boundary conditions, the SPH
method has enabled the simulation of engineering problems such as the flow of water
waves (Dalrymple and Rogers, 2006; Rogers and Dalrymple, 2008), metal forming
(Cleary et al., 2006) and phase change problems (Zhang et al., 2007; Xiong and Zhu,
2010). SPH can be used for modelling fusion processes in which liquid or vaporised
metals can be handled easily.
SPH modelling of laser processing is still in its early stage and needs significant
development since very few (less than ten) publications have been conducted in this
field. Demuth et al. (2012) simulated laser interference patterning of metallic surfaces
using SPH, following Tong and Browne (2011) who modelled laser spot welding of Al
181
with a very primitive model and limited resolution. Gross’s (2008b) model on laser
cutting of metals was further developed by Muhammad et al. (2013) who simulated
both dry and wet cutting of stainless steel stents used for medical applications, while
Yan et al. (2011) modelled CO2 laser underwater machining of alumina ceramics. Chen
and Beraun (2001) presented their preliminary work on SPH modelling of ultrashort
laser pulse interactions with a metal film, in which electron heating and cooling
phenomena had to be taken into account. Cao and Shin (2015) simulated the phase
explosion phenomena in laser ablation of Al and Cu at very high laser fluences of up to
36 J/cm2. The authors presented the ejected material’s size distribution during the
ablation, but their model did not predict the ablation depth for Al at different fluences
and did not show the simulation progression to the end of the ablation process although
the material’s temperature at that time was still above the boiling point and more
material was still to be ejected. Hence, the authors did not show the end of the ablation
process at which the final ablation depth must be measured and compared with the
experiments. Moreover, the proposed SPH model was not able to predict the entire
process but relied on molecular dynamics and hydrodynamics models to calculate the
initial position of the ejected material, such that the SPH model could not simulate the
interaction between the vaporised material and the solid substrate.
Laser material processing depends greatly on the temperature distribution that
determines the phase change locus and the change in temperature-dependent properties
of the material. None of the previous investigations on SPH modelling of laser heating
evaluated the accuracy of temperature values which were mainly used for calculating
the melt ejection velocity, kerf width and depth. Cleary and Monaghan (1999) and
Jeong et al. (2003) studied pure heat conduction problems using SPH and considered
the temperature dependence of material properties using a Dirichlet boundary condition.
Although the authors investigated different geometries and initial temperature profiles,
they only studied the steady-state problem and did not include any heat source terms in
their formulation. Additionally, all geometries studied were 2-D closed domains with no
treatment of the free surfaces where the adiabatic condition applies as in laser heating of
metal surfaces. Hence, there is a need to investigate the accuracy of transient solutions
of problems with rapid heating.
182
In the present research, thermal conduction of an Al sample was investigated using 1-D
and 3-D SPH models to evaluate SPH particles’ behaviour in laser heating of a metallic
free surface. Since the temperature distribution follows an error function distribution,
the temperature gradient across the sample was compared with the analytical solution
and the effect of two kernel correction methods (used to compensate for potential errors
at the free surface) were evaluated. Subsequently, a 3-D SPH model is developed to
further understand the laser ablation process after comparing the modelling with the
analytical solutions in a quasi 1-D model for heat conduction. The results of the
simulation were validated with both experimental and numerical data reported in the
literature.
8.3 Model Description
8.3.1
Physical Phenomena
8.3.1.1 Heat transfer governing equations
The mechanism of material removal in laser ablation of metallic alloys will be mainly
based on thermal ablation processes when nanosecond laser pulses are used (Turner et
al., 2007). Figure 8.1 shows a schematic of a typical laser ablation/cleaning process.
Analytically speaking and taking into account the external heat sources, the differential
equation of the heat transfer can be expressed as follows:
cp
dTi
   k T   Q  Qv
dt
( 8-1 )
where k [W/mK] thermal conductivity, T [K] Temperature, Q [W/m2] the heat source,
Qv heat loss due to convection.
The heat loss Qv can be formulated using (Muhammad et al., 2013):
Qv  hc Ts  T0     Ts4  T04 
( 8-2 )
183
where hc =20 [W/m2K] convection factor, ɛ=0.09 the emissivity, σ=5.67x10-8 [W/m2K4]
the Stefan-Boltzmann constant, Ts and T0 are the surface and the initial temperatures
respectively.
Vaporised Material
Scanning direction
Laser Beam
Sample
Un-processed
Laser ablated
Surface
Surface
Z
X
Metal Sample
Figure 8.1 Simplified schematic laser ablation of a metallic sample
The penetration of the electromagnetic wave (laser light) into the material is neglected
in this work since the particles’ spacing is significantly larger than the optical
penetration depth (1/D), where: D is the optical absorption coefficient for Al (1.23 x108
m-1 (Bäuerle, 2011)). This means that the electromagnetic wave will be absorbed only
by the first layer of particles and the subsequent layers will be mainly heated by
conduction
Using the Beer-Lambert law, the laser intensity at depth z can be calculated as follows
(Bäuerle, 2011):
3
 D
I ( z)
D

e
 0.049
I0
( 8-3 )
184
where I(z) is the laser intensity (power per square area) at depth z, I0 is the laser
intensity at the surface and z is the depth. If z is assumed to be three times the optical
penetration (z= 3/D), then:
I ( z )  I 0 e D z
( 8-4 )
Hence, the intensity will be reduced to about 5% of the original intensity within 24.4 nm
from the surface, which is significantly smaller than the particles’ spacing used in the
simulations. Using the same calculations, the laser intensity at 250 nm depth will drop
to about 0.004% of the nominal intensity due to the attenuation and the electromagnetic
wave will not reach the second layer of particles. Therefore, the optical penetration of
the laser beam can be neglected.
8.3.1.2 Vapour Pressure
The vapour pressure produced by laser ablation of metal can be calculated using the
Clausius-Clapeyron equation depending on the surface temperature (Atkins and De
Paula, 2002):
 L  T T 
Pvap  Patm exp  v  s b  
 R  Ts Tb  
( 8-5 )
where Patm, Lv, R, Tb and Ts are atmospheric pressure, latent heat of vaporisation, the gas
constant, the boiling temperature and the surface temperature respectively.
8.3.1.3 Vapour velocity
Assuming that the vapour particles are expelled from the surface with a onedimensional Maxwellian velocity, the vapour velocity can be approximated using the
average velocity in the normal direction at temperature Ts as follows (Kelly, 1990):
vvap , s 
2 k B Ts
m
( 8-6 )
185
where kB is Boltzmann constant (1.38 × 10-23 m2 kg/ s2 K), and m is the atomic mass. It
should be noted that this equation gives a good approximation when the experiments are
conducted in a vacuum or when the vapour pressure is much greater than the
surrounding pressure.
8.4
SPH Methodology
8.4.1
SPH Interpolation
In SPH, the computational domain is divided into arbitrarily distributed particles where
each has its unique properties including mass mi, volume ωi, pressure Pi, and velocity vi
(Muhammad, 2012). The value of a function A(r) at location r can be found by a local
interpolation for a set of surrounding particles at a specific time step. In continuous
form, the interpolated value of the function can be estimated (Liu and Liu, 2003):
Ai (r )   W  r  r ', h  A  r ' d 
( 8-7 )

where r is a position vector,
denotes approximation, W is the weighting function
called the smoothing kernel, h is the smoothing length (a characteristic length scale of
the kernel). For the interaction between two SPH particles i and j the smoothing kernel
can be written in a general form as follows (Morris et al., 1997):
Wij  W  rij , h  
1  ri  rj

hn  h





( 8-8 )
where n is the number of spatial dimensions, f is a function of h and rij  ri  rj the
distance between two particles i and j.
The function A(r) can be written in discrete SPH form as follows:
Ai (r )   m j
j
Aj
j
Wij
where mj and ρj are the particles mass and density respectively.
186
( 8-9 )
Accordingly, the gradient of the considered function can be calculated by taking the
kernel gradient into account in the approximation:
Ai (r )   m j
j
Aj  Ai
j
iWij
( 8-10 )
To conserve momentum in SPH, a slightly different form is used to calculate the
pressure gradient as shown later in Equation ( 8-22 ):
Ai (r )
i
 A Aj 
  m j  2i  2  iWij


j
 i j 
( 8-11 )
Equation ( 8-10 ) is used for divergence operators, such as in the conservation of mass
equation introduced below, while Equation ( 8-11 ) is used to ensure an equal and
opposite reaction between particles as in the case of calculating the pressure gradient.
The latter formula conserves linear and angular momentum (for further information see
(Liu and Liu, 2003; Mayrhofer et al., 2013))
The smoothing kernel can take different forms; Gaussian, Quadratic, cubic spline (Bspline), or higher order kernels fourth & fifth order. The cubic spline was selected to be
used in our model since it approximates the Gaussian function and is commonly used
throughout SPH. The cubic spline can be expressed by the following equation (Liu and
Liu, 2003):
 3 2 3 2
1  2 q  4 q

3
1
W  rij , h    D   2  q 
4
 0


0  q 1
1q 2
( 8-12 )
2q
where q  rij h , αD is a normalisation parameter to ensure the unity integral of the
kernel, with a value of 2h 3 for 1-D model and 1  2 / 3h3  for 3-D. Figure 8.2 shows
particle i and its neighbour particles j within a smoothing kernel of a radius 2h.
187
i
j
Figure 8.2 Particles i and j within the smoothing kernel support
8.4.2
Kernel gradient correction
The kernel shown in Figure 8.2 is said to have complete support since the considered
particle, i, is surrounded by particles which contribute to its interpolation during the
simulation. However, there are cases where the kernel support is incomplete such as
when the particle approaches the boundaries or when the particle is at a free surface.
Different measures have been introduced to SPH to compensate for such loss of support
(Monaghan and Kos, 1999; Yildiz et al., 2009; Fourtakas et al., 2015), such as kernel
gradient correction KGC (Liu et al., 1997) and the Schwaiger operator (Schwaiger,
2008) that were developed to correct the variables’ gradients and the second derivatives
respectively. This is potentially important for laser processing where the specimen is
heated by the laser acting on a free surface with the temperature evolution controlled by
thermal diffusion which is described mathematically using a second-order diffusion
term (Equation ( 8-1 )).
The kernel gradient correction method replaces the normal kernel gradient ∇iWij by a
~
corrected kernel gradient i Wij expressed using the following set of equations (GómezGesteira et al., 2012):
~
 Wij  LiiWij
( 8-13 )
Li  M i1
( 8-14 )
188
num
Mi  
mj
j
j
iWij   xi  x j 
( 8-15 )
The corrected gradient is therefore given by:
Ai (r )
KGC
  mj
j
Aj  Ai
j
~
i Wij
( 8-16 )
To correct the values of the second derivative, a modified Laplacian operator should be
used along with the corrected kernel gradient according to the following approximation
(Schwaiger, 2008):
 .f i 
tr   
n
1

 fi  f j
 mj
  i   j  
2
 j j
 rij

 ~
 mj ~


 rij i Wij    i fi   fii  ifi  .   i Wij  

j
j




( 8-17 )
where µ is a physical constant such as a diffusion coefficient or thermal conductivity, n
is the number of dimensions, f is the value of the function, and Γ is a tensor:
   
rij .Wij
rij2
x x
( 8-18 )
where subscripts β, γ denote the computational domain directions.
As known, the second derivative (Laplacian) of a temperature is an essential quantity in
the heat conduction Equation ( 8-1 ) to calculate the rate of change in temperature over
time, and hence to calculate temperature values at a specific time.
8.4.3
Governing Equations and SPH Discretisation
SPH uses a Lagrangian formulation for the SPH particles moving in space and time. In
this work, the specimen will be modelled as a viscous fluid. The Lagrangian form of
conservation of momentum and continuity equations for fluids can be expressed from
the general form of Navier-Stokes equations as:
189
Dv
1
  P  . 2v  g
Dt

( 8-19 )
D
  .v
Dt
( 8-20 )
where v is the velocity, P is pressure,  is viscosity and g   0,0, 9.81 m.s 2 is the
gravitational force.
The previous Equations ( 8-19 ) and ( 8-20 ) can be discretized in SPH as (Monaghan,
1992):
mj
d i
  j 
vij .iWij
dt
j j
( 8-21 )
 Pi  Pj
dvi
  m j 
 
dt
j
 i j
( 8-22 )

 iWij

where vij  vi  v j . Note that Equation ( 8-22 ) uses the anti-symmetric form of the
gradient ( 8-11 ) to conserve momentum.
In this work, the viscosity introduced by Monaghan (1992) was used to eliminate any
unphysical instability in the model (Liu and Liu, 2003). These forces were expressed by
inserting artificial viscosity into the momentum equation to become:
 Pi  Pj

dvi
  m j 
  ij  iWij  g
 

dt
j
 i j

( 8-23 )
where  ij is the artificial viscosity:
190
  cij ij

 ij   ij
 0

ij 
cij 
vij .rij  0
( 8-24 )
vij .rij  0
h vij .rij
( 8-25 )
rij2  ij2
1
 ci  c j 
2
,
ij 
1
 i   j 
2
( 8-26 )
In order to simplify the model, α was selected “5.0” to simulate the solid phase as a high
viscous liquid, and “0.1” for the liquid phase.
Additionally, the pressure in this model was expressed by Tait’s equation of state
(Monaghan et al., 2005):
   
P  B    1
 0 

where B 
( 8-27 )
0 c02
, 0  1000 kg / m3 is the reference density, c0 speed of sound, and

  7 is a constant.
To simulate the thermal behaviour of particles, an SPH form of the heat conduction
Equation ( 8-1 ) was introduced into the model to include the laser beam heating of the
surface (Monaghan et al., 2005):
cp
mj
dTi

dt
j i  j
 4ki k j

 ki  k j
 Ti  T j

 r 2
 ij

 rij iWij  Q  Qv

( 8-28 )
In laser ablation processes such as laser cleaning, the laser beam acts only on the free
surface of the sample and the attenuation of the electromagnetic field into the material is
neglected as discussed in section 8.3.1.1.
191
In SPH, the location of free surface can be determined by computing the divergence of
the particle position using the following equation (Muhammad, 2012):
.r  
j
mj
j
rij . Wij
( 8-29 )
The truncated kernel support of any surface particle gives a non-zero value for the
particle position divergence. In 3-D cases, an empirical value .r  2.4 was used to
indicate a free-surface particle (Lee, 2007; Lind et al., 2012).
8.4.4
Time-step scheme and CFL number
The Predictor-Corrector scheme was used in this work to evaluate the parameters over
time. In this scheme, the variables values are calculated in time as follows (GómezGesteira et al., 2012):
t n
t
Fi ; in 1/ 2  in  Din
2
2
t
t
 rin  Vi n ; ein 1/ 2  ein  Ein
2
2
t
Ti n 1/ 2  Ti n  Ti n
2
vin 1/ 2  vin 
ri n 1/ 2









( 8-30 )
where n superscript denotes the current time step.
The values are then corrected at the half step and finally calculated at the end of the
time step as follows:
vin 1  2vin+1/2  vin ; in 1  2 in 1/ 2  in
ri
n 1
 2ri
n+1/2
Ti
n 1
n 1
i
n 1/ 2
n 1/ 2
i
- ri ; e
 2e
 2Ti
 Ti
n
n
e
n
i





( 8-31 )
Usually, a variable time step is selected to replace the constant time step when some
physical variables vary drastically over time. This requires a change in time-step
according to CFL (Courant–Friedrich–Lewy) condition, the forcing terms, the viscous
diffusion term (Monaghan and Kos, 1999) and importantly for the application presented
192
herein the thermal diffusion term (Monaghan et al., 2005). The variable time step can be
calculated according to:

t  CFL .min  t f , tcv , t D  ; t f  min 


tD 
h
fi

;




h

tcv  min 
 c  max h vij .rij
j
 s
rij2








( 8-32 )
 c p x 2
( 8-33 )
k
where ∆tf is based on the specific force, ∆tcv combines Courant and the viscous terms
and ∆tD is based on the thermal diffusion term.
8.4.5
Boundary Conditions
The model utilises dynamic boundary conditions in which the boundaries consist of
three layers of particles arranged in a staggered position. The boundary particles were
created to be fixed fluid particles with zero velocity to simulate the solid walls
surrounding the studied sample. The conservation equations are calculated for all
particles including the boundary, and this provides the interactions between the moving
and the boundary particles. More details can be found in (Crespo et al., 2007).
8.5 Results and Discussion
The laser ablation process involves repeated cycles of heating and cooling. For simple
cases (without ablation or phase change), these heating and cooling processes can be
modelled analytically (1-D) when the temperature distribution is known to be
characterised by an error function (Nath et al., 2012).
Before simulating the laser ablation process using a 3-D model, several steps have to be
completed in order to validate the results obtained from this model. As mentioned
earlier, the SPH interpolation procedure has incomplete support at or near a surface
which can introduce numerical errors. Firstly, a preliminary study was carried out to
determine whether or not any kernel corrections were needed at the sample surface
where the kernel support was incomplete. Secondly, a convergence study was
performed to specify the required resolution and the time step for the simulation.
193
8.5.1
Kernel Gradient Correction (KGC)
Using a 1-D SPH model, two test cases were performed to investigate the performance
of the SPH model by evaluating the errors in the functions’ gradient and the second
derivative at the free surface for functions encountered in heating. The gradient and the
second derivative were calculated for six different functions: three polynomial functions
(first, second and third order), hyperbolic, logarithmic and complementary error
functions. Since the temperature distribution for a surface heat source follows an error
function, this is of particular interest herein. The ordinary SPH results with no
corrections were examined and compared to the same cases with the Kernel Gradient
Correction (KGC) and Schwaiger operator correction. Figure 8.3 shows the different
functions plotted for a variable x=[1,5] m, using a particle spacing of 0.25 m, and
h=1.5Δx. The main aim of testing various functions with/without correction is to
observe the SPH model behaviour and sensitivity to approximating functions of
different orders, especially interpolating the error functions that describe the
temperature evolution during laser heating of metals. The analytical values of the
gradient and Laplacian can be easily obtained by differentiating the six functions once
and twice respectively.
Figure 8.3 Functions used to validate the kernel truncation correction at the free surface
in SPH
Figure 8.4 depicts the values of the functions’ gradient across the sample calculated
using an uncorrected SPH gradient (Equation ( 8-10 )) and a corrected kernel gradient
194
(Equation ( 8-16 )) both numerically and analytically. It is pertinent to mention that with
h=1.5Δx, the kernel support was truncated over three layers of particles starting from
each end of the domain corresponding to 2h=3Δx distance.
In order to quantify the error generated during the simulations, the L2 norm error
(Wolfram, 2016), that is used to evaluate the discrepancy in the SPH results from the
analytical solution, can be calculated as follows:
L2   
 
N
i 1
SPH
  Theory 
2
( 8-34 )
where  SPH is SPH temperature value,  Theory is the analytical value and N is the number
of particles.
As expected for the uncorrected SPH, Figure 8.4 shows that the agreement is
satisfactory away from the surface, but poor within 2h of the surface where the kernel
support is incomplete. Whereas, the kernel gradient correction succeeds in
compensating for the loss of support at the boundaries and produces an excellent
correlation with the analytical solution for the polynomial and the error functions. This
can also be seen in Figure 8.5 which shows that KGC eliminated the error for the linear
and quadratic functions and reduced the error for the cubic and the error functions
significantly. However, the influence of KGC on the logarithmic and hyperbolic
functions’ gradient was not as significant as on the other functions since their numerical
and analytical results are still far from each other at the boundaries.
195
Figure 8.4 Kernel gradient Correction (KGC) effect on SPH results at the free ends of
the 1-D domain
Although KGC was not able to eliminate the error for the logarithmic and hyperbolic
functions entirely, it managed to reproduce the analytical solution for particles located
at x = 0.5m achieving a gradient profile in closer agreement with the analytical solution
than the one produced without correction.
196
Figure 8.5 L2 Norm Errors in SPH results for the gradient of different functions
with/without Kernel Gradient Correction
Generally, the difference between SPH with/without correction results should clearly
appear at each end of the computational domain that suffers from a lack of support.
However, the values of the gradients in some cases (Figure 8.4 c, d, e and g) tend to
zero at one end due to the nature of the function, making the results of SPH
with/without KGC difficult to distinguish. Therefore, it is sufficient to evaluate the
efficiency of the correction method only at one end where the difference in the results is
evident.
According to the aforementioned discussion, the kernel gradient correction may be a
suitable method for correcting the gradients of physical parameters that follow linear or
error functions at the free surfaces such as thermal heating and cooling during laser
processing of metals.
8.5.2
Laplacian Operator Correction (Schwaiger correction)
Since the thermal behaviour is described by a second-order derivative (Equation ( 8-1 ))
which in SPH requires a combined use of two first-order operators (Equation ( 8-28 )),
another test case was conducted to examine the effect of the Schwaiger operator
with/without the KGC on the second derivative of the same functions. Figure 8.6 and
Figure 8.7 show the SPH results of the functions’ second derivative without and with
KGC respectively, while Figure 8.8 depicts the L2 Norm errors produced in each test
case.
197
Figure 8.6 Effect of Schwaiger operator without KGC on SPH results at the boundaries
of the 1-D domain
Figure 8.6 clearly shows that without KGC, the standard SPH second-order derivative
performs poorly near a free surface and the Schwaiger operator helps to reduce the error
but not eliminate it.
It can be seen from Figure 8.8 that Schwaiger operator without KGC has reduced the
errors in the numerical results by about 10% to 60% for all different functions.
However, the second derivative correction had less impact on the logarithmic and
hyperbolic functions than on the other four functions. This can be attributed to the large
difference between d2f/dx2 values in the first and the second particles caused by the
nature of the functions’ derivatives since they tend to infinity when x value tends to
zero. This, therefore, makes the drastic change in d2f/dx2 difficult to be captured using
SPH formulations without further corrections. Moreover, it is already known (Lee,
2007; Lind et al., 2012) that the kernel gradient correction should be coupled with the
Schwaiger operator in order to obtain satisfactory results for the second derivative.
198
After applying Schwaiger correction with KGC, Figure 8.7 shows that the second-order
derivative values improved significantly with a closer agreement with the analytical
solution for all functions. This can be justified in Figure 8.8 which indicates that the
error for the linear function was eliminated and the errors were reduced by about 70% to
90% for the quadratic, cubic and the error functions. However, less impact on the errors
for the logarithmic and hyperbolic functions can be observed with only 10% to 15%
reduction in comparison with Schwaiger correction without KGC.
From Figure 8.6,Figure 8.7 and Figure 8.8, it is important to note that this method of
correction achieved an excellent correlation with the analytical solution for the “Erfc”
function type that can be used to describe the temperature change during laser heating of
metals. However, any variations in the formulation of the Erfc function given in Figure
8.7 (g) require an evaluation as to whether or not a Schwaiger correction is needed. This
comes from the fact that the function formulation has a significant impact on the
function’s behaviour in space and time. The sign of the second-order derivative depends
on the positive direction of the axis of concern in Figure 8.6 and Figure 8.7. However,
the bottom end of the substrate in the 3-D simulation is supported by the boundary
particles limiting the kernel truncation problem to the free top surface. The large jump
in temperature can be modified by selecting the correct particle spacing and time step
(which will be done in Section 8.5.3.2) and the appropriate correction method (KGC,
Schwaiger) if needed. However, since the surface temperature is effectively specified by
the external heat source, Q, on particles that would otherwise have required these
corrections, this obviates the need for corrections. This matter is thoroughly discussed in
Section 8.5.3.2.
199
Figure 8.7 Effect of Schwaiger operator with KGC on SPH results at boundaries of the
1-D domain
Figure 8.8 L2 Norm Errors in SPH results for the second derivative of different
functions with/without correction
To conclude, the combination of the KGC and Schwaiger corrections was able to
correct the errors caused by the truncated support at the domain borders for the gradient
and Laplacian of the studied functions respectively. However, for non-linear functions
such as hyperbolic and logarithmic functions both correction methods were only able to
reduce the deviation at the boundaries without matching the analytical solution.
200
Nevertheless, these methods are essential for correcting the gradient and the second
derivatives of the polynomial and the error functions.
Therefore, it is recommended in SPH modelling to take into account the nature of the
functions that describe the physical phenomena, especially when modelling free
surfaces at which the kernel support is incomplete.
8.5.3
Transient Heat Transfer Test Cases
8.5.3.1 Numerical setup
In this work, a 3-D SPH model of 20 × 20 × 200 µm3 was created with 0.2 µm spacing
and a free surface on the top to simulate pulsed laser ablation of Al and its alloys (see
Figure 8.1). At the beginning of the simulation, the sample was assumed to be at room
temperature. The model was created using a modified SPHysics (Gómez-Gesteira et al.,
2012) serial code and was run using an Intel® Core i7 CPU (3.4 GHz) with 8 GB RAM
on an Ubuntu 14.4 LTS operating system.
Additionally, a 1-D SPH code was compiled using Matlab 2014a and the numerical
results were compared with the analytical solution of the 1-D heat transfer partial
differential equation (PDE). The 1-D analytical solution of the heat conduction PDE for
multi-pulses is given in Appendix B.
The pulsed laser was simulated during the validation studies using 100 ns pulse duration
(laser-ON), 100 ns relaxation time (laser-OFF) and a 9.6 x 1011 W/m2 laser intensity.
Two different reflectivity values were selected to cover the different surface conditions
that may be encountered when ablating Al alloys. Pure Al or polished Al alloys surfaces
have a very high reflectivity of about 95%, while oxidised, contaminated or coated Al
surfaces have lower reflectivity reaching up to 75% (AlShaer et al., 2014). The thermophysical properties of the base material are listed in Table 8.1 and are assumed to be
temperature-independent during the simulations.
201
Table 8.1 Aluminium alloy AA6014 thermo-physical properties (AlShaer et al., 2015a)
Density
3
ρ [kg/m ]
Thermal
Initial
Reflectivity
diffusivity
Temperature
2
R [%]
D [m /s]
[⁰K]
75% and 95%
6.89 x 10-5
300
Thermal Conductivity
Specific Heat
k [W/m.⁰K]
Cp [J/kg]
167
896
Emissivity
0.09
200 μm
2705
Surface Optical
Z
X
Y
20 μm
Figure 8.9 SPH computational domain (not to scale)
8.5.3.2 Convergence Studies
In laser-metallic interactions, the laser beam can only act on the metallic surface, and
the heat dissipates into the sample to the lower layers by heat conduction. The laser
removes, melts or evaporates the material from the surface. Accordingly, gaining the
correct temperature values at the surface is essential in understanding the laser thermal
ablation processes. With the results for different functions in section 8.5.2, a test case
was conducted to examine the temperature values at the surface of a metallic sample
where the laser beam heating is active over typical pulse duration of 100 ns.
As mentioned previously in sections 8.5.1 and 8.5.2, the effect of corrections is
dependent on the type and the behaviour of the function being evaluated at the surface.
The error function that was used in sections 8.5.1 and 8.5.2 was only a function of x and
did not vary with time. Therefore, error function cases that are dependent on variables
202
including time will show a different response to the truncation of the kernel at the
surface and hence their values should also be evaluated against the corresponding
analytical solution through new test cases.
The investigation in Section 8.5.2 showed that both kernel gradient correction and the
Schwaiger correction are necessary to obtain satisfactory agreement with the analytical
result for typical functions describing heat transfer, for example in the form of a
complementary error function. For surface laser application, since the surface
temperature is effectively specified by the external heat source, Q, on particles that
would otherwise have required these corrections, this obviates the need for both the
kernel gradient and Schwaiger corrections. Moreover, Figure 8.6 a, b, d and g show that
the second layer of particles has errors, but when the temperature of the surface particles
is determined by the applied laser, the error at the second layer of particles is also
reduced as will be demonstrated in the temperature profiles presented in section
8.5.3.3). For the analytical cases presented herein for pulsed lasers with heat loss, during
the very short laser-off period (100 ns), the surface particles transfer heat to interior
particles only which do not require the corrections. This case is different from the
internal (volume) heating in which the heat diffuses from the inside of the domain
towards the free surface at which the thermal boundary conditions and the kernel
support will be the main factor to determine the temperature values. As a result, no
corrections are applied in the laser ablation model.
The particles were kept stationary during the simulation to observe the heat transfer
behaviour of the solid particles and the temperature produced at the surface. It should be
noted that the model dimensions were selected to enhance the heat flow in one direction
(z-direction) in order for the results to be validated with a separate SPH 1-D code
results. A 3-D model will represent an Al rod being heated at one end and will be
referred to as “quasi 1-D” model. After validation, the dimensions will be changed to
reflect the 3-D aspects of the problems in the real applications as will be presented in
the “laser ablation model” section of this paper.
First, two convergence studies were conducted to determine the correct resolution
(particles’ spacing) and the time step. To determine the initial particles’ spacing, an
203
initial value of 5 ns for the time step was selected to calculate the thermal penetration
caused by pulsed laser heating of the surface:
z  D. t p  0.689 104  5  109  0.6  m
( 8-35 )
where D is thermal diffusivity, tp is laser pulse duration and z is thermal penetration
depth.
Equation ( 8-35 ) shows that the heat wave will travel 0.6 µm inside the sample within 5
ns of the heating time. Hence, a value of 1 µm (that is larger than 0.6 µm) was chosen as
an initial particle spacing to start the convergence study for the selected model. The
initial value of 1µm was selected to be of the same order of the calculated thermal
penetration depth and to generate an integer number of particles (20 particles) along the
smallest dimension in the computational domain, namely 20µm. To examine numerical
convergence, three different resolutions were used: 1µm, 0.5µm and 0.25µm
respectively.
Figure 8.10 shows the temperature evolution on the top surface using three different
resolutions for a single-pulse laser ablation, and using the analytical solution as given in
Appendix B. It is important to note that some of the literature on pulsed laser beam
heating used these equations to describe the temperature during multi-pulses heating by
only replacing T0 with the temperature value from the preceding pulse. This treatment is
incorrect due to the different boundary conditions associated with equations (B.10) and
(B.13) which are different from the conditions applied in multi-pulses heating i.e. the
temperature profile across the sample after the first pulse is not identical to the constant
temperature distribution across the sample at the beginning of the process. Therefore,
the temperature increase produced by each laser pulse should count for all preceding
heating and cooling cycles of the preceding pulses. Hence, the number of heating and
cooling terms in the previous equations will change accordingly for each pulse.
204
Temperature [K]
Figure 8.10 Surface temperature using different particles spacing for single Laser pulse
tp=100 ns
It can be seen from Figure 8.10 that reducing the particle spacing from 1 µm to 0.25 µm
led to smaller deviations of the SPH results with the analytical solution. This can be
attributed to the increase in the number of particles within the thermal penetration depth
(~3 particles at 0.25 µm spacing). Herein, to quantify the rate of convergence, the errors
are quantified using the L2 error norm since a uniform particle refinement ratio is used.
Figure 8.11 shows that reducing the spacing from 1 µm to 0.25 µm decreased the error
in temperature by approximately 85% from 20 K to only 3 K. Moreover, plotting the L2
error norm in Figure 8.11 indicates a first order convergence which is consistent with
the order of convergence calculated using the expression suggested by Roache (1997)
for a three-resolution system with constant refinement ratio (r):
f f 
log  3 2 
 f 2  f1   1.12
P
log  r 
( 8-36 )
where f3, f2, and f1 are the values of the temperature using the finest resolution to
coarsest resolution.
Taking into account the exact temperature values and the numerical results for the finest
resolution, the relative error and the Grid Convergence Index (GCI) for this case can be
calculated from:
205
 
f1  f exact
  0.0086
f exact
GCI 21  Fs
r

P
 1
( 8-37 )
 0.019
( 8-38 )
where Fs is a safety factor taken as “1.25” and is based on experience by applying GCI
in different applications (Roache, 1994).
The value of GCI indicates that the SPH results lie within a 1.9% deviation band from
the exact solution with a 95% confidence level.
Figure 8.11 L2 Norm Error for SPH results at different resolutions
In order to obtain good simulation results over time, the time step should be selected to
capture all parameters’ changes during the simulation, without increasing the CPU time
at no gain. It can be seen from the analytical solution in Figure 8.10 that the temperature
changes sharply at the beginning of the heating and cooling phases by about 150-200 K
within 3-4 nanoseconds i.e. about 50 K/ns. Using the material properties and the
numerical parameters, the set of equations ( 8-33 ) show that the largest time step to be
used in the simulation should be 0.9 ns in order to capture the sharp change in
temperature.
Figure 8.12 shows that the use of 0.5 ns time step with CFL number of 0.1 enables the
simulation to capture the drastic change in the surface temperature, especially at the
beginning of both the heating and the cooling phases. However, the selection of longer
206
time steps such as 1 ns and 5 ns destabilises the simulation and terminates it at the
beginning.
Figure 8.12 Effect of time step in SPH modelling using the step-predictor corrector
time scheme
8.5.3.3 Pulsed Laser Transient Heating
 Single Heating and Cooling Cycle
The ability of SPH to predict the cooling effect once the heat source is removed was
assessed by simulating a single pulse of 100 ns duration and a relaxation time up to 8 µs
as illustrated in Figure 8.13. These temporal values correspond to a 20 kHz pulse
frequency, which is commonly used in laser ablation processes as will be demonstrated
in the following sections. The 3-D model predicted the heating and cooling of the Al
sample precisely over time after the surface temperature reached more than 4500 K at
the end of the pulse. It is pertinent to mention that laser pulses heat and cool the material
rapidly as shown in Figure 8.13 in which the temperature dropped to less than the
melting point within less than 2 µs. As will be discussed in the following sections, rapid
heating and cooling are very beneficial in metal laser ablation because it leads to a
smaller heat affected zone HAZ and less distortion.
207
Figure 8.13 Single Laser pulse of 100 ns and 75% surface reflectivity on the Al target
using the 3-D SPH code
 Multiple Heating and Cooling Cycles: Surface Temperature
A second stage of validating the SPH model is to make a comparison between the
temperature evolution over time between the 1-D and the quasi-1D models comparing
with an analytical solution for cyclic heating. Figure 8.14 shows the surface temperature
change due to pulsed laser heating using two reflectivity values (a) 95%, and (b) 75%,
in which equal heating and relaxation periods were selected to observe the effect of
multiple pulses within a short time of laser heating. The correct temperature distribution
during the frequent pulses heating will generate the correct crater depth, ejected
material’s characteristics and its behaviour.
208
(a)
(b)
Figure 8.14 SPH modelling of three consecutive laser pulses of 100 ns pulse duration
and 100 ns relaxation time (a) 95% surface reflectivity (b) 75% surface reflectivity
From Figure 8.14, the 1-D model predicts the surface temperature with only 0.8% error
for both high and low reflectivity, while the 3-D model showed a slight deviation from
the theory by about 2.5% in the peak temperature at the end of each heating stage. This
small difference between the two models can be attributed to the existence of the
dynamic boundary (DB) particles on the sides of the computational domain in the 3-D
model. These particles were kept at the room temperature as would occur in the real
applications since this model will be later modified to simulate the laser ablation
process. These boundary particles slightly cool the adjacent particles by conduction
causing the surface temperature to drop by about 2.5% in comparison with the analytical
value. The DB particles existence is very important to imitate the solid Al medium that
surrounds either molten or vaporised matter that will be seen in the proceeding sections
of this paper. Although some of the accuracy is sacrificed by introducing the DB
particles, their physical significance justifies the need for them.
 Multiple Heating and Cooling Cycles: Temperature Distributions
Figure 8.15 shows (a) the temporal variation in temperature at different depths from the
sample surface, and (b) the temperature profile across the sample over time.
Additionally, Figure 8.16(a) and (b) plot the time derivative of temperature, dT dt (that
is the heating or cooling rates), at the surface and at different depths respectively.
During all heating phases in Figure 8.15(a), it is evident that the surface temperature
climbs rapidly to more than 800 K within 20-40 ns at the beginning of the pulse, which
209
corresponds to a very high heating rate (~61010 K/s) distinguished by the positive
value shown in Figure 8.16(a). At the start of the pulse, an instant high value of the
heating rate appears immediately due to the instant application of the laser pulse. After
each occasion that the laser is turned off, the heating rate reduces over time due to the
heat being conducted to the lower layers (shown at depths of 5 μm and 7 μm in Figure
8.15(a)) whose temperatures gradually increase, reducing the difference in the surface
temperature.
Figure 8.15 Temperature history during laser heating (a) Temperature variation over
time (b) Temperature profile across the sample
Figure 8.16 Heating/Cooling rates at the surface during three consecutive laser pulses
At the beginning of the cooling phases (t = 100, 300, 500, … ns), dT dt becomes
instantaneously negative indicating that only cooling is taking place at the surface and
therefore the surface of the sample is transferring the heat rapidly to the lower layers
without gaining or losing heat from or to any external sources. The cooling rate then
reduces with time since the lower layers’ temperature is tending to the surface
210
temperature to achieve thermal equilibrium. This is very clear from Figure 8.16(a)
where dT dt is tending to zero at the end of each cooling phase (at 200 ns and 400 ns)
and in Figure 8.15(a) the temperature at 2 µm depth is tending towards the temperature
of the surface. Moreover, it should be recognised that the peak temperature at 2 µm, 5
µm, and 7 µm are delayed relative the surface temperature by time shifts of 10-60 ns as
shown in Figure 8.15(a). This is due to the time needed for the heat wave to propagate
into the sample body, which is dependent on the thermal conductivity and other thermal
properties of the base material. Additionally, this peak also depends on the depth at
which the temperature is calculated, i.e. the deeper the layer, the greater the time by
which the peak is shifted.
8.5.4
Laser Ablation Model
With the satisfactory agreement of the SPH solution with a 1-D analytical solution, the
model is now applied to laser ablation cases. In order to evaluate the performance of the
SPH model for laser ablation prediction, different test results were compared against
published data on laser ablation of Al. To simulate the material ablation, the boiling
temperature of Al (2730 K) was set as a thermal criterion to eject the SPH particles from
the surface (Equation ( 8-28 )) assuming that a small portion of the heat delivered by the
laser is being wasted due to convection and radiation. It is important to mention that the
fluences used in those studies lay within the low to medium fluence ranges in
comparison with the high regime (order of 103 J/cm2) in which the phase explosion
(Xu, 2002) is the predominant mechanism in the process. The surface temperature at
high fluences may exceed the critical temperature for Al (~8000 K (Faussurier et al.,
2009)), at which the vapour phase volume breaks down and starts interacting with the
incident laser beam. This range of fluences is beyond the scope of this paper.
It can be noted that the ablated surface approximates the shape of a flat plane following
the same spatial distribution of the laser pulse (Top-Hat). If a Gaussian pulse is used, a
more bell-like shape can be seen (see Figure 8.17) due to the concentrated energy at the
centre rather than at the circumference.
211
8.5.4.1 Ablation depth.
The process parameters used in (Lutey et al., 2013) experimental work were introduced
into the SPH model as given in Table 8.2. Running the simulation for 15 ns using 0.2
µm particle spacing, Figure 8.18 shows the temporal progression of the ejected material
as well as the temperature profile across the sample within the active beam zone. It is
pertinent to mention that a Top-Hat beam profile and a square laser pulse were used
during the simulation to reproduce the experimental conditions.
Table 8.2 Material properties and process parameters used in the SPH model
Material
Aluminium
Density
Initial
Thermal
Specific
Pulse
ρ
Temperature
Conductivity
Heat
duration
3
[kg/m ]
[⁰K]
[W/m.⁰K]
[J/kg]
[ns]
2705
300
167
896
10
Fluence
[J/cm2]
10
Repetition
Simulation
rate
timestep
[kHz]
[ns]
30
0.05
The laser pulse was activated instantly at the beginning of the simulation and was
deactivated at 10 ns leaving the top surface to cool naturally due to the conducted heat
to the rest of the bulk material. Due to the very high irradiance (1 GW/cm2) acting on
the top surface, the temperature of the surface exceeded the boiling temperature of Al
within only 1 ns, reaching about 3744 K.
Laser Intensity
Top hat
Gaussian
Figure 8.17 Spatial distribution of the Laser intensity as a function of time
Once the surface particles are ejected in reality, they lose the interaction with the laser
beam (apart from particle scattering) allowing the laser to heat up the newly exposed
layer. Therefore, it is assumed that there will be no interaction with the laser beam once
particles abandon the surface. Hoffman and Szymanski (2002) calculated the optical
212
penetration for different metallic vapours at different temperatures. For Al vapour with
a temperature less than 4000 K, the calculated absorption coefficient was about 410-2
m-1 which corresponds to 25 nm optical penetration at 10 µm laser wavelength. For
shorter wavelengths such as 1.064 µm, the optical penetration will be even higher.
Using the calculated optical penetration, the Al vapour at distances 2h=610-7m from
the surface will absorb only 2.410-6 % of the incident laser beam intensity while the
rest will be delivered to the sample’s surface. This negligible value clearly justifies the
aforementioned assumption.
Therefore, the particles-beam interactions were ignored in this simulation to allow the
heating of the underlying layers. By comparing the snapshots at t= 8 ns and t= 10 ns, the
bottom layer showed a higher temperature which can be clearly seen in a darker red
colour after the preceding layer had left the surface. At the end of the pulse, the surface
temperature drops gradually over time until a second pulse starts again and the heating
cycle is repeated.
213
t=0.5 ns
t= 1.5 ns
t= 5.5 ns
t= 8 ns
Solid
Tm=925 K
t= 2.5 ns
t= 10 ns
Liquid T =2730 K
b
t= 3.5 ns
t= 14 ns
Vapour
Figure 8.18 3-D view of the ablated surface showing the temperature evolution and
phase change with time. Particles are ejected within 1.5 ns time (single shot at 10 J/cm2
and 10ns pulse duration)
In order to determine the ablation depth, a vertical slice in the Z-Y plane of thickness
2∆x is shown in Figure 8.19 to display the ablation depth. The surface particles were
ejected within 1.5 ns when their temperature exceeded the boiling threshold leaving a
0.2 µm crater at the top surface. Once the particles become distant from the surface (∇.r
is greater than 2.4), the heating of the next layer begins until reaching the boiling
temperature where the ejection process is repeated. The particles located at the inclined
edges are ejected normal to the inclined surface reproducing a similar behaviour to that
observed in the real experiments (Horn et al., 2001). As mentioned previously in section
8.4.3, a criterion of (∇.r < 2.4) is used to identify the surface particles and the vapour
214
velocity components are calculated according to the normal vector components in all
three directions.
At the end of the pulse and when the free surface starts to cool towards the ambient
temperature, the taper effect that is usually associated with the ablation process becomes
evident at each edge of the crater, leaving a concave shape on the surface.
t=0.5 ns
t= 8 ns
Solid
t= 1.5 ns
t= 3.5 ns
t= 8.5 ns
t= 10 ns
Tm=925 K
Liquid T =2730 K
b
t= 5.5 ns
t= 14 ns
Vapour
Figure 8.19 Cross section of the Al sample showing the ablation with taper effect
(using single shot at 10 J/cm2 and 10 ns pulse duration)
The ablation depth predicted by the SPH simulation was found to be “0.6 µm”, which
correlates satisfactorily with the reported value by Lutey et al., namely “0.8 µm”. The
small discrepancy can be due to the different conditions of the actual sample which have
not been explained by the authors in comparison with the conditions assumed in the
SPH model. In reality, most Al surfaces suffer from oxidation. This oxidation
215
phenomenon; however, was not taken into consideration in the SPH model. Moreover,
the surface topography (roughness) of the actual sample may promote more laser
absorbance than the flat surface assumed in the model.
8.5.4.2 Temperature and Vapour Pressure.
To track the change in the thermo-physical quantities with time, an SPH particle at the
centre of the free surface was selected to plot its temperature evolution with time until it
loses its connection with the surface. Within region I in Figure 8.20, the particle’s
temperature increases gradually with time and the heat generated by the laser is
transferred into the bulk material due to conduction. The temperature then exceeds the
melting point at about t= 150 ps and the boiling temperature at t= 1 ns where region II
starts.
In the second region, the particle starts to gain velocity due to the recoil pressure that
pushes the vapour particle away from the surface. Considering the small time step (50
ps) at which all the physical quantities and the particles’ coordinates are being
calculated, the particle during phase II travels a very small distance within which the
particle is still considered as part of the surface. Therefore, despite the fact that the
particle starts to leave the surface at this stage, the particle still belongs to the surface
and hence it receives more energy from the laser beam until it abandons the surface
completely. This is also consistent with the condition (∇.r < 2.4) that is used to identify
surface particles.
After losing the connection with the lower layers, the particle’s temperature begins to
decrease as it becomes transparent to the laser light (see Figure 8.19 at t= 1.5 ns).
Additionally, the particle gives up some of its heat to the adjacent particles until it
becomes isolated along with ejected particles of the same temperature, and this is when
the temperature stabilises with time as shown in region III. This happens because the
temperature gradient for a particle surrounded by particles of the same temperature will
equal to zero and no heat exchange will take place between those particles.
216
Figure 8.20 Temperature change with time for a surface particle obtained by SPH
model (Region I: conduction heating, Region II: further heating, Region III: partial
cooling after ejection)
The temperature profile is very important since it controls the ejection process and the
vapour pressure associated with it. The vapour pressure values can be calculated using
Equation ( 8-5 ) with the following parametric values: Patm= 101.325 103 Pa, Lv= 10.53
106 J/kg, R= 308.17 J/K.kg.
Figure 8.21 depicts the recoil pressure values with temperature in the range between the
boiling point, at which the vapour begins to form, and the maximum temperature
obtained in the simulation’s results. The depicted values were calculated using Equation
( 8-5 ) according to the particle’s temperature. The high recoil pressure values which
reached up to 35 bar indicate that the erupted vapour is capable of leaving the surface
without any help by the assist gases that are usually necessary for laser cutting and
drilling processes. In the ablation processes (especially laser cleaning), a fume
extraction unit is typically used to remove the ablated material so that it does not fall
back to the adjacent cleaned/ablated surfaces; thus, the extraction effect does not
contribute to the material ejection during the process but only to keep the vapour away
from the substrate.
217
Figure 8.21 SPH values of vapour pressure for surface particles a function of
temperature
8.5.4.3 Vapour Velocity.
As mentioned in section 8.2, finite element (FE) simulations of laser drilling, cutting,
and ablation are unable to produce information on the ejected material, its behaviour
and its interaction with the surrounding environment. Deleting the elements whose
temperatures exceed the boiling point will not count for the interaction between the
expelled particles and the solid walls of the crater, which may result in an inaccurate
prediction of the process outcome. However, the Lagrangian nature of SPH makes it
possible for predicting the non-linear behaviour of the physical quantities that belong to
a particle wherever it travels within the domain of study.
Previous works on laser cutting/drilling calculated the recoil pressure using the
Clausius-Clapeyron, which was then fed into the melt ejection velocity that was derived
from Bernoulli’s equation after a number of simplifying assumptions. Although the melt
ejection velocity should be only assigned to the molten ejected material, the calculated
values were assigned to all ejected particles regardless their temperature and phase type.
Moreover, some authors (Muhammad et al., 2013) claimed that the recoil pressure
effect becomes predominant in laser cutting due to the vapour particles build-up in the
kerf; despite that, Bernoulli’s equation was still being used to describe the vapour
velocity although it is not the correct formulation to be used in such cases. This
produces inaccurate results in the ejected material behaviour and velocity since the
218
vapour velocity can be one or two orders of magnitude larger than the velocity
calculated using Bernoulli’s equation.
Half-section
B-B
(a)
(b)
(c)
(d)
Figure 8.22 Vapour velocity vectors at the workpiece surface (a) 3-D view at 2 ns (b)
half-section to be considered in the following subfigures (c) cross section B-B at 1 ns
(d) cross section B-B at 3.0 ns
Figure 8.22(a) shows the velocity vectors of the vapour particles after reaching the
boiling point and how the vectors are normal to the top surface of a magnitude of about
380 m/s. In order to have a clearer view of the velocity vectors, a small section of the
studied domain was isolated and projected on the front plane as shown in Figure 8.22(b,
c and d).
Due to the slight decrease in the particles’ temperature at the edges of the domain, these
particles will have slightly lower speed than those closer to the centre and their speed
vectors will be slightly inclined outwards as shown in Figure 8.22(b and c). Consistent
with the normal-to-surface condition, particles that belong to the tapered surfaces will
have their velocity vector inclined towards the inside of the domain as seen in Figure
8.22(b).
Once the particle is ejected and becomes transparent to the beam, its temperature
decreases significantly causing its velocity to drop. Hence the ejected material will
219
accumulate close to the surface and block any new material from being removed. This
has been resolved by specifying that all ejected particles maintain a constant speed until
the end of the simulation or leaving the domain. This serves two purposes: firstly, this
will prevent the aggregation of removed particles above the surface and continues to
allow the direct line of action of the laser onto the lower particles to be heated and
removed. Secondly, this condition realistically resembles the vacuum effect in the real
application which sucks all removed material away from the ablated surface.
Figure 8.23 depicts the evolution of velocity of a surface particle with time. It can be
seen that at 1 ns (in Region II) the particle’s temperature passes through the boiling
point and the particle starts to gain a speed of about 380 m/s. This velocity increases
gradually with the rising temperature in Region III to reach up to 450 m/s at the end of
this stage. Afterwards, the particle leaves the surface completely and maintains its
velocity during the rest of the simulation time (Region IV).
Figure 8.23 SPH results for the ejected particles’ velocity with time (Region I:
stationary state, Region II: instant ejection, Region III: velocity change with
temperature, Region IV: stable speed)
According to Tam et al. (1992), the particle can be ejected without melting the surface
if it has an ejection acceleration greater than (1010 cm/s2), that is much greater than
gravitational acceleration. Taking the average increase in velocity during phase III over
one timestep, the SPH particles’ acceleration will be about 1.41011 m/s2, which is
greater than the threshold given by Tam et al. (1992) experimentally.
220
Due to the lack of experimental data on the vapour velocity of Al during laser ablation,
the very few modelling results reported in the literature can be considered for
comparison. Hamadi et al. (2008) created a finite volume model using “Fluent” code to
estimate the vapour velocity of Al when ablated with nanosecond UV laser beam. It was
assumed that the ambient pressure is 102 Pa and the Al target was irradiated with a 25 ns
pulse. Their results showed that the maximum particle velocity was about 1110 m/s at
the centre of the ablated area, and it reduces to 167 m/s away from the centre, with an
average speed of about 640 m/s. This average value is in good agreement with the SPH
results obtained from this work taking into consideration the different boundary
conditions of the two models. Moreover, the authors indicated that the vapour velocity
reduces when the ambient pressure increases, and taking into account that the ambient
pressure in this work is 105 Pa, the SPH particles velocity is then expected to be lower
than the value reported by Hamadi et al. (2008).
Rajendran et al. (2007) studied a similar test case using a numerical model based on the
kinetic description of the Knudsen layer. Their numerical results, which according to the
authors showed a good agreement with other analytical models, showed that the
maximum velocity reached up to 750 m/s after 15 ns at the surface of the target. These
values indicate that SPH prediction of the vapour velocity lies within a satisfactory
range of values for the considered model and material.
8.5.4.4 Further Validation of Laser Ablation Depth.
After considering one set of parameters to validate the SPH model, a wider range of
fluences and pulse durations are tested to further validate the behaviour of the proposed
model.
Lutey et al. (2013) conducted experiments using nanosecond laser with 10 ns pulse
duration, 30 kHz repetition rate and beam fluence of 4-20 J/cm2. A so-called
“unidimensional” numerical model was created to predict the ablation depth of the
studied Al sheet. Figure 8.24 shows a good agreement in the predicted ablation depth
between the SPH results, the experimental and numerical data by Lutey et al. at
different fluence values, achieving about 1.0-1.2 µm/pulse at 20 J/cm2.
221
Figure 8.24 Ablation depths at different fluences 4-23 J/cm2 using 10 ns pulse with 30
kHz repetition rate. SPH results are compared with experimental and numerical values
reported in (Lutey et al. 2013)
From Figure 8.24, the difference between the SPH modelling results and Lutey’s
numerical data can be attributed to the fact that their unidimensional model only
calculates the heat conduction in one direction without taking the heat losses into
account, while the 3-D nature of the SPH model accounts for the heat flow in the other
two directions of the domain and for the convection and radiation losses at the sample’s
free surface. The discrepancy between the experimental and the simulation results can
be explained by the fact that the experimental values of the ablation depth per pulse
were taken as the average of the total depth over the aggregate number of pulses.
Additionally, the deep ablated surfaces during the experiments enhance the internal
reflection of the laser beam on the side walls of the hole and promote more absorption
of the beam in comparison with the single-shot ablation achieved in the simulation.
Furthermore, the surface conditions of the sample during the experiments may differ
from those in the simulation since there was no mentioning of such information by the
authors. For instance, oxidation layers usually form on the free surface of Al and their
low thermal conductivity (35 W/mK) and the greater density (3750 kg/m3) makes it
more difficult to be ablated in comparison with pure Al.
222
(a)
(b)
Figure 8.25 SPH results of Ablation depth compared to literature experimental and
numerical data at different fluences 4-23 J/cm2 using 6 ns pulse.
Figure 8.25(a) shows a different case created using shorter pulse durations of 6 ns and
laser fluences 10-25 J/cm2 on an Al target. In this case, the SPH results were compared
with both experimental and numerical data from different sources (Horn et al., 2001;
Cao et al., 2013). As mentioned previously, all experimental data are averaged over the
total number of pulses (400 pulses in this case) in order to obtain the ablation depth per
pulse.
In Figure 8.25(a), a good correlation between the SPH prediction and the reported data
can be noted with a linear increase of the ablation depth with the energy density.
223
Despite that, the experimental work has reported higher ablation rates compared to the
simulations over the studied fluence regime. This can be attributed to the so-called
“incubation effect” that takes place during multi-pulses laser ablation as well as the
different surface conditions on the actual samples (Xiao et al., 2012).
The same SPH results also achieved a good correlation with the ablation rates predicted
by a numerical model (Gusarov and Smurov, 2003) of supercritical ablation with an
optical breakdown in the volume of the vapour phase (see Figure 8.25(b)). This model
was different from the previous numerical models as it includes the interaction between
the laser radiation and the gas phase produced when the vapour temperature exceeds the
critical temperature of Al.
The good agreement with both the experimental work and the different simulation
models indicates that the thermal ablation mechanism can be applied successfully to
predict the ablation depth in the low to medium fluence range, and the SPH model was
able to not only predict the ablation depth but also to give an insight into the vapour
phase and its characteristics.
8.6 Summary
A 3-D SPH model has been presented to predict the characteristics of the laser ablation
process after evaluating the heat conduction behaviour in such processes, as well as the
need and the sensitivity of the smoothing kernel correction at the free surfaces in SPH
modelling. According to the results reported in this work, the following conclusions can
be drawn:

The proposed model in this paper showed an excellent agreement with the
analytical solution using different levels of power intensity and surface
reflectivity and produced credible temperature values which were then utilised in
laser ablation simulation.

KGC and Schwaiger's correction were able to correct the deviation caused by
the truncated support at the domain borders for the gradient and Laplacian of the
studied functions respectively. However, for non-linear functions such as
hyperbolic and logarithmic functions both correction methods were only able to
reduce the deviation at the boundaries without matching the analytical solution.
224

SPH predictions of the ablation depth for different process parameters were in a
good agreement with both experimental and numerical data reported in the
literature. Unlike other mesh-based methods in which no information can be
provided on the ejected material, the SPH model was not only able to provide
the temperature and velocity of the ejected particles, but also the effect of the
interaction between them as well as the direction and the pattern of the ejection.
225
 Brief Introduction
In the previous chapter, an SPH model of laser cleaning/ablation was introduced, and
the SPH prediction of the ablation depths and the ejected material’s behaviour were
validated against the published numerical and experimental data.
In Appendix A, a preliminary SPH model was created to investigate the molten material
flow inside the weld pool during laser welding of Al alloys. A successful SPH welding
model is intended to be coupled with the SPH laser cleaning model in order to
understand the mechanism behind porosity reduction using laser cleaning. The initial
results on the molten material flow and the temperature field during laser welding are
presented in Appendix A.
In the next chapter, conclusions remarks are given along with the recommendations for
the future work. In addition, the limitations of the proposed SPH model are also briefly
presented.
226
9 CHAPTER 9
CONCLUSIONS AND FUTURE WORK
9.1 Conclusions
The primary focus of this work was on porosity formation and the practical methods for
its reduction or elimination. Laser cleaning was successfully implemented to reduce
porosity during laser welding of AC-170PX Al alloy to less than 1% for two different
weld configurations. Furthermore, a novel SPH model on laser ablation was created for
the first time and was able to predict the behaviour of the ejected material and the crater
dimensions and shape produced on the base material.
In Chapter 6, the effect of welding parameters and the gap between the sheets on
porosity was thoroughly investigated. Chapter 7 focused on porosity elimination by the
alteration in the filler wire composition while Chapter 8 concerned the use of laser
cleaning as a surface preparation method for porosity elimination. In addition to the
experimental work, a numerical model using SPH was created to further understand
porosity formation on the effect of laser cleaning on the laser welding quality.
Conclusions were given at the end of each chapter while the key findings and the
scientific contribution of this project can be summarised as follows.
9.1.1
Experimental work
9.1.1.1 Effect of heat input and sheet gap on porosity formation

The effect of the laser welding parameters on porosity content in AC170PX Al alloys laser welding was investigated for two types of welding
configurations. Also, the introduction of a 0.2 mm gap between the
welded sheets and its effects on the welding quality were studied.
227

Porosity in fillet edge joints increased from circa 4% to circa 90% when
the laser power and welding speed were increased simultaneously from 2
kW to ~5 kW and from 20 mm/s to 50 mm/s respectively. However,
porosity was significantly less affected by the welding parameters for
flange couch joints in which porosity only increased by 4% when the
laser power and welding speed were increased simultaneously from ~2
kW to 4.5 kW and from 20 mm/s to 50 mm/s respectively. The
introduction of a 0.2 mm gap between the welded sheets significantly
reduced porosity in the fillet welds while its effects were marginal for
flange couch joints.

It was observed that the increase in the traverse speed decreased the
ability of the pores to escape from the welding pool surface. This
reduced ability can be attributed to the energised flow of the molten
material inside the weld which leads the pores towards the bottom of the
welding pool and prevents it from leaving before solidification occurs.
This phenomenon was experienced in both types of welds.

The ultimate tensile strength of the fillet welds was reduced despite the
increase in the welds’ penetration and width when the welding
parameters were increased. This can be related to the significant increase
in porosity which reduces the cross sectional area under the tensile load.
Although the application of the 0.2 mm gap reduced porosity in the cross
section, this promoted the molten metal to flow into the existing gap and
caused lack of fusion and reduction in the weld dimensions. A similar
reduction in the tensile strength of the flange joints was measured due to
the slight increase in porosity.

Although the use of relatively lower traverse speeds reduces porosity,
this jeopardises the production throughput by increasing the process time
and resources required. The sheet gap solution was approved to
significantly reduce porosity in the welded joints and improve the weld
appearance. However, the gap led to other defects such as lack of fusion
which caused the tensile strength to drop in comparison of welds with no
gaps.
228
9.1.1.2 Effect of filler wire composition on porosity reduction

The use of filler wires of different chemical compositions was also
demonstrated as a method for porosity reduction in Al alloys welding.

Porosity in fillet edge joints decreased from 90% using the AlSi5 filler
wire to less than 1.5% when AlMgMn filler wire was used. For flange
couch joints, the maximum porosity proportion was observed in the
AlSi12 joints with 3%, while it was reduced to about 1% using
AlMg4,5Mn filler wires. In both types of joint, filler wires with higher
Mg and Mn content led to a significant reduction in porosity and
significantly improved the welding quality

The maximum tensile strength was achieved in the fillet and flange joints
when AlMgMn and AlMg4,5Mn fillers were used respectively. This
higher strength in comparison to the use of Si filler can be attributed to
the high magnesium and manganese content which adds to the strength
of the material, as well as the reduction of porosity associated with the
use of such filler wires.

The filler wires mentioned above can be used to obtain better tensile
strength and porosity-limited welds which can be a very feasible solution
for porosity problems in automotive applications.

The good weld surface appearance, however, can be an essential
requirement for the welding process to achieve for automotive
applications. The use of Mg and Mn filler wires produces some
appearance defects such as spatter, smut and dark colour surface which
cannot be accepted if the welded area is meant to be painted without any
further processing or repair.
9.1.1.3 Effect of laser cleaning on porosity reduction

Laser cleaning was found to remove most of the lubricant and
contaminants placed on the surface of AC-170PX Al alloys. Moreover,
laser cleaning reduced the surface roughness from 982 nm to 882 nm and
removed about 19 microns from the substrate at the used cleaning
parameters.
229

Depending on welding parameters, laser cleaning was found to achieve a
significant improvement in the weld quality by reducing porosity level
from about 10-90% to less than 0.5% for the fillet edge joint, and from
0.5-4.1% to less than 1% for flange couch joint.

It is believed from this work that the laser cleaning reduced porosity
during the welding of coated AC170 PX Al alloy due to the elimination
of hydrogen and other gases produced from the coating, lubricant and the
contaminants on the surface.

This novel scientific and industrial achievement of applying the laser
cleaning to improve the welding quality led to a full and successful
introduction of the approved system to one of the largest British and
international automobile companies whose laser cleaned and welded cars
are currently driven on roads across the globe.
9.1.2
SPH modelling

SPH simulations were performed using and adding to SPHysics_3D open
source code to model the laser cleaning/ablation process to understand its
effect on the welding quality. Additionally, the need for the free-surface
kernel-related corrections was also investigated in the case of the secondorder spatial behaviour of the numerical solutions expected for thermal
processes. The laser beam was considered as a surface heat source since
laser metal processing takes place at the sample’s surface and the optical
penetration depth at the given wavelength is negligible.

Initially, a 3-D laser cleaning/ablation model was created to
systematically investigate for the first time the laser-induced temperature
field distribution across the studied samples for nanosecond laser pulses.
This step was needed because the thermal effect of nanosecond laser
heating is believed to be the primary source for developing thermal
ablation, thermomechanical shear stresses and other nanosecond lasermaterial interaction phenomena. Moreover, the absence of such study
and validation on the SPH particles thermal behaviour under the laser
230
heating emphasised the need to understand the capabilities of SPH
method and the accuracy of the solution obtained by such method.

The laser ablation model achieved an excellent agreement with the 1-D
analytical solutions during single and multiple ns laser pulses heating of
AC-170PX Al sample. The good correlation with the analytical solution
included the temporal and spatial temperature distribution across the
sample, the maximum surface temperature for different reflectivity
values and the heating and cooling rates at various depths within three
consecutive laser pulses.

Moreover, the ablation depth, vapour pressure and velocity and the crater
profile were also demonstrated and validated with the experimental and
numerical results reported in the literature at different fluences. Unlike
other mesh-based methods in which no information can be obtained on
the removed material, SPH provides information on not only the ejected
material’s physical characteristics such as temperature, pressure and
velocity but also its behaviour and interaction with other expelled parts
as well as the base material.
9.2 Limitations of the SPH model
The physical phenomena captured by the proposed model were simplified using the
assumptions mentioned in Chapters 8. In this section, the limitations imposed by these
assumptions and by some methodology-related factors are addressed as follows:

The SPH simulations ignored the air and the shielding gases to avoid
discontinuity in density at the interface between the metal and the gas. This
sharp change in density (2700:1 ratio) requires special treatment to prevent
numerical instabilities and to conserve all physical quantities. Although the
static air particles may not significantly influence the high-speed ejected particle
in laser ablation, the consideration of the air phase makes the simulation closer
to the real phenomena and makes it more applicable to laser welding modelling.

Also, the proposed model neglected the plasma interaction with the laser beam
and assumed that the full beam energy is reaching the sample’s surface.
However, in real life cases, the laser-generated plasma absorbs a portion of the
231
incident beam energy and hence reduces the laser energy delivered to the
material. For a more realistic model, the laser-plasma interaction should be taken
into account.

The free-surface identification is still a challenging task to accomplish in the
current SPH model if keyhole laser welding is to be simulated. As mentioned in
Chapter 3 and Appendix A, some authors (Hu et al., 2016) are still using
additional approaches to identify the free surface shape created by the metallic
vapour. To date, there are no robust techniques for detecting the free surface
within SPH for complex phenomena such as keyhole welding. Therefore, this
area needs further investigation and improvement to eliminate the need for other
software output.

The resolution (particle size) of the simulations was limited by using a nonparallel code, which in turn limited the computational domain dimensions. Cases
such as laser welding modelling require the simulations of relatively large areas
on the order of centimetres to predict the physical phenomena in the process.

The temperature dependence of reference density and viscosity has been
represented by equations from experiments. However, these equations may
involve sharp changes in the physical quantities at phase transition and hence
may cause numerical instabilities. Equations, that completely describe the
fundamental physics, can produce a clear agreement with the observed physical
behaviour. For example, phase transition could be better represented by using
enthalpy in the energy balance to give a better approximation of the thermal
expansion. The use of the enthalpy helps to conserve energy especially in
transient heat transfer problems such as laser-material interaction.
To expand the model’s capability to simulate complex phenomena such as keyhole laser
welding, the previous limitations should be overcome by introducing the neglected
physical aspects of the process.
9.3 Future work recommendations
Throughout the work in this project, the author aimed to investigate different
methodologies for porosity reduction in Al alloys for automotive applications.
Supported by experimental evidence, three different procedures achieved a significant
232
reduction in porosity. This experimental work was also developed with numerical
modelling using SPH to simulate the laser cleaning process.
The future work recommended to further extend this study includes:
9.3.1

Experimental work
The use of different gap sizes to evaluate the gap’s performance and ability to
reduce porosity, in particular for the fillet edge welds.

Investigating the laser cleaning effectiveness in reducing porosity in thick Al
alloys plates. Thicknesses between 3 mm to 6 mm, which are commonly used in
automotive and aerospace applications, can be cleaned and welded autogenously
or heterogeneously (using filler wires). In thick sections, the effect of the surface
oxide layer or contaminations on the surface may have less effect on porosity in
comparison with hydrogen gas molecules released during the melting process.

Examining the use of other characterisation techniques such as Glow Discharge
Optical Spectroscopy (GDOS) and X-Ray Photoelectron Spectroscopy (XPS) to
quantify the oxygen , hydrogen and other elements contents in the laser welds
with and without laser cleaning. The use of such methods will give a clear idea
about the most affecting elements in porosity generation during Al alloys laser
welding.

Studying the effect of laser cleaning on different seam configurations such as
butt- and T-joints for aerospace applications.

Examining the effect of laser cleaning of the filler wire before welding on
porosity. It is believed that the filler wire is more prone to interact with moisture
during its storage.
9.3.2

SPH modelling
Developing an SPH model that can resolve or represent the oxide layer since it is
usually on the order of nano- to micrometres in scale. Modelling of such a
problem requires two-phase flow considerations and a significant number of
particles (millions) to quantify the effect of oxide layers on laser welding.
233

Taking into account the air phase at the top of the welding surface or/and the
shielding gases effect to give an improved representation of the surface tension
effect on the welding pool shape and dynamics.

Application to keyhole laser welding by including the recoil pressure effect with
and without the use of filler materials.

Modelling of pores generation in the welding pool that originates from the
keyhole collapse, the hydrogen content in the alloys and the oxide layer.

Reducing the high computational cost associated with the currently used
SPHysics FORTRAN code. This can be achieved by the use of hardware
performance accelerators such as Graphic Processing Unit (GPUs) which
significantly reduce the computational time and the energy consumed by the
system. Running SPH simulations using other codes such as DualSPHysics,
which operates on GPUs, has already shown a significant improvement in the
computational performance (Crespo et al., 2011). Higher resolutions and larger
computational domains can be applied for obtaining more accurate predictions
and more realistic representation of the studied problems. However, including
the thermal physics in DualSPHysics using CUDA is a new implementation
which needs time and development to be completed.

Considering a variable smoothing length with temperature which provides a
complete support when particle’s volume increases, reducing the number of
particles contributing to the interpolation.

Combining the laser cleaning and welding models in one code to simulate the
entire process in one model. This combination will allow the boundary
conditions created at the end of the laser cleaning process to be directly used in
the welding SPH model without the need for an intermediate transfer stage, in
which valuable information may accidently be lost. This procedure will create a
closer model to the real application.

As part of the future work, a new, but a primitive, laser welding model was
created using SPH to simulate the laser melting of Al as well as modelling the
surface tension and Marangoni effect. For more details, see Appendix A.
234
REFERENCES
Ait Oumeziane, A., Liani, B. and Parisse, J. D. (2016). 'Non-equilibrium modelling of
UV laser induced plasma on a copper target in the presence of Cu2+', Physics of
Plasmas, 23(3), 033502.
Alfaro, I., Fratini, L., Cueto, E., Chinesta, F. and Micari, F. (2007). 'Meshless
Simulation of Friction Stir Welding', AIP Conference Proceedings, 908(1), 203208.
Alfaro, I., González, D., Bel, D., Cueto, E., Doblaré, M. and Chinesta, F. (2006a).
'Recent advances in the meshless simulation of aluminium extrusion and other
related forming processes', Archives of Computational Methods in Engineering,
13(1), 3.
Alfaro, I., Racineux, G., Poitou, A., Cueto, E. and Chinesta, F. (2009). 'Numerical
simulation of friction stir welding by natural element methods', International
Journal of Material Forming, 2(4), 225.
Alfaro, I., Yvonnet, J., Cueto, E., Chinesta, F. and Doblare, M. (2006b). 'Meshless
methods with application to metal forming', Computer Methods in Applied
Mechanics and Engineering, 195(48–49), 6661-6675.
Alfieri, V., Cardaropoli, F., Caiazzo, F. and Sergi, V. (2011). 'Investigation on Porosity
Content in 2024 Aluminum Alloy Welding by Yb:YAG Disk Laser', Advanced
Materials Research, 383-390, 6265-6269.
Allen, C. M., Verhaeghe, G., Hilton, P. A., Heason, C. P. and Prangnell, P. B. (2006).
Laser and hybrid laser-MIG welding of 6.35 and 12.7mm thick aluminium
aerospace alloy. International Conference on Aluminium Alloys (ICAA 10).
Vancouver.
Allen, S. D., Miller, A. S. and Lee, S. J. (1998). 'laser cleaning of critical surface',
Materials Chomiatry and Physics, 51, 181-185.
Allmen, M. V. (1976). 'Laser drilling velocity in metals', Journal of Applied Physics,
47(12), 5460-5463.
AlShaer, A. W., Li, L. and Mistry, A. (2014). 'The effects of short pulse laser surface
cleaning on porosity formation and reduction in laser welding of aluminium
alloy for automotive component manufacture', Optics & Laser Technology,
64(0), 162-171.
AlShaer, A. W., Li, L. and Mistry, A. (2015a). 'Effect of filler wire properties on
porosity formation in laser welding of AC-170PX aluminium alloy for
lightweight automotive component manufacture', Proceedings of the Institution
of Mechanical Engineers, Part B: Journal of Engineering Manufacture.
Alshaer, A. W., Li, L. and Mistry, A. (2015b). 'Understanding the Effect of Heat Input
and Sheet Gap on Porosity Formation in Fillet Edge and Flange Couch Laser
Welding of AC-170PX Aluminum Alloy for Automotive Component
Manufacture', Journal of Manufacturing Science and Engineering, 137(2),
021011.
Alshaer, A. W., Rogers, B. D. and Li, L. (2017). 'Smoothed particle hydrodynamics
(SPH) modelling of transient heat transfer in pulsed laser ablation of Al and
associated free-surface problems', Computational Materials Science, 127, 161179.
235
Amoruso, S. (1999). 'Modeling of UV pulsed-laser ablation of metallic targets', Applied
Physics A, 69(3), 323-332.
Ancona, A., Sibillano, T., Tricarico, L., Spina, R., Lugarà, P. M., Basile, G. and
Schiavone, S. (2005). 'Comparison of two different nozzles for laser beam
welding of AA5083 aluminium alloy', Journal of Materials Processing
Technology, 164-165, 971-977.
Andersen, M. M. and Jensen, T. A. (2001). Hybrid Nd:Y AG laser + MIG welding in
aluminium. In: 8th Nordic Conference Laser Materials Processing, 13-15 Aug
2001 Copenhagen, Denmark, 371- 380.
Anderson, A. (1995). Computational fluid dynamics: The basics with application:
McGraw-Hill.
Anon (2011a). AC-170Px Aluminium technical specifications. In: (N. D. GmbH.).
Stuttgart, Germany.
Anon (2011b). FONTARGEN A 405 M Aluminium–silicon wire specifications. In: (F.
Brazing). Germany.
Antonov, A. A., Kozlov, G. I., Kuznetsov, V. A. and Masyukov, V. A. (1977). 'A
steady-state crater formed as a result of interaction of high-power CO2 laser
radiation with metals and liquids. 7/8', Sov. Journal Quantum Electron, 4, 17471753
Anyalebechi, P. N. (1995). 'Analysis of the effects of alloying ele- ments on hydrogen
solubility in liquid aluminum alloys', Scripta Metallurgica et Materialia, 33(8),
1209-1216.
Arata, Y. (1980). 'What happens in high energy density welding and cutting? Welding
Research Institute, Osaka (Film).'.
Arnold, N. (2003). 'Theoretical description of dry laser cleaning', Applied Surface
Science, 208, 15-22.
Arnold, N., Schrems, G., Mühlberger, T., Bertsch, M., Mosbacher, M., Leiderer, P. and
Bäuerle, D. (2001). Dynamic particle removal by nanosecond dry laser cleaning:
Theory. In: Dubowski, I., ed. Second Internatuional Symposium on Laser
Precision Microfabrication, Proceedings of SPIE 2001 Singapore, Singapore.
340-346.
Arronte, M., Neves, P., Vilar, R. and Lazaroyl, S. (2003). Modelling of laser cleaning of
metallic particles on silicon substrates. In: Lasers and Electro-Optics Europe,
2003. CLEO/Europe. Conference on, 22-27 June 2003, 561.
Ascari, A., Fortunato, A., Orazi, L. and Campana, G. (2012). 'The influence of process
parameters on porosity formation in hybrid laser-GMA welding of AA6082
aluminum alloy', Optics & Laser Technology, 44(5), 1485-1490.
Ashby, M. F., Shercliff, H. and Cebon, D. (2010). Materials: Engineering, Science,
Processing and Design (2nd ed.): Butterworth-Heinemann.
Askeland, D. R. (1985). The Science and Engineering of Materials, PWS, Boston, USA.
ASM International. (1993). ASM Specialty handbook: Aluminum and aluminum alloys:
American Society for Metals (ASM).
ASM International. (2004a). ASM Handbook 6: Welding Brazing and Soldering:
American Society for Metals (ASM).
ASM International. (2004b). ASM Handbook 9: Metallography and Microstructures:
American Society for Metals (ASM).
Atkins, P. W. and De Paula, J. (2002). Atkins' Physical Chemistry: Oxford University
Press.
236
AudiWorld. (2004). Audi Starts Production of the New A6 Limousine [Online].
Available: http://www.audiworld.com/news/04/021704/content.shtml [Accessed
13/05/2014 2004].
Autrique, D., Alexiades, V. and Khanal, H. (2013). Hydrodynamic modelling of
ns−laser ablation. Ninth MSU-UAB Conference on Differential Equations and
Computational Simulations, Electronic Journal of Differential Equations. San
Marco, USA.
Baba, N. and Watanabe, I. (2005). 'Penetration depth into dental casting alloys by
Nd:YAG laser', Journal of Biomedical Materials Research Part B: Applied
Biomaterials, 72B(1), 64-68.
Bachmann, M., Avilov, V., Gumenyuk, A. and Rethmeier, M. (2013). 'About the
influence of a steady magnetic field on weld pool dynamics in partial penetration
high power laser beam welding of thick aluminium parts', International Journal
of Heat and Mass Transfer, 60, 309-321.
Bahemmat, P., Haghpanahi, M., Besharati, M. K., Ahsanizadeh, S. and Rezaei, H.
(2010). 'Study on mechanical, micro-, and macrostructural characteristics of
dissimilar friction stir welding of AA6061-T6 and AA7075-T6', Proceedings of
the Institution of Mechanical Engineers, Part B: Journal of Engineering
Manufacture, 224(12), 1854-1864.
Bardy, S., Aubert, B., Berthe, L., Combis, P., Hébert, D., Lescoute, E., Rullier, J.-L. and
Videau, L. (2016). 'Numerical study of laser ablation on aluminum for shockwave applications: development of a suitable model by comparison with recent
experiments', Optical Engineering, 56(1), 011014-011014.
Bäuerle, D. W. (2000). Laser Processing and Chemistry. Berlin Springer-Verlag.
Bäuerle, D. W. (2011). Laser Processing and Chemistry (Fourth Edition ed.): SpringerVerlag Berlin Heidelberg.
Belytschko, T., Lu, Y. Y. and Gu, L. (1994). 'Element-free Galerkin methods',
International Journal for Numerical Methods in Engineering, 37(2), 229-256.
Berger, P., Hügel, H. and Graf, T. (2011). 'Understanding pore formation in laser beam
welding', Physics Procedia, 12, 241-247.
Beyer, E., Caiazzo, F., Alfieri, V., Cardaropoli, F., Sergi, V. and Morris, T. (2012).
'Bead characterization of disk-laser butt welding of thin AA 2024 sheets', Proc.
SPIE 8239, High Power Laser Materials Processing: Lasers, Beam Delivery
Diagnostics, and Applications, California (USA).
Birol, Y. (2005). 'Pre-aging to improve bake hardening in a twin-roll cast Al-Mg-Si
alloy', Materials Science and Engineering A, 391, 175-180.
Bityurin, N. (2005). 'Studies on laser ablation of polymers', Annual Reports Section "C"
(Physical Chemistry), 101(0), 216-247.
Bityurin, N., Luk’yanchuk, B. S., Hong, M. H. and Chong, T. C. (2003). 'Models for
Laser Ablation of Polymers', Chemistry Review, 103(519).
Black, J. T., DeGarmo, E. P. and Kohser, R. A. (2008). DeGarmo's materials and
processes in manufacturing (10th ed.): Wiley.
Blackburn, J. E., Allen, C. M., Hilton, P. A. and Li, L. (2009). Statistical analysis of low
porosity laser welding of TI alloys using a directed gas jet. Proc. ICALEO.
Orlando (USA).
Blackburn, J. E., Allen, C. M., Hilton, P. A. and Li, L. (2010a). 'Nd:YAG laser
welding of titanium alloys using a directed gas jet', Journal of Laser
Applications, 22 (2), 71-78.
237
Blackburn, J. E., Allen, C. M., Hilton, P. A. and Li, L. (2010b). Dual focus Nd:YAG
laser welding of titanium alloys. In: (eds. Hinduja, S. & Li, L.) Proceedings of
the 36th International MATADOR Conference. Springer London.
Blackburn, J. E., Allen, C. M., Khan, A. H., Hilton, P. A. and Li, L. (2012). 'Dual focus
Nd:YAG laser welding of titanium alloys: effect on porosity formation', Lasers
in Engineering, 22 (5/6), 319.
Bloisi, F., Di Blasio, G., Vicari, L. and Zoncheddu, M. (2006). 'One-dimensional
modelling of ‘verso’ laser cleaning', Journal of Modern Optics, 53(08), 11211129.
Bolt, P. J., Lamboo, N. A. P. M. and Roizier, P. J. C. M. (2001). 'Feasibility of warm
drawing of aluminum products', Journal of Materials Processing Technology,
115, 118-121.
Bonet, J. and Kulasegaram, S. (2000). 'Correction and stabilization of smooth particle
hydrodynamics methods with applications in metal forming simulations',
Internation Journal of Numerical Mathemetics and Engineering, 47, 1189-1214.
Brackbill, J. U., Kothe, D. B. and Zemach, C. (1992). 'A continuum method for
modeling surface tension', Journal of Computational Physics, 100(2), 335-354.
Brandes, E. A. and Brook, G. B. (1992). Smithells Metals Reference Book (7th Edition
ed.): Reed Educational and Pmfessiond Publishing Ltd.
British Standards Institution. (2001). BS EN 10002-1:2001 - Tensile testing of metallic
materials. Method of test at ambient temperature.
Brookshaw, L. (1986). The Stability of Binary Systems and Rotating Stars. Ph.D. thesis,
Monash University.
Bulgakova, N. M., Zhukov, V. P., Mirza, I., Meshcheryakov, Y. P., Tomáštík, J.,
Michálek, V., Haderka, O., Fekete, L., Rubenchik, A. M., Fedoruk, M. P. and
Mocek, T. (2016). Ultrashort-pulse laser processing of transparent materials:
insight from numerical and semi-analytical models. In Proc. SPIE 9735, Laser
Applications in Microelectronic and Optoelectronic Manufacturing (LAMOM)
XXI, 97350N (March 14, 2016). 97350N-97350N-16.
Caiazzo, F., Alfieri, V., Cardaropoli, F. and Sergi, V. (2012). 'Butt autogenous laser
welding of AA 2024 aluminium alloy thin sheets with a Yb:YAG disk laser',
The International Journal of Advanced Manufacturing Technology, 67(9-12),
2157-2169.
Campbell, S. W., Galloway, A. M., McPherson, N. A. and Gillies, A. (2012).
'Evaluation of gas metal arc welding with alternating shielding gases for use on
AA6082T6', Proceedings of the Institution of Mechanical Engineers, Part B:
Journal of Engineering Manufacture, 226(6), 992-1000.
Cao, Y. and Shin, Y. C. (2015). 'Multi-scale modeling of phase explosion in high
fluence nanosecond laser ablation and clarification of ablation depth prediction
criterion', Applied Surface Science, 357, Part A, 74-85.
Cao, Y., Zhao, X. and Shin, Y. C. (2013). 'Analysis of nanosecond laser ablation of
aluminum with and without phase explosion in air and water', Journal of Laser
Applications, 25(3), 032002.
Casalino, G., Mortello, M., Contuzzi, N. and Minutolo, F. M. C. (2015). 'Finite Element
Model for Laser Welding of Titanium', Procedia CIRP, 33, 434-439.
Chang, B., Allen, C., Blackburn, J. and Hilton, P. (2013). 'Thermal and Fluid Flow
Characteristics and their Relationships with Porosity in Laser Welding of
AA5083', Physics Procedia, 41, 478-487.
238
Chang, D. U. (1987). ' An analytical model for laser reflow soldering of an electronic
component', Welding Journal, 66(11), 323-331.
Chen, H. C. (2010). Fibre Laser Welding of Dissimilar Materials. PhD Thesis, The
University of Manchester.
Chen, H. C., Pinkerton, A. J., Li, L., Liu, Z. and Mistry, A. T. (2011). 'Gap-free fibre
laser welding of Zn-coated steel on Al alloy for light-weight automotive
applications', Materials & Design, 32(2), 495-504.
Chen, J.-S., Pan, C., Roque, C. M. O. L. and Wang, H.-P. (1998a). 'A Lagrangian
reproducing kernel particle method for metal forming analysis', Computational
Mechanics, 22(3), 289-307.
Chen, J.-S., Roque, C. M. O. L., Pan, C. and Button, S. T. (1998b). 'Analysis of metal
forming process based on meshless method', Journal of Materials Processing
Technology, 80–81, 642-646.
Chen, J. K. and Beraun, J. E. (2001). 'Numerical study of ultrashort laser pulse
interactions with metal films', Numerical Heat Transfer, Part A: Applications,
40(1), 1-20.
Chen, L. and Gong, S. L. (2011). 'The research on YAG laser welding porosity of Al-Li
alloy', Advanced Materials Research, 287-290, 2175-2180.
Cheng, C. W., Wang, S. Y., Chang, K. P. and Chen, J. K. (2016). 'Femtosecond laser
ablation of copper at high laser fluence: Modeling and experimental
comparison', Applied Surface Science, 361, 41-48.
Cherepy, N. J., Shen, T. H., Esposito, A. P. and Tillotson, T. M. (2005).
'Characterization of an effective cleaning procedure for aluminum alloys:
surface enhanced Raman spectroscopy and zeta potential analysis', Journal of
Colloid and Interface Science, 282(1), 80-86.
Cho, S.-K., Yang, Y.-S., Son, K.-J. and Kim, J.-Y. (2004). 'Fatigue strength in laser
welding of the lap joint', Finite Elem. Anal. Des., 40(9-10), 1059-1070.
Cirpici, B. K. (2015). Simulating the expansion process of intumescent coating fire
protection. PhD., University of Manchester.
CleanLaser (2013). cleanLASER CL600 Technical sheet. In: (T. K. S. E. GmbH.).
Germany.
Cleary, P. W. and Monaghan, J. J. (1999). 'Conduction Modelling Using Smoothed
Particle Hydrodynamics', Journal of Computational Physics, 148(1), 227-264.
Cleary, P. W., Prakash, M. and Ha, J. (2006). 'Novel applications of smoothed particle
hydrodynamics (SPH) in metal forming', Journal of Materials Processing
Technology, 177, 41-8.
Cole, G. S. and Sherman, A. M. (1995). 'Light weight materials for automotive
applications', Materials Characterization, 35(1), 3-9.
Colina, M., Molpeceres, C., Morales, M., Aliens-Perkins, F., Guadaño, G. and Ocaña, J.
L. (2011). 'Laser ablation modelling of aluminium, silver and crystalline silicon
for applications in photovoltaic technologies', Surface Engineering, 27, 414-423.
Colombier, J., Combis, P., Bonneau, F., Harzic, R. and Audouard, E. (2006).
'Hydrodynamic simulations of metal ablation by femtosecond laser irradiation',
Physical Review B, 71, 165406.
Committee, A. M. H. (1979). Metals handbook. Volume 2, Properties and selection :
nonferrous alloys and pure metals. Metals Park, Ohio: American Society for
Metals.
Couso, E. V. and Gómez, J. V. (2012). 'Laser Beam Welding and Automotive
Engineering', Advanced Structural Materials, 8, 59-84.
239
Crespo, A. J. C., Dominguez, J. M., Barreiro, A., Gomez-Gesteira, M. and Rogers, B.
D. (2011). DualSPHysics, new GPU computing on SPH models. Proc. 6th
International SPHERIC Workshop TUHH Hamburg.
Crespo, A. J. C., Gómez-Gesteira, M. and Dalrymple, R. (2007). 'Boundary conditions
generated by dynamic particles in SPH methods', Computers, Materials, and
Continua, 5(3), 173-184.
Cueto, E. and Chinesta, F. (2015). 'Meshless methods for the simulation of material
forming', International Journal of Material Forming, 8(1), 25-43.
Cueto, E., Sukumar, N., Calvo, B., Martínez, M. A., Cegoñino, J. and Doblaré, M.
(2003). 'Overview and recent advances in natural neighbour galerkin methods',
Archives of Computational Methods in Engineering, 10(4), 307-384.
Dahotre, N. B. and Harimkar, S. P. (2008). Laser fabrication and machining of
materials. New York, N.Y: Springer.
Dai, Y. J., Wu, X. H. and Tao, W. Q. (2011). 'Weighted Least-Squares Collocation
Method (WLSCM) for 2-D and 3-D Heat Conduction Problems in Irregular
Domains', Numerical Heat Transfer, Part B: Fundamentals, 59(6), 473-494.
Dalrymple, R. A. and Rogers, B. D. (2006). 'Numerical modeling of water waves with
the SPH method', Coastal Engineering, 53(2–3), 141-147.
Das, R., Bhattacharjee, K. S. and Rao, S. (2011). 'Welding Heat Transfer Analysis
Using Element Free Galerkin Method', Advanced Materials Research, 410, 298301.
David, S. A. and DebRoy, T. (1992). 'Current issues and problems in welding science',
Science and Technology of Welding & Joining, 27.
Davis, J. R. (1993). Aluminum and Aluminum Alloys (Asm Specialty Handbook)
(06610G): ASM International.
Delaporte, P., Gastaud, M., Marine, W., Sentis, M., Uteza, O., Thouvenot, P., Alcaraz,
J. L., Le Samedy, J. M. and Blin, D. (2003). 'Dry excimer laser cleaning applied
to nuclear decontamination', Applied Surface Science, 208-209, 298-305.
Demuth, C., Bieda, M., Lasagni, A. F., Mahrle, A., Wetzig, A. and Beyer, E. (2012).
'Thermal simulation of pulsed direct laser interference patterning of metallic
substrates using the smoothed particle hydrodynamics approach', Journal of
Materials Processing Technology, 212(3), 689-699.
Dimogerontakis, T., Oltra, R. and Heintz, O. (2005). 'Thermal oxidation induced during
laser cleaning of an aluminium-magnesium alloy', Applied Physics A, 81(6),
1173-1179.
Ding, P. J., Liu, Q. C., Lu, X., Liu, X. L., Sun, S. H., Liu, Z. Y., Hu, B. T. and Li, Y. H.
(2012). 'Hydrodynamic simulation of silicon ablation by ultrashort laser
irradiation', Nuclear Instruments and Methods in Physics Research Section B:
Beam Interactions with Materials and Atoms, 286, 40-44.
Dowden, J., Postacioglu, N., Davis, M. and Kapadia, P. (1987). 'A keyhole model in
penetration welding with a laser', Journal of Physics D., 20, 36-44.
Ducharme, R., Kapadia, P. and Dowden, J. (1993). In: Proc. ICALEO ‘93: ‘Laser
materials processing’, (ed. P. Denney et al.), 1993 Orlando, Florida, Laser
Institute of America.
Efunda.
(2016).
Aluminum
Alloy
AA
3004
[Online].
Available:
http://www.efunda.com/Materials/alloys/aluminum/show_aluminum.cfm?ID=A
A_3004&show_prop=all&Page_Title=AA%203004 [Accessed 02/11/2013].
240
El-Batahgy, A. and Kutsuna, M. (2009). 'Laser Beam Welding of AA5052, AA5083,
and AA6061 Aluminum Alloys', Research Letters in Materials Science, 2009, 19.
Engler, O. and Hirsch, J. (2002). 'Texture control by thermomechanical processing of
AA6xxx Al-Mg-Si sheet alloys for automotive applications', A review journal
Materials Science and Engineering A, 336, 249−262.
ESAB Knowledge Center. (2014). Should I use 4043 or 5356 filler alloy? [Online].
ESAB. Available: http://www.esabna.com/us/en/education/blog/should-i-use4043-or-5356-filler-alloy.cfm [Accessed on 2014].
Fang, J., Zhao, G.-F., Zhao, J. and Parriaux, A. (2009). 'On the Truly Meshless Solution
of Heat Conduction Problems in Heterogeneous Media', Numerical Heat
Transfer, Part B: Fundamentals, 55(1), 1-13.
Fang, Q. T. and Anyalebechi, P. N. (1988). 'Effects of Solidification Conditions on
Hydrogen. Porosity Formation in Aluminum Alloy Castings', Light Metals, The.
Minerals, Metals & Materials Society, Warrendale, PA,, 477-486.
Faussurier, G., Blancard, C. and Silvestrelli, P. L. (2009). 'Evaluation of aluminum
critical point using an \textit{ab initio} variational approach', Physical Review B,
79(13), 134202.
Feliu, J. S., Pardo, A., Merino, M. C., Coy, A. E., Viejo, F. and Arrabal, R. (2009).
'Correlation between the surface chemistry and the atmospheric corrosion of
AZ31, AZ80 and AZ91D magnesium alloys', Applied Surface Science,
255(4102-8).
Filice, L., Alfaro, I., Gagliardi, F., Cueto, E., Micari, F. and Chinesta, F. (2009). 'A
preliminary comparison between finite element and meshless simulations of
extrusion', Journal of Materials Processing Technology, 209(6), 3039-3049.
Fourtakas, G., Vacondio, R. and Rogers, B. D. (2015). 'On the approximate zeroth and
first-order consistency in the presence of 2-D irregular boundaries in SPH
obtained by the virtual boundary particle methods', International Journal for
Numerical Methods in Fluids, 78(8), 475-501.
Gale, W. F. and Totemeier, T. C. (2004). Smithells Metals Reference Book (8th ed.):
Elsevier.
García, J. A., Gascón, L., Cueto, E., Sánchez, F. and Chinesta, F. (2007). 'Meshless
methods with application to Resin Transfer Molding simulation', AIP
Conference Proceedings, 907(1), 1378-1383.
Gillner, A., Bosse, L., Kramer, T., Olowinsky, A. M. and Wild, M. (2000). 'Laser
joining of microcomponents with fiber lasers and diode lasers', 260-263.
Gingell, A. B. D. and Gooch, T. G. (1997). 'Review of factors influencing porosity in
aluminium arc welds', TWI report, 625.
Gingold, R. A. and J.J., M. (1977). 'Smoothed particle hydrodynamics - Theory and
application to non-spherical stars', Monthly Notices of the Royal Astronomical
Society, 181, 375-389.
Gingold, R. A. and Monaghan, J. J. (1977). 'Smoothed particle hydrodynamics-theory
and application to non-spherical stars', Monthly Notices of the Royal
Astronomical Society 181 375-89.
Gómez-Gesteira, M., Rogers, B. D., Crespo, A. J. C., Dalrymple, R. A.,
Narayanaswamy, M. and Dominguez, J. M. (2012). 'SPHysics - development of
a free-surface fluid solver- Part 1: Theory and Formulations', Computers &
Geosciences, 48, 289-299.
241
Gouret, N., Ollivier, E., Dour, G., Fortunier, R. and Miguet, B. (2004). 'Assessment of
the origin of porosity in electron-beam-welded TA6V plates', Metallurgical and
Materials Transactions A, 35(3), 879-889.
Gref, W., Russ, A., Leimser, M., Dausinger, F. and Huegel, H. (2003). Double focus
technique: influence of focal distance and intensity distribution on the welding
process. In, 2003. 289-294.
Greses, J., Barlow, C. Y., Steen, W. M. and Hilton, P. A. (2001). Spectroscopic studies
of plume/plasma in different gas environments. In: ICALEO Proceeding, 2001
LIA, Orlando. 808.
Gross, M. (2008a). Smooth particle hydrodynamics (SPH) modelling of laser cutting.
International Congress on Aplications of Lasers & Electro Optics (ICALEO).
University of Cambridge; Cambridge Great Britain.
Gross, M. (2008b). Smooth particle hydrodynamics (SPH) modelling of laser cutting.
Int. Congress on Aplications of Lasers & Electro Optics (ICALEO). University
of Cambridge; Cambridge Great Britain.
Gualini, M. M. S. (2001). Laser welding of zinc coated steel sheets. An old problem
with a possible solution. 20th International Congress on Applications of Lasers
& Electro–Optics, ICALEO. Jacksonville-Florida.
Guedes, C. F. and César de Sá, J. M. A. (2007). 'Coupling Finite Element And Meshless
Methods To Deal With Contact And Friction In Forging Processes', AIP
Conference Proceedings, 908(1), 1507-1511.
Guo, W., Dong, S., Guo, W., Francis, J. A. and Li, L. (2015). 'Microstructure and
mechanical characteristics of a laser welded joint in SA508 nuclear pressure
vessel steel', Materials Science and Engineering: A, 625, 65-80.
Guo, Y. M. and Nakanishi, K. (2003). 'A backward extrusion analysis by the rigid–
plastic integralless–meshless method', Journal of Materials Processing
Technology, 140(1–3), 19-24.
Gupta, S. R. (2015). Manufacturing processes (Welding) [Online]. NPTL Porject, India.
Available:
http://nptel.iitm.ac.in/courses/Webcourse-contents/IITROORKEE/MANUFACTURING-PROCESSES/welding/lecture13.htm.
[Accessed 04/03/2013].
Gusarov, A. V. and Smurov, I. (2003). 'Near-surface laser–vapour coupling in
nanosecond pulsed laser ablation', Journal of Physics D: Applied Physics,
36(23), 2962.
Gusarov, A. V. and Smurov, I. (2005). 'Thermal model of nanosecond pulsed laser
ablation:Analysis of energy and mass transfer', Journal of Applied Physics,
97(1), 014307.
Haberstroh, E. and Hoffmann, W. M. (2008). 'Laser transmission welding of complex
micro plastic parts', Proceedings of the Institution of Mechanical Engineers,
Part B: Journal of Engineering Manufacture, 222(1), 47-54.
Haboudou, A., Peyre, P. and Vannes, A. B. (2003). Study of keyhole and melt pooi
oscillations in dual beam welding of aluminium alloys: effect on porosity
formation. In: Proceedings of SPIE, 2003a.
Haboudou, A., Peyre, P. and Vannes, A. B. (2004). 'Influence of surface preparation and
process parameters on the porosity generation in aluminum alloys', Journal of
Laser Applications, 16(1), 20-24.
Haboudou, A., Peyre, P., Vannes, A. B. and Peix, G. (2003b). 'Reduction of porosity
content generated during Nd:YAG laser welding of A356 and AA5083
aluminium alloys', Materials Science and Engineering: A, 363(1-2), 40-52.
242
Hallo, L., Riou, O., Stenz, C. and Tikhonchuk, V. T. (2006). 'Infrared emissivity studies
of melting thresholds and structural changes of aluminium and copper samples
heated by femtosecond laser pulses', Journal of Physics D: Applied Physics,
39(24), 5272-5279.
Hamadi, F., Amara, E. H., Kerdja, T. and Mezaoui, D. (2008). 'Numerical modelling of
aluminium vapour produced by nanosecond UV laser beam ablation', physica
status solidi (c), 5(10), 3314-3318.
Harooni, M., Kong, F., Carlson, B. and Kovacevic, R. (2012). Studying the effect of
laser welding parameters on the quality of ZEK100 magnesium alloy sheets in
lap joint configuration. ICALE012. Anaheim, CA.
Hartel, U., Ilin, A., Bantel, C., Gibmeier, J. and Michailov, V. (2016). 'Finite Element
Modeling for the Structural Analysis of Al-Cu Laser Beam Welding', Physics
Procedia, 83, 1404-1414.
Hayashi, T., Matsubayashi, K., Katayama, S., Abe, N., Matsunawa, A. and Ohmori, A.
(2003). 'Reduction mechanism of porosity in tandem twin-spot laser welding of
stainless steel', Welding International, 17(1), 12-19.
Hayat, F. (2011). 'Comparing Properties of Adhesive Bonding, Resistance Spot
Welding, and Adhesive Weld Bonding of Coated and Uncoated DP 600 Steel',
Journal of Iron and Steel Research, International, 18(9), 70-78.
He, B., Huang, S. M. and He, X. (2014). 'Numerical simulation of gear surface
hardening using the finite element method', The International Journal of
Advanced Manufacturing Technology, 74(5), 665-672.
Hoffman, J. and Szymanski, Z. (2002). 'Absorption of the laser beam during welding
with CO2 laser', Optica Applicata, 32(1-2), 129-145
Hongping, G. and Duley, W. W. (1996). 'A statistical approach to acoustic monitoring
of laser welding', Journal of Physics D: Applied Physics, 29 556
Horn, I., Guillong, M. and Günther, D. (2001). 'Wavelength dependant ablation rates for
metals and silicate glasses using homogenized laser beam profiles- implications
for LA-ICP-MS', Applied Surface Science, 182(1–2), 91-102.
Howden, G. D. (1971). 'An up-to-date look at porosity formation in aluminium
weldments', Welding Journal.
Hozoorbakhsh, A., Ismail, M. I. S., Sarhan, A. A. D. M., Bahadoran, A. and Aziz, N. B.
A. (2016). 'An investigation of heat transfer and fluid flow on laser microwelding upon the thin stainless steel sheet (SUS304) using computational fluid
dynamics (CFD)', International Communications in Heat and Mass Transfer,
75, 328-340.
Hu, B. T., Hu, S., Shen, J. and Li, Y. (2015). 'Modeling of keyhole dynamics and
analysis of energy absorption efficiency based on Fresnel law during deeppenetration laser spot welding', Computational Materials Science, 97, 48-54.
Hu, H. and Eberhard, P. (2016). 'Thermomechanically coupled conduction mode laser
welding simulations using smoothed particle hydrodynamics', Computational
Particle Mechanics, 1-14.
Hu, H., Fetzer, F., Berger, P. and Eberhard, P. (2016). 'Simulation of laser welding
using advanced particle methods', GAMM-Mitteilungen, 39(2), 149-169.
Hu, X. Y. and Adams, N. A. (2006). 'A multi-phase SPH method for macroscopic and
mesoscopic flows', Journal of Computational Physics, 213(2), 844-861.
Hussein, A. E., Diwakar, P. K., Harilal, S. S. and Hassanein A. (2013). 'The role of laser
wavelength on plasma generation and expansion of ablation plumes in air',
Journal of Applied Physics, 113, 143305.
243
Imen, K., Lee, S. J. and Allen, S. D. (1991). 'Laser-assisted micron scale particle
removal', Applied Physics Letter, 58, 203.
Ion, J. C. (2005a). Laser processing of engineering materials: principles, procedure and
industrial application. Amsterdam: Boston.
Ion, J. C. (2005b). Laser processing of engineering materials: principles, procedure
and industrial application. Amsterdam: Boston.
Ito, M., Izawa, S., Fukunishi, Y. and Shigeta, M. (2012). SPH simulation of gas arc
welding process. Seventh International Conference on Computational Fluid
Dynamics (ICCFD7). Big Island, Hawaii.
Jain, M., Allin, J. and Bull, M. J. (1998). 'Deep drawing characteristics of automotive
aluminum alloys', Materials Science and Engineering, A256, 69-82.
Jeong, J. H., Jhon, M. S., Halow, J. S. and van Osdol, J. (2003). 'Smoothed particle
hydrodynamics: Applications to heat conduction', Computer Physics
Communications, 153(1), 71-84.
Jeong, S. H., Greif, R. and Russo, R. E. (1998). 'Numerical modeling of pulsed laser
evaporation of aluminum targets', Applied Surface Science, 127–129, 177-183.
Jerniti, A., El Ouafi, A. and Barka, N. (2016). 'Single Track Laser Surface Hardening
Model for AISI 4340 Steel Using the Finite Element Method', Modeling and
Numerical Simulation of Material Science, 6, 17-27.
Jorgensen, M. (1980). 'Increasing energy-absorption in laser-welding', Metal
Construction, 12, 88.
Joshi, S., Fahrenholtz, W. G. and O’Keefe, M. J. (2011). 'Effect of alkaline cleaning and
activation on aluminum alloy 7075-T6', Applied Surface Science, 257(6), 18591863.
Kamat, R. G., Butler, J. F., Murtha, S. J. and Bovard, F. S. (2002). 'Alloy 6022-T4E29
for automotive sheet applications', Materials Science Forum, 396(402),
1591−1596.
Kamimuki, K., Inoue, T., Yasuda, K., Muro, M., Nakabayashi, T. and Matsunawa, A.
(2002). 'Prevention of welding defect by side gas flow and its monitoring
method in continuous wave Nd:YAG laser welding', Journal of Laser
Applications, 14(3), 136.
Kaplan, A. (1994). 'A model of deep penetration laser welding based on calculation of
the keyhole profile', Journal of Physics D: Applied Physics, 27, 1805-1814.
Katayama, S. and Kawahito, Y. (2009). Elucidation of phenomena in high-power fiber
laser welding and development of prevention procedures of welding defects.
Proc. SPIE 7195, Fiber Lasers VI: Technology, Systems, and Applications.
Katayama, S., Kawahito, Y. and Mizutani, M. (2007a). Collaboration of physical and
Metallurgical viewpoints for understanding and process development of Laser
Welding. ICALEO 2007 Congress Proceedings (Proc. of 26th International
Congress on Applications of Lasers & Electro-Optics). Orlando, LIA.
Katayama, S., Kawahito, Y. and Mizutani, M. (2010). 'Elucidation of laser welding
phenomena and factors affecting weld penetration and welding defects', Physics
Procedia, 5, 9-17.
Katayama, S., Matsunawa, A. and Kojima, K. (1998). 'CO2 laser weldability of
aluminium alloys: Defects formation and causes', Welding International, 12(2),
774-789.
Katayama, S., Mizutani, M. and Matsunawa, A. (2003). Development of porosity
prevention procedures during laser welding. In: SPIE Proceedings, 2003. 281288.
244
Katayama, S., Naito, Y., Uchiumi, S. and Mizutani, M. (2006). 'Physical Phenomena
and Porosity Prevention Mechanism in Laser-Arc Hybrid Welding',
Transactions JWRI, Research Organization Report, 35(1), 13-18.
Katayama, S., Tanaka, K., Mizutani, M. and Matsunawa, A. (2000). Pulsed YAG laser
spot welding under microgravity. In: Proceedings of SPIE, 2000.
Katayama, S., Uchiumi, S., Mizutani, M., Wang, J. and Fuji, K. (2007b). 'Penetration
and porosity prevention mechanism in YAG laser-MIG hybrid welding',
Welding International, 21(1).
Kawahito, Y., Kito, M. and Katayama, S. (2007). 'Adaptive Gap Control in
ButtWelding with a Pulsed YAG Laser', Transactions of JWRI, 36(2), 5-10.
Kelly, R. (1990). 'On the dual role of the Knudsen layer and unsteady, adiabatic
expansion in pulse sputtering phenomena', The Journal of Chemical Physics,
92(8), 5047-5056.
Kersten, H., Steffen, H. and Behnke, J. F. (1996). 'Investigations on plasma-assisted
surface cleaning of aluminum contaminated with lubricants', Surface and
Coatings Technology, 86-87, 762-768.
Khan, M. I. (2007). Welding Science and Technology: New Age International.
Khan, M. M. A., Romoli, L., Fiaschi, M., Dini, G. and Sarri, F. (2012). 'Laser beam
welding of dissimilar stainless steels in a fillet joint configuration', Journal of
Materials Processing Technology, 212(4), 856-867.
Kim, J. D., Kim, Y. H. and Oh, J. S. (2004). 'Diagnostics of Laser-Induced Plasma in
Welding of Aluminum Alloy', Key Engineering Materials, 261-263, 1671-1676.
Kim, M. J. (2011). 'Meshfree isoparametric point interpolation method (IPIM) for
evaporative laser drilling', Applied Mathematical Modelling, 35(6), 2639-2649.
Kim, M. J. (2012). 'Meshfree isoparametric finite point interpolation method (IFPIM)
with weak and strong forms for evaporative laser drilling', Applied Mathematical
Modelling, 36(4), 1615-1625.
Klages, K., Gillner, A., Olowinsky, A. M., Sandra, F. and Studt, A. (2003). 'laser beam
micro-welding of dissimilar materials', Proceedings of SPIE, 5063, 303-307.
Klein, T., Vicanek, M., Kroos, J., Decker, I. and Simon, G. (1994). 'Oscillations of the
keyhole in penetration laser beam welding', Journal of Physics D: Applied
Physics, 27(10), 2023.
Klein, T., Vicanek, M. and Simon, G. (1996). 'Forced oscillations of the keyhole in
penetration laser beam welding', Journal of Physics D: Applied Physics, 29(2),
322.
Kou, S. (2002). Welding Metallurgy (2nd ed.): Wiley-Interscience.
Kroos, J., Gratzke, U. and Simon, G. (1993). 'Towards a self-consistent model of the
keyhole in penetration laser beam welding', Journal of Physics D: Applied
Physics, 26, 474-480.
Kudryashov, S. and Allen, S. D. (2006). 'Experimental and Theoretical Studies of Laser
Cleaning Mechanisms for Submicrometer Particulates on Si Surfaces',
Particulate Science and Technology, 24(3), 281-299.
Kuo, T. Y. and Lin, H. C. (2006a). Material Science and Enigneering A, (A416), 281289.
Kuo, T. Y. and Lin, H. C. (2006b). 'Effects of pulse level of Nd-YAG laser on tensile
properties and formability of laser weldments in automotive aluminum alloys',
Materials Science and Engineering: A, 416(1-2), 281-289.
245
Kuo, T. Y. and Lin, H. C. (2006c). 'Effects of pulse level of Nd-YAG laser on tensile
properties and formability of laser weldments in automotive aluminum alloys',
Materials Science and Engineering, A 416, 281-289.
Kutnusa, M., Suzuki, J., KIimura, S., Sugiyama, S., Yuhki, M. and Yamaoka, H.
(1993). Weld. World, 31(2), 126-135.
Kutsuna, M., Kitamura, S., Shibata, K., Sakamoto, H. and Tsushima, K. (2006).
'Improvement of the joint performance in laser welding of aluminium alloys',
Welding in the World, 50(1/2).
Kwok, C. T., Fong, S. L., Cheng, F. T. and Man, H. C. (2006). 'Pitting and galvanic
corrosion behavior of laser-welded stainless steels', Journal of Materials
Processing Technology, 176(1–3), 168-178.
Kwon, K.-C., Park, S.-H. and Youn, S.-K. (2005). 'The least-squares meshfree method
for elasto-plasticity and its application to metal forming analysis', International
Journal for Numerical Methods in Engineering, 64(6), 751-788.
Kwon, K.-C. and Youn, S.-K. (2006). 'The least-squares meshfree method for rigidplasticity with frictional contact', International Journal of Solids and Structures,
43(25–26), 7450-7481.
Labibzadeh, M. (2016). 'Voronoi based discrete least squares meshless method for heat
conduction simulation in highly irregular geometries', Chinese Journal of
Mechanical Engineering, 29(1), 98-111.
Larsson, J. K. (1999). 'Lasers for various materials processing. A review of the latest
applications in automotive manufacturing. . Lappeenranta: Acta Universitatis
Lappeenrantaensis', Proceedings of the 7th Nordic Conference on Laser
Materials Processing (NOLAMP-7).
Lappeenranta: Acta Universitatis
Lappeenrantaensis., 26-37.
Lee, E. S. (2007). Truly incompressible approach for computing incompressible flow in
SPH and comparisons with the traditional weakly compressible approach. PhD
Thesis, The University of Manchester.
Leidich, H. (2016). Autogenous Laser Welding of Aluminum [Online]. TRUMPF Inc.
Available:
http://www.us.trumpf.com/fileadmin/DAM/us.trumpf.com/Brochures/Laser_Tec
hnology/Autogenous_Laser_Welding_of_Aluminum_Leidich.pdf [Accessed 1512-2016].
Li, G. and Belytschko, T. (2001). 'Element free Galerkin method for contact problems
in metal forming analysis', Engineering Computations, 18(1/2), 62-78.
Lima, M. S. F., Wagnière, J., Morato, S. P. and Vieira, J. (2002). 'Elimination of
Lubricants from Aluminum Cold Rolled Products Using Short Laser Pulses',
Materials Research, 5(2), 205-208.
Lind, S. J., Xu, R., Stansby, P. K. and Rogers, B. D. (2012). 'Incompressible smoothed
particle hydrodynamics for free-surface flows: A generalised diffusion-based
algorithm for stability and validations for impulsive flows and propagating
waves', Journal of Computational Physics, 231(4), 1499-1523.
Liu, G. R. and Liu, M. B. (2003). Smoothed particle hydrodynamics: a meshfree
particle method. New Jersey: World Scientific.
Liu, L., Song, G. and Wang, J. (2005a). Material Science., 488, 361-364.
Liu, L. H. (2006). 'Meshless Local Petrov-Galerkin Method for Solving Radiative
Transfer Equation', Journal of Thermophysics and Heat Transfer, 20(1), 150154.
246
Liu, L. M., Song, G., Liang , G. and Wang, J. F. (2005b). Materials Science and
Engineering A, A390, 76-80.
Liu, L. M., Zhang, Z. D., Shen , Y. and Wang, L. (2006). Acta Metallurgica Sinica, 42,
399-404.
Liu, L. M. and Zhao, X. (2008). 'Study on the weld joint of Mg alloy and steel by laserGTA hybrid welding', Materials Characterization, 59(9), 1279-1284.
Liu, W., Li, S. and Belytscho, T. (1997). 'Moving least square Kernel Galerkin method
(I) methodology and convergence', Computer Methods in Applied Mechanics
and Engineering., 143:113.
Liyang, Y., Zengbo, W. and Lin, L. (2012). 'Multiphysics modelling and simulation of
dry laser cleaning of micro-slots with particle contaminants', Journal of Physics
D: Applied Physics, 45(13), 135401.
Lu, Y. F., Song, W. D., Zhang, Y., Hong, M. H. and Chong, T. C. (2001). 'Laser
removal of particles from solid surfaces', RIKEN Review, 32 64-70.
Lucy, L. B. (1977). 'A numerical approach to the testing of the fission hypothesis',
Astronom J., 82 1013-24.
Luk‘yanchuk, B. S., Arnold, N., Huang, S. M., Wang, Z. B. and Hong, M. H. (2003).
'Three-dimensional effects in dry laser cleaning', Applied Physics A, 77(2), 209215.
Lunney, J. G. and Jordan, R. (1998). 'Pulsed laser ablation of metals', Applied Surface
Science, 127–129, 941-946.
Lutey, A. H. A. (2013a). 'An improved model for nanosecond pulsed laser ablation of
metals', Journal of Applied Physics, 114(8), 083108.
Lutey, A. H. A. (2013b). 'Modeling of Thin-Film Single and Multilayer Nanosecond
Pulsed Laser Processing', Journal of Manufacturing Science and Engineering,
135(6), 061003-061003.
Lutey, A. H. A., Sozzi, M., Selleri, S., Cucinotta, A. and Molari, P. G. (2013).
Picosecond and nanosecond pulsed laser ablation of aluminium, polypropylene,
polyethylene, and their thin-film combinations. In, 2013. 860804-860804-11.
Lutey, H. A. (2013c). 'An improved model for nanosecond pulsed laser ablation of
metals', Journal of Applied Physics, 114(8), 083108-10.
Mäkikangas, J., Mäntyjärvi, K., Keskitalo, M., Karjalainen, J. A., Niemela, J. and Ojala,
J. (2007). Laser welding of coated sheet metal constructions. 11th NOLAMP
Conference in Laser Processing of Materials. Lappeenranta
Marczak, J., Jach, K. and Sarzyński, A. (2005). Numerical Modelling of Laser Cleaning
and Conservation of Artworks. In: Dickmann, K., Fotakis, C. & Asmus, J. (eds.)
Lasers in the Conservation of Artworks. Springer Berlin Heidelberg.
Marimuthu, S., Kamara, A. M., Whitehead, D., Mativenga, P. T. and Li, L. (2010).
'Laser Removal of TiN Coatings from WC Micro-tools and In-process
Monitoring', Optics and Laser Technology, 42(8), 1233-1239.
Marimuthu, S., Mhich, A., Molchan, I. S., Whitehead, D., Wang, Z. B., Mativenga, P.,
Li, L., Liu, Z., Grafton-Reed, C., Cheetham, S. and Dilworth, S. (2013).
'Numerical simulation of excimer laser cleaning of film and particle
contaminants', Journal of Heat Transfer, 135(12), 121301-121301.
Marla, D., Bhandarkar, U. V. and Joshi, S. S. (2013). 'Transient Analysis of Laser
Ablation Process With Plasma Shielding: One-Dimensional Model Using Finite
Volume Method', Journal of Micro and Nano-Manufacturing, 1(1), 011007011007.
247
Martukanitz, R. P. (2005). 'Critical Review: Industrial Lasers and Applications',
Proceedings of the SPIE, 5706, 11-24.
Massalski, T. B. (1990). Binary Alloy Phase Diagrams (2nd Edition ed. Vol. 1). OH,
U.S.A.: ASM International, Materials Park.
Mathers, G. (2002). The welding of aluminium and its alloys. Boca Raton, FL: CRC
Press/Woodhead Publishing.
Matrukanitz, R. P. (1990). Selection and weldability of heat-treatable aluminum alloys.
ASM Handbook-Welding, Brazing and Soldering (Vol. 6).
Matsunawa, A., Katayama, S., Ikeda, H. and Nishizawa, K. (1992). Effect of pulse
shaping on defect reduction in pulsed laser welding. In: Proc. ICALEO ’92:
‘Laser materials processing’, (ed. D. Farson et al.), 1992 Orlando, FL, Laser
Institute of America.
Matsunawa, A., Kim, J.-D., Seto, N., Mizutani, M. and Katayama, S. (1998). 'Dynamics
of keyhole and molten pool in laser welding', Journal of Laser Applications,
10(6), 247.
Matsunawa, A., Mizutani, M. and Katayama, S. (2003). 'Porosity formation mechanism
and its prevention in laser welding', Welding International, 17 (6), 431-437.
Matsunawa, A. and Semak, V. (1971). 'The Simulation of Front Keyhole Wall
Dynamics During Laser Welding', Journal of Physics: D, 30, 798-809.
Mayrhofer, A., Rogers, B. D., Violeau, D. and Ferrand, M. (2013). 'Investigation of wall
bounded flows using SPH and the unified semi-analytical wall boundary
conditions', Computer Physics Communications, 184(11), 2515-2527.
Mazur, M. (1992). 'Porosity in aluminium welds', Welding International, 6(12), 929931.
Meja, P., Autric, M., Delaporte, P. and Alloncle, P. (1999). 'Dry laser cleaning of
anodized aluminium', Applied Physics A, 69 [Suppl.], S343-S346.
Melik, J. and Kouzmenkov, A. (2010). Aluminium demand takes its toll [Online]. BBC
World Service. Available: http://www.bbc.co.uk/news/business-11692804
[Accessed 13/03/2013].
Meng, W., Li, Z., Huang, J., Wu, Y. and Cao, R. (2013). 'Effect of gap on plasma and
molten pool dynamics during laser lap welding for T-joints', The International
Journal of Advanced Manufacturing Technology, 69(5), 1105-1112.
Messler, R. W. (1999). Principles of welding: processes, physics, chemistry, and
metallurgy. New York: John Wiley.
Metcalfe, G. E. (1945). Journal of the Institute of Metals, 71, 618.
Mills, K. C. (2002). Recommended values of thermopysical properties for selected
commercial alloys. Cambridge: Wodhead Publishing Limited.
Minnick, W. H. (2008). Gas Metal Arc Welding Handbook (5th ed.). Tinley Park, IL,
USA: The Goodheart-Willcox Co.
Mirzaei, D. and Schaback, R. (2014). 'Solving heat conduction problems by the Direct
Meshless Local Petrov-Galerkin (DMLPG) method', Numerical Algorithms,
65(2), 275-291.
Miyamoto, I., Kosumi, T., Park, S.-J., Uragishi, H., Watanabe, K. and Ooie, T. (2004).
'Applications of single-mode fiber lasers to novel microwelding', 507-514.
Mohanty, S., Laldas, C. K. and Roy, G. G. (2012). 'A new model for keyhole mode
laser welding using FLUENT', Transactions of the Indian Institute of Metals,
65(5), 459-466.
Monaghan, J., Huppert, H. and Grae Worster, M. (2005). 'Solidification using smoothed
particle hydrodynamics', Journal of Computational Physics, 206, 684–705.
248
Monaghan, J. and Kos, A. (1999). 'Solitary Waves on a Cretan Beach', Journal of
Waterway, Port, Coastal, and Ocean Engineering, 125(3).
Monaghan J.J. (2005). ' Smoothed particle hydrodynamics', Rep. Prog. Phys., 68, 17031759.
Monaghan, J. J. (1992). 'Smoothed Particle Hydrodynamics', Ann. Rev. Astron. Astro,
30, 543.
Mondolfo, L. F. (1976). Aluminum alloys: structure and properties (Vol. 5):
Butterworths London.
Morel, V., Bultel, A. and Chron, B. G. (2008). Modelling of the creation phase of an
aluminium laser-induced plasma. In: Gas Discharges and Their Applications,
2008. 17th International Conference on, 7-12 Sept. 2008. 581-584.
Morris, J. P., Fox, P. J. and Zhu, Y. (1997). 'Modeling low reynolds number
incompressible flows using SPH', Journal of Computational Physics, 136(1),
214-226.
Muhammad, N. (2012). Laser micromachining of coronary stents for medical
applications. Doctor of Philosophy (PhD), The Univeristy of Manchester.
Muhammad, N., Rogers, B. D. and Li, L. (2013). 'Understanding the behaviour of
pulsed laser dry and wet micromachining processes by multi-phase smoothed
particle hydrodynamics (SPH) modelling', J. Phys. D: Appl. Phys., 46 095101
Murtha, S. J. (2000). USA patent application 6129792.
Naito, Y., Katayama, S. and Matsunawa, A. (2002). Keyhole behaviour and liquid flow
in molten pool during laser-arc hybrid welding. In: Proc. 1st Int. Symp. on
‘High-power laser macroprocessing’, 2002 Osaka, Japan, Osaka University.
357-362.
Narikiyo, T., Miura, H., Fujinaga, S. and Ohmori, A. (1996). 'Plume aspects and
penetration shape with inclined YAG laser beams', Opt Lett, 21(19), 1562-3.
Nath, A. K., Gupta, A. and Benny, F. (2012). 'Theoretical and experimental study on
laser surface hardening by repetitive laser pulses', Surface and Coatings
Technology, 206(8–9), 2602-2615.
Nijhuis, V. M. J. (2011). Adaptive mesh refinement for neutron transport. Bachelor, TU
Delft University of Technology.
Novelis Deutschland GmbH (2011). AC-170Px Aluminium Technical Specifications.
Stuttgart, Germany.
Oderji, H. Y., Farid, N., Sun, L., Fu, C. and Ding, H. (2016). 'Evaluation of explosive
sublimation as the mechanism of nanosecond laser ablation of tungsten under
vacuum conditions', Spectrochimica Acta Part B: Atomic Spectroscopy, 122, 18.
Ohnishi, N., Ishii, A., Kuramitsu, Y., Morita, T., Sakawa, Y. and Takabe, H. (2015).
'Numerical analysis of hydrodynamic instability in magnetized laser ablation
flow', High Energy Density Physics, 17, Part A, 18-23.
Okon, P., Dearden, G., Watkins, K., Sharp, M. and French, P. (2002). Laser welding of
aluminium alloy 5083. 21st International Congress on Applications of Lasers &
Electro-OpticsOctober 14-17 (ICALEO 2002). Scottsdale.
Ol'shanskii, A. N. and Morozov, B. P. (2004). 'Evaluation of the effect of increased
pressure of shielding gas on the decrease in porosity of welded joints in
aluminium alloys', Welding International, 18(4), 304-306.
Olabode, M., Kah, P. and Martikainen, J. (2013). 'Aluminium alloys welding processes:
challenges, joint types, and process selection', Proceedings of the Institution of
Mechanical Engineers, Part B: Journal of Engineering Manufacture.
249
Oltra, R., Yavaş, O., Cruz, F., Boquillon, J. P. and Sartori, C. (1996). 'Modelling and
diagnostic of pulsed laser cleaning of oxidized metallic surfaces', Applied
Surface Science, 96–98, 484-490.
Pakdil, M., Çam, G., Koçak, M. and Erim, S. (2011). 'Microstructural and mechanical
characterization of laser beam welded AA6056 Al-alloy', Materials Science and
Engineering: A, 528(24), 7350-7356.
Palmeri, D., Buffa, G., Barcellona, A. and Fratini, L. (2007). 'Friction stir spot welding
of AA6082-T6: Influence of the most relevant process parameters and
comparison with classic mechanical fastening techniques', Proceedings of the
Institution of Mechanical Engineers, Part B: Journal of Engineering
Manufacture, 221(7), 1111-1118.
Pan, W., Li, D., Tartakovsky, A. M., Ahzi, S., Khraisheh, M. and Khaleel, M. A.
(2013). 'A new smoothed particle hydrodynamics non-Newtonian model for
friction stir welding: Process modeling and simulation of microstructure
evolution in a magnesium alloy', International Journal of Plasticity, 48, 189204.
Pan, Y. and Richardson, I. M. (2007). Study of laser welding of zinc coated steel sheets
in overlap configuration. 11th NOLAMP Conference in Laser Processing of
Materials. Lappeenranta
Park, Y. W., Yu, J. and Rhee, S. (2010). 'A study on the weld characteristics of 5182
aluminum alloy by Nd:Yag laser welding with filler wire for car bodies',
International Journal of Automotive Technology, 11(5), 729-736.
Pashby, I. R., Hoult, A. P. and Nyiau, C. C. (1997). Precleaning and spot welding of an
automative body steel using a Nd:YAG laser. In, 1997. 28-36.
Pastor, M. (1998). Master thesis. The Pennsylvania State University.
Pastor, M., Zhao, H., Martukanitz, R. P. and Debroy, T. (1999). 'Porosity, Underfill and
Magnesium Loss during Continuous Wave Nd:YAG Laser Welding of Thin
Plates of Aluminum Alloys 5182 and 5754', Welding Research Supplement,
207S-216S.
Pham, X.-T. (2013). 'Two-Dimensional Rosenthal Moving Heat Source Analysis Using
the Meshless Element Free Galerkin Method', Numerical Heat Transfer, Part A:
Applications, 63(11), 807-823.
Pham, X.-T., Pradinc, G., Cao, X. and Fihey, J.-L. (2012). Modeling of moving heat
sources using meshless element free Galerkin method Conference: 9th
International Conference on Trends in Welding Research American Society for
Metals.
Picasso, M. and Hoadley, A. F. A. (1994). 'Finite element simulation of laser surface
treatments including convection in the melt pool', Interrnatonal Journal of
Numerical and Mathmatical Heat Fluid Flow, 4, 61-83.
Pinto, L. A., Luisa, Q., Rosa, M. M. and Phill, C. (2010). 'Laser welding of dissimilar
aluminium alloys with filler materials', Welding in the World, 54(11/12).
Povarnitsyn, M. E., Fokin, V. B. and Levashov, P. R. (2015). 'Microscopic and
macroscopic modeling of femtosecond laser ablation of metals', Applied Surface
Science, 357, Part A, 1150-1156.
Psyllaki, P. and Oltra, R. (2000). 'Preliminary study on the laser cleaning of stainless
steels after high temperature oxidation', Mater Sci Eng, 282(1-2), 145.
Punkari, A., Weckman, D. C. and Kerr, H. W. (2003). 'Effects of magnesium content on
dual beam Nd : YAG laser welding of Al – Mg alloys', Science and Technology
of Welding and Joining, 8(4), 269-281.
250
Qiang, W., Yingchun, G., Baoqiang, C. and Bojin, Q. (2016). 'Laser cleaning of
commercial Al alloy surface for tungsten inert gas welding', Journal of Laser
Applications, 28(2), 022507.
Quintino, L., Costa, A., Miranda, R., Yapp, D., Kumar, V. and Kong, C. J. (2007).
'Welding with high power fiber lasers – A preliminary study', Materials and
Design, 28(4), 1231-1237.
Rai, R. (2009). Modeling of heat transfer and fluid flow in keyhole mode welding. PhD,
The Pennsylvania State University.
Rajendiran, S., Rossall, A. K., Gibson, A. and Wagenaars, E. (2014). 'Modelling of laser
ablation and reactive oxygen plasmas for pulsed laser deposition of zinc oxide',
Surface and Coatings Technology, 260(0), 417-423.
Rajendran, S., Keidar, M. and Boyd, I. (2007). Ablation and Plasma Formation Due to
Laser Irradiance. AIAA-2007-4378, 38th AIAA Plasmadynamics and Lasers
Conference In conjunction with the 16th International Conference on MHD
Energy Conversion. Miami, Florida.
Rajput, R. K. (2007). A Textbook of Manufacturing Technology: Manufacturing
Processes. First Edition. Laximi Publications Ltd.
Ramasamy, S. and Albright, C. E. (2000). J. Laser Appl., 12 101.
Ransley, C. E. and Neufeld, H. (1989). 'The solubility of hydrogen in molten aluminum
alloys', Metallurgical Transactions A, 20(9), 1785-1791.
Rapp, J., Glumann, C., Dausinger, F. and Hogel, H. (1995). 'Laser welding of
aluminium lightweight materials: problems, solutions, readiness for application',
Optical and Quantum Electronics, (27), 1203-1211.
Rechner, R., Jansen, I. and Beyer, E. (2010). 'Influence on the strength and aging
resistance of aluminium joints by laser pre-treatment and surface modification',
International Journal of Adhesion and Adhesives, 30(7), 595-601.
Reinhold, B., Günther, R. and Johannes, A. (2002). 'Nd: YAG Laser Beam Welding of
6013 Aluminium Alloy Sheet Using Different Filler Powders', Materials Science
Forum, 396-402, 1691-1696.
Reutzel, E. W., Sullivan, M. J. and Mikesic, D. A. (2006). Weld. J., 85, 66-71.
Ribic, B., Palmer, T. A. and DebRoy, T. (2009). 'Problems and issues in laser-arc hybrid
welding', International Materials Reviews, 54(4), 223-244.
Roache, P. J. (1994). 'Perspective: A method for uniform reporting of grid refinement
studies', Journal of Fluid Engineering, 116, 405-413.
Roache, P. J. (1997). 'Quantification of uncertainty in computational fluid dynamics',
Annual Review of Fluid Mechanics, 29(1), 123-160.
Rogers, B. D. and Dalrymple, R. A. (2008). 'SPH modeling of Tsunami waves',
Advances in Coastal and Ocean Engineering , Advanced Numerical Models for
Tsunami Waves and Runup (Singapore: World Scientific), 10.
Russo, R. E., Mao, X., Liu, H. J., Gonzalez, J. and Mao, S. S. (2002). 'Laser ablation in
analytical chemistry-a review', Talanta, 57(3), 425-451.
Sabatino, M. D. (2005). Fluidity of aluminium foundry alloys. PhD, Norwegian
University of Science and Technology (NTNU).
Sabbaghzadeh, J., Hamedi, M. J., Ghaini, F. M. and Torkamany, M. J. (2008). 'Effect of
Process Parameters on the Melting Ratio in Overlap Pulsed Laser Welding',
Metallurgical and Materials Transactions B, 39(2), 340-347.
Sánchez-Amaya, J. M., Delgado, T., De Damborenea, J. J., Lopez, V. and Botana, F. J.
(2009a). 'Laser welding of AA 5083 samples by high power diode laser', Science
and Technology of Welding and Joining, 14(1), 78-86.
251
Sánchez-Amaya, J. M., Delgado, T., González-Rovira, L. and Botana, F. J. (2009b).
'Laser welding of aluminium alloys 5083 and 6082 under conduction regime',
Applied Surface Science, 255(23), 9512-9521.
Sands, D. (2011). Heat Transfer - Engineering Applications: Pulsed Laser Heating and
Melting [Online]. InTech. [Accessed.
Schauer, D. A. and Giedt, W. H. (1978). 'Prediction of Electron Beam Welding. Spiking
Tendency', Weld J., 57(7), 189-s.
Schneider, A., Avilov, V., Gumenyuk, A. and Rethmeier, M. (2013). 'Laser beam
welding of aluminum alloys under the influence of an electromagnetic field',
Physics Procedia, 41 (4 - 11).
Schubert, E., Klassen, M., Zerner, I., Walz, C. and Sepold, G. (2001). 'Light-weight
structures produced by laser beam joining for future applications in automobile
and aerospace industry', Journal of Materials Processing Technology, 115(1), 28.
Schwaiger, F. (2008). 'An implicit corrected SPH formulation for thermal diffusion with
linear free surface boundary conditions', International Journal for Numerical
Methods in Engineering, 75(6), 647-671.
Sekhar, N. C. 'Review of laser welding of aerospace aluminium alloys', CEMWAM
report, CEMWAM II/ PT1B/TWI/038.
Seto, N., Katayama, S. and Matsunawa, A. (2000). 'High-speed simultaneous
observation of plasma and keyhole behavior during high power CO[sub 2] laser
welding: Effect of shielding gas on porosity formation', Journal of Laser
Applications, 12(6), 245.
Shalahim, N. S. M., Mon, T. T., Ismail, M. F., Rashid, M. F. F. and Rejab, M. R. M.
(2010). Finite element simulation of laser-micromachining. In:
Proc.
International MultiConference of Engineers and Computers Scientists (IMECS),
2010 Hong Kong.
Shangwu, X., Liu, W. K., Cao, J., Rodrigues, J. M. C. and Martins, P. A. F. (2003). 'On
the utilization of the reproducing kernel particle method for the numerical
simulation of plane strain rolling', International Journal of Machine Tools and
Manufacture, 43(1), 89-102.
Shibahara, M. and Atluri, S. N. (2011). 'The meshless local Petrov-Galerkin method for
the analysis of heat conduction due to a moving heat source, in welding',
International Journal of Thermal Sciences, 50(6), 984-992.
Shibata, K., Iwase, T., Sakamoto, H., Dausinger, F. H., Hohenberger, B., Mueller, M.,
Matsunawa, A. and Seto, N. (2003). Process stabilization by dual focus laser
welding of aluminum alloys for car body. In: Proceedings of SPIE, 2003 Naoki
Setoc.
Shusser, M. (2009). 'Quasi-one-dimensional kinetic theory model of laser ablation of
carbon', Journal of Applied Physics, 105(9), 093534.
Singh, A., Singh, I. V. and Prakash, R. (2007). 'Meshless element free Galerkin method
for unsteady nonlinear heat transfer problems', International Journal of Heat
and Mass Transfer, 50(5–6), 1212-1219.
Singh, I. V. (2004). 'Meshless EFG method in three-dimensional heat transfer problems:
a numerical comparison, cost and error analysis', Numerical Heat Transfer, Part
A: Applications, 46(2), 199-220.
Singh, K. S. and Sharma, A. K. (2016). 'Effect of variation of magnetic field on laser
ablation depth of copper and aluminum targets in air atmosphere', Journal of
Applied Physics, 119(18), 183301.
252
Sladek, J., Sladek, V. and Hon, Y. C. (2006). 'Inverse heat conduction problems by
meshless local Petrov–Galerkin method', Engineering Analysis with Boundary
Elements, 30(8), 650-661.
Song, W. D., Lu, Y. F., Hong, M. H., Chong, T. C. and Low, T. S. (1996). Theoretical
and experimental investigation on laser dry cleaning of particles from substrate
surfaces. In:
Advanced Applications of Lasers in Materials
Processing/Broadband Optical Networks/Smart Pixels/Optical MEMs and Their
Applications. IEEE/LEOS 1996 Summer Topical Meetings:, 5-9 Aug. 1996. 6970.
Song, Y., Cross, M. D. J., Rainforth, W. M. and Wynne, B. P. (2007). 'Observations of
strain induced precipitation during the thermomechanical processing of AA6111
alloy', Materials Science Forum, 550, 211−216.
Sophy, T., Silva, A. and Kribèche, A. (2012). 'A Space-Time Meshless Method That
Removes Numerical Oscillations When Solving PDEs with High
Discontinuities', Numerical Heat Transfer, Part B: Fundamentals, 62(1), 50-70.
Spalding, I. (1987). 'Modern laser applications', ARCHIVE: Proceedings of the
Institution of Mechanical Engineers, Part B: Management and Engineering
Manufacture 1983-1988, 197-202(201), 165-74.
Steen, W. M. (1998). Laser Material Processing (2nd ed.): Springer-Verlag.
Steen, W. M., Dowden, J., Davis, M. and Kapadia, P. (1988). 'A point and line source
model of laser keyhole welding', J. Phys. D: Appl. Phys. , 21 (1255).
Steen, W. M. and Mazumder, J. (2010). Laser Material Processing (4th ed.): Springer.
Story, J. M., Jarvis, G. W., Zonker, H. R. and Murtha, S. J. (1993). 'Issues and trends in
automotive aluminum sheet forming,' SAE Publication, SP-944, 1-25.
Sun, Z. and Ion, J. C. (1995). 'Laser welding of dissimilar metal combinations', Journal
of Materials Science, 30(17), 4205-4214.
Sun, Z. and Karppi, R. (1996). 'The application of electron beam welding for the joining
of dissimilar metals: an overview', Journal of Materials Processing Technology,
59(3), 257-267.
Szekely, J. (1979). Fluid flow phenomena in metals processing (Vol. 237). Toronto,
Ont., Canada: Academic Press.
Taban, E., Gould, J. E. and Lippold, J. C. (2010). 'Dissimilar friction welding of 6061T6 aluminum and AISI 1018 steel: Properties and microstructural
characterization', Materials & Design, 31(5), 2305-2311.
Tam, A. C., Leung, W. P., Zapka, W. and Ziemlich, W. (1992). 'Laser cleaning
techniques for removal of surface particulates', Journal of Applied Physics, 71
(7), 3515.
Tangwarodomnukun, V. (2016). 'Cavity formation and surface modeling of laser
milling process under a thin-flowing water layer', Applied Surface Science, 386,
51-64.
Tani, G., Ascari, A., Campana, G. and Fortunato, A. (2007). 'A study on shielding gas
contamination in laser welding of non-ferrous alloys', Applied Surface Science,
254(4), 904-907.
Tao, W., Yang, Z., Chen, Y., Li, L., Jiang, Z. and Zhang, Y. (2013). 'Double-sided fiber
laser beam welding process of T-joints for aluminum aircraft fuselage panels:
Filler wire melting behavior, process stability, and their effects on porosity
defects', Optics & Laser Technology, 52, 1-9.
253
Thakur, H., Singh, K. M. and Sahoo, P. K. (2010). 'Meshless local Petrov-Galerkin
method for nonlinear heat conduction problems', Numerical Heat Transfer, Part
B: Fundamentals, 56(5), 393-410.
Thomas, P. M. and Gruzleski, J. E. (1978). 'Threshold hydrogen for pore formation
during the solidification of aluminum alloys', Met. Trans., Set. B, 9(3), 139-145.
Thomy, C. and Vollertsen, F. (2009). 'Laser-MIG hybrid welding of aluminium to steel
- effect of process parameters on joint properties', Xii, 1958-09.
Tian, J. and Rao, S. S. (2012). 'Meshless local Petrov-Galerkin method for threedimensional heat transfer analysis', Journal of Heat Transfer, 134(11), 112701112701.
Tong, M. and Browne, D. J. (2011). 'Smoothed particle hydrodynamics modelling of the
fluid flow and heat transfer in the weld pool during laser spot welding', Mater.
Sci. Eng., 27, 1-7.
Tran, H. S., Brusten, R., Jardin, R. a., Duchene, L., Paydas, H., Lecomte Becker, J. and
Habraken, A. M. (2016). 3-D finite element analysis of laser cladding process.
Application to Ti-6Al-4V. Computational methods in Manufacturing Processes
2016 Liège,Belgium. .
Trumpf-Lasers. (2013). Trudisk 5302 Laser Technical sheet [Online]. Available:
http://www.trumpf-laser.com/en/products/solid-state-lasers/disklasers/trudisk.html [Accessed Sep 2013].
Turner, M. W., Crouse, P. L. and Li, L. (2006a). 'Comparison of mechanisms and
effects of Nd:YAG and CO2 laser cleaning of titanium alloys', Applied Surface
Science, 252(13), 4792-4797.
Turner, M. W., Crouse, P. L. and Li, L. (2007). 'Comparative interaction mechanisms
for different laser systems with selected materials on titanium alloys', Applied
Surface Science, 253(19), 7992-7997.
Turner, M. W., Crouse, P. L., Li, L. and Smith, A. J. E. (2006b). 'Investigation into CO2
laser cleaning of titanium alloys for gas-turbine component manufacture',
Applied Surface Science, 252(13), 4798-4802.
Turner, M. W., Schmidt, M. J. J. and Li, L. (2005). 'Preliminary study into the effects of
YAG laser processing of Ti-6Al-4V alloy for potential aerospace component
cleaning application', Applied Surface Science, 247(1–4), 623-630.
Turner, R. P., Panwisawas, C., Sovani, Y., Perumal, B., Ward, R. M., Brooks, J. W. and
Basoalto, H. C. (2016). 'An Integrated Modeling Approach for Predicting
Process Maps of Residual Stress and Distortion in a Laser Weld: A Combined
CFD–FE Methodology', Metallurgical and Materials Transactions B, 47(5),
2954-2962.
University of Manchester. (2014). X-ray Diffraction [Online]. Available:
http://www.materials.manchester.ac.uk/our-research/facilities/x-ray-diffraction/
[Accessed 01/07/2014].
Vaccari, J. H. (1991). American Machinist, February, 139.
Varanasi, C., Murthy, J. Y. and Mathur, S. (2012). 'Numerical Schemes for the
Convection-Diffusion Equation Using a Meshless Finite-Difference Method',
Numerical Heat Transfer, Part B: Fundamentals, 62(1), 1-27.
Vargel, C. (2004). Chapter G.6 - Cleaning of Aluminium. Corrosion of Aluminium.
Amsterdam: Elsevier.
Verhaeghe, G. (2007). Low-porosity laser welding of 12.7mm thickness aerospace
aluminium [Online]. TWI. Available: http://www.twi-global.com/technical254
knowledge/published-papers/low-porosity-laser-welding-of-127mm-thicknessaerospace-aluminium-october-2007/ [Accessed 15-12-2016].
Verhaeghe, G. and Hilton, P. (2004). Laser welding of low-porosity aerospace
aluminium alloy. 34th International MATADOR Conference. UMIST,
Manchester, UK.
Vertes, A., Juhasz, P., De Wolf, M. and Gijbels, R. (1989). 'Hydrodynamic modelling of
laser plasma ionization processes', International Journal of Mass Spectrometry
and Ion Processes, 94(1), 63-85.
Vilar, R., Lima, M. S. F., Riva, R., de Oliveira, A. C., Siqueira, G. R., Conde, O.,
Fajardo, M., Silva, L. O., Pires, M. and Utkin, A. (2008). 'Laser beam welding
aerospace aluminum using fiber lasers', 7131, 713128-713128-4.
Violeau, D. (2012). Fluid Mechanics and the SPH Method: Theory and Applications:
OUP Oxford.
Violeau, D. and Issa, R. (2007). 'Numerical modelling of complex turbulent free-surface
flows with the SPH method: an overview', International Journal for Numerical
Methods in Fluids, 53(2), 277-304.
Vollertsen, F. and Thomy, C. (2006). 'Magnetic stirring during laser welding of
aluminum', Journal of Laser Applications, 18(1), 28.
Wang, H., Li, G., Han, X. and Zhong, Z. H. (2007). 'Development of parallel 3D RKPM
meshless bulk forming simulation system', Advances in Engineering Software,
38(2), 87-101.
Wang, M. Z. and Kassner, M. E. (2002). 'Tensile and fatigue properties of aluminum
alloy sheet 6022', Journal of Materials Engineering and Performance, 11(2),
166-168.
Wang, W., Xu, G., Wang, X., Lei, Z., Qin, G. and Lin, S. (2008). 'Porosity prevention
of 1420 Al-Li alloy for laser welding', Hanjie Xuebao/Transactions of the China
Welding Institution, 29(2), 5-7+26.
Watkins, K. G., Curran, C. and Lee, J. M. (2003). 'Two new mechanisms for laser
cleaning using Nd:YAG sources', Journal of Cultural Heritage, 4, 59s-64s.
Wei, P. S., Lin, C. L., Liu, H. J. and Ting, C. N. (2012). 'Transient Thermocapillary
Convection in a Molten or Weld Pool', Journal of Manufacturing Science and
Engineering, 134(1), 011001-011001.
Weman, K. (2003). Welding Procsses Handbook. Cambridge: Woodhead.
Wiberg, E. and Wiberg, N. (2001). Inorganic Chemistry: ACADEMIC PRESS.
Wolfram.
(2016).
L2
Norm
Error
[Online].
Available:
http://mathworld.wolfram.com/L2-Norm.html [Accessed 20/12/2015].
Wu, B., Shin, Y. C., Pakhal, H., Laurendeau, N. M. and Lucht, R. P. (2007). 'Modeling
and experimental verification of plasmas induced by high-power nanosecond
laser-aluminum interactions in air', Physical Review E, 76(2), 026405.
Wu, J., Yu, S.-T. and Jiang, B.-N. (1998). 'Simulation of two-fluid flows by the leastsquares finite element method using a continuum surface tension model.',
International Journal for Numerical Methods in Engineering, 42(4), 583-600.
Wu, S. C., Yu, X., Zuo, R. Z., Zhang, W. H., Xie, H. L. and Jiang, J. Z. (2013).
'Porosity, Element Loss, and Strength Model on Softening Behavior of Hybrid
Laser Arc Welded Al-Zn-Mg-Cu Alloy with Synchrotron Radiation Analysis',
Welding Journal, 92, 64-71.
Wu, S. C., Zhang, H. O., Wang, G. L. and Xia, W. S. (2009). Heat Transfer
Applications of Meshless Local Petrov-Galerkin (MLPG) Method during
Plasma Spray. Computational Mechanics: Proceedings of “International
255
Symposium on Computational Mechanics” July 30–August 1, 2007, Beijing,
China. Berlin, Heidelberg: Springer Berlin Heidelberg.
Xiao, S., Gurevich, E. and Ostendorf, A. (2012). 'Incubation effect and its influence on
laser patterning of ITO thin film', Applied Physics A, 107(2), 333-338.
Xiao, Y., Zhan, H., Gu, Y. and Li, Q. (2017). 'Modeling heat transfer during friction stir
welding using a meshless particle method', International Journal of Heat and
Mass Transfer, 104, 288-300.
Xiong, H. B. and Zhu, J. (2010). Study of droplet deformation, heat conduction and
solidification using incompressible smoothed particle hydrodynamics method.
9th Int. Conf. on Hydrodynamics. Shanghai, China.
Xiong, S., Liu, W. K., Cao, J., Li, C. S., Rodrigues, J. M. C. and Martins, P. A. F.
(2005). 'Simulation of bulk metal forming processes using the reproducing
kernel particle method', Computers and Structures, 83(8-9), 574-587.
Xiong, S., Rodrigues, J. M. C. and Martins, P. A. F. (2007). 'On background Cells
during the Analysis of Bulk Forming Processes by the Reproducing Kernel
Particle Method', Computational Mechanics, 40(2), 233-247.
Xu, X. (2002). 'Phase explosion and its time lag in nanosecond laser ablation', Applied
Surface Science, 197-198, 61-66.
Xu, X., Meade, A. and Bayazitoglu, Y. (2010). 'Fluence rate distribution in laserinduced interstitial thermotherapy by mesh free collocation', International
Journal of Heat and Mass Transfer, 53(19–20), 4017-4022.
Yan, J., Zeng, X., Gao, M., Lai, J. and Lin, T. (2009). 'Effect of welding wires on
microstructure and mechanical properties of 2A12 aluminum alloy in CO2 laserMIG hybrid welding', Applied Surface Science, 255(16), 7307-7313.
Yan, Y., Li, L., Sezer, K., Wang, W., Whitehead, D., Ji, L., Bao, Y. and Jiang, Y.
(2011). 'CO2 laser underwater machining of deep cavities in alumina', Journal of
the European Ceramic Society, 31(15), 2793-2807.
Yang, J., Chen, L., Gong, S. and Li, X. (2008). Study on the surface treatment for
controlling the aluminum-lithium alloy weld porosity of laser welding. In: 3rd
Pacific International Conference on Applications of Lasers and Optics, PICALO
2008 - Conference Proceedings, 2008. 499-504.
Yao, W. and Gong, S. L. (2011). 'Porosity Formation Mechanisms and Controlling
Technique for Laser Penetration Welding', Advanced Materials Research, 287290, 2191-2194.
Yildiz, M., Rook, R. A. and Suleman, A. (2009). 'SPH with the multiple boundary
tangent method', International Journal for Numerical Methods in Engineering,
77(10), 1416-1438.
Yong-Feng, L., Wen-Dong, S. and Teck-Seng, L. (1998). 'Laser cleaning of microparticles from a solid surface-theory and applications', Materials Chemistry and
Physics, 51, 181-185.
Yu, D., Jiang, L., Wang, F., Shi, X., Qu, L. and Lu, Y. F. (2015). 'Hydrodynamic
simulation of ultrashort pulse laser ablation of gold film', Applied Physics A,
119(3), 1047-1052.
Yu, Y., Wang, C., Hu, X., Wang, J. and Yu, S. (2010). 'Porosity in fiber laser formation
of 5A06 aluminum alloy', Journal of Mechanical Science and Technology,
24(5), 1077-1082.
Yue, L., Wang, Z. and Li, L. (2012). 'Material morphological characteristics in laser
ablation of alpha case from titanium alloy', Applied Surface Science, 258(20),
8065-8071.
256
Zhang, M., Zhang, H. and Zheng, L. (2007). 'Application of smoothed particle
hydrodynamics method to free surface and solidification problems ', Numerical
Heat Transfer A., 52, 299-314.
Zhang, P., Su, Q. and Luan, D. (2014). Twist Rotation Deformation of Titanium Sheet
Metal in Laser Curve Bending Based on Finite Element Analysis. In: Wong, W.
E. & Zhu, T. (eds.) Computer Engineering and Networking: Proceedings of the
2013 International Conference on Computer Engineering and Network
(CENet2013). Cham: Springer International Publishing.
Zhang, Y., Lu, F., Wang, H.-P., Wang, X., Cui, H. and Tang, X. (2015). 'Reduced hot
cracking susceptibility by controlling the fusion ratio in laser welding of
dissimilar Al alloys joints', Journal of Materials Research, 30(7), 993-1001.
Zhang, Z., Wang, J., Cheng, Y. and Liew, K. M. (2013). 'The improved element-free
Galerkin method for three-dimensional transient heat conduction problems',
Science China Physics, Mechanics and Astronomy, 56(8), 1568-1580.
Zhao, H. and DebRoy, T. (2011). 'Pore formation during laser beam welding of die cast
magnesium alloy AM60B –mechanism and remedy', Weld. J., 80 (8), 204S210S.
Zhao, H., White, D. R. and DebRoy, T. (1999a). Int. Mater. Rev., 44 238.
Zhao, H., White, D. R. and DebRoy, T. (1999b). 'Current issues and problems in laser
welding of automotive aluminium alloys', International Materials Reviews, 44
(6), 238-266.
Zhao, X. and Shin, Y. C. (2013). 'Femtosecond laser ablation of aluminum in vacuum
and air at high laser intensity', Applied Surface Science, 283, 94-99.
Zheng, Y. W., Luk’yanchuk, B. S., Lu, Y. F., Song, W. D. and Mai, Z. H. (2001). 'Dry
laser cleaning of particles from solid substrates: Experiments and theory',
Journal of Applied Physics, 90(5), 2135-2142.
Zhou, J. and Tsai, H. (2007a). 'Effects of electromagnetic force on melt flow and
porosity prevention in pulsed laser keyhole welding', International Journal of
Heat and Mass Transfer, 50, 2217-2235.
Zhou, J. and Tsai, H. (2007b). 'Porosity formation and prevention in pulsed laser
welding', ASME Journal of Heat Transfer, 129, 1014-1024.
Zweig, A. D. (1991). 'A thermo-mechanical model for laser ablation', Journal of
Applied Physics, 70(3), 1684-1691.
257
10 APPENDIX A
Laser welding is increasingly applied in the production of lightweight structures, high
precision components and jewellery. The laser welding mechanism depends on the
transport phenomena, which involves heat and mass transfer and is related to the
temperature field as well as the fluid flow inside the welding pool. It is important to
understand the heat effects in this process in order to anticipate and control the
microstructure and mechanical properties (Mackwood and Craferb, 2005). Figure A.1
illustrates a typical flow chart of a keyhole simulation as one of the many existing
algorithms.
Figure A.1 Flow chart of a typical keyhole simulation approach (Moraitis and Labeas,
2008)
CFD has been used extensively to model the keyhole welding dynamics and the
transport phenomenon. Turner et al. (2016) combined an FE and a CFD model to
predict the residual stresses and thermal distortion in laser welding of titanium alloy Ti6Al-4V. The information obtained from the CFD model such as the weld pool shape
and dimensions were coupled with the temperature field in the FE model to finally
simulate the thermomechanical stresses and distortion. Although the OpenFOAM model
has predicted the weld pool shape, the author did not present any information on the
dynamics inside the pool and how the transport phenomenon occurs. Also, the model
overpredicted the distortion of the workpiece in comparison with the experimental data.
Hu et al. (2015) and Mohanty et al. (2012) used FLUENT software to create a model
258
for simulating deep penetration laser welding and to monitor the fluid flow dynamics.
Both models focused only on the steady state keyhole welding in which the keyhole
cavity is assumed to be steady and balanced with the recoil pressure. However, the
transient case of the keyhole formation was not studied and Hu et al. (2015) showed
only the results until 40 ms. Hozoorbakhsh et al. (2016) used the volume of fraction
(VOF) method to predict the free surface during laser micro welding of thin stainless
steel sheets. The model achieved a good agreement for the welding pool dimensions and
predicted the temperature history across the sample.
CFD models have been applied and validated in a large number of studies that cannot be
fully covered in this appendix. However, many challenges are still to be overcome with
these methods such as the expensive computational cost associated with the use of VOF
method for surface tracking (Mohanty et al., 2012). Moreover, mesh-based methods
may not be able to completely simulate the keyhole welding process since most of the
reported works neglect the metal vapour phase in their calculations. This is due to the
inherent connection between the elements that prevent them from leaving the mesh.
SPH as a particle-based method depends on independent particles for representing the
computational domain. SPH can simulate multi-phase flows in which different phases
can be taken into consideration in the same model. Moreover, SPH method eliminates
the need for explicitly computing the thermal convection forces and velocities since
those physical quantities are already computed in the SPH formulations. The use of SPH
for modelling laser welding allow for considering the metallic vapour created within the
keyhole as well as the fluid movement around the keyhole. This will give a better
representation of the process and reduce the need for multiple models to simulate one
process. Moreover, modelling welding using the model proposed by Alshaer et al.
(2017), gives the opportunity to model the combination of different processes, i.e. laser
cleaning before laser welding, to investigate its effect on porosity generation in Al
welding.
In this Appendix, a multi-phase SPH model for laser welding is presented along with
some preliminary results. It should be noted that this model is still under development,
and the results presented in the “Results and Discussion” section are to show the reader
the potential that SPH method has for modelling the transport phenomena.
259
1. Model Description
1.1 Physical phenomena
1.1.1 Heat transfer equation in laser welding
The laser welding process is mainly based on the laser heating of the solid surface
which conducts that heat into the internal parts. The melting process and surface tension
play an essential role during welding in forming the shape of the weld and its
dimensions as well as in the heat conduction to the solid boundaries. Figure A.2
represents a schematic of the laser welding pool.
Taking into account the external heat sources, the differential equation of the heat
transfer can be expressed as follows:
cp
dTi
   k T   Q  Qv
dt
(A.1)
where k [W/m.K] thermal conductivity, T [K] Temperature, Q [W/m2] the heat source,
Qv heat loss due to convection.
The heat loss Qv can be formulated using:
Qv  hc Ts  T0     Ts4  T04 
(A.2)
where hc=20[W/m2.K] convection factor, ɛ=0.09 the emissivity, σ=5.67x10-8 [W/m2.K4]
the Stefan-Boltzmann constant, Ts and T0 are the surface and the initial temperatures
respectively.
260
Laser
Molten pool
HAZ
Z
Solid Sample
X
Figure A.2 Simplified schematic laser welding of a metallic sample
The penetration of the electromagnetic wave (laser light) into the material is neglected
in this work since the particles’ spacing is significantly larger than the optical
penetration depth (1/D), where D is the optical absorption coefficient for Al (1.23108
m-1). This means that the electromagnetic wave will be absorbed only by the first layer
of particles and the subsequent layers will be mainly heated by conduction
1.1.2 Phase transition
In laser welding, if the incident laser beam energy absorbed by the solid material
exceeds a particular threshold, phase change takes place, and the solid metal transforms
into a liquid. The difference in density associated with temperature elevation leads to a
change in the welding pool volume which is also affected by the surface tension. These
factors not only affect the shape of the welding pool around the laser beam but also
induce the material’s flow as a result of convection.
In SPH, the change in density and thermal expansion is a complex issue which needs
rigorous and careful consideration. Since the mass of SPH particles is constant, the
change in density introduces change in the particles’ volume, which in turn affects the
number of particles located in the interpolation region. The change in the number and
volume of the particles may lead to a significant reduction in accuracy without
appropriate treatment.
The phase change can be simulated by expressing the material density dependence on
temperature for Al 6000 series as follows (Mills, 2002):
  2705  0.201T  300  if T  Tmelting
(A.3)
261
  2415  0.280 T  925 if T  Tmelting and T  Tvapourisation


1000
,   10
0.562 
567
T
(A.4)
(A.5)
where T is the particle’s temperature [K],  is the dynamic viscosity of liquid Al [Pa.s].
Equation (A.3) and Equation (A.4) give the change in the solid density and molten Al
density respectively, while Equation (A.5) describes the dynamic viscosity change with
temperature.
1.1.3 Surface tension
The surface tension model employed in this work is based on the continuum surface
force (CSF) model (Brackbill et al., 1992), in which the surface tension force is
transformed from a surface force to a volumetric force by the introduction of a surface
delta function for the convenience of modelling and programming. This force can be
expressed as:
Fs  f s . s
(A.6)
where fs is the surface tension force exerted at the liquid phase surface and consists of
the following terms:
f s    n + s
(A.7)
where σ is the surface tension coefficient for molten Al, κ is the surface curvature
  n , n is the normal vector to the liquid surface and  s is the surface gradient.
The second term in Equation (A.7) takes into account the temperature dependence of the
surface tension which induces the molten material flow according to Marangoni effect.
In this work, the surface tension coefficient follows a linear function of temperature
with a negative slope:
 s  0.88  0.35 103 T  Tm  , Tm  925K
(A.8)
where T is the particle’s temperature [K],  s is surface tension coefficient [N.m-1].
262
1.2 Governing equations and SPH discretisation
As presented in Alshaer et al. (2017) (Chapter 8), the discretised Lagrangian form of
Navier-Stokes equations is used to describe the particles’ movement and interaction
using the conservation of mass and momentum equations. These equations in their SPH
discretised forms are:
mj
d i
  j 
vij .iWij
dt
j j
(A.9)
 Pi  Pj

dvi
  m j 
  ij  iWij  g  f s
 

dt
j
 i j

(A.10)
where fs is surface tension force which will be explained in the following sections.
In order to enclose the system, the pressure in this model is related to the density using
Tait’s equation of state:
   
P  B    1
 0 

(A.11)
0 c02
where B 
, 0  1000 kg / m3 is the reference density, c0 speed of sound, and

  7 is a constant.
1.2.1 Heat transfer
The main mechanisms for the heat transfer in this model are the conduction and
convection, while radiation only takes place only at the top surface of the sample. One
of the attractive advantages of the SPH method is that the movement of the heated
particles inherently simulates the natural convection behaviour without the need to
explicitly include the convection force in the SPH formulation.
To simulate the thermal behaviour of particles, an SPH form of the unsteady heat
conduction equation (A.1) was introduced into the model to include the laser beam
heating of the surface (Monaghan et al., 2005; Alshaer et al., 2017):
263
cp
mj
dTi

dt
j i  j
 4ki k j

 ki  k j
 Ti  T j

 r 2
 ij

 rij iWij  Q  Qv

(A.12)
Every half time step, the SPHysics_3D code calculates the temperature gradient and the
density value before calculating the particle’s pressure which correlates to the current
density value. Once the temperature is updated for the next half time step, the density is
updated to reflect the change in temperature and the pressure is subsequently modified
according to the density. The change in pressure induces the particles’ movements by
creating acceleration in the momentum equation, simulating the thermal expansion of
the material.
Using this method, the system will automatically update the density and pressure values
according to the phase of the particle which is determined by the particle’s temperature.
1.2.2 SPH model for surface tension
Following Hu and Adams (2006) to calculate the surface curvature for the surface
tension force, a colour function is used to compute the surface tension in Equation (A-7)
and the CSF shear stresses. The values of this function are as follows:
1 if particle belongs to liquid phase
Cis  
0 otherwise
(A.13)
Ci   C sj Wij V j
(A.14)
j
Although the second phase, which can be the air or the shielding gas, is neglected in this
model, the interface can still be detected by assigning the zero value to the missing
phase and the value of unity to the liquid metal particles.
The vector normal to the surface can be calculated using the colour function gradient
and its magnitude:
n
C
C
(A.15)
264
where
C is evaluated using (Cirpici, 2015):
C  
j
mj
j
C
j
 C j  iWij
(A.16)
where Ci and Cj are the colour function values for particles i and j respectively.
Wu et al. (1998) expressed the shear stresses induced by the surface tension as a
function of the colour function’s gradient. Hu and Adams (2006) further developed the
previous model in order to eliminate the influence of far particles on the shear forces
calculations in the momentum equation. The shear stresses contribution to the
momentum equation can accordingly be writing as:
 i   j
Dvi
  mj 
  
Dt
j
 i j
where


 iWij

(A.17)
is the continuum surface force given in the tensor form as in (Wu et al., 1998):
s  1
2

 I C  CC 
C  d

where CC 
(A.18)
C C
, d is the spatial dimension, I is the unit vector and  and  are
x x
the Einstein notations.
Therefore, the momentum equation takes the form of:
 Pi  Pj

 i   j
Dvi
  m j 
  ij  iWij   m j 
  

  
Dt
j
j
 i j

 i j

 iWij  g

(A.19)
1.2.3 Surface detection
In SPH, the location of free surface can be determined by computing the divergence of
the particle position using the following equation (Muhammad, 2012):
265
.r  
j
mj
j
rij . iWij
(A.20)
The truncated kernel support of any surface particle gives a non-zero value for the
particle position divergence. In 3-D cases, an empirical value .r  2.4 was used to
indicate a free-surface particle (Lee, 2007; Lind et al., 2012).
1.2.4 Time step for surface tension
Since the surface tension is introduced to the model formulation, the variable time step
criteria should be updated using the surface tension condition provided in (Brackbill et
al., 1992):
  h3 
tst  

 2  s 
(A.21)
where h is the smoothing kernel support and σs is the surface tension coefficient. When
the surface tension coefficient is changing with temperature, the smallest value within
the computational domain is used for evaluating the time step condition.
The time step is updated every time step in the simulation by taking the minimum value
from the gravity force, thermal, viscous force and surface tension conditions:
t  CFL min  t f , tcv , th , tst 
(A.22)
where CFL is taken as “0.1” in this work (Monaghan and Kos, 1999).
1.3 Numerical model
One of the advantages for modelling thermo-physical problems using SPH method is
thermal convection can be inherently represented by the particles’ movement without
explicitly defining the convection forces acting on them. SPHysics_3D open source
code was modified to include the thermo-physical relations given in the previous
sections. Figure A.3 illustrates the workflow for computing the laser-particle interaction
in the SPHysics_3D code.
266
Figure A.3 Flow chart representing the laser-particle interaction in SPHysics 3-D
Assuming that the solid sample is made of Al AA6014, a Gaussian distribution of the
laser intensity was applied on its surface with a beam diameter of 1 mm and a nominal
power of 3500 W. Table A.1 gives the thermo-physical properties of the studied alloy.
The computational domain dimensions were taken as 1.2  1.2  0.25 mm3 with 12.5
μm spacing for the initial uniform particles distribution. This produced 183,798
particles in total. The smoothing kernel is taken as h=1.87510-5m which corresponds to
1.5Δx, covering three neighbouring particles in each direction. The initial time step was
set to 0.1 μs with a total simulation time of 3010-5 s and a CFL number of “0.1”.
Figure A.4 shows the computational domain with all dimensions and the laser Gaussian
distribution at the centre of the workpiece.
267
0.25 mm
1.2 mm
Laser
Intensity
100%
1.2 mm
0%
Figure A.4 SPH computational domain showing the Gaussian distribution of the laser
beam
Table A.1 Aluminium alloy AA6014 thermo-physical properties (AlShaer et al., 2015a)
Density
3
ρ [kg/m ]
2705
Surface Optical
Thermal
Initial
Thermal
Reflectivity
diffusivity
Temperature
Conductivity
2
R [%]
D [m /s]
[⁰K]
k [W/m.⁰K]
70%
6.89 x 10-5
300
167
Specific Heat
Cp [J/kg]
896
Emissivity
0.09
In laser welding, the CW beam travels at a given speed in order to weld the area of
interest. However, due to the limitations of the computation domain’s dimensions and
resolution, simulating the laser beam movement can be computationally expensive.
However, the laser beam movement can be imitated by reducing the laser power linearly
over a calculated period of time. This will give a similar effect to switching off the laser
on particles that are no longer under the laser influence. The laser power used in this
simulation as a function of time is given in Figure A.5
268
Laser power [kW]
3.5
Time [μs]
10
12
Figure A.5 Laser power ramping over 2 μs to simulate the moving beam effect (not to
scale)
2. Results and Discussion
2.1 Thermal expansion of the welding pool
In this test case, the surface tension effect is ignored in order to show the thermal
expansion of the material and the density change with temperature during laser heating.
Figure A.6 shows the temperature and density distribution across the sample at different
time frames.
The period from 0 μs to 40 μs is not shown since is solely thermal conduction and the
solid phase thermal expansion is small despite the density reduction with temperature.
This can be attributed to the high viscosity assigned to the solid phase which limits the
particles movement and the slower change in the solid density in comparison with the
fluid density change with temperature (note the slope difference value between
Equations A.3 and A.4).
Once the particles transform to liquid, the viscosity value drops drastically giving the
liquid particles an acceleration from the momentum equation. With uniform heating, the
temperature continues to climb causing more particles to expand due to volume
increase.
When the laser beam power starts to drop at 100 μs, the acceleration of the particle
decreases until the laser effect vanishes, and heat conduction herein dominates the
transient process. It is worth mentioning that a small amount of the laser power is
wasted to the surrounding due to convection and radiation.
269
From Figure A.6, the right-hand column shows the density drop with temperature with a
minimum value of 2340 kg/m3 correlating to the maximum temperature 1534 K. The
maximum velocity recorded is 0.175 m/s, and the velocity magnitude gradually drops
from the centre, where the maximum temperature is located, to the circumference of the
weld pool.
2.2 Surface tension effect on the welding pool
Surface tension is the main factor that shapes the welding pool and induces the fluid
movement inside it. Figure A.7 shows 3-D views of the sample with the temperature
distribution at the top surface with a maximum temperature value of 1655 K.
In order to demonstrate the fluid movement and the temperature field inside the welding
pool, the sample is sliced in the middle part, and one layer of particles is demonstrated
in Figure A.8. From Figure A.8 (b) and (c), the heated surface of the welding pool starts
to expand when its temperature exceeds the melting point with a slight movement of the
circumferential particles towards the inside of the welding pool. When the welding pool
becomes deeper, the surface tension force starts to have a substantial contribution in the
momentum equation. This can be clearly seen by the significant movement of the
circumferential particles towards the inside of the welding pool and by the
rearrangement of the inner particles within the molten pool.
270
(a) 0 s
(b) 0 s
(c) 40 s
(d) 40 s
(e) 60 s
(f) 60 s
(g) 90 s
(h) 90 s
(j) 110 s
(i) 110 s
(k) 130 s
(l) 130 s
(m)
(n) 160 s
160
(p) 300 s
(q) 300 s
0.175
m/s
Figure A.6 Temperature and density distribution and the velocity vectors in an Al
sample without the application of surface tension.
271
(a) 0 s
(b) 50 s
(c) 90 s
(d) 110 s
(e) 130 s
(f) 160 s
(g) 210 s
(h) 300 s
Figure A.7 3D view of the welding pool at the top surface of the Al sample
It should be noted that the maximum temperature calculated in this test case is slightly
higher (by approximately 100 K) than the value obtained without the surface tension.
This may be explained due to the limited movement of the particles in the first case,
which aids thermal conduction, in comparison with the second test case in which the hot
particles are driven towards the centre, raising the centre’s temperature value.
272
(a) 0 s
(c) 100 s
(b) 50 s
(d) 120 s
(e) 130 s
(f) 140 s
(g) 150 s
(h) 160 s
(i) 170 s
(j) 200 s
Figure A.8 Welding pool cross section showing the surface tension effect on the
welding pool shape
When the cooling phase begins at 130 μs, the particles starts to decelerate due to the
temperature drop until a complete solidification is reached at 170 μs. From Figures A.8
(e) to (j), an unphysical gap can be seen between the first layer of particles and the rest
of sample’s body. This gap becomes slightly larger towards the end of the simulation at
170 μs before the particles solidify and lose their momentum. The unphysical gap may
be produced when the surface particles, which have the maximum momentum and
temperature, expand rapidly while the underlying particles are not able to compensate
and fill the gap due to their smaller momentum. This resolution issue may be fixed by
reducing the particles’ spacing by 2-3 times. However, the computational time will
increase significantly by about 16-24 times taking into account the small time step
required for the simulation. Figure A.9 gives the velocity vector field at selected times
of the simulation. At 70 microseconds, the top layer of particles has a small velocity
magnitude in comparison with the particles at the circumference which are under the
surface tension force. This velocity slightly increases with temperature leading the
particles upwards due to the thermal expansion. Particles inside the welding pool,
273
however, move along the solid boundary of the molten pool towards the centre where
surface tension is lesser. This movement continues according to Marangoni effect,
which is shown in Figure A.10 until solidification takes place and particles decelerate
and stop when their temperature drops below the melting point.
Figure A.10 clearly shows where the particles converge at the bottom of the welding
pool and then rise upwards towards the surface where the least surface tension force and
material concentration is.
(a) 70 s
(b) 90 s
(c) 110 s
(d) 120 s
(e) 130 s
(f) 140 s
Enlarged area
Velocity scale
1.343
m/s
Figure A.9 Velocity field in the welding pool at different time intervals
It is difficult, at this stage of developing this model, to quantitatively validate the model
until the resolution and the computational domain dimensions issues are solved.
However, Figure A.11 gives a qualitative comparison of the weld top surface shape for
autogenous laser welded Al alloys. It should be noted that the real alloys presented in
Figure A.11 are welded under different laser density and travel speeds, from those used
in the simulation. However, it can be seen that the weld top surface generated by the
274
model is displaying qualitatively the correct behaviour towards a real representation of
the final shape of the weld.
Figure A.10 Marangoni effect on particles at 130 microsecond
(a)
(b)
1 mm
Figure A.11 (a) Al sheet lap welded at 3.5 kW (Leidich, 2016) and (b) 7000 series Al
sheet of 12.7 mm sheet thick welded at 7 kW (Verhaeghe, 2007).
3. Summary
In this appendix, a new but preliminary SPH model is introduced to simulate
autogenous laser welding including the transient heating with a continuous wave laser,
coupled with the presence of surface tension and Marangoni effect in the weld pool.
The proposed model produced a good qualitative prediction of the weld pool surface
geometry and correctly predicted the density change with temperature across the
275
sample. However, issues associated with the surface tension and the viscosity forces on
the first layer of particles are still to be resolved.
There are many challenges to be overcome with this model and resolution is considered
to be the most difficult. Moreover, some of the physics are to be included in the model
such as the use of laminar viscosity, which may produce a smoother behaviour for the
particles. This, in addition to the recommendations given in Chapter 9, is going to be
considered in the near future.
276
11 APPENDIX B
In this appendix, the analytical solution of the heat conduction equation is briefly
discussed.
Consider a volumetric cube of Al receiving a heat  Q  Qv  generated by the heat
source Q and the losses due to convection and radiation Qv. In this section, the volume
referred to as a “volumetric element” has no relation to the elements used in FE method.
Due to conduction, the heat is carried by the element from one side to the opposite to
satisfy the conservation principle expressed in Equation (B.1). Figure B.1 illustrates the
heat conduction quantities in the studied element.
The pressure and the shear stress responsible for the volume change of the element are
ignored in this analysis for the sake of simplicity. However, these will be taken into
account in the SPH 3-D model in order to produce more realistic results.
The conservation of energy can be expressed as follows:
A
Rate of change of
=
B
+
Net volumetric
=
heat into
Energy
+
(B.1)
C
Net Work due to
change in volume
element
Term C in the previous equation can be https://www.you
neglected since only the heat transfer due to
tube.com/watch
conduction is considered.
?v=zexH3xebqs
The net volumetric heat into the element can be expressed
as follows:
Q
qz   Q  Qv   qz  qz  dx dy dz   Q  Qv   qz  dx dy dz
where:
.
q z is the volumetric heat in the z direction.
Q is the external heat input generated by the laser (or any other heating sources).
Qv is the heat losses due to convection to the ambient gases and due to emission.
qz change in the heat in the z direction.
277
(B.2)
and dx, dy, dz are the volumetric element dimensions.
y
Q  Qv  dx dy
Volumetric
Element
.
q z dx dy
Shear stresses τ= 0.0
Pressure = 0.0
x
z
qz  qz
Figure B.1 The heat flow into a volumetric element of the metal due to heat conduction
(amended from (Anderson, 1995)).
Taking into account that:
qz   k 
T
z
(B.3)

  T  
B   Q  Qv    k    dx dy dz
z  z  

(B.4)
where k is material thermal conductivity and T is temperature.
Knowing that:
278
A
dH
V
dt
or
A
de
V
dt
(B.5)
where e is the internal energy, H is enthalpy, 𝜌 is the density and V is the element
volume. By substituting Equation (B.5) into the main Equation (B.1), we obtain:

dH
V  B
dt
and 
(B.6)
dH

  T  
 dx dy dz   Q  Qv    k    dx dy dz
dt
z  z  

(B.7)
It is known from the heat conduction law that:
dH
dT
 cp
dt
dt
(B.8)
That yields to:
cp
dTi
   k T   Q  Qv
dt
(B.9)
where c p is specific heat under constant pressure and  is the gradient operator.
By assuming a temperature-independent thermal conductivity and by taking into
account the pulse laser heating source, the solution to heat conduction equation as a
function of time and depth becomes as follows (Sands, 2011):
 For a time shorter than the pulse length:
T  z, t  |t t p  T0 

 z 
2 I 0 1  R 
D. t  ierfc 
 

k
2
D
.
t

 


(B.10)
where D [m2.s-1] is the thermal diffusivity, R is the reflectivity of the Al surface,
Dt
[m] is the thermal penetration depth, I0 is the laser intensity, tp is the pulse length and z
is the distance from the surface.
The integrated complementary error function is given by (Sands, 2011):
279

ierfc( x)   erfc( x) dx
(B.11)
x
And:
erfc( x)  1  erf ( x)  1 
x
2
e
 
t 2
dt
(B.12)
0
 For a time longer than the pulse length:



 z 
2 I 0 1  R  
z


T  z, t  |t t p  T0 
D. t ierfc 
  D.  t  t p  ierfc



k
2
D
.
t


2 D.  t  t p   




(B.13)
If the laser beam diameter is of the same order of the thermal penetration given by D t ,
then the heat conduction in the radial direction should be taken into account and the
beam diameter a should be included in the analytical solution. Otherwise, the terms
containing the parameter a should be ignored.
Taking into account the aforementioned note, a corrected form of the analytical solution
produced by Nath et al. (2012) is now applied as two sets of equations to describe the
heating and the cooling phases during multi-pulse laser heating.
During the heating phase (laser-ON), the temperature at depth z is given by:

2 I 0 1  R  D 
T  z, t  |t t p  T0 
 t   N  1  t p  tr 
k



z 2  a2

ierfc 
 2 D. t   N  1  t p  tr 



   t   n  1  t p  tr 
n 1 

N 1
 t   n t p   n  1 tr 



z


 ierfc 

 2 D. t   N  1  t p  tr 










 


z


 ierfc 

 2 D. t   n  1  t p  tr 




z


 ierfc 

 2 D. t   n t p   n  1 tr 


280



 



z 2  a2



ierfc



 2 D. t   n  1  t p  tr 



z 2  a2



ierfc



 2 D. t   n t p   n  1 tr 




 
 
 
  

(B.14)





where a is the beam diameter, N is the total number of pulses, n is the pulse number,
tp is the pulse duration and tr is the relaxation time. During the cooling phase (laserOFF), the temperature at z depth is given by:
T  z, t  |t t p  T0 
2 I 0 1  R  D 


k



 z 2  a2
 z 
t  ierfc 
 ierfc 

 2 D. t 
 2 D. t




 





z
z 2  a2  




 t  t p  ierfc
 ierfc
 2 D. t  t 
 2 D. t  t  



 p   
p








   t  n  t p  tr 
n 1 

N 1


z


 ierfc 

 2 D t  n  t p  tr 

 t    n  1 t p  ntr 




z 2  a2



ierfc



 2 D t  n  t p  tr 



z


ierfc



 2 D t    n  1 t p  ntr 


281





 


z 2  a2



ierfc



 2 D t   n t p   n  1 tr 

(B.15)

 
 
 
  





