Upload
others
View
6
Download
0
Embed Size (px)
Citation preview
In the name of god, the merciful
WIRE-ARC SPRAYING SYSTEM: Particle Production, Transport, and Deposition
by
AmirHossein Pourmousa Abkenar
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy
Graduate Department of Mechanical and Industrial Engineering University of Toronto
© Copyright by AmirHossein Pourmousa Abkenar 2007
ISBN: 978-0-494-39723-7
ii
Abstract
WIRE-ARC SPRAYING SYSTEM: Particle Production, Transport, and Deposition
AmirHossein Pourmousa Abkenar Doctor of Philosophy
Graduate Department of Mechanical and Industrial Engineering University of Toronto
2007
Protective coatings are important to metal working. Thermal spray is a rapidly growing
market, and wire-arc spraying is gaining a significant share of this market because of its low
operating/equipment costs and high material/energy efficiency. Although wire-arc spraying is
widely used, many of its underlying processes are not yet fundamentally understood. This work
examines and explains different aspects of a wire-arc system.
In wire-arc spraying, two consumable wires are continuously fed into the gun. An electric
arc is struck between the tips of these two wires and continuously melts their material. A cross-
flow gas removes the molten material from the wire-tips and accelerates them towards a
substrate, over which the detached particles form a protective coating layer.
An imaging system was developed to take pictures of the arc, and determine its length
and shape. Using the information extracted from such pictures, a computational fluid dynamic
model of the wire-arc torch was developed to estimate the shear stresses on the wire-tips and also
sizes of primary breakups from the two electrodes.
iii
Shortly after primary breakups, the detached particles break up into smaller particles
(secondary atomization). The size and velocity of such particles were measured in-flight using a
DPV-2000 system for a range of operating parameters. A technique was developed to identify
and separate the size distributions of particles produced by atomization of molten metal at either
the anode or cathode by assuming that both follow a log-normal distribution. (This assumption
was also verified experimentally). It was shown that particles produced by the anode are almost
two times larger than those originating from the cathode. Furthermore, effect of operating
parameters on size distribution of anodic and cathodic particles was investigated.
Experiments were also conducted to study the effect of impact velocity and substrate
temperature on the properties of individual wire-arc splats and coatings. Aluminum was sprayed
onto polished stainless-steel coupons maintained at temperatures ranging from 25°C to 450°C.
At low substrate temperature, droplets splashed, forming irregular splats; at higher temperatures
there was no splashing and splats formed circular disks. The temperature at which the transition
occurred decreased with increasing impact velocity.
iv
To my love, Malahat …
without whom this work could have never been done, or could have been done much sooner...
v
Acknowledgements
I would like to take this opportunity to acknowledge my supervisors and mentors,
Professor Javad Mostaghimi and Professor Sanjeev Chandra, for their invaluable guidance,
encouragement, and support throughout this study. I would like to express my sincere
appreciations to Professor Javad Mostaghimi for his patience and understanding. His concise but
insightful comments fueled me with ideas.
I extend thanks to my supervisory committee members, Professor Bendzsak, Professor
Sullivan, Professor Ashgriz, and Professor Bussmann for their advice and helpful suggestions. I
am also grateful to Dr. Larry Pershin and Mr. Tiegang Li for their assistance in the lab, and
Ms. Brenda Fung for her excellent administrative support at the office of graduate studies.
In addition, I would like to thank all my colleagues at the Center for Advanced Coating
Technologies for making this journey enjoyable. My special thanks goes to Ali Abedini, my lab
partner, Hanif Montazeri, my numerical handyman, Rajeev Dhiman, my heat transfer expert,
Hamid Salimi, my industrial advisor, Fardad Azarmi, my political rival, Hamed Samadi, my
financial advisor, and many more including Liming, Libing, Michelle, Bob, Ken, Ala, Mehdi,
Afsoon, Nikoo, Andre, Reza, and many more friends in the department.
I am grateful to my family, especially my parents for their never ending love and support.
I would like to extend my special thanks to my father-in-law who was not only my teacher, but
also my mentor in difficult times.
To my wife, Malahat: Without you, your energetic essence, enthusiastic support, and
unconditional kindness, I could have never completed this work. I love you and I gladly dedicate
this thesis to you.
vi
Table of Contents
ABSTRACT ................................................................................................................................................................II
ACKNOWLEDGEMENTS .......................................................................................................................................V
CHAPTER 1 INTRODUCTION ..............................................................................................................................1
1.1 BACKGROUND, MOTIVATION, AND LITERATURE SURVEY ...............................................................................1
1.1.1 Thermal spray.......................................................................................................................................1
1.1.2 Twin-Wire-Arc Spray............................................................................................................................5
1.1.2.1 Description of the Wire-Arc spraying process.................................................................................. 5
1.1.2.2 Operating parameters........................................................................................................................ 8
1.1.3 Brief Literature Review ......................................................................................................................10
1.1.3.1 Previous Work on Droplet Production and Transport..................................................................... 10
1.1.3.2 Previous Work on Bimodal Size Distribution of In-flight Particles................................................ 12
1.1.3.3 Previous Work on Particle Deposition............................................................................................ 13
1.2 STATEMENT OF OBJECTIVES ..........................................................................................................................14
1.3 SCOPE OF THE PRESENT WORK .......................................................................................................................15
1.4 OUTLINE OF THE THESIS.................................................................................................................................16
CHAPTER 2 EXPERIMENTAL APPARATUS AND PROCEDURES.............................................................17
2.1 COATING AND PROCESS DIAGNOSTICS ...........................................................................................................17
2.1.1 Coating Characterization...................................................................................................................17
2.1.2 Process Characterization ...................................................................................................................18
2.2 VALUARC 200 SPRAYING SYSTEM AND ITS CHARACTERISTICS.....................................................................26
2.2.1 Volume-Flow-Rate..............................................................................................................................32
2.2.2 Arc Current.........................................................................................................................................33
vii
CHAPTER 3 PARTICLE BREAKUP: THERMAL SPRAY GUN ....................................................................38
3.1 EXPERIMENTAL STUDIES ...............................................................................................................................38
3.1.1 Imaging system ...................................................................................................................................39
3.1.2 Current and Voltage Fluctuations......................................................................................................45
3.2 NUMERICAL STUDIES.....................................................................................................................................49
3.2.1 Flow dynamics of the nozzle geometry ...............................................................................................49
3.2.2 Simplified Arc Solution.......................................................................................................................61
3.2.3 Arc Heating in a cross flow ................................................................................................................66
3.3 SIMPLIFIED BREAKUP MODEL........................................................................................................................72
CHAPTER 4 PARTICLE TRANSPORT: IN-FLIGHT PARTICLES...............................................................75
4.1 BACKGROUND................................................................................................................................................75
4.2 SPATIAL CHARACTERISTICS OF THE SPRAY ...................................................................................................77
4.3 BIMODAL PARTICLE SIZE DISTRIBUTION AND SEPARATION TECHNIQUE .......................................................82
4.3.1 Size Distribution of Anodic and Cathodic Particles ...........................................................................84
4.3.2 Separation Technique.........................................................................................................................86
4.3.3 Error Estimation.................................................................................................................................87
4.3.4 Effect of Varying Wire-Arc Parameters .............................................................................................92
4.4 AXIAL VARIATION OF PARTICLE PROPERTIES .................................................................................................96
4.4.1 Drag Force and Force Balance Relation ...........................................................................................97
4.4.2 Heat Transfer and Exothermic Oxidation of Particles .......................................................................97
CHAPTER 5 PARTICLE DEPOSITION: SPLAT AND COATING FORMATION.....................................102
5.1 EFFECT OF SUBSTRATE TEMPERATURE ON SPLAT FORMATION....................................................................102
5.1.1 Experimental Procedure...................................................................................................................104
5.2 SPLAT MORPHOLOGY ..................................................................................................................................108
viii
5.3 MODEL FOR TRANSITION TEMPERATURE.....................................................................................................112
5.4 COATING PROPERTIES..................................................................................................................................118
CHAPTER 6 CLOSURE.......................................................................................................................................122
6.1 CONCLUSIONS..............................................................................................................................................122
6.2 RECOMMENDATIONS FOR FUTURE WORK .....................................................................................................124
REFERENCE ..........................................................................................................................................................125
APPENDIX A: METAL PROPERTIES ...............................................................................................................132
APPENDIX B: TRANSPORT PROPERTIES OF AIR.......................................................................................133
ix
List of Figures
Figure 1.1 Basic principles underlying the thermal spray processes: Production, Transport, and Deposition of molten particles. ......................................................................................3
Figure 1.2 Schematics of wire-arc spraying system and its major components .................................5
Figure 2.1 (a) Picture of DPV-2000 scanning unit alongside the wire-arc spraying gun detecting the in-flight particles (b) The computer system containing the DPV-2000 operating system and CPS-2000 modules; the two are connected to the scanning unit via fiber-optic cables. ...................................................................................................19
Figure 2.2 Schematic diagram of the DPV’s optical sensing head and its field of view [ 9] .............20
Figure 2.3 (a) A schematic diagram showing the signal sensed by the DPV-2000 sensing head when a particle passes through its field of view. (b) Picture of the two slits in P4590170 photo mask...........................................................................................................20
Figure 2.4 Wire-arc sprayed stainless-steel particles are approximately spherical. ........................20
Figure 2.5 Size distribution of particles was measured using two additional methods (optical picture measurements, and PSA measurements) to calibrate DPV’s particle size measurements. ......................................................................................................................23
Figure 2.6 Pyrometer’s calibrated reference curve. High voltages on λ = 780 nm and λ = 850 nm photomultipliers were 700 V and 1000 V, respectively...............................................24
Figure 2.7 Surface profile of polished AISI 304L stainless steel substrate obtained using a Surface Profiling Microscope (Wyko Optical Profilometer, Veeco Instruments Inc., Woodbury, NY). Surface roughness is 7.90 nm.................................................................25
Figure 2.8 Picture of ValuArc 200 Twin Wire Arc spray system and the spray gun manufactured by Sulzer-Metco. Picture is adapted from [ 63].........................................27
Figure 2.9 Picture of ValuArc 200 Twin Wire Arc spray system during operation.........................28
Figure 2.10 Schematics of ValuArc 200 Twin-Wire-Arc Spraying System and its components. Picture is adapted from [ 2, 72]. ............................................................................................28
Figure 2.11 Exploded rear view of the ValuArc 200 twin-wire-arc gun and its components. The gun can be mounted on the handle and hand operated, or on a separate mount or robot and remotely operated. Picture is adapted from [ 72] .............................................29
Figure 2.12 Exploded front view of the ValuArc 200 twin-wire-arc gun and its components. Picture is adapted from [ 72]................................................................................................30
x
Figure 2.13 Standard volumetric flow-rate of the atomizing gas (dry air) as a function of the upstream pressure. The data points represent 20 psig (239 kPa), 30 psig (308 kPa), 40 psig (377 kPa), 50 psig (446 kPa), and 60 psig (515 kPa) in the system’s pressure setting. ...................................................................................................................................32
Figure 2.14 Current-voltage characteristic of the arc for different wire-feed-rates and pressures. Dry-air as the atomizing gas and aluminum wires were used. The error-bars represent current fluctuation and standard deviation of 5 to 10 measurements. ..34
Figure 2.15 Current that passes through the arc increases when the feed rate of aluminum wires is increased. Solid curves represent quadratic fits to the datapoints.....................36
Figure 2.16 Current that passes through the arc increases when the feed rate of copper wires is increased. Solid curves represent quadratic fits to the datapoints. .................................36
Figure 2.17 Input power versus feed rate of aluminum wires. Slope of 39.5V, 32.1V, and 25V curves represent 7.90, 7.01, and 4.53 MJ/kg (mega joules per kilogram of aluminum), respectively. Solid curves represent quadratic fits to the datapoints. ........37
Figure 2.18 Input power versus feed rate of copper wires. Slope of 39.8V, 32.1V, and 26.9V curves represent 3.01, 2.40, and 1.54 MJ/kg (mega joules per kilogram of copper), respectively. Solid curves represent quadratic fits to the datapoints. .............................37
Figure 3.1 Pictures of the wire-tip region and the arc during operation of the wire-arc spraying system. These pictures are taken using a visible-wavelength Nikon E3 camera with different optical filters. ..................................................................................40
Figure 3.2 Black body radiation curves at different temperatures, scaled to a maximum of one. The presented curves represent radiation at the melting and boiling temperatures of Stainless Steel, Copper, and Aluminum. ................................................41
Figure 3.3 Optical system used to transmit laser beam from the laser system to the region to be photographed: (a) Laser beam output, (b) coupler, and (c) fiber optic cable............43
Figure 3.4 Schematic diagram of the position of the UV-intensified CCD camera and illuminating laser beams. Laser beams are coupled into and transmitted through the optical fibers along positive and negative Y axes. Cylindrical lenses focus the beams onto the YZ plane. The CCD camera, attached to the lens system and aligned with the X axis, takes a picture of the wire tips....................................................43
Figure 3.5 Photographs of the two wires and the detached particles. These pictures are the negative of what was captured by the UV-intensified COHU camera. Two lasers illuminate the area from top and bottom simultaneously. Aluminum wires and a High-Velocity cap were used; wire-feed-rate = 8 m/min, voltage = 29.1 V, pressure = 45 psig (412 kPa)................................................................................................................44
Figure 3.6 The average period of arc voltage fluctuations versus wire-feed-rate for aluminum wires, atomizing gas (air) pressure of 45 psig (412 kPa), Voltage of 29.4 V. The data was collected by taking several snapshots of the voltage fluctuations and counting/averaging the number of cycles in a time interval of about 10 or 20 milliseconds. The error bars are the standard deviation of the collected data. ..............47
xi
Figure 3.7 Average volume of metal detachment, calculated from equation (3-1) and data in Figure 3.6, presented as a function of wire-feed-rate. Material vaporization is neglected in this analysis......................................................................................................48
Figure 3.8 Volume of the gun in which atomizing gas flows. Four tubular inlets carry pressurized atomizing gas (mainly dry air) into the gun chamber. The opening on top (and its bottom mirror-image) is where contact tips and wire guides are located....................................................................................................................................50
Figure 3.9 Inner components of the ValuArc 200 wire arc gun. Contact-tips guide the wires towards the nozzle. Diameter of the wires is 1.6 mm.........................................................50
Figure 3.10 Contours of gas velocity (a) and pressure (b). Numbers are in m/s and Pa, respectively. Mass-flow-rate of air is 12.3 gr/s. κ-ε turbulent modeling.........................52
Figure 3.11 Reduced geometry of the gun included contact-tips and wire-tips. .................................53
Figure 3.12 Contours of gas velocity (a) and pressure (b) for mass-flow-rate of 25.3 gr/s. Numbers are in m/s and Pa, respectively. κ-ε turbulent model. Arc heating is not considered. ............................................................................................................................54
Figure 3.13 Contours of gas temperature (a) and mass-flux density (b) for mass-flow-rate of 25.3 gr/s. Numbers are in K and kg m-2s-1, respectively. κ-ε turbulent model. Arc heating is not considered......................................................................................................55
Figure 3.14 Fluid streamlines parallel to YZ (a) and XZ (b) planes in the vicinity of wire-tips. Streamlines are coloured by Y-component (a) and X-component (b) of gas velocity. Mass-flow-rate of gas is 25.3 gr/s. Arc heating is not considered.....................................56
Figure 3.15 Contours of shear stress on the wire tips. Numbers are in Pa. Mass-flow-rate of air is 12.3 gr/s. Arc heating is not considered. .........................................................................57
Figure 3.16 Numerical predictions of volumetric-flow-rate of air as a function of atomizing gas pressure compare relatively well with experimental measurements. Experiments correspond to pressure settings of 20, 30, 40, 50, and 60 psig. Numerical results of κ-ε model under-predict the flow rate by about 8%. Dash-line takes into account the pressure drop in the connecting hose. LES data points and their error-bars represent time-averaged and RMS values of time-dependent flow-rate. ........................59
Figure 3.17 Shear stress on the surface of the wire-tips for different mass-flow-rates of air. Arc heating is not considered. LES turbulence modeling. .......................................................60
Figure 3.18 Grayscale picture of arc, taken at P = 30 psig (a) and P = 40 psig (b), wfr = 7 m/min, V = 30.1 V, with aluminum wires, and air as atomizing gas. Radiation intensity is translated to grayscale intensity (a number between 0 and 255). The region of higher radiation intensity is then found by stratifying the picture. Shutter speed setting: 750 (a) and 500 (b). .................................................................................................62
Figure 3.19 Grayscale picture of arc, taken at P = 45 psig (a) and P = 60 psig (b), wfr = 7 m/min, V = 30.1 V, with aluminum wires, and air as atomizing gas. Radiation intensity is translated to grayscale intensity (a number between 0 and 255). The region of
xii
higher radiation intensity is then found by stratifying the picture. Shutter speed setting: 500 for both (a) and (b). .........................................................................................63
Figure 3.20 At each pressure setting, arc length from different images was measured and averaged. The error bars represent the standard deviation of the measurements. .......64
Figure 3.21 Arc radius, current density, and electric field as functions of axial distance in a 4-mm long arc with current of 200A. .....................................................................................65
Figure 3.22 Contours of gas velocity (a) and pressure (b) for mass-flow-rate of 25.3 gr/s. Numbers are in m/s and Pa, respectively. κ-ε turbulent model. Arc heating is considered. ............................................................................................................................68
Figure 3.23 Contours of gas temperature (a) and mass-flux density (b) for mass-flow-rate of 25.3 gr/s. Numbers are in K and kg m-2s-1, respectively. κ-ε turbulent model. Arc heating is considered. ...........................................................................................................69
Figure 3.24 Fluid streamlines parallel to YZ (a) and XZ (b) planes in the vicinity of wire-tips. Streamlines are coloured by Y-component (a) and X-component (b) of gas velocity. Mass-flow-rate of gas is 25.3 gr/s. Arc heating is considered. Divergence of gas flow is more than that in Figure 3.14, where arc heating was not considered. .......................70
Figure 3.25 Shear stress on the surface of the wire-tips for different mass-flow-rates of air, with the consideration of arc heating. Turbulence was modeled using κ-ε. ............................71
Figure 4.1 Optical (a) and SEM (b) pictures of aluminum particles collected by spraying into water; P=30 psig (308 kPa), V=32.1 V, wire-feed-rate=7 m/min ......................................78
Figure 4.2 Velocity, diameter and Mass-flow-rate of the spray particles as a function of y and x, with z = 50mm. Center of the spray is located at x = y = 0mm. The error-bars in the graphs represent the standard deviation of 3 to 5 measurements. ............................79
Figure 4.3 Frequency-distribution (a) and volumetric-distribution (b) histograms of measured particle diameter are shown by grey histograms; P=60 psig (515 kPa), V=37.9V, and wire-feed-rate=7m/min. The curve in (a) is a Log-Normal function (μ=56μm, σ=0.451) matching the maximum and full-width-half-maximum of the measured distribution. The curve in (b) is the volumetric Log-Normal function with same μ and σ as in (a) and scaled with the same scaling factor as the measured volumetric-distribution. The black bar-histogram represents the difference between the measured volumetric-distribution and the volumetric log-normal function. .................83
Figure 4.4 An optical picture of magnetically-agglomerated stainless-steel particles before being demagnatized..............................................................................................................85
Figure 4.5 A log-normal function fits well within the error-bars of the size-distribution of anodic particles. Stainless steel and copper wires were used as anode and cathode, respectively. The error bars represent the systematic error of the size measuring device. ....................................................................................................................................86
Figure 4.6 The separation technique was applied to the addition of two known log-normal functions (LN1: µ1=50µm, σ=0.45 and LN2: µ2=90µm, σ=0.45) to reconstruct the original functions. (a) frequency-distribution (b) volumetric-distribution.....................90
xiii
Figure 4.7 Two peaks in the measured diameter distribution were separated and presented in frequency (a) and volumetric (b) forms. LN1 and LN2 represent log-normal distribution functions of cathodic and anodic particles respectively. vLN1 and vLN2 are the volumetric representation of LN1 and LN2. Experimental particle size statistics was obtained by DPV-2000 system at a stand-off distance of 50 mm, voltage of 32.1 V, wire-feed-rate=7 m/min, and P = 60 psig (515 kPa). These distributions represent statistics of about 8000 aluminum particles. ..............................91
Figure 4.8 Mean-Diameter and Mass-Mean-Diameter of cathodic and anodic particles decrease as the pressure of the atomizing gas increases. Anodic particles are more significantly affected by atomizing gas pressure than the cathodic particles. Error-bars represent standard deviation of 3 to 5 measurements of about 8000 particles. Operating parameters: Aluminum wires, V=32.1V, wire-feed-rate=7m/min, stand-off distance=50mm. ..............................................................................................................93
Figure 4.9 Mean-Diameter and Mass-Mean-Diameter of cathodic and anodic particles as a function of the wire-feed-rate. Error-bars represent standard deviation of 3 to 5 measurements of about 8000 particles. Operating parameters: Aluminum wires, P = 60 psig (515 kPa), V = 32.1V, stand-off distance = 50 mm. ............................................94
Figure 4.10 Mean-Diameter and Mass-Mean-Diameter of cathodic and anodic particles as a function of the applied voltage. Error-bars represent standard deviation of 3 to 5 measurements of about 8000 particles. Operating parameters: Aluminum wires, P = 60 psig (515 kPa), wire-feed-rate = 7 m/min, stand-off distance = 50mm.....................95
Figure 4.11 Axial velocity profile of particles in the spray. ..................................................................96
Figure 4.12 Axial temperature profile of particles in the spray with air and nitrogen as atomizing gas. Temperature of aluminum particles sprayed with nitrogen increases by about 130°C as they travel a distance of 20 cm. ...........................................................97
Figure 4.13 Average temperature of aluminum particles in the spray as a function of lateral distance from the centerline of the spray. Axial distance from the gun is 3" (76 mm). V=31V, wfr = 7 m/min, P=30 psig (308 kPa)............................................................100
Figure 5.1 Experimental setup to obtain distinct splats on the substrate........................................106
Figure 5.2 Splat morphology and corresponding coating microstructure of wire-arc sprayed aluminum deposited onto polished stainless steel (type AISI304L) held at various temperatures. ......................................................................................................................109
Figure 5.3 Frequency of disk-shape splats increases with increasing substrate temperature. High-velocity and low-velocity data are adapted from [ 1] and [ 2]. Mid-velocity data are measured solely by the author of this thesis. .............................................................111
Figure 5.4 Degree of splashing decreases with increasing substrate temperature. Adapted from [ 1]................................................................................................................................112
Figure 5.5 Experimental and theoretical spread factor values for both high velocity (143 m/s) and low velocity (109 m/s) tests. The curves are the theoretical predictions from equation (5-6). .....................................................................................................................114
xiv
Figure 5.6 Prediction of transition temperature for aluminum droplets impacting a stainless steel surface. The three experimental data points do not necessarily have similar contact resistances due to the growth of an oxide layer on the substrate......................116
Figure 5.7 Plot of elemental composition of the stainless steel substrates heated to various temperatures. Adapted from [ 1]. ......................................................................................117
Figure 5.8 Effect of substrate temperature on porosity of the produced coating. ..........................118
Figure 5.9 Measured deposition efficiency for high velocity (143m/s) and low velocity (109m/s) test conditions. Curves represent the best fit. ..................................................................120
Figure 5.10 Measured coating adhesion versus substrate temperature for particles having an average velocity of 143m/s. [ 1, 2] .......................................................................................121
1
Chapter 1
Introduction
This opening chapter introduces the concepts of thermal spray as well as the twin-wire-
arc spray process and its spraying gun. A summary of the scientific literature on this topic, and
also the motivation behind the current work are presented. The scope of the work is defined and
an outline for the structure of the thesis is provided.
1.1 Background, Motivation, and Literature Survey
1.1.1 Thermal spray
Thermal spraying is a group of elevated-temperature, high-velocity material processing
techniques in which molten or semi-molten particles are accelerated towards and deposited onto
a prepared surface, on which the deposited particles are solidified or sintered and form a
protective coating layer. The particles in this process are either introduced in the form of solid
ceramic or metallic particles that are heated/melted in a hot flame, or, are atomized off the
molten tip of an electrically-conductive metallic wire.
The coating layer that is formed in a thermal spray process is a collection of many
individual solid flattened particles, splats, that pile on top of each other. The coating layer
formed, depending on its microstructure, can be used in various industries (including electronic,
automotive, aeronautic, and aerospace) to provide:
2
• Resistance to wear, abrasion and erosion
• Thermal barrier coating to protect structures and materials
• Corrosion resistance in air and marine environments
• Protection against high temperature oxidation, erosion and corrosion
• Electrical resistance, electrical conductivity, or electro-magnetic shielding
• Layer-by-layer manufacturing of shaped components
• Dimension restoration for worn surfaces
• Building composite structures of metals and ceramics
• Adhesive base for bone ingrowth in medical implants
Because protective coatings are becoming a widespread part of metal working, thermal
spray is a rapidly growing market.
The methods classified under thermal spraying include flame spraying, plasma spraying,
High-Velocity-Oxyfuel (HVOF), and twin-wire-arc spraying, all of which share the following
common features:
1) Production of molten particles: molten particles are formed in the spraying gun region
2) Transport of molten particles: molten particles are propelled towards the substrate to
be coated
3) Particle deposition: molten particles are deposited and flattened on the substrate,
forming solidified or sintered splats that form the coating layer.
These three common features are the basic principles underlying the thermal spray
process and are schematically shown in Figure 1.1.
3
Figure 1.1 Basic principles underlying the thermal spray processes: Production, Transport, and
Deposition of molten particles.
Despite these common features, there are major differences between thermal spraying
methods; namely:
• Type of feedstock: the spray material can be introduced as powder, wire, suspension,
or solution.
• Heating method: the spray material can be heated with an electric arc or a flame.
• Cost: The equipment and operation costs involved vary from one method to another.
• Microstructure of the coating produced: Microstructure of the coating produced
varies from one thermal spraying method to another, making them suitable for
different applications, such as in corrosion protection or wear-resistant coatings.
• Heat transfer to the substrate: Deposition and solidification of molten particles on a
solid substrate warms up the substrate. (This may even cause melting and re-
solidification of a layer of the substrate). In some applications, where heat treatment
of the substrate changes its desired mechanical properties, it is required that the heat
transfer to the substrate be minimized by proper selection of spray technique.
4
Table 1.1 Comparison of three thermal spray processes [ 8, 33, 66]
Thermal Spray Processes
Criteria Wire-Arc HVOF Atmospheric Plasma
Type of feedstock Wire Powder Powder, suspension, solution
Choice of material Restricted to electrically-conductive wires: metals, metal alloys, metallic composites
Metallic alloys, carbides and composites; Limitation for refractory materials
Metallic alloys, carbides, ceramics, composites, and refractory materials
Versatility in the choice of material
Low Medium High
Reliability Medium High High
Heat transfer to the substrate
Low Very High Medium
Process Cost Low Medium High
Capital Cost 20000$ 100000$ 100000$
Deposition Rate 1 – 50 kg/hr 1 – 10 kg/hr 0.5 – 10 kg/hr
Adhesion Strength Low High High
Coating Porosity 10% – 25% 0.5% – 5% 0.5% – 10%
Maximum Temperature
4000oC – 6500oC 2600oC – 3100oC Up to 20000oC
Particle Temperature Low Medium High
Particle Velocity 80 – 150 m/s 550 – 1000 m/s 100 – 300 m/s
Particle Diameter 30-50 μm Powder (10-100 μm) Powder (10-100 μm)
Surface Roughness 2 - 10 μm 1 - 4 μm 1 - 4 μm
Wire-Arc, HVOF, and Atmospheric Plasma spray systems are extensively used and
compared against each other in industrial applications. For example, Barbezat [ 8] has compared
the performance of these systems in deposition of protective coating layers on engine cylinder
bores. Table 1.1 summarizes such results and compares coating characteristics for these three
thermal spray processes.
Twin-Wire-Arc Spray, also termed as Wire-Arc Spray, is one of the most cost-effective
methods of thermal spraying and is described in more detail in the next section.
5
1.1.2 Twin-Wire-Arc Spray
Twin-wire-arc spraying is an economical technique of thermal spraying and has become
popular in the industry because it combines low operating and equipment costs with high
material and energy efficiencies. The coatings produced in the wire-arc spraying process usually
have a greater porosity and lower adhesion strength than those obtained from other thermal spray
processes, making them of relatively poorer quality. Nevertheless, in some applications, higher
amounts of porosity are acceptable or even desired (e.g. to hold lubricants [ 23]). Twin-Wire-Arc
spray has a wide range of industrial applications, many of which are listed in [ 56], categorized
according to different wire materials and coating thicknesses.
1.1.2.1 Description of the Wire-Arc spraying process
Figure 1.2 schematically shows the twin-wire-arc spraying process and the spraying
system. The material that is used in this process is introduced into the process in the form of two
electrically-conductive consumable wires. The wires commonly used in industry are usually
made of electrically-conductive materials, such as Aluminum, Zinc, Stainless Steel, and Copper.
In some recent wire-arc developments, intermetallic compound coatings and metal-ceramic
composite coatings were prepared by using pre-alloyed wires [ 25, 68] and cored wires [ 18, 30, 60],
in which non-conductive materials are used as the core of a conductive wire.
Figure 1.2 Schematics of wire-arc spraying system and its major components
6
In this process, the two electrically-conductive consumable wires are continuously fed
into the wire-arc spray gun. The gun geometry is designed such that the wire tips are separated
by a distance of about a few tenths of a millimeter from one another. An electric voltage of about
40 V, applied on the two wires, causes an arc in the gap between the wire tips. The arc heats the
tips of the wires, and produces a thin layer of molten material. A stream of atomizing gas, as
shown in the figure, strips the molten layer off the wire tips and propels it towards the substrate.
The detached molten droplets will not only accelerate, but also undergo further atomization and
produce smaller droplets. The molten droplets are then deposited on the substrate one over
another, where they solidify and form a coating layer.
Some of the advantages that make wire-arc spraying an attractive process to industrial
users are [ 23, 26]:
1) The simple design of the equipment, the low cost of producing wires, and the low
cost of electrical power, makes this process cost-efficient. The capital and the
operation costs of a typical wire-arc spraying gun are respectively five times and nine
times less than those of a typical plasma spraying system. [ 33, 58]. Besides, because
of its light weight and portable design, wire-arc system is an ideal tool for on-site
application of coatings.
2) The use of non-flammable gases and the possibility of using dry-air as the atomizing
gas make the process safer and even more cost-efficient.
3) Since the gas heating region is very small, the atomizing gas has a relatively low
temperature, reducing the rate of heat transfer to the substrate. Keeping the substrate
temperature at a low level prevents damage, distortion, and metallurgical changes to
the substrate surface.
7
4) The materials introduced to the system are entirely melted and consumed in the
process. Complete melting of the input material makes the wire-arc spraying a
material- and energy-efficient process. This also eliminates the problems caused by
partially-melted particles. (Incomplete melting of particles is associated with almost
all other thermal spray processes that use solid particles or powder as feedstock. In
such processes, heating and melting starts from the periphery of the particle and the
inner portion may not reach the melting temperature).
5) It has been experimentally shown that the wire-arc spraying process has the highest
coating rate among all other thermal spray processes. Also, the deposition efficiency
(mass ratio of coated material to fed material) of this process is shown to be
comparable and, in some cases, better than that of other thermal spray processes.
Despite the advantages that make wire-arc spraying an attractive process to industry, it
has certain disadvantages too. Its major disadvantage is the poor quality of the coatings
produced: Wire-arc coatings are usually characterized by their high porosity and low adhesion
strength, which are undesirable in most applications.
Another disadvantage of wire-arc spraying is the lack of control over the size of the
produced particles. In most thermal spray processes, the molten droplet sizes are determined by
solid particle sizes in the fed powder, whereas, the molten droplet sizes in wire-arc are
determined by the operating parameters. It should be noted that particle size controls dynamic
and thermal behavior within the spray and the splashing and spreading behavior during
deposition. Therefore, lack of control over particle size effectively limits the ability to adjust
coating characteristics according to wide range of industrial requirements, thereby narrowing the
range of wire-arc applications.
8
1.1.2.2 Operating parameters
The industrial wire-arc spraying systems that are designed and manufactured by different
companies are all based on concepts discussed in section 1.1.2.1. However, they differ in the
geometry of the torch, the wire-diameter, the feed mechanism for the wires, and the location
where the tips of the wires are located with respect to the torch geometry. In this study, the
ValuArc 200 Twin-Wire-Arc Spraying Gun, manufactured by Sulzer-Metco (Westbury, NY)
was used. Detailed engineering drawings of this gun are presented in Chapter 2, section 2.2.
The parameters that can be controlled in most wire-arc spray systems and affect the
microstructure of the produced coatings include [ 54, 55, 70]:
1) Material of the fed wires: The wide range of materials being used in industry include
aluminum, copper, stainless steel, tin, titanium, and zinc. Each one of these materials
or a combination of two of them can be used for different applications. In some
applications, cored wires are used, in which non-conductive materials, such as
carbide, nitride, or cermet, are wrapped inside an electrically conductive outer layer.
2) Type of the atomizing gas: The types of the atomizing gas used in industrial
applications include dry-air, nitrogen, and argon. The atomizing gas strips molten
material off the tips of the wires and therefore needs not be flammable. Use of gas
mixtures that do not contain oxygen reduces the oxide content in the coating
produced.
3) Pressure of the atomizing gas: The upstream pressure of the atomizing gas determines
the volume-flow-rate and the velocity of the atomizing gas. The gas velocity directly
affects the velocity of detached droplets.
4) Voltage applied on the wires: The applied voltage controls the input power of the arc
and indirectly affects the rate at which the tips of the wires are heated and melted.
9
5) Wire-feed-rate (only in type 1 systems), also termed as wfr: This parameter
determines the rate at which material is introduced into the system, which is
equivalent to the rate at which the introduced material is melted. In type 1 systems
that allow control over wire-feed-rate, there is no control over the arc-current. Arc
current, which is directly proportional to the input power, is determined based on
melting rate and other system settings. An example of a type 1 spraying system is the
ValuArc system (manufactured by Sulzer-Metco).
5) Arc current (only in type 2 systems): This parameter controls the input power of the
arc, affecting the rate at which the tips of the wires are heated and melted. In type 2
systems, there is no control over the wire-feed-rate. It is therefore determined based
on other system settings, including the arc current. An example of a type 2 system is
TAFA 9000 Wire Arc (manufactured by Praxair/Tafa, Inc.).
In addition to the abovementioned control parameters, there are other factors that may
affect the coating quality or its characteristics which are not controlled by the knobs and switches
on the spraying system: The state of the ambient gas (e.g. oxygen and moisture content), and the
state of the substrate (e.g. temperature and roughness) may affect the coating characteristics.
10
1.1.3 Brief Literature Review
The idea of wire-arc spraying was first introduced by Schoop in 1910 and quickly found
commercial applications in Germany, France and United States [ 26]. Since then, there have been
many improvements in its design in accordance with its rapidly expanding applications.
However, due to the relatively poor coating quality (as compared to other thermal spray
processes) and its high-efficiency high-deposition-rate, it has mostly been used in applications
that require thick coatings and are less-demanding, such as corrosion or wear protection coatings
[ 33].
With its increasing applications in the industry in the last decade [ 19], the wire-arc spray
process is seeing an upswing in terms of research interest. However, the research work done in
this field is mainly experimental and little work has focused on fundamental modeling. In this
section, previous experimental and numerical work on production of molten droplets using the
wire-arc technique is summarized first, and studies on in-flight particle characteristics and
droplet deposition follow.
1.1.3.1 Previous Work on Droplet Production and Transport
One of the earliest studies on the performance of wire-arc spray was conducted by
Steffens in 1966, in which he used an oscilloscope and high speed cinematography to show 1 to
2 kHz fluctuations in the spray process and that the process is unsteady in nature [ 33]. The high
speed photographs also revealed asymmetric melting of anode and cathode material: while
localized melting was observed in cathode, the molten material of the anode formed a sheet
before breaking up. A complete explanation of this phenomenon, however, was not given. Later,
Marantz in 1974 showed that size of the sheets produced at the anode decreased with increasing
pressure of the atomizing gas.
11
The past two decades have witnessed more research studies on wire-arc. However, most
of these studies have focused on relating coating properties with operating parameters of wire-
arc, and few of them focused on improving the process design. One of the major improvements
in the design of the wire-arc system was suggested by Russ in 1993: he proposed using a
converging-diverging nozzle (instead of the conventional straight bore nozzle) to increase gas
velocities in the spray. Using converging-diverging nozzle results in a much weaker diamond
shock structure in the gun, and results in a higher gas flow velocity, and therefore, higher particle
velocity.
One of the earliest theoretical descriptions of arc heating is given by Steffens in 1990 for
single-wire-arc process, in which molten metal is directly transferred from the consumable wire
(that acts as one electrode) to the substrate (that acts as the second electrode). In this process,
gravity, rather than an external gas flow, is used to transport molten particles to the substrate.
In another study, Varacalle et al [ 64] modeled the arc, jet, and particle transport and
heating in the plume of a wire-arc system using already developed codes. These codes solved
two-dimensional simplified models of an arc with a parabolic laminar cross-flow. They did not
model particle breakup and assumed a single size for the molten droplets.
The most comprehensive attempt to date to model the wire-arc spraying process has been
performed by Kelkar [ 33- 35] and Hussary [ 28], in which they modeled both the fluid flow and
the arc in presence of fluid flow. Assuming the traditional particle breakup model of Amson [ 4]
and Arai [ 5], they modeled primary and secondary atomization. In a separate study, Hussary et al
[ 29] studied the mechanisms involved in primary atomization of molten metal from the wire tips
and the effect of process parameters on these mechanisms. In their study, they presented
quantitative results about sheet, extrusion and membrane lengths, and breakup times.
12
1.1.3.2 Previous Work on Bimodal Size Distribution of In-flight Particles
It has long been known that the anode and cathode are heated differently in a wire-arc
process. The arc attaches to the anode over a larger area than the cathode where heating is more
localized at the cathode spot [ 67, 28]. At the tip of the anode-wire a large area is heated due to
diffuse arc-anode attachment, melting a layer of metal that is pushed off the edge of the wire-tip
by the atomizing gas, creating an “anode sheet”. At the cathode, constricted arc attachment
causes much more localized heating and melting. Also, since the current passes through a smaller
area the current density (j) at the cathode surface is much higher, producing a large magnetic
pinch force (or j B×v uv
force, where Buv
is the induced magnetic field). Molten metal droplets
ejected into the arc from the cathode due to both drag and magnetic forces are observed to be
smaller than those that detach from the anode.
Using laser strobe photography Hussary et al [ 28] and Watanabe et al [ 68, 69] clearly
illustrated the differences between molten metal detachment at the tips of the anode and cathode
wires.
To date, no numerical work has been performed to model the size and shape of droplets
from anode and cathode. Kelkar et al [ 33, 34, 35], who numerically modeled the wire-arc process,
used a simple breakup model and a simplified secondary atomization model to determine the
particle size distribution in a wire-arc plume. Although they showed a bimodal distribution for
particle size, their graphs do not predict equal mass-feed-rates of anode and cathode material.
Inhomogeneity in the microstructure of wire-arc coatings was also observed by Zhu et al
[ 72]. By spraying two different materials as anode and cathode, they demonstrated that particles
originating from anode and cathode are distributed in an asymmetric way about the centerline of
the wire-arc spray.
13
1.1.3.3 Previous Work on Particle Deposition
Determining the effects wire-arc operating parameters have on coating characteristics is
the most practical part of research on wire-arc spraying. Such practical studies have been
conducted using a wide variety of techniques. Wang et al [ 66], for example, used optical
microscopy and Auger electron spectroscopy to study the effect of the atomizing gas pressure on
porosity and oxide content of the produced coating. Their work indicated an increase in oxide
content and decrease in porosity with increasing atomizing gas pressure. They also sprayed wire-
arc particles into an ice block to freeze the molten particles with minimal deformation, and
measured in-flight particle size distribution, in which they found two peaks.
Other work by Varacalle et al [ 63, 64] involved a systematic experimental design along
with two-color pyrometry technique and image analysis to study and determine the effects arc
voltage, gas pressure and spray distance have on roughness, oxide content and porosity of the
produced coating. They found that increasing arc current, lowering atomizing gas pressure, and
shortening spray distance results in higher coating roughness, lower oxide content and lower
porosity.
Another subject that has captured the interest of researchers in the field of thermal
spraying is the effect of substrate conditions (e.g. temperature, roughness, and contaminants) on
the quality of the produced coatings. However, most studies have focused on plasma spray
particles. Pershin et al [ 50], for example, have shown that by increasing the substrate
temperature from 20oC to 650oC, the adhesion strength of plasma-sprayed nickel powder
increases by almost an order of magnitude. Also, effect of substrate temperature on the shape of
individual splats has been studied by several researchers and has been summarized by Fauchais
et al [ 19]: A thermal spray particle landing on an unheated surface will splash and form a
fragmented splat. A particle landing on a heated surface, however, forms a circular disk with no
14
irregular edges. Fukumoto et al [ 20] introduced a “transition temperature” (Tt) for the substrate
above which most of the deposited droplets are disk-shape splats. In another attempt, Jiang et al
[ 32] showed that removing contaminants from the surface of the substrate increases the
probability of obtaining a disk-shape deposit.
Therefore, it can be seen that, to date, there has not been a single comprehensive
numerical treatment of the wire-arc spray system. The development of an acceptably complete
model of the wire-arc process requires that all parts of this complicated process, including
particle production, particle transport, and particle deposition, be analyzed and modeled.
1.2 Statement of Objectives
Advantages of wire-arc spray system, including its low costs and high material/energy
efficiency, have made its market grow rapidly in the thermal spray industry. Despite its simple
design, the physical phenomena underlying its operation are very complex and have yet to be
understood. Fundamental understanding of the wire-arc spray process is needed to better control
the spray particle properties and optimize its performance for different industrial applications.
Better understanding of the wire-arc spraying process will also provide us with some guidelines
as to how existing wire-arc spray equipment designs can be modified to improve the quality of
the produced protective coating layers.
The objective of this research is to quantitatively analyze and model how the wire
material is heated, melted, atomized, transported, and deposited onto the substrate: Knowledge of
fluid flow in the wire-arc gun is necessary to predict the size and shape of the primary
detachments from the wire-tips. Knowledge of particle properties and their distributions is
needed because it directly affects the quality of the produced coating. Knowledge of individual
15
particle deposition (and splat formation) is necessary because it determines the microstructure of
the produced coating layer.
As different industrial applications of thermal sprays require different ranges of particle
velocities, particle temperatures, and particle sizes, the ultimate goal of this research is to use the
obtained knowledge and suggest methods of controlling/enhancing the wire-arc process and
make it more suitable for each application.
1.3 Scope of the present work
Due to the complex physical phenomena underlying the operation of the wire-arc system,
certain simplifications have to be made:
• Although the particle detachment, arcing, and therefore, arc heating are highly
oscillatory, all processes are assumed to occur at their average properties. In other
words, time dependence is not considered.
• The arc is modeled with a simplified semi-2D model, originally developed by
Lowke [ 38].
• Pictures of the arc are used to identify the shape of the arc. Pictures are then used to
prescribe arc info in the CFD solution of the fluid flow in the gun.
• The arc is solved independent of the cross-flow, except that the arc shape/length is
determined from the arc pictures.
• A simplified model of primary atomization is used to determine the size of primary
molten metal detachments from the wire-tips.
16
1.4 Outline of the thesis
This thesis presents the results obtained from numerical and experimental investigations
of three main segments of a wire-arc spraying process: Particle production, Particle transport,
and Particle deposition.
Chapter 2 introduces the experimental apparatus and procedures used in this research
work, including coating-characterization and process diagnostics tools. Section 2.2 of this
chapter analyzes the experimentally measured arc current to estimate metal evaporation rate.
Chapter 3 introduces the imaging system used to take pictures of the arc and wire-tips
during operation. It also presents the post-processed pictures of the arc, which are used in solving
simplified arc-equations. It also describes how this arc solution is used in modeling arc-heating
in the wire-arc gun. Finally, a simplified model of liquid metal breakup is used to estimate the
size of primary molten metal detachments from electrodes.
Chapter 4 discusses the size-distribution of particles in the wire-arc spray. A method is
presented to identify and distinguish the size-distribution of particles originating from anode
from those originating from cathode. Oxidation of aluminum particles and its effect on axial
temperature profile is also discussed.
Chapter 5 summarizes the experiments conducted to study the effect of substrate
temperature on splat formation and coating properties.
Chapter 6 summarizes the conclusions drawn from the findings of this study and presents
recommendations for future work.
17
Chapter 2
Experimental Apparatus and Procedures
This chapter describes the experimental apparatuses used in this study for
experimentation and characterization of the wire-arc spraying process. The major experimental
setup that was used in this study included a wire-arc spraying system (ValuArc 200, Sulzer
Metco, Westbury, NY), a metallic substrate (placed in front of the spray gun), a heater (to heat
the substrate), and diagnostic tools (to monitor the process parameters during operation).
Operating this spraying system produced a coating layer on the substrate. Diagnostics methods
were also employed to characterize the properties of the produced coating layer.
2.1 Coating and Process diagnostics
Process characterization and product characterization are essential steps in all industrial
applications. In thermal spray processes, in-flight particle characteristics, as well as coating
characteristics can be determined using well-developed methods that are briefly described here.
2.1.1 Coating Characterization
Coating characterization is a well-established and well-documented field of science [ 30,
33, 63]. Different material properties, e.g. density, porosity, oxide content, and elemental
composition, can be obtained for any thermal spray coating.
18
In a typical procedure for determining the microstructure, a cross-section of the coating is
obtained by cutting the coating and mounting it into epoxy. The cross-section is then finely
polished to expose its microstructure. Pictures of the microstructure can be taken with either an
optical microscope or a Scanning Electron Microscope (SEM). The pictures that can be easily
digitized are then imported into an image analysis program, with which the porosity of the
microstructure is determined.
A typical procedure for determining the elemental composition in a coating layer is to use
X-Ray Photoelectron Spectroscopy (XPS), described in detail in [ 24]. In this technique, which is
carried out in ultra high vacuum conditions (UHV, P < 10-9Pa), surface contaminants are first
removed from the surface of sample by means of argon-ion-sputtering. The sample is irradiated
with a beam of X-Rays, resulting in electron excitations and ejection of electrons from the
material (photoelectric effect). The energy signature of the emitted electrons (counts per energy
intervals, magnitude and width of the peaks in its spectrum) is then translated to elemental
composition of the sample.
Coating densification can be easily evaluated by measuring its weight in air and in
distilled water. This method, recommended by American Society for Testing and Materials, is
coded as ASTM B 311-93. [ 7]
2.1.2 Process Characterization
Real-time monitoring of in-flight particle characteristics (such as temperature, velocity,
and size) provides a useful tool for the operator of a thermal spray system to control the quality
of the produced coating as it is being produced.
The apparatus used for in-flight particle characterization in the present work is the DPV-
2000 system (manufactured by Tecnar Automation Ltd., Montreal, QC, Canada). This optical
19
monitoring device has a sensing head that consists of a focusing lens, a two-slit photomask, and
optical fibers (Figure 2.2). This sensing head is aimed perpendicular to the spray particle flow
(Figure 2.1) and can be moved (with two degrees of freedom) to scan a cross section of the spray
plume. This device measures properties (velocity, temperature and size) of individual particles
by analyzing the infrared radiation emitted by each particle passing through the field-of-view of
its sensing head. A photomask with two vertical slits is fixed in front of the optical sensor so that
two peaks are recorded whenever a particle is detected. (Figure 2.3). Two photomasks are
supplied with the device allowing to measure particle velocities below and over 400 m/s. The
low velocity mask, P4590170, was used in these experiments.
(a) (b)
Figure 2.1 (a) Picture of DPV-2000 scanning unit alongside the wire-arc spraying gun detecting the
in-flight particles (b) The computer system containing the DPV-2000 operating system and CPS-
2000 modules; the two are connected to the scanning unit via fiber-optic cables.
20
Figure 2.2 Schematic diagram of the DPV’s optical sensing head and its field of view [ 9]
(a) (b)
Figure 2.3 (a) A schematic diagram showing the signal sensed by the DPV-2000 sensing head
when a particle passes through its field of view. (b) Picture of the two slits in P4590170 photo mask.
Figure 2.4 Wire-arc sprayed stainless-steel particles are approximately spherical.
21
The following information can then be deduced from the recorded signal [ 53]:
• Particle velocity is measured by recording the time taken for a particle to traverse the
known distance between the two slits (multiplied by magnification of the lens);
Measurable range: between 10 m/s and 1500 m/s depending on the photomask. Error
of accuracy of velocity measurements is less than 0.5%. [ 36, 53]
• Temperature is determined using principles of two-color-pyrometry. Temperature is
directly related to the ratio of the strength of emission at one wavelength to the other.
It is assumed that the detected particle is a spherical (see Figure 2.4 as an example)
gray-body emitter. Minimum measurable temperature depends on the emissivity and
size of the particle. For a typical wire-arc-sprayed aluminum particle
( 0.3 , 50pd mε μ≈ ≈ ), this minimum temperature is about 1500 ̊C. Temperature
measurements are 3% accurate. [ 36, 53]
• Diameter is determined by measuring the total radiation emitted by each particle.
(The total emission is proportional to the square of diameter). Diameter
measurement range is between 10µm and 300µm [ 53, 62]. Although relative values
of diameter measurements are precise (about 1% precision, [ 53]), the error of
accuracy associated with its measurement can be as much as 7%. [ 36]
To correctly measure the above properties, it is essential that the device be properly
calibrated by modifying the calibration factors in the DPV software. The velocity calibration
factor has to be modified each time the two-slit photomask is changed. The diameter calibration
factor has to be calibrated whenever a new material (with new emissivity) is being measured.
Temperature measurements are calibrated using a supplied calibration module (consisting of a
pre-calibrated lamp) whenever the filters of the two-color-pyrometer are replaced.
22
To calibrate the diameter measurements of DPV, wire-arc particles were frozen and
captured by spraying them into a dry-ice block or water. After washing and drying the particles
with acetone, they were spread on a white piece of paper and their pictures were taken using an
optical microscope and a digital camera. Measuring diameters of several hundred particles
provided an estimate of size distribution of wire-arc particles (histogram in Figure 2.5). The size
distribution of such collected particles were also found by using MasterSizer S (Malvern
Instruments, UK) Particle Size Analyzer (PSA), which is a single lens laser diffraction system
that evaluates the size distribution of a powder (of solid particles) by measuring its laser
scattering data. Particle-size detection range of this device is from 0.05µm to 880µm.
The size-distributions obtained from these two methods (optical picture measurements
and PSA measurements) were then used to modify the diameter factor in the DPV software and
calibrate the device. Figure 2.5 compares these size distributions after calibration for aluminum
particles. It shows that DPV diameter-measurements satisfactorily matched measurements of the
other two methods (except for very small particles, 10pd mμ< ). Furthermore, Mauer et al [ 40]
compared DPV-2000 in-flight particle measurements against Accuraspray-g3 diagnostics system
and have confirmed the measurement accuracy of both systems. On the other hand, Vaidya [ 62]
has reported that DPV is unable to accurately measure properties of very small particles (less
than 5µm) and Biancaniello [ 9] discusses that the gray-body assumption introduces some errors
in measuring temperature of particles whose emissivity is highly dependent on wavelength of the
emitted beam (such as Molybdenum). However, these shortcomings of the DPV-2000 system
will not be of great importance to this study.
23
Figure 2.5 Size distribution of particles was measured using two additional methods (optical picture
measurements, and PSA measurements) to calibrate DPV’s particle size measurements.
In addition to particle properties, it is also necessary to measure the temperature of the
stationary wire tips during operation. Since this temperature cannot be measured using the
pyrometer system of the DPV, a separate custom-made two-color pyrometer system was used. It
consists of two photomultiplier tubes measuring the intensity of thermal radiation at two
different wavelengths, and finds the intensity ratio using an electronic circuit. In theory, the
intensity ratio is related to the temperature of the object, following Wien’s law:
1 2 1
2
525
1
( ) B B
hc hck T k TI
R T eI
λ λ λ
λ
λλ
−
= = (2-1)
where kB , h and c are the Boltzmann constant, Planck’s constant and the speed of light in
vacuum, respectively. The pyrometer output signal, however, depends on the voltages that are
applied on the photomultipliers, and has to be calibrated. The pyrometer output signal was
therefore measured and plotted against different temperatures of a heating element (Figure 2.6).
24
The obtained plot was then used as the calibrated reference curve to measure the unknown
temperature of the wire-tips by aiming the pyrometer at the wire-tips during operation and
measuring the pyrometer’s output signal voltage.
Figure 2.6 Pyrometer’s calibrated reference curve. High voltages on λ = 780 nm and λ = 850 nm
photomultipliers were 700 V and 1000 V, respectively.
Another factor that greatly influences the coating quality is the state of the substrate on
which the coating is being applied. Substrate temperature, its elemental composition, and its
surface roughness, are shown to have huge effects on the coating microstructure and physical
properties [ 19]. Therefore, it is necessary to characterize the substrate properties before and
during the spraying process.
To measure and control the substrate temperature, two or three J-type thermocouples
were attached to the substrate: one on top of the substrate (on its exposed face), and one behind
the substrate, where it was attached to a controlled-temperature heater block. A custom-designed
25
temperature controller unit [ 2] controlled the substrate temperature based on the temperature of
the unexposed surface of the substrate (because the temperature measurement of exposed surface
was unreliable, due to extreme air flow around it).
Surface roughness measurements were done prior to spraying using PDI Surfometer
Series 400 (Precision Devices, Inc., Milan, MI), which records the surface profile of a
component by running a stylus over it. This instrument then gives an average surface roughness
value, which will be denoted as Ra.
Also, substrate surface profile was looked at using a Surface Profiling Microscope (Wyko
Optical Profilometer, Veeco Instruments. Inc., Woodbury, NY). A sample surface profile is
illustrated in Figure 2.7.
Figure 2.7 Surface profile of polished AISI 304L stainless steel substrate obtained using a Surface
Profiling Microscope (Wyko Optical Profilometer, Veeco Instruments Inc., Woodbury, NY). Surface
roughness is 7.90 nm.
26
Another characterizing parameter that was measured for the wire-arc-spraying process in
this study was the deposition efficiency, which was calculated as follows. The mass of the
substrate was measured prior to spraying (mi) and after the spraying (mf). Also, prior to any
testing, the wire-feed-rate setting on the wire arc was measured and calibrated to the desired
setting. With the wire-feed-rate known, and by measuring the duration for which the wire-arc
was sprayed, the mass of the material of the two consumed wires (mc) was determined:
22cm wfr r tρ π= ⋅ ⋅ ⋅ (2-2)
In equation (2-2), ρ is the density of the wire material sprayed, wfr is the calibrated wire-
feed-rate, r is the radius of the wire, and t is the measured spray duration (time in seconds).
Subsequently, the deposition efficiency is given by:
100%f i
c
m meff
m−
= × (2-3)
Numerous tests were carried out at each setting and each time the efficiency was
calculated.
2.2 ValuArc 200 Spraying System and Its Characteristics
The ValuArc 200 Twin-Wire-Arc Spraying System manufactured by Sulzer-Metco
(Westbury, NY) was used to conduct all the experimental work presented in this dissertation.
Figure 2.8 and Figure 2.9 show pictures of this system and Figure 2.10 schematically shows its
components: This system is comprised of
• Power supply and control unit: The ValuArc 200 system uses Sulzer-Metco’s
LCARE power supply, which is a constant-voltage SCR (Silicon Controlled
Rectifier) controlled DC voltage supply. It requires a three phase 440 V 50/60 Hz
27
electrical input. This power supply is installed inside a control unit with switches,
control knobs, indicator lights and digital voltage/current readout. The back of the
control unit contains an air pressure gauge and a regulator to adjust the pressure and
flow-rate of the atomizing gas.
• Wire spool tower to hold the wire spools
• Wire-feed chassis with a mechanism to push the wires towards the gun
• Spraying gun (front and rear views are shown in Figure 2.11 and Figure 2.12). The
front of the gun houses the arc shield, air cap, contact tips and tubes, electrode posts,
and DC power cable connections on the bottom. The rear gun body houses the wire
feed cable connections and the air hose connection.
A handle (Figure 2.11) was also supplied with the system to mount and hand-operate the
gun. However, for the purpose of this study, this handle was removed and the gun was mounted
on an adjustable bracket at the desired positions. The handle was then used to operate the gun
from outside the spraying booth.
Figure 2.8 Picture of ValuArc 200 Twin Wire Arc spray system and the spray gun manufactured by
Sulzer-Metco. Picture is adapted from [ 61]
28
Figure 2.9 Picture of ValuArc 200 Twin Wire Arc spray system during operation.
Figure 2.10 Schematics of ValuArc 200 Twin-Wire-Arc Spraying System and its components.
Picture is adapted from [ 2, 70].
29
Figure 2.11 Exploded rear view of the ValuArc 200 twin-wire-arc gun and its components. The gun
can be mounted on the handle and hand operated, or on a separate mount or robot and remotely
operated. Picture is adapted from [ 70]
30
Figure 2.12 Exploded front view of the ValuArc 200 twin-wire-arc gun and its components. Picture
is adapted from [ 70]
The gun is connected to the control unit with a hose and cable package that includes two
wire-feed cables, two DC power cables (anode and cathode connections), and an air hose. The
wire-feed cables are used to guide the consumable wires from the wire-spool to the gun. The DC
power cables transmit the voltage supplied by the power supply and apply it on the consumable
wires. The air hoses are used to deliver the pressurized atomizing gas intake (dry-air or nitrogen)
from a primary compressor unit (Airtower26, KAESER) or a nitrogen tank to the back of the
control unit, where it is regulated, and also to deliver the regulated gas from there to the gun.
In addition to the above components, an external flow-meter (MEM Thru View,
manufactured by Meter Equipment Manufacturing Inc., OH, USA), and a pressure gauge were
installed upstream of the gas-inlet hose of the spaying gun to measure the volume-flow-rate and
pressure of the atomizing gas. The pressure gauge was installed downstream of the flow-meter
and upstream of the spraying gun.
The spraying system is setup inside an industrial sized spray booth equipped with a
programmable robotic arm (to control the motion of the gun with respect to the substrate), a large
exhaust hood and ventilation system. The sound-proof walls of the booth are designed to reduce
31
the noise level during operation. Since wire-arc operates at 116 dBA noise level and produces
large quantities of harmful dust, fumes and particulates, ear protection and respiratory protection
are required during operation inside the booth.
Parameters that can be directly or indirectly controlled in this system and their operating
range for stable spray conditions are listed here:
1) Material of the fed wires: The following 14-gauge wires (diameter = 1.63 mm)
supplied by Sulzer-Metco (NY, USA) were primarily used in this study:
• Stainless Steel Metcoloy® 2: Fe 13Cr 0.5Si 0.5Ni 0.5 Mn 0.35C
• Stainless Steel Metcoloy® 5: Fe 18Cr 8.5Mn 5Ni 1Si 0.15C
• Metco Al: 99% aluminum
• Metco Copper AW: 99.8% copper
2) Type of atomizing gas: In this study, dry air and nitrogen were used.
3) Pressure of the atomizing gas: The pressure of the atomizing gas determines its
volume-flow-rate and thereby its velocity in the nozzle region. The maximum gas
pressure that can be used is limited by the laboratory’s air compressor unit
(Airtower26, KAESER). The minimum gas pressure is restricted by the stability of
the spray system. Overall, the gas pressure in the ValuArc 200 spray system can be
varied between 20 psig (239 kPa) and 65 psig (550 kPa).
4) Voltage applied on the wires: The applied voltage in the ValuArc 200 spraying
system can be varied from 20 V to 40 V.
5) Wire-feed-rate: wfr in ValuArc 200 spray system can be varied from 3 m/min to 11
m/min. (5 cm/s to 18 cm/s)
32
6) Arc current (indirectly controlled): The current that passes through the wires and
maintains the arc, ranges from about 150 Amperes to about 300 Amperes in the
ValuArc 200 spraying system. This current is determined by other operating
parameters.
The next two sub-sections discuss how these operating parameters interrelate.
2.2.1 Volume-Flow-Rate
Volume-flow-rate of the atomizing gas and its pressure were measured using the external
flow-meter (MEM Thru View, Meter Equipment Manufacturing, Inc., OH) and pressure gauge
installed between the flow-meter and the gun. These measurements are plotted in Figure 2.13 for
air as the atomizing gas. It can be observed that volume-flow-rate increases linearly with
pressure. Since the arcing occurs downstream of the nozzle, this relationship is not significantly
affected by other operating parameters (i.e. voltage, wfr, wire material).
Figure 2.13 Standard volumetric flow-rate of the atomizing gas (dry air) as a function of the
upstream pressure. The data points represent 20 psig (239 kPa), 30 psig (308 kPa), 40 psig (377
kPa), 50 psig (446 kPa), and 60 psig (515 kPa) in the system’s pressure setting.
33
2.2.2 Arc Current
Although the voltage applied on the wires can be directly set using the voltage knob on
the control unit, the current that passes through the arc depends on other operating parameters,
namely, voltage, gas pressure, wfr, and wire material. Increasing arc voltage increases the
electric field within the arc plasma, causing more current to flow. Increasing gas pressure
increases the velocity of the cross-flow gas, increases the cooling rate of the arc region,
decreases the arc temperature and conductivity, and thereby decreases the current that passes
through the arc. Increasing wire-feed-rate shortens the arc length, increases the electric field
within the arc plasma, and thereby increases the current that passes through the arc. Besides,
increasing the wire-feed-rate increases the amount of heat required to melt the fed material and
that is generated by the increased current that passes through the arc. (This is one of the factors
that stabilize the spray operation). Moreover, changing the wire material can affect the arc
current: a material with a higher specific heat or latent heat of fusion requires more power to
melt, thereby the spray gun draws more current from the power supply.
Figure 2.14 shows the voltage-current characteristic relationship of the arc for different
operating conditions when aluminum (Metco Al) wires and dry-air (as the atomizing gas) are
used. It can be observed that increasing voltage or wire-feed-rate significantly increase the
current through the arc. It can also be observed that increasing the atomizing gas pressure
decreases the current, but the change in current is insignificant (less than 4%) and of the same
order of magnitude as the current fluctuations.
34
Figure 2.14 Current-voltage characteristic of the arc for different wire-feed-rates and pressures. Dry-
air as the atomizing gas and aluminum wires were used. The error-bars represent current fluctuation
and standard deviation of 5 to 10 measurements.
Figure 2.15 and Figure 2.16 show that current through the arc changes linearly
(approximately) with wire-feed-rate for different operating voltages, and for both aluminum and
copper wires. Changes in arc current due to higher gas pressures are, as stated earlier,
insignificant and therefore not plotted.
To understand the trend of Figure 2.15 and Figure 2.16, one should notice that more
power is required to melt material that is fed at a higher rate: This is why the system draws more
current. Figure 2.17 and Figure 2.18 show the input power (voltage multiplied by current) as a
function of wire-feed-rate, which is the same as the rate at which material is melted. The slopes
of these graphs determine the additional power required to heat and melt an extra 1 m/min of fed
wire. These graphs can be used for an energy balance analysis of the system.
The main source of energy input into the system is joule heating of arc. A small fraction
(less than 2%) of this energy is propagated in the form of electromagnetic radiation, a
35
considerable portion increases the temperature of the atomizing gas, and the rest heats up (and
melts) the wire-material. Increasing the rate at which wire-material is introduced to the system,
requires more power to melt it, whereas the power required to maintain the arc (by heating the
gas) is approximately constant (and its variation can be neglected). Therefore, any additional
power for an increased wire-feed-rate is approximately equal to the power required to heat up the
additional wire-material. In mathematical terms, the energy required to heat up solid aluminum
(at room temperature) to molten aluminum at its boiling point is:
melting room f liquid boiling melting( ) ( ) 3.10 /solidQ c T T L c T T MJ kgm
= − + + − =&
& (2-4)
where, fL is the latent heat of fusion of aluminum and c is the average specific heat of solid and
liquid aluminum. This means that 3.1 MJ is required to heat up 1 kg of solid aluminum to its
boiling point (liquid form). However, the additional energy provided by the arc is 4.53 MJ/kg
(for V = 25 V; see caption of Figure 2.17). It is speculated that the difference (4.5 – 3.1 = 1.4
MJ/kg) vaporizes a fraction of the fed aluminum. Percentage of the vaporized material can be
estimated from the following relation:
melting room f liquid boiling melting v( ) ( ) 4.53 /solidc T T L c T T f L MJ kg− + + − + = (2-5)
or v3.10 / 4.53 /MJ kg f L MJ kg+ =
where vL and f are latent heat of vaporization and fraction of material that is vaporized,
respectively. Equation (2-5) estimates f to be about 13%. The fraction of vaporized material at
higher applied voltages, 32.1 V and 39.5 V, is estimated at 36% and 44%, respectively. These
estimated fractions explain the considerably large quantity of white powdery-form fume
produced during operation of wire-arc spray with aluminum wires. For copper material, this
estimated fraction of vaporization is considerably smaller (e.g. f = 3% for V = 26.9 V).
36
Figure 2.15 Current that passes through the arc increases when the feed rate of aluminum wires is
increased. Solid curves represent quadratic fits to the datapoints.
Figure 2.16 Current that passes through the arc increases when the feed rate of copper wires is
increased. Solid curves represent quadratic fits to the datapoints.
37
Figure 2.17 Input power versus feed rate of aluminum wires. Slope of 39.5V, 32.1V, and 25V curves
represent 7.90, 7.01, and 4.53 MJ/kg (mega joules per kilogram of aluminum), respectively. Solid
curves represent quadratic fits to the datapoints.
Figure 2.18 Input power versus feed rate of copper wires. Slope of 39.8V, 32.1V, and 26.9V curves
represent 3.01, 2.40, and 1.54 MJ/kg (mega joules per kilogram of copper), respectively. Solid curves
represent quadratic fits to the datapoints.
38
Chapter 3
Particle Breakup: Thermal Spray Gun
This chapter discusses the physical phenomena of arc heating and particle breakup in a
wire-arc spray gun. The custom-made laser-illumination visualization technique used to visualize
the process is discussed first, numerical modeling of fluid dynamics of the gun is presented next,
and discussions on breakup mechanisms (from anode and cathode) and secondary atomization
follow.
3.1 Experimental Studies
One of the most influential components in any thermal spraying process is the production
of small molten droplets. In most thermal spraying processes, solid particles, in the form of
powder, are introduced to the high-temperature high-velocity stream of gas or plasma. The high-
temperature ensures that the particles melt in the process and the high-velocity of gas ensures
that the particles reach a proper speed before impacting the substrate.
In the wire-arc process, however, the small particles are produced throughout the process.
The material is introduced in the form of two solid wires that are continuously fed into the
system and act as anode and cathode for a DC circuit. The tips of the two wires are positioned
close to each other and form an angle ranging from 30 degrees to 45 degrees [ 26]. The DC
voltage applied on the wires causes an electric current to flow between the tips of the two wires
and produces an arc. The heat generated by the electric arc heats the tips of the two wires and
39
melts a thin layer on them. The shear force due to the cross-flow atomizing gas strips the molten
material off the tip of the wires. The detached molten material that is produced in the wire-arc
spray system will then accelerate and undergo secondary atomization.
3.1.1 Imaging system
A custom-made high-speed imaging system was developed to take pictures of the wire
tips and identify mechanisms of particle detachment. Any imaging system requires a light source
to illuminate the region to be photographed, and a light capturing device to record the reflected
light. However, excessive amounts of high-intensity radiation at visible wavelengths exist in the
wire-tip region that are emitted from the arc and high-temperature metal. Therefore, it is not
possible to discern the borders of the wire-tips in a picture taken with a visible-wavelength
camera. However, such pictures can be used to determine the general shape of the arc. Examples
of such photographs are shown in Figure 3.1.
40
(a)
(b)
Figure 3.1 Pictures of the wire-tip region and the arc during operation of the wire-arc spraying
system. These pictures are taken using a visible-wavelength Nikon E3 camera with different optical
filters.
41
To avoid the effect of thermal radiation on the imaging system, two pulsed nitrogen
lasers ( 337nmλ = , UV radiation) were used to illuminate the wire-tip region, while a UV-
intensified CCD camera (Cohu 4910 RS-170, COHU Inc.), originally used by Masri [ 39],
captured the reflected beam behind a narrow-band filter at 337 nm. To show that the thermal
emission at this wavelength is negligible, Planck radiation of a gray body material is plotted as a
function of wavelength in Figure 3.2. These curves are plotted at the melting and boiling
temperatures of the materials used as the fed wires in this study (stainless steel, copper, and
aluminum). Since the temperature of the wire material lies somewhere between its melting and
boiling temperatures (closer to the melting point), it can be observed, from the curves, that its
radiation at 337 nm is relatively small. Therefore, the CCD camera will mostly capture the
reflected laser beam, rather than the thermal radiation.
Figure 3.2 Black body radiation curves at different temperatures, scaled to a maximum of one. The
presented curves represent radiation at the melting and boiling temperatures of Stainless Steel,
Copper, and Aluminum.
42
Other obstacles were also addressed in developing this imaging system:
• The shutter of the CCD camera has to be synchronized with the laser pulse; the
shutter has to be opened before the laser pulse is fired and has to remain open while
the laser pulse is illuminating the region. Synchronizing the lasers with the CCD
camera was done by using the parallel port of a personal computer and parallel-port
programming: consecutive TTL pulses (+5V square wave) were sent to different
pins of the parallel port of a desktop computer by assigning 0 or 1 to the
corresponding port number. The parallel port, connected to the lasers and the CCD
camera, could then fire the lasers and open/close the shutter on the CCD camera at
the programmed times.
• There was a time delay between the sent signal and the actual laser pulse (about 750
ns [ 39]), and also between the sent signal and the closing/opening of the shutter
(about 100 ns [ 39]). These delays were considered in programming the parallel port
signal.
• Since the laser system was relatively large in size, the laser beam was transmitted
from the laser exit to the wire-tip region through an optical fiber. Figure 3.3 shows
the laser output, the optical fiber, and the coupler used to transmit the laser beam
into the fiber.
• Since the CCD camera with its lens system attachment was relatively large in size, it
was not possible to illuminate the region (with the laser) and shoot (with the CCD
camera) from the same angle. Therefore, the two lasers were positioned to illuminate
the region simultaneously from top and bottom (Figure 3.4). Improved illumination
and visibility were obtained.
43
(a) (b) (c)
Figure 3.3 Optical system used to transmit laser beam from the laser system to the region to be
photographed: (a) Laser beam output, (b) coupler, and (c) fiber optic cable.
Figure 3.4 Schematic diagram of the position of the UV-intensified CCD camera and illuminating
laser beams. Laser beams are coupled into and transmitted through the optical fibers along positive
and negative Y axes. Cylindrical lenses focus the beams onto the YZ plane. The CCD camera,
attached to the lens system and aligned with the X axis, takes a picture of the wire tips.
Figure 3.5 shows a picture of the two wires and the detached particles. The main reason
for the poor quality of the picture is that the output of the COHU camera was analog and could
only be recorded on a VCR tape. Still, it is possible to clearly see the detached particles and the
anode sheet.
44
Figure 3.5 Photographs of the two wires and the detached particles. These pictures are the negative
of what was captured by the UV-intensified COHU camera. Two lasers illuminate the area from top
and bottom simultaneously. Aluminum wires and a High-Velocity cap were used; wire-feed-rate = 8
m/min, voltage = 29.1 V, pressure = 45 psig (412 kPa).
Originally, we planned to use the imaging system to measure particle velocities based on
Particle Image Velocimetry (PIV) methodology. However, it was later decided to measure the
particle properties using the DPV-2000 system, and therefore, the pictures taken with the
imaging system were only analyzed for observation of the detachment mechanisms.
30º
1.63 mm
1.63 mm
45
3.1.2 Current and Voltage Fluctuations
In the wire-arc spray gun, melting of the wires occurs at the wire tips, where they are
attached to the arc. When the molten material is pushed by the atomizing gas and breaks away,
the arc extinguishes and then reignites between the points of closest distance between the two
electrodes. This periodic behavior translates into fluctuations of arc voltage and current, because
a change in the distance between the wire-tips results in a change in the arc voltage/current.
The effects of the flow rate of atomizing gas and the arc current on the amplitude and the
peak frequency of the arc voltage fluctuations (with different nozzle configurations) has been
studied by several researchers [ 46, 51, 60, 65, 69]. Steffens and Sheard experimentally showed that
there is a relationship between metal atomization from the wire tips and voltage fluctuations
[ 26]. It was shown that molten metal atomization occur shortly after a minimum voltage is
reached. Steffens and Babiak [ 60] measured the arc voltage fluctuations and identified
frequencies between 500 Hz and 2 kHz in its power spectrum. Watanabe et al [ 69] found this
frequency peak to be between 500 Hz and 1 kHz. Newbery and Grant [ 46], also, analyzed the
FFT (Fast Fourier Transform) of the voltage fluctuation signal. Furthermore, they correlated the
frequency of such fluctuations to the operating parameters of the wire-arc gun and showed that
this frequency increases with increasing atomizing gas pressure and increasing wire-feed-rate.
Planche et al [ 51] also analyzed the amplitude of voltage fluctuations and found that it decreases
with increasing atomizing gas pressure. Wang et al [ 67] used a laser strobe high speed vision
system to record images of metal detachment from the wire tips, synchronized with arc voltage
measurements. They used those images to correlate the fluctuating voltage trace to various stages
of metal detachment. Finally, Hussary et al [ 27, 29] studied synchronized detachment images and
identified different mechanisms of primary breakup: membrane, axisymmetric, and non-
46
axisymmetric. They clearly illustrated that primary particle breakup mechanisms from the anode
and cathode are different, which results in a bimodal (dual peak) particle size distribution.
To further understand primary atomization in the wire arc gun and to focus on the
bimodal particle size distribution, the voltage trace of the ValuArc 200 spray system was
recorded and analyzed in this study. If particles are detached from the anode and cathode with
different average sizes, the rate of detachment from the anode and cathode should be different,
and therefore, theoretically, the power spectrum (FFT) of the voltage trace should contain two
peaks.
To investigate the possibility of having two or more peaks in the power spectrum of the
arc voltage fluctuations, the signal was traced on a TDS 220 Tektronix Oscilloscope (Beaverton,
OR, USA). The observed cyclic signal was then analyzed using the Fast Fourier Transform
(FFT) utility of the same oscilloscope. The FFT spectrum of the signal showed several peaks;
however, these peaks moved back and forth in the frequency domain. This uncertainty in the
peak frequency can be attributed to: 1) the voltage signal noise level, 2) the randomness in the
time the arc is extinguished or reignited, and 3) the inherent errors in estimating the power
spectrum with FFT analysis (with limited sampling rate). Therefore, it was technically
impossible to distinguish frequency peaks.
Instead, the average period of these not-so-uniform voltage fluctuations was calculated by
taking several snapshots of the voltage signal and measuring the time duration of about 100
cycles, several times. The average period of voltage fluctuation provides an estimate of the time
interval between consequent metal detachments. This time period is of the order of half a
millisecond.
The dependence of the average period of fluctuation on the operating parameters of the
wire-arc spray system was also studied. Changing the applied voltage did not cause a
47
considerable change in the main period of the voltage fluctuations. In contrast, this period
decreases with increasing wire-feed-rate or increasing pressure of the atomizing gas (Figure 3.6).
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
6 6.5 7 7.5 8 8.5 9 9.5 10Wire Feed Rate (m/min)
mai
n pe
riod
of fl
uctu
atio
n (m
s)
Figure 3.6 The average period of arc voltage fluctuations versus wire-feed-rate for aluminum wires,
atomizing gas (air) pressure of 45 psig (412 kPa), Voltage of 29.4 V. The data was collected by
taking several snapshots of the voltage fluctuations and counting/averaging the number of cycles in a
time interval of about 10 or 20 milliseconds. The error bars are the standard deviation of the collected
data.
Based on the abovementioned hypotheses and applying conservation of mass, the average
period of arc voltage fluctuations can be related to the wire-feed-rate (wfr) and the average size
of detached molten droplets: Neglecting vaporization of material during primary atomization, the
total volume of detached molten droplets is equal to the volume of metal that is fed into the gun
in the form of wires:
2
12
2
4
4
4
Nw
ii
w
w
dV NV wfr t
d tV wfrN
dV wfr T
π
π
π
=
= = ⋅ ⋅
⇒ = ⋅ ⋅
⇒ = ⋅ ⋅
∑
(3-1)
where wd , Vi, V , and N are, respectively, wire diameter, volume of the ith detached molten
48
particle, the average volume of detached particles, and the number of detached particles during
sampling time interval t. T is the average period of the main oscillation of the arc voltage
fluctuation.
0.0E+00
2.0E-11
4.0E-11
6.0E-11
8.0E-11
1.0E-10
1.2E-10
1.4E-10
1.6E-10
6 6.5 7 7.5 8 8.5 9 9.5 10Wire Feed Rate (m/min)
Exp
ecte
d av
erag
e vo
lum
e of
met
al
deta
chm
ents
(cub
ic m
eter
s)
Figure 3.7 Average volume of metal detachment, calculated from equation (3-1) and data in Figure
3.6, presented as a function of wire-feed-rate. Material vaporization is neglected in this analysis.
Using equation (3-1) and the measured average period of voltage fluctuations (presented
in Figure 3.6), the average volume of molten metal detachments can be estimated. The results are
presented in Figure 3.7. This figure suggests that the size of the molten metal droplets detached
in the primary atomization stage has a maximum in the mid-wfr values, although it is not
significantly affected by wire-feed-rate. Although slightly overestimated, the calculated sizes of
the metal droplets (due to primary atomizations) are of the same order of magnitude as the sizes
of metal droplets observed via the CCD camera. It should be noted that the size of the particles
downstream of the arc is reduced due to secondary atomizations.
Due to unreliability of the voltage fluctuation analysis in identifying the dual-peak
particle-size-distribution, no further work was conducted on this method. Instead, the dual peak
nature of particle-size-distribution was extensively studied using another method that will be
discussed in Chapter 4.
49
3.2 Numerical Studies
One of the most powerful tools available today for design and diagnostics of technical
systems is numerical modeling and it was therefore used in this study to estimate the size of
primary metal detachments from the wire tips in the twin-wire-arc spray system. The numerical
modeling was conducted using FLUENT software (finite-volume based code; FLUENT Inc.,
Lebanon, NH). Solution of fluid flow without consideration of arc-heating is presented first, an
approximate arc solution and the effect of arc-heating on the shear stresses on the wire-tips is
discussed next, and a breakup model to predict the size of primary detachment follows.
3.2.1 Flow dynamics of the nozzle geometry
The geometry of the ValuArc 200 wire-arc gun was created in GAMBIT software (Fluent
Inc., Lebanon, NH, USA) based on the engineering drawings provided by the Sulzer-Metco Inc.
[ 70] and measurements of the components of the gun. The created geometry was then meshed
and introduced to FLUENT. Figure 3.8 and Figure 3.9 show the outer and inner components of
the meshed geometry. This geometry consists of four tubular inlets that carry pressurized
atomizing gas into the main torch chamber. There is a regulating rod around which the atomizing
gas flows and reaches the nozzle exit. Also, the two contact-tips shown in Figure 3.9 guide the
wires towards the nozzle, where the tips are kept at a distance of about 1 mm. Although the wires
are constantly moving (because they are consumed), their motion is not modeled in this study.
50
Figure 3.8 Volume of the gun in which atomizing gas flows. Four tubular inlets carry pressurized
atomizing gas (mainly dry air) into the gun chamber. The opening on top (and its bottom mirror-
image) is where contact tips and wire guides are located.
Figure 3.9 Inner components of the ValuArc 200 wire arc gun. Contact-tips guide the wires towards
the nozzle. Diameter of the wires is 1.6 mm.
In addition to the shown geometry, a cylindrical volume (or a frustum) was attached to
the numerical domain downstream of the nozzle to study the flow downstream of the nozzle exit.
It also increases the distance of the nozzle region (which is of primary importance to this study)
from the boundaries, and therefore, reduces the numerical errors associated with boundary
conditions. The boundary conditions assigned for the model geometry included:
24 mm
Contact-tip Wire-tips
Regulating Rod
Tubular inlets
51
• No-slip boundary condition for the solid surfaces in the geometry, including the wire
surfaces and contact tips.
• Pressure-Inlet for the four tubular inlets
• Pressure-Outlet boundary condition was used for the faces exposed to ambient
atmosphere.
Turbulent fluid flow was solved within the chamber and also downstream of the nozzle
(in the attached cylindrical region). Since the Mach number of the gas flow in the gun exceeds
0.3, compressible flow equations were solved. Arc heating was not considered at this stage.
Turbulence in the flow was initially modeled using the κ-ε model and turbulent intensities
of 10% were assigned to the inlets. Since the κ-ε solution in this geometry is symmetric with
respect to YZ and XZ planes, only one quarter of the geometry was introduced to the FLUENT
software. Also, since the transport properties of air are strongly dependent on temperature, this
dependence was manually assigned in FLUENT. The flow was then solved with second-order
solvers and numerical results were obtained with residual levels below 10-5%. Some of these
results are presented in Figure 3.10 for a mass-flow-rate of 1.23×10-2 kg/s: Figure 3.10 shows
pressure and velocity contours in a cross section of the gun, parallel to the jet. The pressure
contours clearly illustrate that most of the pressure drop occurs in the nozzle region and, that
pressure is almost constant upstream of the contact-tips. Gas velocity and temperature are also
observed to be approximately constant upstream of the contact-tips. Hence, there is no need to
include this region in the numerical domain, as only the shear stresses on the wire-tips are of
concern to this study.
52
Figure 3.10 Contours of gas velocity (a) and pressure (b). Numbers are in m/s and Pa, respectively.
Mass-flow-rate of air is 12.3 gr/s. κ-ε turbulent modeling.
Contact-tip Wire-tip Regulating Rod
(a)
(b)
Velo
city
(m
/s)
Pres
sure
(Pa)
53
A new reduced geometry (Figure 3.11) was therefore created, meshed with a finer mesh,
and solved with similar boundary and flow conditions: “Pressure-Inlet” for the inlet; “pressure-
outlet” for the outlet, no-slip velocity boundary condition for the solid walls, contact tips and
wires.
Figure 3.11 Reduced geometry of the gun included contact-tips and wire-tips.
A cross-section of this new geometry and contours of the solved flow properties
(velocity, pressure, temperature, and mass-flux density) are shown in Figure 3.12 and Figure
3.13. As shown, supersonic velocities create shockwaves (that are captured by the refined mesh),
and a sudden temperature drop of 130K (arc-heating is not yet considered). It is also observed
that maximum velocity occurs few millimeters downstream of the nozzle exit. This high velocity
region also accommodates local minima of pressure and temperature.
Figure 3.14 shows the fluid streamlines in YZ and XZ planes, which clearly demonstrate
that, without consideration of arc-heating, divergence of the atomizing gas stream is almost zero.
54
Figure 3.12 Contours of gas velocity (a) and pressure (b) for mass-flow-rate of 25.3 gr/s. Numbers
are in m/s and Pa, respectively. κ-ε turbulent model. Arc heating is not considered.
(a)
(b)
Velo
city
(m
/s)
Pres
sure
(Pa)
55
Figure 3.13 Contours of gas temperature (a) and mass-flux density (b) for mass-flow-rate of 25.3
gr/s. Numbers are in K and kg m-2s-1, respectively. κ-ε turbulent model. Arc heating is not considered.
(a)
(b)
Tem
pera
ture
(K)
Mas
s-Fl
ux d
ensi
ty (k
g m
-2s-1
)
56
Figure 3.14 Fluid streamlines parallel to YZ (a) and XZ (b) planes in the vicinity of wire-tips.
Streamlines are coloured by Y-component (a) and X-component (b) of gas velocity. Mass-flow-rate
of gas is 25.3 gr/s. Arc heating is not considered.
(a)
(b)
V x (
m/s
) V y
(m
/s)
57
Figure 3.15 Contours of shear stress on the wire tips. Numbers are in Pa. Mass-flow-rate of air is
12.3 gr/s. Arc heating is not considered.
Figure 3.15 illustrates the distribution of shear stress on the surface of the wire tip. It can
be observed that when arc heating is not considered, the shear stress (skin friction) on the wire-
tips has a maximum that is located on the side that is facing away from the other wire-tip, a
region that is not directly heated by the arc. This can be explained by the fact that the flow area
between the two wire-tips is much smaller that the opening between the wires and the nozzle.
This limits the amount of flow that passes between the wire tips, and therefore, decreases the
flow velocity in that region. Consequently, velocity gradients and shear stresses will be lower
than those in the rest of the nozzle cross section.
The second approach used to model turbulence in the gun was LES. As powerful
computers become more affordable, Large Eddy Simulation (LES) is emerging as a viable and
powerful alternative tool in turbulence computations. In recent years, LES has been applied to an
increasing number of problems in the field of engineering. This is made possible through the use
of parallel computing over under-utilized distributed machines in an industry setting, and the
availability of relatively cheap processors. The challenge in carrying out LES is that a three-
58
dimensional, unsteady calculation must be carried out on a grid capable of resolving the larger
scales of the motion. For flow geometries and Reynolds numbers of engineering interest, this
implies that the mesh-file is usually large. Hence, the CPU time required to perform the
calculation is substantially larger than that for an analogous RANS calculation.
Unlike κ-ε, implementation of LES turbulence modeling technique in FLUENT has not
been extensively tested by independent researchers. Therefore, a benchmark case, based on [ 21]
was created, meshed and solved in FLUENT. The case involved a rectangular channel, open on
two ends. FLUENT satisfactorily predicted the velocity profile in the channel, and was therefore
used in evaluating the shear stresses on the wire-tips of the ValuArc 200 gun.
To be able to resolve relatively large scales of motion, the full geometry of the gun was
meshed with a finer grid, and LES equations were solved in the domain. Pressure-inlet and
pressure-outlet boundary conditions were assigned to the inlet and outlet of the domain. Time-
averaged flow properties were found, and also time-variations of pressure, velocity, and shear-
stress at a few points in the domain were tracked. The FFT spectrum of the time-variations
revealed a few strong peaks between 40 kHz (for 201 kPa pressure setting) and 88 kHz (for 351
kPa pressure setting). It should be noted that an estimate of the frequency of vortex shedding can
also be found using the Strouhal-Reynolds relation: ( )0.198 1 19.7 /St Re= − [ 15]; the main
frequency of vortex shedding of a flow of 250 m/s past a cylinder of diameter 1.6 mm is
estimated to be 31 kHz, which is of the same order of magnitude as that predicted by LES
modeling.
It can be concluded that the frequency of vortex shedding (about 40 kHz) is about 20
times the frequency of particle detachment from the wire-tips (about 2 kHz). This means that a
molten droplet will experience about twenty oscillations of shear stress; therefore, molten
particle atomization can be modeled using a time-average turbulent model (such as κ-ε).
59
To verify that the numerical calculations correctly solve the fluid flow in the experiments,
the total pressure drop in the gun+hose system was numerically calculated and experimentally
measured for different pressure settings. The experimental and numerical findings are plotted
and compared in Figure 3.16. It can be observed that the numerical model provides an acceptable
estimate of the total pressure drop.
Figure 3.16 Numerical predictions of volumetric-flow-rate of air as a function of atomizing gas
pressure compare relatively well with experimental measurements. Experiments correspond to
pressure settings of 20, 30, 40, 50, and 60 psig. Numerical results of κ-ε model under-predict the flow
rate by about 8%. Dash-line takes into account the pressure drop in the connecting hose. LES data
points and their error-bars represent time-averaged and RMS values of time-dependent flow-rate.
Furthermore, shear stresses were determined on the surface of the wire-tips. The results
are shown in Figure 3.17 for different mass-flow-rates of air. Increasing the mass-flow-rate of
atomizing gas (air) increases the shear stress on the wire-tip surfaces, and without consideration
of arc heating, the maximum shear stress on the wire-tip surface occurs at a point that will not be
heated by the arc. This reduces the effectiveness of the current design of the wire-arc gun in
atomizing the molten material off the wire-tips.
60
(a) 0.008 kg/s (b) 0.012 kg/s
(c) 0.018 kg/s (d) 0.025 kg/s
(e) 0.032 kg/s (f) 0.038 kg/s
Figure 3.17 Shear stress on the surface of the wire-tips for different mass-flow-rates of air. Arc
heating is not considered. LES turbulence modeling.
61
3.2.2 Simplified Arc Solution
To incorporate the effect of arc heating in the fluid flow, the arc was solved theoretically
with some simplifying assumptions, and based on the experimentally observed shape of the arc.
About five pictures of the arc were taken at different system settings (P= 30, 40, 45, and
60 psig) using a visible-wavelength Nikon E3 camera with different optical filters. These digital
pictures, as shown in Figure 3.18 and Figure 3.19, were processed in MATLAB software (The
MathWorks Inc., Natick, MA): the radiation intensity at all points were translated to grayscale
level (a number between 0 and 255), and the grayscale levels were stratified to identify the
region of maximum intensity. Then, the arc-shape information (including the length and bending
angle of the centerline of the arc) were extracted from the processed pictures. It should be noted
that these pictures were taken with different shutter speeds to optimally use the sensitivity range
of the camera pixels (sensors). Therefore, maximum grayscale intensities will not indicate
maximum radiation level.
The presented pictures of arc (Figure 3.18 and Figure 3.19), clearly illustrate that the arc
is deflected in the downstream direction (due to convective effects of the cross-flow air). Also, it
can be observed that the bending angle and length of the arc increase with increasing atomizing
gas pressure. The bending angle and length of the arc were measured from the processed
pictures, and then averaged at each pressure setting (Figure 3.20). The average arc length and
shape, along with a simplified 2-D arc solution were used to model the arc-heating in FLUENT.
62
(a)
(b)
Figure 3.18 Grayscale picture of arc, taken at P = 30 psig (a) and P = 40 psig (b), wfr = 7 m/min, V =
30.1 V, with aluminum wires, and air as atomizing gas. Radiation intensity is translated to grayscale
intensity (a number between 0 and 255). The region of higher radiation intensity is then found by
stratifying the picture. Shutter speed setting: 750 (a) and 500 (b).
63
(a)
(b)
Figure 3.19 Grayscale picture of arc, taken at P = 45 psig (a) and P = 60 psig (b), wfr = 7 m/min, V =
30.1 V, with aluminum wires, and air as atomizing gas. Radiation intensity is translated to grayscale
intensity (a number between 0 and 255). The region of higher radiation intensity is then found by
stratifying the picture. Shutter speed setting: 500 for both (a) and (b).
64
0
1
2
3
4
5
6
20 30 40 50 60Pressure (psi)
Ave
rage
Arc
Len
gth
(mm
)
Figure 3.20 At each pressure setting, arc length from different images was measured and averaged.
The error bars represent the standard deviation of the measurements.
In order to obtain a simplified solution to the arc in cross-flow, a simple theory of free-
burning high-current arcs, developed by Lowke [ 38], was used and the following simplifying
assumptions were made:
• The arc was considered two dimensional: although the arc is deflected due to
convective effects of the cross-flow, the arc was considered symmetrical about its
maximum-temperature centerline. The arc was solved as if it was stretched along a
straight line.
• Arc radius (radius of the current carrying column) was assumed to be the same as the
radius of the radiating column (this radius can be measured from the arc images).
• The arc plasma was assumed to be in local thermodynamic equilibrium (LTE).
• Transport properties of air at atmospheric conditions were used [ 12] and gas
compressibility effects were neglected.
Lowke’s method [ 38] solves for the centerline temperature, electric field, plasma
velocity, and voltage as functions of location along the arc axis, while considering that
temperature is constant in the radial direction within the arc. It also assumes that pressure and
65
axial plasma velocity profiles are parabolic in the radial direction. An approximate arc solution
can be obtained using the resultant simplified electromagnetic and Navier-Stokes equations.
However, due to strong energy convection effects of the atomizing gas, the simplified energy
equation will not be a proper approximation of the physics of the arc in cross-flow. Therefore,
only the simplified electromagnetic equations were used. Such equations were coded and solved
in MATLAB for different current settings and arc lengths (measured from the processed arc
images). Results are presented in Figure 3.21 for a current of 200 A and an arc length of 4 mm.
Figure 3.21 Arc radius, current density, and electric field as functions of axial distance in a 4-mm
long arc with current of 200A.
66
3.2.3 Arc Heating in a cross flow
The simplified arc model that was introduced in the previous section provides an
approximate solution to current density and electric filed within the arc region. It is therefore
possible to estimate the “joule heating” term ( .j Err
) in the general energy equation.
To estimate the arc-heating term in the arc region, it was assumed that the arc centerline
is aligned with the curved centerline obtained from images (local maximum intensity points on
the arc pictures). It was also assumed that at any point on the centerline, the arc has a circular
cross-section on a plane that is normal to the centerline at that point. Such a simplified arc shape
allows determination of the local rate of heating at any point in the arc region.
FLUENT was again used to solve the fluid flow equations in the wire-arc-gun geometry.
This time, however, an additional volume source of energy, corresponding to the local rate of
heating, was added to the energy equation by means of a User Defined Function (UDF) script. A
UDF is a C-language routine (programmed by the user) that can be invoked at each iteration of
the FLUENT solver. At each iteration of the energy-equation, FLUENT executed a UDF that
determined the heat source terms: the amount of energy generated by joule heating, and also the
energy loss due to radiation.
The new model of fluid flow in the wire-arc-gun, with these heat sources, was run and
solved in FLUENT. The resultant velocity, pressure, temperature, and mass-flux density
contours are shown in Figure 3.22 and Figure 3.23. It can be observed that due to the increased
temperature, the supersonic shockwave effects have subsided. This is mainly because the speed
of sound increases with the square root of temperature, and therefore Mach number decreases
substantially when arc-heating is considered.
67
In addition, as shown in Figure 3.24, the fluid streamlines in YZ and XZ planes are
significantly different than those obtained without the effect of arc-heating (Figure 3.14). Arc
heating causes the atomizing gas stream to diverge by an angle of about 15 degrees in YZ plane
and 20 degrees in XZ plane. This divergence is expected to affect the initial velocity of the
produced particles.
Furthermore, the resultant shear stresses on the surface of the wire-tips were found at
different pressure settings. Figure 3.25 shows the distribution of the shear stress on the wires
with the consideration of arc heating. It can be observed that inclusion of arc-heating increases
the maximum value of shear-stress on the wire-tip (compare with Figure 3.17). In addition, the
point of maximum shear-stress is no longer on the outside face of the wire-tip: it is located
between the wires, and moves forward with increasing atomizing gas pressure.
68
Figure 3.22 Contours of gas velocity (a) and pressure (b) for mass-flow-rate of 25.3 gr/s. Numbers
are in m/s and Pa, respectively. κ-ε turbulent model. Arc heating is considered.
(b)
(a)
Velo
city
(m
/s)
Pres
sure
(Pa)
69
Figure 3.23 Contours of gas temperature (a) and mass-flux density (b) for mass-flow-rate of 25.3
gr/s. Numbers are in K and kg m-2s-1, respectively. κ-ε turbulent model. Arc heating is considered.
(b)
(a)
Tem
pera
ture
(K)
Mas
s-Fl
ux d
ensi
ty (k
g m
-2s-1
)
70
Figure 3.24 Fluid streamlines parallel to YZ (a) and XZ (b) planes in the vicinity of wire-tips.
Streamlines are coloured by Y-component (a) and X-component (b) of gas velocity. Mass-flow-rate
of gas is 25.3 gr/s. Arc heating is considered. Divergence of gas flow is more than that in Figure
3.14, where arc heating was not considered.
(b)
V x (
m/s
) V y
(m
/s)
71
(a) 0.008 kg/s (b) 0.012 kg/s
(c) 0.018 kg/s (d) 0.025 kg/s
(e) 0.032 kg/s (f) 0.038 kg/s
Figure 3.25 Shear stress on the surface of the wire-tips for different mass-flow-rates of air, with the
consideration of arc heating. Turbulence was modeled using κ-ε.
72
3.3 Simplified Breakup Model
The surface of the wire-tips, where they are in contact with the arc, is heated by the arc
and therefore, a layer of molten material is formed at this contact region. This molten layer
experiences several forces:
• Shear stress by the atomizing gas, which was estimated in the previous section.
• Pressure drag (form drag): this force is due to existence of high pressure gradients in
the region.
• Magnetic pinch force ( j B×v uv
), which, for a particle of diameter d detaching from a
wire of diameter wd is estimated to be 4
2044 w
d Id
μπ
by Amson [ 4]. This force becomes
very small in the anode region, because anode cathodej j<v v
(size of arc-anode
attachment is much bigger than the size of arc-cathode attachment).
• Surface tension force
• Weight of the molten layer: this force is about two orders of magnitude smaller than
other forces and, therefore, is neglected.
Although, theoretically, it may be possible to use a multi-phase fluid code (such as
FLUENT) to track the surface of the molten layer and model the primary breakup from the wire-
tips, FLUENT was not used because of two reasons: poor performance, and high accumulated
errors. Numerical studies conducted in FLUENT were found to produce an unacceptable
prediction of interface motion in a two-phase fluid flow. Besides, interface tracking becomes
particularly difficult (and inaccurate) at high flow velocities.
73
In addition, even with a validated numerical code, the errors in shear stresses, magnetic
pinch forces, and pressure drag forces, are too high to produce an adequately accurate particle
shape and size during primary atomization process.
Therefore, it was decided to use the simplified primary-breakup model developed by
Amson [ 4], and used by Kelkar and Heberlein [ 34]. In this model, a force balance equation is
solved to determine the size of the detached material:
420
4 cos( ) 04 D
w
d I F dd
μ θ πσπ
+ − = (3-2)
where σ is the surface tension constant and θ is the angle between the aerodynamic drag force
( DF ) and the direction of current flow (normal to wire-tip’s surface). The aerodynamic drag
force, DF , can be estimated by multiplying the average shear stress on the wire-tip (in the arc-
electrode attachment region) by the surface area of a hemisphere of size d. This equation was
solved for droplets detaching from aluminum wires (dw = 1.6 mm), for a wire-arc spraying gun
operating at a current I = 202 A at different pressure settings. The results are summarized in
Table 3.1.
Table 3.1 Size of primary metal detachments from wire-tips, using the simplified model of
Amson [ 4].
Pressure Setting Droplet Diameter (μm) (primary atomization)
Cathode
Droplet Diameter (μm) (primary atomization)
anode 20 psig (239 kPa) 480 510 30 psig (308 kPa) 355 365 40 psig (377 kPa) 280 285 50 psig (446 kPa) 220 225 60 psig (515 kPa) 210 210
74
The estimated size of the primary droplets can be used to determine a Weber number
2 /gWe U dρ σ= , where gρ , U , and σ are gas density, relative gas-droplet velocity, and surface
tension of the liquid metal. The Weber number for the above particles varies between 20 and 50.
Metallic droplets with Weber numbers above a critical Weber number 13critWe ≈ , are
unstable and subject to secondary atomization [ 71]. Since the range of Weber numbers of the
primary droplets in the wire-arc spray is estimated to be more than critWe , all of the primary
droplets are subject to secondary atomizations.
Particle sizes after secondary atomization and their distribution functions are discussed in
the next chapter.
75
Chapter 4
Particle Transport: In-flight Particles
The previous chapter of this thesis considered the physical phenomena of arc heating and
particle breakup in a wire-arc spray gun. This chapter discusses the characteristics of these
detached particles as they are transported in the spray plume. The results presented in this
chapter have been summarized and published in the Journal of Thermal Spray Technology [ 55].
4.1 Background
The quality of thermal sprayed coatings is directly related to the properties of the in-flight
molten particles, namely, size, temperature and velocity [ 30, 52, 54, 67]. These properties are
interdependent because the diameter of a particle determines the magnitude of both heat transfer
and the drag force acting on it, and hence its temperature and velocity. In powder based spray
techniques, such as plasma spraying or HVOF, particle size is determined by the size distribution
of the powder fed into the gun. Wire arc spraying is different since no powder is used; rather, the
heating of the wire tips by the arc, and the detachment of molten metal droplets due to drag and
magnetic forces determines the shape and size of spray particles.
The heating of the anode and cathode in a wire-arc process, and the difference between
them, were extensively discussed in Chapter 3: The arc attaches to the anode over a larger area
than the cathode, and so heating is more localized at the cathode spot [ 28, 67]. At the tip of the
76
anode wire a large area is heated due to diffuse arc-anode attachment, melting a layer of metal
that is pushed off the end of the wire-tip by the atomizing gas, creating an “anode sheet”. At the
cathode, constricted arc attachment causes much more localized heating and melting. Also, since
the current passes through a smaller area the current density (j) at the cathode surface is much
higher, producing a large magnetic pinch force (the j B×v uv
force, where Buv
is the induced
magnetic field). Molten metal droplets ejected into the arc from the cathode due to both drag and
magnetic forces are observed to be smaller than those that detach from the anode. Using laser
strobe photography Hussary et al [ 28, 29] and Watanabe et al [ 68, 69] clearly illustrated the
differences between molten metal detachment at the tips of the anode and cathode wires.
Inhomogeneity in the microstructure of wire-arc coatings was also observed by Zhu et al [ 72].
They demonstrated that particles originating from anode and cathode are distributed in an
asymmetric way about the centerline of the wire-arc spray.
It is interesting to note that although several different researchers have observed and
discussed the bimodal size distribution of wire-arc spray particles [ 28, 34, 35], no attempt has
been made to quantitatively analyze the particle-size-distribution graph and observe its variation
with different operating parameters. This chapter addresses this issue and includes the following:
• Spatial variation of average particle properties (temperature, size, and velocity) are
presented and discussed.
• The dual peak feature of the size distribution of wire-arc particles is analyzed. A new
technique to separate the individual peaks of the size distribution of in-flight
particles (pertaining to anodic and cathodic particles) is presented. The effect of
process parameters on the size distribution is also presented.
77
• The temperature distribution within particles and heat production due to exothermic
oxidation are discussed for in-flight aluminum spray particles. A numerical model of
the heat transfer and temperature distribution within a particle is also presented.
• The drag force and acceleration of particles in the plume are also discussed.
4.2 Spatial Characteristics of the Spray
The DPV-2000 monitoring system (Tecnar Ltd., Montreal, QC, Canada) was used to
measure particle characteristics at different spatial locations during the time of traversing their
path from the wire-arc spray gun towards the substrate. DPV particle size measurements were
calibrated by spraying wire-arc particles into water, drying them and washing with acetone.
Figure 4.1 shows a typical sample of such particles. The size distribution of the powder collected
was measured using a particle size analyzer (MasterSizer S; Malvern Instruments, UK) with a
detection range from 0.05µm to 880µm, and then fitted to the DPV-2000 size distribution by
varying the assumed diameter coefficient of the DPV software, a factor that takes the effect of
material emissivity into account [ 53]. Particle size distributions were plotted as both frequency-
histogram and volumetric-histogram of particle diameter, assuming particles were spherical.
78
(a) (b)
Figure 4.1 Optical (a) and SEM (b) pictures of aluminum particles collected by spraying into water;
P=30 psig (308 kPa), V=32.1 V, wire-feed-rate=7 m/min
The wire-arc spraying system with a high-velocity air cap has a divergence angle of about
15o, giving a coated area of 50 mm × 50 mm when the substrate is placed 200 mm away from the
gun. The deposited coating is not a perfect circle; rather the spray pattern is in the shape of an
oval whose shorter radius lies in the plane of the two wires. For example, the minor and major
radii of the coated area are 25mm and 30mm respectively for a substrate at a stand-off distance
of 200 mm (with V=32.1 V, P = 30 psig (308 kPa), wire-feed-rate=7 m/min). As discussed in
Chapter 3, this is because the atomizing gas stream is diverted by the wires producing a larger
spray divergence along the x-axis than the y-axis (the coordinate system is shown in Figure
4.2a).
79
Figure 4.2 Velocity, diameter and Mass-flow-rate of the spray particles as a function of y and x,
with z = 50mm. Center of the spray is located at x = y = 0mm. The error-bars in the graphs represent
the standard deviation of 3 to 5 measurements.
80
Particle properties were measured at different positions in the spray to examine the
uniformity of the spray. Figure 4.2 shows the variation of particle velocity (Figure 4.2-a) and
diameter (Figure 4.2-b) at different x and y positions in the spray at a distance of 50 mm from the
spray nozzle. Results are shown for aluminum particles with the operating parameters of the gun
kept constant with a gas pressure of 30 psig (308 kPa), arc voltage of 32.1 V, and wire-feed-rate
of 7 m/min. The points in Figure 4.2 represent the average of 3 to 5 measurements of 7000 to
10000 particles in the spray and the error bars show the standard deviation of the values
recorded.
Particle velocity diminished with distance from the centerline of the spray (see Figure
4.2-a), following the gas velocity profile as discussed in Chapter 3 and in [ 11]. The average
particle diameter increased with increasing radial distance from the centerline of the spray
(Figure 4.2-b). This is because the larger/heavier particles that gain their initial momentum at a
region with diverging gas flow (Figure 3.24), tend to retain their momentum for a longer period
of time, and therefore are deposited farther from the centerline of the spray. Particle temperatures
(not plotted here) were practically constant in the x-y plane (about 2433 K) and any variations
were within the experimental error range. Particle temperature was much higher than the melting
point of aluminum, indicating that surface oxidation produced a highly exothermic reaction on
the surface of particles.
Mass flow-rate of particles ( m& ) at any axial location in the plume can be calculated as:
( )
3
6v dAverage
m nAverage v
π
ρ
⎛ ⎞⋅⎜ ⎟⎝ ⎠= ⋅ ⋅& & (4-1)
where d and v are diameter and speed of a particle, ρ is the density of the particles, and n& is the
number flow-rate (number of particles per second that pass through the field of view of the DPV-
81
2000 sensing head). Since the DPV-2000 system only records a certain fraction of particles
(depending on the settings and detection criteria) passing through its sensing volume, the number
flow-rate of particles that is measured is proportional to the actual value. Therefore, the
calculated mass-flow-rate will be a relative (or scaled) value. Figure 4.2c illustrates the relative
mass-flow-rate profile of the in-flight particles at a cross-section of the spray.
A measure of the spatial dispersion of particles is the full-width-half-maximum (FWHM)
of the mass-flow-rate curve, defined as the distance between points on the mass-flow-rate curve
at half the peak value. The full-width-half-maximum of particle mass-flow-rate in the y-direction
is about 28 mm and 35 mm in the x-direction (Figure 4.2-c), producing the elliptical deposit on
the substrate observed in experiments.
When calibrating the DPV-2000 the size of particles sprayed into water was matched to
those measured by the DPV along the axis of the spray. Radial variations of particle size, such as
shown in Figure 4.2a were ignored in this process. Neglecting particle size variations does not
create a large error because the particle density in the spray is concentrated along the spray axis
(see Figure 4.2-c). It was estimated that the error in total mass-flow-rate introduced by assuming
uniform particle size distribution was less than 4%.
The variation of particle properties along the axis of the spray gun was also investigated.
Velocity of particles decreased with distance form the nozzle, falling from 160 m/s at z = 70 mm,
to about 100 m/s at z = 200 mm. Particle size distribution and temperature showed no variation in
the z-direction. It is speculated that the heating of aluminum particles due to surface oxidation
offsets the cooling due to convection and radiation to the surroundings.
82
4.3 Bimodal Particle Size Distribution and Separation Technique
In the wire-arc spraying system, the arc attaches differently to the anode and cathode
[ 28, 31, 34, 35, 57, 67, 68, 69], so that the two wires do not melt in the same way. Photographs of
arcs have shown that cathode heating is confined to a small area (cathode spot) while the anode
is heated more uniformly, though in both cases the area of arc attachment is smaller than the
diameter of the wire [ 28, 29, 69]. Droplets detaching from the anode are therefore typically larger
than those from the cathode [ 28, 68], producing a dual-peak particle size distribution. In some
recent studies [ 28, 34], bimodal size-distributions have been reported when operating with low
atomizing gas pressures. However, at higher pressures the two peaks overlap so that it is not easy
to distinguish between them. Figure 4.3a shows the diameter frequency-histogram of aluminum
particles produced with atomizing gas pressure 60 psig (515 kPa), arc voltage 37.9 V and wire-
feed-rate=7 m/min. The same data is presented as a volumetric-histogram (volume fraction) in
Figure 4.3b. The y-axes in the size distributions are shown with arbitrary units since only a
fraction of all particles in the spray were recorded by the DPV-2000. Only one peak is obvious in
both cases. The technique used to separate cathodic and anodic particles is described below.
83
(a)
(b)
Figure 4.3 Frequency-distribution (a) and volumetric-distribution (b) histograms of measured
particle diameter are shown by grey histograms; P=60 psig (515 kPa), V=37.9V, and wire-feed-
rate=7m/min. The curve in (a) is a Log-Normal function (μ=56μm, σ=0.451) matching the maximum
and full-width-half-maximum of the measured distribution. The curve in (b) is the volumetric Log-
Normal function with same μ and σ as in (a) and scaled with the same scaling factor as the measured
volumetric-distribution. The black bar-histogram represents the difference between the measured
volumetric-distribution and the volumetric log-normal function.
84
4.3.1 Size Distribution of Anodic and Cathodic Particles
The size distribution of particles produced by atomization of liquid jets typically follows
a log-normal probability distribution function defined by [ 37]:
21 ln( ) ln( )( )21 1( )
2
DdNf D edD D
μσ
πσ
−−
= = (4-2)
where σ and μ are the geometric standard deviation and geometric mean drop size [ 37, 22]. This
curve fits best to the experimental data of Figure 4.3a with μ=56 μm and σ=0.451. However,
there are more than expected large particles in the experiments (see Figure 4.3a, d>100 μm). The
difference becomes more obvious when the volumetric log-normal probability distribution
function defined as:
21 ln( ) ln( )3 ( )2 21
1( ) ( )6 72
DdV Dvdf D f D D edD
μσπ
π σ
−−
−= = = (4-3)
is plotted on the experimental volumetric-distribution data in Figure 4.3b. In this figure the
volumetric log-normal function is produced with same μ and σ as in Figure 4.3a and is scaled in
the same way as the experimental data. Here the difference between the number of large particles
measured and those expected from typical atomization theory becomes obvious. The difference
(black bar-histograms in Figure 4.3b) is evidence of there being two sets of particle, produced by
the cathode and anode, respectively, which have different but overlapping size distributions.
Since particles in each of the two sets are produced by simple atomization process from one
electrode, their individual size distributions are expected to follow a log-normal function.
To confirm that particles produced by melting and atomization of each of the wire tips
follow a log-normal probability distribution function it was necessary to physically separate the
anodic and cathodic particles. For this purpose, Stainless Steel Metcoloy 2 and Metco Copper
wires were used as anode and cathode, respectively. After spraying into water, particles were
85
collected, dried, washed with acetone, and then separated using a magnet. Stainless steel and
copper are distinguishable by their color under a microscope and inspection showed that the
number of copper particles present in the stainless steel particles after separation was less than
1%. To avoid particle agglomeration (see Figure 4.4), stainless steel particles were placed in a
demagnetizer, (DEMAG, Nortronics, NY).
Figure 4.4 An optical picture of magnetically-agglomerated stainless-steel particles before being
demagnatized.
Size distribution of stainless steel particles was measured using a Particle Size
Analyzer™. Figure 4.5 shows the size distribution of anodic stainless steel particles. A log-
normal distribution curve fits the data well; discrepancies were less than the uncertainty of the
measuring instrument (represented by error bars in Figure 4.5). The experiment was repeated
with the polarity of wires switched so that the cathode was stainless steel. Cathodic stainless steel
particles, too, follow a log-normal distribution function.
86
Figure 4.5 A log-normal function fits well within the error-bars of the size-distribution of anodic
particles. Stainless steel and copper wires were used as anode and cathode, respectively. The error
bars represent the systematic error of the size measuring device.
4.3.2 Separation Technique
The size-distribution of wire-arc particles can be represented by superposing two log-
normal distribution functions. Since anode and cathode wires are fed into the spray gun at the
same rate in experiments the total mass of cathodic and anodic particles in any sample collection
are equal. Therefore, even though the volumetric-distribution curves of anodic and cathodic
particles are different, the area under the curves must be the same. Since cathodic particles are
smaller they must be more numerous than the larger anodic particles. A method for determining
the size distributions of anodic and cathodic particles produced by a wire-arc spray is
summarized in the following steps:
1. Plot the experimental volumetric-distribution of particles and determine the area (A0)
under the curve;
87
2. Fit a log-normal function, using the least square method, to the ascending portion of the
experimental frequency-distribution curve of particle diameter (from 0 to the most
frequent particle diameter - e.g., from 0 to 45 μm in Figure 4.3a). Since this fit represents
the frequency-distribution of cathodic particles, the area under its volumetric-distribution
curve must equal A0/2. We assume here that the anodic particles are larger and much
fewer in number and hence contribute little to the population of small particles.
3. Subtract the fitted cathodic distribution curve from the measured distribution to obtain the
distribution of anodic particles.
4. Fit a log-normal function through the calculated diameter frequency-distribution of
anodic particles using method of least-squares. The area under its volumetric-distribution
curve must also be A0/2.
5. Add the cathodic and anodic distributions and compare with the experimental size
distribution to evaluate errors.
4.3.3 Error Estimation
To estimate the errors associated with the experimental instrumentation for measuring
powder size distributions, we mixed two powders of known size distribution, measured the size
distribution of the mixture and then calculated the size distribution of one of the original
powders. Metco 54NS-1 powder (Al 99%, particle size: -75+45µm) was sieved to isolate batches
with diameter ranges of -53+45µm and -75+63µm. A particle size analyzer (MasterSizer S;
Malvern Instruments, UK) was used to measure the size distributions of both samples. Equal
weights of both powders were then mixed and the size distribution of the mixture determined.
Since the volume of the mixture was twice the volume of the samples, their size distributions
were scaled similar to steps 1 and 2 of the algorithm outlined above. By subtracting the size
88
distribution of the smaller particle sample from that of the mixture, the size distribution of the
larger powder was determined. The relative error (ε ) was calculated from
reconsty y dx
y dxε
−= ∫
∫ (4-4)
where y and reconsty are the experimental and calculated size distribution of the larger diameter
powder respectively. The relative error in reconstructing the size distribution was less than 4%.
To estimate the errors associated with the proposed technique for separating anodic and
cathodic particle size distributions, we reconstructed two peaks from the mathematical sum of
two known log-normal functions. Typical values of μ1=50μm and σ1=0.45 were assumed for the
cathodic log-normal particle size distribution function and μ2=120μm and σ2=0.45 for the anodic
distribution function. These curves and their summations are shown in Figure 4.6, both as
frequency-distribution (Figure 4.6a) and volumetric-distribution (Figure 4.6b) of particle
diameter. Following the procedure outlined above the anodic and cathodic particle diameter
distributions were reconstructed from the combined curve; the calculated size distributions are
also shown in Figure 4.6. The relative errors of both reconstructed functions were calculated to
be less than 0.5%. This error varies with the shape and peak-to-peak spacing of the anodic and
cathodic distribution curves. Table 4-1 lists five different anodic log-normal functions and the
error in reconstructing them while holding the cathodic particle size distribution constant. Errors
increase when the peaks are closer together and there is greater overlap of the two distribution
curves.
89
Table 4-1 Different log-normal functions and the relative error in reconstructing them from
their sum.
Cathodic log-normal function Anodic log-normal function μ1 (μm) σ1 μ2 (μm) σ2 Relative error
50 0.45 150 0.45 0.2% 50 0.45 120 0.45 0.4% 50 0.45 90 0.45 1.6% 50 0.45 80 0.45 3% 50 0.45 70 0.45 8%
90
(a)
(b)
Figure 4.6 The separation technique was applied to the addition of two known log-normal functions
(LN1: µ1=50µm, σ=0.45 and LN2: µ2=90µm, σ=0.45) to reconstruct the original functions. (a)
frequency-distribution (b) volumetric-distribution.
91
(a)
(b)
Figure 4.7 Two peaks in the measured diameter distribution were separated and presented in
frequency (a) and volumetric (b) forms. LN1 and LN2 represent log-normal distribution functions of
cathodic and anodic particles respectively. vLN1 and vLN2 are the volumetric representation of LN1
and LN2. Experimental particle size statistics was obtained by DPV-2000 system at a stand-off
distance of 50 mm, voltage of 32.1 V, wire-feed-rate=7 m/min, and P = 60 psig (515 kPa). These
distributions represent statistics of about 8000 aluminum particles.
92
4.3.4 Effect of Varying Wire-Arc Parameters
To investigate the effect of wire-arc operating parameters such as atomizing gas pressure,
wire-feed-rate and operating voltage on particle size distribution, a series of experiments was
done in which each of these was varied. Particle size distributions were measured using the
DPV-2000 and anodic and cathodic particles identified using the separation technique described
above. Figure 4.7 shows a typical result for a spray gun operated with gas pressure 60 psig (515
kPa), wire-feed-rate 7 m/min and arc voltage 32.1 V. Particle sizes are shown as both a
frequency-distribution (Figure 4.7a) and volumetric-distribution (Figure 4.7b) and calculated
anodic and cathodic particle size distributions are also shown. Similar experiments were done for
atomizing gas pressures ranging from 20 psig (239 kPa) to 60 psig (515 kPa), wire-feed-rates of
6 to 10 m/min and arc voltages of 25 to 40 V. Repeated experiments were done at each setting.
The results shown are the average of 4 measurements and error bars represent the standard
deviation.
Figure 4.8 shows the variation of both mean-diameter and mass-mean-diameter of both
anodic and cathodic particles with gas pressure. Mass-mean-diameter is defined as
( ) /i i iMMD m d m= ∑ ∑ where im and id are mass and diameter of thi particle, and the
summation is over all particles [ 28]. Anode particles were always significantly larger than
cathodic particles. Particle size increased as gas pressure was reduced. Drag forces exerted by the
gas are the main reason for atomization of molten material from the wire tips. Droplets of molten
metal are formed when the drag and magnetic forces that tear liquid off the wire tip exceed
surface tension forces attaching it to the wire. As gas pressure decreases so does its velocity and
therefore drag forces. Molten metal droplets grow larger before detaching from the wire tip when
drag forces diminish. Moreover, secondary atomization that breaks the molten material into
smaller droplets tends to produce smaller particles with increasing gas flow velocity [ 71].
93
Figure 4.8 Mean-Diameter and Mass-Mean-Diameter of cathodic and anodic particles decrease as
the pressure of the atomizing gas increases. Anodic particles are more significantly affected by
atomizing gas pressure than the cathodic particles. Error-bars represent standard deviation of 3 to 5
measurements of about 8000 particles. Operating parameters: Aluminum wires, V=32.1V, wire-feed-
rate=7m/min, stand-off distance=50mm.
The primary dimensionless parameter on which secondary atomization depends is the gas
Weber number defined as 2 /gWe U Dρ σ= Δ , where gρ , U , and σ are gas density, relative gas-
droplet velocity, and surface tension of the liquid metal. For all metallic melts, there exists a
critical Weber number ( 13critWe ≈ ) below which droplets remain stable [ 71]. Weber number in
the nozzle region of the spraying gun, where the primary breakups occur, is estimated to be 70
(assuming typical values for gas velocity and spherical drop diameter as in [ 34]). Thus the
metallic detachments are unstable and subject to further disintegration. However, five
centimeters downstream from the nozzle exit, the Weber number dramatically drops to an
approximate low value of 0.5, below the critical value. Therefore, the effect of secondary
disintegrations can be neglected in the spray region. This is confirmed by the experimental
94
measurements of the average size of particles that showed no significant variation along the
spray plume.
Since the size of droplets resulting from a secondary breakup is proportional to the size of
the original particle [ 71], and since the primary breakups from anode and cathode are different in
size [ 34], the size of anodic and cathodic droplets after the secondary atomization should be
distinct.
Figure 4.9 Mean-Diameter and Mass-Mean-Diameter of cathodic and anodic particles as a function
of the wire-feed-rate. Error-bars represent standard deviation of 3 to 5 measurements of about 8000
particles. Operating parameters: Aluminum wires, P = 60 psig (515 kPa), V = 32.1V, stand-off
distance = 50 mm.
Figure 4.9 shows the variation in particle size with wire-feed-rate. There was a small
increase in particle diameter as feed rate increases, though it was so small that it was difficult to
see the effect on mass mean diameter. Increasing the wire-feed-rate shortens the arcing distance,
which results in an increase in the current that passes through the arc, and therefore the heat flux
to the electrodes. In the case of the cathode, where heating is localized at the cathode spot, an
increase in heat flux results in faster detachment of same size droplets from the wire material that
95
is fed at a higher rate. In the anode region, where heating is spread over a larger area, the
increased heat flux results in a thicker layer of molten material, which is pushed away by the
drag force of the gas flow. The drag force, however, does not increase with the thickness of the
anode sheets and therefore slightly larger particles are produced in the anode.
Varying operating voltage, too, had only a small effect on particle size. Figure 4.10
shows the variation of mean particle diameter with voltage. A mid-value voltage, at about 32V,
appeared to maximize particle size. A full explanation of this behavior will likely require a
detailed analysis of changes in heat transfer and magnetic forces at the wire tip caused by
varying voltage.
Figure 4.10 Mean-Diameter and Mass-Mean-Diameter of cathodic and anodic particles as a function
of the applied voltage. Error-bars represent standard deviation of 3 to 5 measurements of about 8000
particles. Operating parameters: Aluminum wires, P = 60 psig (515 kPa), wire-feed-rate = 7 m/min,
stand-off distance = 50mm.
96
4.4 Axial variation of particle properties
As discussed earlier in this chapter, aluminum particle properties (temperature, velocity
and size) produced by the ValuArc200 twin-wire-arc spray were measured at different stand-off
distances from the gun. The experimental results showed that while the velocity of wire-arc
aluminum particles diminishes along the axis of particle spray, particle temperature does not
noticeably change along the axis of particle spray. The behavior exhibited by the velocity profile
(Figure 4.11) is explained in accordance with the velocity profile of the atomizing gas. The
behavior exhibited by the temperature profile (Figure 4.12) is explained by the exothermic
oxidation of the aluminum surface.
y = 135.03e-3.6867x
0
20
40
60
80
100
120
140
0 0.05 0.1 0.15 0.2 0.25Distance (m)
Parti
cle
Velo
city
(m/s
)
Figure 4.11 Axial velocity profile of particles in the spray.
97
2200
2250
2300
2350
2400
2450
0 0.05 0.1 0.15 0.2 0.25Distance (m)
Tem
pera
ture
(K)
Sprayed with AirSprayed with Nitrogen
Figure 4.12 Axial temperature profile of particles in the spray with air and nitrogen as atomizing gas.
Temperature of aluminum particles sprayed with nitrogen increases by about 130°C as they travel a
distance of 20 cm.
4.4.1 Drag Force and Force Balance Relation
Upon primary atomization of molten material from the wire-tips, the molten detachments
are exposed to transonic flow of the atomization gas. This gas flow produces a positive drag
force on the particles and accelerates them towards the substrate. A few centimeters downstream,
the gas velocity reaches its maximum and then decelerates due to its expanded flow area (see
Figure 3.10). This causes the relative gas-particle velocity to become negative and therefore the
spray particles decelerate. This deceleration is experimentally measured and presented in Figure
4.11.
4.4.2 Heat Transfer and Exothermic Oxidation of Particles
In order to model the heat transfer processes during and after primary particle breakup in
the wire-arc spray gun, it is necessary to have an estimate of the timescales involved in such
processes. It is also useful to distinguish between dominant and negligible terms in the heat
balance equation. This section qualitatively describes the variation of particle temperature based
on some theoretical calculations and experimental results.
98
As shown in Figure 4.12, the surface temperature of aluminum particles, when sprayed
with air, is about 2400K, which is below the boiling point of aluminum, and slightly above the
melting point of aluminum oxide. Therefore, due to rapid oxidation of aluminum, a typical in-
flight aluminum particle will have a liquid shell of aluminum oxide containing a mixture of
molten aluminum and molten aluminum oxide.
Formation of the aluminum oxide layer is a very rapid process. The thickness of the oxide
layer has a parabolic behavior with time: Depth of the oxide layer can be estimated by δ=(Dt)1/2,
where D is the effective diffusion coefficient. During the typical particle-travel-time (from the
gun to the substrate), which is about 2ms, the depth of the oxide level can reach a few
nanometers, at room temperature. Diffusion rate at elevated temperatures is probably much
higher (no reference was found for this data), and therefore, the oxide layer on the in-flight wire-
arc particle is likely much thicker. In addition, the motion of molten aluminum inside the particle
can mix the aluminum and aluminum oxide, which increases the oxide content of the particle
even further.
Oxidation of aluminum is an exothermic process and produces 399kcal of heat per mole.
The released energy can be calculated by estimating the number of aluminum oxide molecules
and the heat capacity of the particle:
2 3
2 /Al OH G d Mρ π δ= Δ (4-5)
where GΔ , 2 3Al Oρ , d, δ, and M are molar heat of formation, density of aluminum oxide, diameter
of in-flight particle (about 50μm), thickness of oxide layer (considered to be 10nm), and the
molar mass of aluminum oxide (102 gr/mol), respectively.
99
A heat balance equation can then be used to relate the released energy to the temperature
increase experienced by the particle (assuming that this energy will completely dissipate within
the particle):
ΔT = H / (cp Alρ πd3/6) (4-6)
where cp and Alρ are specific heat and density of aluminum. For a typical aluminum particle, this
temperature increase can be expressed as:
11 2 8
5
1.7 10 4.2 101.16 10 / 23
H d JT d K
δ
δ
−= × ≅ ×
Δ = × ≅ (4-7)
This ΔT likely underestimates the actual increase in temperature and can be orders of
magnitude higher if the diffusion rate at the high temperatures (@ 2400K) and the mixing of
aluminum and aluminum oxide were considered.
To further investigate the effect oxidation has on the temperature of a particle, the
atomizing gas (usually dry air) was replaced with nitrogen. If exothermic oxidation counteracts
the convective cooling effects of the surrounding gas, eliminating oxidation should reduce the
particle temperature. Figure 4.12 shows the experimental measurements of particle temperature
at different axial locations for two cases of air and nitrogen as atomizing gas. It can be observed
that for the case of nitrogen (as the atomizing gas), the temperature of aluminum particles
immediately downstream of the gun is about 150 K lower than when dry-air is used. As the
particles move towards the substrate, air is entrained in the particle+nitrogen plume, and
oxidation starts to occur. This is more evident from the lateral variation of particle temperature
shown in Figure 4.13; Off-center nitrogen-sprayed particles in the plume are more exposed to the
oxygen molecules in the ambient air, and therefore, oxidize more (and will have a higher
temperature than the centerline particles). It is interesting to note that the average temperature of
100
particles sprayed with air and nitrogen both reach a value of 2350 K in the outer layer of the
spray.
2200
2250
2300
2350
2400
2450
-30 -20 -10 0 10 20 30Distance from centerline of the spray (mm)
Tem
pera
ture
(K)
Sprayed with AirSprayed with Nitrogen
Figure 4.13 Average temperature of aluminum particles in the spray as a function of lateral distance
from the centerline of the spray. Axial distance from the gun is 3" (76 mm). V=31V, wfr = 7 m/min,
P=30 psig (308 kPa).
Relaxation Times for Heat Transfer: Since formation of the protective aluminum-oxide
layer is a rapid process (about 1ns at room temperature, and less than 1ns at elevated
temperatures), oxidation can be assumed as an instantaneous process (compared to the typical in-
flight time of about 2 ms). During and after the oxidation process, the produced heat will be
conducted to the center of the particle in a very short time, causing it to reach an equilibrium
temperature. The characteristic time during which the generated heat is dissipated through the
spherical particle can be estimated using temperature-response-time [ 15]:
2
12pc dk
ρτ = (4-8)
where τ , k , pc , and d are characteristic temperature-response-time, thermal conductivity, heat
capacity, and diameter of the particle, respectively. This characteristic time for aluminum
particles is about 2μs which is much less than the in-flight time period. This characteristic time
101
would be the same for any other process that changes the temperature on the surface of the
particle. For example, heat conduction from the surface will cool down the whole particle in a
time interval that is much shorter than the in-flight time.
Detached Aluminum Particles: The above discussion provides us with a clearer view of
in-flight particles. Roughly speaking, every 1 ms, the atomizing gas strips molten material off the
wire tips. In just a couple of nanoseconds, when the molten material (an undetached drop) is still
in the process of detachment, its outer layer will be oxidized. In addition to the heat gained from
the arc, the exothermic process of oxidation will also contribute to the heating of the outer layer
of the particle. Shortly after detachment (in few microseconds), the particle reaches thermal
equilibrium: convective heat transfer and radiation losses cool down the particle, while the
exothermic oxidation counteracts these effects and maintains the temperature of particles.
102
Chapter 5
Particle Deposition: Splat and coating formation
Previous chapters of this thesis considered heating, melting, atomization and transport of
wire-arc particles. This chapter discusses how these particles are deposited on the substrate. The
results presented in this chapter have been summarized and published in Journal of Surface and
Coating Technologies [ 1]
Generally, a thermal spray particle is partially or fully melted before deposition. Upon
touching the substrate, it spreads on the substrate and starts to cool down and solidify. This
solidified splat (or deposit) and the ones that will be deposited on top of it, will form the building
blocks of the produced coating layer. The quality of the coating depends on the shape, size and
material of these splats and also on how strong they are attached together and to the substrate.
Shape and size of the solid deposits depend on substrate condition and also on in-flight particle
properties, which, in turn, depend on operating parameters of the spray system. Effects of
substrate temperature and particle velocity on shape of the deposited splat are discussed first, and
their effects on coating quality follow.
5.1 Effect of Substrate Temperature on Splat Formation
Increasing substrate temperature is shown to have a strong effect on the properties of the
thermal spray coating that is applied on it: Pershin et al [ 50] plasma sprayed nickel powder onto
103
a stainless steel plate and found that coating adhesion strength increased by almost an order of
magnitude as surface temperature was raised from room temperature to 650°C. Several
explanations were proffered: heating the surface clears volatile contaminants adsorbed on the
surface, improving contact between impinging particles and the substrate; reducing the cooling
rate and solidification time of droplets allows them to flow into surface cavities before freezing,
enhancing mechanical bonding. The most visible effect of increasing substrate temperature,
though, was to change the shape of splats formed by solidified droplets after impact on the
surface.
The effect of substrate temperature on splat shape has been well established in a number
of studies, reviewed in detail by Fauchais et al [ 19]. Typically, a thermal spray particle landing
on an unheated surface will splash, forming a fragmented splat with irregular edges. If surface
temperature is increased splat shape changes, so that impacting droplets form circular discs with
no evidence of splashing. Fukumoto et al [ 20] did a statistical analysis of splat shapes deposited
on a surface and defined a “transition temperature” (Tt) as the substrate temperature where equal
numbers of splat and disk splats were visible on the surface. Surface temperature, though, is not
the only factor affecting splat shape. Jiang et al [ 32] plasma sprayed ZrO2 particles onto polished
stainless steel coupons and found that the presence of condensates/adsorbates on the substrate
enhanced splashing; removing adsorbed volatile compounds on the surface reduced splashing.
Computer simulations of impacting molten metal droplets [ 43, 47] provide insight into a
mechanism for solidification induced splashing. A spreading drop begins to freeze along its
edges, where it first contacts the colder substrate. The solid rim formed obstructs further flow,
forcing liquid to jet off the surface so that it becomes unstable and breaks up into satellite
droplets. The rate of solidification depends on both properties and temperatures of the substrate
and droplet, and also the thermal contact resistance between the two (which varies with surface
104
roughness and the presence of any surface contaminants). The transition temperature is therefore
a complex function of all these parameters.
Dhiman and Chandra [ 16, 17] proposed a simple criterion to determine if solidification
induced splashing will not occur: a droplet will form a disk splat if the solid layer in the droplet
does not grow as thick as the splat, in the time the droplet takes to spread. A one-dimensional
heat conduction analysis was used to develop a model to predict the transition temperature.
Most studies on the effect of surface temperature on the properties of single splats and
coatings have investigated plasma spraying [ 10, 44], and little work has been done on splats
generated by wire-arc spraying. This section of the thesis focuses on the morphology of the
splats formed by wire-arc spraying aluminum droplets onto stainless-steel substrate.
5.1.1 Experimental Procedure
ValuArc 200 wire-arc spray system (Sulzer-Metco, Westbury, NY, USA) with High-
Velocity cap and air as its atomizing gas was used to spray aluminum (Metco Al wires) onto
polished stainless steel substrates for the purpose of determining the effects in-flight particle
properties have on the splat formation
The experimental results presented in Chapter 4 showed how in-flight particle properties
(velocity, temperature, and diameter) and their distributions are affected by operating parameters
of ValuArc 200 wire-arc spray system: Although average particle velocity inhibits a significant
change when operating parameters are altered, average temperature and diameter inhibit only a
slight change. Experiments also showed that particle velocity is strongly dependent on the
atomizing gas pressure (and not voltage or wfr). Therefore, arc voltage and wire-feed-rate were
kept constant in the experiments presented in this section of the thesis (for the purpose of finding
the effects of particle properties on the splat formation). Voltage and wire-feed-rate were held at
105
32.1 V and 7 m/min, respectively, while atomizing gas pressures of 30, 45, and 60 psig were
used. Table 5.1 summarizes particle diagnostics in the spray at these three pressure settings.
Since increasing pressure of the atomizing gas increases particle velocity, these settings are
hereafter referred to as low-velocity, mid-velocity, and high-velocity settings, respectively.
Table 5.1 Wire-Arc Operating Parameters and Particle Properties
High velocity
setting (V=143 m/s)
Mid velocity setting
(V=131 m/s)
Low velocity setting
(V=109 m/s)
Voltage (V) 32.1 32.1 32.1
wire-feed-rate (m/min) 7 7 7
Operating Gas Pressure (psig , kPa) 60 , 515 45 , 411 30 , 308
Gas Flow Rate (scfm) 64 53 40
Arc Current (A) 202 205 212
Mean Velocity (m/s) Standard deviation (m/s)
143 ± 3 36
131 ± 3 36
109 ± 2 28
Mean Temperature (oC) Standard deviation (oC)
2132 ± 10 131
2135 ± 12 133
2140 ± 7 120
Mean Diameter (μm) Standard deviation (μm)
71 ± 3 42
70 ± 3 40
71 ± 3 40
Substrates were 16-gauge AISI 304L stainless steel coupons, 75 mm × 40 mm × 2.5 mm
in size. These coupons were polished to a mirror finish (Ra = 0.01-0.04 μm), using progressively
finer grades of emery paper and finishing with alumina suspension applied with a soft cloth.
As shown schematically in Figure 5.1, the prepared substrates were mounted 200 mm
from the spray gun exit on a heater block whose temperature could be controlled with an
accuracy of ±5°C. A steel restriction plate with a 1.6 mm opening was placed approximately 125
mm downstream from the gun, the hole aligned with the centerline of the spray. A second
blocker plate was used to close the opening while the gun was started until the spray process
106
stabilized. It was then removed very briefly (for about 1 to 2 seconds), to allow a few particles to
pass through the hole, land on the substrate and form distinct splats. Removing the
restriction/blocker plates for 3 to 5 seconds allows a coating layer with 200 μm to 400 μm
thickness to build up.
Figure 5.1 Experimental setup to obtain distinct splats on the substrate.
DPV-2000 diagnostics system (described in 2.1.1) was used to measure size, velocity,
and temperature distribution of in-flight particles. Although DPV-2000 measures the properties
of individual particles, it was not possible to relate each splat with its corresponding in-flight
particle. (Obtaining such information requires a more sophisticated setup similar to [ 41]).
Instead, a statistical approach was taken: mean and mode diameter of in-flight particles were
compared to mean and mode diameter of splats.
107
Optical analysis of splats and coating cross-sections was done using Clemex Vision Pro
image analysis software (Clemex Technologies Inc., Longueuil, QC, Canada). The area (Asplat)
and the length of the periphery (Psplat) of each splat were determined using the software. An
equivalent splat diameter (Dmax) was then calculated using the relation:
Dmax =4Asplat
π (5-1)
For irregular splats a “degree of splashing” (DS) was also calculated, as defined by
Sampath et al [ 58]
DS =Psplat
2
4πAsplat
(5-2)
For perfectly circular splats DS = 1 and the value increases as splat splashing increases.
Coating deposition efficiency was measured by recording the difference in coupon
weight before and after spraying, and dividing that by the weight of the wire sprayed. Adhesion
strength of the wire-arc sprayed aluminum coatings onto stainless steel substrates have been
previously measured by Abedini [ 2]: He used the procedure recommended by ASTM standard
C633-01 (Standard Test Method for Adhesion or Cohesion Strength of Thermal Spray Coatings).
Coatings were deposited on the flat ends of cylindrical coupons, 50 mm long and 24.7 mm in
diameter, held in a heater block. Adhesion strength was measured by attaching another identical
coupon to the coating using epoxy, after which an Instron Model 1331 hydraulic tensile test
machine used to pull the coating off the substrate. Coating adhesion strength was defined as the
force required to detach the coating, divided by the cross-sectional area of the coupon.
X-ray Photoelectron Scanning (XPS) was used to determine the elemental composition of
the stainless steel substrates and to detect how much oxidation was caused by increasing surface
temperature. A PHI 5500 Multi-Technique System (Physical Electronics, Eden Prairie,
108
Minnesota, USA) was used to measure the fraction of oxygen, iron and chromium and other
elements present on substrates. Prior to XPS analysis the test specimens were argon-ion sputtered
to remove surface contaminants.
5.2 Splat Morphology
Figure 5.2 shows images of aluminum splats formed on stainless steel substrates held at
temperature (Ts) ranging from 25°C to 300°C, and corresponding cross-sections through coatings
formed at the same conditions. The two columns on the left show splats and coatings formed
with a higher atomizing gas pressure, where mean particle velocity was 143 m/s, while those on
the right were formed with a lower gas pressure and mean impact velocity of 109 m/s. At surface
temperatures below 100°C splats showed signs of having undergone extensive splashing, with
long fingers radiating out from a central core of solidified metal. Computer simulations of
droplet splashing [ 43, 47] have shown that solidification starts at the periphery of the spreading
droplet, creating a solid rim that forces the liquid to jet off the surface, where it becomes unstable
and breaks into fingers. Voids between fragments of the drop and in the central splat itself create
pores in the coating: cross-sections through coatings formed on surfaces at 25°C show large
voids and pores (see Figure 5.2). The voids were largest at the lower impact velocity (V=109
m/s), especially at the substrate-coating interface, and decreased when impact velocity increased
to 143 m/s.
109
Figure 5.2 Splat morphology and corresponding coating microstructure of wire-arc sprayed
aluminum deposited onto polished stainless steel (type AISI304L) held at various temperatures.
110
Increasing substrate temperature produced a change in splat shape. As Ts was increased
above 100°C splats became rounder and fingers became shorter until they disappeared almost
entirely. The change was progressive, but there was a sharp transition in the range
100°C<Ts<150°C for splats with mean velocity 143 m/s. At lower impact velocity (109 m/s) the
transition temperature was in a higher range 215°C<Ts<250°C. The number of voids in splats
decreased as substrate temperature was elevated (see Figure 5.2) and as a consequence the
density of pores in the coating also decreased.
Judging the transition temperature from photographs of individual splats is subjective,
since the decrease in splashing is gradual. A more reliable technique for determining transition
temperature, proposed by Fukomoto [ 20], is to photograph an area of the substrate with 20-30
splats on it and count the fraction of disk splats. The frequency of disk splats was counted for
three different substrates at each temperature and their average calculated. Figure 5.3 shows the
variation of disk splat frequency with substrate temperature, at three different impact velocities.
Less than 10% of droplets landing with average velocity 143 m/s formed disk splats at Ts=100°C,
increasing to more than 80% at Ts=200°C. The transition temperature, corresponding to a disk
splat frequency of 50%, was 140°C. When impact velocity was reduced to 131 m/s and 109 m/s,
the transition temperature increased to 200°C and 230°C, respectively.
111
Figure 5.3 Frequency of disk-shape splats increases with increasing substrate temperature. High-
velocity and low-velocity data are adapted from [ 1] and [ 2]. Mid-velocity data are measured solely
by the author of this thesis.
An alternate method of identifying the transition temperature was discussed by Abedini
et al [ 1, 2], in which they plotted the degree of splashing (DS) as a function of substrate
temperature (Figure 5.4). As splashing decreases so does DS approaching DS = 1 for perfectly
circular splats. DS values for some splats at different substrate temperatures are shown in Figure
5.2. Selecting a threshold of DS = 1.5 to 2, above which splats were noticeably distorted, gives
values of transition temperature close to those obtained by counting the frequency of disk splats.
However, selection of this threshold value is subjective, and therefore, this may not be a good
method of identifying the transition temperature.
112
Figure 5.4 Degree of splashing decreases with increasing substrate temperature. Adapted from [ 1].
5.3 Model for Transition Temperature
Dhiman and Chandra [ 16, 17] developed a simple model to predict the transition
temperature. The model assumes that a spherical droplet with diameter Do and velocity Vo lands
on the substrate and forms a disk shaped splat with diameter Dmax. The thickness of the splat, h,
is given by conservation of volume to be
h =2D0
3
3Dmax2 (5-3)
The time (tc) taken for the droplet to flatten onto the substrate is estimated to be [ 48]:
tc =8Do
3Vo
, (5-4)
To trigger splashing the solid layer at the edge of the spreading splat has to be thick
enough to obstruct flow [ 43, 47]. Dhiman and Chandra [ 16, 17] postulated that a droplet will
splash if the solidified layer thickness (s) grows to that of the splat itself (h) in time tc. Numerical
calculations by Mehdizadeh et al [ 43] have shown that the temperature gradients and therefore
heat transfer parallel to the surface are much smaller than those in the normal direction. Thus, the
113
solid layer thickness was calculated using a one-dimensional model. It was assumed that the drop
was at its melting point, that heat transfer was due to conduction alone, and that the temperature
drop across the solid layer was negligible. All non-equilibrium effects such as undercooling or
nucleation delay were neglected. The model accounted for the thermophysical properties of both
the droplet and substrate and the thermal contact resistance between the two. In dimensionless
form it predicted that the solid layer thickness grows as:
s∗(t∗) =s
D0
=2π
Ste ρsksCst∗
ρd kdCd Pe1−
1Bi
ρsksCsPeρd kdCdπt∗ ln 1+ Bi ρd kdCdπt∗
ρsksCsPe
⎡
⎣ ⎢ ⎢
⎤
⎦ ⎥ ⎥
⎧ ⎨ ⎪
⎩ ⎪
⎫ ⎬ ⎪
⎭ ⎪ (5-5)
where the dimensionless Peclet number (Pe=VoDoρdCd/kd), Stefan number (Ste=Cp(Tm-Ts.o)/Hf)
and Biot number (Bi=Do/(Rckd) ) have been introduced.
The maximum spread diameter of the splat can be estimated by a simple energy
conservation model [ 48] which equates the final surface energy of the splat to its initial kinetic
and surface energy, less the energy dissipated in droplet deformation and solidification. It
predicts that the droplet spread factor is
eRWeWes
WeD
D
a 4)cos1(383
12*0
maxmax
+−+
+==
θξ (5-6)
where the dimensionless Weber number (We=ρdVo2Do/σ) and Reynolds number (Re= ρdVoDo/μ)
have been used along with θa, the advancing contact angle of the liquid.
To test the accuracy of equation (5-6) we compared its predictions with measured values
of spread factors. Diameters of individual spray droplets were measured in-flight using the DPV-
2000 monitoring system, while the equivalent diameter of all splats on a coupon – defined in
equation (5-1) was measured using the image analysis. The most frequently observed splat
diameter (i.e., the mode diameter) was divided by the modal particle diameter to get a
representative spread factor at each value of substrate temperature. Figure 5.5 shows measured
114
values of spread factors, compared with predictions from equation (5-6). The contact angle for
aluminum drops on a steel surface is θa=170°C [ 3]. We assumed contact resistance Rc=0, so that
the term in parentheses in equation (5-5) equaled 1. Results in Figure 5.5 are shown only for
Ts>150°C: at lower substrate temperatures, below the transition temperature, there was too much
splashing and loss of mass (not accounted for in the model) to obtain meaningful splat diameters.
Equation (5-6) predicts disk splat diameters with reasonable accuracy.
Figure 5.5 Experimental and theoretical spread factor values for both high velocity (143 m/s) and
low velocity (109 m/s) tests. The curves are the theoretical predictions from equation (5-6).
The criterion postulated for formation of disk splats [ 16, 17] is that hs ≤ during the
spreading time, tc. The transition temperature, Tt , is the lowest substrate temperature at which
this will happen; for Ts>Tt the droplet spreads fully before the solid layer has grown sufficiently
to obstruct flow. Combining equations (5-3) to (5-6) we obtain an explicit expression for the
transition temperature:
115
Tt = Tm −Pe A H f
1−ln(1+ Bi A)
Bi A⎡
⎣ ⎢
⎤
⎦ ⎥
(1− cosθa ) +43
WeRe
2 We +16( )Cp
⎡
⎣
⎢ ⎢ ⎢ ⎢
⎤
⎦
⎥ ⎥ ⎥ ⎥
(5-7)
where
PeCkCkA
sss
ddd
38π
ρρ
= (5-8)
The transition temperature depends on the properties of the impinging droplet and the
substrate, impact velocity and contact resistance. Figure 5.6 shows the variation of transition
temperature with impact velocity for aluminum droplets impacting stainless steel surfaces.
Results are shown for four different values of Rc= 0, 10-7, 1.4x10-7, and 1.7x10-7 m2K/W, as well
as the three experimentally determined values of Tt. Transition temperature decreased
significantly with increasing contact resistance. For Rc=0 there was little effect of impact
velocity on transition temperature; at higher values of contact resistance transition temperature
decreased with impact velocity.
116
Figure 5.6 Prediction of transition temperature for aluminum droplets impacting a stainless steel
surface. The three experimental data points do not necessarily have similar contact resistances due to
the growth of an oxide layer on the substrate.
As impact velocity increases droplet spreading time and splat thickness both decrease. As
a consequence of shorter droplet spreading time the solid layer has to grow faster to obstruct
flow, resulting in a lower transition temperature. However, reduced splat thickness means that
the solid layer has to grow less to obstruct flow, which increases transition temperature. The
contact resistance determines the relative magnitude of these two competing effects. For Rc=0,
solidification progresses rapidly and changes in spreading time have little impact; for higher Rc,
solid layer growth is slow and therefore transition temperature decreases with increasing droplet
velocity.
The experimentally observed decrease in transition temperature, from 230°C at Vo=109
m/s to 140°C at Vo=143 m/s, was more than that predicted by the model if Rc was assumed the
same at all three temperatures. However, it seems quite likely that contact resistance changes
with surface temperature due to the growth of an oxide layer when steel plates are heated in air
[ 13]. Heating test coupons to 300ºC changed their color to a golden hue, which turned brown
117
under further heating. Figure 5.10 shows the surface composition of coupons heated in air to a
given temperature and then allowed to cool. The oxygen content increased from 35% of the total
at room temperature to over 60% at 350°C, indicating increased oxidation.
Figure 5.7 Plot of elemental composition of the stainless steel substrates heated to various
temperatures. Adapted from [ 1].
Prolonged heating in air can create a significant amount of oxide scale at the surface,
sufficient to increase not only thermal contact resistance but also surface roughness. A mirror-
polished substrate, initially with Ra=0.01 μm, was found to have surface roughness as high as
0.40 μm after being heated for 25 min at 350°C. Increased roughness also promotes splashing of
molten metal droplets [ 59]. Figure 10 shows images of splats collected on coupons at 350°C. The
first (Figure 10a) is a disk splat formed on a surface heated for 7 min, in which time the surface
roughness did not show significant change. In the second (Figure 10b), the surface was held at
the elevated temperature for over 20 min, increasing surface roughness significantly. Splats
collected on this surface showed clear evidence of splashing. The mechanism for this is different,
being caused not by solidification, but by fluid instabilities at the edges of the spreading liquid
droplet, which are promoted by surface roughness [ 42].
118
5.4 Coating Properties
Figure 5.8 and Table 5-3 show the porosity, determined using optical analysis, in coatings
produced at various substrate temperatures, for both impact velocities. All coatings had less than
5% porosity. Coatings produced at low substrate temperatures frequently showed voids at the
substrate-coating interface; both their number and size reduced significantly for Ts>100°C and
were no longer detectable for substrate temperatures of about 200°C. Conversely, it was
observed that increasing the substrate temperature above 200°C increased the porosity level.
Although both velocity conditions illustrated similar patterns, the presence of large experimental
errors in porosity measurements makes it impossible to draw definitive conclusions about
porosity behavior from this study.
Figure 5.8 Effect of substrate temperature on porosity of the produced coating.
119
Table 5-3 Microstructure Analysis
High velocity setting (V=143 m/s)
Low velocity setting (V=109 m/s) Substrate
Temperature (ºC) Porosity % Pore Size [μm] Porosity % Pore Size [μm]
25 2.4 ± 1.1 5.3 2.6 ± 1.1 4.8
100 1.8 ± 1.7 5.8 1.7 ± 0.8 3.0
150 - - 1.9 ± 1.2 3.3
200 1.1 ± 0.2 2.5 1.9 ± 1.7 4.1
250 1.9 ± 1.1 2.6 2.8 ± 1.0 2.8
300 2.3 ± 0.1 2.3 3.3 ± 0.8 4.2
Average 1.9 ± 0.8 3.7 2.4 ± 1.1 3.7 Measurements of the cross-sectional hardness (10 from each sample) were also taken for
each coating using a micro hardness tester (Zwick Hardness Tester, Zwick, Germany) applying
500 g-force. Neither substrate temperature nor spray impact velocity had a significant effect on
hardness. The average hardness (Vickers) for the 109 m/s coatings was 51.2 ± 4.2, while the
average hardness for the 143 m/s coatings was 52.1 ± 4.0.
Coating roughness was measured with a Surfometer (Precision Devices Inc., Milan,
Michigan, USA) that ran a stylus over the surface. An average of 10 measurements was made for
each sample. Again, substrate temperature and spray velocity did not have a measurable effect on
coating roughness, which was Ra=15±2 µm in all cases.
Coating deposition efficiency increased with substrate temperature. Figure 5.9 shows the
variation of deposition efficiency, determined by weighing the coupons before and after
spraying, with substrate temperature. Deposition efficiency was lowest (approximately 40%)
when the substrate was at room temperature and increased with substrate temperature, reaching a
maximum value of 52% at Ts=300ºC. Further increases in substrate temperature did not result in
greater deposition efficiency. It was also shown that within the velocity range of these
experiments, impact velocity had little effect on deposition efficiency.
120
Figure 5.9 Measured deposition efficiency for high velocity (143m/s) and low velocity (109m/s)
test conditions. Curves represent the best fit.
Figure 5.10 shows the variation of coating adhesion strength with substrate temperature.
Each data point on the figure shows the average of five measurements and error bars represent
their standard deviation. Average spray velocity in all tests was 143 m/s. For coatings produced
at room temperature the average coating strength measured was approximately 9.5 MPa,
increasing to 12.1 MPa at approximately 100ºC. Raising substrate temperature to almost 200ºC –
above the transition temperature – increased adhesion strength by 86% to an average value of
17.9 MPa. Further increases in substrate temperature led to a rise in adhesion strength, though
not by as great an amount.
121
Figure 5.10 Measured coating adhesion versus substrate temperature for particles having an average
velocity of 143m/s. [ 1, 2]
122
Chapter 6
Closure
This concluding chapter summarizes the major results, findings, and contributions of the
present work. Also, some ideas are suggested as to how this research topic can be explored in
more details.
6.1 Conclusions
In a wire-arc spray system (such as Sulzer-Metco’s ValuArc 200, which was used in this
study), particles are formed by atomization of molten metal from the tips of two consumable
wires between which an electric arc is struck. A cross-flow atomizing gas accelerates the
detached particles towards a substrate on which a protective coating is formed by deposition of
these particles.
In this study, arc voltage and arc current were experimentally measured at different
operating conditions. The measured data were analyzed to find the energy delivered to unit mass
of the fed material, and to estimate aluminum evaporation. Arc voltage fluctuations were also
looked at, and analyzed to obtain an estimate of the size of primary atomizations from the wires.
It is well known that the arc attaches to the anode over a much larger area that the
cathode and, consequentially, particles separating from the anode are larger than those from the
cathode. The sizes of these primary detachments were estimated using computational fluid
123
dynamics and simplified models of arc and particle-breakup. These simplified models were
based on the information obtained from pictures of the arc (taken with Nikon E3 digital camera)
and pictures of metal detachment (taken with a custom-made UV camera with Nitrogen laser
illumination).
Shortly after primary atomization, the detached particles break up into smaller particles
(undergo secondary atomizations). These in-flight particles are a mixture of cathodic and anodic
particles. Although mixed, an algorithm was presented to identify the size distributions of the
two sets of particles. The presented algorithm assumes that both anodic and cathodic particles
follow a log-normal distribution. This assumption was also validated by spraying magnetic and
non-magnetic materials (as anode and cathode), and separating the resultant particles.
The presented separation-algorithm provides a tool to study effects of operating
parameters on each of the anodic or cathodic set of particles. Experiments showed that increasing
the atomizing gas pressure decreased the size of both anodic and cathodic particles, while
changing wire-feed-rate and operating voltage did not change particle size significantly.
Axial variations (along the spray plume) of particle velocity and temperature were also
investigated. While aluminum particles decelerate as they move towards the substrate (as
expected), their temperature remains almost constant. This was explained by analyzing the
exothermic oxidation of the surface of aluminum particles. The presented explanation was also
verified by spraying aluminum with nitrogen as atomizing gas (to prevent oxidation).
Also, effects of substrate temperature and spray velocity on the properties of aluminum
splats and coatings deposited on mirror-polished steel surfaces were studied. As substrate
temperature was increased droplets no longer splashed, but formed disk shaped splats.
Aluminum particles sprayed with an average velocity of 109 m/s had a transition temperature of
230°C; particles sprayed with an average velocity of 131 m/s had a transition temperature of
124
200°C; particles sprayed with an average velocity of 143 m/s had a transition temperature of
140°C.
In addition, coating porosity levels were measured (less than 5% for all coatings
produced). Raising substrate temperature reduced the size and density of voids at the
coating/substrate interface, and also increased deposition efficiency and adhesion strength of
coatings.
By conducting the studies on a mirror-polished surface, the effect of surface roughness on
splashing was eliminated, allowing focusing on other phenomena that promote splashing.
6.2 Recommendations for future work
The findings of this study can be used as a basis for future work. The followings are
suggestions that would prove beneficial to the advancement of scientific research in the field of
thermal spray coatings.
• Profile of axial variation of velocity, and particle deceleration rate, can be used to
calculate gas velocities, which can be used to verify numerical modeling of fluid
flow in the gun and spray.
• Profile of axial variation of temperature, and particle cooling rates due to convection
and radiation, can be used to calculate rate of oxidation of the particles. This can be
verified by measuring the oxide content of substrates placed at different distances
from the gun.
• Shear stresses on the wire-tips can be used in a two-fluid CFD code to model shape
and size of primary detachments.
125
Reference
1. Abedini, A., A. Pourmousa, S. Chandra, J. Mostaghimi, Journal of Surface and Coating
Technology, Vol. 201 (6), 2006, p 3350-3358
2. Abedini, A., M.A.Sc., Masters Thesis, University of Toronto, Toronto, Canada, 2004
3. Amada, S., I. Imakawa, S. Aoki, Proceedings of the International Thermal Spray
Conference 2003, 2003, p 857
4. Amson, J.C., British Welding Journal, April 1962, p 232–249
5. Arai, T., and H. Hashimoto, Proceedings of the 3rd International Conference on Liquid
Atomization and Spray Systems, 1985 (London), The Institute of Energy, p VIB/1/1–7
6. Assael, M.J., Journal of Physical Chemistry Reference Data, Vol. 35, No. 1, 2006
7. Bailey, S.J. (Editor), Annual Book of ASTM Standards, Section 2, Vol. 02.05, ASTM
International, West Conshohocken, PA, USA
8. Barbezat, G., Surface & Coatings Technology, 200 (2005), p 1990–1993
9. Biancaniello, F.S., S.D. Ridder, Thermal Spray Coatings Workshop: Sensors, Modeling
and Control Strategies, National Institute of Standards and Technology, NISTIR6460,
Gaithersburg, MD, 1998
126
10. Bianchi, L., F. Blein, P. Lucchese and M. Vardelle, Proceedings of the 7th National
Thermal Spray Conference, 1994, p 569-574
11. Bolot, R., R. Bonnet, G. Jandin, C. Coddet, Application of CAD to CFD for the Wire Arc
Spray Process, Thermal Spray 2001: New Surfaces for a New Millennium, C.C. Berndt,
K.A. Khor, and E.F. Lugscheider, Ed., May 28-30, 2001 (Singapore), ASM International,
2001, 1381, p 889-894
12. Capitelli, M., G. Colonna1, C. Gorse1, and A. D'Angola, Eur. Phys. J. D 11, 279-289
(2000)
13. Cedelle, J., M. Vardelle, B. Pateyron, P. Fauchais, M. Fukumoto, I. Ohgitani, Proceedings
of the International Thermal Spray Conference 2005, p 656
14. COHU Inc. Installation and Operation Instructions, 4910 series RS-170 and CCIR CCD
Cameras, 6X-924(B).
15. Crowe, C.T., J. Sommerfeild, Y. Tsuji, Multiphase Flows with Droplets and Particles,
CRC Press, 1998
16. Dhiman, R., “Coating Formation by Sequential Impact of Molten Tin Droplets, MASc
Thesis”, Department of Mechanical and Industrial Engineering, University of Toronto,
2003
17. Dhiman, R. and S. Chandra, Proceedings of the ASME Heat Transfer/Fluids Engineering
Summer Conference, Paper HT-FED 2004-560096, Charlotte, NC, 2004.
18. Drzeniek, H., and H.D. Steffens, Proceedings of 1st National Thermal Spray Conference,
Orlando, FL., 1987, Ed. D.L. Houck, Pub. ASM International, Ohio, 1987
19. Fauchais, P., M. Fukomoto, A. Vardelle & M. Vardelle, Journal of Thermal Spray
Technology, 13 (3), p 337-360, 2004
127
20. Fukumoto, M., S. Katoh, and I. Okane, Proceedings of the 14th International Thermal
Spray Conference, Ed.: A. Ohmori, Vol 1, p 353-359 , 1995
21. Germano, M, U. Pomelli, P. Moin, W. Cabot, Phys. Fluids A 3 (7), July 1991
22. Ghafouri-Azar, R., J. Mostaghimi, S. Chandra, and M. Charmchi, Journal of Thermal
Spray Technology, 2003 12(1), p 53-69
23. Guillen, D.P., Oxidation Behavior of In-Flight molten Aluminum Droplets in the Twin-Wire
Electric Arc Thermal Spray Process”, Ph.D. Thesis, 2005
24. Handbooks of Monochromatic XPS Spectra, Volumes 1-5, B.V.Crist, XPS International
LLC, 2004, Mountain View, CA, USA
25. Hedges, M.K., A.P. Newbery and P.S. Grant, Materials Science and Engineering A,
Volume 326, Issue 1 , 15 March 2002, p 79-91
26. Hussary, N.A., M.A.Sc. Thesis, University of Minnesota, 1999
27. Hussary, N.A., Ph.D. Thesis, University of Minnesota, 2003
28. Hussary, N.A., and J. Heberlein, Thermal Spray: Surface Engineering via Applied
Research, C.C. Berndt, Ed., May 8-11, 2000 (Montréal, Québec, Canada), ASM
International, 2000, p 737-742
29. Hussary, N.A., and J. Heberlein, Thermal Spray 2003: Advancing the Science and Applying
the Technology, B.R. Marple and C. Moreau, Ed., May 5-8, 2003 (Orlando, FL), ASM
International, 2003, p 1023-1032
30. Jandin, G., H. Liao, Z.Q. Feng, and C. Coddet, Materials Science and Engineering J.,
A349, 2003, p 298-305
128
31. Jenista, J., J. Heberlein, and E. Pfender, IEEE Transactions On Plasma Science, Vol 25,
No. 5, October 1997, p 883-890
32. Jiang, X., Y. Wan, H. Herman, and S. Sampath, Thin Solid Films, 385, p 132-141, 2001
33. Kelkar, M. , Ph.D. Thesis, University of Minnesota, 1998
34. Kelkar, M., and J. Heberlein, Plasma Chemistry and Plasma Processing, Vol 22, No.1,
March 2002, p 1-25
35. Kelkar, M., N. Hussary, J. Schein, and J. Heberlein, Thermal Spray: Meeting the
Challenges of the 21st Century, C. Coddet, Ed., May 25-29, 1998 (Nice, France), ASM
International, 1998, p 329-334
36. Krauss, M., D. Bergmann, U. Fritsching, and K. Bauckhage, Materials Science and
Engineering A, Vol. 326, Issue 1, 15 March 2002, p 154-164
37. Lefebvre, A.H., Atomizations and Sprays, Taylor & Francis, 1989
38. Lowke, J.J., Journal of Physics D: Applied Physics, Vol. 12, 1979
39. Masri, M., M.A.Sc. Thesis, Department of Mechanical and Industrial Engineering,
University of Toronto, 1996
40. Mauer, G., R. Vaßen, and D. Stover, Journal of Thermal Spray Technology, Vol 16 (3),
2007, p 414-424
41. McDonald A., M. Lamontagne, C. Moreau & S. Chandra, Thin Solid Films, Vol 514, Issues
1-2 , 30 August 2006, p 212-222
42. Mehdizadeh, N.Z., S. Chandra, and J. Mostaghimi, Journal of Fluid Mechanics, 510
(2004), p 353
129
43. Mehdizadeh, N. Z., M. Raessi, S. Chandra, and J. Mostaghimi, Journal of Heat Transfer,
126, p 445-452 (2004)
44. Mostaghimi, J., M. Pasandideh-Fard, and S. Chandra, Plasma Chemistry and Plasma
Processing, 22(1), p 59-84 (2002)
45. Newbery, A.P., B. Cantor, R.M. Jordan, S.E. Singer, J. Mater., Synth. Processing, 4 (1996)
46. Newbery, A.P. and P.S. Grant, Powder Metallurgy, 2003, Vol. 46, No. 3, p 229-235
47. Pasandideh-Fard, M., V. Pershin, S. Chandra and J. Mostaghimi, Journal of Thermal Spray
Technology, 11, p 206-217 (2002)
48. Pasandideh-Fard, M., R. Bhola, S. Chandra, and J. Mostaghimi, International Journal of
Heat and Mass Transfer, 41 (1998), p 2929
49. Perry, R.H., D. Green, Perry’s Chemical Engineer’s Handbook, McGraw-Hill, 1984
50. Pershin, V. I., M. Lufitah, S. Chandra and J. Mostaghimi, Journal of Thermal Spray
Technology, 12, p 370-376, 2003
51. Planche, M.P., H. Liao, T. Le Ba, and C. Coddet, Proceedings of the 16th International
Symposium on Plasma Chem., p 22-27, June 2003 (Taormina, Italy), 2003
52. Planche, M.P., H. Liao, and C. Coddet, Surface and Coatings Technology, Vol 182, 2004, p
215–226
53. Pouliot. L., J. Blain, and F. Nadeau, DPVOS Reference Manual, Revision 5.0, TECNAR
Automation Ltd., St-Hubert, QC, Canada, July 1999
54. Pourmousa, A., A. Abedini, J. Mostaghimi, and S. Chandra, Thermal Spray 2004:
Advances in Technology and Application, ASM International, May 10-12, 2004 (Osaka,
Japan), ASM International, 2004, Process Diagnostics (I).
130
55. Pourmousa, A., J. Mostaghimi, A. Abedini, and S. Chandra, “Particle Size Distribution in a
Wire-Arc Spraying System”, Journal of Thermal Spray Technology, Volume 14(4),
December 2005
56. Praxair Surface Technologies, Wire Solutions Catalog, 2004
57. Saevarsdottir, G., M.T. Jonsson, and J.A. Bakken, Proceedings of the 16th International
Symposium on Plasma Chem., p 22-27 June 2003 (Taormina, Italy), 2003
58. Sampath, S., G. Montavon, C.C. Berndt, H. Herman , and C. Coddet, Surface and Coatings
Technology, 91 (1997), p 107
59. Shakeri, S., S. Chandra, International Journal of Heat and Mass Transfer, 24 (2002), 4561
60. Steffens, H.D., Z. Babiak, and M. Wewel, IEEE Transactions on plasma science, Vol 18,
No 6, December 1990
61. Sulzer-Metco’s official webpage: http://www.sulzermetco.com
62. Vaidya, A., T. Streibl, S. Sampath, H. Zhang, Thermal Spray 2004: Advances in
Technology and Application, ASM International, May 10-12, 2004 (Osaka, Japan), ASM
International, 2004, Consumables for thermal spraying (I)
63. Varacalle, D.J., D.P. Zeek, V. Zanchuck, E. Sampson, K.W. Couch, D. Benson, G.S. Cox,
Journal of Thermal Spray Technology, 7, 4, 1998, p 513-520
64. Varacalle, D.J., G.C. Wilson, L.B. Lundberg, D.L. Hale, V. Zanchuck, W. Kratochvil and
Irons, G., Proceedings of 8th National Thermal Spray Conference, Houston, TX, 1995,
Eds: Bemdt, C. C. & Sampath, S., ASM International, Ohio, pp. 373-380
65. Wang, X, PhD Thesis, University of Minnesota, 1996
131
66. Wang, X, J. Heberlein, E. Pfender, and W. Gerberich, Thermal Spray: Practical Solutions
for Engineering Problems; Cincinnati, Ohio; USA; 7-11 Oct. 1996, p 807-811, 1996
67. Wang, X., D. Zhuang, E. Pfender, J. Heberlein, and W. Gerberich, Thermal Spray
Industrial Applications, C.C. Berndt and S. Sampath, Ed., June 20-24, 1994 (Boston, MA),
ASM International, 1994, p 587-592
68. Watanabe, T., T. Sato, and A. Nezu, Thin Solid Films, Vol 407, 2002, p 98-103
69. Watanabe, T., X. Wang, E. Pfender, and J. Heberlein, Thin Solid Films, Vol 316, 1998, p
169-173
70. Wire-Arc Manual, Sulzer-Metco, Westbury, NY
71. Yule, A.J., and J.J. Dunkley, Atomization of Melts for Powder Production and Spray
Deposition, Oxford, 1994, p 43
72. Zhu, Y.L., H.L. Liao, C. Coddet, B.S. Xu, Surface and Coatings Technology, Vol. 162,
2003, p 301–308
132
Appendix A: Metal Properties
Table 6.1. Properties of Copper, Aluminum and Aluminum Oxide (Al2O3) [ 6, 27, 49, 68]
Stainless Steel
Copper Aluminum Aluminum Oxide
Melting Point (oC) 1536 1083 660 2015
Boiling Point (oC) 2861 2567 2494 2980
Specific Heat (J/kg/K) @ 30 oC 385 900 1637
Density (kg/m3) @ 30 oC 7015 8960 2700 3970
Latent Heat of Fusion (kJ/kg) 246.6 205 388 1100
Latent Heat of evaporation (kJ/kg) 10800
Thermal Conductivity @30oC (W/m/K) 393 273 36
Surface Tension (N/m) 1.872 1.3 0.91
Viscosity in liquid form (Pa.s.) 6.0×10-3 1.3×10-3
133
Appendix B: Transport Properties of Air
Table 6.2. Transport properties of air at atmospheric pressure [ 12]
Temperature Viscosity Thermal
Conductivity Electric
Conductivity Enthalpy Entropy Specific
Heat Density (K) (kg/m/s) (W/m/K) (S/m) (cal/g) (cal/g/K) (cal/g/K) (g/m3) 50 3.479E-06 5.085E-03 0 121.77 1.2467 0.2419 7018.1
100 7.048E-06 1.031E-02 0 133.87 1.3984 0.2419 3509 200 1.302E-05 1.906E-02 0 158.06 1.5579 0.2421 1754.5 300 1.801E-05 2.642E-02 0 182.28 1.6533 0.2428 1169.6 400 2.240E-05 3.312E-02 0 206.62 1.7219 0.2449 877.22 500 2.639E-05 3.961E-02 0 231.27 1.776 0.2487 701.75 600 3.008E-05 4.607E-02 0 256.37 1.8212 0.2538 584.77 700 3.355E-05 5.252E-02 0 282.03 1.8603 0.2596 501.21 800 3.685E-05 5.893E-02 0 308.28 1.895 0.2654 438.53 900 4.001E-05 6.527E-02 2.443E-23 335.11 1.9263 0.2711 389.77
1000 4.304E-05 7.152E-02 2.882E-20 362.5 1.955 0.2746 350.76 1500 5.694E-05 0.1035 5.593E-07 502.9 2.0702 0.2955 233.83 2000 6.949E-05 0.1382 8.832E-07 663.38 2.1705 0.3479 175.79 2500 8.153E-05 0.2264 1.120E-02 848.62 2.274 0.4079 140.58 3000 9.501E-05 0.5232 1.986E-02 1099.3 2.4205 0.64 113.7 3500 1.117E-04 0.7599 0.4896 1508.1 2.6709 0.9428 91.276 4000 1.291E-04 0.5507 2.162E+00 1954.3 2.9104 0.7661 76.625 4500 1.436E-04 0.5548 7.197E+00 2281.9 3.0255 0.5964 66.395 5000 1.566E-04 0.781 2.273E+01 2597.4 3.1122 0.7156 58.223 6000 1.855E-04 2.665E+00 9.857E+01 3804.7 3.5402 1.9643 44.181 7000 2.153E-04 4.267E+00 3.277E+02 6692.6 4.5879 3.4157 30.852 8000 2.338E-04 2.179E+00 1.060E+03 9448.3 5.3614 1.8803 23.213 9000 2.509E-04 1.262E+00 2.146E+03 10785 5.5171 1.0318 19.659
10000 2.614E-04 1.377E+00 3.264E+03 11812 5.5557 1.1142 17.23 11000 2.540E-04 1.775E+00 4.372E+03 13148 5.6688 1.6212 15.113 12000 2.219E-04 2.335E+00 5.478E+03 15185 5.9293 2.5217 13.083 13000 1.748E-04 2.948E+00 6.535E+03 18304 6.3853 3.75 11.101 14000 1.253E-04 3.462E+00 7.536E+03 22651 7.026 4.8648 9.2751
134
15000 8.617E-05 3.665E+00 8.418E+03 27767 7.733 5.1913 7.7689 16000 5.724E-05 3.531E+00 9.205E+03 32708 8.3343 4.5586 6.6522 17000 4.003E-05 3.317E+00 9.886E+03 36741 8.7321 3.4915 5.868 18000 2.929E-05 3.143E+00 1.052E+04 39735 8.9404 2.5437 5.3119 19000 2.418E-05 3.142E+00 1.111E+04 41937 9.0234 1.9157 4.8967 20000 2.110E-05 3.225E+00 1.168E+04 43663 9.0409 1.5704 4.5673 22000 1.955E-05 3.650E+00 1.277E+04 46658 9.0283 1.5337 4.0505 24000 1.945E-05 4.264E+00 1.365E+04 50274 9.0949 2.2194 3.6281 26000 1.804E-05 5.000E+00 1.405E+04 56186 9.384 3.8729 3.228 28000 1.475E-05 5.779E+00 1.402E+04 66273 10.021 6.2045 2.8072 30000 1.125E-05 6.494E+00 1.399E+04 80033 10.905 7.1159 2.4137