1-s2.0-S0017931009003925-main

8/9/2019 1-s2.0-S0017931009003925-main

1/14

Formation of weld crater in GMAW of aluminum alloys

H. Guo a , J. Hub, * , H.L. Tsai aa Department of Mechanical and Aerospace Engineering, Missouri University of Science and Technology (formerly University of Missouri Rolla), 400 W. 13th Street, Rolla,MO 65409, USAb Department of Mechanical Engineering, University of Bridgeport, Bridgeport, CT 06604, USA

a r t i c l e i n f o

Article history:Received 9 October 2008Received in revised form 17 June 2009Accepted 17 June 2009Available online 5 August 2009

Keywords:GMAWAluminum WeldingCraterMicrostructure

a b s t r a c t

Both mathematical modeling and experiments have been conducted on the formation of the craterformed in a GMAW of aluminum alloy 6005-T4. Transient weld pool shape and the distributions of tem-perature and velocity were calculated by a three-dimensional numerical model. The nal weld beadshape and dimensions were obtained. Corresponding experiments were conducted and in good agree-ment with modeling predictions. Metallurgical characterizations were also performed on the experimen-tal samples. It was found that due to the fast solidication of the weld pool after the termination of thewelding arc, there is no timefor the molten metal to ow back towards the weld pool center and close upthecrater. Thus a crater was formed at theend of theweld bead. Solidication cracking was formed at thecenter of weld crater. A back-up technique was proposed to allow extra molten metal to ow back tothe crater and ll it up. The crater was successfully lled and the crater cracking was eliminated.

2009 Elsevier Ltd. All rights reserved.

1. Introduction

Gas metal arc welding (GMAW) is an arc welding process whichuses a metal wire as a combined electrode and ller metal in aplasma arc of inert shielding gas. Filler metal is added to the weldpool automatically and continuously. GMAW has some advantagesover other welding methods, such as high productivity, better pen-etration, no need for ux, little spatter, and ability to weld in allpositions. Therefore, it has been one of the most popular weldingmethods in industry. Due to the demands for a lower environmen-tal impact through improved fuel efciency, weight reduction, andload capacity, aluminum is being more and more widely used inthe auto industry because of its light weight. More automobilebody structures, such as engine cradles, are being manufacturedwith aluminum instead of steel.

One feature of these aluminum welds is that they are muchshorter than normal welds. They are usually less than 10 cm inlength, and many are only around three or four centimeters. There-fore, the terminating end (weld crater) makes up a great portion of short welds ( Fig. 1 ). At the crater, the welding arc and wire feed areterminated. Unlike the middle portion of a long weld, where thewelding process is in a quasi-steady state, the energy, mass andmomentum transfer varies sharply from moment to moment in acold weld, thus creating very unsteady temperature and uid owelds. For GMAW of aluminum, the weld pool solidies very fastdue to the very high heat conductivity of aluminum alloys, leading

to the formation of cracking, which is a major defect in weldcraters. Cracks are likely to occur where the metal lacks ductilityand the tensile stress develops after the shut off of heat and massinput at the end of the welding process [1,2] .

Hot cracking occurs when metal is above solidus temperatureand a tensile stress is applied [2,3] . It usually happens when weld-ing the heat-treatable aluminum alloys, resulting from the use of incorrect ller metal, excessive base alloy dilution of a weld, an im-proper joint design or excessive joint spacing [2] . Aluminum alloysare sensitive to solidication cracking as a result of high thermalexpansion combined with a brittle structure at and just belowthe solidication temperature [4] . According to the studies of Pereira et al. [5] , the development of ne grain structure in theweldment helps to reduce the solidication cracking tendency.The shape of the weld pool also plays an important role in thedevelopment of solidication cracking. The liquation cracking inthe HAZ was investigated by Kerr and Katoh [6] . The crack lengthincreased linearly with the increase of augmented strain or theheat input. The simulation of liquation cracking in 7017 aluminumalloy [7] found it is determined by both the applied stress level andthe temperature at which stress is applied. Increasing the coolingrate may minimize the cracking. The relationships between therestraint intensity of weld grooves in aluminum alloy and hotcracking of weld metals were investigated [8] . To prevent cracking,the welding parameters should be selected carefully to obtain aproper cross-sectional shape of the weld bead [8] .

Numerous studies have been carried out on the welding pro-cess, but no detailed analysis has been reported on crater in gasmetal arc welding of aluminum alloys. Many experimental studies

0017-9310/$ - see front matter 2009 Elsevier Ltd. All rights reserved.doi: 10.1016/j.ijheatmasstransfer.2009.06.028

* Corresponding author. Tel.: +1 203 576 4757; fax: +1 203 576 4765.E-mail address: [email protected] (J. Hu).

International Journal of Heat and Mass Transfer 52 (2009) 55335546

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

j ou r na l homepag e : www.e l s ev i e r. com/ loca t e / i j h m t
http://dx.doi.org/10.1016/j.ijheatmasstransfer.2009.06.028mailto:[email protected]://www.sciencedirect.com/science/journal/00179310http://www.elsevier.com/locate/ijhmthttp://www.elsevier.com/locate/ijhmthttp://www.sciencedirect.com/science/journal/00179310mailto:[email protected]://dx.doi.org/10.1016/j.ijheatmasstransfer.2009.06.028

8/9/2019 1-s2.0-S0017931009003925-main

2/14

have been conducted on the GMAW process [916] . Since GMAWis a very complicated process involving many coupled parameters,such as welding current, voltage, welding speed, electrode feedspeed, base metal material, electrode material, electrode size, andshielding gas, and also because welding is a transient process athigh temperatures, it is very hard to use experimental methods

alone to understand its mechanisms. Mathematical modeling pro-vides a convenient way to get insightful information. Many studieshave been performed on the mathematical modeling of the weld-ing process in the past. It was found that uid ow within the weldpool is driven by forces such as buoyancy, electromagnetic forces,surface tension, and arc plasma drag forces. Effect of gravity onweld pool shape was simulated. A high gravitational eld causesan enhanced buoyancy-driven radially outward ow in the weldpool [17] . Surface tension is one of the major driving forces[1823] . Electromagnetic forces drive uid in a downward motionand cause a deeper penetration [18,24] . In the welding process thearc plasma drag force induces an outward uid ow on the weldpool surface [21] .

Since the middle of the 1980s, many theoretical models havebeen establishedon thesimulationof thegas metalarc weldingpro-cess. Ushioand Wuproposed a model to calculate the three-dimen-sional heat anduidowin a moving gasmetal arc weldpool [25] . Aboundary-ttednonorthogonalcoordinate systemwasadoptedandit was found that the size and prole of the weld pool are strongly

inuenced by the volume of molten wire, impact of droplets, andheat content of droplets [25] . A at weld pool surface is assumedin Jaidi and Dutta [26] . Park and Rhee reported that the kinetic en-ergy of the transferring droplets produces a depression on the weldpool surface [27] . According to the computational investigations byDavies et al. [28] , the impinging droplet momentum dominates theow pattern andoverrides anysurface tension effects at a relativelyhighcurrent. Arghode et al. [29] simulated theeffect of moltendrop-let addition to theweldpoolas volumetric heat and species sources.Wang and Tsai [23] and Hu and Tsai [3033] developed detailedmodelson the impingement of ller droplets andweld pool dynam-ics and calculated the combined effect of droplet impingement andsurface tension. However, all the researchefforts were focused onlyon the quasi-steady state part of the GMAW process.

The objectives of this project are to obtain a better understand-ing of the formation mechanism of crater and to nd a properimprovement procedure. The uid ow and heat transfer were cal-culated when droplet carries mass, momentum, and thermal en-ergy into the weld pool. The transient deformed weld poolsurface was handled by the volume of uid (VOF) technique [34]and the fusion and solidication in the liquid region, the mushyzone and the solid region were handled by the continuum formu-lation [35] . In experimental research, bead-on-plate experimentswere performed. Experimental samples were then characterizedusing metallurgraphical methods.

2. Mathematical model

A sketch of a moving GMAW for a plate is shown in Fig. 1 . Thethree-dimensional xyz -coordinate system is xed to the baseFig. 1. A crater with cracking at the end of a weld.

Nomenclature

c specic heat f mass fraction g gravitational accelerationh enthalpyhc convective heat transfer coefcientk thermal conductivity~n normal vector to the local surface p pressure pv vapor pressure or any other applied external pressurer z cylindrical coordinate system

~s local surface tangential vectort timeu velocity in x-directionuw arc voltagev velocity in y-directionw velocity in z -direction B*

magnetic induction vectorC inertial coefcientD mass diffusion coefcientF volume of uid functionH v latent heat of vaporizationI welding current

~ J current density vectorK permeability functionP max maximum arc pressure at the arc centerT temperature

~V velocity vector~V r relative velocity vector between the liquid phase and

solid phase.W melt mass evaporation rate

Greek symbolsbT thermal expansion coefcientc surface tension coefciente surface radiation emissivityj free surface curvatureg arc thermal efciencygd ratio of droplet thermal energy to the total arc energyr StefanBoltzmann constantq densityl dynamic viscosityr p arc pressure distribution parameterr q arc heat ux distribution parameters~s Marangoni shear stress

Subscripts0 initial conditiond dropletl liquid phasem melting point of aluminums solid phase

5534 H. Guo et al. / International Journal of Heat and Mass Transfer 52 (2009) 55335546

8/9/2019 1-s2.0-S0017931009003925-main

3/14

metal. The arc is moving in the positive x-direction, and dropletsimpinge onto the base metal in the negative z -direction whilemoving at the same velocity along the x-direction as the arc.

2.1. Governing equations

The differential equations governing the conservation of mass,momentum, and energy based on continuum formulation givenby Diao and Tsai [35] were modied and employed in the study.The equations are given below:

(1) Continuity,

o qo t r q V

* 0: 1

(2) Momentum,

o

o t qu r q V

*u r l l

qq l

r uo po x

l lK

qq l

u us

C q 2

K 1=2q l ju usju us

r q f s f l V *

r ur J

* B*

j x; 2

o

o t q v r q V

*v r l l

qq l

r v o po y

l lK

qq l

v v s

C q2

K 1=2q ljv v sjv v s

r q f s f l V *

r v r J

* B*

j y; 3

o

o t qw r q V

*w r l l

qq l

r wo po z

l lK

qq l

w ws

C q 2

K 1=2q ljw wsjw ws

r q f s f l V *

r w r q g q g bT T T 0

b s f al f al;0 J

* B*

j z F drag : 4

(3) Energy,

o

o t qh r q V

*h r

kc s

r h r kc s

r hs h

r q V *

V *

s

h l h : 5

(4) Species,

o

o t q f a r q V

* f a r qDr f a r qDr f al f a

r q V *

V *

s f al f

a ; 6

where t is thetime; q is the density; ~V is the velocityvector; u , v andw are the velocities in the x-, y-, and z -directions, respectively;

V *

r V

*

lV *

sis the relative velocity vector between the liquid phase

andthe solid phase; thesubscripts s and l refer to thesolid and liquidphases, respectively; p is thepressure; l is thedynamic viscosity; f isthe mass fraction; K is the permeability, a measure of the ease withwhich uids pass through the porous mushy zone; C is the inertial

coefcient; bT is the thermal expansion coefcient; g is the gravita-

tionalacceleration; T is the temperature; B*

is themagnetic induction

vector; J *

is the current densityvector; the subscript 0 represents theinitial condition; h is the enthalpy; k is the thermal conductivity; c isthe specic heat; f a is the fraction of constituent; and D is the massdiffusivity. The detailed descriptions of the terms in Eqs. (1)(5) canbefound in Refs. [23,3033] and will not be repeated here. The solid/liquid phase-change is handled by the continuum formulation [35] .The third, fourth and fth termsin theright-hand side of Eqs. (2)(4)vanish at the solid region because of u = us = v = v s = w = w s = 0 and f l = 0 for thesolid. In theliquid region, K goes to innity, and alltheseterms also vanish [23,3033] . Those terms are only effective for themushy zone where 0 < f l < 1 and 0 < f s < 1. Therefore, the liquid re-gion, mushy zone and solidregion can behandledby thesame equa-tions. During the fusion and solidication process, latent heat isabsorbed or released in the mushy zone.

2.2. Tracking of free surfaces

Volume of uid (VOF) technique [34] was employed to track thedynamic free surfaces. The uid conguration is dened by a vol-ume of uid function, F(x, y, z , t ), which represents the volume of liquid metal per unit volume and satises the following conserva-tion equation:

dF dt

o F o t

V *

r F 0: 7

When averaged over the cells of a computing mesh, the averagevalue of F in a cell is equal to the fractional volume of the cell occu-pied by uid. A unit value of F corresponds to a cell full of uid,whereas a zero value indicates a cell contains no uid. Cells withF values between zero and one are partially lled with uid andidentied as surface cells.

2.3. Boundary conditions

The boundary conditions for the previous Eqs. (1)(6) are givenbelow.

2.3.1. Normal to the local free surfaceFor cells containing a free surface, that is, cells that contain uid

but have one or more emptyneighbors, the followingpressurecon-ditions must be satised [34] :

p pv cj ; 8

where p is the pressure at the free surface in a direction normal tothe local free surface and pv is the plasma arc pressure which is as-sumed to have a radial distribution in the following form [25] :

pv P max exp r 2

2r 2 p !; 9where P max is the maximum arc pressure at the arc center, which iscalculated from the welding current [25] , r is the distance from thearc center, and r p is the arc pressure distribution parameter [25] . jin Eq. (8) is the free surface curvature given by

j r n*

j n*

j !" # 1j n* j n*

j n*

jr !j n* j r n*" #; 10

where n*

is the normal vector to the local surface, which is the gra-dient of VOF function n

*

r F .

2.3.2. Tangential to the local free surfaceThe Marangoni shear stressat the free surface in a direction tan-

gential to the local free surface is given by

H. Guo et al. / International Journal of Heat and Mass Transfer 52 (2009) 55335546 5535

8/9/2019 1-s2.0-S0017931009003925-main

4/14

s~s l lo V

*s*

o n*

o co T

o T o~s

; 11

where s*

is the local surface tangential vector. Since there is no sur-face tension coefcient data available for 6005-T4, the property of pure aluminum was used instead. For pure aluminum, surface ten-sion coefcient c is a function of temperature [36] .

c 868 0:152 T T m; 12

where T is the temperature and T m is the melting temperature of aluminum.

2.3.3. Top surfaceAt the moving arc center, in addition to the droplet impinge-

ment, arc heat ux is also impacting on the base metal. Since thearc heat ux is relatively concentrated, it is assumed that the heatux is perpendicular to the base metal (i.e., neglecting the inclina-tion of current andheat ux). Therefore, the temperature boundaryconditions at the top surface of the base metal are

ko T

o z

g1 gdIuw

2p r2q

exp r 2

2r2q

!qconv qradi qev ap ; 13

where I is the welding current, g is the arc thermal efciency, gd isthe ratio of droplet thermal energy to the total arc energy, uw is thearc voltage, and r q is the arc heat ux distribution parameter [20] .The heat loss due to convection, radiation, and evaporation can bewritten as

qconv hc T T 1 ; qradi r eT 4 T 41 ; qev ap WH m; 14

where h c is the convective heat transfer coefcient [20] , T 1 is theroom temperature, r is StefanBoltzmann constant, e isthe surface radiation emissivity [36] , H v is the latent heat for theliquidvapor phase-change [37] , and W is the melt mass evapora-tion rate [38] .

2.3.4. Symmetrical y = 0 plane

o T o y

0 ;o uo y

0 ; v 0 ;o wo y

0 ;o f ao y

0: 15

2.3.5. Other surfaces

ko T o~n

qconv ; u 0 ; v 0 ; w 0: 16

2.4. Electromagnetic force

In Eqs. (2)(4) , there are three terms caused by the electromag-

netic force, J *

B

*

, which should be calculated rst before the calcu-lation of velocity. Assuming the electric eld is a quasi-steady-stateand the electrical conductivity is constant, the scalar electric po-tential, / , satises the following Maxwell equation [22] in the localrz-coordinate system:

r 2 / 1r

o

o r r

o /o r

o 2 /o z 2

0: 17

The required boundary conditions for the solution of Eq. (17)are

r eo /o z

I

2p r 2c exp

r 2

2r 2c ; at the top free surface ; 18

o /o z 0; at z 0; 19

o /o r

0; at r 0; 20

/ 0; at r 10 r c ; 21

where r e is the electrical conductivity and r c is the arc current dis-tribution parameter [22] . After the distribution of electrical poten-tial is solved, the current density in the r- and z-directions can becalculated via

J r r eo /o r ;

J z r eo /o z :

22

The self-induced azimuthal magnetic eld is derived from Am-peres law via the following equation [22] :

Bh l 0r Z

r

0 J z r dr ; 23

where l 0 is the magnetic permeability in free space. Finally, thethree components of electromagnetic force in Eqs. (2)(4) are calcu-lated via the following equations [22] :

J *

B*

j x Bh J z x xa

r ; J

* B*

j y Bh J z yr ;

J *

B*

j z Bh J r : 24

2.5. Numerical considerations

The governing equations were solved iteratively at each timestep using nite volume method [39] . At each time step, the conti-nuity and momentum equations were solved iteratively with atwo-step projection method involving the time discretization of the momentum equations to get the velocity andpressure distribu-tions [23] . Then the energyequation was solvedexplicitly to obtainthe enthalpy and temperature eld. The species equation wassolved in a similar way. This process was repeated for each itera-

tion step. Iteration within a time step was terminated when thesolutions of velocity, pressure, temperature, and species distribu-tions converged. Then the VOF function equation was solved to ob-tain the newfree surface and liquidpool domain. The temperature-dependent material properties were updated. The time step wasthen advanced and the above procedure was repeated until the de-sired time was reached.

Since the governing equations are valid for the entire computa-tional domain including the liquid phase, the solid phase, and themushy zone, there is no need to track the shape and extent of eachphase. Therefore, a xed grid system was used in the calculationwith rened grid cells in the weld pool zone to improve accuracy.Due to the symmetry of the xz plane of the domain, a grid systemof 408 66 56 points was used for the half computational do-main to save computational time. The ner grids, concentratingon and around the weld pool, move with the weld pool as thewelding proceeds. Time step length varied during the calculationto ensure the convergence and save computational time. The com-putation was performed on the Dell Precision 650 workstationswith 3.2 GHz Pentium 4 processors. It took about 71 h of CPU timeto simulate 1.4 s of real-time welding. The average time step isaround 2 10 5 s.

2.6. Material properties in simulation

The thermophysical properties shown in Table 1 were inputsinto the numerical simulation model. When temperature-depen-dent material properties at high temperature were not available,

constant values for the solid metal at room temperature wereused.


8/9/2019 1-s2.0-S0017931009003925-main

5/14

3. Experiments

3.1. Experimental setup

The experimental setup is shown in Fig. 2 . In the experiments,bead-on-plate welds were made on aluminumalloy 6005-T4 plates203.2 mm 38.1 mm 5 mm in dimension, which were extrudedby Hydro Raufoss Automotive. Every weld coupon was chemicallycleaned and degreased. The electrode material was 4043 producedby Alcoa. The diameter of the electrode wire was 1.6 mm in allexperiments.

The welding machine was a Lincoln PowerWave 455 program-mable waveform controlled welding machine made by LincolnElectric. The weld torch was xed onto a small cart on a rail. Thetravel speed of the cart could be adjusted. Argon was used as theprotecting gas, which has a ow rate of 40 CFH. To provide an ade-quate protection of the weld pool, a welding gun leading angle of 15 was used in the experiments. The weld bead was made underconstant current mode of the welding machine at the center of theplate along the x-direction as shown in Fig. 2 . All welds startedfrom 30 mm to the left end of the weld coupon. Before welding,the upper surface of the plate was brushed with a stainless steelbrush to remove the oxide layer. Three major parameters could

be adjusted during the process: welding current, wire feed speed,and arc/cart travel speed. Arc voltage was automatically set bythe machine once other parameters were xed.

3.2. Monitoring the welding process

It is very important that the experiments are closely monitoredduring the process by connecting the port on the PowerWave455front panel to the serial port of a computer and using WaveDesign-er software from Lincoln Electric. The welding parameters, suchas arc current, voltage, and welding time, were stored in the com-puter and input into the mathematical model later on.

3.3. Metallurgical characterizations of samples

3.3.1. Sample preparationWelded samples were sectioned, grinded, polished, and then

etched for metallurgical characterizations. Sectioning was per-formed on a Leco CM-15 cut-off machine. Coolant was used dur-ing the cutting process. The sectioned samples were hot mountedusing Bakelite powder. The sample grinding and polishing wereperformed on a Leco Spectrum System2000 grinder/polisher. Ingrinding, 240-, 400-, 600-, 800-, and 1200-grit abrasive disks wereused consecutively, and in polishing, diamond compounds of 6, 3,and 1 l m and 0.05 l m gamma alumina solution were used. Thepolished samples were then etched by Tuckers reagent and Kel-lers reagent [41] for macroscopy and microscopy analysis,respectively.

3.3.2. Metallurgical characterizationsMetallographic analysis was performed under stereoscopes and

optical microscopes. An image acquisition system including a dig-ital camera and a computer was used to capture and store theimages. Adobe Photoshop and Image Processing Tool Kit wereused for the processing of sample images. The weld penetration,width, and reinforcement were measured.

4. Results and discussion

4.1. Normal weld

The formation of the crater for GMAW of 6005-T4 aluminumalloy was calculated. The uid ow pattern, temperature

Table 1

Thermophysical properties used in the model.

Property, unit Value

Specic heat of solid phase, J/kg-K 900 a

Specic heat of liquid phase, J/kg-K 900 a

Thermal conductivity of solid phase, W/m-K 167 b

Thermal conductivity of liquid phase, W/m-K 167 b

Density of solid phase, kg/m 3 2700 b

Density of liquid phase, kg/m 3 2300 bCoefcient of thermal expansion, K 1 2.34 10 5 b

Dynamic viscosity, kg/m-s 0.0012 a

Heat of fusion, J/kg 0.397 a

Heat of vaporization, J/kg 1.08 10 7 a

Solidus temperature, K 880 b

Liquidus temperature, K 927 b

Electrical conductivity, X 1 m 1 2.5 10 7 b

Plasma density, kg/m 0.06 b

Plasma viscosity, kg/m-s 2.5 10 4 b

a Property of pure aluminum [36] .b Property of 6005 [40] .

Fig. 2. Experimental setup and simulation domain of a GMAW system.


8/9/2019 1-s2.0-S0017931009003925-main

6/14

8/9/2019 1-s2.0-S0017931009003925-main

7/14

diameter [42] . Further aging leads to the formation of the b 00 phase,which is a needle-shaped precipitation larger than the GP zones.The next product of the aging process is the b 0 Mg 2Si phase, whichis formed from the b00 phase by growth in precipitate length anddiameter [36] . Theequilibrium phase b -Mg

2Si is formed in the nal

stage of aging by diffusionless transformation of the b0 phase.

Among the four phases, GP zones and b 0 phases have only moder-ate strengthening effects, and the contribution of the b phase isvery small. b00 phase is the primary strengthening phase in 6xxxseries alloys [43] . According to the research works of Malin [44] ,in the HAZ of the weld joint, the b 00 phase begins to coarsen at tem-peratures below 523 K, which lowers the metal hardness. Further

t = 1.4017 s

t = 1.4007 s

t = 1.4022 s

Z ( m m

)

2

4

6

8 t = 1.3900 s

Z ( m m

)

2

4

6

8 t = 1.4017 s

X (mm)

t = 1.4040 s

2014 16 18 22

X (mm)

Z ( m m

)

2

4

6

8 t = 1.4024 s

2014 16 18 22

T

927880811738665519

T (K)

Z ( m m

)

2

4

6

8 t = 1.3900 s t = 1.4007 s

Z ( m m

)

2

4

6

8 t = 1.4017 s t = 1.4022 s

X (mm)

t = 1.4040 s

14 2216 18 20

X (mm)

Z ( m m

)

2

4

6

8 t = 1.4024 s

14 2216 18 20

)b()a(

0.5 m/s

Z ( m m

)

2

4

6

8 t = 1.3900 s

Z ( m m )

2

4

6

8 t = 1.4017 s t = 1.4022 s

t = 1.4007 s

0.5 m/s

0.5 m/s

0.5 m/s

0.5 m/s

Z ( m m

)

2

4

6

8 t = 1.4024 s t = 1.4040 s 0.5 m/s

(c)X (mm)

2014 16 18 22

X (mm)

2014 16 18 22

Fig. 4. Side view of the formation of a crater: (a) weld pool (the darkest color) and weld bead (the second darkest color), (b) temperature eld, and (c) velocity eld.


8/9/2019 1-s2.0-S0017931009003925-main

8/14

increase of temperature causes the b 00 phase to transform to the b 0

phase between 523 K and 653 K. The b 00 phase is reported to con-tinue to coarsen in this temperature range [45] . When the temper-ature is elevated above 653 K, which is the solvus temperature of precipitations, both b 00 and b 0 phases dissolve in the aluminum ma-trix [45] . Although the composition of alloy in this research is dif-ferent from the 6061 used by Dumolt [45] and Malin [44] ,according to Ceresara et al. [46] , an excess of Mg or Si does nothave any effect on the sequence and the structure of the normal

precipitation process. Only the precipitation rate and extent are af-fected. The corresponding temperatures of 6005 alloy found byMaitland and Ried [47] are very close to the above data. Therefore,the temperatures organized by Malin [44] will be used as bench-marks for convenience.

It is observed from Fig. 9 that for the whole micro-hardnessmeasurement zone, the peak temperature is above 523 K every-where. When the temperature is between 523 K and 653 K, theb 00 phase coarsens and also transforms to the b 0 phase, causing low-

Z ( m m

)

2

4

6

8t =1.3900 s

Z ( m m

)

2

4

6

8t = 1.4017 s t =1.4022 s

Y (mm)

Z ( m m

)

-5 0 5

2

4

6

8t = 1.4024 s

Y (mm)-5 0 5

t = 1.4040 s

t = 1.4007 s

T927880811738665519

T (K)

Z ( m m

)

2

4

6

8t =1.3900 s

Z ( m m

)

2

4

6

8

t = 1.4017 s t =1.4022 s

Y (mm)

Z ( m m

)

-5 0 5

2

4

6

8t = 1.4024 s

Y (mm)-5 0 5

t = 1.4040 s

t = 1.4007 s

)b()a(

0.5 m/s0.5 m/s

Z ( m m

)

2

4

6

8t =1.3900 s t = 1.4007 s

0.5 m/s

Z ( m m

)

2

4

6

8t = 1.4017 s 0.5 m/st =1.4022 s

0.5 m/s

Y (mm)

Z ( m m

)

- 5 0 5

2

4

6

8

t = 1.4024 s

Y (mm)- 5 0 5

t = 1.4040 s 0.5 m/s

(c)

Fig. 5. Front view of the formation of a crater: (a) weld pool (the darkest color) and weld bead (the second darkest color), (b) temperature eld, and (c) velocity eld.


8/9/2019 1-s2.0-S0017931009003925-main

9/14

er hardness than that of the base metal. At 653 K, the size of the b 00

phase and the amount of the b 0 phase reach the maximum and thestrength of the metal decreases to a minimum value [44] . This cor-responds to the lowest hardness near 6 mm. In the region between5.86 mm and 3.56 mm to the fusion line, where the temperature isbetween 653 K and 773 K, the dissolution of b 00 and b 0 occurs be-cause the precipitations are held at temperatures higher than thesolvus. The dissolution process enriches the solid solution of thealuminum matrix with alloying element Mg [45] . Therefore thiszone may undergo a solution-hardening heat treatment duringthe heating and cooling of the welding process, which contributesto the rise of local hardness. Thus a local hardness increase is foundat about 3.5 mm. Another contribution to the local hardness rise isthat during the post-weld natural aging (>3000 h), new precipi-tates are formed, which can be either GP zones [47] or b 00 phases.When the distance to the fusion line is less than 3.56 mm, the tem-

perature is higher than 773K. There areno precipitates in this zonebecause of the dissolution of b 00 and b 0 phases [45] . The aluminummatrix is therefore enriched with Mg. The possible reason for thehardness drop may be the diffusion of alloy element Mg betweenthe solid and liquid metals at the interface between the weld pooland solid metal. In the electrode material 4043, the Mg content isfar lower than in the base metal [40] . Since, during the weldingprocess, areas adjacent to the fusion zone undergo high tempera-tures, the diffusion of Mg may not be negligible, inducing thedepletion of Mg in this zone and consequently resulting in thehardness drop near the fusion line.

4.2. Crater lling

One approach to reduce or even eliminate the existence of solidication cracking is to maintain a sufcient amount of mol-

ten metal during the terminal stage of solidication, which can

ow into the cracking and thus ll and heal the cracks. To reducethe crater cracking of GMAW of aluminum alloys, it is desirable

Fig. 6. Zones near the fusion lineat the cross-section of the weld. (a) Zones near thefusion line and (b) base metal. Fig. 7. Comparison of the experimental and calculated cross-sections for a crater at

x = 20 mm. (a) Cracking at the crater center and (b) comparison of the results.

Table 3

Dimensions of cross-sections at x = 20 mm.

Normal crater Crater lling

Experiment Simulation Experiment Simulation

Pb , mm 1.83 a 1.89 1.97 a 2.03W b , mm 7.38 a 7.58 7.66 a 7.80Db , mm 0.83 a 0.92 R b , mm 0.57 a 0.51

a Average values are used for experimental results.b P, penetration; W, width; D, crater depth; R, reinforcement.

0.4 mm

hardness measurement line

weld bead

base metal

fusion line

Fig. 8. Knoop hardness measurement positions.


8/9/2019 1-s2.0-S0017931009003925-main

10/14

that the weld pool is kept longer at the nal stage of welding toallow enough liquid metal to ll the cracking. Therefore, a back-up crater lling technique was employed in this research tomaintain the weld pool and ll up the crater. The crater llingparameters are shown in Table 2 . When t < 1.4 s, the normalwelding parameters are used. When t > 1.4 s, a smaller set of parameters is used while the weld arc travel direction is reversedtowards the left-hand side. There is only heat input from thewelding arc after t = 1.4 s and no electrode material depositions.The crater lling procedure lasts for 0.6 s. The purpose is to llthe crater with small heat input to prevent unnecessary heat-af-fected zone (HAZ) softening. The change of welding current isshown in Fig. 10 .

Fig. 11 is the side views of the simulated crater lling processshowing the weld bead shape change, temperature eld, andvelocity distribution, respectively. When t < 1.4 s, the simulatedresults are the same as those of normal parameters. Aftert = 1.4s, the arc travels in the negative x-direction and the weld-ing current is reduced from 183 A to 83 A. In addition, there is nomaterial deposition onto the workpiece. Since there is no droplet

impingement and the arc pressure is also decreased signicantly,the force driving the uid to ow away from the arc center isvastly decreased. Therefore, under the inuence of gravity, the li-quid metal tends to ow back towards the center of the puddleand closes up the crater. At the beginning of the crater lling,e.g., at t = 1.4190 s, both uid on the left-hand side and right-hand side of the arc center ow back. After colliding with eachother at the center of the weld pool, the uid ows upwards.With the arc moving to the negative x-direction, e.g., att = 1.4670 s, the uid on the right-hand side of the arc solidiessince it receives insufcient heat input. Only the left-hand sideuid ows downward to close up the crater. Because of the iner-tia of the metal, the liquid continues to ow upward to the right-hand side after reaching the bottom of the crater. The molten me-tal owing to the right-hand side slows down near the right endof the weld pool, while at the same time at the left-hand side of the arc, downhill owing liquid material still has signicantmomentum. Therefore, some uid is pushed upward and pilesup, forming a higher uid level in the weld pool. Furthermore,as the arc proceeds to the left-hand side, e.g., at t = 1.7070 s, somepreviously solidied weld bead metal is re-melted, leading to anincrease of the size of the weld pool. While the arc continues tomove to the left-hand side, after a certain moment, the weld poolsize begins to decrease. The arc is turned off at t =2.0 s. Att = 2.0040 s, the weld pool has completely solidied.

Fig. 12 is the front views of the simulated crater lling processshowing the weld bead shape change, temperature eld, andveloc-ity distributions at x = 20 mm, respectively. When the welding pro-cess enters the crater lling procedure by employing low currentand no electrode material transportation, the driving force pushingliquid metal down at the arc center decreases signicantly. Aftert = 1.4 s, due to the continuous heat input, the weld pool doesnot solidify immediately. Instead, it decreases gradually in size asthe welding arc moves away from x = 20 mm. As the crater llingproceeds, e.g., at t = 1.4270 s, the liquid metal ows from theperiphery back to the center, lling up the crater. The uid velocityalso decreases signicantly. When the weld pool completely solid-ies, a at-top weld bead cross-section is formed, since there is noforce pushing the weld pool center downward.

To validate the numerical model, experiments were performedfor the crater-lling technique, the parameters of which areshown in Table 2 . The cross-section of the weld bead of thelled-up crater at x = 20 mm is compared with the simulated re-sults in Fig. 13 and Table 3. It is observed from Fig. 13 (a) thatthere is no cracking in the weld bead. As demonstrated inFig. 13 (b) and Table 3 , a good agreement in weld dimensions be-tween the experimental and calculated results was obtained. Mi-cro-hardness measurements were also conducted on crater-llingsamples. The results are shown in Fig. 14 . The hardness prole of the normal crater is also included for easy comparison. The peak

temperature along the hardness measurement line was calculatedin the model and is presented in Fig. 14 . Compared with the nor-mal crater, the positions corresponding to temperatures 653 K and773 K are 6.31 mm and 4.07 mm from the fusion line, respec-tively, instead of 5.8 mm and 3.56 mm. The wider high tempera-ture zones led to a generally lower hardness in the lled cratersince the coarsening of the b00 phase and its transformation to b0

precipitates are more serious when held at higher temperaturesfor a longer time. The lled crater hardness shows a similar trendto that of a normal crater. There is a signicant hardness dropwhen the distance from the fusion line is between 6 mm and7 mm. The lowest hardness is found around 6 mm. Beside thelow hardness zone and nearer to the fusion line, there is an in-crease in hardness, the peak of which is near 4 mm. And then

the hardness drops again toward the fusion line. The relationshipbetween the hardness prole and peak temperature and the

T ( K )

600

700

800

900

773 K

653 K

Distance from fusion line (mm)

H a r

d n e s s

( H K )

0 1 2 3 4 5 6 7

50

60

70

80

90

base metal hardness

hardness in the HAZ

3.55 5.88

Fig. 9. Knoop hardness measurement results and peak temperature along thehardness measurement line on cross-section at x = 20 mm for a crater.

Time ( s)

C u r r e n

t ( A m p

)

0 500 1000 1500 20000

100

200

300

400

500

183 Amp 82 Amp

Fig. 10. Welding current in a crater lling technique.


8/9/2019 1-s2.0-S0017931009003925-main

11/14

mechanisms are similar to those of the normal crater and will notbe repeated here.

5. Conclusions

The uid ow and heat and mass transfer in the weld pool for amoving GMAW of aluminum alloy 6005 were analyzed using the

VOF technique and the continuum formulation. Weld pool andweld bead shapes, temperature eld, and velocity distributionwere obtained for the terminating stage of the welding process.Experiments were conducted on the formation of the stoppingend of the weld. Metallurgical characterizations were performedon the welded samples. It was found that the crater is formed be-cause of the depression at the weld pool center as a result of

X (mm)

t = 2.0040 s

8 12 16 20

Z ( m m

)

5

10

t = 1.3900 s

X (mm)

Z ( m m

)

5

10

t = 1.7070 s

8 12 16 20

t = 1.4190 s

t = 1.4670 s

Z ( m m

)

5

10

t = 1.4270 s

Z ( m m

)

5

10

t = 1.3900 s t = 1.4190 s

Z ( m m

)

5

10

t = 1.4270 s

X (mm)

t = 2.0040 s

8 2012 16X (mm)

Z ( m m

)

5

10

t = 1.7070 s

8 2012 16

T

927880811738665519

t = 1.4670 s T (K)

)b()a(

Z ( m m

)

5

10

t = 1.3900 s 0.5 m/s

Z ( m m )

5

10

t = 1.4270 s 0.5 m/s

Z ( m m

)

5

10

t = 1.7070 s 0.5 m/s t = 2.0040 s 0.5 m/s

t = 1.4670 s 0.5 m/s

t = 1.4190 s 0.5 m/s

(c)X (mm)

8 12 16 20X (mm)

8 12 16 20

Fig. 11. Side view of the crater lling: (a) weld pool (the darkest color) and weld bead (the second darkest color), (b) temperature eld, and (c) velocity eld.


8/9/2019 1-s2.0-S0017931009003925-main

12/14

droplet impingement effect and arc pressure. The weld pool solid-ies very quickly once the weld process stops. Due to the rapidheat dissipation, there is no time for the molten metal to ow backtowards the weld pool center and close up the crater. Thus, a crateris formed at the end of the weld bead. Solidication cracking is alsoformed at the center of the weld crater due to the fast solidica-tion. To ll the crater andeliminate the cracking, a back-up crater

lling technique was proposed. During the crater lling stage, thewelding procedure switches from normal parameters to a smallercurrent for 0.6 s. At the same time, the welding arc moving direc-tion is reversed. As a result, the weld pool is maintained for a long-er time, and while its size decreases gradually, the crater issuccessfully lled and no cracking was found on the crater-llingexperimental results.

Y (mm)-5 0 5

t = 2.0040 s

Z ( m m

)

2

4

6

8t =1.3900 s t =1.4190 s

Z ( m m

)

2

4

6

8

t = 1.4270 s t =1.4670 s

Y (mm)

Z ( m m

)

-5 0 5

2

4

6

8t = 1.7070 s

Z ( m m

)

2

4

6

8

t = 1.4270 s

Y (mm)

Z ( m m

)

-5 0 5

2

4

6

8t = 1.7070 s

Y (mm)-5 0 5

t = 2.0040 s

Z ( m m

)

2

4

6

8t =1.3900 s t =1.4190 s

T927880811738665519

t =1.4670 sT (K)

)b()a(

Z ( m m

)

2

4

6

8t = 1.4270 s 0.5 m/s

t =1.4190 s 0.5 m/s

Y (mm)

t = 2.0040 s 0.5 m/s

t =1.4670 s 0.5 m/s

Z ( m m

)

2

4

6

8t =1.3900 s 0.5 m/s

Y (mm)

Z ( m m

)

- 5 0 5- 5 0 5

2

4

6

8t = 1.7070 st = 1.7070 s 0.5 m/s

)c(

Fig. 12. Cross-sectional view at x = 20 mm of the crater lling: (a) weld pool (the darkest color) and weld bead (the second darkest color), (b) temperature eld, and(c) velocity eld.


8/9/2019 1-s2.0-S0017931009003925-main

13/14

References

[1] H.L. Saunders, Welding Aluminum: Theory and Practice, 3rd ed., TheAluminum Association, 1997, pp. 1.29.5.

[2] P.B. Dickerson, Weld discontinuities causes and cures, Weld. J. 77 (6) (1998)3742.

[3] P. Dickerson, Quality control in aluminum arc welding, in: Proceedings of theAluminum Joining Seminar Aluminum Association, Washington, DC, LandmarkMotor Hotel, Metarie, Louisiana, 1986, pp. 331359.

[4] O. Runnerstam, K. Persson, The importance of a good quality gas shield,Svetsaren 50 (3) (1995) 2427.

[5] M. Pereira, C. Taniguchi, S. Brandi, S. Machida, Analysis of solidication cracksin welds of AlMgSi A6351 type alloy welded by high frequency pulsed TIGprocess, Q. J. Jpn. Weld. Soc. 12 (3) (1994) 342350.

[6] H. Kerr, M. Katoh,Investigation of heat-affectedzone cracking of GMA welds of AlMgSi alloys using the varestraint test, Weld. J. 66 (9) (1987) 251s259s.

[7] Z. Lu, W. Evans, J. Praker, S. Birley, Simulation of microstructure and liquationcracking in 7017 aluminum alloy, Mater. Sci. Eng. A 220 (1996) 17.

[8] M. Mizuno, S. Takeno, T. Teramoto, Y. Sakei, Relations between restraintintensity and weld cracking in welding of aluminium alloy, in: Proceedings of the Third International Conference Aluminium-Verlag, Duesseldorf, Munich,FRG, 1985, pp. I.5.1I.5.13.

[9] E.C.Partington, Control of metal transfer in modulatedpulseM.I.G.welding, in:IIW Asian Pacic Regional Welding Congress, 1988, pp. 970988.

[10] A.J. Sunwoo, E.L. Bradley III, J.W. Morris Jr., Effects of heat-affected zone peaktemperature on the microstructure and properties of 2090 Al alloy, Metall.Trans. A 21A (10) (1990) 27952804.

[11] L.A. Guittnerez, G. Neye, E. Zschech, Microstructure, hardness prole andtensile strength in welds of AA6013T6 extrusions, Weld. J. 75 (4) (1996) 115s121s.

[12] M.J. Lu, S. Kou, Power inputs in gas metal arc welding of aluminum part 1,Weld. J. 68 (9) (1989) 382s388s.

[13] A.O. Kluken, B. Bjorneklett, A study of mechanical properties for aluminumGMA weldments, Weld. J. 76 (2) (1997) 3944.

[14] T. Ma,G. Ouden, Heat-affected zone softening during arc welding of AlZnMgalloys, Int. J. Join. Mater. 8 (3) (1996) 105110.

[15] R.P. Martukanitz, C.A. Natalie, J.O. Knoefel, The weldability of an AlLiCualloy, J. Met. 39 (11) (1987) 3842.

[16] V.P. Budnik, Effect of the type of inert gas on pool temperature and fracture of the oxide lm in welding aluminium, Paton Weld. J. 6 (12) (1994) 2325.

[17] J. Domey, D.K. Aidun, G. Ahmadi, L.L. Regel, W.R. Wilcox, Numericalsimulation of the effect of gravity on weld pool shape, Weld. J. 74 (8)(1995) 263s268s.

[18] W.H. Kim, H.G. Fan, S.J. Na, Effect of various driving forces on heat and masstransfer in arc welding, Numer. Heat Transfer A 32 (1997) 633652.

[19] K. Hong, D.C. Weckman, A.B. Strong, The inuence of thermouids phenomenain gas tungsten arc welds in high and low thermal conductivity metals, Can.Metall. Q. 37 (34) (1998) 303393.

[20] Y. Wang, Q. Shi, H.L. Tsai, Modeling of the effects of surface-active elements onow patterns and weld penetration, Metall. Mater. Trans. B 32B (2) (2001)145161.

[21] R.T.C. Choo, J. Szekely, R.C. Westhoff, On the calculation of the free surfacetemperature of gas-tungsten-arc weld pools from rst principles. Part I.

Modeling the weld arc, Metall. Trans. B 23B (6) (1992) 357369.[22] R.T.C. Choo, J. Szekely, S.A. David, On the calculation of the free surface

temperature of gas-tungsten-arc weld pools from rst principles. Part II.Modeling the weld pool and comparison with experiments, Metall. Trans. B23B (6) (1992) 371384.

[23] Y. Wang, H.L. Tsai, Impingement of ller droplets and weld pool dynamicsduring gas metal arc welding process, Int. J. Heat Mass Transfer 44 (2001)20672080.

[24] S.-Y. Lee, S.-J. Na, A numerical analysis of a stationary gas tungsten welding arcconsidering various electrode angles, Weld. J. 75 (9) (1996) 259s269s.

[25] M. Ushio, C.S. Wu, Mathematical modeling of three-dimensional heat and uidow in a moving gas metal arc weld pool, Metall. Mater. Trans. B 28B (6)(1995) 509516.

[26] J. Jaidi, P. Dutta, Modeling of transport phenomena in a gas metal arc weldingprocess, Numer. Heat Transfer A 40 (2001) 543562.

[27] H. Park, S. Rhee, Analysis of weld geometry considering the transferringdroplets in gas metal arc welding, JSME Int. J. Ser. C 44 (3) (2001) 856862.

[28] M.H. Davies, M. Wahab, M.J. Painter, An investigation of the interaction of amolten droplet with a liquid weld pool surface: a computational andexperimental approach, Weld. J. 79 (1) (2000) 18s23s.

[29] V.K. Arghode, A. Kumar, S. Sundarraj, P. Dutta, Computational modeling of GMAW process for joining dissimiliar aluminum alloys, Numer. Heat TransferA 53 (4) (2008) 432455.

[30] J. Hu, H.L. Tsai, Heat and mass transferin gas metal arc welding. PartI. The arc,Int. J. Heat Mass Transfer 50 (2007) 833846.

[31] J. Hu, H.L. Tsai, Heat and mass transfer in gas metal arc welding. Part II. Themetal, Int. J. Heat Mass Transfer 50 (2007) 808820.

[32] J. Hu, H.L. Tsai, Effects of current on droplet generation and arc plasma in gasmetal arc welding, J. Appl. Phys. 100 (2006) 053304.

[33] J. Hu, H.L. Tsai, Metal transfer andarc plasma in gas metal arc welding, ASME J.Heat Transfer 129 (2007) 10251035.

[34] D.B. Kothe, R.C. Mjolsness, M.D. Torrey, Ripple: A Computer Program forIncompressible Flows with Free Surfaces, LA-12007-MS, Los Alamos NationalLaboratory, 1991.

[35] Q.Z. Diao, H.L. Tsai, Modeling of soluteredistribution in themushyzone duringsolidication of aluminumcopper alloys, Metall. Trans. A 24A (4) (1993) 963973.

[36] J.E. Hatch (Ed.), Aluminum: Properties and Physical Metallurgy, AmericanSociety for Metals, Metals Park, OH, 1984, pp. 1319.

Fig. 13. Comparison of the experimental and calculated cross-sections for a lledcrater at x = 20 mm: (a) center of the lled-up crater and (b) comparison of theresults.

T ( K )

600

700

800

900

773 K

653 K

Distance from fusion line (mm)

H a r d n e s s

( H K )

0 1 2 3 4 5 6 7

50

60

70

80

90

normal craterfilled crater

base metal hardness

4 .0 7 6 .3 1

Fig. 14. Knoop hardness measurement results and peak temperature along thehardness measurement line on x = 20 mm cross-section for a crater lling.


8/9/2019 1-s2.0-S0017931009003925-main

14/14

[37] J. Hu, H. Guo, H.L. Tsai, Weld pool dynamics and the formation of ripples in 3Dgas metal arc welding, Int. J. Heat Mass Transfer 51 (2008) 25372552.

[38] T. Zacharia, S.A. David, J.M. Vitek, Effect of evaporation and temperaturedependentmaterial properties on weld pool development, Metall. Trans. B 22B(2) (1992) 233241.

[39] S.V. Patankar, Numerical Heat Transfer and Fluid Flow, rst ed., Hemisphere,New York, NY, 1980.

[40] ASM, Properties and Selection: Nonferrous Alloys and Pure Metals, 9th ed,Metals Handbook, vol. 2, American Society for Metals, Metals Park, OH, 1985,p. 113.

[41] ASM, Metallography and Microstructures, 9th ed, Metals Handbook, vol. 9,American Society for Metals, Metals Park, OH, 1985, pp. 352354.

[42] S. Kou, Welding Metallurgy, Wiley, New York, 1987. p. 278.

[43] T. Enjo, T. Kuroda, Microstructure in weld heat-affected zone of AlMgSialloy, Trans. JWRI 11 (1) (1982) 6166.

[44] V. Malin, Study of metallurgical phenomena in the HAZ of 6061-T6 aluminumwelded joints, Weld. J. 74 (9) (1995) 305s318s.

[45] S.D. Dumolt, Metallurgical Transformations in the Heat-Affected Zone of Aluminum Alloys by Transmission Electron Microscopy, Carnegie-MellonUniversity, Pittsburgh, PA, 1983.

[46] S. Ceresara, E. Di Russo, P. Fiorini, A. Giarda, Effect of Si excess on theaging behavior of AlMg 2 Si 0.8% alloy, Mater. Sci. Eng. 5 (19691970)220227.

[47] A.H. Maitland, A. Ried, Metallurgical events in the heat affected zone of AlMgSialloys, in: International Aluminum Welding Conference, Cleveland, OH, 1981,pp. 106114.


Documents

1-s2.0-S0017931009003925-main