See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/317376637
Dissimilar steels laser welding: Experimental and numerical assessment of
weld mixing
Article in Journal of Laser Applications · May 2017
DOI: 10.2351/1.4983168
CITATIONS
READS
4
234
5 authors, including:
Alexandre Métais
Simone Matteï
ArcelorMittal Atlantique et Lorraine
University of Burgundy
5 PUBLICATIONS 8 CITATIONS
70 PUBLICATIONS 901 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Assemblage Acier-Aluminium par Faisceau Laser (A3FL) View project
Modelling of dissimilar steels laser welding View project
All content following this page was uploaded by Alexandre Métais on 22 June 2017.
The user has requested enhancement of the downloaded file.
SEE PROFILE
Dissimilar steels laser welding: Experimental and numerical assessment of weld
mixing
Alexandre MétaisSimone Matteï, Iryna Tomashchuk, and Eugen CicalaSadok Gaied
Citation: Journal of Laser Applications 29, 022420 (2017); doi: 10.2351/1.4983168
View online: http://dx.doi.org/10.2351/1.4983168
View Table of Contents: http://lia.scitation.org/toc/jla/29/2
Published by the Laser Institute of America
Articles you may be interested in
Laser cladding and wear testing of nickel base hardfacing materials: Influence of process parameters
Journal of Laser Applications 29, 022504 (2017); 10.2351/1.4983160
JOURNAL OF LASER APPLICATIONS
VOLUME 29, NUMBER 2
MAY 2017
Dissimilar steels laser welding: Experimental and numerical assessment
of weld mixing
tais
Alexandre Me
Laboratoire Interdisciplinaire Carnot de Bourgogne (ICB), UMR 6303 CNRS/Universit
e de Bourgogne
Franche-Comt
e, 12 rue de la fonderie, 71200 Le Creusot, France and ArcelorMittal, Global Research and
Development, 1 route de Saint Leu, 60761 Montataire, France
Simone Matte€ı, Iryna Tomashchuk, and Eugen Cicala
Laboratoire Interdisciplinaire Carnot de Bourgogne (ICB), UMR 6303 CNRS / Universit
e de Bourgogne
Franche-Comt
e, 12 rue de la fonderie, 71200 Le Creusot, France
Sadok Gaied
ArcelorMittal, Global Research and Development, 1 route de Saint Leu, 60761 Montataire, France
(Received 28 March 2017; accepted for publication 28 March 2017; published 6 June 2017)
Upcoming strict CO2 regulations lead car manufacturers to look for mass saving solutions. The use
of advanced high strength steel (AHSS) solutions enable optimizing both crash performances and
mass saving. Particularly, the use of laser welded blanks made of dissimilar high strength steels is
an efficient weight optimization solution. To support the joining of AHSS in car body design, a 3D
model of heat transfer, turbulent flow and transport of species in the laser weld pool has been
developed. It aims at providing a better understanding of diffusive-convective mixing in the weld
and its influence on the weld mechanical properties. The presented model allows predicting the
weld geometry and the element distribution. To validate the model, experimental tests were carried
out. Welding of two dissimilar steels with different laser beam offset from the joint line was performed. Numerical and experimental investigations of dissimilar butt laser welding between high
Mn and dual phase steels were carried out. The cross sections of the welds were characterized by
scanning electron microscope (SEM) with energy-dispersive X-ray spectroscopy (EDX) elemental
analysis. Quantitative mappings of Mn distribution in the melted zone offer an overview of mixing
intensity. The results of the simulation have been found in good agreement with the experimental
data. To go further and to assess the effect of weld mixing on mechanical performances, tensile
tests were done. It was found that tensile behavior of the welds is determined by level of Mn and C
dilutions. For attaining maximal joint performances, it is necessary to comprehend the elements
distribution in the melted zone and to be able to control it through an accurate choice of operational
C 2017 Laser Institute of America. [http://dx.doi.org/10.2351/1.4983168]
parameters. V
Key words: laser welding, dissimilar steel joint, multiphysical modeling, melted pool mixing,
mechanical properties
I. INTRODUCTION
Due to its performances, flexibility, and high production
rates, laser welding is more and more used in the industry
and particularly in automotive applications. The main benefits for laser welded blanks are mass and cost savings. Mass
savings are achieved through the higher strength level–lower
thicknesses combinations. When welding dissimilar steel
grades, the weld composition depends on the parent steel
grades and on the welding parameters. A 3D modeling of
butt laser welding allows to clarify the mechanism of weld
formation and to attain important savings of cost and time
during the determination of optimal operational parameters.
Indeed, the welding parameters impact the alloying elements
dilution in the weld and the dilution rate influences the
mechanical properties of the weld. Few numerical investigations of laser welding of dissimilar combinations have been
proposed by the scientific community over the last 10 years.
In 2007, Chakraborty and Chakraborty1 discussed the role of
1938-1387/2017/29(2)/022420/9/$28.00
turbulence in 3D modeling of conduction mode laser dissimilar welding of Cu-Ni. Tomashchuk et al. in 2010 (Ref. 2)
showed that heterogeneous welding is more complicated
than homogeneous welding because of the great difference
in thermophysical properties of welded materials. Then in
2012, Hu et al.3 developed a model studying heat and mass
transfer in dissimilar laser welding of a stainless steel and
nickel and concluded that the mass transport is strongly
dependent of convection. In 2015, Esfahani et al.4 investigated the microstructure and service performance of dissimilar joint between a low carbon steel and an austenitic
stainless steel. They showed that the alloying element concentration has a significant influence on the microstructure
and service performance of the weld. In 2014, Behm et al.5
studied the dissimilar high manganese (TWIP)/low carbon
steels laser welding configuration for automotive applications. They demonstrate the importance of identifying process windows to attain high reliable welds. Metallurgical
properties of dissimilar welding highly depend on the laser
022420-1
C 2017 Laser Institute of America
V
022420-2
tais et al.
Me
J. Laser Appl., Vol. 29, No. 2, May 2017
welding process. More recently, Behm et al.6 have developed an experiment based on the Charpy impact test in order
to evaluate the mechanical properties of the dissimilar joints.
They showed the role of the weld metallurgy on mechanical
strength.
In the present paper, the 3D multiphysical model was
developed and applied to the prediction of the alloying element
distribution in the melted zone formed during laser welding
between the dual phase steel (DP) and TWIP steel. The temperature field, velocity field, and manganese concentration
map were studied for various beam offsets from the joint line.
In parallel, a number of welding experiments were carried out
to provide the basis for model validation and to clarify the
effect of melted zone composition on tensile properties.
II. EXPERIMENTAL DESIGN
A continuous Yb:YAG laser welding of 1.5 mm thick
sheets of DP and TWIP (Table I) was performed in the butt
configuration with a beam power of 4 kW and a welding
speed of 6 m/min. A 600 lm laser spot was focused on the
top surface of the joint. These processing conditions led to
deep-penetration welding, when laser energy is transferred
via bulk absorption at the walls of a vapor capillary (a keyhole).7 To assess the mixing in the molten zone, two steel
grades with large difference in terms of chemical composition were selected. These steels are DP and TWIP steels (see
Table I). Furthermore, several welding conditions were
tested. The laser beam was perpendicular to the sheet surface
and different laser beam offsets from the joint line were
tested: 200 lm offset to DP steel (weld 1, Fig. 1(a)), zero
offset (weld 2, Fig. 1(b)) and 200 lm offset to TWIP steel
(weld 3, Fig. 1(c)).
The experimental tests were performed through several
phases. The cross sections of the welds underwent polishing
and etching with Bechet-Beaujard solution, optical microscopy, and Vickers microhardness measurements with 100 g
load. The energy-dispersive X-ray spectroscopy (EDX) element analysis was performed with a JEOL JSM-6610LA field
emission scanning electron microscope (SEM) equipped with
a JED-2300F EDX system. Three cross sections per weld condition were studied; average values of melted zone dimension
and Mn concentration were used for validation of the numerical model. Finally, the welds underwent tensile test with deformation speed of 1 mm/min and fracture surfaces were
subjected to X-ray diffraction (XRD) analysis (PANalytical
X’Pert PRO using Co target) and examination by SEM.
III. EXPERIMENTAL WELDABILITY OF DP/TWIP STEEL
COMBINATION
FIG. 1. Cross section of butt joints with different offsets. (a) Weld 1:
200 lm offset toward DP; (b) weld 2: zero offset joint; (c) weld 3: 200 lm
offset toward TWIP.
mismatch in physical properties of welded couple, the dilution rate is highly dependent on the laser beam offset (see
Fig. 1 and Table II). For example, the liquidus temperature
of the TWIP steel having 22 wt. % Mn content is 150 K
TABLE II. Materials dilution and average composition of alloying elements
for different welding compositions.
For all tested operational conditions, fully penetrated
and defect-free welds were obtained. Because of important
Average composition,wt. %
Weld
TABLE I. Chemical composition of DP, and TWIP steels used in this study.
Wt. %
C
Mn
Si
Cr
Fe
DP
TWIP
0.14
0.601
1.4
22.0
0.2
0.2
0.4
0.1
Balance
Balance
1
2
3
a
DP steel dilution
Mna
Cb
0.50
0.38
0.12
10
16
21
0.22
0.40
0.52
EDS analysis.
Estimated from dilution of two steels.
b
tais et al.
Me
J. Laser Appl., Vol. 29, No. 2, May 2017
022420-3
FIG. 3. XRD diagrams of the welds for 1, 2, and 3 weld configurations.
FIG. 2. Metastable phase diagram of the Fe-Mn-C system, quenching from
700 C to ambient temperature (Ref. 8).
lower compared to the DP steel. In consequence, the joint
with centered beam position (configuration 2, Fig. 1(b)) contains only 38% of melted DP. On the other hand, in the case
of 200 lm laser beam offset toward DP (configuration 1, Fig.
1(a)), 50% of each metal is melted. 200 lm laser beam offset
toward TWIP (configuration 3, Fig. 1(c)) leads to a DP dilution of only 12% against 88% TWIP.
DP steel showed the ferrite-martensite microstructure
and the TWIP steel exhibited a fully austenitic microstructure. According to the metastable Fe-Mn-C phase diagram
(Fig. 2), after quenching of mixed steels, two different martensite structures may form along with c-Fe:a0 martensite
(body-centered tetragonal structure with a lattice parameter
varying with the carbon content from 0.287 to 0.300 nm) and
e martensite (hexagonal close packed structure with lattice
parameters a ¼ 0.2538 nm and c ¼ 0.4080 nm).8 Considering
the high cooling rate of laser welding process, the hypothesis
that weld presents a quenched microstructure is assumed.
The average composition of the melted zones (Table II)
allows supposing that c-Fe will be the main phase in weld 3.
Weld 2 will contain c-Fe and e martensite and weld 1 both a0
and e martensite phases along with c-Fe.
To confirm the presence of these phases in the weld,
XRD analyses were performed. Analysis of weld surface of
welds 1, 2, and 3 revealed the important differences in the
microstructure (Fig. 3): weld 1 sample shows the presence of
c-Fe phase and important quantity of a0 martensite, when
welds 2 and 3 present c-Fe and small peaks attributed to the
e martensite phase.
Tensile test showed that the difference in weld composition has a significant impact on the fracture mode and ultimate tensile strength (UTS) (Table III). The dissimilar weld
with laser offset of 200 lm toward DP steel (weld 1) failed
in the middle of the melted zone and exhibited the lowest
UTS. Murica et al. showed in Ref. 9 that bands of a0 martensite are responsible for rupture in the fusion zone formed
between TWIP and transformation induced plasticity (TRIP)
steels. Here, the same effect was observed. According to FeC-Mn metastable diagram (Fig. 2), weld 1, which fractures
in the fusion zone, exhibits a0 martensite phase in the weld.
Welds 2 and 3, having higher Mn concentration and thus no
a0 martensite, resisted to tensile test: fracture took place in
DP far away from the weld.
Vickers’ microhardness measurements of the joint (1)
also indicate important hardening of the melted zone comparing to two other conditions, which can be attributed to the
formation of a0 martensite (Fig. 4). For welds 2 and 3, maximal hardness was observed in HAZ at DP side, where the
quantity of a0 martensite rises under the effect of the laser
welding thermal cycle.
To resume, the tensile properties of welded joints
depend highly on Mn and C content. The prediction of local
chemical composition in the weld formed between dissimilar
steels as a function of the welding parameters opens a real
perspective of improving welding process conditions.
IV. MODEL DESCRIPTION
The 3D model has been developed to simulate butt welding of dissimilar sheets of steel with a laser beam (Fig. 5).
Heat transfer, flow and transport of diluted species problems
are solved with COMSOL MultiphysicsV 5.1. A number of
assumptions have been made to develop the following tridimensional model in order to reduce the computation time:
R
•
•
•
a steady keyhole with a conical geometry;
heat source as limit condition of temperature inside the
keyhole. Temperature is assumed to be uniform
Tkeyhole ¼ Tvap;
top and bottom surfaces of the weld are assumed to be flat;
TABLE III. Tensile properties and spatial localization of the fractures.
200 lm on DP (1)
Zero offset (2)
Ultimate tensile strength
610 MPa
660 MPa
Fracture location
Fusion zone
Base material (DP side)
Ultimate tensile strength of base materials: UTSDP ¼ 634 MPa and UTSTWIP ¼ 1100 MPa
200 lm on TWIP (3)
660 MPa
Base material (DP side)
022420-4
tais et al.
Me
J. Laser Appl., Vol. 29, No. 2, May 2017
FIG. 6. Coarse mesh (a) and fine mesh (b).
FIG. 4. Vickers microhardness line profiles on weld cross sections.
•
•
•
equations are strongly coupled;
liquid metal is assumed to
incompressible;
quasisteady approach is used.
be
Newtonian
and
A. Geometry and mesh
The keyhole is represented by a cone-shaped geometry;
it is sized to obtain a weld geometry in good agreement with
the experimental weld. The sheet thickness is conserved
(1.5 mm). Two meshes are used for the calculation; a coarse
mesh (Fig. 6(a)) to obtain quickly a “good” initial condition
and a fine mesh (Fig. 6(b)) to solve the whole of equations
using the previously found solution. A representative volume
of the welded blanks is meshed. Tetrahedral mesh of 60 lm
maximal size was additionally refined around the keyhole.
B. Material properties
The material properties are defined as the functions of
temperature and manganese concentration. Smoothed
Heaviside step functions are employed to smooth sharp
changes of material properties at the boundary of dissimilar
metals and in the phase change zone (an example in Fig. 7).
The material properties are given in Table IV.
FIG. 5. Full penetrated dissimilar laser welding.
One of the main difficulties of numerical simulation of
welding resides in a lack of data on materials properties,
especially in the liquid state. Moreover, several input parameters related to the keyhole cannot be determined with high
precision (like keyhole’s temperature, top and bottom diameters and angle of inclination). To quickly optimize these
parameters, the experimental design method was used.10
After defining the variation intervals for every uncertain
parameter, the estimation of their impact on target functions
(here, melted zone dimensions and Mn concentration) using
factionary plans 215–11 and 23 allows to quickly converge
toward experimental results. The factorial experiments’
design allows significant reduction of a number of numerical
experiments compared to full plans. In the present case, 24
parameters were studied through only 35 numerical experiments. It was concluded that only few parameters in the
model have a great impact on objectives; the material properties at the liquid state first and at the solid state then, the
heat source parameters and Marangoni coefficients are the
most influencing parameters in this simulation. An important
result is that the material properties at the liquid state and
Marangoni coefficients are, in the present model, very influent parameters on target functions.
The preliminary model optimization study has been performed in a centered beam configuration (configuration 2).
Then, the obtained set of parameters was applied to both
configurations with beam offset from the joint line. The
closeness of obtained results with experimental data confirmed the robustness of the model.
C. Governing equations
In this model, all equations are strongly coupled and
solved by a segregated solver. In segregated resolution,
direct PARDISO solvers are used to calculate variable T and
FIG. 7. Thermal conductivity of the DP steel with Heaviside step function.
tais et al.
Me
J. Laser Appl., Vol. 29, No. 2, May 2017
022420-5
TABLE IV. Material properties.
Property
Fusion temperature
Heat of fusion
Heat capacity
Density
Thermo-density coefficient
Thermal conductivity
Viscosity
Symbol (unit)
DP steel [solid, liquid]
TWIP steel [solid, liquid]
Tf (K)
Lf (kJ Kg1)
Cp (J kg1. K1)
q (kg m3)
b (K1)
k (W m1 K1)
l (Pa s)
1798
270
[500, 600]
7800/(1 þ b T)
15 106
[30, 45]
[100, 0.003]
1648
270
[550, 650]
7500/(1 þ b T)
20 106
[40, 50]
[100, 0.003]
an iterative GMRES solver is used to calculate fluid flow and
transport species variables. The use of iterative solver allows
reducing the use of RAM.
1. Heat transfer
The energy equation solved with convection term provides
the temperature field over the entire domain. Because of the
quasisteady approach, the velocity field u has a constant term V
in the welding direction added at the calculated velocity field
qCp u rT þ rðkrTÞ ¼ 0;
(1)
where q is the density, Cp is the heat capacity, k is the conductivity, T is the temperature, and u is the velocity field.
The fusion is taken into account by using the equivalent
enthalpy method
Cp ðTÞ ¼ Cp ðTÞ þ d : Lf ;
(2)
where CP ðTÞ is the heat capacity function of temperature, Lf
is the melting latent heat, and d is the Gaussian distribution
defined as
d¼
TTf
1
pffiffiffi : e DT ;
DT: p
(3)
where DT is the temperature range of 100 K and Tf is the
melting point of materials.
qðu : rÞk ¼ r : ðl þ lT rk Þrk þ pk b0 qxk;
(6)
x
qðu : rÞx ¼ r : ðl þ lT rx Þrx þ a pk qb0 x2 ;
k
(7)
k
;
x
h
i
pk ¼ lT ru:ðru þ ðruÞT Þ :
lT ¼ q
(8)
(9)
The moving solid-liquid interface is modeled through
changing the viscosity of the fluid. Dynamic viscosity is temperature dependent: solid is assumed as an equivalent liquid
with a viscosity of 100 Pa s.
3. Fick law for diluted species
To study the transport of species in the weld pool, the
Fick law has been used
r ðDi rCi þ uCi Þ ¼ 0;
(10)
where Di the diffusion coefficient of the element i is defined
by the following equation:
Di ðT Þ ¼ Dthermal ðT Þ þ Dturbulent ðT Þ
pffiffiffi
kB T
2
¼
T :
þ
2
6pri l
(11)
Density, conductivity, and heat capacity are considered
as dependent on Mn content through the following equation:
2. Fluid flow
The Reynolds-Averaged Navier-Stokes turbulent model
k x has been used in order to take into account turbulent
mixing caused by eddy diffusivity. The k x model is useful in many cases such as internal flows, flows that exhibit
strong curvature, separated flows, and jets. The governing
equations for mass conservation, momentum, and energy
transport in steady-state formulation are as follows as they
are implemented within the simulation framework of
COMSOL MultiphysicsV 5.1. The following equations are
about the k x model:
ðMnTWIP Mn Þ
þ fTWIP
ðMnTWIP MnDP Þ
ðMn MnDP Þ
;
ðMnTWIP MnDP Þ
f ¼ fDP (12)
where f can be q, k, and CP.
R
qr ðuÞ ¼ 0;
h
i
qðu : rÞu ¼ r : pI þ ðl þ lT Þðru þ ðruÞT Þ
qg þ FM ;
(4)
(5)
4. Boundary conditions
Energy supply is introduced through imposing uniform
temperature on the keyhole walls. Symmetric condition is
applied on lateral walls. The heat is exchanged between the
weld plates and environment, consequently heat losses due
to convection and radiation are taken into account
022420-6
tais et al.
Me
J. Laser Appl., Vol. 29, No. 2, May 2017
TABLE V. Boundary conditions.
Heat transfer
Keyhole
ABCD
BCFG - ADHE
ABFE - DCGH - EFGH
Turbulent flow
Keyhole - BCGF ADHE - ABFE - CDHG
ABCD
EFGH
Transport of diluted species
Keyhole - BCGF ADHE - ABFE
- CDHG
ABCD
EFGH
T ¼ Tvap
T ¼ Tambiente
~ T :~
kr
n ¼ hg ðT T0 Þ
~
n :q ¼ 0
u :~
n¼0
u ¼ u0 ; k ¼ k0 ; x ¼ x0
½pI þ ðl þ lT Þðru þ ðruÞT Þ
23 ðl þ lT Þðr: uÞI 23 qKI ~
n
n ¼ p^0 ~
~
n : ðDMn rMn þ u : MnÞ ¼ 0
FIG. 8. Boundary conditions of the numerical domain.
Mn ¼ Mn0
~
n : ðDMn rMn Þ ¼ 0
throughout a global heat exchange coefficient on top and bottom surfaces.
Inflow condition at welding velocity is imposed on the
front surface and outflow condition at constant pressure on
the back surface. Sliding wall condition is applied to all surfaces. Due to the variation of the surface tension coefficient
with temperature, the Marangoni effect, which is one of the
driving forces in the weld pool, is taken into account on top
and bottom surfaces
c ¼ c0 þ
dc
ðT T f Þ:
dT
(13)
Full-penetrated welding produces a high velocity jet of
gas out of the keyhole. In the literature, a velocity of the
metallic vapor gas, named plume, higher than 100 m s1 can
be found.11 The shear stress induced by the interaction of the
plume on the liquid metal cannot be neglected and has been
introduced to surfaces of the keyhole as a weak contribution.
This phenomenon is still misunderstood and in this model
the shear stress is assumed to be a constant. Considering that
the flow inside the vapor plume is laminar, the DarcyWeisback equation gives an order of magnitude for the shear
stress between the liquid metal and the vapor plume
1
sp ¼ f qp Vp2 ;
8
temperature coefficient of surface tension. In this case, liquid
metal flows from the higher-temperature keyhole boundary
to the lower-temperature molten pool boundary and leads to
the expansion of the fusion zone at the top and the bottom
surfaces (Fig. 9). Taking into account the hourglass shape of
weld cross section, three characteristic widths of the melted
zone were chosen to control the concordance of calculated
and experimental solidus line: on top (L1), in the middle
(L2), and on the bottom (L3).
A relative error <10% has been obtained (Table VI)
between experimental and numerical weld measurements of
the joint with zero beam offset. This good agreement between
experiments and simulation results is due to an optimization of
the tridimensional model by factorial experiments’ design. In
two other configurations, where a lateral shift of 6200 lm was
used, maximal relative error increases up to 22%. Numerical
modeling is able to return results in good agreement with
experiments without any parameters adjustment. This demonstrates a good robustness of the model regarding the high number of dissimilar joint that can be made.
B. Flow field in the weld pool
Calculated velocity fields illustrate that eddies are
formed near the top and bottom surfaces due to Marangoni
convection (Fig. 10). It can be noticed that the velocity in the
weld pool is higher in these eddies, in particular, next to top
(14)
where f ¼ 64/Re for a supposed laminar flow of the plume. A
velocity of the plume VP ¼ 100 m s1 gives a shear stress of
50 N m2. This value is currently used in our numerical
models without experimental validation.
Boundaries conditions are summed up in Table V where
surfaces are described in Fig. 8.
V. RESULTS AND DISCUSSION
A. Weld geometry
An hourglass weld shape is generally observed for fullpenetrated laser welds, which corresponds to a negative
FIG. 9. Weld width measurement on three characteristic lines (sample 1).
tais et al.
Me
J. Laser Appl., Vol. 29, No. 2, May 2017
022420-7
TABLE VI. Comparison between experimental and calculated dimensions
of melted zones. Boldface indicates the relative error between experimental
and calculated values.
Dimensions (lm)
L1
L2
L3
Exp
Calc
d(%)
Exp
Calc
d(%)
Exp
Calc
d(%)
200 lm on DP (1) Zero offset (2) 200 lm on TWIP (3)
1002
779
22
681
735
28
1123
1019
9
783
733
6
655
724
210
974
913
6
745
788
6
664
780
18
995
933
6
and bottom surfaces. Thermocapillary effects accelerate the
fluid up to 0.7 m s1 (i.e., seven times the laser welding
velocity). Because of the important mismatch in thermal
FIG. 11. Stream lines in a longitudinal section having maximal melted zone,
color scale is the velocity magnitude (m s1), in red arrow the velocity field.
200 lm offset on DP (1), 0 lm offset (2), and 200 lm offset on TWIP (3).
properties of welded steels, asymmetric flow field in the
weld pool is observed (Fig. 10). The highest velocity magnitude has been found near the bottom surface and close to the
TWIP steel. Weld geometry is highly dependent on the
Marangoni effect; the maximum weld width (Table VI) has
been obtained for weld (1) near surfaces where the flow is
the most accelerated.
In longitudinal observation, two eddies can be observed
in the y-z plane just behind the keyhole (Fig. 11). The plume,
a high velocity jet of gas, generates a strong shear stress and
sets the liquid in motion. These eddies seem to be the main
driving force for mixing in z direction. The vortex size and
the velocity field inside the vortex depend on this plume.
The highest velocity of 1.2 m s1 is attained in the bottom
surface where plume shear stress and the Marangoni effect
accelerate the fluid together. Because of a lower melting
point of the TWIP, the weld pool length increases with a
laser beam offset on TWIP (Fig. 11(3)).
C. Weld chemistry
FIG. 10. Stream lines in a cross section having maximal melted zone, color
scale is the velocity magnitude (m s1), in black the melting point and red
arrow the velocity field. 200 lm offset on DP (1), 0 lm offset (2), and
200 lm offset on TWIP (3).
The chemical composition plays a major role in microstructure and mechanical properties of the joints. The challenge in modeling is to validate numerical flow field with the
help of experimental observations. Post-mortem X-mapping
of alloying element (here, Mn) is used to conclude on matter
transport mechanisms. Weld chemical composition is found
homogeneous in global. Thus, the manganese mass fraction
in the weld depends mainly on the dilution rate of each steel.
Because of very different thermal properties (melting point,
thermal conductivity, etc.), the dilution rate is very dependent on laser beam offset (Table VII).
Numerical results always show smoother profile compared to experimental results (Fig. 12). In fact, the mesh size
is too large to simulate heterogeneity at microscale. For macroscale, the numerical results have been found in good agreement with experimental data with the k x model (Fig. 12).
For all cases considered, a good representation of alloying
element distribution is observed.
022420-8
tais et al.
Me
J. Laser Appl., Vol. 29, No. 2, May 2017
TABLE VII. Average Mn content in cross section.
200 lm
on DP (1)
Zero
offset (2)
Experimental Mn content Numerical 10.1 wt. % 16.2 wt. %
Mn content Relative error
10.0 wt. % 15.7 wt. %
3%
1%
200 lm
on TWIP (3)
21.0 wt. %
19.2 wt. %
8%
Manganese mass fractions (wt. %) found numerically
are sometimes higher than the experimental one in the weld
(Fig. 12). It may be explained by the manganese evaporation.
Mujica et al.9 found by EDX analysis, of an homogeneous
joint of Fe-25Mn-3Al-3Si steels, a decrease of the manganese content in the fusion zone up to 2.5 at. %.
VI. CONCLUSION
In order to provide a better understanding of diffusiveconvective mixing of alloying elements during keyhole laser
welding, a 3D multiphysical model of heat transfer, turbulent
fluid flow, and transport of species is proposed.
Immediate application of this model consists in prediction of global and local composition of the melted zone
formed between the dual phase steel and the TWIP steel. An
associated experimental study showed that proper choice of
operational parameters allows maintaining high Mn content
in the melted zone and thus stabilize c-Fe matrix. It was
found that tensile properties of DP/TWIP joints are highly
affected by accumulation of brittle a-martensite that can be
completely avoided by proper choice of laser beam offset
from the joint line.
The developed multiphysical model shows a good
robustness to the variation of beam offset from the joint line.
A maximal relative error <22% has been obtained between
experimental and calculated dimensions of melted zones performed with different offsets. Velocity field in the weld pool
has been found highly dependent in plume shear stress and
Marangoni convection. Maximal velocity magnitude is
attained in the flow surface just behind the keyhole where
the plume shear stress and Marangoni effect act together.
Calculated average Mn content in melted zone is found in
good agreement with EDX analysis. Maximal relative error
<8% is observed. Calculated concentration maps allow following the Mn distribution in weld cross sections; however,
it remains smoothed compared with experimental Xmapping.
In perspective, this model will be applied for optimizing
laser welding of various dissimilar combinations of novel
steels, in view of identification of optimal operational parameters that allow high profitability (high process rate) and
good tensile properties (avoiding the formation of brittle
phases).
FIG. 12. Experimental and numerical distributions of alloying elements (wt. %) along the line L1, L2, and L3. 200 lm offset on DP (1), 0 lm offset (2), and
200 lm offset on TWIP (3).
tais et al.
Me
J. Laser Appl., Vol. 29, No. 2, May 2017
1
N. Chakraborty and S. Chakraborty, “Modelling of turbulent molten pool
convection in laser welding of a copper–nickel dissimilar couple,” Int. J.
Heat Mass Transfer 50, 1805–1822 (2007).
2
I. Tomashchuk, P. Sallamand, J., Jouvard, and D. Grevey, “The simulation
of morphology of dissimilar copper–steel electron beam welds using level
set method,” Comput. Mater. Sci. 48, 827–836 (2010).
3
Y. Hu, X. He, G. Yu, Z. Ge, C. Zheng, and W. Ning, “Heat and mass transfer in laser dissimilar welding of stainless steel and nickel,” Appl. Surf.
Sci. 258, 5914–5922 (2012).
4
M. Esfahani, J. Coupland, and S. Marimuthu, “Numerical simulation of
alloy composition in dissimilar laser welding,” J. Mater. Process. Technol.
224, 135–142 (2015).
5
V. Behm, M. H€ofemann, A. Hatscher, A. Springer, S. Kaierle, D. Hein, M.
Otto, and L. Overmeyer, “Investigations on laser beam welding dissimilar
material combinations of austenitic high manganese (FeMn) and ferrite
steels,” Phys. Procedia 56, 610–619 (2014).
View publication stats
6
022420-9
V. Behm, G.-A. Hoffmann, J. V. Trzebiatowski, A. Springer, S. Kaierle,
and L. Overmeyer, “Method for evaluating the behavior of laser welded
dissimilar material seams in single overlap configuration under impact
loads by the example of twinning induced plasticity steels,” J. Laser Appl.
27(2), S29006 (2015).
7
R. Fabbro, S. Slimani, F. Coste, and F. Briand, “Study of keyhole behaviour for full penetration Nd–Yag CW laser welding,” J. Phys. D. Appl.
Phys. 38, 1881–1887 (2005).
8
H. Schumann, “Distribution of phases in Fe-Mn-C system after
deformation,” Neue Hutte 17, 605–609 (1972).
9
L. Mujica, S. Weber, and W. Theisen, “Welding of twinning-induced plasticity steels,” Scr. Mater. 66, 997–1001 (2012).
10
E. Cicala, Metoda experimentelor factoriale: proiectarea experimentelor,
modelare, optimizare, Editura Politechnica (Timisoara, Romania, 2005).
11
E. Amara and A. Bendid, “Modelling of vapour flow in deep penetration
laser welding,” J. Phys. D: Appl. Phys. 35, 272–280 (2002).