www.elsevier.com/locate/commatsci
School of Material Science and Engineering, Beijing Polytechnic University, 100 Ping Le Yuan, Chaoyong District,
Beijing 100022, People's Republic of China
b
Joining and Welding Research Institute of Osaka University, Ibaraki, Osaka 567, Japan
Abstract
A one-domain mixture continuum model is introduced to simulate numerically solid/liquid phase transformation
with a mushy region in laser surface remelting process of a type 304 stainless steel. Emphasis is given to the competitive
inuence of laser-induced alloying element vaporization and Marangoni ow on the heating surface maximum temperature and its distribution as well as the molten pool shapes. The molten pool shapes and uid ow, the temperature
distribution and its peak values on the heating surface have been computed for six calculation cases corresponding to
six dierent pool-surface heat ux balances. The results show that the Langmuir-type vaporization heat loss due to Fe,
Mn, Cr, Ni elements can signicantly reduce the heating surface peak temperature and the pool prole, while jor=oT j is
small or the buoyancy force is considered as one and only driving force. However, when vaporization-type heat loss and
Marangoni ow are coexisting, the free surface temperature distribution and its peak value are markedly aected by the
magnitude order of surface tension gradient and the nature of the relationship between surface tension temperature
coecient and temperature. When jor=oT j > 3:0 10 4 N/m K, the inuence of Langmuir-type vaporization heat loss
is very small and can be ignored. 2001 Elsevier Science B.V. All rights reserved.
Keywords: Laser surface remelting; Marangoni ow; Temperature distribution; Fluid ow; Molten pool shape; Vaporization; Stainless
steel
1. Introduction
During laser surface remelting of many important engineering alloys, the transient evolution of
temperature and uid ow elds in molten pool,
and then the pool shapes, are evidently aected by
*
Corresponding author. Tel.: +86-10-673-92523; fax: +86-10673-92523.
E-mail addresses: yplei@bjpu.edu.cn (Y.P. Lei), shyw@
solaris.bjpu.edu.cn (Y.W. Shi).
0927-0256/01/$ - see front matter 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 7 - 0 2 5 6 ( 0 1 ) 0 0 1 4 3 - 4
Nomenclature
ai
activity of element i
C
heat capacity
e
emissivity
f
mass fraction
g
gravitational acceleration or liquid and
solid volume fraction
h
sensible enthalpy
hc
convection heat transfer coecient
k
thermal conductivity
K0
permeability coecient
L
latent heat of fusion
Lr ; Lz radius and thickness of the workpiece
Lvap:i heat of vaporization of species i
Mi
atomic weight of species i
p
pressure
pi
partial pressure of species i
pi0
equilibrium vapor pressure of species i
qlaser heat transfer ux from laser
qnet net heat input into workpiece
qloss total heat loss from molten pool surface
r; z
radial and axial coordinates
277
rb
R
t
T
Ta
u; v
Greek
b
r
l
q
rb
g
symbols
thermal coecient of volume expansion
surface tension
molecular dynamic viscosity
density
StefanBolzmann constant
eective coecient of laser beam
Subscripts
i
representation of certain element
LIQ liquidus
SOL solidus
l
liquid phase
s
solid phase
ref
reference value
278
surface temperature of a weld pool and also govern the temperature distribution of the pool surface. Mundra and Debroy [20] calculated the
vaporization rate and the change of composition
in weld pool during a laser weld process. Their
results have revealed that the heat loss due to
evaporation has a signicant inuence on the peak
temperature in a range of dierent laser powers,
and the vaporization rate predicted by Langmuir
equation was very much higher than the actual
value. In most previous studies on heat transfer
and uid ow for laser surface remelting and laser
welding, a constant surface tension temperature
coecient has been assumed and dierent values
are even adopted for the same material. However,
it should be emphasized that the predicted results
are sensitive to the value of surface tension temperature coecient.
In the present paper, a one-domain mixture
continuum model for the conservation of total
mass, momentum and energy equations is introduced to simulate the development of molten pool,
uid ow and temperature eld in laser surface remelting. Attention is concentrated on the competitive inuence of Marangoni ow and evaporation
on heating surface peak temperature and surface
temperature distributions as well as the molten pool
size and prole. The purpose of computing is to
obtain an insight and understand the complex
processes that occur in laser surface remelting.
Fig. 1. Schematic illustration of the numerical model for laser surface remelting process. The computational boundaries are represented by ci ; i 1 to 6.
0:
ot r or
oz
ot
r
or
oz
op 1 o
ou
o
ou
rl
l
or r or
or
oz
oz
u q
l q
l 2
u:
r q 1 K q1
ot
r
or
oz
1 o
k oh
o k oh
r
Sh ;
r or Cs or
oz Cs oz
href
;
Cs
oqf1 dH 1 orqu dH
ot
r
or
qref gb
Sh
oqv dH
:
oz
5
The detailed derivations of the one-domain mixture continuum equations can be found in [21,22].
The denitions of all symbols are presented in the
nomenclature.
2.3. Auxiliary equations
ot
r
or
oz
op 1 o
ov
o
ov
rl
l
oz r or
or
oz
oz
l q
v Sz :
K q1
279
280
hSOL 6 h 6 hLIQ ;
fl hhLIQ hSOL
9
hSOL
:
0
h < hSOL ;
Z
dH
Tref
Cl
Cs dT L:
10
13
14
qloss :
15
Ta rb eT 4
Ta4 :
16
:
17
oz
oT
or
Table 1 shows the overall boundary conditions
corresponding to the boundaries in Fig. 1.
c1
c2
c3
c4
c5
c6
ou
oz
or
or
ov
0
or
0
T
k
oT
qsrf
oz
oT
qsrf
or
oT
qsrf
oz
oT
qnet
oZ
oT
0
or
oT
0
or
3. Numerical method
The advantage of the developed model is that
the governing equations are valid in liquid, solid
and mushy regions, and there is no need to track
the molten pool geometrical shapes and the extent
of each region. Hence, a xed-grid system can be
used in the numerical calculation. Furthermore,
the governing equations are in the general format
as suggested by Patankar [25] for the numerical
solution of heat and uid ow problems. Therefore, any established numerical procedure for
solving coupled elliptic partial dierential equations could be used with slight modication to the
source terms. In the present study, the numerical
method using a control volume approach is employed to solve the governing equations. In this
case, the dierential equation is integrated over the
control volume to yield the discretization equation. A fully implicit formulation is used for timedependent terms, and power-law scheme is employed to evaluate the combined convection/diffusion coecient. A staggered grid arrangement is
introduced for which the velocities are stored at a
location midway between the grid points on the
control volume faces. All other variables including
pressure are determined at the grid points. This
arrangement enables to handle the pressure linkages through the continuity equation and is known
281
282
Case 2:
or=oT 0;
n
X
qi 6 0:
i0
Case 3:
or=oT
10 4 ;
n
X
qi 0:
i0
Case 4:
or=oT 10 4 ;
n
X
i0
qi 0:
283
Case 5:
or=oT
10 4 ;
n
X
qi 6 0:
i0
Case 6:
or=oT 10 4 ;
n
X
qi 6 0;
i0
284
Table 2
Thermophysical properties of the material and input parameters for laser surface remelting [17,18]
Symbol and value
Cl 735
Cs 735
L 2:47 105
Lvap:Fe 6091
Lvap:Mn 4014
Lvap:Cr 6577
Lvap:Ni 6388
MFe 55:85
MMn 54:94
MNi 58:71
MCr 52:00
Tl 1723
Ts 1523
Unit
1
(J kg K )
(J kg 1 K 1 )
(J kg 1 )
(kJ kg 1 )
(kJ kg 1 )
(kJ kg 1 )
(kJ kg 1 )
(g mole 1 )
(g mole 1 )
(g mole 1
(g mole 1 )
(K)
(K)
Unit
Ta 300
kl 20
ks 20
Lr 4
Lz 2
rb 0:25
(K)
(W m
(W m
(mm)
(mm)
(mm)
ql 7200
qs 7200
l 0:006
e 0:4
b 1:0 10
Q 500
(kg m 3 )
(kg m 3 )
(kg m 1 s 1 )
4
1
1
K 1)
K 1)
(K 1 )
(W)
Fig. 4. (a) The equilibrium vapor pressures. (b) Vaporization mass ux. (c) Heat loss rate for various vaporizing species as a function
of temperature.
Fig. 5. Heating surface temperature distribution of laser surface remelting for the six calculation cases: Case 1: or=oT 0;
Pn
Pn
Case
Case 3:
i0 qi 0;
i0 qi 6 0;
Pn 2: or=oT 0;
4
qi 0;
Case
4:
or=oT 10 4 ;
or=oT 10 ;
i0
Pn
Pn
5: or=oT 10 4 ;
i0 qi 0; Case
i0 qi 6 0; Case 6:
Pn
4
or=oT 10 ;
i0 qi 6 0.
285
eld is responsible for dissipating the thermal energy within the pool. Furthermore, through comparing case 3 with case 4, it is seen that for
negative value of or=oT , the temperature at every
point on the surface is higher than that for positive
value of or=oT . It is the same with the results
predicted by Choo [24]. If the inuences of alloying element vaporization heat losses are considered
together with Marangoni convection, the cases 5
and 6 are obtained. For the case 5, the surface
temperature distribution is similar to the case 3,
but the value of temperature at each corresponding point is slightly lower than that of case 3, and
there still exists a temperature plateau near the
front of mushy region. Because in real practice of
laser surface remelting Marangoni convection and
evaporation heat loss always coexist, the dierence
between the two plots reects the actual eect of
alloying element vaporization heat loss on heating
surface temperature distribution. For cases 4 and
6, the temperature distribution exhibits a similar
trend with the cases 3 and 5.
4.3. Molten pool shape and velocity elds of the
calculation cases
The dierences of the molten pool circulation
and of the pool surface energy balance must aect
the pool proles. Fig. 6(a)(f) show the molten
pool proles and velocity elds of the six calculation cases at the heating time of 0.1 s after laser
beam irradiating. Comparing Fig. 6(a) with (b), it
can be seen that in the calculation condition of
case 2 both the pool radius and depth are signicantly less than those in the case 1, but the pool
prole is similar to each other and looks like a
half-ellipse. It is revealed that heat loss due to alloying element vaporization can only reduce the
heat input and not change the circulation patterns
in the pool. For the case 3 (Fig. 6(c)), the negative
surface-tension temperature coecient or=oT
< 0 induces an outward ow as mentioned in the
preceding section. This ow with a higher temperature collides with the interface of solid/liquid
at the cold corner of the pool and results in a
relatively higher melting rate there than at the pool
base. Consequently, the molten pool prole revealed a shallow and wide shape. For the case 4
286
Fig. 6. Molten pool size, prole and velocity elds corresponding to the calculation cases in Fig. 5 for laser surface remelting. Solid line
is solidus, and dashed line is liquidus.
287
Fig. 7. The heating surface temperature distributions with dierent values of or=oT , (a) or=oT < 0 and (b) or=oT > 0.
288
Fig. 8. The inuence of surface tension temperature coecient on the peak surface temperature with and without vaporization heat
loss for laser surface remelting. (a) or=oT < 0 and (b) or=oT > 0.
Acknowledgements
This research is supported by National `973'
Project Foundation of China under grant no.
G1998061500.
Appendix A
The equilibrium vapor pressures of vaporizing
species, Fe, Mn, Cr and Ni, over the respective
pure liquids are calculated using the following
equations:
0
ln pFe
atm
5:58 10 4 T
8:381 10 7 T 2
9:290 10 3 T
87:077
18:042 10 3 T
214:297
289
290