ANALYTICAL MODELS FOR TEMPERATURE PREDICTION ON WELDING

Elena SCUTELNICU1, Mihaela IORDACHESCU2, Bogdan GEORGESCU1

1
Robotics and Welding Department, Dunarea de Jos University of Galati - Romania
2
Dep. Ciencia de Materiales, Universidad Politécnica de Madrid – España
E-mail: [email protected]

Abstract

Temperature field has an important influence on the phase changes and, finally, on the
microstructure and mechanical properties, residual stresses and strains development in the welded
joints. During the welding, temperature field shape and its profile depend on many factors but,
assuming some assumptions regarding homogeneity and materials isotropy, the mathematical
relations for different practical cases can be obtained. Besides, welded bodies can be infinite or
semi-infinite as the plates, bars and massive bodies are considered. The instantaneous or
permanent thermal sources, fixed or mobile have an important influence on the temperatures
distribution. Therefore, the paper presents some analytical solutions for the temperatures
prediction in the welded joints in case of the mobile sources with 2D and 3D Gauss distribution.

1. Equations of the temperature field

The general equation of the energy which is the start point for the temperature field analysis has
the following expression:
∂T ∂ ⎛ ∂T ⎞ ∂ ⎛ ∂T ⎞ ∂ ⎛ ∂T ⎞ ⎛ ∂T ∂T ∂T ⎞
ρ ⋅c = ⎜λ ⎟ + ⎜⎜ λ ⎟⎟ + ⎜ λ ⎟ − ρ ⋅ c⎜⎜ v x + vy + vz ⎟ + Qv (1)
∂t ∂x ⎝ ∂x ⎠ ∂y ⎝ ∂y ⎠ ∂z ⎝ ∂z ⎠ ⎝ ∂x ∂y ∂z ⎟⎠
where vx, vy, vz are the speed components on the three directions.

Considering the thermal source moving on the direction x (vx=v, vy=vz=0) and the coordinates
system mobile solidary with the thermal source, the energy balance equation can be written as:
∂ ⎛ ∂T ⎞ ∂ ⎛ ∂T ⎞ ∂ ⎛ ∂T ⎞ ∂T
⎜λ ⎟ + ⎜⎜ λ ⎟⎟ + ⎜ λ ⎟ − ρ ⋅c⋅v + Qv = 0 (2)
∂x ⎝ ∂x ⎠ ∂y ⎝ ∂y ⎠ ∂z ⎝ ∂z ⎠ ∂x
Due to the existence of the solid (s) and liquid (l) phases in the weld pool, the energy balance
equations, in the quasi-stationary period, can be expressed as:
- in case of solid phase:
∂ ⎛ ∂T ⎞ ∂ ⎛ ∂T ⎞ ∂ ⎛ ∂T ⎞ ∂T
⎜ λs ⎟ + ⎜⎜ λ s ⎟⎟ + ⎜ λ s ⎟ − ρ ⋅ cs ⋅ v + Qv = 0 , (3)
∂x ⎝ ∂x ⎠ ∂y ⎝ ∂y ⎠ ∂z ⎝ ∂z ⎠ ∂x
- in case of liquid phase:
∂ ⎛ ∂T ⎞ ∂ ⎛ ∂T ⎞ ∂ ⎛ ∂T ⎞ ∂T
⎜ λl ⎟ + ⎜⎜ λl ⎟⎟ + ⎜ λl ⎟ − ρ ⋅ cl ⋅ v + Qv = 0 , (4)
∂x ⎝ ∂x ⎠ ∂y ⎝ ∂y ⎠ ∂z ⎝ ∂z ⎠ ∂x
at solid-liquid interface:

(n ⋅ v )⋅ ρ ⋅ ∆H = λ s ⎛⎜ ∂T ⎞⎟ ⎛ ∂T ⎞
− λl ⎜ ⎟; T = Ttop . (5)
⎝ ∂n ⎠ s ⎝ ∂n ⎠ l
The upper surface of the piece is under the influence of the thermal source and also of the heat
lost in the environment. Therefore, the contour condition can be written as:
∂T
−λ =q (6)
∂z z =0

313
2. Mobile thermal source of 2D Gauss distribution

2.1. Temperature field analysis in the butt welded joints

Considering the mobile thermal source of Gauss distribution, Eagar and Tsai changed the
Rosenthal’s theory [2]. Their solution represents an important step for the temperatures’
approximation in the vicinity of the thermal source. Assuming an exponential repartition of the heat
flow, its mathematical expression can be written as:
q = q max e −k ⋅r ,
2
(7)
where k is the concentration factor which depends on the thermal source type.

Figure 1. Gauss distribution of the heat flow.

Taking account of Gauss distribution of the heat flow, Pavelic et al. [1], developed the calculus
expressions of the heat flow for the contact area and outside of the contact area:

q=
3Q
exp
x2 + y2
(
for: x 2 + y 2 )
12
≤r, (8)
π ⋅r2 − r2 3
where Q=η·U·I and r=r0 is the effective radius of the thermal source when the heat flow decreases
at 5% of maximum value 3Q/πr2.
Outside of the area, which is under the influence of the thermal source, the heat flow’s expression
is:
[ (
q = − α ⋅ (T − T0 ) + ε ⋅ C ⋅ T 4 − T04 )]
(
for: x 2 + y 2 )
12
fr. (9)

Figure 2. 2D Gauss distribution of heat flow.

2.2. Temperature field analysis in the fillet welds

Fillet welds represent 80% of the total naval production. Therefore, taking account of two-
dimensional distribution of heat flow, Jeong and Cho [3] analysed temperature field in fillet welds
and found the analytical solutions for the temperatures calculus. (Figure 2).

314
Since the geometrical shape and the electric arc distribution are complex, the authors replaced the
real system with an equivalent one (Figure 3).

Figure 3. The real and equivalent systems in case of fillet welds [3].

Finally, the mathematical model of the heat transfer in the changed coordinates system is given by
the following equations (10)...(16).
¾ Conduction equation in case of solids:
∂Tu ⎛ ∂2 ∂2 ∂2 ⎞
= a⎜⎜ 2 + 2 + 2 ⎟Tu
⎟ (10)
∂t ⎝ ∂u ∂v ∂w ⎠
¾ Contour condition on the upper surface:
∂Tu (U ,V ,0 )
−λ + α 1 (Tu (U , V ,0) − T0 ) = 0 (11)
∂W
¾ Contour condition on the lower surface:
∂Tu (U ,V , d )
λ + α 2 (Tu (U , V , d ) − T0 ) = 0 (12)
∂W
¾ Boundary condition at the infinite distance from the thermal source:
lim Tu (U , V , W , t ) = T0 , r= (U − U ')2 + (V − V ')2 + W 2 , (13)
r →∞
Where r is the distance from the thermal source located in the coordinates point (U’, V’, 0).
¾ Temperatures distribution at time t=t0:
Tu (U ,V ,W ,0) = T0 (14)
¾ Heat flow two-dimensional distribution is given by the equation:
q (t ) ⎡ ⎛ u2 v2 ⎞⎤
Q(u , v, t ) = exp ⎢− ⎜ + ⎟⎥ , (15)
2πσ u σ v ⎢⎣ ⎜⎝ 2σ u 2σ v ⎟⎥
2 2
⎠⎦
if σ2= σu x σv, the equation (15) can be written as:
q(t ) ⎛ σ v 2u 2 + σ u 2 v 2 ⎞
Q(u, v, t ) = exp ⎜ − ⎟. (16)
2πσ 2 ⎜ 2σ 4 ⎟
⎝ ⎠

315
In the equation (16) σu, σv are the distribution parameters and q(t)=ηUI(t) is the useful power of the
electric arc. Solving the equations 10)...(16), the analytical solutions of the temperatures in the
changed coordinates system can be obtained for three analyzed cases:

1. Temperature at t moment, in the point P of (U,V,W) coordinates, in the infinite plate which is
under the influence of the thermal source qi, located in the point of (U ’, V ’, 0) coordinates at t1
moment
⎛ (U − U ')2 + (V − V ')2 ⎞ ∞
Tu (U , V ,W , t ) =
qi
2π ⋅ λ ⋅ d ⋅ (t − t1 )
⋅ exp⎜ −
⎜ 4a ⋅ (t − t1 ) ⎟
( )
⎟ ⋅ ∑ An exp − µ n2 (t − t1 ) ⋅
⎝ ⎠ n =0
(17)
⎡ ⎛µ ⎞ β a ⎛ µ n ⎞⎤
⋅ ⎢cos⎜⎜ n W ⎟⎟ + 1 sin ⎜⎜ W ⎟⎟⎥
⎢⎣ ⎝ a ⎠ µn ⎝ a ⎠⎥⎦
µ n2 α1 α2
where: An = , β1 = , β2 = , λ = a⋅c⋅ ρ
2 aβ 1 λ λ
µ n2 + aβ12 +
d
⎛ µnd ⎞ a µ (β + β 2 )
and µn is a positive value which satisfies the equality: tg ⎜⎜ ⎟⎟ = 2 n 1 .
⎝ a ⎠ µ n − β1 ⋅ β 2 ⋅ a

2. Temperature in the infinite plate which is under the influence of the two-dimensional Gauss
thermal source located in a fixed point at t1 moment and obtained by overlapped instantaneous
thermal sources
q(t1 ) ⋅ d (t1 ) σ2
dTu (t1 ) = ⋅ ⋅
π ⋅ ρ ⋅ c ⋅ d σ 4 + 2a ⋅ [t − t ]σ 2 ⋅ σ 4 + 2a ⋅ [t − t ]σ 2
1 v 1 u

( ) (
⎡ σ v2 + 2a ⋅ (t − t1 ) ⋅ U 2 + σ u2 + 2a ⋅ (t − t1 ) ⋅ V 2 ⎤
⋅ exp ⎢− ⎥⋅
)
⎣⎢ ( )(
2 σ u2 + 2a ⋅ (t − t1 ) ⋅ σ v2 + 2a ⋅ (t − t1 ) ) ⎦⎥
(18)

⎡ ⎛µ ⎞ β a ⎛ µ n ⎞⎤
( )
∞
⋅ ∑ An exp − µ n2 (t − t1 ) ⋅ ⎢cos⎜⎜ n W ⎟⎟ + 1 sin ⎜⎜ W ⎟⎟⎥
n =0 ⎣⎢ ⎝ a ⎠ µn ⎝ a ⎠⎦⎥
3. Temperature in the infinite plate which is under the influence of the two-dimensional Gauss
thermal source, located in the point of (Ua, Va, 0) coordinates at t1 moment and obtained by
overlapped instantaneous thermal sources
q(t1 ) ⋅ d (t1 ) σ2
dTu (t1 ) = ⋅ ⋅
π ⋅ ρ ⋅ c ⋅ d σ 4 + 2a ⋅ [t − t ]σ 2 ⋅ σ 4 + 2a ⋅ [t − t ]σ 2
1 v 1 u

( ) (
⎡ σ 2 + 2a ⋅ (t − t1 ) ⋅ (U − U a )2 + σ u2 + 2a(t − t1 ) ⋅ (V − Va )2
⋅ exp ⎢− v
)⎤⎥ ⋅
⎢⎣ ( )(
2 σ u2 + 2a(t − t1 ) ⋅ σ v2 + 2a (t − t1 ) ) ⎥⎦
(19)

⎡ ⎛µ ⎞ β a ⎛ µ n ⎞⎤
( )
∞
⋅ ∑ An exp − µ n2 (t − t1 ) ⋅ ⎢cos⎜⎜ n W ⎟⎟ + 1 sin ⎜⎜ W ⎟⎟⎥
n =0 ⎣⎢ ⎝ a ⎠ µn ⎝ a ⎠⎦⎥
In case of mobile system, solidary with the electrode, the temperature Tu(u,v,w), at t moment, can
be computed with the following mathematical relation:

316
 q(t1 ) σ2
Tu (u, v, w, t ) = ∫
t
⋅ ⋅
0π ⋅ ρ ⋅c⋅d
σ 4 + 2a[t − t1 ] ⋅ σ v2 ⋅ σ 4 + 2a[t − t1 ] ⋅ σ u2
( ) (
⎡ σ v2 + 2a ⋅ (t − t1 ) ⋅ (u + U a (t ) − U a (t1 ))2 + σ u2 + 2a(t − t1 ) ⋅ (v + Va (t ) − Va (t1 ))2 ⎤
⋅ exp⎢− ⎥⋅
)
⎣⎢ ( )(
2 σ u2 + 2a(t − t1 ) ⋅ σ v2 + 2a(t − t1 ) ) ⎦⎥
(20)

⎡ ⎛µ ⎞ β a ⎛ µ n ⎞⎤
( )
∞
⋅ ∑ An exp − µ n2 (t − t1 ) ⋅ ⎢cos⎜⎜ n W ⎟⎟ + 1 sin⎜⎜ W ⎟⎟⎥ ⋅ dt1
n =0 ⎢⎣ ⎝ a ⎠ µn ⎝ a ⎠⎥⎦
According to Jeong and Cho [3], the analytical solution for the temperatures calculus in the fillet
joints is:

T ( x, y, z , t ) = Tu (u , v, w, t ) ⋅
d (e A
⋅ cos B + a) + (e
2 A
⋅ sin B ) 2

, A=
π ⋅v
, B=
π ⋅w
(21)
(e ⋅ cos B − 1) + (e ⋅ sin B )
2
H A 2 A
d d

3. Mobile thermal source of 3D Gauss distribution

Due to the important penetration in case of the semi-infinite welded bodies, Goldak, Chakravarti
and Bibby proposed the three-dimensional mobile source to predict the temperatures.

3.1. Semi-ellipsoidal thermal source

Firstly, Goldak et al. proposed a semi-ellipsoidal thermal source of the heat flow in any point of (x,
y, z) which can be computed with the equation [4]:
6 3η ⋅ U ⋅ I ⎛ 3x 2 3 y 2 3z 2 ⎞
Q ( x, y , z , ) = ⋅ exp ⎜⎜ − 2 − 2 − 2 ⎟⎟ , (22)
a h ⋅ bh ⋅ c h ⋅ π π ⎝ ch ah bh ⎠
where ah, bh, ch are the parameters of the semi-ellipsoidal thermal source (Fig. 4) and x, y, z are
the coordinates of the thermal source.

Figure 4. 3D Gauss distribution of the heat flow (chf=chb=ch).

3.2. Temperature field developed by the semi-ellipsoidal thermal source

Taking account of the solution obtained by Carslaw and Jaeger in case of the instantaneous
source, the researchers [2] established the temperature field produced by the semi-ellipsoidal
thermal source:
δQ ⋅ dt ' ⎛ ( x − x ' )2 + ( y − y ' )2 + ( z − z ' )2 ⎞
dTt ' = ⋅ exp⎜− ⎟. (23)
ρ ⋅ c ⋅ [4π ⋅ a ⋅ (t − t ')]3 2 ⎜
⎝ 4 a ⋅ (t − t ' ) ⎟
⎠

317
Overlapping more instantaneous thermal sources and taking account of semi-ellipsoidal thermal
source distribution, it can be written the equation (24) [2]:
∞ ∞ ∞
1 dt ' 6 3η ⋅ U ⋅ I ⎛ 3 x' 2 3 y ' 2 3 z ' 2 ⎞
⎜− ⎟⋅
2 −∫∞ −∫∞ −∫∞ ρ ⋅ c ⋅ [4π ⋅ a ⋅ (t − t ')]3 2 a h ⋅ bh ⋅ c h ⋅ π π
dTt ' = dx ' dy ' dz ' ⋅ ⋅ exp⎜ c2 − a2 − b2 ⎟
⎝ h h h ⎠
(24)
⎛ ( x − x ' )2 + ( y − y ' )2 + ( z − z ' )2 ⎞
⋅ exp⎜ − ⎟
⎜
⎝ 4a ⋅ (t − t ') ⎟
⎠
and rewritten as:
3 3η ⋅ U ⋅ I ⋅ dt ' 1 1 1
dTt ' = ⋅ ⋅ ⋅ ⋅
ρ ⋅ c ⋅π π 12a ⋅ (t − t ') + a h2 12a ⋅ (t − t ') + bh2 12a ⋅ (t − t ') + c h2
(25)
⎛ 3x 2 3y 2 3z 2 ⎞
⋅ exp ⎜⎜ − − − ⎟
⎟
⎝ 12a ⋅ (t − t ') + c h 12a ⋅ (t − t ') + a h 12a ⋅ (t − t ') + bh
2 2 2
⎠
Considering that the thermal source moves with constant speed from initial moment t’=0 to the
moment t’=t, the increase of temperature in this range time can be computed as [2]:
3 3η ⋅ U ⋅ I
t
dt '
T − T0 = ⋅∫ ⋅
ρ ⋅ c ⋅π π 0 12a ⋅ (t − t ') + a h2 12a ⋅ (t − t ') + bh2 ⋅ 12a ⋅ (t − t ') + c h2
(26)
⎛
⎜ 3( x − vt ')2 3y 2 3z 2 ⎞
⎟
⋅ exp − − −
⎜ 12a ⋅ (t − t ') + c 2 12a ⋅ (t − t ') + a 2 12a ⋅ (t − t ') + b 2 ⎟
⎝ h h h ⎠
Appling the equation from above, the analytical solutions can be obtained for three specific cases
as [2]:

1. If ah=bh=ch=rh, the thermal source becomes a semi-sphere with radius of rh, and the relation (26)
has the expression:
3 3η ⋅ U ⋅ I
t
dt ' ⎛ 3( x − vt ')2 + 3 y 2 + 3 z 2 ⎞
T − T0 = ⋅∫ ⋅ exp ⎜ − ⎟.
[12a ⋅ (t − t ') + r ]
(27)
ρ ⋅ c ⋅π π 2 32 ⎜ 12a ⋅ (t − t ') + rh2 ⎟
0 h ⎝ ⎠
When rh = 3σ the equation (2.40) becomes:
η ⋅U ⋅ I
t
dt ' ⎛ ( x − vt ')2 + y 2 + z 2 ⎞
T − T0 = ⋅∫ ⋅ exp⎜ − ⎟.
[ ]
(28)
ρ ⋅ c ⋅ π π 0 4a ⋅ (t − t ') + σ 2 32 ⎜ 4a ⋅ (t − t ') + σ 2 ⎟
⎝ ⎠
2. If the distribution parameter is σ=0, the thermal source becomes a point one, the analytical
solution confirming the solution of Carslaw and Jaeger:

η ⋅U ⋅ I
t
dt ' ⎛ ( x − vt ')2 + y 2 + z 2 ⎞
T − T0 = ⋅∫ ⋅ exp ⎜− ⎟. (29)
ρ ⋅ c 0 [4πa ⋅ (t − t ')]3 2 ⎜
⎝ 4 a ⋅ (t − t ' ) ⎟
⎠

3. If bh=0, then the semi-ellipsoidal thermal source becomes a surface semi-elliptical thermal
source and the equation (26) is the following:
3η ⋅ U ⋅ I
t
dt '
T − T0 = ⋅∫ ⋅
π ⋅ ρ ⋅ c 0 4π ⋅ a ⋅ (t − t ') 12a ⋅ (t − t ') + a 2 12a ⋅ (t − t ') + c 2
h h
(30)
⎛
⎜ 3( x − vt ')2 3y 2 z2 ⎞
⎟
⋅ exp − − −
⎜ 12a ⋅ (t − t ') + c 2 12a ⋅ (t − t ') + a 2 4a ⋅ (t − t ') ⎟
⎝ h h ⎠

318
When a h = c h = 6 ⋅ σ , the thermal source changes in a circular one. In this case, the equation
(28) is similar to the solution of Eagar and Tsai:
η ⋅U ⋅ I
t
dt ' ⎛ ( x − vt ')2 + y 2 z2 ⎞
T − T0 = ⋅∫ ⋅ ⎜− − ⎟ (31)
[
π ⋅ ρ ⋅ c 0 4π ⋅ a ⋅ (t − t ') ⋅ 4a ⋅ (t − t ') + 2σ 2
exp
⎝ ]
⎜ 4a ⋅ (t − t ') + 2σ 2
4 a ⋅ (t − t ' ) ⎟
⎠

3.3. Double ellipsoidal thermal source

Combining two semi-ellipses, Goldak, Chakravarti and Bibby proposed the double ellipsoidal
thermal source (Figure 4) [4]. For the semi-ellipse located in front of the welding arc, the heat flow
can be computed in each point of (x, y, z) coordinates using the following mathematical relation:
6 3 ⋅ r f ⋅η ⋅ U ⋅ I ⎛ 3x 2 3 y 2 3z 2 ⎞
Q ( x, y , z , ) = ⋅ exp ⎜ − 2 − 2 − 2 ⎟ . (32)
a h ⋅ bh ⋅ c hf ⋅ π π ⎜ c hf ah bh ⎟⎠
⎝
For the second semi-ellipse located behind of the welding arc, the heat flow can be computed in
each point of (x, y, z) coordinates using the following mathematical relation:
6 3 ⋅rb ⋅η ⋅ U ⋅ I ⎛ 3x 2 3 y 2 3z 2 ⎞
Q ( x, y , z , ) = ⋅ exp ⎜⎜ − 2 − 2 − 2 ⎟⎟ . (33)
a h ⋅ bh ⋅ c hb ⋅ π π ⎝ c hb ah bh ⎠
In the equations (32) and (33) rf and rb represent the repartition coefficients of the heat in front and
behind of the thermal source and their mathematical relations are the following [4]:
r f = 2c hf c hf + c hb , ( )
rb = 2c hb c hf + c hb , ( )
(34)
where the thermal source dimensions are similar to the weld pool geometrical characteristics [4].

3.4. Temperature field developed by the double ellipsoidal thermal source

In case of the double ellipsoidal distribution of the thermal source, for the volume corresponding to
the areas located in front and behind of the thermal source, the following equation can be written:

1 6 3 ⋅η ⋅ U ⋅ I ⋅ dt '
∞ ∞ ∞
⎛ ( x − x')2 + ( y − y ')2 + ( z − z ')2 ⎞
dTt ' = ⋅
4 ρ ⋅ c ⋅ a h ⋅ bh ⋅ π π [4π ⋅ a ⋅ (t − t ')]3 2 ∫ ∫ ∫ exp ⎜⎜ − 4a ⋅ (t − t ')
⎟⋅
⎟
−∞ −∞ −∞ ⎝ ⎠
(35)
⎛ rf ⎛ 3 x' 2 3 y ' 2 3 z ' 2 ⎞ r ⎛ 2 2 2 ⎞ ⎞⎟
⋅ ⎜ ⋅ exp ⎜ − 2 − 2 − 2 ⎟ + b ⋅ exp ⎜ − 3 x' − 3 y ' − 3 z ' ⎟ ⋅ dx'⋅dy '⋅dz '
⎜ c hf ⎜ c hf ⎟ c hb ⎜ c2 a h2 bh2 ⎟⎟
⎝ ⎝ ah bh ⎠ ⎝ hb ⎠⎠

The equation (35) can be adapted as:

⎛ ⎞
3 3 ⋅ η ⋅ U ⋅ I ⋅ dt' ⎜ A B ⎟
dTt ' = ⋅⎜ + ⎟ (36)
( )( )
2π ⋅ ρ ⋅ c ⋅ π ⋅ 12a ⋅ (t − t ') + ah2 ⋅ 12a ⋅ (t − t ') + bh2 ⎜⎝ 12a ⋅ (t − t ') + chf2 12a ⋅ (t − t ') + chb
2 ⎟
⎠
⎛ 3x 2 3y 2 3z 2 ⎞
where A = r f ⋅ exp ⎜ − − − ⎟
⎜ 12a ⋅ (t − t ') + c hf2 12a ⋅ (t − t ') + a h2 12a ⋅ (t − t ') + bh2 ⎟
⎝ ⎠.
⎛ 3x 2 3y 2 3z 2 ⎞
B = rb ⋅ exp ⎜⎜ − − − ⎟
⎟
⎝ 12a ⋅ (t − t ') + c hb 12a ⋅ (t − t ') + a h 12a ⋅ (t − t ') + bh
2 2 2
⎠
Similarly, considering that the thermal source moves with constant speed from initial moment t’=0
to the moment t’=t, the increase of temperature in this range time can be computed as [2]:

319
 ⎛ ⎞
3 3 ⋅η ⋅U ⋅ I
t
dt' ⎜ A B ⎟
T −T0 = ⋅∫ +
2π ⋅ ρ ⋅ c ⋅ π 0 ( 2
)( )
2 ⎜
12a ⋅ (t −t') + ah ⋅ 12a ⋅ (t −t') + bh ⎜⎝ 12a ⋅ (t −t') + chf
2 ⎟
12a ⋅ (t −t') + chb ⎟⎠
2

⎛ 3(x − vt') ⎞
2 (37)
3y 2 3z 2
where A = rf ⋅ exp ⎜ − − − ⎟
⎜ 12a ⋅ (t − t') + c2 12a ⋅ (t −t') + a2 12a ⋅ (t −t') + b2 ⎟
⎝ hf h h ⎠

⎛ 3(x − vt') ⎞
2
3y 2 3z 2
B = rb ⋅ exp ⎜ − − − ⎟
⎜ 12a ⋅ (t −t') + c 12a ⋅ (t − t') + a 12a ⋅ (t −t') + b ⎟
2 2 2
⎝ hb h h ⎠
The equation from above represents the analytical solution of temperature calculus in the transitory
regime in case of the semi-infinite body which is under the influence of the double ellipsoidal
thermal source. If chf=chb=ch, the equations (36) and (37) can be written as the equations (25) and
(26) which describe the temperature variation of the semi-infinite body under the influence of the
semi-ellipsoidal thermal source.

4. Conclusion
ƒ Investigations concerning the analytical solutions of the thermal transfer in the welded joints
in case of 2D and 3D thermal sources are presented in this paper. The analytical models
don’t take account neither of thermo-physical properties dependent of temperature nor of
heat lost by convection and radiation.
ƒ Goldak [4] proposes the different distribution of heat flow in front and behind the thermal
source and its real dimensions in the analytical modelling of the thermal transfer in the
welded joints.

Notations
a – thermal diffusivity;
ah, bh, chf, chb – ellipsoidal thermal source;
c – specific heat at ambient temperature;
cS – specific heat for the solid phase;
cL – specific heat for the liquid phase;
∆H – fusion heat per mass unit;
λ - material thermal conductivity;
λx, λy, λz – material thermal conductivities on x, y, z directions;
rf, rb – repartition coefficients of the heat in front and behind of the thermal source;
σ – distribution parameter specific for the thermal source;
Vx, Vy, Vz – speed components on x, y, z directions.

References

[1] Kou, S.: Welding Metallurgy, University of Wisconsin, John Wiley & Sons, Inc., 1987.
[2] Nguyen, N., T., Ohta, A., Matsuoka, K., Suzuki, N., Maeda, Y. – Analytical solutions for
transient temperature of semi-infinite body subjected to 3-D moving heat sources, Welding Journal,
August, 1999, pag. 265s-273s.
[3] Jeong, S., K., Cho, H., S. – An analytical solution to predict the transient temperature
distribution in fillet arc welds, Welding Journal 76(6), 1997, pag. 223s-232s.
[4] Goldak,J., Chakravarti, A., Bibby, M. – A double ellipsoid finite element model for welding heat
sources, IIW Doc. 212-603-85, 1985.

320