Documente Academic
Documente Profesional
Documente Cultură
93-101 (2004) 93
Zong-Yi Lee
Key words: multilayered spheres, laplace transform, similarity transforma- thermoelasticity problem. The solution technique uses
tion. Laplace transform. The inversion to real domain is
obtained by means of Cauchy’s theorem of residues and
the convolution theorem. Sherief and Anwar [6] dis-
ABSTRACT cussed the problem of an infinitely long elastic circular
cylinder whose inner and outer surfaces are subjected to
This paper deals with quasi-static coupled thermoelastic prob-
known temperature and are traction free. They have
lems for multilayered spheres. Using the Laplace transform with
respect to time, the general solutions of the governing equations are neglected both the inertia terms and the relaxation effects.
obtained in transform domain. The solution is obtained by using the Takeuti and Furukawa [7] discussed the thermal shock
matrix similarity transformation and inverse Laplace transform. We problem in a plate; they included the inertia and
obtain solutions for the temperature and thermal deformation distri- thermoelastic coupling terms in the governing equation
butions for a transient state. It is demonstrated that the computational
and obtained the exact solution for the thermal shock
procedures established in this paper are capable of solving the gener-
alized thermoelasticity problem of multilayered spheres. problem of a plate. Chen and Yang [1] discussed the
transient response of one-dimensional quasi-static
INTRODUCTION coupled thermoelasticity problems of an infinitely long
cylinder composed of two different materials. They
The increasing use of composite materials in engi- applied the Laplace transform with respect to time and
neering application has resulted in considerable re- used the Fourier series and matrix operations to obtain
search activity in this area in recent years. An under- the solution. Chen and Chen [2] presented a new
standing of thermally induced stresses in isotropic bod- numerical technique hybrid numerical method for the
ies is essential for a comprehensive study of their re- problem of a transient linear heat conduction system.
sponse due to an exposure to a temperature field, which They applied the Laplace transform to remove the time-
may in turn occur in service or during the manufacturing dependence from the governing equations and boundary
stages. While the deformation and stress fields in a shell conditions, and solved the transformed equations with
caused by thermal effects have commonly been ob- the finite element and finite difference method. Jane
tained on the basis of uncoupled thermoelasticity theory, and Lee [4] considered the same problem by using the
the thermo-mechanical coupling effect is included in Laplace transform and the finite difference method.
the present study. The cylinder was composed of multilayers of different
Vollbrecht [9] has analysed the stresses in both materials. There is no limit to the number of annular
cylindrical and spherical walls subjected to internal layers of the cylinder in the computational procedures.
pressure and stationary heat flow. Kandil [5] has stud- Laplace transform and finite difference methods were
ied the effect of steady-state temperature and pressure used to analyze problems. They obtained solutions for
gradient on compound cylinders fitted together by shrink the temperature and thermal stress distributions in a
fit. Ghosn and Sabbaghian [3] investigated a one- transient state.
dimensional axisymmetric quasi-static coupled In this paper, the one-dimensional quasi-static
coupled thermoelastic problem of a multilayered sphere
with time-dependent boundary conditions is considered.
Paper Submitted 01/19/04, Accepted 04/13/04. Author for Correspondence:
Zong-Yi Lee. E-mail: zylee@mail.hit.edu.tw. The medium is without body forces and heat generation.
*Assistant Professor, Department of Mechanical Engineering, Hsiuping Laplace transform and finite difference methods are
Institute of Technology Taichung, Taiwan 412, R.O.C. used, which are quite effective and powerful, to obtain
94 Journal of Marine Science and Technology, Vol. 12, No. 2 (2004)
r = r*/R 1
Rout
t = ( k ) τ /R 21
R1 ρC ν 1
β
u = U( ) /R
ρC ν 1 1
ai = ( k ) / ( k )
ρC ν i ρC ν 1
σ ri = σ ri* /(β 1Θ 0)
σ θi = σ θ*i /(β 1Θ 0)
3Q i = [ Eα ] / β 1
(1 – 2ν ) i
1Q i =[
E(1 – ν )
] / [β (
β
)Θ ] (∂u ) – (∂u )
(1 + ν ) (1 – 2ν ) i 1 ρC ν i 0 ∂u ∂t ∂t j – 1
= r1
j j +1
+ (10)
j ∂t 2∆r i
2E ν β
2Q i = [ ] / [β ( )Θ ]
(1 + ν ) (1 – 2ν ) i 1 ρC ν i 0 u j + 1 – 2u j + u j – 1 uj +1 – uj –1 1
+ r1 – 2 uj
Eν β 2 ∆r i rj
] / [β (
j
1R i =[ )Θ ] (∆r i )
(1 + ν ) (1 – 2ν ) i 1 ρC ν i 0
Tj +1 – Tj –1
E β = gi (11)
2R i = [ ] / [β ( )Θ ] 2∆r i
(1 + ν ) (1 – 2ν ) i 1 ρC ν i 0
β uj +1 – uj –1 uj
g i = [(ρc ) α (1 + ν )] Θ 0 (5) σ ri = 1Q i + 2Q i r – 3Q i T j (12)
ν i 2∆r i j
σ θi = 1R i ∂u + 2R i ur – 3R i T
c + i∞
(9) Φ(t) = L –1[ Φ (s)] = 1 e st Φ (s) ds
∂r 2π i c – i∞
The nondimensional boundary and interface con- Taking the Laplace transform for equations (10),
ditions can be written as: (11), (12) and (13), we obtain the following equations:
1. Boundary conditions
T –2T j + T T – T
+ r1
j +1 j –1 j +1 j –1
p(t) ai ( )
σ r(r, t) = , 2 ∆r i
β 1Θ 0 T 1 = f(t)/Θ 0 at r = r 1 (∆r i ) j
u i (r, s) = u i + 1(r, s) r = ri + 1 Fj = 1
2∆r i
σ i (r, s) = σ ri + 1
(r, s) r = ri + 1 G j =0
q i (r, s) = q i + 1(r, s) r = ri + 1 Q –1 am a a
A N = [1 + 3 m ] [ – + rm – rm 1 ]
Q
1 m (∆r m) 2 k k ∆r m
T i (r, s) = T i + 1(r, s) r = ri + 1
Q – 1 2a m
B N = – [1 + 3 m ] [ ]
i = 1, 2, ..., m − 1 layer (17) 1Q m (∆r m)2
Q –1 Q
Substituting equations (16), (17) into equations E N = [1 + 3 m ] [– r1 + 2 m r1 ]
1Q m k 1Q m k
(14), (15), we obtain the following equation in matrix
form Q –1 a a1 p
G 1 = [1 + 3 m ] [( r 1 1 – )T1 + ]
Q
1 m 1 ∆r 1 (∆r 1) 2
1 1β Θ 0
Q
B1 C 1 T 1 Q –1 a a
G = – [1 + 3 m ] [( m 2 + r m 1 ) T N + 1]
k ∆r m
A2 B2 C 2 T N
2 Q
1 m (∆r m)
– s[I]
AN – 1 BN – 1C N – 1 m denotes the last layer, k the last point and i denotes ith
T N –1
layer for j = 2, 3, ..., N − 1
AN BN T N
I1 J 1 T 1
E1 F1 u G 1
H 2 I2 J2 T 2
1
D 2 E2 F2 u 2 G 2
+s = H N – 1 IN – 1 J N – 1 T N – 1
D N – 1 EN – 1FN – 1 u N –1 G H N IN TN
N –1
D N EN u N G N
L 1 M1 u 1
V 1
(18) K 2 L 2 M2 u 2 V 2
+ =
where K N – 1 L N – 1 MN – 1 u N –1 V N –1
Q a –1 KN LN uN V
B 1 = – [1 + 3 1 ] [ 1 2 ] N
1Q 1 (∆r 1)
(19)
–1
3Q 1 a a
C 1 = [1 + ] [ 1 2 + r1 1 ] where
Q
1 1 (∆r 1) 1 ∆r 1
Q
I 1 = 3 1 (r2 – 2 )
∆r 1
–1
3Q 1 Q 1Q 1
E 1 = [1 + ] [r1 2 1 – r1 ] 1
Q
1 1 1 1Q 1 1
g
J1 = – 1
2∆r 1
F1 = DN = 0
Q
L 1 = 2 1 ( 2 – 22 ) – 2 2 – 12
ai a 1Q 1 r 1∆r 1 r 1 (∆r 1) r 1
Aj = – ri 1
(∆r 1) 2 j ∆r i 2
M1 =
2a i (∆r 1)2
Bj = –
(∆r i )2 g
V 1 = – 1 T1
2∆r 1
ai a
Cj = + ri 1
(∆r 1) 2 j ∆r i Ij = 0
gi
Dj = 1 Jj = –
2∆r i 2∆r i
Z.Y. Lee: Coupled Problem of Thermoelasticity for Multilayered Spheres with Time-Dependent Boundary Conditions 97
gi
Hj = where λ j(j = 1, 2, ..., N) are the eigenvalues of matrix
2∆r i
[A].
K j = 1 2 – r2 1 Substituting equation (23) into (22), we obtain the
(∆r i ) j 2∆r i
equation
L j = – 2 2 + 12
(∆r i ) rj { [P]– 1[A] [P] – s[P]– 1[I][P]} [P]– 1{ T j } = [P]– 1{ F j }
M j = 1 2 + r1 1
(∆r i ) j ∆r i (24)
V j =0 Equation (24) can be rewritten as
gm * *
HN = {diag[A] – s[I]}{ T j } = { F j } (25)
2∆r m
Q
I N = 3 m (r2 + 2 ) where
1Q m m ∆r m
*
KN = 2 T j = [P]– 1{ T j }
(∆r m)2
*
Q –2 F j = [P]– 1{ F j }
LN = 2 m ( – 22 ) – 2 2 – 12
1Q m r m ∆r m r m (∆r m) r m
From equation (25), the following solutions can be
g obtained immediately.
V N = – m TN +1
2∆r m
*
Equations (18) and (19) can be rewritten in the * Fj
Tj= j = 1, 2, ..., N (26)
following matrix forms λj – s
{[M] – s[I]} { T j } + s[N] { u j } = { G j } (20) By applying the inverse Laplace transform to equa-
*
tion (26), we obtain the solution for T j . After we have
*
[R] { T j } + [Q] { u j } = { V } (21) obtained T j , we can then use equations (27) and (28)
given below to obtain the solutions for T j and u j.
where the matrices [M], [N], [R] and [Q] are the corre-
sponding matrices in equations (18) and (19). Substi- {T j } = [P] {T *j } (27)
tuting equation (21) into (20), we have
{u j } = [Q]– 1[ V ] – [Q]– 1[R]{T j } (28)
{[A] – s[I]} { T j } = { F j } (22)
Substituting Tj and uj into equations (8) and (9), we
where obtain the radial and circumferential stresses.
–1
[A] = {[N]– 1 + [Q]– 1[R]} [N]– 1[M] NUMERICAL RESULTS AND DISCUSSIONS
–1
[F j ] = {[N]– 1 + [Q]– 1[R]} {[N]– 1[ G ] – s[Q]– 1[ V ]} In this study, we present some numerical results
for the temperature distributions in multilayered com-
Since the (N × N) matrix [A] is a nonsingular real posite spheres, subjected to the considered boundary
matrix, the matrix [A] possesses a set of N linearly conditions and the resulting displacement and thermal
independent eigenvectors, hence the matrix [A] is stresses. For the multilayered spheres, the geometry
diagonalizable. Therefore, there exists a nonsingular and material quantities of the sphere are shown in Table
transition matrix [P] such that [P] −1[A][P] = diag [A], 1. The table 2 and table 3 show the computation time for
that is, the matrices [A] and diag [A] are similar, where different number of finite difference gird points. The
the matrix diag [A] is defined as inner and outer radii of the sphere are assumed to be
1.0 and 4.5, respectively. The boundary conditions at
λ1 inner and outer surfaces are assumed to be f(t) and
λ1 convective respectively. Each layer is assumed to have
diag [A] = (23) a different thickness (nondimensional thickness of each
λN layer is taken for ∆h 1 = 1, ∆h 2 = 1, and ∆h 3 = 1.5,
respectively). In all the figures, the grid points assumed
98 Journal of Marine Science and Technology, Vol. 12, No. 2 (2004)
Watt
sphere (Rout / R1 = 4.5, h = 200 2 , Θ0 = Θ∞ = 298 Water vapor
m ⋅K 8.00
K)
6.00
layer 1 layer 2 layer 3
Pressure
Titanium Al2O3 Steel(1025)
4.00
Water vapor
0.00
Table 3. Compuation time for uncoupled theory 0.00 10.00 20.00 30.00 40.00 50.00
t
15t, 0≤t≤20
Grids = 69 Grids = 99 Grids = 129 Grids = 219 f(t) =
300, t<20
t=1 2.103024 4.065846 6.259000 17.57527
t = 10 2.633787 5.377733 8.762600 25.53672
Fig. 3. Temperature distribution with time in inner boundary.
t = 30 5.397762 10.96577 18.39645 54.08778
(CPU P3_700, unit = second)
10.00
Water vapor
219 (N = 219).
8.00
Fig. 2 shows water vapour temperature and pres-
sure relation assumed for the inner boundary. The water
6.00
vapour temperature and pressure data were obtained
Pressure
ture gradient varies in each layer because of the differ- 0.00 10.00 20.00 30.00 40.00 50.00
t
ence in the thermal conductivity coefficients. Fig. 6a
-0.00013995t + 0.00461523t2
shows the variation of the radial displacement as a p(t) = -0.00112473t3 + 0.00060406t4, 0≤t≤20
function of the radial coordinate and time. Fig. 6b 8.585 , t<20
shows the displacement distributions with time and
different radial locations. From these figures, we can Fig. 4. Pressure distribution with time in inner boundary.
Z.Y. Lee: Coupled Problem of Thermoelasticity for Multilayered Spheres with Time-Dependent Boundary Conditions 99
see where the maximum radial displacement may occur. modulus. The circumferential stress varies characteris-
Fig. 7a shows the thermal radial stress distribution σ r tically in each layer in view of the occurrence of
along the radial direction and with time. Fig. 7b shows discontinuities at all interfaces shown in the Fig. 8a.
the thermal radial stress distribution σ r with time and Fig. 8b shows the circumferential stress distribution
different radial locations. As expected, the circumfer- with time and different radial locations. In all the
ential stress distribution exhibits significant jumps at figures, the dotted lines express the results of uncoupled
all interfaces as shown in Fig. 8a. These discontinuities cases. It is obvious that there is distinct difference
are due to the differences in material properties such as between the coupled and uncoupled treatment. From
the coefficient of linear thermal expansion and Young’s these figures, it should be concluded that the coupling
(b)
(a) 350.00
350.00
coupled
coupled
300.00 uncouple
300.00 uncouple
250.00
250.00 r=1.5
200.00
200.00
T 150.00
T
150.00 r=2.5
100.00
t=30
50.00 0.00
t=1
0.00 -50.00
1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 0.00 10.00 20.00 30.00 40.00 50.00
r t
Fig. 5. (a) Temperature distribution along radial direction and different times, (b) Temperature distribution with time and different radial locations.
(a) (b)
1.6 1.40
coupled
1.4
uncouple 1.20 r=2.5
1.2
t=30
1.00
1
r=3.5
0.8 0.80 r=1.5
T u
0.6
0.60
t=10
0.4
0.40 coupled
0.2 uncouple
t=1
0 0.20
-0.2
0.00
1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 0.00 10.00 20.00 30.00 40.00 50.00
r t
Fig. 6. (a) Radial displacement distribution along radial direction and different times, (b) Radial displacement distribution with time and different
radial locations.
100 Journal of Marine Science and Technology, Vol. 12, No. 2 (2004)
(a) (b)
350.00 20.00
coupled coupled
uncouple 0.00 uncouple
300.00
-20.00
250.00
r=3.5
-40.00
200.00
σr σr -60.00
150.00
t=30 -80.00
r=1.5
100.00 t=10
-100.00
50.00 r=2.5
-120.00
t=1
0.00 -140.00
1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 0.00 10.00 20.00 30.00 40.00 50.00
r t
Fig. 7. (a) Radial stress distribution along radial direction and different times, (b) Radial stress distribution with time and different radial locations.
(a) (b)
500.00 500.00
coupled coupled
uncouple uncouple
400.00 400.00
t=30 r=2.5
300.00
300.00
200.00
200.00 r=3.5
t=10
σθ 100.00 σθ
t=1 100.00
0.00
0.00
-100.00
r=1.5
-100.00
-200.00
-300.00 -200.00
1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 0.00 10.00 20.00 30.00 40.00 50.00
r t
Fig. 8. (a) Circumferential stress along radial direction and different times, (b) Circumferential stress for time and different radial locations.