ASYMPTOTIC BEHAVIOR OF THE TRANSMISSION PROBLEM

OF THE BRESSE BEAM IN THERMOELASTICITY

WAEL YOUSSEF

Abstract. We study the stabilization of the transmission problem of the

Bresse beam in thermoelasticity. A non-exponential stability is established
by using a result extended from the Weyl theorem. On the other hand, a
polynomial stability is proved by using a frequency domain method.

1. Introduction
We consider the transmission problem of the following Bresse beam which is
composed of two parts, elastic and thermoelastic. It is governed by the following
differential equations:



 ρ1 φtt − k(φx + ψ + lw)x − k0 l(ωx − lφ) = 0 in (c, L) × (0, ∞),







 ρ2 ψtt − bψxx + k(φx + ψ + lω) = 0 in (c, L) × (0, ∞),







 ρ1 ωtt − k0 (ωx − lφ)x + kl(φx + ψ + lω) = 0 in (c, L) × (0, ∞),






 δ1 Φtt − k1 (Φx + Ψ + lΩ)x − k01 l(Ωx − lΦ) + lαθ1 = 0 in (0, c) × (0, ∞),



 δ2 Ψtt − b1 Ψxx + k1 (Φx + Ψ + lΩ) + αθ2x = 0 in (0, c) × (0, ∞),







 δ1 Ωtt − k01 (Ωx − lΦ)x + k1 l(Φx + Ψ + lΩ) + αθ1x = 0 in (0, c) × (0, ∞),







 θ1t − θ1xx + α(Ωx − lΦ)t = 0 in (0, c) × (0, ∞),





θ2t − θ2xx + αΨxt = 0 in (0, c) × (0, ∞),
(1.1)
with t > 0, 0 < x < L, and 0 < c < L. The unknowns θ1 and θ2 represent
the temperature deviations from a reference temperature along the longitudinal
and vertical directions (see [20], [7]). ρ1 , ρ2 , k, b, k0 , δ1 , δ2 , k1 , b1 , and k01 are
positive physical constants. L represents the length of the beam. α ≥ 0 is the
coupling parameter. Assume that the system satisfies the following boundary and
transmission conditions, respectively:
{
Φ(0, t) = Ψ(0, t) = Ω(0, t) = θ1 (0, t) = θ2 (0, t) = 0,
(1.2)
φ(L, t) = ψ(L, t) = ω(L, t) = 0,
1
2 WAEL YOUSSEF

and


 kφ(c, t) = k1 Φ(c, t), bψ(c, t) = b1 Ψ(c, t),


 ω(c, t) = Ω(c, t) = θ1 (c) = θ2 (c) = 0,
−M Φtt (c, t) = k1 (Φx + Ψ + lΩ)(c, t) − k(φx + ψ + lω)(c, t), (1.3)


 −M Ψtt (c, t) = b1 Ψx (c, t) − bψx (c, t),


k01 (Ωx − lΦ)(c, t) = k0 (ωx − lφ)(c, t),

where M > 0.

c
b

L
0

The initial conditions for this problem are:

(φ, φt , ψ, ψt , ω, ωt )(x, 0) = (φ0 , φ1 , ψ, ψ1 , ω0 , ω1 ) on (c, L),
(1.4)
(Φ, Φt , Ψ, Ψt , Ω, Ωt , θ1 , θ2 ) = (Φ0 , Φ1 , Ψ0 , Ψ1 , Ω0 , Ω1 , θ10 , θ20 )(x) on (0, c).
The energy of solutions of the system (1.1) is defined by
M M
E(t) = E1 (t) + E2 (t) + |Φt (c, t)|2 + |Ψt (c, t)|2 ,
2 2
where  ∫ L(

 E1 (t) = 1 ρ1 |φt |2 + ρ2 |ψt |2 + ρ1 |ωt |2 + b|ψx |2
2 c ) (1.5)

 +k|φx + ψ + lω|2 + k0 |ωx − lφ|2 dx
and
 ∫ c(

 E2 (t) = 1 δ1 |Φt |2 + δ2 |Ψt |2 + δ1 |Ωt |2 + b1 |Ψx |2
2 0 ) (1.6)

 +k1 |Φx + Ψ + lΩ|2 + k01 |Ωx − lΦ|2 + |θ1 |2 + |θ2 |2 dx.
Moreover, the dissipation relation is
∫ c
d ( )
E(t) = − |θ1x |2 + |θ2x |2 dx ≤ 0. (1.7)
dt 0
One of the main issues in the study of the stabilization of a beam is the case
when the damping is locally distributed on a part of this beam. In the case of
thermoelasticity, especially when temperature is locally distributed on the beam, we
have a mathematical complication and difficulty to prove the well-posedness of the
system of differential equations which describes the motion. In our present work, we
overcome this difficulty by considering a transmission problem in thermoelasticity.
The thermoelastic part is damped by the temperature. However, the elastic part
is indirectly damped by the thermoelastic part through the transmission data and
conditions. From a physical viewpoint, the terms
k0 (ωx − lφ), k(φx + ψ + lω), and bψx
 3

denote the axial force, the shear force and the bending moment, respectively. The
functions φ, ψ and ω represent, respectively, the transverse displacement of a curved
beam, the rotation angle of the filament and the longitudinal displacement. We de-
note by k0 = EH, k = GH, and b = EI. ρ1 , ρ2 , l, G, E, and H are positive
constants characterizing physical properties of the beam and the filament. In ad-
dition, l = R1 , where R is the radius of curvature. The condition (1.3)3 describes
the equality between the acceleration of the transversal displacement and the shear
forces acting on the particle c which separetes the elastic and thermoelastic parts
during the motion. (1.3)4 describes the equality between the acceleration of the
shear angle displacement and the bending moments. However, (1.3)5 condition
describes the equality between the axial forces. Note that the Bresse system is a
spacial case of a general model for 3-d nonlinear thermoelastic beams derived by
Lagnese, Leugering and Schmidt [18]. The isothermal system is exactly the system
obtained by Bresse [5] in 1856 (see [20] for more details).
The study of the stabilization of the Bresse beam (or the circular arch problem) was
attracted by the researchers since 2009. Rao and Liu [20] was the first started who
considered the Bresse system damped by two temperature equations. Precisely, the
two wave equations about the longitudinal displacement and shear angle displace-
ment are effectively damped by the dissipation from the two heat equations. They
showed the exponentially decay rate when the wave speed of the vertical displace-
ment coincides with the wave speed of longitudinal displacement or of the shear
angle displacement. Otherwise, they obtained a polynomial type decay rate.
Later, many researchers proved the stabilization of the Bresse beam under differ-
ent kind of damping such that frictional, boundary, thermoelastic, viscoelastic, and
thermoviscoleastic [1, 2, 3, 9, 17, 26]. M.L. Santos [23] considered the thermoelastic
beam system where the oscillations are defined by the Bresse model and the heat
conduction is given by Green and Naghdi theories. He established that the system is
exponentially stable if and only if the wave speeds associated to the hyperbolic part
of the system are equal. In the case of of nonexponential stability, he showed that
the solution decays polynomially and he proved that the rate of decay is optimal.
Z-J. Han and G-Q Xu [14] studied the energy decay rate of a thermoelastic Bresse
system with variable coefficients. He assumed that the thermo-propagation in the
system satisfies the Cattaneo’s law, which can eliminate the paradox of infinite
speed of thermal propagation in the assumption of the Fourier’s law in the classical
theory of thermoelasticity. Also, they discussed the effect of a boundary viscoelastic
damping on the stability of this system and proved that the energy of the system
with a boundary viscoelastic damping decays exponentially. However, it no longer
decays exponentially if there is no boundary damping. We also refer to [6], where
F. Dell’Oro provided a comprehensive stability analysis of the thermoelastic Bresse
system by assuming a temperature evolution of Gurtin-Pipkin type. He proved a
necessary and sufficient condition for exponential stability in terms of the structural
parameters of the problem. In addition, he obtained a complete characterization of
the longtime behavior of Bresse-type systems with Fourier, Maxwell-Cattaneo and
Coleman-Gurtin thermal laws. We also mention some transmission problems in
thermoelasticity. In [15], Z-J. Han and G-Q Xu considered a transmission problem
between elastic and thermoelastic material. They assumed that these two materi-
als are connected by a vibrating concentrated mass. They obtained an asymptotic
expressions of the eigenvalues of the system. They established the non-exponential
4 WAEL YOUSSEF

stability. However, by the frequency domain method together with some multiplier
techniques, the showed the polynomial decay of the system. In [25], the transmis-
sion problem of the Timoshenko system in thermoelasticity with two concentrated
masses is investigated. The nonexponential stability and a polynomial stability are
proved. From an other viewpoint, F. Gallego and J. E. Munoz Rivera [11] studied
the energy decay for the thermoelastic Bresse system in the whole line with two
dissipative mechanisms, given by heat conduction (Types I and III). They proved
that the decay rate of the solutions are very slow and the wave speeds of propa-
gation have influence on the decay rate with respect to the regularity of the initial
data. In [24], B. Said-Houari and A. Soufyane also treated the stabilization of the
Bresse system in the whole line. Their system is coupled with the Fourier law of
heat conduction. They proved that the decay rate of the solution is very slow. In
addition, they showed that the wave speed of the first two equations still control
the decay rate of the solution with respect to the regularity of the initial data.
On an other side, some notations and definitions are needed to state the method
used to show our nonexponential stability result . Let X be a Banach space. A
bounded operator S ∈ L (X) is called a Fredholm operator if there is T ∈ L (X)
such that IdX − T S and IdX − ST are compact. We denote by
σess (S) := C\{λ ∈ C; λ IdX − S is a Fredholm operator}
the essential spectrum of S. Moreover, we define the essential spectral radius by
ress (S) := max{|λ|; λ ∈ σess (S)}.
For more details see [8].
In this work, to establish our nonexponantial stability we use the theorem intro-
duced by J. Rivera and R. Racke which is based on the Weyl theorem ([21], Theorem
3.3).
Theorem 1.1. Let H1 be a Hilbert space with norm ∥ . ∥H1 , and let ∅ ⊈ H2 ⊂ H1
be a closed subspace with orthogonal projection P : H1 −→ H2 . Let Sj = (Sj (t))t≥0
be a C0 -semigroup on Hj for j = 1, 2. If there exists t0 > 0 such that for t ≥ t0 > 0
we have
(1) ress (S2 (t)) ≥ 1,
(2) S1 (t) − S2 (t) : H2 −→ H1 is compact,
then S1 is not exponentially stable.
However, our result on the polynomial stability, it is based on the Borichev and
Tomilov theorem [4].
Theorem 1.2. Let A be the generator of C 0 -semigroup of contractions on a Hilbert
space X. Then,
iR ⊆ ρ(A), ∥(iλI − A)−1 ∥ < C|λ|s , (1.8)
for some s > 0 and for all λ ∈ R if and only if we have
C
∥S(t)W 0 ∥X ≤ 1/s ∥W 0 ∥D(A) , (1.9)
t
and for some constant C > 0 and for all W0 ∈ D(A).
The remaining part of this paper is organized as follows. In section 2 we prove
the well posedness of the system (1.1)-(1.4) by formulating an appropriate Hilbert
state space setting. Section 3 is devoted to the nonexponential stability result. In
section 4 we establish the polynomial stability.
 5

2. Well-Posedness
For well-posedness, let us introduce the following spaces:
{ }
VL1 (c, L) = f ∈ H 1 (c, L); f (L) = 0 ,
{ }
V0j (0, c) = f ∈ H j (0, c); f (0) = 0 , j = 1, 2,
( )2 ( )2 ( )2
T := VL1 (c, L) × H01 (c, L) × V01 (0, c) × H01 (0, c) × L2 (0, c) ,
{ }
H1 := (φ, ψ, ω, Φ, Ψ, Ω, θ1 , θ2 ) ∈ T ; kφ(c, t) = k1 Φ(c, t), bψ(c, t) = b1 Ψ(c, t) ,
and ( ) ( )
3 3
L := L2 (c, L) × L2 (0, c) .
{ }
(2n1 + 1)π (2n2 + 1)π
Lemma 2.1. For lc ∈ / nπ, , lL ̸= , and lL − lc ̸= n3 π,
2 2
where n, n1 , n2 , n3 ∈ N, the space
( )2 ( )3 ( )3
H := H1 × L2 (0, c) × L2 (c, L) × L2 (0, c) × C2 .
is a Hilbert space.
Proof. Let us consider the bilinear form h defined over H × H by:
 ∫ L[



 h(W, f ) :=
W ρ1 ueu + ρ2 ve
v + ρ1 zez + bψx ψex + k(φx + ψ + lω)(φ ex + ψe + le
ω)

 ]


c

 +k0 (ωx − lφ)(e ωx − lφ)
e dx+
 ∫ c[
 e + δ2 V Ve + δ1 Z Z
δ1 U U e + b1 Ψx Ψ
e x + k1 (Φx + Ψ + lΩ)(Φ
ex + Ψ e + lΩ)
e



 0 ]

 e + θ1 θe1 + θ2 θe2 dx
e x − lΦ)

 +k01 (Ωx − lΦ)(Ω


+M pe p + M qeq.
for all
W := (φ, ψ, ω, Φ, Ψ, Ω, θ1 , θ2 , u, v, z, U, V, Z, p, q) ∈ H
and
f := (φ,
W e ψ,eω e Ψ,
e , Φ, e θe1 , θe2 , u
e Ω, e , Ve , Z,
e, ve, ze, U e pe, qe) ∈ H.
h is positive-definite over H. Indeed,
∫ L[ ]
h(W, W ) = 0 ⇒ ρ1 |u|2 + ρ2 |v|2 + ρ1 |z|2 + b|ψx |2 + k|φx + ψ + lω|2 + k0 |ωx − lφ|2 dx
c
∫ c[
+ ρ1 |U |2 + ρ2 |V |2 + ρ1 |Z|2 + b|Ψx |2 + k|Φx + Ψ + lΩ|2 + k0 |Ωx − lΦ|2
0
]
+|θ1 |2 + |θ2 |2 dx + M |p|2 + M |q|2 = 0
⇒ p = q = 0 in C, u = v = z = 0 on (c, L), U =V =Z=0 on (0, c),
Φx + Ψ + lΩ = Ωx − lΦ = Ψx = θ1 = θ2 = 0 on (0, c),
ψx = φx + ψ + lω = ωx − lφ = 0 on (c, L).
So Ψ = cte on (0, c). But, Ψ(0) = 0. Hence Ψ = 0 and so
{
Φx + lΩ = 0 on (0, c),
(2.1)
Ωx − lΦ = 0 on (0, c).
6 WAEL YOUSSEF

Therefore,
Ωxx + l2 Ω = 0, on (0, c). (2.2)

Thus,
Ω = A cos(lx) + B sin(lx). (2.3)

As Ω(0) = Ω(c) = 0, then Ω = 0 on (0, c) if lc ̸= nπ. Consequently Φ = 0 on (0, c).

b1
On the other hand, by the transmission conditions, we have ψ(c) = Ψ(c). Then
b
ψ = 0 on (c, L) and we get
{
φx + lω = 0 on (c, L),
(2.4)
ωx − lφ = 0 on (c, L)

and so
ω = C cos(lx) + D sin(lx).

Since by the transmission condition ω(c) = Ω(c) = 0 and the boundary condition
ω(L) = 0 then
{
C cos(lc) + D sin(lc) = 0,
C cos(lL) + D sin(lL) = 0.

(2n1 + 1)π (2n2 + 1)π

Thus, for lc ̸= and lL ̸= , we have
2 2
{
C + D tan(lc) = 0,
C + D tan(lL) = 0.

Therefore,
( )
D tan(lc) − tan(lL) = 0.
n3 π
Hence, D = 0 since lL − lc ̸= i.e. lL ̸= lc + n3 π. Thus, C = 0, ω = 0, and φ
l
on (c, L). Therefore, W = 0H . Consequently, H is a Hilbert space under the scalar
product h associated to norm energy.
□

Now, Set φt = u, ψt = v, ωt = z, Φt = U , Ψt = V , and Ωt = Z. Then for

W := (φ, ψ, ω, Φ, Ψ, Ω, θ1 , θ2 , φt , ψt , ωt , Φt , Ψt , Ωt , p, q)

we define the linear unbounded operator A by:

 

 (φ, ψ, ω, Φ, Ψ, Ω, θ1 , θ2 , u, v, z, U,[V, Z, p, q) ∈ H/ 


 ( )3 ( )5 ] 

(φ, ψ, ω, Φ, Ψ, Ω, θ1 , θ2 ) ∈ H1 ∩ H 2 (c, L) × H 2 (0, c) ,
D(A) := ,

 (u, v, z, U, V, Z, θ1xx − α(Zx − lU ), θ2xx − αVx ) ∈ H1 , 


 

p = U (c), q = V (c)
 7

and  
u
 v 
 
 z 
 
 U 
 
 V 
 
 Z 
 
 θ − α(Z − lU ) 
 1xx x 
 θ − αV 
 2xx x 
 k lk 
 (φxx + ψx + lωx ) +
0
(ωx − lφ) 
 ρ ρ 
 1 1 
 b k 
 ψxx − (φx + ψ + lω) 
AW =  ρ ρ .
 2 2 
 k0 lk
(ωxx − lφx ) − (φx + ψ + lω) 
 
 ρ1 ρ1 
 k lk0 lα 
 
 δ (Φxx + Ψx + lΩx ) + δ (Ωx − lΦ) − δ θ1 
 1 1 1 
 b k α 
 Ψ − (Φ + Ψ + lΩ) − θ 
 xx x 2x 
 k δ2 δ2 δ2 
 0 lk α 
 (Ωxx − lΦx ) − (Φx + Ψ + lΩ) − θ1x 
 δ1 δ1 δ1 
 [ ] 
 − 1 k1 (Φx + Ψ + lΩ)(c) − k(φx + ψ + lω)(c) 
 M 
 1 [ ] 
− b1 Ψx (c) − bψx (c)
M
Then the system (1.1) can be reformulated into an evolution problem of first order
on H in the form {
W ′ (t) = AW (t), t > 0
. (2.5)
W (0) = W 0 ∈ D(A),
Using a classical semigroup theorem [13, 16], we have the following well-posedness
result:
Theorem 2.1. The operator A generates a C 0 - semigroup S(t) of contractions on
H. Then, for any ( ∈ D(A), the
( initial )data W
0
) W (t) is a strong solution of (2.5)
i.e. W (t) ∈ C 1 [0, ∞), H ∩ C 0 [0, ∞), D(A) .

Proof. For W = (φ, ψ, ω, Φ, Ψ, Ω, θ1 , θ2 , u, v, z, U, V, Z, p, q) ∈ D(A), a straightfor-

ward calculations leads to
∫ c
( ) ( )
Re⟨AW, W ⟩H := Re h(AW, W ) = − |θ1x |2 + |θ2x |2 dx ≤ 0. (2.6)
0
Next, let us prove that
AW = (f1 , f2 , ..., f16 ) (2.7)
has a unique solution, for any (f1 , f2 , ..., f16 ) ∈ H.
In order, the first forth equations of (2.7) imply that
u = f1 , v = f2 , z = f3 , p = f3 , q = f4 .
From an other side, substituting of U, V , and Z in the seventh and eighth equations
of (2.7) and using the standard elliptic PDE theory yield to the existence of the
unique solution
θ1 , θ2 ∈ H02 (0, c)
8 WAEL YOUSSEF

of the equation {
θ1xx = α(f6x − lf4 ) + f7 ,
θ1 (0) = θ1 (c) = 0
and {
θ2xx = αf5x + f8 .
θ2 (0) = θ2 (c) = 0,
It remains to establish the existence of (φ, ψ, ω, Φ, Ψ, Ω) satisfying
k(φxx + ψx + lωx ) + lk0 (ωx − lφ) = ρ1 f9 on (c, L), (2.8)
bψxx − k(φx + ψ + lω) = ρ2 f10 on (c, L), (2.9)
k0 (ωxx − lφx ) − lk(φx + ψ + lω) = ρ1 f11 on (c, L), (2.10)
k1 (Φxx + Ψx + lΩ) + lk01 (Ωx − lΦ) = lαθ1 + δ1 f12 on (0, c), (2.11)
b1 Ψxx − k1 (Φx + Ψ + lΩ) = αθ2x + δ2 f13 on (0, c), (2.12)
k0 (Ωxx − lΦx ) − lk(Φx + Ψ + lΩ) = αθ1x + δ1 f14 on (0, c), (2.13)
k1 (Φx + Ψ + lΩ)(c) − k(φx + ψ + lω)(c) = −M f15 on C, (2.14)
b1 Ψx (c) − bψx (c) = −M f16 on C. (2.15)
For this aim, let us consider the continuous and coercive linear form
 ( )
 e ,ω
e ψ,
A (φ, ψ, ω, Φ, Ψ, Ω), (φ, e Ψ,
e , Φ, e Ω)
e

 ∫ L[

 ]

 := bψx ψex + k(φx + ψ + lω)(φ ex + ψe + le
ω ) + k0 (ωx − lφ)(e
ωx − lφ)
e dx
c



 ∫ [ ]


c
e x + k1 (Φx + Ψ + lΩ)(Φ
ex + Ψ
e + lΩ)
e + k01 (Ωx − lΦ)(Ω
e x − lΦ)
e dx
 + b1 Ψx Ψ
0
eω
e ψ,
for (φ, ψ, ω, Φ, Ψ, Ω) and (φ, e Ψ,
e , Φ, e Ω)
e belong to the Hilbert space
 ( )2 ( )2 
 (φ, ψ, ω, Φ, Ψ, Ω) ∈ VL1 (c, L) × H01 (c, L) × V01 (0, c) × H01 (0, c); 
V := .
 
kφ(c, t) = k1 Φ(c, t), bψ(c, t) = b1 Ψ(c, t)
Let us denote by F the continuous linear form defined by
 ∫ L ∫ L ∫ L

 e e e e e

 e ψ, ω
F(φ, e , Φ, Ψ, Ω) := −ρ1 e − ρ2
f9 φdx f10 ψdx − ρ1 e dx
f11 ω




c c c
 ∫ ∫ c ∫ c
c( ) ( ) ( )
 − e
lαθ1 + δ1 f12 Φdx − e
αθ2x + δ2 f13 Ψdx + e
αθ1x + δ1 f14 Ωdx



 0 0 0




e + M f16 ψ(c).
+M f15 φ(c) e
By applying the Lax-Milgram theorem, there exists a unique (φ, ψ, ω, Φ, Ψ, Ω) in V
such that
( )
eω
e ψ,
B (φ, ψ, ω, Φ, Ψ, Ω), (φ, e Ψ,
e , Φ, e Ω)
e = F(φ, eω
e ψ, e Ψ,
e , Φ, e Ω)
e
eω
e ψ,
for every (φ, e Ψ,
e , Φ, e Ω)
e ∈ V. Therefore, the equation (2.7) has a unique solution
W ∈ D(A) and so 0 ∈ ρ(A). Thus, by the resolvant identity, for λ > 0, we have
R(λI − A) = H (see Theorem 1.2.4 in [19]). So, by applying the Lumer-Philips
theorem the proof is achieved. □
 9

3. Non-exponential stability
We start this section by stating some observability results in the two following
lemmas. These estimations will be the main ingredient for the proof of the main
theorem.
Lemma 3.1. Let T > 0 be arbitrary and (φ, ψ, ω) the first three components of
(φ, ψ, ω, Φ, Ψ, Ω, θ1 , θ2 ) ∈ H1 be the solution of the system


 ρ1 φtt − k(φx + ψ + lω)x − k0 l(ωx − lφ) = 0 in (c, L) × (0, ∞),






 ρ2 ψtt − bψxx + k(φx + ψ + lω) = 0 in (c, L) × (0, ∞),
(3.16)



 ρ ω
1 tt − k 0 (ω x − lφ) x + kl(φ x + ψ + lω) = 0 in (c, L) × (0, ∞),





(φ, φt , ψ, ψt , ω, ωt )(., 0) = (φ0 , φ1 , ψ 0 , ψ 1 , ω 0 , ω 1 ) in (c, L),
then, there exist three positive constants C1 > 0, C2 > 0, and C3 > 0 depending on
T, ρ1 , ρ2 , b, k, k0 , and γ > 0 such that
∫ T ∫ T ∫ T
|φt (c)| dt +
2
|(φx + ψ + ω)(c)| dt +
2
|φx (L)|2 dt ≤ C1 E(0), (3.17)
0 0 0
∫ T ∫ T ∫ T
|ψt (c)|2 dt + |ψx (c)|2 dt + |ψx (L)|2 dt ≤ C2 E(0), (3.18)
0 0 0
∫ T ∫ T
|(ωx − lφ)(c)| dt + 2
|ωx (L)|2 dt ≤ C3 E(0). (3.19)
0 0
The constant γ will be determined in the proof.
Lemma 3.2. Let T > 0 be arbitrary and (Φ, Ψ, Ω, θ1 , θ2 ) the last five components
of (φ, ψ, ω, Φ, Ψ, Ω, θ1 , θ2 ) ∈ H1 be the solution of the system

 δ1 Φtt − k1 (Φx + Ψ + lΩ)x − k01 l(Ωx − lΦ) + lαθ1 = 0 in (0, c) × (0, ∞),







 δ2 Ψtt − b1 Ψxx + k1 (Φx + Ψ + lΩ) + αθ2x = 0 in (0, c) × (0, ∞),




δ1 Ωtt − k01 (Ωx − lΦ)x + k1 l(Φx + Ψ + lΩ) + αθ1x = 0 in (0, c) × (0, ∞),







 θ1t − θ1xx + α(Ωx − lΦ)t = 0 in (0, c) × (0, ∞),





θ2t − θ2xx + αΨxt = 0 in (0, c) × (0, ∞).
(3.20)
Then, there exist three positive constants C4 > 0, C5 > 0, and C6 > 0 depending
on T, δ1 , δ2 , b1 , k1 , k01 , and γ > 0 such that
∫ T ∫ T ∫ T
|Φt (c)| dt +
2
|(Φx + Ψ)(c)| dt +
2
|Φx (0)|2 dt ≤ C4 E(0), (3.21)
0 0 0
∫ T ∫ T ∫ T
|Ψt (c)|2 dt + |Ψx (c)|2 dt + |Ψx (0)|2 dt ≤ C5 E(0), (3.22)
0 0 0
∫ T ∫ T
|(Ωx − lΦ)(c)| dt + 2
|Ωx (0)|2 dt ≤ C6 E(0). (3.23)
0 0
10 WAEL YOUSSEF

Now, we are going to prove the lemma 3.1. The proof of the lemma 3.2 is
concluded by repeating the same arguments.

Proof. Let Υ ∈ C 1 ([c, L]) be a given function satisfying Υ(c) = −Υ(L) = 2γ, where
γ > 0. Multiplying the first equation of (3.16) by Υ(x)(φx +ψ +lω) and integrating
the resulting equation over [0, T ] × (c, L) we get
 ∫ T∫ L ∫ T∫ L


 ρ1 φtt Υ(x)(φx + ψ + lω)dx − k (φx + ψ + lω)x Υ(x)(φx + ψ + lω)dx
0∫ c∫ 0 c

 −k
T L
 0 (ωx − lφ)Υ(x)(φx + ψ + lω)dx = 0.
0 c
(3.24)
Therefore,
 ∫ T∫ L ∫ L

 −ρ
T

 1 Υ(x)φ t (φxt + ψ t + lω t )dxdt + ρ 1 Υ(x) [φt (φx + ψ + lω)]0 dx

 ∫ 0T ∫ cL ∫ T[ c ]L

Υx Υ(x)
+k |φx + ψ + lω| dxdt − k
2
|φx + ψ + lω| 2
dt

 2 2

 ∫0 T ∫c L 0 c


 −k0 (ωx − lφ)Υ(x)(φx + ψ + lω)dx = 0.
0 c
(3.25)
Thus,
 ∫ T∫ L ∫ T[ ]L

 Υx Υ(x)

 ρ |φ | 2
dxdt − ρ |φ |2
dt


1
0∫ c∫ 2
t 1
2
t

 0 ∫ c


T L L

 −ρ1 Υ(x)φt (ψt + lωt )dxdt + ρ1
T
Υ(x) [φt (φx + ψ + lω)]0 dx
∫ 0T ∫ cL ∫ Tc [ ]L

 Υ Υ(x)

 +k
x
|φ + ψ + lω| 2
dxdt − k |φ + ψ + lω|2
dt

 2
x
2
x

 ∫
0 ∫
c 0 c

 T L

 −k0 (ωx − lφ)Υ(x)(φx + ψ + lω)dx = 0.
0 c
(3.26)
Hence,
 ∫ T ∫ T



 ρ1 γ |φt (c)| dt + ρ1 γ
2
|φt (L)|2 dt



 ∫
0 0 ∫ T


T

 +kγ |(φx + ψ + lω)(c)| dt + kγ
2
|(φx + ψ + lω)(L)|2 dt =



 ∫ T∫ L
0 ∫ T∫ L 0
Υx
−ρ1 |φt |2 dxdt + ρ1 Υ(x)φt (ψt + lωt )dxdt

 2

 ∫ 0 c 0 c ∫ T∫ d


L
Υx

 −ρ Υ(x) [φ (φ + ψ + lω)]
T
dx − k |φx + ψ + lω|2 dxdt


1 t x 0
2

 ∫c T ∫ L 0 c



 +k0 (ωx − lφ)Υ(x)(φx + ψ + lω)dx.
0 c
(3.27)
So, by the Cauchy-Shwarz inequality, the boundary conditions, and the definition
of E we get (3.17).
By a similar way we establish the estimations (3.18) and (3.19). In fact, we multiply
the second and the third equations of (1.1) by Υ(x)ψx and Υ(x)(ωx − lφ). □
 11

Now, let us proceed as Rivera and Racke in [21] and define the semigroup S0 by
the following initial boundary value problem over (0, L):


 ρ1 φett − k(φ ex + ψe + le ω )x − k0 (e
ωx − lφ) e = 0 in (c, L) × (0, ∞),






 ρ2 ψett − bψexx + k(φ
 ex + ψe + le
ω ) = 0 in (c, L) × (0, ∞),
(3.28)


 ρ1 ω
 ett − k0 (eωx − lφ) e x + kl(φex + ψe + le ω ) = 0 in (c, L) × (0, ∞),






e φ
(φ, e ψet , ω
et , ψ, e, ω
et )(x, 0) = (φ e1 , ψe0 , ψe1 , ω
e0 , φ e0 , ω
e1 )(x) in (c, L),
with boundary conditions

φ(x, e t) = ω
e t) = ψ(x, e (x, t) = 0 (x = c, L). (3.29)
Therefore, the above system is conservative i.e.

E(φ, eω
e ψ, e Ψ,
e , Φ, e θe1 , θe2 )(t) =
e Ω, E(φ, eω
e ψ, e , 0, 0, 0, 0, 0)(t)
= E(φ, e
e ψ, ω
e , 0, 0, 0, 0, 0)(0)
= E(φ, e e Ψ,
e ψ, Φ, e θe1 , θe2 )(0), ∀t > 0.

Note that in (c, L) the problem is purely elastic. Thus, the associated energy is
conservative i.e.
E1 (t) = E1 (0), ∀t > 0. (3.30)
Hence, the semigroup S0 associated to (3.28) and (3.29), defined on the Hilbert
space
( )3 ( )5 ( )3 ( )3 ( )2
H0 := H01 (c, L) × {0} × L2 (c, L) × {0} × {0C }

by
{
e0 , ψe0 , ω
S0 (φ e1 , ψe1 , ω
e0 , 0, 0, 0, 0, 0, φ e1 , 0, 0, 0, 0, 0)
:= (φ, e
e ψ, ω et , ψet , ω
e , 0, 0, 0, 0, 0, φ et , 0, 0, 0, 0, 0)
( )
is a unitary semigroup ad so ress S0 (t) = 1. Thereby, after the introduction of the
semigroup S0 , the statement and the proof of the non-exponential stability result
is ready.
Theorem 3.1. The semigroup S with S(t) = eAt is not exponentially stable.
Proof. Let
(φn0 , ψ0n , ω0n , 0, 0, 0, 0, 0, φn1 , ψ1n , ω1n , 0, 0, 0, 0, 0, 0, 0, 0)
be a bounded sequence of initial data in H0 and
(φn , ψ n , ω n , Φn , Ψn , Ωn , θ1n , θ2n , φnt , ψtn , ωtn , Φnt , Ψnt , Ωnt , pn , q n )
be the corresponding solution to the transmission problem (1.1)-(1.2) with asso-
ciated semigroup S(t) = eAt . Let (φen , ψen , ω
e n ) be the solution of the uncoupled
system (3.28)-(3.29) with
e n, Ψ
(Φ e n , θe1n , θe2n ) := (0, 0, 0, 0, 0)
e n, Ω
12 WAEL YOUSSEF

e
with associated semigroup S0 (t) = eAt . We denote by
χn := φn − φ en , ζ n := ψ n − ψen , τ n := ω n − ω
en,
n e n n n n e
η := Φ − Φ = Φ , ξ := Ψ − Ψ = Ψ , ϵn := Ωn − Ω
n n n e n = Ωn ,
σ1 := θ1 − θe1 = θ1 , σ2 := θ2 − θe2 = θ2 .
n n n n n n n n

The aim is to show that there exists a subsequence of

Λn := (χn , ζ n , τ n , η n , ξ n , ϵn , σ1n , σ2n , χnt , ζtn , τtn , ηtn , ξtn , ϵnt , P n , Qn )
converges in H. Indeed (χn , ζ n , τ n , η n , ξ n , ϵn , σ1n , σ2n ) satisfies the following system


 ρ1 χntt − k(χnx + ζ n + lτ n )x − k0 l(τxn − lχn ) = 0 in (c, L) × (0, ∞),







 ρ2 ζttn
− bζxx n
+ k(χnx + ζ n + lτ n ) = 0 in (c, L) × (0, ∞),







 ρ1 τttn − k0 (τxn − lχn )x + kl(χnx + ζ n + lτ n ) = 0 in (c, L) × (0, ∞),








n
δ1 ηtt − k1 (ηxn + ξ n + lϵn )x − k01 l(ϵnx − lη n ) + ασ1n = 0 in (0, c) × (0, ∞),




 n
δ2 ξtt − b1 ξxx n
+ k1 (ηxn + ξ n + lϵn ) − k01 (ϵnx − lη n ) + ασ2x n
= 0 in (0, c) × (0, ∞),





 δ1 ϵntt − k01 (ϵnx − lη n )x + k1 l(ηxn + ξ n + lϵn ) + ασ1n = 0 in (0, c) × (0, ∞),








n
σ1t − σ1xxn
+ α(ϵnxt − lηtn ) = 0 in (0, c) × (0, ∞),








n
σ2t − σ2xxn
+ αξxt n
= 0 in (0, c) × (0, ∞),







 (η n , ηtn , ξ n , ξtn , ϵn , ϵnt , σ1n , σ2n )(0, .) = (0, 0, 0, 0, 0, 0, 0, 0) in (0, c),

(χn , χnt , ζ n , ζtn , τ n , τtn )(0, .) = (0, 0, 0, 0, 0, 0) in (c, L).
(3.31)
The corresponding boundary and transmition conditions satisfied
 n

 η (t, 0) = ξ n (t, 0) = ϵn (t, 0) = σ1n (t, 0) = σ2n (t, 0) = 0,
 n

 η (t, c) = Φn (t, c), ξ n (t, c) = Ψn (t, c), ϵn (t, c) = Ωn (t, c) = 0,
σ1n (t, c) = θ1n (t, c) = 0, σ2n (t, c) = θ2n (t, c) = 0, (3.32)



 χ n
(t, c) = φ n
(t, c), ζ n
(t, c) = ψ n
(t, c), τ n
(t, c) = ω n
(t, c) = 0,
 n
χ (t, L) = ζ n (t, L) = τ n (t, L) = 0.
The associated energy to the system (3.31)-(3.32)
M n M
Fn (t) = F1n (t) + F2n (t) + | Φ (c) |2 + 2| Ψnt (c) |2 , (3.33)
2 | t{z } | {z }
:=P n :=Qn

where
 ∫ L (

 1
F1n (t) = ρ1 |χnt |2 + ρ2 |ζtn |2 + ρ1 |τtn |2 + b|ζxn |2 + k|χnx + ζ n + lτ n |2
2 c )

 +k0 |τxn − lχn |2 dx
and
 ∫ c (

 1
F2n (t) = δ1 |ηtn |2 + δ2 |ξtn |2 + δ1 |ϵnt |2 + b1 |ξxn |2 + k1 |ηxn + ξ n + lϵn |2
2 0 )

 +k01 |ϵnx − lη n |2 + |σ1n |2 + |σ2n |2 dx.
 13

By a straightforward calculation, we have the identity

 ∫
 d d 1 c( n 2 )

 F1n (t) + F2n (t) + |σ1x | + |σ2x n 2
| dx

 dt dt 2

  0  
 = k(χn + ζ n + lτ n )χn L + bζ n ζ n L + k (τ n − lχn )τ n L
x t  x t  0 x t 
cc cc cc (3.34)

 n n n n n n

 +k (η n
+ ξ n
+ lϵ )η  + b ξ ξ  + k (ϵ n
− lη )ϵ t


1 x
c
t
c 0
1 x t
c 0
01
c
x
0
   
 +αξ n σ n  + σ n σ n  + αϵn σ n  + σ n σ n  . 
t 2 2x 2 t 1 1x 1
0 0 0 0

Next, thanks to (3.32), we obtain from (3.34):

 ∫
 d d 1 c( n 2 )

 F1n (t) + F2n (t) + |σ1x | + |σ2x n 2
| dx

 dt dt 2 0



 = k(φnx + ψ n + lω n )(L)φnt (L) − k(φnx + ψ n + lω n )(c)φnt (c)








 −k φenx (L)φnt (L) + k φ enx (c)φnt (c) + bψxn (L)ψtn (L) − bψexn (L)ψtn (L)







 −bψxn (c)ψtn (c) + bψexn (c)ψtn (c) + k0 (ωxn − lφnx )(L)ωtn (L)







 −k0 (ωxn − lφnx )(c)ωtn (c) − k0 ω exn (L)ωtn (L) + k0 ω exn (c)ωtn (c)




+k1 (Φnx + Ψn + lΩn )(c)Φnt (c) − k1 (Φnx + Ψn + lΩn )(0)Φnt (0) (3.35)





 e nx (c)Φnt (c) + k1 Φ
e nx (0)Φnt (0) + b1 Ψxn (c)Ψnt (c) − b1 Ψ e nx (c)Ψnt (c)

 −k1 Φ





 e nx (0)Ψnt (0) + k01 (Ωnx − lΦnx )(c)Ωnt (c)

 −b1 Ψnx (0)Ψnt (0) + b1 Ψ







 −k01 (Ωnx − lΦnx )(0)Ωtn (0) − k01 Ω e xn (c)Ωnt (c) + k01 Ω e nx (0)Ωnt (0)







 +αΩnt (c)θ1n (c) − αΩnt (0)θ1n (0) + αΨnt (c)θ2n (c) − αΨnt (0)θ2n (0)






+θe1x
n
(c)θ1n (c) − θe1x
n
(0)θ1n (0) + θe2xn
(c)θ2n (c) − θe2x
n
(0)θ2n (0).
14 WAEL YOUSSEF

Thus, using the fact that (Φ e n, Ψ e n , θen ) := (0, 0, 0, 0), (3.35) gives
e n, , Ω
 ∫
 d d 1 c( n 2 )

 F (t) + F (t) + |σ1x | + |σ2x n 2
| dx

 dt
1n
dt
2n
2 0



 = k(φnx + ψ n + lω n )(L)φnt (L) − k(φnx + ψ n + lω n )(c)φnt (c)








 −k φenx (L)φnt (L) + k φenx (c)φnt (c) + bψxn (L)ψtn (L) − bψexn (L)ψtn (L)







 −bψxn (c)ψtn (c) + bψexn (c)ψtn (c) + k0 (ωxn − lφnx )(L)ωtn (L)



−k0 (ωxn − lφnx )(c)ωtn (c) − k0 ω exn (L)ωtn (L) + k0 ω exn (c)ωtn (c) (3.36)







 +k1 (Φnx + Ψn + lΩn )(c)Φnt (c) − k1 (Φnx + Ψn + lΩn )(0)Φnt (0)







 +b1 Ψnx (c)Ψnt (c) − b1 Ψnx (0)Ψnt (0) + k01 (Ωnx − lΦnx )(c)Ωnt (c)







 −k01 (Ωnx − lΦnx )(0)Ωnt (0) + αΩnt (c)θ1n (c) − αΩnt (0)θ1n (0)





+αΨnt (c)θ2n (c) − αΨnt (0)θ2n (0).
∫ T ∫ T
Now, for simplicity, we shall denote f dt by f . Then, integrating over [0, T ],
0 0
we obtain from (3.36)
 ∫ T ∫ T



 F n (t) ≤ −k (φ n
+ ψ n
)(c)φ n
(c) + k enx (c)φnt (c)
φ


x t


0 0

 ∫ T ∫ T ∫ T

−b ψxn (c)ψtn (c) + b ψexn (c)ψtn (c) + k1 (Φnx + Ψn )(c)Φnt (c) (3.37)




0 0 0



 ∫ T


 +b1 Ψnx (c)Ψnt (c).
0

Thus, using Young’s inequality, (3.17), (3.18), (3.19), (3.21), (3.22), (3.23), the
boundary conditions (1.2), and the transmission condition (1.3)1 , the estimation
(3.37) gives
(∫ ∫ T )
T
Fn (t) ≤ C7 |Φnt (c)|2 + |Ψnt (c)|2 , (3.38)
0 0

where C7 > 0 depends on T, ρ1 , ρ2 , δ1 , δ2 , b, k, b1 , k1 , γ, α, E(0), and E1 (0).

Due to the lemma 3.2 and the conditions (1.3)3 and (1.3)4 , the sequences Φnt (c)
and Ψnt (c) are bounded in H 1 (0, T ). Therefore, the compactness of the embedding
n
H 1 (0, T ) ,→ L2 (0, T ) affirms the existence of a convergent subsequences Φt j (c) and
nj
Ψt (c) in L2 (0, T ). Then (3.38) implies that the sequence Fnj (t) is bounded in R.
Moreover, we have
n n n n n n n n
∥(χnj , ζ nj , τ nj , η nj , ξ nj , ϵnj , σ1 j , σ2 j , χt j , ζt j , τt j , ηt j , ξt j , ϵt j , P nj , Qnj )∥2H = 2Fnj (t).
Thus, by the classical Bolzano-Weierestrass theorem, the sequence Fnj (t) has a
convergent subsequence in R and so
n n n n n n n n
(χnj , ζ nj , η nj , ξ nj , σ1 j , σ2 j , χt j , ζt j , τt j , ηt j , ξt j , ϵt j , P nj , Qnj )
 15

has a convergent subsequence in H. Consequently, S(t) − S0 (t) is compact and the

proof of the theorem 3.2 is complete. □

4. Polynomial Stability
The polynomial stabilisation is summarized in the following theorem.
Theorem 4.1. There exists a positive constant C > 0 such that
C
∥S(t)W 0 ∥H ≤ 1/80 ∥W 0 ∥D(A) , (4.1)
t
for all W 0 ∈ D(A).
Proof. We must establish the following two conditions:
iR ⊂ ρ(A), (4.2)
and
1
lim ∥(iλI − A)−1 ∥ < +∞. (4.3)
λ−→∞λ80
Let us start by proving (4.2). Indeed, D(A) has a compact embedding into H,
then A−1 is compact in H. Hence, to prove (4.2) it is sufficient to check that A
has no pure imaginary eigenvalue. Then, suppose that there exists λ ∈ R∗ such
that iλ is an eigenvalue and W = (φ, ψ, ω, Φ, Ψ, Ω, θ1 , θ2 , u, v, z, U, V, Z, p, q) be the
normalized eigenfunction, i.e.
AW = iλW. (4.4)
Therefore, we have
u(x) = iλφ(x), x ∈ (c, L), (4.5)
v(x) = iλψ(x), x ∈ (c, L), (4.6)
z(x) = iλω(x), x ∈ (c, L), (4.7)
U (x) = iλΦ(x), x ∈ (0, c), (4.8)
V (x) = iλΨ(x), x ∈ (0, c), (4.9)
Z(x) = iλΩ(x), x ∈ (0, c), (4.10)
θ1xx − α(Zx − lU ) = iλθ1 (x), x ∈ (0, c), (4.11)
θ2xx − αVx = iλθ2 (x), x ∈ (0, c) (4.12)
k lk0
λ2 φ + (φxx + ψx + lωx ) + (ωx − lφ) = 0, x ∈ (c, L) (4.13)
ρ1 ρ1
b k
λ2 ψ + ψxx − (φx + ψ + lω) = 0, x ∈ (c, L) (4.14)
ρ2 ρ2
k0 lk
λ2 ω + (ωxx − lφx ) − (φx + ψ + lω) = 0, x ∈ (c, L) (4.15)
ρ1 ρ1
k lk0 lα
λ2 Φ + (Φxx + Ψx + lΩx ) + (Ωx − lΦ) − θ1 = 0, x ∈ (0, c) (4.16)
δ1 δ1 δ1
b k α
λ Ψ + Ψxx − (Φx + Ψ + lΩ) − θ2x = 0, x ∈ (0, c)
2
(4.17)
δ2 δ2 δ2
k 0 lk α
λ2 Ω + (Ωxx − lΦx ) − (Φx + Ψ + lΩ) − θ1x = 0, x ∈ (0, c) (4.18)
δ1 δ1 δ1
16 WAEL YOUSSEF

−M iλp = k1 (Φx + Ψ + lΩ)(c) − k(φx + ψ + lω)(c), (4.19)

−M iλq = b1 Ψx (c) − bψx (c), (4.20)
(4.21)
Using (4.4) in (2.6), we obtain
∫
( ) c ( )
Re⟨AW, W ⟩H = Re iλ∥W ∥2H = − |θ1x |2 + |θ2x |2 dx = 0. (4.22)
0
Therefore, θ1x = θ2x = 0 on (0, c). So, since θ1 (0) = θ2 (0) = 0 by (1.2) we get
θ1 = θ2 = 0 on (0, c). Next, thanks to (4.12), Vx = 0 on (0, c) and so V = 0 on
(0, c) because V (0) = 0. Thus, (4.9) implies that Ψ = 0 on (0, c). Thereby, (4.17)
gives
Φx + lΩ = 0, x ∈ (0, c). (4.23)
In addition, using (4.8) and (4.10) in (4.11), we obtain
Ωx − lΦ = 0, x ∈ (0, c). (4.24)
Next, repeating the same argument in the proof of lemma 2.1, we get Φ(x) =
Ψ(x) = Ω(x) = 0, for all x ∈ (0, c) Next, the conditions (1.3)3 , (1.3)4 , and (1.3)5 ,
give
φ(x) = φx (x) = ψ(x) = ψx (x) = ω(x) = ωx (x) = 0, x = c, L. (4.25)
Denoting by Y = (φ, φx , ω, ψ, ψx , ωx ), then from equations (4.13), (4.14), (4.15),
and the boundary conditions (4.25), we obtain the following system
dY
= MY and Y (c) = Y (L) = 0, (4.26)
dx
where M is is a real square matrix of order three. Therefore the ordinary differential
system (4.26) has the unique solution Y = 0. Consequently, W = 0 which is a
contradiction since W ̸= 0. Thereby, (4.2) is proved.
It remains to complete the proof of the polynomial stability by applying the theorem
1.2 by establishing that
1
lim ∥(iλI − A)−1 ∥ < +∞. (4.27)
λ−→∞ λ80
Let us argue by contradiction. Then, we suppose that the condition (4.27) is false.
Then, there is a real sequence (λn ) and a sequence
W n = (φn , ψ n , ω n , Φn , Ψn , Ωn , θ1n , θ2n , un , v n , z n , U n , V n , Z n , pn , q n ) ∈ D(A)
such that
λn −→ +∞, (4.28)
∥ W n ∥H = 1, (4.29)
and
n ∥(iλn I − A)W ∥H = 0.
lim λ80 n
(4.30)
n−→∞
So [ ] n
n (iλn I − A) W = (f1 , f2 , ..., f16 ) −→ 0 in
λ80 H.
n n n
(4.31)
Therefore, in H 1 (c, L) we have the following convergence
[ ]
n iλn φ − u
λ80 = f1n −→ 0,
n n
(4.32)
[ ]
n iλn ψ − v
λ80 = f2n −→ 0,
n n
(4.33)
[ ]
n iλn ω − z
λ80 = f3n −→ 0.
n n
(4.34)
 17

In H 1 (0, c) we have the following convergenve

[ ]
λn80 iλn Φn − U n = f4n −→ 0, (4.35)
[ ]
n iλn Ψ − V
λ80 = f5n −→ 0,
n n
(4.36)
[ ]
n iλn Ω − Z
λ80 = f6n −→ 0.
n n
(4.37)
In L2 (0, c) we have the following convergence
[ ]
n iλn θ1 − θ1xx + α(Zx − lU ) = f7 −→ 0,
λ80 n n n n n
(4.38)
and [ ]
n iλn θ2 − θ2xx + αVx = f8 −→ 0.
λ80 n n n n
(4.39)
2
In L (c, L) we have the following convergence
[ k n ]
n iλn u −
λ80 (φxx + ψxn + lwxn ) − k0 l(wxn − lφn ) = f9n −→ 0,
n
(4.40)
ρ1
[ b n k ]
n iλn v −
λ80 −→ 0,
n
ψxx + (φnx + ψ n + lwn ) = f10 n
(4.41)
ρ2 ρ2
[ k0 n lk ]
n iλn z −
λ80 (ωxx − lφnx ) + (φnx + ψ n + lω n ) = f11 −→ 0,
n n
(4.42)
ρ1 ρ1
In L2 (0, c) we have the following convergence
[ k1 n k01 α ]
n iλn U −
λ80 (Φxx + Ψnx + lΩnx ) − l(Ωnx − lΦn ) + θ1n = f12 −→ 0, (4.43)
n n
δ1 δ1 δ1
[ b1 n k1 α n]
n iλn V −
λ80 −→ 0,
n
Ψxx + (Φnx + Ψn + lΩn ) + θ2x n
= f13 (4.44)
δ2 δ2 δ2
[ k01 n lk1 n α n]
n iλn Z −
λ80 (Ωxx − lΦnx ) + −→ 0, (4.45)
n
(Φx + Ψn + lΩn ) + θ1x n
= f14
δ1 δ1 δ1
Also, in C we have the following convergence
 [

 λ80 1 ( )]
n iλn p +
n
k1 (Φnx + Ψn + lΩn )(c) − k(φnx + ψ n + lω n )(c)
[ M
1 ( )] (4.46)

 = λ80 n
k1 (Φnx + Ψn )(c) − k(φnx + ψ n )(c) f15 n
−→ 0
n iλn p +
M
and [ 1 ( )]
λ80
n iλn q +
n
b1 Ψnx (c) − bψxn (c) = f16 n
−→ 0. (4.47)
M
Note that
∫ L[
∥ W ∥H =
n 2
ρ1 |un |2 + ρ2 |v n |2 + ρ1 |z n |2 + b|ψxn |2 + k|φnx + ψ + lω n |2
c
]
+k0 |ωxn − lφn |2 dx
∫ c[
+ δ1 |U n |2 + δ2 |V n |2 + δ1 |Z n |2 + b1 |Ψnx |2 + k1 |Φnx + Ψn + lΩn |2
0
]
+k01 |Ωnx − lΦn |2 + |θ1n |2 + |θ2n |2 dx
+|pn |2 + |q n |2 .
The main goal is to prove that ∥ W n ∥2H −→ 0 to get a contradiction with (4.29).
Then, several steps are required to achieve the proof.
18 WAEL YOUSSEF

Step 1. Thanks to (4.31), we obtain

∫ c ( )
Re⟨(λn (iλn − A))W , W ⟩H =
80 n n
|λ40
n θ1x | + |λn θ2x |
n 2 40 n 2
dx −→ 0. (4.48)
0
Hence
n θ1x , λn θ2x −→ 0 in
λ40 n 40 n
L2 (0, c). (4.49)
Thanks to the Poincaré inequality, we have also
n θ1 , λn θ2 −→ 0 in
λ40 n 40 n
L2 (0, c). (4.50)

Step 2. In this step, we shall prove that

un , v n , z n −→ 0 in L2 (c, L), (4.51)
ψxn −→ 0 in 2
L (c, L), (4.52)
φnx n
+ ψ + lω n
−→ 0 in 2
L (c, L), (4.53)
ωxn − lφ n
−→ 0 in 2
L (c, L). (4.54)
Firstly, dividing (4.32), (4.33), (4.34), (4.35), (4.36), and (4.37) by and using λ81
n
the fact that un , v n , z n , U n , V n , and Z n are bounded from (4.29) lead to
φn , ψ n , ω n −→ 0 in L2 (c, L). (4.55)
Φn , Ψn , Ωn −→ 0 in L2 (0, c). (4.56)
∫ l2 ∫ l2
Note that the term f dx will be denoted by f.
l1 l1
n
Hence, eliminating u in (4.40) by (4.32) then taking the inner product of resulting
ρ1
equation with 80 (L − x)φnx in L2 (c, L) and integrating by parts lead to
λn
 ∫ L ∫

 ρ1 L−c k L n2

 − |λ n φ |
n 2
+ ρ 1 |λn φ n
(c)| 2
− |φx |

 2 c 2 ∫ L 2 c ∫ L
 {
(L − c)k n
+ |φx (c)|2 + Re − k (L − x)ψxn φnx − lk (L − x)ωxn φnx (4.57)

 2

 ∫ L c∫
} c


L
 −lk0 (L − x)ωxn φnx + l2 k0 (L − x)φn φnx −→ 0.
c c
Moreover, the sequence ∥φnx ∥L2 (c,L) is bouded. Indeed,
∥φnx ∥L2 (c,L) ≤ ∥φnx + ψ n + lω n ∥L2 (c,L) + ∥ψ n ∥L2 (c,L) + l∥ω n ∥L2 (c,L) ,
∥φnx + ψ n + lω n ∥L2 (c,L) is bounded from (4.29) and ∥ψ n ∥L2 (c,L) , ∥ω n ∥L2 (c,L) are
bounded from (4.55). Therefore, the last term in (4.57) tends to zero thanks to
(4.55). Thus, (4.57) gives
 ∫ ∫

 ρ1 L L−c k L n2

 − |λ n φ |
n 2
+ ρ 1 |λn φn
(L)| 2
− |φx |

 2 c 2 ∫ 2 c ∫ L
 {
(L − c)k n L
+ |φx (c)|2 + Re − k (L − x)ψxn φnx − lk (L − x)ωxn φnx

 2

 ∫ L } c c


 −lk0 (L − x)ωx φx −→ 0.
n n
c
(4.58)
 19

Similarly, eliminating v n in (4.41) by (4.33) and z n in (4.42) by (4.34), then taking

ρ2 ρ1
the inner product of resulting equations with 80 (L − x)ψxn and 80 (L − x)ωxn in
λn λn
L2 (c, L), respectively, taking in consideration that ω n ∈ H01 (c, L), and integrating
by parts lead to
 ∫ ∫

 ρ2 L L−c (L − c)b L n 2

 − |λ n ψ |
n 2
+ ρ 2 |λn ψ n
(c)| 2
− |ψx |

 2 c ∫ 2 ∫2
 { L L
c
b
+ |ψxn (c)|2 + Re k (L − x)ψxn φnx + k (L − x)ψxn φn (4.59)

 2

 ∫ }c c


L
 +k (L − x)ψxn ω n −→ 0
c
and  ∫ ∫

 ρ1 L k0 L n 2 (L − c)k0 n

 − |λn ω | −
n 2
|ωx | + |ωx (c)|2

 2 c ∫ 2 c ∫ 2
 { L L
+Re lk0 (L − x)ωxn φnx + lk (L − x)ωxn φnx (4.60)



 ∫ L c ∫ L c
}


 +lk (L − x)ωxn ψ n + lk (L − x)ωxn ω n −→ 0.
c c
On the other hand, the sequence ∥ωxn ∥L2 (c,L) is bouded. Indeed,
∥ωxn ∥L2 (c,L) ≤ ∥ωxn − lφn ∥L2 (c,L) + l∥φn ∥L2 (c,L) ,
∥ωxn − lφn ∥L2 (c,L) is bounded from (4.29) and ∥φn ∥L2 (c,L) is bounded from (4.55).
Moreover, ∥ψxn ∥L2 (c,L) is bounded by (4.29). Therefore, the last two terms in (4.59)
and those in (4.60) tend to zero thanks to (4.55). Hence, (4.59) and (4.60) give
 ∫ ∫

 ρ2 L L−c b L n2
 − |λn ψ n |2 + ρ2 |λn ψ n (c)|2 − |ψx |
2 c 2 ∫ L 2 c
 (L − c)b n { } (4.61)

 + |ψx (c)|2 + Re k (L − x)ψxn φxn −→ 0.
2 c
and  ∫ ∫

 ρ1 L L−c k0 L n 2

 − |λ n ω |
n 2
+ ρ 1 |λn ω n
(c)| 2
− |ωx |

 2 c 2 ∫ 2 c
 {
(L − c)k0 n L
+ |ωx (c)|2 + Re lk0 (L − x)ωxn φnx (4.62)

 2

 ∫ L } c


 +lk (L − x)ω φ −→ 0.
n n
x x
c
Adding (4.58), (4.61), and (4.62) leads to
 ∫ ∫ ∫

 ρ1 L ρ2 L ρ1 L

 − |λn φ | −
n 2
|λn ψ | −
n 2
|λn ω n |2

 2 2 2

 ∫ L
c ∫ c ∫ c

 −k b L n 2 k0 L n 2
|φnx |2 − |ψx | − |ωx |
2 c 2 c 2 c (4.63)

 L−c L−c

 +ρ1 |λn φn (c)|2 + ρ2 |λn ψ n (c)|2




2 2

 + k |φn (c)|2 + b |ψ n (c)|2 + k0 |ω n (c)|2 −→ 0.
2 x 2 x 2 x
In the following steps, we shall denote by ∥ . ∥c := ∥ . ∥L2 (0,c) .
20 WAEL YOUSSEF

Step 3. In this step, the main goal is the proving that all the terms defined
at x = c in (4.63) tend to zero. Only, the transmission conditions will give back
information from the thermoelastic part of the beam to the elastic part. Let us
start with the term λn ψ n (c). So by the Gagliardo-Nirenberg inequality, we have
b21
|λn ψ n (c)|2 = |λn Ψn (c)|2 ≲ ∥Ψn ∥c ∥λn Ψnx ∥c , (4.64)
b2
where the notation ≲ means: For G1 , G2 ∈ R+ , G1 ≲ G2 if there exists a constant
K > 0 such that G1 ≤ K G2 .
Eliminating Vn in (4.39) by (4.36) then taking the inner product of resulting equa-
1
tion with 61 Ψnx in L2 (0, c) and integrating by parts lead to
λn
∫ c ∫ c ∫ c
[ ]
n − λ19 θ n Ψn c + iα |λ10
λ20 n Ψx | −→ 0.
n n 19 n n 2
i θ
n 2 x Ψ + λ θ Ψ
n 2x xx n 2x x 0 (4.65)
0 0 0

The first term tends to zero since λ20n θ2 −→ 0 by (4.49) and Ψx is bounded in
n n
2
L (0, c) due to (4.29). The second term can be written as
∫ c n
n Ψxx
λ20
n θ2x .
0 λn
Ψnxx
So this term tends to zero since is bounded from the dividing of (4.44) by
λn
λ79 40 n 2
n and λn θ2x tends to zero in L (0, c) by (4.49). Next, for x = 0, c, by the
Gagliardo-Nirenberg inequality we have
  12 n 12
1 n 1
 19 n  40 n 2 θ2xx n 2 Ψxx
λn 2x
θ (x)Ψ n (x) ≲ ∥λ θ ∥
n 2x c ∥Ψ ∥
x c .
x
λn c λn c
Vxn
Now, dividing (4.36) by λ81
n implies that is bounded in L2 (0, c). Then, dividing
λn
n
θ2xx
(4.39) by λ81
n yields to is bounded in L2 (0, c). Consequently, the first three
λn
terms in (4.65) tend to zero and so
∥λ10
n Ψx ∥c −→ 0.
n
(4.66)
Therefore, Ψn is bounded in L2 (0, c) from the Poincaré inequality and since Ψnx is
bounded in L2 (0, c) from (4.29). Thus, using (4.66) in (4.64) leads to
λn ψ n (c) −→ 0. (4.67)
Repeating the same arguments to (4.38), (4.35), and (4.37) yields to
∥λ2n (Ωnx − lΦn )∥c −→ 0. (4.68)
Now, let us prove that
∥λ2n (Φx + Ψn + lΩn )∥c −→ 0. (4.69)
To reach to this goal, eliminating Vn in (4.44) by (4.36) then taking the inner
1
product of resulting equation with 76 (Φx + Ψn + lΩn ) in L2 (0, c) and integrating
λn
 21

by parts lead to
 { ∫ c ∫ c

 b

 Re − 6 n n n
λn Ψ (Φx + Ψ + lΩ ) + λ4 Ψn (Φx + Ψn + lΩn )x

 ρ2 0∫ n x
 b1 [ 4 n
0 ]c k c
1
− λn Ψx (Φx + Ψn + lΩn ) + |λ2n (Φnx + Ψn + Ωn )|2 (4.70)

 δ2 ∫ 0 } δ2 0

 α c 4 n


 + λ θ (Φx + Ψn + lΩn ) −→ 0.
δ2 0 n 2x
By the poincaré inequality we have
∥λ6n Ψn ∥c ≤ ∥λ6n Ψnx ∥c .
Hence, thanks to (4.66)
λ6n Ψn −→ 0 in L2 (0, c). (4.71)
So (4.71) and the boundedness of Φx + Ψn + lΩn in L2 (0, c) due to (4.29) give
{∫ c }
Re λn Ψ (Φx + Ψ + lΩ ) −→ 0.
4 n n n (4.72)
0
Next, we have
{∫ ∫ }
c c
(Φx + Ψn + lΩn )x
Re 4 n n n
λn Ψx (Φx + Ψ + lΩ )x = λ4n Ψnx −→ 0 (4.73)
0 0 λn

(Φx + Ψn + lΩn )x
since λ3n Ψnx −→ 0 in L2 (0, c) thanks to (4.66) and is bounded in
λn
1
L2 (0, c) by dividing (4.43) by . The last term in (4.70)
λ81
n
{∫ c }
Re λn θ2x (Φx + Ψ + lΩ ) −→ 0.
4 n n n (4.74)
0

Thanks to (4.49) and the fact that Φx + Ψn + lΩn is bounded due to (4.29). It
remains to establish that the boundary terms in (4.70) tends to zero. Indeed, for
x = 0, c, by the Gagliardo-Nirenberg inequality, we have


 |λn Ψx (x)(Φx + Ψn1 + lΩn )(x)| ≲
4 n
n 2 21
1 1 n n (4.75)
 10 n 2 Ψxx n 2 (Φx + Ψ + lΩ )x
 ∥λ n Ψ ∥
x c ∥Φ x + Ψ n
+ lΩ ∥c .
λn c λn c

Ψnxx (Φx + Ψn + lΩn )x

Thus, (4.66) and the boundedness of , Φx + Ψn + lΩn , and
λn λn
yield to
[ ]c
λ2n Ψnx (Φx + Ψn + lΩn ) −→ 0. (4.76)
0
Consequently, using (4.71), (4.72), (4.73), (4.74), and (4.76) in (4.70) implies (4.69).
Next, by the Gagliardo-Nirenberg, we have
1 1
|λn φn (c)| = |λn Φn (c)| ≲ ∥λn Φnx ∥c2 ∥λn Φn ∥c2 (4.77)
nπ
In addition, since l ̸= , for n ∈ N, there exists a positive constant C such that
c
( )
∥Φx ∥2c + ∥Ψx ∥2c + ∥Ωx ∥2c ≤ C ∥Φx + Ψn + lΩn ∥2c + ∥Ψx ∥2c + ∥Ωx − lΦn ∥2c . (4.78)
22 WAEL YOUSSEF

Therefore, due to (4.78), (4.66), (4.68), and (4.69), the estimation (4.77) gives
∥λn Φnx ∥c −→ 0. (4.79)
Thereby, from (4.77), we get
|λn φn (c)| −→ 0. (4.80)
Step 4. In this step, the aim is the proving that the last three terms in (4.63) tend
to zero. Firstly, dividing (4.35) by λ79 , we obtain
iλ2n Φn − λn U n −→ 0 in L2 (0, c) (4.81)
and
iλ2n Φnx − λn Uxn −→ 0 in L2 (0, c). (4.82)
Therefore, due to (4.66), (4.67), and (4.69), we conclude from (4.78), (4.81), and
(4.82) that
λn U n , λn Uxn −→ 0 in L2 (0, c). (4.83)
Using the Gagliardo-Nirenberg inequality and (4.83), we obtain
1 1
λn pn = λn U n (c) ≲ ∥λn U n ∥c2 ∥λn Uxn ∥c2 −→ 0 in C. (4.84)
79
Again, dividing (4.36) and (4.37), respectively, by λ , we get
iλ2n Ψn − λn V n −→ 0 in L2 (0, c) (4.85)
and
iλ2n Ψnx − λn Vxn −→ 0 in L2 (0, c), (4.86)
So, using (4.66), (4.67), and (4.69), the expressions (4.85) and (4.86) imply
λn V n , λn Vxn −→ 0 in L2 (0, c). (4.87)
Next, applying the Gagliardo-Nirenberg inequality yields
1 1
λn q n = λn V n (c) ≲ ∥λn V n ∥c2 ∥λn Vxn ∥c2 in C. (4.88)
Consequently, using (4.87) in (4.88), we obtain
λn q n −→ 0 in C. (4.89)
Now, let us overcome the most difficult point of the proof witch is
|φnx (c)| −→ 0 in L2 (0, c). (4.90)
First, using (4.84) in (4.46), we get
k1 (Φnx + Ψn + lΩn )(c) − k(φnx + ψ n + lω n )(c) = o(1). (4.91)
On the other hand, by the Gagliardo-Nirenberg inequality, we have
1
1 (Φn + Ψn + Ωn ) 2
2 x
|(Φx + Ψ + Ω )(c)| ≲ ∥λ (Φx + Ψ + Ω )∥c
n n n n n n n x
. (4.92)
λn c
(Φnx + Ψn + Ωn )x
Since is bounded in L2 (0, c), thanks to (4.69), the expression
λn
(4.92) gives
(Φnx + Ψn + Ωn )(c) = (Φnx + Ψn )(c) −→ 0 in C. (4.93)
Thereby, (4.91) leads to
(φnx + ψ n + ω n )(c) = (φnx + ψ n )(c) −→ 0 in C. (4.94)
Next, we have
|φnx (c)| ≤ |(φnx + ψ n )(c)| + |ψ n (c)|. (4.95)
 23

So, using (4.94) and (4.67) in (4.95) leads to

|φnx (c)| −→ 0 in C. (4.96)
Repeating the same process to establish that
|ψxn (c)| −→ 0 in C. (4.97)
Indeed, using (4.89) in (4.47), we get
b1 Ψnx (c) − bψxn (c) = o(1). (4.98)
Next, by the Gagliardo-Nirenberg inequality, we have
n 12
1
n n 2 Ψxx
|Ψx (c)| ≲ ∥λ Ψx ∥c n .
n
(4.99)
λ c
Ψnxx
Since is bounded in L2 (0, c), thanks to (4.66), the expression (4.99) gives
λn
|Ψnx (c)| −→ 0 in C. (4.100)
Now, from (4.98), we conclude (4.97).
Again, the Gagliardo-Nirenberg inequality gives
n 1
1 (Ωx − lΦ)x 2
|(Ωnx − lΦ)(c)| ≲ ∥λ n
(Ωnx
− lΦ))∥c
2 . (4.101)
λn
c
(Ωnx − lΦ)x
Due to (4.45), the sequence is bounded in L2 (0, c). Therefore, using
λn
(4.68) in (4.101) implies
(Ωnx − lΦ)(c) −→ 0 in C. (4.102)
Then, thanks to the transmission condition (1.3)5 , we have
(ωxn − lφ)(c) −→ 0 in C. (4.103)
Moreover, we have
|φn (c)| −→ 0 in C. (4.104)
Indeed, φnx is bounded in L2 (c, L) and
1 1
|φn (c)| ≲ ∥φn ∥L2 2 (c,L) ∥φnx ∥L2 2 (c,L) .
Therefore, (4.104) is obtained from (4.55). Now, using (4.103) and (4.104) in the
following inequality
|ωxn (c)| ≤ |(ωxn − lφ)(c)| + l|φn (c)|, (4.105)
we get
ωxn (c) −→ 0 in C. (4.106)
Now, using (4.67), (4.80), (4.96), (4.97), and (4.106), the expression (4.63) implies
that  ∫ ∫ ∫

 ρ1 L ρ2 L ρ1 L
 − |λn φ | −
n 2
|λn ψ | −
n 2
|λn ω n |2
2∫ c ∫ 2 c ∫ 2 c (4.107)

 k L n 2 b L n 2 k0 L n 2
 − |φx | − |ψx | − |ωx | −→ 0.
2 c 2 c 2 c
Consequently,
λn φn , λn ψ n , λn ω n −→ 0 in L2 (c, L) (4.108)
24 WAEL YOUSSEF

and
φnx , ψxn , ωxn −→ 0 in L2 (c, L). (4.109)
Thus, (4.108), (4.32), (4.33), and (4.34) imply that
un , v n , z n −→ 0 in L2 (c, L). (4.110)
Also, (4.109) with (4.55) leads to
φnx + ψ n + lω n , ψxn , ωxn − lφn −→ 0 in L2 (c, L). (4.111)
Thanks (4.66), (4.68), and (4.69) we have
Φnx + Ψn + lΩn , Ψnx , Ωnx − lΦn −→ 0 in L2 (0, c). (4.112)
Also, the poincaré inequality, (4.66), (4.68), (4.69), and (4.78) give
λn Φn , λn Ψn , λn Ωn −→ 0 in L2 (0, c). (4.113)
Therefore, dividing (4.35), (4.36), and (4.37) by λ80
n , we obtain
U , V , Z −→ 0 in
n n n 2
L (0, c). (4.114)
In addition, thanks to (4.50), (4.84), and (4.89), respectively, we have
θ1n , θ2n −→ 0 in L2 (0, c), (4.115)
pn −→ 0 in C, (4.116)
and
q n −→ 0 in C. (4.117)
Finally, (4.110), (4.111), (4.112), (4.114), (4.115), (4.116), and (4.117) imply
∥ W n ∥H −→ 0. (4.118)
This is a contradiction with (4.29) and the proof is completed. □
Remark 4.1. The polynomial decay of the rate is really a slow one. The sharpness
of the obtained result under the considered boundary conditions for this system is
an open problem. Indeed, if we argue by contradiction and suppose that the rate
t− 80 can be improved i.e. we assume that the rate is t− 80−ϵ , for some 0 < ϵ < 80,
1 1

it is difficult to get a contradiction since it is not easy to show the existence of

a sequence of values (λn ) ⊆ R with limn−→∞ |λn | = ∞ and a sequence of data
(Wn ) ⊆ D(A) for (Fn ) ⊆ H such that (iλn I − A)Wn = Fn is bounded in H and
1
limn−→∞ ∥Wn ∥H = ∞. We refer the readers to [10] for more details.
|λn |80−ϵ
Acknowledgment. The author would like to express his gratitude to the anony-
mous referees for very careful reading and punctual suggestions.

References
[1] F. Alabau-Boussouira, J.E. Muñoz Rivera, D.S. Almeida Júnior, Stability to weak dissipative
Bresse system, J. Math. Anal. Appl., 374 (2011), pp. 481-498.
[2] T. El Arwadi; M.I.M. Copetti; W. Youssef, On the theoretical and numerical stability of the
thermoviscoelastic Bresse system, Z. Angew. Math. Mech., 99 (2019), pp. 1-20.
[3] F. Abdallah, M. Ghader, A. Wehbe, Stability results of a distributed problem involving Bresse
system with history and/or Cattaneo law under fully Dirichlet or mixed boundary conditions,
Math. Methods Appl. Sci., 41 (2018), pp.1876-1907 .
[4] A. Borichev and Y. Tomilov, Optimal polynomial decay of functions and operator semi-
groups, Math. Ann., 347 (2009), pp. 455478.
[5] J. A.C. Bresse, Cours de Mécanique Appliquée, Mallet Bachelier, 1859.
 25

[6] F. Dell’Oro, Asymptotic stability of thermoelastic systems of Bresse type, J. Differ. Eq., 258
(2015), pp. 39023927
[7] T. El Arwadi; W. Youssef, On the Stabilization of the Bresse Beam with Kelvin-Voigt Damp-
ing, Appl. Math. and Opt., 83 (2021), pp.1831-1857.
[8] K. Engel, R. Nagel, One-Parameter Semigroups for Linear Evolution Equations, Graduate
Texts in Mathematics. New York, Springer Verlag, 2000.
[9] L. H. Fatori and J.E. Muñoz Rivera, Rates of decay to weak thermoelastic Bresse system,
IMA J. Appl. Math., 75 (2010), pp. 881-904.
[10] L.H. Fatori, R. N. Monteiro, The optimal decay rate for a weak dissipative Bresse system,
Appl. Math. Lett., 25 (2012), pp. 600604.
[11] F. Gallego and J. E. Muñoz Rivera, Decay rates for solutions to thermoelastic Bresse systems
of types I and III, Elect. J. Diff. Eq., 73 (2017), pp. 1-26.
[12] F. Huang, Characteristic conditions for exponential stability of linear dynamical systems in
Hilbert spaces, Ann. Differential Equations 1 (1985) 4356.
[13] A. Haraux, Semi-groupes lineéaires et équations d’évolution linéaires peériodiques, Publica-
tion du Laboratoire d’Analyse Numeérique No. 78011, Universiteé Pierre et Marie Curie,
Paris, 1978.
[14] Z-J. Han and G-Q Xu, Spectral analysis and stability of thermoelastic Bresse system with
second sound and boundary viscoelastic damping, Math. Methods Appl. Sci., 13 (2013), pp.
13951406.
[15] Z-J. Han and G-Q Xu, Spectrum and stability analysis for a transmission problem in ther-
moelasticity with a concentrated mass, Z. Angew. Math. Phys., 66 (2015), pp. 1717-1736.
[16] T. Kato, Pertubation Theory of Linear Operators, New York, Springer-Verlag, 1980.
[17] A.A. Keddi, T.A. Apalara, and S.A. Messaoudi, Exponential and polynomial decay in a
thermoelastic-Bresse system with second sound, Appl. Math. Optim., 77 (2018), pp. 315341.
[18] J. E. Lagnese, G. Leugering, and E. J. P. G. Schmidt, Modelling of dynamic networks of thin
thermoelastic beams, Math. Meth. Appl. Sci., 16 (1993), pp. 327-358.
[19] Z. Liu and S. Zheng, Semigroups Associated With Dissipative Systems, 398 Research Notes
in Mathematics, Chapman & Hall/CRC, 1999.
[20] Z. Liu and B. Rao, Energy Decay of the thermoelastic Bresse System, Z.angew. Math. Phys.
60 (2009), pp. 54 - 69.
[21] J. E. Muñoz Rivera and R. Racke, Transmission Problems in (Thermo)Viscoelasticity with
Kelvin-Voigt Damping: Nonexponential, Strong, and Polynomial Stability, SIAM J. Math.
Analysis 49 (2017), pp. 3741-3765.
[22] J. Prüss, On the spectrum of C0-semigroups, Trans. Amer. Math. Soc. 284 (1984) 847857.
[23] M.L. Santos, Bresse System in Thermoelasticity of Type III Acting on Shear Force, J.
Elast.,125 (2016), pp. 185216.
[24] B. Said-Houari and A. Soufyane, The Bresse system in thermoelasticity, Math. Meth. App.
Sci., 38 (2015), DOI: 10.1002/mma.3305.
[25] W. Youssef, Stabilization for the transmission problem of the Timoshenko system in ther-
moelasticity with two concentrated masses, Math. Meth. Appl. Sci., 43 (2020), pp. 3965-3981.
[26] A. Wehbe and W. Youssef, Exponential and Polynomial Stability of an Elastic Bresse System
with two Locally Distributed Feedbacks, Jornal of Math. Phy., 51 (2010), pp. 1-17.

Wael Youssef, Department of Mathematics, Faculty of Sciences 1, Lebanese Uni-

versity, Hadath, Beyrouth, Lebanon
E-mail address: [email protected]