This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
 Author's personal copy

J. Differential Equations 253 (2012) 2715–2733

Contents lists available at SciVerse ScienceDirect

Journal of Differential Equations

www.elsevier.com/locate/jde

The stability number of the Timoshenko system with second

sound
M.L. Santos a , D.S. Almeida Júnior a,∗ , J.E. Muñoz Rivera b
a
Department of Mathematics, Federal University of Pará, Augusto Corrêa Street, Number 01, CEP 66075-110, Belém, Pará, Brazil
b
National Laboratory for Scientific Computation, Getúlio Vargas Street, Number 333, CEP 25651-075, Petrópolis, Rio de Janeiro, Brazil

a r t i c l e i n f o a b s t r a c t

Article history: In this work, we consider the Timoshenko beam model with
Received 18 April 2012 second sound. We introduce a new number χ0 that characterizes
Revised 26 July 2012 the exponential decay. We prove that the corresponding semigroup
Available online 9 August 2012
associated to the system is exponentially stable if and only if
MSC:
χ0 = 0. Otherwise there is a lack of exponential stability. In this
74F05 case we prove that the semigroup decays as t −1/2 . Moreover we
74F20 show that the rate is optimal.
© 2012 Elsevier Inc. All rights reserved.
Keywords:
Exponential stability
Polynomial stability
Optimality

1. Introduction

Here we will consider a fully hyperbolic thermoelastic beam model. That is to say, instead of the
classical thermoelastic system defined by the Euler–Bernoulli equation coupled with the (parabolic)
heat equations, we will use the Timoshenko system coupled to a hyperbolic heat model defined by
the Cattaneo law. From our point of view this model explains better the thermoelastic phenomenon.
Next we will explain the reasons of these changes.
The well-known Euler–Bernoulli model for the transverse vibration of a beam is not suitable for
all applications. The model permits transmission of energy at speeds approaching infinity. Rayleigh
solved this problem in principle by introducing rotatory inertial but the corrections were insufficient.
In 1921, Timoshenko made a further improvement by which shear deformation is taken into account.
The mathematical model consists of two partial differential equations presented as follows (see [22])

* Corresponding author.
E-mail address: [email protected] (D.S. Almeida Júnior).

0022-0396/$ – see front matter © 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jde.2012.07.012
 Author's personal copy

2716 M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733

ρ A ϕtt = S x , (1.1)

ρ I ψtt = M x − S . (1.2)

By t we denote the time variable, x is the space variable, ϕ is the transverse displacement, ψ is
the rotation of the transversal section of the neutral axes, ρ is the mass density, M is the curvature
moment, S is the stress, A is the area of the transversal section and I is the inertial moment of the
transversal section. The corresponding constitutive laws are given by

M = E I ψx − δθ, (1.3)

S = k AG (ϕx + ψ). (1.4)

In these equations δ denotes the density, E and G are the elastic constants and k is the shear
coefficient. Finally by θ we are denoting the difference of temperature.
On the other hand, it is well known that the model for the temperature using Fourier’s law results
in a physical discrepancy of infinite heat propagation speed of signals. In others words, any thermal
disturbance at a single point has an instantaneous effect everywhere in the continuous medium. For
some applications like laser cleaning computer chips with very short laser pulses, see the references
in [17]. It is worth while thinking of another model removing this paradox, but still keeping the es-
sentials of a heat conduction process. One such model is given by the simplest Cattaneo law replacing
Fourier’s law q = −k ∇θ (see [24]), that is

τ q t + q + k  θ x = 0,

now regarding the heat flux vector as another function to be determined through the differential
equation. The positive parameter τ is the relaxation time describing the time lag in the response of
the heat flux to a gradient in the temperature. To see the hyperbolic nature of this model, let us
combine the balance of the energy θt + q x = 0 with the Cattaneo law to obtain

τ θtt − k θxx + θt = 0,

which is the damped wave equation, once more we obtain the well-known exponential stability. That
is, both models, Fourier and Cattaneo, exhibit the same qualitative behavior, with both leading to
exponentially stable systems for pure heat conduction.
Therefore, from the constitutive laws (1.1)–(1.4) and the balance of the energy given by the Catta-
neo law, we get the system

1 ϕtt − κ (ϕx + ψ)x = 0 in ]0, l[ × ]0, ∞[, (1.5)

2 ψtt − bψxx + κ (ϕx + ψ) + δθx = 0 in ]0, l[ × ]0, ∞[, (1.6)

3 θt + qx + δψxt = 0 in ]0, l[ × ]0, ∞[, (1.7)

τ qt + β q + θx = 0 in ]0, l[ × ]0, ∞[, (1.8)

where we are assuming 1 = ρ A, κ = kG A, 2 = ρ I and b = E I . The positive constants 3 , τ , δ and

β relate to hypotheses in thermoelasticity.
Here we consider the following boundary conditions

ϕ (0, t ) = ϕ (l, t ) = ψx (0, t ) = ψx (l, t ) = θ(0, t ) = θ(l, t ) = 0, ∀t  0, (1.9)

 Author's personal copy

M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733 2717

and the following initial conditions

ϕ (x, 0) = ϕ0 (x), ϕt (x, 0) = ϕ1 (x), ψ(x, 0) = ψ0 (x), ψt (x, 0) = ψ1 (x),

θ(x, 0) = θ0 (x), q(x, 0) = q0 (x), ∀x ∈ ]0, l[. (1.10)

In all the literature concerning Timoshenko system, when only one dissipation is taking into ac-
count over the two-by-two Timoshenko system, the conclusion is that the exponential stability holds
if and only if the velocities of propagations are equal. That is to say, denoting by χ the difference of
velocities of propagations

κ b
χ= − ,
ρ1 ρ2
this conclusion can be written as: The semigroup associated to Timoshenko system is exponentially stable
if and only if χ = 0. Therefore χ is an important number that characterizes the asymptotic behavior
of the solutions to Timoshenko system. This was proved to memory viscoelastic constitutive laws in
[1,3,4,6,8,12]; to thermoelastic constitutive laws with Fourier law and also to thermoelastic dissipation
of type III in [10,12,23]; to Timoshenko system with boundary dissipation in [15,16] and also with
locally distributed dissipation [18–20]. Of course, since Timoshenko system is a two-by-two system
of hyperbolic equations, if there exist two dissipative mechanisms, we always get the exponential
stability, no matter if the velocities of propagations are equal or not, see [13,14,21].
Here we consider the Timoshenko beam model with thermal dissipation given by Cattaneo law. As
proved in [5] this dissipation is different to all studied in the above cited references. Moreover, the
stability number χ does not say anything about the exponential decay of the corresponding semi-
group. We can ask, why this fails?
It fails because all the thermal dissipations considered before [5], were of parabolic type. That is
to say, Timoshenko beam model was coupled to a heat equation of type

3 θt − k θxx + δψxt = 0,

or with thermoelasticity of type III

3 θtt − κ1 θxx − κ1 θxxt + γ ψxtt = 0.

The latter model for the temperature was considered in [10]. Of course it is not 100% parabolic
in the classical sense, but the semigroup associated to this model (γ = 0) is analytical as proved
in [9]. Cattaneo’s law turns the Timoshenko beam model coupled with the heat equation, into a fully
hyperbolic system, this means that the temperature also has a finite speed of propagation which
plays an important role in the exponential stability. Here we prove that Cattaneo’s law modifies the
stability number χ . Since the new hyperbolic system is now 3 × 3 more conditions appear, therefore
we introduce a new stability number

  
1 b 1 τ 1 δ 2
χ0 := τ − 2 − − .
3 κ κ 3 κ

In this paper we prove that the exponential decay holds if and only if χ0 = 0. Note that when
τ = 0, Cattaneo’s law turns into the Fourier law and the conditions over the new number χ0 are
equivalent to the old stability number. That is, we get the same result as such proved in [23] to
Timoshenko system with thermal dissipation given by the Fourier law. Moreover in case that χ0 = 0,
we prove that the system decays polynomially to zero as t −1/2 . Finally, we prove that this rate of
decay is optimal. That is, this rate cannot be improved for any initial data in D (A).
 Author's personal copy

2718 M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733

The method we use to show the lack of exponential stability is based on Gearhart–Herbst–Prüss–
Huang theorem to dissipative systems. See also [7,11].

Theorem 1.1. Let S (t ) = e At be a C 0 -semigroup of contractions on Hilbert space. Then S (t ) is exponentially

stable if and only if i R ⊂ (A) and

 
lim|λ|→∞ (i λ I − A)−1 L(H) < ∞. (1.11)

On the other hand, to show the polynomial stability we use the result [2].

Theorem 1.2. Let S (t ) be a bounded C 0 -semigroup on a Hilbert space H with generator A such that i R ⊂
(A). Then

1     c
(i λ I − A)−1   C, ∀λ ∈ R ⇔  S (t )A−1   .
|λ| α L ( H) L ( H) t 1/α

The paper is organized as follows. In Section 2 we will find a necessary and sufficient condition to
get exponential stability. In Section 3 we will
√ show that in general the Timoshenko–Cattaneo system is
polynomially stable, with rate of decay 1/ t. Additionally we show that this rate of decay is optimal.

2. The semigroup approach

In this section we will show that the system (1.5)–(1.10) is well posed using the semigroup tech-
niques. Let us denote for H = H 01 (0, l) × L 2 (0, l) × H ∗1 (0, l) × L 2∗ (0, l) × L 2 (0, l) × L 2 (0, l) the Hilbert
space with internal product given by

l
   
(U , V ) = 1 Φ 1 Φ 2 + κ ϕx1 + ψ 1 ϕx2 + ψ 2 + 2 Ψ 1 Ψ 2 + bψx1 ψx2 + 3 θ 1 θ 2 + τ q1 q2 dx,
0

where U = (ϕ 1 , Φ 1 , ψ 1 , Ψ 1 , θ 1 , q1 ) T and V = (ϕ 2 , Φ 2 , ψ 2 , Ψ 2 , θ 2 , q2 ) T . Here

l 
L 2∗ (0, l) = f ∈ L 2 (0, l); f dx = 0 , H ∗1 (0, l) = H 1 (0, l) ∩ L 2∗ (0, l),
0

and


H ∗2 (0, l) = f ∈ H ∗1 (0, l); f x ∈ H 01 (0, l) .

Let us define the operator A as

⎛ Φ ⎞
κ
⎜
⎜ 1 (ϕx + ψ)x ⎟
⎟
⎜ Ψ ⎟
⎜ b ⎟
AU = ⎜ ψx − (ϕx + ψ) − θx ⎟ .
κ δ
⎜ 2 2 2 ⎟
⎜ ⎟
⎝ − 13 q x − δ3 Ψx ⎠
− βτ q − τ1 θx
 Author's personal copy

M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733 2719

Under this conditions it is not difficult to see that

D (A) = H 2 (0, l) ∩ H 01 (0, l) × H ∗2 (0, l) ∩ H ∗1 (0, l) × H 01 (0, l) × H 1 (0, l).

Since the operator A is dissipative and

l
Re(AU , U )H = −β |q|2 dx, (2.1)
0

we conclude that

l
|q|2 dx  U H  F H , (2.2)
0

where F = ( f 1 , f 2 , f 3 , f 4 , f 5 , f 6 ) T .
As showed in [5] we have the following theorem:

Theorem 2.1. The operator A is the infinitesimal generator of a C 0 -semigroup of contractions over the Hilbert
space H.

3. Exponential stability

Here we will show the exponential stability. To do this we will show, following [11], that the
resolvent is uniformly bounded over the imaginary axes. Note that the resolvent system in terms of
the coefficients is given by

i λϕ − Φ = f 1 , (3.1)

i λ 1 Φ − κ (ϕx + ψ)x = 1 f 2 , (3.2)

i λψ − Ψ = f 3 , (3.3)

i λ 2 Ψ − bψxx + κ (ϕx + ψ) + δθx = 2 f 4 , (3.4)

i λ 3 θ + q x + δΨx = 3 f 5 , (3.5)

i λτ q + β q + θ x = τ f 6 , (3.6)

where λ ∈ R. Our starting point is to show that i R ∩ σ (A) = ∅, where σ (A) is the spectrum of A.
Note that 0 ∈ (A) therefore A−1 is bounded and it is a bijection between H and the domain D (A).
Since D (A) has compact embedding into H it follows that A−1 is a compact operator, which implies
that the spectrum of A is discrete.

Lemma 3.1. Under the above notations we have that

i R ⊂ (A).
 Author's personal copy

2720 M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733

Proof. Let us suppose that A has an imaginary eigenvalue. Then we have that

AW = iλW , λ ∈ R.

From (2.1) we get q = 0, which implies that θ = 0. From Eqs. (3.3) and (3.5) we conclude that
ψ = Ψ = 0. Therefore from (3.4) we get ϕ = 0. This implies that W = 0. But this is a contradiction,
therefore there are not imaginary eigenvalues. 2

Remark 3.1. In particular this result implies that the semigroup is strongly stable, that is

S (t )U 0 → 0,

where S (t ) := e At is the C 0 -semigroup of contractions on Hilbert space H and U 0 is the initial data.

Lemma 3.2. Under the above conditions we have that there exists a positive constant c such that

l
|θ|2 dx  c Ψ q + c U H  F H . (3.7)
0

Proof. From Eq. (3.5) we get

l l
  
i λ 3 θ ds + q(l) − q(x) + δ Ψ (l) − Ψ (x) = 3 f 5 ds.
x x

l
Multiplying the above equation by 0
q dx we have

l l l l l l l

i λ 3 θ ds q dx + q(l) + δΨ (l) q dx − q(x) q dx − δΨ (x) q dx = 3 f 5 ds q dx.
x 0 0 0 0 x 0

Using Eq. (3.6) we obtain

l l l l
3
i λ 3 θ q dx = − θ ds β q − τ f 6 dx.
τ
x 0 x 0

From there it follows that

l l l l l l l
 3
q(l) + δΨ (l) q dx = − θ ds β q − τ f 6 dx + q(x) q dx − δΨ q dx + 3 f 5 ds q dx.
τ
0 x 0 0 0 x 0

Integrating over [0, l] and using (2.2) we can conclude that

 l 
 
 
 q(l) + δΨ (l) q dx  c θq + C Ψ q + C U H  F H . (3.8)
 
0
 Author's personal copy

M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733 2721

On the other hand, integrating over [0, x] ⊂ [0, l] Eq. (3.6), after multiplying by θ and integrating
over [0, l] we obtain

 l x  l x l  l x
2
i λτ q ds θ dx + β q ds θ dx + |θ| dx = τ f 6 ds θ dx.
0 0 0 0 x 0 0

From (3.5) we have

 l x  l x  l x
τ τ
i λτ q ds θ dx = − q ds i λ 3 θ dx = q ds (q x + δΨx − 3 f 5 ) dx
3 3
0 0 0 0 0 0
 l l l  l x 
τ 
= q dx q(l) + δΨ (l) − |q|2 dx − qΨ dx + 3 q ds f 5 dx .
3
0 0 0 0 0

From (3.8) we have

  l x 
 
 
 i λ τ q ds θ dx  c θq + c Ψ q + c U H  F H . (3.9)
 
0 0

Therefore we obtain

l
|θ|2 dx  c θq + c Ψ q + c U H  F H .
0

Using (2.2) our conclusion follows. 2

Lemma 3.3. Under the above conditions we have that there exists a positive constant c such that

l
c
|Ψ |2 dx  c U H  F H + U H θ. (3.10)
|λ|
0

Moreover for any ε > 0 there exists c ε > 0 such that

l l
2 cε ε
|ψx | dx  U H θ + |ϕx + ψ|2 dx + c ε U H  F H , (3.11)
|λ| |λ|2
0 0

for |λ| large enough.

x
Proof. Multiplying Eq. (3.5) by 0
Ψ ds we have

l l l x l l x
2
i λ 3 θ Ψ ds dx + qx Ψ ds dx + δ |Ψ | dx = 3 f5 Ψ ds dx.
0 0 0 0 0 0 0
 Author's personal copy

2722 M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733

From (3.4) we have

l l x l x
δ |Ψ |2 dx = 3 f5 Ψ ds dx − qx Ψ ds dx
0 0 0 0 0

l x
3 
+ θ −bψxx + κ (ϕx + ψ) + δθx − 2 f 4 dx.
2
0 0

From there it follows that

l
c
|Ψ |2 dx  c U H  F H + U H θ + c θψx  + c θ2 .
|λ|
0

Using Lemma 3.2 we arrive at

l
c
|Ψ |2 dx  c ε U H  F H + U H θ + c θψx . (3.12)
|λ|
0

Multiplying Eq. (3.4) by ψ we get

l l l l l
2
b |ψx | dx = −i λ 2 Ψ ψ dx − k (ϕx + ψ)ψ dx − δ θx ψ dx + 2 f 4 ψ dx
0 0 0 0 0

l l l
2 ε 2
 cε |Ψ | dx + |ϕx + ψ| dx + c |θ|2 dx
|λ|2
0 0 0

l
b
+ |ψx |2 dx + c U H  F H .
2
0

From Lemma 3.2 we arrive at

l l l
2 2 
|ψx | dx   |Ψ | dx + |ϕx + ψ|2 dx + c U H  F H . (3.13)
|λ|2
0 0 0

From (3.12), (3.13) and Lemma 3.2 we conclude that

l
c
|Ψ |2 dx  c U H  F H + U H θ.
|λ|
0

The proof is now complete. 2

 Author's personal copy

M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733 2723

Remark 3.2. From (2.2) and (3.7) we get

l  
2 1/2 1/2 c 1/2 1/2
|θ| dx  c c U H  F H + U H θ q + c  U H  F H
|λ|1/2
0

 U H θ + c  U H  F H
|λ|

 U 2H + c  U H  F H
|λ|2

for |λ| large enough.

Here we will introduce two important numbers we call as dissipative numbers, associated to
Timoshenko–Cattaneo systems:

  
1 b 1 τ 1 δ 2 1
χ0 := τ − 2 − − , χ1 := τ − . (3.14)
3 κ κ 3 κ κ 3

The above numbers will characterize the exponential and polynomial stability of the system.

Lemma 3.4. Let us suppose that χ1 = 0, then we have that

  l  l 
 c   
κχ1 −  |ϕx + ψ|2 dx  |χ0 | Ψ Φx dx + c U H θ + c U H  F H .
 |λ|2    |λ|
0 0

If χ1 = 0 then we have

l  l 
 
  c
|ϕx + ψ|2 dx  c  Ψ Φx dx + U H θ + c |λ|2 U H  F H ,
  |λ|
0 0

for |λ| large enough.

Proof. Multiplying Eq. (3.4) by ϕx + ψ and integrating by parts over [0, l] we get

l l l l
2
κ |ϕx + ψ| dx = − i λ 2 Ψ ϕx dx − i λ 2 Ψ ψ dx − b ψx [ϕx + ψ]x dx
0 0 0 0
        
:= I 1 := I 2 := I 3

l l
+δ θx [ϕx + ψ]x dx + 2 f 4 [ϕx + ψ] dx.
0 0
  
:= I 4

Substituting ϕ , ψ and [ϕx + ψ]x given, respectively, by (3.1), (3.3), (3.3) into I 1 , I 2 , I 3 and I 4 ,
respectively, we get
 Author's personal copy

2724 M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733

l l l l l
λ 1 b λ 1 δ
κ |ϕx + ψ|2 dx = −λ 2 Ψ ϕx dx − λ 2 Ψ ψ dx + ψx Φ dx − θΦ dx
κ κ
0 0 0 0 0

l
+ f 4 [ϕx + ψ] dx
0

  l l l
1 b λ 1 δ
= 2 − Ψ Φx dx − i θΦ dx − λ 2 Ψ ψ dx
κ κ
0 0 0

l l l l
δ 1 b
− θ f 2 dx + 2 Ψ f 1,x dx − f 3,x Φ dx + f 4 [ϕx + ψ] dx .
κ κ
0 0 0 0
  
:= R 1

From where we have

l   l l
1 b λ 1 δ
κ |ϕx + ψ|2 dx = 2 − Ψ Φx dx − i θ Φ dx + 2 Ψ 2 + R 1 . (3.15)
κ κ
0 0 0

On the other hand, multiplying Eq. (3.5) by Φ and integrating by parts we get

l l
λ 1 δ 1 δ
i θ Φ dx = [−q x − δΨx + 3 f 5 ]Φ dx
κ 3 κ
0 0

l l l
1 δ 1 δ 2 1 δ
= qΦx dx − Ψx Φ dx + f 5 Φ dx
3 κ 3 κ κ
0 0 0

l l
1 δ 2 1 δ
=− Ψx Φ dx − i λqϕx dx + R 2 ,
3 κ 3 κ
0 0

where

l l
1 δ 1 δ
R2 = − q f 1,x dx + f 5 Φ dx.
3 κ κ
0 0

Using (3.3) we get

l l l
λ 1 δ 1 δ 2 1 δ
i θΦ dx = − Ψx Φ dx − i λq[ϕx + ψ] dx
κ 3 κ 3 κ
0 0 0

l l
1 δ 1 δ
+ qΨ dx − q f 3 dx + R 2 . (3.16)
3 κ 3 κ
0 0
 Author's personal copy

M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733 2725

If χ1 = 0, then the substitution of (3.16) into (3.15) implies the second inequality. Let us suppose
that χ1 > 0. Then using (3.5) and (3.1) we have

l l l
λ 1 δ 1 δ 1 δ 2
θΦ dx = − i λqϕx dx − Ψx Φ dx + R 2 , (3.17)
κ 3 κ 3 κ
0 0 0

where R 2  c U H  F H . From (3.6) we get

l l l l
1 δ 1 δ 1 δ 1 δ
− i λqϕx dx = β qϕx dx + θx ϕx dx − f 6 ϕx dx,
3 κ τ 3 κ τ 3 κ 3 κ
0 0 0 0
  
:= R 3

that is to say,

l l l
1 δ 1 δ β 1 δ
− i λqϕx dx = θx ϕx dx + qϕx dx + R 3 , (3.18)
3 κ τ 3 κ τ 3 κ
0 0 0

and using (3.4) we have

l l
1 δ 1 
θx ϕx dx = − i λ 2 Ψ − bψxx + κ (ϕx + ψ) − 2 f 4 ϕx dx
τ 3 κ τ 3 κ
0 0

l l l
1 1 1
= 2 Ψ λϕx dx + bψxx ϕx dx − κ (ϕx + ψ)ϕx dx
τ 3 κ τ 3 κ τ 3 κ
0 0 0

l
1 2
− f 4 ϕx dx
τ 3 κ
0

l l l
1 1 1
= 2 Ψ Φx dx − 2 Ψ f 1,x dx + bψxx [ϕx + ψ] dx
τ 3 κ τ 3 κ τ 3 κ
0 0 0

l l l
1 κ 1 2 1
− bψxx ψ dx − |ϕx + ψ| dx + κ (ϕx + ψ)ψ dx
τ 3 κ τ 3 κ τ 3 κ
0 0 0

l
1 2
− f 4 ϕx dx.
τ 3 κ
0

From there it follows

l l l l
1 δ 1 1 b 1
θx ϕx dx = 2 Ψ Φx dx − 2 Ψ f 1,x dx − ψx [ϕx + ψ]x dx
τ 3 κ τ 3 κ τ 3 κ τ 3 κ
0 0 0 0
 Author's personal copy

2726 M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733

l l l
b 1 κ 1 1
+ |ψx |2 dx − |ϕx + ψ|2 dx + κ (ϕx + ψ)ψ dx
τ 3 κ τ 3 κ τ 3 κ
0 0 0

l
1 2
− f 4 ϕx dx.
τ 3 κ
0

Finally, using (3.2) we get

l l l l
1 δ 1 1 b 12
θx ϕx dx = 2 Ψ Φx dx − 2 Ψ f 1,x dx − ψx λΦ dx
τ 3 κ τ 3 κ τ 3 κ τ 3 κ 2
0 0 0 0

l l l
b 1 2 κ 1 2 1
+ |ψx | dx − |ϕx + ψ| dx + κ (ϕx + ψ)ψ dx
τ 3 κ τ 3 κ τ 3 κ
0 0 0

l l
b 1 1 2
− ψx f 2 dx − f 4 ϕx dx,
τ 3 κ 2 τ 3 κ
0 0

or

l   l l
1 δ 1 2 b 12 b 1
θx ϕx dx = − Ψ Φx dx + |ψx |2 dx
τ 3 κ τ 3 κ τ 3 κ 2 τ 3 κ
0 0 0

l l
κ 1 2 1
− |ϕx + ψ| dx + κ (ϕx + ψ)ψ dx
τ 3 κ τ 3 κ
0 0

l l l l
b 1 1 b 12 1 2
− ψx f 2 dx − 2 Ψ f 1,x dx + f 1 Φ dx + f 4 ϕx dx .
τ 3 κ 2 τ 3 κ τ 3 κ 2 τ 3 κ
0 0 0 0
  
:= R 4

Substituting the above equality into (3.18) we get

l   l
1 δ 1 2 b 12
− i λqϕx dx = − Re Ψ Φx dx
3 κ τ 3 κ τ 3 κ 2
0 0

l l
κ 1 b 1
− |ϕx + ψ|2 dx + |ψx |2 dx
τ 3 κ τ 3 κ
0 0

l l
1 β 1 δ
+ κ (ϕx + ψ)ψ dx + qϕx dx + R 3 + R 4 .
τ 3 κ τ 3 κ
0 0
 Author's personal copy

M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733 2727

Now, substituting the above equation into (3.17) we have

l   l
λ 1 δ 1 2 b 12 1 δ 2
i θΦ dx = − + Re Ψ Φx dx
κ τ 3 κ τ 3 κ 2 3 κ
0 0

l l
κ 1 2 b 1
− |ϕx + ψ| dx + |ψx |2 dx
τ 3 κ τ 3 κ
0 0

l l
1 β 1 δ
+ κ (ϕx + ψ)ψ dx + qϕx dx + R 2 + R 3 + R 4 .
τ 3 κ τ 3 κ
0 0

Inserting the above inequality into (3.15) we have

  l     l
κ 1 1 b 1 τ 1 δ 2
τκ − |ϕx + ψ|2 dx = τ− 2 − − Ψ Φx dx
3 κ 3 κ κ 3 κ
0 0

l l l
β 1 δ b 1 2
+ qϕx dx − |ψx | dx + τ 2 |Ψ |2 dx
3 κ 3 κ
0 0 0

l
1
− κ (ϕx + ψ)ψ dx + τ R 1 + τ R 2 + τ R 3 + τ R 4 .
3 κ
0

Taking the real part and then absolute value of the above expression and using

l l l
β 1 δ β 1 δ β 1 δ
qϕx dx = q[ϕx + ψ] dx − qψ dx,
3 κ 3 κ 3 κ
0 0 0

we get

l  l  l
  
2  
κχ1 |ϕx + ψ| dx  χ0  Ψ Φx dx + c qU H + c |ψx |2 + |Ψ |2 dx
 
0 0 0

l
c
+ |ϕx + ψ|2 dx + c U H  F H .
|λ|2
0

Using Lemma 3.3 and recalling the definition of χ0 and χ1 we get

  l  l 
 
c   c
χ1 − |ϕx + ψ|2 dx  χ0  Ψ Φx dx + 2 U H θ + c U H  F H , (3.19)
|λ|2   |λ|
0 0

from where our conclusion follows. 2

 Author's personal copy

2728 M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733

Theorem 3.1. The system (1.5)–(1.8) is exponentially stable if and only if χ0 = 0.

Proof. Multiplying Eq. (3.2) by ϕ we get

 l l l
2 2
1 |Φ| dx = κ |ϕx + ψ| dx − κ (ϕx + ψ)ψ dx + 1 f 2 ϕ dx.
0 0 0

From there it follows

 l l
1 |Φ|2 dx  c |ϕx + ψ|2 dx + c |Ψ |2 dx + c U H  F H .
0 0

Since χ0 = 0 we have that

  
1 b 1 τ 1 δ 2
τ− 2 − = > 0.
3 κ κ 3 κ
  
:=χ1

Therefore χ1 = 0. Using the first inequality in Lemma 3.4 for |λ| large enough we get

l
|ϕx + ψ|2 dx  c qU H + c U H θ + c U H  F H . (3.20)
0

Then the above inequality together with Lemma 3.3 implies that

1 |Φ|2 dx  c qU H + c U H θ + c U H  F H . (3.21)

From Lemma 3.3 we have

l
|Ψ |2 dx  c U H  F H + c U H θ. (3.22)
0

Lemma 3.3 together with inequality (3.20) also implies that

l
|ψx |2 dx  c U H θ + c qU H + c U H  F H . (3.23)
0

Finally, Lemma 3.2 implies

l
|θ|2 dx  c U H q + c U H  F H . (3.24)
0
 Author's personal copy

M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733 2729

Using inequalities (3.20)–(3.24) and recalling the definition of the norm of U H we get that

U 2H  c U H q + c U H  F H + c U H θ,

or

U 2H  c 0 q2 + c 0 θ2 + c 0  F 2H

 c 1 U H  F H + c 1 U H q + c 0  F 2H ,

which implies that there exists a positive constant c 2 that does not depend on |λ| such that

U H  c 2  F H .

From there the exponential decay holds.

To show that the condition is also necessary let us assume that there exists U ∈ H such that
U H = 0. Without loss of generality we can take f 1 = f 3 = f 4 = f 5 = f 6 = 0 and choose f 2 (x) =
sin(μx)/ρ1 in system (3.1)–(3.6) with l = π and such that F = (0, f 2 , 0, 0, 0, 0) is limited in H. Be-
cause of the boundary conditions (1.9) we can suppose that

ϕ = A sin(μx), ψ = B cos(μx), θ = D sin(μx), q = E cos(μx). (3.25)

Therefore, system (3.1)–(3.6) is equivalent to


− 1 λ2 + κμ2 A + κμ B = 1, (3.26)

κμ A + − 2 λ2 + bμ2 + κ B + δ μ D = 0, (3.27)

i 3 λ D − μ E − i δλμ B = 0, (3.28)

(i τ λ + β) E + μ D = 0, (3.29)

where μ ∈ N. So we have that

μ
E =− D.
iτ λ + β

Therefore we can rewrite the above system as

    
p1 κμ 0 A 1
κμ p2 δμ B = 0 ,
0 −i δλμ p3 D 0

where

μ2
p 1 = − 1 λ2 + κμ2 , p 2 = − 2 λ2 + bμ2 + κ , p 3 = i 3 λ + .
iτ λ + β

Therefore

p 2 p 3 + i λδ 2 μ2 K
A= = , (3.30)
p 1 p 2 p 3 + i λδ 2 μ2 p 1 − κ 2 μ2 p 3 p 1 K − κ 2 μ2
 Author's personal copy

2730 M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733

where K = p 2 + i λδ 2 μ2 / p 3 . Then we have

−τ λ2 + i βλ
K = p 2 + δ 2 μ2 .
− 3 τ λ2 + i β 3 λ + μ2

Now, let us take λ such that p 1 (λ) = d, that is λ2 = κμ2 / 1 − d/ 1 , where d ∈ R is going to be
fixed later. Then we get

−τ κμ2 / 1 + τ d/ 1 + i βλ
K = p 2 + δ 2 μ2 . (3.31)
(1 − 3 τ κ / 1 )μ2 + 3 τ d/ 1 + i β 3 λ

Let us denote χ2 := 1 − 3 τ1 κ . If χ1 = 0, then we have that χ2 = − 3 τ1 κ χ1 = 0. In this case we take

d = 0, so K can be written as

 
2 κ δ 2 μ2 δ 2 μ2
K =− − b μ2 + κ + τ λ+ . (3.32)
1 i β 3 3

Therefore since d = 0 we have that

 
K τ δ2 1 2 b δ2 1 τ δ2
A=− 2 2 =− 2 λ+ − − − ≈− 2 λ,
κ μ i κ β 3 κ 1 κ κ 3 κμ2 i κ β 3

for λ large. Therefore

π
 
U μ 2H  1 Φμ 2 = A 2 |λ|2 1 sin(μx)2 dx
0

π π τ 2 δ4
= | A |2 |λμ |2 1 ≈ |λμ |4 1 . (3.33)
2 2 κ 4 β 2 32

This implies that

U μ H → ∞,

as μ → ∞. Finally, suppose that χ1 = 0 and χ0 = 0. Then from (3.31) we get

−τ κμ2 / 1 + τ d/ 1 + i βλ
K = p 2 + δ 2 μ2
χ2 μ2 + 3 τ d/ 1 + i β 3 λ
(1 − 2χ2 )τ d/ 1 − i βλ
(τ κ / 1 )
= p 2 − δ 2 μ2 − δ 2 μ2
χ2 χ2 (χ2 μ2 + 3 τ d/ 1 + i β 3 λ)
 
κ 2 τ κ / 1 2 d (1 − 2χ2 )τ d − i βλ
= − 2 +b−δ μ2 + k + − δ 2 μ2 .
1 χ2 1 χ2 (χ2 μ2 + 3 τ d/ 1 + i β 3 λ)

We will take d such that

 
κ τ κ / 1 1 χ1
− 2 + b − δ2 d = κ2 ⇒ d= .
1 χ2 κχ0
 Author's personal copy

M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733 2731

Therefore we have that

κ2 2 d (1 − 2χ2 )τ d − i βλ
K= μ2 + κ + − δ 2 μ2
d 1 χ2 (χ2 μ2 + 3 τ d + β 3 λ)
κ2 2 d βδ 2 (1 − 2χ2 )τ dδ 2
≈ μ2 + κ + +i 2 λ−
d 1 χ2 χ22

for λ large. From (3.30) and recalling that λ is such that p 1 (λ) = d we have

κ 2 μ2 + κ + 2 d + i βδ 2 λ − (1−2χ2 )τ dδ 2
d 1 χ22 χ22 χ22 κ
A= ≈ − λ.
2 2 (1−2χ2 )τ d2 δ 2 2 2
κ d + 2 d1 + i βδ
χ22
d λ −
χ22
1 βδ d

Therefore

π
 
U μ 2H  1 Φμ 2 = A 2 |λ|2 1 sin(μx)2 dx
0
π
= | A |2 |λμ |2 1 ≈ c 0 |λμ |4 . (3.34)
2

This implies that

U μ H → ∞

as μ → ∞. Therefore, our conclusion follows. 2

4. Polynomial decay

√In this section we will show that in general the Timoshenko system goes to zero polynomially as
1/ t.

Theorem 4.1. Let us suppose that χ0 = 0, then the semigroup is polynomially stable and

  1
 S (t )U 0   √  U 0  D (A ) .
H
t

Moreover, this rate of decay is optimal.

Proof. Let us suppose that χ1 > 0. Using Lemma 3.4 we get

l  l 
 
  c
|ϕx + ψ|2 dx  c |χ0 | Ψ Φx dx + U H θ + c U H  F H
  |λ|
0 0
 l 
 
  c
 c |λ| Ψ ϕx dx + U H θ + c U H  F H . (4.1)
  |λ|
0
 Author's personal copy

2732 M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733

Multiplying Eq. (3.2) by ϕ we get

  l 
 
  c c
1 |Φ|2 dx  c |λ| Ψ ϕx dx + U H θ + U H Ψ  + c U H  F H .
  |λ| |λ|
0

From Lemma 3.3 and by the inequalities (2.2), (3.7), (3.10), (3.11) we get for |λ| large enough

 l 
 
  c c
U 2H  c |λ| Ψ ϕx dx + U H θ + U H Ψ  + c U H  F H
  |λ| |λ|
0

l
1 2 2
 U H + c |λ| |Ψ |2 dx + c θ2 + c  F 2H
8
0

1
 U 2H + c |λ|U H θ + c |λ|4  F 2H
6
1
 U 2H + c |λ|2 θ2 + c  |λ|4  F 2H
4
1
 U 2H + c  |λ|4  F 2H .
2

From there it follows that

1
U H  C  F H ,
|λ|2

which is equivalent to

 
(λ I − A)−1   C |λ|2 .
L ( H)

Then using Theorem 2.4 in [2], we obtain

   1   C
 S (t )A−1  = O t− 2 ⇒  S (t )A−1 F   √  F H .
L ( H) L ( H)
t

Since 0 ∈ ρ (A), it follows that A is onto over H, then taking AU 0 = F , we get

  C
 S (t )U 0   √  U 0  D (A ) .
H
t

Therefore the solution decays polynomially. In case of χ0 = 0 and χ1 = 0 we use the same ideas
as above. So the polynomial decay holds.
Finally, to show the optimality we follow the same ideas of the proof of Theorem 4.1 Note that in
case of χ1 = 0 or χ0 = 0, it is valid inequality (3.34), therefore

U 2H  c 0 |λ|4 (4.2)

 Author's personal copy

M.L. Santos et al. / J. Differential Equations 253 (2012) 2715–2733 2733

for |λ| large enough. If we assume that the rate of decay can be improved from 1/t 1/2 to 1/t 1/(2− )
for some  > 0, then we will have that

1
U H
|λ|2−

must be bounded. But this is not possible because of the inequality (4.2). The proof is now com-
plete. 2

Acknowledgments

The authors are grateful to the referees for the valuable suggestions that improved this paper. This
work has been done thanks to the support of the Faperj: E-26/102.812/2008.

References

[1] M. Alves, C.A. Raposo, M. Sepulveda, O. Vera, Uniform stabilization for transmission problem for Timoshenko’s system with
memory, J. Math. Anal. Appl. 369 (1) (2010) 323–345.
[2] A. Borichev, Y. Tomilov, Optimal polynomial decay of functions and operator semigroups, Math. Ann. 347 (2) (2009) 455–
478.
[3] Farid Ammar Khodja, Assia Benabdallah, J.E. Muñoz Rivera, Reinhard Racke, Energy decay for Timoshenko systems of mem-
ory type, J. Differential Equations 194 (1) (2003) 82–115.
[4] H.D. Fernandez Sare, Exponential decay of Timoshenko systems with indefinite memory dissipation, Adv. Differential Equa-
tions 13 (7–8) (2009) 733–752.
[5] H.D. Fernandez Sare, Reinhard Racke, On the stability of damped Timoshenko systems: Cattaneo versus Fourier law, Arch.
Ration. Mech. Anal. 194 (1) (2009) 221–251.
[6] Salim A. Messaoudi, Muhammad I. Mustafa, A stability result in a memory-type Timoshenko system, Dynam. Systems
Appl. 18 (3–4) (2009) 457–468.
[7] F. Huang, Characteristic conditions for exponential stability of linear dynamical systems in Hilbert spaces, Ann. Differential
Equations 1 (1985) 43–56.
[8] M. Grasselli, V. Pata, G. Prouse, Longtime behavior of a viscoelastic Timoshenko beam, Discrete Contin. Dyn. Syst. 10 (1–2)
(2004) 337–348.
[9] Z. Liu, S. Zheng, Semigroups Associated with Dissipative Systems, CRC Res. Notes in Math., vol. 398, Chapman & Hall, 1999.
[10] Salim A. Messaoudi, Belkacem Said-Houari, Energy decay in a Timoshenko-type system of thermoelasticity of type III,
J. Math. Anal. Appl. 348 (1) (2008) 298–307.
[11] J. Prüss, On the spectrum of C 0 -semigroups, Trans. Amer. Math. Soc. 284 (1984) 847–857.
[12] Salim A. Messaoudi, Belkacem Said-Houari, Energy decay in a Timoshenko-type system with history in thermoelasticity of
type III, Adv. Differential Equations 14 (3–4) (2009) 375–400.
[13] C.A. Raposo, J. Ferreira, M.L. Santos, N.N.O. Castro, Exponential stability for Timoshenko system with two weak dampings,
Appl. Math. Lett. 18 (2005) 535–541.
[14] Jong Uhn Kim, Yuriko Renardy, Boundary control of the Timoshenko beam, SIAM J. Control Optim. 25 (6) (1987) 1417–
1429.
[15] F. Ammar-Khodja, S. Kerbal, A. Soufyane, Stabilization of the nonuniform Timoshenko beam, J. Math. Anal. Appl. 327 (1)
(2007) 525–538.
[16] Salim A. Messaoudi, Abdelaziz Soufyane, Boundary stabilization of solutions of a nonlinear system of Timoshenko type,
Nonlinear Anal. 67 (7) (2007) 2107–2121.
[17] Reinhard Racke, Thermoelasticity with second sound—exponential stability in linear and non-linear 1-d, Math. Methods
Appl. Sci. 25 (5) (2002) 409–441.
[18] Abdelaziz Soufyane, Ali Wehbe, Uniform stabilization for the Timoshenko beam by a locally distributed damping, Electron.
J. Differential Equations 2003 (29) (2003) 1–14.
[19] Ali Wehbe, Wael Youssef, Stabilization of the uniform Timoshenko beam by one locally distributed feedback, Appl.
Anal. 88 (7) (2009) 1067–1078.
[20] Salim A. Messaoudi, Muhammad I. Mustafa, On the stabilization of the Timoshenko system by a weak nonlinear dissipation,
Math. Methods Appl. Sci. 32 (4) (2009) 454–469.
[21] Salim A. Messaoudi, Michael Pokojovy, Belk Said-Houari, Nonlinear Damped Timoshenko Systems with Second Sound—
Global Existence and Exponential Stability, Konstanzer Schr. Math. Inform., vol. 246, Universität Konstanz, 2008.
[22] S.P. Timoshenko, On the correction for shear of the differential equation for transverse vibrations of prismatic bars, Philos.
Magazine 6 (1921) 744–746.
[23] Jaime E. Muñoz Rivera, Reinhard Racke, Mildly dissipative nonlinear Timoshenko systems—global existence and exponential
stability, J. Math. Anal. Appl. 276 (1) (2002) 248–278.
[24] T. Sabri Öncü, T. Bryant Moodie, On the constitutive relations for second sound in elastic solids, Arch. Ration. Mech.
Anal. 121 (1) (1992) 87–99.