Abstract
In this paper, we consider a new Timoshenko beam model with thermal and mass diffusion effects where heat and mass diffusion flux are governed by Cattaneo’s law. Necessary and sufficient conditions for exponential stability are provided in terms of the physical parameters of the model. Firstly, by the \(C_0\)-semigroup theory, we prove the well-posedness of the considered problem. Then, we prove the lack of exponential stability of the system when one of these conditions is not valid. Finally, we prove in this case that the semigroup decays to zero polynomially as \(1/\sqrt{t}\). Moreover, we show that the rate is optimal.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Beams represent the most common structural element in engineering and mechanics structures. Because of their ubiquity, they are widely studied, from mechanical and mathematical points of view. In materials mechanics, vibration has been known for a long time for a source of disturbance, discomfort, damage and destruction. However, the requirement for a more precise control of its vibrations has led to consider all the possible effects that a beam can undergo in a structure. One might perhaps think that the classical theory of thermoelasticity is a good model for explaining thermal conduction in this kind of problem. For a long time, the effects of diffusion have been ignored in the frame of the classical linear theory of thermoelasticity. Maybe we can think that the classical theory of thermoelasticity is a good model to explain the thermal conduction in contact problems. However, the research conducted in the development of high technologies after the second world war, confirmed that the fields of temperature and diffusion in solids cannot be ignored. So, the obvious question is what happens when the diffusion effect is considered with the thermal effect in contact problems. Diffusion can be defined as the random walk of a set of particles from regions of high concentration to regions of lower concentration. Thermodiffusion in an elastic solid is due to coupling of the fields of strain, temperature and mass diffusion. The processes of heat and mass diffusion play an important role in many contact engineering applications, such as satellites problems, returning space vehicles and aircraft landing on water or land.
Recently, Aouadi et al. [3] have considered the effect of mass diffusion effect in a thermo-Timoshenko beam. If the mass diffusion is taken into account in the Timoshenko equations, the evolution equations are given by
where \(\varphi \) is the transverse displacement and \(\psi \) is the rotation of the neutral axis due to bending. Here, \(\rho _{1}=\rho A\) and \(\rho _{2}=\rho I\), where \(\rho >0\) is the density, A is the cross-sectional area and I is the second moment of the cross-sectional area. By S, we denote the shear force and M is the bending moment, \(\Psi \) is the entropy, q is the heat flux, C is the concentration of the diffusive material in the elastic body, and \(\eta \) is the mass diffusion flux. In this case, the constitutive equations with temperature and mass diffusion are given by [3]
where b and \(\kappa \) stand for \(b=EI\) and \(\kappa =k_{1}GA\) where E, G and \(k_1\) represent the Young’s modulus, the modulus of rigidity and the transverse shear factor, respectively. Here, P is the chemical potential, \(\varpi \) is a measure of the thermodiffusion effect, \(\varrho \) is a measure of the diffusive effect, and \(\gamma \) and \(\beta \) are the coefficients of thermal and mass diffusion expansions, respectively. Substituting (1.2) into (1.1), we get the Timoshenko equations with thermodiffusion effects
where \((x,t)\in (0,l)\times {\mathbb {R}}^{+}.\) We shall now formulate a different alternative form where the chemical potential P is considered as a state variable instead of the concentration C. This alternative form is obtained by substituting the last equation of (1.2) into (1.3)\({}_{2-4}\)
where
are physical positive constants.
Aouadi et al. [3] proved the well posedness of (1.4) with Dirichlet or Neumann boundary conditions when the temperature and the mass diffusion follow the Fourier’s law and the Fick’s law, respectively,
Then, they showed, without assuming the well-known equal wave speeds condition, the lack of exponential stability for the Neumann problem; meanwhile, one linear frictional damping is strong enough to guarantee the exponential stability for the Dirichlet problem.
The drawback of the Fourier law lies in the physical paradox of infinite propagation speed of (thermal) signals, a typical side-effect of parabolicity. A different model, removing this paradox, is the Cattaneo’s law [8], namely the differential perturbation of (1.5)
where the relaxation time \(\tau _0\) describes the time lag in the response of the heat flux to a gradient in the temperature, while \(\tau _{1}\) is the diffusion relaxation time, which will ensure that the equation satisfied by the concentration will also predict finite speeds of propagation of matter from one medium to the other. In [2, 6], Cattaneo’s law is applied instead to Fick’s law in order to remove the physical paradox that affects such a model.
Inserting (1.6) into (1.4), we get the thermodiffusion Timoshenko beam equations with second sound in \((x,t)\in (0,l)\times {\mathbb {R}}^{+}\)
with the initial conditions
and the boundary conditions
A natural question is under what necessary and sufficient conditions, the semigroup generated by Timoshenko systems is exponentially stable. Since there is not much literature on Timoshenko system with mass diffusion, the answer is known only in the case of Timoshenko systems with or without thermal effect. Indeed, in the case where the thermal effect is absent, the Timoshenko system is uniformly stable for weak solutions if
Consequently, the number \(\chi \) plays an important role in the asymptotic behavior of solutions to Timoshenko systems with or without thermal effect. Of course, since the Timoshenko system is a two by two system hyperbolic equations, many authors showed that the dissipation given by some damping terms is strong enough to stabilize the system exponentially regardless of whether the propagation velocities are equal or not (see, e.g., [10, 11, 14,15,16, 21]).
When we consider the thermal effect in Timoshenko beam according to Fourier’s law, \(\tau _{0}=0\), Fernández Sare et al. [12] proved that the exponential stability can never occur when \(\chi =0\). Muñoz Rivera and Racke [17] proved several exponential decay results for the linearized system and a non-exponential stability result for the case of different wave speeds. Moreover, when \(\chi \not =0\), the authors showed the polynomial stability. Aouadi and Soufyane [5] showed that the dissipation product by the memory effect working at the boundary is sufficiently strong to produce a general decay obtained without imposing the condition (1.15). In [1], Almeida Júnior et al. considered a Timoshenko beam acting on shear force and proved that the resulting model is exponentially stable if and only if (1.15) holds.
Another option to remove the infinity speed of propagation is to consider the Cattaneo’s law for the heat flux. In this case, the Timoshenko system is given by
The above model was studied by several authors, see, for example, [12, 20, 22] to quote a few. Fernández Sare and Racke [12] showed that, in the absence of the extra frictional damping, the coupling via Cattaneo’s law causes loss of the exponential decay usually obtained in the case of coupling via Fourier’s law. Precisely, it has been shown that (1.16) is no longer exponentially stable even if (1.15) holds. However, Santos et al. [22] proved that (1.18) is exponentially stable if and only if
Note that when \(\tau _0 = 0\), the Cattaneo’s law (1.6)\({}_1\) turns into the Fourier’s law (1.5)\({}_1\) and the condition (1.17) is equivalent to (1.15), which tells at once that
It is worth noting that Aouadi and Boulehmi [4] designed one feedback controller to make the solutions to non-uniform Timoshenko beam acting on shear force within Cattaneo’s law decay exponentially regardless the value of \(\chi _{\tau _0}\).
Later, Dell’Oro and Vittorino Pata [9] studied the same problem for the system
describing a Timoshenko beam coupled with Gurtin–Pipkin heat conduction law for the heat flux. Then, they introduced the stability number
They proved that the corresponding semigroup is exponentially stable if and only if \(\chi _{g}=0\). In particular, they showed that the corresponding semigroup remains stable (although not exponentially stable) also when \(\chi _{g}\not =0\). Moreover, they generalized previously known results on the Fourier–Timoshenko and the Cattaneo–Timoshenko beam models.
In this present work, we consider the Timoshenko system with mass diffusion and second sound effects, given by (1.7)–(1.14). We introduce some new numbers by the physical coefficients that characterize the exponential decay
where \(\xi =1-\frac{\rho _{1}\Gamma }{\delta \kappa }\), \(\delta =cr-d^2>0\) and \(\Gamma =\big (d\gamma _{2}-r\gamma _{1}\big )\frac{K}{\tau _0\gamma _{1}}=\big (d\gamma _{1}-c\gamma _{2}\big )\frac{\hbar }{\tau _1\gamma _{2}}\) if \(\chi _{0}=0\). If \(\chi _{0}=0\), we prove that the semigroup associated with (1.7)–(1.14) is exponentially stable if and only if \(\chi _{1}=0\). Otherwise, there is a lack of exponential stability. In this case, we prove that the semigroup decays as \(1/\sqrt{t}\).
The method we use to show the lack of exponential stability is based on Gearhart–Herbst–Prüss–Huang theorem to dissipative systems. See also [13, 19].
Theorem 1.1
Let \(S(t) = e^{{\mathcal {A}}t}\) be a \(C_0\)-semigroup of contractions on Hilbert space. Then, S(t) is exponentially stable if and only if \(i{\mathbb {R}}\subset \varrho ({\mathcal {A}})\) and
where \(\varrho ({\mathcal {A}})\) is the resolvent set of the linear operator \({\mathcal {A}}\).
On the other hand, to show the polynomial stability we use Theorem 2.4 in [7].
Theorem 1.2
Let \(S(t) = e^{{\mathcal {A}}t}\) be a \(C_0\)-semigroup of contractions on a Hilbert space with generator \({\mathcal {A}}\) such that \(i{\mathbb {R}}\subset \varrho ({\mathcal {A}})\). Then,
Throughout the paper, C will always stand for a generic positive constant.
2 Well-posedness of the problem
In this section, we prove the existence and uniqueness of solutions for (1.7)–(1.14) using semigroup theory. Introducing the vector function \(U = (\varphi , v, \psi , \phi , \theta , q, P,\eta )^T\), where \(v=\varphi _{t}\) and \(\phi =\psi _{t}\), we consider the following Hilbert space
where
provided with the following inner product
with \(cr-d^2>0\) for all \(U_{1} = (\varphi _{1}, v_{1}, \psi _{1}, \phi _{1}, \theta _{1}, q_{1}, P_{1}, \eta _{1})^{T}\) and
\(U_{2} = (\varphi _{2}, v_{2}, \psi _{2}, \phi _{2}, \theta _{2}, q_{2}, P_{2}, \eta _{2})^{T}\) in \({\mathcal {H}}\) and norm given by
In order to prove the existence and uniqueness of solutions, we will use the semigroup theory [18]. Then, the system (1.7)–(1.12) can be rewritten as follows:
where \(U=\big (\varphi ,\varphi _t,\psi ,\psi _t, \theta , q, P, \eta \big )^{T}\), \(U_0=\big (\varphi _0,\varphi _1,\psi _0,\psi _1, \theta _{0}, q_{0}, P_{0},\eta _{0}\big )^{T}\) and \({\mathcal {A}}:D({\mathcal {A}})\subset {\mathcal {H}}\rightarrow {\mathcal {H}}\) is the operator defined by
where \(\delta :=cr-d^2>0\) and \(I(\cdot )\) is the identity operator. The domain of \({\mathcal {A}}\) is given by
Clearly, \(D({\mathcal {A}})\) is dense in \({{\mathcal {H}}}\). We have the following existence and uniqueness result.
Theorem 2.1
Let \(U_0 \in {\mathcal {H}}\), then there exists a unique solution \(U \in C {({\mathbb {R}}^+, {\mathcal {H}}})\) of problem (2.2). Moreover, if \(U_0 \in D({\mathcal {A}})\), then \(U \in C {({\mathbb {R}}^+, D({\mathcal {A}}))}\cap C^{1} {({\mathbb {R}}^+, {\mathcal {H}})}\).
Proof
The result follows from Lumer–Phillips Theorem provided we prove that \({\mathcal {A}}\) is a maximal monotone operator. In what follows, we prove that \({\mathcal {A}}\) is monotone. For any \(U \in D({\mathcal {A}})\), and using the inner product, we obtain
Since \(K>0\) and \(\hbar >0\), it follows that \(\langle {\mathcal {A}}U,{U}\rangle _{{\mathscr {H}}} \le 0\), which implies that \({\mathcal {A}}\) is dissipative. Next, we prove that the operator \(I - {\mathcal {A}}\) is surjective. Given \(G = (g_1, g_2, g_3, g_4, g_5, g_6, g_7,g_8)^{T} \in {\mathcal {H}}\), we prove that there exists \(U \in D({\mathcal {A}})\) satisfying
that is,
Suppose \(\varphi \), \(\psi \), q and \(\eta \) are found with the appropriate regularity. Then, (2.5)\({}_1\), (2.5)\({}_3\), (2.5)\({}_6\) and (2.5)\({}_8\) yield
From (2.6)\({}_{3,4}\), we have
then \(\theta (0)=\theta (1)=0\) and \(P(0)=P(1)=0\). By using (2.6) and (2.7), it can easily be shown that \(\varphi ,\ \psi ,\ q\) and \(\eta \) satisfy
where
To solve (2.8) we consider the variational formulation
where \(B: {\big [H_0^{1}(0, l) \times H^{1}_{*}(0, l) \times L_{*}^{2}(0, l) \times L_{*}^{2}(0, l)\big ]}^{2}\rightarrow {\mathbb {R}}\) is the bilinear form defined by
and \(F: \big [H_0^{1}(0, l) \times H^{1}_{*}(0, l) \times L_{*}^{2}(0, l) \times L_{*}^{2}(0, l) \big ]\rightarrow {\mathbb {R}}\) is the linear form
Now, for \(V= H_0^{1}(0, l) \times H^{1}_{*}(0, l) \times L_{*}^{2}(0, l) \times L_{*}^{2}(0, l)\), one can easily see that B and F are bounded and B is coercive. Consequently, by Lax–Milgram Lemma, system (2.8) has a unique solution
Substituting \(\varphi ,\ \psi \), q and \(\eta \) in (2.5)\(_1\), (2.5)\(_3\), (2.5)\(_6\), and (2.5)\(_8\), respectively, we obtain
Now, if \(({\tilde{\varphi }}, {\tilde{q}}, {\tilde{\eta }}) \equiv (0, 0,0) \in H_0^{1}(0, l) \times L_{*}^{2}(0, l) \times L_{*}^{2}(0, l)\), then (2.10) reduces to
which implies
Consequently, by the regularity theory for the linear elliptic equations, it follows that
Moreover, (2.11) is also true for any \(u\in C^{1}([0, l]) \subset H_{*}^{1}(0, l)\). Hence, we have
for all \(u \in C^{1}([0, l])\). Thus, using integration by parts and bearing in mind (2.12), we obtain
Therefore, \(\psi _{x}(0)=\psi _{x}(l)=0\). In the same way, if \(({\tilde{\psi }}, {\tilde{q}}, {\tilde{\eta }}) \equiv (0, 0,0) \in H_*^{1}(0, l) \times L_{*}^{2}(0, l) \times L_{*}^{2}(0, l)\), then we obtain
Recalling \(\delta =cr-d^2>0\), the resolution of the system
gives
Finally, the application of the regularity theory for the linear elliptic equations guarantees the existence of a unique \(U \in D({\mathcal {A}})\) such that (2.4) is satisfied. Consequently, \({\mathcal {A}}\) is a maximal operator. Hence, the result of Theorem 2.1 follows from Lumer–Phillips Theorem (see [18]). \(\square \)
The previous lemmas lead to the next theorem.
Theorem 2.2
The operator \({\mathcal {A}} \) generates a \(C_0\)-semigroup of contractions on the phase-space \({\mathcal {H}}\).
Proof
Since the operator \({\mathcal {A}}\) is maximal dissipative in \({\mathcal {H}}\) and \(D({\mathcal {A}})\) is densely defined in \({\mathcal {H}}\), the proof follows from the Lumer–Phillips Corollary to the Hille–Yosida Theorem [18].\(\square \)
3 Exponential stability
We introduced the total energy of the system (1.7)–(1.14) given by
which is positive definite since \(d^2<cr\) and satisfies the dissipation law
Here, we will provide the necessary conditions on the coefficients ensuring the exponential stability of our model. From [19], we have to show that the resolvent is uniformly bounded over the imaginary axes. Note that, for any \(U = (\varphi , v, \psi , \phi , \theta , q, P, \eta )^{T}\in D({\mathcal {A}})\), the resolvent of our model is given by the equation \((i\lambda I - {\mathcal {A}})U=F\), i.e.,
where \(F=(f_{1},f_{2},f_{3},f_{4},f_{5},f_{6},f_{7},f_{8})^{T}\in {\mathcal {H}}\) and \(\lambda \in {\mathbb {R}}\). From (2.3), it is easy to see that we have
Furthermore,
Our starting point is to show that \(i{\mathbb {R}}\subset \varrho ({\mathcal {A}})\). Note that from Theorem 2.1, one can deduce that \(0\in \varrho ({\mathcal {A}})\), therefore \({\mathcal {A}}^{-1}\) is bounded and it is a bijection between \({\mathscr {H}}\) and the domain \(D({\mathcal {A}})\). Since \(D({\mathcal {A}})\) has compact embedding into \({\mathcal {H}}\), it follows that \({\mathcal {A}}^{-1}\) is a compact operator, which implies that the spectrum of \({\mathcal {A}}\) is discrete.
Lemma 3.1
Under the above notations, we have that \(i{\mathbb {R}}\subset \varrho ({\mathcal {A}})\).
Proof
Let us suppose that \({\mathcal {A}}\) has an imaginary eigenvalue. Then, we have that \({\mathcal {A}}U = i\lambda U\), \(\lambda \in {\mathbb {R}}\). From (2.3), we get \(q=\eta = 0\), which implies that \(\theta =P= 0\). From Eqs. (3.5) and (3.7), we conclude that \(\psi = \phi = 0\). Therefore, from (3.7) we get \(\varphi = 0\). This implies that \(U = 0\). But this is a contradiction, therefore there are not imaginary eigenvalues. \(\square \)
Before we prove the technical lemmas that will support our propositions, we present the following definition.
Definition 3.1
Let \(U=(u_{1},u_{2},u_{3},u_{4},u_{5},u_{6},,u_{7},u_{8})^{T}\in D({\mathcal {A}})\) and \(F=(f_{1},f_{2},f_{3},f_{4},f_{5},f_{6},f_{7},f_{8})^{T}\in {\mathcal {H}}\). Then, we define the functional class \({\mathfrak {R}}\) given by
Lemma 3.2
Let \((\varphi ,v,\psi ,\phi , \theta , q, P, \eta )\) be a solution of system (1.7)–(1.14). There is a positive constant C independent of \(\lambda \) such that
Proof
Integrating Eq. (3.7) over \([x,l]\subset [0,l]\) and multiplying by \(\int \limits _{0}^{l}{\overline{q}}\mathrm{d}x\), we get
where \({\mathcal {R}}\in {\mathfrak {R}}\). Now integrating Eq. (3.8) and multiplying by \(\frac{\delta }{\tau _0}\int \limits _{x}^{l}\theta ds\)
From there, it follows that
Integrating over [0, l] and using (3.12), we can conclude that
Similarly, integrating over \([0,x]\subset [0,l]\) Eq. (3.8), after multiplying by \(\theta \) and integrating over [0, l] we obtain
Using Eq. (3.7), we get
Substituting I in (3.18), we have
From (3.17) and (3.12), we have
which reads
On the other hand, combining Eqs. (3.7) and (3.9), we have
Similarly, one can get
This concludes the proof. \(\square \)
Lemma 3.3
Let \((\varphi ,v,\psi ,\phi , \theta , q, P, \eta )\) be a solution of system (1.7)–(1.14). There is a positive constant C independent of \(\lambda \) such that for all \(\varepsilon >0\)
and
Proof
Combining Eqs. (3.7) and (3.9), we have
Multiplying by \(\int \limits _{0}^{x}{\overline{\phi }}ds\), we get
where \({\mathcal {R}}\in {\mathfrak {R}}\). From (3.6), we have
By using Cauchy–Schwarz inequality, we have
From Eqs. (3.3) and (3.5), we infer that
and
Now using Young’s inequality, Lemma 3.2 and the estimate (3.12) we obtain
On the other hand, multiplying Eq. (3.6) by \({\overline{\psi }}\)
Substituting \({\overline{\psi }}\) by Eq. (3.5) and using the Cauchy–Schwarz and Young inequalities, we get
for all \(\varepsilon >0\). Finally, using Lemma 3.2 and Young’s inequality we obtain
Substituting (3.28) in (3.26) and using Lemma 3.2, we have
Note that
Using the two estimates above with Young’s inequality, we get (3.23). \(\square \)
Now, we introduce two important stability numbers, associated with our model (1.7)–(1.14):
where \(\xi =1-\frac{\rho _{1}\Gamma }{\delta \kappa }\), \(\delta =cr-d^2>0\) and \(\Gamma =\big (d\gamma _{2}-r\gamma _{1}\big )\frac{K}{\tau _0\gamma _{1}}=\big (d\gamma _{1}-c\gamma _{2}\big )\frac{\hbar }{\tau _1\gamma _{2}}\) if \(\chi _{0}=0\). Otherwise, if \(\chi _{0}\ne 0\), then \(\chi _{1}\) does not exist.
Lemma 3.4
Let \((\varphi ,v,\psi ,\phi , \theta , q, P, \eta )\) be a solution of system (1.7)–(1.14). Assuming that \(\chi _{0}=0\), there exists a positive constant C independent of \(\lambda \) such that
where \(\xi \ne 0\) and for \(|\lambda |\) large enough. Otherwise, if \(\xi =0\), then we have
Proof
Multiplying Eq. (3.6) by \(\overline{\varphi _{x}+\psi }\) and integrating by parts over [0, l], we get
where \({\mathcal {R}}\in {\mathfrak {R}}\). Substituting \({\overline{\varphi }}_{x}\), \({\overline{\psi }}\) and \(\overline{(\varphi _{x}+\psi )}_{x}\), respectively, by (3.3), (3.5) and (3.4), we get
Here, our aim is to obtain an expression for \(H_{1}\) and \(H_{2}\). For this, we multiply Eq. (3.7) by \(\frac{\rho _{1}\gamma _{1}}{\kappa \delta }{\overline{v}}\) and integrating by parts we get
Substituting \({\overline{v}}_{x}\) by Eq. (3.3), we have
Analogously defining \(H_{2}:=i\lambda \frac{\gamma _{2}\rho _{1}}{\kappa }\int \limits _{0}^{l}P{\overline{v}}\mathrm{d}x\). Multiplying Eq. (3.9) by \(\frac{\rho _{1}\gamma _{2}}{\delta \kappa }{\overline{v}}\) and using Eq. (3.3), we have
From here, the proof is divided into two steps to help the reader follow the arguments and the calculations.
Step 1. By adding and subtracting the terms \(i\lambda \frac{r\gamma _{1}\rho _{1}}{\delta \kappa }\int \limits _{0}^{l}q{\overline{\psi }}\mathrm{d}x\) and \(i\lambda \frac{d\gamma _{1}\rho _{1}}{\delta \kappa }\int \limits _{0}^{l}\eta {\overline{\psi }}\mathrm{d}x\) in (3.34) we get
Then, substituting Eq. (3.5) in \({\overline{\psi }}\) we get
Similarly, we have for Eq. (3.35)
Substituting \(H_1\) and \(H_2\) in (3.33), we have
where \(c_{1}:=\bigg (\rho _{2}-\frac{\alpha \rho _{1}}{\kappa }\bigg )-\bigg [\bigg (r-\frac{d\gamma _{2}}{\gamma _{1}}\bigg )\gamma _{1}^{2}+\bigg (c-\frac{d\gamma _{1}}{\gamma _{2}}\bigg )\gamma _{2}^{2}\bigg ]\frac{\rho _{1}}{\delta \kappa }. \) By using the Young’s inequality and estimate (3.12), we get
and then
By using Lemma 3.3, (3.32) follows immediately.
Step 2. We aim to obtain the expressions of \(H_{1}\) and \(H_{2}\). Bearing in mind \(I_{1}\) in Eq. (3.34), we multiply Eq. (3.8) by \(\frac{r\gamma _{1}\rho _{1}}{\tau _0\delta \kappa }{\overline{\varphi }}_{x}\) to get
Now, we add and subtract the term \(\frac{r\gamma _{1}\rho _{1}}{\tau _0\delta \kappa }\int \limits _{0}^{l}q{\overline{\psi }}\mathrm{d}x\) to get
Performing the same procedure in \(I_{2}\) in Eq. (3.34), we have
Substituting \(I_{1}\) and \(I_{2}\) in (3.34), we obtain
Performing the same procedure in \(H_{2}\) in Eq. (3.35), we have
Adding the simplifications \(H_{1}\) and \(H_{2}\), we get
where \(\Gamma :=\big (r\gamma _{1}-d\gamma _{2}\big )\frac{K}{\gamma _{1}\tau _0}=\big (c\gamma _{2}-d\gamma _{1}\big )\frac{ \hbar }{\gamma _2\tau _1}\) since \(\chi _{0}=0\). From J and (3.6), we obtain
Substituting \({\overline{\varphi }}_{x}\), \(\overline{(\varphi _{x}+\psi )}_{x}\) and \({\overline{\psi }}_{x}\), respectively, by (3.3), (3.4) and (3.5), and making some algebraic manipulations, we get
Here, we replace J in (3.37) to get
where \(\Delta _{r}:=r-d\gamma _{2}/\gamma _{1}\) and \(\Delta _{c}:=c-d\gamma _{1}/\gamma _{2}\). Substituting (3.38) in (3.33), we obtain
where \(\xi :=1-\frac{\rho _{1}}{\delta \kappa }\Gamma \) and \(\chi _{1}\) is given in (3.30). Taking the absolute value and using Young and Poincaré inequalities, we get
From \(\varepsilon :=\kappa |\xi |/2C\), we have
By using Lemma 3.3, we obtain
Choosing, \(\varepsilon _{1}:=\kappa |\xi |\), (3.31) follows immediately. This completes the proof of the lemma. \(\square \)
Lemma 3.5
Let \((\varphi ,v,\psi ,\phi , \theta , q, P, \eta )\) be a solution of system (1.7)–(1.14). Then, there exists a positive constant C such that
Proof
Multiplying Eqs. (3.24) and (3.20) by \({\overline{P}}\) and \({\overline{\theta }}\), respectively, we get
and
where \({\mathcal {R}}\in {\mathfrak {R}}\). Adding Eqs. (3.40) and (3.41), we have
Substituting \({\overline{\theta }}_{x}\) and \({\overline{P}}_{x}\), respectively, by (3.8) and (3.9), we obtain
Taking the real part and using Eq. (3.6), we get
Adding and subtracting \(v_x\) and using Poincaré inequality yields
Substituting \({\overline{\phi }}_{x}\) and \(\overline{(\varphi _{x}+\psi )}_{x}\), respectively, by (3.5) and (3.4), we get
Combining Eqs. (3.3) and (3.5) we obtain
Substituting (3.44) into (3.43), we have
which leads to
On the other hand, taking the imaginary part we have
By using the Young’s inequality, we obtain
Using \({1+\lambda ^{2}\tau _0^{2}}>{\lambda ^{2}\tau _0^{2}}\) and \({1+\lambda ^{2}\tau _1^{2}}>{\lambda ^{2}\tau _1^{2}}\) yields
Substituting (3.45) in (3.46) and using Lemma 3.3, we obtain (3.39). \(\square \)
Lemma 3.6
Let \((\varphi ,v,\psi ,\phi , \theta , q, P, \eta )\) be a solution of system (1.7)–(1.14). There then exists a positive constant \(\varepsilon \) independent of \(\lambda \) such that
for \(|\lambda |> 1\) large enough.
Proof
Multiplying Eq. (3.4) by \(-i\lambda ^{-1}{\overline{v}}\), we have
where \({\mathcal {R}}\in {\mathfrak {R}}\). From (3.3) we have \({\overline{v}}_{x}=-(i\lambda {\overline{\varphi }}_{x}+{\overline{f}}_{1,x})\) and consequently,
By using the Young’s inequality and Lemmas 3.3 and 3.5, we obtain (3.47). \(\square \)
Theorem 3.1
If \(\chi _{0}=0\), then system (1.7)–(1.14) is exponentially stable if and only if \(\chi _{1}=0\).
Proof
(i) Sufficiency: From Lemmas 3.2–3.6 and estimate (3.12), we get if \(\chi _{0}=0\)
Since \(\chi _{1}=0\) we have
Choosing \(|\lambda |\) large enough and \(\varepsilon \) sufficiently small, we concluded the sufficiency condition.
(ii) Necessity: we show that the semigroup S(t) is not exponentially stable when the stability number \(\chi _{1}\) is different from zero. The proof is based on Theorem 1.1. The strategy consists of verifying that condition (1.20) fails to hold. To this end, let us assume that there exists \(U=(\varphi ,v,\psi ,\phi , \theta , q, P, \eta )\in {\mathcal {H}}\) such that \(||U||_{{\mathcal {H}}}\ne 0\). Without loss of generality we can take \(f_{1}=f_{3}=f_{4}=f_{5}=f_{6}=f_{7}=f_{8}=0\) and choose \(f_{2}(x)=\rho _{1}^{-1}\sin (\beta _{n}x)\) in system (3.3)–(3.10) such that \(F=\big (0, f_{2}, 0, 0, 0, 0, 0, 0\big )\) is limited in \({\mathcal {H}}\). Because of the boundary conditions (1.14), we can suppose that
where \({\widehat{A}}_{i}\) (\(i=1,\ldots ,5\)) are constant and \(\beta _{n}:=n\pi /l\). Therefore, system (3.3)–(3.10) is equivalent to
By using elementary row operation in (3.54) and (3.56), we obtain
where \(\Delta _{r}:=r-d\gamma _{2}/\gamma _{1}\) and \(\Delta _{c}:=c-d\gamma _{1}/\gamma _{2}\). So we have that
Therefore, we can rewrite the above system as
where
Now we choose \(\lambda ^{2}\equiv \lambda ^2_{n}=\frac{\kappa }{\rho _{1}}\beta _{n}^{2}-\frac{\sigma }{\rho _{1}}\) which gives \(p_{1}(\lambda )=\sigma \), where \(\sigma \in {\mathbb {R}}\) is going to be fixed later. The resolution of Eq. (3.65) gives us
where
For this case, we consider the following asymptotic equivalences
Because of this, note that the following asymptotic equivalences hold:
and
Suppose that \(\chi _{0}=0\), \(\gamma _{1}\ne \gamma _{2}\), \(\xi \ne 0\) and \(\chi _{1}\ne 0\). Then, from (3.67), (3.74) and (3.75) we get
where we have used \(\gamma _{1}\ne \gamma _{2}\) and \( \Upsilon _3=\Bigg [\displaystyle \xi -\frac{\rho _{1}}{\kappa }\frac{d}{\delta }\bigg (\frac{\hbar \gamma _{1}}{\tau _{1}\gamma _{2}}+\frac{K\gamma _{2}}{\tau _{0}\gamma _{1}}\bigg )\Bigg ]\big (\gamma _{1}^{2}-\gamma _{2}^{2}\big )\). From \(\displaystyle p_{2}(\lambda ):=-\lambda ^{2}\rho _{2}+\alpha \beta _{n}^{2}+\kappa \) we have
Taking \(-\frac{\kappa \rho _{2}}{\rho _{1}}+\alpha +\frac{\big (\Delta _{r}\gamma _{1}^{2}+\Delta _{c}\gamma _{2}^{2}\big )}{\delta \xi }=\frac{\kappa ^{2}}{\sigma }\) we have
where \(\sigma :=-\rho _{1}\kappa \xi /\chi _{1}\). From (3.66), we have
since \(\xi \ne 0\) and \(\chi _{1}\ne 0\). Therefore,
which implies that \( ||U||_{{\mathcal {H}}} \rightarrow \infty \) as \(\lambda \rightarrow \infty \). \(\square \)
Remark 3.1
One can also show the lack of exponential stability of system (1.7)–(1.14) under the conditions \(\chi _{0}=0\), \(\xi =0\) and \(\gamma _{1}\ne \gamma _{2}\). In fact, if one chooses \(\lambda ^{2}\equiv \lambda _{n}^2=\frac{\kappa }{\rho _{1}}\beta _{n}^{2}\) which gives \(p_{1}(\lambda )=0\). The resolution of Eq. (3.65) gives us
Since \(\lambda ^2=\frac{\kappa }{\rho _{1}}\beta _{n}^{2}\), we consider the following asymptotic equivalences
Consequently, by setting
we get
Taking \(\chi _{0}=0\) and \(\xi =0\), we have
since \(\gamma _{1}\ne \gamma _{2}\). Consequently,
Therefore,
which implies that \( ||U||_{{\mathcal {H}}} \rightarrow \infty \) as \(\lambda \rightarrow \infty \).
4 Polynomial decay
In this section, we will show the solutions to system (1.7)–(1.14) decay to zero polynomially as \(1/\sqrt{t}\) by using the Borichev and Tomilov’s result [7]:
Theorem 4.1
Let us suppose that \(\chi _{0}=0\), \(\xi \ne 0\) and \(\chi _{1}\ne 0\). Then, the semigroup associated with system (1.7)–(1.14) is polynomially stable, i.e.,
In the particular case \(\xi =0\), (4.1) holds as well. Moreover, this rate of decay is optimal.
Proof
Let us suppose that \(\xi \ne 0\) and \(\chi _{1}\ne 0\). It follows from Lemma 3.4 and Eq. (3.3) that
By using the Young’s inequality, Lemma 3.2 and the estimate (3.12), we get
On the other hand, combining Lemmas 3.2, 3.3 and the estimate (3.12) yields
From Lemmas 3.2, 3.3, 3.5 and 3.6 and by the inequalities (4.2), we have
Since \(\int \limits _{0}^{l}|\varphi _{x}|^2\mathrm{d}x\le 2\int \limits _{0}^{l}|\varphi _{x}+\psi |^2\mathrm{d}x+2c_p\int \limits _{0}^{l}|\psi _x|^2\mathrm{d}x\) yields
From (4.3), we obtain
Using the fact that \(C||U||_{{\mathcal {H}}}||F||_{{\mathcal {H}}}\le \lambda ^{2}C_{\varepsilon }||U||_{{\mathcal {H}}}||F||_{{\mathcal {H}}})\), the above estimation becomes
Therefore,
Choosing \(|\lambda |\) large enough and \(\varepsilon \) sufficiently small, we get
which is equivalent to
Then, using Theorem 1.2 (see (1.21)), we obtain
From Theorem 2.1, we conclude that \(0 \in \rho ({\mathcal {A}})\), it follows that \({\mathcal {A}}\) is onto over \({\mathcal {H}}\), then taking \({\mathcal {A}}U_0 = F\), we get
Therefore, the solution decays polynomially.
In case \(\xi =0\), 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 3.1 (ii) or Remark 3.1. Note that in case \(\chi _{0}=0\) with \(\xi =0\) or in case \(\chi _{0}=0\) with \(\xi \ne 0\) and \(\chi _{1}\ne 0\), we have the inequality
for \(|\lambda |\) large enough. If we assume that the rate of decay can be improved from \(1/t^{1/2}\) to \(1/t^{1/(2-\varsigma )}\) for some \(\varsigma >0\), then we will have that
must be bounded. But this is not possible because of the inequality (4.7). The proof is now complete.
\(\square \)
5 Conclusion
(a) Note that when \(\tau _0=\tau _1 = 0\), Cattaneo’s law turns into the Fourier’s law for heat transmission and Fick’s law for diffusion transmission. In that case the number \(\chi _0\) does not exist and consequently \(\Gamma \equiv 0\) and the condition of exponential stability over the new number \(\chi _1=0\) is equivalent to the one over the old stability number \(\chi =0\). That is, we get the same result as proved by Aouadi et al. [3] for a Timoshenko system with thermodiffusion effects in the case \(\tau _0=\tau _1 = 0\).
(b) Table 1 summarizes the different types of decay obtained for the system (1.7)–(1.12) for different numbers of stability.
References
Almeida Júnior, D.S., Santos, M.L., Muñoz Rivera, J.E.: Stability to 1-D thermoelastic Timoshenko beam acting on shear force. Z. Angew. Math. Phys. 65, 1233–1249 (2014)
Aouadi, M., Copetti, M.I.M.: Exponential stability and numerical analysis of a thermoelastic diffusion beam with rotational inertia and second sound. Math. Comput. Simul. 187, 586–613 (2021)
Aouadi, M., Campo, M., Copetti, M.I.M., Fernández, J.R.: Existence, stability and numerical results for a Timoshenko beam with thermodiffusion effects. Z. Angew. Math. Phys. 70, 744–746 (2019)
Aouadi, M., Boulehmi, K.: Asymptotic behavior of non-uniform Timoshenko beam acting on shear force with feedback controller. Z. Angew. Math. Mech. 97, 1579–1599 (2017)
Aouadi, M., Soufyane, A.: Decay of the Timoshenko beam with thermal effect and memory boundary conditions. J. Dyn. Contr. Syst. 19, 33–46 (2013)
Aouadi, M.: Theory of generalized micropolar thermoelastic diffusion under Lord–Shulman model. J. Therm. Stress. 32, 923–942 (2009)
Borichev, A., Tomilov, Y.: Optimal polynomial decay of functions and operator semigroups. Math. Ann. 347, 455–478 (2010)
Cattaneo, C.: Sulla conduzione del calore. Atti del Seminario Matematico e Fisico dell’ Universitá di Modena e Reggio Emilia 3, 83–101 (1948)
Dell’Oro, F., Pata, V.: On the stability of Timoshenko systems with Gurtin–Pipkin thermal law. J. Differ. Equ. 257, 523–548 (2014)
Fernández Sare, H.D.: Exponential decay of Timoshenko systems with indefinite memory dissipation. Adv. Differ. Equ. 13, 733–752 (2009)
Fernández Sare, H.D., Muñoz Rivera, J.E.: Stability of Timoshenko systems with past history. J. Math. Anal. Appl. 339, 482–502 (2008)
Fernández Sare, H.D., Racke, R.: On the stability of damped Timoshenko systems: Cattaneo versus Fourier law. Arch. Ration. Mech. Anal. 194, 221–251 (2009)
Huang, F.: Characteristic conditions for exponential stability of linear dynamical systems in Hilbert spaces. Ann. Differ. Equ. 1, 43–56 (1985)
Khodja, F.A., Benabdallah, A., Muñoz Rivera, J.E., Racke, R.: Energy decay for Timoshenko systems of memory type. J. Differ. Equ. 194, 82–115 (2003)
Messaoudi, S.A., Mustafa, M.I.: A stability result in a memory-type Timoshenko system. Dyn. Syst. Appl. 18, 457–468 (2009)
Messaoudi, S.A., Pokojovy, M., Said-Houari, B.: Nonlinear Damped Timoshenko Systems with Second Sound-Global Existence and Exponential Stability, Konstanzer Schr. Math. Inform., vol. 246. Universität Konstanz (2008)
Muñoz Rivera, J.E., Racke, R.: Global stability for damped Timoshenko systems. Discrete Contin. Dyn. Syst. 9, 1625–1639 (2003)
Pazy, A.: Semigroups of Linear Operators and Applications to Partial Differential Equations. Springer, New York (1983)
Prüss, J.: On the spectrum of \(C_0-\)semigroups. Trans. Am. Math. Soc. 284, 847–857 (1984)
Racke, R.: Thermoelasticity with second sound-exponential stability in linear and nonlinear 1-d. Math. Methods Appl. Sci. 25, 409–441 (2002)
Raposo, C.A., Ferreira, J., Santos, M.L., Castro, N.N.O.: Exponential stability for Timoshenko system with two weak dampings. Appl. Math. Lett. 18, 535–541 (2005)
Santos, M.L., Almeida Júnior, D.S., Muñoz Rivera, J.E.: The stability number of the Timoshenko system with second sound. J. Differ. Equ. 253, 2715–2733 (2012)
Acknowledgements
The authors would like to thank the Editor Prof. Laurent Chupin and the anonymous reviewer for their critical reviews, helpful and constructive comments that greatly contributed to improving the final version of the paper.
Funding
A. J. A. Ramos thanks the CNPq for financial support through the Grant 310729/2019-0.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Aouadi, M., Ramos, A. & Castejón, A. Stability conditions for thermodiffusion Timoshenko system with second sound. Z. Angew. Math. Phys. 72, 151 (2021). https://doi.org/10.1007/s00033-021-01580-0
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s00033-021-01580-0