Abstract
In this work, we analyze a truncated version for the Timoshenko beam model with thermal and mass diffusion effects derived by Aouadi et al. (Z Angew Math Phys 70:117, 2019). In particular, we study some issues related to the second spectrum of frequency according to a procedure due to Elishakoff (in: Advances in mathematical modelling and experimental methods for materials and structures, solid mechanics and its applications, Springer, Berlin, 2010). In Aouadi et al. (2019), the lack of exponential stability for the classical Timoshenko beam with thermodiffusion effects without assuming the nonphysical condition of equal wave speeds has be proved. By using the classical Faedo–Galerkin method combined with the a priori estimates, we prove the existence and uniqueness of a global solution of the truncated version of this problem. Then we prove that this solution is exponentially stable without assuming the condition of equal wave speeds.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Recently, Aouadi et al. [5] introduced a new Timoshenko beam model with thermal and mass diffusion effects given by
where \(\varphi \) is the transverse displacement, \(\psi \) is the rotation of the neutral axis due to bending, \(\theta \) is the temperature, and P is the chemical potential. The constants \(\rho _{1}\), \(\rho _{2}\), \(\kappa \), \(\alpha \), \(\gamma _{1}\), \(\gamma _{2}\), c, r, d, \(\hbar \), and K are physical positive parameters. They showed, without assuming the well-known equal wave speeds condition \(\chi :=\kappa /\rho _{1}-b/\rho _{2}=0\), the lack of exponential stability for the problem. Based on [5] and the recent studies due to Almeida Júnior et al. [1,2,3,4], we consider the truncated version given by
with the initial conditions
and boundary conditions of Dirichlet-Neumann-type
The truncated version (1.5)–(1.10) is obtained by following the procedure of Elishakoff [7] which involves replacing the term \(\psi _{tt}\) in (1.2) by \(-\varphi _{xtt}\) based on d’Alembert’s principle for dynamic equilibrium. This eliminates the second spectrum of frequency and its damaging consequences for wave propagation speed (see the first results in [1] and also in [9]). Therefore, the goal of this work is to prove the well-posedness of problem (1.5)–(1.10) and the exponential stability of solutions without assuming the nonphysical condition of equal wave speeds.
In order to derive the dissipative nature of the system (1.5)–(1.10), we define its functional energy of solutions
which preserves its positivity property for
2 Well-posedness
In this section, the existence and uniqueness of weak and strong solutions to (1.5)–(1.10) will be proved. To this end, we will use the Faedo–Galerkin approximations and pass to the limit by using compactness arguments (see also [6]).
We introduce the phase space
and
where
and
In order to state our main result, we begin with a precise definition of a weak solution to (1.5)–(1.10).
Definition 2.1
Given initial data \((\varphi _0,\varphi _1,\varphi _2,\psi _0,\theta _0,P_0)\in {\mathcal {H}}\), a function \(U=(\varphi ,\varphi _t,\varphi _{tt},\psi ,\theta ,P)\in C(0,T;{\mathcal {H}})\) is said to be a weak solution of (1.5)–(1.10) if for almost every \(t\in [0,T]\),
for all \(u,v,\xi ,\zeta \in H_0^1(0,L),\) \(w\in H_*^1(0,L)\), and
Theorem 2.2
Suppose that condition (1.12) holds. Then we have:
(i) If the initial data \((\varphi _{0}, \varphi _{1},\varphi _{2},\psi _0,\theta _0,P_0) \in {\mathcal {H}}\), then problem (1.5)–(1.10) has a weak solution satisfying
(ii) If the initial data \((\varphi _{0}, \varphi _{1},\varphi _{2}, \psi _0,\theta _0,P_0) \in {\mathcal {H}}_1\), then problem (1.5)–(1.10) has a unique stronger weak solution satisfying
(iii) In both cases, the solution \((\varphi ,\varphi _{t},\varphi _{tt},\psi ,\theta , P)\) depends continuously on the initial data in \({\mathcal {H}}.\) In particular, problem (1.5)–(1.10) has a unique weak solution.
Proof
The proof is given by the Faedo–Galerkin method. We only briefly present the main (six) steps.
Step 1 – Approximate problem. Let us consider the initial data \((\varphi _{0}, \varphi _{1},\varphi _{2},\) \(\psi _0,\theta _0,P_0) \in {\mathcal {H}} \). Let \(\{\omega _j\}^\infty _{j=1}\) and \(\{\mu _j\}^\infty _{j=1}\) be orthogonal bases for \(H^2(0,L)\cap H^1_0(0,L)\) and \(H^2_*(0,L)\), respectively, which are both orthonormal in \(L^2(0,L)\). Now we denote the finite-dimensional subspaces, for any integer \(n\in {\mathbb {N}}\), by
We will find an approximate solution of the form
to the following approximate problem
for all \(u,v,\xi ,\zeta \in H_n\), \(w\in V_n\) with initial conditions
satisfying
From the application of the standard ODE theory, we can obtain a local solution \(\big (\varphi ^n(t),\varphi ^n_t(t),\varphi ^n_{tt}(t),\psi ^{n}(t),\theta ^n(t)\), \(P^n(t)\big )\) on the maximal interval \([0,t_n)\) with \(0<t_n\le T\) for every \(n\in {\mathbb {N}}\).
Step 2 – A priori estimate. Replacing u by \(\varphi ^n_{t}\) in (2.8), v by \(\varphi _{tt}^n\) in (2.9), w by \(\psi ^n_t\) in (2.10), \(\xi \) by \(\theta ^n\) in (2.11), and \(\zeta \) by \(P^n\), we obtain
where
Then integrating (2.14) from 0 to \(t<t_n\), we obtain from our choice of initial data that for all \(t\in [0,T]\) and for every \(n\in {\mathbb {N}}\),
where \(C_1\) is a positive constant depending on the initial data. Thus, approximate solutions are defined on the whole range [0, T].
Step 3 – Passing to the limit. From (2.15) and definition of \(E^n(t)\), we deduce that
Then we can extract a subsequence of \(\{\varphi ^n\}\), \(\{\psi ^n\}\), \(\{\theta ^n\}\), and \(\{P^n\}\) still denoted by \(\{\varphi ^n\}\), \(\{\psi ^n\}\), \(\{\theta ^n\}\), and \(\{P^n\}\), such that
Therefore the above limits allow us to pass to the limit in the approximate problem (2.8)–(2.12) to get a weak solution satisfying
Step 4 – Initial data. By using Aubin-Lions lemma, see [8], we arrive at
Consequently
Now, we multiply (2.9) by a test function
and integrate the result over [0, T] to obtain
for all \(v\in H_0^1(0,L)\). Taking the limit \(n\rightarrow \infty \), we obtain
for all \(v\in H_0^1(0,L)\). On the other hand, multiplying (2.2) by \(\eta \) and integrating the result over [0, T], we obtain
for all \(v\in H_0^1(0,L)\). Combining (2.18) and (2.19), we conclude that \(\varphi _{tt}(0)=\varphi _2\). Analogously, we obtain
Step 5—Stronger solutions. Suppose that the initial data in the approximate problem (2.8)–(2.12) satisfies \((\varphi _{0}, \varphi _{1},\varphi _{2}, \psi _0,\theta _0,P_0) \in {\mathcal {H}}_1\) and
Replacing u by \(-\varphi ^n_{xxt}\) in (2.8), v by \(-\varphi ^n_{xxtt}\) in (2.9), w by \(-\psi _{xxt}^n\) in (2.10), \(\xi \) by \(-\theta _{xx}^n\) in (2.11), and \(\zeta \) by \(-P_{xx}\) in (2.12), we see that
where
Then from (2.21), we obtain that for all \(t\in [0,T]\), \(n\in {\mathbb {N}}\),
where \(C_2\) is a positive constant independent of t and n but depending on the initial data. From (2.22), we deduce that
This implies that
From the above limits, we conclude that \((\varphi ,\varphi _t,\varphi _{tt},\psi ,\theta ,P)\) is a stronger weak solution satisfying
Step 6 – Continuous dependence. Firstly, we consider the case of stronger solutions. Let \(U(t)=(\varphi ,\varphi _t,\varphi _{tt},\psi , \theta , P)\) and \(V(t)=({\tilde{\varphi }},{\tilde{\varphi }}_t,{\tilde{\varphi }}_{tt},{\tilde{\psi }},{\tilde{\theta }}, {\tilde{P}})\) be the stronger weak solutions of the problem (1.5)–(1.8) corresponding to the initial data \(U(0)=(\varphi _0,\varphi _1,\varphi _2,\psi _0,\) \(\theta _0,P_0)\), \(V(0)=({\tilde{\varphi }}_0,\tilde{\varphi _1},\tilde{\varphi _2},{\tilde{\psi }}_0,{\tilde{\theta }}_{0},{\tilde{P}}_{0})\in {\mathcal {H}}_1\), respectively. Then \((\Phi ,\Phi _t,\Phi _{tt},\Psi ,\Theta , \Upsilon )=U(t)-V(t)\) satisfies the following equations:
with initial data \((\Phi (0),\Phi _t(0),\Phi _{tt}(0), \Psi (0), \Theta (0),\Upsilon (0))=U(0)-V(0)\).
We multiply (2.23) by \(\Phi _t\), (2.24) by \(\Psi _t\), (2.25) by \(\Theta \), and (2.26) by \(\Upsilon \) and integrate the result over (0, L) to derive
where \({\widehat{E}}(t)\) is the energy corresponding to \(U(t)-V(t)\) defined by
Integrating (2.27) over (0, t), we get that there exists a constant \(C_T>0\) such that for any \(t\in [0,T]\),
which implies the continuous dependence of stronger weak solutions on the initial data. Then we know that the stronger weak solution of problem (1.5)–(1.10) is unique. The continuous dependence and uniqueness for weak solutions can be proved by using density arguments (weak solutions are limits of stronger weak solutions). Combining the above analysis, we complete the proof of Theorem 2.2. \(\square \)
3 Exponential decay
In this section, we use the energy method to prove that E(t), the energy of system (1.5)–(1.10) given by (1.11), decays exponentially. For this, we assume that \((\varphi , \varphi _{t},\varphi _{tt},\psi ,\theta , P)\) is a solution of the system (1.5)–(1.10) with the regularity stated in Theorem 2.2 and we suppose that condition (1.12) holds true.
Theorem 3.1
Suppose that the hypotheses of Theorem 2.2 hold. Then there exist two positive constants M and \(\eta \) such that
The proof of Theorem 3.1 will be established through several lemmas. We have the first lemma regarding the dissipative nature of the energy.
Lemma 3.2
The energy E(t) of the system (1.5)–(1.10) satisfies the energy dissipation law given by
Proof
Multiplying Eq. (1.5) by \(\varphi _t\), (1.6) by \(\psi _t\), (1.7) by \(\theta \), (1.8) by P, and integrating the result over [0, L], we obtain the desired result. \(\square \)
We set
Lemma 3.3
Suppose that the hypotheses of Theorem 2.2 hold. Then we have
where \(c_p>0\) is the Poincaré constant.
Proof
Multiplying Eq. (1.5) by \(\varphi \), integrating over [0, L] using integration by parts, and taking into account the boundary conditions (1.10), we have
Taking into account the identity \(\varphi _{tt}\varphi =\displaystyle \frac{\partial }{\partial t}(\varphi _{t}\varphi )-|\varphi _{t}|^2\) and Young’s inequality, we arrive at
Moreover, we consider the inequality given by \( \mathop \int \nolimits _{0}^{L}|\varphi _{x}|^2dx\le 2\mathop \int \nolimits _{0}^{L}|\varphi _{x}+\psi |^2dx+2c_{p}\mathop \int \nolimits _{0}^{L}|\psi _x|^2dx, \) we complete the proof. \(\square \)
Lemma 3.4
Suppose that the hypotheses of Theorem 2.2 hold. Then we have
where
Proof
Multiplying Eq. (1.6) by \(\psi \) and integrating by parts, we obtain
It follows from Eq. (1.5) that \(\psi _{x}=\displaystyle \frac{\rho _{1}}{\kappa }\varphi _{tt}-\varphi _{xx}\). Then, substituting \(\psi _x\) into (3.9), we obtain
and using Young’s and Poincare’s inequalities, we arrive at the desired result.
\(\square \)
Let us introduce one more functional which is given by
Lemma 3.5
Suppose that the hypotheses of Theorem 2.2 the hold. Then for all \(\varepsilon >0\), there are constants \(C_i>0\) \((i=1,2,3)\) such that
Proof
Multiplying Eq. (1.6) by \((\varphi _{x}+\psi )\), integrating over [0, L], and using integration by parts, we have
Now, using Young’s inequality, it follows that
On the other hand, it follows from Eq. (1.5) that \((\varphi _{x}+\psi )_{x}=\displaystyle \frac{\rho _{1}}{\kappa }\varphi _{tt}\) and then we can rewrite the above inequality as
Moreover, we consider the identity given by \(\varphi _{xtt}(\varphi _{x}+\psi )=\displaystyle \frac{\partial }{\partial t}\big [\varphi _{xt}(\varphi _{x}+\psi )\big ]-\varphi _{xt}(\varphi _{x}+\psi )_{t}\) from where we obtain
where \(C_{1}\displaystyle :=\alpha \big (\rho _{1}/\kappa +\rho _{2}/\alpha \big )\). On the other hand, multiplying Eq. (1.7) by \(C_{1}\gamma _{1}^{-1}\varphi _{t}\), we have
Taking into account the identities \(\displaystyle \theta _{t}\varphi _{t}=\frac{\partial }{\partial t}(\theta \varphi _{t})-\theta \varphi _{tt}\), \(\displaystyle P_{t}\varphi _{t}=\frac{\partial }{\partial t}(P\varphi _{t})-P\varphi _{tt}\) and \(\displaystyle \psi _{xt}\varphi _{t}=\frac{\partial }{\partial t}(\psi _{x}\varphi _{t})-\psi _{x}\varphi _{tt}\), we have
Adding (3.14) and (3.15), we obtain
and using again Young’s inequality, we arrive at the desired result. \(\square \)
Lemma 3.6
Suppose that the hypotheses of Theorem 2.2 hold. Then there are constants \(\lambda , \, N_{0}, \, \delta >0\) such that
Proof
First, we define the constant \(\lambda _{0}:=\max \big \{d^2c_p/Kr, \ rc_{p} / \hbar \big \}\). Then, choosing \(\lambda >\lambda _{0}\) and using (3.2), we have
Then, using the Poincaré inequality, we get
Since \(\frac{d^{2}\lambda K}{2rcc_{p}}\mathop \int \nolimits _{0}^{L}|\theta |^2dx>0\) and \(-\frac{\lambda \hbar }{2c_{p}}\mathop \int \nolimits _{0}^{L}|P|^2dx<0\), we get the following estimate
Next, we add the term \(\frac{d^2}{2r}\mathop \int \nolimits _{0}^{L}|\theta |^2dx>0\)
Since \(\lambda >\lambda _{0}=\max \big \{d^2c_p/Kr, \ rc_{p} / \hbar \big \}\), we have \( \zeta _{1}:=\frac{\lambda Kr}{d^2c_{p}}-1>0 \quad \text{ and } \quad \zeta _{2}:=\frac{\lambda \hbar }{rc_{p}}-1>0. \) Using Young’s inequality, we have
Replacing (3.19) in (3.18), we have
Therefore,
where \(\displaystyle \delta :=\min \{1, \ \lambda K/cc_{p}, \,\zeta _{1}, \, \zeta _{2}\}\). \(\square \)
Now we are ready to prove the main result of this paper.
Proof of Theorem 3.1
We consider the following Lyapunov functional defined by
where \(\lambda >\lambda _{0}:=\max \big \{d^2c_p/Kr, \ rc_{p} / \hbar \big \}\) and \(N_{i},\ i=0,2,3\), are positive constants to be fixed later. Moreover, the coefficients \( N_0\) and \(\lambda \) will be chosen large enough such that \({\mathcal {L}}(t)\) and E(t) are equivalent. Indeed, from Young’s and Poincaré’s inequalities, we infer that there exists a constant \(0<c<N_0+\lambda \) such that
Consequently,
Substituting the results of Lemmas 3.3, 3.4, and 3.5 in the time derivative of \({\mathcal {L}}(t)\), we obtain after selecting \( \varepsilon :=\frac{\gamma _{1}\rho _{1}\rho _{2}}{2\kappa (c+d) C_{1}N_{3}} \),
By choosing \(N_{2}>\max \Big \{1/2, \, 2c_{p}\kappa /\alpha \Big \}\), \(N_{3}>\max \Big \{2N_{2}, \, 3+3\kappa c_{p}N_{2}/\alpha \Big \}\) and \(N_{0}\) large enough such that \( N_{0}>\max \Big \{\frac{3\gamma _{1}^{2}c_{p}}{2K\alpha }N_{2}+\frac{C_{2}}{K}N_{3}, \, \frac{3\gamma _{2}^{2}c_{p}}{2\hbar \alpha }N_{2}+\frac{C_{3}}{\hbar }N_{3}\Big \}, \) one can obtain that \(\xi _{1}:=2N_{2}-1>0\), \(\xi _{2}:=N_{3}-2N_{2}>0\), \(\xi _{3}:=N_{2}-\frac{2c_{p}\kappa }{\alpha }>0\), \(\xi _{4}:=N_{3}-3-\frac{3\kappa c_{p}}{\alpha }N_{2}>0\), \(\xi _{5}:=KN_{1}-\frac{3\gamma _{1}^{2}c_{p}}{2\alpha }N_{2}-C_{2}N_{3}>0\), \(\xi _{6}:=\hbar N_{1}-\frac{3\gamma _{2}^{2}c_{p}}{2\alpha }N_{2}-C_{3}N_{3}>0\).
Now, we can conclude that there exists a positive constant \(\omega :=\min \Big \{2,\) \( \, \xi _{1}, \, \xi _{2}, \, \xi _{3}, \, \xi _{4}, \, \delta \Big \}>0\) such that
Combining (3.21) with (3.22), we obtain
From this and using (3.21) again, we have after integrating over (0, t),
which gives (3.1) with \(M:=\frac{N_0+\lambda +c}{N_0+\lambda -c}\) and \(\eta :=\frac{\beta }{N_0+\lambda +c}\). This completes the proof.\(\square \)
References
Almeida Júnior, D.S., Ramos, A.J.A.: On the nature of dissipative Timoshenko systems at light of the second spectrum. Z. Angew. Math. Phys. 68(6), Paper No. 145, 31 pp. (2017)
Almeida Júnior, D.S., Ramos, A.J.A., Santos, M.L., Miranda, L.G.R.: Asymptotic behavior of weakly dissipative Bresse–Timoshenko system on influence of the second spectrum of frequency. Z. Angew. Math. Mech. 98(8), 1320–1333 (2018)
Almeida Júnior, D.S., Elishakoff, I., Ramos, A.J.A., Miranda, L.G.R.: The hypothesis of equal wave speeds for stabilization of Bresse–Timoshenko system is not necessary anymore: the time delay cases. IMA J. Appl. Math. 84(4), 763–796 (2019)
Almeida Júnior, D.S., Ramos, A.J.A., Soufyane, A., Cardoso, M.L., Santos, M.L.: Issues related to the second spectrum, Ostrogradsky’s energy and the stabilization of Timoshenko–Ehrenfest-type systems. Acta Mech. 231(9), 3565–3581 (2020)
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, 117 (2019)
Choucha, A., Ouchenane, D., Zennir, K., Feng, B.: Global well-posedness and exponential stability results of a class of Bresse–Timoshenko-type systems with distributed. Math. Methods Appl. Sci. 1–26 (2020)
Elishakoff, I.: An equation both more consistent and simpler than the Bresse–Timoshenko equation. In: Gilat, R., Banks-Sills, L. (eds.) Advances in Mathematical Modeling and Experimental Methods for Materials and Structures. Solid Mechanics and Its Applications, vol. 168. Springer, Dordrecht (2009)
Lions, J.L.: Quelques Méthodes de Résolution des Problèmes aux Limites Non Linéaires. Dunod Gauthier-Villars, Paris (1969)
Ramos, A.J.A., Almeida Júnior, D.S., Miranda, L.G.R.: An inverse inequality for a Bresse–Timoshenko system without second spectrum of frequency. Arch. Math. (Basel) 114(6), 709–719 (2020)
Acknowledgements
The authors would like to thank the referee for his critical review and valuable comments which allowed to improve the paper.
Funding
A. J. A. Ramos thanks the CNPq for support through Grant 310729/2019-0. M. L. Araújo is grateful to Fundação de Amparo á Pesquisa do Estado do Pará (FAPESPA) for financial support.
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
Ramos, A., Aouadi, M., Almeida Júnior, D.S. et al. A new stabilization scenario for Timoshenko systems with thermo-diffusion effects in second spectrum perspective. Arch. Math. 116, 203–219 (2021). https://doi.org/10.1007/s00013-020-01526-4
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00013-020-01526-4