In this paper, we show that the high frequency modes of a thin clamped plate and the associated eigenfunctions converge, as the thickness of the plate goes to zero, to the eigenvalues and the eigenfunctions of a two-dimensional eigenvalue problem associated to the stretching displacements of the plate.
Citation: Nabil Kerdid. Asymptotic analysis of high frequency modes for thin elastic plates[J]. AIMS Mathematics, 2023, 8(8): 18618-18630. doi: 10.3934/math.2023948
[1] | Nabil Kerdid . Asymptotic analysis of stretching modes for a folded plate. AIMS Mathematics, 2023, 8(10): 23974-23988. doi: 10.3934/math.20231222 |
[2] | Maha M. Helmi, Ali M. Mubaraki, Rahmatullah Ibrahim Nuruddeen . Analysis of horizontally polarized shear waves on a highly inhomogeneous loaded bi-material plate. AIMS Mathematics, 2023, 8(1): 2117-2136. doi: 10.3934/math.2023109 |
[3] | Jianan Qiao, Guolin Hou, Jincun Liu . Analytical solutions for the model of moderately thick plates by symplectic elasticity approach. AIMS Mathematics, 2023, 8(9): 20731-20754. doi: 10.3934/math.20231057 |
[4] | Mohammed Alkinidri, Rab Nawaz, Hani Alahmadi . Analytical and numerical investigation of beam-spring systems with varying stiffness: a comparison of consistent and lumped mass matrices considerations. AIMS Mathematics, 2024, 9(8): 20887-20904. doi: 10.3934/math.20241016 |
[5] | H. Thameem Basha, R. Sivaraj, A. Subramanyam Reddy, Ali J. Chamkha, H. M. Baskonus . A numerical study of the ferromagnetic flow of Carreau nanofluid over a wedge, plate and stagnation point with a magnetic dipole. AIMS Mathematics, 2020, 5(5): 4197-4219. doi: 10.3934/math.2020268 |
[6] | Maxime Krier, Julia Orlik . Solvability of a fluid-structure interaction problem with semigroup theory. AIMS Mathematics, 2023, 8(12): 29490-29516. doi: 10.3934/math.20231510 |
[7] | Yu Yuan, Chen Chen . Fault detection of rolling bearing based on principal component analysis and empirical mode decomposition. AIMS Mathematics, 2020, 5(6): 5916-5938. doi: 10.3934/math.2020379 |
[8] | Wenhui Feng, Xingfa Zhang, Yanshan Chen, Zefang Song . Linear regression estimation using intraday high frequency data. AIMS Mathematics, 2023, 8(6): 13123-13133. doi: 10.3934/math.2023662 |
[9] | Ahmed SM Alzaidi, Ali M Mubaraki, Rahmatullah Ibrahim Nuruddeen . Effect of fractional temporal variation on the vibration of waves on elastic substrates with spatial non-homogeneity. AIMS Mathematics, 2022, 7(8): 13746-13762. doi: 10.3934/math.2022757 |
[10] | Peipei Wang, Yanting Wang, Fei Wang . Indirect stability of a 2D wave-plate coupling system with memory viscoelastic damping. AIMS Mathematics, 2024, 9(7): 19718-19736. doi: 10.3934/math.2024962 |
In this paper, we show that the high frequency modes of a thin clamped plate and the associated eigenfunctions converge, as the thickness of the plate goes to zero, to the eigenvalues and the eigenfunctions of a two-dimensional eigenvalue problem associated to the stretching displacements of the plate.
The purpose of this article is to study the asymptotic behavior of the high frequency modes of thin plates when the thickness of the plates goes to zero. The asymptotic methods were used to study a large variety of problems in thin elastic structures. Let us refer to Ciarlet and Destunder [1] for the justification of the two-dimensional linear plate model, Ciarlet et al. [2] for the junctions between three-dimensional and two-dimensional linear elastic structures, Le Dret [3,4] for modeling of a folded plate and [5] for modeling of the junction between two rods, and Trabucho and Viaño [6] for asymptotic analysis of linearized elastic beams.
The problem of modeling the vibrations of thin elastic structures using a rigourous asymptotic technique was first done by Ciarlet and Kessavan [7] in the case of a clamped plate. The techniques introduced in this article were adapted and used to study different spectral problems: Le Dret [8] for the vibrations of a folded plate, Bourquin and Ciarlet [9] and Lods [10] for a plate inserted in a three-dimensional body, Kerdid [11,12] for a single rod and junction between two rods. All these works are concerned with the convergence of low frequency modes of the three-dimensional linear elasticity, as the thickness of the body tends to zero. The limit problems obtained are the classical spectral problems associated with the flexural displacement of the structure. However, these techniques fail to obtain the limit problem for higher frequency modes.
We also refer to some interesting works related to this article, which deal with the asymptotic analysis of the eigenvalue problem in different thin linear elastic structures: Jumbo and Rodriguez Mulet [13] and Jumbo et al. [14] for thin elastic rod with non-uniform cross-section, Serpilli and Lenci [15] for laminated beams, Tamba˜ca [16] for curved rods and Qaudiello et al. [17] for a thin T-like shaped structure.
A nonstandard technique has been proposed in Irago et al. [18,19] to study the behavior of the high frequency modes and their associated eigenfunctions in thin elastic rods. The limit problem obtained is the coupled one-dimensional spectral problem giving the classical equations for torsion and stretching vibrations.
In this work, we combine the techniques of [7,19] to study the asymptotic behavior of high frequency modes in a thin clamped plate when the thickness of the plate goes to zero. Indeed, we will construct a suitable families of index {ℓmε}ε>0, which varies with ℓ and ε, and for which the high frequency eigenvalues ηℓmε converge when the thickness of the plate approaches zero towards the eigenvalues ηm of a two-dimensional spectral problem. The limit problem is identified to be the standard eigenvalue problem associated with the classical equations for stretching vibrations of the plate. The limit eigenfunctions are determined by the couples (ζm1,ζm2) of functions of the longitudinal variables of the plate that are the unique solution of the limit problem and correspond to the stretching displacements of the plate.
Let ω be an open bounded set of R2 and γ=∂ω its boundary which is assumed to be sufficiently smooth. Given ε≥0 we define
Ωε=ω×(−ε,ε),Γε=γ×(−ε,ε). | (2.1) |
Ωε is assumed to be the reference configuration of the plate under consideration. The plate is clamped on its boundary Γε. The material that constitute the plate is assumed to be homogeneous and isotropic with Young's modulus E and Poisson's ratio ν, all independent of ε.
We will also use the Lame's coefficients λ and μ related to E and ν by the formulas:
λ=νE(1+ν)(1−2ν),μ=E2(1+ν). | (2.2) |
In the sequel, we shall use the repeated index convention, the Greek indices take their values in the set {1,2} and the Latin indices take their values in the set {1,2,3}.
The classical eigenvalue problem for the plate under consideration consists in finding pairs (ηε,uε) satisfying:
{−∂jσεij=ηεuεi in Ωε,σεij(uε)=λeεpp(uε)δij+2μeεij(uε) in Ωε,uε=0 on Γε,σεnε=0 on ∂Ωε∖Γε, | (2.3) |
where σε(uε) is the stress tensor, nε is the outer unit normal vector to ∂Ωε, and eε(uε) is the linearized strain tensor corresponding to the displacement uε:
eεij(uε)=12(∂uεj∂xεi+∂uεi∂xεj). | (2.4) |
In order to put the above problem in variational form, we introduce the space
Vε={vε=(vεi)∈[H1(Ωε)]3,vε=0 on Γε}. | (2.5) |
So, problem (2.3) is equivalent to finding (ηε,uε)∈R×Vε satisfying
∫Ωεσεij(uε)eεij(vε)dxε=ηε∫Ωεuεivεidxε∀vε∈Vε. | (2.6) |
Thanks to Korn inequality, the bilinear form
(uε,vε)∈Vε×Vε⟼∫Ωεσεij(uε)eεij(vε)dxε | (2.7) |
is Vε-elliptic. From spectral theory, it is known that problem (2.6) has a sequence of eigenvalues (ηεm)m≥1 satisfying
0<ηε1≤ηε2≤ηε3≤...≤ηεm≤... | (2.8) |
with
limm⟶∞ηεm=+∞, | (2.9) |
associated with a family of eigenfunctions (uε,m)m≥1, that is
∫Ωεσεij(uε,m)eεij(vε)dxε=ηεm∫Ωεuε,mivεidxε∀vε∈Vε, | (2.10) |
which can be orthonormalized as
∫Ωεuε,miuε,nidxε=δmn∀m,n≥1, | (2.11) |
and which make a basis in both Hilbert spaces Vε and [L2(Ωε)]3.
In order to define a problem equivalent to problem (2.6) but posed over a domain which does not depend on ε, we let
Ω=ω×(−1,1),Γ=γ×(−1,1), | (3.1) |
and
V=H1Γ(Ω;R3). | (3.2) |
We introduce the following mapping:
ϕε:Ω→Ωε,x↦xε=(x1,x2,εx3), |
and the scaling functions v(ε)∈V defined as:
vα(ε)(x)=ε−1vεα(xε);v3(ε)(x)=vε3(xε). | (3.3) |
Due to scaling, we have
{eεαβ(vε)=eαβ(v(ε)),eεα3(vε)=ε−1eα3(v(ε)),eε33(vε)=ε−2e33(v(ε)). | (3.4) |
Substituting (3.4) in (2.6), we obtain the following scaled variational formulation:
Find (ηm(ε),um(ε))∈R×V such that for, all v∈V
∫Ωbε(um(ε),v)dx=ηm(ε){∫Ωumα(ε)vαdx+ε−2∫Ωum3(ε)v3dx}, | (3.5) |
where
ηm(ε)=ηεm, | (3.6) |
and
bε(u,v)=2μeαβ(u)eαβ(v)+λeαα(u)eββ(v)+ε−2[4μeα3(u)eα3(v)+λ(eαα(u)e33(v)+e33(u)eαα(v))]+ε−4(λ+2μ)e33(u)e33(v), | (3.7) |
with the normalization condition
∫Ωumα(ε)unα(ε)dx+ε−2∫Ωum3(ε)un3(ε)dx=δmn∀m,n≥1. | (3.8) |
We define the space of Kirchhoff-Love on Ω as
VKL={v∈V,ei3(v)=0}. | (3.9) |
Elements of this space are characterized by
{vα(0)(x)=ζα(x1,x2)−x3∂αζ3(x1,x2),v3(0)(x)=ζ3(x1,x2), | (3.10) |
where ζα∈H1(ω) and ζ3∈H2(ω).
A first convergence analysis of the low frequencies of the three-dimensional linearized elasticity system in a thin plate, when the thickness of the plate approaches zero, was done in [7]. It has been shown that the standard biharmonic two-dimensional eigenvalue problem associated with the flexural displacements of the plate can be derived mathematically from the standard three-dimensional eigenvalue problem of linear elasticity through a rigourous convergence analysis as the thickness of the plate tends to zero.
More precisely, it has been proven that for each integer m≥1,
λm(ε)=ε−2ηm(ε)→λm(0) | (4.1) |
and
um(ε)→um(0)strongly inV | (4.2) |
with
um(0)=(−x3∂1um3(x1,x2),−x3∂2um3(x1,x2),um3(x1,x2)), |
where (λm(0),um3)∈R×H20(ω) are eigensolutions of the limit spectral problem:
Find (λ,u)∈R×H20(ω) such that, for all v∈H20(ω)
E3(1−ν2)∫ωΔuΔvdx1dx2=λ∫ωuvdx1dx2. | (4.3) |
The pairs (λm(0),um3) are solutions of the eigenvalue problem for the biharmonic operator Δ2
E3(1−ν2)Δ2u=λu, | (4.4) |
corresponding to the classical equations for the flexural vibrations.
Our objective in this work is to characterize the limit problem associated to high frequency modes when the thickness of the plate goes to zero. Unfortunately, the techniques used in [7] are not adapted for the asymptotic analysis of higher frequency modes. So, we will be inspired by the idea proposed in [18,19] for the convergence analysis of high frequencies in a thin rod in order to characterize the limit problem for high frequency modes in a thin plate.
Let us start with the following lemma:
Lemma 5.1. There exists an increasing sequence of constants Km>0,m≥1, independent of ε such that
ηm(ε)≤Kmε2. | (5.1) |
Proof. From [7] Lemma 1 we have, for each integer m≥1,ε−2ηm(ε)≤Km where Km is a constant independent of ε, which gives (5.1).
So, if we fix the index m and we make ε tend to zero, all the sequence ηm(ε) goes to zero. indeed, the high frequency modes are concentrated at infinity when ε approaches zero and cannot be obtained using such a passage to the limit. So, the idea in order to characterize this kind of frequencies, consists in associating to each integer m≥1, a family of index {ℓmε}ε>0 that depend on ε and such that
limε→0ℓmε=+∞, | (5.2) |
and
ηℓmε(ε)<Km. | (5.3) |
This family of index can be defined by
ℓmε=max{j∈N∗:ηj(ε)≤Km}. | (5.4) |
It is clear that (5.4) satisfies (5.2) and (5.3).
The family of index {ℓmε} varies with m and ε, and for each ε>0, {ℓmε}m≥1 is an increasing subsequence of positive integers satisfying ℓmε≥m,∀m≥1. It contains the indices of the stretching modes among all the modes {ηm(ε)}m≥1 of the plate.
To illustrate this idea and show the layout of the stretching modes {ηℓmε(ε)} when m and ε vary, let us represent the family {ηℓmε(ε)} in a double-entry table. Consider a decreasing sequence {εn}n≥1 converging to 0, the elements of the sequences {ηm(εn)}m≥1 are arranged in rows while the elements of the sequences {ηm(εn)}n≥1 are arranged in columns.
![]() |
Since for each m≥1 the family {ℓmεn}n≥1 is increasing, that is ℓmεn≥ℓmεn′ for n>n′, the elements of the sequence {ηℓmεn(εn)}n≥1 corresponding to the modes associated to the stretching vibrations of the plate are arranged diagonally. As the stretching vibrations are high frequency modes and are concentrated at infinity when ε approaches zero, they can only be reached through such a family of indices.
The following theorem summarize the results obtained when passing to the limit on these families of sequences and constitute the main result of this paper.
Theorem 5.2. For each integer m≥1, there exists a sequence {ℓmε} such that
ηℓmε(ε)→ηm(0) | (5.5) |
where ηm(0) is an eigenvalue of the limit spectral problem:
Find (η,ζ)∈R×[H10(ω)]2 such that, for all ξ∈[H10(ω)]2,
E1+ν∫ωeαβ(ζ)eαβ(ξ)dx1dx2+Eν1−ν2∫ωeαα(ζ)eββ(ξ)dx1dx2=η(0)∫ωζαξαdx1dx2. | (5.6) |
In addition, there exists a subsequence (still denoted ε) and um(0)∈V such that
uℓmε(ε)⇀um(0)weaklyinV, | (5.7) |
where
umα(0)=ζmα(x1,x2) | (5.8) |
and
um3(0)=0, | (5.9) |
with
(ζm1,ζm2)∈[H10(ω)]2. |
If (ζm1,ζm2)≠(0,0) then it is an eigenfunction associated to ηm(0).
To prove this theorem we combine the techniques in [7,19]. First, we start by establishing an appropriate bound for the eigenfunctions.
Lemma 5.3. For each m≥1, there exists a constant Cm>0 independent of ε, such that
∥uℓmε(ε)∥H1(Ω;R3)≤Cm. | (5.10) |
Proof. Let us define the scaled strain tensors
{κℓmεαβ(ε)=eαβ(uℓmε(ε)),κℓmεα3(ε)=ε−1eα3(uℓmε(ε)),κℓmε33(ε)=ε−2e33(uℓmε(ε)). | (5.11) |
Taking v=uℓmε(ε) in (3.5) and using (3.7) and (3.8), we obtain
2μ‖ |
So, we have
\begin{equation} \|\kappa_{ij}^{{\ell_\varepsilon^m}}({{\varepsilon}})\|_{L^2(\Omega)} \leq C_m, \end{equation} | (5.12) |
and consequently, since 0 < {{\varepsilon}} \leq 1 ,
\begin{eqnarray} && \|e_{\alpha\beta}(u^{{\ell_\varepsilon^m}}({{\varepsilon}}))\|_{L^2(\Omega)} \leq C_m, \\ && \|e_{\alpha3}(u^{{\ell_\varepsilon^m}}({{\varepsilon}}))\|_{L^2(\Omega)} \leq C_m {{\varepsilon}} \leq C_m, \\ && \|e_{33}(u^{{\ell_\varepsilon^m}}({{\varepsilon}}))\|_{L^2(\Omega)} \leq C_m {{\varepsilon}}^2 \leq C_m. \end{eqnarray} | (5.13) |
Therefor, (5.10) is obtained using Korn inequality in H_\Gamma^1(\Omega; {{\mathbb R}}^3) .
Lemma 5.4. For each m\geq 1 , there exists a subsequence (still denoted {{\varepsilon}} ) such that
\begin{equation} \eta_{{\ell_\varepsilon^m}}({{\varepsilon}}) \rightarrow \eta_{m}(0) \end{equation} | (5.14) |
and
\begin{equation} u^{{\ell_\varepsilon^m}}({{\varepsilon}}) \rightharpoonup u^{m}(0) \;\;\;\:\;\; { weakly}\; { in }\; V \end{equation} | (5.15) |
where
\begin{equation} u_\alpha^m(0)(x) = \zeta_\alpha^m(x_1, x_2)-x_3\partial_\alpha \zeta_3^m(x_1, x_2), \\ \end{equation} | (5.16) |
\begin{equation} u_3^m(0)(x) = \zeta_3^m(x_1, x_2), \end{equation} | (5.17) |
with \zeta_\alpha \in H^1_0(\omega) and \zeta_3 \in H^2_0(\omega).
Proof. Convergences (5.14) and (5.15) come from (5.3) and (5.10). Now using (5.13) we have
\begin{equation*} e_{i3}(u^{{\ell_\varepsilon^m}}({{\varepsilon}})) \rightarrow 0 \; \; \; \hbox {strongly}\; \hbox { in}\; L^2(\Omega), \end{equation*} |
and since
\begin{equation*} u_i^{{\ell_\varepsilon^m}}({{\varepsilon}}) \rightharpoonup u^m_i(0) \; \; \hbox { weakly}\; \hbox { in}\; H^1(\Omega), \end{equation*} |
then
\begin{equation*} e_{i3}(u^{{\ell_\varepsilon^m}}({{\varepsilon}})) \rightharpoonup e_{i3}(u^{m}(0)) \; \hbox { weakly}\; \hbox { in} \;L^2(\Omega). \end{equation*} |
Thus,
e_{i3}(u^{m}(0)) = 0. |
Therefore, u^{m}(0) \in V_{KL} and consequently, we deduce (5.16) and (5.17) from (3.10).
Lemma 5.5. For each m\geq 1 , if \eta_m(0)\neq0 then
\begin{equation} \zeta_3^m = 0. \end{equation} | (5.18) |
Proof. Let v = (0, 0, v_3), \; v_3 \in H_0^1(\omega) we have
e_{\alpha\beta}(v) = 0, \; e_{\alpha3}(v) = \frac{1}{2}\partial_\alpha v_3, \hbox{ and} \; e_{33}(v) = 0. |
Substituting in (3.5) and multiplying the equation by {{\varepsilon}}^2 we have
\begin{equation*} {{\varepsilon}}\int_\Omega2\mu\kappa_{\alpha3}^{{\ell_\varepsilon^m}}({{\varepsilon}})\partial_\alpha v_3 dx = \eta_m({{\varepsilon}}) \int_\Omega u_3^{{\ell_\varepsilon^m}}({{\varepsilon}}) v_3 dx. \end{equation*} |
Passing to the limit as {{\varepsilon}} \rightarrow 0 , we obtain
\int_\omega \zeta_3^m v_3 dx_1dx_2 = 0\;\;\;\;\;\; \forall v_3 \in H_0^1(\omega). |
Therefore,
\zeta^m_3 = 0. |
Lemma 5.6. For each m\geq 1 , there exists a subsequence (still denoted {{\varepsilon}} ) such that
\begin{equation} \kappa_{\alpha i}^{{\ell_\varepsilon^m}}({{\varepsilon}}) \rightharpoonup \kappa_{\alpha i}^m(0) \;\;\;\:\;\; {weakly\; in}\; L^2(\Omega), \end{equation} | (5.19) |
with
\begin{equation} \kappa_{\alpha\beta}^{m}(0) = e_{\alpha\beta}(\zeta^m), \end{equation} | (5.20) |
and
\begin{equation} \kappa_{\alpha3}^{m}(0) = 0. \end{equation} | (5.21) |
Proof. Convergence (5.19) comes from (5.12) and from (5.15) we have
\begin{equation} \kappa_{\alpha\beta}^{{\ell_\varepsilon^m}}({{\varepsilon}}) \rightharpoonup e_{\alpha\beta}(u^{m}(0)) \;\;\;\:\;\; \hbox { weakly}\; \hbox { in }\; L^2(\Omega). \end{equation} | (5.22) |
Using (5.16) and (5.18) we obtain
\begin{equation*} u_{\alpha}^{m}(0) = \zeta^m_\alpha(x_1, x_2), \end{equation*} |
and then
\begin{equation*} \kappa_{\alpha\beta}^{m}(0) = e_{\alpha\beta}(\zeta^m). \end{equation*} |
Taking now v = \left(v_1, v_2, 0 \right) \hbox{in} (3.5), \hbox{with} \; v_\alpha\in H_\Gamma^1(\Omega) , and multiplying the equation by {{\varepsilon}} , we obtain
\begin{eqnarray} && {{\varepsilon}} \int_\Omega \left[ 4\mu\kappa^{{\ell_\varepsilon^m}}_{\alpha\beta}({{\varepsilon}})e_{\alpha\beta}(v) + \kappa^{{\ell_\varepsilon^m}}_{\alpha\alpha}({{\varepsilon}})e_{\beta\beta}(v) \right]dx + 4\mu\int_\Omega\kappa^{{\ell_\varepsilon^m}}_{\alpha3}({{\varepsilon}})e_{\alpha3}(v)dx + {{\varepsilon}} \lambda \int_\Omega \kappa^{{\ell_\varepsilon^m}}_{33}({{\varepsilon}})e_{\alpha\alpha}(v)dx \\ && = {{\varepsilon}} \eta_m({{\varepsilon}}) \int_\Omega u_\alpha^{{\ell_\varepsilon^m}}({{\varepsilon}})v_\alpha dx. \end{eqnarray} | (5.23) |
Passing to the limit as {{\varepsilon}} \rightarrow 0 , we obtain
\begin{equation*} 4\mu\int_\Omega\kappa^m_{\alpha3}(0)\partial_3v_\alpha dx = 0, \;\;\;\;\;\; \forall v_\alpha \in H_\Gamma^1(\Omega), \end{equation*} |
which has as unique solution (see [8])
\kappa^m_{\alpha3}(0) = 0. |
Lemma 5.7. For each m\geq 1 , there exists a subsequence, still denoted {{\varepsilon}} , such that
\begin{equation} \kappa_{33}^{{\ell_\varepsilon^m}}({{\varepsilon}}) \rightharpoonup \kappa_{33}^{m}(0) \; \; { weakly}\; {in }\; L^2(\Omega) \end{equation} | (5.24) |
where
\begin{equation} \kappa_{33}^{m}(0) = \frac{-\lambda}{\lambda+2\mu}\kappa_{\alpha\alpha}^{m}(0). \end{equation} | (5.25) |
Proof. Let v = (0, 0, v_3), \; v_3 \in H_\Gamma^1(\Omega) , we have
e_{\alpha\beta}(v) = 0 \;\;\; \hbox{and}\;\;\; e_{\alpha3}(v) = \frac{1}{2}\partial_\alpha v_3. |
Substituting in (3.5) and multiplying the equation by {{\varepsilon}}^2 , we obtain
\begin{eqnarray*} {{\varepsilon}}\int_\Omega2\mu\kappa_{\alpha3}^{{\ell_\varepsilon^m}}({{\varepsilon}})\partial_\alpha v_3 dx + \lambda \int_\Omega\kappa_{\alpha\alpha}^{{\ell_\varepsilon^m}}({{\varepsilon}})\partial_3 v_3 dx&+& (\lambda+2\mu)\int_\Omega \kappa_{33}^{{\ell_\varepsilon^m}}({{\varepsilon}})\partial_3 v_3 dx = \eta_m({{\varepsilon}}) \int_\Omega u_3^{{\ell_\varepsilon^m}}({{\varepsilon}}) v_3 dx. \end{eqnarray*} |
Which gives by passing to the limit as {{\varepsilon}} \rightarrow 0
\begin{equation*} \int_\Omega\left[\lambda\kappa^m_{\alpha\alpha}(0)+ (\lambda+2\mu)\kappa_{33}^m(0)\right]\partial_3 v_3 dx = 0, \;\;\;\;\forall v_3 \in H_\Gamma^1(\Omega), \end{equation*} |
and consequently,
\begin{equation*} \lambda\kappa^m_{\alpha\alpha}(0)+ (\lambda+2\mu)\kappa_{33}^m(0) = 0. \end{equation*} |
Lemma 5.8. The stretching displacements (\zeta_1^m, \zeta_2^m) \in [H_0^1(\omega)]^2 satisfy for all (\xi_1, \xi_2) \in [H_0^1(\omega)]^2 ,
\begin{eqnarray} && \frac{E}{1+\nu} \int_\omega e_{\alpha\beta} (\zeta^m) e_{\alpha\beta}(\xi)dx_1dx_2 + \frac{E\nu}{1-\nu^2} \int_\omega e_{\alpha\alpha} (\zeta^m) e_{\beta\beta} (\xi) dx_1dx_2 \\ & = & \eta_m(0) \int_\omega \zeta_\alpha^m \xi_\alpha dx_1dx_2. \end{eqnarray} | (5.26) |
Proof. Taking v = (\xi_1, \xi_2, 0) in (3.5) with (\xi_1, \xi_2) \in [H_0^1(\omega)]^2 , we have
\begin{eqnarray*} \int_\Omega \left[2\mu\kappa^{{\ell_\varepsilon^m}}_{\alpha\beta}({{\varepsilon}}) e_{\alpha\beta}(v)+\lambda \kappa^{{\ell_\varepsilon^m}}_{\alpha\alpha}({{\varepsilon}}) e_{\beta\beta}(v)\right]dx &+& \int_\Omega \lambda \kappa^{{\ell_\varepsilon^m}}_{33}({{\varepsilon}}) e_{\beta\beta}(v)dx = \eta_m({{\varepsilon}}) \int_\Omega u_\alpha^{{\ell_\varepsilon^m}}({{\varepsilon}}) v_\alpha dx. \end{eqnarray*} |
Passing to the limit when {{\varepsilon}} \rightarrow 0 , we obtain
\begin{array}{l} 2\mu\int_\omega \kappa^m_{\alpha\beta}(0) e_{\alpha\beta}(\xi) dx_1dx_2 +\lambda \int_\omega \kappa^m_{\alpha\alpha}(0) e_{\beta\beta}(\xi)dx_1dx_2 \\ \;\; \;\;\;\; + \lambda\int_\Omega \kappa^m_{33}(0) e_{\beta\beta}(\xi)dx_1dx_2 = \eta_m(0) \int_\omega u_\alpha^m(0) \xi_\alpha dx_1dx_2, \end{array} |
which can be written, using (5.25)
\begin{eqnarray*} 2\mu\int_\omega \kappa^m_{\alpha\beta}(0) e_{\alpha\beta}(\xi) dx_1dx_2 +\frac{2\mu\lambda}{\lambda+2\mu} \int_\omega \kappa^m_{\alpha\alpha}(0) e_{\beta\beta}(\xi)dx_1dx_2 = \eta_m(0) \int_\omega u_\alpha^m(0) \xi_\alpha dx_1dx_2. \end{eqnarray*} |
Replacing \kappa^m_{\alpha\beta}(0) by their expressions (5.20), we obtain
\begin{eqnarray*} 2\mu\int_\omega e_{\alpha\beta} (\zeta^m) e_{\alpha\beta}(\xi) dx_1dx_2 +\frac{2\mu\lambda}{\lambda+2\mu} \int_\omega e_{\alpha\alpha} (\zeta^m) e_{\beta\beta} (\xi) dx_1dx_2 = \eta_m(0) \int_\omega \zeta_\alpha^m \xi_\alpha dx_1dx_2. \end{eqnarray*} |
(5.26) comes using relations (2.2).
Lemma 5.9. For each m\geq 1 , the whole family (\eta_{{\ell_\varepsilon^m}}({{\varepsilon}}))_{{{\varepsilon}} > 0} converges as {{\varepsilon}} \rightarrow 0 . In addition, if \eta_m(0) is a simple eigenvalue of (5.26), then \eta_{{\ell_\varepsilon^m}}({{\varepsilon}}) is also a simple eigenvalue of (3.5) for {{\varepsilon}} < {{\varepsilon}}_0 small enough.
Proof. See [7].
Proposition 5.10. For each m\geq 1 , the limit eigensolutions (\eta_m(0), \zeta_1^m, \zeta_2^m) verify the classical equations of stretching vibrations:
\begin{equation} \left\{ \begin{array}{l} \frac{E}{2(1-\nu^2)}\left[2\frac{\partial^2\zeta^m_1}{\partial x_1^2} + (1-\nu)\frac{\partial^2\zeta^m_1}{\partial x_2^2} +(1+\nu) \frac{\partial^2\zeta^m_2}{\partial x_1 \partial x_2}\right] = \eta_m(0)\zeta^m_1, \\ \frac{E}{2(1-\nu^2)}\left[2\frac{\partial^2\zeta^m_2}{\partial x_2^2} + (1-\nu)\frac{\partial^2\zeta^m_2}{\partial x_1^2} + (1+\nu)\frac{\partial^2\zeta_1}{\partial x_1 \partial x_2}\right] = \eta_m(0)\zeta^m_2, \end{array} \right. \end{equation} | (5.27) |
with
\begin{equation} \zeta^m_1 = \zeta^m_2 = 0 \;\;\; on \; \gamma. \end{equation} | (5.28) |
Proof. Performing an integrating by part in the left-hand side of Eq (5.26) we obtain, for all (\xi_1, \xi_2) \in [H_0^1(\omega)]^2
\begin{array}{l} \frac{E}{1+\nu}\int_\omega\left[ \frac{\partial^2\zeta^m_1}{\partial x_1^2} + \frac{1}{2}\frac{\partial^2\zeta^m_1}{\partial x_2^2} + \frac{1}{2}\frac{\partial^2\zeta^m_2}{\partial x_1\partial x_2} \right]\xi_1 dx_1dx_2\\ \;\;\;\; +\frac{E}{1+\nu}\int_\omega\left[ \frac{\partial^2\zeta^m_2}{\partial x_2^2} + \frac{1}{2}\frac{\partial^2\zeta^m_2}{\partial x_1^2} + \frac{1}{2}\frac{\partial^2\zeta^m_1}{\partial x_1\partial x_2}\right]\xi_2 dx_1dx_2 \\ \;\;\;\;\;\; +\frac{E\nu}{1-\nu^2}\int_\omega\left[ \frac{\partial^2\zeta^m_1}{\partial x_1^2} + \frac{\partial^2\zeta^m_2}{\partial x_1\partial x_2}\right]\xi_1 dx_1dx_2\\ \;\;\;\;\;\;\;\;+ \frac{E\nu}{1-\nu^2}\int_\omega\left[ \frac{\partial^2\zeta^m_2}{\partial x_2^2} + \frac{\partial^2\zeta^m_1}{\partial x_1\partial x_2}\right]\xi_2dx_1dx_2\\ \;\;\;\;\;\;\;\;\;\; = \eta_m(0)\int_\omega\zeta^m_\alpha\xi_\alpha dx_1dx_2 \end{array} |
which gives, for all (\xi_1, \xi_2) \in [H_0^1(\omega)]^2
\begin{array}{l} \frac{E}{2(1-\nu^2)}\int_\omega\left[2\frac{\partial^2\zeta^m_1}{\partial x_1^2} + (1-\nu)\frac{\partial^2\zeta^m_1}{\partial x_2^2} +(1+\nu) \frac{\partial^2\zeta^m_2}{\partial x_1 \partial x_2}\right]\xi_1\; dx_1dx_2\\ \;\;+\frac{E}{2(1-\nu^2)}\int_\omega\left[2\frac{\partial^2\zeta^m_2}{\partial x_2^2} + (1-\nu)\frac{\partial^2\zeta^m_2}{\partial x_1^2} + (1+\nu)\frac{\partial^2\zeta^m_1}{\partial x_1 \partial x_2}\right]\xi_2\; dx_1dx_2\\ \;\;\;\;\;\;\;\; = \eta_m(0)\int_\omega\zeta^m_1\xi_1\; dx_1dx_2 +\eta_m(0)\int_\omega\zeta^m_2\xi_2\; dx_1dx_2. \end{array} |
(5.27) is obtained by taking respectively \xi_2 = 0 and \xi_1 = 0 .
In this work, we have proved that the stretching frequencies of an elastic thin plate is the limit of a family of high frequencies of the three-dimensional elastic model of the plate, as the thickness approaches zero. We have also shown, that the standard spectral problem associated to stretching modes in linear elastic plates can be derived mathematically from the standard three-dimensional eigenvalue problem of linear elasticity through a non-standard asymptotic analysis technique. This technique can be used to study a wide variety of problems of modeling vibrations for thin structure like folded plate and junction between different thin structures.
The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.
The author gratefully acknowledges the financial support of the Deanship of research of Imam Mohammad Ibn Saud Islamic University (IMSIU).
The author declares no conflict of interest.
[1] | P. G. Ciarlet, P. Destuynder, A justification of the two-dimensional linear plate model, J. Mec., 18 (1979), 315–344. |
[2] | P. G. Ciarlet, H. Le Dret, R. Nzengwa, Junctions between three-dimensional and two-dimensional linear elastic structures, J. Math. Pures Appl., 68 (1989), 261–295. |
[3] |
H. Le Dret, Folded plates revisited, Comput. Mech., 5 (1989), 345–365. http://dx.doi.org/10.1007/BF01047051 doi: 10.1007/BF01047051
![]() |
[4] |
H. Le Dret, Modeling of a folded plate, Comput. Mech., 5 (1990), 401–416. https://doi.org/10.1007/BF01113445 doi: 10.1007/BF01113445
![]() |
[5] | H. Le Dret, Modeling of the junction between two rods, J. Math. Pure Appl., 68 (1989), 365–397. |
[6] |
L. Trabucho, J. M. Viaño, Existence and characterization of higher-order terms in an asymptotic expansion method for linearized elastic beams, Asymptotic Anal., 2 (1989), 223–255. https://doi.org/10.3233/ASY-1989-2303 doi: 10.3233/ASY-1989-2303
![]() |
[7] |
P. G. Ciarlet, S. Kesavan, Two-dimensional approximations of three-dimensional eigenvalue problems in plate theory, Comput. Methods Appl. Mech. Eng., 26 (1981), 0045–7825. http://dx.doi.org/10.1016/0045-7825(81)90091-8 doi: 10.1016/0045-7825(81)90091-8
![]() |
[8] |
H. Le Dret, Vibrations of a folded plate, ESAIM: M2AN, 24 (1990), 501–521. https://doi.org/10.1051/m2an/1990240405011 doi: 10.1051/m2an/1990240405011
![]() |
[9] |
F. Bourquin, P. G. Ciarlet, Modeling and justification of eigenvalue problems for junctions between elastic structures, J. Funct. Anal., 87 (1989), 392–427. https://doi.org/10.1016/0022-1236(89)90017-7 doi: 10.1016/0022-1236(89)90017-7
![]() |
[10] |
V. Lods, Modeling and justification of an eigenvalue problem for a plate inserted in a three-dimensional support, ESAIM: M2AN, 30 (1996), 413–444. http://dx.doi.org/10.1051/m2an/1996300404131 doi: 10.1051/m2an/1996300404131
![]() |
[11] | N. Kerdid, Asymptotic behavior of the eigenvalue problem in a thin linearly elastic clamped rod when its thickness tends to zero, C. R. Acad. Sci. Ser. I: Math., 316 (1993), 755–758. |
[12] |
N. Kerdid, Modeling the vibrations of a multi-rod structure, ESAIM: M2AN, 31 (1997), 891–925. https://doi.org/10.1051/m2an/1997310708911 doi: 10.1051/m2an/1997310708911
![]() |
[13] |
S. Jimbo, A. Rodríguez Mulet, Asymptotic behavior of eigenfrequencies of a thin elastic rod with non-uniform cross-section, J. Math. Soc. Japan, 72 (2020), 119–154. https://doi.org/10.2969/jmsj/81198119 doi: 10.2969/jmsj/81198119
![]() |
[14] |
S. Jimbo, E. Ushikoshi, H. Yoshihara, Asymptotic behavior of the eigenfrequencies of a thin elastic rod with non-uniform cross-section of extremely oblate shape, Calc. Var. Partial Differ. Equ., 62 (2022), 11. https://doi.org/10.1007/s00526-022-02325-1 doi: 10.1007/s00526-022-02325-1
![]() |
[15] |
M. Serpilli, S. Lenci, Asymptotic modelling of the linear dynamics of laminated beams, Int. J. Solids Struct., 49 (2012), 1147–1157. http://dx.doi.org/10.1016/j.ijsolstr.2012.01.012 doi: 10.1016/j.ijsolstr.2012.01.012
![]() |
[16] |
J. Tamba\tilde{{\rm{c}}}a, One-dimensional approximations of the eigenvalue problem of curved rods, Math. Methods Appl. Sci., 24 (2001), 927–948. https://doi.org/10.1002/mma.249 doi: 10.1002/mma.249
![]() |
[17] |
A. Gaudiello, D. Gómez, M. E. Pérez-Martínez, Asymptotic analysis of the high frequencies for the Laplace operator in a thin T-like shaped structure, J. Math. Pures Appl., 134 (2020), 299–327. http://dx.doi.org/10.1016/j.matpur.2019.06.005 doi: 10.1016/j.matpur.2019.06.005
![]() |
[18] |
H. Irago, N. Kerdid, J. M. Viaño, Asymptotic analysis of high frequency modes in thin rods, C. R. Acad. Sci. Ser. I: Math., 326 (1998), 1255–1260. https://doi.org/10.1016/S0764-4442(98)80238-3 doi: 10.1016/S0764-4442(98)80238-3
![]() |
[19] |
H. Irago, N. Kerdid, J. M. Viaño, Asymptotic analysis of torsional and stretching modes of thin rods, Quart. Appl. Math., 58 (2000), 495–510. http://dx.doi.org/10.1090/QAM/1770651 doi: 10.1090/QAM/1770651
![]() |