Abstract
The response of a body described by a quasi-linear viscoelastic constitutive relation, whose material moduli depend on the mechanical pressure (that is one-third the trace of stress) is studied. The constitutive relation stems from a class of implicit relations between the histories of the stress and the relative deformation gradient. A-priori thresholding is enforced through the pressure that ensures that the displacement gradient remains small. The resulting mixed variational problem consists of an evolutionary equation with the Volterra convolution operator; this equation is studied for well-posedness within the theory of maximal monotone graphs. For isotropic extension or compression, a semi-analytic solution of the quasi-linear viscoelastic problem is constructed under stress control. The equations are studied numerically with respect to monotone loading both with and without thresholding. In the example, the thresholding procedure ensures that the solution does not blow-up in finite time.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Implicit constitutive relations were introduced by Rajagopal [32] to describe elastic response that could not be described by classical constitutive relations for the same. In that study, he also considered implicit relations for viscoelastic response between the stress and its various time derivatives, and kinematical variables and their various time derivatives. Implicit constitutive relations for viscoelastic response had been introduced much earlier by Burgers [2] and Oldroyd [28]. While the constitutive relation introduced by Maxwell [24] relates the stress and the time rate of stress to the symmetric part of the velocity gradient, it is not an implicit relation as the symmetric part of the velocity gradient can be expressed in terms of the stress and the time rate of stress. Later, Průša and Rajagopal [31] introduced implicit relations between the history of the stress and the history of the relative deformation gradient. Recently, Murru et al. [27] derived a subclass of constitutive relations which are appropriate to describe the elastic response of materials such as rocks, bone, ceramics, concrete, intermetallic alloys and other porous bodies. Rajagopal and Wineman [35] put into place implicit constitutive relations for viscoelastic porous bodies.
Let us consider the implicit response of elastic bodies introduced in [32]. The challenge consists in the fact that the implicit constitutive relation between the Cauchy stress \(\varvec{\sigma }\) and the deformation gradient \({\textbf{F}}\) is given in the form
and cannot be inverted to express the stress as a function of the strain, and vice versa.
A particular subclass of these constitutive relations was introduced by Rajagopal and co-authors [34, 35] in which both the stress and the linearized strain \(\varvec{\varepsilon }\) appear linearly. Such material response is nonlinear as it contains the mutual product of stress and strain, allowing one to have linearly inhomogeneous material moduli. The material moduli depend on the density in contrast to the classical linearized elastic model with constant coefficients. In virtue of the balance of mass, the density can be replaced by \(\rho =\rho _{\textrm{R}}/(1+\textrm{tr}\varvec{\varepsilon })\), where \(\rho _{\textrm{R}}\) is the density in the reference configuration. Well-posedness for the corresponding problems was established within the variational theory by Itou et al. [14,15,16] by thresholding the moduli, thus preventing them from becoming unbounded.
The first integral constitutive relation to describe viscoelastic response was developed by Boltzmann [1], and this has been followed by integral constitutive relations due to Green and Rivlin [8], Lockett [23], Pipkin and Rogers [29] and others. Fung [7] developed a one-dimensional approximation, referred to as a quasi-linear constitutive relation, to describe the viscoelastic behavior of biological tissues. Muliana et al. [26] developed a quasi-linear viscoelastic model in three-dimensions, wherein the relationship between the stress and strain is nonlinear. These constitutive relations have been extended to implicit constitutive relations, in the context of strain-limiting approach by Bulíček et al. [6] and Itou et al. [10, 11]. The quasi-linear viscoelastic model has been used to study the Boussinesq problem by Itou et al. [12, 13] for the indentation of half-space by a rigid punch with unknown contact zones. We cite the references [18,19,20,21, 30] for nonlinear and unilateral boundary conditions which is appropriate to problems concerning the response of viscoelastic materials.
In order to introduce the quasi-linear constitutive relation, we begin with a consideration of the implicit constitutive relation for elastic response due to Rajagopal [34] in which both the stress \(\varvec{\sigma }\) and the linearized strain \(\varvec{\varepsilon }^\textrm{e}\) appear linearly:
where \({\textbf{I}}\) is the identity transformation, \(E_1, E_2\) and \(\lambda _1, \lambda _2, \lambda _3\) are constants. Equation (1.1) is nonlinear since it involves the products of variables \(\varvec{\sigma }\) and \(\varvec{\varepsilon }^\textrm{e}\). When all \(\lambda _1 =\lambda _2 =\lambda _3 =0\), the above constitutive relation reduces to that for classical linearized elasticity. Let \(\lambda _1 =\lambda _2 =0\) such that (1.1) can be inverted as an explicit constitutive expression for the strain in terms of the stress
where the parameters \(E_1>0\) and \(E_2>0\), and \(\lambda _3\in {\mathbb {R}}\) is a material moduli. Notice that the linearized strain is a nonlinear function of the stress, but in (1.1), the linearized strain and stress are such that they appear linearly. Below we make the assumptions in order to get from (1.2) a specific model suitable for mathematical analysis.
Decomposing the stress and strain tensors into its deviatoric and spherical parts
we assume that the nonlinearity in the deviatoric part of (1.2) is negligible, for example, for isotropic extension or compression in Sect. 4, such that
where \(E_3:=E_1/3 +E_2\). This assumption eliminates the mixed term \(\varvec{\sigma }^*/ (1 +\lambda _3 \textrm{tr}\varvec{\sigma })\) from our consideration. It is worth noting that \(\textrm{tr}\varvec{\sigma }\) determines the mechanical pressure
thus implying the pressure-dependent factor in front of \(\textrm{tr}\varvec{\sigma }\) (see Rajagopal [33] for a discussion of the concept of pressure).
The next assumption allows us to guarantee properties of ellipticity and boundedness for the constitutive relation. It can be observed that small \(|1 +\lambda _3 \textrm{tr}\varvec{\sigma }|\) leads to the equation (1.4), wherein the strain \(\varvec{\varepsilon }^\textrm{e}\) becomes unbounded. For consistency with the assumption of small displacement, according to [14,15,16], we prescribe the lower and upper thresholds
where \({\underline{M}}\) may be small, and the cut-off function
Using (1.6) and (1.7), the thresholding equation is introduced as
This includes equation (1.4) if \({\underline{M}}\le 1 +\lambda _3 \textrm{tr}\varvec{\sigma }\le {\overline{M}}\) in (1.7). Otherwise, if \(1 +\lambda _3 \textrm{tr}\varvec{\sigma }< {\underline{M}}\), then (1.8) turns into the linearized relation
on the other hand, if \(1 +\lambda _3 \textrm{tr}\varvec{\sigma }> {\overline{M}}\), then it is linearized as well
Furthermore, if \(\lambda _3=0\), then \(B(\textrm{tr}\varvec{\sigma }) =\textrm{tr}\varvec{\sigma }\) in (1.7) due to (1.6), and we recover the equation of classical linearized elasticity
From (1.11), the material moduli are identified as
where \(E>0\) and \(\nu \in (0, 1/2)\) are the Young’s modulus and Poisson’s ratio, which determine the Lame parameter \(\mu \) and the bulk modulus K, whereas \(\lambda _3\) in (1.4) and the thresholds \({\underline{M}}, {\overline{M}}\) in (1.8) are fitting parameters.
For time \(t\ge 0\), following Rajagopal and co-authors [26, 35], we introduce the viscoelastic constitutive relation corresponding to that in the thresholding equation (1.8):
It corresponds to the constitutive equation (1.4) when
The kernel \(J\ge 0\) that appears in the Volterra convolution equation (1.13) is typically given by the exponential sum
with creep parameters \(J_1,\ldots ,J_N\), \(\tau _1,\ldots \tau _N \ge 0\). It is worth noting that the integral operator \({\mathcal {I}}\) cannot be inverted in general, even if \({\mathcal {F}}\) would be linear in \(\varvec{\sigma }\). In the particular case of \(N=1\) in (1.15), equation (1.13) reads
On differentiating (1.16) with respect to t, we calculate \(J^{\prime \prime }(t) =-J^\prime (t)/ \tau _1\) and can invert the equation, thus arriving at the generalized Kelvin–Voigt model:
where dot denotes the time derivative. The classical linearized Kelvin–Voigt model \(\varvec{\varepsilon }^\textrm{v} +\tau _1 \dot{\varvec{\varepsilon }}^\textrm{v} =E_1 \varvec{\sigma }\) is recovered from equation (1.17) with the parameters chosen as \(J_1 =1\) and \(\tau _1 =\alpha E_1\), where \(\alpha >0\) is the viscosity.
From the mathematical point of view, the principal difficulty of quasi-linear viscoelastic equations (1.13) and (1.14) concerns the fact that the Volterra convolution operator \({\mathcal {I}}\) is not monotone or coercive. Therefore, the monotone operator theory is inapplicable to guarantee solvability for the corresponding variational problems that will be stated. To remedy this difficulty, following [11,12,13], we introduce an auxiliary relation between the strains \(\varvec{\varepsilon }^\textrm{e}\) and \(\varvec{\varepsilon }^\textrm{v}\) by the integral formula
Indeed, inserting into (1.18) the strain \(\varvec{\varepsilon }^\textrm{e}\) satisfying the constitutive relation for elastic response (1.8), we find \(\varvec{\varepsilon }^\textrm{v} ={\mathcal {I}}\bigl [{\mathcal {F}}[\varvec{\sigma }]\bigr ]\) which reduces to the constitutive equation for viscoelastic response (1.13).
The nonlinear equation (1.8) can be described by a graph \({\mathfrak {G}}\) between \(\varvec{\sigma }\) and \(\varvec{\varepsilon }^\textrm{e}\). The theory of graphs is well-suited to study implicit and multi-valued functions. Following Bulíček et al. [4, 5], we will prove that \({\mathfrak {G}}\) is maximal monotone and coercive on the appropriate selection \(\varvec{\varepsilon }^\textrm{e} =\varvec{\varepsilon }({\textbf{w}})\), in this manner justifying well-posedness for the underlying variational problem. The concept of maximal monotony for nonlinear operators was established by Browder [3] and Minty [25]. We also cite relevant results using the pseudo-monotone operators in [9, 17] and hemi-variational inequalities in [22, 36].
The structure of the paper is as follows. In Sect. 2, we document a boundary-value problem for the quasi-linear viscoelastic model for unknown \(\varvec{\sigma }\) and \(\varvec{\varepsilon }^\textrm{v} =\varvec{\varepsilon }({\textbf{u}})\) within the context of the thresholding equation (1.13). Well-posedness for the corresponding mixed variational problem is proved in Sect. 3 based on the representation (1.18). In Sect. 4, we apply the theory to the semi-analytic example of isotropic extension or compression when deformation is controlled by the pressure. The numerical simulation tests are presented first applying monotone load, and second maintaining and then removing the load. The thresholding model (1.13) demonstrates an interesting feature when compared to the unthresholded model (1.14), whereas in the latter, a solution may blow-up under finite pressure when subject to monotone loading.
2 The quasi-linear viscoelastic problem
Let \(\Omega \) be a bounded domain in the Euclidean space \({\mathbb {R}}^d\), where the spatial dimensions \(d=2\) and \(d=3\) are physically relevant. Let its boundary \(\partial \Omega \) be Lipschitz continuous, and the unit normal vector \({\textbf{n}} =(n_1,\ldots ,n_d)\) at \(\partial \Omega \) be dire cted outward \(\Omega \). We assume that the boundary is comprised of two mutually disjoint sets \(\partial \Omega =\overline{\Gamma _{\textrm{N}}}\cup \overline{\Gamma _{\textrm{D}}}\) corresponding to the Neumann \(\Gamma _\textrm{N}\) and Dirichlet \(\Gamma _{\textrm{D}}\not =\emptyset \) boundary conditions. For spatial points \({\textbf{x}}=(x_1,\ldots ,x_d)\) in the closure \({\overline{\Omega }} =\Omega \cup \partial \Omega \) and times \(t\in [0,T]\) with some final time \(T>0\) fixed, we denote the right time-space cylinder by \(\Omega ^T=(0,T)\times \Omega \) with the side consisting of two parts \(\Gamma _{\textrm{N}}^T=(0,T)\times \Gamma _{\textrm{N}}\) and \(\Gamma _{\textrm{D}}^T=(0,T)\times \Gamma _{\textrm{D}}\).
We look for the displacement \({\textbf{u}} =(u_1,\ldots ,u_d)(t,{\textbf{x}})\) in \(\overline{\Omega ^T}\). It determines the linearized strain \(\varvec{\varepsilon }({\textbf{u}})\) valued in the space of second-order symmetric tensors \({\mathbb {R}}^{d\times d}_{\textrm{sym}}\) as the symmetric gradient with the entries
For the given body force \({\textbf{f}} =(f_1,\ldots ,f_d)(t,{\textbf{x}})\), \({\textbf{f}}\in C([0,T]; L^2(\Omega ; {\mathbb {R}}^d))\), we look also for the stress tensor \(\varvec{\sigma }(t, {\textbf{x}})\in {\mathbb {R}}^{d\times d}_\textrm{sym}\) satisfying the equilibrium equation omitting the inertia terms:
Prescribing the boundary force \({\textbf{g}} =(g_1,\ldots ,g_d)(t,{\textbf{x}})\), \({\textbf{g}}\in C([0,T]; L^2(\Gamma _{\textrm{N}}; {\mathbb {R}}^d))\), we augment (2.2) with the mixed Dirichlet–Neumann boundary conditions
where \(\varvec{\sigma }{\textbf{n}} =(\sum _{j=1}^d \sigma _{1j} n_j, \ldots , \sum _{j=1}^d \sigma _{dj} n_j)\). The quasi-static equilibrium problem (2.1)–(2.4) is rendered complete with the constitutive equation for viscoelastic response (1.13) for \(\varvec{\varepsilon }^\textrm{v} =\varvec{\varepsilon }({\textbf{u}})\) as
Following the argument for the use of (1.18) which was articulated in the Introduction, we look for the displacement \({\textbf{u}}\) in the form
where the selection \(\varvec{\varepsilon }^\textrm{e} =\varvec{\varepsilon }({\textbf{w}})\) satisfies equation (1.8). The relation (1.8) between stress and strain can be generalized to an implicit relation on graph \({\mathfrak {G}}\subset ({\mathbb {R}}^{d\times d}_{\textrm{sym}})^2\) which we define by the inclusion
Lemma 2.1
(Selection on the graph) The quasi-linear equation (2.5) for viscoelastic response completing the equilibrium problem (2.1)–(2.4) can be generalized to the identity (2.6) and the selection \(\varvec{\varepsilon }^\textrm{e} =\varvec{\varepsilon }({\textbf{w}})\) on the graph \({\mathfrak {G}}\) in (2.7) such that
Proof
If the selection \(\varvec{\varepsilon }({\textbf{w}}) ={\mathcal {F}}[\varvec{\sigma }]\) in (2.8) holds, then the variable \({\textbf{w}}\) can be reduced by use of (2.6). Indeed, applying to the both sides of (2.6) the symmetric gradient from (2.1), we can interchange the linear operators \(\varvec{\varepsilon }\) and \({\mathcal {I}}\), thus obtaining
and resulting in (2.5).
Conversely, \(\varvec{\varepsilon }({\textbf{u}}) =\varvec{\varepsilon }({\mathcal {I}}[{\textbf{w}}])\) follows (2.6) excluding rigid motions, which vanish here due to the homogeneous Dirichlet condition (2.3). Then, (2.5) and (2.6) together with (2.9) justify the selection (2.8). \(\square \)
In the next lemma, we establish some useful properties of \({\mathfrak {G}}\).
Lemma 2.2
(Properties of the graph)
-
(i)
The graph \({\mathfrak {G}}\) in (2.7) includes the origin:
$$\begin{aligned} ({\textbf{0}}, {\textbf{0}})\in {\mathfrak {G}}. \end{aligned}$$(2.10) -
(ii)
For all \((\varvec{\sigma }, \varvec{\varepsilon }^\textrm{e})\in {\mathfrak {G}}\), the graph is coercive with the uniform estimate:
$$\begin{aligned} \varvec{\varepsilon }^\textrm{e} \cdot \varvec{\sigma }\ge C_{\varvec{\sigma }} \Vert \varvec{\sigma }\Vert ^2 +C_{\varvec{\varepsilon }} \Vert \varvec{\varepsilon }^\textrm{e}\Vert ^2, \end{aligned}$$(2.11)where the dot stands for the scalar product of tensors \(\varvec{\varepsilon }^\textrm{e}\cdot \varvec{\sigma } =\sum _{i,j=1}^d \varepsilon ^\textrm{e}_{ij} \sigma _{ij}\), Frobenius norm \(\Vert \varvec{\sigma }\Vert =\sqrt{\varvec{\sigma }\cdot \varvec{\sigma }}\), and the constant factors are
$$\begin{aligned} C_{\varvec{\sigma }}:=\frac{1}{2}\min \Bigl (E_1, \frac{E_3 d}{{\overline{M}}^2}\Bigr ),\quad C_{\varvec{\varepsilon }}:=\frac{1}{2}\min \Bigl (\frac{1}{E_1}, \frac{{\underline{M}}^4}{(E_3 d) {\overline{M}}^2}\Bigr ). \end{aligned}$$(2.12) -
(iii)
For all pairs \((\varvec{\sigma }^1, \varvec{\varepsilon }^1), (\varvec{\sigma }^2, \varvec{\varepsilon }^2)\in {\mathfrak {G}}\) the graph is monotone:
$$\begin{aligned} (\varvec{\varepsilon }^1 -\varvec{\varepsilon }^2) \cdot (\varvec{\sigma }^1 -\varvec{\sigma }^2)\ge 0. \end{aligned}$$(2.13) -
(iv)
For \((\varvec{\sigma }^1, \varvec{\varepsilon }^1)\in ({\mathbb {R}}^{d\times d}_{\textrm{sym}})^2\), the graph is maximal monotone:
$$\begin{aligned} \text {if } (\varvec{\varepsilon }^1 -\varvec{\varepsilon }^2) \cdot (\varvec{\sigma }^1 -\varvec{\sigma }^2)\ge 0\text { for all } (\varvec{\sigma }^2, \varvec{\varepsilon }^2)\in {\mathfrak {G}},\quad \text {then } (\varvec{\sigma }^1, \varvec{\varepsilon }^1)\in {\mathfrak {G}}. \end{aligned}$$(2.14)
Proof
For \(\varvec{\sigma }^1, \varvec{\sigma }^2\in {\mathbb {R}}^{d\times d}_{\textrm{sym}}\), it is straightforward to check that the function \(B: {\mathbb {R}}\mapsto {\mathbb {R}}\) defined in (1.7) is Lipschitz continuous:
and strongly monotone:
hence bounded and coercive, too.
According to (1.3) extended to \({\mathbb {R}}^d\) and (1.8), the strain \(\varvec{\varepsilon }^\textrm{e}\) admits the decomposition
Forming the scalar product of (2.17) with \(\varvec{\sigma }\), using estimates (2.15) and (2.16) yields the lower bound
In virtue of the norm identity \(\Vert \varvec{\sigma }\Vert ^2 =\Vert \varvec{\sigma }^*\Vert ^2 +\textrm{tr}^2\varvec{\sigma }/d\) and the notation for factors in (2.12), the lower estimate (2.11) follows.
For points on the graph \((\varvec{\sigma }^1, \varvec{\varepsilon }^1), (\varvec{\sigma }^2, \varvec{\varepsilon }^2)\in {\mathfrak {G}}\) the definition (2.7) implies that
Subtracting (2.18) for \(n=1\) and \(n=2\), with the help of (2.16) and using the notation \(C_{\varvec{\sigma }}\) from (2.12), we estimate
This justifies the monotone property (2.13) for \({\mathfrak {G}}\). Moreover, this estimate establishes strong monotonicity of \({\mathcal {F}}\), that is, uniqueness of the solution.
Now, for fixed \((\varvec{\sigma }^1, \varvec{\varepsilon }^1)\in ({\mathbb {R}}^{d\times d}_\textrm{sym})^2\), we assume that the inequality in (2.14) holds for all \((\varvec{\sigma }^2, \varvec{\varepsilon }^2)\in {\mathfrak {G}}\) satisfying (2.18). For arbitrary \(\varvec{\sigma }\in {\mathbb {R}}^{d\times d}_{\textrm{sym}}\) and small \(\delta >0\), let
which belongs to the graph: \((\varvec{\sigma }^\delta , \varvec{\varepsilon }^\delta )\in {\mathfrak {G}}\). Testing (2.14) with \((\varvec{\sigma }^2, \varvec{\varepsilon }^2) =(\varvec{\sigma }^\delta , \varvec{\varepsilon }^\delta )\) and substituting (2.19) yields
which after division by \(\delta \) yields
On taking the limit as \(\delta \rightarrow 0\), the continuity of B in (2.15) leads to the variational equality
for all \(\varvec{\sigma }\in {\mathbb {R}}^{d\times d}_{\textrm{sym}}\). This justifies the equation (2.18) for \((\varvec{\sigma }^1, \varvec{\varepsilon }^1)\), thus \((\varvec{\sigma }^1, \varvec{\varepsilon }^1)\in {\mathfrak {G}}\), and the maximal monotone property (2.14) of the graph \({\mathfrak {G}}\) is valid.
The inclusion (2.10) is evident. The proof is completed. \(\square \)
In the next section, we provide a variational formulation for the quasi-linear viscoelastic problem (2.1)–(2.4), (2.6) and (2.8).
3 Variational formulation within maximal monotone and coercive graphs
We start by recalling the Korn–Poincaré inequality:
with constant \(C_{\textrm{KP}}\ge 1\), and the boundary trace theorem
We apply standard variational arguments, namely the equilibrium equation (2.2) is multiplied by \({\textbf{v}}=(v_1,\ldots ,v_d)(x)\) and integrated by parts over \(\Omega \) using Green’s formula
where \(\varvec{\varepsilon }({\textbf{v}})\) is defined according to (2.1). Applying the Neumann boundary condition (2.4), this yields the variational equation
for all test functions \({\textbf{v}}\in H^1(\Omega ; {\mathbb {R}}^d)\) such that \({\textbf{v}} ={\textbf{0}}\) on \(\Gamma _{\textrm{D}}\). Conversely, for \(H^1\)-smooth stress \(\varvec{\sigma }\), pointwise relations (2.2) and (2.4) follow from (3.3).
For the stress \(\varvec{\sigma }\in C([0,T]; L^2(\Omega ; {\mathbb {R}}^{d\times d}_{\textrm{sym}}))\) and displacements \({\textbf{u}}, {\textbf{w}}\in C([0,T]; H^1(\Omega ; {\mathbb {R}}^d))\) such that \({\textbf{u}} ={\textbf{w}} ={\textbf{0}}\) on \(\Gamma _{\textrm{D}}^T\), for every \(t\in [0,T]\), we express the variational equation (2.6) in the form:
for all test functions \(\varvec{\xi }\in L^2(\Omega ; {\mathbb {R}}^d)\), and set a selection on the graph \({\mathfrak {G}}\) which fulfill the inclusion in (2.7) as
for all test functions \(\varvec{\eta }\in L^2(\Omega ; {\mathbb {R}}^{d\times d}_{\textrm{sym}})\).
We prove existence of the weak solution \((\varvec{\sigma }, {\textbf{u}}, {\textbf{w}})\) satisfying the variational relations (3.3)–(3.5) based on Lemma 2.2. It is worth noting that the variable \({\textbf{w}}\) is redundant and later can be reduced from the solution due to Lemma 2.1.
Theorem 3.1
(Well-posedness) The unique solution \(\varvec{\sigma }\in C([0,T]; L^2(\Omega ; {\mathbb {R}}^{d\times d}_{\textrm{sym}}))\) and \({\textbf{u}}, {\textbf{w}}\in C([0,T]; H^1(\Omega ; {\mathbb {R}}^d))\) such that \({\textbf{u}} ={\textbf{w}} ={\textbf{0}}\) on \(\Gamma _{\textrm{D}}^T\) satisfying for all \(t\in [0,T]\) the Dirichlet boundary condition (2.3) and variational relations (3.3)–(3.5) exists and fulfills the following a-priori estimate:
where the constant \(C_{\varvec{\sigma }}\) and \(C_{\varvec{\varepsilon }}\) are defined in (2.12), and the forces determine
Proof
The theorem is proved in three steps: from the Galerkin approximation, we derive uniform estimate, and then pass to the limit.
k-dimensional Galerkin approximation. Let subspaces \(S^k\) and \(V^k\) of finite dimensions \(k\in {\mathbb {N}}\) build the conforming approximation of the admissible stress \(\varvec{\sigma }\) and displacement \({\textbf{w}}\) preserving \({\textbf{w}} ={\textbf{0}}\) on \(\Gamma _{\textrm{D}}\). We assume that \(\cup _{k=1}^\infty S^k\) is dense in \(L^2(\Omega ; {\mathbb {R}}^{d\times d}_{\textrm{sym}})\), the union \(\cup _{k=1}^\infty V^k\) is dense in \(H^1(\Omega ; {\mathbb {R}}^d)\), and inclusion \({\textbf{v}}^k\in V^k\) implies that \(\varvec{\varepsilon }({\textbf{v}}^k)\in S^k\).
First, we look for a discrete solution \(\varvec{\sigma }^k\in C([0,T]; S^k)\) and \({\textbf{w}}^k\in C([0,T]; V^k)\) fulfilling for all \(t\in [0,T]\) the semi-discrete in space problem (3.3):
which is endowed with the constitutive equation according to (3.5):
for all test functions \({\textbf{v}}^k\in V^k\) and \(\varvec{\eta }^k\in S^k\). Its unique solution exists in virtue of the Browder–Minty theorem, because the Lipschitz continuity (2.15) and strong monotony (2.16) of the nonlinear function B provide the coercive, strictly monotone, bounded, and hemi-continuous properties of the operator \({\mathcal {F}}\) in the mixed variational problem (3.8) and (3.9).
Uniform in k estimate. Since the stress \(\varvec{\sigma }^k\) and strain \(\varvec{\varepsilon }({\textbf{w}}^k)\) are connected by the relation (3.9), the selection \((\varvec{\sigma }^k, \varvec{\varepsilon }({\textbf{w}}^k))\in {\mathfrak {G}}\) holds according to definition (3.5). Therefore, applying the coercivity (2.11) and using the Korn–Poincaré inequality (3.1) yields the lower bound
On the other side, we test the variational equation (3.8) with \({\textbf{v}}^k ={\textbf{w}}^k\), apply the Cauchy–Schwarz inequality and trace theorem (3.2), then the weighted Young inequality provides the upper bound, where the constant \(C( {\textbf{f}}, {\textbf{g}})\) is from (3.7):
Combining together (3.10) and (3.11) and taking maximum over \(t\in [0,T]\) leads to the uniform in k estimate:
Passage to the limit \(k\rightarrow \infty \). From the uniform estimate (3.12), we obtain a weakly convergent subsequence still denoted by k such that as \(k\rightarrow \infty \):
Taking the limit of the linear equation (3.8), we get the equilibrium equation (3.3) for \(\varvec{\sigma }\) and \({\textbf{w}}\) from (3.13). Next, we derive the nonlinear equation in (3.5).
Testing the equilibrium equation (3.3) with the limit function \({\textbf{v}} ={\textbf{w}}\) implies
For finite k, inserting \({\textbf{v}}^k ={\textbf{w}}^k\) into (3.8), we obtain
By the virtue of weak convergences in (3.13), from (3.14) and (3.15), we conclude that
Arbitrary \((\varvec{\sigma }^2, \varvec{\varepsilon }^2)\in {\mathfrak {G}}\) and \((\varvec{\sigma }^k, \varvec{\varepsilon }({\textbf{w}}^k))\in {\mathfrak {G}}\) on the graph fulfill (2.13), that is
This allows us to estimate from below the scalar product
On taking the limit based on the convergences in (3.13) and (3.16) leads to
Then, the maximal monotone property (2.14) of the graph guarantees the inclusion \((\varvec{\sigma }, \varvec{\varepsilon }({\textbf{w}}))\in {\mathfrak {G}}\), i.e., relation (3.5) holds for the limit functions from (3.13).
The uniqueness of \(\varvec{\sigma }\) and \({\textbf{w}}\) can be derived from the strong monotone property (2.16) of B entering the term \({\mathcal {F}}\) as shown in the estimate between (2.18) and (2.19). The displacement \({\textbf{u}}\) is determined uniquely from the equation (3.4). This completes the proof. \(\square \)
4 Semi-analytical solution for isotropic extension or compression
In dimension \(d=3\), our consideration is simplified under the assumption of isotropic extension or compression independent of \({\textbf{x}}\) such that
Formula (4.1) implies the deviatoric parts \(\varvec{\sigma }^* =\varvec{\varepsilon }^* ={\textbf{0}}\). Therefore, the unknowns are the scalar time-dependent functions for pressure p according to (1.5) and dilatation \(e =\textrm{tr}\varvec{\varepsilon }\) according to the deviatoric-spherical decomposition (1.3). The sign \(p<0\) implies extension, and \(p>0\) compression. Since the stress tensor is space-independent, the equilibrium equation (2.2) is satisfied identically with the body force \({\textbf{f}}\equiv {\textbf{0}}\).
In the representation (4.1), we can distinguish between elastic and viscoelastic dilatation response through
Inserting (4.2) into the governing equation (2.6) with the Volterra convolution operator \({\mathcal {I}}\) from (1.13) implies that
and the governing relation (2.8) using the cut-off function B from (1.7) takes the following form:
We will compare (4.4) with the unthresholded constitutive equation which exhibits unlimited strain, introduced according to (1.4) as
The stress control formulation is studied. Namely, from the prescribed pressure p(t) as \(t\in [0,T]\), we find the evolution of dilatation \(e^\textrm{e}(t)\) and \(e^\textrm{v}(t)\) satisfying equations (4.3) and (4.4) (respectively, (4.5) in the case when there is no thresholding).
To solve (4.3), we discretize the problem on the time-grid of \(M+1\) points
and pick the piecewise-affine approximation
for \(k=1,\ldots , M\), where the differences
Inserting (4.6) into (4.3) and using \((e^\textrm{e}_M)^\prime (t) =\delta e^\textrm{e}_{k}\) as \(t\in [t_{k-1}, t_{k}]\) provides the numerical quadrature for the Volterra convolution operator
which after integration by parts using \(J^\prime (t-s) =-\hbox {d}J(t-s)/\hbox {d}s\) yields
For the numerical example, we take the kernel \(J(t) =J_1 (1 -\exp (-t/\tau _1) )\) as \(N=1\) in (1.15) such that
and (4.8) takes the explicit form of the piecewise-exponential function
The elastic moduli are set for concrete [27]: \(E = 30\) (GPa) and \(\nu = 0.2\) such that \(E_1 = 0.04\), \(E_3 =-E_2 = 0.00{\overline{6}}\) (1/GPa) in (1.12), \(\lambda _3 = 0.1\) (1/GPa) in (4.4) and (4.5), the lower threshold is taken to be \({\underline{M}} = 0.2\). The creep parameters in (1.15) are \(J_1 = 0.04\) and \(\tau _1 = 0.4\) (h).
4.1 Monotone loading by pressure
First, we prescribe the pressure by the linearly increasing function
where the final time \(T=1.5\) (h), and the loading rate \(g =3.{\overline{3}}\) (GPa/h) as portrayed in the plot (a) of Fig. 1.
Inserting p(t) defined as (4.10) into (4.4) (respectively, (4.5)) we obtain \(e^\textrm{e}(t)\). Then, the discrete solution \(e^\textrm{v}_M(t)\) to the corresponding equation is computed with the help of the quadrature formulas (4.7) and (4.9) on the uniform mesh with time-step \(\delta t =0.02\) as \(M=75\). The quasi-linear dilatation for the viscoelastic response is portrayed versus time in Fig. 1 in the plot (b) for the response (4.5), and in plot (c) for the thresholded equation (4.4). Blue lines indicate the part of \({\underline{M}}\le (1-3\lambda _3 p) \le {\overline{M}}\), red lines indicate the part of \((1-3\lambda _3 p)<{\underline{M}}\).
We observe that approaching the critical pressure \(p_{\textrm{cr}}:=1/(3 \lambda _3)\) the right-hand side of equation (4.5) becomes unbounded. The rate g in (4.10) is chosen such that the critical pressure \(p_{\textrm{cr}} =3.{\overline{3}}\) (GPa) is attained in time \(t=1\) hour as seen in Fig. 1 in plot (a). Therefore, approaching from the left \(t =1\), which is marked by the vertical dashed line, the solution \(e^\textrm{v}_M(t)\) to (4.5) marked by the solid line in the plot (b) blows up to minus infinity. Whereas this singularity is avoided within the equation (4.4) with thresholding, the solution \(e^\textrm{v}_M(t)\) continues after \(t=1\) as indicated by the marked dashed line in the plot (c). Both the solutions coincide for \(t\in [0,0.8]\) (h) and are distinguished only after reaching the critical pressure \(p_\textrm{cr}(1-{\underline{M}}) =2.{\overline{6}}\) (GPa), which is marked by the dotted lines in Fig. 1.
4.2 Creep test
In order to simulate the creep behavior, in the second test, the loading undergoes the three stages portrayed in the plot (a) of Fig. 2. Within \(t\in [0,0.9]\) (h), the pressure increases from zero linearly with the constant rate \(g=3.{\overline{3}}\) (GPa/h) as in (4.10). After reaching \(0.9 g =3\) (GPa/h) at \(t=0.9\) before it reaches \(p_{\textrm{cr}}\), the pressure is maintained within the time interval \(t\in [0.9,2]\), then immediately removed and kept zero for \(t\in [2,3]\), that is
The quasi-linear solutions \(e^\textrm{v}_M(t)\) to the discrete Volterra convolution equation (4.7) and (4.9) with the uniform time-step \(\delta t =0.02\) are portrayed versus time in Fig. 2 in the plot (b) corresponding to the equation (4.5) without thresholding, and in the plot (c) to the equation (4.4) with thresholding. The responses are marked by the solid line for monotone loading, dashed line while the pressure is maintained constant, and dash-dotted line on the removal of the pressure. Here the variation of the solution to the equation that is thresholded is moderate compared to the one that is not thresholded.
Data availability
No datasets were generated or analyzed during the current study.
References
Boltzmann, L.: Zur Theorie der elastischen Nachwirkung. Sitzungsberichte der kaiserlichen Akademie der Wissenschaften Wien, Mathematisch-Naturwissenschaftliche Classe 70, Bd.II (1874), 275–306
Burgers, J.M.: Mechanical considerations—model systems—phenomenological theories of relaxation and viscosity. In: Burgers, J.M. (ed.) First Report on Viscosity and Plasticity, pp. 5–67 (Chapter 1). Nordemann Publishing, New York (1939)
Browder, F.E.: Nonlinear maximal monotone operators in Banach space. Math. Annalen 175, 89–113 (1968)
Bulíček, M., Gwiazda, P., Málek, J., Rajagopal, K.R., Świerczewska-Gwiazda, A.: On flows of fluids described by an implicit constitutive equation characterized by a maximal monotone graph. In: Robinson, J., Rodrigo, J., Sadowski, W. (eds.) Mathematical Aspects of Fluid Mechanics, pp. 23–51. Cambridge University Press (2012)
Bulíček, M., Málek, J., Süli, E.: Existence of global weak solutions to implicitly constituted kinetic models of incompressible homogeneous dilute polymers. Commun. Part. Differ. Equ. 38, 882–924 (2013)
Bulíček, M., Patel, V., Şengül, Y., Süli, E.: Existence of large-data global weak solutions to a model of a strain-limiting viscoelastic body. Commun. Pure Appl. Anal. 20, 1931–1960 (2021)
Fung, Y.C.: Biomechanics: Mechanical Properties of Living Tissues. Springer, New York (1981)
Green, A.E., Rivlin, R.S.: The mechanics of non-linear materials with memory. Part III. Arch. Rational Mech. Anal. 4, 387–404 (1959)
Gwinner, J., Jadamba, B., Khan, A.A., Raciti, F.: Uncertainty Quantification in Variational Inequalities. Chapman and Hall/CRC, Boca Raton (2021)
Itou, H., Kovtunenko, V.A., Rajagopal, K.R.: On the states of stress and strain adjacent to a crack in a strain-limiting viscoelastic body. Math. Mech. Solids 23, 433–444 (2018)
Itou, H., Kovtunenko, V.A., Rajagopal, K.R.: Crack problem within the context of implicitly constituted quasi-linear viscoelasticity. Math. Models Methods Appl. Sci. 29, 355–372 (2019)
Itou, H., Kovtunenko, V.A., Rajagopal, K.R.: The Boussinesq flat-punch indentation problem within the context of linearized viscoelasticity. Int. J. Eng. Sci. 151, 103272 (2020)
Itou, H., Kovtunenko, V.A., Rajagopal, K.R.: Lagrange multiplier approach to unilateral indentation problems: well-posedness and application to linearized viscoelasticity with non-invertible constitutive response. Math. Models Methods Appl. Sci. 31, 649–674 (2021)
Itou, H., Kovtunenko, V.A., Rajagopal, K.R.: On an implicit model linear in both stress and strain to describe the response of porous solids. J. Elasticity 144, 649–674 (2021)
Itou, H., Kovtunenko, V.A., Rajagopal, K.R.: Investigation of implicit constitutive relations in which both the stress and strain appear linearly, adjacent to non-penetrating cracks. Math. Models Methods Appl. Sci. 32, 1475–1492 (2022)
Itou, H., Kovtunenko, V.A., Rajagopal, K.R.: A generalization of the Kelvin-Voigt model with pressure-dependent moduli in which both stress and strain appear linearly. Math. Meth. Appl. Sci. 46, 14765–15918 (2023)
Itou, H., Kovtunenko, V.A., Rudoy, E.M.: Three-field mixed formulation of elasticity model nonlinear in the mean normal stress for the problem of non-penetrating cracks in bodies. Appl. Eng. Sci. 7, 100060 (2021)
Itou, H., Kovtunenko, V.A., Tani, A.: The interface crack with Coulomb friction between two bonded dissimilar elastic media. Appl. Math. 56, 69–97 (2011)
Itou, H., Tani, A.: Existence of a weak solution in an infinite viscoelastic strip with a semi-infinite crack. Math. Models Methods Appl. Sci. 14, 975–986 (2004)
Khludnev, A.M.: Contact viscoelastoplastic problem for a beam. In: Antontsev, S.N., Hoffmann, K.-H., Khludnev, A.M. (eds.) Problems in Continuum Mechanics, pp. 159–166. Birkhäuser, Basel (1992)
Khludnev, A.M., Kovtunenko, V.A.: Analysis of Cracks in Solids. WIT-Press, Southampton, Boston (2000)
Kovtunenko, V.A.: A hemivariational inequality in crack problems. Optimization 60, 1071–1089 (2011)
Lockett, F.J.: Non-linear Viscoelastic Solids. Academic Press, London (1972)
Maxwell, J.C.: On the dynamical theory of gases. Proc. R. Soc. A: Math. Phys. Eng. Sci. 157, 26–78 (1867)
Minty, G.J.: Monotone (nonlinear) operators in Hilbert space. Duke Math. J. 29, 341–346 (1962)
Muliana, A., Rajagopal, K.R., Wineman, A.S.: A new class of quasi-linear models for describing the nonlinear viscoelastic response of materials. Acta Mech. 224, 2169–2183 (2013)
Murru, P.T., Torrence, C., Grasley, Z., Rajagopal, K.R., Garboczi, E.: Density-driven damage mechanics (D3-M) model for concrete II: fully coupled chemo-mechanical damage. Int. J. Pavement Eng. 23, 1175–1185 (2022)
Oldroyd, J.G.: On the formulation of rheological equations of state. Proc. R. Soc. A: Math. Phys. Eng. Sci. 200(1063), 523–541 (1950)
Pipkin, A.C., Rogers, T.G.: A non-linear integral representation for viscoelastic behaviour. J. Mech. Phys. Solids 16, 59–72 (1968)
Popova, T.S.: Problems of thin inclusions in a two-dimensional viscoelastic body. J. Appl. Ind. Math. 12, 313–324 (2018)
Průša, V., Rajagopal, K.R.: On implicit constitutive relations for materials with fading memory. J. Non-Newton. Fluid Mech. 181–182, 22–29 (2012)
Rajagopal, K.R.: On implicit constitutive relations. Appl. Math. 48, 279–319 (2003)
Rajagopal, K.R.: Remarks on the notion of “pressure”. Int. J. Nonlinear Mech. 71, 165–172 (2015)
Rajagopal, K.R.: An implicit constitutive relation for describing the small strain response of porous elastic solids whose material moduli are dependent on the density. Math. Mech. Solids 26, 1138–1146 (2021)
Rajagopal, K.R., Wineman, A.S.: A note on viscoelastic bodies whose material properties depend on the density. Math. Mech. Solids 26, 1726–1731 (2021)
Sofonea, M., Migórski, S.: Variational-Hemivariational Inequalities with Applications. Chapman and Hall/CRC, Boca Raton (2017)
Acknowledgements
H. Itou is partially supported by JSPS KAKENHI Grant Number JP23KK0049. V.A. Kovtunenko thanks the University of Graz for support. K.R. Rajagopal thanks the Office of Naval Research and the National Science Foundation for their support.
Funding
Open access funding provided by University of Graz.
Author information
Authors and Affiliations
Contributions
The authors contributed equally to this work.
Corresponding author
Ethics declarations
Conflict of interest
The authors declare no competing interests.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Itou, H., Kovtunenko, V.A. & Rajagopal, K.R. Well-posedness of the governing equations for a quasi-linear viscoelastic model with pressure-dependent moduli in which both stress and strain appear linearly. Z. Angew. Math. Phys. 75, 22 (2024). https://doi.org/10.1007/s00033-023-02160-0
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s00033-023-02160-0
Keywords
- Viscoelasticity
- Implicit constitutive response
- Creep behavior
- Volterra convolution operator
- Mixed variational problem
- Thresholding
- Maximal monotone graph