betaFeedback

WorksPhysics and Astronomy

Piecewise linear constitutive relations for stretch-limited elastic strings

Abstract

This study proposes a simple and novel class of stretch-limiting constitutive relations for perfectly flexible elastic strings drawing from modern advances in constitutive theory for elastic bodies. We investigate strings governed by constitutive relations where stretch is a bounded, piecewise linear function of tension, extending beyond the traditional Cauchy elasticity framework. Our analysis includes explicit solutions for both catenaries and longitudinal, piecewise constant stretched motions.

Introduction

In the past 20 years, the term “elastic” has experienced a significant evolution, expanding its definition and scope. Unlike Cauchy elasticity [6, 7], wherein the stress is given explicitly as a function of the deformation gradient, or Green elasticity [14] where one assumes the existence of a stored energy in terms of the deformation gradient, the second listed author [20] suggested an implicit relationship between the density, the stress, and the deformation gradient. The second listed author and co-workers have shown that the ambit of elastic bodies, namely bodies incapable of dissipation (converting mechanical working into heat), is far larger than that of Cauchy elastic bodies and includes bodies described by such implicit constitutive relations which do not belong to the class of Cauchy elastic bodies (see [27, 29, 30] [4, 5, 20, 21, 22]).

A special sub-class of such implicit elastic bodies are isotropic elastic bodies wherein the right Cauchy-Green tensor is a function of the stress and density (or isotropic bodies wherein the left Cauchy-Green tensor is a function of the stress and density). When displacement gradients are restricted to be small, we obtain the linearized strain as a function of the stress and the density. In general, this relationship between the linearized strain and the stress can be nonlinear, a sharp contrast to classical constitutive theory that only produces linear relations between linearized strain and stress. Recent studies expanding these theories into viscoelastic and dispersive settings can be found in, e.g., [33, 34, 35] [9, 10, 11, 12].

This paper focuses on the physically important but simpler setting in which classical and modern advances in constitutive theory for elastic bodies can be explored: perfectly flexible strings. Interest in the study of the mechanics of strings and fibers has a storied history; it attracted the attention of da Vinci, Galileo, Euler, the Bernoullis, Savart, Lagrange, Leibniz, Riccati, Huygens, Navier, Poisson, Lamé and other distinguished scientists. The interested reader can find a detailed account of the history of investigations into the mechanics of strings/fibers/wires in the work of Todhunter [36]. All of these early studies were within the context of special theories and not within the purview of general nonlinear theories of elasticity. Studies within the context of nonlinear theories are too numerous to mention and one can find copious references to both recent and older studies mentioned above in the treatise by Antman [1].

K. R. Rajagopal and C. Rodriguez gratefully acknowledge support from NSF DMS-2307562. For a perfectly flexible string, a constitutive relation is specified using only two scalar quantities. The frame indifferent kinematic variable of a string is the stretch ν=∣∂sr∣\nu= \lvert\partial_s \mathbf{r} \rvert of the curve in space,

s∋[0,L]↦r(s,t),s \ni[0,L] \mapsto\mathbf{r}(s,t),

defining its configuration at time tt. The contact force n\mathbf{n} is assumed to be proportional to the unit tangent to the curve defining the configuration of the string, n=Nν−1∂sr\mathbf{n} = N\nu^{-1}\partial_s\mathbf{r}. Thus, there is only one stress variable, the tension NN. Classical constitutive relations take the form

N=N^(ν),(1)N = \widehat{N}(\nu), \tag*{(1)}

in keeping with the traditional approach of Cauchy elasticity wherein one specifies the stress variable in terms of the deformation gradient, a kinematic variable (see § 2 for more details).

Recent works have re-examined the mechanics of perfectly flexible strings within the modern framework introduced in [20]. The works [31, 32] initiated the study of elastic stretch-limited strings satisfying constitutive relations of the form

ν=ν^(N),\nu= \widehat{\nu}(N),

where ν^:R→[ν0,ν1]\widehat{\nu} : \mathbb{R} \to[\nu_0,\nu_1] is a bounded, non-decreasing, function, and there exist N0,N1>0N_0,N_1 > 0 such that

ν^(N)={ν0if N<−N0,ν1if N>N1.\widehat{\nu}(N) = \begin{cases} \nu_0 & \text{if } N < -N_0,\\ \nu_1 & \text{if } N > N_1. \end{cases}

In [31], the author established results on the existence and multiplicity of equilibria for stretch-limited catenaries, strings suspended between two points, a problem that attracted the interests of the likes of Galileo, Bernoulli, Huygens, Leibniz, among others. He found that the number of solutions for such stretch-limited strings deviates significantly from the case of traditional classical elastic strings satisfying (1). The work [32] studied the propagation of longitudinal waves in a semi-infinite stretch-limited string with the finite end held fixed and the other end asymptotically in tension. There, he proved the orbital asymptotic stability of an explicit two-parameter family of piecewise constant stretched motions.

The latter two authors recently explored Cosserat rods within frameworks arising from modern advances in constitutive theory, considering both elasticity and viscoelasticity. In [25], they studied Cosserat rods satisfying strain-limiting constitutive relations between the rod’s strains and stress relevant to modeling biological fibers such as DNA. They derived explicit equilibrium states for such rods that exhibited a variety of rich characteristics including shearing instabilities and Poynting effects. In [24] they studied the response of an inextensible, unshearable, viscoelastic rod that moves on a plane and possesses an evolving natural configuration characterized by its curvature (see [19] for a discussion of the notion of natural configuration, a concept studied in detail earlier by Eckart [8]; see also [28] for the connection between the evolution of the natural configuration and the rate of entropy production). They established global well-posedness and asymptotic dynamics for solutions of the governing quasi-static equations of motion. The model studied in [24] was extended in [26] to encompass Cosserat rods moving in three-dimensional space that possess rate-dependent evolving natural configurations. In an interesting departure from the usual imposition of the second law of thermodynamics in a point-wise strong sense, the work [26] enforces a weaker requirement, namely that the Clausius-Duhem inequality hold over the whole body in an integral sense. They showed that one can have two possible constitutive relations, one that meets the stricter point-wise inequality and another that meets the global integral inequality.

In this work, we introduce a simple class of stretch-limiting, piecewise linear constitutive relations for elastic strings of the form ν=ν^(N)\nu=\hat{\nu}(N). The organization of the paper is as follows. In § 2, we briefly review the kinematics, balance laws and constitutive theory governing perfectly flexible elastic strings. In § 3, we present a simple form of ν^(N)−1\hat{\nu}(N)-1 that limits the amount of compression and extension the string can undergo, is linear in NN when the string is extensible, and has different tangent moduli (i.e., proportionality constants) for extension versus compression (see (3.1)). In § 4, we formulate simple boundary value problems, which are then solved from a semi-inverse perspective in § 5 (for catenaries) and § 6 (for piecewise constant stretched motions). Finally, in § 7, we discuss potential applications of the proposed model.

The problems solved in § 5 and § 6 illustrate key differences between the present work and previous studies while also offering potential applications. § 5 provides explicit solutions for catenaries, unlike the more abstract results on the existence and multiplicity of solutions obtained in [31]. Similarly, the explicit piecewise constant stretch motions obtained in § 6 involve a compressed segment and an elongated segment with distinct tangent moduli, whereas the solutions in [32] involve a maximally stretched segment and an elongated but still flexible segment. One application of the solutions in § 5 is in modeling quasi-static deformations of stretch-limited strings, which may be relevant in civil engineering. In such cases, allowing for some elasticity while imposing a stretch limit offers insight into how these strings behave differently from the standard inextensible assumption often used in structural analysis (see, for example, Section 5/8 of [17]). The solutions in § 5 and § 6 also serve as useful benchmarks for numerical methods, which are essential for studying more realistic problems involving these new constitutive theories. Future work will explore some of these numerical methods. In summary, the solutions obtained in this work offer new concrete perspectives on stretch-limited strings and open the door to further analytical and computational developments.

Perfectly flexible strings

In this section, we provide a brief summary of the kinematics and balance laws relevant to the modeling of perfectly flexible strings. A careful and detailed description of the same can be found in the treatise [1].

Kinematics

2.1. Kinematics. Let E3\mathbb{E}^{3} be three-dimensional Euclidean space. We identify its associated translation space with R3\mathbb{R}^{3} via a fixed, right-handed, orthonormal basis {i,j,k}\{\mathbf{i},\mathbf{j},\mathbf{k}\}.

Let L>0L>0 be the reference length of the string which we assume coincides with the undeformed length of the string. We parameterize the material points of the string by the interval [0,L][0,L]. The current configuration of the string at time tt is parameterized by a curve

[0,L]∋s↦r(s,t)∈E3.[0,L]\ni s\mapsto\mathbf{r}(s,t)\in\mathbb{E}^{3}.

The tangent to the curve r(⋅,t)\mathbf{r}(\cdot,t) at ss is rs(s,t):=∂sr(s,t)\mathbf{r}_{s}(s,t):=\partial_{s}\mathbf{r}(s,t), and the velocity of the material point ss is rt(s,t):=∂tr(s,t)\mathbf{r}_{t}(s,t):=\partial_{t}\mathbf{r}(s,t). The stretch ν(s,t)\nu(s,t) of the string at (s,t)(s,t) is

ν(s,t):=∣rs(s,t)∣.\nu(s,t):=\lvert\mathbf{r}_{s}(s,t)\rvert.

We require that the configuration is regular so that ν>0\nu>0 throughout the motion of the string.

Balance laws and constitutive theory

The classical equations of motion expressing local balance of linear momentum for a string are given by:

(ρA)(s)rtt(s,t)=ns(s,t)+f(s,t),(s,t)∈I×[0,T].(\rho A)(s)r_{tt}(s,t)=n_s(s,t)+f(s,t),\qquad(s,t)\in I\times[0,T].

Here (ρA)(s)(\rho A)(s) is the mass per unit reference length at ss, f(s,t)f(s,t) is an external body force per unit reference length at (s,t)(s,t), and n(s,t)n(s,t) is the contact force at (s,t)(s,t).

The defining characteristic of a perfectly flexible string is that the contact force is proportional to the unit tangent of the string’s configuration: there exists a scalar-valued function N(s,t)N(s,t), the tension, such that

n(s,t)=N(s,t)rs(s,t)∣rs(s,t)∣.(2)n(s,t)=N(s,t)\frac{r_s(s,t)}{|r_s(s,t)|}. \tag*{(2)}

We note that NN satisfies

N(s,t)=sgn⁡(rs(s,t)⋅n(s,t))∣n(s,t)∣.(3)N(s,t)=\operatorname{sgn}(r_s(s,t)\cdot n(s,t))|n(s,t)|. \tag*{(3)}

The mechanical properties of the string are modeled by specifying a relation between the stretch ν\nu and tension NN. Classically, a string is said to be elastic if there exists a function N^(ν,s)\widehat{N}(\nu,s) such that

N(s,t)=N^(ν(s,t),s),N(s,t)=\widehat{N}(\nu(s,t),s),

with N^(1,s)=0\widehat{N}(1,s)=0, ν↦N^(ν,s)\nu\mapsto\widehat{N}(\nu,s) increasing, lim⁡ν→0N^(ν,s)=−∞\lim_{\nu\to0}\widehat{N}(\nu,s)=-\infty and lim⁡ν→∞N^(ν,s)=∞\lim_{\nu\to\infty}\widehat{N}(\nu,s)=\infty for all ss. These mathematical restrictions on N^\widehat{N} reflect the physical assumptions:

  • A configuration of the string is natural (N=0N=0) if and only if it is unstretched (ν=1\nu=1).

  • An increase in tension coincides with an increase in stretch.

  • Total compression (ν=0\nu=0) requires infinite compressive force and infinite stretch requires infinite tensile force.

The previous mathematical assumptions imply that N^(⋅,s)\widehat{N}(\cdot,s) has an inverse function ν^(⋅,s)\widehat{\nu}(\cdot,s), and the constitutive relation between stretch and tension can be recast as

ν(s,t)=ν^(N(s,t),s).\nu(s,t)=\widehat{\nu}(N(s,t),s).

A new constitutive relation for elastic strings

In this section we propose a new constitutive relation for perfectly flexible elastic strings of the form:

ν=ν^(N),\nu=\widehat{\nu}(N),

where we have suppressed the dependence of the variables on ss and tt. As discussed in the previous section, ν>0\nu>0 is the stretch and N∈RN\in\mathbb{R} is the tension of the string. We will impose the following weaker restrictions than in the classical setting:

  • A configuration of the string is natural (N=0N=0) if and only if it is unstretched (ν=1\nu=1).

  • Tensile forces (N>0N>0) correspond to extended configurations (ν>1\nu>1).

  • Compressive forces (N<0N<0) correspond to compressed configurations (0<ν<10<\nu<1).

A piecewise linear constitutive relation

3.1. A piecewise linear constitutive relation. From the physical point of view, it is very reasonable that there are limits to the amount of compression and extension that a real string can endure before failure. In Figure 1 on the left, we have a schematic depiction for ν−1\nu-1 versus NN, while on the right we have a simplified model for ν^−1\hat{\nu}-1 for the approximation of the above behavior.

Figure 1. Stretch limiting model for the string.

The simplified constitutive relation that we propose in this work is the piecewise differentiable model:

ν^(N)−1={ν1−1if N>N1,αTNif 0≤N≤N1,αCNif −N0≤N<0,ν0−1if N<−N0.(4)\hat{\nu}(N)-1 = \begin{cases} \nu_1-1 & \text{if } N>N_1,\\ \alpha_TN & \text{if } 0\le N\le N_1,\\ \alpha_CN & \text{if } -N_0\le N<0,\\ \nu_0-1 & \text{if } N<-N_0. \end{cases} \tag*{(4)}

where N0N_0, N1N_1, and 0<ν0<1<ν10<\nu_0<1<\nu_1 are fixed positive constants. If N>N1N>N_1, then the extension of the string has reached its limit ν1\nu_1, while if N<−N0N<-N_0, compression has reached its limit ν0\nu_0. When 0≤N≤N10\le N\le N_1 (respectively, −N0≤N<0-N_0\le N<0), extension (respectively, compression) and tension are proportional. We assume that αT<αC\alpha_T<\alpha_C so that the string can deform more with relatively small internal load when in compression. In particular, strings modeled by (3.1) have different tangent moduli for extension versus compression.

It is important to note that the tension is not a function of the stretch, however, the stretch is a function of the tension. Thus, the constitutive relation under consideration does not belong to the class of classical Cauchy elastic bodies.

Boundary value problems

In this section, we set-up a boundary value problem for a uniform string satisfying the constitutive relation (3.1). For certain calculations, we will use the alternative expression for the constitutive relation (2.1):

rs=ν^(N)Nn.(5)\mathbf{r}_s=\frac{\hat{\nu}(N)}{N}\mathbf{n}. \tag*{(5)}

We seek to solve the governing field equations

(ρA)rtt(s,t)=ns(s,t)+f(s,t),∣rs∣=ν=ν^(N),n=Nrs∣rs∣(6)(\rho A)\mathbf{r}_{tt}(s,t)=\mathbf{n}_s(s,t)+\mathbf{f}(s,t),\qquad|\mathbf{r}_s|=\nu=\hat{\nu}(N),\qquad\mathbf{n}=N\frac{\mathbf{r}_s}{|\mathbf{r}_s|} \tag*{(6)}

for r(s,t)\mathbf{r}(s,t) and n(s,t)\mathbf{n}(s,t) where f\mathbf{f} is a given external body force and (ρA)(\rho A) is a constant. For the above equations, we impose boundary and initial conditions

r(0,t)=0,r(L,t)=Lk,r(s,0)=u(s),rt(s,0)=v(s),(7)\mathbf{r}(0,t)=\mathbf{0},\qquad\mathbf{r}(L,t)=L\mathbf{k},\qquad\mathbf{r}(s,0)=\mathbf{u}(s),\qquad\mathbf{r}_t(s,0)=\mathbf{v}(s), \tag*{(7)}

where

u(0)=0,u(L)=Lk,v(0)=0,v(L)=0.(8)\mathbf{u}(0)=\mathbf{0},\qquad\mathbf{u}(L)=L\mathbf{k},\qquad\mathbf{v}(0)=\mathbf{0},\qquad\mathbf{v}(L)=\mathbf{0}. \tag*{(8)}

Alternative forms of the field equations

In this section, we assume that N≠0N\ne0. If N>0N>0, then N=∣n∣N=|\mathbf{n}|, and (5) and (3) imply that

rs=ν^(∣n∣)n∣n∣.\mathbf{r}_s=\hat{\nu}(|\mathbf{n}|)\frac{\mathbf{n}}{|\mathbf{n}|}.

Thus, by (7),

r(s,t)=∫0sν^(∣n(ξ,t)∣)n(ξ,t)∣n(ξ,t)∣ dξ.(9)\mathbf{r}(s,t)=\int_0^s\hat{\nu}(|\mathbf{n}(\xi,t)|)\frac{\mathbf{n}(\xi,t)}{|\mathbf{n}(\xi,t)|}\,\mathrm{d}\xi. \tag*{(9)}

Inserting (9) into (6), we obtain

(ρA)∂2∂t2{∫0sν^(∣n(ξ,t)∣)n(ξ,t)∣n(ξ,t)∣ dξ}=∂n∂s(s,t)+f(s,t).(10)(\rho A)\frac{\partial^2}{\partial t^2}\left\{\int_0^s\hat{\nu}(|\mathbf{n}(\xi,t)|)\frac{\mathbf{n}(\xi,t)}{|\mathbf{n}(\xi,t)|}\,\mathrm{d}\xi\right\}=\frac{\partial\mathbf{n}}{\partial s}(s,t)+\mathbf{f}(s,t). \tag*{(10)}

In the case that N<0N<0, the only modification we need to consider to the above equation is to use ν^(−∣n(ξ,t)∣)\hat{\nu}(-|\mathbf{n}(\xi,t)|) instead of ν^(∣n(ξ,t)∣)\hat{\nu}(|\mathbf{n}(\xi,t)|).

By (9), the boundary condition {(4.3)}1_1 is satisfied automatically, and {(4.3)}2_2 becomes

∫0Lν^(∣n(ξ,t)∣)n(ξ,t)∣n(ξ,t)∣ dξ=Lk.\int_0^L\hat{\nu}(|\mathbf{n}(\xi,t)|)\frac{\mathbf{n}(\xi,t)}{|\mathbf{n}(\xi,t)|}\,\mathrm{d}\xi=L\mathbf{k}.

The initial conditions {(4.3)}3,4_{3,4} become

∫0sν^(∣n(ξ,0)∣)n(ξ,0)∣n(ξ,0)∣ dξ=u(s),∂∂t[∫0sν^(∣n(ξ,t)∣)n(ξ,t)∣n(ξ,t)∣ dξ]∣t=0=v(s).\int_0^s\hat{\nu}(|\mathbf{n}(\xi,0)|)\frac{\mathbf{n}(\xi,0)}{|\mathbf{n}(\xi,0)|}\,\mathrm{d}\xi=\mathbf{u}(s),\qquad \left.\frac{\partial}{\partial t}\left[\int_0^s\hat{\nu}(|\mathbf{n}(\xi,t)|)\frac{\mathbf{n}(\xi,t)}{|\mathbf{n}(\xi,t)|}\,\mathrm{d}\xi\right]\right|_{t=0}=\mathbf{v}(s).

Dimensionless equations

We define the following dimensionless variables

sˉ=sL,tˉ=gLt,nˉ=αTn,rˉ=1Lr,ζ=(ρA)gLαT,fˉ=LαTf,Nˉ=αTN,\bar{s}=\frac{s}{L},\qquad\bar{t}=\sqrt{\frac{g}{L}}t,\qquad\bar{\mathbf{n}}=\alpha_T\mathbf{n},\qquad\bar{\mathbf{r}}=\frac{1}{L}\mathbf{r},\qquad\zeta=(\rho A)gL\alpha_T,\qquad\bar{\mathbf{f}}=L\alpha_T\mathbf{f},\qquad\bar{N}=\alpha_TN,

where gg is the gravitational constant. Then (6) and (3) become

ζ∂2rˉ∂tˉ2=∂nˉ∂sˉ+fˉ,(11)\zeta\frac{\partial^2\bar{\mathbf{r}}}{\partial\bar{t}^2}=\frac{\partial\bar{\mathbf{n}}}{\partial\bar{s}}+\bar{\mathbf{f}}, \tag*{(11)}
∣rˉsˉ∣=ν^(Nˉ),nˉ=Nˉrˉsˉ∣rˉsˉ∣,Nˉ:=sgn⁡(rˉsˉ⋅nˉ)∣nˉ∣,(12)|\bar{\mathbf{r}}_{\bar{s}}|=\hat{\nu}(\bar{N}),\qquad\bar{\mathbf{n}}=\bar{N}\frac{\bar{\mathbf{r}}_{\bar{s}}}{|\bar{\mathbf{r}}_{\bar{s}}|},\qquad\bar{N}:=\operatorname{sgn}(\bar{\mathbf{r}}_{\bar{s}}\cdot\bar{\mathbf{n}})|\bar{\mathbf{n}}|, \tag*{(12)}

where rˉsˉ=∂rˉ∂sˉ\bar{\mathbf{r}}_{\bar{s}}=\frac{\partial\bar{\mathbf{r}}}{\partial\bar{s}}. The boundary and initial conditions {(4.3)}, {(4.4)} are

rˉ(0,tˉ)=0,rˉ(1,tˉ)=k,rˉ(sˉ,0)=uˉ,rˉtˉ(sˉ,0)=vˉ,(13)\bar{\mathbf{r}}(0,\bar{t})=\mathbf{0},\qquad\bar{\mathbf{r}}(1,\bar{t})=\mathbf{k},\qquad\bar{\mathbf{r}}(\bar{s},0)=\bar{\mathbf{u}},\qquad\bar{\mathbf{r}}_{\bar{t}}(\bar{s},0)=\bar{\mathbf{v}}, \tag*{(13)}

where

uˉ=1Lu,vˉ=1Lgv\bar{\mathbf{u}}=\frac{1}{L}\mathbf{u},\qquad\bar{\mathbf{v}}=\frac{1}{\sqrt{Lg}}\mathbf{v}

and rˉtˉ=∂rˉ∂tˉ\bar{\mathbf{r}}_{\bar{t}}=\frac{\partial\bar{\mathbf{r}}}{\partial\bar{t}}.

In the case of the alternative representation (9) we obtain (in the case N>0N>0)

rˉ(sˉ,tˉ)=∫0sˉν^(∣nˉ(ξˉ,tˉ)∣)nˉ(ξˉ,tˉ)∣nˉ(ξˉ,tˉ)∣ dξˉ,\bar{\boldsymbol{r}}(\bar{s},\bar{t})=\int_{0}^{\bar{s}}\hat{\nu}(|\bar{\boldsymbol{n}}(\bar{\xi},\bar{t})|)\frac{\bar{\boldsymbol{n}}(\bar{\xi},\bar{t})}{|\bar{\boldsymbol{n}}(\bar{\xi},\bar{t})|}\,\mathrm{d}\bar{\xi},

and from (10) we have

C∂2∂tˉ2[∫0sˉν^(∣nˉ(ξˉ,tˉ)∣)nˉ(ξˉ,tˉ)∣nˉ(ξˉ,tˉ)∣ dξˉ]=∂nˉ∂sˉ+fˉ.\mathcal{C}\frac{\partial^{2}}{\partial\bar{t}^{2}}\left[\int_{0}^{\bar{s}}\hat{\nu}(|\bar{\boldsymbol{n}}(\bar{\xi},\bar{t})|)\frac{\bar{\boldsymbol{n}}(\bar{\xi},\bar{t})}{|\bar{\boldsymbol{n}}(\bar{\xi},\bar{t})|}\,\mathrm{d}\bar{\xi}\right]=\frac{\partial\bar{\boldsymbol{n}}}{\partial\bar{s}}+\bar{\boldsymbol{f}}.

The boundary condition (13)2_2 becomes

∫01ν^(∣nˉ(ξˉ,tˉ)∣)nˉ(ξˉ,tˉ)∣nˉ(ξˉ,tˉ)∣ dξˉ=k,\int_{0}^{1}\hat{\nu}(|\bar{\boldsymbol{n}}(\bar{\xi},\bar{t})|)\frac{\bar{\boldsymbol{n}}(\bar{\xi},\bar{t})}{|\bar{\boldsymbol{n}}(\bar{\xi},\bar{t})|}\,\mathrm{d}\bar{\xi}=\boldsymbol{k},

while for the initial conditions (13)3,4_{3,4}, we have

∫0sˉν^(∣nˉ(ξˉ,0)∣)nˉ(ξˉ,0)∣nˉ(ξˉ,0)∣ dξˉ=uˉ(sˉ),\int_{0}^{\bar{s}}\hat{\nu}(|\bar{\boldsymbol{n}}(\bar{\xi},0)|)\frac{\bar{\boldsymbol{n}}(\bar{\xi},0)}{|\bar{\boldsymbol{n}}(\bar{\xi},0)|}\,\mathrm{d}\bar{\xi}=\bar{\boldsymbol{u}}(\bar{s}),
∂∂tˉ[∫0sˉν^(∣nˉ(ξˉ,tˉ)∣)nˉ(ξˉ,tˉ)∣nˉ(ξˉ,tˉ)∣ dξˉ]∣tˉ=0=vˉ(sˉ).\left.\frac{\partial}{\partial\bar{t}}\left[\int_{0}^{\bar{s}}\hat{\nu}(|\bar{\boldsymbol{n}}(\bar{\xi},\bar{t})|)\frac{\bar{\boldsymbol{n}}(\bar{\xi},\bar{t})}{|\bar{\boldsymbol{n}}(\bar{\xi},\bar{t})|}\,\mathrm{d}\bar{\xi}\right]\right|_{\bar{t}=0}=\bar{\boldsymbol{v}}(\bar{s}).

Finally, our model (4) becomes

ν^(Nˉ)−1={ν1−1if Nˉ>Nˉ1,NˉNˉif 0≤Nˉ≤Nˉ1,αCαTNˉif −Nˉ0≤Nˉ<0,ν0−1if Nˉ<−Nˉ0.(14)\hat{\nu}(\bar{N})-1= \begin{cases} \nu_{1}-1 & \text{if }\bar{N}>\bar{N}_{1},\\ \frac{\bar{N}}{\bar{N}} & \text{if }0\leq\bar{N}\leq\bar{N}_{1},\\ \frac{\alpha_C}{\alpha_T}\bar{N} & \text{if }-\bar{N}_{0}\leq\bar{N}<0,\\ \nu_{0}-1 & \text{if }\bar{N}<-\bar{N}_{0}. \end{cases} \tag*{(14)}

where Nˉ1=αTN1\bar{N}_{1}=\alpha_TN_{1} and Nˉ0=αTN0\bar{N}_{0}=\alpha_TN_{0}.

Going forward we work with the dimensionless variables and equations derived in this section. To decrease the amount of necessary notation, we will drop the overbars.

Catenaries

In this section, we assume that the different variables do not depend on time and obtain solutions to the boundary value problem from the semi-inverse perspective.

Since the variables are time independent, rtt=0r_{tt}=0 and rt=0r_t=0. From (11), we obtain ns=−fn_s=-f, and as a result,1

n=−∫0sf(ξ) dξ+n0,(15)\boldsymbol{n}=-\int_{0}^{s}\boldsymbol{f}(\xi)\,\mathrm{d}\xi+\boldsymbol{n}_{0}, \tag*{(15)}

where n0\boldsymbol{n}_{0} is a constant dimensionless vector. Inserting (15) into (12) and assuming that for 0≤s≤10\leq s\leq1 we have rs⋅n≥0\boldsymbol{r}_s\cdot\boldsymbol{n}\geq0, we obtain

rs(s)=ν^(∣n0−∫0sf(ξ) dξ∣)∣n0−∫0sf(ξ) dξ∣{n0−∫0sf(ξ) dξ}.\boldsymbol{r}_s(s)=\frac{\hat{\nu}\left(\left|\boldsymbol{n}_{0}-\int_{0}^{s}\boldsymbol{f}(\xi)\,\mathrm{d}\xi\right|\right)}{\left|\boldsymbol{n}_{0}-\int_{0}^{s}\boldsymbol{f}(\xi)\,\mathrm{d}\xi\right|}\left\{\boldsymbol{n}_{0}-\int_{0}^{s}\boldsymbol{f}(\xi)\,\mathrm{d}\xi\right\}.

As a result, we obtain

r(s)=∫0sν^(∣n0−∫0ξf(η) dη∣)∣n0−∫0ξf(η) dη∣{n0−∫0ξf(η) dη} dξ+r0.(16)\boldsymbol{r}(s)=\int_{0}^{s}\frac{\hat{\nu}\left(\left|\boldsymbol{n}_{0}-\int_{0}^{\xi}\boldsymbol{f}(\eta)\,\mathrm{d}\eta\right|\right)}{\left|\boldsymbol{n}_{0}-\int_{0}^{\xi}\boldsymbol{f}(\eta)\,\mathrm{d}\eta\right|}\left\{\boldsymbol{n}_{0}-\int_{0}^{\xi}\boldsymbol{f}(\eta)\,\mathrm{d}\eta\right\}\,\mathrm{d}\xi+\boldsymbol{r}_{0}. \tag*{(16)}

From the boundary conditions (4.9) we have r(0)=0\mathbf{r}(0)=\mathbf{0} that implies that r0=0\mathbf{r}_0=\mathbf{0}, and from r(1)=k\mathbf{r}(1)=\mathbf{k} we can obtain n0\mathbf{n}_0.

We consider the particular case of the force due to gravity (catenaries) which is equivalent to f=−ζev\mathbf{f}=-\zeta\mathbf{e}_v where ev\mathbf{e}_v is the unit vector in the vertical direction. The relation (5.2) and the fact that r0=0\mathbf{r}_0=\mathbf{0} imply

r(s)=∫0sν^(∣n0+ξζev∣)∣n0+ξζev∣{n0+ξζev} dξ.(17)\mathbf{r}(s)=\int_0^s\frac{\hat{\boldsymbol{\nu}}\left(\lvert\mathbf{n}_0+\xi\zeta\mathbf{e}_v\rvert\right)}{\lvert\mathbf{n}_0+\xi\zeta\mathbf{e}_v\rvert}\{\mathbf{n}_0+\xi\zeta\mathbf{e}_v\}\,\mathrm{d}\xi. \tag*{(17)}

The case of tensile internal forces and inextensibility is not reached.

To advance in the analysis of the boundary value problem, let us assume that k=e1\mathbf{k}=\mathbf{e}_1 and that ev=e2\mathbf{e}_v=\mathbf{e}_2, as depicted in Figure 2.

Figure 2. A string under the force of gravity.

We use the constitutive relation (4.12), and in this first case, we assume that N=(rs⋅n)∣n∣N=(\mathbf{r}_s\cdot\mathbf{n})\lvert\mathbf{n}\rvert is such that 0≤N≤N10\leq N\leq N_1 for 0≤s≤10\leq s\leq1. Then from (4.12) and (5.3) we obtain

r(s)=∫0s(1+1∣n0+ξζe2∣){n0+ξζe2} dξ.\mathbf{r}(s)=\int_0^s\left(1+\frac{1}{\lvert\mathbf{n}_0+\xi\zeta\mathbf{e}_2\rvert}\right)\{\mathbf{n}_0+\xi\zeta\mathbf{e}_2\}\,\mathrm{d}\xi.

For each component rir_i of r\mathbf{r} (the components of n0\mathbf{n}_0 are n0,in_{0,i}), we obtain

r1(s)=n0,1ℓ(s),r3(s)=n0,3ℓ(s),r_1(s)=n_{0,1}\ell(s),\qquad r_3(s)=n_{0,3}\ell(s),
r2(s)=12ζ[2n0,12+(n0,2+ζs)2+n0,32+ζs(2n0,2+ζs)−2n0,12+n0,22+n0,32],r_2(s)=\frac{1}{2\zeta}\left[2\sqrt{n_{0,1}^2+(n_{0,2}+\zeta s)^2+n_{0,3}^2}+\zeta s(2n_{0,2}+\zeta s)-2\sqrt{n_{0,1}^2+n_{0,2}^2+n_{0,3}^2}\right],

where we have defined

ℓ(s)=s+1ζln⁡[n0,2+ζs+n0,12+(n0,2+ζs)2+n0,32n0,2+n0,12+n0,22+n0,32].\ell(s)=s+\frac{1}{\zeta}\ln\left[\frac{n_{0,2}+\zeta s+\sqrt{n_{0,1}^2+(n_{0,2}+\zeta s)^2+n_{0,3}^2}}{n_{0,2}+\sqrt{n_{0,1}^2+n_{0,2}^2+n_{0,3}^2}}\right].

From the boundary condition (4.9)2_2, we have r(1)=e1\mathbf{r}(1)=\mathbf{e}_1. Then

r1(1)=1,r2(1)=0,r3(1)=0,r_1(1)=1,\qquad r_2(1)=0,\qquad r_3(1)=0,

which can be used to find n0,1n_{0,1}, n0,2n_{0,2} and n0,3n_{0,3}.

The case of tensile external forces and inextensibility is reached for a part of the string.

In this section we study the case when there exist s~\tilde{s} and s˘\breve{s} such that 0<s~<1−s˘<10 < \tilde{s} < 1 - \breve{s} < 1 and

N>N1if 0≤s≤s~,0≤N≤N1if s~<s<1−s˘,N>N1if 1−s˘<s<1.\begin{aligned} N &> N_1 && \text{if } 0 \le s \le\tilde{s},\\ 0 &\le N \le N_1 && \text{if } \tilde{s} < s < 1-\breve{s},\\ N &> N_1 && \text{if } 1-\breve{s} < s < 1. \end{aligned}

i.e., two parts of the string reach inextensibility.

In the case 0≤s≤s~0 \le s \le\tilde{s} since N>N1N > N_1 from (14) we have ν^(N)=ν1\hat{\nu}(N) = \nu_1 and from (17) we have

r(s)=∫0sν1∣n0+ξζe2∣{n0+ξζe2} dξ,r(s) = \int_0^s \frac{\nu_1}{\lvert n_0+\xi\zeta e_2\rvert}\{n_0+\xi\zeta e_2\}\,\mathrm{d}\xi,

and for the components of rr we obtain

r1(s)=n0,1ℓA(s),r3(s)=n0,3ℓA(s),r_1(s)=n_{0,1}\ell_A(s),\qquad r_3(s)=n_{0,3}\ell_A(s),
r2(s)=ν1ζ[n0,12+(n0,2+ζs)2+n0,32−n0,12+n0,22+n0,32],r_2(s)=\frac{\nu_1}{\zeta}\left[\sqrt{n_{0,1}^2+(n_{0,2}+\zeta s)^2+n_{0,3}^2}-\sqrt{n_{0,1}^2+n_{0,2}^2+n_{0,3}^2}\right],

where we have defined

ℓA(s)=ν1ζln⁡[n0,2+ζs+n0,12+(n0,2+ζs)2+n0,32n0,2+n0,12+n0,22+n0,32].\ell_A(s)=\frac{\nu_1}{\zeta}\ln\left[\frac{n_{0,2}+\zeta s+\sqrt{n_{0,1}^2+(n_{0,2}+\zeta s)^2+n_{0,3}^2}}{n_{0,2}+\sqrt{n_{0,1}^2+n_{0,2}^2+n_{0,3}^2}}\right].

In the case s~<s<1−s˘\tilde{s}<s<1-\breve{s} from (14) we have ν^(N)=N+1\hat{\nu}(N)=N+1 and from (17) we get

r=∫s~s(1+1∣n0+ξζe2∣){n0+ξζe2} dξ+r0,r=\int_{\tilde{s}}^s\left(1+\frac{1}{\lvert n_0+\xi\zeta e_2\rvert}\right)\{n_0+\xi\zeta e_2\}\,\mathrm{d}\xi+r_0,

and for each component of rr we obtain

r1(s)=n0,1ℓB(s)+r0,1,r3(s)=n0,3ℓB(s)+r0,3,r_1(s)=n_{0,1}\ell_B(s)+r_{0,1},\qquad r_3(s)=n_{0,3}\ell_B(s)+r_{0,3},
r2(s)=12ζ[2n0,12+(n0,2+ζs)2+n0,32+ζs(2n0,2+ζs)−2n0,12+(n0,2+ζs~)2+n0,32−ζs~(2n0,2+ζs~)]+r0,2,r_2(s)=\frac{1}{2\zeta}\left[2\sqrt{n_{0,1}^2+(n_{0,2}+\zeta s)^2+n_{0,3}^2}+\zeta s(2n_{0,2}+\zeta s)-2\sqrt{n_{0,1}^2+(n_{0,2}+\zeta\tilde{s})^2+n_{0,3}^2}-\zeta\tilde{s}(2n_{0,2}+\zeta\tilde{s})\right]+r_{0,2},

where

ℓB(s)=s−s~+1ζln⁡[n0,2+ζs+n0,12+(n0,2+ζs)2+n0,32n0,2+ζs~+n0,12+(n0,2+ζs~)2+n0,32].\ell_B(s)=s-\tilde{s}+\frac{1}{\zeta}\ln\left[\frac{n_{0,2}+\zeta s+\sqrt{n_{0,1}^2+(n_{0,2}+\zeta s)^2+n_{0,3}^2}}{n_{0,2}+\zeta\tilde{s}+\sqrt{n_{0,1}^2+(n_{0,2}+\zeta\tilde{s})^2+n_{0,3}^2}}\right].

For the case 1−s˘<s<11-\breve{s}<s<1 we again have N>N1N>N_1 thus ν^(N)=ν1\hat{\nu}(N)=\nu_1, and as a result from (17) we have

r(s)=∫1−s˘sν1∣n0+ξζe2∣{n0+ξζe2} dξ+r^0,r(s)=\int_{1-\breve{s}}^s\frac{\nu_1}{\lvert n_0+\xi\zeta e_2\rvert}\{n_0+\xi\zeta e_2\}\,\mathrm{d}\xi+\hat{r}_0,

from which we get

r1(s)=n0,1ℓC(s)+r^0,1,r3(s)=n0,3ℓC(s)+r^0,3,r_1(s)=n_{0,1}\ell_C(s)+\hat{r}_{0,1},\qquad r_3(s)=n_{0,3}\ell_C(s)+\hat{r}_{0,3},
r2(s)=ν1ζ{n0,12+(n0,2+ζs)2+n0,32−n0,12+[n0,2+ζ(1−s˘)]2+n0,32}+r^0,2.r_2(s)=\frac{\nu_1}{\zeta}\left\{\sqrt{n_{0,1}^2+(n_{0,2}+\zeta s)^2+n_{0,3}^2}-\sqrt{n_{0,1}^2+[n_{0,2}+\zeta(1-\breve{s})]^2+n_{0,3}^2}\right\}+\hat{r}_{0,2}.

where

ℓC(s)=ν1ζln⁡{n0,2+ζs+n0,12+(n0,2+ζs)2+n0,32n0,2+ζs+n0,12+[n0,2+ζ(1−sˇ)]2+n0,32}.\ell_C(s)=\frac{\nu_1}{\zeta}\ln\left\{\frac{n_{0,2}+\zeta s+\sqrt{n_{0,1}^{2}+(n_{0,2}+\zeta s)^{2}+n_{0,3}^{2}}}{n_{0,2}+\zeta s+\sqrt{n_{0,1}^{2}+[n_{0,2}+\zeta(1-\check{s})]^{2}+n_{0,3}^{2}}}\right\}.

In the above results for ri(s)r_i(s) we have the following list of 11 unknown parameters: s~\tilde{s}, sˇ\check{s}, r0,1r_{0,1}, r0,2r_{0,2}, r0,3r_{0,3}, r^0,1\hat{r}_{0,1}, r^0,2\hat{r}_{0,2}, r^0,3\hat{r}_{0,3}, n0,1n_{0,1}, n0,2n_{0,2} and n0,3n_{0,3}. These can be found from the 6 continuity conditions

ri(s~−)=ri(s~+),ri(1−sˇ−)=ri(1−sˇ+),i=1,2,3,(18)r_i(\tilde{s}^{-})=r_i(\tilde{s}^{+}),\qquad r_i(1-\check{s}^{-})=r_i(1-\check{s}^{+}),\qquad i=1,2,3, \tag*{(18)}

where the superscripts −- and ++ mean approaching from the left and the right, respectively. As well as this we have the boundary conditions (13)2_2, which in our case means

r1(1)=1,r2(1)=0,r3(1)=0.(19)r_1(1)=1,\qquad r_2(1)=0,\qquad r_3(1)=0. \tag*{(19)}

Finally, from (14) we have that if s=s~s=\tilde{s} and s=1−sˇs=1-\check{s} we have that N=N1N=N_1, which is equivalent to

n0,12+(n0,2+ζs~)2+n0,32=N12,n0,12+[n0,2+ζ(1−sˇ)]2+n0,32=N12.(20)n_{0,1}^{2}+(n_{0,2}+\zeta\tilde{s})^{2}+n_{0,3}^{2}=N_1^{2},\qquad n_{0,1}^{2}+[n_{0,2}+\zeta(1-\check{s})]^{2}+n_{0,3}^{2}=N_1^{2}. \tag*{(20)}

From (18), (19), and (20), we have 11 equations determining the list of 11 unknowns.

Motions with piecewise constant stretch

In this section we consider the motion of a straight string r(s,t)=χ(s,t)e1\mathbf{r}(s,t)=\chi(s,t)\mathbf{e}_1 satisfying (14). We assume that the string has piecewise constant stretch separated by a shock front σ(t)∈[0,1]\sigma(t)\in[0,1]: there exist ν−>0\nu_->0 and ν+>0\nu_+>0 with ν−≠ν+\nu_-\ne\nu_+ such that

χs(s,t)={ν−if s<σ(t),ν+if s>σ(t).\chi_s(s,t)= \begin{cases} \nu_- & \text{if }s<\sigma(t),\\ \nu_+ & \text{if }s>\sigma(t). \end{cases}

In this setting, we do not prescribe the placement of r(1,t)\mathbf{r}(1,t) but, instead, the terminal tension N+:=N(1,t)N_+:=N(1,t) is given.

We will assume that the terminal tension is a tensile force N+>0N_+>0 and the string is separated into two extensible parts: a compressed part parameterized by [0,σ(t)][0,\sigma(t)] and an elongated part parameterized by [σ(t),1][\sigma(t),1]. In particular, it follows that

ν0<ν−=αCαTN−+1<1<ν+=N++1<ν1,\nu_0<\nu_-=\frac{\alpha_C}{\alpha_T}N_-+1<1<\nu_+=N_++1<\nu_1,

where we have used (14) to relate stretches to the (constant) tensions to the left and right of σ(t)\sigma(t). The case that both parts are elongated with one part of the string reaching inextensibility was treated in [32].

Solution of the shock problem

6.1. Solution of the shock problem. We seek configurations with piecewise constant stretch of the form

χ(s,t)={(αCαTN−+1)sif s∈[0,σ(t)],(N++1)(s−s0)+v+t+σ0if s∈[σ(t),1].(21)\chi(s,t)= \begin{cases} \left(\frac{\alpha_C}{\alpha_T}N_-+1\right)s & \text{if }s\in[0,\sigma(t)],\\ (N_++1)(s-s_0)+v_+t+\sigma_0 & \text{if }s\in[\sigma(t),1]. \end{cases} \tag*{(21)}

Here, σ0∈(0,1)\sigma_0\in(0,1) is the initial position of the shock front with reference position s0∈(0,1)s_0\in(0,1), and v+≠0v_+\ne0 is the (dimensionless) velocity of the particles comprising the elongated segment. N+∈(0,N1)N_+\in(0,N_1) is the terminal tension and N−∈(−N0,0)N_-\in(-N_0,0) is the unknown constant tension in the compressed part of the string.

We observe that the boundary conditions, χ(0,t)=0\chi(0,t)=0 and N(1,t)=χs(1,t)−1=N+<N1N(1,t)=\chi_s(1,t)-1=N_+<N_1, are automatically satisfied, and the equations of motion hold on the segments [0,σ(t)][0,\sigma(t)] and [σ(t),1][\sigma(t),1]. The Rankine-Hugoniot jump conditions guaranteeing that r(s,t)=χ(s,t)e1\mathbf{r}(s,t)=\chi(s,t)\mathbf{e}_1 is a weak solution to (11) reduce to

⟦N⟧+ζσ˙⟦∂tχ⟧=0.(22)\llbracket N\rrbracket+\zeta\dot{\sigma}\llbracket\partial_t\chi\rrbracket=0. \tag*{(22)}

See, e.g., [1]. Above, ⟦y⟧(t)=y(σ+(t),t)−y(σ−(t),t)\llbracket y\rrbracket(t)=y(\sigma^+(t),t)-y(\sigma^-(t),t) is the jump across s=σ(t)s=\sigma(t) at time tt. We now solve (22). Continuity of χ\chi requires that χ(σ(t)−,t)=χ(σ(t)+,t)\chi(\sigma(t)^-,t)=\chi(\sigma(t)^+,t) and differentiating in time yields

σ′χs(σ−,t)+χt(σ−,t)=σ′χs(σ+,t)+χt(σ+,t)⟺σ′=v+(αCαTN−−N+)−1.(23)\sigma'\chi_s(\sigma^-,t)+\chi_t(\sigma^-,t)=\sigma'\chi_s(\sigma^+,t)+\chi_t(\sigma^+,t)\Longleftrightarrow\sigma'=v_+\left(\frac{\alpha_C}{\alpha_T}N_- -N_+\right)^{-1}. \tag*{(23)}

Since ⟦∂tχ⟧=v+\llbracket\partial_t\chi\rrbracket=v_+, (23) implies that (22) is equivalent to

(N−−N+)(N−−αTαCN+)=ζv+2αTαC.(N_--N_+)\left(N_--\frac{\alpha_T}{\alpha_C}N_+\right)=\zeta v_+^2\frac{\alpha_T}{\alpha_C}.

The assumption that the segment parameterized by [0,σ(t)][0,\sigma(t)] is extensible and compressive, −N0<N−<0-N_0<N_-<0, implies that

N−=12(1+αTαC)N+−[14(1−αTαC)2N+2+ζv+2αTαC]1/2,(24)N_-=\frac{1}{2}\left(1+\frac{\alpha_T}{\alpha_C}\right)N_+-\left[\frac{1}{4}\left(1-\frac{\alpha_T}{\alpha_C}\right)^2N_+^2+\zeta v_+^2\frac{\alpha_T}{\alpha_C}\right]^{1/2}, \tag*{(24)}

with

N+2ζ<v+2<N+2ζ+1ζ(αCαT+1)N+N0+αCζαTN02.(25)\frac{N_+^2}{\zeta}<v_+^2<\frac{N_+^2}{\zeta}+\frac{1}{\zeta}\left(\frac{\alpha_C}{\alpha_T}+1\right)N_+N_0+\frac{\alpha_C}{\zeta\alpha_T}N_0^2. \tag*{(25)}

Then the velocity of the shock front can be written as

σ′=v+{12(αCαT−1)N+−[14(αCαT−1)2N+2+αCαTζv+2]1/2}−1.(26)\sigma'=v_+\left\{\frac{1}{2}\left(\frac{\alpha_C}{\alpha_T}-1\right)N_+-\left[\frac{1}{4}\left(\frac{\alpha_C}{\alpha_T}-1\right)^2N_+^2+\frac{\alpha_C}{\alpha_T}\zeta v_+^2\right]^{1/2}\right\}^{-1}. \tag*{(26)}

In conclusion, the Rankine-Hugoniot jump condition (22) is satisfied by χ\chi with N−∈(−N0,0)N_-\in(-N_0,0) if and only if N−N_- is given by (24), the inequalities (25) hold, and the shock speed is given by (26).

The dimensionless speed of propagation in the elongated section is c+=1c_+=1 while the dimensionless speed of propagation in the compressed part is c−=αCαT<1=c+c_-=\frac{\alpha_C}{\alpha_T}<1=c_+. Consequently, we must have v+>0v_+>0 and

−1<σ′<−αTαC(27)-1<\sigma'<-\frac{\alpha_T}{\alpha_C} \tag*{(27)}

for Lax’s geometric shock inequalities to hold [15]. These conditions ensure the structurally stability of (21) as a weak solution to (11) [15,16]. By (26) and (25), the condition v+>0v_+>0 implies that elongated section parameterized by [σ(t),1][\sigma(t),1] is growing in time.

We now determine the implications of (27). Using (26), we see that σ′>−1\sigma'>-1 if and only if

[v++12(αCαT−1)N+]2<14(αCαT+1)2N+2+ζv+2αCαT⟺\left[v_++\frac{1}{2}\left(\frac{\alpha_C}{\alpha_T}-1\right)N_+\right]^2<\frac{1}{4}\left(\frac{\alpha_C}{\alpha_T}+1\right)^2N_+^2+\zeta v_+^2\frac{\alpha_C}{\alpha_T} \Longleftrightarrow
(1−ζαCαT)v+2+(αCαT−1)N+v+<0.\left(1-\zeta\frac{\alpha_C}{\alpha_T}\right)v_+^2+\left(\frac{\alpha_C}{\alpha_T}-1\right)N_+v_+<0.

Thus, in order for the above solutions to satisfy Lax’s shock inequality σ′>−1\sigma' > -1, we must have that 1−ζαCαT<01-\zeta\frac{\alpha_C}{\alpha_T}<0 and

(ζαCαT−1)−1(αCαT−1)N+<v+\left(\zeta\frac{\alpha_C}{\alpha_T}-1\right)^{-1}\left(\frac{\alpha_C}{\alpha_T}-1\right)N_+<v_+

The second condition σ′<−αTαC\sigma'<-\frac{\alpha_T}{\alpha_C} can similarly be seen to be equivalent to

0<(1−ζαTαC)v+2+(1−αTαC)N+v+.(28)0<\left(1-\zeta\frac{\alpha_T}{\alpha_C}\right)v_+^2+\left(1-\frac{\alpha_T}{\alpha_C}\right)N_+v_+. \tag*{(28)}

If 1−ζαTαC≥01-\zeta\frac{\alpha_T}{\alpha_C}\geq0, then (28) holds automatically, but if 1−ζαTαC<01-\zeta\frac{\alpha_T}{\alpha_C}<0, then (28) holds if and only if

v+<(ζαTαC−1)−1(1−αTαC)N+.v_+<\left(\zeta\frac{\alpha_T}{\alpha_C}-1\right)^{-1}\left(1-\frac{\alpha_T}{\alpha_C}\right)N_+.

In summary, piecewise constant stretch solutions of the form (21), which satisfy (24), (25), and (26), verify Lax’s geometric shock inequalities if and only if v+>0v_+>0, i.e., the elongated segment is growing in time. Additionally, the following conditions must be met: either

αTαC<ζ≤αCαTand(ζαCαT−1)−1(αCαT−1)N+<v+,(29)\frac{\alpha_T}{\alpha_C}<\zeta\leq\frac{\alpha_C}{\alpha_T} \quad\text{and}\quad \left(\zeta\frac{\alpha_C}{\alpha_T}-1\right)^{-1}\left(\frac{\alpha_C}{\alpha_T}-1\right)N_+<v_+, \tag*{(29)}

or

ζ>αCαTand(ζαCαT−1)−1(αCαT−1)N+<v+<(ζαTαC−1)−1(1−αTαC)N+.(30)\zeta>\frac{\alpha_C}{\alpha_T} \quad\text{and}\quad \left(\zeta\frac{\alpha_C}{\alpha_T}-1\right)^{-1}\left(\frac{\alpha_C}{\alpha_T}-1\right)N_+<v_+< \left(\zeta\frac{\alpha_T}{\alpha_C}-1\right)^{-1}\left(1-\frac{\alpha_T}{\alpha_C}\right)N_+. \tag*{(30)}

In particular, if ζ=αTαC[1+β(αC2αT2−1)]\zeta=\frac{\alpha_T}{\alpha_C}\left[1+\beta\left(\frac{\alpha_C^2}{\alpha_T^2}-1\right)\right], β∈(0,1)\beta\in(0,1), αCαT\frac{\alpha_C}{\alpha_T} and β\beta sufficiently close to 1, then

(ζαCαT−1)−1(αCαT−1)N+<N+ζ1/2.\left(\zeta\frac{\alpha_C}{\alpha_T}-1\right)^{-1}\left(\frac{\alpha_C}{\alpha_T}-1\right)N_+<\frac{N_+}{\zeta^{1/2}}.

Therefore, (25) and (29) are then satisfied for all v+v_+ sufficiently close to N+ζ1/2\frac{N_+}{\zeta^{1/2}}, establishing that the conditions (25) and either (29) or (30) are satisfied by an open set of values for the parameters αCαT\frac{\alpha_C}{\alpha_T}, ζ\zeta, and v+v_+.

Final remarks

In the present work, we proposed a new type of piecewise linear constitutive relation for strings that limits the amount of extension and compression the string can undergo. Moreover, the tangent moduli relating tension to stretch are different for extension versus compression. We solved static and dynamic problems from the semi-inverse perspective with solutions to the latter having piecewise constant stretch. Numerical methods for solving more realistic static problems will be explored in future work, while dynamic problems will be studied using the method developed in [3].

The constitutive model developed in this paper provides a simple framework for describing the behavior of stretch-limited strings, which appear in various physical and biological settings and cannot be modeled using the classical approach. As shown in Figure 1 of [21], in general, implicit constitutive theories are particularly useful in situations where the conventional approach of expressing the stress (or, in this case, the tension) explicitly as a function of the strain is inadequate. This is precisely the case for the constitutive relation (4) introduced here: when the stretch reaches a critical threshold, it remains fixed, yet the tension can continue to either decrease or increase without bound. This behavior fundamentally defies modeling within the classical framework, which expresses tension as a function of stretch. The model developed in this study has potential applications in biological systems, where stretch-limited strings arise in different contexts (see, e.g., [13], particularly Figure 1–3, and the references therein). More broadly, the model presented here offers a simple alternative to classical elastic string models where standard constitutive assumptions fail to incorporate natural stretch-limiting behavior.

References

R. Bustamante

Departamento de Ingeniería Mecánica, Universidad de Chile Beauchef 851, Santiago Centro, Santiago, Chile [email protected]

K. R. Rajagopal

Department of Mechanical Engineering, Texas A&M University College Station, TX 77843, USA [email protected]

C. Rodriguez

Department of Mathematics, University of North Carolina Chapel Hill, NC 27599, USA [email protected]

References

  1. [25]Rajagopal, K. R. and Rodriguez, C. On a strain-limiting special Cosserat rod model, Math. Models Methods Appl. Sci., 33: 1-30, 2023.arxiv.org/abs/2205.10875
  2. [26]Rajagopal, K. R. and Rodriguez, C. Special Cosserat rods with rate-dependent evolving natural configurations, Int. J. Eng. Sci., 191, 103890, 2023.arxiv.org/abs/2304.04633
  3. [27]Rajagopal, K. R. and Saccomandi, G. Implicit nonlinear elastic bodies with density dependent material moduli and its linearizations, Int. J. Solids Struct., 234: 111255, 2022.DOI
  4. [28]Rajagopal, K. R. and Srinivasa, A. R. A thermodynamic framework for rate type fluid models. J. Non-Newton. Fluid Mech. 88: 207–227, 2000.DOI
  5. [29]Rajagopal, K. R. and Srinivasa, A. R. On the response of non-dissipative solid. Proc. Roy. Soc. London A, 463: 357–367, 2007.
  6. [30]Rajagopal, K. R. and Srinivasa, A. R. On a class of non-dissipative materials that are not hyperelastic. Proc. Roy. Soc. London A, 465: 493–500, 2009.DOI
  7. [31]Rodriguez, C. On stretch-limited elastic strings. Proc. R. Soc. Lond. A, 477 (2249): Paper No. 20210181, 15, 2021.DOI
  8. [32]Rodriguez, C. Longitudinal shock waves in a class of semi-infinite stretch-limited elastic strings. Math. Mech. Solids, 27: p. 474–490, 2022.arxiv.org/abs/2102.07600
  9. [33]Şengül, Y. One-dimensional strain-limiting viscoelasticity with an arctangent type nonlinearity. Appl. Eng. Sci., 7: 100058, 2021DOI
  10. [34]Şengül, Y. Viscoelasticity with limiting strain, Discrete Contin. Dyn. Syst.-S, 14: 57–70, 2021.DOI
  11. [35]Şengül, Y. Global existence of solutions for the one-dimensional response of viscoelastic solids within the context of strain-limiting theory, in Research in Mathematics of Materials Science, Springer, pp. 319–332, 2022.DOI
  12. [36]Todhunter, I. (edited and completed by Pearson K). A history of the theory of elasticity and strength of materials, Volume 1: From Galilei to St. Venant. Printed by C. J. Clay, M. A. and Sons, at the University Press, Cambridge, 1886.

Paper details

vlnc:W_uwxaat7t
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents