Introduction
From the time carbon nanotubes (CNT) were first discovered by Iijima (
1991), the development of nanoscience and nanotechnologies has proceeded at a fast rate, as has the application of CNT to various advanced industries because of their superior mechanical, electrical, thermal, and chemical properties. To assess the performance of CNT and extend their lifetimes, a number of studies carried out the structural analyses of nonembedded and embedded single- and multiwalled (SW and MW) CNT. Basically, there are four kinds of simulation methods used to conduct such analyses: (1) atomistic modeling, (2) hybrid atomistic-continuum mechanics, (3) experimental methods, and (4) continuum mechanics, with the first three approaches being especially time-consuming and expensive. However, classical continuum mechanics-based theories are not able to capture the small-length scale effect and, therefore, some modified theories have been developed, such as the strain gradient elasticity, coupled-stress elasticity, surface elasticity, and Eringen’s nonlocal elasticity theories (ENET) (
Eringen 1972,
1983,
2002), with ENET being the most popular and widely used in various nonlocal continuum mechanics theories. Some comprehensive literature surveys with regard to the vibration characteristics of CNT and the various simulation methods for the analysis of these also have been carried out (
Arash and Wang 2012;
Brischetto 2014;
Eltaher et al. 2016;
Hu et al. 2012;
Gibson et al. 2007;
Wang et al. 2016).
Some theoretical and experimental investigations (
Fu et al. 2006;
Walgraef 2007) have examined the behaviors of CNT under large external loads, in which their deformations are nonlinear, rather than linear. However, there remain relatively few articles carrying out the large deflection and large amplitude vibration analyses of CNT, as compared with those examining their linear counterparts. Among the various ENET-based nonlocal beam theories used for the analysis of CNT, the nonlocal Timoshenko beam theory (TBT) commonly is used in the open literature and has been shown to produce satisfactory results as seen in applications, such as the nonlocal classical beam theory (CBT), TBT, Reddy’s beam theory (RBT), Levinson’s beam theory (LBT), generalized shear deformation beam theory (GSDBT), and sinusoidal-shear deformation beam theory (SSDBT) (
Aydogdu 2008,
2009;
Murmu and Pradhan 2009a,
b;
Reddy 2007;
Reddy and Pang 2008;
Thai 2012;
Thai and Vo 2012;
Wang et al. 2008). The literature survey in the current work thus will focus on articles related to the geometrically nonlinear analysis of SW/MW CNT that are nonembedded and embedded in an elastic medium, and different boundary conditions.
On the basis of the principle of virtual displacement (PVD), Yang et al. (
2010) developed a nonlocal TBT combined with the von Kármán geometrical nonlinearity (VKGN) for the nonlinear free vibration analysis of SWCNT, with combinations of simply-supported and clamped edges, in which the effects of transverse shear deformation and rotary inertia were considered. Within the framework of the nonlocal CBT and VKGN, Ghayesh (
2014) investigated the nonlinear dynamic responses of SWCNT with various boundary conditions and subjected to the primary and superharmonic excitations. A parametric study with regard to the effects of the length-to-diameter ratio, forcing amplitude and excitation frequency on the dynamic responses of SWCNT also was undertaken. Ke et al. (
2009) extended the work of Yang et al. (
2010) to the large amplitude vibration analysis of double-walled (DW) CNT embedded in an elastic medium. In their formulation, the nonlocal TBT-based motion equations were derived using Hamilton’s principle; the interactions between the DWCNT and its surrounding medium were considered, and the van der Waals (vdW) interaction between adjacent walls were simulated using the Winkler-type foundation model with different spring constants. The differential quadrature (DQ) method (
Bert and Malik 1997;
Du et al. 1994;
Wu and Lee 2001) in conjunction with a direct iterative method then was used to solve the resulting nonlinear equations. This issue then was reexamined by Ansari et al. (
2012), in which a two-parameter Pasternak foundation model was used to simulate the interaction between the DWCNT and its surrounding medium. Moreover, an alternative approach, the strain gradient theory, also has been used to study various structural analyses of nanoscale beams, such as linear bending analysis (
Lim and Wang 2007) and wave propagation analysis (
Lim et al. 2015), in which the local higher-order strain gradients were considered in a global sense.
Because almost all of the two-dimensional (2D) nonlocal beam theories used for the analysis of SW/MW CNT were derived on the basis of the PVD rather than Reissner’s mixed variational theorem (RMVT) (
Reissner 1984,
1986), Wu and Lai (
2015a,
b) and Wu and Liou (
2016) thus developed an RMVT-based nonlocal TBT for the linear bending, free vibration, and buckling analyses of SWCNT, embedded and non embedded in an elastic medium, and with different boundary conditions, in which the DQ and differential reproducing kernel methods (
Wang et al. 2010) were used. In the PVD-based beam theories, the generalized displacement components (i.e., the normal displacement and rotation) were regarded as the primary variables; however, except for the generalized displacement components, the generalized resultant forces (i.e., shear force and bending moment) also were regarded as the primary variables in the RMVT-based beam theories. In Wu and Lai (
2015a,
b), the highest differential orders of the Euler-Lagrange equations in the RMVT-based strong and weak formulations were shown to be one-half lower than those in the PVD-based ones. The performance of the RMVT-based nonlocal TBT was shown to be superior to that of the PVD-based one, in that the convergence rate of the RMVT-based nonlocal TBT was faster than that of the PVD-based nonlocal one. Moreover, the predictions of the generalized force resultants obtained using the RMVT-based nonlocal TBT were more accurate than those obtained using the PVD-based nonlocal one. Finally, the RMVT-based nonlocal TBT was less time-consuming than the PVD-based one. Because of these benefits of the RMVT-based nonlocal TBT, and the fact that there are relatively few articles regarding the nonlinear analysis of MWCNT as compared with those related to the linear analysis of these, the authors of this paper extended the earlier work regarding the linear static analysis of embedded SWCNT to the geometrically nonlinear static analysis of embedded MWCNT with combinations of free, simply-supported, and clamped edge conditions, in which the effect of vdW interaction was considered. In the formulation, the Euler-Lagrange equations of a typical wall of the MWCNT and the associated boundary conditions were derived by using the stationary principle of Reissner’s energy functional, in which the ENET and VKGN were used. The vdW interaction forces between any pair of walls constituting the MWCNT and the Pasternak-type foundation model used to simulate the interaction between the MWCNT and its surrounding medium were taken into account. A parametric study related to the influence of some crucial effects on the nonlinear static behavior of the embedded MWCNT was conducted, such as the small-length scale effect, vdW interaction, aspect ratios, different boundary conditions, and the stiffness of the foundation.
Applications
In this section, the DQ method was used to construct the weighting coefficients for the first- and higher-order derivatives of the primary variables in the spatial (
) domain, such that
where
; and
=
,
,
,
,
, and
which are the nodal values of in- and out-of-plane deflections, rotation, axial force, bending moment, and shear force components of the
th-wall of the MWCNT at
.
Applying the DQ formulae [i.e., Eq. (
37)] to the strong formulation of the RMVT-based nonlocal TBT for the nonlinear bending problem of an embedded MWCNT, which consists of the governing Eqs. (
22)
–(
24) and (
33)
–(
35) associated with the appropriate boundary conditions given in Eqs. (
28a)
–(
28c) and (
29a)
–(
29c), the authors obtained a set of simultaneous nonlinear algebraic equations. In these, four different boundary conditions were considered: (1) the simple-simple (SS), (2) clamped-simple (CS), (3) clamped-clamped (CC), and (4) clamped-free (CF) supports.
Applying the DQ formulae to the governing equations at the sampling nodes in the interior domain, the following is obtained:
where
.
For a typical
th-wall, Eqs. (
38)
–(
43) represent
equations in terms of
unknowns, whereas there are only three boundary conditions at each boundary point, and, thus, the authors needed to supplement three more governing equations at each of these. Four different cases of boundary conditions were considered and given as follows:
Finally, the sets of resulting equations considering the previously noted boundary conditions of Cases 1–4 are
simultaneous algebraic equations in terms of
unknowns, when
, which represents a standard nonlinear boundary value problem. Because a direct iterative method was used to solve these nonlinear equations, they are rewritten in matrix form as follows:
where superscript
th-iteration; coefficients
and
are in terms of the determined variables
and
;
, which means the linear solutions will be used as the initial guess of the unknowns to obtain the nonlinear ones, i.e., the convergent solutions obtained from the direct iterative method with an allowable error
less than
, in which the allowable error
is defined as
, in which the symbol,
, denotes the inner product operation for two vectors; and
.
Once Eq. (
48) is solved, the nonlinear solutions of the nodal generalized displacement and force resultant components for each layer can be obtained.
Concluding Remarks
In this article, the authors developed the RMVT-based nonlocal TBT for the nonlinear static analysis of MWCNTs with combinations of free, simply-supported, and clamped edge conditions embedded in an elastic medium. The strong formulations of the RMVT-based nonlinear nonlocal TBT and its associated possible conditions were derived using the stationary principle of Reissner’s energy functional, in which the vdW interaction forces between any pair of walls constituting the MWCNT were considered and the interaction between the MWCNT and its surrounding medium was simulated using the Pasternak-type foundation model. The DQ method was used to discretize the generalized displacement and force resultant components, and a direct iteration method was used to solve the resulting system equations. The results showed that in the linear cases, the variation of the dimensionless midspan deflection with the dimensionless loading intensity was linear, whereas it appeared to be a higher-order polynomial function with a decreasing slope in the nonlinear cases. The geometrical nonlinear effect on the deflection functions of MWCNTs was more significant than the small-scale effect on these. The geometrical nonlinear effect on the static behavior of MWCNTs with different boundary conditions was significant for the SS, CS, and CC cases, rather than the CF case, and were , in which > means more significant.