Abstract
A coupled electro-thermo-mechanical discontinuous Galerkin (DG) method is developed considering the non-linear interactions of electrical, thermal, and mechanical fields. In order to develop a stable discontinuous Galerkin formulation the governing equations are expressed in terms of energetically conjugated fields gradients and fluxes. Moreover, the DG method is formulated in finite deformations and finite fields variations. The multi-physics DG formulation is shown to satisfy the consistency condition, and the uniqueness and optimal convergence rate properties are derived under the assumption of small deformation. First the numerical properties are verified on a simple numerical example, and then the framework is applied to simulate the response of smart composite materials in which the shape memory effect of the matrix is triggered by the Joule effect.
Similar content being viewed by others
Notes
The contributions on \(\partial _{\mathrm{D}} \varOmega _{0\mathrm{h}}\) can be directly deduced by removing the factor (1/2) accordingly to the definition of the average flux on the Dirichlet boundary and by using \({{Z_0(F_\text{h}, \bar{M})}}\) instead of \(Z_0(F_\text{h}, {M_\text{h}})\). However, there is one more additional term in \({{\mathbf{F}}}_{\mathbf{u}{\text {I1}}}^{a\pm}\) in the Dirichlet boundary, which is \(\sum _{\mathrm{s}} \int _{\left( \partial _{\mathrm{D}} \varOmega _0\right) ^{\text {s}}} (-{{N}}^a_{{\mathbf{u}}})\left( \pmb {\mathcal {Y}}({{{\mathbf{M}}}}){{{\mathbf{M}}}}-\pmb {\mathcal {Y}}({{{\mathbf{M}}}_0}){{{\mathbf{M}}}_0}\right) \cdot \varvec{{N}}^{-} {\text {dS}}_0\).
In this case \({{\mathbf{G}}}^{{\text{T}}} {\mathbf{o}}^{{\text{T}}} ({{\mathbf{G}}}^{\prime } )\delta {{\mathbf{G}}}_{{\mathbf{n}}} = {{\mathbf{G}}}_{{\mathbf{n}}}^{{\text{T}}} \widetilde{{\mathbf{o}}}^{{\text{T}}} ({{\mathbf{G}}}^{\prime } )\delta {{\mathbf{G}}} = - \frac{{3K}}{{f_{{\text{T}}}^{{2^{\prime } }} }}f_{{\text{T}}} \alpha _{{{\text{th}}}} n_{{\text{X}}}^{ - } \delta u_{{\text{X}}} - \frac{{3K}}{{f_{{\text{T}}}^{{2^{\prime } }} }}f_{{\text{T}}} \alpha _{{{\text{th}}}} n_{{\text{y}}}^{ - } \delta u_{{\text{y}}} - \frac{{3K}}{{f_{{\text{T}}}^{{2^{\prime } }} }}f_{{\text{T}}} \alpha _{{{\text{th}}}} n_{{\text{z}}}^{ - } \delta u_{{\text{z}}}\)
References
Ainsworth M, Kay D (1999) The approximation theory for the p-version finite element method and application to non-linear elliptic pdes. Numer Math 82(3):351–388
Ainsworth M, Kay D (2000) Approximation theory for the hp-version finite element method and application to the non-linear laplacian. Applied numerical mathematics 34(4):329–344
Amestoy P, Duff I, L’Excellent JY (2000) Multifrontal parallel distributed symmetric and unsymmetric solvers. Comput Methods Appl Mech Eng 184(2):501–520. doi:10.1016/S0045-7825(99)00242-X
Anand L, On H (1979) Hencky’s approximate strain-energy function for moderate deformations. J Appl Mech 46:78–82. doi:10.1115/1.3424532
Arnold DN, Brezzi F, Cockburn B, Marini LD (2002) Unified analysis of discontinuous galerkin methods for elliptic problems. SIAM J Numer Anal 39(5):1749–1779
Arnold DN, Brezzi F, Marini LD (2005) A family of discontinuous galerkin finite elements for the reissner-mindlin plate. J Sci Comput 22–23:25–45
Babuška I, Suri M (1987) The \(hp\) version of the finite element method with quasiuniform meshes. RAIRO-Modélisation mathématique et analyse numérique 21(2):199–238
Becker G, Noels L (2013) A full-discontinuous galerkin formulation of nonlinear kirchhofflove shells: elasto-plastic finite deformations, parallel computation, and fracture applications. Int J Numer Meth Eng 93(1):80–117. doi:10.1002/nme.4381
Behl M, Lendlein A (2007) Shape-memory polymers. Mater Today 10(4):20–28
Bonet J, Burton A (1998) A simple orthotropic, transversely isotropic hyperelastic constitutive equation for large strain computations. Comput Methods Appl Mech Eng 162(1):151–164
Chung DD (1994) CHAPTER 4—properties of carbon fibers. In: Chung DD (ed) Carbon fiber composites. Butterworth-Heinemann, Boston, pp. 65–78. doi:10.1016/B978-0-08-050073-7.50008-7. URL www.sciencedirect.com/science/article/pii/B9780080500737500087
Ciarlet P (2002) Conforming finite element methods for second-order problems, chapter 3, pp. 110–173. SIAM. doi:10.1137/1.9780898719208.ch3
Cockburn B, Karniadakis GE, Shu CW (2000) The development of discontinuous Galerkin methods. Springer, New York
Culebras M, Gómez CM, Cantarero A (2014) Review on polymers for thermoelectric applications. Materials 7(9):6701–6732
Douglas J, Dupont T (1976) Interior penalty procedures for elliptic and parabolic Galerkin methods. Springer, Berlin, pp 207–216. doi:10.1007/BFb0120591
Engel G, Garikipati K, Hughes TJR, Larson MG, Mazzei L, Taylor RL (2002) Continuous/discontinuous finite element approximations of fourth-order elliptic problems in structural and continuum mechanics with applications to thin beams and plates, and strain gradient elasticity. Comput Methods Appl Mech Eng 191(34):3669–3750
Fabré M, Binst J, Bocsan I, De Smet C, Ivens J (2012) Heating shape memory polymers with alternative ways: microwave and direct electrical heating. In: 15th European Conference on composite materials. University of Padova, pp. 1–8
Ferreira ADBL, Nóvoa PRO, Torres Marques A (2016) Multifunctional material systems: a state-of-the-art review. Compos Struct 151:3–35. doi:10.1016/j.compstruct.2016.01.028. Smart composites and composite structures In honour of the 70th anniversary of Professor Carlos Alberto Mota Soares
Georgoulis EH (2003) Discontinuous Galerkin methods on shape-regular and anisotropic meshes. University of Oxford D. Phil, Thesis
Geuzaine C, Remacle JF (2009) Gmsh: a 3-d finite element mesh generator with built-in pre- and post-processing facilities. Int J Numer Meth Eng 79(11):1309–1331. doi:10.1002/nme.2579
Gilbarg D, Trudinger NS (2015) Elliptic partial differential equations of second order. Springer, New York
Gudi T (2006) Discontinuous Galerkin methods for nonlinear elliptic problems. Ph.D. thesis, Indian Institute of Technology, Bombay
Gudi T, Nataraj N, Pani AK (2008) hp-Discontinuous Galerkin methods for strongly nonlinear elliptic boundary value problems. Numer Math 109(2):233–268
Hansbo P, Larson MG (2002) A discontinuous Galerkin method for the plate equation. Calcolo 39(1):41–59
Hansbo P, Larson MG (2002) Discontinuous Galerkin methods for incompressible and nearly incompressible elasticity by Nitsche’s method. Comput Methods Appl Mech Eng 191(17–18):1895–1908
Homsi L (2017) Development of non-linear electro-thermo-mechanical discontinuous Galerkin formulations. Ph.D. thesis, University of Liège, Belgium
Homsi L, Geuzaine C, Noels L (2017) A coupled electro-thermal discontinuous Galerkin method. J Comput Phys 348:231–258
Houston P, Robson J, Süli E (2005) Discontinuous Galerkin finite element approximation of quasilinear elliptic boundary value problems I: the scalar case. IMA J Numer Anal 25(4):726–749
Huang W, Yang B, An L, Li C, Chan Y (2005) Water-driven programmable polyurethane shape memory polymer: demonstration and mechanism. Appl Phys Lett 86(11):114,105
Issi JP (2003) Electronic and thermal properties of carbon fibers. World of Carbon. CRC Press, Boca Raton, pp 207–216. doi:10.1201/9780203166789.ch3
Karypis G, Kumar V (1998) A fast and high quality multilevel scheme for partitioning irregular graphs. SIAM J Sci Comput 20(1):359–392
Kaufmann P, Martin S, Botsch M, Gross M (2009) Flexible simulation of deformable models using discontinuous galerkin fem. Graphical Models 71(4):153–167. doi:10.1016/j.gmod.2009.02.002. http://www.sciencedirect.com/science/article/pii/S15240703090%00125. Special Issue of ACM SIGGRAPH / Eurographics Symposium on Computer Animation 2008
Keith JM, Janda NB, King JA, Perger WF, Oxby TJ (2005) Shielding effectiveness density theory for carbon fiber/nylon 6, 6 composites. Polym Compos 26(5):671–678
Langer R, Tirrell DA (2004) Designing materials for biology and medicine. Nature 428(6982):487–492
Lendlein A, Jiang H, Jünger O, Langer R (2005) Light-induced shape-memory polymers. Nature 434(7035):879–882
Leng J, Lan X, Liu Y, Du S (2011) Shape-memory polymers and their composites: stimulus methods and applications. Prog Mater Sci 56(7):1077–1135
Liang C, Rogers C, Malafeew E (1997) Investigation of shape memory polymers and their hybrid composites. J Intell Mater Syst Struct 8(4):380–386
Liu L (2012) A continuum theory of thermoelectric bodies and effective properties of thermoelectric composites. Int J Eng Sci 55:35–53
Liu R, Wheeler M, Dawson C (2009) A three-dimensional nodal-based implementation of a family of discontinuous galerkin methods for elasticity problems. Comput Struct 87(3–4):141–150. doi:10.1016/j.compstruc.2008.11.009
Liu R, Wheeler M, Dawson C, Dean R (2009) Modeling of convection-dominated thermoporomechanics problems using incomplete interior penalty Galerkin method. Comput Methods Appl Mech Eng 198(9–12):912–919. doi:10.1016/j.cma.2008.11.012
Lu H, Liu Y, Leng J, Du S (2010) Qualitative separation of the physical swelling effect on the recovery behavior of shape memory polymer. Eur Polymer J 46(9):1908–1914
Mahan GD (2000) Density variations in thermoelectrics. J Appl Phys 87(10):7326–7332
McBride A, Reddy B (2009) A discontinuous Galerkin formulation of a model of gradient plasticity at finite strains. Comput Methods Appl Mech Eng 198(21–26):1805–1820. doi:10.1016/j.cma.2008.12.034. Advances in Simulation-Based Engineering Sciences Honoring J. Tinsley Oden
Meng H, Li G (2013) A review of stimuli-responsive shape memory polymer composites. Polymer 54(9):2199–2221
Meng Q, Hu J (2009) A review of shape memory polymer composites and blends. Compos A Appl Sci Manuf 40(11):1661–1672
Miehe C (1994) Aspects of the formulation and finite element implementation of large strain isotropic elasticity. Int J Numer Meth Eng 37(12):1981–2004
Muliana A, Li KA (2010) Time-dependent response of active composites with thermal, electrical, and mechanical coupling effect. Int J Eng Sci 48(11):1481–1497
Noels L (2009) A discontinuous galerkin formulation of non-linear Kirchhoff–Love shells. Int J Numer Meth Eng 78(3):296–323
Noels L, Radovitzky R (2006) A general discontinuous Galerkin method for finite hyperelasticity. Formulation and numerical applications. Int J Numer Meth Eng 68(1):64–97
Noels L, Radovitzky R (2008) An explicit discontinuous Galerkin method for non-linear solid dynamics: formulation, parallel implementation and scalability properties. Int J Numer Meth Eng 74(9):1393–1420
Pilate F, Toncheva A, Dubois P, Raquez JM (2016) Shape-memory polymers for multiple applications in the materials world. Eur Polymer J 80:268–294
Prudhomme S, Pascal F, Oden J, Romkes A (2000) Review of a priori error estimation for discontinuous Galerkin methods. Technical report, TICAM, UTexas
Reed W, Hill T (1973) Triangular mesh methods for the neutron transport equation. Technical report LA-UR-73-479, Los Alamos Scientific Laboratory. http://www.osti.gov/scitech/servlets/purl/4491151
Romkes A, Prudhomme S, Oden J (2003) A priori error analyses of a stabilized discontinuous Galerkin method. Comput Math Appl 46(8):1289–1311
Rothe S, Schmidt JH, Hartmann S (2015) Analytical and numerical treatment of electro-thermo-mechanical coupling. Arch Appl Mech 85(9–10):1245–1264
Schmidt AM (2006) Electromagnetic activation of shape memory polymer networks containing magnetic nanoparticles. Macromol Rapid Commun 27(14):1168–1172
Spencer A (1982) The formulation of constitutive equations for anisotropic solids. Nijhoff, Amsterdam, pp. 3–26
Spencer AJM (1986) Modelling of finite deformations of anisotropic materials. In: John G, Joseph Z, Siavouche N-N (eds) Large deformations of solids: physical basis and mathematical modelling. Springer, Dordrecht, pp 41–52
Srivastava V, Chester SA, Anand L (2010) Thermally actuated shape-memory polymers: experiments, theory, and numerical simulations. J Mech Phys Solids 58(8):1100–1124
Sun S, Wheeler MF (2005) Discontinuous Galerkin methods for coupled flow and reactive transport problems. Appl Numer Math 52(2):273–298
Ten Eyck A, Celiker F, Lew A (2008) Adaptive stabilization of discontinuous Galerkin methods for nonlinear elasticity: motivation, formulation, and numerical examples. Comput Methods Appl Mech Eng 197(45–48):3605–3622. doi:10.1016/j.cma.2008.02.020
Ten Eyck A, Lew A (2006) Discontinuous Galerkin methods for non-linear elasticity. Int J Numer Meth Eng 67(9):1204–1243. doi:10.1002/nme.1667
Truster TJ, Chen P, Masud A (2015) Finite strain primal interface formulation with consistently evolving stabilization. Int J Numer Meth Eng 102(3–4):278–315. doi:10.1002/nme.4763
Truster TJ, Chen P, Masud A (2015) On the algorithmic and implementational aspects of a discontinuous Galerkin method at finite strains. Comput Math Appl 70(6):1266–1289. doi:10.1016/j.camwa.2015.06.035. http://www.sciencedirect.com/science/article/pii/S08981221150%03211
Vilčáková J, Sáha P, Quadrat O (2002) Electrical conductivity of carbon fibres/polyester resin composites in the percolation threshold region. Eur Polymer J 38(12):2343–2347
Wells GN, Dung NT (2007) AC 0 discontinuous Galerkin formulation for Kirchhoff plates. Comput Methods Appl Mech Eng 196(35–36):3370–3380
Wells GN, Garikipati K, Molari L (2004) A discontinuous Galerkin formulation for a strain gradient-dependent damage model. Comput Methods Appl Mech Eng 193(33–35):3633–3645
Wen S, Chung D (1999) Seebeck effect in carbon fiber-reinforced cement. Cem Concr Res 29(12):1989–1993
Wheeler MF (1978) An elliptic collocation-finite element method with interior penalties. SIAM J Numer Anal 15(1):152–161
Wu L, Tjahjanto D, Becker G, Makradi A, Jérusalem A, Noels L (2013) A micro-meso-model of intra-laminar fracture in fiber-reinforced composites based on a discontinuous galerkin/cohesive zone method. Eng Fract Mech 104:162–183
Yadav S, Pani A, Park EJ (2013) Superconvergent discontinuous Galerkin methods for nonlinear elliptic equations. Math Comput 82(283):1297–1335
Yang Y, Xie S, Ma F, Li J (2012) On the effective thermoelectric properties of layered heterogeneous medium. J Appl Phys 111(1):013,510
Zheng XP, Liu DH, Liu Y (2011) Thermoelastic coupling problems caused by thermal contact resistance: a discontinuous Galerkin finite element approach. Sci China Phys Mech Astron 54(4):666–674
Zhupanska OI, Sierakowski RL (2011) Electro-thermo-mechanical coupling in carbon fiber polymer matrix composites. Acta Mech 218(3–4):319–332
Acknowledgements
This project has been funded with support of the European Commission under the grant number 2012-2624/001-001-EM. This publication reflects the view only of the author, and the Commission cannot be held responsible for any use which may be made of the information contained therein. Computational resources have been provided by the supercomputing facilities of the “Consortium des Équipements de Calcul Intensif en Fédération Wallonie Bruxelles (CÉCI)” funded by the “Fond de la Recherche Scientifique de Belgique (FRS-FNRS)”.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Appendices
Appendix 1: Constitutive behaviors
The objective of this section is to summarize large deformation constitutive theories in order to model the response of Shape Memory Polymers composites (SMPC) subjected to a variety of ETM histories. The composite material system is obtained by defining two separate models, one for carbon fiber and another one for SMPs. For carbon fibers, a transversely isotropic hyperelastic model is considered while an elasto-visco-plastic model is considered for the shape memory polymers.
1.1 Material model of carbon fiber
Carbon fiber is a transversely isotropic material and subsequently the number of mechanical constants is reduced to 5 because of the in-plane isotropy:
The missing in-plane shear modulus \({{G}}^{\text {TT}}\) is obtained from \({\nu}^{\text {TT}}\) and \({{{E}}^{\text {T}}}\), with
In the previous relations, the subscript 3 or the superscript \(\text {L}\) refers to the fiber direction and 1, 2, or \(\text {T}\) is a direction transverse to the fiber direction. Along the longitudinal direction the Poisson ratios are not symmetric but instead satisfy \(\frac{{\nu}_{ij}}{{{E}}_{i}}=\frac{{\nu}_{ji}}{{{E}}_{j}}\).
In order to model the carbon fiber, we have considered the equation proposed by Bonet et al. [10], with some modifications proposed by Wu et al. [70], since the original formulation considered that \({\nu}^{\text {TL}} ={\nu}^{\text {TT}}\), to describe the isotropic hyperelastic solids in the large strain regime. In addition, we have added the thermal contribution, characterized by the thermal expansion term \(\alpha _{\mathrm{th}}\). In this formulation, the strain energy density \(\psi\) consists of an isotropic component \(\psi ^{\text {is}}\) and of an orthotropic transversely isotropic component \(\psi ^{\text {tr}}\) such that \(\psi =\psi ^{\text {is}}+\psi ^{\text {tr}}\). The Neo–Hookean equation is used to model the isotropic part, such that
where this energy density function has been defined by C. Miehe in [46], and where \(\alpha '_{\mathrm{th}} = \alpha _{\mathrm{th}} \frac{\lambda +2/3G^{\text {TT}}}{\lambda}\) in order to recover the usual dilation coefficient definition of isotropic materials. The orthotropic transversely isotropic component is obtained from a generalization of the model proposed by Bonet et al. [10] and enhanced by Wu et al. [70]:
where \({\text {I}}_4\) and \({\text {I}}_5\) denote the two new pseudo invariants of \(\mathbf{C}\) expressed as [57, 58],
with the unit vector \(\varvec{A}\) defining the main direction of orthotropy (fiber direction) in the undeformed configuration.
The parameters of the model Eq. (116), \(\lambda ,\,{{G}}^{\text {TT}},\, \alpha ^{\text {tr}},\,\beta ^{\text {tr}}\) and \(\gamma ^{\text {tr}}\) are obtained from the measured properties Eqs. (113, 114) as
The second Piola–Kirchhoff stress tensor can be obtained by differentiating the free energy in terms of the right Cauchy–Green strain tensor \(\pmb {S}=2\frac{\partial \psi}{\partial \pmb {{C}}}\) leading to
where \(\pmb {I}\) is the identity tensor, and with
Then the first Piola–Kirchhoff stress tensor is evaluated from the second Piola–Kirchhoff stress tensor as
The stiffness is computed following [10, 70].
1.2 Elasto-visco-plastic formulation for SMP
In this section, we summarize the work of Srivastava et al. [59] to model the SMP behavior above and below glass transition.
1.2.1 Kinematics
To model the inelastic response of the amorphous polymeric materials, it is assumed that the deformation gradient \(\pmb {{F}}\) may be multiplicatively decomposed into elastic and plastic parts
where \(\pmb {{F}}^{\text {e}({\alpha})}\) is the elastic distortion and \(\pmb {{F}}^{\text {p}({\alpha})}\) is the inelastic distortion with \(\pmb {{F}}^{\text {p}{(\alpha )}}(\varvec{X},\,0)=\pmb { I}\). In these equations we have considered the possibility to account for several mechanisms \({\alpha}= 1,2,3\). Moreover, the elastic decomposition of the deformation gradient can be written as
with the elastic right and left Cauchy–Green strain tensors respectively equal to
1.2.2 Elasto-visco-plasticity
The material may be idealized to be isotropic. Accordingly, all constitutive functions are presumed to be isotropic in character. Let us assume that the free energy has the separable form
where \(\varPhi _{\pmb {c}^{{\text {e}}(\alpha )}}\) represents a list of the principle invariants of \({\pmb {{C}}^{{\text {e}}(\alpha )}}\) and T is the temperature. The Cauchy stress is decomposed in terms of the mechanisms \(\pmb {\sigma}= \sum _{(\alpha )}\pmb {\sigma}^{(\alpha )}\) with
where \(\pmb {{S}}^{e(\alpha )}\) is the symmetric elastic second Piola–Kirchhoff stress
Moreover, the first Piola–Kirchhoff stress tensor can be computed following
The driving stress of the plastic flow is the symmetric Mandel stress, which is defined as
where \(\pmb { {M}}^{{\text {e}}(\alpha )}\) is the elastic Mandel stress, \(\pmb {{R}}^{{\text {e}}(\alpha )}\) is the rotation matrix, and where it has been assumed that \(\pmb {{C}}^{{\text {e}}(\alpha )}\) and \(\pmb {{S}}^{{\text {e}}(\alpha )}\) permute. The corresponding equivalent shear stress is given by
where \(\pmb {{M}}^{{\text {e}}(\alpha )}_0\) is the deviatoric part of the Mandel stress
and \(|\pmb {{M}}^{{\text {e}}(\alpha )}_0 |=\sqrt{\pmb {{M}}^{{\text {e}}(\alpha )}_0 :\pmb {{M}}^{{\text {e}}(\alpha )}_0}\) is the norm of the deviatoric part of the Mandel stress.
In order to account for the major strain-hardening and softening characteristics of polymeric materials observed during visco-plastic deformation, Srivastava et al. [59] have introduced macroscopic internal variables \(\pmb {\xi}^{(\alpha )}\) to represent important aspects of the microstructural resistance to plastic flow. The plastic flow follows
where each \(\pmb { {F}}^{\text{p}(\alpha )}\) is to be regarded as an internal variable part of \(\pmb {\xi}^{(\alpha )}\), and which is defined as a solution of the differential equation
where \(\pmb {D}^{\text{p}}\) is the plastic stretching tensor, and \(\dot{\epsilon}^{\text{p}(\alpha )}=\sqrt{2}|\pmb {D}^{\text{p}(\alpha )}|\) is the equivalent plastic shear strain rate.
Therefore for given \(\bar{\tau}^{(\alpha )}\) and \(\pmb {\varLambda}^{(\alpha )}= (\pmb { {C}}^{{\text {e}}(\alpha )},\,\pmb {\xi}^{(\alpha )},\, T)\) a list of constitutive variables, the equivalent plastic shear strain rate \(\dot{\epsilon}^{\text{p}(\alpha )}\) is obtained by solving a scalar strength relation such as
where the strength function \(\varUpsilon ^{(\alpha )}(\pmb {\varLambda}^{(\alpha )},\dot{\epsilon}^{\text{p}(\alpha )})\) is an isotropic function of its arguments.
1.2.3 Partial differential governing equations
In order to complete Eq. (7), y, the internal energy per unit mass, is defined as \(y=c_{\text{v}} T\), where the volumetric heat capacity per unit mass is a function of the glass transition temperature, and is defined as follows
Moreover, \(\bar{F}\), the body source of heat, is expressed as
where \({Q}_{{\text{r}}}\) is the scalar heat supply measured per unit reference volume and the last term of the right hand side is the thermo-elastic damping term which is neglected. Instead it is assumed that only a fraction \(\text{v}\) of the rate of the plastic dissipation contributes to the temperature change
where \(0\le v\le 1\).
The glass transition in amorphous polymers depends on the equivalent shear strain rate \(\dot{\epsilon}=\sqrt{2}|\pmb { D}_0|\) to which the material is subjected, where \(\pmb {{D}}_0=\pmb {{D}}-\frac{1}{3}\text{tr} \pmb {{D}}\; \pmb {{I}}\) denotes the total deviatoric stretching tensor, with
Eventually, the glass transition \(T_{\mathrm{g}}\) is calculated from the following expression
where \({T}_{\text{r}}\) is the reference glass transition temperature at low strain rate, \(\dot{\epsilon}\) is the shear strain rate, and \(\epsilon _{\text{ r }}\) is the reference strain rate.
1.2.4 The first micromechanism (\(\alpha =1\))
The first micromechanism (\(\alpha = 1\)) represents an elastic resistance due to intermolecular energetic bond-stretching and a dissipation due to the thermally-activated plastic flow following chain segment rotation and relative slippage of the polymer chains between neighboring cross-linkage points.
The following simple generalization of the classical strain energy function of infinitesimal isotropic elasticity is considered, and uses a logarithmic measure
of the finite elastic strain [4]. The form of the elastic free energy is thus defined as
where the deviatoric part of the logarithmic strain is denoted by \(\pmb {E}^{{\text {e}}}_0\), \(\tilde{f}(T)\) is an entropy contribution to the free energy related to the temperature dependent specific heat of the material, and where the temperature dependent parameters \({G}(T), \;K (T),\; \alpha _{\text{th}} (T)\) are respectively the shear modulus, bulk modulus, and the coefficient of thermal expansion.
The Mandel stress is thus obtained from
if \(\pmb { C}^{{\text {e}}{(1)}}\) and \(\pmb {{M}}^{{\text {e}}(1)}\) permute. It should be noted that in this work \(\pmb { E}^{{\text {e}}{(1)}}\) is computed by using a Taylor series approximation of Eq. (141), and not through the eigenvalue decomposition. Substituting Eq. (142) in Eq. (143), as \(|\pmb { {E}}^{{\text {e}}(1)}_0|= \pmb { {E}}^{{\text {e}}(1)}_0: \pmb { {E}}^{{\text {e}}(1)}_0\) one can get directly \(\pmb {{M}}^{{\text {e}}(1)}\) as
The thermal expansion is taken to have a bilinear temperature dependence, with the slope \(\alpha _{{{\text{th}}}} = \alpha _{{\text{r}}} {\text{}}\) above the glass transition temperature and the slope \(\alpha _{\mathrm{th}}=\alpha _{\text{gl}}\) below it.
Moreover, the evaluation equation for \({\pmb { {F}}}^{\text{p}(1)}\) follows Eqs. (133–134) with the thermally-activated relation for the equivalent plastic strain rate following
where \(\dot{\epsilon}^{\text{p}(1)}\) is the plastic strain rate, the parameter \(\epsilon _{0}^{\pmb .(1)}\) is a pre-exponential factor with units of 1/time, \(K_{{\text{B}}}\) is Boltzmann’s constant, \(V'\) is an activation volume, \(m^{(1)}\) is the sensitive parameter for the strain rate, Q(T) is the temperature dependent activation energy, and \(\tau ^{e(1)}\) denotes a net shear stress for the thermally activated flow
with \(\alpha _{\text{p}}\geqslant 0\) a parameter introduced to account for the pressure sensitivity. The term \(\exp {(-\dfrac{1}{\xi})}\) in Eq. (145) represents a concentration of flow defects, with
For the first micromechanism, besides the plastic strain gradient, the list \(\varvec{\xi}^1=( \varphi ,{S}_{{\text{a}}},{S}_{{\text{b}}})\) of internal variables consists of three positive scalars, where the variable \(\varphi \ge 0\) and \({S}_{{\text{a}}} \ge 0\) are introduced to model the yield peak which is observed in the stress-strain response of glassy polymers and \({S}_{{\text{b}}} \ge 0\) is introduced to model the isotropic hardening at high strain. The evolution equations of \(\dot{S}_{{\text{a}}}\) and \(\dot{\varphi}\) are governed by
with \(\varphi ^*\) as
which represents the temperature and strain rate dependency of \(\varphi\), where \(z,\,r, \,{h}_{\mathrm{g}}\), and s are taken to be constants. In particular we have introduced \(h_{\mathrm{g}}\) to get a small value for \(\varphi ^*\) when \(T>T_{{\text {g}}}\), instead of 0 in order to improve the convergence of the numerical model. Then the evolution of \(S_{{\text{b}}}\) is governed by
where \(\bar{\lambda}\) is an effective stretch which increases or decreases as the overall stretch increases or decreases, and where \(H_{{\text{b}}}(T)\) is a temperature dependent hardening parameter.
The temperature evolutions of \(H_{{\text{b}}}(T)\), Q(T) , G(T), and of the Poisson ratio \({\nu}({T})\) follow a law
where \(\mathbf{\cdot}_{\mathrm{gl}}\) and \(\mathbf{\cdot}_{\text{r}}\) are the values in glassy and rubbery regions, and where \(L_\mathbf{\cdot}\) represents the slope of the temperature variation of \(\mathbf{\cdot}\), and takes the value of \(L_\mathbf{\cdot} =L_{\mathbf{\cdot} _{\mathrm{gl}}}\,\text{if}\, T\le {T}_{\mathrm{g}}\) and \(L_\mathbf{\cdot} =L_{\mathbf{\cdot} _{\mathrm{r}}}\,\text{if}\, T> {T}_{\mathrm{g}}\). The temperature dependence of the bulk modulus K(T) is then obtained by using the standard relation for isotropic materials \({K}(T)= {G}(T)\frac{2(1+\nu (T))}{3(1-2\;\nu (T))}\).
1.2.5 The second micromechanism (\(\alpha =2\))
The second micromechanism (\(\alpha = 2\)) represents the molecular chains between mechanical crosslinks. At temperatures below \(T_{\mathrm{g}}\) the polymer exhibits a significant amount of mechanical crosslinking which disintegrates when the temperature is increased above \(T_{\mathrm{g}}\).
Only deviatoric contributions are considered in the free energy function
where \(\bar{\pmb { {C}}}^{{\text {e}}(2)}=\bar{\pmb { {F}}}^{{\text {e}}(2)\text {T}} \bar{\pmb { \text {F}}}^{{\text {e}}(2)}=\text {J}^{-\frac{2}{3}} \pmb {{C}}^{{\text {e}}(2)}\) denotes the distrotional (or volume preserving) right Cauchy strain tensor, the parameter \(I^{(2)}_{{\text{m}}}\) is taken to be temperature constant, and where \(\mu ^{(2)}\) is the rubbery shear modulus, which follows
with \(\mu _{\mathrm{g}}\) the value of \(\mu ^{(2)}\) at the glass transition temperature, and N a parameter that represents the slope of temperature variation on a logarithmic scale.
The corresponding Mandel stress is evaluated from Eqs. (130) and (153) as
where \(\bar{\pmb { {C}}}^{{\text {e}}(2)}_0=\bar{\pmb { {C}}}^{{\text {e}}(2)}-\frac{1}{3}\text {tr} \bar{\pmb { {C}}}^{{\text {e}}(2)} \pmb { I}\) is the deviatoric part of \(\pmb { {C}}^{{\text {e}}(2)}\) the right Cauchy Green tensor. Clearly, as \(\bar{\pmb {{C}}}^{{\text {e}}(2)}\) and \({\pmb {{C}}}^{{\text {e}}(2)}\) permute, \({\pmb { {M}}^{{\text {e}}(2)}}\) and \(\bar{\pmb { {C}}}^{{\text {e}}(2)}\) permute as well.
For the second mechanism, the equivalent plastic strain rate follows
where \(\dot{\epsilon}^{(2)}_0\) is a reference plastic shear strain rate, \(m^{(2)}\) is the positive valued strain rate sensitivity parameter, and \(S^{(2)}\) is a temperature dependent parameter which follows (152).
1.2.6 The third micromechanism (\(\alpha =3\))
The third micromechanism (\(\alpha = 3\)) introduces the molecular chains between chemical crosslinks and represents the resistance due to changes in the free energy upon stretching of the molecular chains between the crosslinks.
The free energy is a function of the deviatoric tensor \(\bar{\pmb {{C}}}=\bar{\pmb { {F}}}^{\text {T}} \bar{\pmb {{F}}}= {J}^{-\frac{2}{3}}\pmb {{C}}\), and is given by a deviatoric form
where the material constants \(\mu ^{(3)}>0\) and \(I^{(3)}_{{\text{m}}}>0\) are assumed to be temperature-independent.
The free energy (157) yields the corresponding second Piola stress \(\pmb {{S}}^{(3)}\) as
1.2.7 Finite increment form of the shape memory polymer constitutive law
The constitutive laws are formulated in a finite strain setting and solved following the predictor-corrector scheme during the time interval [\(t_{n} ;\;t_{n+1}\)], where we use the subscript \(n\) for the previous time \(t_{n}\) and \(n+1\) for the current time \(t_{n+1}\). The formulation can be summarized as follows:
-
Prediction step The plastic deformation gradient is initialized to the value at the previous step \(\pmb {{F}}^{\text {p}({\alpha})}_{(\text {pr})}=\pmb {{F}}^{\text {p} ({\alpha})}_{n}\), and the elastic deformation follows
$$\begin{aligned} \pmb {{F}}^{\text {e}({\alpha})}_{n+1}=\pmb {{F}}_{n+1}\pmb { {F}}^{\text {p}(\alpha )^{-1}}_{n}\,. \end{aligned}$$(159) -
Correction step In this step we solve the system of equations that has been presented for each mechanism. To extract the plastic increment using the evaluation equation of the plastic deformation gradient during the time step between the configurations n and n+1, we consider
$$\begin{aligned} \pmb {{F}}^{\text{p}({\alpha})}_{n+1}=\exp (\Delta {\pmb {\text {D}}}^{\text{p}({\alpha})})\pmb { {F}}^{\text{p}({\alpha})}_{n}\,, \end{aligned}$$(160)with
$$\begin{aligned} {\Delta}\pmb {{D}}^{\text{p}({\alpha})}=(\epsilon ^{\text{p}({\alpha})}_{n+1}-\epsilon ^{\text{p}({\alpha})}_{n})\frac{\pmb {{M}}^{{\text {e}}({\alpha})}}{2\bar{\tau}^{({\alpha})}}=\Delta {\epsilon}^{\text{p}({\alpha})} \left( \dfrac{\pmb { {M}}^{{\text {e}}({\alpha})}}{2\bar{\tau}^{({\alpha})}} \right) \,. \end{aligned}$$(161)
More details about the predictor-corrector algorithm and the stiffness computation can be found in [26].
Appendix 2: The finite element formulation
Using the interpolations (33–34), the gradients are readily obtained by:
where \(\nabla _0{{N}}^{a}_{{\mathbf{u}}}\) and \(\nabla _0{\mathbf{N}}^{a}_{{\mathbf{M}}}= \left( \begin{array}{cc} \nabla _0{{N}}^{a}_{{\text {f}}_{{\text {V}}}} &{} \varvec{0}\\ \varvec{0} &{} \nabla _0{{N}}^{a}_{{\text {f}}_{\mathrm{T}}} \end{array} \right)\) are the gradients of the shape functions at node a.
1.1 Nodal forces
The expressions of the nodal forces (36) are obtained by substituting the interpolations (33–34) and (162–163) in the weak formulation (31).
First the mechanical contribution reads
with the three mechanical contributions to the interface forces related to the degrees of freedom of the nodes \(a\pm\) on each side of the interface elements readingFootnote 1
Secondly, the electro-thermal contributions read
with the three electric contributions to the interface forces\(^{1}\)
1.2 Tangent operators
In order to derive the tangent matrix \(\pmb {\mathbb {K}}_{{\mathbf{G}}}^{{ab}}=\frac{ \partial {\mathbf{F}}_{\mathrm{ext}}^{a}}{\partial {{\mathbf{G}}}^{b}}-\frac{\partial {\mathbf{F}}_{\mathrm{int}}^{a}}{\partial {{\mathbf{G}}}^{b}}-\frac{\partial {\mathbf{F}}_{\mathrm{I}}^{a}}{ \partial {{\mathbf{G}}}^{b}}\), the system (38) is rewritten
The stiffness matrix has been decomposed into four sub-matrices as shown in Eq. (176) with respect to the discretization of the five independent field variables (d for displacement \({\mathbf{u}}\), and 2 for \({\mathbf{M}}\) (for \({{\text {f}}_{\mathrm{V}}}\), and \({{\text {f}}_{\mathrm{T}}}\))), and can be obtained in a straighforward way from the internal forces, see details in [26].
Appendix 3: Derivation of the numerical properties
1.1 Taylor’s remainders
The remainder terms of Eqs. (60–60) are obtained by defining \(\mathbf{V}^{{\text {t}}}={\mathbf{G}}+ t(\mathbf{V}-{\mathbf{G}})\), \(\nabla \mathbf{Q}^{{\text {t}}}=\nabla {\mathbf{G}}+ t(\nabla \mathbf{Q}-\nabla {\mathbf{G}})\). They can thus be evaluated by
and by
with the partial derivatives \(\mathbf{w}_{{\mathbf{G}}{} {\mathbf{G}}}({\mathbf{G}},\nabla {\mathbf{G}})=\mathbf{v}_{{\mathbf{G}}{} {\mathbf{G}}}({\mathbf{G}}) \nabla {\mathbf{G}}\) and \(\mathbf{w}_{{\mathbf{G}}\nabla {\mathbf{G}}}({\mathbf{G}})=\mathbf{v}_{{\mathbf{G}}}({\mathbf{G}})\) of \({\mathbf{w}}({\mathbf{G}},\nabla {\mathbf{G}})\), since \(\mathbf{w}_{\nabla {\mathbf{G}}\nabla {\mathbf{G}}}({\mathbf{G}})=0\).
The remainder terms of Eqs. (62–63) read
and
Finally, the remainder terms of Eqs. (64–65) read
1.2 Application of the Taylor’s expansion
The first term of Eq. (70) is rewritten, using the Taylor’s expansion defined in Eq. (60), as
Similarly, the second term of Eq. (70) is rewritten, using the Taylor’s expansion defined in Eq. (62), as
Likewise, the third term is rewritten, using the Taylor’s expansion defined in Eq. (60), as
The fifth term of Eq. (70) is developed by using the definition of \(\mathbf{p}^{\text {T}}({\mathbf{G}})={\mathbf{G}}^{\text {T}}{} \mathbf{o}^{\text {T}}({\mathbf{G}})\) and using the Taylor’s expansion defined in Eq. (64), leading to
However, as \(\mathbf{p}^{\text {T}}_{{\mathbf{G}}}=-{{\mathbf{o}}}^{\text {T}}({\mathbf{G}})\), using Eq. (66), this last term is also rewritten as
Finally, using the definition of the \({\tilde{ \mathbf{\cdot} }}\) operator \({\mathbf{G}}^{\text {T}}{{\mathbf{o}}}^{\text {T}}({\mathbf{G}}^{\prime}) \delta {\mathbf{G}}_{{\mathbf{n}}}={\mathbf{G}}^{\text {T}}_{{\mathbf{n}}}\tilde{{\mathbf{o}}}^{\text {T}}({\mathbf{G}}^{\prime}) \delta {\mathbf{G}}\) Footnote 2, and Eq. (186) is rewritten as
1.3 General properties of the finite element method and Hilbert spaces
The following properties will be used in the numerical properties derivation.
Lemma 5
(Interpolation inequality) For all \({\mathbf{G}}\in \left( \text {H}^{s}(\varOmega ^{{\text {e}}})\right) ^{n}\) there exists a sequence \({\mathbf{G}}^{\text {h}}\in \left( \mathbb {P}^{k}(\varOmega ^{{\text {e}}})\right) ^{n}\) and a positive constant \({\text{C}}_{{\mathcal {D}}}^{k}\) depending on s and k but independent of \({\mathbf{G}}\) and \(h_{\mathrm{s}}\), such that
-
1.
for any \(0\le n\le s\)
$$\begin{aligned} \begin{aligned}&\parallel {\mathbf{G}}-{\mathbf{G}}^{\text {h}}\parallel _{\mathrm{H}^{n}(\varOmega ^{{\text {e}}})} \le {\text{C}}_{{\mathcal {D}}}^{k} h_{\mathrm{s}}^{\mu -n} \parallel {\mathbf{G}}\parallel _{\mathrm{H}^{s}(\varOmega ^{{\text {e}}})}, \end{aligned} \end{aligned}$$(188) -
2.
for any \(0\le n\le s-1+\frac{2}{r}\)
$$\begin{aligned} \begin{aligned}&\parallel {\mathbf{G}}-{\mathbf{G}}^{\text {h}}\parallel _{\text{W}^{n}_{r}(\varOmega ^{{\text {e}}})} \le {\text{C}}_{{\mathcal {D}}}^{k} h_{\mathrm{s}}^{\mu -n-1+\frac{2}{r}} \parallel {\mathbf{G}}\parallel _{\mathrm{H}^{s}(\varOmega ^{{\text {e}}})},\text { if } d=2, \end{aligned} \end{aligned}$$(189) -
3.
for any \(s>n+\frac{1}{2}\)
$$\begin{aligned} \begin{aligned}&\parallel {\mathbf{G}}-{\mathbf{G}}^{\text {h}}\parallel _{\mathrm{H}^{n}(\partial \varOmega ^{{\text {e}}})} \le {\text{C}}_{{\mathcal {D}}}^{k} h_{\mathrm{s}}^{\mu -n-\frac{1}{2}} \parallel {\mathbf{G}}\parallel _{\mathrm{H}^{s}(\varOmega ^{{\text {e}}})}, \end{aligned} \end{aligned}$$(190)
where \(\mu ={\text {min}}\left\{ s,\, k+1\right\}\).
The proof of the first and third properties can be found in [7], and the proof of the second property in the particular case of \({d}=2\) can be found in [1, 2], see also the discussion by [23].
Remarks
-
(1)
The approximation property in (2) is still valid for \(r=\infty\), see [1].
-
(2)
For \({\mathbf{G}}\in \text{X}_{s}\), let us define the interpolant \({\text {I}}_{\mathrm{h}}{} {\mathbf{G}}\in \text{X}^{k}\) by \({\text {I}}_{\mathrm{h}}{} {\mathbf{G}}|_{\varOmega ^{{\text {e}}}}={\mathbf{G}}^{\text {h}}( {\mathbf{G}}|_{\varOmega ^{{\text {e}}}})\), which means \({\text {I}}_{\mathrm{h}}{} {\mathbf{G}}\) satisfies the interpolant inequality property provided in Lemma 5 on \(\varOmega _{\mathrm{h}}\), see [28].
Lemma 6
(Trace inequality) For all \({\mathbf{G}}\in \left( \text {H}^{s+1}(\varOmega ^{{\text {e}}})\right) ^{n}\), there exists a positive constant \({\text{C}}_{{\mathcal {T}}}\), such that
where \(s=0, 1\) and \(r=2, 4\), or in other words
The first equation, \(s=0\) and \(r=2\), is proved in [52], and the second one, \(r=4\) and \(s=0\), is proved in [22].
Lemma 7
(Trace inequality on the finite element space) For all \({\mathbf{G}}_{\mathrm{h}}\in \left( \mathbb {P}^{k}(\varOmega ^{{\text {e}}})\right) ^{n}\) there exists a constant \({\text{C}}_{{\mathcal {K}}}^{k}>0\) depending on k, such that
where \({\text{C}}_{{\mathcal {K}}}^{k}=\text {sup}_{{\mathbf{G}}_{\mathrm{h}}\in {(\text{P}_{k}(\varOmega ^{\text {e}})})^n}\frac{{h}_{\text s}\parallel \nabla {\mathbf{G}}_{\mathrm{h}}\parallel ^2_{\mathrm{L}^2(\partial \varOmega ^{{\text {e}}})}}{\parallel \nabla {\mathbf{G}}_{\mathrm{h}}\parallel ^2_{\mathrm{L}^2(\varOmega ^{\text {e}})}}\) is a constant which depends on the degree of the polynomial approximation only with \({h}_{\text s}=\frac{\vert \varOmega ^{\text {e}}\vert}{\vert \partial \varOmega ^{\text {e}}\vert}\), see [25] for more details.
Lemma 8
(Inverse inequality) For \({\mathbf{G}}_{\mathrm{h}}\in \left( \mathbb {P}^{k}(\varOmega ^{{\text {e}}})\right) ^{n}\) and \(r\ge 2\), there exists \({\text{C}}_{{\mathcal {I}}}^{k}>0\), such that
The proof of these properties can be found in [12, Theorem 3.2.6]. Note that Eqs. (194–195) involve the space dimension \({d}\).
Lemma 9
(Relation between energy norms on the finite element space) From [69], for \({{\mathbf{G}}_{\mathrm{h}}}\in \text{X}^{k}\), there exists a positive constant \({\text{C}}^{k}\), depending on k, such that
The demonstration directly follows by bounding the extra terms \(\sum _{{\text {e}}} {h}_{\text s}\parallel {\mathbf{G}}\parallel ^2_{\mathrm{H}^1(\partial \varOmega ^{{\text {e}}})}\) of the norm defined by Eq. (77), in comparison to the norm defined by Eq. (76), using successively the trace inequality, Eq. (192), and the inverse inequality, Eq. (196), for the first term, and the trace inequality on the finite element space, Eq. (193), for the second term.
Lemma 10
(Energy bound of interpolant error) Let \({\mathbf{G}}^{{\text {e}}}\in \text{X}_{s}, s\ge 2\), and let \({\text {I}}_{\mathrm{h}}{} {\mathbf{G}}\in \text{X}^k\), be its interpolant. Therefore, there is a constant \({\text{C}}^{k}>0\) independent of \(h_{\mathrm{s}}\), such that
with \(\mu =\text {min}\left\{ {s,\, k+1}\right\}\). The proof follows from Lemma 5, Eq. (188), and Eq. (190), applied on the mesh dependent norm (77).
1.4 The bound of the non-linear term \({\mathcal {N}}({{\mathbf{G}}}^{{\text {e}}},\mathbf{y};\delta {\mathbf{G}}_{\mathrm{h}})\)
1.4.1 Intermediate bounds
To bound the terms of \({\mathcal {N}}({{\mathbf{G}}}^{{\text {e}}},\mathbf{y};\delta {\mathbf{G}}_{\mathrm{h}})\), we have recourse to the following intermediate bounds, which are derived for the particular case of \(d=2\).
Lemma 11
(Intermediate bounds) Let \(\pmb \xi ={\text {I}}_{\mathrm{h}}{} {\mathbf{G}}-{\mathbf{y}},\,\delta {\mathbf{G}}_{\mathrm{h}}\in \text{X}^{k}\), \(\pmb \eta ={\mathbf{G}}^{{\text {e}}} -{\text {I}}_{\mathrm{h}}{} {\mathbf{G}}\in \text{X}\) and \(\pmb \zeta =\pmb \xi + \pmb \eta\). The terms in \(\pmb \xi\) can be bounded by recourse to the trace inequality, see Lemma 6, and to the inverse inequality, see Lemma 8, while bounding the terms in \(\pmb \eta\) makes use of the trace inequality, see Lemma 6 and of the interpolation inequality, see Lemma 5 for \(d=2\). Using the definition of the ball (91–92) thus leads to the different contributions, see [23, 26] for details,
with \(\mu =\text {min} \left\{ s,\, k+1\right\}\). Moreover, using the inverse inequality, see Lemma 8, one has
1.4.2 Bounds of the different contributions
The bound of \({\mathcal {N}}({{\mathbf{G}}}^{{\text {e}}},\mathbf{y};\delta {\mathbf{G}}_{\mathrm{h}})\) follows from the argumentation reported in [23] and is nominated by the term with the largest bound.
The first term of \({\mathcal {N}}({{\mathbf{G}}}^{{\text {e}}},\mathbf{y};\delta {\mathbf{G}}_{\mathrm{h}})\), defined in Eq. (74), can be expanded using Eq. (61) as
The two term of the right hand side of Eq. (206) are bounded by using the generalized Hölder’s inequality, the generalized Cauchy–Schwartz’ inequality, the definition of \(C_{{\text {y}}}\) in Eq. (67), and the bounds (199, 200, 201, and 202) as
Combining the above results leads to
The second term of \({\mathcal {N}}({{\mathbf{G}}}^{{\text {e}}},\mathbf{y};\delta {\mathbf{G}}_{\mathrm{h}})\), defined in Eq. (74), becomes by using Eq. (61),
The two terms of the right hand side of Eq. (210) are bounded by using the generalized Hölder's inequality, the generalized Cauchy–Schwartz’ inequality, the definition of \(C_{{\text {y}}}\) in Eq. (67), and the bounds (202, 204), yielding
We now substitute Eqs. (211, 212) in Eq. (210) to obtain the final bound of the second term of \(\mathcal {N}({{\mathbf{G}}}^{{\text {e}}},\mathbf{y};\delta {\mathbf{G}}_{\mathrm{h}})\) as
Furthermore, for the third term of \({\mathcal {N}}({{\mathbf{G}}}^{{\text {e}}},\mathbf{y};\delta {\mathbf{G}}_{\mathrm{h}})\) as decomposed in Eq. (74), using Taylor’s expansion as in Eq. (60–60), the generalized Hölder’s inequality, the generalized Cauchy–Schwartz’ inequality, the definition of \(C_{{\text {y}}}\) in Eq. (67), and the bounds (202, 203), leads to
Likewise, the fourth term of \({\mathcal {N}}({{\mathbf{G}}}^{{\text {e}}},\mathbf{y};\delta {\mathbf{G}}_{\mathrm{h}})\) defined in Eq. (74) is bounded using a Taylor’s expansion as in Eqs. (60–60), the generalized Hölder's inequality, the generalized Cauchy–Schwartz’ inequality, the definition of \(C_{{\text {y}}}\) in Eq. (67), and the bounds (202, 203) leading to
Then the bound of the fifth term of \({\mathcal {N}}({{\mathbf{G}}}^{{\text {e}}},\mathbf{y};\delta {\mathbf{G}}_{\mathrm{h}})\) defined in Eq. (74) is derived using Eq. (231) developed in Appendix section “Declaration related to the fifth term of \({\mathcal{N}}({{\mathbf{G}}}^{{\text{e}}} ,{\mathbf{y}};\delta {{\mathbf{G}}}_{{\text{h}}} )\) ”, following
where \(\mathbf{I}\) is a matrix of unit norm and has the same size of \(\bar{{\mathbf{o}}}_{{\mathbf{G}}}^{\text {T}}\). Using the generalized Hölder's inequality, the generalized Cauchy–Schwartz’ inequality, and the bounds (202, 203) one has
Combining Eqs. (217 and 218) leads to the final bound
Finally to bound the last term of \({\mathcal {N}}({{\mathbf{G}}}^{{\text {e}}},\mathbf{y};\delta {\mathbf{G}}_{\mathrm{h}})\) defined in Eq. (74), we rewrite it using Eq. (63) as
The two contributions are bounded using the generalized Hölder’s inequality, the generalized Cauchy–Schwartz’ inequality, the definition of \(C_{{\text {y}}}\) in Eq. (67), the bounds (200, 201), and the inverse inequality of Lemma 8, following
Substituting Eqs. (221, 222) in Eq. (220) leads to
Combining Eqs. (209, 213, 214, 215, 216, and 223), yields
Finally, using the definition of the energy norm (77), this results yields the bound (95) of \({\mathcal {N}}({{\mathbf{G}}}^{{\text {e}}},\mathbf{y};\delta {\mathbf{G}}_{\mathrm{h}})\).
1.4.3 Declaration related to the fifth term of \({\mathcal{N}}({{\mathbf{G}}}^{{\text{e}}} ,{\mathbf{y}};\delta {{\mathbf{G}}}_{{\text{h}}} )\)
Using the identities \(\left[\kern-0.15em\left[ {{\text{ab}}} \right]\kern-0.15em\right] = \left[\kern-0.15em\left[ {\text{a}} \right]\kern-0.15em\right]\left\langle {\text{b}} \right\rangle + \left\langle {\text{a}} \right\rangle \left[\kern-0.15em\left[ {\text{b}} \right]\kern-0.15em\right]\) and \(\left\langle {\text{a}} \right\rangle \left\langle {\text{b}} \right\rangle = \left\langle {{\text{ab}}} \right\rangle - \frac{1}{4}\left[\kern-0.15em\left[ {\text{a}} \right]\kern-0.15em\right]\left[\kern-0.15em\left[ {\text{b}} \right]\kern-0.15em\right]\) on \({\partial _{{\text {I}}}\varOmega _{\mathrm{h}}}\), the term
\(\llbracket ({\mathbf{G}}^{{\text {e}}}-{\mathbf{G}}_{\mathrm{h}})^{\text {T}} \bar{{\mathbf{o}}}^{\text {T}}_{{\mathbf{G}}}({{\mathbf{G}}}_{\mathrm{h}}) ({\mathbf{G}}^{{\text {e}}}-{\mathbf{G}}_{{\text {h}}}) \rrbracket \left\langle \delta {\mathbf{G}}_{\mathrm{h}_{\mathbf{n}}}\right\rangle\) can be rewritten with an abuse of notations on the product operator as
Now, we need to solve explicitly the term \(\llbracket ({\mathbf{G}}^{{\text {e}}}-{\mathbf{G}}_{\mathrm{h}})^{\text {T}} \bar{{\mathbf{o}}}^{\text {T}}_{{\mathbf{G}}}({{\mathbf{G}}}_{\mathrm{h}})\rrbracket\), where \(\bar{{\mathbf{o}}}_{{\mathbf{G}}}({{\mathbf{G}}}_{\mathrm{h}})\) corresponds to \(-\bar{\mathbf{p}}_{{\mathbf{G}}{} {\mathbf{G}}}({{\mathbf{G}}}_{\mathrm{h}})\) defined in Eq. (181) with \({\mathbf{p}}_{{\mathbf{G}}{} {\mathbf{G}}}(\mathbf{V}^{{\text {t}}})=-{\mathbf{o}}_{{\mathbf{G}}}(\mathbf{V}^{{\text {t}}})\), yielding
with \(\mathbf{V}^{{\text {t}}}={\mathbf{G}}^{{\text {e}}}+{\text {t}}({\mathbf{G}}_{\mathrm{h}}-{\mathbf{G}}^{{\text {e}}})\). As \({\mathbf{o}}_{\mathbf{{G}}}\) only involves terms in \(\frac{2}{{\text {f}}_{\mathrm{T}}^3}\), we compute \(\bar{\alpha}\) the nonzero component.
For simplicity, let us define \(\lambda\) as
It can be noticed that to evaluate \(\llbracket ({\mathbf{G}}^{{\text {e}}}-{\mathbf{G}}_{\mathrm{h}})^{\text {T}} \bar{{\mathbf{o}}}^{\text {T}}_{{\mathbf{G}}}({{\mathbf{G}}}_{\mathrm{h}})\rrbracket\), we need \(\lambda ({\text {f}}_{\mathrm{T}}^{\text {e}}-{\text {f}}_{\mathrm{T}})\) which reads
and the jump of the last result is
Hence considering this equation in the matrix form, then substituting it in Eq. (225), and using the definition of \(C_{{\text {y}}}\) in Eq. (67), lead to
where \({\mathbf{I}}\) is a matrix of unit norm and has the same size of \(\overline{{\mathbf{o}}} _{{{\mathbf{G}}}}^{{\text{T}}}\).
Rights and permissions
About this article
Cite this article
Homsi, L., Noels, L. A discontinuous Galerkin method for non-linear electro-thermo-mechanical problems: application to shape memory composite materials. Meccanica 53, 1357–1401 (2018). https://doi.org/10.1007/s11012-017-0743-9
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11012-017-0743-9