The decomposition \(\mathbf F = \mathbf F^{\text e}\mathbf F^{\text b}\), material symmetry, and plastic irrotationality for solids that are isotropic-viscoplastic or amorphous (Q2485961): Difference between revisions

We present a full description of the paper, our comments concerning the proposed model and we make references to other results already existing in the literature. The paper deals with two types of materials: isotropic and amorphous, which were defined within the constitutive framework of finite elasto-plasticity based on the multiplicative decomposition of the deformation gradient \({\mathbf F}\) into its elastic and plastic components, \({\mathbf F}= {\mathbf F}^e {\mathbf F}^p.\) Three configurations are involved in the modeling: (i) the reference configuration and (ii) the deformed configuration, which are global and (iii) \textit{relaxed space} which is local. \({\mathbf F}^p\) characterizes the passage from the reference configuration to the \textit{relaxed space}, at every material point. The authors develop a constitutive framework to describe the constitutive equations for \(\psi-\) the \textit{free energy}, \({\mathbf T},\) which plays the role of \textit{Cauchy stress} and \({\mathbf T}^p -\) \textit{microstress}, in terms of \({\mathbf F}^e, {\mathbf F}^p\) and \({\mathbf D}^p -\) the \textit{inelastic stretch-rate}. The \textit{inelastic spin} \({\mathbf W}^p\) is not included among the arguments of the constitutive functions and no supposition about zero plastic spin is made. The internal power is expanded by the stress \({\mathbf T}\) power conjugate to \({\mathbf L}^e = \dot{\mathbf F}^e ({\mathbf F}^e)^{-1}\) and the microstress \({\mathbf T}^p\) has been introduced under the assumption that is power conjugate to \({\mathbf L}^p = \dot{\mathbf F}^p ({\mathbf F}^p)^{-1}.\) Here \({\mathbf L}^p\) is traceless field. \textit{The principle of virtual} power requires that the external and internal powers be balanced, given for any part \({\mathcal P}\) of the body, for all \textit{generalized virtual velocity}. The consequences of the principle of virtual power are proved in the paper. (1) The \textit{local force balance} (or linear momentum balance equation) together with the traction condition follow for \(\tilde{\mathbf L}^p =0,\) but for any arbitrary given \(\tilde{\mathbf v},\) while (2) \textit{microforce balance} \({\mathbf S}_{0} = {\mathbf T}^p - {\mathbf M}^p,\) is derived for \(\tilde{\mathbf v} = 0,\) but \(\tilde{\mathbf L}^p\) arbitrary given. \({\mathbf M}^p\) is the tensor of virtual external microforces conjugated to \({\mathbf L}^p,\) that can be added in the expression of the external power. The argument ''\textit{various kinematical fields are independently varied} without consideration of forces need to incur such variations'' was fundamental in deriving these balance equations. Microforce balance characterizes the interaction between virtual forces associated with elastic response and inelastic response of the material. Reviewer's comments: Kröner (1960) made reference to multiplicative decomposition in connection with the incompatibility of the plastic deformation in material with continuously distributed dislocations. The multiplicative decomposition is generally related with a pure elastic strain during unloading process by Lee and Liu (1968). \({\mathbf S}_{0}\) generally non-symmetric stress tensor, appears in Mandel (1971), Teodosiu (1975), Van der Giessen (1993), Lubliner (1986) and so on. Several restrictions on the constitutive functions were found in the paper, by exploring several assumptions that will be briefly listed below. (a) \textit{The frame indifference assumption}, which requires \({\mathbf F}^p\) (''since observers view only the deformed configuration'') and internal power be invariant under a change of frame; (b) \textit{The free energy imbalance}, which means that the temporal increase in free energy of any part of the body be less than or equal to the power expended on \({\mathcal P};\) (c) \textit{The material symmetry,} discussed for two classes of transformations, of the \textit{reference} and \textit{relaxed space}, respectively, configurations. Both of them were viewed in unimodular group, {U}\(_{nim}\), and assuming that under such kind of transformations \({\mathbf T}: {\mathbf D}^e -\) the internal elastic power and \({\mathbf T}^p : {\mathbf L}^p -\) the internal irreversible power are invariant. Consequences of (a): \({\mathbf T}^p\) is invariant and \({\mathbf T}\) is objective and symmetric. Consequences of (b) are: (1) the free energy function \(\hat{\psi}\) is not dependent on \({\mathbf D}^p;\) (2) \(\hat{\psi}\) is potential with respect to \({\mathbf F}^e\) for \({\mathbf T},\) and with respect to \({\mathbf F}^p\) for \({\mathbf T}^p_{en}\) the part of \({\mathbf T}^p\) independent of \({\mathbf D}^p,\) respectively ( called \textit{energetic} part); (3) the part of microstress dependent on \({\mathbf D}^p,\) \({\mathbf T}^p_{dis}\) (called the \textit{dissipative} part) has to be symmetric and generate non-zero dissipation. The microforce balance equation together with the previously derived representation for the microstress \({\mathbf T}^p\) can be viewed as a \textit{flow rule}, with the energetic part of the miscrostress \({\mathbf T}^p_{en}\) defining the \textit{back stress}, while \({\mathbf D}^p\) enters the arguments of the constitutive function defining the dissipative part \({\mathbf T}^p_{dis}\) from the microstress. The discussion of the \textit{referential symmetry} is based on the \textit{change of the reference configuration} by \({\mathbf H} \in\) U\(_{nim},\) assuming that \({\mathbf F}^e\) \textit{remains invariant}. Hence \({\mathbf F}\) and \({\mathbf F}^p\) are transformed into \({\mathbf F}{\mathbf H}\) and \({\mathbf F}^p{\mathbf H},\) respectively. The definition of the referential symmetry group follows similarly with symmetry group for finite elasticity, elaborated by Noll. The \textit{relaxational symmetry} group is based on the \textit{transformation} of \({\mathbf F}^e\) into \({\mathbf F}^e {\mathbf H},\) together with the hypothesis \({\mathbf F}\) is \textit{invariant}. Hence \({\mathbf F}^p\) is changed into \({\mathbf H}^{-1}{\mathbf F}^p.\) Under the hypothesis that \({\mathbf H}\) is unimodular (and constant) \({\mathbf L}^p\) is changed into \({\mathbf H}^{-1}{\mathbf L}^p {\mathbf H},\) denoted here by us \(\overline{\mathbf L}^p.\) The derived symmetry group is characterized by the relationships written in (10.10). Comments: Let us remark that \({\mathbf H}^{-1}{\mathbf D}^p {\mathbf H}\) in the formulae (10.10) stands for the symmetric part of \(\overline{\mathbf L}^p.\) But \({\mathbf H}^{-1}{\mathbf D}^p {\mathbf H}\) for \({\mathbf H}\) unimodular is symmetric if and only if \({\mathbf H}{\mathbf H}^T\) is identity, i.e. \({\mathbf H}\) is orthogonal. Consequently the resulting relaxational symmetry group \textbf{has to be a subgroup of orthogonal transformations}, and \({\mathcal G}^{rel}\) \textbf{never could be equal to} U\(_{nim}.\) Hence to refer to the material at X as \textbf{relaxationally fluid-like} (as the authors do) seems to not make sense (see 10.3.2 (iv)). The \textit{isotropic, viscoplastic solids} are defined in the paper by the reduced constitutive equations, that follows from their definition to be \textit{referentially fluid-like,} i.e. \({\mathcal G}^{ref}\) is U\(_{nim}\), and \textit{relaxationally isotropic solid-like}, i.e. \({\mathcal G}^{rel}\) is the proper orthogonal group, respectively. Comments: Let us remark that the authors in order to find the consequences inferred into the model by the adopted definition, first of all choose \({\mathbf H}= ({\mathbf F}^p)^{-1}.\) The previously result written in (10.6) could not be applied, without calculating for \({\mathbf F}^p{\mathbf H},\) denoted here by us \(\overline{\mathbf F}^{p},\) the symmetric part of the appropriate \(\overline{\mathbf L}^{p}.\) Consequently we found that \(\overline{\mathbf D}^{p} = 0,\) which means that in formula (11.1) zero stands on the position of \({\mathbf D}^{p},\) and so it is for the subsequently represented formulae, (11.4)\(_{3},\) (11.3)\(_{3}\). Hence \textit{isotropic, viscoplastic solid} is \textbf{reduced to an elastic type representation}, in terms of the elastic part of the deformation gradient. We have in mind that the irreversible behaviour would be used to define \({\mathbf F}^p,\) via the evolution equations, as dependent on the history of the deformation gradient, and thus at any time t and at any fixed material point \({\mathbf F}^e= {\mathbf F} ({\mathbf F}^p)^{-1}\) would be derived for the class of elastic-plastic material described above. In section 12, the \textit{amorphous material} is by definition \textit{both referentially and relaxationally} an isotropic solid. In order to derive the reduced form of the constitutive equations, the consequences that follows from \({\mathcal G}^{ref}= Ort^+\) are first investigated by the authors. Comments: Let us remark that when \({\mathbf Q} = ({\mathbf R}^p)^T,\) the symmetry transformation will be \textbf{time- dependent}, and again we ought to calculate the expression of \(\overline{\mathbf D}^p\) corresponding to \(\overline{\mathbf F}^p= {\mathbf F}^p ({\mathbf R}^p)^T \equiv {\mathbf V}^p.\) Proceeding with the calculus we arrive at the expression \(\overline{\mathbf D}^p =\) sym \(\{ \dot{\mathbf V}^p ({\mathbf V}^p)^{-1}\},\) i.e. the symmetric part of the appropriate tensor. Let us remark that \(\overline{\mathbf W}^p\) is equal to the skew-symmetric part of \(\dot{\mathbf V}^p ({\mathbf V}^p)^{-1},\) generally non-zero. Thus in formulae (12.2) and in the subsequently derived (12.3) and (12.5), sym\(\{ \dot{\mathbf V}^p ({\mathbf V}^p)^{-1}\}\) stands for \({\mathbf D}^{p}.\) Comments: In the paper by Cleja-Tigoiu and Soós (1990) a possible axiomatic reconstruction of elasto-plastic model for metals were proposed, based on the local, current, relaxed configurations (lcrc). Although the elastic and plastic deformations are local deformation, they do not represent pure geometric or kinematic concepts. The mathematical description of the (lcrc) was introduced simultaneously with the constitutive and evolution equations, taking into account the physical origin and mechanical significance (see Teodosiu (1970), Kratochvil (1971), Mandel (1971), Rice(1971)) of the elastic and plastic deformations, which means the specific role played by the crystalline structure. On the basis of the relaxed and objective assumptions, the material symmetry group relative to the local current relaxed configuration is a subgroup of the proper orthogonal group. When the arbitrary current relaxed configurations, associated with the \textit{same local motion} of the body, have to be also \textit{isoclinic}, the material symmetry concept leads to a fixed symmetry group, which characterizes the pre-existing anisotropy in the initial reference configuration for the crystalline materials. Comments: On the other hand, appealing the Symmetry Lemma written in (11.12) the authors state (see (12.11) and (12.13)) that \({\mathbf S}_{0}\) and \({\mathbf T}^p_{en}\) are symmetric tensor fields, starting from the definitions (12.10) and (12.12). When we look at the Symmetry Lemma it is clear that the isotropic function involved in it is dependent on one symmetric field only. In contrast, the formulae (12.10) and (12.12) refer to an isotropic function with respect to two symmetric tensors, \(({\mathbf B}^{p}, {\mathbf C}^{e}),\) generally with different principal directions. The symmetry of the derived fields \({\mathbf S}_{0}\) and \({\mathbf T}^p_{en}\) holds only if, either \({\mathbf B}^{p}\) and \({\mathbf C}^{e}\) have the same principal vectors, or either \({\mathbf B}^{p}\) or \({\mathbf C}^{e}\) are spherical tensors. The non-symmetric fields are generated through the formulae (14.2), (14.3) and (14.9), written for the free isotropic energy function \(\overline{\psi},\) expressed in (14.1) through two tensors \(({\mathbf B}^{p}\) and \({\mathbf C}^{e}).\) Moreover the microforce balance cannot be expressed through the form of a Mises flow rule (8.17), because of the lack of symmetry for the appropriate stress tensors. Thus the formula (14.7) is misleading. The authors refer to the maximum dissipation postulate, formulated as the \textit{material microstability} assumption, in the case of zero microforce, when a prescribed flow rule is given. The direction \({\mathbf N}^p\) (\textit{the so- called flow direction}) for the \textit{inelastic stretch-rate} \({\mathbf D}^p\) is introduced. \textit{Material microstability} assumption stipulates that the \textit{power expend by microstress} in any other flow obtained by a slight change in flow direction (but the modulus of \({\mathbf D}^p\) being kept constant) \textit{must be greater than or equal} the power expend, during the prescribed flow rule. As a consequence of the of material stability (and supposing in addition that \({\mathbf S}_{0}\) and \({\mathbf T}^p_{en}\) are symmetric, the appropriate dissipation function is independent of the flow direction, and so on), the microforce balance takes the form of a Mises flow rule, i.e. the flow direction is defined by the direction of \({\mathbf S}_{0} - {\mathbf T}^p_{en}.\) Comments: We note that the dissipation postulate, as well as the maximum dissipation postulate, cannot be regarded without any \textit{definite constitutive framework}. The consequences are rather different within the finite elasto-plasticity and the comparison between them is not already evident. We mention several forms in Lubliner (1986, 1990), in Marigo (1989) for general mechanical system with domain of the reversibility, in Lucchesy and Podio-Guidugli (1990), Luchhesi and Šilhavy (1991) within of the material with elastic range, Cleja-Tigoiu (2003), etc. In the section 13. the authors \textit{state} that for isotropic, viscoplastic materials and amorphous materials, without loss in generality, we could assume that plastic spin is zero. The results have been summarized in the Irrotationality Theorem. If the basic equations have solutions then every change \({\mathbf Q}({\mathbf X}, t)\) in relaxational frame for the body also yield a solution. Comments: Let us remark that the proof of the theorem is only sketched. From our point of view, a coherent and mathematically correct description of the body in a given material point, means: given an history of the deformation in a neighborhood of the material point, find the \textit{irreversible} (plastic) part of deformation via the evolution equation (i.e. \({\mathbf F}^e = {\mathbf F} ({\mathbf F}^p )^{-1}\) can be derived too) and the \textit{stress state}, characterized through the constitutive equations. We remark that the list of the constitutive relations (13.2) together with (13.3) are dependent not only on \({\mathbf C}^e = ({\mathbf F}^e)^T {\mathbf F}^e,\) and \({\mathbf B}^p = ({\mathbf F}^p)^T {\mathbf F}^p,\) but on the elastic rotation \({\mathbf R}^e.\) It is possible to construct a model with non zero plastic spin (defined in a certain manner), and after that to choose the relaxational time-dependent frame-change \({\mathbf Q}\) that leads to vanishing plastic spin. On the other hand we can build at the very beginning a model in which the plastic spin is stipulated to be zero, at any time and at any material point, i.e. the irrotationality theorem obviously hold. There is a largely debated point in the literature concerning the models in finite elasto-plasticity. Conclusions: In a classical framework of finite elasto-plasticity an yield criteria or an appropriate elastic range is generally introduced, and a state characterized by \({\mathbf F}^p = {\mathbf I}\) makes sense and the behaviour of the material is reduced to an elastic one. Here no yield criteria has been introduced, and the irreversible behaviour seems to develop at the very beginning, due to the microforce balance equation. On the other hand the inelastic stretch-rate is included among the arguments of the constitutive functions. The evolution equations do not pay a special role in finding the constitutive restrictions. The authors proposed a powerful methodology to define constitutive restrictions within a non-usual constitutive framework for finite elasto-plasticity.