An enhanced data-driven constitutive model for predicting strain-rate and temperature dependent mechanical response of elastoplastic materials (Q6105206)

The authors state that the novelty of their data-driven approach to elasto-plastic materials consists in the strain reconfiguration strategy, which is realized via Formula (3), to ``\textit{enhance the strain rate and temperature effect}.'' Formula (3) is not mathematically comprehensible. Concerning the tension-compression asymmetry (see Sections 2.3 and 4), the stress triaxiality, \(\eta,\) which is introduced in terms of mean stress and second invariant of the stress-deviator, is misleading for 1D stress state. We notice that the presence of \(\eta\) inside the constitutive relation, see for instance [\textit{B. Erice} and \textit{F. Gálvez}, Int. J. Solids Struct. 51, No. 1, 93--110 (2014; \url{doi:10.1016/j.ijsolstr.2013.09.015})], emphasizes the dependence of material behaviour on the mean stress. The basic formulation of data-driven constitutive model is declared to be Equation (1), which expresses the equivalent stress in terms of equivalent strain, i.e., an 1D-elastic type constitutive relation. Their modified forms, (5) and (14), compute the equivalent stress, but being defined in terms of sign of the triaxiality (i.e., the stress defined in terms of stress!). Formula (14) is derived from (5) by taking into account the composition with the re-configurated strain Expression (3), which involves also the so-called cumulative effective strain. The curves (1D-type) of true stress in terms of true strain (about 40 graphs), are plotted using the experimental database when different (but constant) tension-rate, compression-rate and temperature, respectively are imposed. In Figure 6 there are curves plotted for different triaxiality. To compare and validate their results the Johnson-Cook (J-C) model is considered but only one constitutive equation of (J-C) model, i.e., Equation (19), is written. Thus the formulae allowing the computation of the so-called effective cumulative plastic strain, or of the plastic strain, involved in Formulae (14) and (19), ought to be written. Formulae (12) are not correct. There are only verbal statements about the feasibility of ``\textit{simulating history-dependent behaviors of elasto-plastic materials}.'' Concerning the numerical implementation of data-driven model (summarized the Box 1), the elastic trial solution, the trial stress factor, stress triaxiality, and \(C_{\mathrm{trial}}\) together with update the stress factor \(C_{\mathrm{uptrial}},\) involving a variable \(C_{\eta},\) are computed, without any reference to plastic behavior. \(C_{\eta},\) which denotes the ``hydrostatic pressure effect on the plastic behavior,'' is introduced with the mention that this coefficient can be iteratively determined based on the experimental curves, sending to the paper [\textit{Y. Bai} and \textit{T. Wierzbicki}, Int. J. Plast. 24, No. 6, 1071--1096 (2008; Zbl 1421.74016)], in which the plasticity and fracture with pressure and Lode dependence are considered (i.e., another type of constitutive framework). The numerical incremental-type algorithm ought to be developed within a well-defined constitutive framework (elasto-plasticity in this case) altogether with conservation laws.