Sei sulla pagina 1di 42

Accepted Manuscript

A simplified approach to high strain rate effects in cold deformation of polycrystalline


FCC metals: Constitutive formulation and model calibration

Tiago dos Santos, Pedro A.R. Rosa, Samir Maghous, Rodrigo Rossi

PII: S0749-6419(16)30013-4
DOI: 10.1016/j.ijplas.2016.02.003
Reference: INTPLA 2024

To appear in: International Journal of Plasticity

Received Date: 21 September 2015


Revised Date: 24 December 2015
Accepted Date: 6 February 2016

Please cite this article as: Santos, T.d., Rosa, P.A.R., Maghous, S., Rossi, R., A simplified approach to
high strain rate effects in cold deformation of polycrystalline FCC metals: Constitutive formulation and
model calibration, International Journal of Plasticity (2016), doi: 10.1016/j.ijplas.2016.02.003.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT

A simplified approach to high strain rate effects in cold


deformation of polycrystalline FCC metals: Constitutive
formulation and model calibration

PT
Tiago dos Santosa,c , Pedro A. R. Rosad , Samir Maghousb , Rodrigo Rossia,∗

RI
a
Departamento de Engenharia Mecânica, Universidade Federal do Rio Grande do Sul, Rua Sarmento
Leite, 425, Porto Alegre, RS, 90046-902, Brazil.
b
Departamento de Engenharia Civil, Universidade Federal do Rio Grande do Sul, Porto Alegre, RS,
Brazil.

SC
c
CAPES Foundation, Ministry of Education of Brazil, Brasília, DF, 70040-020, Brazil.
d
Instituto de Engenharia Mecânica, Instituto Superior Técnico, Universidade de Lisboa, Lisboa,
Portugal.

U
AN
Abstract
This work describes a macroscopic finite elastic-viscoplastic constitutive formulation
M

that accounts for the individual contributions of strain hardening, strain rate induced
hardening and viscous behavior in cold deformation of polycrystalline FCC metals. The
constitutive modeling follows a Perzyna-type formulation within the thermodynamics
D

with internal variables framework. The approach introduces a single phenomenological


TE

internal variable defining an effective microstructural feature, which comprises strain


hardening contributions related to large strain processes. Internal variable evolution
is based on physical considerations and gives rise to a modified Voce hardening law,
EP

which allows for predicting both the strain rate history effects and the linear hardening
at large strains. Instantaneous rate-sensitivity is accounted for by a phenomenological
overstress function. Material parameters are adjusted using experimental data available
C

in literature for annealed high purity copper. Comparisons with experimental results
AC

and other viscoplastic models have demonstrated the model aptitude to properly predict
high strain rate effects in cold deformation of polycrystalline FCC metals.
Keywords: A. Yield condition, High strain rate effects, B. Constitutive behavior, B.
Elastic-viscoplastic material, B. Finite strain.

Corresponding author. Phone: +55 51 3308 3790


Email addresses: tsantos.mec@gmail.com (Tiago dos Santos),


pedro.rosa@tecnico.ulisboa.pt (Pedro A. R. Rosa), samir.maghous@ufrgs.br (Samir Maghous),
rrossi@ufrgs.br (Rodrigo Rossi)

Preprint submitted to International Journal of Plasticity February 11, 2016


ACCEPTED MANUSCRIPT

1. Introduction

The strain rate is known to have a major influence on the mechanical behavior of
some metallic alloys, especially above 103 s−1 when an upturn in the rate-sensitivity is

PT
observed (Lindholm, 1964; Campbell, 1972). The stress response presents a complex
nonlinear behavior, resulting from the combination of strain, strain rate, and temper-

RI
ature histories, what still is the subject of current investigations. Furthermore, even
mechanical testing presents difficulties in reaching the range of strains, strain rates,
and their combined evolution in real-time. These limitations can cause serious draw-

SC
backs in understanding plastic flow mechanics and to characterize material hardening
occurring at real manufacturing processes, such as machining (Neugebauer et al., 2011;

U
Silva et al., 2014), forming (Geier et al., 2014) or compaction (Mamalis et al., 2004) of
metals.
AN
Although the dislocation cell structure formation still is a matter under investiga-
tion (Hansen et al., 2008; Zhang and Shim, 2010; Huang and Tao, 2011; Leff et al.,
2015), some authors claim that the upturn in the rate-sensitivity, for FCC metals as
M

copper and aluminum, is due to strong rate dependency of the material microstructure
at strain rates exceeding 103 s−1 (Follansbee and Kocks, 1988; Molinari and Ravichan-
D

dran, 2005; Gao and Zhang, 2012), in a manner that an increase in strain rate can
increase the density of stored dislocations and decrease the average cell size (Nes, 1997;
TE

Rusty Gray III, 2012). It is known that, during severe plastic deformations (SPD),
dislocation cell structures may originate new grains by means of dynamic recrystalliza-
tion (Humphreys, 2012). Thereby, by increasing the strain rate smaller grains can be
EP

obtained in a SPD grain refinement process, as evidenced by Zhang and Shim (2010) for
OFHC copper and Huang and Tao (2011) for pure aluminum. These microstructural
C

rate dependencies are responsible for the well-known strain rate history effect observed
in FCC metals, which comes from the fact that the current material response depends
AC

not only on its strain history, but also on the strain rate history of the material (Chiem
and Duffy, 1983; Klepaczko and Chiem, 1986). From this brief discussion it is possible
to note that the strain rate has an important influence on metal microstructural evolu-
tion, i.e., strain rate plays an important role on dislocation storage, cell formation, and
grain refinement induced by deformation processes and, in turn, these features have
direct relations with material properties as strength and ductility. Thus, a constitutive
formulation suitable to predict material behavior during high strain rate, large strain
processes should account not only for deformation history effects, but also for the effects

2
ACCEPTED MANUSCRIPT

of previous strain rates on material properties and response.


A variety of Internal State Variable (ISV) constitutive models accounting for strain
rate history effects have been proposed in the literature. In this context, relevant ref-

PT
erences include the works of Bodner and Partom (1975), Anand (1985), Klepaczko
and Chiem (1986), Follansbee and Kocks (1988), Tong et al. (1992), Bodner and Ru-
bin (1994), Molinari and Ravichandran (2005), and recently those of Rusinek et al.

RI
(2010) and Gao and Zhang (2012). Most of physically-based models, e.g. those based
on the Mechanical Threshold Stress (MTS) (Follansbee and Kocks, 1988; Rusinek and

SC
Klepaczko, 2001; Rusinek et al., 2010; Gao and Zhang, 2012), have proven suitable to
predict the behavior of metals at high strain rates and large deformation. This class of
physically-based models allows for detailed description of both the material behavior

U
and its current state, thus providing an appropriate framework for capturing loading
AN
history effects. However, these models require complex algorithms and significant com-
putational resources to find the material properties during calibration procedures, see
Gao and Zhang (2012).
M

Bodner and Partom (1975) proposed an elastic-viscoplastic model (BP model) to


predict the behaviour of metals at different strain rate and temperature histories within
the context of continuum mechanics. Estrin and Mecking (1986) extended the BP
D

model to predict strain rate induced microstructural evolution by letting the internal
TE

variable saturation and hardening rate to be rate dependent. Some years after, Bodner
and Rubin (1994) proposed an isothermal modified BP model to account for extensive
hardening and strain rate induced hardening in copper at room temperature.
EP

Tanner and McDowell (1999) investigated loading history effects on experimental


responses of annealed high purity copper at various strain rates and temperatures. In
a subsequent work, Tanner et al. (1999) calibrated and compared the JC (Johnson and
C

Cook, 1983), the MTS (Follansbee and Kocks, 1988), and the BCJ (Bamman et al.,
AC

1996) models. These authors concluded that the performance of the BCJ model was
better than the MTS one. However, specifically on room temperature decremental
strain rate test, comparison results have shown some scattering between predictions
and experiments, for both MTS and BCJ models. Aiming at providing a simplified
viscoplastic model, Molinari and Ravichandran (2005) proposed a high-strain-rate con-
stitutive formulation based on a single internal variable, which represents an effective
microstructural feature. The choice of the internal variable and its evolution were
grounded on the cell size parameter and in the dislocation generation and annihilation
processes, following ideas developed in physically-based models as those of Klepaczko

3
ACCEPTED MANUSCRIPT

and Chiem (1986), Estrin and Mecking (1984), and Nes (1997). Thereby, the model
proposed by Molinari and Ravichandran (2005) is capable of accounting for rate effects
on the microstructural evolution, providing reasonable results for the reproduction of

PT
experimental data for copper tests under different strain rates, temperatures, and strain
rate jumps. However, as the internal variable evolution saturates at a given strain level,
the model is not able in its original form to model strain hardening at advanced strains,

RI
i.e. during deformation Stage IV.
In Table 1 some single internal variable models are highlighted, namely the models of

SC
Bodner and Rubin (1994) (BPR), Molinari and Ravichandran (2005) (MR), Follansbee
and Kocks (1988) (MTS). This table also presents, in anticipation, some characteristics
related to the model that will be presented in this work. The first two models consist

U
in simplified formulations widely employed in viscoplastic modeling of loading history
AN
effects. The MTS model is introduced to compare the simplified ones with a physically-
based model. The comparison in Table 1 consists in assessing some constitutive features
related to these models, namely the ability of representing deformation Stage IV, strain
M

rate (SRH), and temperature (Temp.) history effects, as well as the simplicity of the
internal variable evolution equation, i.e., if it allows an analytic solution (AS). Fur-
thermore, the number of material parameters, which are classified as physical (Phys.),
D

adjustable (Adjust.), and non-adjustable or reference (Ref.) paramenters, are also as-
TE

sessed. Unidimensional constitutive formulations associated with each of the selected


models are shown in Tab. A.7 in AppendixA.
EP

Table 1: Constitutive characteristics of selected models.

Number of parameters
Model SRH Stage IV Temp. AS
Total Phys. Ref. Adjust.
C

Present   ×  10 0 2 8
MR  ×   14 0 4 10
AC

BPR  × × × 9 0 2 7
MTS  /×  × 14 3 2 9

All models of Tab. 1 are theoretically able to represent strain rate history effects.
However, as it will be demonstrated in this work, only the present model is capable of
accounting for constant strain hardening at larger strains (Stage IV). The MTS model,
in a first stage can prevents the stress saturation at given strain level. But, as discussed
by Kok et al. (2002), this model saturates with strain. Only the MR and MTS models
are able of modelling thermal effects. Furthermore, among the models analyzed, the

4
ACCEPTED MANUSCRIPT

present constitutive formulation and that of Molinari and Ravichandran (2005) allow
an analytical expression for the ISV evolution. Concerning the number of parameters,
the BPR model exhibits the lower number of constants to be adjusted. However, as

PT
discussed earlier, this model does not account for hardening in Stage IV, and at least
one parameter is needed to perform this task. Among the simplified models, the present
and the MR one, under isothermal condition, have 8 parameters to be calibrated, since

RI
the latter has 2 parameters associated with temperature dependence. It is expected that
the model proposed in this work fills some gaps regarding the viscoplastic constitutive

SC
modeling, while accounting in a simple way for strain hardening, strain rate hardening,
and instantaneous rate sensitivity.
In summary, the goal of present work is to provide a continuum formulation of a

U
simple, but well-founded, finite elastic-viscoplastic constitutive model capable of ac-
AN
counting for strain rate effects on cold deformation of polycrystalline FCC metals. The
proposed constitutive model follows a phenomenological internal variable approach in
which different material constitutive features can be identified separately: quasi-static
M

response, hardening dependence on strain rate history, and instantaneous viscous ef-
fects. This approach allows the model calibration to be performed in subsequent steps,
using more efficient local optimization algorithms as an alternative to global strategies.
D

The proposed model is formulated within the thermodynamics with internal vari-
TE

ables framework following the widely employed elastic-viscoplastic framework of Perzyna


(1971). Aiming at simulating monotonic loading conditions we adopt, for simplicity, a
von Mises yield criterion with a nonlinear isotropic hardening. The representation of
EP

irreversible material evolution follows the idea of Molinari and Ravichandran (2005),
and thus it is accounted for by a single phenomenological internal variable α defining an
effective microstructural feature. However, differently from the groundwork (Molinari
C

and Ravichandran, 2005), in order to account for linear hardening at larger strains, here
AC

the internal variable comprises two strain hardening contributions related to large strain
processes. The first, α1 , is associated with dislocations storage and its arrangement in
dislocation cells, which govern hardening mechanisms in deformation Stages II and III.
Evolution of α1 is based on physical aspects, moreover its evolution is a specialization
of the Bailey-Orowan equation (Bailey, 1926; Orowan, 1945) and can be related to the
Mecking-Kocks model (Kocks and Mecking, 2003). Similar procedures were employed
by Tjøtta and Mo (1993) for high strain rate cold deformations of aluminum, and by
Anand (1982) for rate-dependent hot deformation of metals. The second hardening
contribution, α2 , arises from geometric hardening related to misorientations between

5
ACCEPTED MANUSCRIPT

dislocation cells, and mainly between grains, and is associated with deformation Stage
IV. In a phenomenological way, we assume α2 as being proportional to the accumulated
viscoplastic strain. From these physical considerations an evolution equation for the

PT
isotropic hardening A is proposed. Then, assuming a constant strain rate loading, from
direct integration a modified Voce hardening law is obtained. In this hardening rule, the
saturation hardening is assumed to be a rate-dependent parameter, what also impacts

RI
on the hardening rate at large strains. This rate dependence allows the prediction of
strain rate history effects. Furthermore, instantaneous rate-sensitivity is accounted for

SC
by a phenomenological overstress function in the sense of Perzyna (Perzyna, 1966, 1971).
Although the present constitutive formulation is intended to high strain rate model-
ing, in which adiabatic conditions are generally adopted, in this simplified approach we

U
will assume isothermal conditions1 , what renders this approach suitable to high strain
AN
rate process, but restricted to moderate strain levels (see discussion in section 2.6).
The proposed constitutive formulation is embedded into a finite strain framework, in
which we adopt a total Lagrangian description and employ the classical multiplicative
M

decomposition of the deformation gradient into its viscoplastic and elastic parts. An
isotropic material is considered, whose constitutive formulation is given in terms of the
logarithmic deformation measure and the rotated Kirchhoff stress.
D

The work is organized as follows. In section 2, some continuum mechanics aspects


TE

are described, e.g., the deformation gradient multiplicative decomposition, the specific
Helmholtz free energy and non-negative inelastic dissipation. In subsection 2.3, the
inelastic evolution equations are described in a general form, following the normality
EP

structure of Rice (1971) and the constitutive equation of Perzyna (1966, 1971). In sub-
section 2.4, the effective microstructural variable α is introduced and its evolution rule
is formulated. In section 3, rigid-viscoplastic analytical solutions are given, considering
C

a finite strain simple uniaxial tensile/compression problem. Analytical expressions are


AC

used in section 4, where the model parameters are adjusted. Calibration procedure has
as reference the experimental data of Tanner and McDowell (1999), Nemat-Nasser and
Li (1998), and Jordan et al. (2013), which were obtained from large strain compres-
sion testing performed on annealed high purity copper within a strain rate range of
4 × 10−4 − 9 × 103 s−1 . In section 5 the accuracy of proposed model is assessed through
comparisons with decremental strain rate experimental data as well with the models

1
Isothermal flow stress curves at high strain rates can be obtained by performing incremental loading
(Nemat-Nasser and Isaacs, 1997).

6
ACCEPTED MANUSCRIPT

of Follansbee and Kocks (1988), Bamman et al. (1996), Bodner and Rubin (1994), and
Molinari and Ravichandran (2005). Our conclusions and comments are given in section
6.

PT
2. Mechanics and thermodynamics of elastic-inelastic continuum

RI
This section provides the main features of the macroscopic formulation adopted for
modeling an elastic-viscoplastic material at finite strains. The starting point is the
classical multiplicative2 decomposition of the deformation gradient (Lee, 1969; Mandel,

SC
1972)
F = F e F vp , with det (F e ) > 0 and det (F vp ) = 1, (1)

U
where F e and F vp are the elastic and viscoplastic parts of F . Decomposition given in
Eq. (1) supposes the existence of a stress-free local state Oξ , which defines an intermedi-
AN
ate configuration. The stress-free local state is mapped from reference configuration Ω0
by F vp and from current configuration Ω by F e . The displacement function ϕ maps
−1
M

an initial point X ∈ Ω0 onto a current one x ∈ Ω at time t, such that x = ϕ (X, t)


and F = ∂ϕ(X,t)
∂X
.
D

2.1. State equations


Irreversible behavior is defined from a single phenomenological internal variable α.
TE

Since the present model is phenomenological, the hidden variable has no quantitative
relationship with physical parameters, but it represents an effective microstructural
EP

variable which is seen as a dominant material strain hardening feature (see also discus-
sion of Molinari and Ravichandran (2005)).
Adopting multiplicative decomposition of Eq. (1) and assuming isothermal condi-
C

tions, specific Helmholtz free-energy ψ can be split into (Lubliner, 1984)


AC

ψ = ψ e (E e ) + ψ vp (α) , (2)

2
where E e = ln (U e ) is the Hencky elastic strain with U e = (F e )T F e and F e =
F F vp . The terms ψ e and ψ vp are the elastic and viscoplastic parts of ψ. It is
−1

implicitly assumed from Eq. (2) of free energy that instantaneous material response
is uniquely determined by elastic strain E e and internal variable α (see for instance

2
Along this work single contractions between second-order tensors are omitted, i.e., S · T = ST , in
components (ST )ij = Sik Tkj .

7
ACCEPTED MANUSCRIPT

references (Coleman and Gurtin, 1967; Rice, 1971; Lubliner, 1984)). The following
forms are assumed for both ψ e and ψ vp

1 1

PT
ρ0 ψ e = E e : De : E e and ρ0 ψ vp = Hα2, (3)
2 2

where ρ0 is the reference mass density, De is a symmetric positive-definite constant

RI
fourth-order tensor and H ≥ 0 is a constant material parameter. We assume an elasti-
cally isotropic material, for which

SC
2
 
e
D = 2µI + κ − µ I ⊗ I, (4)
3

being I, I, µ, and κ the fourth-order and the second-order identity tensors, the shear

U
1
and bulk modulus, respectively. Components of I are Iijkl = 2
(δik δjl + δil δjk ) in which
AN
δij denotes the Kronecker’s symbol.

2.2. Complementary relationships


M

The material model considered here must satisfy thermodynamic restrictions im-
posed by first and second laws, whose combination provides the dissipation inequality
(Eterovic and Bathe, 1990)
D

vp
Φ = τ̄ : D̄ − Aα̇ ≥ 0, (5)
TE

defined on intermediate configuration, and associated constitutive relations


EP

∂ψ e e e ∂ψ vp
τ̄ = ρ0 = D : E and A = ρ0 = Hα, (6)
∂E e ∂α

in which τ̄ is the rotated Kirchhoff stress tensor. The latter is related to the Kirchhoff
C

stress tensor τ by means of the rotation tensor R = F U −1 with U 2 = F T F , such that


AC

τ̄ = RT τ R. It is recalled that τ and the Cauchy stress tensor σ are related through
τ = Jσ with J = det (F ). Parameter A ≥ 0 stands for the isotropic hardening
vp
 vp 
associated with α and D̄ = sym Ḟ F vp
−1
is the viscoplastic strain rate.
The current elastic domain is defined by a closed convex set

K = {(τ̄ , A) |f (τ̄ , A) ≤ 0} , (7)

where the yield function


f = f (τ̄ , A) (8)

8
ACCEPTED MANUSCRIPT

depends on both τ̄ and A. Condition f = 0 defines the yield surface Y , which is the
boundary of set K ,
Y = {(τ̄ , A) |f (τ̄ , A) = 0} . (9)

PT
In the context of rate-independent plasticity the set K stands for the space of
admissible stresses. Accordingly, all stress states (τ̄ , A) must belong to K . In contrast,

RI
condition (τ̄ , A) ∈
/ K is physically possible in the context of viscoplasticity.
For sake of simplicity, in subsequent analysis we adopt a von Mises yield criterion
together with an isotropic hardening,

SC
s
2
f (τ̄ , A) = τ̄ D − (σy + A) , (10)

3

U
q
where τ̄ D = τ̄ijD τ̄ijD , τ̄ D = τ̄ − 13 tr (τ̄ ) I is the deviatoric part of τ̄ and σy is the
AN
initial yield stress, which can be the yield stress of an annealed material. It is observed
that in FCC metals the initial yield is not significantly rate-dependent, since in these
materials Peierls-Nabarro stress has small contribution on the strength, and thus the
M

slip resistance is mainly due to forest dislocations (Dieter, 1986), then we consider σy
as being constant.
D

2.3. Inelastic evolution equations


TE

In addition to the yield criterion, the internal variables evolution must be formu-
lated. Inelastic evolution can be given by general constitutive equations,
EP

vp ∂φ ∂φ
D̄ = λ̇ and α̇ = −λ̇ , (11)
∂ τ̄ ∂A

in which φ = φ (τ̄ , A) is a flow potential (Rice, 1971). In order to satisfy the dissipation
C

inequality (5), potential φ is required to be a non-negative convex function of both τ̄


AC

and A, and zero-valued at origin, i.e., φ (0, 0) = 0. If φ is a non-smooth function, a more


general formulation can be obtained inside the convex analysis framework employing
sub-gradient concepts, see e.g. (de Angelis, 2000). In Perzyna (1966, 1971) formulation
the flow potential is given by the yield function itself, φ (τ̄ , A) = f (τ̄ , A), which corre-
sponds to an associative flow rule. Moreover, differently from a rate-independent model,
instead of satisfying the Karush-Kuhn-Tucker and consistency conditions, viscoplastic
multiplier λ̇ must satisfy the viscoplastic constitutive equation

1
λ̇ = Θ (hf i , A) , (12)
ϑ
9
ACCEPTED MANUSCRIPT

1
where hxi ≡ 2
(x + |x|) denotes the Macaulay brackets, ϑ > 0 is the material viscosity
parameter and Θ ≥ 0 is the overstress function which should be convex. Usually, the
overstress function is given only in terms of positive value of yield function f . However,

PT
in subsequent analysis, we assume that Θ depends explicitly on both f and hardening
parameter A, and as usual, complies with condition Θ (0, A) = 0.
Many constitutive models assume as scalar internal variable α the accumulated

RI
viscoplastic strain ǫ, i.e., α ≡ ǫ with
s

SC
2 vp
ǫ̇ = D̄ ≥ 0, (13)
3

which is related to isotropic hardening A by constant parameters. However, as men-

U
tioned previously, this class of models do not account for strain rate influence on hard-
AN
ening mechanisms, the rate effects are accounted for only by means of the overstress
function Θ. Then, with the aim of incorporating strain rate effects into hardening
mechanisms, instead of Eq. (13), an alternative evolution law for internal variable α
M

shall be adopted.

2.4. Internal variable evolution


D

It is commonly accepted that overall strain rate sensitivity on constitutive response


TE

of metals is composed by two main contributions: (i) the instantaneous rate sensitivity,
which is related to the waiting-time of thermally activated dislocation motion (this
effect can be accounted for by function Θ); and (ii) the rate dependence of dislocation
EP

generation and annihilation acting over the hardening A. These two mechanisms can be
strongly influenced by both the strain and strain rate histories (Klepaczko and Chiem,
1986; Rashid et al., 1992).
C

Many of the viscoplastic models (e.g. (Perzyna, 1966, 1971; Perić, 1993)) can ac-
count for instantaneous rate effects during plastic deformation, but they disregard the
AC

hardening dependence on strain rate. Therefore, accounting for rate-dependent mi-


crostructural evolution becomes an important feature into a high strain rate constitu-
tive formulation. This task can be achieved by introducing a phenomenological internal
variable, α, defining an effective microstructural feature (Molinari and Ravichandran,
2005). Even though this macroscopic parameter does not allow a quantitative rela-
tionship with physical identities, it follows materials science fundamentals and some
microstructural evidences (see for instance (Klepaczko and Chiem, 1986; Estrin and
Mecking, 1984)). In what follows, our proposal will address these rising trends.

10
ACCEPTED MANUSCRIPT

In the present phenomenological isotropic hardening model the current microstruc-


tural configuration is responsible for actual material strength σy + A (α, . . .). Then,
since the model is intended to account for deformation processes at large strains, we

PT
propose that the internal variable α be the result of two contributions, each one related
to strain hardening mechanisms observed at large strains: (i) the storage of disloca-
tions and its arrangement in dislocation cells. These features are governing hardening

RI
mechanisms observed in deformation Stages II and III (Nes, 1997; Kocks and Mecking,
2003); (ii) the geometric hardening related to misorientations between dislocation cells,

SC
and mainly between grains, which gives rise to the well-known Stage IV observed at
later stages of straining. This deformation stage is characterized by a significant reduc-
tion in the hardening rate at the Stage III saturation (Rollett et al., 1989; Nes, 1997;

U
Kocks and Mecking, 2003; Hansen et al., 2008). Thus, we will assume that the internal
AN
variable is given as (we suppose the two hardening contributions are uncoupled)

α = α1 + α2 . (14)
M

Thus, from Eq. (6)2 we can also split the hardening variable A into two terms,
D

A = A1 + A2 , (15)
TE

where
A1 = Hα1 and A2 = Hα2 . (16)
EP

Term α1 is associated to dislocation storage and can be qualitatively related to the


mean dislocation density ρd by assuming


C

α1 ∝ ρd . (17)
AC

From this proportionality condition, considering a polycrystalline metal, the term α1 is


related to dislocation cell size and also to grain size effects. Since deformation-induced
dislocation storage is given from a competition between dislocation generation and
annihilation mechanisms, a constitutive equation widely employed in literature is the
Mecking-Kocks model (Kocks and Mecking, 2003),


ρ̇d = (k1 ρd − k2 ρd ) ǫ̇, (18)

11
ACCEPTED MANUSCRIPT

in which k1 corresponds to athermal dislocation storage and k2 is related to annihilation


(dynamic recovery) mechanisms. Commonly, k1 is assumed to be constant and k2
rate-dependent, since plastic deformation velocity plays an important role on dynamic

PT
recovery processes, in a manner that the strain rate increasing inhibits the dislocation
annihilation and thus increases the dislocation saturation limit, see e.g. Rashid et al.
(1992), Nes (1997), and Kocks and Mecking (2003).

RI
A heuristic reasoning based upon relation (17) and law (18) indicates that

ρ̇d √

SC
α̇1 ∝ √ = (k1 − k2 ρd ) ǫ̇. (19)
ρd

A simple expression for α̇1 meeting relation (19) can be taken as

U
α1
 
α̇1 = h 1 − ǫ̇, (20)
AN
α∞

in which h and α∞ are respectively the rate and saturation parameters related to α1 ,
see (Kocks, 1976). In view of Eq. (16)1 we obtain3
M

A1
 
Ȧ1 = H1 1 − ǫ̇, (21)
A∞
D

where H1 = Hh is the hardening rate and A∞ = Hα∞ is the saturation hardening. In


TE

order to account for monotonic loading applications, one can readily integrate evolution
equation (21) for ǫ̇ constant, obtaining a Voce hardening law (Voce, 1948),
EP

A1 − A∞
= exp [−δ (ǫ − ǫi )] , (22)
A1i − A∞
C

where δ = AH∞1 , A1i , and ǫi are initial values of A1 and ǫ, respectively. Generally, both
parameters H1 and A∞ can depend positively on strain rate (Follansbee and Kocks,
AC

1988). Hence, in present formulation we assume the ratio δ = AH∞1 as being constant,
thus we shall attribute a rate dependence only to parameter A∞ .
Second hardening contribution α2 is linked to geometrical hardening related to de-
formation Stage IV, which is not so well understood as the others deformation Stages

3
Law (21) is a particular case of the Bailey-Orowan equation (Bailey, 1926; Orowan, 1945), without
the static recovery term. A more general model was employed by Brown  et al. (1989), for hot working
  a 
A1 A1
of metals, where they assume the evolution Ȧ1 = H1 1 − A∞ sign 1 − A∞ ǫ̇ − ṙ, in which a ≥ 1

and ṙ is the static recovery evolution.

12
ACCEPTED MANUSCRIPT

(I, II and III). Stage IV can be related to substructure morphology or dimensions within
grains in the later stages of straining (Hughes et al., 1997; Hughes and Hansen, 2000), or
it can be associated with dislocation debris acting as obstacles to dislocation glide, re-

PT
sulting in work hardening even after Stage III saturation (Rollett et al., 1989). Thereby,
based upon monotonic loading applications, we will assume a simple phenomenological
relationship between α2 and accumulated viscoplastic strain ǫ, yielding

RI
α2 = H2 ǫ, (23)

SC
where H2 ≥ 0 is a proportionality parameter. Furthermore, even approximately, it is
possible to write a scaling law between H2 and saturation parameter α∞ (Rollett et al.,

U
1989; Kocks and Mecking, 2003), what yields
AN
α2 = cα∞ ǫ, (24)

where c ≥ 0 is a material constant. Also, from Eqs. (24) and (16)2 we can write A2 in
M

terms of the accumulated viscoplastic strain ǫ,

A2 = cA∞ ǫ. (25)
D
TE

Observing Eq. (25), we notice that the hardening rate at large strains is given by term
cA∞ , which is dependent on strain rate since A∞ is a rate-dependent parameter.
Combination of Eqs. (15), (22), and (25) results
EP

A = Ai + A∞ c (ǫ − ǫi ) + [A∞ (1 + cǫi ) − Ai ] {1 − exp [−δ (ǫ − ǫi )]} , (26)


C

where Ai is the initial value of A. Further, for Ai = ǫi = 0, Eq. (26) reduces to


AC

A = A∞ [1 + cǫ − exp (−δǫ)] , (27)

which is a modified Voce hardening law appended with a linear term, this relation is
slightly different from those proposed by Tome et al. (1984) or Simo and Armero (1992).
The latter is recovered when A∞ is constant. However, here the hardening rule (27)
should account for strain rate history effects by means of the rate dependence of A∞ .

13
ACCEPTED MANUSCRIPT

We attribute this dependence by expressing it as

A∞ = Alwr
∞ fǫ̇ (ǫ̇) , (28)

PT
where Alwr
∞ is the lower bound of A∞ associated with the lower reference strain rate
ǫ̇lwr ≪ 1 and with (room) temperature θR . Higher strain rates inhibits dynamic recovery

RI
mechanisms, causing higher strain hardening. Therefore, the function fǫ̇ (ǫ̇) should
dfǫ̇
increases as ǫ̇ increases, leading to the condition dǫ̇
> 0. Also, the function fǫ̇ (ǫ̇) is

SC
defined in the following interval

fǫ̇ (ǫ̇lwr ) = fǫ̇lwr = 1 and fǫ̇ (ǫ̇up ) = fǫ̇up > 1. (29)

U
The values of the lower, ǫ̇lwr ≪ 1, and upper, ǫ̇up ≫ 1, reference strain rates are set a
AN
priori, while the value of fǫ̇up has to be determined from experiments. Functional forms
of fǫ̇ (ǫ̇) must also be determined from laboratory tests, consisting in finite rate loading,
unloading, and subsequent low strain rate reloading, performed at various strain rates
M

until different strain values. As a preliminary approach, the following form can be
proposed for fǫ̇ (ǫ̇):
D

fǫ̇ (ǫ̇) = [1 − β (ǫ̇)] fǫ̇lwr + β (ǫ̇) fǫ̇up , (30)

where function β is given by


TE


ǫ̇ − ǫ̇lwr
β (ǫ̇) = , (31)
ǫ̇up − ǫ̇lwr
EP

which obviously satisfies


C

β (ǫ̇lwr ) = 0 and β (ǫ̇up ) = 1, (32)


AC

in order to comply with conditions (29). Scalar ξ > 0 is a material parameter. Function
β versus inelastic strain rate ǫ̇ is displayed in Fig. 1 for different values of ξ , with
ǫ̇lwr = 10−4 s−1 and ǫ̇up = 5 × 104 s−1 . Increasing parameter ξ causes a more sudden
increase in function β as ǫ̇ approaches ǫ̇up .
Combining Eqs. (28) and (30) yields

A∞ = [1 − β (ǫ̇)] Alwr up
∞ + β (ǫ̇) A∞ , (33)

14
ACCEPTED MANUSCRIPT

1
1
ξ = 0.2
ξ = 0.2
ξ = 0.5
ξ = 0.5
0.8
0.8 ξ=1
ξ=1
ξ=2
ξ=2
ξ=5

PT
ξ=5 0.6
0.6

β
β

0.4 0.4

RI
0.2 0.2

0 0 −4 −2 0 2 4
0 1 2 3 4 5 10 10 10 10 10

SC
ǫ̇ [1/s] x 10
4 ǫ̇ [1/s]

(a) (b)
Figure 1: Influence of parameter ξ on β vs. ǫ̇ curves. ǫ̇lwr = 10−4 s−1 and ǫ̇up = 5 × 104 s−1 : (a) linear

U
scale; (b) logarithm scale.
AN
where Aup lwr
∞ = fǫ̇up A∞ is the saturation value of A associated with the upper reference
rate ǫ̇up . Equation (33) represents a convex combination of the lower Alwr
∞ and upper
up
A∞ values of A∞ , where the weights are given in terms of ǫ̇. An illustration of the
M

strain rate history effect on saturation hardening A∞ , and thus on the stress−strain
response, is given in Fig. 2. In this figure a high strain rate loading (at a finite inelastic
D

strain rate ǫ̇) followed by an unloading and subsequent quasi-static-reloading is shown.


This curve (solid-line) is compared with a quasi-static loading (dashed-line). From this
TE

comparison we can note the previous strain rate influence on the saturation hardening
A∞ , where A∞ (ǫ̇) is greater than the associated quasi-static value Alwr
∞ , such that the
EP

quasi-static loading curve is not recovered by the quasi-static-reloading. Furthermore,


since we consider ǫ̇ < ǫ̇up , obviously we have A∞ (ǫ̇) < Aup
∞.

Remark 1. Some authors, as Qi et al. (2009) and Yu et al. (2013), advocate that
C

the strain rate-induced strengthening of engineering materials tends to a saturation


AC

limit. Notice, on one hand, that the functional form of Eq. (31) presented in this
work does not envisage the saturation of parameter A∞ as a function of strain rate.
Such an assumption becomes reasonable into the strain rate regime adopted in this
work. Also, some authors as Armstrong et al. (2007) and Meyers et al. (2003) have
shown that the flow stress (controlled by dislocation generation at the shock front) of
ductile FCC metals as pure copper still increases with strain rate for values exceeding
107 s−1 , see also discussion of Huang et al. (2009) and references cited therein. In
fact, there are few works (see Borodin et al. (2014)) pointing towards a saturation in
the strain rate effect on the dynamic yielding strength of metals at ultra-high strain

15
ACCEPTED MANUSCRIPT

τ 11 ε&lwr < ε& < ε&up


σ y + A∞up
high-rate ε&
ε&lwr

PT
loading
σ y + A∞ (ε& )
ε&lwr
σ y + A∞lwr
quasi-static

RI
loading
σy quasi-static
unloading
reloading

SC
E11

U
Figure 2: Schematic representation of the strain rate history effect on saturation hardening A∞ .
AN
rates (> 107 s−1 ). On the other hand, in a material modeling task in which the strain
rate effect saturation becomes to be an important behavior into the desired strain
M

rate regime, this saturation behavior can be accounted for by adopting, instead of
the functional form proposed in Eq. (31), a S-shaped functional form for parameter
D

β (ǫ̇). However, the model presented in this manuscript is not intended to model very
high strain rate processes (> 104 s−1 ), where thermally activated dislocation motion is
TE

no longer the most important mechanism governing the plastic deformation, and the
hardening behavior can be contolled by shock wave propagation (Meyers, 1994).
EP

2.5. Viscoplastic models


In order to complete the formulation, expressions for the constitutive function
Θ (hf i , A) given in Eq. (12) must be introduced. Functional forms of Θ (hf i , A)
C

should incorporate material physical aspects as the instantaneous rate dependence,


AC

while keeping it simple enough to be mathematically tractable. For this purpose, we


employ explicit forms of function Θ (hf i , A) following two widely applied viscoplastic
models, namely the nonlinear model of Perzyna (1966, 1971) and that proposed by
Perić (1993), see also the descriptions, applications, and discussions of Alfano et al.
(2001).

16
ACCEPTED MANUSCRIPT

2.5.1. Perzyna-based nonlinear model


Constitutive function Θ (hf i , A) is given by (Perzyna, 1966, 1971)
!m
hf i

PT
ϑλ̇ = Θ (hf i , A) = , (34)
R

where R (A) is a characteristic size of the yield locus. In the context of von Mises

RI
criterion, Eq. (10), we have

SC
s
2
R (A) = (σy + A) . (35)
3
 
For f ≥ 0, the inverse function Θ−1 λ̇, A is given by

U
AN
   1
m
f = Θ−1 λ̇, A = R ϑλ̇ . (36)

2.5.2. Perić-based nonlinear model


M

Viscoplastic models should recover the rate-independent behavior as ϑ → 0 or as


m → ∞. Unfortunately, the rate-independent feature is not recovered when m → ∞
D

by the Perzyna nonlinear model. To overcome this problem Perić (1993) proposed a
viscoplastic model which retrieves the inviscid behavior in both the limit cases. In this
TE

model, function Θ (hf i , A) is expressed as


!m
hf i + R
ϑλ̇ = Θ (hf i , A) = − 1, (37)
EP

R
 
and the inverse function Θ−1 λ̇, A , for f ≥ 0, reads
C

   1 
m
f =Θ −1
λ̇, A = R 1 + ϑλ̇ −1 . (38)
AC

2.6. Comments
Before further theoretical developments, a series of comments related to proposed
constitutive modeling deserve to be stated. In this respect, some main features of the
model are emphasized below:
• Starting from additive decomposition of hardening into two contributions, respec-
tively associated with dislocation storage and geometry changes at microscale
level, a hardening law described by Eq. (27) has been formulated. Referred to

17
ACCEPTED MANUSCRIPT

as “modified Voce hardening law”, it allows for prediction of nonlinear hardening


until related saturation (A ≤ A∞ ), as well as for linear hardening prevailing at
subsequent stages of straining. Interestingly, the linear component in the hard-

PT
ening law is also rate-dependent since the saturation hardening A∞ explicitly
depends on the strain rate;

RI
• The ability of the model to predict for strain rate history effects stems directly
from dependence of saturation hardening A∞ on strain rate, as clearly expressed
by Eq. (33) through parameter β (ǫ̇). The specific expression (31) seems suit-

SC
able for capturing significant hardening rate sensitivity observed beyond a certain
critical strain rate level, as illustrated in Fig. 1(b);

U
• Apart from elastic constants, the whole proposed constitutive model depends on
AN
eight independent scalar parameters whose physical interpretation is clear, making
it possible to proceed by subsequent steps to identify their values from experi-
mental data. Actually, the model calibration is rather simple and can be achieved
M

considering three separate constitutive aspects (as detailed in Sec. 4): quasi-static
response, strain and strain rate induced hardening, and viscous behavior.
D

Remark 2. In general finite strain applications, temperature dependence of FCC poly-


crystalline metals becomes important at high strain rates. However, the present sim-
TE

plified viscoplastic model does not explicitly account for the heat generation during
high strain rate deformation process. The aim of the present proposal is to provide
a simple constitutive tool focused mainly on the rate effects on constitutive response,
EP

in idealized isothermal conditions. The present isothermal model, under high strain
rate conditions, should be calibrated from incremental deformation tests, in order to
C

minimize heat generation effects in experimental results, as done e.g. by Follansbee and
Kocks (1988) and Nemat-Nasser and Isaacs (1997). In its current form, the proposed
AC

model can therefore be employed to predict high strain rate deformation processes,
but at moderate strain levels, e.g. < 0.5. Simulations performed in both adiabatic
and isothermal settings at high strain rates have shown the difference between isother-
mal and adiabatic conditions for copper is not significant as long as moderate imposed
strains are respected (Follansbee and Kocks, 1988; Molinari and Ravichandran, 2005).
Moreover, an isothermal viscoplastic model can be used in high strain rate conditions
when the material calibration procedure reproduces the loading history associated with
the desired application (Silva et al., 2014). In this situation, thermal effects are im-

18
ACCEPTED MANUSCRIPT

plicitly accounted for through material parameters. Otherwise, in cases where thermal
effects have to be explicitly considered, within the present framework temperature in-
fluence can be accounted for by adopting adiabatic conditions at high strain rates, and

PT
assuming temperature dependences on the initial yield stress σy , instantaneous rate
sensitivity parameter m, and on evolution of hardening variable A, through the satu-
ration value A∞ , which is somehow associated with temperature dependent dynamic

RI
recovery mechanisms.

SC
3. Simulation of uniaxial tensile/compression test

For illustrative purposes and in order to calibrate the model from experimental

U
data, let us considering a homogeneous cylindrical specimen, geometrically defined in
reference configuration by its diameter
q 0
d and length l0 , with symmetry axis coinciding
AN
with X1 , i.e., 0 ≤ X1 ≤ l0 and 0 ≤ X22 + X32 ≤ d20 (Fig. 3). The specimen is placed
between two rigid and smooth platens subjected to respective vertical displacements
u1 = 0 at X1 = 0 and u1 = ū1 at X1 = l0 . The constitutive homogeneous material is
M

vp
considered as rigid-viscoplastic, implying that D̄ = D̄ .

X1
D

u1 (t )
TE

− E11

d0
l0
EP

−K
1
0
R t
C

Figure 3: Schematic representation of uniaxial boundary value problem.


AC

The displacement vector u and deformation gradient F are given by


   




(a1 − 1) X1 


 
a1 0 0 
u= (a2 − 1) X2  and F = 

0 a2 0 .

(39)

   
 (a − 1) X 

0 0 a2
2 3


Accordingly, the right stretch tensor U = F T F and the logarithmic strain measure

19
ACCEPTED MANUSCRIPT

E = ln (U ) are
   

a1 0 0  
ln (a1 ) 0 0 

PT
U = 0 a2 0  and E =  0 ln (a2 ) 0 , (40)
   
   
0 0 a2 0 0 ln (a2 )

RI
where the principal stretches must complies with a1 ≤ 1 and a2 ≥ 1 in the compression
case. Furthermore, a1 is defined from boundary condition u1 = ū1 at X1 = l0 ,

SC
ū1
a1 = 1 + . (41)
l0

The rate of axial strain is

U
ȧ1 ū˙ 1
Ė11 = = . (42)
a1 l0 + ū1
AN
Notice that for the specific case considered herein, the modified strain rate D̄ =
 
sym Ḟ F −1 and the rate of strain Ė are coincident. The analysis will be restricted to
M

a constant rate of strain

Ė11 = D̄11 = K < 0 (compression) . (43)


D

The prescribed displacement ū1 (t) of upper cylinder face is actually controlled through
TE

the value of axial strain E11 applied at a constant rate K. Thus, starting from
ū1 (t = 0) = 0, the prescribed displacement takes the form
EP

ū1 (t) = l0 [exp (Kt) − 1] . (44)

For this simple compression case, the uniaxial stress state reads
C

   
2
τ̄ 0 0  τ̄ 0 0
AC

 11  3 11 
D
τ̄ =  0 0 0  , τ̄ =  0 − 13 τ̄11 0 , (45)
   
   
1
0 0 0 0 0 − 3 τ̄11

and s

D
2
τ̄ = |τ̄11 | , (46)
3
where |τ̄11 | denotes the absolute value of τ̄11 . Considering a rigid-viscoplastic material

20
ACCEPTED MANUSCRIPT

and assuming an associative evolution, it comes from Eqs. (11)1 and (12):
 
s s 1 0 0
3 τ̄ D 2  
0 − 12

PT
D̄ = λ̇ = λ̇  0  sign (τ̄11 ) . (47)
 
2 |τ̄11 | 3  
0 0 − 12

RI
It follows that s
2
D̄11 = λ̇sign (τ̄11 ) , (48)
3

SC
where λ̇ = ϑ1 Θ (hf i , A) and sign (τ̄11 ) denotes the signal of τ̄11 . Moreover, in view of
Eqs. (13) and (47), the accumulated viscoplastic strain rate becomes

U
ˆ t
ǫ̇ = D̄11 = |K| and ǫ =

D̄11 dt = |K| t,

(49)
AN
0

since ǫ̇ is constant and ǫ (t = 0) = 0. We remember that, from Eq. (43), Ė11 = D̄11 and
thus E11 = Kt.
M

The hardening variable can be calculated from Eq. (27),

A = A∞ [1 + cǫ − exp (−δǫ)] , (50)


D
TE

with A∞ given by, see Eqs. (33) and (31),



ǫ̇ − ǫ̇lwr  
A∞ = Alwr
∞ + Aup lwr
∞ − A∞ . (51)
ǫ̇up − ǫ̇lwr
EP

In what follows we examine the cases of Perzyna and Perić based models.
C

3.1. Perzyna-based nonlinear model


AC

In the case of inelastic flow, the current yield function must obey f = Θ−1 , where
 
the constitutive function Θ−1 λ̇, A is given in Eq. (36). Combining Eqs. (10), (46),
(48), and (49) leads to the following expression for axial stress
s 1 
m
3 
|τ̄11 | = (σy + A)  ϑǫ̇ + 1 , (52)
 
2

where hardening parameter A is computed from Eqs. (50) and (51). The above equation
shows how stress τ̄11 depends on strain ǫ and its rate ǫ̇ = |K| (or E11 and Ė11 = K).

21
ACCEPTED MANUSCRIPT

3.2. Perić-based nonlinear model


 
In a similar way, in view of Perić viscoplastic function Θ−1 λ̇, A defined in Eq.
(38), over again combining Eqs. (10), (46), (48), and (49) yields the expression of axial

PT
stress  1
s m
3
|τ̄11 | = (σy + A) 1 +
 ϑǫ̇  . (53)
2

RI
As already mentioned, Eq. (52) shows that in Perzyna nonlinear model |τ̄11 | =
2 (σy + A) as m → ∞, which corresponds to twice the non-viscous response. This

SC
shortcoming of Perzyna nonlinear model is solved by using Perić viscoplastic model,
since the latter leads to |τ̄11 | → (σy + A) as m → ∞. Thus, in the subsequent calibration
we will adopt only the viscoplastic model of Perić.

U
AN
4. Calibration of material parameters

The objective of this section is outline the calibration strategy employed to deter-
M

mine the adjustable parameters of the proposed constitutive model. Due to lack of
own specific experimental data the calibration procedure is performed using data re-
ported in literature, for annealed high purity copper in a wide strain rate range at
D

room temperature. First, the data given in Tanner and McDowell (1999), in a strain
TE

rate range of 4 × 10−4 − 6 × 103 s−1 , are selected to perform the material parameters
adjustment. Furthermore, in order to increase the strain rate range, we supplement
the data of Tanner and McDowell (1999) with high strain rate experimental curves of
EP

Nemat-Nasser and Li (1998) (strain rate of 8 × 103 s−1 ) and Jordan et al. (2013) (strain
rate of 9 × 103 s−1 ).
We remark that the experimental data available in the above works consists in flow
C

stress-strain curves. However, in the absence of experimental data, in order to account


AC

for the current hardening state (σy + A) associated with each strain level and previous
strain rate, suitable experimental needs to be “generated”. These data consist in what
is called static reload stress. For this purpose the Mechanical Threshold Stress (MTS)
model (Follansbee, 1986; Follansbee and Kocks, 1988), with parameters adjusted by
Tanner et al. (1999), is used.
For sake of simplicity, we perform directly the calibration reasoning on the rigid-
viscoplastic analytical equations derived in section 3 for Perić viscoplastic model. Cali-
bration procedure is carried out by means of a nonlinear least-square method. Although,
the present model has 8 parameters to be adjusted, this gradient-based procedure is

22
ACCEPTED MANUSCRIPT

justified since this simple viscoplastic model enables calibration to be performed in sub-
sequent steps, allowing to adjust separately each constitutive contribution, and thus to
verify graphically the adequacy of each calibrated material constant.

PT
4.1. Non-viscous response
The non-viscous material response is given by equation4

RI
|τ̄11 | = σy + A∞ [1 + cǫ − exp (−δǫ)] . (54)

SC
Experimental data of Tanner and McDowell (1999) were obtained from compression
testing performed at room temperature (298 K) for different strain rates, namely: 4 ×
10−4 s−1 , 0.01 s−1 , 0.1 s−1 , 1.0 s−1 , and 6 × 103 s−1 . Following similar procedures Nemat-

U
Nasser and Li (1998) and Jordan et al. (2013) obtained the supplementary high strain
AN
rate curves, at strain rates equal to 8×103 s−1 and 9×103 s−1 , respectively. Experimental
flow stress-true strain curves are depicted in Fig. 4(a).
First, the initial yield stress σy of Eq. (54) is set to be equal to the athermal stress
M

σ̂a given in Tab. 3: σy = 35 MPa. In sequence, to perform the non-viscous equation


calibration, we consider first the quasi-static response to adjust parameters A∞ , δ, and
D

c related to ǫ̇ = 4 × 10−4 s−1 . Calibrated parameters A∞ |4×10−4 , δ|4×10−4 , and c|4×10−4


are given in Tab. 2. The adjustment result is compared with reference experimental
TE

data (Tanner and McDowell, 1999) in Fig. 5. For the higher strain rates, the flow stress
curves given in Fig. 4(a) exhibit viscous effects, thus making it difficult to proceed with
direct calibration of Eq. (54) from experimental data.
EP

Table 2: Saturation hardening A∞ and hardening rates δ and c obtained from calibration.
C

ǫ̇ [s−1 ] 4 × 10−4 0.01 0.1 1.0 6 × 103 8 × 103 9 × 103


A∞ [MPa] 226.5 230.6 236.1 240.4 264.3 333.5 362.8
AC

δ [−] 6.458 − − − − − −
c [−] 0.4207 − − − − − −

To proceed with the calibration procedure, we make use of the static reload stress
σsr = σy + A, which is related to a prestrain strain rate and to a given strain level, and

4
In fact, experiments report true stress-true strain curves. However, since we consider a rigid-
viscoplastic material and a simple compression test, the true stress σ11 and rotated Kirchhoff stress
τ̄11 measures are coincident.

23
ACCEPTED MANUSCRIPT

600

500

PT
400
|τ̄11 | [MPa]

RI
300

200
0.0004 1/s

SC
0.01 1/s
0.1 1/s
100 1.0 1/s
6000 1/s
8000 1/s
9000 1/s

U
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
|E11 |
AN
(a)
600
M

500

400
D

Comparison
|τ̄11 | [MPa]

300
TE

200
sr: 0.01 1/s
sr: 0.1 1/s
sr: 1.0 1/s
EP

100 sr: 6000 1/s


sr: 8000 1/s
sr: 9000 1/s
exp.: 6000 then 0.0004 1/s
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
|E11 |
C

(b)
AC

Figure 4: Experimental
 flow stress-true strain curves of Tanner  and McDowell (1999)
4 × 10−4 −6 × 103 s−1 , Nemat-Nasser and Li (1998) 8 × 103 s−1 , and Jordan et al. (2013)
9 × 103 s−1 ; (b) Calculated and experimental (Tanner and McDowell, 1999) quasi-static reload
curves. Static reload stresses (σsr ) are calculated from Eqs. (56) and (57).

it can be measured/calculated from a quasi-static reloading after high strain rate pre-
straining. For example, Follansbee (1986) measured the static reload stress at various
temperatures to obtain the well-known Mechanical Threshold Stress (MTS) σ̂, which is
the flow stress at temperature of 0 K. In the MTS model (Follansbee, 1986; Follansbee

24
ACCEPTED MANUSCRIPT

600

500

PT
400
|τ̄11 | [MPa]

RI
300

0.0004 1/s
200
0.01 1/s

SC
0.1 1/s
1.0 1/s
100 6000 1/s
8000 1/s
9000 1/s
fitted reload

U
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
|E11 |
AN
Figure 5: Calculated and adjusted quasi-static reload curves.

and Kocks, 1988), internal state variable σ̂ represents the material response related to
M

current microstructural configuration at 0 K. Since we do not have sufficient exper-


imental data regarding static reload stress associated with flow stress curves of Fig.
D

4(a), then, in order obtain this information and to proceed with the calibration of the
non-viscous response, the static reload stresses are calculated making use of the MTS
TE

model.

4.1.1. Determination of σsr using MTS model


EP

By knowing the value of σ̂ the flow stress (σ) related to given temperature θ and
strain rate ǫ̇ can be computed as
C

 1
 1 p

ǫ̇0 q
kθ ln ǫ̇


 
AC

σ = σ̂a + (σ̂ − σ̂a ) 1 −   , (55)




 g0 µb3 

where k is the Boltzmann constant, ǫ̇0 is a reference strain rate, µ is the shear modu-
lus, b is the material Burgers vector, σ̂a is the athermal stress, g0 is the dimensionless
activation energy, q and p are material parameters. Tanner et al. (1999), considering
the MTS model and experimental data of Tanner and McDowell (1999), obtained the
adjusted parameters given in Tab. 3. In the present analysis, starting from a known
value of flow stress σ at 298 K for each prestrain strain rate, we proceed to the deter-

25
ACCEPTED MANUSCRIPT

mination of associated static reload stress σsr for each prestrain level. To this end, we
first derive σ̂ from σ by using Eq. (55),
 1

PT
  1 − p

 
kθ ln ǫ̇ǫ̇0 q


 
σ̂ = σ̂a + (σ − σ̂a ) 1 −   , (56)

 g0 µb3 

RI
and then we compute σsr as

SC
 1
  1  p
kθ ln ǫ̇ǫ̇sr0 q

 
 
σsr = σ̂a + (σ̂ − σ̂a ) 1 −   , (57)


 g0 µb3 

U
where ǫ̇sr = 4 × 10−4 s−1 is the quasi-static reload strain rate. Computed quasi-static
AN
reload points are displayed in Fig. 4(b), where an experimental quasi-static reload
curve associated with a prestrain strain rate equal to 6 × 103 s−1 and a prestrain of
ǫ = 0.34 is shown. From Fig. 4(b), a good agreement is observed between experimental
M

measurements and the calculated point (associated with ǫ̇ = 6 × 103 s−1 and ǫ = 0.34)
using Eqs. (56) and (57). From calculated quasi-static reload curves of Fig. 4(b) and
D

knowing the values of δ|4×10−4 and c|4×10−4 , the saturation hardening A∞ of Eq. (54)
is adjusted for each strain rate. Calibrated parameters are given in Tab. 2 and related
TE

curves are compared with calculated static reload curves in Fig. 5 (previously shown
in Fig. 4(b)). It is first observed the model predictions are in good agreement with the
reference data. It is also observed that increasing the strain rate leads to an increase
EP

in the reload stress level and in the hardening rate at large strains.
C

Table 3: MTS model parameters obtained and properties used by Tanner et al. (1999) for high purity
copper.
AC

g0 p q ǫ̇0 k/b3 σ̂a µ


[−] [−] [−] [s−1 ] [MPa K ] −1
[MPa] [GPa]
2.98 0.318 1.2 107 0.848 35 42

4.1.2. Saturation hardening calibration - dependence of A∞ on ǫ̇


In sequence, Eq. (51) is calibrated from values of Tab. 2 in terms of parameters
Alwr
∞ , Aup
∞ , and ξ for fixed values ǫ̇lwr = 10 s and ǫ̇up = 104 s−1 . Notice that, we
−4 −1

set the reference strain rate ǫ̇lwr to be lower than the experimental quasi-static strain

26
ACCEPTED MANUSCRIPT

rate (ǫ̇ = 4 × 10−4 s−1 ). The adjusted values of Alwr up


∞ , A∞ , and ξ are given in Tab. 4
and comparison of calibrated curves with values (ǫ̇, A∞ ) of Tab. 2 are displayed in Fig.
6, where the adopted function given in Eq. (31) proves to be suitable to model the

PT
strain rate dependence of saturation hardening A∞ . Observing Fig. 6, we note that
the model can represent the increased microstructural strain rate-sensitivity observed in
FCC metals at strain rates between 103 − 104 s−1 (Follansbee and Kocks, 1988; Molinari

RI
and Ravichandran, 2005; Gao and Zhang, 2012). In the present model, this strain rate
dependence is accounted for by means of the rate-sensitivity of parameter A∞ and, as

SC
will be seen in the sequel, this feature has direct influence on the predicted hardening
response.

U
Table 4: Parameters Alwr up
∞ , A∞ , and ξ obtained in calibration, for ǫ̇lwr = 10 s and ǫ̇up = 104 s−1 .
−4 −1
AN
Alwr
∞ [MPa] Aup
∞ [MPa] ξ [−]
233.0 419.6 3.155
M

450
Points of Tab. 2
Fitted curve
400
D
A∞ [MPa]

350
TE

300

250
EP

200 −4 −2 0 2 4
10 10 10 10 10
ǫ̇ [1/s]
C

Figure 6: Adjusted saturation hardening curve compared with values of Tab. 2.


AC

4.2. Viscous response


Regarding the non-viscous calibration, we must adjust parameters ϑ and m related
to viscous effects. For this purpose, the stress flow curves of Fig. 4(a) are used. Adjusted
parameters are
ϑ = 1.194 × 103 s, and m = 105.37. (58)

The predicted flow surface is displayed in Fig. 7(a) as a function of strain ǫ and
strain rate ǫ̇ (or E11 and Ė11 ). The flow stress rate-sensitivity for a given strain of

27
ACCEPTED MANUSCRIPT

 
20% is shown in Fig. 7(b), where calibrated flow stress curve |τ̄11 | v.s. Ė11 is also
compared with experimental data of Follansbee (1986) and Jordan et al. (2013). In
order to demonstrate the strain rate influence on the hardening response, in Fig. 7(b)

PT
 
we also display the yield stress curve σy + A v.s. Ė11 , where becomes clear the strong
microstructural strain rate-sensitivity at high strain rates, i.e., up to strain rates close
to 103 s−1 the quasi-static strength (σy + A) is practically rate-insensitive and for strain

RI
rates exceeding 103 s−1 it becomes strongly rate-dependent. Both Figs. 7(a) and 7(b)
indicate the calibrated model is capable to account for drastic flow stress increasing at

SC
strain rates within 103 − 104 s−1 . Figure 7(b) also indicates the model predictions are
in qualitative agreement with high strain rate experiments of Follansbee (1986) and
Jordan et al. (2013). Nevertheless, it is worth noting that there is a scatter between

U
these two experimental results. This discrepancy is mainly due to differences in initial
AN
grain size and inertia effects associated with the size of work pieces (Jordan et al., 2013).
In Fig. 7(c) the adjusted model is compared with experimental true stress−true strain
curves of Tanner and McDowell (1999), Nemat-Nasser and Li (1998), and Jordan et al.
M

(2013). In a general manner, the proposed constitutive model fits with experiments.

5. Decremental strain rate testing comparisons


D

In this section, the constitutive capabilities of the proposed model are assessed by
TE

means of its “ability” to reproduce the decremental strain rate testing for annealed
high purity copper, in this test instantaneous rate sensitivity and strain rate history
effects can be observed. In the decremental strain rate testing the material sample is
EP

subjected to a monotonic loading with a given initial strain rate D̄111 , which is then
abruptly decreased to a value D̄112 . Instantaneous strain rate sensitivity is characterized
C

by a sharp drop in the flow stress, while strain rate history effect is evidenced when
comparing the decremental response with that obtained from a monotonic loading under
AC

a constant strain rate D̄11 = D̄112 during overall deformation process.


The proposed model is compared with other models, namely the MTS (Follansbee
and Kocks, 1988), BCJ (Bamman et al., 1996), BPR (Bodner and Rubin, 1994), and
MR (Molinari and Ravichandran, 2005). First, the experimental data, MTS, and BCJ
model adjustments presented by Tanner et al. (1999) are considered. The comparison
with the proposed model is presented in subsection 5.1, in which we make use of the
parameters adjusted in section 4. They are summarized in Tab. 5. For the BRP and
MR models, the calibration presented by Bodner and Rubin (1994) and Molinari and

28
ACCEPTED MANUSCRIPT

PT
450
Flow stress: exp. Jd
Experimental Flow stress: exp. F
Flow stress: exp. Ref.

|τ̄11 |, (σy + A) [MPa]


400

RI
800 Flow stress: model
Yield stress: model
600
|τ̄11 | [MPa]

350

400

SC
300
200

0 250
5
10 3
10 1

U
1 0.8
10 −1 0.6
10 200 −5
−3 0.4 10 10
−3
10
−1 1 3 5
¯ ¯ 10 −5
0
0.2 ¯ ¯ 10 10 10
¯Ė11¯ [1/s]
¯ ¯ 10 |E11 | ¯Ė11¯ [1/s]
AN ¯ ¯

(a) (b)
700
M

600

500
D
|τ̄11 | [MPa]

400
TE

300
0.0004 1/s
0.01 1/s
200 0.1 1/s
EP

1.0 1/s
6000 1/s
100 8000 1/s
9000 1/s
fitted curves
0
C

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1


|E11 |
AC

(c)
Figure 7: (a) Model flow surface and reference experimental data; (b) Adjusted model rate-sensitivity
for ǫ = 0.2 compared with reference experimental data (Ref.) and those of Follansbee (1986) (F) and
Jordan et al. (2013) (Jd); (c) Comparison of calculated flow stress and reference experimental data

of Tanner and McDowell (1999) 4 ×10−4 − 6 × 103 s−1 , Nemat-Nasser and Li (1998) 8 × 103 s−1 ,
and Jordan et al. (2013) 9 × 103 s−1 .

29
ACCEPTED MANUSCRIPT

Ravichandran (2005) are used. However, this calibration was based on the experiments
of Follansbee and Kocks (1988). Thus, in order to compare these adjusted model
curves with those of the present proposal, we also have used the experimental data and

PT
MTS model parameters of Follansbee and Kocks (1988). The calibrated parameters,
accordingly with the procedure in section 4, are in Tab. 6, and model comparisons are
described in subsection 5.2.

RI
5.1. Decremental strain rate test of Tanner and McDowell (1999)

SC
The decremental strain rate test performed by Tanner and McDowell (1999) con-
sisted of:

• QS : quasi-static test. Material is subjected to a total strain equal to 92% at a

U
strain rate of D̄11 = 4 × 10−4 s−1 ;
AN
• DSR: decremental strain rate test. Material is subjected to a high strain rate of
D̄111 = 6 × 103 s−1 until a partial strain of 32% is reached, then the strain rate is
abruptly changed to a lower value D̄112 = 4 × 10−4 s−1 while strain reaches 79%.
M

Table 5: Material parameters adjusted using experiments of Tanner and McDowell (1999).
D

E ν σy δ c Alwr Aup ǫ̇lwr ǫ̇up ξ ϑ m


TE

∞ ∞
[GPa] [−] [MPa] [−] [−] [MPa] [MPa] [s−1 ] [s ] [−]
−1
[s] [−]
4
70 0.33 35 6.46 0.42 233 420 10−4 10 3.16 1.2 × 103 105
EP

The results of each model are depicted in Fig. 8(a). Based on this results one
concludes, on one hand, that the MTS model overestimates the hardening rate after
flow stress drop while BCJ model underestimates the experimental dropped flow stress.
C

The latter estimation is closer to the low strain rate response. On the other hand, the
AC

present model produces estimations closer to the experimental data for both high strain
rate loading and flow stress drop responses.
Note that no jump is observed in hardening response, as illustrated by Fig. 8(b).
This behavior could be expected, since parameter A is related to current microstructural
configuration, which does not undergo an instantaneous change by abruptly shifting
strain rate (see Klepaczko and Chiem (1986)). Moreover, QS curves should only be
recovered asymptotically by both stress and hardening responses in DSR simulation.
This is attributed to the fact that the flow stress does not depend only on instantaneous
values of strain rate, but also on strain rate history. In other words, a higher previous

30
ACCEPTED MANUSCRIPT

400 350
350
300
300
250

PT
|τ̄11 | [MPa]

250

A [MPa]
200
200
150
150 Exp.: 0.0004 1/s

RI
Exp.: 6000 then 0.0004 1/s
100 Model: 0.0004 1/s 100
Model: 6000 then 0.0004 1/s
50 MTS−model 50 Model: 0.0004 1/s
BCJ−model

SC
Model: 6000 then 0.0004 1/s
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
|E11 | ǫ
(a) (b)

U
Figure 8: Decremental strain rate test results: (a) Comparison of present model estimation with
experiments (Tanner and McDowell, 1999), MTS, and BCJ model predictions presented by Tanner
AN
and McDowell (1999); (b) Stress hardening vs. accumulated viscoplastic strain.

strain rate induces a larger hardening when compared to a lower strain rate imposed
M

during the whole deformation process, what can be physically related to the rate-
dependence of dislocation storage (Chiem and Duffy, 1983; Klepaczko and Chiem, 1986;
D

Rashid et al., 1992). This rate-sensitivity is captured by the present model through the
rate dependence attributed to saturation hardening A∞ (see Eq. (33)).
TE

5.2. Decremental strain rate test of Follansbee and Kocks (1988)


The aim of this subsection is to compare the present constitutive model with those
EP

of Molinari and Ravichandran (2005), Fig. 9(a), and Bodner and Rubin (1994), Fig.
9(b). The calibrated constitutive results are confronted with experiments reported by
Follansbee and Kocks (1988) and Follansbee and Gray III (1991). Here, the experiments
C

consist of:
AC

• QS test: A total strain equal to 70% is imposed at a strain rate of D̄11 = 1.5 ×
10−3 s−1 ;

• DSR test: High strain rate of D̄111 = 6 × 103 s−1 until a partial strain of 51%.
Lower strain rate of D̄112 = 1.5 × 10−3 s−1 to a strain level of 70%.

Figure 9(a) shows that the present model provides a better estimation than the
MR model. Notice that, the low strain rate response of the latter overestimates the
experimental data, since the behavior predicted by the MR model is higher even than

31
ACCEPTED MANUSCRIPT

Table 6: Material parameters adjusted using experiments of Follansbee and Kocks (1988).

E ν σy δ c Alwr
∞ Aup
∞ ǫ̇lwr ǫ̇up ξ ϑ m
[GPa] [−] [MPa] [−] [−] [MPa] [MPa] [s−1 ] [s ] [−]
−1
[s] [−]

PT
70 0.33 40 7.15 0.54 230 290 10−4 104 0.35 2.7 × 104 215

RI
the dropped flow stress after the strain rate reduction. Further, decremental behavior
estimated by the MR model is slightly distant from the experimental response, when
compared with the present model estimation. The results presented in Fig. 9(b) also

SC
show that the proposed model is able to produce better estimations than the BPR
model. The latter approximates a low strain rate response far from the experiments
of Follansbee and Gray III (1991). Still, the BPR model presents a higher hardening

U
rate than that observed in experimental flow stress after strain rate decrease. As for
AN
subsection 5.1, the results provided by the proposed model present estimations closer
to the experimental data for both high strain rate loading and flow stress dropping
responses.
M

450 450

400 400
D

350 350

300 300
TE
|τ̄11 | [MPa]

|τ̄11 | [MPa]

250 250

200 200
Exp−FG: 0.001 1/s Exp−FG: 0.001 1/s
150 Exp−FK: 2500 then 0.0015 1/s 150 Exp−FK: 2500 then 0.0015 1/s
EP

100 Model: 0.0015 1/s 100 Model: 0.0015 1/s


Model: 2500 then 0.0015 1/s Model: 2500 then 0.0015 1/s
50 MR: 0.0015 1/s 50 BPR: 0.0015 1/s
MR: 2500 then 0.0015 1/s BPR: 2500 then 0.0015 1/s
0 0
C

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
|E11 | |E11 |
(a) (b)
AC

Figure 9: Decremental strain rate test results. Comparison of present model estimation with (a) MR
model; and (b) BPR model. Experiments of Follansbee and Kocks (1988) (FK) and Follansbee and
Gray III (1991) (FG).

6. Conclusions

A finite strain elastic-viscoplastic model with consideration of strain rate effects on


material hardening has been formulated in this paper. The material hardening evolution

32
ACCEPTED MANUSCRIPT

is described through introduction of a single internal variable, whose evolution leads


to a modified Voce hardening law, endowing the model with the capability of properly
predicting the prevailing linear hardening at latter deformation stages and to account

PT
for strain rate history effects. The calibrated model proved successful to reproduce the
observed upturn in rate sensitivity of flow stress at high strain rates (103 − 104 s−1 ).
Analytical comparisons with experimental data indicate the proposed simplified con-

RI
stitutive model is relevant for capturing the main features of high strain rate effects in
deformation of polycrystalline FCC metals at large strains. Restricting to isothermal

SC
processes, the simplified model formulated in this work allows for a semi-analytic formu-
lation that involves a low number of parameters while revealing constitutive features
that are suitable to model high strain rate effects. The promising results that have

U
been obtained encourage future developments including: (i) thermo-mechanical cou-
AN
pling, which is expected to enhance the rate effects on viscoplastic deformations, and
(ii) formulation of micromechanics-based models aiming at accounting for microstruc-
tural driving mechanisms responsible for the overall macroscopic strain rate effects.
M

Acknowledgments
D

The author Tiago dos Santos wishes to acknowledge the doctoral scholarship support
of CAPES, Coordenação de Aperfeiçoamento de Pessoal de Nível Superior of Brazil.
TE

Process number BEX 7023/15-4. The author Rodrigo Rossi wishes to acknowledge the
support of CNPq, Conselho Nacional de Desenvolvimento Científico e Tecnológico of
Brazil. Grant number 302605/2012-6.
EP

References
C

Alfano, G., Angelis, F.D., Rosati, L., 2001. General solution procedures in
AC

elasto/viscoplasticity. Computer Methods in Applied Mechanics and Engineering


190, 5123–5147.

Anand, L., 1982. Constitutive equations for the rate-dependent deformation of metals
at elevated temperatures. Journal of Engineering Materials and Technology, Trans-
actions of the ASME 104, 12–17.

Anand, L., 1985. Constitutive equations for hot-working of metals. International Jour-
nal of Plasticity 1, 213 – 231.

33
ACCEPTED MANUSCRIPT

de Angelis, F., 2000. An internal variable variational formulation of viscoplasticity.


Computer Methods in Applied Mechanics and Engineering 190, 35–54.

Armstrong, R., Arnold, W., Zerilli, F., 2007. Dislocation mechanics of shock-induced

PT
plasticity. Metallurgical and Materials Transactions A 38, 2605–2610.

Bailey, R., 1926. Note on the softening of strain hardened metals and its relation to

RI
creep. J. Inst. Metals 25, 27.

Bamman, D.J., Chiesa, M.L., Johnson, G.C., 1996. Modeling large deformation and

SC
failure in manufacturing process. Theory of Applied Mechanics , 359–376.

Bodner, S.R., Partom, Y., 1975. Constitutive Equations for Elastic-Viscoplastic Strain-

U
Hardening Materials. Journal of Applied Mechanics 42, 385.
AN
Bodner, S.R., Rubin, M.B., 1994. Modeling of hardening at very high strain rates.
Journal of Applied Physics 76, 2742–2747.
M

Borodin, E., Mayer, A., Petrov, Y., Gruzdkov, A., 2014. Maximum yield strength under
quasi-static and high-rate plastic deformation of metals. Physics of the Solid State
56, 2470–2479.
D

Brown, S.B., Kim, K.H., Anand, L., 1989. An internal variable constitutive model for
TE

hot working of metals. International Journal of Plasticity 5, 95 – 130.

Campbell, J., 1972. Dynamic Plasticity of Metals: Course Held at the Department for
EP

Mechanics of Deformable Bodies, July 1970. CISM International Centre for Mechan-
ical Sciences, Springer Vienna.
C

Chiem, C., Duffy, J., 1983. Strain rate history effects and observations of dislocation
substructure in aluminum single crystals following dynamic deformation. Materials
AC

Science and Engineering 57, 233–247.

Coleman, B.D., Gurtin, M.E., 1967. Thermodynamics with Internal State Variables.
Reports on Progress in Physics 47, 597–613.

Dieter, G., 1986. Mechanical metallurgy. McGraw-Hill series in materials science and
engineering, McGraw-Hill.

Estrin, Y., Mecking, H., 1984. A unified phenomenological description of work harden-
ing and creep based on one-parameter models. Acta Metallurgica 32, 57 – 70.

34
ACCEPTED MANUSCRIPT

Estrin, Y., Mecking, H., 1986. An extension of the bodner-partom model of plastic
deformation. International Journal of Plasticity 2, 73 – 85.

Eterovic, A.L., Bathe, K.J., 1990. A hyperelastic-based large strain elasto-plastic con-

PT
stitutive formulation with combined isotropic-kinematic hardening using the loga-
rithmic stress and strain measures. International Journal for Numerical Methods in

RI
Engineering 30, 1099–1114.

Follansbee, P., Gray III, G., 1991. Dynamic deformation of shock prestrained copper.

SC
Materials Science and Engineering: A 138, 23 – 31.

Follansbee, P., Kocks, U., 1988. A constitutive description of the deformation of copper

U
based on the use of the mechanical threshold stress as an internal state variable. Acta
Metallurgica 36, 81 – 93.
AN
Follansbee, P.S., 1986. 24 - high strain rate deformation of {FCC} metals and alloys, in:
Murr, L.E., Staudhammer, K.P., , Meyers, M.A. (Eds.), Metallurgical Applications
M

of Shock-Wave and High-Strain Rate Phenomena. Marcel Dekker Inc., New York,
pp. 451–479.
D

Gao, C., Zhang, L., 2012. Constitutive modelling of plasticity of fcc metals under
extremely high strain rates. International Journal of Plasticity 32-33, 121 – 133.
TE

Geier, M., José, M., Rossi, R., Rosa, P., Martins, P., 2014. Interference-fit joining
of aluminium tubes by electromagnetic forming. Advanced Materials Research 853,
EP

488–493.

Hansen, N., Huang, X., Winther, G., 2008. Grain orientation, deformation microstruc-
C

ture and flow stress. Materials Science and Engineering: A 494, 61 – 67. Advances in
microstructure-based modeling and characterization of deformation microstructures
AC

held at the {TMS} Annual Meeting 2007, Orlando, Florida.

Huang, F., Tao, N., 2011. Effects of strain rate and deformation temperature on mi-
crostructures and hardness in plastically deformed pure aluminum. Journal of Mate-
rials Science & Technology 27, 1–7.

Huang, M., del Castillo, P.E.R.D., Bouaziz, O., van der Zwaag, S., 2009. A constitu-
tive model for high strain rate deformation in {FCC} metals based on irreversible
thermodynamics. Mechanics of Materials 41, 982 – 988.

35
ACCEPTED MANUSCRIPT

Hughes, D., Hansen, N., 2000. Microstructure and strength of nickel at large strains.
Acta Materialia 48, 2985 – 3004.

Hughes, D., Liu, Q., Chrzan, D., Hansen, N., 1997. Scaling of microstructural param-

PT
eters: Misorientations of deformation induced boundaries. Acta Materialia 45, 105 –
112.

RI
Humphreys, F., 2012. Recrystallization and Related Annealing Phenomena. Elsevier
Science.

SC
Johnson, G.R., Cook, W.H., 1983. A constitutive model and data for metals subjected
to large strains, high strain rates and high temperatures, in: International Symposium

U
on Ballistics, The Hague, The Netherlands. pp. 541–547.
AN
Jordan, J., Siviour, C., Sunny, G., Bramlette, C., Spowart, J., 2013. Strain rate-
dependant mechanical properties of ofhc copper. Journal of Materials Science 48,
7134–7141.
M

Klepaczko, J., Chiem, C., 1986. On rate sensitivity of f.c.c. metals, instantaneous rate
sensitivity and rate sensitivity of strain hardening. Journal of the Mechanics and
D

Physics of Solids 34, 29–54.


TE

Kocks, U., 1976. Laws for work-hardening and low-temperature creep. Journal of
Engineering Materials and Technology, Transactions of the ASME 98 Ser H, 76–85.
EP

Kocks, U., Mecking, H., 2003. Physics and phenomenology of strain hardening: the
{FCC} case. Progress in Materials Science 48, 171–273.

Kok, S., Beaudoin, A., Tortorelli, D., 2002. On the development of stage {IV} hardening
C

using a model based on the mechanical threshold. Acta Materialia 50, 1653 – 1667.
AC

Lee, E.H., 1969. Elastic-plastic deformation at finite strains. Journal of Applied Me-
chanics 36, 1–6.

Leff, A., Weinberger, C., Taheri, M., 2015. Estimation of dislocation density from
precession electron diffraction data using the nye tensor. Ultramicroscopy 153, 9 –
21.

Lindholm, U., 1964. Some experiments with the split hopkinson pressure bar. Journal
of the Mechanics and Physics of Solids 12, 317 – 335.

36
ACCEPTED MANUSCRIPT

Lubliner, J., 1984. A maximum-dissipation principle in generalized plasticity. Acta


Mechanica 52, 225–237.

Mamalis, A., Manolakos, D., Kladas, A., Koumoutsos, A., 2004. Electromagnetic form-

PT
ing and powder processing: Trends and developments. Applied Mechanics Reviews
57, 299–324.

RI
Mandel, J., 1972. Plasticité classique et viscoplasticité: course held at the Department
of Mechanics of Solids, September-October, 1971. Courses and lectures - International

SC
Centre for Mechanical Sciences, Springer-Verlag.

Meyers, M., 1994. Dynamic Behavior of Materials. Wiley-Interscience publication,

U
Wiley. AN
Meyers, M., Gregori, F., Kad, B., Schneider, M., Kalantar, D., Remington, B.,
Ravichandran, G., Boehly, T., Wark, J., 2003. Laser-induced shock compression
of monocrystalline copper: characterization and analysis. Acta Materialia 51, 1211 –
M

1228.

Molinari, A., Ravichandran, G., 2005. Constitutive modeling of high-strain-rate de-


D

formation in metals based on the evolution of an effective microstructural length.


Mechanics of Materials 37, 737 – 752.
TE

Nemat-Nasser, S., Isaacs, J., 1997. Direct measurement of isothermal flow stress of
metals at elevated temperatures and high strain rates with application to ta and taw
EP

alloys. Acta Materialia 45, 907 – 919.

Nemat-Nasser, S., Li, Y., 1998. Flow stress of f.c.c. polycrystals with application to
C

OFHC Cu. Acta Materialia 46, 565–577.


AC

Nes, E., 1997. Modelling of work hardening and stress saturation in {FCC} metals.
Progress in Materials Science 41, 129–193.

Neugebauer, R., Bouzakis, K.D., Denkena, B., Klocke, F., Sterzing, A., Tekkaya, A.,
Wertheim, R., 2011. Velocity effects in metal forming and machining processes.
{CIRP} Annals - Manufacturing Technology 60, 627 – 650.

Orowan, E., 1945. The creep of metals. J. of the West Scotland Iron and Steel Institutive
54, 1633.

37
ACCEPTED MANUSCRIPT

Perić, D., 1993. On a class of constitutive equations in viscoplasticity: Formulation and


computational issues. International Journal for Numerical Methods in Engineering
36, 1365–1393.

PT
Perzyna, P., 1966. Fundamental problems in viscoplasticity, Elsevier. volume 9 of
Advances in Applied Mechanics, pp. 243–377.

RI
Perzyna, P., 1971. Thermodynamic theory of viscoplasticity, Elsevier. volume 11 of
Advances in Applied Mechanics, pp. 313–354.

SC
Qi, C., Wang, M., Qian, Q., 2009. Strain-rate effects on the strength and fragmentation
size of rocks. International Journal of Impact Engineering 36, 1355 – 1364. Seventh

U
International Conference on Shock and Impact Loads on Structures.
AN
Rashid, M.M., Gray, G.T., Nemat-Nasser, S., 1992. Heterogeneous deformations in
copper single crystals at high and low strain rates. Philosophical Magazine A 65,
707–735.
M

Rice, J., 1971. Inelastic constitutive relations for solids: An internal-variable theory
and its application to metal plasticity. Journal of Mechanics Physics of Solids 19,
D

433–455.
TE

Rollett, A., Kocks, U., Stout, M., Embury, J., Doherty, R., 1989. Strain hardening at
large strains, in: KETTUNEN, P., LEPISTO, T., LEHTONEN, M. (Eds.), Strength
of Metals and Alloys (ICSMA 8). Pergamon, Oxford, pp. 433 – 438.
EP

Rusinek, A., Klepaczko, J., 2001. Shear testing of a sheet steel at wide range of strain
rates and a constitutive relation with strain-rate and temperature dependence of the
C

flow stress. International Journal of Plasticity 17, 87 – 115.


AC

Rusinek, A., Rodríguez-Martínez, J., Arias, A., 2010. A thermo-viscoplastic constitutive


model for {FCC} metals with application to {OFHC} copper. International Journal
of Mechanical Sciences 52, 120 – 135. {SPECIAL} ISSUE: Advances in Modeling
and Evaluation of Materials in Honor of Professor Tomita.

Rusty Gray III, G., 2012. High-strain-rate deformation: Mechanical behavior and
deformation substructures induced. Annual Review of Materials Research 42, 285–
303.

38
ACCEPTED MANUSCRIPT

Silva, C., Rosa, P., Martins, P., 2014. Innovative testing machines and methodologies
for the mechanical characterization of materials. Experimental Techniques , n/a–n/a.

Simo, J.C., Armero, F., 1992. Geometrically non-linear enhanced strain mixed methods

PT
and the method of incompatible modes. International Journal for Numerical Methods
in Engineering 33, 1413–1449.

RI
Tanner, A.B., McDowell, D.L., 1999. Deformation, temperature and strain rate se-
quence experiments on {OFHC} {Cu}. International Journal of Plasticity 15, 375–

SC
399.

Tanner, A.B., McGinty, R.D., McDowell, D.L., 1999. Modeling temperature and strain
rate history effects in {OFHC} {Cu}. International Journal of Plasticity 15, 575–603.

U
Tjøtta, S., Mo, A., 1993. A constitutive model for cold deformation of aluminium at
AN
large strains and high strain rates. International Journal of Plasticity 9, 461 – 478.

Tome, C., Canova, G., Kocks, U., Christodoulou, N., Jonas, J., 1984. The relation
M

between macroscopic and microscopic strain hardening in f.c.c. polycrystals. Acta


Metallurgica 32, 1637 – 1653.
D

Tong, W., Clifton, R.J., Huang, S., 1992. Pressure-shear impact investigation of strain
rate history effects in oxygen-free high-conductivity copper. Journal of the Mechanics
TE

and Physics of Solids 40, 1251 – 1294.

Voce, E., 1948. The relationship between stress and strain for homogeneous deforma-
EP

tion. Journal of Institute of Metals 74, 537–562.

Yu, S.S., Lu, Y.B., Cai, Y., 2013. The strain-rate effect of engineering materials and
C

its unified model. Latin American Journal of Solids and Structures 10, 833–844.
AC

Zhang, B., Shim, V., 2010. Effect of strain rate on microstructure of polycrystalline
oxygen-free high conductivity copper severely deformed at liquid nitrogen tempera-
ture. Acta Materialia 58, 6810–6827.

AppendixA. Models summary

Table A.7 summarizes the constitutive formulations related to the MTS, BRP, MR,
and the present model. Constitutive features outlined are the flow stress, internal state
variable evolution, strain rate history, and the model parameters.

39
ACCEPTED MANUSCRIPT

Table A.7: Viscoplastic models summary.

PT
Present model (using model of Perić (1993))
 q 1
σf low = (σy + A) 1 + 32 ϑǫ̇
m
Flow stress

RI
Ȧ = Ȧ1 + Ȧ2
ISV evolution Ȧ1 = δ (A∞ − A1 ) ǫ̇ and A2 = cA∞ ǫ
A = A∞ [1 + cǫ − exp (−δǫ)]

SC
 ξ  
ǫ̇−ǫ̇lwr
Strain rate history A∞ = Alwr
∞ + ǫ̇up −ǫ̇lwr
Aup lwr
∞ − A∞
n o
Parameters (10) σy , ϑ, m, δ, c, ǫ̇lwr , ǫ̇up , Aup lwr
∞ , A∞ , ξ

Molinari-Ravichandran model (Molinari and Ravichandran, 2005)

U
  θ
δ0
ǫ̇ A
Flow stress σf low = σ̂ (d)
AN
ǫ̇0δ
δ̇ = − δδrs (δ 2 − δs δ) ǫ̇
ISV evolution
δ =  δs δs
1− 1− δ exp(−δr ǫ)
0
M
  ξr  −νr 
ǫ̇ θ
δr = δr0 1 + ar
Strain rate history  
ǫ̇r0
ξs 
θ0
−νs 
ǫ̇ θ
δs = δs0 1 + as ǫ̇s0 θ0
D

Parameters (14) {σ̂, δ0 , A, ǫ̇0 , ǫ̇r0 , ǫ̇s0 , δr0 , δs0 , ar , as , ξr , ξs , νr , νs }


Bodner-Partom-Rubin model (Bodner and Rubin, 1994)
TE

h  i− 1

Flow stress σf low = 2 ln √2ǫ̇30ǫ̇ 2n


Z
Ż = m (Z1 − Z) σ ǫ̇
ISV evolution
m = mb + (ma + mhb ) exp [−m (Z − Z0 )]
EP

qci
ǫ̇
Strain rate history ma = Ma 1 + ǫ̇s
0
Parameters (9) {ǫ̇0 , n, Z1, ma , mb , mc , ǫ̇s0 , q, Z0}
C

MTS model (Follansbee and Kocks, 1988)


 1
ǫ̇ 1 p
( )

kθ ln ǫ̇0
AC

 q
Flow stress σ = σ̂a + (σ̂ − σ̂a ) 1 − g0 µb3
 
h  i
σ̂ = h0 1 − tanh 2 σ̂σ̂−σ̂ a
s −σ̂a
ǫ̇
ISV evolution
n
ǫ = h tanh2 (2)−1 (σ̂ − σ̂a ) tanh (2) + (σ̂s −σ̂
tanh(2) a)
×
0[ ] 2
h    io σ̂f
× ln tanh (2) cosh 2 σ̂σ̂−σ̂a
− sinh 2 σ̂σ̂−σ̂ a

s −σ̂a s −σ̂a σ̂i
h  i
kθ ǫ̇
Strain rate history σ̂s = σ̂s0 exp Aµb 3 ln ǫ̇s0
and h0 = c1 + c2 ln (ǫ̇) + c3 ǫ̇
Parameters (14) {σ̂a , k, µ, b, g0, p, q, ǫ̇0 , σ̂s0 , ǫ̇s0 , A, c1 , c2 , c3 }

40
ACCEPTED MANUSCRIPT
Highlights

 A simple effective constitutive proposal is presented to model


viscoplastic metals.

 Model calibration is performed in subsequent steps in a wide strain

PT
rate range.

 Proposed model captures the sudden raising of flow stress at high

RI
strain rates.

SC
 Model predicts strain rate history effects and linear hardening at large
strains.

U
AN
M
D
TE
C EP
AC

Potrebbero piacerti anche