You are on page 1of 9

International Journal of Non-Linear Mechanics 95 (2017) 10–18

Contents lists available at ScienceDirect

International Journal of Non-Linear Mechanics


journal homepage: www.elsevier.com/locate/nlm

A dynamic flow rule for viscoplasticity in polycrystalline solids under high


strain rates
Md. Masiur Rahaman a , Bensingh Dhas a , D. Roy a , J.N. Reddy b, *
a
Computational Mechanics Lab, Department of Civil Engineering, Indian Institute of Science, Bangalore 560012, India
b
Advanced Computational Mechanics Lab, Department of Mechanical Engineering,Texas A&M University, College Station, TX 77843-3123, United States

a r t i c l e i n f o a b s t r a c t

Keywords: For visco-plasticity in polycrystalline solids under high strain rates, we introduce a dynamic flow rule (also called
Micro-inertia the micro-force balance) that has a second order time derivative term in the form of micro-inertia. It is revealed
High strain rate that this term, whose physical origin is traced to dynamically evolving dislocations, has a profound effect on the
Visco-plasticity macro-continuum plastic response. Based on energy equivalence between the micro-part of the kinetic energy
Hyperbolic flow rule
and that associated with the fictive dislocation mass in the continuous dislocation distribution (CDD) theory,
Micro-inertial length scale
an explicit expression for the micro-inertial length scale is derived. The micro-force balance together with the
classical momentum balance equations thus describes the viscoplastic response of the isotropic polycrystalline
material. Using rational thermodynamics, we arrive at constitutive equations relating the thermodynamic forces
(stresses) and fluxes. A consistent derivation of temperature evolution is also provided, thus replacing the
empirical route. The micro-force balance, supplemented with the constitutive relations for the stresses, yields a
locally hyperbolic flow rule owing to the micro-inertia term. The implication of micro-inertia on the continuum
response is explicitly demonstrated by reproducing experimentally observed stress–strain responses under high
strain-rate loadings and varying temperatures. An interesting finding is the identification of micro-inertia as the
source of oscillations in the stress–strain response under high strain rates.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction with experimental observations under high strain-rates and varying


temperatures still remains elusive and attempts in this direction are
Viscoplasticity is a very intricate manifestation of highly nonlinear often mired in controversies. To quote an instance, most existing
dynamical processes associated with microscopic defects (e.g. disloca- continuum viscoplastic models are unable to capture the experimentally
tions, micro-voids etc.). Typically, the underlying microscopic defects, observed oscillations in the stress–strain response of metals (e.g. molyb-
like dislocations for metals, tend to self organize in the form of patterns denum, tantalum etc.) under high strain rates and such phenomena are
and render the microscale deformation field heterogeneous, even as sometimes dismissed as mere experimental artefacts [10]. Indeed, the
the macroscopic field of interest in continuum viscoplasticity may
fact that certain transient features of viscoplasticity under high strain
remain homogeneous [1]. Despite the demonstrable success of classical
rates may not be revealed without experiments aided by an adequately
viscoplasticity models for a large class of problems (especially with
time-resolved sensing facility also underlines the importance of rational
length scale above several microns), their inadequacy in modelling
problems that involve mesoscopic length scales (typically in the range of underpinnings of a predictive model.
a tenth of micron to a few tens of micron) or lower is well documented. In introducing the desired capability within continuum viscoplas-
Such failure has been put down to the absence of microstructure-related ticity, one must start by grasping the limitations of the existing ap-
length scale information in the model to capture the size dependency proaches in incorporating changes in the mechanical behaviour of
of the meso scale viscoplastic response. As a remedy, several length materials (e.g., yield strength, ductility, etc.) under different strain-rates
scale dependent plasticity models [2–9] have also been developed in and temperatures. There are both phenomenological and physically
the last few decades. Unfortunately, a rationally grounded continuum motivated models to predict material behaviour under high strain
model with the capability of reproducing viscoplastic response in accord rates and elevated temperatures. Initial attempts were in developing

* Corresponding author.
E-mail addresses: masiur@civil.iisc.ernet.in (Md.M. Rahaman), bensingh@civil.iisc.ernet.in (B. Dhas), royd@civil.iisc.ernet.in (D. Roy), jnreddy@tamu.edu (J.N. Reddy).

http://dx.doi.org/10.1016/j.ijnonlinmec.2017.05.010
Received 18 April 2017; Received in revised form 21 May 2017; Accepted 21 May 2017
Available online 16 June 2017
0020-7462/© 2017 Elsevier Ltd. All rights reserved.
Md.M. Rahaman et al. International Journal of Non-Linear Mechanics 95 (2017) 10–18

empirical models using simple uniaxial stress–strain data [11] and one in the equation of motion for dislocation, especially at high strain
dimensional stress wave propagation approaches taking into account rates, may significantly alter the predicted micro-structural evolution
the effect of temperature, strain and strain rate on the flow stress. For and the macroscopic response compared to the oft-used over-damped
the von Mises type of materials, these one dimensional models could equation of motion. This motivates us to trace the dislocation inertia to
be used within the general three dimensional case by replacing the one its physical roots and build up a rational set-up to investigate its effect
dimensional quantities by equivalent invariant measures of stress, strain on the continuum response.
and strain rate tensors. Johnson and Cook [12] proposed a phenomeno- Specifically, by extending the micro force balance of Gurtin [23],
logical model to predict the material flow stress subjected to high strain, we propose a dynamic flow rule that contains a micro inertia term to
strain rate and temperature. Although, the Johnson–Cook (JC) model is account for dislocation inertial effects on the viscoplastic response of
very popular thanks to its simplicity and known parameter values for polycrystalline materials under high strain rates. A thermodynamically
various materials of interest, the assumed uncoupling of strain, strain consistent evolution equation for the temperature associated with the
rate and temperature effects on the flow stress is rarely applicable to plastic flow is also derived. A probable expression for the micro-inertial
certain materials (e.g. molybdenum) where rate sensitivity changes with length scale is arrived at based on energy equivalence between the
varying temperature. Moreover, the JC model does not account for micro-kinetic energy and the kinetic energy pertaining to the dislocation
thermal or strain history effects, accounting of which underlines the mass in the CDD framework [21]. We demonstrate that the inclusion of
relative benefits of an internal-variable-based theory [13]. the micro-inertia term enables capturing the experimentally observed
In order to consider the coupling among strain, strain rate and oscillations in the stress–strain responses of molybdenum and high-
temperature, Zerilli and Armstrong [14] have proposed a dislocation strength low-alloy steel (HSLA-65) at high strain rates. We also briefly
mechanics based constitutive model, which incorporates a thermal comment on how the micro-inertia term could be seamlessly incorpo-
activation analysis for overcoming obstacles to dislocation motion. The rated in several other existing models.
main idea in the Zerilli–Armstrong (ZA) model is that the constitutive
behaviour could be different depending on the rate-controlling mecha- 2. Micro-inertial length scale
nism specific to the material structure type [body-centred cubic (bcc),
face-centred cubic (fcc) and so on]. A very strong dependence of the We begin by describing a continuum representation of the disloca-
yield stress on strain rate and temperature is observed in bcc metals, tion inertia to be included in our thermoviscoplasticity formulation. In
even though the yield stress of fcc metals is mainly affected by strain order to incorporate dislocation acceleration effects in the continuum
hardening. The rationale behind this disparity is given based on the response, an additional form of kinetic energy, called the micro-kinetic
dislocation characteristics. Specifically, while the cutting of dislocation energy K𝑚𝑖𝑐𝑟𝑜 , is introduced. It is given by
forests is the main mechanism in fcc metals, it is the overcoming of
1 2 2
Peierls–Nabarro barriers in bcc metals. Although, the ZA model provides K𝑚𝑖𝑐𝑟𝑜 = 𝜌 𝑙 𝛾̇ (1)
2 0𝑚 𝑝
a very good correlation of the yield stress of metals with experiments,
where 𝛾𝑝 is the equivalent plastic strain introduced as an internal vari-
prediction of work-hardening behaviour for bcc metals is not very
able, 𝛾̇ 𝑝 is the rate of the equivalent plastic strain, 𝑙𝑚 is the micro-inertial
accurate. This drawback arises from the inaccurate assumption that
length scale and 𝜌0 is the reference material density. In the present
work-hardening behaviour in bcc metals does not depend on the strain
formulation, 𝑙𝑚 𝛾̇ 𝑝 is a continuum scalar representation of dislocation
rate and temperature [15]. Moreover, material parameters in the ZA
velocity. To characterize 𝑙𝑚 , we take recourse to the CDD theory wherein
model lose their physical meaning at high strain rates and temperatures
dislocations are represented using line like objects. There, stress and
because of the approximation used in the derivation of the constitutive
strain fields generated by dislocations are typically described using
relations (ln(1 + 𝑥) ≈ 𝑥, 𝑥 a function of strain rate and temperature).
linear elasticity [21,24]. The force on a dislocation f is thus given by
Voyiadjis and Abed [16] have proposed a constitutive relation for
the Peach–Koehler equation.
metals based on a modified ZA model, wherein the exact value of the
function ln(1 + 𝑥) replaces its truncated expansion. The Voyiadjis–Abed f = −l × 𝝈b. (2)
(VA) model reportedly predicts the flow stress better at relatively higher
strain rates. In addition to thermally activated dislocation interactions, Here 𝝈 is the stress field, b the Burgers vector and l the line direction
the effect of dislocation drag at very high strain-rates of order 107 associated with the dislocation. As the CDD theory is adopted within
has been considered important, resulting in the Preston–Tonks–Wallace a linear set up, it allows an additive decomposition of the stress into
(PTW) model. However, even after incorporating additional physics an external component 𝝈 𝑒 and a self-stress 𝝈 𝑠 . The external stress 𝝈 𝑒 is
associated with dislocation motion, these models do not always show a caused by the applied boundary conditions and body forces, whereas
very good agreement with the experiment, viz. for molybdenum over a the self-stress is due to the presence of dislocations and their motion.
wide range of strain, strain rate and temperature [17]. Many other such The self-stress is again additively decomposed into a static part 𝝈 0 and a
attempts at continuum modelling of response of materials across differ- dynamic part 𝝈 𝑑 . The force on the dislocation loop due to the dynamic
ent strain rate and temperature regimes have also been reported [18– part of the self-stress is given by,
20]. Even so, to our knowledge, few are able to offer a rationally
grounded route to capturing certain finer features in the experimentally f𝑑 = 𝝁(l, l′ )v̇ 𝑑 (l′ )𝑑l. (3)

observed stress–strain plots, e.g. oscillations in the stress–strain response
In the above expression, 𝝁 is the mass per unit length of the dislocation
of polycrystalline materials (e.g. molybdenum, high-strength low-alloy
and v𝑑 its velocity. l and l′ are the local tangents to the dislocation line
steel etc.) under high strain rates. Thus the question arises as to whether
respectively at points 𝑠 and 𝑠′ . The expression for mass per unit length
these models have consistently overlooked any important mechanism
of the dislocation line is given by (see [21]),
related to dislocation motion which could be very important at high ( )
( )4
strain-rate loading and possibly responsible for oscillations in the stress– 1 𝑐𝑡
strain response. 𝝁 = 𝜌0 ‖b‖2 [I − (l ⊗ l′ )] 1 + sin2 𝜃
∫𝛤 ∫𝛤 8𝜋 𝑐𝑙
In search for an answer to this question, we recognize an important ( )
𝜔(𝜷)𝜔(𝜷 ′ )
aspect of dislocation motion, viz. dislocation inertia, which is not incor- × 𝑑𝑎𝑑𝑎′ 𝑑𝑙𝑑𝑙′ (4)
∫𝐴 ∫𝐴 ‖𝜷 − 𝜷 ′ ‖
porated in existing viscoplasticity models. Kosevich [21] has identified
that the dislocation inertia plays a pivotal role in the evolution of micro 𝜔 is an averaging kernel with the property ∫𝐴 𝜔𝑑𝐴 = 1, 𝐴 is the
structure when the dislocation acceleration is large. More recently, dislocation core area often assumed to√be circular with diameter
√ 2𝑟0
Wang et al. have shown that [22] the inclusion of dislocation inertia and 𝛤 is the dislocation line. 𝑐𝑙 = (𝜆 + 2𝐺)∕𝜌0 and 𝑐𝑡 = 𝐺∕𝜌0

11
Md.M. Rahaman et al. International Journal of Non-Linear Mechanics 95 (2017) 10–18

are the longitudinal and transverse wave velocities in the media. 𝜷 Comparison of Eq. (12) with the proposed micro-kinetic energy given
and 𝜷 ′ are dislocation core cross section variables respectively at 𝑠 by Eq. (1) yields the expression for the micro inertial length scale as,
and 𝑠′ . 𝜃 = b.(x(𝑠) − x(𝑠′ ))∕‖(x(𝑠) − x(𝑠′ )‖ ‖b‖), x(𝑠) and x(𝑠′ ) are the ( ( )4 )
2 1 𝑐𝑡 2
position vector of the dislocation line 𝑠 and 𝑠′ respectively. I is the 𝑙𝑚 = 1+ sin 𝜃 log(1∕𝑟0 ). (13)
4𝜋 𝑐𝑙
second order identity tensor. The averaging kernel is introduced to avoid
the singularity of the elastic energy associated with the zero thickness It is important to note that this expression for the micro-inertial length
of dislocation line by smearing the Burgers vector over an area 𝐴. The scale is limited to straight dislocations. This assumption can be re-
non-local nature of the integral in Eq. (4) poses difficulties in finding laxed by adopting a geometric theory for curved dislocations presented
a closed form expression for the dislocation mass. To make the integral in [26,27]. Moreover the calculation of dislocation mass is based on
tractable, the following assumptions are made. (a) The dislocation line is the infinitesimal strain assumption; such assumptions are not required
assumed to be straight, i.e. l = l′ . (b) The Burgers vector associated with if one adopts a geometric setting for describing dislocations [28]. As
the dislocation is uniformly smeared over an area of radius 𝑟0 . In [25], our current interest is mainly to show the effect of micro inertia on
a cubic kernel is used to smear the Burgers vector over the dislocation plastic deformation, derivation of a micro-inertial length scale using a
core area so as to avoid the unboundedness of the elastic energy. Such geometric theory of curved dislocations is not attempted here.
kernels would require numerical methods for calculating the non-local
dislocation mass. With the uniform smearing adopted in this work, the 3. Thermoviscoplasticity model
mass of dislocation per unit length for a straight dislocation is found to
be, We now describe a thermoviscoplasticity model with the aid of a
( ) micro inertia driven dynamic flow rule. The theoretical background
( )4
𝜌 ‖b‖2 𝑐𝑡 needed for this development could be found in the work of Gurtin [23],
𝝁𝑠𝑡 = 0 (I − l ⊗ l) 1 + sin2 𝜃 log(1∕𝑟0 ). (5)
4𝜋 𝑐𝑙 though the latter is essentially limited to the isothermal quasi-static case.

Having discussed the notion of mass associated with dislocation, we now 3.1. Kinematics
proceed to calculate the kinetic energy associated with its motion. The
expression for the kinetic energy per unit length of the dislocation line Consider a body , the reference configuration of which could be
is given by, identified with an open set 𝑈 ⊂ R3 . The deformation of the body at time
1 𝑑 𝑡 ∈ R+ is defined by 𝝋𝑡 ∶ 𝑈 → 𝑉 ⊂ R3 where the deformed configuration
K𝐶𝐷𝐷 = v .𝝁𝑠𝑡 v𝑑 . (6) of the body is identified with the open set 𝑉 . The spatial derivative of
2
the deformation 𝝋𝑡 is called the deformation gradient, which is given
The present continuum formulation is related to the CDD through the
by F = 𝜕𝝋∕𝜕X in a Cartesian co-ordinate system with X denoting the
following relation,
co-ordinates of the reference configuration. The deformation gradient F
should satisfy the constraint 𝑑𝑒𝑡(F) > 0 to preserve the orientation and
Sym(J) = D𝑝 (7)
local invertibility of the deformation. In order to characterize the plastic
J is the dislocation flux density tensor and Sym(.) denotes the symmetric deformation, Kröner–Lee decomposition of the deformation gradient
part of a tensor. In the CDD theory, J is expressed in terms of the into an elastic distortion F𝑒 and a plastic distortion F𝑝 is adopted here.
dislocation velocity, line direction and the Burgers vector b as,
F = F𝑒 F𝑝 . (14)
J = (l × v𝑑 ) ⊗ b (8)
It should be noted that even though the deformation gradient is inte-
D𝑝 , the plastic strain rate tensor (see Section 3 for details), is the rate of grable, the elastic and plastic parts of it are in general non-integrable;
plastic slip occurring in a slip plane. The slip plane and the direction of that is one cannot find deformations whose derivatives will be F𝑒 and F𝑝 .
slip are given by m and s respectively. The magnitude of the plastic slip For incompressible plastic deformation, F𝑝 needs to satisfy the constraint
rate is given by the equivalent plastic strain rate 𝛾̇ 𝑝 , so that we have 𝑑𝑒𝑡(F𝑝 ) = 1 which implies that 𝑑𝑒𝑡(F) = 𝑑𝑒𝑡(F𝑒 ). The rate of deformation
is the velocity field v, gradient of which is defined as,
D𝑝 = 𝛾̇ 𝑝 Sym(s ⊗ m). (9) ̇ −1 .
L = FF (15)
In the equation above, it is assumed that plastic slip occurs only in
Using Kröner–Lee decomposition in Eq. (15) leads to the following
one slip plane at a given instant of time. However, the slip plane and relation,
the slip direction could be time dependent which is accounted for in
the proposed formulation (see Section 3 for details). Expressing the L = L𝑒 + F𝑒 L𝑝 F−1
𝑒 . (16)
dislocation velocity in the orthonormal basis {l, m, s}, we may write,
For brevity, we have introduced the following notations
v 𝑑 = 𝑣𝑙 l + 𝑣𝑚 m + 𝑣𝑠 s (10)
L𝑒 ∶= Ḟ 𝑒 F−1
𝑒 ; L𝑝 ∶= Ḟ 𝑝 F−1
𝑝 . (17)
𝑣𝑙 , 𝑣𝑚 and 𝑣𝑠 are the velocity components of dislocation in the l, m and
The elastic strain rate and elastic spin tensors are defined to be the
s directions respectively. The slip direction is typically taken the same
symmetric and antisymmetric parts of L𝑒 .
as Burgers vector direction. Employing Eqs. (9) and (10) in Eq. (7), we
arrive at the following equation, 1 1
D𝑒 = (L + L𝑇𝑒 ); W𝑒 = (L𝑒 − L𝑇𝑒 ). (18)
2 𝑒 2
v𝑑 = 𝑣𝑠 s; 𝑣𝑠 = 𝛾̇ 𝑝 ∕‖b‖. (11) The symmetric and anti-symmetric parts of L𝑝 are respectively called
the plastic strain rate tensor and the plastic spin tensor.
From these expressions, it may be seen that 𝛾̇ 𝑝 is related to the dislo-
cation velocity in the slip direction s. Using Eq. (11), we may express 1 1
D𝑝 = (L + L𝑇𝑝 ); W𝑝 = (L𝑝 − L𝑇𝑝 ). (19)
K𝐶𝐷𝐷 in terms of the internal variable 𝛾̇ 𝑝 . 2 𝑝 2
( ) In the present formulation, plastic flow is assumed to be only due to
( )4
1 𝑐𝑡 2 the gliding motion of dislocation. Thus the plastic spin tensor W𝑝 is set
K𝐶𝐷𝐷 = 𝜌 1+ sin 𝜃 log(1∕𝑟0 )𝛾̇ 𝑝2 . (12)
8𝜋 0 𝑐𝑙 to be zero, which implies that L𝑝 = D𝑝 . Now, from the second part of

12
Md.M. Rahaman et al. International Journal of Non-Linear Mechanics 95 (2017) 10–18

Eq. (17), one can arrive at the evolution rule for F𝑝 given by Ḟ 𝑝 = D𝑝 F𝑝 , plastic deformation. Associated with the micro-traction 𝝌(n), a vector-
the solution of which is the matrix exponential function given below. valued stress called the micro-stress 𝝃 is introduced. Now, using the
second transport theorem and using 𝝌(n) = 𝝃 ⋅ n in Eq. (25), one may
F𝑝 (𝑡) = exp((𝑡 − 𝑡0 )D𝑝 )F0𝑝 ; F𝑝 (𝑡0 ) = F0𝑝 . (20) arrive at,
Since D𝑝 is symmetric, exp((𝑡 − 𝑡0 )D𝑝 ) is symmetric and positive definite. 2
We accordingly conclude that the plastic distortion tensor remains 𝜌0 𝑙𝑚 𝛾̈ 𝑝 𝑑𝑉 = (∇X .𝝃 + 𝜋)𝑑𝑉
̄ . (26)
∫ ∫
symmetric and positive definite for all time if F0𝑝 = I. Thus, the
Applying the localization theorem on Eq. (26) leads to the following
assumption of irrotational plastic flow i.e. W𝑝 = 0 renders the plastic
local form for the micro-force balance.
distortion tensor a pure stretch and allows us to characterize the plastic
deformation through the plastic strain rate tensor D𝑝 , which is written 2
∇X .𝝃 + 𝜋̄ = 𝜌0 𝑙𝑚 𝛾̈ 𝑝 . (27)
as a product of its magnitude 𝛾̇ 𝑝 and a unit tensor N𝑝 .
D𝑝 The local micro-force balance given by Eq. (27) acts as a dynamic
D𝑝 = 𝛾̇ 𝑝 N𝑝 ; 𝛾̇ 𝑝 = ‖D𝑝 ‖; N𝑝 ∶= (21) flow rule and provides for the evolution of the equivalent plastic strain
‖D𝑝 ‖
in a yield-free set up.
𝛾̇ 𝑝 is treated as an internal variable. It may be observed that 𝛾̇ 𝑝 ≥ 0 by
its definition itself and N𝑝 determines the direction of the plastic flow.
3.3. Thermodynamics first law and internal energy evolution
The equivalent plastic strain is defined as,
𝑡
𝛾𝑝 = 𝛾̇ 𝑝 (𝑠)𝑑𝑠. (22) In the framework of rational thermodynamics, existence of an in-
∫−∞ ternal energy density is assumed and the evolution of it is derived
The internal variable 𝛾𝑝 , which is positive and non-decreasing by based on the conservation of energy. The first law of thermodynamics,
definition, is taken as an additional field variable in this theory. In a version of the law of energy conservation, could be stated as follows
order to quantify the elastic part of total deformation, a strain tensor E𝑒 , (see Reddy [29]):
which is a measure of the change in length of an infinitesimal material
𝑑𝐸
line element between the intermediate and deformed configurations, 𝑑𝐸 = 𝛿𝑊 + 𝛿𝑄; = 𝑊 ◦ + 𝑄◦ (28)
is introduced on the intermediate configuration. Recall that the inter- 𝑑𝑡
mediate configuration is the collection of tangent spaces pulled back where 𝑊 is the work done by the external forces on the system and 𝑄 the
using F−1𝑒 . The expression for E𝑒 in terms of the Green Lagrangian strain
heat supply to the system. Note that 𝑊 and 𝑄 are path dependent and
E = 12 (F𝑇 F − I) and the plastic distortion F𝑝 is given as, hence not state functions; (.)◦ denotes the rate of an inexact differential,
( ) signifying path dependence of 𝑊 and 𝑄. The total energy 𝐸 equals the
1 −𝑇 −1
E𝑒 = F−𝑇 −1
𝑝 EF𝑝 + F F −I . (23) sum of the internal energy 𝑈 and the kinetic energy 𝐾, that is, 𝐸 = 𝐾+𝑈 .
2 𝑝 𝑝
For an arbitrary subdomain  ⊂ , 𝑈 may be written in terms of the
It is interesting to observe that the assumption of incompressible specific internal energy 𝑒 as,
plastic deformation i.e. 𝑑𝑒𝑡(F𝑝 ) = 1 implies that Tr(D𝑝 )(𝑡) = 0 which can
be seen from the following arguments. From Eq. (20), it is easy to find 𝑈= 𝜌0 𝑒𝑑𝑉 . (29)
that 𝑑𝑒𝑡(F𝑝 ) = 1 implies det(exp(((𝑡 − 𝑡0 )D𝑝 ))) = 1. Also, we know that ∫
exp(Tr((𝑡 − 𝑡0 )D𝑝 )) = det(exp(((𝑡 − 𝑡0 )D𝑝 ))) is the solution to the scalar The rate of the internal energy is given by
differential equation 𝑦̇ = Tr(D𝑝 )𝑦, with initial condition 𝑦0 = 1. Now,
consider the following cases. If Tr(D𝑝 ) > 0, the solution 𝑦(𝑡) → ∞ as 𝑑𝑈
= 𝜌 𝑒𝑑𝑉
̇ . (30)
𝑡 → ∞. On the other hand, if Tr(D𝑝 ) < 0, 𝑦(𝑡) → 0 as 𝑡 → ∞. However, 𝑑𝑡 ∫ 0
for the case of Tr(D𝑝 ) = 0, every point 𝑦 in the phase space is a neutral Similarly the rate of the kinetic energy could be written, considering
fixed point of the differential equation and the solution 𝑦(𝑡) = 𝑦0 , ∀𝑡. contributions from its conventional macroscopic and the presently
Thus, the condition det F𝑝 (𝑡) = 1 is satisfied by imposing Tr(D𝑝 )(𝑡) = 0 introduced microscopic constituents. First, we write 𝐾 in terms of its
on D𝑝 for all time 𝑡. macro and micro components:
( )
3.2. Balance laws 1 1 2 2
𝐾= 𝜌 v.v + 𝜌0 𝑙𝑚 𝛾̇ 𝑝 𝑑𝑉 . (31)
∫ 2 0 2
With the kinematics available entirely from geometrical considera- One thus obtains the rate of the kinetic energy as,
tions without a reference to the external forcing, (macro) momentum ( )
balance laws enable a description of the continuum dynamics under 𝑑𝐾 1 1 2
= ̇ + 𝜌0 𝑙 𝑚
𝜌 v.v 𝛾̈ 𝑝 𝛾̇ 𝑝 𝑑𝑉 . (32)
𝑑𝑡 ∫ 2 0 2
the action of the external forces. Local forms of linear and angular
momentum balance, which hold for each material point in the reference Now an expression for the external power 𝑊 ◦ may be written for
configuration, are as follows (see Reddy [29]): an arbitrary sub-domain  ⊂  with boundary 𝜕, considering the
macroscopic part corresponding to macro-traction Pn and body force
̇
∇X .P + b0 = 𝜌0 v; FP𝑇 = PF𝑇 (24) b0 , and the microscopic part corresponding to the micro-traction 𝝌(n).
P is the first Piola stress tensor, b0 the body force and ∇X the gradient
operator in the reference configuration. 𝑊◦ = Pn.v𝑑𝐴 + b0 .v𝑑𝑉 + 𝝌(n).𝛾̇ 𝑝 𝑑𝐴. (33)
∫𝜕 ∫ ∫𝜕
A micro force balance is proposed based on the premise that the ‘‘rate
Employing the relation 𝝌(n) = 𝝃.n and invoking Stokes’ theorem for the
of change of momentum associated with the rate of plastic deformation
boundary terms in Eq. (33), the following is obtained.
equals all forces arising from (and maintaining) the effect of plastic
deformation’’ [30]. For any arbitrary sub-domain  ⊂  with boundary ( )
𝑊◦ = (∇X .P + b0 ).v + P ∶ Ḟ + ∇X .𝝃 𝛾̇ 𝑝 + 𝝃.∇X 𝛾̇ 𝑝 𝑑𝑉 . (34)
𝜕, we may write, ∫
𝑑 Considering the heat source ℎ and the heat flux vector q, the thermal
𝜌 𝑙2 𝛾̇ 𝑑𝑉 = 𝝌(n)𝑑𝐴 + 𝜋𝑑𝑉
̄ (25)
𝑑𝑡 ∫ 0 𝑚 𝑝 ∫𝜕 ∫ power 𝑄◦ may be expressed as follows.
where n is the unit normal to the boundary 𝜕, 𝝌(n) is the micro-traction ( )
and 𝜋̄ is the net micro body force which acts as the driving force for the 𝑄◦ = 𝜌0 ℎ − ∇X .q 𝑑𝑉 . (35)
∫𝜕

13
Md.M. Rahaman et al. International Journal of Non-Linear Mechanics 95 (2017) 10–18

Substituting Eqs. (32) and (34)–(36) in Eq. (28) and using macro In terms of the Helmholtz free energy 𝜓 ∶= 𝑒 − 𝜃𝑠, Eq. (44) is rewritten
and micro force balances given by Eqs. (24) and (27), we obtain the as,
following equation. ( )
̇ − q .∇x 𝜃 ≥ 0.
− 𝜌0 𝜓̇ + P ∶ Ḟ − 𝜋̄ 𝛾̇ 𝑝 + 𝝃.∇x 𝛾̇ 𝑝 − 𝜃𝑠 (45)
( ) 𝜃
𝜌0 𝑒𝑑𝑉
̇ = P ∶ Ḟ − 𝜋̄ 𝛾̇ 𝑝 + 𝝃.∇X 𝛾̇ 𝑝 + 𝜌0 ℎ − ∇X .q 𝑑𝑉 . (36)
∫ ∫ Employing Eqs. (16), (17) and (43) into Eq. (45) and using the relation
𝑇
An evolution equation for the internal energy density results upon Ė 𝑒 = (1∕2)(Ḟ 𝑒 F𝑒 + F𝑇𝑒 Ḟ 𝑒 ) we arrive at,
localization of Eq. (36) which can be written as ( )
−𝜌0 F𝑒 𝜕E𝑒 𝜓 + PF𝑇𝑝 ∶ Ḟ 𝑒 + (𝝃 − 𝜌0 𝜕∇𝛾𝑝 .𝜓).∇X 𝛾̇𝑝
𝜌0 𝑒̇ = P ∶ Ḟ − 𝜋̄ 𝛾̇ 𝑝 + 𝝃.∇X 𝛾̇ 𝑝 + 𝜌0 ℎ − ∇X .q. (37) ( ) ( ) q
− 𝜌0 𝜕𝜃 𝜓 + 𝜌0 𝑠 𝜃̇ + F𝑇𝑒 PF𝑇𝑝 ∶ N𝑝 − 𝜋̄ 𝛾̇ 𝑝 − .∇X 𝜃 ≥ 0. (46)
𝜃
3.4. Material frame indifference and constitutive restrictions Applying the Coleman–Noll procedure [31] to the equation above, we
obtain the following constitutive relations.
Material frame indifference is an important principle in continuum
mechanics to impose restrictions on constitutive functions [29]. A P = 𝜌0 F𝑒 𝜕E𝑒 𝜓F−𝑇
𝑝 (47)
change of reference frame is given by an element from the Euclidean
group, which is characterized by a rotation Q ∈ SO(3), a translation in
𝝃 = 𝜌0 𝜕∇𝛾𝑝 .𝜓 (48)
space c ∈ R3 and a translation in time 𝑎 ∈ R. In general, Q and c are
functions of time. The change of reference frame is given by,
𝑠 = −𝜕𝜃 𝜓. (49)
x∗ = Qx + c; 𝑡∗ = 𝑡 + 𝑎 (38)
Then the inequality given by Eq. (46) boils down to requiring that the
x∗ and 𝑡∗ are the co-ordinate and time measured from an arbitrary
last two terms are greater than or equal to zero, i.e.
moving frame. A variable with the superscript ∗ denotes a quantity from
( ) q
the arbitrary moving frame. The transformation rule for the deformation F𝑇𝑒 PF𝑇𝑝 ∶ N𝑝 − 𝜋̄ 𝛾̇ 𝑝 − .∇X 𝜃 ≥ 0. (50)
gradient under a change of reference frame is given by, 𝜃
This constraint can be satisfied by choosing
F∗ = QF. (39)
F𝑇𝑒 PF𝑇𝑝 ∶ N𝑝 − 𝜋̄ = g1 𝛾̇ 𝑝 ; g1 ≥ 0 (51)
As F𝑝 is defined on the reference configuration, F∗𝑃 = F𝑝 . From Eq. (39)
it follows that F∗𝑒 = QF𝑒 . The velocity vector in the arbitrary rotating
frame is given by, q = −g2 ∇X 𝜃; g2 ≥ 0 (52)

ẋ ∗ − Qẋ = ċ + W(x∗ − c); ̇ 𝑇.


W = QQ (40) where g1 and g2 are constitutive functions. Since N𝑝 is symmetric, only
the symmetric part of F𝑇𝑒 PF𝑇𝑝 does work on D𝑝 . Moreover the pressure
The rate quantities transform according to the following rules.
part of F𝑇𝑒 PF𝑇𝑝 does no work on D𝑝 ; this is because D𝑝 is deviatoric. The
L∗ = W + QLQ𝑇 ; L∗𝑝 = L𝑝 ; L∗𝑒 = W + QL𝑒 Q𝑇 . (41) validity of the last statement may be gauged from the fact that the inner
product between a deviatoric tensor and an arbitrary tensor equals the
Since the Cauchy stress T is a frame indifferent quantity, its transfor- inner product between the deviatoric tensor and the deviatoric part of
mation under a change of reference frame is given by T∗ = QTQ𝑇 . the arbitrary tensor. For clarity of notation, we introduce the following
Using this, the transformation rule for the first Piola stress tensor may stress,
be deduced as P∗ = QP. With the vector micro stress 𝝃 defined on the ( )
reference configuration, frame indifference demands 𝝃 ∗ = 𝝃. Suppose T𝑝 ∶= Sym F𝑇𝑒 PF𝑇𝑝 . (53)
̃ 𝑒 , 𝜶) is a constitutive function that depends on the elastic
that 𝚵(F
distortion F𝑒 and an array of internal variables 𝜶. Then the requirement Assuming co-directionality between the plastic flow and Tdev
𝑝 = T𝑝 −
of the constitutive function being frame indifferent is reduced to the (1∕3)Tr(T𝑝 ), we arrive at the following conclusion.
̂ 𝑒 , 𝜶).
existence of the function 𝚵(E
Tdev
𝑝
N𝑝 = ; 𝜏 ∶= F𝑇𝑒 PF𝑇𝑝 ∶ N𝑝 = ‖Tdev
𝑝 ‖. (54)
3.5. Second law of thermodynamics and constitutive modelling ‖Tdev
𝑝 ‖

We now present the constitutive equations obtained upon imposing


the second law of thermodynamics. The Helmholtz free energy 𝜓 is 3.6. Temperature evolution
assumed to depend constitutively on the elastic strain E𝑒 , the gradient
of the equivalent plastic strain ∇X 𝛾𝑝 and temperature 𝜃. With this, the The evolving visco-plastic response typically entails significant
free energy is expressed as, changes in the resulting temperature and a visco-plasticity model is
ideally required to reproduce these changes faithfully. We accordingly
𝜓 = 𝜓(E
̂ 𝑒 , ∇X 𝛾𝑝 , 𝜃). (42) aim at deriving a coupled, thermodynamically consistent temperature
evolution equation. We start by considering the free energy as the sum
Using chain rule, the rate of Helmholtz free energy is written as,
of the elastic energy 𝜌0 𝜓 el , a defect energy 𝜌0 𝜓 de and the free energy
𝜓̇ = 𝜕E𝑒 𝜓 ∶ Ė 𝑒 + 𝜕∇𝛾𝑝 𝜓.∇𝛾̇𝑝 + 𝜕𝜃 𝜓 𝜃.
̇ (43) due to temperature 𝜓 𝜃 . The decomposition, so accomplished, may be
thought of as a generalization over Gurtin’s [23] isothermal model.
In the equation above, 𝜕(.) with a suffix represents the derivative of a
function with respect to one of its arguments keeping the others fixed. 𝜌0 𝜓 = 𝜌0 𝜓 el + 𝜌0 𝜓 de + 𝜓 𝜃 . (55)
The second law of thermodynamics requires that the rate of entropy
production be greater than or equal to zero, which is enforced using the Using the Legendre transform 𝜓 ∶= 𝑒 − 𝜃𝑠 along with Eqs. (49) and (55),
local form of Clausius–Duhem inequality, the internal energy can be written as,
( )
(q) 𝜌 ℎ 𝑑𝜓 𝜃
𝜌0 𝑠̇ + ∇X . − 0 ≥ 0. (44) 𝜌0 𝑒 = 𝜌0 𝜓 el + 𝜌0 𝜓 de + 𝜓 𝜃 − 𝜃 . (56)
𝜃 𝜃 𝑑𝜃

14
Md.M. Rahaman et al. International Journal of Non-Linear Mechanics 95 (2017) 10–18

For zero mechanical strains i.e. F𝑒 = I and F𝑝 = I, considerations of the stress–strain response, so that any reasonable choice of the dynamic
first law of thermodynamics as in Eqs. (28) and (56) give the following: yield function should suffice. As we know that such a function should
contain terms reflecting strain hardening, strain rate hardening and
𝑑2𝜓 𝜃
𝜌0 𝑑𝑒 = −𝜃 𝑑𝜃 = 𝛿𝑄 ∶= 𝜌0 𝐶𝑉 𝑑𝜃 (57) temperature softening, we choose the functional form of g1 as,
𝑑𝜃 2
where 𝐶𝑉 is the specific heat. Consideration of Eq. (57) in the differen- ( ) ( 𝛾̇ 𝑝 )𝑚1 −1 ( )
1 ̄ 𝑚2
g1 = 𝑆0 + 𝐻0 𝛾𝑝𝑛 𝜃 (64)
tial form of Eq. (56) leads to the following: 𝑑0 𝑑0
( )
𝜃−𝜃ref
𝜌0 𝑒̇ = 𝜌0 𝜓̇ el + 𝜌0 𝜓̇ de + 𝜌0 𝐶𝑉 𝜃.
̇ (58) where 𝐻0 is the hardening modulus, 𝜃̄ = 𝜃 −𝜃 the homologous
melt ref
temperature, 𝑛 the hardening parameter, 𝑑0 the reference strain rate, 𝑚1
Employing Eqs. (37), (47), (48), (60) and (62) in Eq. (58), the temper- the strain rate hardening parameter and 𝑚2 the temperature softening
ature evolution may be obtained as, parameter which depends on the initial temperature 𝜃0 . In this study 𝑚2
1 [ ] is chosen as a quadratic function of 𝜃0 , i.e. 𝑚2 = 𝑎𝜃02 + 𝑏𝜃0 + 𝑐. Use of
𝜃̇ = (𝜏 − 𝜋)
̄ 𝛾̇ 𝑝 + 𝜌0 ℎ − ∇X .q . (59)
𝜌0 𝐶 𝑉 Eqs. (54) and (64) in Eq. (52) provides the expression of 𝜋. ̄
( ) ( 𝛾̇ 𝑝 )𝑚1
𝜋̄ = ‖Tdev
𝑝 ‖ − 𝑆0 + 𝐻0 𝛾𝑝
𝑛
𝜃̄ 𝑚2 . (65)
3.7. Specialization of constitutive functions 𝑑0

Our model so far is quite general. However, a quantification of the


3.8. Dynamic flow rule
thermodynamic forces requires a specific choice for the constitutive
functions, viz. 𝜓 el , 𝜓 de , g1 and g2 . From the application perspective,
Augmentation of the constitutive equations for the micro-forces
quadratic forms for 𝜓 el and 𝜓 de could be used to quantify the first
given by Eqs. (63) and (65) with the micro-force balance as in Eq. (27)
Piola stress and the vector micro-stress respectively. Computation of the
leads to the following second order partial differential equation.
driving force 𝜋̄ is accomplished through a proper choice of g1 based on
( ) ( 𝛾̇ 𝑝 )𝑚1
the physical understanding of the governing factors (e.g. applied strain 2
𝜌0 𝑙 𝑚 𝛾̈ 𝑝 = 𝑆0 𝑙12 ∇2X 𝛾𝑝 + ‖Tdev
𝑝 ‖ − 𝑆0 + 𝐻 0 𝛾𝑝
𝑛
𝜃̄ 𝑚2 . (66)
rate, temperature etc.) which affect the plastic flow. The constitutive 𝑑0
function g2 is taken as the thermal conductivity 𝜅 (a material constant) Eq. (66) acts as a flow rule which incorporates the dislocation accel-
so Eq. (52) is identifiable as the Fourier law of heat conduction. eration effect through the term involving second order time derivative
of the equivalent plastic strain. Although, for low strain rate loading,
3.7.1. Elastic free energy and constitutive equation for macro stress the left hand side of Eq. (66) may be negligible, it assumes a much
Presently, we limit our attention to isotropic materials and consider more important role under high strain rate loading as the dislocation
the following quadratic form of the elastic free energy. acceleration affects the plastic flow. For an inhomogeneous plastic flow,
1 Eq. (66) can be seen as a generalization of the micro-force balance by
𝜌0 𝜓 el = 𝐺(E𝑒 ∶ E𝑒 ) +𝜆Tr(E𝑒 )2 (60)
2 Gurtin [23] for high strain rates. In general, solution to Eq. (66) requires
where 𝐺 and 𝜆 are the Lamé parameters. Now, utilizing Eqs. (47) and specified initial conditions on 𝛾𝑝 and 𝛾̇ 𝑝 and boundary conditions.
(55), we obtain the expression for the first Piola stress tensor P in terms However, for homogeneous plastic deformation, the first term on the
of the kinematic variables as, right hand side of Eq. (66) vanishes and hence Eq. (66) reduces to a
nonlinear ordinary differential equation for 𝛾𝑝 . In such a case, Eq. (66)
( )
P = F𝑒 2𝐺E𝑒 + 𝜆Tr(E𝑒 )I F−𝑇
𝑝 . (61) may be thought as a generalization of the flow rule of the consistency
visco-plasticity theory [32].
Substituting Eq. (61) in Eq. (53), we can estimate T𝑝 , the deviatoric part
of which determines the plastic flow direction (see Eq. (54)).
4. Numerical results and discussions

3.7.2. Defect free energy and constitutive equation for vector micro-stress
We now undertake an assessment of the predictive capability of
Free energy associated with the defect is assumed to be of the form,
our model through numerical simulations. Specifically, we validate
1 the simulated results against high strain rate experimental data for
𝜌0 𝜓 de = 𝑆 𝑙2 ‖∇ X 𝛾𝑝 ‖2 (62)
2 01 molybdenum and high-strength low-alloy steel (HSLA-65) [33,34]. Our
where 𝑆0 is the initial yield strength and 𝑙1 is an energetic length scale. focus will be on how the proposed model captures the experimen-
Employing Eqs. (48) and (55), the vector micro-stress 𝝃 may be obtained tally observed oscillations in the stress–strain response. The relevant
as, macroscopic experiments correspond to a uniform state of stress and
homogeneous plastic deformation. Uniform state of stress implies that
𝝃 = 𝑆0 𝑙12 ∇ X 𝛾𝑝 . (63) the macroscopic force balance equation (24) is trivially satisfied in the
absence of body force. Homogeneous plastic deformation also implies
This vector micro-stress may be identified as the back stress that enables
that the first term of the right hand side of Eq. (66) is zero. Moreover,
the model to capture the Bauschinger effect.
experiments under high strain rate loading typically correspond to the
3.7.3. Constitutive equation for 𝜋̄ adiabatic condition; so the divergence of the heat flux term appearing in
Quantification of the driving force 𝜋̄ for plastic deformation needs a Eq. (59) vanishes. Thus, in practice, we solve a set of non-linear ordinary
specific choice of the constitutive function g1 . It is greatly interesting differential equations in order to compute the uni-axial strain and stress
and useful to observe that g1 may be identified with the so called respectively using the kinematic as well as constitutive relations.
dynamic yield stress. This identification makes possible to interface Computation of stress–strain response is done by choosing the
the most suitable dynamic yield function with the proposed model following deformation gradient which corresponds to a homogeneous
for a given strain rate and temperature. Thus, with the best available deformation.
evidence and/or physical understanding of the dynamic yield, g1 may be
F = (1 − 𝛼)(e1 ⊗ e1 ) + (e2 ⊗ e2 ) (67)
judiciously chosen from amongst the existing flow-stress models (e.g. JC,
ZA, VA etc.). This, by itself, constitutes an interesting study that we 𝛼 is a time dependent parameter, e1 and e2 are the standard basis
leave out of the scope of this work. Our present interest is only in vectors. Since we are interested in constant strain rate loading, 𝛼
assessing the extent to which micro-inertia influences the continuum is assumed to be a linearly varying function of time. The generated

15
Md.M. Rahaman et al. International Journal of Non-Linear Mechanics 95 (2017) 10–18

(a) Molybdenum. (b) HSLA-65.

Fig. 1. Stress–strain response for different strain rate loading 𝜖.̇ The solid line represents the stress–strain response for non-zero 𝑙𝑚 and the line with circular marker for 𝑙𝑚 = 0.

Table 1
Material parameters for Molybdenum and HSLA-65.a
𝐶𝑣 (J/K) 𝐻0 (MPa) 𝑆0 (MPa) 𝑛 𝑚1 𝑎 𝑏 𝑐 𝑑0 (1∕s) 𝜃ref (K) 𝑙𝑚 (m)
Mo 277 300 448 0.45 0.1 −2.56𝑒−7 4.24𝑒−4 0.11 100 398 1.0𝑒−3
HSLA-65 460 500 790 0.3 0.1 1.47𝑒−7 −6.71𝑒−5 0.18 100 77 3.0𝑒−3
a
The parameters mentioned in the table cannot be obtained by fitting the flow curve with the constitutive function for 𝜋.
̄ Since 𝜋̄ depends on the rate of equivalent plastic strain
which is only obtainable through solution of dynamic flow. Hence, an inverse problem is needed to determine the material parameters.

stress associated with the deformation is computed based on the elastic rates. We now aim at assessing how the computed response through our
part of the deformation gradient. The elastic and plastic parts of the model corresponds with available experimental evidence, specifically
deformation gradient are determined based on the flow rule Eq. (66), for molybdenum and HSLA-65 at high strain rates with varying tem-
which is a nonlinear ordinary differential equation for 𝛾𝑝 . A second order perature. These are shown in Figs. 2 and 3. Fig. 2(b), whilst providing
implicit integration scheme is used to integrate the flow rule from which adequate evidence of a good correspondence of the oscillation patterns
the stress–strain response for molybdenum and HSLA-65 is recovered. in the predicted with 𝑙𝑚 ≠ 0 and experimental response of molybdenum,
In high temperature applications, a frequently used metal is molyb- is perhaps also indicative of a slightly degraded performance of our
denum (𝜌0 = 10 200 kg∕m3 ) thanks to its ability to maintain reason- model for lower temperature. As anticipated and reported in Fig. 2(a),
ably high strength even at high temperatures. Melting temperature of the predicted oscillations disappear for 𝑙𝑚 = 0. Indeed, for HSLA-65, the
molybdenum is as high as 2885 K which makes it useful even in non- prediction 𝑙𝑚 ≠ 0 versus experiment match-up appears to be still better;
oxidizing conditions (above 1273 K). HSLA-65 (𝜌0 = 7850 kg∕m3 ) is see Fig. 3(b) and (a), the latter showing the predicted solutions with
extensively used in naval surface vessels and submarines owing to its 𝑙𝑚 = 0.
higher strength, better impact toughness and easier weldability. Mate- Prior to concluding this section, a word about the discrepancies in the
rial parameters for molybdenum and HSLA-65 used for our simulations prediction–experiment match-up especially at low temperatures. In such
are given in Table 1. Nemat-Nasser et al. [33,34] have carried out uni- cases, neglect of the heat flux consequent upon the assumed adiabatic
axial compression tests on molybdenum and HSLA-65 at high strain condition may no longer hold. This is also clearly suggestive of a future
rates and temperatures and these data are used here for assessing the possibility to improve upon the current model.
predictive capability of the proposed model.
We investigate the source of oscillations in the continuum stress– 5. Conclusions
strain response at high strain rates. For a given micro-inertial length
scale 𝑙𝑚 , stress–strain response is computed for different strain rate In an effort to incorporate the effect of dislocations acceleration in
loading. Fig. 1 shows the stress–strain response for HSLA-65 and molyb- the continuum response of isotropic polycrystalline solids under high
denum for different strain rates. The graphs are plotted with 𝑙𝑚 = 0 as strain rate loading, we have presented a micro-inertia driven dynamic
well as with non-zero 𝑙𝑚 . As shown in Fig. 1, while non-zero 𝑙𝑚 does flow rule in the framework of rational thermodynamics. The dynamic
not make much of a difference in the response for low strain rates, it flow rule is in the form of a second order partial differential equation,
does bring in the additional feature of oscillations in the stress–strain solution to which gives the evolution of the equivalent plastic strain.
response at high strain rates. This is proof positive of the significance Towards this, the equivalent plastic strain rate is expressed through the
the micro-inertia term has as a continuum representation of dislocation average dislocation velocity in the slip direction. This enables writing
inertia. This is also in line with Kosevich’s [21] observation that, though the kinetic energy due to dislocation motion in terms of the equivalent
the material response is mainly influenced by the dissipative forces plastic strain rate. We are then led to an expression for the micro-inertial
for small dislocation acceleration, dislocation inertia starts playing an length scale by an energy equivalence argument involving our proposed
important role as the dislocation acceleration is larger. micro-kinetic energy and the kinetic energy due to dislocation motion,
It is thus clear that, for small strain rate loading, our model re- as in the CDD theory. Through numerical simulations and a comparative
produces the same result as given by a flow rule without the micro- assessment of predicted solutions with the experimental evidence for
inertia, the effect micro-inertia showing up only for higher applied strain molybdenum and HSLA-65, we have explicated the significant role

16
Md.M. Rahaman et al. International Journal of Non-Linear Mechanics 95 (2017) 10–18

Fig. 2. Stress–strain response of molybdenum at strain rate 8000/s. Solid lines indicate the prediction from the proposed model; the circular marker denotes the experimental data
reproduced from [33]. (a) Comparison of stress–strain response computed using flow rule with 𝑙𝑚 = 0 m against experimental data. (b) Comparison of stress–strain response computed
using flow rule with 𝑙𝑚 = 1 × 10−3 m against experimental data.

Fig. 3. Stress–strain response of HSLA-65 at strain rate 3000/s. Solid lines indicate prediction from the proposed model, circular markers represent the experimental data reproduced
from [34]. (a) Comparison of stress–strain response computed with flow rule 𝑙𝑚 = 0 m against experimental data. (b) Comparison of stress–strain response computed with flow rule
𝑙𝑚 = 3 × 10−3 m against experimental data.

of micro-inertia in reproducing the oscillatory stress–strain responses choosing the right flow model prior to the application of micro-inertia
under high strain rates. However, the significance of the micro-inertia should in itself be a useful study.
term recedes as the applied strain rate tends lower. Thus, augmentation
with a micro-inertia term enables the flow rule to be applicable at higher Acknowledgements
strain rates and hence should qualify as a non-trivial addition that has
been overlooked so far. This work is funded by the Defence Research and Development
Although our micro-inertia based flow rule is in the set-up of Organization, Government of India, through Grant No. DRDO/0642. The
a micro-force balance in line with the work of Gurtin and his col- last author gratefully acknowledges the support by the Oscar S Wyatt
leagues [7,23,35], our identification of the constitutive function g1 as Endowed Chair in Mechanical Engineering at Texas A&M University,
the so called dynamic yield stress affords an accommodation of the College Station. The authors express their sincere thanks to Dr. Kari
proposed micro-inertia term within several other flow models, viz. Santaoja of Aalto University, Finland, for his constructive comments on
ZA, VA, mechanical threshold stress (MTS) or Preston–Tonks–Wallace the contents of the paper.
(PTW) models. Specifically, as our numerical work indicates, the micro-
inertia does a good job of replicating the oscillatory pattern in the
References
experimental stress–strain response if the flow model reproduces well
the overall flow curve minus the oscillations. A good agreement with
[1] H.M. Zbib, T.D. de la Rubia, A multiscale model of plasticity, Int. J. Plast. 18 (9)
the overall flow curve is in turn possible by choosing g1 appropriately, (2002) 1133–1163.
which is determined by the flow model itself (such as ZA, VA, PTW, MTS [2] O. Dillon, J. Kratochvil, A strain gradient theory of plasticity, Int. J. Solids Struct.
etc.). Indeed, given the material, applied strain rate and temperature, 6 (12) (1970) 1513–1533.

17
Md.M. Rahaman et al. International Journal of Non-Linear Mechanics 95 (2017) 10–18

[3] N. Fleck, J. Hutchinson, A phenomenological theory for strain gradient effects in [19] S. Nemat-Nasser, T. Okinaka, L. Ni, A physically-based constitutive model for bcc
plasticity, J. Mech. Phys. Solids 41 (12) (1993) 1825–1857. crystals with application to polycrystalline tantalum, J. Mech. Phys. Solids 46 (6)
[4] N. Fleck, J. Hutchinson, A reformulation of strain gradient plasticity, J. Mech. Phys. (1998) 1009–1038.
Solids 49 (10) (2001) 2245–2271. [20] C. Gao, L. Zhang, Constitutive modelling of plasticity of fcc metals under extremely
[5] H. Gao, Y. Huang, W. Nix, J. Hutchinson, Mechanism-based strain gradient plastici- high strain rates, Int. J. Plast. 32 (2012) 121–133.
tyi. Theory, J. Mech. Phys. Solids 47 (6) (1999) 1239–1263. [21] A. Kosevich, Dynamical theory of dislocations, Phys.-Usp. 7 (6) (1965) 837–854.
[6] H. Gao, Y. Huang, Taylor-based nonlocal theory of plasticity, Int. J. Solids Struct. [22] Z. Wang, I. Beyerlein, R. LeSar, Dislocation motion in high strain-rate deformation,
38 (15) (2001) 2615–2637. Phil. Mag. 87 (16) (2007) 2263–2279.
[7] M.E. Gurtin, A gradient theory of single-crystal viscoplasticity that accounts for [23] M.E. Gurtin, On the plasticity of single crystals: free energy, microforces, plastic-
geometrically necessary dislocations, J. Mech. Phys. Solids 50 (1) (2002) 5–32. strain gradients, J. Mech. Phys. Solids 48 (5) (2000) 989–1036.
[8] Y. Huang, S. Qu, K. Hwang, M. Li, H. Gao, A conventional theory of mechanism-based [24] T. Mura, Micromechanics of Defects in Solids, Springer Science & Business Media,
strain gradient plasticity, Int. J. Plast. 20 (4) (2004) 753–782. 2013.
[9] G.Z. Voyiadjis, D. Faghihi, Thermo-mechanical strain gradient plasticity with ener- [25] W. Cai, A. Arsenlis, C.R. Weinberger, V.V. Bulatov, A non-singular continuum theory
getic and dissipative length scales, Int. J. Plast. 30 (2012) 218–247. of dislocations, J. Mech. Phys. Solids 54 (3) (2006) 561–587.
[10] D. Jia, K. Ramesh, E. Ma, L. Lu, K. Lu, Compressive behavior of an electrodeposited [26] T. Hochrainer, M. Zaiser, P. Gumbsch, A three-dimensional continuum theory of
nanostructured copper at quasistatic and high strain rates, Scr. Mater. 45 (5) (2001) dislocation systems: kinematics and mean-field formulation, Phil. Mag. 87 (8–9)
613–620. (2007) 1261–1282.
[11] J. Campbell, A. Eleiche, M. Tsao, Strength of metals and alloys at high strains and [27] T. Hochrainer, Thermodynamically consistent continuum dislocation dynamics, J.
strain rates, in: Fundamental Aspects of Structural Alloy Design, Springer, 1977, Mech. Phys. Solids 88 (2016) 12–22.
pp. 545–563. [28] A. Yavari, A. Goriely, Riemann–Cartan geometry of nonlinear dislocation mechanics,
[12] G.R. Johnson, W.H. Cook, A constitutive model and data for metals subjected to Arch. Ration. Mech. Anal. 205 (1) (2012) 59–118.
large strains, high strain rates and high temperatures, in: Proceedings of the 7th [29] J.N. Reddy, An Introduction to Continuum Mechanics with Applications, second ed.,
International Symposium on Ballistics, Vol. 21, 1983, The Hague, The Netherlands, Cambridge University Press, 2013.
1983, pp. 541–547. [30] P. Naghdi, A. Srinivasa, A dynamical theory of structured solids. I basic develop-
[13] D.J. Bammann, An internal variable model of viscoplasticity, Internat. J. Engrg. Sci. ments, Philos. Trans. R. Soc. Lond. Ser. A Math. Phys. Eng. Sci. 345 (1677) (1993)
22 (8–10) (1984) 1041–1053. 425–458.
[14] F.J. Zerilli, R.W. Armstrong, Dislocation-mechanics-based constitutive relations for [31] B.D. Coleman, W. Noll, The thermodynamics of elastic materials with heat conduc-
material dynamics calculations, J. Appl. Phys. 61 (5) (1987) 1816–1825. tion and viscosity, Arch. Ration. Mech. Anal. 13 (1) (1963) 167–178.
[15] R. Liang, A.S. Khan, A critical review of experimental results and constitutive models [32] W. Wang, L. Sluys, R. De Borst, Viscoplasticity for instabilities due to strain softening
for BCC and FCC metals over a wide range of strain rates and temperatures, Int. J. and strain-rate softening, Internat. J. Numer. Methods Engrg. 40 (20) (1997) 3839–
Plast. 15 (9) (1999) 963–980. 3864.
[16] G.Z. Voyiadjis, F.H. Abed, Microstructural based models for bcc and fcc metals with [33] S. Nemat-Nasser, W. Guo, M. Liu, Experimentally-based micromechanical modeling
temperature and strain rate dependency, Mech. Mater. 37 (2) (2005) 355–378. of dynamic response of molybdenum, Scr. Mater. 40 (7) (1999) 859–872.
[17] J.-B. Kim, H. Shin, Comparison of plasticity models for tantalum and a modification [34] S. Nemat-Nasser, W.-G. Guo, Thermomechanical response of HSLA-65 steel plates:
of the PTW model for wide ranges of strain, strain rate, and temperature, Int. J. experiments and modeling, Mech. Mater. 37 (2) (2005) 379–405.
Impact Eng. 36 (5) (2009) 746–753. [35] M.E. Gurtin, On a framework for small-deformation viscoplasticity: free energy,
[18] P. Follansbee, U. Kocks, A constitutive description of the deformation of copper microforces, strain gradients, Int. J. Plast. 19 (1) (2003) 47–90.
based on the use of the mechanical threshold stress as an internal state variable,
Acta Metall. 36 (1) (1988) 81–93.

18

You might also like