2nd Reading

January 14, 2015 12:56 WSPC/245-JMM 1350011

Journal of Multiscale Modelling

Vol. 5, No. 3 (2013) 1350011 (14 pages)

c Imperial College Press
DOI: 10.1142/S175697371350011X

Finite Element Modeling to Simulate

the Elasto-Plastic Behavior of Polycrystalline in 718

E. A. Bonifaz
Universidad San Francisco de Quito
Casilla Postal: 17-12-841
by WEIZMANN INSTITUTE OF SCIENCE on 02/26/16. For personal use only.
J. Multiscale Modelling 2013.05. Downloaded from www.worldscientific.com

Quito-Ecuador, Ecuador
ebonifaz@usfq.edu.ec

Published 19 January 2015

A 3D strain gradient plasticity finite element model was developed to simulate the
elasto-plastic behavior of polycrystalline IN 718 alloys. The proposed model constructed
in the basis of the so-called Kocks-Mecking model is used to determine the influence of
microstructure attributes on the inelastic stress-strain distribution. Representative Vol-
ume Elements (RVEs) of different edge size but similar grain morphology and affordable
computational meshes were tested to investigate the link between micro and macro
variables of deformation and stress. The virtual specimens subjected to continuous
monotonic straining loading conditions were constrained with random periodic boundary
conditions. The difference in crystallographic orientation (which evolves in the process
of straining) and the incompatibility of deformation between neighboring grains were
accounted by the introduction of averaged Taylor factors and the evolution of geometri-
cally necessary dislocation density. The effect of plastic deformation gradients imposed
by the microstructure is clearly observed. Results demonstrate a strong dependence of
flow stress and plastic strain on phase type and grain size. A main strategy for consti-
tutive modeling of individual bulk grains is presented. The influence of the grain size
on the aggregate response, in terms of local stress variations and aggregate elastic mod-
uli was analyzed. It was observed that the elastic modulus in the bulk material is not
dependent on grain size.

Keywords: Geometrically necessary dislocations; representative volume elements; grain

size effect; plastic strain gradients; slip systems.

1. Introduction
Polycrystalline materials (metals, alloys or ceramics) are commonly used in engi-
neering applications. Their microstructure, at the grain scale, is characterized by
the grain morphology, size distribution, anisotropy and crystallographic orienta-
tion, by the presence of flaws and porosity and by physical and chemical properties
of the intergranular interfaces,1 which also have a direct effect on the initiation
and evolution of damage. The simulation of polycrystalline superalloys with a

1350011-1
 2nd Reading
January 14, 2015 12:56 WSPC/245-JMM 1350011

E. A. Bonifaz

microstructural basis is limited to few studies. Their analysis has been until recently
restricted to two dimensional cases, due to high computational requirements. In the
last decade, however, the wider affordability of increased computational capability
has promoted the development of fully three-dimensional models.2
The link between microstructure and material macroscopic properties, the
structure-property relationship, is technologically interesting as it may provide valu-
able information for the design of enhanced materials.3–7 The analysis of material
microstructures requires the generation of reliable micro-morphologies and afford-
able computational meshes as well as the description of the mechanical behavior of
the elementary constituents and their interactions.2 The problem of the generation
of a suitable virtual microstructure, morphology and mesh, is particularly relevant,
by WEIZMANN INSTITUTE OF SCIENCE on 02/26/16. For personal use only.

especially when the analysis of a relevant number of grains in the three-dimensional

J. Multiscale Modelling 2013.05. Downloaded from www.worldscientific.com

case is of interest.2 Analyses of this kind are of interest, for example, for predicting
the strong dependence of flow stress and plastic strain on phase type and grain size.
In order to incorporate grain size effects, the mechanical plastic model should be
length dependent.
Polycrystalline aggregates are idealized as simple three-dimensional arrange-
ments of grains, where many elements per grain are used to represent nonuniform
deformations within individual grains, seen as domains separated by boundaries
of high misorientation. To investigate the link between micro and macro variables
of deformation and stress, the polycrystal is represented by a finite element mesh
loaded by periodic boundary conditions. Component level phenomenological mod-
els may not always be able to predict complex materials behaviors, especially if
damage initiation and evolution are of concern.2 It is today widely recognized
that these aspects may be better understood if the features of the material
microstructure are considered and brought into the modeling framework.2 It is
for this reason that three-dimensional grain-scale mechanical modeling of poly-
crystalline materials may provide valuable information for the design of enhanced
materials. In this work, a 3D strain gradient plasticity finite element model was
developed to simulate the elasto-plastic behavior of polycrystalline IN 718 alloys.
Three virtual microstructures (RVEs) of different edge size (L0 = 100 um, 10 um
and 1 um) but similar grain morphology and affordable computational meshes were
tested to study the strong dependence of flow stress and plastic strain on phase
type and grain size.

2. Basic Equations
To include first order effects (grain size, γ
precipitate size distribution, γ
precipi-
tate volume fraction and the evolution of dislocation densities) on the stress–strain
response of a polycrystalline IN 100 at 650◦ C, physically based hardening laws such
as Eq. (1) were employed in Shenoy et al.8 model,

α
κα α
λ = κ0,λ + αt µmix b ρλ . (1)

1350011-2
 2nd Reading
January 14, 2015 12:56 WSPC/245-JMM 1350011

Finite Element Modeling to Simulate the Elasto-Plastic Behavior of Polycrystalline in 718

Here, κα α
λ is the threshold slip resistance on slip system α for λ = oct, cub; ρλ is
the dislocation density, b is the Burgers’ vector, αt is a statistical coefficient that
accounts for the deviation from the regular spatial arrangements of the dislocation
pollution



αt = (0.1 − 0.68fp1 + 1.1fp2 ) (2)




fp1 , fp2 and fp3 are the normalized precipitate volume fractions of the primary fp1 ,
secondary fp2 , and tertiary fp3 γ
precipitates respectively,
fp1 fp2 fp3
fp1
= , fp2
= , fp3
= (3)
fp1 + fm fp2 + fm fp3 + fm
by WEIZMANN INSTITUTE OF SCIENCE on 02/26/16. For personal use only.

fm is the volume fraction of the matrix phase, µmix is the volume averaged shear
J. Multiscale Modelling 2013.05. Downloaded from www.worldscientific.com

modulus

µmix = (fp1 + fp2 + fp3 )µγ
+ fm µm . (4)

Here, uγ
is shear moduli for the precipitate and µm is shear moduli for the matrix
phase. The initial critical resolved shear stress (CRSS) κα 0,λ is microstructure sen-
sitive and its strength is determined for octahedral and cube slip systems based
on the volume fractions of the precipitates.9 The anti-phase boundary and Hall-
Petch relation that takes into account the increase in the CRSS with decrease in
grain size was also included in this term.9 The Shenoy’s et al.8 model implies that
the strength of the material is determined by the presence of precipitates and by
dislocation-dislocation interactions. The microstructure-dependence of the stress–
strain response is completely embedded in the evolution equations for the internal
state variables (ISV’s): the dislocation density, ρα and the backstress variable χα .
It is also mentioned that the use of the Taylor relation in the second term of Eq. (1)
is warranted by the strongly planar nature of slip of IN 100 under cyclic loading at
the temperature of 650◦ C, with strength dominated by particle shearing.9
Initial yield stress varies from sample to sample depending on, among several
factors, the relation between the crystal lattice to the loading axis (i.e., orientation).
The applied stress resolved along the slip direction on the slip plane (to give a shear
stress τ α ) initiates and controls the extent of plastic deformation. Yield begins on a
given slip system when the shear stress on this system reaches a critical value, called
the critical resolved shear stress (CRSS) τcα , independent of the tensile stress or any
other normal stress on the lattice plane. The magnitude of the yield stress depends
on the density and arrangement of obstacles to dislocation flow such as precipitates.
If the critical condition is reached simultaneously in several slip systems into each
single grain, the slip is called multiple. A second-rank tensor (mij = bi nj ) can be
associated with each slip system, formed from the outer product of slip direction
and normal. The Taylor factor M = m1ij which evolves in the process of straining
is commonly used to account for polycrystallinity of the material. In multiple slip,
we can form the resolved shear stress as the sum over all the contributions from
the external stress components.

1350011-3
 2nd Reading
January 14, 2015 12:56 WSPC/245-JMM 1350011

E. A. Bonifaz

The resolved shear stress on a slip system is then given by the inner product of
the Schmid and the stress tensors:
τ α = mij σij (5)
or, τ = Eσ. (6)
Note that one is entitled to invert the matrix E, provided that its determinant is
nonzero, which it will only be true if the slip systems chosen are linearly indepen-
dent.
σ = E −1 τ . (7)
The direct form of the stress equation means that, if we assume a fixed critical
by WEIZMANN INSTITUTE OF SCIENCE on 02/26/16. For personal use only.
J. Multiscale Modelling 2013.05. Downloaded from www.worldscientific.com

resolved shear stress (τ = τc ), then we can compute all the possible multislip stress
ms
states σij , based on the set of linearly independent combinations of slip.
ms
σij = E −1 τc . (8)
For multiaxial stress states, one may use an effective multislip stress q ms , to account
for all the slip systems contributions, so that
q ms
M̄ = . (9)
τc
Here, q ms includes the effect of the active slip systems. In other words, given a set of
grains (orientations) comprising a polycrystal, and assuming six active slip systems
per grain, one can calculate an average Taylor factor M̄ (for each grain) using the
Von Mises effective multislip stress expression. If more than six active slip systems
are considered, then another effective multislip stress expression is necessary.
In the present work, to describe the work-hardening process in polycrystalline
nickel base alloys, a dislocation-density-related model constructed in the basis of
the so-called Kocks-Mecking model (Voce type constitutive equation) is proposed
(e.g., Kocks,10 Estrin and Mecking,11 Narutani and Takamura12 ). The benefit of
dislocation modeling in which the dislocation density plays the role of an internal
variable representing the microstructural state of the material, is the relatively small
number of adjustable parameters. To a first approximation, single slip plasticity
will be considered, that is, only one slip system reach the critical shear stress τcα ,
remaining the others inactivate. It means that the grain with the largest resolved
shear stress yields first and other (less favorably oriented) grains yield later. Of
this manner, the shear stress is reduced to the active slip system, and tacitly the
influence of other stress components such as: hydrostatic stresses, stress normal to
the slip plane are excluded.
According Dunne and Petrinic,13 if there is not preferred crystallographic orien-
tation, but that the orientation changes randomly from one grain to the next, and
if the material sample contains a sufficiently large amount of grains, we can get a
reasonable physical feel that macroscale yielding of the material will be isotropic.
This is a further cornerstone of the Von Mises yield criterion. The specifics of the

1350011-4
 2nd Reading
January 14, 2015 12:56 WSPC/245-JMM 1350011

Finite Element Modeling to Simulate the Elasto-Plastic Behavior of Polycrystalline in 718

model enter through a particular form of the equation relating the equivalent plastic
strain rate and the equivalent stress; this form is referred to as the kinetic equation.
That is, if the material is rate dependent, the relationship is the uniaxial flow rate
definition of Eq. (10). For continuous monotonic straining, a one-internal variable
σ 0 (that represents the microstructural state of a material) model is sufficient. If
σ 0 is constant, the kinetic equation
 q  m1
pl
ē˙ = ε̇0 , (10)
σ0
refers to a fixed microstructure. Here, q is the Mises equivalent stress (the resolved
shear stress τ α in ref. 9), ε̇0 (T ) and m(T ) are user-defined temperature-dependent
pl
material parameters and σ 0 (ē˙ , T ) is the yield stress (the slip resistance, τcα or κα
λ
by WEIZMANN INSTITUTE OF SCIENCE on 02/26/16. For personal use only.
J. Multiscale Modelling 2013.05. Downloaded from www.worldscientific.com

in ref. 9). Integrating this relationship by the backward Euler method gives q as a
function of ēpl at the end of the increment
 q  m1
∆ēpl = ∆tε̇0 , (11)
σ0
 m 
0 ∆ēpl
q=σ . (12)
∆tε̇0
The plasticity requires that the material satisfy a uniaxial-stress plastic-strain
strain-rate relationship.14 So, the yield condition for a rate independent material is
q − σ 0 = 0. (13)
And for a rate dependent material is
 m 
∆ēpl
q − σ0 = 0. (14)
∆tε̇0
Thus, both the rate-independent model and the integrated rate-dependent model
give the general uniaxial form q = σ̄(ēpl ). The above equations define the material
behavior in any increment when plastic flow is occurring.
Next, we will examine the evolution of slip resistance. As a point of departure,
we will discuss the specific relation between flow stress and dislocation density that
is in common usage. According to the experimental observations carried out by
Narutani and Takamura12 and other investigators (Dingley and McLean,15 Meakin
and Petch,16 Mecking and Kocks,17 Estrin and Mecking,11 Bonifaz18 ), the flow
stress is proportional to the square root of dislocation density ρ irrespective of
the grain size, amount of strain and test temperature. For a coarse-grained, single-
phase material, which can be regarded as “structureless”, the flow stress at zero
temperature is set equal to
√
σ = M αGb ρ, (15)
√
σ = α̂Gb ρ. (16)
Here, M is the average Taylor factor, which evolves in the process of straining
(in what follows, M = M̄ will be considered constant for simplicity), α a scalar

1350011-5
 2nd Reading
January 14, 2015 12:56 WSPC/245-JMM 1350011

E. A. Bonifaz

coefficient of order 0.3, b is the magnitude of the Burgers vector, G is an appropriate

shear modulus and α̂ is a constant of order unity which depends, in part, on the
strength of the dislocation/dislocation interaction.17 Thermal activation may lower
this effective obstacle strength so that the flow stress at a finite temperature and
strain rate becomes
√
σ = s(ε̇, T )α̂Gb ρ, (17)
where s(ε̇, T ) is a function that goes to 1 as T → 0. From Eq. (17), it is apparent
that the flow stress is a product of a rate sensitivity term and a structure sensitive
term. The flow stress, as given by Eqs. (15) to (17), relates only to the impediment
to dislocation motion that is provided by other dislocations. In most materials,
there are other contributions to the plastic resistance. In some cases (e.g., lattice
by WEIZMANN INSTITUTE OF SCIENCE on 02/26/16. For personal use only.
J. Multiscale Modelling 2013.05. Downloaded from www.worldscientific.com

resistance, solution hardening, some grain size effects), these are additive to the
contributions discussed above.17 The flow stress dependence on a rate sensitivity
term and on a structure sensitive term (including lattice resistance and solution
hardening) is accounted for in equation (18).


σ 0 = σ0 (ε̇, T ) + α̂(ε̇, T )Gb ρT . (18)
Here, the rate dependence of σ0 may be more important than that of α̂ (or s), or it
may be negligible; the less rate sensitive term (contained into the σ0 term) is often
called an “internal stress” (Pierls or friction stress). The total dislocation density
ρT is defined by Ashby19 as the sum of geometrically necessary dislocations ρG and
statistically stored dislocations ρS . The statistically stored dislocations are accumu-
lated in pure crystals during straining and are responsible for the normal 3-stage.19
Plastic strain gradients are caused by the geometry of deformation, by local bound-
ary conditions or by the microstructure.20 These strain gradients require, for com-
patibility reasons, the presence of geometrically necessary dislocations of density ρG ,
which are introduced to accommodate the incompatibility of deformation between
neighboring grains. The step of translating from the simple dislocation equations
to a continuum formulation is not obvious. The statistically stored dislocations, ρS ,
are assumed to be dependent on the plastic strain εp , while the geometrically neces-
sary dislocations, ρG , are assumed to be dependent on strain gradient ∂εp /∂x21 or
in a linear manner with the reciprocal of the grain size.12,19 In its present state, dis-
location density-related constitutive modelling is considered mature enough to be
broadly used in finite element codes including viscoplasticity.22 In order to formu-
late the grain-size or local boundary conditions dependence of the total dislocation
density, it is necessary to derive an equation to describe the accumulation of dis-
locations during deformation, but the constitutive equation to describe the work
hardening process in polycrystalline materials has not been well established.

3. The Model
In this work, we assume that σ 0 identifies the dislocation/dislocation interaction
component of the flow stress through the evolution of the ρ term. It is related with

1350011-6
 2nd Reading
January 14, 2015 12:56 WSPC/245-JMM 1350011

Finite Element Modeling to Simulate the Elasto-Plastic Behavior of Polycrystalline in 718

the impediment to dislocation motion that is provided by other dislocations.

σ 0 = σ0 + M αGb ρS + ρG . (19)

Lattice resistance and solution hardening are accounted for the σ0 term. σ0 considers
the additive contribution from lattice resistance at zero temperature σp (Peierls or
friction stress), solution hardening stress from alloying elements σss , and strength-



ening due to microstructure (grain size) σ, i.e., σ0 = σp + σss + σ. Note the simil-
itude of Eqs. (1) and (19). However, to be comparable, the shear √ flow-stress has
to be related to √ the macroscopic tensile flow-stress through the 3 value, so, the
expression σ0 = 3κα 0,λ can be used in Eq. (19) to account the effect of octahedral
and cube slip systems based on the volume fractions of the precipitates. As the
by WEIZMANN INSTITUTE OF SCIENCE on 02/26/16. For personal use only.
J. Multiscale Modelling 2013.05. Downloaded from www.worldscientific.com

microstructure varies in the process of plastic deformation, a separate equation is

needed to describe the evolution of σ 0 .
dσ 0
= f (σ 0 ; ε̇p ; T ). (20)
dεp
For a coarse-grained (or monocrystalline) single-phase material which can be
regarded as “structureless”, the evolution equation in the form of Eq. (21) describes
materials where dislocation storage during deformation is controlled by the total
dislocation density ρ (Estrin,23 Narutani and Takamura12 ).
dρ √
= M (K1 ρ − K2 ρ). (21)
dε
Here, K1 and K2 are constants. Because Ashby’s total dislocation density ρT def-
inition assumes that the geometrically necessary dislocations, ρG , have no direct
influence on the accumulation of the statistically stored dislocations ρS , the vari-
ation of ρS in the process of plastic deformation can be described by the same
evolution equation:
dρs √
= M (K1 ρs − K2 ρs ). (22)
dε
After integration of equation (22) with appropriate limits (i.e., ρS = ρ0 at ε = 0,
and ρS = ρS at ε = ε̄), expression (23) is derived
 2
K1  −M K2 ε̄

−M K2 ε̄
ρS = 1−e 2 + ρ0 e 2 . (23)
K2

Here, ρ0 is the initial dislocation density, ε̄ is the effective strain, K1 and K2 char-
acterize the processes of dislocation storage and concurrent dislocation annihilation
by dynamic recovery, respectively.23 The process of dislocation storage is athermal,
so that K1 is a constant. By contrast, the coefficient K2 represents a thermally
activated process of dynamic recovery by dislocation cross-slip (low temperature
case) or dislocation climb (high temperature case). The boundary between the two
temperature regimes lies at approximately two thirds of the melting temperature.

1350011-7
 2nd Reading
January 14, 2015 12:56 WSPC/245-JMM 1350011

E. A. Bonifaz

In both cases the strain rate and temperature dependence of K2 can be expressed
as23 :

p −1/n
ε̇
K2 = K20 , (24)
ε̇∗0
where K20 is a constant. The temperature dependence is contained either in n (in
the low-temperature case when n is inversely proportional to temperature T , while
ε̇∗0 can be considered constant) or in ε̇∗0 (in the high-temperature case when it is
given by an Arrhenius-type equations, n being a constant ranging from 3 to 5).23
In the present work, the temperature and the strain rate dependence in K2 are not
considered, so, the term associated with the dynamic recovery is assumed to be a
constant.
by WEIZMANN INSTITUTE OF SCIENCE on 02/26/16. For personal use only.
J. Multiscale Modelling 2013.05. Downloaded from www.worldscientific.com

As mentioned above, as a result of the tensile straining, inhomogeneous defor-

mation occurs adjacent to grain boundaries, leading to a concept of a harder ‘man-
tle’ close to the boundary and a softer ‘core’ in the grain interior as proposed by
Thompson et al. (1973).24 The difference in crystallographic orientation between
neighboring grains can be corrected by introducing geometrically necessary disloca-
tions ρG , which are introduced to accommodate the incompatibility of deformation
between grains.19 To quantify the geometrically necessary dislocations density ρG ,
Fleck et al.20 proposed the following equation:
χeq
ρG = C , (25)
b
where, C is a constant ranging from 1 to 2, χeq represents the magnitude of the cur-
vature tensor χ used as the scalar measure of the density of geometrically necessary
dislocations ρG , and b the Burgers’ vector.

2
χeq = χij χij .
3
Here, χni = enkj εij,k , enkj is the alternating tensor, and εij,k is the strain gradient
tensor.
Of this manner, the elasto-plastic behavior of polycrystalline IN 718 alloys is
defined by using equations (13), (19), (23) and (25). Note that the microstructure
evolution during the macroscopic straining process is represented by the evolution
of the total dislocation density ρT , and that the grain size effect is captured by the
ρG term.

4. The Finite Element Model

Following the recommendations of Ref. 9, a cubic box with an edge size (L0 ) is used
to represent the polycrystalline aggregate with
dgr (ngr )1/3
L0 = , (26)
0.7
where dgr  is the average grain size, and ngr is the number of grains. The random
polycrystalline aggregate generated with the software DREAM 3D25 consists of 12

1350011-8
 2nd Reading
January 14, 2015 12:56 WSPC/245-JMM 1350011

Finite Element Modeling to Simulate the Elasto-Plastic Behavior of Polycrystalline in 718

by WEIZMANN INSTITUTE OF SCIENCE on 02/26/16. For personal use only.
J. Multiscale Modelling 2013.05. Downloaded from www.worldscientific.com

Fig. 1. A dual-phase representative volume element constrained with 3D periodic boundary condi-
tions. The continuous monotonic straining loading is applied in the top face. The length dependent
term L0 is the cubic box edge size.

grains distributed in two phases: 76% matrix phase–24% precipitate phase. Fig. 1
shows a dual-phase representative volume element constrained with 3-D periodic
boundary conditions.
The calculations were performed with ABAQUS r code using an 8 node C3D8
representative finite element mesh. Virtual specimens with grain size dimensions
in the range of 0.3 to 30 um were used in the finite element simulations. The sin-
gle slip plasticity assumption was considered in all the aggregate. To observe the
strong dependence of flow stress on CRSS and/or grain size (L0 ) itself, the follow-
ing literature reported values were used in the simulations: Nickel Young’s modu-
lus E = 170 GPa, Poisson’s modulus ν = 0.3, K1 = 9.47 × 105 cm−1 , K2 = 6.12,
M = 2.74 (for the matrix), M = 2.83 (for the precipitate), CRSS κα0,oct = 85.1 MPa
(for the matrix) and, κα0,cub = 170.2 MPa (for the precipitate).

5. Results and Discussions

Figure 2 shows the Misses stress–strain curves for the two selected elements (E: 3
and E: 4). Element 3 belongs to the precipitate phase, and element 4 belongs to
the matrix phase. Results demonstrate a strong dependence of flow stress on phase
type. It is clearly observed the influence of the yield stress (different CRSS for each
phase) on the mechanical behavior.
The grain size effect is included in the formulation through the length dependent
term L0 . The macroscopic stress σ and strain ε in the traction direction are deduced
from the resulting force on the face where the displacement is applied (for the stress)

1350011-9
 2nd Reading
January 14, 2015 12:56 WSPC/245-JMM 1350011

E. A. Bonifaz
by WEIZMANN INSTITUTE OF SCIENCE on 02/26/16. For personal use only.
J. Multiscale Modelling 2013.05. Downloaded from www.worldscientific.com

Fig. 2. Misses stress–strain curves for the two selected elements. Element 3 belongs to the precip-
itate phase, and element 4 belongs to the matrix phase.

Fig. 3. Macroscopic curves (stress σ versus strain ε) for the three different reported virtual speci-
men initial lengths after an applied displacement of 0.4 L0 . Note that the elastic modulus in the
bulk material is not dependent on grain size.

u2
σ = A1 A σ22 dA and from the displacement itself (for the strain) ε = ln(1 + L 0
).
Here, u2 is the applied displacement and L0 is the virtual specimen initial length.
Figure 3 shows the strong dependence of flow stress on grain size (L0 ). At the
polycrystalline level, all homogenized grains are considered as local contributors to

1350011-10
 2nd Reading
January 14, 2015 12:56 WSPC/245-JMM 1350011

Finite Element Modeling to Simulate the Elasto-Plastic Behavior of Polycrystalline in 718

by WEIZMANN INSTITUTE OF SCIENCE on 02/26/16. For personal use only.
J. Multiscale Modelling 2013.05. Downloaded from www.worldscientific.com

Fig. 4. Mises stress versus strain curves for the two selected elements (3 and 4) at three different
mesh sizes (different L0 values) strained at 0.4 L0 .

the macroscopic response of polycrystalline structure. Note that the elastic modulus
in the bulk material is not dependent on grain size (Fig. 3).
Figure 4 shows the Misses stress–strain curves for the two selected elements
tested at three different mesh sizes (different L0 values). In all the situations, after

Fig. 5. Geometrically Necessary Dislocation Density (GNDD) versus true strain curves for the
reported L0 values after a 0.4 L0 applied displacement.

1350011-11
 2nd Reading
January 14, 2015 12:56 WSPC/245-JMM 1350011

E. A. Bonifaz
by WEIZMANN INSTITUTE OF SCIENCE on 02/26/16. For personal use only.
J. Multiscale Modelling 2013.05. Downloaded from www.worldscientific.com

Fig. 6. Stress (Mises and S22) and displacement contours for the three tested specimen initial
lengths (100 um, 10 um and 1 um) after an applied displacement of 0.4 L0 .

1350011-12
 2nd Reading
January 14, 2015 12:56 WSPC/245-JMM 1350011

Finite Element Modeling to Simulate the Elasto-Plastic Behavior of Polycrystalline in 718

an applied displacement of 0.4 L0 , element 3 that belongs to the precipitate hard

phase, deform to a true strain value of approximately 0.24.
On the other hand, after the same applied displacement, element 4 that belongs
to the matrix soft phase, deform distinctly in the three studied situations. The
deformation of the matrix phase (represented by the element 4) reaches true strain
values of 0.27, 0.33 and 0.39 when calculated in specimen’s lengths of 1 um, 10 um
and 100 um respectively. It can be inferred that for identical simulation conditions,
the bigger the matrix grain size, the bigger the final local plastic strain.
Figure 5 shows the effect of the length dependent term L0 in the generation of
geometrically necessary dislocation density GNDD. The contribution of the GNDD
(ρG term in equations 25 and 19) included into the ABAQUS-UMAT finite element
by WEIZMANN INSTITUTE OF SCIENCE on 02/26/16. For personal use only.

analysis is very important in the inelastic stress–strain distribution as observed in

J. Multiscale Modelling 2013.05. Downloaded from www.worldscientific.com

Figs. 2, 3 and 4.
Figure 6 shows the stresses (Mises and S22) and displacements contours for
the three tested cubic mesh sizes (100 um, 10 um and 1 um) after an applied dis-
placement of 0.4 L0 . Note the heterogeneity and great difference in stress contour
values. This important difference reveals that for identical simulation conditions, as
the grain size decreases, the maximum stress value (and corresponding difference
in stress contour values) increases.

6. Conclusions
At the polycrystal level, all homogenized grains are considered as local contribu-
tors to the macroscopic response of polycrystalline structure. The polycrystalline
mechanical behavior of nickel alloys was analyzed from a composite point of view,
that is, the mechanical behavior of the aggregate (polycristal with different ori-
ented grains) is represented by the individual contribution of a single phase com-
posed of various single oriented crystals distributed in a randomly manner. The
microstructural characteristics of the alloy were captured on the stress–strain
response of the aggregate. Three virtual microstructures (RVEs) of different edge
size (L0 = 100 um, 10 um and 1 um) but similar grain morphology and affordable
computational meshes were tested to study the strong dependence of flow stress
and plastic strain on phase type and grain size. The difference in crystallographic
orientation (which evolves in the process of straining) and the incompatibility of
deformation between neighboring grains were accounted by the introduction of aver-
aged Taylor factors and geometrically necessary dislocations. The effect of plastic
deformation gradients imposed by the microstructure is clearly observed. Results
demonstrate a strong dependence of flow stress and plastic strain on phase type
and grain size. A main strategy for constitutive modeling of individual bulk grains
is presented. The influence of the grain size on the aggregate response, in terms of
local stress variations and aggregate elastic moduli was analyzed. It was observed
that the elastic modulus in the bulk material is not dependent on grain size.

1350011-13
 2nd Reading
January 14, 2015 12:56 WSPC/245-JMM 1350011

E. A. Bonifaz

References
1. B. Adams and T. Olson, The mesostructure — Properties linkage in polycrystals,
Progress in Materials Science 43 (1998) 1–88.
2. I. Benedetti and F. Barbe, Modeling polycrystalline materials: An overview of three-
dimensional grain-scale mechanical models, J. Multiscale Modeling, 5 (2013) 1350002
(51 pages).
3. Z. Hashin, Analysis of composite materials — A survey, ASME J. Appl. Mech. 50
(1983) 481–505.
4. T. Mura, Micromechanics of Defects in Solids, Mechanics of Elastic and Inelastic
Solids (Kluwer Academic Publishers, Dordrecht, The Netherlands, 1987).
5. S. Nemat-Nasser and M. Hori, Micromechanics: Overall Properties of Heterogeneous
Materials, 2nd rev. edn. (North-Holland, Elsevier, Amsterdam, The Netherlands,
1999).
by WEIZMANN INSTITUTE OF SCIENCE on 02/26/16. For personal use only.
J. Multiscale Modelling 2013.05. Downloaded from www.worldscientific.com

6. A. Needleman, Computational mechanics at the mesoscale, Acta Materialia 48 (2000)

105–124.
7. T. Watanabe and S. Tsurekawa, The control of brittleness and development of desir-
able mechanical properties in polycrystalline systems by grain boundary engineering,
Acta Materialia 47 (1999) 4171–4185.
8. M. Shenoy, Y. Tjiptowidjojo and D. McDowell, Microstructure-sensitive modeling of
polycrystalline, In 100. Int. J. Plasticity, 24 (2008) 1694–1730.
9. M. Shenoy, J. Zhang and D. L. McDowell, Estimating fatigue sensitivity to polycrys-
talline Ni-base superalloy microstructures using a computational approach, Fatigue
and Fracture of Engineering Materials and Structures, 30 (2007) 889–904.
10. U. F. Kocks, Laws for work-hardening and low-temperature creep. J. Eng. Mater.
Technol. 98 (1976) 76–85.
11. Y. Estrin and H. Mecking, A unified phenomenological description of work hardening
and creep based on one-parameter models. Acta Metall. 32 (1984) 57–70.
12. T. Narutani and J. Takamura, Acta Metall. Mater. 39 (1991) 2037.
13. F. Dunne and N. Petrinic. Introduction to Computational Plasticity (Oxford Univer-
sity Press, 2005).
14. ABAQUS documentation manual, V. 6.12. Dassault Systèmes Simulia Corp., Provi-
dence, RI, USA.
15. D. J. Dingley and D. McLean, Components of the flow stress of iron. Acta Metall. 15
(1967) 885.
16. J. D. Meakin and N. J. Petch, Phil. Mag. 30 (1974) 1149.
17. H. Mecking and U. F. Kocks, Acta Metallurgica 29 (1981) 1865.
18. E. A. Bonifaz, Modelling of mechanical properties and local deformation of high
strength multi-phase steels. ECSC Steel RTD Annual Report, March 2000.
19. M. F. Ashby, Phil. Mag. 21 (1970) 399.
20. N. A. Fleck, G. M. Muller, M. F. Ashby and J. W. Hutchinson, Acta Metall. Mater.
42 (1994) 475.
21. J. W. Hutchinson, Int. J. Solids Structures 37 (2000) 225.
22. Y. Estrin, In: Unified Constitutive Laws of Plastic Deformation, pp. 69–106, Krausz
A. S. and Krausz K. (Eds.) (Academic Press, New York, 1996).
23. Y. Estrin, J. Mat. Processing Tech. 33 (1998) 80–81.
24. A. W. Thompson, M. I. Baskes and W. F. Flanagan, The dependence of polycrystal
work hardening on grain size. Acta Metall. 21 (1973) 1017–1028.
25. Dream 3D Software. http://www.immijournal.com/content/3/1/5/abstract.

1350011-14