International Journal of Plasticity 26 (2010) 617–633

Contents lists available at ScienceDirect

International Journal of Plasticity

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

The effect of microstructural representation on simulations of microplastic

ratcheting
Rémi Dingreville a,*, Corbett C. Battaile a, Luke N. Brewer a, Elizabeth A. Holm a, Brad L. Boyce b
a
Computational Materials Science and Engineering Department, Sandia National Laboratories, P.O. Box 5800, Albuquerque, NM 87185-1411, USA
b
Materials Science and Engineering Center, Sandia National Laboratories, P.O. Box 5800, Albuquerque, NM 87185-1411, USA

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

Article history: This paper assesses the sensitivity of cyclic plasticity to microstructure morphology by
Received 31 March 2009 examining and comparing the microplastic ratcheting behavior of different idealized
Received in final revised form 6 August 2009 microstructures (square, hexagonal, tessellated, and digitized from experimental data).
Available online 1 October 2009
This analysis demonstrates the sensitivity of computational accuracy to the various
approximations in microstructural representation. The methodology used to perform this
Keywords: study relies on a coupling between microstructural characterization, mechanical testing
Finite element method
and numerical simulations to investigate the influence of the microstructure on the purely
Crystal plasticity
Fatigue
tensile uniaxial microplastic ratcheting behavior of pure nickel polycrystals. The morphol-
Ratcheting ogy and deformation behavior of polycrystals were characterized using electron back-scat-
Microstructure ter diffraction (EBSD), while a finite element model (FEM) of crystal plasticity was used in a
computational framework. The predicted cyclic behavior is compared to experimental
results both at the macroscopic and microstructural scales. The stress–strain response is
less sensitive to the details of the microstructural representation than might be expected
with all representations displaying similar macroscopic constitutive response. However,
the details of the plastic strain distribution at the microstructural scale and the related esti-
mations of damage mechanics vary substantially from one microstructural representation
to another.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Ratcheting, otherwise known as cyclic creep, refers to the cycle-dependent accumulation of plastic strain that is com-
monly observed in many metals and alloys. Ratcheting is often associated with an asymmetry in the yield surface (i.e. kine-
matic hardening or Bauschinger effect) and is typically observed under reversed loading conditions, e.g. with a negative
stress ratio R = rmin/rmax < 0, although it can also be observed under purely tensile loading at stress ratios R > 0 (Besson
et al., 2001; Chen et al., 2008). Unlike conventional ratcheting behavior, some microplastic ratcheting can occur well below
the material’s nominal yield strength or without a reversal in the macroscopic stress state, such as during purely tensile load-
ing where R > 0. Materials with a low elastic limit (yield strength) are particularly susceptible to this phenomenon. Accumu-
lated plastic deformations can lead to non-negligible permanent strain and eventually damage by crack nucleation.
Therefore, predicting and understanding ratcheting are important to the design of mechanical devices subjected to cyclic
loading. Nevertheless, since ratcheting is the progressive accumulation of plastic deformation cycle after cycle, it is difficult
both to simulate accurately and to link to the material’s microstructure. Numerous and various investigations have been con-
ducted over the past 20 years in an attempt to address this phenomenon.

* Corresponding author. Tel.: +1 718 260 3461.

E-mail address: rdingre@poly.edu (R. Dingreville).

0749-6419/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijplas.2009.09.004
618 R. Dingreville et al. / International Journal of Plasticity 26 (2010) 617–633

Many experimental studies (Ahmed and Wilkinson, 2001; Chaboche and Nouailhas, 1989; Chen et al., 2006; Kulkarni
et al., 2004; Yaguchi and Takahashi, 2005; Yoshida et al., 1989) describe the investigation of ratcheting behavior and the links
between the observations of fatigue life to various physical phenomena. Most of these associate the accumulation of plastic
strain to parameters such as cyclic frequency, hold time, cyclic loading history, etc. Despite the fact that the microstructure is
intrinsically influencing these observations, it is difficult to assess its role in a way that decouples it from the aforementioned
mechanism. Complementary to experimental work, many models have been proposed to describe ratcheting. They can be
classified into two groups: macroscopic phenomenological models and crystallographic models.
Most of the phenomenological models either couple or decouple the plastic modulus with some non-linear kinematic
hardening rule through the yield surface consistency condition (Abdel-Karim, 2009; Abdel-Karim and Ohno, 2000; Arm-
strong and Frederick, 1966; Bari and Hassan, 2000, 2001; Chaboche, 1991; Chen and Jiao, 2004; Chen et al., 2005; Daf-
alias and Popov, 1975; McDowell, 1995; Mroz, 1967; Ohno and Wang, 1993a,b; Zhang and Jiang, 2008). For example, in
a recent paper by Abdel-Karim (2009), the author proposed modified rules based on the Ohno–Wang kinematic harden-
ing rule in which the accumulated plastic strain increment contributes to the dynamics recovery term. These modified
kinematic hardening rules simulate multiaxial ratcheting with good agreement with experimental measurements. As an-
other example, Zhang and Jiang (2008) compared cyclic plasticity experiments with a model using a yield surface com-
bined with the Armstrong–Frederick type kinematic hardening rule. This constitutive model was in good agreement with
the experiments to describe the transient behavior during cyclic hardening and non-proportional hardening of polycrys-
talline copper. Although the macroscopic ratcheting behavior predicted by these types of models validates well with
experimental measurements, they generally ignore the microstructural details of the material during deformation. These
models consider a Representative Volume Element of material subjected to a uniform macroscopic state. This continuum
approach neglects the local heterogeneities of stresses and strains within the microstructure and reflects the inhomoge-
neity of the material state through average quantities cast in internal state variables. A second kind of phenomenological
models is based on a multi-mechanism and multi-criteria method (Cailletaud and Sai, 1995; Hassan et al., 2008; Sai and
Cailletaud, 2007; Taleb et al., 2006) by representing the different physical mechanisms involved through different inelas-
tic strains.
The crystallographic models, on the other hand, introduce internal state variables to represent and characterize the
microstructure (Cailletaud and Sai, 1995; Kwofie, 2005; Sai and Cailletaud, 2007; Sinha and Ghosh, 2006; Xie et al.,
2004). Micromechanical models constitute one class of approach accounting for the complex local state of stress and
strain between various scales (Feaugas and Gaudin, 2004; Johansson et al., 2005; Kang and Kan, 2007). Feaugas and Gau-
din (2004), for example, showed the dependence of ratcheting on the dislocation substructure within the microstructure.
Grain-based approaches represent an alternative to the micromechanical models. Sinha and Ghosh (2006), for example,
used a crystal plasticity based finite element model (FEM) to predict cyclic ratcheting of HSLA steels. Their model incor-
porates crystallographic orientation from microstructural information, while each element of the FE model represents
one grain of the polycrystalline aggregate. Similarly, Cailletaud and Sai (2008) performed a systematic study using a
two scale approach combining the constitutive equations for single crystal behavior with a scale transition rule crystal
to show the effect of intergranular and intragranular hardening on the ratcheting behavior. However, all these models
lack key details at the microstructural level such as the grains morphology, topology and size distribution. For more de-
tailed surveys of these topics, the reader is referred to the reviews by Kang (2008), Ohno (1997, 1990) and by Chaboche
(2008).
Whereas ratcheting is typically thought of as a continuum phenomenon (Abdel-Karim, 2009; Armstrong and Frederick,
1966; Bari and Hassan, 2000, 2001; Cailletaud and Sai, 1995; Chaboche, 1991; Chaboche and Nouailhas, 1989; Dafalias
and Popov, 1975; Kwofie, 2005; McDowell, 1995; Mroz, 1967; Ohno and Wang, 1993a,b; Sai and Cailletaud, 2007; Taleb
et al., 2006; Zhang and Jiang, 2008), the present study investigates the possible microstructural origins of microplastic rat-
cheting (purely tensile cyclic loading where R > 0) and proposes a combination of experiments and FEM simulations for pre-
dicting microstructural plasticity and evaluating its sensitivity to the microstructure to answer the following question: ‘‘how
detailed of a microstructural representation is necessary to capture microplastic ratcheting?” In other words, we are interested in
assessing the effect of the description of the microstructure on the simulation of microplastic ratcheting. A number of key
physical characteristics are taken into account in the modeling and simulations to accurately represent the onset and devel-
opment of microplastic ratcheting. As illustrated in Fig. 1, we use a coupling between microstructural characterization,
mechanical testing and numerical simulations to compare experimental and numerical results at both macroscopic and
microscopic level. While the present study considers one isolated problem in cyclic plasticity (purely tensile microplastic
ratcheting), the approach could be extended to address other cyclic plasticity issues, such as microstructural crack initiation
and small crack growth (McDowell, 2007) or cyclic evolution of residual stresses (Boyce et al., 2003).
The computational model explicitly addresses the effects of microstructure by including realistic topological information
(grain morphology, topology and crystallography) taken from electron back-scatter diffraction (EBSD) data. The physics of
deformation of the face centered cubic (fcc) polycrystal are incorporated through a classical crystal plasticity formulation
that explicitly considers the different dislocation slip systems, and the elasto-plastic crystallographic anisotropies at the
grain level. A computer-generated, representative microstructure is used to calibrate the crystal plasticity parameters to
the experimental data.
To assess and isolate the sensitivity of cyclic FEM-based polycrystalline plasticity to microstructure morphology repre-
sentation, we examine the microplastic ratcheting behavior of different idealized microstructures (square, hexagonal, tessel-
 R. Dingreville et al. / International Journal of Plasticity 26 (2010) 617–633 619

Fig. 1. Methodology for coupling microstructure characterization, mechanical testing, and numerical simulations.

lated, and EBSD-generated). This analysis demonstrates the sensitivity of numerical simulations to the various approxima-
tions in microstructure representation.

2. Methods for comparing experiments and simulations

2.1. Microstructure characterization

The microstructure-based approach presented in this paper is validated against deformation experiments performed in
uniaxial cyclic tension on high purity nickel. The mechanical testing was conducted using nickel sheet stock purchased from
ESPI Metals Corp. (Ashland, Oregon) at a thickness of 1.6 mm and a nominal purity of 99.9%. After being annealed at 1071 °C
for 30 min under flowing Ar, these sheets of material had an average grain size of 33 lm. Subsequently, the material spec-
imen was sectioned into tensile bars using electro-discharge machining (EDM). The bars were cut both longitudinal and
transverse to the rolling direction of the sheet with a nominal length of 10 cm in conformation to the ASTM E8-04 standard
for preparing tensile specimens. The tensile samples were prepared for EBSD measurements, by mechanically grinding and
polishing one surface to a 1 lm diamond finish followed by a final vibratory polish using 0.05 lm colloidal silica suspension.
To monitor representative areas of interest throughout the mechanical testing, the microstructural areas of interest were
marked with a rectangular fiduciary by ion milling with a FEI focused ion beam (FIB), DB235, using a gallium beam at
30 keV (see Fig. 2(a) and (c)). The three areas of interest tracked in this study were separated from one another by 1 mm
along the tensile axis of the specimen.
EBSD was used to spatially map the grain structure, microcrystallography, and local plastic deformation within each area
of interest periodically during the cyclic tensile testing. These EBSD measurements were performed using a Zeiss Supra 55
VP-FEG SEM taken at 20 keV equipped with a Nordlys II Charge Coupled Device (CCD) for EBSD diffraction pattern collection.
The acquisition and analysis of the EBSD data was performed with the HKL Channel 5 software suite. The three areas of inter-
est were collected on each tensile bar with a grid of 500  500 points at 0.5 lm per step to ensure resolution of microstruc-
tural features. Details of the various analysis algorithms can be found in Brewer et al. (2006).

2.2. Stress-controlled cyclic testing

The cyclic stress–strain tests were performed on the previously described 99.9% pure Ni rectangular dogbone tensile bar
with a gage section of 1.33  6.25  31.75 mm consistent with the ASTM E8-04 ‘‘subsize specimen” specification. Stress-con-
trolled saw-tooth cycles (linear ramp load and unload) were performed on an MTS servohydraulic loadframe with a mini-
mum and maximum stress of 1 and 112 MPa, respectively (R = 0.01) at a stress rate of 11 MPa/s corresponding to a cycle
duration of 20 s (0.05 Hz). The maximum stress was chosen to be 112 MPa to correspond to 1% deformation after the first
cycle while the R-ratio is kept constant to R = 0.01 to correspond to pure tensile loading/unloading with no reversal in the
macroscopic load. Strain was measured with a high-resolution extensometer with 4% full-scale range over a 25.0 mm gage
620 R. Dingreville et al. / International Journal of Plasticity 26 (2010) 617–633

Fig. 2. Microstructure of the areas of interest (a) revealed by EBSD measurements and (b) digitized for the numerical simulations. (c) The inverse pole
figures (IPFs) represent the raw grain orientations measured by EBSD with respect to the loading axis. The colors of the IPFs correspond directly to the colors
of the grains. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this paper.)

length. The specimen was removed from and mounted back on the test machine after cycles 1, 2, 4, 8, 16, 32, 128, 1000, and
10,000 for EBSD analysis.

2.3. Crystal plasticity

To examine microplastic ratcheting at the microstructural level, the material deformation was treated with a crystal plas-
ticity rate-dependent formulation (Buchheit et al., 2005). The foundations of the constitutive model assume that the elasto-
viscoplastic response of single crystals is dominated by slip deformation mechanisms. Other mechanisms such as twinning,
grain boundary sliding and diffusion are not considered. The total deformation of a single crystal consists of slip-based plas-
tic deformation, elastic lattice rotations and rigid body rotations. Following Kröner (1960) and Lee (1969), the single crystal
kinematics are described by means of a multiplicative decomposition of the total deformation gradient F into a plastic com-
ponent, Fp, representative of the motion of dislocations on active slip systems leaving the crystal lattice unrotated, and an
elastic component, Fe, depicting the rotation and elastic stretching of the lattice. For isothermal conditions,
 R. Dingreville et al. / International Journal of Plasticity 26 (2010) 617–633 621

F ¼ Fe  Fp : ð1Þ
A rate-form expression of the deformation is used to express the kinematics of the crystalline solid from this decompo-
sition. The velocity gradient L in the current configuration can be written as,

L ¼ F_  F1 ; ð2Þ
where F_ is the rate of change of the total deformation gradient. Subsequently, the velocity gradient L can be additively
decomposed into its elastic Le and plastic Lp counterparts such as,
1 1 1
L ¼ F_ e Fe þ Fe F_ p Fp ¼ Le þ Fe Lp Fe ð3Þ
p
The plastic velocity gradient L is assumed to be solely the result of crystallographic slip over the active slip systems
which corresponds to the sum of the plastic shearing rate c_ a on N active slip systems a (Asaro, 1983; Hill and Rice, 1972;
Rice, 1971),
X
N
Lp ¼ c_ a Pa ; Pa ¼ m
an
 a; ð4Þ
a¼1

where Pa is the generalized Schmid tensor resulting from the dyadic product of the crystallographic slip direction unit vector
m a and the slip plane normal unit vector n
 a in the so-called intermediate configuration of the conventional Kröner kinemat-
ics (Kröner, 1960). The rate of plastic deformation gradient F_ p resulting from crystallographic slip is governed by,

F_ p ¼ Lp Fp ð5Þ
A classical Hookean law gives the constitutive stress–strain relation under isothermal conditions such that,
1  eT e 
T ¼ C : Ee e; with Ee ¼ F F I ; ð6Þ
2
where T is the second Piola–Kirchoff stress tensor, Ee is the Green–Lagrange tensorial elastic strain measure, C is the aniso-
tropic fourth-order elastic stiffness tensor, and I the second order identity tensor. The Cauchy stress tensor r is related to the
second Piola–Kirchoff stress tensor through,
T
r ¼ ½detðFe Þ1 Fe TFe : ð7Þ
The flow and evolutionary equations describing the behavior of each individual slip system complete the formulation. The
kinetic equation used for the crystallographic slip rate c_ a follows a power law viscoplastic flow rule (Hutchinson, 1976) such
that,
 
 sa 1=m
c_ a ¼ c_ 0  a  sgnðsa Þ; ð8Þ
sCRSS
where c_ 0 is the reference shearing rate, m is the rate sensitivity, and sa and saCRSS are the resolved shear stress and total slip
resistance (or critical resolved shear stress), respectively, on slip system a. The sign of the resolved shear stress sgn(sa) ac-
counts for either positive or negative slip on the system.
To ensure numerical stability, we introduce a shear stress threshold s ¼a saCRSS such that,
8 a
< c_ ¼ 0; for sa < s;
 1=m ð9Þ
: c_ a ¼ c_ 0  asa  sgnðsa Þ; for sa  s:
s CRSS

This shear stress threshold, or active stress, might be viewed as a barrier to dislocation motion below which no slip occurs,
thereby limiting the accumulation of plastic deformation during the cyclic loading.
Three hardening laws are considered to describe the evolution of the overall resistance to slip saCRSS . The first is based on a
hardening saturation law proposed by Voce (1948, 1955) and used in an integrated form by Buchheit et al. (2005),
" !#
HVoce
saCRSS ¼ s0 þ HVoce
1 1  exp 2
Voce
ep ; ð10Þ
H1

where s0 is the initial slip resistance, HVoce

1 and HVoce
2 are material hardening parameters and ep is the equivalent plastic strain
defined by,
Z t rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 p p
p
e ¼ F : F dt: ð11Þ
0 3
The second hardening model considers the classical monotonic power law hardening to represent slip resistance,
pwr
saCRSS ¼ s0 þ Hpwr p H2 ;
1 ðe Þ ð12Þ
622 R. Dingreville et al. / International Journal of Plasticity 26 (2010) 617–633

where Hpwr1 and Hpwr

2 are material hardening parameters. Both the Voce and power law constitutive formulations assume
that all slip systems harden equally and that this isotropic hardening evolves according to the accumulation of plastic defor-
mation represented by scalar equivalent plastic strain. Expressing the hardening evolution through equivalent plastic strain
(as opposed to the algebraic sum of crystallographic shears) facilitates the determination of the slip system hardening
parameters from a single polycrystalline stress–strain curve. It should also be noted that the values of the isotropic harden-
ing parameters s0 ; HVoce
1 ; HVoce
2 ; Hpwr
1 and Hpwr
2 are dependent on the texture of the experimental data from which they are
derived. This dependence is taken into account through a Taylor factor, which depends on the crystal structure and on
the slip systems activated during deformation. For fcc metals with random texture, the appropriate Taylor factor M is
3.06. In calculating the hardening parameters for the Voce and power law hardening models, we have applied the appropri-
ate Taylor factor. It should be mentioned that, although these two hardening models give rather crude approximations of the
physical mechanisms involved, they offer the advantage of being efficient, simple and convenient to implement.
The third model is more physically based on self-hardening, and assumes that dislocations act as obstacles to dislocation
motion and contribute to the total slip resistance according to the Taylor equation (Taylor, 1934) given by,
pffiffiffiffiffiffi
saCRSS ¼ kG½110 b qa ; ð13Þ

where k is a statistical coefficient accounting for the deviation from regular spatial arrangements of dislocations, b is the
magnitude of the Burgers vector, qa is the forest dislocation density for a given slip system a and G[110] is the shear modulus
on the [1 1 0] plane. Note that the bulk shear modulus G is instead commonly used in the literature. However, at the grain
level, the directional shear modulus seems a better choice due to the local anisotropy of the material. The evolution of the
dislocation density is based on the phenomenological approach proposed by Kocks (1976) and Mecking and Kocks (1981)
where dislocation sources and sinks are considered:

dqa pffiffiffiffiffiffi
¼ HKM
1 qa  HKM a
2 q ; ð14Þ
dca

where ca is the accumulated slip on the slip system a, and HKM 1 and HKM2 are material hardening parameters. The first term
KM
H1 reflects that dislocation storage is inversely proportional to the dislocation mean free path, while the second hardening
term HKM2 describes the annihilation and cross-slip of dislocations. The initial dislocation density is found by inverting equa-
tion (13) and assuming that the initial slip resistance is equal to s0. An important limitation of the current approach is that it
employs a monotonic evolution equation (14) that ignores complexities such as back stress and slip band formation. Nev-
ertheless, note that Eq. (14) could have been written in terms of a slip system back stress with an Armstrong–Frederick form,
which is very similar but facilitates treatment of reversals. Finally, it should be mentioned that the focus of this work was on
the role of microstructure representation on simulation of microplastic ratcheting, rather than the physical models them-
selves. In other words, the aforementioned hardening models are targeted at studying the variability in ratcheting response
at both the macroscale and microscale for different representations of the same underlying microstructure.

2.4. Finite element implementation

The crystal plasticity model described above was incorporated into Sandia’s finite element analysis code, JAS3D (Biffle,
1987) as a standard subroutine. The simulations use eight node (hexahedral) 3D isoparametric elements with a single inte-
gration point at the element centroid. To deal with zero energy modes that may arise as a result of the single point integra-
tion scheme, the code uses an hourglass control, based on the work of Flanagan and Belytschko (1981). Numerical
integration of the constitutive model is performed using a forward Euler scheme. Restrictions are placed on the time step
to ensure that the forward integration scheme remains stable. If the step size requested by the user is larger than the allow-
able size, the step is divided into sub-increments of allowable size within the constitutive subroutine.

2.5. Parameter calibration

Parameters for the constitutive models introduced in Section 2.3 were calibrated using a representative microstructure
that is meant to approximate the actual specimen’s microstructure to not only provide a larger population of grains than
provided by the experimental data, but also to overcome the artifacts generated by the pixelization of the grain boundaries
induced by the EBSD-generated microstructure. This representative microstructure, illustrated in Fig. 3(a), was obtained by
means of a two-dimensional Potts model of grain growth (Holm and Battaile, 2001) with an imposed periodicity constraint
at the boundaries of the polycrystalline mesh. The grain growth simulation used to generate the representative microstruc-
ture employed the Monte Carlo Potts model to simulate ideal normal grain growth with isotropic grain boundary energy and
mobility. Hereafter, this 199-grain fictitious polycrystal is designated as the representative microstructure. The representa-
tive microstructure consists of 35,873 elements, extends only a single element into the third dimension (perpendicular to the
plane in Fig. 3(a)) and contains linear (as opposed to stepped) grain boundaries. Periodic boundary conditions were applied
in all directions, which is equivalent to simulating a columnar microstructure of infinite extent in all directions.
The crystallographic orientations assigned to each grain of the representative microstructure were assigned from orien-
tation distribution functions generated from the EBSD experimental data. For every pixel in the experimental digitized
 R. Dingreville et al. / International Journal of Plasticity 26 (2010) 617–633 623

Fig. 3. (a) Representative microstructure composed of 199 grains with periodicity constraints on the edges. (b) Stress vs. strain from experiment (black
symbols), and from a simulation using best-fit constitutive parameters and grain orientations on the microstructure in (a).

images from each of the three areas of interest, a set of three Euler angles (u1 ; U; u2 ; 0 6 u1 < 360 ; 0 6 U < 90 ; 0 6

u2 < 90 for a crystal with cubic symmetry) were generated from the EBSD measurements. A cumulative probability density
function Fh(x) was then defined from this experimental dataset for each Euler angle such that,
8 P
< x ! F h ðxÞ  Pðh 6 xÞ ¼ pðxi Þ;
xi 6x
ð15Þ
:
pðxi Þ ¼ NNi ;

where Ni is the total number of pixels having an orientation ranging from hi to hi + Dh, N is the total number of data points,
and p(xi) is the probability of observing an orientation h in the angular interval hi to hi + Dh. Subsequently, 199 sets of Euler
angles were randomly generated and chosen from the experimental cumulative probability functions Fh(x) and assigned to
each grain of the representative microstructure. Equivalent orientations based on crystal symmetry have been used to assure
that the desired orientation distribution was actually imposed. To guarantee this artificial polycrystal was the closest rep-
resentation of the actual experimental polycrystal, 100 different random orientation sets (each one of them containing
199 sets of three Euler angles) were generated. Each orientation set was then deformed to 1% elongation with the Taylor
model resulting in 100 stress–strain responses. The average response of these 100 different stress–strain curves was subse-
quently determined. As shown in Fig. 3(b) the simulated stress–strain response obtained from the average orientation dis-
tributions agrees well with the experimental data. The orientation set giving the stress–strain response closest to this
average behavior was chosen as the representative orientation set and, although not shown here, the deviations between
the 100 different stress–strain curves are relatively small. Consequently, the representative microstructure coupled with
the chosen orientation set was considered as statistically representative of the macroscopic experimental tensile specimen.
624 R. Dingreville et al. / International Journal of Plasticity 26 (2010) 617–633

This constitutes a valid assumption given the fact that the experimental data is sparse and the microstructure appears to
have a random texture. Furthermore, it should be noted that because this work is using a local model (i.e. the length scale
of the microstructure is implicitly accounted for in the materials fitting parameters), the effect of the grain size distribution is
not primary. Therefore, correspondence between the modeled and experimental grain size distributions is not a concern in
the present work.
The representative microstructure with the chosen orientations was used to calibrate the material parameters to the mac-
roscopic experimental response. These parameters include two flow parameters ðm; c_ 0 Þ, five hardening parameters
(s0, H1, H2, k, b) for three different hardening laws (only the first three are defined for the power and Voce hardening laws),
and one threshold parameter a  . The flow parameters and the hardening parameters were initially fitted analytically and then
adjusted to the experimental tensile test to further agreement at 1% deformation (prior to any stress-controlled cycling or
first unloading phase), while the threshold parameter was evaluated from the experimental cyclic response and its ratchet-
ing behavior. Results are shown in Fig. 3(b).
Tables 1 and 2 list the parameters used for the different constitutive models employed in this study. Note that the differ-
ent sets of hardening parameters used in this work are characteristic of the average grain size of the experimental samples.

2.6. Initial microstructure and boundary conditions

The deformation behavior of the nickel polycrystalline specimen under cyclic uniaxial tension was studied using the crys-
tal plasticity model described above in Section 2.3 for comparison against experimental results. As such, the geometry and
test conditions of the tensile specimen were based on the grain-level experimental work detailed in Sections 2.1 and 2.2.
The three EBSD-generated microstructures were represented by meshes of square, hexahedral, three-dimensional, iso-
parametric elements. These meshes consisting of 120  115  1 elements representing 36 grains for the first area of interest
(Area 1), 120  115  1 elements representing 45 grains for second area of interest (Area 2) and 97  94  1 elements rep-
resenting 35 grains for the third area of interest (Area 3). As shown in Fig. 2, the mesh is superimposed onto the microstruc-
ture, and grain orientations obtained from the EBSD texture analysis are assigned to the corresponding elements. Note that
the square grid used in both the EBSD and finite element representations, necessitates that most grain boundaries contain
right-angle segments to conform to the topology of the grid. However, during the course of this study, it was found that mesh
refinement did not substantially alter the results of the macroscopic ratcheting response and consequently the analysis was
carried out with the coarser mesh.
Even in a pure, nominally undeformed metal, the crystallographic data collected by EBSD is not uniform within a single
grain. This is due to several factors, including random error associated with the measurement itself and any significant lattice
imperfections present in the material. On the one hand, the nickel material used in this study is not expected to contain any
significant deformation. On the other hand, it is not straightforward to ascertain the specific source of crystallographic inho-
mogeneity in practice (e.g. ‘‘noise” in experimental measurements or real local misorientation). Therefore, we will consider
two treatments for each EBSD data set: one in which all intragranular crystallographic inhomogeneity is assumed to be
important (i.e. ‘‘raw” orientations), and another in which the orientation of each grain is assigned as the average value from
the EBSD data within the grain (i.e. averaged orientations). The orientation maps of the three area of interest are shown in
Fig. 2(c) in the case of the raw orientation.
To simulate the loading conditions of the experiments, a triangular waveform uniform traction was applied on both X
faces of the reconstructed microstructures, while the Y faces were constrained to remain planar and parallel (see Fig. 2
for coordinate system). The applied traction was chosen for each area of interest in order to attain 1% deformation after
the first tensile cycle, as in the experiments. Periodic boundary conditions were applied to the front and the back faces equiv-

Table 1
Elastic and viscoplastic parameters.

C11 (GPa) C12 (GPa) C44 (GPa) m c_ 0 ð1=sÞ a

246.5 147.3 124.7 0.04 1 0.832

Table 2
Hardening parameters.

Power law hardening Voce hardening Taylor hardening

s0 76.0a (MPa) 76.0a (MPa) 24.84 (MPa)
H1 2314 (MPa) 398 (MPa) 4.906 (108/m)
H2 0.8768 4298 (MPa) 26.65
k – – 0.5
b – – 2.517 (Å)
M 3.06 3.06 –
a
Note that the initial slip resistance s0 is weighed by the Taylor factor M.
 R. Dingreville et al. / International Journal of Plasticity 26 (2010) 617–633 625

alent to simulating a columnar microstructure. Note that the EBSD data, and therefore the simulated microstructures, rep-
resent only a small fraction of the test specimen. The real microstructures extend well beyond the boundaries of the regions
considered in this study, and this surrounding material undoubtedly imposes non-uniform boundary conditions unlike those
used here. It is obvious that the present boundary conditions are constraining and affecting the mechanical behavior of some
of the grains near these boundaries. This issue will be addressed in a subsequent publication (Dingreville et al., in prepara-
tion). In this paper, we therefore neglect grains close to the edges of the computational domains.

3. Cyclic ratcheting: numerical simulations vs. experiments

The comparisons of the experimental data to the representative microstructure and the predicted cyclic stress–strain re-
sponses for the three areas of interest are illustrated in Fig. 4 in the case of the Taylor hardening model with average grain
orientation. The microstructure-based approach adopted in this work provides good prediction of the microplastic ratcheting
response. The cyclic behavior of the representative microstructure and the three digitized microstructures agrees well with
the experimental results. The three digitized microstructures exhibit a similar cycle-dependent accumulation of plastic
strain at a similar rate. The differences between the three different areas and the experiment mainly stem from the small

Fig. 4. Comparison for (a) the representative microstructure and simulated stress–strain responses for stress-controlled cyclic tests, (b) the accumulation of
plastic deformation as a function of the cyclic history. The anisotropic Taylor hardening model, with averaged orientations per grain, was used in this set of
simulations. The different areas are labeled in Fig. 2.
626 R. Dingreville et al. / International Journal of Plasticity 26 (2010) 617–633

population of grains in each area, the effect of the local texture (the three digitized microstructures are softer than the rep-
resentative microstructure) and the nature of the microstructural representation (square elements for the digitized micro-
structures as opposed to paved elements for the representative microstructure). As a consequence of the aforementioned
discrepancies, the microplastic ratcheting for the digitized microstructures occurs mainly in the first three cycles while it
occurs across five or six cycles in the case of the representative microstructure and the experiments.
In all cases, both experimental and computational, two different stages can be observed in terms of strain accumulation
and ratcheting rate. The first stage consists of a rapid accumulation of the plastic strain for the first few cycles, while the
second stage exhibits a steady state where the accumulation of plastic strain is very small after each cycle. Both stages
are illustrated in Fig. 5, where only equivalent plastic deformations greater than 1% are shown. In the first stage of ratcheting,
the material hardens rapidly and pockets of plastic deformation form in grains with preferred orientations. In the second
stage, where ratcheting stabilizes, the cyclic hardening is very small after each cycle resulting in a slowly dissipating
(approximately constant) ratcheting rate.
The EBSD technique is capable of providing information about deformation-induced crystallographic reorientation with
subgrain resolution, which is a subset of the information available from our computational approach. However, just as the
accumulation of plastic strain saturates after only a very few cycles in Fig. 4(b), the evolution of subgrain-scale deformation
beyond the first cycle is not significant relative to the measurement uncertainty (approximately 0.5° local misorientation).
Therefore, while the agreement between the ratcheting simulations and experimental response is good at the macroscopic
level, a comparison between simulation and experiment at a subgrain level does not provide additional benefit to the present
analysis. Nonetheless, it should be noted that these simulations explicitly consider the microstructure morphology and
micromechanical evolution, and therefore produce more accurate representations of the microstructure’s response than con-
tinuum based models such as the Armstrong–Frederick kinematic hardening law or earlier FEM studies (Sinha and Ghosh,
2006; Xie et al., 2004).

4. Perspective on the effects of microstructural features

Given the differences in behavior between the three different areas, and the illustration of ratcheting at the microstruc-
tural level displayed in Fig. 5, it is evident that the sensitivity of the accumulation of plastic strains and ratcheting stems from
microstructural features such as plastic strain distribution, misorientation distribution and microstructure morphology/rep-
resentation (e.g. grain shape). In this section, we explore the impact of different attributes (constitutive and microstructural)

Fig. 5. Illustration of the two stages of ratcheting. Equivalent plastic deformations higher than 1  102 are plotted for Area 1 (i.e. no coloring corresponds
to plastic strain lower than this threshold) in the case of the Taylor hardening model with averaged orientation per grain. Note that the regions close to the
edges of the simulation box presented in Fig. 2(b) have been omitted. Area 1 is labeled in Fig. 2.
 R. Dingreville et al. / International Journal of Plasticity 26 (2010) 617–633 627

on the microplastic ratcheting behavior, not only to provide insight into the influence of microstructure on cyclic deforma-
tion, but also to emphasize some of the limitations of the present work.

4.1. Effect of work hardening model

Based on the observations made in Section 3 about the different stages of ratcheting, it is clear that the evolution of hard-
ening of the different slip systems controls the amount of ratcheting and its development through the cyclic history. In the
results shown in the previous section, the strain-hardening behavior used in the simulations is based on the anisotropic Tay-
lor hardening model given by Eqs. (13) and (14). In this section, we focus on understanding the influence of the hardening
model on the ratcheting behavior by comparing the predicted cyclic stress–strain responses at the macroscopic and micro-
scopic level for the three different hardening models (Voce, Power, Taylor) presented in Eqs. (10)–(14). The fundamental dif-
ference between the three hardening models is that, in the cases of the Voce and power law hardening, where hardening
evolves isotropically on all the slip systems, the anisotropic hardening behavior of the polycrystal as a whole is mainly
due to the crystallographic orientations of the individual grains and their immediate vicinities. In the case of the Taylor
anisotropic hardening model, the dependence of polycrystal hardening on crystallography is exacerbated by the different
evolution of the critical resolved shear stress saCRSS on the different slip systems. Therefore, the isotropic and anisotropic hard-
ening models used in this work influence the texture evolution differently.
As seen in Fig. 6, the macroscopic ratcheting response given by the isotropic hardening models differs from that obtained
by the anisotropic Taylor model. The isotropic hardening models tend to overpredict the accumulation of plastic strain com-
pared to the Taylor model. This overprediction can be interpreted as a direct consequence of the isotropy of hardening on
each slip system. When comparing the isotropic hardening to the anisotropic hardening at the microstructural level in
Fig. 7, it is clear that the localization and magnitude of the equivalent plastic strain ep are different, while the two isotropic
hardening models, i.e. Voce and power law, exhibit similar distribution and magnitude of plastic deformation within the
polycrystal over its cyclic history. In the isotropic hardening case, the partitioning and distribution of plastic deformation
is uniform throughout all the active slip system, while it is more localized in the case of the Taylor hardening model. This
difference at the subgrain level highlights the respective contributions of the different mechanisms (lattice rotations and
anisotropic slip systems hardening) involved in microplastic ratcheting. Another interesting point to be made is that this
quick survey of using simple hardening models demonstrates that, in the case of microstructural-based formulation (i.e.
crystal plasticity), there is no need for more advanced physical models to qualitatively capture microplastic ratcheting,
the simplest models such as Voce or power law hardening are sufficient. Nonetheless, more advanced physical hardening
models such as the Armstrong–Frederick kinematic hardening law might certainly improve quantitatively the details of
the simulations results.

4.2. Effect of orientation scatter

As noted previously, the crystallographic data collected by EBSD is not uniform within a single grain, even for the unde-
formed microstructure. While the nickel microstructure used in this study is not expected to contain any significant initial

Fig. 6. Effect of hardening model on the accumulation of plastic deformation as a function of the cyclic history for the different hardening. In all three cases,
the orientation of each grain is assigned as the averaged value from the EBSD data within the grain. The representative microstructure uses the anisotropic
Taylor hardening model. Area 1 is labeled in Fig. 2.
628 R. Dingreville et al. / International Journal of Plasticity 26 (2010) 617–633

Fig. 7. Effect of isotropic and anisotropic hardening on the plastic deformation localization (Area 2 with averaged orientation per grain). The circled areas
correspond to the highest equivalent plastic strain within the microstructure. Note that the regions close to the edges of the simulation box presented in
Fig. 2(b) have been omitted. Area 2 is labeled in Fig. 2.

deformation, and therefore intragrain rotation; it is not straightforward to ascertain the specific source of crystallographic
inhomogeneity in practice, i.e. scatter in experimental measurements or local misorientation. In this section, we consider
the impact of the initial intragranular crystallographic orientation on the cyclic behavior by evaluating two treatments
for each EBSD data set: one in which all crystallographic inhomogeneity is assumed to be important (i.e. ‘‘raw” orientations),
and another in which the orientation of each grain is assigned as the average value from the EBSD data within the grain (i.e.
averaged orientations). Fig. 8 illustrates the ratcheting behaviors of the uniform orientation per grain (averaged orientations)
and the non-uniform orientation within each grain (raw orientation). There is little difference between the two consider-
ations both at the macroscopic level and at the microscopic level (not shown here). In fact, at the microstructural level,
the localization and magnitude of the equivalent plastic strain are almost identical. Therefore, the treatment of the EBSD data
set in which the orientation of each grain is assigned as the average value from the EBSD data within the grain is a sufficient
assumption to describe the crystallography of the microstructure. It is worth noting that this statement is expected to hold
only when crystallographic inhomogeneities or intragranular misorientations are small (less than 2°). Materials that have
been previously deformed, e.g. by cold work, could contain significant intragrain misorientation that must be represented
in polycrystalline plasticity simulations.

4.3. Effect of microstructure

The effort dedicated to a realistic and detailed description of the microstructure in numerical simulations is important as
morphology/microstructure representation can have a significant influence on the mechanical response. The difference in
 R. Dingreville et al. / International Journal of Plasticity 26 (2010) 617–633 629

Fig. 8. Effect of misorientation on the accumulation of plastic deformation as a function of the cyclic history. The symbols represent ‘‘raw” orientations
while the dotted lines represent orientations averaged within each grain. The anisotropic Taylor hardening model was used in this set of simulations. The
different areas are labeled in Fig. 2.

the microplastic ratcheting behavior between the three different areas presented in Section 3 suggests the sensitivity of cyc-
lic plasticity to the details of the microstructure.
To have a better assessment of this sensitivity and to ascertain the level of microstructural detail required to ade-
quately capture ratcheting behavior both at the macroscopic and microscopic scales, we considered several different ide-
alized microstructures that follow a natural ‘‘microstructural evolution” towards the representative microstructure and
compare their behavior to the representative microstructure and the three digitized areas. As illustrated in Fig. 9, this pro-
gression starts from a microstructure with square grains comprised of one element each (Fig. 9(a)), to square grains with a
100 elements per grain (Fig. 9(b)), to a regular array of hexagonal grains containing several elements each (Fig. 9(c)), to a
microstructure of distorted hexagonal grains containing several elements each (Fig. 9(d)). The distorted hexagonal micro-
structure was obtained by randomly displacing the corners of the hexagonal grains by a small amount (less than 15% dis-
placements relative to the regular hexagonal grains). These idealized geometries allow for periodic boundary conditions to
be applied in all directions. For the square grains with 100 elements per grain, and the hexagonal grains, the crystallo-
graphic orientations assigned to each grain were obtained through the same procedure used in the case of the represen-
tative microstructure presented in Section 2.5. The orientations for the square grains with one element per were the same
as for the square grains with 100 elements per, while the orientations for the distorted hexagonal microstructure were the
same as for the regular hexagonal grain microstructure. Additionally to these four microstructures, we also considered a
‘‘shuffled” digitized microstructure (Fig. 9(f)). We used the Fisher–Yates shuffling algorithm (Fisher and Yates, 1948) to
randomly permute the orientation set of three Euler angles between elements of one of the digitized microstructures (Area
1), i.e. the shuffled digitized microstructure consists of a same square mesh with the same population of elements and
orientation set as the digitized microstructure except that the grain topology is lost in the permutation process. The Taylor
hardening model was chosen to describe the evolution of the overall resistance to slip saCRSS , while the orientation per grain
is assumed to be uniform.
The comparison between the macroscopic ratcheting behaviors of the different idealized microstructures is illustrated in
Fig. 10. The microstructural effects of the grain morphology do not bring any substantial improvement to the numerical pre-
dictions of ratcheting. The idealized microstructures give roughly the same amount of ratcheting as the digitized microstruc-
tures but deviate from the representative microstructure. This is partly due to the fact that the digitized microstructure
possesses jagged right-angle grain boundaries, which provides the same effect as the square microstructure. In terms of rat-
cheting rates, the differences between the very simple representations of the microstructure and the digitized microstruc-
tures are smaller than the differences they have with the experiment/representative microstructure. From a continuum point
of view, as long as the orientation distribution is reasonable, adding microstructural details slows down the simulations but
does not significantly improve macroscopic predictions over phenomenological models.
Despite the apparent similarity in ratcheting response at the macroscale, significant differences in the microscale distri-
bution of plasticity were observed for the different microstructural representations. To illustrate these differences, a simple
pointwise (local) crack nucleation model is used as a common metric at the microstructural level (microscopic scale) to eval-
uate the sensitivity of the cyclic behavior to the microstructure. Based on a slowly dissipating accumulation of plastic
630 R. Dingreville et al. / International Journal of Plasticity 26 (2010) 617–633

Fig. 9. Different idealized microstructure morphologies used to assess the influence of the microstructure on ratcheting: (a) square microstructure with one
element per grain, (b) square microstructure with 100 elements per grains, (c) regular hexagonal microstructure, (d) distorted hexagonal microstructure, (e)
digitized microstructure (Area 1), and (f) shuffled digitized microstructure (Area 1). Colors are assigned randomly and do not correspond to any properties.
Area 1 is labeled in Fig. 2. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this paper.)

Fig. 10. Effect of microstructure on the accumulation of plastic deformation as a function of the cyclic history. The anisotropic Taylor hardening model was
used in this set of simulations. Area 1 is labeled in Fig. 2.

deformation (assumed linear for simplicity), this crack nucleation model depends on the magnitude of the plastic strain at
the microstructural level. In this model, a crack is considered to nucleate when a critical maximum equivalent plastic defor-
mation enucleation is locally attained (Lemaitre and Chaboche, 1985) such that,
 R. Dingreville et al. / International Journal of Plasticity 26 (2010) 617–633 631

 
enucleation  ep;max
0
Nnucleation ¼ N0 þ hoep;max i ; ð16Þ
0
oN
N0

where N0 is the number of cycles for ratcheting to stabilize, ep;max

0 is the maximum effective plastic strain reached during the
N0th cycle, and is the rate of maximum effective plastic strain after N0 cycles. The nucleation strain enucleation is assumed to be
independent of the morphology considered and was arbitrarily chosen to be enucleation = 0.1.
Table 3 presents the crack nucleation predictions for the different idealized microstructures. As expected, a notable dif-
ference can be observed between the different cases. The square microstructure with one element per grain tends to under-
predict both the magnitude and rate of increase of the local maximum plastic strain, and therefore overpredicts the number
of cycles to crack nucleation. As the microstructural morphology becomes more complex and the mesh more refined, more
localized plastic deformation occurs in the microstructure, thereby lowering the predicted number of cycles to crack nucle-
ation. By increasingly adding details to the microstructure, one starts gradually resolving localized phenomena such as inho-
mogeneous deformation. It is interesting to note that both the representative microstructure and the digitized
microstructures predict local maximum equivalent plastic strains of the same order of magnitude. Furthermore, despite
the small population of grains and the disparity in grain morphology, the three different digitized microstructures show sim-
ilar nucleation predictions.
Even though the description of the microstructure has a weak influence on the macroscopic prediction of ratcheting,
the grain morphology strongly impacts the behavior at the microscopic scale. The level of detail necessary to describe
a microstructure is therefore closely related to the scale and phenomenon of interest. From a continuum perspective ide-
alized microstructures might be sufficient to describe the cyclic behavior of a polycrystalline material. However, to ade-
quately describe microscopic mechanisms such as crack nucleation, a more faithful depiction of the microstructure might
be necessary.

4.4. Limitations of the present approach

Although the procedure presented here provides good predictions of the ratcheting behavior and an assessment of sen-
sitivity to the microstructure description, several factors should be kept in mind when comparing experimental results to the
numerical simulations.
First of all, the simulated stress–strain responses soften slightly as the mesh refinement is increased since the number of
degrees of freedom within each grain is also increased. This is a direct consequence of the lack of length scale in the physical
models. No mechanism exists within the present framework to define a microstructural length scale. A non-local crystal
plasticity formulation would provide a better resolution of the texture evolution, and localization and distribution of stresses
and strains, at the microstructural level. Incorporating a length scale would intensify the role of the microstructure’s mor-
phology on microplastic ratcheting.
As previously mentioned, due to the limited microstructural information obtained from EBSD measurements, the FE mod-
el does not extend beyond the areas of interest studied here and consequently the boundary conditions applied to simulate
the uniaxial cyclic deformation of these regions do not properly account for the surrounding microstructure. This constraint
can induce local errors (Dingreville et al., in preparation) in the predictions of the behavior of the grains close to the edges of
the mesh.
A related issue concerns the influence of the underlying microstructure on the surface fields since deformation mecha-
nisms activated in each grain are three-dimensional. The microstructure below the surface cannot be determined with
the experimental procedures used in this work. Experiments combining focused ion beam and electron back scattered dif-
fraction may be able to tackle this issue by their ability to map in three dimensions the grain shape, grain size and crystal-
lography within a microstructure.

Table 3
Crack nucleation predictions for different microstructure morphology.
h i
Microstructure morphology N0 ep;max
0
ep;max
o0 Nnucleation
oN N0

Square grain (1 element/grain) 6 1.3662  102 9.70  106 89,057

Square grain (100 elements/grain) 6 1.7680  102 3.30  106 24,951
Regular hexagonal grain 6 1.7592  102 1.73  106 47,640
Distorted hexagonal grain 6 1.7819  102 1.78  106 46,175
Representative microstructure 6 1.9446  102 8.77  106 9191
Shuffled digitized microstructure (Area 1) 6 1.8782  102 1.48  106 54,840
Area 1 6 2.1897  102 3.21  106 24,337
Area 2 6 2.2226  102 4.74  106 16,418
Area 3 6 1.0622  102 3.77  106 21,054
632 R. Dingreville et al. / International Journal of Plasticity 26 (2010) 617–633

5. Conclusions

This paper investigates the sensitivity of ratcheting simulations to the various approximations in plastic hardening model
and microstructure morphology. The present model predicts the microplastic ratcheting phenomenon under pure tension
(no load reversal), a phenomenon that is not captured by traditional continuum FE models without a kinematic shift of
the entire yield surface into the purely tensile regime. Further, the model captures the onset of localized microstructural
plasticity prior to generalized yielding, a phenomenon important to microplastic ratcheting that is typically ignored in con-
tinuum simulations. A variety of representation of microstructures ranging from the experimentally determined to the sim-
plest computer-generated microstructures were incorporated into an FE model using a crystal plasticity theoretical
framework, and the predicted cyclic behavior was compared against experimental results both at the macroscopic and
microstructural levels.
As long as the orientation distribution is reasonable, the predicted stress–strain and ratcheting curves were in agreement
with the experimental results regardless of the morphology/representation of the microstructure. Examination of the influ-
ence of the hardening model on the ratcheting behavior highlighted the sensitivity to different hardening mechanisms, while
the consideration of experimental scatter in the initial crystallographic configurations showed no significant influence. In
fact, this work demonstrates that in the case of microstructural-based formulation (i.e. crystal plasticity) there is no need
for advanced physical models to qualitatively capture microplastic ratcheting, and the simple models such as Voce or power
law hardening are more than sufficient. However, the prediction of the detailed distribution of plastic strain at the micro-
structural level was quite sensitive to the specific microstructural representation. This observation may have important con-
sequences for damage predictions. The level of details necessary to adequately simulate the microplastic ratcheting response
of a microstructure was found to be dependent on the scale and phenomenon of interest. While having microstructural de-
tails is necessary to completely capture the ratcheting response, simple idealized representation of microstructures might
just be sufficient to describe the mechanical behavior of a polycrystalline material at the macroscopic scale, but a more real-
istic description of the microstructure is essential to describe localized mechanical behavior at the microscopic scale. The
chosen microstructural representation can induce limitations in the modeling of a physical phenomenon such as ratcheting.
The coupling methodology presented in this paper offers such a description in a self-contained form.

Acknowledgments

The authors gratefully acknowledge T.E. Buchheit and R.A. Roach for their useful discussions and valuable contributions.
Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States
Department of Energy, under Contract No. DE-AC04-94AL85000.

References

Abdel-Karim, M., 2009. Modified kinematic hardening rules for simulations of ratcheting. International Journal of Plasticity 25, 1560–1587.
Abdel-Karim, M., Ohno, N., 2000. Kinematic hardening model suitable for ratcheting with steady-state. International Journal of Plasticity 16, 225–240.
Ahmed, J., Wilkinson, A.J., 2001. ECCI characterisation of dislocation structures associated with extrusion/intrusion systems and fatigue cracks in copper
single crystals. Philosophical Magazine 81, 1473–1488.
Armstrong, P.J., Frederick, C.O., 1966. A mathematical representation of multiaxial Bauschinger effect. In: G.E.G.B. Report RD/B/N, p. 731.
Asaro, R.J., 1983. Crystal plasticity. Journal of Applied Mechanics 50, 921–934.
Bari, S., Hassan, T., 2000. Anatomy of coupled constitutive models for ratcheting simulation. International Journal of Plasticity 16, 381–409.
Bari, S., Hassan, T., 2001. Kinematic hardening rules in uncoupled modeling for multiaxial ratcheting simulation. International Journal of Plasticity 17, 885–
905.
Besson, J., Cailletaud, G., Chaboche, J.L., Forest, S., 2001. Mécanique Non-Linéaire des Matériaux. Hermès, Paris.
Biffle, J.H., 1987. A three-dimensional finite element computer program for the nonlinear quasi-static response of solids with the conjugate gradient
method. In: SAND Report, Albuquerque, NM, p. SAND87-1305.
Boyce, B.L., Chen, X., Peters, J.O., Hutchinson, J.W., Ritchie, R.O., 2003. Mechanical relaxation of localized residual stresses associated with foreign object
damage. Materials Science and Engineering A 349, 48–58.
Brewer, L.N., Othon, M.A., Young, L.M., Angeliu, T.M., 2006. Misorientation mapping for visualization of plastic deformation via electron back-scattered
diffraction. Microscopy and Microanalysis 12, 85–91.
Buchheit, T.E., Wellman, G.W., Battaile, C.C., 2005. Investigating the limits of polycrystal plasticity modelling. International Journal of Plasticity 21, 221–249.
Cailletaud, G., Sai, K., 1995. Study of plastic/viscoplastic models with various inelastic mechanisms. International Journal of Plasticity 11, 991–1005.
Cailletaud, G., Sai, K., 2008. A polycrystalline model for the description of ratchetting: Effect of intergranular hardening and intragranular hardening.
Materials Science and Engineering A 480 (1–2), 24–39.
Chaboche, J.L., 1991. On some modifications of kinematic hardening to improve the description of ratcheting effects. International Journal of Plasticity 7
(661–678).
Chaboche, J.L., 2008. A review of some plasticity and viscoplasticity constitutive theories. International Journal of Plasticity 24, 1642–1693.
Chaboche, J.L., Nouailhas, D., 1989. Constitutive modeling of ratchetting effects: part I. Experimental facts and properties of classical models. Transactions of
the ASME: Journal of Engineering Materials and Technology 111, 384–392.
Chen, X., Jiao, R., 2004. Modified kinematic hardening rule for multiaxial ratcheting prediction. International Journal of Plasticity 20, 871–898.
Chen, X., Jiao, R., Kim, K.S., 2005. On the Ohno–Wang kinematic hardening rules for multiaxial ratcheting modeling of medium carbon steel. International
Journal of Plasticity 21, 161–184.
Chen, G., Chen, X., Niu, C.D., 2006. Uniaxial ratcheting behavior of 63Sn37Pb solder with loading histories and stress rates. Materials Science and Engineering
A 421, 238–244.
Chen, X., Li, R., Qi, K., Lu, G.Q., 2008. Tensile behaviors and ratcheting effects of partially sintered chip-attachment films of a nanoscale silver paste. Journal of
Electronic Materials 37 (10), 1574–1579.
Dafalias, Y.E., Popov, V.P., 1975. A model of nonlinear hardening materials for complex loading. Acta Mechanica 21 (173–192).
Dingreville, R., Battaile, C.C., Brewer, L.N., in preparation.
 R. Dingreville et al. / International Journal of Plasticity 26 (2010) 617–633 633

Feaugas, X., Gaudin, C., 2004. Ratchetting process in the stainless steel AISI 316L at 300 K: an experimental investigation. International Journal of Plasticity
20 (4–5), 643–662.
Fisher, R.A., Yates, F., 1948. Statistical Tables for Biological, Agricultural and Medical Research, third ed. Oliver & Boyd, London.
Flanagan, D.P., Belytschko, T., 1981. A uniform strain hexahedron and quadrilateral with orthogonal hourglass control. International Journal for Numerical
Methods in Engineering 17, 597–609.
Hassan, T., Taleb, L., Krishna, S., 2008. Influence of non-proportional loading on ratcheting responses and simulations by two recent cyclic plasticity models.
International Journal of Plasticity 24 (10), 1863–1889.
Hill, R., Rice, J.R., 1972. Constitutive analysis of elastic–plastic crystals at arbitrary strain. Journal of the Mechanics and Physics of Solids 20, 401–413.
Holm, E.A., Battaile, C.C., 2001. The computer simulation of microstructural evolution. Journal of Metals 53, 20–23.
Hutchinson, J.W., 1976. Bounds and self-consistent estimates for creep of polycrystalline materials. Proceedings of the Royal Society of London A 348, 101–
127.
Johansson, G., Ekh, M., Runesson, K., 2005. Computational modeling of inelastic large ratcheting strains. International Journal of Plasticity 21 (5), 955–980.
Kang, G., 2008. Ratchetting: Recent progresses in phenomenon observation, constitutive modeling and application. International Journal of Fatigue 30,
1448–1472.
Kang, G., Kan, Q., 2007. Constitutive modeling for uniaxial time-dependent ratcheting of SS304 stainless steel. Mechanics of Materials 39 (5), 488–499.
Kocks, U.F., 1976. Laws for work hardening and low temperature creep. Transactions of the ASME: Journal of Engineering Materials and Technology 98, 76–
85.
Kröner, E., 1960. Allgemeine kontinuumstheorie der versetzugnen und eigenspannungen. Archive for Rational Mechanics and Analysis 4, 273–334.
Kulkarni, S.C., Desai, Y.M., Kant, T., Reddy, G.R., Prasad, P., Vaze, K.K., Gupta, C., 2004. Uniaxial and biaxial ratcheting in piping materials – experiments and
analysis. International Journal of Pressure Vessels and Piping 81, 609–617.
Kwofie, S., 2005. Cyclic creep behaviour described in terms of one parameter kinetic model. Materials Science and Engineering A 392, 23–30.
Lee, E.H., 1969. Elastic–plastic deformation at finite strains. Journal of Applied Mechanics 36, 1–6.
Lemaitre, J., Chaboche, J.L., 1985. Mecanique des materiaux solides, Paris.
McDowell, D.L., 1995. Stress state dependence of cyclic ratchetting behavior of two rail steels. International Journal of Plasticity 11, 397–421.
McDowell, D.L., 2007. Simulation-based strategies for microstructure-sensitive fatigue modeling. Materials Science and Engineering A, 4–14.
Mecking, H., Kocks, U.F., 1981. Kinetics of flow and strain hardening. Acta Metallurgica 29, 1865–1875.
Mroz, Z., 1967. On the description of anisotropic work hardening. Journal of the Mechanics and Physics of Solids 15, 163–175.
Ohno, N., 1990. Recent topics in constitutive modeling of cyclic plasticity and vicsoplasticity. Applied Mechanics Review 43, 283–295.
Ohno, N., 1997. Recent progress in constitutive modeling for ratchetting. Materials Science Research International 3, 1–9.
Ohno, N., Wang, J.D., 1993a. Kinematic hardening rules with critical state of dynamic recovery: I. Formulation basic features for ratchetting behaviour.
International Journal of Plasticity 9, 373–389.
Ohno, N., Wang, J.D., 1993b. Kinematic hardening rules with critical state of dynamic recovery: II. Application to experiments of ratchetting behaviour.
International Journal of Plasticity 9, 390–403.
Rice, J.R., 1971. Inelastic constitutive relations for solids: an internal-variable theory and its application to metal plasticity. Journal of the Mechanics and
Physics of Solids 19, 433–455.
Sai, K., Cailletaud, G., 2007. Multi-mechanism models for the description of ratchetting: Effect of the scale transition rule and of the coupling between
hardening variables. International Journal of Plasticity 23, 1589–1617.
Sinha, S., Ghosh, S., 2006. Modeling cyclic ratcheting based fatigue life of HSLA steels using crystal plasticity FEM simulations and experiments. International
Journal of Fatigue 28, 1690–1704.
Taleb, L., Cailletaud, G., Blaj, L., 2006. Numerical simulation of complex ratcheting tests with a multi-mechanicsm model type. International Journal of
Plasticity 22, 724–753.
Taylor, G.I., 1934. Mechanism of plastic deformation of crystals. Part I – theoretical. Proceedings of the Royal Society of London A 145, 362–387.
Voce, E., 1948. The relationship between stress and strain for homogeneous deformation. Journal of the Institute of Metals 74, 537–562.
Voce, E., 1955. A practical strain-hardening function. Metallurgica 51, 219–226.
Xie, C.L., Ghosh, S., Groeber, M., 2004. Modeling cyclic deformation of HSLA steels using crystal plasticity. Transactions of the ASME: Journal of Engineering
Materials and Technology 126, 339–352.
Yaguchi, M., Takahashi, Y., 2005. Ratcheting of viscoplastic material with cyclic softening, part 1: experiments on modified 9Cr–1Mo steel. International
Journal of Plasticity 21, 43–65.
Yoshida, F., Kondo, J., Kikuchi, Y., 1989. Viscoplastic behaviour of SUS 304 stainless steel under cyclic loading at room temperature. JSME International
Journal Series A: Solid Mechanics and Material Engineering 32, 136–141.
Zhang, J., Jiang, Y., 2008. Constitutive modeling of cyclic plasticity deformation of a pure polycrystalline copper. International Journal of Plasticity 24 (10),
1890–1915.