A Constitutive Model of Cyclic

N. Olmo Plasticity for Nonlinear Hardening

Department of Energy Engineering,
Toyohashi University of Technology,
Tempaku-cho, Toyohashi,440 Japan
Materials
A constitutive model is proposed for cyclic plasticity of nonlinear hardening
materials. The concept of a cyclic nonhardening range, which enables us to describe
Y. Kachi the dependence of cyclic hardening on the amplitude of cyclic straining or stressing,
Sumitomo Electric Industries, Ltd., is employed together with the idea of a two-surface plasticity model. Results of the
Shimaya, Konohana-ku, proposed model are compared with experiments of 304 and 316 stainless steels in
Osaka, 554 Japan several cases of cyclic loading in which mean strain is zero or nonzero and strain
limits are fixed or variable. Thus, it is shown that the model successfully describes
both the cyclic hardening phenomenon and the transient elastoplastic behavior after
initial and reverse yields of these materials. The capability of the model to provide
nonlinear cyclic stress-strain curves is also discussed.

Introduction
For detailed inelastic analysis of structural components sub- imaginary stress surface called the bounding surface is in-
jected to cyclic loading, we need a constitutive model which troduced to specify a stress region for the translation and ex-
can describe cyclic plastic behavior of materials accurately. pansion of a yield surface. It must be noticed, however, that,
Since cyclic hardening or softening is one of the most fun- without appropriate evolution equations of the bounding sur-
damental characteristics in cyclic plasticity of metals, the con- face, the two-surface models cannot describe cyclic hardening
stitutive model must describe this phenomenon by any means. or softening of materials well. Dafalias [7] proposed one type
However, it is not sufficient for the model to be valid for one of such evolution equations on the basis of the maximum
specific cyclic strain or stress range only, because generally plastic strain range, which Chaboche et al. [8] originally used
strain and stress ranges distribute nonuniformly in structural to describe cyclic plastic behavior of 316L stainless steel.
components. The constitutive model, therefore, must be valid Another idea to express the transient elastoplastic behavior
for any cyclic strain or stress range occurring in the compo- is to take account of a memory erasure term (a recovery term)
nent to be analyzed. The dependence of cyclic hardening on in the evolution equation of a kinematic hardening variable
the size of cyclic strain or stress range is significant especially [9]. However, this idea and that of two-surface plasticity
in 304 and 316 stainless steels, as seen later in Figs. 5 and 10. models are not completely different from each other, because
Ohno [1] proposed a constitutive model of cyclic plasticity a mathematical equivalence can be shown between them in a
by introducing the concept of a cyclic nonhardening range, special case [10], as quoted in [11].
which enabled us to describe the dependence of cyclic harden- In the present paper, by applying the concept of a cyclic
ing mentioned above. He postulated that a plastic strain range nonhardening range to the evolution equations of the
called the cyclic nonhardening range develops as cyclic strain- bounding surface in a two-surface model, we construct a con-
ing proceeds, and that the plastic strain increment inside this stitutive model of cyclic plasticity which can describe both the
range does not contribute to cyclic hardening. Some details of cyclic hardening phenomenon and the transient elastoplastic
the concept will be reviewed later. The validity of the model behavior after yielding. The validity of the resulting model is
was ascertained for 304 stainless steel subjected to cyclic verified on the basis of experiments of 304 and 316L stainless
straining between variable, as well as fixed, strain limits at steels subjected to several kinds of cyclic straining at room
room temperature [1]. temperature. Besides, we discuss the capability of the pro-
However, the model above has a limitation that it cannot posed model to provide nonlinear cyclic stress-strain curves.
describe transient elastoplastic behavior observed just after in-
itial yielding and reyielding under reverse loading.
The transient elastoplastic behavior can be expressed suc- Previous Model Based on Cyclic Nonhardening Range
cessfully by two-surface plasticity models [2-6], in which an (Model I)
First let us review the constitutive model proposed by Ohno
[1] in order to explain the description of cyclic hardening
Contributed by the Applied Mechanics Division for publication in the JOUR- based on the cyclic nonhardening range. We consider here for
NAL OF APPLIED MECHANICS.
Discussion on this paper should be addressed to the Editorial Department, simplciity uniaxial loading of stress a, strain e, and plastic
ASME, United Engineering Center, 345 East 47th Street, New York, N.Y. strain e^.
10017, and will be accepted until two months after final publication of the paper
itself in the JOURNAL OF APPLIED MECHANICS. Manuscript received by ASME Cyclic Nonhardening Range. We assume that isotropic
Applied Mechanics Division, August 8, 1984; final revision, September 5, 1985. hardening of materials does not develop, while the plastic

Journal of Applied Mechanics JUNE 1986, Vol. 53/395

Copyright © 1986 by ASME
Downloaded From: http://appliedmechanics.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use
 i(2//3)e' developing. Therefore, 2p increases to AiP ( = e£,ax - eg,in).
The increase of p as a function of n is obtained as equation
044) in the Appendix.
It is seen from equation 044) that when c is a constant as in
the present paper, the number of cycles for the saturation of
progressive development of the range g < 0 does not depend
on AtP. If c is a function of p and so on, however, it will de-
pend on Ae?.
A special case of the range g < 0 with c = 1/2 respresents
the maximum plastic strain range, which was originally used
by Chaboche et al. [8], and subsequently by Dafalias [7], for
the description of strain-controlled cyclic plastic behavior of
316 stainless steel. However, it was pointed out that the use of
this special case is not always appropriate even under strain-
controlled cyclic loading [1], see also [12]. Moreover, Ohno
and Kachi [13] recently showed that the progressive develop-
Fig. 1 Evolution of cyclic nonhardening region g < 0 in plastic strain
space ment of the cyclic nonhardening range, i.e.,, c < < 1/2, is im-
portant for the description of stress-controlled cyclic plastic-
ity. Incidentally, constitutive models of creep with the range
strain point moves inside a certain plastic strain range after a g < 0 were proposed by Murakami and Ohno [14] and Ohno
reversal of the straining direction, for example, the range [P,, et al. [15],
P 2 ] along the strain path in Fig. 1. We refer to this plastic Model I. Assuming combined isotropic-kinematic harden-
strain range as the cyclic nonhardening range, because the ing of materials, we consider the following constitutive rela-
plastic strain increment inside it does not contribute to the ac- tions incorporating T defined by equation (4):
cumulation of cyclic hardening. It is possible to interpret such
a plastic strain range from the microscopic point of view [1];
f=(a-vy-K2 =0 (5)
dislocations piled up to obstacles may be remobilized under K = K(Q), q = T\e"\ (6)
reverse straining, and thus a certain amount of plastic strain r) = [K+(l -T)dK/dq]eP {la)
may proceed without marked increase of the dislocation densi-
ty after a reversal of the straining direction. eP=(K+dK/dq)~la (8a)
The evolution of the cyclic nonhardening range, which can where/ = 0 is the yield condition, K and t\ denote half the size
be a plastic strain region under multiaxial loading, is il- and the center of the elastic stress r a n g e / < 0, q is an isotropic
lustrated in Fig. 1. Under mono tonic or near-monotonic hardening variable, and ^ is a constant for kinematic
straining as along the path OPu the region expands and hardening.
translates progressively under the condition that the plastic The concept of the cyclic nonhardening range is embodied
strain point remains on the boundary of the region and moves through T in equations (6) and (7a) as follows:
in an outward direction. Under reverse straining, on the other
r = 1: When the plastic strain point c? is located on the
hand, the region is assumed to remain unchanged as long as
bound g = 0 and moves outward, equations (5)-(8) represent
the plastic strain point moves inside it (i.e., along the path
the combined hardening model of isotropic hardening
P\P2)- But it starts to expand and translate again, when the
modulus dK/dq and kinematic hardening modulus K.
plastic strain point reaches the bound (i.e., the point P 2 )-
T = 0: On the other hand, when <? moves inside the range g
In the case of uniaxial loading, the cyclic nonhardening
< 0, the isotropic hardening variable q does not change, and
range is represented as follows:
equations (5)-(8) are reduced to those for the kinematic
2
g=(eP-a) -p2<0 (1) hardening model of hardening modulus K + dK/dq.
where a and p denote the center and half the size of this range. It is seen from equation (8a) that the plastic tangent
As assumed in the above, the range g < 0 expands and modulus is equal to K + dK/dq independently of the value of
translates, only when the plastic strain point eP is located on T, and hence it is not disturbed by the discrete change of T in
the bound g = 0 and moves outward: equation (4). It is because the change of the isotropic harden-
ing modulus from dK/dq to zero due to T in equation (6) is
p = cr\e"}, p(0)=Po (2) compensated with the change of the kinematic hardening
a=(l-c)TeP (3) modulus expressed in terms of (1 -Y)dKldq in equation (la).
where c is a constant specifying the expansion rate of the range Figure 2 illustrates the response of equations (5)-(8). In this
g < 0, and r represents the evolution condition of this range: figure, dK/dq is taken to be a constant. The evolution of the
cyclic nonhardening range g < 0 is also shown by hatched
fl, g = 0 and (dg/deP)e">0 ranges on the axis of e", and it occupies the ranges RA, RB,
F=] (4) and Rc when plastic deformation proceeds to A, B, and C on
[0, g<0 or (dg/deP)eP<0 the a - (P diagram, respectively. (It is assumed that the range
The initial value of p is taken to be p 0 , whereas that of a is g < 0 is initially reduced to the origin of the axis of eP, i.e., p 0
zero. Here and from now on, ( ) stands for the derivative = 0). Then, the model gives the following cyclic response:
with respect to a certain loading parameter. It is ascertained f) Under the incipient tensile loading to B, T = 1.
that when g = 0 and (dg/de»/)e" > 0 ( i . e . , r = 1), equations Therefore, the range g < 0 expands and translates continuous-
(2) and (3) satisfy the consistency condition g = 0. ly, and stress a increases owing to the expansion and transla-
Under cyclic straining between fixed strain limits eg,in and tion of the yield surface.
epmm, the cyclic nonhardening range g < 0 develops pro- if) Under the unloading and reverse loading from BtoB,ep
gressively with an increase of the number of cycles, n (see Ap- moves inside the range RB, and hence T = 0. Therefore, the
pendix A). Eventually, the range g < 0 occupies the cyclic range g < 0 remains unchanged, and the yield surface does
range of plastic strain [eg,in, e£ ax ], and the condition g < 0 not expand but simply translates. This translation of the yield
becomes satisfied in a whole cycle. Then, equation (4) gives Y surface is accompanied with an increase of the kinematic
the value of zero only, so that the range g < 0 ceases from hardening modulus expressed by (1 - T)dK/dq in equation

396/Vol. 53, JUNE 1986 Transactions of the ASME

Downloaded From: http://appliedmechanics.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 /3a

Fig. 3 Bounding surface f* = 0 and yield surface f = 0

2p o 1

r i q i
r a n g e g < 0 : w//\///i
a
Fig. 2 Stress-strain diagram by Model I, together with evolution of
cyclic nonhardening range g < 0
(7a), as mentioned already (see the dashed line in the figure).
<*
Hi) Under the reverse loading from B to C, eP keeps located
on the bound g — 0 and moves outward (T = 1). Thus the
development of the range g < 0 and the expansion of the yield / •
surface occur in the same manner as under the incipient
loading. 0
iv) The response under the second reverse loading is the
same as under the first, except that the plastic strain range of Y
= 0 under the second reverse loading, Rc, is larger than the s /
first, RB.
The model describes the saturation of cyclic hardening
which depends on the amplitude of cyclic straining as follows:
As cyclic straining proceeds between fixed plastic strain limits,
the cyclic nonhardening range eventually occupies the cyclic
^ - " " " " ^ a * (Mode2 l I
range of plastic strain, so that the condition g < 0 for Y = 0
becomes satisfied in a whole cycle, as was mentioned earlier
and will be seen from Fig. 2. Then, the yield surface ceases Fig. 4 Stress-strain diagram by Model II
from expanding and only translates. The saturated size of the
yield surface is given by 2K(qs), where qs, the saturated value
of q, is expressed as qs = (1/c) (AeV2 - p0) (see equation cyclic recovery of mean stress, this term has only a small effect
(Al) in Appendix). Therefore, Model I describes the on the transient elastoplastic behavior after yielding.
dependence of cyclic hardening on the size of cyclic strain
range. If K(q) is a dcreasing function of q, Model I expresses Two-Surface Plasticity Model
cyclic softening.
A two-surface model is briefly explained by following Krieg
By taking account of the recovery of kinematic hardening [3] and Dafalais and Popov [4, 5] based on Mroz [16].
[9], we can generalize the kinematic hardening equation (7 a) to In order to describe the transient elastoplastic behavior of
•i) = [K+ (l-Y)dK/dq\ep -Kri}\ep\ (76) materials observed just after initial and reverse yieldings, a
where Kr is a constant. Then, by use of the condition/ = 0, bounding surface/* = 0 is introduced inside which a yield sur-
equation (8a) is modified as face/ = 0 is allowed to translate and expand, as schematically
shown in Fig. 3, where 77,* and JJ^- denote the centers of these
e"= [K + dK/dq-sgn[o-ri]Krri)-l0 (8b) surfaces, K* and K being their radii. After the onset of yielding,
Introduction of such a recovery term seems to give a general the yield surface is assumed to translate so that stress ay ap-
model which can describe nonlinear hardening of materials, proaches the stress point afj o n / * = 0 at which df/dafj is co-
cyclic relaxation of mean stress, and cyclic creep at least directional with df/ddjj. In this change of the yield surface, the
qualitatively. Nevertheless, we will employ a two-surface plastic tangent modulus is taken to be larger when <jy is more
plasticity model, because this model has an advantage that distant from <T,*, and it gets to depend on the evolution of the
transient elastoplastic behavior after yielding can be described bounding surface as cr,-, comes near a,*. In other words, the
separately from cyclic hardening behavior by means of change of the yield surface within the bounding surface
bounding and yield surfaces, as explained below. Moreover, it describes the transient elastoplastic behavior after yielding,
will be discussed later that when the coefficient Kr in the while the evolution of the bounding surface is responsible for
recovery term in equation (lb) is determined from the data of the plastic behavior succeeding this transient one.

Journal of Applied Mechanics JUNE 1986, Vol. 53/397

Downloaded From: http://appliedmechanics.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Formulation of New Constitutive Model (Model II) Then, the uniaxial evolution equations of the bounding sur-
face, (9) and (10), can be generalized as
We are in a position to formulate a new constitutive model
of cyclic plasticity, Model II, by combining the cyclic K* = K*(q), q^V^P (20)
nonhardening range model with the two-surface model. 7lfj = (2/3){K+(l-r)dK*/dq}efj-Kri1tjP' (21)
Let us specify evolution equations of the bounding surface
on the basis of the cyclic nonhardening range; i.e., we regard Similary, the translation equation of the yield surface, (12),
the yield surface in Model I as the boundary surface (Fig. 4). can be extended as
Then the radius and center of the bounding surface, K* and ?,*, 71y=A(Sy-Sy)£P
change as in equations (6) and (7): (22o)
S
K* = K*(q), q = V\iP\ (9) *U=7lfj+ (K*/K)(SU~^i^
rt* = [K+(\-T)dK*/dq]P-Kri)*\eP\ (10) Assuming the normality of eg to the yield surface, and using
the consistency condition/ = 0 together with equations (19)2
where V is given by equations (l)-(4). The recovery term in and (22a), we obtain
equation (10) plays a role in describing cyclic recovery of mean
stress. q = Q/2)nunklsM/H (23)
Next let us specify the evolution of the yield surface for H= (3/2)[/2A (Sy-Sy)ny (24a)
Model II. We assume for simplicity the yield surface of con-
where n,y = (df/do^/Mf/doW and H denotes the plastic
stant size
tangent modulus.
/=(<T-T,)2-K2=0, K = /CQ (11)
Elimination of sfj. Although we have completed the for-
where K0 is a constant. In regards to the translation of the yield mulation of the model, we eliminate sfj from the constitutive
surface inside the bounding surface, let us employ the rule [3] relations, i.e., equations (22a) and (24a). Equations (22a)2 and
(23) give, respectively,
\ (12) Sy Sy = I,} - , , „ + ( K* IK - \)(Sy ~ 7,y ) (25)
a* = r,* + sgn[o-T)]K* J efj/iP = (3/2)(Sy-Vy)/K (26)
where a* is the stress state on the bounding surface which Then, equations (22a), and (24a) are rewritten as
stress a approaches (Fig. 4), and A is a constant, possibly a
function of a*, a and so on. The above equation is a simple ex- yy=A[(2/3)(K*-K)i>!j- dto-1,5)^] (226)
ample, and it is possible to use other translation rules, e.g., [4, H=A[K* U2
-K~(3/2) (Vy-7,fj)ny] (24b)
5]- It is seen from equation (22b) that the translation equation of
Then, using the consistency condition/ = 0 together with the yield surface of Model II has a recovery term. Since it has
equations (11)2 and (12), we obtain the form (i,,y - i,,*)e", the recovery of rjy takes place around
iP = sgnhr - -q] {A (a* - a)} ~' a (13) the center of the boundary surface. It is obvious from the
character of Model II that this recovery term describes the
Extension of Model II to Multiaxial States transient elastoplastic behavior of materials just after yielding.
The new constitutive model expressed by equations (9)-(13)
together with (l)-(4) is extended to multiaxial stress states. Comparison With Experiments and Discussion
Cyclic Nonhardening Region. Restricting our concerns to Cyclic Torsional Tests of 304 Stainless Steel. Let us discuss
radial or near-radial loading, we approximate the cyclic the validity of the proposed model, Model II, on the basis of
nonhardening region by the hyper-sphere of radius p and cyclic torsional tests of 304 stainless steel at room temperature
center a,y (Fig. 1). Then, equations (l)-(4) are extended as [1]. Total (engineering) shear strain y is assumed to be the sum
follows [1]:
g=(2/3)(.efj-aij)(efj-aij)-p2<0 (14)
p = cT& (15) 1 1 ^ ^
in
au = Q/2) {\-c)Y&vu (16) .300
o
fViM/\\iP\\,
V=\
g = 0andc, y eP>0
(17)
CL
€>*^
[0, g<0 or ^ e « < 0
1/2
%
where e* = [(2/3)egeg] , v-y = (dg/defj)/\\dg/d<J>\\, and IItil
~~*r^,——* —
= (tjjtjj)1'2 for a second rank tensor ty. It is seen from equa- ^200
tion (17) that T is equal to the cosine of the angle between the
vectors vtj and eg when g = 0 and j>,yeg > 0, and thus 0 < T < Cyclic torsion test (15th cycle)
1. Kb o zero mean-strain
Bounding and Yield Surfaces. We consider the following • [0.0,0.03]
multiaxial forms of the bounding and yield surfaces: v [0.015,0.03]
100 --
r =(3/2)(^.-^.)(4-^)-/c*2 =0 (18)
Monotonic torsion test •
/ = (3/2)(^ - ViJ) (SiJ - no) ~ K2 = 0, K = K0 (19)
where Sy and Sy are the deviators of a,* and aijt respectively. -— Present model
The centers ?,,* and i)y are assumed to be in the deviatoric stress Bounding stress T*, Eqs.(28),(29)
space, and thus the axial components in uniaxial stress states, I 1 1 ...
T/*, and T)U , are related with T)* and t\ in uniaxial equations (10) 0 0.01 0.02 0.03
and (12) as follows: 7 P , A7 p s /2
Fig. 5 Torsional cyclic and monotonic curves of stress versus plastic
ijf,= ( 2 / 3 V , Vn=(2/3)v. strain (304 stainless steel [1])

398 / V o l . 53, JUNE 1986 Transactions of the ASME

Downloaded From: http://appliedmechanics.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 of elastic and plastic components claculated by Hooke's law T (MPa)
and equations (9)-(13) together with (l)-(4), respectively.
Equations (9)—(13) and (l)-(4) are used by simply replacing a 300
and (P with shear stress T and shear plastic strain yp. The rate-
dependence is disregarded, which this material shows at room
temperature [17, 18], although the present model can be ex-
tened to viscoplastic deformations (see Concluding Remarks).
Let us choose the form of K* (q) as
K*(q) =K$ + Lq (KQ,L: constants) (27)
Then, if the recovery term in equation (10) is negligibly small,
equations (A 9) and (.410) in the Appendix give the following
relations for monotonic loading and cyclic loading between
fixed strain limits, respectively:
T* = KS + (K + L)JP (28)

^ = KS + (.K+L)PO+[K+ (I+-^-)L](^--P0) (29)

where A designates the sizes of cylic ranges, and the subscript s • OAD Experiment
stands for the stabilized state of stress-strain hysteresis loops.
Present model
Moreover, the monotonic stress-strain relation is obtained as
equation (^411). These relations helped us obtain the following Fig. 7(a)
values of material constants from the experimental data in T(MPa)
Figs. 5 and 6 (see Appendix D): K0* = 197 MPa, K0 = 150
MPa, ,4 = 350, K = 900 MPa, L = 265 MPa, Kr = 7.0, p0 =
200 >-?*«h^3^^st7
0.0036, and c = 0.08.
Figures 7(a)-7(c) show the stress-strain hysteresis loops for
cyclic strain ranges [-0.03, 0.03], [0.0, 0.3], and [0.015,
0.03], respectively. The changes of peak stresses during the
tests are shown in Fig. 6. The 15th loops can be regarded as 100
stabilized (Fig. 6).
It is seen in Figs. 7(a)-7(c) that the proposed model
simulates fairly well the stress-strain hysteresis loops obtained
from the experiments. In regards to Fig. 1(a), however, the 0
F
b 0.01
1
0.02 J 0.03'
Bauschinger effect does not become so large in the calculation
as in the experiment, as cyclic hardening proceeds. A better I 4
description may be obtained by letting the coefficient A in r » o A Experiment^ J
equation (12) be a function of K* , for example, as A = -100
~fi Present /
A0[IL + (\-IX)K.Q/K*] (A0, II: constants). Then, since A
becomes smaller with increase of K*, larger Bauschinger effect
f model &/
will be calculated as cyclic hardening proceeds.
-200

1 1
[-0.03,0.03]
300-
^-—5 A &
D /6"
Q. 200
2

P250 [0.0,0.03]
(+) r, r, O O O . O O 6
O Q U
^j -is IT ~&~
/ A__. A. A--
A .» (-)
>^o [0.015,0.03]
0 O O o
• ° ° ° O O O o l
/ ~7+)
200 A _ _ A - - A - - A - -T\ 3 A" A -
A A
A A A
A

A
o A Experiment
ZZZ. Present model
• 200
I O A Experiment
150 1 I ZTZ Present model
0 5 10 15
Fig. 7(c)
Cycles Fig. 7 Stress-strain loops for constant cyclic strain ranges (304
Fig. 6 Changes of positive and negative peak stresses, stainless steel [1]): (a) cyclic strain range [-0.03, 0.03]; (b) cyclic strain
r peak and
peak> < 304 stainless steel [1]) range [0.0, 0.03]; (c) cylic strain range [0.015, 0.03]

Journal of Applied Mechanics JUNE 1986, Vol. 53/399

Downloaded From: http://appliedmechanics.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 The values of A T / 2 and Ayps/2 in the experiments for the
strain ranges [0.0, 0.03] and [0.015, 0.03] shown in Figs. 1(b)
and 7(c) are plotted in Fig. 5, too. Noting that mean strain is
not zero in these experiments, we see that the saturated level of
cyclic hardening is expressed as a function of strain amplitude
and hence does not depend on mean strain. This tendency was
Strain observed for 2017 aluminum alloy, too, subjected to several
range (7o)
kinds of plastic prestrain, up to 10.0 percent, followed by
°((1J Experiment
cyclic straining of plastic strain range 1.0 or 2.0 percent [19].
The experimental tendency above is described by the present
100 ~ ! i Present model as follows: We have assumed that the expansion of the
model bounding surface is controlled by the development of the
Cycles
cyclic nonhardening range g < 0. Since the range g < 0 with c
< < 1/2 develops little by little with increase of the number of
cycles, it eventually occupies the cyclic range of plastic strain
10 20 30 50 irrespective of the value of mean strain. Thus, equations (A 10)
Cycles and (A7) in the Appendix are valid in this case, too.
Fig. 8 Changes of peak stresses under variation of cyclic strain range Therefore, mean strain does not affect the saturated level of
(304 stainless steel [1])
cyclic hardening.
Chaboche et al. [18] proposed a model of cyclic plasticity by
representing isotropic hardening in terms of the maximum
T (MPa) plastic strain range given by a special case of equations (l)-(4)
11
with c = 1/2. However, their model is not supported by the
experimental observation above: It can be shown that for
cyclic straining between e^in and e£,ax (0 < e£,in < eg,ax), this
maximum plastic strain range takes [0, e£,ax]. Therefore, ac-
cording to their model, cyclic hardening in this case depends
on e£,ax rather than Ae? ( - e^ax - e^in), no matter how small
Ae^ is. It does not agree with the experimental results shown in
Figs. 1(b) and 7(c) for the cyclic strain ranges [0.0, 0.03] and
[0.015, 0.03], where e£ax is the same.
We now discuss the recovery term in the translation rule (10)
-0.03 of the bounding surface, i.e., the term Krr\* I e" I. When Kr =
0, this rule is essentially a linear one and similar to Prager's
rule; hence, the present model does not describe cyclic relaxa-
tion of mean stress under nonzero mean strain. But, for a
positive value of Kr, it describes the cyclic relaxation, which
becomes rapid with increase of Kr. Therefore, the value of Kr
was determined from the best fitting of the data of cyclic
relaxation (cyclic equlaization of positive and negative peak
stresses) obtained from the experiments of [0.0, 0.03] and
[0.015, 0.03] (see Fig. 6). Then, it was found in equation (10)
Fig. 9(a) that the magnitude of the recovery term is at most about twen-
ty percent of that of the hardening term. In other words, when
the recovery term in equation (10), or (lb) for Model I, is
determined from the data of cyclic relaxation of mean stress, it
T (MPa) is not large enough to describe the transient elastoplastic
A
behavior appearing just after yielding.
Figures 8 and 9 show predicted and experimental results
under cyclic straining between varying strain limits. The cyclic
history consists of Step I through V with ten cycles each, as
shown in Fig. 8. The stress-strain loops in Fig. 9 are the ones at
the last cycles in these steps.
It is seen in Fig. 9 that the predicted peak stresses coincide
well with the experimental ones. The degree of coincidence is
nearly the same as that in the previous paper [1]. It is because
the bounding surface in the present model is just the yield sur-
face of Model I proposed in [1], In regards to the description
of the transient response after yielding, on the other hand, the
present model is definitely better than Model I and simulates
the experimental stress-strain loops fairly well (Fig. 9).
Incidentally, the test results in Fig. 8 shows weak cyclic
softening after the reduction of the strain range size from Step
IV to V, but the present model does not describe it. This
phenomenon may be described by adding a memory erasure
IV -300 term to equation (2), see [30].

Fig. 9(b) Cyclic Uniaxial Tests of 316L Stainless Steel. The cyclic
Fig. 9 Stress-strain loops at the last cycles in Step I through V: (a) stress-strain curve of 316L stainless steel at room temperature
present model; (b) experiment (304 stainless steel [1]) [8] shown in Fig. 10 is considerably nonlinear, in contrast with

400/Vol. 53, JUNE 1986 Transactions of the ASME

Downloaded From: http://appliedmechanics.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 = 0.01 becomes nearly stabilized after about ten cycles [8]. It
a is seen in Fig. 10 that the present model simulates well the ex-
a. perimental results.
2 600- Figure 11 shows the stabilized a — e loops under stepwise
expanding of strain range. It is seen that the present model
%
describes very well the transient response after yielding ob-
served in the experiment. It is also seen that the nearly-elastic
400
stress range of large plastic stiffness expands with increase of
the strain range size, although we have assumed no expansion
of the yield surface, i.e., equation (11)2-
200 Concluding Remarks
We proposed a constitutive model of cyclic plasticity,
Model II, by combining Model I based on the cyclic
nonhardening range with a two-surface model. It was shown
M 2 Q 0 3 that the resulting model describes well the cyclic hardening
^(2 phenomenon and the transient elastoplastic behavior after
Fig. 10 Uniaxial cyclic and monotonic curves of stress versus strain yielding of 304 and 316 stainless steels in several cases where
(316L stainless steel [8])
mean strain is zero or nonzero and strain limits are fixed or
variable.
Recently, Ohno and Kachi [13] applied the present con-
stitutive model to stress-controlled cyclic loading, and showed
that the cyclic stress-strain curve calculated for constant stress
ranges coincides with that for constant strain ranges, as
observed experimentally for 304 and 316 stainless steels.
Therefore, they pointed out the importance of the progressive
development of the cyclic nonhardening range under stress-
controlled cyclic loading.
The present model can be extended to viscoplastic deforma-
tions in the same way as in [20, 9]: Regarding the yield surface
in the rate-independent model as the quasi-static yield surface,
we replace equation (23) with
defj/dt = li<a>do/doij (33)
5 = I (3/2)(J„ - ViJ) (s,j -ViJ)} "2 - K (34)
where a denotes the overstress measured from the quasi-static
yield surface a = 0 , t denotes the time, and JX<X> is ix(x) if
x>0, and zero if x<0. Equation (22b) is appropriate as the
translation rule of the quasi-static yield surface. On the other
•.03 -.02 .01 0 hand, when equation (22a) is used, Sy in this equation should
be replaced with the projection of Sy onto the quasi-static yield
Strain surface, su = 7?,y + (sy — ?j,y)K/(K-fff). We notice that the
Fig. 11 Stabilized stress-strain loops under stepwise expanding of bounding surface is defined for the quasi-static yield surface.
strain range as Ac/2 = 1.0 — 1.5 — • • • • — 3.0 percent (316L stainless
steel [8]) Cyclic hardening of 304 and 316 stainless steels is more
significant under nonproportional cyclic loading than under
that in Fig. 5, although this cyclic curve up to Ae/2 = 0.01 or proportional one [21-25], see also [26] for OFHC copper and
0.02 can be approximated by a linear relation as in Fig. 5. [27] for 1 percent Cr-Mo-V steel. Recently, Tanaka et al. [28]
Let us describe the nonlinearity mentioned above by assum- showed an extension of the present model to the case of non-
ing n(q) as proportional cyclic loading. Nouailhas et al. [29] discussed
K* (q) =«o [1 +Lq/(mK$)}"< (*„*, L, m: constants) (30) this case on the basis of the cyclic nonhardening region and the
Baltov-Sawczuk yield surface. McDowell [30] proposed a
which is reduced to equation (27) when m = 1 or q is suffi-
measure of nonproportional cyclic straining and formulated a
ciently small. Since we are not concerned with cyclic relaxation
constitutive model of cyclic plasticity.
of mean stress under nonzero mean strain here, we neglect the
recovery term for -q*, i.e., Kr = 0 in equation (10). Then the Acknowledgment
monotonic and cyclic relations between a* and eP are obtained
from equations (A9) and 0410), respectively, as follows: The authors are grateful to Professor S. Murakami of
Nagoya University for his encouragement and discussion. The
a* = K0* + (K+L)eP (for eP < <m^/L) (31) second author, Y. Kachi, was associated with the present work
A<7* while studying as a graduate student at Toyohashi University
of Technology.
References

<Kf-"«) +1 ]' (32) 1 Ohno, N., " A Constitutive Model of Cyclic Plasticity With a Nonharden-
ing Strain Region," ASME JOURNAL OF APPLIED MECHANICS, Vol. 49, 1982, pp.
721-727.
where u> = L/(mcnS). Using the above relations and the 2 Mroz, Z., "An Attempt to Describe the Behavior of Metals Under Cyclic
monotonic a - e " relation 0411), we determined the material Loads Using a More General Workhardening Model," Acta Mechanica, Vol. 7,
constants as explained in the Appendix D; K0* = 325 MPa, K0 1969, pp. 199-212.
= 230 MPa, A = 230, K = 350 MPa, L = 1750 MPa, m = 3 Kreig, R. D., " A Practical Two Surface Plasticity Theory," ASME JOUR-
NAL OF APPLIED MECHANICS, Vol. 42, 1975, pp. 641-646.
0.40, p 0 = 0.0034, and c = 0.08, where the value of c was 4 Dafalias, Y. F., and Popov, E. P., " A Model of Nonlinearly Hardening
determined from the observation that the a ~ e loop for Ae/2 Materials for Complex Loading," Acta Mechanica, Vol. 21, 1975, pp. 173-192.

Journal of Applied Mechanics JUNE 1986, Vol. 53/401

Downloaded From: http://appliedmechanics.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 5 Dafalias, Y. F., and Popov, E. P., "Plastic Internal Variables Formalism Cyclic Plasticity: Part 1—Development of Appropriate Equations: Part
of Cyclic Plasticity," ASME JOURNAL OF APPLIED MECHANICS, Vol. 43, 1976, 2—Comparison of Theory With Experiments," ASME JOURNAL OF APPLIED
pp. 645-651. MECHANICS, Vol. 52, 1985, pp. 298-308.
6 Hashiguchi, K., "Constitutive Equations of Elastoplastic Materials With
Anisotropic Hardening and Elastic-Plastic Transition," ASME JOURNAL OF A P -
PLIED MECHANICS, Vol. 48, 1981, pp. 297-301.
7 Dafalias, Y. F., " A Novel Bounding Surface Constitutive Law for the APPENDIX
Monotonic and Cyclic Hardening Response of Metals," Transactions of the 6th
International Conference on Structural Mechanics in Reactor Technology, A Evolution of Cyclic Nonhardening Range. Let us con-
Paper No. L 3/4, Vol. L, Paris, Aug. 1981. sider cyclic straining between fixed limits eg,in and eg,ax. The
8 Chaboche, J. L., Dang Van, K., and Cordier, G., "Modelization of the
Strain Memory Effect on the Cyclic Hardening of 316 Stainless Steel," Transac- range g < 0 at the beginning of the nth cycle is [e£,ax - 2p„_ l,
tions of the 5th International Conference on Structural Mechanics in Reactor 6^ ax ]. (In Fig. 2, RB is the range g < 0 at the beginning of the
Technology, Paper No. L 11/3, Vol. L, Berlin, Aug. 1979. first cycle.) Then, in the first half of the nth cycle where ep
9 Chaboche, J. L., "Viscoplastic Constitutive Equations for the Description changes from e£,ax to e£,in, equation (2) becomes
of Cyclic and Anisotropic Behaviour of Metals," Bulletin de TAcademie
Polonaise des Sciences, Serie des Sciences and Techniques, Vol. 25, 1977, pp.
33-42.
10 Marquis, D., "Etude Theorique et Verification Experimental d'un
f> = \ (,41)
Modele de Plasticite Cyclique," These Paris, VI, 1979. [clF\, eg l i n <e"<eg l a x -2 / 0 „_ 1
11 Chaboche, J. L., and Rousselier, G., "On the Plastic and Viscoplastic With AtP = eg,ax - e£,in, the above equation is rewritten as
Constitutive Equations—Part I: Rules Developed With Internal Variable Con-
cept—Part II: Application of Internal Variable Concepts to the 316 Stainless P„-o.5-P„-i=c(A^-2p„_1) 042)
Steel," ASME Journal of Pressure Vessel Technology, Vol. 105, 1983, pp.
153-164. In 041) and 042), pn_1 and p„_0.5 denote the values of p at the
12 Dafalias, Y. F., "Modeling Cyclic Plasticity: Simplicity Versus beginning and end of this first half cycle. For the latter half,
Sophistication," Mechanics of Engineering Materials, ed. by Desai, C. S., and
Gallagher, R. H., Wiley, 1984, pp. 153-178.
we have
13 Ohno, N., and Kachi, Y., "Description of Stress- and Strain-Controlled
Pn-P„-o.5=c(A^-2Pn-o.5) ^43)
Cyclic Plasticity Using the Cyclic Nonhardening Region Model," Transactions
of the 8th International Conference on Structural Mechanics in Reactor If 0 < c < < 1, we obtain from 042) and 043) the approxima-
Technology, Vol. L, Brussels, Aug. 1985, pp. 57-64; Res Mechanica (in print). tion dp/dn = p„ — p„_ ( = 2c(Ae" — 2p). Thus, assuming
14 Murakami, S, and Ohno, N., "A Constitutive Equation of Creep Based
on the Concept of a Creep-Hardening Surface," International Journal of Solids
p„ = 0 = 0, we have
and Structures, Vol. 18, 1982, p. 597-609. 2p/Ae" = l - e x p ( - 4 c n ) 044)
15 Ohno, N., Murakami, S., and Ueno, T., " A Constitutive Model of Creep
Describing Ceep Recovery and Material Softening Caused by Stress Reversals,"
ASME Journal of Engineering Materials and Technology, Vol. 107, 1985, pp.
B Monotonic and Cyclic Relations (Model I). Under
1-6. monotonic tensile loading, equation (6)2 together with (l)-(4)
16 Mr6z, Z., "On the Description of Anisotropic Workhardening," Journal gives q = 0 for e" < p 0 and q = eP - p0 for (P > p 0 .
of the Mechanics and Physics of Solids, Vol. 15, 1967, pp. 163-175. Therefore, if Kr = 0, we obtain from equation (8) the mon-
17 Krempl, E., "An Experimental Study of Room-Temperature Rate Sen- tonic tensile relation of Model I:
sitivity, Creep and Relaxation of Type 304 Stainless Steel," Journal of the
Mechanics and Physics of Solids, Vol. 27, 1979, pp. 363-375.
fK(0) + [K+(dK/dq)0]eP, e"<p0
18 Ikegami, K., and Nittsu, Y., "Experimental Evaluation of the Interaction
Effect Between Plastic and Creep Deformation," Engineering Fracture a=] 045)
Mechanics, Vol. 21, 1985, pp. 897-907. [KeP+(dK/dq)0p0 + K(q), e">p0
19 Kawashima, K., and Yoshida, T., "Cycle-Dependent Change in Plastic
Stress-Strain Curves," Proceedings of the 1982 Joint Confeence on Experimen- where (dK./dq)0 denotes the value of dn/dq at q = 0.
tal Mechanics, Part I, SESA and JSME, Hawaii, May 1982, pp. 371-376. Let us consider cyclic straining between fixed strain limits.
20 Perzyna, P., "Thermodynamic Theory of Viscoplasticity," Advances in The stabilized size of cyclic stress range is expressed as
Applied Mechanics, ed. by Yih, C. S., Vol. 11, 1971, pp. 313-354.
21 Nouailhas, D., Policella, H., and Kaczmarek, H., "On the Description of Aos = Ais + 2K(qs) (A6)
Cyclic Hardening Under Complex Loading Histories," Proceedings of Interna-
tional Conference on Constitutive Laws for Engineering Materials, ed. by where the subscript 5 denotes the stabilized state. Using (2) and
Desai, C. S., and Gallagher, R. H., Tucson, Arizona, Jan. 1983, p. 45-49. (6)2, we have q = p/c, i.e., q = (p — p0)/c. In the stabilized
22 McDowell, D. L., "On the Path Dependence of Transient Hardening and state, the range g < 0 occupies the cyclic range of plastic
Softening to Stable States Under Complex Biaxial Cyclic Loading," Pro- strain, i.e., 2ps = Aef, as seen from 044). Therefore,
ceedings of International Conference on Constitutive Laws for Engineering
Materials, ed. by Desai, C. S., and Gallagher, R. H., Tuscson, Arizona, Jan. qs=(l/c)(AeP/2-p0) 047)
1983, p. 125-132.
23 Krempl, E., and Lu, H., "The Hardening and Rate-Dependent Behavior Noting that T = 0 in a whole cycle in the stabilized state, we
of Fully Annealed AISI Type 304 Stainless Steel Under Biaxial In-Phase and obtain Ar^ from (J a) as
Out-of-Phase Strain Cycling at Room Temperature," ASME Journal of
Engineering Materials and Technology, Vol. 106, 1984, pp. 376-382. Ar,s = [K+(dK/dq)s]AeP (AS)
24 Ohashi, Y., Kawai, M., and Kaito, T., "Inelastic Behaviour of Type 316
Stainless Steel Under Multiaxial Nonproportional Cyclic Stressings at Elevated where (dn/dq)s denotes the value of dn/dq with respect to qs.
Temperature," ASME Journal of Engineering Materials and Technology, Vol.
107, 1985, pp. 101-109.
C Monotonic and Cyclic Relations (Model II). Since the
25 Tanaka, E., Murakami, S., and Ooka, M., "Effects of Plastic Strain yield surface in Model I is regarded as the bounding surface in
Amplitudes on Non-Proportional Cyclic Plasticity," Acta Mechanica, Vol. 57, Model II, equations 045)-048) for Model I result in the rela-
1985, pp. 167-182. tions for the bounding stress a* in Model II by replacing a, K,
26 Lamba, H. S., and Sidebottom, O. M., "Cyclic Plasticity for Nonpropor- and ij with a*, K*, and i\*, respectively. Therefore, for
tional Paths: Part I—Cyclic Hardening, Erasure of Memory, and Subsequent
Strain Hardening Experiments: Part 2—Comparison With Predictions of Three monotonic tensile loading,
Incremental Plasticity Models," ASME Journal of Engineering Materials and
Technology, Vol. 100, 1978, pp. 96-111. rK*(0) + [K+(dK*/dq)0]e>>, e"<p0
27 Kanazawa, K., Miller, K. J., and Brown, M. W., "Cyclic Deformation of • o* = \ 049)
1% Cr-Mo-V Steel Under Out-of-Phase Loads," Fatigue of Engineering lKeP+(dK*/dq)0p0 + K*(q), e">p0
Materials and Structures, Vol. 2, 1979, pp. 217-228.
28 Tanaka, E., Murakami, S., and Ooka, M., " A Constitutive Model of
Cyclic Plasticity in Multiaxial Non-Proportional Loading," Transactions of the and for cyclic straining between fixed strain limits,
8th International Conference on Structural Mechanics in Reactor Technology,
Vol. L, Brussels, Aug. 1985, pp. 71-77. Aas* = [K+ (dK*/dq)s]AeP + 2K*(qs) (AW)
29 Nouailhas, D., Chaboche, J. L., Savalle, S., and Cailletaud, G., "On the where qs is given by 047).
Constitutive Equations of Cyclic Plasticity Under Non-Proportional Loading,"
International Journal of Plasticity, Vol. 1, 1985, pp. 317-330. Under monotonic tensile loading, we have ij = a — K0 from
30 McDowell, D. L., " A Two Surface Model for Transient Nonproportional (11), so that (12)! results in da/dtp = A (a* - a). Therefore,

402/Vol. 53, JUNE 1986 Transactions of the ASME

Downloaded From: http://appliedmechanics.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 for the linear a* - ep relation (28) or (31), we obtain the Kr = 0 tentatively, we determined K*(q), p0, and K by ap-
monotonic a — ep relation proximating the experimental data of monotonic and cyclic
a=K0 + (K+L)eP stress-strain curves with the bounding stress relations (28) and
(29), or (31) and (32), (dashed lines in Fig. 5 or 10). The values
+ [KS-Ko-(K+L)A-l][l-ex.p(-Ae?y\ (All) of K0 and A were obtained by simulating the transient
elastoplastic deformation after initial yielding with (A 11) (Fig.
D Determination of Material Constants. Constitutive rela- 5 or 10).
tions (9)-(13) together with (l)-(4) have a material function Finally, the value of Kr was adjusted so that (9)-(13)
K* (q) and constants c, p 0 , K, K0, A, and Kr. together with (l)-(4) gave the best fitting of the experimental
The function and constants above, except Kr, can be deter- data of cyclic equalization of positive and negative peak
mined from monotonic and cyclic stress-strain curves together stresses under nonzero mean strain (Fig. 6). Incidentally, it
with ns, the number of cycles for the saturation of strain- was found that the tentative assumption Kr = 0 influenced lit-
controlled cyclic hardening. By using (.44), the value of c was tle on the values of the other constants as far as the data in
first estimated from the data of ns (Fig. 6). Then, supposing Figs. 5 and 6 were concerned.

.Readers Of_
The Journal Of Applied Mechanics
Will Be Interested In:
The books listed here contain papers presented at the 11th Design Automation Conference/10th Biennial Conference on Mechanical
Vibration and Noise/6th Biennial Conference on Failure Prevention and Reliability, September 10-13,1985, Cincinnati, Ohio.

Vibration Analysis to Improve Reliability and Reduce Failures

Editors: P.M. Niskode, RE. Doepker

Contents Include: Vibration Studies of Freestanding Electronic Enclosures, Resonant Vibration of Rotor Teeth in Induction
Motors, and Analysis of the Velocity Response of a Spherical Roller Bearing.
1985 Book Number H00331 List Price: $20.00 ASME Members: $10.00

Failure Prevention and Reliability—1985

Editor: P.A. Doepker

Contents Include: The Role of Fracture Mechanics Relative to Brittle Tool Failure, Damage Detection and Location in
Complex Structures, and Composite Unit Accelerated Life Testing.
1985 Book Number H00332 List Price: $30.00 ASME Members: $15.00
Descriptions of other volumes of interest appear on pages 332, 346, 353, 360, 385, 394, 411, 423, and 449

Address Orders To:

AM159
ASME Order Department/22 Law Drive/Box 2300/Fairfield, NJ 07007-2300

Journal of Applied Mechanics JUNE 1986, Vol. 53/403

Downloaded From: http://appliedmechanics.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use