You are on page 1of 17

.I. Mech.Phys.Solids Vd33,No. 6,pp.559-575.

1985 0022-5096185
$3.00+O.OO
Printed in Great Britain. 01985 PergamonPressLtd.

EFFECTS OF STRAIN PATH SHAPES ON NON-


PROPORTIONAL CYCLIC PLASTICITY

E. TANAKA, S. MURAKAMI and M. OOKA


Department of Mechanical Engineering, Nagoya University, Chikusa-ku, Nagoya, 464 Japan

(Received 28 November 1984)

ABSTRACT

THE TITLEproblem was discussed to facilitate the formulation of constitutive models of cyclic plasticity under
general states of loading. A series of plastic strain controlled cyclic tests was performed by applying combined
axial force and torque to thin-walled tubular specimens of Type 316 stainless steel at room temperature.
These tests consist of cyclic loading along uniaxial, torsional, cruciform, stellate in eight directions, square
and circular plastic strain paths with a constant amplitude of equivalent plastic strain.
The results of these tests showed that the strain-hardening depends markedly on the shape of the plastic
strain path, and that the strain-hardening (measured by equivalent stress amplitudes) in the saturated state is
significant in the order of circular, square, stellate, cruciform and proportional paths. It was also observed
that these saturated values were independent of the less significant plastic strain cycles experienced in the
past. Finally, the characteristic features of strain-hardening mechanisms under non-proportional loadings
were discussed in some detail on the basis of the present results.

1. INTRODUCTION

CYCLIC plastic behaviour of polycrystalline metals under general multiaxial loading


conditions has been one of the most important objectives of modern solid mechanics,
expecially in relation to the current demand for reliable life predictions of inelastic
structures. MARJANOVIC and SZCZEPI~SKI (1974, 1975), for example, discussed the
change of yield surfaces under multiaxial non-proportional cyclic straining, whereas
LAMBA and SIDEBOTTOM (1978), KANAZAWA, MILLER and BROWN (1979), MCDOWELL
(1983) and OHASHI, TANAKA and OOKA (1985) elucidated the dependence of multiaxial
cyclic hardening on the phase angles between component strain cycles. Cyclic plastic
behaviour under different stress paths in stress controlled tests, furthermore, was
observed by SLIWOWSKI (1979) and OHASHI, KAWAI and KAITO (1985). These studies
succeeded in showing that non-proportional cyclic loadings, in general, induce much
larger hardening than proportional ones, and that cyclic loadings along a circular path
erase the history of the preceding cycles of smaller hardening. Nevertheless, the
information available from these tests is rather limited and is far from systematic for the
rational modelling of the complicated phenomena of cyclic plasticity.
A systematic approach to the formulation of constitutive equations of cyclic
plasticity under general multiaxial loading conditions will first require exploration of
the microstructural mechanisms of strain-hardening under such cycles, and then the
representation of the predominant microstructural changes responsible for the
hardening in terms of proper mechanical variables. In order to facilitate this approach,
559
560 E. TANAKA. S. MURAKAMI and M. OOKA

we must perform a series of systematic and carefully organized experiments, and


discuss thoroughly the essential features of cyclic plastic behaviour of the materials ; for
example :

(1) Dependence of cyclic hardening on the shape of loading path, as well as on the
history of different paths ;
(2) Dependence of cyclic hardening on the strain or stress amplitude for a specific
loading path, and on the history of variable amplitude;
(3) Equivalence of cyclic hardening among the strain or the stress paths of different
shapes.

The present paper is concerned with the detailed experiments on the first feature
mentioned above, with a view to establrshing a constitutive equation of cyclic plasticity
applicable to arbitrary multiaxial loading conditions. Namely, by choosing seven
representative strain paths in a plastic strain space of combined tension-compres-
sion and torsion, and by specifying a constant amplitude of equivalent plastic strain
(AeP/2 = 0.279, we first performed a series of plastic strain controlled cyciic tests for
Type 316 stainless steel tubes at room temperature. The strain paths include
tension-compression, torsional, cruciform (1) cruciform (II). stellate in eight direc-
tions, square and circular path and the effects of different sequences of these paths
were also elucidated.
The second feature of cyclic plasticity mentioned above has hecn already discussed in
a separate paper of the present authors (TANAKA. MUKARAMI and OOKA. 1085).

The specimens employed in the present tests were ‘lype 316 stainless steel tubes of outside‘
diameter 21 mm, thickness 1 mm and gauge length 60 mm. The specimen maternal was subjected
to solution heat treatment at 1100°C before machining, and its chemical compositions were 0.05
C, 0.27 Si, 1.65 Mn, 0.034 P, 0.028 S, IO. I2 Ni, 16.24 Cr, 2.04 MO and Fe balance in weight “o
The multiaxial cyclic tests were performed at room temperature by use of a minicomputer
controlled combined load testing machine (TANAKA :’ 1 trl.. 1985). The ;~xial force and torque
acting on a specimen were detected by the corresponding load cells incorporated in the machine,
while the axial, circumferential and torsional shear strain components on the outer surface of
specimens were detected by use of three-axes rosette type metal foil gauges. On the basis of them.
we calculated the mean values of the axial and torsional shear components of the Euler stress
tensor (true stress tensor), and those of the axial. circumferential and shear components of the
Almansi strain tensor.
For convenience in the following discussion, the strain and the stress state are represented in
terms of the plastic strain vector ep and the stress vect.or u in the plastic: strain space (6”. ;;” L .?J
and the deviatoric stress space (IYJ,
,,/5 T), respectively. detined as follows (1L’~I!SHIN, 196.7):

II =: m, t (v’3~)n,. (-?i

where the plastic strains are assumed to he isochoric. fhc 3vmbols n , and nz m equations ( I j and
(2) denote a set of orthonormal base vectors in the deviatoric vector spaces. and :,p. ;“I, 0 and 5 arc
axial plastic strain, twice torsional plastic strain. axial stress and torsmnai stress. respcctivelv.
Non-proportional cyclic plasticity 561
Then, the magnitudes of ep and u, i.e. ]ePland ]u), correspond to the equivalent plastic strain and
the equivalent stress of von Mises type. The trajectory of the tip of the vector eP with respect to
time t (piastic strain path) furthe~ore represents uniquely the plastic strain history, and the
length of the path sp corresponds to the accumulated plastic strain ; this is defined by the equation

sp = ’ IdeP/dt( dt. (3)


s0
The plastic strain components were evaluated from the total strain by subtracting the elastic
strain, which in turn was calculated on the basis of Wooke’s law.
According to the separate paper of the present authors (TANAKAet al., 19SS), it has been
elucidated that cyclic hardening under proportional plastic strain cycles is independent of the
orientation of the loading path, and can be expressed adequately in terms of the equivalent stress
and the equivalent plastic strain of von Mises type. Furthermore, comparison between the stress
amplitude of these proportional cycles with that of circular cycles of the same plastic strain
amplitude showed that the circular cycles bring about cyclic hardening N 1.5-1.8 times as large
as that of the corresponding proportional cycles.
Thus, in view of these results, we will employ the seven fundamental plastic strain paths of Fig.
I (a)-fg) and the seven histories A-G of the strain paths entered in Table 1 for the foilowing tests.
Then, the plastic strain amplitudes of these paths are prescribed to a constant value A&‘/2 =
0.2x, so that the plastic strains attain to their maximum values on the circles of radius 0.2%
entered in Fig. 1.
By comparing Fig. 1 (a) or (b) with (g), the essential features of the difference between the
proportional and the circular strain path can be found in ;
(I) The difference in the frequencies of the peak values of the plastic strain, as well as in ;
(2) The difference in relative direction of plastic strain rate vector to that of the plastic strain
vector, observed in the process of these strain cycles.

(a) Tension-Compression

(e)SteIlate ( f ) Square (g)Circular

FIG 1. Cyclic plastic strain paths.


E. TANAKA, S. MUKAKAMIand M. OOKA

TABLE 1. Histories c$‘&ferent cyclic plastic strain paths

Thus, the simple cycles (Fig. 1 (a), (b)), the cruciform cycles (Fig. 1(c),(d)) and the stellate cycles in
eight directions (Fig. 1 (e)) are employed to discuss the effects of the first point mentioned above,
and the square and the circular cycles of Fig. 1 (f) and (g) illustrate the second point. The two
simple cycles of Fig. 1 (a),(b), on the other hand, are adopted to confirm the isotropy of the plastic
strain space of Fig. 1, while the two cruciform cycles of Fig. (c),(d) are to elucidate the effect of the
frequency of the change in plastic strain rate vector direction per one cycle.
In this context, the tests A- G of Table I are designed to elucidate the relation of the cyclic
hardening properties to the shapes of the plastic strain paths as well as to the history of the
sequences of these paths. Then, the cyclic loading in each stage (i.e. in each strain path) of the tests
A-G was continued until stabilized behaviour was almost attained. The corresponding
accumulated plastic strains .sJ’for the proportional cycles of Figs 1 (a) and (b) were more than
30’:,,, while those of the more complex cycles in Fig. 1 were more than 605,. The plastic strain rate
throughout the tests A- G was prescribed as a constant value .9’ = 3 x 10 ’ s ‘.
In the following discussion, we will denote each stage of the tests A G by symbols A- 1,A-2, etc.
as indicated in Table 1. The cyclic tests of Table I, however. are insufficient to discuss the effect of
preceding strain paths of larger cyclic hardening (cyclic recovery), and the effects of asymmetric
plastic strain paths (cyclic relaxation). These problems will be discussed in our future work.
Finally, we need the values of Young’s modulus E and the elastic shear modulus G to perform
plastic strain controlled tests : we evaluated them from elastic tests before performing tests A--G.
Their mean values are
E = 203 GPa, G = 78 GPa. (4)
Non-proportionalcyclic plasticity 563

3. DEPENDENCE OF CYCLIC HARDENINGON THE SHAPE OF PLASTIC STRAIN PATHS

Reproducibility of plastic strain paths and a measure of cyclic hardening

In order to clarify the cyclic property of virgin materials under simple plastic cycles,
we first performed three cyclic torsion tests for strain amplitudes of AeP/2 = 0.1,0.2 and
0.4%. Figure 2 shows the stress-plastic strain curves for these tests. As observed from
this figure, the relevant material is a cyclic hardening material, but the hardening is not
very significant.
The discussion of the dependence of cyclic behaviour on strain path shapes requires
us, first of all, to confirm the reproducibility of the prescribed strain paths, as well as to
define a proper measure of strain hardening. Figure 3 (aHe) shows the plastic strain
paths and the corresponding stress paths for the cyclic tests of C-l, D-l, E-l, F-l and G.
The numbers 0, 0, @ etc. in these figures represent the peak values of the equivalent
plastic strain on each strain path and the corresponding points on stress paths, and thin
circles in these strain spaces are those with a constant radius AeP/2 = 0.2%.
As observed in these figures, the square and the circular path of Fig. 3, (d), (e) are
realized accurately, while some deviations are observed on the strain paths of the
cruciform and the stellate cycles of Fig. 3 (aHc). However, the peak values of the
equivalent plastic strain on the latter paths take the prescribed value of de”/2 = 0.2%
and these peak values correspond accurately to the peak values in the stress space.
Hence, the deviations observed in the strain paths of Fig. 3 (aHc) have no significant
influence on the following discussion.

G5z MPa
400
r
Virgin Material

FIG.2. Stress+plasticstrain curvesin torsion cyclesfor plasticstrain amplitudesAeqi2= 0.1, 0.2 and 0.47”.
xi4
tipa
*to

; -p
._

L
-0.3

(a)

3
(-4
i.. ‘%_._
-0.3 -0.2

/ ..
/ “@
ib) -0.3’-
Non-proportional cyclic plasticity 565

(d)

(e) -0.3L -6OOL

FIG. 3 (continued).

Now, let us pass to the discussion of the measure of cyclic hardening. The cyclic
hardening in the simple strain paths of Fig. 1 (a), (b) can be reasonably defined by the
stress amplitude Aa,/ or A.a2/2 (da,, Aa, ; range of a stress component cr or o2 in each
cycle). As regards the strain cycles of Fig. 1 (c)-(g), on the other hand, the peak values of
the equivalent stress observed in every one cycle of the corresponding stress path (e.g.
four peak values in the case of Figs 3 (a) and (b)) are almost identical. Thus, we will
adopt the average value of the equivalent stress at peak points throughout one cycle as
a measure of cyclic hardening under non-proportional cycles; this definition is in
conformity with the above-mentioned definition of stress amplitude for the simple
cycles. In the following, we designate them as the stress amplitude Aa/2.
566 E. TANAKA,S. MCIKAKAMI
and M. OOKA

Dependence of’cvclic hurdeniny on the shape of’plustic struin path

Figure 4 (a)-(g) shows the variation of the equivalent stress observed in the
experiments of Table 1. Then, the cyclic hardening properties under strain paths of Fig. 1
can be discussed by establishing the relations between the stress amplitude Aa12 and the
accumulated plastic strain sp by use of the experimental results A- I, R- 1, C- 1.. . . G of
Fig. 4; this is shown in Fig. 5. One of the essential features of this figure is the marked
dependence of the cyclic hardening on the strain path shape, although every curve
starts from an almost identical value of Aoj2 in the incipient stage. Namely, the cyclic
hardening is the least significant under simple torsion, then becomes more saiient in the
order of tension-compression, cruciform (I) or (II), stellate and square path, and the

0 20 40 60 a0 100 120 140

(a) sp %

1 “C 40 i$C 80 100 IC? IJC i6r? 18 0 '2130 220 rsi; 2 t;17 ;5 : -r

(b) sp %

FIG. 4. Variations in equivalent stress for tests A-G (A@‘/2 = 0.27,). (a) test A,(b) test B, (c) test C, (d) test D,
(e) test E, (f) test F. (g) test G.
600
r 567

0 20 40 60 80 100 120 140


Cc) sp 70

600

500
r --

D-2

,400
a
I
300
5
200

100

-I

0 20 40 60 80 100 120 140 160 180


fd) sp %

600

0 20 40 60 80 100 120 140 160 180 200 220


(e) sp %
FIG. 4 (continued).
568 MUKAKAMIand M. OOKA
E.TANAKA,S.

500

*400
2
300
I
e?
200 I
/

0 20 40 60 80 100 120 140 16n 180 290 ‘33 1-j


__c
(0 sp %

600 - _-._____- ______ ~.._ ~____._ .__.___~ _ _ ._____..~ .~ _

7
500

@a400
CL
z:
I
,30°

200

loo-

I I I I I i L I I
0 10 20 30 40 50 60 70 80 90 100 Ii0
(9) sp %

FE. 4 (mirinued).

largest hardening is attained by the circular path. The difference of about 4.57(,
observed in do/2 of the two simple paths A-l and B-l is attributable to the effect of the
third invariant of the deviatoric stress tensor (OHASHI,TOKUDA and YAMASHITA 1975).
Ifwe disregard this second order effect, it has been ascertained that the proportional
cycles of combined tension-compression and torsion show the identical hardening
with that of the simple tension-compression and the simple torsion cycles (LAMBA and
SIDEROTTOM, 1978; BROWN and MIL.LER, 1979; NOUAILHAS, POLICELLA and
KACZMAREK ~%~;TANAKA et al., 1985).
Non-proportional cyclic plasticity 569

The second salient feature of Fig. 5 is that the cyclic hardening curves in this figure
can be classified according to the degree of hardening into three groups : (1) simple
paths, (2) cruciform and stellate paths, and (3) square and circular paths. Namely, the
cyclic hardening induced by the cruciform and the stellate cycles is nearly the same ; this
is about 1.5 times as large as for the simple cycles. Regardless ofsome differences in their
path shapes, the square cycles on the other hand, show nearly the same hardening as
that of the circular cycles; they lead to about 1.7 times the hardening as for the simple
cycles.
Now, let us examine the strain paths of the second group mentioned above in some
detail. The experimental results for the strain paths C-l and D-l, first of all, imply that
intersection of the radial paths brings about much larger hardening than the simple
proportional paths of A-l and B-l. If we compare the results for the cruciform paths
C-l (or D-l) with those of the stellate path E-l, it will be noticed that the significant
hardening observed in C-l (or D-l) cannot be accounted for merely by the numbers of
the peak points appearing in one cycle, because the difference in the hardening curves of
these paths is insignificant regardless of the large difference in the number of the peak
points in them. The coincidence of the results for C-l and D-l, furthermore, implies that
the frequency of the change in strain rate vector direction per one cycle has no effect,
provided that the strain path shapes are identical. By recapitulating these results, we
will see that the intersection of radial strain paths brings about marked cyclic
hardening, but the number of changes in the strain rate vector direction per one cycle
for a specific path shape or the number of the intersecting segments have rather small
effects on the hardening.

600

500 Virgin Material, AeP12 = 0.2 %

0 10 20 30 LO 50 60 70 80 90 l(
sp %

FIG. 5. Effects of the shapes of plastic strain paths on cyclic hardening.


570 IJ.TANAKA. S. MURAK.O.U and M. OOKA

Moreover, in view of the experimental curves of the square and the stellate paths in
Fig. 5, wc should note that these hardenings again cannot be accounted for by the
number of the intersecting segments or by the number of changes in strain rate vector
direction in one cycle alone, because the square path apparently shows large1
hardening although it has a smaller number of intersecting segments and changes ot
strain rate vector direction. The fact that the square path shows nearly the same
hardening as that of circular path suggests that the marked hardening due to these
paths should be attributed to another mechanism; this is related to the relatilc
direction of the plastic strain rate vccto~. to that of the plastic strain vector itself.
To summarize the above discussion, the cyclic hardening under non-proportional
loading is governed mainly by three types of loading: (1) strain cycles in one fixed
orientation, (2) strain cycles along the intersecting segments. and (3) strain cycles with
local non-colinearity between the plastic strain rate vector and the corresponding
plastic strain vector in each instant. These three types of loading. of course. correspond
to the three groups of the cyclic stress-strain curves of Fig. 5 mentioned above. Since
the cruciform and the stellate cycles also include the rotation of the strain rate vector at
their origins. the hardening observed in these cycles can be related to the third type 01
loading to some extent.
The above observations concerning cyclic hardening will give essential information
for the mechanical modelling of the cyclic plastic behaviour of polycrystalline
materials. From the microstructural point of view, the salient hardening under non-
proportional loadings can be accounted for by the formation of a highly immobilized
dislocation structure in cotnparison with the stable structure due to proportional
cycles. The former structure results from the j/ariation of actIke slip systems and the
complicated dislocation interactions in these systems (KAXUAW,~ it I//.. 1979:
MCDOWELL, 1983). Therefore, the formulation of an appropriate constitutive equation
of cyclic plasticity for practical USCwill require us to incorporate the salient features
discussed above by taking account of these microscopic mechanisms.

4. DEPENDENCE 01: CY(T’LIC:HAKIXNING ON TN: IHISTORYw SRAIN PATH SHAPI:S

Cyc~lic burdening under srquences qt’d[# bent strain path shtrprs

We are now in a position to elucidate the effects of the past plastic strain cycles of
different path shapes on the succeeding hardening under non-proportional strain
cycles, For this purpose, we performed a series of cyclic tests summarized in Table 1.
where the plastic strain cycles in each stage (i.e. in each strain path) were repeated until a
stabilized state was almost attained. Figure 6 shows the relation between the stress
amplitude, Aoj2 and the accumulated plaslic strain s P for test B; this is a typical
example of the present tests. The corresponding relations for the other tests in Table 1
can be constructed from Fig. 4.
As seen from this figure, a marked discontinuous increase In AC/~ is observed
immediately after the torsion cycles B-l were changed to the tension- compression
cycles B-2. This cross hardening is found also in test A when the tensionPcompression
cycles A-l were changed to the torsion cycles A-2 (see Fig. 4 (a)). This discontinuous
Non-proportional cyclic plasticity 571

hardening caused by a sudden change in the strain cycle direction has been reported
also in Type 304 stainless steel at room temperature (KREMPL and Lu, 1984) as well as in
Type 316 stainless steel at 600°C (OHASHI et al., 1985); this cross effect can be accounted
for by a large resistance brought about by a stable dislocation structure formed in the
preceding proportional cycles to the dislocation movement in the subsequent cycles in
another direction. Comparison of this phenomenon with the cyclic hardening in test
C-l suggests the essential role of the stable dislocation structure and the dislocation
intersections in cyclic plastic behaviour under non-proportional loadings.
In stage B-2 after the path change, strain cycles are again proportional. Thus, except
the first few cycles, even the cyclic softening, rather than the cyclic hardening, occurs in
the major part of this stage. The material behaviour of stage B-l and B-2 clearly shows
that even if two certain cyclic strain paths are congruent but different only in their
orientations, their combination causes additional transient hardening.
Let us now consider the strain path change from the tension-compression cycles B-2
to the cruciform cycles (I). The increase in k/2 due to the path change is insignificant,
and the cyclic hardening advances gradually in the succeeding stage, B-3. This
hardening process can be accounted for by the formation of another dislocation
structure more complicated than that of the stage B-2. In view of the fact that the path
B-3 is composed of paths B- 1 and B-2, the cyclic hardening observed in B-3 implies that
the essential factors governing the cyclic hardening are not mere superpositions of the
component paths, but the relevant strain path of the cycles itself.
Furthermore, as regards the transition from stage B-3 to B-4, no significant effect on
the cyclic behaviour is observed. This fact suggests the similarity in the internal

FIG. 6. Effects ofchangesin plastic strain paths on cyclic hardening (test B).
572 E. TANAKA, S. MI~RARAMI and M. OOKA

Circular Cycles, AeP12 = 0.2 %

l B-5 I
z!
t
0 C-2

0 D-2

0 E-2

ID F-2

I
0 G ( For Virgin Material )
I I 1 I I / I I i
0 10 20 30 40 50 60 70 80 90 100
(a) sp 01.

500
Cruciform Cycles, AeD/ = 0.2 %
t

I? 400
I
I
,N 300 m C-l ( For Virgin Material )
b
-3
200
0 D-l ( For Virgin Material )
100 l B-4

0 10 20 30 40 50 60 70 80 90
(b) sp “/.

FIG. 7. Effects of the preceding cyclic histories on cyclic hardening. (a) Circular cycles, (h) cruciform cycles.

structures formed in these cycles, regardless of the difference in the number of changes
in strain rate vector direction per one cycle in these paths. In other words, as far as the
strain path shapes are identical, minor differences in the order of loading along the
component segments have no significant effects on the cyclic behaviour.
Finally, the significant increase in Aaj2 caused by the transition from B-4 to B-5
means the formation of another microstructure providing further resistance for the
dislocation motions. This path change, furthermore, induces a temporary softening of
the material, as seen in Fig. 6. This phenomenon is found also in the other cases of the
transition to the circular cycles (see Fig. 4 (btif)). Hence. when the strain-hardening
Non-proportional cyclic plasticity 573

caused by preceding cycles has attained a high level, the abrupt change in the direction
of the strain rate vector due to path changes can induce a temporary but significant
softening of materials.

Efects of preceding strain cycles on the saturated strain-hardening of succeeding cycles

Now, let us discuss the effects of the preceding strain cycles of smaller hardening on
the saturated state of hardening in the succeeding cycles of larger hardening. For this
purpose, we take a set of circular cycle tests with various past histories. Figure 7 (a)
summarizes the relations between the stress amplitude A0/2 and the accumulated
plastic strain .sJ’for the tests A-3 through G, where sp is measured from the start of these
circular cycle tests. As observed from this figure, the curves for different histories
approach the curve G for the virgin material and coincide with it at the saturated state,
although their initial values differ according to the variety of their preceding cycles.
This fact indicates that the dislocation structures finally formed by these circular cycles
tend to coincide with each other irrespective of their preceding histories; in other
words, the effects of the preceding histories tend to be erased by the succeeding cycles of
larger hardening. Similar features are observed also in the case of the two cruciform
cycles entered in Fig. 7 (b).
The observations obtained in the present paper will be interesting not only from a
metallurgical point of view, but also furnish quite essential information for the
modelling of appropriate constitutive equations of cyclic plasticity under general
multiaxial loading.
Recently, OHNO (1982) proposed the notion of a non-hardening strain region to
formulate a constitutive equation of cyclic plasticity, and succeeded in describing cyclic
behaviour under arbitrary uniaxial cycles. The results of Fig. 7 (a) and (b) suggest the
existence of a certain strain region, such that arbitrary strain cycles within this region
bring about no isotropic hardening of the material, and that this region is uniquely
determined by the relevant cycles independently of the preceding cycles of smaller
hardening. Thus, the results of the present paper confirm that the notion of non-
hardening strain regions mentioned above is valid also for non-proportional strain
cycles.

5. CONCLUSIONS

In order to establish essential features of the cyclic plastic behaviour of materials


under non-proportional multiaxial loadings, a series of plastic strain controlled cyclic
tests was performed for Type 316 stainless steel tubes at room temperature. Seven
representative strain paths in a plastic strain space and their combination were
adopted, and a constant equivalent plastic strain amplitude AeP/2 = 0.2% was
prescribed for these cyclic tests. The results of the present work are summarized as
follows :

1. When virgin materials are subjected to plastic strain cycles under these paths up to
the saturated states, the strain hardening in the proportional path is the least
significant, then becomes more salient in the order of the cruciform (I) or (II), stellate
514 E. TANAKA, S. MUKAKAMI and M. oOKA

in eight directions and square, and the largest hardening is attained by the circular
path.
The cyclic hardening curves observed from these tests can be classified according to
the degree of hardening into three groups : (I) simple or proportional paths, (2)
cruciform and stellatc paths, and (3) square and circular paths.
?hesc features suggest that the cyclic hardening under non-proportional loading
can bc characterized mainly by the corresponding three types of loading: (I) strain
cycles in one fixed orientation. (2) strain cycles along strain path composed of
intersecting segments. and (3) strain cycles with non-colinearity between the plastic
strain rate vector and the corresponding plastic strain vector in each instant.
The cyclic hardenings observed for the cruciform (I) and cruciform (II) coincide with
one another. Hence, as far as the strain path shapes are identical, minor differences in
the order of loading along the component segments have no significant effects on the
cyclic hardening.
A significant discontinuous increase in stress amplitude is observed when the strain
path is changed from torsion cycles to tension--compression cycles. This cross
hardening implies the essential role of the stable dislocation structures and the
dislocation intersections in cyclic plastic behaviour under non-proportional
Io:\din;g
When ;vc p~~rI_tvrmed‘I set of circular cycle tests with varinlus histories of preceding
cycles. the strain-hardenings of these tests in the saturated 4tatc coincide with thar ot
virgin mutcrial. Similar features were observed also for the cruciform cycie tests. Thi<
bl<>Imeans that the dislvcation structures formed under va:i!~us not-proportional
cycles tend to be rearranged by the subsequent cycles of larger strain-hardening. and
replaced by a new structure inherent to such cycles.

l‘hc cffcc~s of strain path shapes and of the histories of their sequence on the cyclic
bchaviour of materials discussed in this paper were restricted only to symmetric strain
paths in the plastic strain space and to preceding cycles of less significant hardzling.
The elaborate modelling of :I constitutive equation of cyclic plasticity will further need
the discussion of similar effects of asymmetric strain paths with an offset from the origin
and those of preceding cyclic histories of larger hardening. These problems will be the
objectives of future work of the present authors.

The authors would hkc to express their gratitude to Professor Y. Ohash ofAich1 institute of
‘Technology for his cordial encouragement in this study. They also wish to express their
appreciation for the use of F.4COM System M382 of Nagoya University C’omputatlon Center
for the numerical and graphical processing of the present paper.

REFEKENChS

BROWN. M. W. and MIILLK,


K. J.
IL\iL:SIlIN. A. i-2.
KASALAU’A, K., MILLFK.
K. J. and RKOWN. M. W.
Non-proportional cyclic plasticity 57s

KREMPL, E. and Lu, H. 1984 J. Engng Mat. Tech. 106, 376.


LAMBA, H. S. and 1978 J. Testing Ecaluation 6, 260.
SIDEBOTTOM,0. M.
LAMBA, H. S. and 1978 J. Engng Mat. Tech. lOO,96.
SIDEBOTTOM,0. M.
MARJANOVIC, R. and 1974 Arch. Mech. Stos. 26, 311.
SZCZEPII;ISKI,W.
MARJANOVI~, R. and 1975 Acta Mech. 23, 65.
SZCZEPI~~SKI,W.
MCDOWELL, D. L. 1983 Constitutice Laws,for Engineering Materials (edited
by C. S. DESAI and R. H. GALLAGHER), p. 125.
Proceedings of the International Conference,
Tucson, Arizona, U.S.A. January l@-14.
NOUAII~HAS,D., POLICELLA, H. 1983 Constitutive Lawsfor Engilteering Materials (edited
and KACZMAREK, H. by C. S. DESAI and R. H. GALLAGHER), p. 45.
Proceedings of the International Conference,
Tucson, Arizona, U.S.A. January lC-14.
OHASHI, Y., TOKUDA, M. 197.5 J. Mech. Phys. Solids 23, 295.
and YAMASHIT.~, H.
OHASHI, Y., KAWAI, M. 1985 J. Engng Mat. Tech. 107, 101
and KAITO, T.
OHASHI, Y., TANAKA, E. 1985 J. Engng Mat. Tech. 107 (in press).
and OOKA, M.
OHNO, N. 1982 J. appl. Mech. 49, 72 1.
SLIWOWSKI, M. 1979 Bull. Acad. Polonaise Sci. 27, 115.
TANAKA, E., MURAKAMI, S. 1985 Acta Mech. 55 (in press).
and ~OKA, M.

You might also like