Beruflich Dokumente
Kultur Dokumente
217-228
Pergamon Press. Printed in Great Britain.
Fatigue of Engineering Materials Ltd. 1979.
Abstract- The stable cyclic stress strain response of 1°/0 Cr-Mo- V steel has been studied for
combined axial and torsional loads, where the respective sinusoidal strains were applied
both in and out of phase. Typical hysteresis loops are presented, and a correlation is
proposed for the cyclic stress strain curve.
NOMENCLATURE
D Diameter a Axial stress
E Young's modulus r Shear stress
F Rotation factor </> Phase angle
G Modulus of rigidity w Frequency
k, I, m, n Constants
P Axialload Subscripts
R Correlation coefficient a Amplitude
S Estimate of error i Internal
t Time max Maximum
T Torque o External
ex A phase angle oct Octahedral
y Engineering shear strain pmax Maximum plastic
r. Axial strain
Strain ratio, yalc:a Superscript
v Poisson's ratio Plastic strain equation
INTRODUCTION
Cvcuc stress-strain curves are often determined during laboratory fatigue studies, usually
under uniaxial loading conditions. Whilst most engineering applications involve a complex
stress state, there are comparatively few published studies of a materials cyclic response for
biaxial loading. These have been reviewed recently by Brown and Miller [1], who
considered proportional loading only, i.e. with strains controlled in· a constant ratio.
Recently Kanazawa, Miller and Brown [2] investigated the low cycle fatigue strength of
1% Cr-Mo-V steel under out-of-phase combined axial and torsional loads. This paper
presents the stable cyclic deformation behaviour of these tests.
For proportional loading, Brown and Miller [1] concluded that a unique stress-strain
curve could be determined for stable conditions. It could be obtained by plotting the
maximum shear stress against the maximum shear strain. Kanazawa et a/. [2] reviewed
previous work on non-proportional biaxial fatigue, but of all the references quoted, only
Taira eta/. [3] measured the stresses. They also used combined tension and torsion, with a
217
218 K. KANAZAWA eta/.
90° phase angle, and plotted a figure showing the change of magnitude and direction of the
maximum principal stress, although the actual amplitudes were not given.
The deformation behaviour of OFHC copper was investigated by Lamba [ 4] during
transient cyclic hardening, together with studies of the yield surface shape at cyclic stability.
However, he did not conduct sufficient tests to determine a cyclic stress-strain curve, and,
like Taira et al., only 0 and 90° phase angles were considered.
Apart from these two references, the authors are unaware of any work on stable non-
proportional cyclic deformation behaviour.
EXPERIMENTAL PROGRAMME
Details of specimen geometry, testing machine and extensometry are given in Refs [2]
and [5]. Briefly, the specimen was a tube, 25 mm gauge length, 22 mm external diameter
and 16 mm bore. Both torsional and axial strains were sinusoid;ll, with chosen phase
differences,¢, ofO, 30, 45, 90, 135 and 180°.
The material was 1% Cr-Mo-V steel, heat treated to give an isotropic mid-bainite
structure. The composition and monotonic test results are given elsewhere [1, 2, 6]. All
tests were conducted at room temperature, with a net maximum shear strain rate of
1·5ks- 1 .
The strain analysis, described in [2], assumes a Poisson's ratio of 0·5. Stresses were
derived from the axial load, P, and torque, T, assuming a constant stress distribution,
giving the axial stress,
(1)
and the torsional stress
r = 12T/n(D~ -Dt). (2)
A more accurate stress analysis [1] could not be used since Hencky's equations are
invalid for non-proportional loading. Although errors are introduced by assuming a
uniform stress, these errors cancel out for the maximum shear stress, to give calculated
values less than 2% below the true values for in-phase tests. It is assumed that no greater
deviations arose for out-of-phase loading. The maximum shear stress at any time is given
by
(3)
RESULTS
Figures 1 and 2 illustrate some typical hysteresis loops together with plots of both axial
vs. torsional strain (reflecting the accuracy of the strain control) and axial vs. torsional
stress. The axial strain was given by
£ = £0 sin (wt) (5)
Cyclic deformation of I% Cr-Mo-V steel under out-of-phase loads 219
I ~Yr,
-9d'
'1/7'00 ~
f/--rgq /;
I L '?f. _ _ ,.9Cf
-I -I -I
/ 1/ 0/0"a
I
f (
'reo· //
270°
-q ----
'
2~ lBO"
-1 ~ ~-~~
i7o· 27r5-"
~
~
>
z
A= 4 , ¢"' 90°, E'a=05% >
N
1:/la >
~
h %l
180¢
~9<1 >
1/ A
\
. -18d
05l \ 05
/
-go• ~
~
t:l
:-
,.-9d
0.5 0.5
z717{
I
zro~\ /
-05 f-
i k
~
I _j //
o·
II
(j'
- -I
o• o/a..
90"
05 I 0.5
<8. 05
05
270~
180°
/./
1800 h-Q._
/ "'· 90"
180Q
05 05
27(!'
05 9(!'
·05 ;.-,. -05 o/aa
-o5
"0"
270"
_/
05 ~
-----+--~-
1
o•
o•
<!a·
ELASTIC RESPONSE
For linear elastic behaviour, one may calculate the stresses from the applied strains in
equations (5) and (6), giving
CJ = E1-:u sin (wt) (7)
and
r = G{'u sin ((l)t-cf>). (R)
222 K. KANAZA W A eta/.
A common approach to plasticity is to consider the effective stress and strain, and to
correlate them. In Fig. 3, the variations of the maximum shear stress and strain are plotted
for three different phase angles, together with the octahedral shear stress and strain. The
influence of the phase angle is immediately apparent. Although the maximum stress values
do not vary markedly, the minimum stresses are raised as¢ increases, up to approximately
the maximum value when ¢ = 90°. One would expect the maximum value to be attained
when ;_ = 1 + v, the special case discussed above, since no elastic unloading can occur.
Where unloading does occur, shown by decreasing stresses in Fig. 3, it is linear elastic until
reverse yield produces a deviation from that linear response. It can be seen that the amount
of elastic deformation in a cycle decreases as ¢ approaches 90°.
It is worth repeating the conclusion of Kanazawa et a/. [2] that a phase angle of :t. is
equivalent to one of (180- :t.). Thus a similar stress-strain response was obtained for both
45 and 135 ¢ values, even though the hysteresis loops were markedly different (Fig. 2).
Measurement of the loop areas in Fig. 2 showed that the same amount of plastic work was
done on the specimen in these two cases, and analysis of the strains revealed the fact that
the maximum shear strains were also identical (2).
In Fig. 4 the axial and torsional stress amplitudes are compared for tests with ¢ = 90 .
Although there is some scatter, the results suggest that the octahedral shear stress remains
fixed throughout the loading cycle, giving au= firu. This may also be observed in Fig. 3
where ¢ = 90o. If little or no elastic unloading occurs, one must conclude that the yield
surface follows the von Mises criterion when ¢ = 90'. In contrast, Brown and Miller [1]
found the Tresca criterion was obeyed for in-phase tests, predicting au = 2ru.
In order to compare these two biaxial correlations more closely, the test ~ata have been
replotted in Fig. 5, where the stresses were measured at the instants when the maximum
and octahedral shear strains attained their maximum values respectively. The results show
some divergence in both diagrams, due not only to experimental scatter but also to the
variation of¢. Neither approach can be said to give a single cyclic stress-strain curve.
Brown and Miller [I] suggested that a unique cyclic stress-strain relationship could be
obtained from the Tresca flow rule because of the development of specific slip systems in
Cyclic deformation of I% Cr-Mo-V steel under out-of-phase loads 223
1 ""400 MSS-MSS
MOSS
deg "''
OL-------~---------L--------~--------L-----~
0 05 1.0 15 20
MAXJMJM SHEAR STRAIN. MAXJMtM OCTAI-EDRAL 51-£/!.R STRAIN %
1
~ 400
Ill
~
(J) 300
-~
ii\Vl
~~ 200
(J)jJ
~f'!?5
(J) 100
deg w t
1i 400
deg wt
Fig. 3. Maximum and octahedral shear stress and strain variations during a single cycle.
224 K. KANAZAWA e1 a/.
~
<1:100
Fig. 4. Relationship of axial and torsional stress amplitudes for a phase angle, ¢>. of 90.
metals, in the form of either dislocation cells or planer arrays in most instances. These
dislocation substructures are produced within individual grains due to the repeated
shearing of fixed slip planes aligned with the maximum shear plane.
However, under non-synchronous straining, the rotation of the principal strain axes
during each cycle ensures that the plane of maximum shear also rotates, so that the
preferential slip plane changes from one crystallographic slip system to another. This
probably prevents the development of the stable dislocation structures associated with in-
phase cycling, and therefore one would expect the stress-strain curves for out-of-phase
loading to rise above the in-phase stress levels. Such hardening can be attributed to
dislocation jogs and to intersection of active slip planes, etc., that arise from complex
dislocation movements on the many slip systems of each grain. Figure 5 shows hardening
of this nature, although the trend is obscured a little by experimental scatter.
In the special case of¢ = 90° and),= l + v, the maximum shear strain range is the same
on all planes, giving the same degree of plasticity on each. Therefore one might expect the
von Mises criterion to govern plastic flow, since it may be derived from a statistical analysis
of yield in an isotropic polycrystalline material [1]. Thus in Fig. 4 those Jc values close to
( l + 1') fall close to the octahedral prediction, but for ), = 4, one obtains the two points
which lie below the von Mises line, the axial stress amplitude being reduced by elastic
unloading.
This discussion suggests that a unique stress-strain curve might still be expected if one
could correlate the additional hardening due to rotation of the principal axes. One
approach is to define a rotation factor, F, in terms of the amount of slip experienced by
critical planes in the specimen, so that
shear strain range at 45'' to maximum shear plane
F=------=-
maximum shear strain range
Cyclic deformation of I% Cr- Mo-V steel under out-of-phase loads 225
i>;'(e*o"'
A= 0 1.5 2 4 DC
0~--------~--------~~--------~--------~------~
0 05 (5 2
f\AAXIMUM S/-EAR STRAIN AMPLITUDE %
1
~
Lu 400-
¢.:<;;
f-~
<J:f-
(f)
~~300f--
~~
r:r:<i
L;)rl:: 2001-
~~
--'~
~~ 1001- A= 0
"*•*ol!l
1.5 2 4 DO
~~
u><
0
~ o~-------L-L______~IL_______~I--------~~----~
0 05 (5 2
MAXMJM OCTN£0RAL SHEAR STRAIN %
Fig. 5. Cyclic stress-strain curves for maximum and octahedral shear stresses and strains.
For in-phase tests, F = 0, since the principal planes are at 45° to the maximum shear
plane. In the special case of <P = 90°, ;" = 1 + v, F takes its maximum value, 1. For
sinusoidal loading, F may be derived from equation (7) of Ref. [2] giving
z -- ,;_z + (1 + ~·)2 -J[((l + vf -)"2)2 + (2A.(l + v) cos </J)z]
F ( 10)
A. 2 + (1 + v) 2 +J[((l + v) 2 -A. 2 ) 2 + (2A(1 + v) cos </J) 2 ]
since the minimum shear strain range occurs on the 45o plane.
Brown and Miller [1] fitted in-phase data to the equation
( 11)
226 K. KANAZAWA eta/.
400-
300
~200
lL
~
0
+
05 1·0 15 20
Om ax
One may modify equation (II) to account for the hardening effect by introducing a
correction factor for the strength coefficient, k. Thus
'max= k(h'maxl"(l +IF) ( 12)
which can be fitted to the data of Table I to determine the constant I. Linear regression
analysis of the 12 in-phase results gave the values of k and n shown in Table 2. An optimum
value of I= 0'14 was then chosen to fit the out-of-phase results.
Exactly the same procedure was followed for an equation based on plastic strain,
Tmax = k'{!f'pmax)""(l +IF) ( 13)
where the plastic strain was defined by
All data
In-phase results 't"max/(1 +0·14F) Browh and Miller [I]
Number of specimens
in the analyses 12 29 17
the in-phase and the overall data are given, together with some previous results for
comparison. S is the percentage standard deviation about the regression line, as derived
in [!].
DISCUSSION
Both equations (12) and ( 13) correlate the out-of-phase data to within almost the same
scatterband as the in-phase results of Brown and Miller [1], although the data do not
cover such a wide range of shear strains. However, it is unlikely that these equations would
be valid close to yield, since they fail to predict the correct elastic behaviour as 11-+ l. One
might overcome this problem by replacing I with 1'(/'pmaxhmaxt, where /' and m are
constants. Since }'pmax tends to zero as one approaches yield, this would give the correct
elastic behaviour, and if 0 < m < I it may also fit the results of Table 1. As the present
study was limited to 0·53 ~ Ypmaxi'Ymax ~ 0·80, it was not possible to obtain reliable values
for/' and m.
The differences between the k and 11 values of Brown and Miller and the present
conditions are due to several causes. First is the effect of material variations: different
batches of specimens were used. This steel has been shown elsewhere to be quite sensitive to
the final heat treatment [I]. Secondly, the assumption of\'= 0·5 overestimates the values
of ~'max' and thirdly, the assumption of uniform stress in the specimen underestimates 'max·
Overall, the present curve follows the same shape as that of Ref. [I], with the stresses
reduced by about 7%. ·
The constant I may be expected to vary from one material to another, and also with
temperature when, for example, dislocation climb becomes possible. Lamba [ 4] conducted
three tests on OFHC copper where either torsional or axial cycling to the stable state was
followed by non-proportional cycling with cp = 90° and I. = 1·81. This implies a change of
the rotation factor, F, from 0 to 0·83, giving a rise in the peak stress level of about 40'/~,
whence I ::::= 0·5 for copper. One might expect to find a higher value for I in f.c.c. metals,
compared to b.c.c., since the slip systems are easier to activate. Similarly for h.c.p.
structures, the./ value should be reduced.
228 K. KANAZA W A et a/.
CONCLUSIONS
1. Hystere~is loops obtained under out-of-phase cyclic loads are quite different from the
in-phase behaviour of materials.
2. A cyclic stress-strain curve may be drawn in terms of the maximum shear stress and
strain, if the hardening effect due to rotating principal axes is first correlated.
Acknowledgements-The authors are indebted to the Science Research Council for finance, and the Central
Electricity Generating Board for the steel. The experimental work was performed in the Engineering Department
of Cambridge University.
REFERENCES
[I] Brown, M. W. and Miller, K. J. (1979) Biaxial cyclic deformation behaviour of steels. Fatigue
engng. Mater. Struct. 1, 93-106.
[2] Kanazawa, K., Miller, K. J. and Brown, M. W (1977) Low-cycle fatigue under out-of-phase
loading conditions. Trans. Am. Soc. mech. Engrs., J. Engng. Mat. Tech. 99 (H), 222-228.
[3] Taira, S., Inoue, T. and Yoshida, S. ( 1968) Low cyclefatigue under multiaxial stresses (in the
case of combined cyclic tension compression and cyclic torsion out-ofphase at elevated
temperature), Proc. ll th Japan Congress on Materials Research, pp. 60-65.
[4] Lamba, H. S. (1976) Non-proportional cyclic plasticity, Dept. Theoretical and Applied
Mechanics, University of Illinois, Report No. 413.
[5] Brown, M. W. (1975) High temperature multiaxial fatigue. Ph.D. Thesis, University of
Cambridge.
[6] Brown, M. W. and Miller, K. J. (1979) High temperature low cycle biaxial fatigue of two steels.
Fatigue Engng. Mater. Struct. 1, 217-230.