Está en la página 1de 14

Cyclic stress–strain behaviour of

circumferentially notched cylindrical bars at


high temperature

R. P. Skelton1, S. T. Wee2 and G. A. Webster1


1Department of Mechanical Engineering, Imperial College, Exhibition Road, London
SW7 2BX, UK
2School of Artillery, 310 Sembawang Road, 01-01 Khatib Camp, Singapore 758542

Cyclic stress–strain tests were undertaken at temperatures between 550°C and 820°C on plain and
doubly-notched specimens of three alloy steels: 20Cr/25Ni/Nb, 316 and 1CrMoV respectively.
Integrated axial ‘strains’ were measured between the minimum sections of the (semi-circular) notches
using a longitudinal extensometer, while surface hoop strains were measured on one notch by means of
a diametral extensometer. Cyclic hardening occurred in plain specimens of 316 steel, cyclic softening
occurred in the 1CrMoV steel and the 20Cr/25Ni/Nb alloy showed stable behaviour. These effects were
also demonstrated in notched specimens, although to a lesser degree. From hysteresis loops determined
during multiple step tests, the cyclic deformation response of the notched regions was expressed in
terms of an ‘equivalent gauge length’. Comparison was also made between the equivalent (von Mises)
stress–strain curves deduced at a representative (skeletal) point in the minimum section with
stress–strain data obtained from uniaxial (plain) specimens. The equivalent curves calculated from
standard relations were found to be lower than the uniaxial curves and possible reasons for this are
suggested.
Keywords: axial/diametral strain, cyclic hardening/softening, effective Poisson’s ratio, equivalent gauge length, equivalent
stress–strain, failure energy, low cycle fatigue, skeletal point

has been given to the Neuber [8] and similar relations for
1. INTRODUCTION
strain concentration at a surface and whether the local
Stress distributions in components undergoing cyclic strain is over- or underestimated compared with detailed
loading are generally non-uniform and moreover change finite element (FE) calculations [9–12]. With a circum-
as cyclic hardening or softening of materials develops. ferentially notched bar we are concerned with a triaxial
For a given point in the structure, corresponding cyclic stress state (axial, hoop and radial) in the minimum
stress analyses either (i) update behaviour at each stage section and a biaxial state at the notch root. Recently a
using constitutive models for determining the plastic link has been made with a conventional strain concen-
strain range or (ii) assume simplified stabilised (steady- tration factor and the triaxial state [13]. Shatil et al. [14]
state) material behaviour, neglecting the path of approach have further investigated local conditions by attaching
to current conditions. Further, the stress state in many strain gauges in the axial and circumferential directions
components is multiaxial. Since most supporting labora- at the root of a notched bar and comparing subsequent
tory data are generated from uniaxial tests, such data fully reversed cyclic deformation with the approaches
must either be appropriately converted or alternative above. These authors used a nominal axial strain averaged
means of producing multiaxial data must be sought. over a 15 mm gauge length embracing the notch. Circum-
Multiaxial tests in three orthogonal directions are com- ferential surface cracks at the notch root were sometimes
plicated to perform and require great capital outlay in present after 50% of the fatigue life had been consumed.
testing equipment. In the notched bar, however (which The present investigation concerns the generation of
can be tested in a single direction), we have a specimen an equivalent hysteresis loop at a representative point in
capable of generating several triaxial stress states at dif- the minimum section of circumferentially notched bars
ferent points in the minimum cross-section. (for convenience, doubly-notched bars were used). It can
Most previous studies of circumferentially notched be shown [6,15] that the hydrostatic component of stress
bars at high temperature have concerned tension loading (arising from axial, hoop and radial components) is very
only and then principally for creep studies [1–6]. nearly a maximum at this point which is identified as the
Previous studies of cyclic plastic behaviour at a notch skeletal point. Further, it has been shown [5,15–17], in
have generally dealt with conditions at the notch root [7] creep studies at least, that the location of this point is
(for that is where fatigue cracks initiate). Much attention independent of the power law used to describe creep,

0964-3409/01/01/00035–12
© 2001 Science Reviews MATERIALS AT HIGH TEMPERATURES 18(3) 139–152 139
R.P. Skelton et al.: Cyclic stress–strain behaviour of circumferentially notched cylindrical bars at high temperature

though there is some shift in this point according to Minimum section radius, a 2.82 mm
notch acuity. It is thus an extremely useful point to Shank radius, b 4.00 mm
characterise specimen behaviour, though it should be Notch root radius, R 1.17 mm
noted that:
The ‘notch acuity’ a/R was thus 2.41 and the ‘notch
• Strains at this internal position must be inferred from ratio’ b/a was √2. These dimensions were according to a
experimental displacement measurements taken else- Code of Practice for notch bar creep testing [16, 17].
where on the specimen A non-linear FE analysis of the stress and strain distri-
• The equivalent strain is not a maximum here, but at the butions in the notch was conducted using the BEGL
notch root. Fatigue failure is most likely to occur at the BERSAFE stress analysis programme. An axisymmetric
root (initiation of a crack) though damage in creep tests mesh of half the specimen was modelled. Boundary con-
will be a function of stress triaxiality and could well ditions of zero axial displacement on the plane of sym-
form preferentially at an internal site. metry together with a uniformly distributed axial load
were applied to the ends of the specimen. Initial loading
strains were derived from (plain specimen monotonic)
A triaxial state of stresses arises, for example, in pipe-
data characterised by the constants Am = 442 MPa, m =
cylinder intersections or in section changes in steam
0.108 in the following power law at 550°C:
headers [18]. In assessing the integrity of plant it is often
required to employ a multiaxial stress–strain relation in
 = Amp m (1a)
determining local plastic and creep strains. This usually
takes the form of an ‘equivalent’ uniaxial stress–strain
where  is the tension stress and p is the plastic strain.
relation based on the von Mises criterion. It is hoped to
In the FE program the notched specimen was loaded to a
show in this paper that (i) the circumferential notched
net section stress, net, of 357 MPa in 10 approximately
bar provides a simple practical method for generating a
equal steps. In Figure 2 is shown the development of the
cyclic stress–strain curve under a given stress triaxial
equivalent plastic strain across the minimum section
ratio (defined later) when referred to the skeletal point
while the corresponding hoop strain at each loading
and (ii) that the whole notch section can be reduced to an
stage is shown in Figure 3. These Figures also indicate
‘equivalent gauge length’ for ease in applying infor-
the position of the skeletal point, which for a semi-circu-
mation obtained from displacements measured remote
lar notched specimen is at about 0.75a [15]. As the stress
from the notch. Three materials typical of the power
increases, the value of the respective strains at the skele-
generation industry were tested at a temperature of
tal point adopts an ever larger (disregarding the sign)
550°C and above. These were a 316 stainless steel piping
value to that experienced at the notch root. Further
material which is known to cyclically harden, a 1CrMoV
details from the FE analysis are provided in Table 1a,
turbine rotor steel which cyclically softens and a cycli-
which from the 5th loading step onwards shows the fol-
cally stable 20/25/Ni/Nb alloy used in nuclear fuel
lowing ratios (i) p,eq*/p,ax and (ii) p,eq*/p,hoop where
cladding.
p,eq * is the equivalent strain at the skeletal point, p,ax is
the axial strain at the notch root and p,hoop is the hoop
strain at the notch root. Additional validation for this
2. BACKGROUND
Table is provided in Table 1b which compares certain
The starting point for this work was an FE analysis of a load ratios determined upon the 10th loading with those
semi-circular notched specimen of a similar 316 stainless provided by a Code of Practice for notch bar creep test-
steel, undertaken by British Energy Generation Ltd ing [16, 17].
(BEGL). Referring to Figure 1, the dimensions were as Throughout this paper, an asterisk will indicate the
follows: value of any parameter at the skeletal point. It may also

Figure 2 Equivalent plastic strain results from FE analysis in notch


Figure 1 Parameters used in notch analysis. minimum section for loading steps shown.

140 MATERIALS AT HIGH TEMPERATURES 18(3)


R.P. Skelton et al.: Cyclic stress–strain behaviour of circumferentially notched cylindrical bars at high temperature

Following the work of Bridgman [19] on semi-circular


notches, the representative (average) plastic axial strain,
rep, for a fully yielded notch is given by the constant
volume criterion [20–23]:

d
 rep = 2 ln   (2)
 do 

where the term ln(d/do) is the surface hoop strain, d is the


current diameter and do is the original diameter in the
minimum section. It can be shown for the case of a
Bridgman notch [15] that the equivalent strain profile
lies very close to the axial strain profile, diverging by a
maximum factor of 1.03 between the skeletal point and
the notch root. From the argument above, equation (2)
may be rewritten in general:

Figure 3 Hoop strain results from FE analysis in notch minimum sec- 1 d
tion for loading steps shown.  *p,eq = ln   (3)
1 – eff  do 

be noted that, for cyclic loading, calculations will be per- The next step is to relate the skeletal stress with the
formed in terms of total stress and strain ranges so that easily-determined net section stress. This is provided by
equation (1a) becomes: the Bridgman relation [19–22] for a semi-circular notch:
 = Ap (1b)
*eq  a 
–1
=  1 +
2R  
ln 1 + (4)
where A and  are constants. Thus at peak compression the net   a   2R  
sign of the strain profiles in Figures 2 and 3 is reversed.
where for the present specimen, equation (4) gives eq*
= 0.69net, in excellent agreement with the Code of
3. MULTIAXIAL STRESS-STRAIN EXPRESSION Practice [16] value of 0.68 determined by FE analysis,
Values in Column 3 of Table 1a show how the equivalent see also Table 1b.
plastic strain at the skeletal point increases as a function
of the notch root axial surface plastic strain. The ratio is 4. EQUIVALENT GAUGE LENGTH
denoted by the symbol . Values in Column 4 of the
same Table suggest, since the strains are in orthogonal From the foregoing, notch deformation is usually studied
directions, that they represent the inverse of an effective by means of diametral extensometers [24]. Consider now
value of Poisson’s ratio, eff (which increases from 0.3 an axial extensometer to be placed remote from a notch
(elastic) to 0.5 for the fully plastic case). They are thus as shown in Figure 4a and that a permanent extension is
given by the quantity 1/(1–eff), which accordingly measured after loading and unloading. Deformation has
increases from 1.43 to 2. If this is true in the general taken place not only within the minimum section, but in
case, we can develop an expression for multiaxial stress the notch region generally. Assuming no deformation in
and strain (as seen by the skeletal point) in terms of mea- the remote shank and referring all measurements to the
sureable parameters. skeletal point, the notch may be replaced by a cylinder of
original diameter do and an effective gauge length leff* as
indicated in Figure 4a. In terms of engineering strain,
Table 1a Finite element results on loading a notched specimen
this cylinder has extended by some fraction of the notch
Loading Net section p,eq*/p,ax (“”) p,eq*/p,hoop surface extension i.e. by an amount xa/leff*, where xa is
stress (MPa) (“1/(1-eff)”) the measured axial displacement at the surface, see also
Column 3 of Table 1. In terms of true strain this may be
5th 127 0.256 –1.46
6th 169 0.314 –1.69
identified with equation (3) to give:
7th 213 0.371 –1.86
8th 259 0.416 –1.95
1 d  x 
9th 307 0.449 –1.99 ln   =  ln  1 + *a  (5)
10th 357 0.454 –1.98 1 –  eff  do   leff 

Table 1b Finite element load ratios compared with code of practice which solving for leff* gives:
[16, 17] values xa
*
leff =
10th loading eq*/net m*/net 1*/net m*/eq 1*/eq  1  d  (6)
BERSAFE 0.70 0.59 1.04 0.84 1.49 exp  ln    – 1
COP [16, 17] 0.68 0.55 1.04 0.81 1.53   (l –  eff )  do  
Subscipts ‘eq’, ‘net’, ‘m’and ‘1’ denote equivalent, net section, mean In practice, for strains < 10%, the engineering form of
(hydrostatic) and maximum principal stresses respectively. equation (6) may be used:

MATERIALS AT HIGH TEMPERATURES 18(3) 141


R.P. Skelton et al.: Cyclic stress–strain behaviour of circumferentially notched cylindrical bars at high temperature

(a)

(b)

Figure 4 Determination of an equivalent gauge length for (a) a single Figure 5 Schematic diagram for plain and double-notched test speci-
notch and (b) a double notch specimen. mens.

xa
*
leff = do (1 –  eff ) (7)
xd

It may be argued that, when permanent offset measure-


ments are taken at zero load (width of hysteresis loops),
equation (7) may be applied with eff = 0.5. In the fol-
lowing and for simplicity, we take the maximum value of
 as 0.45. Equation (7) is actually fairly insensitive to
current values of  and eff, being dominated by the ratio
Figure 6 Schematic diagram for axial and diametral extensometer
of axial displacement xa to the diametral displacement, layout.
xd. The equation is justified further in Section 8.

To avoid buckling problems in compression, the shank


5. SPECIMENS AND TESTING radius, b, of the notched specimens was made as large as
practicable. To conform with the b/a = √2 recommenda-
5.1 Methodology
tion of the Code [16] the dimensions were as follows:
In addition to notched specimens, plain smooth speci-
mens as shown in Figure 5 were used to establish basic Minimum section radius, a 4.245 mm
uniaxial cyclic stress–strain data. They were of 12.7 mm Shank radius, b 6.00 mm
diameter and 12.5 mm gauge length. All specimens were Notch radius, R 1.76 mm
rigidly gripped and tested in reversed loading, to execute
various hysteresis loops. Tests were performed in axial The notch spacing was closer than recommended [16]
strain control (effectively displacement control for the since it was governed by the axial extensometer design.
notched specimens, see below) using a servo-electric Reference to Figures 4b and 5 shows that the experimen-
testing machine. Heating was by RF induction. The side- tal gauge length between the probes constitutes one half
contacting axial extensometer used for plain and notched of each notch region plus the intervening shank. The
specimens was a development of a previous capacitor deformation behaviour of the remaining notch halves
design [25, 26] and is shown schematically in Figure 6; and outer shanks do not contribute since they are not
it employed ceramic probes which were sprung loaded measured by the axial control system. Moreover, when
onto the surface. The doubly-notched specimens were of the equivalent gauge length of each notch is demon-
novel design as shown in Figures 4b and 5. Their lines of strated to be < 12.5/2 mm i.e. < 6.25 mm, it can be
minimum section were 12.5 mm apart to correspond argued that there is no interaction between the plastic
with the gauge length of the plain specimens. In this way strain zones generated in the respective notches.
the axial extensometer probes could be accommodated, Values of Young’s modulus were obtained from the
with their tips nestling in the respective notch roots. In elastic part of the hysteresis loops produced on the plain
addition these specimens were equipped with a diametral specimens. In what follows, the full range of stress and
extensometer (on the lower notch only) as shown in strain is used to characterise cyclic stress–strain behav-
Figure 6, this again being a development of a previous iour. Two methods were available for doing this [28]. In
design [27]. The signal was the average of the “A” and one, the actual path is traced (curve shape method) tak-
“B” outputs on opposite sides of the root. ing peak compression as the origin and noting the devia-

142 MATERIALS AT HIGH TEMPERATURES 18(3)


R.P. Skelton et al.: Cyclic stress–strain behaviour of circumferentially notched cylindrical bars at high temperature

tion from elastic behaviour. In the other, a locus method was also tested, see Table 3 and Discussion.) An example
is applied whereby multiple step tests are used to create for the 20Cr/25Ni/Nb steel is shown in Figure 7. This
pairs of plastic strain range, p, (width of hysteresis loop was specially generated at 550°C at a rather large strain
at zero stress) and corresponding stress range, , gener- range in order to demonstrate the effect. Similar but nar-
ally in ascending order of loading. The former method rower loops were obtained for all other tests on this alloy
was used for plain specimens while it was considered at different temperatures and also for the other two alloys
safer using the latter method (which involves taking sep- at 550°C. Naturally, with a large number of cycles of dif-
arate hysteresis loops in incremental steps) for the notched fering magnitude being imposed on relatively few speci-
specimens in order to avoid large strains at the outset. mens, history effects are to be expected and these are
commented on where possible.
5.2 Schedule
6.1 Evolutionary behaviour
Eight specimens were available for this project and tests
were conducted to produce the maximum amount of evo- From previous work [29] it is known that the
lutionary and steady-state cyclic information before the 20Cr/25Ni/Nb steel shows stabilised behaviour, and to
initiation of a low-cycle-fatigue (LCF) crack. This was conserve specimen life, evolutionary cycles were not
an important consideration because the volume element explored. The 316 stainless steel hardened, as indicated
of interest lies at the skeletal point whereas fatigue by an increase in stress range with cycles, Figure 8a, and
cracks initiate at the notch root. Reference to Figures 2 the 1CrMoV steel softened, Figure 8b. Results for plain
and 3 shows that the equivalent strain at the notch root and notched specimens are given in these Figures. From
can rise to about 4.5 times the monitored hoop strain (or the diametral hysteresis loops the estimated plastic strain
by using Figure 2 alone, some 2.25 times the skeletal range at the notch root of 316/77 (Figure 8a) was 0.7%,
point equivalent strain). Plain specimen tests were con- and this caused failure in 177 cycles. Thus in specimen
ducted as shown in Table 2 and notched specimen tests 316/78, a notch root plastic strain range of 0.32% was
as shown in Table 3. aimed for, generally matching that of plain specimen
All tests were conducted in the well-established axial 316/76. In specimen 1CMV/23 a notch root plastic strain
control mode. Having regard to the relatively small range of 0.28% was aimed for (approximately matching
notch diameter coupled with high sensitivity of the that of plain specimen 1CMV/22).
diametral extensometer (0.05 mm/V as opposed to 0.1 For both plain and notched specimens, stress or equiv-
mm/V for the axial probes) and the sometimes ‘jerky alent stress range response with cycles was fitted to the
flow’ (dynamic strain ageing) exhibited by the 316 stain- following law [30]:
less steel, it was considered wise at this stage to employ
the diametral system in the monitoring mode only. For  = CN (8)
plain specimens, tests were generally performed at a
specified total strain rate. For notched specimens in where C (the initial stress range) and  are constants
axial control, tests were more conveniently done at a whose values are provided in Table 4. The value of  is
specified frequency, to correspond with those experi- positive for the case of hardening and negative for the
enced by the plain specimens. case of softening.
It is emphasised that the output from the axial exten- Corresponding changes in plastic strain range (decreas-
someter records the integrated response from two half ing for hardening, conversely for softening) with cycles
notches whereas the output from the diametral exten- are given in Figures 9a and 9b. The notched specimen
someter records the hoop strain at one notch root via the values refer to the skeletal point, calculated from equation
term ln(d/do). (2). When hardening or softening occurs within limits of
constant total strain range then changes in plasticity are
gained at the expense of elasticity via the relation [30]:
6. OBSERVATIONS
The basis of this work relies on a comparison of the out- d
puts from the axial and diametral extensometers on dou- d p = – (9)
bly-notched specimens. (One single-notched specimen E

Table 2 Tests on plain specimens

Specimen no. Temperature (°C) Total strain range (%) Reversed strain rate or frequency Order of tests

20/25/Nb/C 750 0.45 0.017 Hz 50 cycles, stable behaviour


820 0.45 0.017Hz 15 cycles, stable behaviour
700 0.45 0.017Hz 50 cycles, stable behaviour
Ambient 1.0 0.00015/s Single hysteresis loop
Ambient 0.40 0.00015/s 87 cycles, stable behaviour
Ambient 1.0 0.00015/s Two hysteresis loops
316/76 550 1.0 0.0004/s Two hysteresis loops, CSS
550 0.6 0.0004/s 175 cycles, cyclic hardening then CSS
550 1.0 0.0001/s Two hysteresis loops, CSS
550 1.0 0.00001/s Two hysteresis loops, CSS
1CMV/22 550 1.0 0.0004/s Two hysteresis loops, CSS
550 0.64 0.017 Hz 331 cycles, cyclic softening
550 1.0 0.02 Hz Two hysteresis loops, CSS

CSS = cyclic stress–strain curve from actual shape

MATERIALS AT HIGH TEMPERATURES 18(3) 143


R.P. Skelton et al.: Cyclic stress–strain behaviour of circumferentially notched cylindrical bars at high temperature

Table 3 Tests on notched specimens

Specimen no. Temperature (°C) Displacement Reversed strain rate Order of tests
(across 12.5 mm gauge) (m) or frequency

20/25/Nb/D 750 ± 6 to ± 25 0.017 Hz Multiple step, increasing strain


820 ± 6 to ± 22 0.017 Hz Multiple step, increasing strain
700 ± 6 to ± 26 0.017 Hz Multiple step, increasing strain
700 ± 26 0.017 Hz 35 cycles, check for cyclic stability
Ambient ± 12 to ± 35 0.017 Hz Multiple step, increasing strain
Ambient ± 35 0.017 Hz 46 cycles, check for cyclic stability
316/77 550 ± 19 to ± 37 Multiple step, increasing strain
550 ± 39 0.02 Hz 172 cycles, cyclic hardening.
Failed specimen
316/78 550 ± 22 0.02 Hz 103 cycles, cyclic hardening
550 ± 22 to ± 16 0.02 Hz Multiple step, decreasing strain
550 ± 22 0.02 Hz Further 393 cycles, cyclic hardening
550 ± 22 to ± 14 0.02 Hz Multiple step, decreasing strain
± 23 to 38 0.02 Hz Multiple step, increasing strain
550 ± 40 0.0005 Hz Single loop
550 ± 19 0.001 Hz Single loop
550 ± 28 0.00075 Hz Single loop
316/94* 550 ± 37 0.02 Hz Two loops, CSS
550 ± 22 0.02 Hz 300 cycles, cyclic hardening
550 ± 37 0.02 Hz Two loops, CSS
550 ± 32 to ± 10 0.02 Hz Multiple step, decreasing strain
550 ± 10 to ± 37 0.02 Hz Multiple step, increasing strain
1CMV/23 550 ± 10 to ± 40 0.02 Hz Multiple step test
550 ± 30 0.02 Hz 333 cycles, cyclic softening.
Signs of cracking

*Single-notched specimen

where E is Young’s modulus. This will apply for the plain


specimens, but not necessarily for the notched specimens (a)
since axial control over a hybrid gauge does not necessar-
ily guarantee constancy of leff* as evolution occurs.

6.2 Uniaxial cyclic stress-strain properties


From the large hysteresis loops (1% total strain range)
taken at the beginning and end of tests (Table 2), the uni-
axial cyclic stress–strain response was established by
curve-fitting to equation (1b) i.e. with the proviso that
the full range of stress and plastic strain was used [28].
The high temperature results are given in Table 5 and in
Figures 10a, 10b and 10c. Hardening or softening are
characterised by an increase or decrease respectively in
the value of the parameter A.

(b)

Figure 8 Evolutionary behaviour with cycles in plain and notched


Figure 7 Typical hysteresis loops in longitudinal and lateral directions specimens of (a) 316 stainless steel at 550°C showing hardening and
for 20Cr/25Ni/Nb steel at 550°C. (b) 1CrMoV steel at 550°C showing softening.

144 MATERIALS AT HIGH TEMPERATURES 18(3)


R.P. Skelton et al.: Cyclic stress–strain behaviour of circumferentially notched cylindrical bars at high temperature

Table 4 Evolutionary cyclic parameters


(a)
Specimen no. C, (MPa)  Initial p Remarks
or peq*
(mm/mm)

20/25/Nb*
316/76 335 0.087 0.0039 Uniaxial
316/77 314 0.109 0.0032 Triaxial
316/78 247 0.056 0.0019 Triaxial
316/94 568 0.037 0.0040 Triaxial
1CMV/22 906 –0.043 0.0025 Uniaxial
1CMV/23 858 –0.053 0.0013 Triaxial

*For Specimens C & D the term C is equal to the initial stress range
and  = 0.

6.3 Axial versus diametral displacements


For notched specimens, the multiple step tests proved a
convenient and unambiguous method of comparing axial
and diametral displacements from each run. From each
hysteresis loop in the ascending or descending series the (b)
axial loop width (at zero load) was plotted against
diametral loop width (at zero load) for the
20Cr/25Ni/Nb, 316 stainless and 1CrMoV alloys respec-
tively in Figures 11a, 11b and 11c. All the results show
clearly the much greater displacement obtained from the
axial probes. Within experimental scatter, straight lines
are obtained which pass through the origin. The data
show a clear banding with respect to temperature
(20Cr/25Ni/Nb) and evolutionary state (316). Data on

(a)

(c)

(b)

Figure 10 Cyclic stress–strain curves obtained by ‘curve-shape’


method for plain specimens of (a) 20Cr/25Ni/Nb, (b) 316 stainless
steel and (c) 1CrMoV steel.

the 1CrMoV steel are inconclusive because of the inter-


vention of surface cracking and the information in the
final state was obtained on a single excursion into com-
pression. (For purposes of obtaining the cyclic
stress–strain relation, the observed loads were simply
doubled.)
Figure 9 Change in plastic strain range accompanying hardening or The gradient of each line determined in Figures 11a–c
softening in: (a) plain specimens and (b) notched specimens. was used in subsequent calculations.

MATERIALS AT HIGH TEMPERATURES 18(3) 145


R.P. Skelton et al.: Cyclic stress–strain behaviour of circumferentially notched cylindrical bars at high temperature

Table 5 Uniaxial power-law constants for cyclic stress-strain tests (curve shape method)

Material Temperature (°C) Strain rate or frequency E (GPa) A (MPa)  Remarks

20/25/Nb/C 700 0.017 Hz 129 772 0.133 Stable


750 0.017 Hz 119 529 0.092 Stable
820 0.017 Hz 129 333 0.066 Stable
316/76 550 0.0004 /s 143 629 0.120 Virgin
550 0.0004 /s 1328 0.147 Hardened
550 0.0001 /s 1085 0.113 Hardened
550 0.00001 /s 1102 0.107 Hardened
1 CMV/22 550 0.0004 192 2934 0.204 Virgin
550 0.0001 1316 0.104 Softened
550 0.0002 1339 0.105 Softened

(a) (b)

(c)

Figure 11 Variation of axial displacement with diametral displacement for (a) 20Cr25Ni/Nb, (b) 316 stainless steel and (c) 1CrMoV steel.

7. CALCULATIONS tracts during cyclic hardening. Results for the softening


of 1CrMoV steel are uncertain and further tests are
7.1 Equivalent gauge length
underway on this class of alloy. Despite the cyclic soft-
Using the data in Figures 11a-c, equivalent gauge ening, the 1CrMoV ferritic steel remains stronger than
lengths were calculated from equation (7) assuming the austenitic alloys, and the smaller axial displacement
eff = 0.5. A minor correction was applied to calculate for a given diametral displacement, Figures 11b and
appropriate values of do at high temperatures from 11c, accounts for a much smaller value of leff* from
known values of expansion coefficients. The gauge equation (7).
length results are given in Table 6. For the 20Cr/25Ni/Nb It is seen from Table 6 that the equivalent gauge
steel it is seen that there is generally an increase in leff* length is rather greater than a value that might be intu-
with increase in temperature. This may be expected, as itively expected as that equal to the notch mouth opening
more of the notch region is deformed as flow becomes (1.76  2 = 3.5 mm). This implies that the deformed
easier, see Figure 10a. Equally, it is noted in the case of notch region makes considerable inroads into the adja-
316 stainless steel that the equivalent gauge length con- cent shank region.

146 MATERIALS AT HIGH TEMPERATURES 18(3)


R.P. Skelton et al.: Cyclic stress–strain behaviour of circumferentially notched cylindrical bars at high temperature

Table 6 Calculation of equivalent gauge length, equation (7)

Specimen Temperature (°C) Strain rate or frequency leff* (mm) Remarks

20/25/Nb/D Ambient 0.017 Hz 4.8 Stable


700 0.017 Hz 5.0 Stable
750 0.017 Hz 4.6 Stable
820 0.017 Hz 6.1 Stable
316/77 550 0.02 Hz 8.6 Virgin
316/78 550 0.02 Hz 7.0 Hardened, 103 cycles
550 0.02 Hz 6.8 Hardened , 393 cycles
316/94 550 0.02 Hz 5.3 Hardened
1CMV/23 550 0.02 Hz 3.8 Virgin
550 0.02 Hz 3.7 Softened

7.2 Prediction of multiaxial stress-strain behaviour below their uniaxial counterparts. This observation is
taken up further in the Discussion Section.
In the multiple step tests used on notched specimens,
plastic strain range was determined from a permanent
offset i.e. the width of each hysteresis loop at zero stress. 8. DISCUSSION AND FURTHER
Thus, insofar as the locus method [28] represents a INVESTIGATIONS
cyclic stress–strain curve, an equivalent triaxial
8.1 Extra tests
stress–strain curve can in principle be deduced from
equation (4) and equation (2). Results from this method The equivalent cyclic stress strain-curves, based on the
are given in Figures 12a–c for the 20Cr/25Ni/Nb, 316 von Mises definition, appear to lie below their uniaxial
and 1CrMoV alloys respectively. The equivalent counterparts for all materials, both in the virgin and
stress–strain curves for all data may also formally be fit- hardened/softened condition, and the difference between
ted to the ‘full range’ equation (1b) and the results are the initial and worked state is not so pronounced for the
given in Table 7. calculated equivalent curves, see Figures 12b and 12c.
It is noted that on comparison with data in the respec- The case of the 1CrMoV steel is inconclusive because
tive Figures 10a–c, the equivalent curves lie somewhat the specimen broke before softening was complete.

(a) (b)

(c)

Figure 12 Calculated equivalent stress–strain curves for notched bars of (a) 20Cr25Ni/Nb steel, (b) 316 stainless steel and (c) 1CrMoV steel.

MATERIALS AT HIGH TEMPERATURES 18(3) 147


R.P. Skelton et al.: Cyclic stress–strain behaviour of circumferentially notched cylindrical bars at high temperature

Table 7 Multiaxial power-law constants

Material Temperature (°C) Strain rate or frequency A (MPa)  Remarks

20/25/Nb/C 700 0.017 Hz 1349 0.235 Stable


750 0.017 Hz 2657 0.356 Stable
820 0.017 Hz 888 0.238 Stable
316/77 550 0.02 Hz 496 0.064 Virgin
550 0.02 Hz 735 0.109 Hardened
316/94 550 0.02 Hz 590† 0.130† Virgin
550 0.02 Hz 940† 0.156† Hardened
1 CMV/2 550 0.02 Hz 3699 0.247 Virgin
550 0.02 Hz 1443* 0.084* Softened

*Data from single compressive loading


†Obtained by ‘curve-shape’ method

With this in mind, a final specimen of 316 stainless • The accompanying plastic strain ranges measured on
steel (No. 94) was available for further experimentation. the diameter, shown as the upper line in Figure 14,
This was provided with a single notch, and the axial were higher than those given in Figure 9b, though the
extensometer was positioned symmetrically about the rate of decease with cycles was similar
notch root, with the probes 12.5 mm apart now sprung- • Neglecting the very first large loops (which were per-
loaded against the shanks. Comparison of Figures 4a and turbed by dynamic strain ageing effects), the
4b shows that the measurements from such a system are axial/diametral strain ratios from all other runs on this
identical to the double-notch arrangement. Diametral specimen were consistent and the results are given in
strains were also monitored as before, but it may be
noted that the specimen was slightly oversize as follows:

Minimum section radius, a 4.48 mm


Shank radius, b 6.35 mm
Notch radius, R 1.86 mm

notch dimensions, nevertheless, being in accord with the


Code of Practice [16, 17]. Referring to Table 3, specimen
No. 94 was put through a slightly different procedure as
follows:

• Two large loops were first executed. The second loop


(which was completely closed) was used to determine
‘virgin’ cyclic stress-stress strain response by the
curve-shape method and equations (2) and (4)
• The specimen was allowed to cyclically harden over
300 cycles at a somewhat lower strain range
• Two large loops were again executed at the same dis-
placement range used at the beginning. ‘Hardened’ Figure 13 Evolutionary hardening curve for notched 316 stainless
cyclic stress–strain properties were again determined steel specimen with different loading history.
from the second loop using the curve-shape method
and equations (2) and (4)
• Finally, the locus method was used to verify the hardened
response by means of the multiple step test, first in
descending and then in ascending order, and also to deter-
mine the ratio of axial to diametral extensometer outputs.

At the outset of the test, massive dynamic strain ageing


was observed in the large loops, manifest as discontinu-
ous yielding with ‘serrated’ load drops. It was also pre-
sent in attenuated form in the two final loops. As such, a
different history effect was experienced compared with
the companion tests (Nos 77 and 78 in Table 3). The
results are now compared:

• Although the same axial displacement range was used


to control the evolutionary test (± 22 m, see Table 3),
the initial stress response was higher, Figure 13 (com-
pare with Figure 8a and Table 4), the microstructure
hardening at an intermediate rate (compare values of 
in Table 4). These effects were almost certainly due to a Figure 14 Change in plastic strain range accompanying hardening:
large hardening effect during the two initial loops comparison of diametral and remote axial extensometer outputs.

148 MATERIALS AT HIGH TEMPERATURES 18(3)


R.P. Skelton et al.: Cyclic stress–strain behaviour of circumferentially notched cylindrical bars at high temperature

Figure 15. The slope of the line is rather less than those 16. It is immediately evident that, in terms of the equiva-
for hardened material in Figure 11b, no doubt because lent strain deduced at the skeletal point, the notched
of the influence of the two large loops imposed at the specimen has not hardened as much as the plain speci-
outset. By means of equations (6) or (7) the equivalent men. A likely explanation is that in plain specimens there
gauge length is reduced to 5.3 mm, see also Table 6. is no strain gradient and the microstructure hardens uni-
These equations also indicate that a factor 2.4 is formly cycle-by-cycle, equation (9) applying throughout
required to superimpose the remote axial and diametral the cross section. In notched specimens, however, there
data as indicated in Figure 14 is a strain gradient at the outset, see Figure 2 for exam-
• Equivalent stress–strain curves for the virgin and hard- ple. It is known that hardening is accomplished more
ened state are given in Figure 16. On comparing with quickly at higher strain ranges in low cycle fatigue [30],
Figure 12b it is noted that the ‘virgin’ curves are very so in circumferentially-notched specimens the outer
similar in response. However it is to be noted that spec- annuli nearer the root harden preferentially as cycling
imen No. 94 has hardened more than companion speci- proceeds. At the end of the test therefore, in addition to
men Nos 77 and 78. It is finally remarked that the the maximum strain gradient at either peak tension or
‘locus’ method appears to predict a slightly higher compression stress there is in addition a microstructural
strength than the ‘curve-shape’ method as shown in hardening gradient. It is suggested that deformation
Figure 16. behaviour is dominated by the softer core which occu-
pies a greater area so that equivalent stress strain behav-
In order to compare equivalent stress–strain data with iour is not fully represented in such a specimen.
(plain-specimen) uniaxial data more closely for the stain- It should be noted, however, that this mismatch
less steel, the virgin and hardened (at a strain rate of 4  between equivalent and uniaxial stress–strain response
10–4/s) results from Figure 10b are re-plotted in Figure appears to be present for the very first loading curves,
taking zero stress and zero strain as origin (“s cycle
data”). The appropriate data are given in Figure 17
where stress is now in terms of semi-range. Further
attempts to reconcile the data could possibly invoke use
of the Tresca criterion for which the equivalent stress
would be a maximum of 15% larger.

8.2 Triaxial stress state


It appears to be a feature of a Bridgman notch (a/R =
2.41) that the value of the principal (1*) stress at the
skeletal point approximates to the net section stress. The
corresponding values from the FE analysis of Figure 2
and 3 are as follows:

net 357 MPa


1* (principal) 372 MPa
2* (hoop) 154 MPa
3* (radial) 100 MPa

Figure 15 Variation of axial displacement with diametral displace-


(As a check, the skeletal point equivalent stress, eq*, is
ment for 316 stainless steel: complete loading history found from the von Mises relation:

Figure 16 Comparison of equivalent with uniaxial stress–strain curves Figure 17 Comparison of equivalent with uniaxial stress–strain curves
for 316 stainless steel in the initial and final states. for 316 stainless steel after first s cycle.

MATERIALS AT HIGH TEMPERATURES 18(3) 149


R.P. Skelton et al.: Cyclic stress–strain behaviour of circumferentially notched cylindrical bars at high temperature

notch minimum diameter and (ii) the working distance


1
( ) +( ) +( )
2 2 2
 eq
*
= 1* –  2*  2* –  3*  3* – 1* (10) between the axial extensometer probes viz. 12.5 mm.
2 Using appropriate vales of leff* from Table 6 (i.e. for the
hardened or softened condition) the average energies of
to be 250 MPa, in agreement with equation (4), which the notched specimens at failure were as follows:
gives 246 MPa.)
Equation (4), nevertheless, severely under-predicts the 316/77 0.46 J/mm3
equivalent stress for a/R values > 2.41, this result being 316/78 0.64 J/mm3
determined from a comparison with FE analyses [16]. 316/94 (unfailed) 0.41 J/mm3
Similarly, other FE work [15] has called into question 1CMV/23 0.63 J/mm3
the asymptotic value of 2 implied in equation (2). The
factor of 2 is only strictly applicable when the hoop and However, specimens failed at the surface and since hys-
radial strains are equal and this does not necessarily teresis loop energy varies as p1+ the above values
apply at the skeletal point [15]. For conditions of steady must be multiplied [32] by a factor (1/)1+. Taking  =
state creep in notched (a/R = 2.41) specimens of 316 0.45 and  = 0.15 and 0.1 respectively for the 316 stain-
steel the coefficient actually reduces to about 1.3 as the less steel and ferritic alloy respectively, the factors are
stresses across the minimum section re-distribute [15]. 2.5 and 2.4 respectively. Failure energies are then in
Neglecting hardening effects, this would thus bring the good agreement with those reported in the literature [32].
‘virgin’ curves in Figure 16 for example closer together
under conditions of creep. 8.5 Equivalent gauge length
The core information for this paper is provided in
8.3 Other work Figures 11 a–c and in Figure 15; these all enable a calcu-
The nearest study to the present work appears to be the lation of strains within the minimum section of the notch
FE analysis of Ohno and Satra [31] who modelled cyclic in terms of remotely measured diametral and axial dis-
notch behaviour of 304 steel at ambient temperature. The placements. Alternative approaches may be used, but we
uniaxial (hardened) cyclic curve was given by the para- have chosen a representative (equivalent) gauge length
meters A = 1961 MPa and  = 0.239. The (single) which has been assumed to be located at the skeletal
notched specimen was characterised by the values b/a = point. It is implicit in equation (7) that:
1.33 and an inclusive notch flank angle of 60°, the notch
acuity however being much sharper than in the present • Equivalent strains at the skeletal point are some fraction,
work (a/R = 32). The specimen was modelled in displace- , of those acting at the notch root surface, see Figure 2
ment control, with a remote stress range of 240 MPa. In • Equivalent strain at the notch root surface is equal to
the model, it took 10 cycles to reach the stabilised state, the axial strain at the same location
and it was shown how, at the notch root, the stress range • Integrated axial strains as measured during this work
increased from 700 MPa to 900 MPa, the plastic strain are a reasonable indication of notch root strains
range meanwhile decreasing from 1.12% to 0.94 %.
As noted in Section 3, finite element analyses have
Other marked similarities with the present experimental
demonstrated [15] a close correspondence between the
investigation were as follows:
equivalent and axial strain profiles. Closer examination
of other data [3, 6] shows that the divergence depends on
Whereas 44% hardening was observed from the
notch acuity, but this can be neglected for the case of a
authors’ basic uniaxial data, from the above informa-
semi-circular notch.
tion only 29% hardening occurred at the notch root
Regarding locally determined strains, intricate tech-
The situation at the notch root was intermediate
niques are available for inserting twin probes (“strain
between constant strain (characterised by equation (9))
pecker”) into a single notch root at high temperatures
and constant stress, since the total strain range
[33] using 0.4 mm as a typical working gauge length at
decreased from 1.47% to 1.27% during the 10 harden-
the surface. Notch diameters in the present work (2R =
ing cycles. This is strongly reminiscent of controlling
2.52 mm) were too small to apply this technique, but jus-
to a ‘Neuber hyperbola’ [8, 30] or constant energy cri-
tification of our approach is made by referring to some
terion
creep tests [15] where dimensions of notch shape were
At a point in the minimum section of depth equal to
determined by shadowgraph techniques before and after
the notch radius (0.19 mm), the model predicted that
testing. In this work [15], two double-notched specimens
the material was stable, with no hardening. In fact,
(geometry as given in Section 2) of a similar 316 steel
deformation here was essentially elastic and all other
were loaded at 550°C to a net section stress of 382 MPa
points between this location and the surface hardened
and allowed to creep. Failure took place after 175 h and
to an intermediate degree. The gradient of the effec-
248 h respectively but the deformation occurring in the
tive stress amplitude was much steeper than that cal-
un-failed notches was almost identical, one example
culated for the present work.
being given in Figure 18. It is noted that the surface pro-
file has deformed into an ellipse. The proposal is to use
8.4 Energy considerations
this observation to predict notch root strains in the mini-
The equivalent gauge length is proposed as a useful con- mum section without enquiring into the precise deforma-
cept for describing overall behaviour and also for deduc- tion mechanisms occurring elsewhere in the surface.
ing the failure energy of the notched specimens. The The profile of the original semi-circular notch in
testing machine was equipped with an ‘expended energy Figure 18 is given by:
monitor’ which integrated the area of all axial hysteresis
loops. The input parameters at the outset were (i) the x2 + y2 = R2i (11)

150 MATERIALS AT HIGH TEMPERATURES 18(3)


R.P. Skelton et al.: Cyclic stress–strain behaviour of circumferentially notched cylindrical bars at high temperature

there was no net displacement after testing so informa-


tion on this aspect has again been provided on another
unidirectional test undertaken on 316 steel. Here a dou-
ble-notched specimen (geometry as given in Section 2)
was loaded to a net section stress of 380 MPa in 6 min.
The results were as follows:

Notch mouth opening displacement (average of two)


0.050 mm
Remote displacement on axial extensometer
0.054 mm
Since experiments in reversed deformation do not pro-
vide easily measurable permanent dimensional changes,
these results in unidirectional loading give further confi-
dence in the interpretation of cyclic data.

9. CONCLUSIONS
The equivalent gauge length of a semicircular notch
given by equation (7) could prove a useful concept for
representing a rather complex strain profile. However it
is based on the widely quoted equation (2) [19-23] as a
limiting value. Similarly, the equivalent stress accompa-
nying the equivalent strain is given by the equally famil-
iar equation (4) [16, 17, 21]. Notwithstanding the pres-
ence of a microstructural hardening gradient, there is
thus a case for studying these relations more closely in
order to superimpose the curves shown in Figure 16.

10. ACKNOWLEDGEMENTS
The authors would like to thank Mr M. W. Spindler of
British Energy Generation Ltd. for supply of the 316
steel and communicating details of the notch ‘first load-
Figure 18 Shadowgraph trace showing final notch shape in 316 steel ing’ analysis. Helpful discussions were also held with Dr
after 248 h of creep at 550°C. I. W. Goodall.

where the initial notch radius Ri = 1.17 mm. After creep REFERENCES
deformation the profile is given by: [1] Loveday, M. S. and Dyson, B. F., Creep deformation and cavitation
damage in Nimonic 80A under a triaxial stress. In: ICM 3, Volume 2,
x 2 y' 2 Pergamon Press, pp. 213–222 (1979).
+ =1 (12) [2] Dyson, B. F. and Loveday, M. S. Creep fracture in Nimonic 80A
R2 Rf2 under triaxial stressing. In: Creep in Structures (Eds ARS Ponter and
D R. Hayhurst), Springer-Verlag, Berlin, pp. 406–421 (1981).
[3] Hayhurst, D. R., Leckie, F. A. and Henderson, J. T. Design of
where the final notch ‘radius’ (ellipse semi-major axis in notched bars for creep-rupture testing under tri-axial stress. Int. J.
the axial “y”direction) Rf = 1.36 mm. Manipulation of Mech. Sci., 19, 147–159 (1977).
these relations gives: [4] Hayhurst, D. R. and Henderson, J. T. Creep stress distributions in
notched bars. Int. J. Mech Sci., 19 133–146 (1977).
[5] Hayhurst, D. R. and Webster, G. A. An overview on studies of stress
y' – y Rf – Ri state effects during creep of circumferentially notched bars. In:
= (13) Techniques for Multiaxial Creep Testing (Eds D. J. Gooch and I. M.
y Ri How), Elsevier Applied Science, pp. 137–175 (1986).
[6] Eggeler, G. and Wiesner, C. A numerical study of parameters con-
trolling stress redistribution in circular notched specimens during
Equation (13) demonstrates that the apparent strain as creep. J. Strain Anal., 28, 13–22 (1993).
measured at any co-ordinate (x, y´) at the probe tips [7] Crews, J. H. and Hardrath, H. F. A study of cyclic plastic stresses at a
depends only on the term Rf which is constant for a given notch root. Exper. Mech., 6, 313–320 (1966).
deformation. In the case of Figure 18, this strain is ~16%. [8] Neuber, H. Theory of stress concentration for shear-strained prismati-
cal bodies with arbitrary non-linear stress–strain law. Trans. ASME,
Thus according to our approximate calculation (which Ser. E, 28, 544–550 (1961).
assumes the same shift in the “x” direction for all points [9] Molski, K. and Glinka, G. A method of elastic-plastic stress
on the surface), this value of ‘strain’, determined at the and strain calculation at a notch root. Mater. Sci. Eng., 50, 93–100
notch mouth at one extreme, is a measure of the actual (1981).
[10] Polak, J. Stress and strain concentration factor evaluation using the
strain experienced by the notch root at the other extreme. equivalent energy concept. Mater. Sci. Eng., 61, 195–200 (1983).
The remaining step is to compare notch mouth dis- [11] Ohji, K. Ogura, K. and Takii, H. Elastic-plastic analysis of notched
placement with that measured (axially and) remotely, as specimens using the finite element method. 14th Japanese Cong.
indicated in Figure 4a, b. In the present cyclic work, Materials Research, Kyoto, Japan Soc. mater. Sci., pp. 170–173 (1971).

MATERIALS AT HIGH TEMPERATURES 18(3) 151


R.P. Skelton et al.: Cyclic stress–strain behaviour of circumferentially notched cylindrical bars at high temperature

[12] Mowbray, D. F. and McConnelee, J. E. Application of finite elastic- [23] Contesti, E., Cailletaud, G. and Levaillant, C. Creep damage in 17–12
plastic stress analysis to notched fatigue specimens. 1st Conf. Reactor SPH stainless steel notched specimens: Metallographical study and
Structural Mechanics, Berlin, Paper M6/8, pp. 425–441 (1971). numerical modelling. Trans. ASME J. Pressure Vessel Technol., 109,
[13] Majima, T. Strain-concentration factor of circumferentially notched 228–235 (1987).
cylindrical bars under static tension. J. Strain Anal., 34, 347–360 [24] Loveday, M. S. Practical aspects of testing circumferential notch
(1999). specimens at high temperature. In: Techniques for Multiaxial Creep
[14] Shatil, G., Ellison, E. G. and Smith, D. J. Elastic-plastic behaviour Testing (Eds D. J. Gooch and I. M. How). Elsevier Applied Science,
and uniaxial low cycle fatigue life of notched specimens. Fatigue London, pp. 177–197 (1986).
Fract. Eng. Mater. Struct., 18, 235–245 (1995). [25] Hales, R. Fatigue testing methods at elevated temperatures. In:
[15] Spindler, M. W., Hales, R. and Skelton, R. P. The multiaxial creep Fatigue at High Temperature (Ed. R. P. Skelton), Applied Science
ductility of an ex-service type 316H stainless steel. In: Creep and Publishers,London, pp. 63–96 (1983).
Fracture in Engineering Materials and Structures (ed. J. D. Parker), [26] Walters, D. J. and Loveday, M. S. Strain measurements by contact
Institute of Materials, London, pp. 679–688 (2001). methods and extensometry. In: Materials Metrology and Standards
[16] Webster, G. A., Aplin, P. F., Cane, B. J., Dyson, B. F. and Loveday, for Structural Performance (eds B. F. Dyson et al.), Chapman &
M. S. A code of practice for notched bar creep rupture testing: proce- Hall, London, pp. 81–113 (1995).
dures and interpretation of data for design, National Physical labora- [27] Kerr, D. C., Nikbin, K. M. Webster, G. A. and Walters, D. J. Creep
tory (1991). Also in: Harmonisation of Testing Practice for High strain determinations across the root of a notch. In: Local Strain and
Temperature Materials (Eds M. S. Loveday and T. B. Gibbons), pp. Temperature Measurements in Non-Uniform Fields at Elevated
295–330, Elsevier Applied Science (1992). Temperatures (Eds J. Ziebs et al.), Woodhead Publishing Ltd., pp.
[17] Al-Abed, B., Timmins, R., Webster, G. A. and Loveday, M. S. 263–273 (1999).
Validation of a code of practice for notched bar creep rupture testing: [28] ESIS P8-99D: Draft code of practice for the determination and inter-
Procedures and interpretation of data for design. Mater. High Temp., pretation of cyclic stress–strain data, ESIS TC11, High Temperature
16, 143–158 (1999). Mechanical Testing, November (1999).
[18] Holt, P. J. and Spindler, M. W. A practical application of continuum [29] Skelton, R. P. Effect of strain rate on the high strain fatigue proper-
damage mechanisms to plant integrity assessment. In: Symposium on ties of a 20%Cr25%Ni/Nb stabilised stainless steel at 1025K. Mater.
Inelasticity and Damage in Solids subject to Microstructural Change. Sci.. Eng., 11, 195–202 (1973).
Faculty of Engineering & Applied Science, Memorial University of [30] Skelton, R. P. Cyclic stress–strain properties during high strain
Newfoundland, Canada, Sept 25–27 (1996). fatigue. In: High Temperature Fatigue: Properties and Prediction (ed.
[19] Bridgman, P. W. Studies in Large Plastic Flow and Fracture, 2nd R. P. Skelton), Elsevier Applied Science, London, pp. 27–112 (1987).
edn, Harvard (1964). [31] Ohno, N. and Satra, M. Detailed and simplified elasto-plastic analy-
[20] Earl, J. C. and Brown, D. K. Distributions of stress and plastic strain ses of a cyclically notched loaded bar. Trans. ASME, J. Eng. Mater.
in circumferentially notched tension members. Eng. Fract. Mech., 8, Tech., 109, 194–202 (1987).
599–611 (1976). [32] Skelton, R. P. Energy criterion for high temperature low cycle fatigue
[21] Lubahn, J. D. and Felgar, R. P. Plasticity and Creep of Metals, failure, Mater. Sci. Technol., 7, 427–439 (1991).
Wiley, New York (1961). [33] Holdsworth, S. R., Holt, A. and Scholz, A. Experience with pecker
[22] Yoshida, M., Levaillant, C. and Pineau, A. Metallographic measure- capacitance gauges for the measurement of local strains at high tem-
ment of creep intergranular damage and creep strains – influence of peratures. In: Local Strain and Temperature Measurements in Non-
stress state on critical damage at failure in an austenitic steel. Proc. Uniform Fields at Elevated Temperatures (Eds J. Ziebs et al.),
Int. Conf. On Creep, Tokyo, Japan, pp. 327–332 (1986). Woodhead Publishing Ltd., pp. 128–137 (1999).

152 MATERIALS AT HIGH TEMPERATURES 18(3)

También podría gustarte