See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/245394759

The stress relaxation and creep behaviour of a manganese-stabilized

austenitic stainless steel

Article  in  The Journal of Strain Analysis for Engineering Design · March 2009

DOI: 10.1243/03093247JSA476

CITATIONS READS

4 209

2 authors, including:

Jonathan Beddoes
University of Manitoba
96 PUBLICATIONS   1,356 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Jonathan Beddoes on 11 August 2015.

The user has requested enhancement of the downloaded file.

 201

The stress relaxation and creep behaviour of a

manganese-stabilized austenitic stainless steel
A Pagliarello and J Beddoes*
Mechanical and Aerospace Engineering, Carleton University, Ottawa, Canada

The manuscript was received on 5 October 2008 and was accepted after revision for publication on 19 January 2009.

DOI: 10.1243/03093247JSA476

Abstract: The stress relaxation behaviour of 21–4N, a manganese-stabilized austenitic

stainless steel, is investigated in terms of the metallurgical state, the application of multiple
strain levels during ‘stepped’ stress relaxation testing at 700 uC, the strain level during isostrain
stress relaxation tests at 538 uC and 700 uC, and the correspondence with results from constant-
load creep tests. The results indicate that for isostrain stress relaxation tests the stress
relaxation rate is similar for strains that span both elastic and plastic strain levels. A transition
in the stress relaxation behaviour occurs at a stress level approximately equivalent to the tensile
stress–strain proportional limit; below this transition the stress–strain rate relationship, or the
time predicted for 1 per cent creep strain, obeys a creep power law type of equation. Stress
relaxation testing successfully delineates the difference between the creep resistances of two
different metallurgical conditions with similar tensile properties using fewer specimens and
requiring less time. The time to 1 per cent creep strain determined from the analysis of stress
relaxation results is always less than the actual time to 1 per cent creep strain during constant-
load creep tests.

Keywords: stress relaxation, creep, austenitic stainless steel

1 INTRODUCTION standing of the underlying mechanisms controlling

creep in a specific alloy is central to designing alloy–
The desired service life of many elevated-tempera- microstructure combinations with improved creep
ture structural components is of the order of tens or resistance.
hundreds of thousands of hours. For these applica- The stress relaxation test, during which strain is
tions, deformation due to creep is frequently life held constant and the stress decreases, potentially
limiting. Consequently, long-term tests, sometimes offers the opportunity to determine creep strain
exceeding 100 000 h, are undertaken to characterize rates in a considerably shorter time period, namely
creep properties. Aside from the requirement for hours to days for stress relaxation versus weeks to
long test times, creep tests are relatively easily months to years for creep testing. Such an economy
undertaken. However, from these relatively easy of time is of obvious value. The stress relaxation
tests it can prove quite difficult to ascertain the technique is predicated on a constant total strain
underlying metallurgical processes that control the during the test, which has both an elastic and plastic
creep deformation strain rate. The controlling component, so that
process varies with temperature and stress as well
as metallurgical state which may undergo micro- eT ~ee zep ~C ð1Þ
structural changes during long-term tests at the
associated elevated temperatures. A clear under- where eT is the total strain, ee is the elastic strain, ep is
the plastic strain, and C is a constant [1, 2]. As the
*Corresponding author: Department of Mechanical and Aero- stress relaxation test proceeds at temperature, the
space Engineering, Carleton University, 1125 Colonel By Drive, elastic stress relaxes and elastic strain is replaced by
Ottawa, Ontario, K1S 5B6, Canada. email: plastic strain through creep deformation processes,
jbeddoes@mae.carleton.ca which may be a function of stress. Substituting

JSA476 F IMechE 2009 J. Strain Analysis Vol. 44

202 A Pagliarello and J Beddoes

Hooke’s law s 5 Eee, where E is Young’s modulus and the yield stress are necessary to avoid anelastic-
s is the stress, into equation (1) and differentiating associated transient behaviour during the initial
with respect to time provide stages of stress relaxation [9, 10]. The short time
required for stress relaxation tests, compared with
dep 1 ds creep tests, allows multiple relaxation tests to be
~{ ð2Þ
dt E dt undertaken on a single sample, with the resulting
creep strength predicted insensitive to thermal and
relating the stress rate during a stress relaxation test mechanical history [9]. An alternative stress relaxa-
to the plastic or creep strain rate. Such stress tion method is the ‘multiple hardening–relaxation’
relaxation tests have been referred to as self- test in which the strain is increased in a stepwise
programmed variable-stress creep tests [3], with manner with a time interval at each step during
the benefit that a wide range of creep strain rates, which relaxation occurs [5].
corresponding to a range of applied stresses, occur In this paper the stress relaxation behaviour of 21–
in a relatively short time. Combining this short test 4N, a manganese-stabilized austenitic stainless steel,
time with the fact that the total strain is constant and is investigated in terms of the metallurgical state, the
the accumulated plastic strain is minimal offers the application of multiple strain levels during ‘stepped’
possibility to undertake several relaxation events on stress relaxation testing, the strain level during
one sample without any significant change in the isostrain stress relaxation tests, and the correspon-
metallurgical state. dence with results from constant-load creep tests.
Stress relaxation results have been used to esti- The purpose is to evaluate the potential to reduce
mate the creep properties of various elevated- the testing time required to characterize the creep
temperature alloys [4–8]. The creep rupture life for behaviour of elevated-temperature structural mate-
industrial gas turbine blades can be estimated within rials by using stress relaxation methods.
a factor of 3 via stress relaxation testing [4]. For a
nickel superalloy gas turbine disc material the creep
curves were computed from stress relaxation tests 2 MATERIALS AND EXPERIMENTAL
[5], albeit for creep rupture lives of less than 10 h. PROCEDURE
Likewise, for a Cr–Mo–V steam turbine rotor steel,
stress relaxation tests accurately predicted creep Material for this study was cut from hot-rolled 21–4N
strain rates of about 1024 h21 at 137 MPa, but at bars of 22 mm diameter with the composition listed
100 MPa the strain rate discrepancy is nearly an in Table 1 as determined by optical emission
order of magnitude [6]. Regardless of the accuracy spectroscopy for the major elements. Three heat-
with which stress relaxation tests predict creep strain treated conditions were prepared according to the
rates, relatively short-duration (about 1 day) stress thermal profiles of Table 2. The microstructure
relaxation tests revealed the deterioration of the resulting from the standard (Std) heat treatment
temperature capability of one group of superalloy has been described in reference [11], and those for
blades [4], a result that would probably require creep the normal grain boundary (NGB) and serrated grain
tests of much longer duration. boundary (SGB) heat treatments have been de-
Differences in stress relaxation test methodology scribed in reference [12].
exist. The amount of pre-strain prior to stress All specimens for tensile, creep, and stress relaxa-
relaxation varies from as low as 0.2 per cent [6, 7, tion tests were machined to a straight gauge length
9] prior to yielding, to in excess of 1.0 per cent [5–9] of 40.6 mm and diameter of 6.3 mm. Grooves were
after yielding. For a directionally solidified nickel machined into specimen shoulders for attachment
superalloy loaded parallel to the directional struc- of a high-temperature extensometer with a strain
ture the stress necessary for a relaxation strain rate accuracy of at least ¡161024 equipped with a linear
of 1029 s21 was considerably lower for a pre-strain of variable-differential transducer (LVDT). For all me-
0.2 per cent than for pre-strains of 0.4–1.5 per cent chanical testing, the temperature from thermocou-
[7]. In contrast, for Cr–Mo–V steel the stress ples attached to the specimen shoulders was con-
necessary for a relaxation strain rate of 1029 s21 trolled to ¡1 uC. The yield and ultimate tensile
was lower at higher pre-strains of 0.8 per cent and 1 strengths were determined at 700 uC with a loading
per cent than for pre-strains of 0.2–0.6 per cent [6], strain rate of 1.761023 s21.
with the higher pre-strains correlating better with Creep testing was conducted in Satec M3 creep
creep rupture data. This concurs with the conclusion frames at a constant load and temperature, corre-
that pre-strains greater than those corresponding to sponding to an initial stress of 196–300 MPa at

J. Strain Analysis Vol. 44 JSA476 F IMechE 2009

 Behaviour of a manganese-stabilized austenitic stainless steel 203

Table 1 Nominal and actual composition of 21–4N

Concentration (wt%)
C N Cr Mn Ni Mo Si P S
Nominal [11] 0.45–0.60 0.30–0.50 20–23 7–10 3–5 – 0.25 maximum – –
This study* 0.52 0.39 20.6 8.3 3.35 0.25 0.19 0.030 0.001

*Also contains 0.21 wt% Cu, 0.089 wt% V, 0.035 wt% Co, 0.029 wt% W, and 0.020 wt% Nb.

Table 2 Heat treatment details

Condition Heat treatment*
Std 2 h at 1175 uC AC + 14 h at 760 uC AC
NGB 1 h at 1200 uC WQ + 30 h at 700 uC AC + 30 h at 1000 uC AC
SGB 1 h at 1200 uC FC to 1030 uC WQ + 30 h at 750 uC AC + 3 h at 1000 uC AC

*FC, furnace cool (6 uC/min); WQ, water quench; AC, air cool.

700 uC. The creep load was applied all at once via a to control continuously the cross-head to ensure
motorized drawhead, with strain measurement in- that constant strain was maintained.
itiated prior to loading. During and immediately
after loading, strain data were recorded every
second, with this time interval gradually increasing 3 RESULTS
to a maximum of 30 min as the test proceeded.
Stress relaxation tests were conducted in Satec Since limited elevated-temperature tensile data are
DLF-20 stress relaxation frames of 98 kN capacity available for 21–4N, samples of the NGB and SGB
equipped with load cells and mechanically driven conditions were tensile tested at 700 uC with the
cross-heads capable of controlled loading or main- results listed in Table 3, as are reference values for
taining a constant strain level. Two types of stress the Std condition at 650 uC and 760 uC [11]. The
relaxation test were undertaken. Stepped stress modification of the heat treatment between the NGB
relaxation tests at 700 uC consisted in loading to a and SGB conditions causes minimal change in
strain level for a period of time (usually 1000 s), tensile properties. The proportional limit in Table 3
followed by stepwise increases in the strain, with is the lowest stress at which the stress–strain
each step 0.5 per cent strain at a strain rate of about response deviates from a linear Hooke’s law relation-
4.361025 s21. These stepped tests are similar to the ship. Young’s modulus E is estimated from the
multiple hardening–relaxation tests reported in average stiffness measured during the elastic portion
reference [5]. The second type of test, namely iso- of the stress–strain response during loading for
strain stress relaxation tests, at 700 uC or 538 uC isostrain stress relaxation tests, consistent with
involved loading at a strain rate of about 1.56 previous stress relaxation testing [3, 7, 9, 10]. The
1023 s21 to a single strain level of 0.3 per cent, 1.0 per number of such tests included in each estimate of
cent, 1.5 per cent, 2.0 per cent, or 2.5 per cent for 20– Young’s modulus is given in parentheses in Table 3.
28 h. Throughout all tests, displacement and load Data for the elevated-temperature Young’s modulus
data were recorded every second from the LVDT- of manganese-stabilized austenitic steel are not
equipped extensometer and a load cell. During stress readily available. Young’s modulus estimated at
relaxation periods the extensometer signal was used 700 uC in Table 3 is below the range 138–147 GPa

Table 3 Elevated-temperature tensile results

Proportional limit 0.2% yield strength Ultimate tensile
Condition Temperature (uC) (MPa) (MPa) strength (MPa) Elongation (%) E, estimate* (GPa)
Std 538 108 (1)
Std [11] 650 331 593 16
Std 700 119 (3)
Std [11] 760 255 427 18
NGB 700 234 306 558 20 113 (4)
SGB 700 214 292 527 21 115 (4)

*The number of tests included in each estimate of Young’s modulus is given in parentheses.

JSA476 F IMechE 2009 J. Strain Analysis Vol. 44

204 A Pagliarello and J Beddoes

for several grades of austenitic stainless steel, none

of which was manganese stabilized [13].
The results of constant-load creep tests are
summarized in Table 4 with example creep strain–
time curves shown in Fig. 1. Despite the similar
tensile properties of Table 3 it is apparent from
Table 4 and Fig. 1(a) that the NGB condition has
better creep resistance than the SGB condition, in
terms of the minimum strain rate, time to 1 per cent
strain, and rupture life. The SGB condition does
exhibit a higher rupture strain due to greater tertiary
creep, but with a corresponding decrease in rupture
life. Figure 1(a) illustrates that, when loaded to
250 MPa, beyond the proportional limit but below
the yield stress, the SGB condition exhibits a higher
minimum strain rate, an earlier onset of tertiary
creep, and a shorter rupture life. The same trends
exist when loaded to 196 MPa, below the propor-
tional limit, or at 900 uC and 925 uC [12]. Figure 1(b)
shows that creep is strongly dependent on stress.
This is to be expected given the relatively high
stresses applied; at 700 uC the 300 MPa creep stress is
probably beyond the yield stress (Table 3), and even
the 230 MPa creep stress is probably close to the
proportional limit. Based on the minimum strain
rates of Table 4, the creep stress exponent n between
230 MPa and 300 MPa for the Std condition accord-
ing to
Fig. 1 Creep properties of (a) 21–4N heat treated to
e_ ~Csn ð3Þ the NGB or SGB state at 250 MPa and 700 uC
and (b) 21–4N heat treated to the Std state and
is 8. For oxide-dispersion-strengthened alloys, stress tested at 700 uC and 230 MPa or 300 MPa
exponents of 8 or higher occur at high stresses [14].
For most metals and alloys at low stresses, n 5 1 and, and (b) respectively. From Fig. 2(a) it is clear that the
at intermediate stresses, n 5 3–5 [15, 16]. At high load frame is capable of applying well-defined strain
creep stresses the stress exponent increases rapidly steps and maintaining constant strain during each
with increasing stress, a phenomenon often referred hold period with minimal strain overshoot. The first
to as power law breakdown [15, 16]. In the current strain increment of 0.5 per cent causes a peak stress
case of the Std condition, n 5 8 and power law of 289 MPa, close to the 0.2 per cent yield stress of
breakdown is consistent with the high creep stress 306 MPa and well above the proportional limit
levels of 230–300 MPa that span the range from the (Table 3). During each constant strain period of
proportional limit to the yield stress. 1000 s the stress relaxes at a decreasing rate. At each
The strain–time and stress–time response for a successive strain level, both the peak stress and the
stepped stress relaxation test are shown in Figs 2(a) stress after 1000 s of relaxation increase, with the
Table 4 Summary of creep results at 700 uC
Condition Stress (MPa) Minimum strain rate (h21) Time to 1% strain t1% (h) Rupture life (h) Rupture strain (%)
24
Std 230 1.2610 44 229 4.2
300 1.661023 3.6 12.5 2.8
NGB 196 4.061025 144 531 6.2
196 3.761025 188 584 7.0
250 9.361025 67 252 4.1
SGB 196 7.161025 80 465 13.5
196 5.861025 114 504 17.6
250 3.661024 10.5 148 11.7

J. Strain Analysis Vol. 44 JSA476 F IMechE 2009

 Behaviour of a manganese-stabilized austenitic stainless steel 205

Fig. 3 Isostrain stress relaxation results for 21–4N in

the Std condition at 538 uC and a constant
strain of 1.0 per cent. The stiffness during the
loading transient was used for estimate of E in
Table 3

Fig. 2 Result of a stepped stress relaxation test on the

NGB condition, in terms of (a) strain–time and
(b) stress–time. D indicates the difference
between the maximum stress and the mini-
mum stress at each strain increment

difference between the peak stress and the 1000 s

relaxed stress increasing with each strain increment Fig. 4 Isostrain stress relaxation results for 21–4N in
(Fig. 1(b)). As such, the average stress relaxation rate the Std condition at 700 uC for three different
increases after each successive strain increment. constant strains. The 1.5 per cent and 2 per cent
Importantly during this stepped stress relaxation test plots are shifted +3 h and +6 h respectively to
the stress never decreases to below 228 MPa or 75 the left for clarity. The average of the stiffnesss
during the loading transient from all three tests
per cent of the 0.2 per cent yield stress.
was used to determine the estimate of E in
The stress–time response for an isostrain (e 5 1 per Table 3
cent) stress relaxation test of the Std state at 538 uC is
shown in Fig. 3. The stress–strain response during
the elastic portion of the initial loading period per cent cause a much larger peak stress of about
generated the estimate of Young’s modulus E 5 108 400 MPa, well above the yield stress, but again
GPa, in Table 3. During the relaxation period the relaxation to about 200 MPa (roughly half the peak
stress monotonically decreases very rapidly from value) is very rapid. Subsequent stress relaxation
about 425 MPa to 375 MPa and, after this, much below 200 MPa occurs at virtually the same rate
more slowly at an ever-decreasing rate. regardless of the strain level, with all three curves of
Figure 4 shows the influence of the strain applied Fig. 4 superimposed on one another. Similarly to the
on isostrain stress relaxation of the NGB condition. isostrain test at 538 uC, at long relaxation times it
At 0.3 per cent strain the peak stress reaches almost appears that the stress asymptotically approaches a
250 MPa, below the yield stress, and rapidly relaxes threshold stress of approximately 325 MPa in Fig. 3
to about 200 MPa. Higher strains of 1.5 per cent or 2 and about 100 MPa in Fig. 4.

JSA476 F IMechE 2009 J. Strain Analysis Vol. 44

206 A Pagliarello and J Beddoes

the time to 1 per cent creep strain. From Table 4 the

time to 1 per cent creep strain during constant-load
creep tests for the NGB condition is only 67 h or
about 166 h at 250 MPa and 196 MPa respectively.
Given that most elevated-temperature applications
involve service times of 10 000–100 000 h, stress
relaxation testing is more useful at stress levels
considerably below those of Fig. 2(b). Previous
stepped stress relaxation results were successfully
used to generate creep strain–time data, but only for
creep lives of a few hours [5]. The second advantage
of stepped stress relaxation tests of generating data
for multiple strain levels is of limited value since, as
discussed next, it appears that the strain level has no
Fig. 5 Isostrain stress relaxation results for the NGB influence on the stress relaxation rate.
and SGB conditions tested at 700 uC and By calculating ds/dt for various stress levels from
constant strain of 2.5 per cent. The stiffnesses the relaxation data of Fig. 4, and using the estimate
during the loading transient of these tests and
of Young’s modulus from Table 3, the relaxation
three other tests for each condition at different
strain levels were used to estimate E in Table 3 strain rate can be calculated according to equation
(2). Using this strain rate the time to 1 per cent creep
Figure 5 compares the isostrain (e 5 2.5 per cent) strain can be estimated (t1% 5 0.01/ė) with the
stress relaxation behaviour of the NGB and SGB resulting stress versus time to 1 per cent strain
conditions. The peak stress on loading of nearly plotted in Fig. 6. For all three strain levels, including
500 MPa is close to the ultimate strength (Table 3) strains corresponding to elastic and plastic initial
for these conditions, and, similarly to the results of loading, the predicted times to 1 per cent creep
Figs 3 and 4, when the peak stress is considerably strain are very similar, suggesting that, for the range
higher than the yield stress, the stress rapidly relaxes of strains investigated, there is no influence on the
to about half the peak stress. Again similarly to the relaxation strain rate, nor consequently on the time
results of Figs 3 and 4, at long times the stress to 1 per cent creep strain, or any detrimental
relaxation of the NGB and SGB conditions asympto- anelastic effects as reported elsewhere [9, 10]. It
tically approaches threshold stresses of about appears that the stress relaxation rate may be
150 MPa and 75 MPa respectively.

4 DISCUSSION

From the standpoint of defining creep properties

there are two potential justifications for undertaking
stress relaxation tests: first, the ability to predict the
creep strain rate as a function of stress from one
sample in a much shorter time than for creep testing;
second, identification of the mechanisms controlling
the creep strain rate, again in a shortened time
period. The results presented are discussed in terms
of these two potential justifications.
The stepped stress relaxation test method allows
data for multiple strain levels to be generated in only Fig. 6 Comparison of the predicted time to 1 per cent
a few hours. However, as evident from Fig. 2(b), for strain for three stress relaxation tests on the Std
the NGB condition the minimum relaxed stress is condition at 700 uC, each at the constant strain
level indicated. The full squares indicate the
228 MPa, about 75 per cent of the yield strength. At actual time to 1 per cent strain from the creep
these relatively high stresses, creep tests will not tests results of Table 4. The open squares
typically require overly extended times. It is often of indicate the time to 1 per cent strain predicted
interest to use stress relaxation test results to predict from the minimum creep strain rate of Table 4

J. Strain Analysis Vol. 44 JSA476 F IMechE 2009

 Behaviour of a manganese-stabilized austenitic stainless steel 207

controlled by the difference between the instanta-

neous stress and the asymptotic threshold stress,
about 100 MPa in Fig. 4. The significance of this
threshold stress will be investigated as part of
ongoing research.
Interpolating from the data of Table 3 it is
estimated that the proportional limit for the Std
heat-treated condition at 700 uC is about 220 MPa.
Interestingly at about this same stress the results of
Fig. 6 exhibit a transition. Below 220 MPa the stress–
t1% function is logarithmically linear, indicating a
power law stress–strain rate relationship (equation
(3)), as would be expected during creep testing.
Above 220 MPa the stress–t1% slope becomes in-
creasingly more negative, consistent with increasing
plasticity. This result may be an indication of power Fig. 7 Comparison of the time to 1 per cent creep
strain from stress relaxation tests on the NGB
law breakdown as suggested by the stress exponent
and SGB conditions at 700 uC and constant
of 8 predicted from the 230 MPa and 300 MPa creep strain of 2.5 per cent: +, 6, actual time to 1 per
of Table 4. However, care in interpreting results cent strain from the creep tests results of
above the proportional limit is required, since Table 4; o, fl, time to 1 per cent strain pre-
equation (2) on which this analysis is based assumes dicted from the minimum creep strain rate of
a Hooke’s law stress–strain relationship, an assump- Table 4
tion that loses validity as the stress increases beyond
the proportional limit. to which relatively short-time stress relaxation test
Despite the fact that both the NGB and the SGB results can be used to ascertain the underlying
conditions exhibit similar tensile properties (see deformation mechanism controlling the strain rate
Table 3), the creep properties of these two condi- during long-term creep tests.
tions differ substantially (Fig. 1(a) and Table 4), with For comparison with the results from stress
the SGB condition less creep resistant. Repeating the relaxation tests, the actual times to 1 per cent strain
foregoing analysis for the stress relaxation data of determined from creep tests of the NGB and SGB
Fig. 5 results in the stress–t1% plot of Fig. 7 for these conditions at 196 MPa and 250 MPa (Table 4) are
two conditions. Importantly, the stress relaxation plotted in Fig. 7. Likewise, in Fig. 6 the actual times
results exhibit the same trend as the creep results, to 1 per cent creep strain for the Std state at 230 Mpa
namely that the time to 1 per cent creep strain for and 300 MPa, also from Table 4, are plotted. The
the SGB condition is less than for the NGB condition. time to 1 per cent creep strain includes a contribu-
Figure 7 summarizes the outcome of two stress tion due to the instantaneous and primary creep,
relaxation tests totalling 40 h duration that generated which is not reflected in equation (2) or the stress
stress–t1% data over a range of several hundred relaxation results. Therefore, using the minimum
megapascals. In contrast, the corresponding six creep strain rate results from Table 4, a modified
creep tests of Table 4 required a total of more than time to 1 per cent creep strain is calculated and also
600 h to determine the time to 1 per cent strain for plotted in Figs 6 and 7.
only two stress levels. This result indicates that stress Figure 8 shows the same analysis based on Fig. 3 for
relaxation testing can be used to delineate the the Std condition tested at 538 uC, and 10 000 h, 1 per
relative creep resistances of different microstructural cent creep data from reference [11]. In Figs 6 to 8,
conditions, even those with similar tensile proper- stress relaxation testing gives a predicted time to 1 per
ties, in a much more effective manner in terms of cent strain that is less than the actual time from creep
time and the number of samples required. By tests or those calculated from the minimum creep
examining the deformation structures developed strain rate by a factor of about 5, slightly greater than
during stress relaxation and creep testing, continu- the factor of 3 reported previously [4], but within the
ing research will investigate whether the same order-of-magnitude range reported in reference [6].
deformation mechanistic processes occur during These results indicate that the strain rate as a function
stress relaxation and creep. It is the long-term of stress during stress relaxation is greater than the
objective of such research to determine the extent strain rate for the same stress during creep. Two

JSA476 F IMechE 2009 J. Strain Analysis Vol. 44

208 A Pagliarello and J Beddoes

nitic stainless steel, provides the following conclu-

sions.

1. For isostrain stress relaxation tests undertaken at

one strain level, the stress relaxation rate is
similar for strains in the range from 0.3 per cent
to 2.0 per cent, a range that spans both elastic and
plastic strain levels.
2. Stepped stress relaxation tests in which the strain
level is incrementally increased are only useful for
determining stress relaxation behaviour at rela-
tively high stresses. Further, given the first
conclusion, there is minimal incremental value
to the application of multiple strain levels to 21–
4N.
3. A transition in the stress relaxation behaviour
Fig. 8 Time to 1 per cent creep strain from the
isostrain stress relaxation test on 21–4N in the occurs at a stress level approximately equivalent
Std condition at 538 uC and constant strain of 1 to the stress–strain proportional limit, i.e. the
per cent. The solid circle indicates the actual point at which Hooke’s law no longer applies.
time to 1 per cent strain from the creep results Below this transition the stress–strain rate or the
of reference [11] time predicted for 1 per cent creep strain obeys a
relationship similar to the creep power law.
factors may contribute to this difference. The first
4. In much less time and using fewer specimens,
factor is the accuracy of the Young’s modulus
stress relaxation testing successfully delineated
equation (2) analysis. Young’s modulus measured
the difference between the creep resistances of
during the elastic loading portion of stress relaxation
different metallurgical conditions despite the fact
tests is lower than might be expected from reference
that these two conditions have similar tensile
[12]. Clearly, to obtain accurate strain rate and
properties.
predicted times to 1 per cent creep strain an accurate
5. For 21–4N the time to 1 per cent creep strain
value for Young’s modulus is required. The second
determined from the analysis of stress relaxation
factor is that, during creep testing, the strain rate
data is always less than the actual time to 1 per
observed is the sum of two components: that due to
cent creep strain during constant-load creep tests.
the underlying mechanism controlling creep, and that
due to the accumulation of creep damage such as the
formation of intergranular voids or cracking. It is
difficult to separate the contributions to the creep ACKNOWLEDGEMENTS
strain rate of each of these components. The constant-
strain characteristic of stress relaxation testing means This research was funded by the Natural Sciences
that damage mechanisms are not active and the strain and Engineering Research Council of Canada
rate observed is only that due to the underlying through the Discovery grant programme. Thanks
mechanism controlling deformation. Therefore, it are due to Crucible Specialty Metals, Syracuse, New
might be expected that stress relaxation testing would York, USA, for supply of the 21–4N bars. The authors
predict a longer time to 1 per cent strain than creep are grateful to F. Barrett and K. Sangster for
testing, i.e. the opposite of the trend in Figs 6 to 8. assistance with some experimental aspects of this
Future research will confirm whether this trend is research.
unique to 21–4N or is observed in other elevated-
temperature alloys, as well as seek the reason for this
REFERENCES
trend through examination of deformed structures.
1 Soderberg, C. R. The intrepretation of creep tests
for machine design. Trans. ASME, 1936, 58,
5 SUMMARY AND CONCLUSIONS 733–743.
2 Oding, I. A., Ivanova, V. S., Burdukskii, V. V., and
The investigation of the stress relaxation and creep Geminov, V. N. Creep and stress relaxation in
behaviour of 21–4N, a manganese-stabilized auste- metals, 1959 (Oliver and Boyd, London).

J. Strain Analysis Vol. 44 JSA476 F IMechE 2009

 Behaviour of a manganese-stabilized austenitic stainless steel 209

3 Woodford, D. A. Test methods for accelerated 9 Woodford, D. A. Advances in the use of stress
development, design and life assessment of high relaxation data for design and life assessment in
temperature materials. Mater. Des., 1993, 14, combustion turbines. JSME Int. J., 2002, 45, 98–103.
231–242. 10 Woodford, D. A., Van Steele, D. R., Amberge, K.,
4 Daleo, J. A., Ellison, K. A., and Woodford, D. A. and Stiles, D. Creep strength evaluation for IN738
Application of stress relaxation testing in metallur- based on stress relaxation. In Superalloys 1992 (Eds
gical life assessment evaluation of GTD111 alloy S. D. Antolovich, R. W. Strusrud, R. A. MacKay, D. L.
turbine buckets. Trans. ASME, J. Engng Gas Anton, T. Khan, R. D. Kissinger, and D. L. Klarstrom),
Turbines Power, 1999, 121, 129–137. 1992, pp. 657–664 (TMS, Warrendale, Pennsylvania).
5 Zhan, Z. and Tong, J. Determination of creep 11 Lula, R. A. Manganese in stainless steels, 1986
curves from multiple hardening–relaxation testing. (Manganese Centre, Paris).
J. Strain Analysis, 2003, 38, 277–280. DOI: 10.1243/ 12 Wisniewski, A. and Beddoes, J. Effect of solution
030932403765310608. treatment cooling rate on creep of a Mn–N
6 Bose, S. C., Singh, K., and Jayaraman, G. Applica- stabilized austenitic stainless steel. J. Mater. Sci.,
tion of stress relaxation test methodology for 2008, 43, 5350–5357.
predicting creep life of a large steam turbine rotor 13 Davis, J. R. (Ed.) ASM specialty handbook: stainless
steel (1CrMoV). J. Test. Evaluation, 2003, 31, 1–13. steels, 1994 (ASM International, Materials Park,
7 Woodford, D. A. Creep analysis of directionally Ohio).
solidified GTD111 based on stress relaxation test- 14 Kassner, M. E. and Pérez-Prado, M. T. Funda-
ing. Mater. High Temperatures, 1997, 14, 413–420. mentals of creep in metals and alloys, 2004 (Else-
8 Bose, S. C., Singh, K., Swaminathan, J., and vier, Oxford).
Sarma, D. S. Prediction of creep life of 15 Nabarro, F. R. N. and de Villiers, H. L. The physics
X10CrMoVNbN-91 (P-91) steel through short term of creep, 1995 (Taylor & Francis, London).
stress relaxation test methodology. Mater. Sci. 16 Evans, R. W. and Wilshire, B. Introduction to creep,
Technol., 2004, 20, 1290–1296. 1993 (Institute for Metals, London).

JSA476 F IMechE 2009 J. Strain Analysis Vol. 44

View publication stats