Professional Documents
Culture Documents
net/publication/247151624
CITATIONS READS
17 1,231
3 authors, including:
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
growing of transition metals-filled Carbon Nanotubes on the metal and non-metal materials View project
All content following this page was uploaded by Dr Balhassn S.M. Ali on 26 February 2014.
Department of Mechanical, reference stress parameters have been investigated. On this basis, specimen
Materials and dimension ratio ranges are recommended. The effects of friction, between the loading
Manufacturing Engineering, pins and the specimen surfaces, on the specimen failure time, are also investigated.
University of Nottingham, Test results obtained from two-bar specimen tests and from corresponding uniaxial
Nottingham, specimen tests, for a P91 steel at 600 oC, are used to validate the test method. These
NG7 2RD, UK results demonstrated that the specimen type is capable of producing full uniaxial
creep strain curves. The advantages of this new, small, creep test specimen, for
determining uniaxial creep data, are discussed and recommendations for future
research are given.
1. Introduction
pipes or bends [e.g. 2, 3]. However, the dimensions of a
As the power industry strives toward higher thermal
typical scoop sample are far too small to allow a standard
efficiency and reduced CO2 the use of advanced alloys in
conventional uniaxial specimen, (see Fig. 2), to be extracted
power plant components is inevitable and new generation
from it. Impression creep test specimen [3] and the recently
coal plants in Europe will be designed to operate using
developed small ring creep test specimen [4] can both be
steam at higher temperatures and pressures [1]. Many
removed from scoop samples. They can then be used to
components in power plants operate at temperatures which
determine secondary creep data for materials, but they
are high enough for creep to occur. However obtaining creep
cannot be used to obtain tertiary creep data. Small disc punch
properties for critical, service exposed components remains
creep tests [5] can be used to obtain the creep deformation
a serious challenge. For this reason small or miniature creep
curves up to rupture. However, interpretation of small punch
test specimen types have been developed and used to obtain
creep test data, in relation to the corresponding uniaxial data,
material creep properties. A small scoop sample, (see Fig.
remains difficult, and to date, a universally accepted
1), can be removed from the surface of a component, for
interpretation procedure is not available [6]. Alternatively,
example, using on-site sampling process on thick-welded
small, sub-sized uniaxial specimen tests have been used to
obtain the full creep strain curves. Electron beam welding may
* have to be used to join the gauge length section to the
Corresponding author. Tel.: +44(0) 11584 67683
specimen ends, due to an insufficient volume of material being obeying a Norton’s power law, i.e. ̇ , these show
available, which makes the preparation of the test specimen that the general form is:-
rather complicated [4, 7]. For these reasons, there is a strong ̇ ( ) ( ) (1)
desire for the development of miniature specimen types and where f1 (n) is a function of the stress index, n, and f2
the associated testing techniques from which reliable creep (dimensions) is a function of the component dimensions and
deformation and creep rupture data can be conveniently and nom is a conveniently determined nominal stress for the
economically derived, from a small amount of material. In this component and loading [13-15]. By introducing an
paper, a recently developed, small-sized specimen type, which appropriate scaling factor, , for the nominal stress, Equ. (1)
is suitable for use in obtaining the full uniaxial creep strain can be rewritten as:
curves, including creep rupture life data, is described. The ( )
̇ ( ) ( ) (2)
specimen design, testing procedure and data conversion
By choosing several, random values till the value of (=
methods are described in this paper. Using finite element
( )
analyses the effects of the specimen dimension ratios on the ) so that the term in Equ. (2), becomes independent
reference parameters and the effects of the friction between the (or approximately independent) of n, where n is the stress
loading pin and the specimen surfaces are assessed. In order to index in Norton’s law, then Equ. (2) can be further
assess the accuracy of the test method, a comparison between simplified, i.e.
the results obtained from creep tests using the new specimen ̇ ̇( ) (3)
type, and those from corresponding uniaxial specimen tests, for where D is the so-called reference multiplier, i.e.
a P91 steel at 600oC were carried out. Discussion of the ( )
( )
practical implications of the work also included.
and ̇ ( ) is the minimum creep strain rate obtained from
x ~ 3mm
a uniaxial creep test at the so-called reference stress, i.e.
x (4)
~ 30mm ~ 20mm
The reference multiplier, D, has the units of length, and can
View on x-x
usually be defined by D = d, where d is a conveniently
(a) (b)
chosen, “characteristic”, component dimension. Therefore,
Fig. 1. (a) Photograph of a scoop sample, and (b) Dimensions for the known loading mode and component dimensions,
of a typical scoop sample [8]. can be conveniently defined, and if the values of and
M12×1.75
are known, the corresponding equivalent uniaxial stress
∅ 6.2
∅6
90
2.2. Determination of Reference Parameters. If an
Fig. 2. Conventional uniaxial creep test specimen dimensions analytical solution for the component steady-state creep
3=
4
5
n
n
Moving part
r 1 (1 ) eq (9)
Loading pins where σ1 is the maximum principal stress, σeq is the von-
Loading pin holders
Mises equivalent stress and is a material constant
representing the multi-axial stress state (0 < < 1). The
The TBS parameter can be determined using experimental and FE
creep rupture analysis results for notched bar specimens at
Fig. 5. (a) TBS experiment setup and loading application; 4.3. Specimen Modeling. The 3D FE analyses were carried
(b) the TBS specimen and (c) tested specimen.
out using meshes which consist of 20-noded brick elements.
4. Finite Element Analysis for the TBS Because of the symmetry, it was only necessary to model
one quarter of the specimen and one half of the specimen
4.1. Scope of Investigation. The continuum damage
thickness, d, as shown in Fig. 6. The boundary conditions,
material behaviour model, proposed by Liu and Murakami,
i.e. ux = 0 on plane A, uy = 0 on plane B and uz = 0 on plane
has been used in the FE investigations, to obtain full creep
C, are also indicated in Fig. 6. The specimen is loaded and
curves and rupture data [10]. A limited study has been also
constrained through two loading pins which are assumed to
carried out, to assess the effects of friction between the
be “rigid” in the FE model. The applied load was calculated
loading pins and the specimen surfaces on the specimen
for the required stress using Equ. (4). The Liu and
failure time. Using Norton's creep law, a series of elastic-
Murakami damage model was used in the FE analyses to
creep analyses have been also performed for a range of n
predict the TBS rupture time, as illustrated in Fig. 7. For the
values to determine the reference stress parameters η and β.
Liu and Murakami damage model, the FE analysis was
The ABAQUS finite element software package [16] was
terminated when the damage parameter, ω , approached
used for the FE analyses.
unity in all of the elements on a path through the cross
4.2. Material Behaviour Models. The Norton material section area of the bar, as illustrated in Fig. 8. The creep
model, i.e. ̇ was used in FE analyses to obtain the time at which ω approaches unity in those elements was
TBS steady state deformation rates, where B and n are considered to be the failure time of the specimen [10]. The
material constants [17]. The Liu and Murakami model, used Liu and Murakami model was also used to assess the effects
to obtain full deformation -time creep curves for the TBS, of the friction between the loading pins and the specimen
consists of a pair of coupled creep/damage equations, i.e. surfaces during loading application. Various coefficients of
friction (μ) were used in the FE analyses, to investigate the
c
d ij
2 effects of the contact properties between the loading pins
2( n 1)
B eq Sij Exp
n1 3 3/2
1
(7) and the specimen surfaces, as described in table (1). It has
dt 2 1 3 / n eq
been found that for a large range of the coefficient of
d M [1 Exp ( q2 )]
friction, the specimen failure times remained practically
( r ) Exp ( q ) (8)
dt q2 2 unchanged indicating that the contact friction does not have
any measurable effect on the creep deformation.
where ω is the damage parameter (0 < ω < 1), where ω = 0
(no damage) and ω = 1 (failure). B and n are the material
4.4. The determination of the reference stress
constants in Norton’s law, M, χ and q2 are also material
parameters for the TBS. FE analysis was used to
constants, which can be obtained by curve fitting to uniaxial
determine the reference stress parameters, i.e. the conversion
factors η and β, for the TBS. Accurate determination of the
reference stress parameters allows the equivalent gauge
length (EGL) and the corresponding uniaxial stress for the
specimen to be obtained. Using a Norton material model,
steady-state creep analyses were performed for the TBS and
the steady-state deformation rates between the two loading
pins were obtained, for a range of n values.
2.5 α=η=0.9866
160MPa 1.5 α=1.2
2 1 α=1.5
170MPa
0.5
1.5 180MPa c
2 ss
Log 0
L B ( α P )n
o
1 0.5 bd -0.5
-1
0.5
-1.5
0 -2
0 200 400 600 800 1000 1200 1400 -2.5
0 1 2 3 4 5 6 7 8
Time (huor) n
Fig. 7. Creep deformation versus time for P91 at 600oC, Fig. 9.
obtained from TBS-FE analyses for different stress levels. Determination of β and η for the TBS.
high ratios. As the ratio increases significantly the
5. Effects of TBS Dimensions on Conversion Factors and
specimen deformation is predominantly uniaxial, where η
Recommended Dimension Ratio Ranges
and β are both close to unity, as illustrated in Fig. 10.
However, as the ratio reduced, the contribution of the
5.1. Effects of Specimen Dimension Ratios on the η and β
Values. The specimen dimensions that can be used are not ̇ to the total specimen deformation rate decreases,
rigidly fixed to allow the specimen to be constructed from resulting in an increase in the β value and a decrease in the η
different shapes and volumes of material samples that may value for small ratios.
be available. The specimen geometry is defined by three
2
dimension ratios: , and (see Fig. 4). Since the η
1.8
conversion factors η and β are geometry dependent, the three β
dimension ratios will have an effect on the conversion 1.6
factors. Consequently, making the most appropriate choice
1.4
η and β
of specimen dimension ratios may affect the conversion
factors, and lead to the most accurate interpretation of the 1.2
bars were kept constant. Rigid pins were used throughout Lo/Di
and the contact conditions, material properties and pin
Fig. 10. Variations of β and η parameters with various
diameter were kept the same for all cases. The total TBS
deformation rate measured at the centres of the loading pins, ratios; the other ratios and for the specimen are 0.62
̇ = ̇ ̇ (10)
where ̇ is the deformation rate of the uniform bars of 5.1.2. The effect of ratio. For this investigation six
length and ̇ is the deformation rate (in the loading
ratios were used, while the other two dimension ratios and
direction) of the supporting material around the loading and
were fixed at values of 2.16 and 0.12, respectively. In
constraining pins. In fact, ̇ is a combination of two
deflections, i.e. ( ̇ and ̇ ( ) ), where ̇ is the creep this case the changes in the total specimen deformation rate,
i.e. Equation (10), will be governed only by the changes in
deformation rate in the contact region and ̇ ( ) is the
the loading pins supporting material deflections. When the
deformation rate as result of the bending in the supporting
supporting material is large enough to prevent significant
material (if k is small enough for the bending to occur, see
bending, the contribution of ̇ to the total specimen
Fig 6).
deformation rate becomes constant; and therefore the
in Fig. 11. close to unity for all ratios used in the study.
2
η 1.6 η
1.7 β β
1.4
1.4
η and β
β and η
1.2
1.1
1
0.8
0.5 0.8
0.5 1 1.5 2 2.5 3 3.5 0.1 0.15 0.2 0.25 0.3 0.35
k/Di b/Di
Fig. 12. Variations of β and η parameters with various
Fig. 11. Variations of β and η parameters with various
ratios; the other ratios and for the specimen are 2.16 ratios; the other ratios and for the specimen are 3.12
and 0.12 respectively. and 1.25 respectively.
5.1.3. The effect of the ratio. Six ratios were used in However, the values of η remain practically constant and
this study, while the other two dimension ratios and close to unity for all ratios used in the study.
Changing the ratio requires modification of the applied 5.2. The effect of the loading pin diameter, , on the
Conversion Factors.
load in order to maintain a constant nominal stress level for
The variations of β and η parameters with have also been
all specimens used in the study. However, increasing the
investigated using five values, (i.e. = 9, 8, 7, 6, and 5
load magnitude, P, with constant contact area between the
mm). The bar width, b, and the supporting material behind
loading pins and the loading pins supporting material, i.e. K
the loading pin change with Di, whereas Lo remains constant
region, (see Fig. 6), results in an increase in the localised
for all cases at 18 mm; the b value was increased with each
deformation rate in the contact region. In this case, ̇ in
reduction in . Similar to the study, the applied load, P,
Equation (10) is proportional to the ratio for a constant
has to be increased as increases in order to maintain same
loading pin diameter. For small ratios, i.e. the
stress levels for all cases. However, increasing the applied
deformation rate in the loading pin supporting material load and reducing the loading pin diameter at the same time
(contact region) is negligible and the contribution of ̇ to result in significant deflections in the supporting material
the total specimen deformation rate in Equation (10) is very behind the loading pin. Therefore, the contribution of the ̇
small compared to ̇ . Therefore, as expected, the in Equ. (10) to the total specimen deformation increases
conversion factors are close to unity. The applied load significantly for small . From Fig. 12 one can deduce that
increases with the ratio , consequently causing large β is inversely proportional to . In contrast, η remains
practically constant and close to unity for this range of . It
deformation in the loading pins supporting material.
was practical to keep the loading pin diameter, Di, constant
Therefore, the contribution of the deformation rate caused
by the loading pin supporting material deflections to the during each dimension ratio study, i.e. , and ,
overall specimen deformation rate increases. Fig. 12 shows because, Di has direct influence on both the supporting
that the β values increase as the ratio of increases.
material behind the loading pin and the ratio, if Lo recommended ranges of specimen dimension ratio are given
in Table (2). The specimens which were experimentally
remained constant during the study.
tested had dimension ratios, see Fig. 5(b), which fall in the
In actual creep testing situations the loading pin diameter,
range of the dimension ratios given in table (2).
Di, should not be very small in relation to other specimen
dimensions in order to (i) avoid high stress concentration
Table 2 The recommended TBS dimension ratio ranges for a
and significant deformation in the pin region, (ii) maintain
constant
sufficient stiffness in order to avoid significant bending in
the loading pins during the test. Dimension ratios Range
2 Lo/ ~ 4.0 - 5.0
η ~ 1.0
k/
1.8 β b/ ~ 0.12 - 0.32
1.6
β and η
Fig. 13. Variations of β and η parameters with , for n
Lo B (12)
specimens with Lo = 18 mm and b = (0.5, 1, 1.5, 2 and
2.5mm). and numerically using FE analyses. The specimen
dimensions, Lo, k, b, d and were (13.0, 6.5, 2.0, 2.0 and
5.3. Recommended Specimen Dimension Ratio Ranges.
5.0 mm), respectively. These specimen dimensions result in
The actual specimen dimensions chosen for a particular
conversion factors β and η values of 1.4557 and 0.9966
application may be dictated by the shape and size of the
respectively. The magnitude of the material constant B in
small sample of the material available. However, the
Norton’s law was 1E-16, and the applied load was
specimen is designed to obtain creep properties from the
corresponded to a constant stress of 50MPa. The load was
overall specimen creep deformation; not from the localised
calculated using Equ. (4). The correlation between the
deformation (which occurs in the contact area between the
steady state displacement rates obtained theoretically and
loading pins and the loading pins supporting material). In
numerically is remarkably good as shown in Fig. 13.
order to make a suitable chose of specimen dimension ratios,
Another FE analysis was carried out using different
with minimum localized deformation in the contact areas,
specimen dimensions and therefore different conversion
the FE analyses results presented in Figs. 10 to 13 can be
factors. The (Lo, k, b, d and ) for the second specimen
used as a guide. It is recommended that specimen dimension
were (20.0, 6.0, 1.0, 1.0 and 6.0 mm), β and η were 1.179
ratios, which minimise the contribution of the deformation
and 0.991 respectively. The material constant B in Norton’s
in the loading pins supporting material, should be used when
law was for the second case was 1E-17 and the applied load
it is possible, as indicated in Figs. 10 to13. The η values, for
corresponded to a stress of 80MPa. The specimen steady
the range of specimen dimension ratios, should be close to
state deformation rates were obtained theoretically using
unity and practically independent of the specimen
Equ. (12) and numerally using FE for a number of n values.
dimensions. However, the β values are always dependent on
Again, very good agreement is found between the two sets
magnitude of to the contribution of the deformation rate in
of results as seen in Fig. 14. The changes in the conversion
the loading pins supporting material to the total specimen
̇ factors may be significant, and are dependent on the
deformation rate, i.e. β is dependent on ̇
. Hence the
specimen dimensions especially for β as can be seen from
Fig. 12 and 13. However, as long as the accurate conversion
factors for the particular specimen dimensions are used in
6. Experimental Creep Testing and Validation for the
the conversion relationship, accurate creep data can be TBS
obtained. From these preliminary validations, it can be
6.1. Specimens and Loading Setup.
observe that despite the significant differences in β values
The TBSs were constructed using P91 steel [22, 23], this
between case one, β =1.4557, and case two, β =1.179, a
material is used extensively in the pipework of the power
remarkable agreement has been achieved between
plants. Due to the small size of the specimen, the electrical
theoretical and numerical results, as shown in Fig. (14) and
discharge method (EDM) was found to be the most suitable
(15). However, any design which involves large deformation
for specimen manufacture. In order to ensure high quality
in the loading pin region should be avoided, because the aim
surface finishing, the specimens were carefully polished to
of the new TBS testing technique is to obtain creep
the final dimensions. The specific specimen dimensions
properties from the overall specimen not from the small
used for Lo, k, b, d and were 13.0, 6.5, 2.0, 2.0 and 4.974
local area of contact between the loading pin and the
mm, respectively. These dimensions were considered
specimen.
because small material samples, with the dimensions of
0.00035
Numerical (FE) approximately (30 × 20 × 3 mm), can be easily removed
0.0003 Theoretical from a component surface using the surface sampling
Minimum displacement rate
0.0020
6.2. Minimum Creep Strain Rates and Creep Rupture
0.0015
Data for the P91 Steel at 600oC. The conversion
0.0010 relationship, i.e. Equ. (13), has been used to convert the TBS
minimum creep deformation rates, measured between the
0.0005
centres of the loading pins, to the equivalent uniaxial
0.0000 minimum creep strain rates, i.e.
6 6.2 6.4 6.6 6.8 7
n c
c ss
= (13)
Fig. 15 The steady state deformation rates obtained from the Lo
TBS theoretically and numerically (FE) using β =1.179
where ̇ is the equivalent uniaxial minimum creep strain
rate, ̇ is the TBS minimum creep deformation rate, is
the distance between the centres of the loading pins and is that the conversion factors remain constant throughout the
the conversion factor, which is dependent on the relative test duration. Hens, the TBS deformation time curves
specimen dimensions. Five experimental TBSs were tested presented in Fig. 16 can be conveniently converted to strain
with loads corresponding to uniaxial stresses of 140, 150, versus time curves using Equ. (6). The converted TBS
160, 170 and 180 MPa, respectively, where Equ. (4) was strain-time curves are compared with the corresponding
used to calculate the applied load for each stress. All tests uniaxial curves in Fig.18. The failure time versus applied
o
were performed at 600 C. The recorded deformation time stress for the TBS and the uniaxial tests (log-log scale) are
curves obtained from the TBS tests are shown in Fig. 16. shown in Fig. 19. Reasonable correlation between the
The equivalent uniaxial minimum creep strain rates together uniaxial and the TBS data is found [10].
with the converted TBS minimum creep strain rates are 40
Uniaxial-Exp
shown in Fig. 17. It can be seen that the correlation between 35 TBS-Exp
the uniaxial and the TBS data is remarkable good [10].
30 170 150
2.5
180MPa 25
Strain %
170MPa 180 160 140
2 20
160MPa
Deformation (mm)
150MPa 15
1.5
140MPa 10
1 5
0
0.5 0 200 400 600 800 1000 1200 1400
Time (hour)
0 Fig. 18. Converted TBS creep strain together with the
0 200 400 600 800 1000 1200
corresponding uniaxial creep strain for P91steel at 600oC,
Time (hour)
stresses in [MPa].
Fig. 16. Deformation versus time curves obtained from the
two bar specimens for a P91 steel at 600oC. 3.25 TBS - EXP
-0.5 uniaxial - EXP
Uniaxial-Exp
TBS-Exp 2.75
-1
Log tf (hour)
2.25
Log (MSR)
-1.5
-2 1.75
-2.5
1.25
2.12 2.15 2.18 2.21 2.24 2.27
-3 Log stress (MPa)
2.12 2.15 2.18 2.21 2.24 2.27
Fig. 19. Creep rupture data obtained from TBSs and uniaxial
Log (stres) specimens for P91steel at 600oC.
Fig. 17. Minimum creep strain rate data for P91steel at The research group have carried out an investigation to
o
600 C. assess the differences between the TBS and the
corresponding uniaxial data for some stress levels. The
Since the changes in the overall specimen dimensions during investigation concluded that the differences are a result of
the test are relatively insignificant, it is reasonable to assume scattering in the material. However, to produce more data
for the TBS validation, new tests using another material at obtaining reliable creep rupture data [5]. It has been shown
different temperature are on-going. This will be published in that the TBS testing technique proposed in this paper is
future papers. capable of obtaining both steady state creep and creep
rupture data.
7. Discussion The limited amount of test data obtained so far, for a
The novel, small TBS is close in its behaviour to that of a P91steel, at 600oC, have shown good agreement between the
miniature uniaxial specimen. The main advantage of the MSRs obtained from the TBS tests and the corresponding
small TBS type, over other miniature creep test specimen conventional creep tests (see Fig. 17), also a good agreement
types, is that a full creep strain versus time curve, till is found between the TBSs failure time and the
rupture, can be obtained by using such a specimen type. corresponding conventional specimens failure time (see Fig.
Another advantage is that the specimen has a simple 19). However, care must be taken when manufacturing and
geometry and can be easily machined and tested. Similar to polishing the specimens to ensure that uniform bar
the small ring creep test specimen [4], the measured thicknesses and depths are achieved. For the purpose of
deformation of the TBS is related to the overall creep further validation, more creep tests using the TBS will be
deformation of the specimen. The conversion factors η and β carried out using different specimen dimensions, materials
depend on specimen dimension ratios and are independent and testing conditions. Statistical analysis on the validation
of the tested material. Loading the specimen through loading results will be carried out in the future in order to
pins makes it possible to achieve accurate alignment and demonstrate repeatability of the TBS testing method. The
hence to avoid possible bending effects as would be failure mechanisms of the TBS test will also be investigated
experienced if a small cylindrical uniaxial specimen was further in future publications.
used, where the loading is applied through “stiff” connection
Acknowledgement
of the specimen ends [10]. Useable specimen dimensions
The authors wish to thank Mr Shane Maskill, Senior
can be readily obtained for a wide range of material sizes.
Engineering Technician at the University of Nottingham, for
Hence, specimens can be easily extracted from small sample
his expert assistant with the experimental work.
of material, for example, the small scoop sample shown in
Fig. 1. The design of the TBS requires a compromise
between minimising the overall specimen dimensions and Notation
producing, as close as possible, a uniaxial state of stress. In χ, M, material constants in the Liu and
the pin connections and bending of the supporting material b, d, , k, two-bar specimen dimensions (Fig. 4)
behind the pins should be avoided. Also, the specimen B, n material constants in Norton’s model
design should be such that the creep deformation of the two and in Liu and Murakami material
take in to consideration metallurgical effects, i.e. the grain EGL equivalent gauge length
size of the tested material should be much smaller than the EDM electrical discharge machining
“bulk” properties are obtained a sufficient number of grains GL, gauge length and diameter of the gauge
should be maintained in the uniform part cross section area length, respectively
of the specimen [26-28]. In order to assess the creep strength HAZ heat-affected zone
of service-aged materials, small specimen creep testing has MSR minimum strain rate
data to help in the process of predicting the remaining life TBS two-bar specimen
time of components [5, 6]. Most miniature creep specimen failure time