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

net/publication/247151624

Analysis and Design of a Small, Two-Bar Creep Test Specimen,

Article  in  Journal of Engineering Materials and Technology · May 2013

DOI: 10.1115/1.4025192

CITATIONS READS

17 1,231

3 authors, including:

Dr Balhassn S.M. Ali

Bursa Orhangazi University
26 PUBLICATIONS   85 CITATIONS   

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.

The user has requested enhancement of the downloaded file.

 Analysis and Design of a Small, Two-
Bar Creep Test Specimen
In this paper, a new small-sized (two-bar) specimen type, which is suitable for use in
obtaining both uniaxial creep strain and creep rupture life data, is described. The
Tom H. Hyde specimen has a simple geometry and can be conveniently machined and loaded
(through pin-connections) for testing. Conversion relationships between the applied
Balhassn S. M. Ali*
load and the corresponding uniaxial stress, and between the measured load-line
eaxba@nottingham.ac.uk
deformations and the corresponding uniaxial minimum creep strain rate, have been
Wei Sun obtained, based on the reference stress method, in conjunction with finite element
analyses. Using finite element analyses the effects of the specimen dimensions on

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.

Keywords: Two-bar specimen; Creep test; Conversion factors; FE analyse; Specimen

Design

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

can be obtained by (=  ), and the corresponding

uniaxial minimum creep strain rate can be obtained using
Equ. (3) if ̇ is known.
20 10 30 10 20

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

are in mm [9]. deformation rates can be obtained, substituting two values of

2. Reference Stress Method and Equivalent Gauge
Length
2.1. Creep Deformation and Reference Stress Method
(RSM). For some components and loading modes, it is
possible to obtain analytical expressions for steady-state
creep deformation rates, ̇ , [e.g. 10- 12]. For a material
( )
n in the expression and equating the two resulting
expressions allow the value of  to be determined. Hence,
ref (=  ), and D can be obtained. This approach was
proposed by MacKenzie [13-15]. However, analytical
solutions for the component steady-state creep deformation
rates only exist for a small number of relatively simple
components and loadings [14]. If computed (e.g. finite
element) solutions to a creep problem are obtained using
several stress index n values, but keeping all other material
properties, loading and component dimensions the same,
then can be obtained. This is done by taking several
values of , normalising the steady-state value of
deformation rate, ̇ , with respect to (
in which D (=  d) is the reference multiplier, which is, in
fact, the EGL for the test. In some cases, the geometric
changes, due to the specimen creep deformation with time,
are small (e.g. for impression creep tests), and in such cases,
) and hence the effects of geometric changes on D (EGL) can be

finding the value of  which renders ̇ neglected [10].

( ( ) independent of n. This process is most easily

̇ 3. The “Two-Bar” Specimen (TBS)
visualised by plotting log ( ( ) for various
3.1. Specimen Geometry and Dimensions. The pin-loaded,
values of  against n, as illustrated in Fig. 3. It can be seen
Two Bar Specimen (TBS) type, shown in Fig. 4, has a
that the straight lines are produced, using all of the  values,
simple geometry; specimen dimensions are defined by , b,
have the same intercept on the log ̇ ( ( ) axis.
d, , and k; where is the uniform “bar” length, i.e. the
This intercept is equal to the logarithm of the reference
distance between the centres of the loading pins, b is the bar
multiplier, D.
width, d is the specimen thickness, is the diameter of the
1 loading pins and k is the length of the loading pin supporting
2 end.

3=

4

5

n
n

Fig. 3. Variation of log  c

Δ ss ( n) 
B  nom 
n
 with n
[10].

2.3. Equivalent Gauge Length (EGL). For a conventional

uniaxial creep test, the creep strain at a given time is usually
determined from the deformation of the gauge length (GL). Fig.4. Two bar specimen geometry and dimensions.
If the gauge length elongation is  and the elastic portion is
neglected, 3.2 Test Procedure. The TBS testing technique is based

(5a) on the principle of converting the specimen load-line

For non-conventional small specimen creep tests, an
equivalent gauge length (EGL) [5] can be defined, if the
measured creep deformation can be related to an equivalent
uniaxial creep strain, in the same form as that of Equ. (5a),
i.e.
The EGL is related to the dimensions of the specimen and in
some cases may be related to the time-dependent
deformation of the test specimen. The creep strain and creep
deformation given in Equ. (5b) may be presented in a form
related to the reference stress, ref, i.e.
( )
deformation -time curves, to the equivalent uniaxial strain-
time curves, using conversion relationships defined by Equs.
(4) and (6). The conversion relationships, which are used to
convert the specimen loading to the equivalent uniaxial
stress and the specimen (pin) displacements to equivalent
uniaxial strains, are functions of specimen dimensions and
(5b)
deformations as described in section 2. The TBS, Figs. 5 (a)
and 5 (b), is loaded by two loading pins and the
displacement between the centres of the two loading pins is
measured during the test. The loading fixtures generally
have a much higher stiffness, compared to the specimen, and
are generally manufactured from a material which has a
(6) much higher creep resistance than the tested material. The
conversion factors η and β are obtained from finite element creep curves [18-20]. The rupture stress, , in Equ. (8) is
(FE) analyses. given by:

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

Constraining different stresses [20]. The material constants for P91 at

pin
Constrained 600oC, which are used in the FE analyses for the Liu and
part
Murakami model, were reported in [21].
(a) (b) (c)

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 
n1 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.

Table 1 Illustrate the effect friction coefficient factor on the

specimen failure time for (P91 steel, T= 600oC, σ =
170MPa)
Friction coefficient factor (μ) tf ( hour) Fig. 8. Contour plot of damage parameter, 𝜔, in the TBS for

169.025 P91 at 600oC, (σ = 180 MPa, tf = 90.81 hour).

0.001
0.01 169.011
0.1 169.067
0.15 169.103 Similar to Fig. 3, the steady state deformation rates,  css are
0.2 169.099
normalised, by L B( α  P )n , where P is the applied load
o
2bd
ux = 0 (Plane A)
[3]. However, because only quarter of the specimen and half
Lo/2 uz = 0 (Plane C)
of the thickness is use in the FE analyses, (see Fig. 8), the
obtained minimum deformation rates have to be doubled and
b
k P
the nominal stress will be . Several α values were
0.5 bd
Y
X considered for all of the deformation rate values, with
different n-values. The value of α which makes
 
Z  c  practically independent of n is the
2  ss
log  
 L B ( α  P )n 
 o 
 0.5 bd 
Rigid pin
uy = 0 (Plane B) required α value. This value (corresponding to the solid line
d/2
in Fig. 9), is the reference stress parameter, η, for the
Fig. 6. Finite element mesh and the boundary conditions. particular TBS geometry and dimensions. The value of β can
then be obtained from the intercept of the solid line in Fig. 9.
3.5 The procedure is described in more detail in reference [3].
3 140MPa
2.5
α=0.6
150MPa 2 α=0.75
Deformation (mm)

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

MSR and rupture data. In this paper, a series of 3D steady 1

state creep analyses using Norton’s law have been carried
0.8
out to assess the effects of the dimension ratios on the η and
β values using meshes which consist of 20-noded brick 0.6
elements as in Fig. 6. The nominal stress in the specimen 0 2 4 6 8 10 12 14

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

̇ , can be approximated by:- and 0.12 respectively.

̇ = ̇ ̇ (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

conversion factors become constant for any ratios above

5.1.1. The effect of ratio. Several ratios were
about 1.25, as illustrated in Fig. 11. However, for very small
considered for this study. During the study the other two
ratios, where significant bending in the k region may
dimension ratios and were fixed at values of 0.62 and
occur, the contribution of ̇ to the total specimen
0.12, respectively. As the ratio increases the contribution
deformation rate will be significant. Therefore, the
of the ̇ to the total specimen deformation rate increases, conversion factors, in particular the β value, change
therefore the conversion factors approach the unity for very
significantly with when k is relatively small, as illustrated However, the values of η remain practically constant and

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.

were fixed at values of 2.16 and 3.12, respectively.

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

1.4 5.4. Preliminary FE validation

Preliminary validation of the testing technique has been
1.2
carried out using 3D- FE analysis and Norton’s law. This
1 was also used to assess the accuracy of the conversion
relationship and conversion factors. The specimen steady
0.8
state deformation rates were obtained for several n values,
4.5 5.5 6.5 7.5 8.5 9.5
Di theoretically, using Equ. (12),

 
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

0.00025 system (SSam), see for example Fig. 1 [25]. Furthermore,

because the specimen dimensions have been given as ratios,
0.0002
i.e. variable rather than fixed values, the specimen can be
0.00015 made with even smaller dimensions. The loading pins
diameter is 4.974 mm, which makes an allowance between
0.0001
the loading pin and the specimen surface of about 0.026
0.00005 mm. The loading pins and the loading pin holders (see Fig.

0 5) were manufactured from a Nickel-base Superalloy

5.5 5.7 5.9 6.1 6.3 6.5 6.7 (Nimonic 80A), which has much higher creep resistance
n
than the tested material; this ensures that the deformation of
Fig. 14 The steady state deformation rates obtained from the the loading fixture is negligible [10]. Using the procedure
TBS theoretically and numerically (FE) using β =1.4557 described in section 4 the η and β values for the tested
specimen were found to be 0.9966 and 1.4557, respectively,
0.0035
for this specific set of dimension ratios, i.e. = 2.6, =
Theoretical
0.0030
Numerical (FE) 0.4 and = 1.3, for a constant loading pin diameter ( =
Minimum displacement rate

0.0025 4.974 mm).

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
different temperature are on-going. This will be published in
future papers.
7. Discussion
The novel, small TBS is close in its behaviour to that of a
miniature uniaxial specimen. The main advantage of the
small TBS type, over other miniature creep test specimen
types, is that a full creep strain versus time curve, till
rupture, can be obtained by using such a specimen type.
Another advantage is that the specimen has a simple
geometry and can be easily machined and tested. Similar to
the small ring creep test specimen [4], the measured
deformation of the TBS is related to the overall creep
deformation of the specimen. The conversion factors η and β
depend on specimen dimension ratios and are independent
of the tested material. Loading the specimen through loading
pins makes it possible to achieve accurate alignment and
hence to avoid possible bending effects as would be
experienced if a small cylindrical uniaxial specimen was
used, where the loading is applied through “stiff” connection
of the specimen ends [10]. Useable specimen dimensions
can be readily obtained for a wide range of material sizes.
Hence, specimens can be easily extracted from small sample
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
producing, as close as possible, a uniaxial state of stress. In
addition, excessive deformation / stress concentration near
the pin connections and bending of the supporting material
behind the pins should be avoided. Also, the specimen
design should be such that the creep deformation of the two
uniform “bars” will dominate the overall specimen
deformation. The minimum dimensions of the TBS should
take in to consideration metallurgical effects, i.e. the grain
size of the tested material should be much smaller than the
smallest dimension of the specimen. In order to ensure that
“bulk” properties are obtained a sufficient number of grains
should be maintained in the uniform part cross section area
of the specimen [26-28]. In order to assess the creep strength
of service-aged materials, small specimen creep testing has
been found to be an extremely useful method for producing
data to help in the process of predicting the remaining life
time of components [5, 6]. Most miniature creep specimen
types available to date do not have the capability of
obtaining reliable creep rupture data [5]. It has been shown
that the TBS testing technique proposed in this paper is
capable of obtaining both steady state creep and creep
rupture data.
The limited amount of test data obtained so far, for a
P91steel, at 600oC, have shown good agreement between the
MSRs obtained from the TBS tests and the corresponding
conventional creep tests (see Fig. 17), also a good agreement
is found between the TBSs failure time and the
corresponding conventional specimens failure time (see Fig.
19). However, care must be taken when manufacturing and
polishing the specimens to ensure that uniform bar
thicknesses and depths are achieved. For the purpose of
further validation, more creep tests using the TBS will be
carried out using different specimen dimensions, materials
and testing conditions. Statistical analysis on the validation
results will be carried out in the future in order to
demonstrate repeatability of the TBS testing method. The
failure mechanisms of the TBS test will also be investigated
further in future publications.
Acknowledgement
The authors wish to thank Mr Shane Maskill, Senior
Engineering Technician at the University of Nottingham, for
his expert assistant with the experimental work.
Notation
χ, M, material constants in the Liu and
Murakami material model
b, d, , k, two-bar specimen dimensions (Fig. 4)
B, n material constants in Norton’s model
and in Liu and Murakami material
model
D reference multiplier
EGL equivalent gauge length
EDM electrical discharge machining
FE finite element
GL, gauge length and diameter of the gauge
length, respectively
HAZ heat-affected zone
MSR minimum strain rate
P applied load
TBS two-bar specimen
failure time
 ̇ ̇ ( ) creep strain, minimum creep strain rate
and minimum creep strain rate at
reference stress, respectively
̇ creep deformation and steady state
deformation rate
̇ total deformation rate in the loading pin
supporting material
̇ strain rate data. 2nd Int. Conf. on Determination of
creep deformation rate in the loading pin
supporting material
̇ ( ) deformation rate in the loading pin
supporting material as result of the
bending
̇ deformation rate which occurs in the
parallel regions of the bars which are of
length
̇ 10. Hyde T. H., Sun W., and Ali B. S. M. Small creep test
total deformation rate measured at the
loading pin
β, η reference parameters (conversion
factors)
stress, nominal stress and reference
stress, respectively
reference stress scaling factor and
material constant in damage model
ω damage parameter
μ coefficient of friction
3D three dimensional
References
1. Maharaj C., Dear J. P. and Morris A. Review of
methods to estimate creep damage in low-alloy steel
power station steam pipes, strain, 45 (4): 316-331,
2009.
2. Parker J. D. and James J. D. Creep behaviour of
miniature disc specimens of low alloy steel, ASME,
PVP 279, Developments in a Progressing Technology
167-172, 1994.
3. Hyde T. H. and Sun W. Evaluation of the conversion
relationship for impression creep testing. Int. J. of
Pres. Ves. & Piping 86 (11): 757-763, 2009.
4. Sun W. and Hyde T. H. Determination of secondary
creep properties using a small ring creep test
technique. Metallurgical Journal, LXIII 185-193,
2010.
5. Li Y. Z. and Šturm R. Determination of creep
properties from small punch test. Proc. ASME
Pressure Vessels and Piping Division Conference.
Chicago, Illinois, USA, 27-31, 2008.
6. Hyde T. H. and Sun W. Some considerations on
specimen types for small sample creep tests.
Materials at High Temperatures 27 (3): 157-165,
2010.
7. Askins M. C. and Marchant K. D. Estimating the
remaining life of boiler pressure parts. EPRI Contract
RP2253-1, Part 2, Miniature specimen creep testing in
tension, CEGB Report, TPRD/3099/R86, CEGB, UK,
1987.
8. Sun W., Hyde T. H. and Brett S. J. Use of impression
creep test method for determining minimum creep
Mechanical Properties of Materials by Small Punch
and Other Miniature Testing Techniques, Ostrava,
Czech Rep. 297-304, 2012.
9. Takamoto I., Yuichi I., Masao S., Yusuke K.,
Takafumi T., Masaharu, F. Development Of Miniature
Creep Testing For High Temperature Materials
Verification Testing. In Determination of mechanical
properties of materials by small punch and other
miniature testing techniques. Ostrava, Czech
Republic. 283-290, 2012.
specimen for use in determining uniaxial creep rupture
data. 2nd Int. Conf. On Determination of Mechanical
Properties of Materials by Small Punch and Other
Miniature Testing Techniques, Ostrava, Czech
Republic, 261-270, 2012.
11. Anderson R. G., Gardener L. R. T. and Hodgkins W.
R. Deformation of uniformly loaded beams obeying
complex creep laws, J. Mech. Engng. Sci. 5: 238-244,
1963.
12. Johnsson A. An alternative definition of reference
stress for creep, Int. J. Mech. Sci. 16 (5): 298-305,
1973.
13. Hyde, T.H., K.A. Yehia, and A.A. Becker,
Interpretation of impression creep data using a
reference stress approach. International Journal of
Mechanical Sciences, 35 (6): 451-462, 1993.
14. Hyde T. H., Yehia K. and Sun W. Observation on the
creep of two-material structures. J. Strain Analysis 31
(6): 441-461, 1996.
15. MacKenzie A. C. On the use of a single uniaxial test
to estimate deformation rates in some structures
undergoing creep. Int. J. Mech. Sci. 10: 441-453,
1968.
16. ABAQUS 6.11.3 manual, Dassault Systèmes, Simulia
Corp., Providence, RI, USA, 2010.
17. Penny R. K. and Marriott D. L. Design for creep, 2nd
edition, (Chapman & Hall), 1995.
18. Hyde T. H., Becker A. A., Sun W. and Williams J. A.
Finite element creep damage analyses of P91 pipes.
Int. J. of Pres. Ves. & Piping 83: 11-12, 2006.
19. Liu Y. and Murakami S. Damage localization of
conventional creep damage models and proposition of a
new model for creep damage analysis. JSME
International Journal 41: 57-65, 1998.
 20. Saber M., Tanner D. W. J., Sun W. and Hyde T. H.
Determination of creep and damage properties for P92
at 675°C. J. Strain Anal. Eng. Des. 46: 842-851, 2011.
21. Li R., Hyde T. H., Sun W. and Dogan B. Modelling
and data interpretation of small punch creep testing.
ASME, Pressure Vessels and Piping Conference,
Maryland. USA, 2011.
22. Hyde, T., Sun, W. Becker A. and Williams J. Creep
properties and failure assessment of new and fully
repaired P91 pipe welds at 923 K. Proceedings of the
Institution of Mechanical Engineers, Part L: Journal
of Materials: Design and Applications,. 218(3): 211-
222), 2004.
23. Shibli I. A. and Hamata N. L. M. Creep crack growth
in P22 and P91 welds — overview from SOTA and
HIDA projects. International Journal of Pressure
Vessels and Piping, 78(11–12): 785-793, 2001.
24. Hyde T.H., M. Saber and W. Sun Testing and
modelling of creep crack growth in compact tension
specimens from a P91 weld at 650°C. Engineering
Fracture Mechanics, 77(15): 2946-2957, 2010.
25. Hyde C. J., Hyde T. H. and Sun W. Small ring
specimen creep testing of a Nickel-based superalloy.
2nd Int. Conf. On Determination of Mechanical
Properties of Materials by Small Punch and other
miniature testing techniques. Ostrava, Czech
Republic. 276-282, 2012.
26. Evans M. and Wang D. Optimizing the sensitivity of
the small-disc creep test to damage and test
conditions. J. Strain Anal. 42(389-409), 2007
27. Dipersio F. and Klapyt J. miniaturised creep testing
using the small punch test technique. Third
International Charles Parsons Turbine Conference,
419-428, 1995.
28. Garzillo A., Guardamagna C., Moscotti L. and
Ranzani L. A technique for the residual life
assessment of high temperature components based on
creep-rupture testing on welded miniature specimens.
International Journal of Pressure Vessels and Piping,
66(1-3): 223-232, 1996.

View publication stats