You are on page 1of 13

Accepted Manuscript

On the determination of the ductile to brittle transition temperature from small punch
tests on Grade 91 ferritic-martensitic steel

M. Bruchhausen, S. Holmström, J.-M. Lapetite, S. Ripplinger

PII: S0308-0161(15)30116-2
DOI: 10.1016/j.ijpvp.2017.06.008
Reference: IPVP 3632

To appear in: International Journal of Pressure Vessels and Piping

Received Date: 9 November 2015


Revised Date: 17 January 2017
Accepted Date: 19 June 2017

Please cite this article as: Bruchhausen M, Holmström S, Lapetite J-M, Ripplinger S, On the
determination of the ductile to brittle transition temperature from small punch tests on Grade 91
ferritic-martensitic steel, International Journal of Pressure Vessels and Piping (2017), doi: 10.1016/
j.ijpvp.2017.06.008.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT

On the determination of the Ductile to Brittle Transition Temperature from Small


Punch tests on Grade 91 ferritic-martensitic steel

M. Bruchhausena,∗, S. Holmströma , J.-M. Lapetitea , S. Ripplingera


a European Commission, Joint Research Centre (JRC), Institute for Energy and Transport (IET), Westerduinweg 3, 1755 LE Petten, The

PT
Netherlands

RI
Abstract
The exposure of structural materials to the environment in nuclear power plants leads to aging through thermal and/or

SC
irradiation effects. One of the consequences of this aging is the shift of the ductile to brittle transition temperature
(DBT T ) towards higher temperatures. Characterising the aging of structural materials is therefore a safety relevant
topic especially in the context of long term operation of nuclear plants.
For testing in service materials and to reduce the health risks and costs associated with handling of irradiated

U
materials miniature testing techniques are being developed. The small punch (SP) technique is one of these methods.
In a tensile/fracture SP test, a punch with a hemispherical tip is pushed with a constant displacement rate through the
AN
center of a small cylindrical disk specimen.
A series of tensile SP tests on the ferritic/martensitic steel Gr. 91 at two different displacement rates show no impact
of the displacement rate on the DBT T . A new method using fracture energies normalised to the maximum force for
deriving the DBT T from force-deflection data is presented.
M

Post-test analysis by means of X-ray computed tomography is used for deriving the DBT T from fracture strains.
Both methods give consistent values for the transition temperature. However, the DBT T determined in this study is
higher than what can be found in the literature. Supplementary tests indicate that the punch diameter has an effect on
the SP defined DBT T .
D

Keywords: Small Punch Testing, Ductile to Brittle Transition Temperature (DBT T ), fracture strain, Grade 91,
Gr. 91, P 91, T 91, miniature testing techniques
TE

1. Introduction One of the possible areas of application of SP testing


in the energy sector is the characterization of structural
EP

Recently the small punch (SP) test technique [1] materials with regard to aging phenomena such as tem-
has received much attention for characterising materi- per embrittlement. In nuclear power plants the expo-
als with regard to their mechanical properties. sure of structural components to neutron fluxes during
The multi-axial, time dependent stress state makes it their lifetime also leads to irradiation induced embrit-
C

difficult to relate the mechanical properties determined tlement. Both embrittlement processes lead to the shift
by SP testing to the results from standard sized spec- of the ductile to brittle transition temperature (DBT T )
AC

imens [1–7]. The small specimen size nevertheless towards higher temperatures. The DBT T is usually de-
makes SP testing a valuable tool for materials character- termined by means of Charpy tests which require quite
ization, especially when only small amounts of material large standard specimens (10×10×55 mm3 ) [10]. Irradi-
are available (as for aged materials or experimental ma- ation embrittlement of the structural components of nu-
terials that are still being developed) or because safety clear power plants is monitored using dedicated Charpy
considerations require the use of small specimens (as specimens that were deposited in the reactor pressure
for irradiated materials) [2, 5, 8, 9]. vessel and have been exposed to the same irradiation
history as the actual reactor components. Especially in
the context of lifetime extension projects for many nu-
∗ Correspondingauthor clear power plants there is a strong interest in develop-
Email address: matthias.bruchhausen@ec.europa.eu ( M.
Bruchhausen)
ing techniques for determining the DBT T with smaller

Preprint submitted to International Journal of Pressure Vessels and Piping June 20, 2017
ACCEPTED MANUSCRIPT

specimens. Consequently, the SP test is also used in that


context [5, 6, 11].
We present SP results regarding the impact of the
punch displacement rate on the measured DBT T . Dif-
ferent approaches for deriving the DBT T from fracture
energies are discussed and compared to the DBT T esti-
mated from fracture strains.

PT
The material selected for the present work was the
ferritic/martensitic steel Grade 91 (Gr. 91). The me-
chanical properties of the alloy such as its tensile and
creep characteristics and its swelling resistance make it

RI
an interesting candidate material especially for nuclear
applications. Consequently, is has been in the focus of
European projects like MATTER [12] and MATISSE.

SC
2. Experimental Details

U
2.1. Material and specimens Figure 1: Sketch of the SP setup. The main dimensions are listed in
Tab. 3
AN
The alloy used in the present study is the fer-
ritic/martensitic steel Gr. 91 (X10CrMoVNb9-1). It has
the same chemical composition as P91/T91 but is sup-
2.2. Test rig
plied in form of a plate rather than as pipe (P91) or tube
M

(T91). The particular batch of Gr. 91 used in the present In a tensile/fracture SP test, a hemispherical punch
work (S50460) as part of the FP7 project MATTER is pushed with a constant displacement rate through the
[12] was originally produced for the FP6 project EURO- disc shaped specimen along its axis of symmetry. Dur-
TRANS (project domain 4: DEMETRA). Its chemical
D

ing the test, the applied force is recorded as a function


composition is listed in Tab. 1. The material was deliv- of punch displacement.
ered as hot rolled and heat treated plate with a thickness The setup used in the current work is shown in
TE

of 15 mm by Industeel, Arcelor group. The steps of the Fig. 1. The specimen holder consists of two dies be-
heat treatment are detailed in Tab. 2. tween which the specimen is clamped. The punch
Disc shaped SP specimens as specified in the Euro- (2 mm diameter) is pushed from the top through the
pean Code of Practice [1] were used. Their diameter specimen. The receiving hole in the lower die has a
EP

was 8 mm and their thickness 0.5 mm. The surface diameter of 4 mm with a 45◦ chamfer to avoid shearing-
roughness was less than Ra = 0.15 µm on both sides as off of the specimen. The movement of the punch is
specified by the workshop. One surface of every speci- controlled by means of an Instron type 5586 universal
men was checked using a surface roughness tester (Mi- testing machine.
C

tutoyo SJ-201) and no surfaces out of specification were Below the specimen a ceramic rod is mounted. This
found. ceramic rod contains a thermocouple in direct contact
AC

All specimens were cut from the same block of mate- with the lower specimen surface for controlling the test
rial but with two different orientations: TS and TL. The temperature. The ceramic rod also transmits the defor-
letters L, T and S respectively refer to the longitudinal mation of the specimen to a linear variable displacement
(rolling), transverse and short transverse directions. The transducer (LVDT). This configuration avoids measure-
designation of the groups follows the principle that the ment uncertainties from the test rig compliance. How-
letter missing in the designations indicates the axis per- ever, due to thinning of the specimen during the test, the
pendicular to the specimen plane. measured deflection is slightly smaller than the actual
Investigations within the MATTER project did not punch tip displacement.
show any anisotropy effect in SP tensile tests, so the dif- The specimen holder is placed in a thermal chamber
ferent orientations in the two test groups are considered (Instron 3119-407 series) that can be cooled down to
having no impact on the test results [13]. −196 ◦ C by means of liquid nitrogen or heated up to
2
ACCEPTED MANUSCRIPT

Fe C Mn P S Si Cu Ni Cr Mo Al Nb V Ti N
bal. 0.097 0.386 0.020 0.0005 0.218 0.080 0.115 8.873 0.871 0.009 0.077 0.195 0.003 0.0440

Table 1: Chemical composition of the DEMETRA Gr. 91 heat [wt.%] according to supplier

step temperature duration cooling

PT

normalization 1050 C 15 minutes water cooling to room temperature
tempering 770 ◦ C 45 minutes air cooling to room temperature

RI
Table 2: Heat treatment of the DEMETRA Gr. 91

300 ◦ C.

SC
2000

1800
For the present work tests were carried out at punch
1600
displacement rates of 0.5 mm/s and 0.005 mm/s as mea-
1400
sured from the cross-head displacement.
1200

U
F [N]
1000

800
2.3. Post-test analysis
AN
600
TL17
400
Post-test determination of the fracture strains was car- du/dt = 0.005 mm/s
200 T = −100 °C
ried out by means of X-ray computed tomography (CT)
0
using a commercial Phoenix Nanotom S research edi- 0 0.5 1 1.5 2
M

u [mm]
tion tomograph. In a CT analysis, the specimen is ro-
tated stepwise and X-ray images are taken under differ- Figure 2: Force-deflection curve for ductile failure.
ent angles. In a post processing step, all the scans taken
D

at the different angles are used to reconstruct a full 3D


model of the specimen. The scans in the present study
3. Experimental results
were carried out with a voltage of 130 kV and a cur-
TE

rent of 120 µA. A 0.5 mm Sn filter was used to cut-off 3.1. Fracture energy
the low-energetic radiation. The configuration lead to a One possibility for determining the DBT T is based
spatial resolution of 4.24 µm. on the temperature dependence of the fracture energy
A commercial software package (VGStudio Max 2.2 Efrac . In the transition region, a small reduction of tem-
EP

from Volume Graphics) was used for visualizing and perature leads to a strong drop of Efrac .
analyzing the data. Specimen thicknesses were deter- Fig. 2 shows the force-deflection curve of a typical
mined by means of the metrology module ”Wall Thick- ductile fracture. In order to determine Efrac , the force F
ness”. For this purpose, first the specimen surface was has to be integrated over the deflection u until specimen
C

determined from the 3D data. During the actual wall rupture occurs. The current Code of Practice suggests
thickness determination, a simulated beam is sent from using the force drop to 80% of its maximum [1] as rup-
AC

each point of the surface in the direction normal to the ture criterion. In the present work, however, the integra-
surface through the specimen and bounced back from tion is carried out up the maximum force as seems to be
the opposite surface. If the reflected beam hits the spot done by most authors [5, 15–17].
from which the original beam was emitted, the distance Fig. 3 shows the force-deflection curve for a typical
between the emission point and the point where the brittle case where failure does not occur in a single event
beam was reflected is considered as the wall thickness but progressively in several steps. In the example three
in that point of the specimen. For the surface normals occasions (as indicated by the black arrows) can be seen
at the points where the beam is emitted and where it is where the force drops before reaching its maximum.
reflected tolerance angles of 20◦ were used. This case of brittle failure is not covered by the cur-
More detailed information on the experimental pro- rent Code of Practice. Three possibilities for defining
cedures can be found in [14]. the fracture energy Efrac in the case of brittle failure are:
3
ACCEPTED MANUSCRIPT

feature symbol size


punch diameter d 2 mm
diameter of receiving hole (lower die) d2 4 mm
chamfer length (lower die) l 0.2 mm
chamfer angle (lower die) α 45◦
specimen diameter d1 8 mm
specimen thickness h 0.5 mm

PT
Table 3: Principle dimensions of the SP specimen and setup

RI
1500
these cases the consecutive force drops added up to a
cumulative drop of 20% before the maximum force was

SC
reached. After reaching its maximum the value the force
1000 drops very quickly so that in the present case maximum
and the 20% cumulative drop criterion lead to virtually
F [N]

identical results. Therefore only the maximum and first


crack criteria are considered further.

U
500
TL14
The fracture energies evaluated according to both def-
initions are plotted in Fig. 4 for the two cross-head dis-
AN
du/dt = 0.500 mm/s
T = −196 °C
placement rates. The data upon which these evaluations
0
0 0.5 1 1.5 2 are based can be found in Tab. 4 in the appendix. The
u [mm]
data sets themselves are identified by digital object iden-
tifiers (DOIs) [20]. They can be accessed through the
M

Figure 3: Force-deflection curve for brittle failure in several steps


European Commission’s online data base MatDB via
the ODIN portal [21].
In both graphs the fracture energy Efrac starts at low
1. Maximum Force
D

temperatures at a low level, rises quickly in a transition


In this case, Efrac is calculated exactly as in the region and then drops again slowly at higher tempera-
ductile case by integrating the force up to its max- tures. There is no discernible effect of the variation in
TE

imum. This procedure has been used by several displacement rate. The scatter in the lower shelf i.e. at
authors on different types of steels [5, 16, 17]. temperatures below the transition temperature for brittle
2. 20% cumulative drop fracture is much higher than for ductile failure for both
This is an extension of the 20% force drop criterion methods of evaluating the fracture energy Efrac . This re-
EP

suggested in the current Code of Practice: consec- flects the strong dependency especially of brittle crack
utive force drops are added until a total cummu- initiation on irregularities in the microstructure or mi-
lative force drop of 20% has been reached. This croscopic scratches even on polished surfaces.
method has been applied on ferritic oxide disper-
C

sion strengthened steel MA956 [18].


3.2. DBT T from fracture energy
3. First Crack
AC

Applying this method considers the first onset of For determining the DBT T the fracture energies are
cracking indicated by the first measurable drop of often considered separately in the upper and the lower
the force as upper limit for the integration. The shelf regions [5, 16]. We use the same approach as in [5]
specimen has not yet failed at this point and is able where Efrac is considered separately in the lower and up-
to support higher stresses as demonstrated by the per shelves and described as a function of temperature
continued increase of the force after initial crack- T in the form:
ing (Fig. 3). This method has been used e.g. for
ELS/US (T ) = ALS/US + BLS/US exp CLS/US T
 
the characterization of the ferritic-pearlitic steel (1)
AE460 [19]. where the fitting parameters ALS/US , BLS/US and CLS/US
In the present study many cases of multi-step fracture are determined independently by least squares fits in the
in the brittle regime were found, however in none of lower (index LS) and upper (index US) shelves where
4
ACCEPTED MANUSCRIPT

the lower shelf includes the transition region. In this ap-


proach two distinctive energies can be defined: Emin , the
fracture energy at the zero-point of the absolute temper-
ature (i.e. T = −273.15 ◦ C), is determined by extrapo-
lating the fitted function for the lower shelf. Emax is the
fracture energy at the intersection point between the fits
for the lower and upper shelves. The transition tempera-

PT
ture DBT T 2fits is then defined as the temperature where:
Emin + Emax
(a) 2.5 Efrac (DBT T 2fits ) = (2)
0.5 mm/s 2

RI
0.005 mm/s
where the index ”2 fits” refers to the 2 functions fitting
2 procedure.
The curves in Fig. 4 show the fitted curves for both

SC
1.5 displacement rates and evaluation methods. The differ-
E frac [J]

ent symbols for the data points identify the points hav-
ing been assigned to the lower and upper shelf regions
1 respectively. In the case of the lower displacement rate
the data point at −130 ◦ C (specimen TS09) was consid-

U
Method: maximum

0.5
E min = 0.7 J E min = 0.1 J ered being at the edge of the upper shelf and has there-
Emax = 1.7 J Emax = 1.6 J
fore been taken into account for fitting both functions
AN
DBTT2 fits = -120.9 °C DBTT2 fits = -171.7 °C
ELS (T ) and EUS (T ).
0 The calculated transition temperatures DBT T 2fits (as
-200 -150 -100 -50 0 50
T [°C] listed in Fig. 4) lie relatively close together with the ex-
ception of the case where the data for a cross-head dis-
M

(b) 3
0.5 mm/s placement of 0.005 mm/s was evaluated with the maxi-
0.005 mm/s
2.5
mum method (red, dashed line in Fig. 4 (a)). A problem
with this two fits method is that the intersection point
D

between the two separately fitted functions for the up-


2
per and lower shelves needs to be determined accurately
E frac [J]

since that defines Emax in eq. 2. In cases where there is


TE

1.5
no data point at the low edge of the upper shelf, this in-
tersection point needs to be determined by extrapolation
1 which can lead to relatively large uncertainties.
Method: first crack
E min = 0.4 J E min = 0.5 J An alternative approach is determining DBT T by fit-
EP

0.5 Emax = 1.7 J Emax = 1.7 J ting a tanh function as the one used in varying formula-
DBTT2 fits = -124.0 °C DBTT2 fits = -122.9 °C
tions for evaluating Charpy tests [11, 22–24]:
0
-200 -150 -100 -50 0 50 EUS + ELS EUS − ELS  T − DBT T 
Efrac = + tanh (3)
C

T [°C]
2 2 2∆T
Figure 4: Fracture energy Efrac as function of temperature evaluated where T is the temperature, EUS and ELS are the upper
AC

using the maximum (a) and the first crack (b) method. and lower shelf energies, DBT T is the transition tem-
perature and ∆T is a fitting parameter characterizing the
slope in the transition region.
However, when looking at the data points in the upper
shelf it is clear that Efrac drops with rising temperature.
This trend has already been observed by several authors
for a variety of materials [17, 25–28]; although some
data indicate this behaviour might be material depen-
dent and not be a universal trend [11, 17]. In the case
of many materials, where the fracture energy Efrac in the
upper shelf drops with temperature the function in eq. 3
5
ACCEPTED MANUSCRIPT

×10 -3 In all cases, the normalization of Efrac with Fmax leads


(a) 1
0.5 mm/s to constant values in the upper shelf and the data points
0.005 mm/s
can be well described by a function as in eq. 3 where
0.8 the energies are replaced by normalized energies. For
the data in the present study, the transition temperatures
E frac / F max [m]

0.6
DBT T n determined by this procedure are 5−10 ◦ C lower
than DBT T 2fits determined above. However, there is one

PT
outlier, namely the DBT T n determined for a displace-
Method: maximum
0.4
DBTT n = -125 °C DBTT n = -135 °C
ment rate of 0.5 mm/s evaluated using the first crack
E/F LS = 0.470 mm E/F LS = 0.369 mm
method. The significantly lower DBT T results from the
data point at −196 ◦ C which is from a specimen that

RI
E/F US = 0.837 mm E/F US = 0.867 mm
0.2
∆T n = 2.72 °C ∆T n = 21.28 °C started cracking at a very low force level Fig. 5 (b)).
Using the tanh function (eq. 3) for calculating the
0 transition temperature from the fracture energies instead
-200 -150 -100 -50 0 50

SC
T [°C]
of the method based on two independent fits (eq. 1) has
×10 -3 the advantages that less fitting parameters are required,
(b) 1
0.5 mm/s
it potentially avoids extrapolation and does not require
0.005 mm/s a manual (and to some extend subjective) attribution of

U
0.8 the data points to the lower or upper shelf.
Regarding the calculation of Efrac by either the max-
imum or the first crack method and respectively com-
AN
E frac / F max [m]

0.6
paring the data points plotted in Fig. 4 (a) and (b) and in
Fig. 5 (a) and (b), it is obvious that there is less scatter in
0.4
Method: first crack the plots (a) in both cases i.e. when Efrac was calculated
DBTT n = -149 °C DBTT n = -128 °C
using the maximum of the force also for brittle failure.
M

E/F LS = 0.053 mm E/F LS = 0.361 mm


Hence, based on the data from the present study, the
E/F US = 0.867 mm E/F US = 0.853 mm
0.2 best method for calculating DBT T from the fracture
∆T n = 25.11 °C ∆T n = 8.80 °C
energies is using the maximum method for calculating
D

0
Efrac , normalizing it by the maximum force Fmax and
-200 -150 -100 -50 0 50 fitting a tanh function according to eq. 3. That corre-
T [°C]
sponds to the evaluation shown in Fig. 5 (a).
TE

Figure 5: Fracture energy Efrac normalized to the maximum force as


function of temperature evaluated using the maximum (a) and the first
3.3. DBT T from fracture strain
crack (b) method. Another possibility for determining the DBT T from
SP tests is by means of the fracture strain εf which is
EP

defined as [1]:
!
t0
is not suited to describe the data in the entire upper shelf εf = ln (4)
region. One possibility to overcome this problem would tf
C

be to use only data on the low temperature side of the where t0 is the initial specimen thickness and tf the final
upper shelf for the fit. We propose a different approach: specimen thickness next to the area of failure. We have
AC

the drop of Efrac with rising temperature is likely caused used X-ray computed tomography (CT) for determining
by the reduction of the ultimate tensile strength σUTS εf for a number of SP specimens. Since the CT scans are
with rising temperature. The maximum force in a small quite time consuming, only the specimens tested with a
punch test Fmax is correlated to σUTS [4, 5, 29]. We displacement rate of 0.5 mm/s have been investigated.
therefore propose normalizing Efrac with Fmax , the max- Fig. 6 shows the reconstructed 3D model of a SP
imum force reached in the test. When the maximum specimen tested at −196 ◦ C. The color coding refers
method is applied for calculating Efrac , Fmax is simply to the specimen thickness which determined as laid out
the maximum force during the entire test; when the first in section 2.3 . The thickness has not changed signif-
crack method is applied, Fmax is the first (local) maxi- icantly in the outer part of the specimen where it was
mum of the force signal. The results are plotted in Fig. 5 clamped during the test. In the central part of the speci-
for both evaluation methods. men, however, significant thinning can be observed with
6
ACCEPTED MANUSCRIPT

1.5
0.5 mm/s

ǫf

PT
0.5
ǫ min = 0.14
ǫ max = 1.06

RI
DBTTǫ = -123.0 °C
Figure 6: Full view of a CT image of specimen TL14 (tested at
∆ Tǫ = 13.1 °C
−196 ◦ C) with automatic thickness determination. In the gray areas
no thickness could be calculated. 0
-200 -150 -100 -50 0 50

SC
T [°C]

Figure 8: Fracture strain εf as function of temperature T for specimens


tested with a displacement rate of 0.5 mm/s.

U of DBT T n :
AN
εUS + εLS εUS − εLS  T − DBT T 
ε
εf = + tanh (5)
2 2 2∆T
The transition temperature DBT T ε determined from
M

the fracture strain is −123 ◦ C which coincides very well


with the DBT T of −125 ◦ C determined from the same
Figure 7: Profile taken from the specimen in Fig. 6. specimens’ normalized fracture energies using the max-
D

ium method (solid black curve in Fig. 5 (a)). In an


earlier study on the micro-alloyed ferritic-pearlitic steel
AE460, the DBT T from fracture strain was found to be
TE

the thinnest area being in a ring around the specimen 15 ◦ C higher than the DBT T from the fracture energies
center. [19].
The current CWA recommends cutting the specimen
through the centre on a line perpendicular to the crack
EP

and measure its thickness ”adjacent to the area of fail- 4. Discussion


ure”. The CT data allows building on that procedure and
analysing the thickness along two perpendicular diamet- Based on a number of SP tests and specimens two
ric cuts. Fig. 7 shows one of these cuts with the colors completely independent evaluations have resulted in
C

again referring to the wall thickness. The thinnest part close estimates for the DBT T . One evaluation was
along the profile has been used to calculate the fracture based on the force and deflection data registered during
AC

strain according to eq. 4. For determining the thinnest the test (DBT T 2fits , DBT T n ) whereas the other method
part of the specimen the areas in violet at the edges used post-test analysis of the specimens (DBT T ε ). The
of the crack have been discarded. E.g. in the case of good agreement between the transition temperatures
the specimen in Figs. 6 and 7 tf was determined to be gives a strong confidence in their accuracy. Neverthe-
407 µm. less, previous studies using SP tests for determining the
In Fig. 8 εf is plotted as a function of temperature DBT T of Gr. 91 came to significantly lower transition
T . The transition from brittle failure at lower temper- temperatures around −164 ◦ C [11] and −167 – −178 ◦ C
atures to ductile behaviour at higher temperatures is in an inter-laboratory comparison [5]. Since the latter
clearly visible. The corresponding transition tempera- study used material from the same block of Gr. 91 as
ture DBT T ε was determined by a least squares fit of a the present work, material variations can be ruled out as
tanh function of the same form as for the determination a possible reason for this discrepancy.
7
ACCEPTED MANUSCRIPT

An explanation might be found in the test setup itself: 3000


while all tests were carried out using 0.5 mm thick spec- 2.0 mm punch
2.5 mm punch
imens with 8 mm diameter and the diameter of the lower 2500

die was 4 mm, the two studies cited above used punches
2000
with a tip diameter of 2.5 mm whereas the punch used

F [N]
in this study had a tip diameter of 2 mm. It seems likely 1500
that the smaller punch could have lead to the higher tran-

PT
1000
sition temperatures. In fact in [5] smaller TEM-type SP
specimens were compared to the ususal SP specimens. 500
The TEM-type specimens had a diameter of 3 mm and
a thickness of 0.25 mm. The punch tip diameter was 0

RI
0 0.5 1 1.5
1 mm. The DBT T determined from the smaller speci- u [mm]
mens showed a larger spread but were generally lower
than the values from classical SP specimens (−170 – Figure 9: Force deflection curves for Gr. 91 specimens at −140 ◦ C
with two different punch diameters.

SC
−193 ◦ C). This supports the hypothesis that the test
setup has an impact on the DBT T .
To confirm or refute this hypothesis a number of addi-
tional small punch tests were performed with two punch were based on the maximum force and the first
diameters (2 mm and 2.5 mm) at −140 ◦ C, a tempera-

U
crack, respectivly, the former method proved to be
ture lying between the DBT T from the literature and the better choice because it led to less scatter in the
the DBT T determined in the present study. The tests data.
AN
have been carried out at a cross-head displacement rate 2. The method proposed in this work for calculating
of 0.005 mm/s on a different melt of Gr. 91 (60 mm the DBT T based on fracture energies normalized
thick plate 20057 (M1) from ArcelorMittal delivered for with the maximum force has significant advantages
the MATTER project). In two cases the hemispherical compared to the method frequently used in the lit-
M

punch was replaced by a ceramic ball bearing ball with erature which directly uses the fracture energies. It
the same diameter (see Tab. 5). This change is not ex- allows using a single fitting function over the entire
pected to have an effect on the test. The advantage of us- temperature region, requires less fitting parameters
D

ing balls is that these can easily be changed after every and avoids a manual attribution of data points to
test which eleminates potential issues related to wear of the lower or upper shelf region.
the punch tip caused by repeated testing with the same
TE

3. X-ray computed tomography is a reliable, non de-


punch. structive means for determining fracture strains.
A criterion for distinguishing between brittle and
4. Fracture strains can be used for calculating the
ductile failure could be whether the force-deflection
DBT T . The DBT T determined independently
curve reaches the ”force reversal point” i.e. the point
EP

from normalised fracture energies and fracture


where dF/du = 0. Applying this criterion to the force-
strains of the same specimens are consistent.
deflection curves from the additional tests (Fig. 9) show
5. Depending on the punch diameter SP tests under
that all three specimens tested with the 2 mm punch in-
otherwise identical conditions can lead to brittle or
curred brittle failure. In the tests with a 2.5 mm punch
C

ductile fracture with a smaller punch diameter fa-


only the single early failure was brittle whereas the three
voring brittle fracture. This explains the difference
other specimens failed in a ductile way. This observa-
AC

between the DBT T determined in the present study


tion confirms the impact of the punch diameter on the
and data from the literature.
DBT T determined by small punch testing.

5. Conclusions 6. Appendix: Test Data


A series of tensile SP tests were carried out on Gr. 91 7. Acknowledgment
specimens with two displacement rates and the ductile
to brittle transition temperatures have been determined. The materials were made available through the MAT-
The following main conclusions can be drawn: TER project within the European Union Seventh Frame-
1. From the two methods used for determining the work Programme (FP7/2007-2013) under grant agree-
fracture energy from force-deflection curves which ment number 269706. The reasearch was carried out
8
ACCEPTED MANUSCRIPT

maximum first crack


Displacement
Specimen T [◦ C] Efrac [J] Efrac
Fmax [m] Efrac [J] Efrac
Fmax [m] εf
rate [mm/s]
TS08 0.005 -196 0.633 0.000420 0.633 0.000420 -

PT
TS15 0.005 -196 0.378 0.000365 0.273 0.000266 -
TS11 0.005 -170 0.681 0.000444 0.681 0.000444 -
TL08 0.005 -150 1.045 0.000582 0.370 0.000338 -
TL23 0.005 -132 1.109 0.000580 1.110 0.000580 -

RI
TS09 0.005 -120 1.407 0.000703 1.399 0.000699 -
TL17 0.005 -100 1.624 0.000816 1.624 0.000816 -
TL22 0.005 -60 1.530 0.000859 1.529 0.000859 -

SC
TS05 0.005 -20 1.482 0.000920 1.482 0.000920 -
TL24 0.005 23 1.287 0.000854 1.287 0.000854 -
TL14 0.5 -196 0.709 0.000517 0.019 0.000053 0.206
TL10 0.5 -196 - - - - 0.142

U
TL16 0.5 -170 0.825 0.000477 0.825 0.000477 0.223
TS13 0.5 -170 0.528 0.000379 0.528 0.000379 0.272
AN
TL12 0.5 -150 0.760 0.000506 0.261 0.000254 0.243
TL20 0.5 -130 0.895 0.000522 0.895 0.000522 0.307
TS07 0.5 -120 1.705 0.000788 1.705 0.000788 0.835
TL26 0.5 -100 1.703 0.000809 1.703 0.000809 0.853
M

TL18 0.5 -60 1.614 0.000861 1.614 0.000861 1.039


TS10 0.5 -20 1.458 0.000839 1.458 0.000839 1.121
TL25 0.5 23 1.332 0.000840 1.332 0.000840 1.027
D

Table 4: Test data for Gr. 91 from the INTEGRITY project. The fracture strain has only been determined for the tests at a cross head displacement
rate of 0.5 mm/s. For specimen TL10 no curve data is available because of a problem with the data acquisition. The full data sets are identified
through digital object identifiers (DOIs) [20]. The data can be accessed through the European Commission’s online data base MatDB via the ODIN
TE

portal [21]. Additional information on the experimental procedures is available in [14].


EP

Specimen T [◦C] Punch diameter [mm] Punch type Failure mode Fmax [N]
C

AW023 -142 2 hemispherical brittle 1946


AW024 -135 2 hemispherical brittle 1824
AC

AW028 -142 2 hemispherical brittle 1833


AW025 -142 2.5 ball ductile 2797
AW033 -141 2.5 hemispherical ductile 2720
AW034 -142 2.5 hemispherical brittle 1684
AW036 -141 2.5 ball ductile 2657

Table 5: Tests on specimens from the 60 mm plate of Gr. 91 from the MATTER project. The full data sets are identified through digital object
identifiers (DOIs) [30]. The data can be accessed through the European Commission’s online data base MatDB via the ODIN portal [21].

9
ACCEPTED MANUSCRIPT

in the framework of the EERA (European Energy Re- 8. References


search Alliance) Joint Programme on Nuclear Materi-
als. [1] Small Punch Test Method for Metallic Materials, CEN Work-
shop Agreement, CWA 15627:2007 E.
[2] G. Lucas, Review of small specimen test techniques for irradi-
ation testing, Metallurgical Transactions A 21A (1990) 1105–
1119.
[3] R. Lacalle, J. Álvarez, F. Guitérrez-Solana, Analysis of key fac-

PT
tors for the interpretation of small punch test results, Fatigue &
Fracture of Engineering Materials & Structures 31 (2008) 841–
849.
[4] T. Garcı́a, C. Rodrı́guez, F. Belzunce, C. Suárez, Estimation of
the mechanical properties of metallic materials by means of the

RI
small punch test, Journal of Alloys and Compounds 582 (2014)
708–717.
[5] E. Altstadt, H. Ge, V. Kuksenko, M. Serrano, M. Houska,
M. Lasan, M. Bruchhausen, J.-M. Lapetite, Y. Dai, Criti-

SC
cal evaluation of the small punch test as a screening proce-
dure for mechanical properties, Journal of Nuclear Materials
(in press) doi:http://dx.doi.org/10.1016/j.jnucmat.
2015.07.029.
[6] M. Saucedo-Muñoz, T. Matsushita, T. Hashida, T. Shoji,
H. Takahashi, Development of a multiple linear regression

U
model to estimate the ductile-brittle transition temperature of
ferritic low-alloy steels based on the relationship between small
AN
punch and charpy V-notch tests, Journal of Testing and Evalua-
tion 25 (2000) 352–358.
[7] S. Rasche, M. Kuna, Improved small punch testing and param-
eter identification of ductile to brittle materials, International
Journal of Pressure Vessels and Piping 125 (2015) 23–34.
[8] D. Finarelli, F. Carsughi, P. Jung, The small ball punch test at
M

FZJ, J. Nucl. Mater. 377 (2008) 65–71.


[9] M. Bruchhausen, K. Turba, F. de Haan, P. Hähner, T. Austin,
Y. de Carlan, Characterization of a 14Cr ODS steel by means of
small punch and uniaxial testing with regard to creep and fatigue
D

at elevated temperatures, J. Nucl. Mater. 444 (2014) 283–291.


[10] ASTM International (Ed.), ASTM standards 2009, Section 3,
E 23 – 07a, Standard Test Methods for Notched Bar Impact
TE

Testing of Metallic Materials, Vol. 03.01 Metals Test Meth-


ods and Analytical Procedures, ASTM International, West Con-
shohocken, PA, USA, 2009.
[11] B. Gülçimen, A. Durmuş, S. Ülkü, R. Hurst, K. Turba,
P. Hähner, Mechanical characterisation of a P91 weldment by
EP

means of small punch fracture testing, International Journal of


Pressure Vessels and Piping 105-106 (2013) 28–35.
[12] T. Lebarbé, S. Marier, P. Agostini, C. Fazio, S. Gavrilov, Pre-
sentation of FP7 MATTER project: General overview, in: Pro-
ceedings of the ASME 2011 Pressure Vessels & Piping Division
C

Conference, Baltimore, Maryland, USA, 2011.


[13] E. Altstadt, Y. Dai, V. Kuksenko, M. Serrano, M. Houska,
K. Turba, M. Lasan, M. Bruchhausen, J.-M. Lapetite, S. Holm-
AC

ström, Small punch collective exercise, Deliverable 2.2 Rev. 2,


FP7-project MATTER (2014).
[14] M. Bruchhausen, J.-M. Lapetite, S. Ripplinger, T. Austin, Small
punch tensile/fracture test data and 3D specimen surface data
on Grade 91 ferritic/martensitic steel from cryogenic to room
temperature, Data in Brief 9 (2016) 245–251.
[15] J.-M. Baik, J. Kameda, O. Buck, Small punch test evaluation of
intergranular embrittlement of an alloy steel, Scripta Matallur-
gica 17 (1983) 1443–1447.
[16] T. Misawa, T. Adachi, M. Saito, Y. Hamaguchi, Small punch
tests for evaluating ductile-brittle transition behavior of irradi-
ated ferritic steels, J. Nucl. Mater. 150 (1987) 194–202.
[17] J. Ha, E. Fleury, Small punch tests to estimate the mechanical
properties of steels for steam power plant: I. Fracture toughness,

10
ACCEPTED MANUSCRIPT

International Journal of Pressure Vessels and Piping 75 (1998)


707–713.
[18] K. Turba, R. Hurst, P. Hähner, Anisotropic mechanical proper-
ties of the MA956 ODS steel characterized by the small punch
testing technique, J. Nucl. Mater. 428 (1–3) (2012) 76–81.
[19] M. A. Contreras, C. Rodrı́guez, F. Belzunce, C. Betegón, Use
of the small punch test to determine the ductile-to-brittle tran-
sition temperature of structural steels, Fatigue & Fracture of
Engineering Materials & Structures 31 (2008) 727–737. doi:

PT
10.1111/j.1460-2695.2008.01259.x.
[20] J.-M. Lapetite, S. Ripplinger, M. Bruchhausen, Small
punch tensile/fracture test data for Gr. 91 material (batch
S50460), JRC Petten, http://dx.doi.org/10.5290/

RI
1900100, http://dx.doi.org/10.5290/1900128
and http://dx.doi.org/10.5290/1900102 to
http://dx.doi.org/10.5290/1900120 inclusive,
v1.0/v1.1, [data set] (2015).
[21] Online data and information network (ODIN). retrieved on 12

SC
october 2015, https://odin.jrc.ec.europa.eu.
[22] A. Rouffié, P. Wident, L. Ziolek, F. Delabrouille, B. Tanguy,
J. Crépin, A. Pineau, V. Garat, B. Fournier, Influences of process
parameters and microstructure on the fracture mechanisms of
ODS steels, J. Nucl. Mater. 443 (2013) 108–115.

U
[23] H. Hadraba, B. Fournier, L. Stratil, J. Malaplate, A.-L. Rouffié,
P. Wident, L. Ziolek, J.-L. Béchade, Influence of microstructure
on impact properties of 9–18%Cr ODS steels for fusion/fission
AN
applications, J. Nucl. Mater. 411 (2011) 112–118.
[24] W. Qiang-mao, W. Rong-shan, S. Guo-ganga, D. Huia, H. Ping,
L. Feng, W. Li-kui, Analysis method of Charpy V-notch impact
data before and after electron beam welding reconstitution, Nu-
clear Engineering and Design 241 (2011) 459–463.
M

[25] T. Misawa, T. Adachi, M. Saito, Y. Hamaguchi, Small punch


tests for evaluating ductile-brittle transition behaviour of irradi-
ated ferritic steels, J. Nucl. Mater. 150 (1987) 194–202.
[26] S.-H. Song, R. Faulkner, P. Flewitt, R. Smith, P. Marmy,
D

Temper embrittlement of a CrMo low-alloy steel eval-


uated by means of small punch testing, Materials Sci-
ence & Engineering: A 281 (2000) 75–81. doi:http:
TE

//dx.doi.org/10.1016/S0921-5093(99)00740-6.
URL http://www.sciencedirect.com/science/
article/pii/S0921509399007406
[27] Y. Dai, P. Marmy, Charpy impact tests on martensitic/ferritic
steels after irradiation in SINQ target-3, J. Nucl. Mater. 343
(2005) 247–252.
EP

[28] J. Kameda, X. Mao, Small-punch and TEM-disc testing tech-


niques and their application to characterization of radiation
damage, Journal of Materials Science 27 (1992) 983–989.
[29] E. Fleury, J. Ha, Small punch tests to estimate the mechani-
cal properties of steels for steam power plant: I. Mechanical
C

strength, International Journal of Pressure Vessels and Piping


75 (1998) 699–706.
AC

[30] J.-M. Lapetite, M. Bruchhausen, Small punch tensile/fracture


test data for Gr. 91 material (batch 20057 (M1)), JRC Pet-
ten, http://dx.doi.org/10.5290/1900121 to http://
dx.doi.org/10.5290/1900127 inclusive, v1.0, [data set]
(2015).

11
ACCEPTED MANUSCRIPT
 Small punch testing was used to determine the DBTT of Gr. 91 steel.
 Evaluation methods based on fracture energy and fracture strain are compared.
 A single type fitting function is proposed for the determination of the DBTT.
 DBTT derived from fracture energy and fracture strain was very similar.
 The data are openly available.

PT
RI
U SC
AN
M
D
TE
C EP
AC

You might also like