S. HERTELÉ et al.: CHARACTERIZATION OF HETEROGENEOUS ARC WELDS ...
571–574
CHARACTERIZATION OF HETEROGENEOUS ARC WELDS
THROUGH MINIATURE TENSILE TESTING AND
VICKERS-HARDNESS MAPPING
KARAKTERIZACIJA HETEROGENIH ZVAROV S POMO^JO
MINIATURNIH NATEZNIH PREIZKUSOV IN MATRI^NIMI
MERITVAMI TRDOTE PO VICKERSU
Stijn Hertelé1, Jonas Bally1, Nenad Gubeljak2, Primo` [tefane2,
Patricia Verleysen3, Wim De Waele1
1Ghent University, Department of Electrical Energy, Systems and Automation, Soete Laboratory, Technologiepark Zwijnaarde 903,
9052 Zwijnaarde, Belgium
2University of Maribor, Laboratory for Machine Parts and Structures, Smetanova 17, 2000 Maribor, Slovenia
3Ghent University, Department of Materials Science and Engineering, 9052 Zwijnaarde, Belgium
stijn.hertele@ugent.be
Prejem rokopisa – received: 2015-07-01; sprejem za objavo – accepted for publication: 2015-07-13
doi:10.17222/mit.2015.157
The heterogeneity of arc-welded connections is often ignored in structural assessments, giving rise to inaccuracies. Improved
assessments taking into account heterogeneity require the characterization of local constitutive properties. We have compared
two methods to do this: Vickers-hardness mapping and miniature tensile testing. Whereas the former is more straightforward to
apply, the latter provides full-range stress-strain data. This paper discusses an experimental comparison of both methods on a
heterogeneous arc weld. Miniature tensile tests were performed, using digital image correlation to measure the strain. The
specimens were indented to compare their stress-strain response with Vickers hardness. Notwithstanding that small natural
flaws invalidated some tests, reliable stress-strain curves were obtained. Vickers hardness testing is a convenient alternative if
the yield and ultimate tensile strength are the only points of interest and the corresponding conversion inaccuracy is acceptable.
Keywords: arc weld, heterogeneity, hardness, miniature tensile testing, digital image correlation
Heterogenost oblo~no zavarjenega zvara pogosto ni upo{tevana pri ocenjevanju celovitosti konstrukcij, kar je lahko razlog za
nepravilno oceno. Za bolj{o oceno, ki upo{teva heterogenost, je potrebno dolo~iti lokalne mehanske lastnosti. Avtorji ~lanka so
opravili primerjavo dveh metod za dolo~anje lokalnih mehanskih lastnosti: matri~no meritev mikrotrdote po Vickersu in
miniaturni natezni preizkus. Za razliko od dosedanje prakse neposredne uporabe mejnih vrednosti, se danes uporablja celotna
natezna krivulja napetost-deformacija. V ~lanku je opravljena primerjava med obema eksperimentalno dobljenima skupinama
podatkov za heterogeni zvarni spoj. Meritve deformacij na miniaturnih nateznih preizku{ancih so bile opravljene s stereo
opti~no digitalno merilno metodo. Na vsakemu od miniaturnih nateznih preizku{ancev je bila opravljena meritev mikrotrdote po
Vickersu, pri ~emer so bile izmerjene vrednost mikrotrdote primerjane z natezno krivuljo napetost-deformacija. Kljub dejstvu,
da so bili preizku{anci z majhnimi napakami v zvarih izklju~eni iz analize, so bile dobljene verodostojne vrednosti meje te~enja
in natezne trdnosti iz krivulje napetost-deformacija. Pokazalo se je, da je za hitro oceno natezne trdnosti in meje te~enja
materiala zvara, matri~na meritev mikrotrdote po Vickersu alternativna metoda nateznemu preizku{anju.
Klju~ne besede: oblo~ni zvar, heterogenost, trdota, miniaturno natezno preizku{anje, digitalne opti~ne meritve
1 INTRODUCTION
Arc welds are prone to defects of which the accept-
ability should be investigated from the viewpoint of
structural integrity. The occurrence of different micro-
structures in a confined area implies potentially strong
local variations of stress-strain behaviour (weld hetero-
geneity), leading to uncontrolled scatter in the accuracy
of defect-tolerance predictions.1 A first step in account-
ing for weld heterogeneity in an improved assessment is
the experimental characterization of local stress-strain
properties. This paper compares two techniques to do
this: Vickers-hardness mapping and miniature tensile
testing.
The first technique is to characterize the spatial
distribution of hardness by applying a large number of
adjacent indents. This method indicates variations of
strength2, but is not linked to ductility and strain-harden-
ing behaviour. When full-range information is desired,
miniature tensile testing is an alternative. The associated
test procedure, however, is characterized by a large num-
ber of technical challenges and requires a careful pre-
paration, execution and analysis of the tests.3
The authors have developed and optimized a proce-
dure for miniature tensile testing, and have compared the
results of a test program with Vickers hardness data. This
paper highlights the challenges that have been addressed
to achieve reliable tensile test results, and the relation
between hardness and local stress-strain properties.
Materiali in tehnologije / Materials and technology 50 (2016) 4, 571–574 571
UDK 621.791.053:620.172:620.178.151.6 ISSN 1580-2949
Original scientific article/Izvirni znanstveni ~lanek MTAEC9, 50(4)571(2016)
2 MATERIAL
Test material was taken from the girth weld of a
914-mm-diameter steel pipeline having a nominal wall
thickness of 10 mm, excavated from a natural-gas trans-
mission grid after nearly five decades of operation. The
specified minimum yield strength of the microalloyed
steel was 415 MPa and the weld was constructed by
multipass shielded metal arc welding (SMAW). This
weld was selected for its severe heterogeneity, as wit-
nessed by the 49 N Vickers (HV5) hardness map shown
in Figure 1. Traversing from root to cap, the hardness
increases by roughly 50 % of its minimum value.
Similarly significant variations are to be expected for the
yield and ultimate tensile strength given their
approximate relation with hardness.
3 TEST PROGRAM
Miniature tensile specimens were sampled from the
weld by means of electrical discharge machining (EDM)
with a 0.25-mm-diameter wire electrode. This technique
is suitable as no forces are applied to the vulnerable spe-
cimens. A dog-bone-shaped block, taken out in the weld-
ing direction, was divided into 0.7-mm-thick slices
(Figure 2). Nine specimens were extracted. These were
then ground on both sides to a thickness of 0.5 mm, thus
removing the brittle heat-affected zone associated with
the EDM. The specimens are numbered in a manner that
indicates their through-thickness position, starting from
the weld root (1) and heading towards the weld cap (9).
The adopted dog-bone geometry has a nominal cross-
sectional area of 1 mm2. It was developed at the GKSS
research centre (Helmholtz-Zentrum Geesthacht)4 and
has been adopted by other institutions.5,6
Note the prismatic section’s ratio of length (9 mm) to
width (2 mm), allowing us to measure strain using an
extensometer whose gauge length is four times the speci-
men width, in accordance with the standards for conven-
tional tensile testing of metallic materials.7 Note a
filleted end corner having a radius of 1 mm, providing a
unique reference for the specimen orientation with res-
pect to the weldment.
Following grinding, the top and bottom surfaces were
polished using 2000-grit sandpaper, on average reducing
the roughness Ra (measured in a longitudinal track) from
0.26 μm to 0.026 μm. The side faces were also polished
(using 1200-grit sandpaper) to remove the burrs resulting
from the grinding process. Without this step, specimen-
width measurements would have been biased, giving rise
to an estimated error of 10 % on the stress calculation.
For all nine specimens, the net section width and
thickness were measured at five equidistant positions.
Small coefficients of variance were observed between all
45 values of the width (0.82 %) and thickness (1.74 %),
indicating a sound dimensional repeatability of the
specimen production. The specimen-specific coefficients
of variance did not exceed 0.5 % for the width and 1.0 %
for the thickness.
To examine the correlation between the Vickers
macrohardness and the tensile-test characteristics, a total
of twelve HV5 measurements were performed in the
clamping faces of each specimen (six at either end of the
specimen). ASTM E3848 was consulted to confirm that a
specimen thickness of nominally 0.5 mm is sufficient to
obtain valid HV5 indentations, and to ensure a sufficient
spacing between the adjacent indentations.
After polishing and hardness testing, specimens were
painted white, followed by applying black speckles for
the purpose of monitoring strain by means of digital
image correlation (DIC). This is an optical measurement
technique that allows us to track full-field displacements
by following the movements of a random pattern. Next, a
švirtual extensometer’ measurement of strain can be out-
put from the obtained displacement data. A stand-alone
system provided by Limess GmbH, adopting the VIC3D
software (version 2009) of Correlated Solutions Inc, was
used for the DIC analysis.
Particularly challenging was the application of a
high-quality speckle pattern. Optimal DIC accuracy
requires an average speckle size of around 3×3 camera
pixels.9 From the camera resolution and the dimensions
S. HERTELÉ et al.: CHARACTERIZATION OF HETEROGENEOUS ARC WELDS ...
572 Materiali in tehnologije / Materials and technology 50 (2016) 4, 571–574
Figure 2: Sampling plan and geometry of miniature tensile specimens
Slika 2: Plan izreza vzorcev in geometrija
Figure 1: Hardness map of the investigated weld
Slika 1: Povr{inska porazdelitev trdote na analiziranem zvaru
of the area of interest, a desired speckle size of
18×18 μm followed. The literature indicated that such a
speckle size can be achieved by non-conventional
methods such as dispersing and heating graphite powder
on the surface, or contact lithography.9 For this project,
however, conventional painting was performed by means
of an in-house-developed spray procedure using a nozzle
airbrush. The optimization of paint viscosity, nozzle
opening, airbrush pressure and spray angle eventually
allowed us to create patterns having the desired speckle
size with sufficient repeatability. Figure 3 shows a
speckled specimen and a close-up of the pattern at a
pixel level, confirming the ability to obtain the desired
fine speckle size.
The miniature tensile tests were executed using a
5-kN Deben stage, made available by TU Delft for the
purpose of this study. Specimens were clamped between
grips that were actuated at a fixed displacement rate,
which was empirically tuned to limit the stress rate in the
linear-elastic region to a value below 11.5 MPa/s (as
required by ASTM E8M7 for the standard tensile testing
of metallic materials). The tests were stopped upon
fracture. The load was recorded and coupled with DIC
images that were taken at a rate of 0.5 Hz. To obtain
stress-strain data, the stress was calculated by dividing
the tensile load by the average values of the net section
width and thickness. The strain was obtained from DIC
by means of a švirtual extensometer’ having a gauge
length of 8 mm, as discussed above.
4 RESULTS AND DISCUSSION
This section focuses on three aspects of the test pro-
gram: the validity of the miniature tensile test results, the
observations related to weld heterogeneity, and the
relation between the Vickers hardness and the strength
characteristics.
Of all nine specimens, three specimens (3, 4 and 9)
clearly showed an anomalous response, characterized by
localized necking at unexpectedly low strain values.
Post-mortem fractography revealed weld porosities in
each of these specimens. The relative size of such porosi-
ties with respect to miniature tensile specimens is clearly
sufficient to invalidate their test result (Figure 4).
No irregularities were observed on the fracture sur-
faces of the other specimens, which behaved in an expec-
table manner. Their stress-strain curves are depicted in
Figure 5. Weld heterogeneity is reflected in many
aspects. Obviously, yield strength (expressed as 0.2 %
proof stress Rp0.2) and tensile strength Rm vary consider-
ably (493 MPa < Rp0.2 < 672 MPa and 570 MPa < Rm <
794 MPa). In addition, there are strong variations in the
ductility and strain-hardening behaviour. Comparing
specimen 1 with 8, the former has a Lüders plateau and a
uniform elongation (strain at Rm) of 10.1 %, whereas the
latter shows round-house yielding and a uniform
elongation of 4.3 %. The extent of the strain hardening is
also variable, as reflected in the ratio Rp0.2/Rm, which
varies between 0.80 and 0.89.
The variety of the stress-strain curve shapes in Fig-
ure 5 indicates the limitation to the use of Vickers hard-
ness as a unique number to characterize local properties.
Nonetheless, the literature and standards suggest that
hardness may be used to estimate Rp0.2 and Rm. For
instance, the following Equation (1), proposed for steel
weld metal in ISO 15653 (in MPa):3
S. HERTELÉ et al.: CHARACTERIZATION OF HETEROGENEOUS ARC WELDS ...
Materiali in tehnologije / Materials and technology 50 (2016) 4, 571–574 573
Figure 5: Test results indicate a significant heterogeneity of the con-
stitutive properties
Slika 5: Rezultati nateznih preizkusov ka`ejo heterogenost v mehan-
skih lastnostih
Figure 3: Speckle pattern for the strain analysis using digital image
correlation
Slika 3: Raster za meritev deformacij z uporabo opti~ne digitalne me-
rilne tehnike
Figure 4: Weld porosities invalidated three miniature tensile test re-
sults
Slika 4: Pore v zvarnem spoju so povzro~ile odstopanje v rezultatih na
nateznih preizku{ancih
R
R
p
m
HV
HV
0 2 235 62 0
3 00 221
. . .
. .
= +
= +
⎧
⎨
⎩
(1)
The six hardness values at one specimen end may
significantly differ from those at the other end (up to 33
HV difference between the average values from either
end). This observation indicates weld heterogeneity in
the weld direction (perpendicular to the section shown in
Figure 1). As the weakest cross-section governs the
failure load of the tensile specimen, the average hardness
value of the weakest specimen end was considered for a
comparison with strength properties (Figure 6). Equa-
tion (1) succeeds in describing the observed relations
between the hardness and the strength characteristics.
For the six valid tests, predicting Rp0.2 and Rm from HV5
results in errors below 25 MPa. The prediction accuracy
is higher for Rm than for Rp0.2 (average absolute value of
error 14.5 MPa and 18.5 MPa, respectively).
5 CONCLUSIONS
This paper has shown the potential merit of miniature
tensile testing to characterize weld stress-strain hetero-
geneity. The careful specimen preparation and execution
of the tests are essential for this. The following attention
points of the experimental procedure have been high-
lighted: specimen polishing, accurately measuring cross-
section dimensions, and applying a suitable speckle
pattern for optical strain monitoring. Test validity re-
quires that the specimen is free of any natural weld
defects such as porosities.
Comparing the miniature tensile test results with the
Vickers hardness testing has pointed out the suitability of
the latter (more straightforward) method, provided the
desired information is limited to the evolution of the
strength variations and, hereto, small strength estimation
errors are allowed. Vickers hardness testing does not
provide direct information with respect to the ductility or
the nature of the strain hardening.
Acknowledgements
The authors wish to acknowledge the financial
support of BOF UGent (grant BOF12/PDO/049), FWO
Vlaanderen and ARRS Slovenia (joint FWO grant
G.0609.15N), and the kind permission of TU Delft to use
their 5-kN test stage.
6 REFERENCES
1 M. Koçak, S. Webster, J. J. Janosch, R. A. Ainsworth, R. Koers
(Eds.), Fitness for service procedure, FITNET, MK8, Vol. 1: Proce-
dure, GKSS Research Centre, Geesthacht, Germany 2008
2 ISO 15653: Metallic materials – method for the determination of
quasistatic fracture toughness of welds, ISO, 2010
3 D. A. LaVan, W. N. Sharpe, Tensile testing of microsamples, Expe-
rimental Mechanics, 39 (1999) 3, 210–216, doi:10.1007/
BF02323554
4 M. Koçak, Structural integrity of welded structures: process – pro-
perty – performance relationship, 63rd Ann. Assembly & Int. Conf.
Int. Instit. Weld., Istanbul 2010, 3–19
5 D. Dobi, E. Junghans, Determination of the tensile properties of spe-
cimens with small dimensions, Kovine zlitine tehnologije, 33 (1999)
6, 451–457
6 S. Hertelé, N. Gubeljak, W. De Waele, Advanced characterization of
heterogeneous arc welds using micro tensile tests and a two-stage
strain hardening (šUGent’) model, International Journal of Pressure
Vessels and Piping, 119 (2014) 9, 87–94, doi:10.1016/j.ijpvp.2014.
03.007
7 ASTM E8M: Standard test method for tension testing of metallic
materials, ASTM International, 2011
8 ASTM E384: Standard test method for Knoop and Vickers hardness
of materials, ASTM International, 2011
9 M. A. Sutton, J. J. Orteu, H. W. Schreier, Image correlation for
shape, motion and deformation measurements, Springer, USA 2009,
doi:10.1007/978-0-387-78747-3
S. HERTELÉ et al.: CHARACTERIZATION OF HETEROGENEOUS ARC WELDS ...
574 Materiali in tehnologije / Materials and technology 50 (2016) 4, 571–574
Figure 6: ISO 15653 describes the observed relations between hard-
ness and strength
Slika 6: ISO 15653 ka`e razmerje med trdoto in trdnostjo ter odsto-
panje od eksperimentalno dobljenih vrednosti