Sie sind auf Seite 1von 7

Applied Mechanics and Materials Vol.

70 (2011) pp 123-128
(2011) Trans Tech Publications, Switzerland
doi:10.4028/www.scientific.net/AMM.70.123

Online: 2011-08-18

Using digital image correlation techniques and finite element models for
strain-field analysis of a welded aluminium structure
Mathias Flansbjer1, a, Torsten Sjgren2, b
1,2

SP Technical Research Institute of Sweden, P.O. Box 857, SE-501 15 Bors, Sweden
a

mathias.flansbjer@sp.se
b

torsten.sjogren@sp.se

Keywords: Digital image correlation (DIC), Micro hardness, Finite element analysis (FEA),
Aluminium, Welding, Heat affected zone (HAZ)

Abstract
The use of aluminium in lightweight structures and corrosive environments is continuously
increasing. Welding is often used to assemble different aluminium components and it is of great
importance to take into account the deteriorating effect the welding operation on the aluminiums
strength when designing these structures. In the present study, the strain-fields of welded aluminium
tensile specimens have been analysed by digital image correlation (DIC) techniques, micro hardness
measurements and finite element (FE) modelling. The stress-strain curve of the weld material, the
heat affected zone (HAZ) and the base material have been determined by tensile tests. The material
properties and the extension of the HAZ were also correlated to the micro hardness. The
experimental data has been used as an input to the material models of the FE analysis. The study
shows that using FE-analysis in combination with strain-field determination by DIC is very
powerful. The resulting strain-fields are easily compared and the FE-analysis is verified in a
straightforward way. Furthermore, the results of the study suggest that micro hardness
measurements could be used to derive the strength of the material affected by the welding
operation.
Introduction
To achieve constructions with low weight and good corrosion resistance, aluminium is a good
material choice. The most common method for joining the components of an aluminium structure is
welding.. To design a welded aluminium structure and estimate its load carrying capacity, different
calculation standards and norms are used which regulate how to dimension the structures and welds.
In these standards, e.g. Eurocode 9 (EC9) [1] which is used in Europe, the materials are often
assumed to be elastic-ideal plastic. This assumption is acceptable when estimating failure load but
is rather inaccurate when used to estimate the change in stiffness when the structure starts to yield
and necking occurs. This lack in knowledge of the stiffness becomes troublesome when assessing
how the failure evolves. For safety reasons, a ductile failure is preferred which warns the user
before a catastrophic failure occurs. To better be able to estimate the change in stiffness, finite
element analysis (FEA) can be used. By using FEA, the plastic deformation behaviour of each
material of the structure can be described in greater detail and the structure can be analysed for high
loads and large deformations. However, the validity of a FEA is strongly dependent on the material
models that are used and how well they describe the actual elastic-plastic behaviour. To accurately
model the weld and the region surrounding the weld, the degradation of the material due to the
welding operation has to be taken into account in the material models. In a previous study [2], the
calculated strength of the welded T-joint specimen shown in Fig. 1 was compared to experimental
results. For the calculated strength, a finite element model was developed where the material
properties and the extension of the different zones of the welds were based on the EC9. The
calculated ultimate strength matched the experimental results well. However, the stiffness was
overestimated by the FEA. It was found that this was mainly due to strain localisations in the HAZ
All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of Trans
Tech Publications, www.ttp.net. (ID: 103.6.236.38, Universiti Teknologi Mara (UiTM), Shah Alam, Malaysia-22/09/15,09:04:11)

124

Advances in Experimental Mechanics VIII

zones that were not accurately resolved. Furthermore, the comparison shows that an FE-model that
does not take into account the reduced strength of the HAZ overestimates the stiffness and the
ultimate strength of the T-joint by about 40 % when compared to the test results.
(a)

(b)

(c)

Figure 1. (a) Schematic representation of set-up for the T-joint test and strain field results obtained
from (b) DIC and (c) FEA [2].
In the present study, welded aluminium tensile specimens have been analysed by mechanical
testing, digital image correlation (DIC), micro hardness measurements and FEA regarding their
mechanical properties and how changes in material properties in the vicinity of the weld affect the
strain-field. The determined mechanical properties have been implemented in an FE-model and the
calculations have been verified by using DIC during testing. Studies analysing the strength reducing
effect of the HAZ and relating it to different standards are found in the literature [3].
Experimental methods
Mechanical testing. Two aluminium plates of material grade EN-AW 5754 H22 were tungsten
inert gas (TIG) welded together and tensile test specimens were machined from the welded
structure. Also, standard tensile test specimens were machined from the plate material in order to
determine the properties of the base material. To determine the strength degradation of the heat
affected zone (HAZ), miniature tensile test specimens were cut at different distances from the
centre of the weld. The specimens were machined by water cutting with the geometries shown in
Fig. 2. Specimens of series BB were cut perpendicular to the weld. The weld is indicated in Fig. 2.

Figure 2. Test specimen geometries.

Applied Mechanics and Materials Vol. 70

125

The tensile tests were performed according to standard SS-EN ISO 6892-1:2009 [4] at a strain rate
of 1 mm/min. The strain was measured with a double sided measuring extensometer. The micro
hardness across the weld was also studied. Hardness tests were performed according to standard SSEN ISO 6507-1:2006 [5]. The tests were performed with an applied weight of 200 g and
indentations were measured at 50 magnification.
Digital Image Correlation (DIC) Technique. To analyse how the strain field evolves during
loading of the tensile test specimen, DIC was used. In order for the DIC technique to work for
metallic specimens, a pattern is applied on the surface of the specimen prior to testing. The pattern
is analysed by the software and facets are identified throughout the image series acquired during the
test. From the changes in facet geometry a strain field is calculated which shows the strains on a
local scale for the tested specimen. For the tests in this study, the Aramis system from GOM
mbH was used. Further information on the DIC technique can be found in [6, 7]. Prior to testing,
the two 4.0 megapixel cameras are positioned at specified angles and distances relative to each
other and the test object and a calibration procedure is performed. This gives a measuring volume
which the test object is positioned in. In this study, the measuring volume was 10010080 mm3.
The image acquisition rate was 1 Hz.
Results and discussion
Test results. The development of strain distribution along the tensile specimen BB2 determined by
DIC is presented in Fig 3. Based on the strain distributions, three characteristic regions
corresponding to base material, heat affected material and weld metal can be identified. Although
the boundaries of the zones are not clearly defined, it is possible to determine the extension of the
HAZ on either side of the weld by studying the strain distribution at different stress levels. From the
highly non-uniform strain distribution it becomes very clear that the stiffness of the tensile
specimen is a result of the stiffness of the three individual regions.
(a)
(b)

Figure 3. Development of strain distribution across the weld during loading to final fracture, as
determined by DIC. In (a) the strain is presented along a section and in (b) as the total strain-field.
To determine the material parameters of the base material, the deteriorated material in the HAZ and
the weld, tensile tests were performed on the miniature specimens shown in Fig. 2. From these tests,
the stress-strain curves shown in Fig. 4 were generated. Stress-strain curves were also derived
directly from the DIC results of specimen BB2 by placing virtual extensometers and measuring
strain locally over each individual zone (
,
,
and
) as indicated in Fig 3. There
was a good correlation between the DIC results and the tensile test results for all zones.
Comparing the experimental strength data from miniature tensile tests with strength values from the
base material tests reduced according to EC9 [1] it is observed that the strength across the weld is
both over- and underestimated by EC9 and the width of HAZ is overestimated by EC9 as shown in
Fig. 4. The strength values of the weld are based on the filler material specified in EC9.

126

(a)

Advances in Experimental Mechanics VIII

(b)

Figure 4. (a) Stress-strain curves of the materials in and around the weld. (b) Ultimate tensile
strength (UTS) and yield stress (YS) vs. distance from centre of weld.
To improve the determination of the strength of a welded aluminium structure it is observed that
hardness measurements can be used to determine the extension of the HAZ and the material
properties in and around the weld. Fig. 5a shows a hardness measurement matrix that was acquired
in the area of the weld and the HAZ. Linear correlations between the hardness measurements
(HV0.2) and the ultimate tensile strength (UTS) and yield strength (YS) were observed (see Fig.
5b). These correlation equations are given in Eqs. 1 and 2. The hardness values used in Fig 5b give
an average based on the data at the specified distance 0.5 mm, i.e. averaging 21 values.
UTS=1.47HV0.2+113.7

(1)

YS=5.84HV0.2-265.2

(2)

Using Eqs. 1 and 2, the strength parameters for EN-AW 5754 H22 structures welded with an
AlMg5 filler can easily be determined by performing a hardness matrix survey (120x7 indentations)
across the weld as illustrated in Fig 5a. The strength parameters determined from Eqs. 1 and 2 are
presented as section profiles in Fig. 4b (UTS Mod and YS Mod).
(b)

(a)

Figure 5. (a) Hardness and strength distribution in and around the weld. (b) Hardness and strength
correlation.

Applied Mechanics and Materials Vol. 70

127

FEA study of welded tensile specimen. In the FE-model, the welded tensile specimen was divided
into three regions; the base material, the HAZ and the weld, see Fig. 6a. The software
ABAQUS/Standard was used for the FEA. The extension of the regions was determined based on
the strain distribution obtained from the test by DIC and the hardness profile as discussed earlier,
see Fig 3 and 4. The material models implemented in the FEA were derived from the miniature
tensile tests and individual deformation curves were used for each region. The transformation from
nominal stress-strain curves (
) into true stress-strain curves (
) is schematically shown
in Fig. 6b. The model considers the material linear elastic until the yield stress is reached (region
A). Between the start of yielding to necking (region B) the material exhibits a uniform yielding.
Region C represents an ideal yielding that takes into account the reduction in area at high strains.
(a)

(b)

Figure 6. (a) Regions of the specimen in the FE-model. (b) Schematic representation of true stressstrain curves used in FEA.
Fig. 7a shows a comparison of the load-deformation curves obtained from tests and FEA. The
premature failure of BB1 was determined by fracture in the weld. It is seen that the FEA gives a
rather good correlation to the experimental data, though it overestimates the strength and
deformation capacity of the welded specimen.
(a)

(b)

Figure 7. (a) Comparison of the load-deformation curves obtained from tests and FEA and (b)
comparison of strain-sections from DIC (test BB2) and FEA at a deformation of 1.4, 4.6 and 8.3
mm.

128

Advances in Experimental Mechanics VIII

Studying the strain development across the weld, as measured by DIC, it is observed (Fig. 3) that
the strains start to concentrate to one side of the weld resulting in one-sided necking. This behaviour
is not captured in an idealized FE-model with the same material properties in both HAZ. Due to the
observation of this one-sided necking, an alternated FE-model, denoted FEM alt, was used in
which the HAZ on one side of the weld was set to have a lower hardening behaviour after the start
of yielding. In this weaker HAZ, the strength was reduced by 5%, which approximately corresponds
to the lower stress-strain curve obtained from the miniature tensile tests. In this way, the strain
distribution in the FEA correlates much better with the test results and the load-deformation curves
show very good agreement as depicted in Fig. 7a.
Concluding remarks
In detailed calculations of a welded aluminium structure it is critical that the deterioration of the
material properties of the weld and the heat affected zone is considered in a relevant way (or maybe
treated accurately). If this is not accurately taken into account, the stiffness and the strength of a
structure might be highly overestimated. Using the assumptions given in EC9 as input to FEcalculations has shown to be an acceptable method for estimating failure loads for welded
aluminium structures, but leads to errors in the estimation of stiffness when the structure starts to
yield and strains tend to concentrate in the heat affected zones. In the present study, welded
aluminium tensile specimens have been analysed by mechanical testing, digital image correlation
(DIC), micro hardness measurements and finite element (FE) modelling regarding their mechanical
properties and how changes in the material properties in the vicinity of the weld affect the strainfield. From the strain field obtained by DIC it is possible to find the heat affected zone distribution
on either side of the weld and to determine the important material parameters of the base material,
the deteriorated material in the HAZ and the weld. The determined mechanical properties have been
implemented in a FE-model and the calculations have been verified by use of DIC. Furthermore, the
study suggests that hardness measurements could be correlated to the strength properties of the
deteriorated material. Hence, by combining DIC and hardness measurements, the correlation
between the hardness and the strength for a given combination of base material and filler can be
derived from standard tensile tests of the base material and tensile tests of welded specimens. This
correlation, along with hardness measurements could then be used to derive the strength of the
material affected by the welding operation in more complex welded structures.
Acknowledgements
The authors are thankful to Egil Bartos who performed some of the tests and calculations as part of
his master thesis work.
References
[1] prEN 1999-1-1:2007 Eurocode 9: Design of aluminium structures, European Committee for
Standardization, Brussels, 2007.
[2] Bartos E.: Modellering av vrmepverkad zon vid svetsning av aluminium (Modelling of the
Heat Affected Zone in Aluminium Welds), Master thesis, KTH, Stockholm, Sweden 2007.
[3] Li J-B., Zhang Q-L. and Ding J-M.: Experiments on Properties of Aluminium Welding Joints,
Structural Engineering International, Vol. 4, 2006, pp. 331-338.
[4] SS-EN ISO 6892-1:2009: Metallic materials - Tensile testing - Part 1: Method of test at room
temperature, European Committee for Standardization, Brussels, 2009.
[5] SS-EN ISO 6507-1:2006: Metallic materials - Vickers hardness test - Part 1: Test method,
European Committee for Standardization, Brussels, 2006.
[6] Sutton M.A., Orteu J-J. and Schreier H.W.: Image Correlation for Shape, Motion and
Deformation Measurements, Springer, New York, 2009.
[7] GOM mbH: ARAMIS, User manual, (v6-1_1st_en_rev_c), 2009.

Advances in Experimental Mechanics VIII


10.4028/www.scientific.net/AMM.70

Using Digital Image Correlation Techniques and Finite Element Models for Strain-Field Analysis of a
Welded Aluminium Structure
10.4028/www.scientific.net/AMM.70.123
DOI References
[3] Li J-B., Zhang Q-L. and Ding J-M.: Experiments on Properties of Aluminium Welding Joints, Structural
Engineering International, Vol. 4, 2006, pp.331-338.
doi:10.2749/101686606778995209
[6] Sutton M.A., Orteu J-J. and Schreier H.W.: Image Correlation for Shape, Motion and Deformation
Measurements, Springer, New York, (2009).
doi:10.1007/978-0-387-78747-3