Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Chapter 23

Out-of-Plane Motion Evaluation and Correction in 2D DIC

Michele Badaloni, Pascal Lava, Marco Rossi, Gianluca Chiappini, and Dimitri Debruyne

Abstract 2D and stereo Digital Image Correlation (DIC) allows to retrieve complex displacement and strain fields on
a specimen’s surface. Although 2D DIC is strongly affected by out-of-plane motions, in many situations, it is preferred
over stereo DIC because of its ease to use and because only one camera is required. The out-of-plane movements can be
ascribed mainly to three causes: the camera positioning, the imperfections of the used test device, and the camera self-
heating. These effects gain importance when the distance between the camera and the specimen is reduced. The positioning
of the camera aims to have its optical axis perfectly perpendicular to the specimen to observe. Nevertheless small but
effective misalignments can easily happen even if suitable devices are used for the alignment. This contribution concerns the
experimental evaluation of these movements considering a cyclic uni-axial tensile test performed on an aluminium specimen.
The study is particularly focused to the out-of-plane motions that occur at every cycle because of the tensile bench, which
are the more critical ones. Finally a compensation method, based on fixed compensation plates, is presented. The method
allows to properly correct the data coming from a 2D DIC set-up.

Keywords Digital Image Correlation • Material identification • Out-of-plane motions • Correction • Simulation

23.1 Introduction

One of the reasons of the great recent diffusion of full-field measurement techniques is that they allow retrieving directly
complicated stress–strain fields from experimental tests. Amongst the different full-field techniques, as the grid method [1, 2],
geometric moiré [2, 3] and speckle laser [4], Digital Image Correlation (DIC), became a largely diffused contactless method
to retrieve heterogeneous states of strain on specimens while tested [5]. These data can be exploited to achieve material
mechanical parameters by using different identification methods [6] as, for instance the, Virtual Fields Method (VFM) [7,
8] or Finite Element Model Updating (FEMU) [9, 10]. In the characterization of the behaviour of complex materials, the
advantages of performing a single test, using full-field measurements, instead of many different tests (tensile, shear, punch,
etc.) are straightforward in terms of costs and time consumption. However, every experiment involving DIC is deeply affected
by the experimental set-up used (cameras, lenses, test device, etc.), which change substantially the outcomes. This brought
to a simulating procedure with the aim to numerically reproduce the experimental chain, in order to find out the most
effective configuration to apply, as proposed in [11] where the grid method was used. An extended literature is available
regarding the errors related to DIC settings [12–16], however, the link between DIC errors and identification errors is still
scarcely addressed. In [17, 18] the simulating procedure has been implemented considering the experimental uncertainties
that affect a real acquisition. This was done by studying the interactions between the identification procedure and the DIC
technique. In that paper, the impacts of experimental uncertainties on material identification, such as noise, rigid motions
and illumination issues, have been disentangled and evaluated. In the present contribution, the attention is given to the

M. Badaloni ()
Università Politecnica delle Marche, via brecce bianche, 61100 Ancona, Italy
Department of Materials Engineering, K.U. Leuven, Campus Gent, Gebroeders Desmetstraat 1, 9000 Ghent, Belgium
e-mail: m.badaloni@univpm.it
P. Lava • D. Debruyne
Department of Materials Engineering, K.U. Leuven, Campus Gent, Gebroeders Desmetstraat 1, 9000 Ghent, Belgium
e-mail: pascal.lava@kuleuven.be; dimitri.debruyne@kuleuven.be
M. Rossi • G. Chiappini
Università Politecnica delle Marche, via brecce bianche, 61100 Ancona, Italy
e-mail: m.rossi@univpm.it; g.chiappini@univpm.it

© The Society for Experimental Mechanics, Inc. 2016 181


H. Jin et al. (eds.), Advancement of Optical Methods in Experimental Mechanics, Volume 3, Conference Proceedings
of the Society for Experimental Mechanics Series, DOI 10.1007/978-3-319-22446-6_23
182 M. Badaloni et al.

out-of-plane movements which turned out to be the most effective experimental error source in a 2D DIC measurement. It
is possible to distinguish between three contributions that can cause out-of-plane motions: a component due to the camera
self-heating, a contribution due to the camera positioning, which is fixed once the set-up is built, and a component that varies
during a test because of the lack of stiffness or defects of the used test device. The first component, treated in [19], is not
considered in the present paper since, before making the experimental test, the used cameras have been turned on in order to
reach a saturation temperature. The misalignments due to the camera positioning have been studied in [20, 21], while in [22]
a frame is attached to the specimen surface to compensate the unwanted effects of the out-of-plane motions. In the present
paper, a variation of this technique is adopted by modifying strains instead of displacements. The soundness of the procedure
is first validated numerically by simulating a uni-axial tensile test performed on an aluminium rectangular specimen and then
validated experimentally. Tests only regard the elastic region, therefore the focus in the identification is given to the Young’s
modulus (E) and the Poisson ratio (). Moreover, the movements of the clamps during the test were monitored during the
test using a dedicated set-up.

23.2 Influence and Solutions to the Out-of-Plane Motions

The influence of the out-of-plane motions on displacements and strains has been deeply studied and described in [23]. Let
us consider a specimen that can translate of a quantity .z/ and rotate around the horizontal axis with an angle theta ( )
or respect to the longitudinal axis with an angle gamma ( ). As described in [23], the influence of the out-of-plane on the
displacement can be written as
     
L z L Y sin  L X sin  cos 
U .z; ;  /   X   X   X cos   1  (23.1)
z z z z z z
     
L z L Y sin  cos  L X sin 
V .z; ;  /   Y   Y cos   1   Y  (23.2)
z z z z z z

where L is the image distance, (X,Y) the object in-plane dimensions and z the object distance. Evidently, if a 2D DIC is
performed without any expedient, the results are compromised if out-of-plane motions occur.
Different solutions can be adopted. For instance, applying stereo DIC instead of 2D, thanks to the calibration between
the used cameras, the movements outside the plane are easily filtered out. However, for many reasons (low costs, longer
time, etc.), sometimes a 2D analysis would be preferred. Another possibility is using telecentric lenses, although a big
disadvantage is the impossibility to change the field of view of a given lens. Therefore other solutions have been studied
in the recent years. Methods that can solve the out-of-planes due to a misalignment of the camera can be a rectification,
numerical or with mechanical devices, of the images taken experimentally. These solutions do not consider the out-of-plane
motions due to the movements that the specimen can have during the test though. Even a simple tensile test performed on
an aluminium rectangular specimen shows that often there is a consistent out-of-plane motion. Thus, even if the components
due to the heating of the camera and the misalignment of the camera itself respect to the specimen’s surface are solved
otherwise, still there is the need to solve the contribution of the out-of-planes happening while the specimen is under loading
conditions. A solution to that is faced in the followings.

23.3 Correction Method

A solution is presented in [22], using an holed frame attached on the back side of the specimen. The displacements of
the region of interest (ROI) on the specimen’s surface are compensated by considering the displacements of a region of
compensation (ROC). The assumptions are that the frame follows the motion of the specimen without any load transmitted
among them. A second order polynomial fitting is applied to the displacements obtained analysing the ROC and used to
correct the data coming from the ROI.
h    2 i
2 0 2 2
uci D a0 C a1 xci C a2 yci C a3 xci C a4 xci yci  k1 xci x 0 ci C y 0 ci  xci xci 2
C yci (23.3)
h    2 i
2 0 2 2
vci D b0 C b1 xci C b2 yci C b3 xci yci C b4 yci  k1 xci x 0 ci C y 0 ci  xci xci 2
C yci (23.4)
23 Out-of-Plane Motion Evaluation and Correction in 2D DIC 183

Fig. 23.1 ROI and ROC on an


experimental image

ROI
ROC

where a0 and b0 refer to the in-plane rigid motions; a1 , a2 , a3 , b1 , b2 , b3 are used to correct the rigid motions such as the
in-plane motions and the out-of-plane translations and rotations; k1 takes into account the lens distortions.
Since the displacements on the ROC are known from the DIC, a vector with these parameters can be retrieved and used
to correct the displacements of the region of interest (ROI).
In the present study a similar approach is considered but the correction is made on strains instead. Basically, this means
that a zero order fitting is used. The correction is made as

N
X
"i
iD1
"corr D "meas  (23.5)
N
where the mean value of the strain measured experimentally "meas , is modified by the contributions of the strains measured in
the compensation areas i. In this study, a frame is attached behind the specimen thus a left and the right area are considered,
as visible in Fig. 23.1. Therefore in this case N D 2.
Compared to the method described in [22], the present method, by averaging the strains, is a simplification. This variation
has been validated in case of homogeneous deformations and further studies have to be done in case of more complex
situations. However, whenever the deformation is homogeneous, this method showed smaller standard deviations compared
to the ones obtained using the second order polynomial fitting.

23.4 Numerical Analysis

Before the experimental part, a numerical analysis have been carried on. The tool used for the analysis is a simulating
procedure introduced in [17] and developed in [18]. Starting from an experimental speckle pattern, synthetic images are
® ®
created using Matlab routines. A finite element model of an aluminium 6061 T6 specimen has been created with Abaqus .
The considered width is L D 20.1 mm and thickness t D 4.1 mm. A load of F D 10 kN is applied. The synthetic image is
deformed by applying the displacements coming from the FEM plus a contribution due to the different out-of-plane motions
described in as

U  D U FEM C U .z; ;  / (23.6)


184 M. Badaloni et al.

Table 23.1 Values for the Minimum value Maximum value


out-of-plane motions used in the
simulation z (mm) 0.05 0.05
 (rad) 0.015 0.015
 (rad) 0.015 0.015

case1 case2 case3

ROI
ROC

Fig. 23.2 The three different regions considered in the numerical analysis

Table 23.2 Values for the three Case 1 Case 2 Case 3


considered cases
PixScale 0.025475 0.034957 0.066556
Pixels in the ROI 762174(71 %) 550275(53 %) 296866(27 %)
Pixels in the ROC 153860(29 %) 245388(47 %) 406351(73 %)

Dependency of n on the out-of-plane translation with correction for case 2 Dependency of n on the out-of-plane rotation q with correction for case 2 Dependency of n on the out-of-plane rotation g with correction for case 2
0.4 1.4 0.8
exp exp exp
corr corr corr
0.38 1.2 0.7

0.36 0.6
1
Poisson ratio n

0.5
Poisson ratio n

Poisson ratio n

0.34
0.8

0.4
0.32
0.6
0.3
0.3
0.4
0.2

0.28
0.2
0.1

0.26
-0.05 -0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04 0.05 0 0
-0.015 -0.012 -0.009 -0.006 -0.003 0 0.003 0.006 0.009 0.012 0.015 -0.015 -0.012 -0.009 -0.006 -0.003 0 0.003 0.006 0.009 0.012 0.015
dz
theta q gamma g

Fig. 23.3 Evaluation of z,  and  on the identification of E and . Comparison between the raw values and the corrected ones

where U(z, ,  ) is defined in Eqs. (23.1) and (23.2). Numerically the translation z and the rotations  and  have been
added separately as resumed in Table 23.1.
For the DIC analysis the MatchID software was used (http://www.matchid.org).
The prescribed out-of-plane motions reproduce what happens during a test, thus they cannot be simply solved by taking
care of the alignment of the camera. In the present study, a cyclic test in the elastic regime of the specimen is performed.
As first study, we made a sensitivity analysis on the dimensions of the ROC to consider. Because of the camera aspect
ratio, a set-up with the camera close to the specimen will cause a small region for the ROC, while if the specimen is placed
further from the camera, a bigger region will be available for the compensation but a smaller amount of data will be available
for the region to analyse. A compromise has to be found. Three cases have been considered and illustrated in Fig. 23.2.
Table 23.2 lists their relative characteristics.
Figure 23.3 describes the results for case 2, similar trends are observed in the other cases. All data in Fig. 23.3 are
numerical, the data set called “exp” refers to the data that one would obtain by performing a normal DIC experiment while
the “corr” refer to the data obtained with the correction described in (23.5).
It is confirmed that even a small amount of out-of-plane motion strongly influence the outcomes. As expected, when z,
 and  are null, the numerical raw values and the corrected ones overlay. The three considered cases can be compared to
assess which is the best compromise between the ROI and the ROC size. An evaluation of the results on the Poisson’s ratio
for the three cases is shown in Fig. 23.4, only the corrected values are reported.
23 Out-of-Plane Motion Evaluation and Correction in 2D DIC 185

Fig. 23.4 Evaluation of the Cases comparison on Poisson ratio


effect of z,  and  on the 0.7
identification  in the three cases
considered 0.6

0.5

Poisson ratio
0.4

0.3

0.2

0.1

0
dz theta gamma
case1 case2 case3 FEM

Fig. 23.5 Evaluation of the Case2 comparison on Poisson ratio


effect of z,  and  on the 0.8
identification  in the second
case. The experimental values are 0.7
compared to the corrected ones
0.6

0.5
Poisson ratio

0.4

0.3

0.2

0.1

0
dz theta gamma

exp corrected FEM

Taking as reference the value input in the FEM, the comparison shows that the so-called case 2 gives results closer to the
reference. A variation of  is the most critical situation, this makes sense since the areas used for the correction are aside of
the specimen, thus they are more effective in the correction of the angle . Figure 23.5 better shows the effect of the single
out-of-plane components in case 2. The effect of the translation z is easily solved, as the effect of the rotation , where in
particular the standard deviation is strongly reduced by compensating. On the other hand, while a variation of the angle 
is less effective in terms of means, the standard deviation still remains rather large even after the correction, for the reason
explained above. When all the three, motions are considered together, basically as sum of the effects can be noticed in the
standard deviation of the results.

23.5 Experimental Analysis


®
An experiment was performed and the used set-up is shown in Fig. 23.6. An electromechanical tensile bench Zwick Z050
®
with a wedge clamping system has been used and images were recorded using four CMOS Pixelink B371F cameras. Here
a second camera was used to apply stereo DIC, in order to have reference values for the specimen. The other two cameras
were mounted so that they could frame also the clamps. On these, circular markers have been placed in order to evaluate
their movement during the test. The experimental set-up and an example of the take images is resumed in Fig. 23.6.
186 M. Badaloni et al.

Fig. 23.6 Experimental set-up applied (a) and two images taken from the central camera (b) and one of the side cameras (c)

Fig. 23.7 The movement of the a b 0


markers in one cycle is shown in def
(a) and in (b) the value of the z ref -0.01

for every cycle is reported 50 -0.02

40 -0.03

30 -0.04

D z [mm]
Y [mm]

20 -0.05

10 -0.06

0 -0.07

-10 -0.08
50
40 -0.09
30 30
20 20
10 -0.1
10 X [mm] 0 5 10 15 20 25 30 35 40 45 50
Z [mm] Cycles

Table 23.3 Identification Experimental values Corrected values


performed on the experimental
E (MPa)  E (MPa) 
test on the raw results and with
the correction method applied Mean value 66111.84 0.3643 65839.64 0.3465
Standard deviation 2553.56 0.0301 2124.50 0.0258

In the experiment, thanks to the stereo set-up, the movement of the markers in Fig. 23.6c was analysed to evaluate the
movement of the clamps during a cyclic test in the elastic regime. As visible in Fig. 23.7a, the markers move in plane,
following the theoretical behaviour of a uni-axial tensile test, but also out-of-plane. Figure 23.7b shows the average out-of-
plane motions evaluated in the experiment for all the performed cycles.
We assume that the out-of-plane movements of the specimen are the same as that of the clamps. In Table 23.3 the identified
values, with or without correction, are listed.
Looking at the results, the correction still works properly along the 50 performed repetitions along the cyclical test. In
particular, it seems that in the actual test a strong component for the gamma angle is present, since a big standard deviation
is still evident after the correction. Further analysis may indeed solve that issue by evaluating precisely the values of z, 
and  affecting the experimental set-up.

23.6 Conclusions

The well-known problem of out-of-plane motions in DIC is treated in the present contribution. Among the different
components of the out-of-planes, here the component that varies during a test is considered. By changing the amount of out-
of-plane movements, the effect on the identification values such as the Young’s modulus and the Poisson’s ratio are studied
thanks to a simulating routine. A correction method is considered to compensate the out-of-plane motions by attaching a
frame on the specimen. It is a modified version of a correction method presents in the literature. It can be seen that the
method is capable to correct the results which become closer to the reference values of E D 65 GPa and  D 0.33.
23 Out-of-Plane Motion Evaluation and Correction in 2D DIC 187

A sensitivity analysis has been performed in order to evaluate the optimal size for the ROC. A compromise has to be
found between having enough data to analyse and enough data for the compensation. Amongst the three considered cases,
the second one is the one that better performs in terms of means and standard deviations. The effect of the single components
for the out-of-plane motion is shown by performing a numerical analysis, evidencing that the actual correction is more
capable to handle variations of  angle more than  one. Afterwards, an experimental test was carried out, using a dedicate
set-up, to evaluate the motions and rotations of a specimen during a real uni-axial tensile test on an aluminium 6061 T6
rectangular specimen. The performed experiment confirms the presence of out-of-plane motions.

References

1. H. Chalal, S. Avril, F. Pierron, F. Meraghni, Experimental identification of a nonlinear model for composites using the grid technique coupled
to the virtual fields method. Compos. Appl. Sci. 37(2), 315–325 (2006)
2. J.-L. Piro, M. Grédiac, Producing and transferring low-spatial-frequency grids of measuring displacement fields with moiré and grid methods.
Exp. Tech. 28, 23–26 (2004)
3. P.M. Boone, A.G. Vinckier, R.M. Denys, W.M. Sys, E.N. Deleu, Application of specimen-grid moiré techniques in large scale steel testing.
Opt. Eng. 21(4), 615–625 (1982)
4. M. Sjödahl, Recent advances in electronic speckle photography. Opt. Lasers Eng. 29, 125–144 (1998)
5. M.A. Sutton, J.-J. Orteu, H.W. Schreier, Image Correlation for Shape, Motion and Deformation Measurements (Springer, New York, 2009)
6. S. Avril, M. Bonnet, A.-S. Bretelle, M. Grédiac, F. Hild, P. Ienny, F. Latourte, D. Lemosse, S. Pagano, E. Pagnacco, F. Pierron, Overview of
identification methods of mechanical parameters based on full-field measurements. Exp. Mech. 38, 381–402 (2008)
7. M. Grédiac, F. Pierron, S. Avril, E. Toussaint, The virtual fields method for extracting constitutive parameters from full-field measurements: a
review. Strain 42, 233–253 (2006)
8. F. Pierron, M. Grédiac, The Virtual Fields Method (Springer, New York, 2012)
9. S. Cooreman, D. Lecompte, H. Sol, J. Vantomme, D. Debruyne, Identification of mechanical material behavior through inverse modeling and
fDICg. Exp. Mech. 48(4), 421–433 (2008)
10. D. Lecompte, S. Cooreman, S. Coppieters, J. Vantomme, H. Sol, D. Debruyne, Parameter identification for anisotropic plasticity model using
digital image correlation: Comparison between uni-axial and bi-axial tensile testing. Eur. J. Comput. Mech. 18, 393–418 (2009)
11. M. Rossi, F. Pierron, On the use of simulated experiments in designing test for material characterization from full-field measurement. Int. J.
Solid. Struct. 49, 420–435 (2012)
12. M. Bornert, F. Brémand, P. Doumalin, J.-C. Dupré, M. Fazzini, M. Grédiac, F. Hild, S. Mistou, J. Molimard, J.-J. Orteu, L. Robert, Y. Surrel, P.
Vacher, B. Wattrisse, Assessment of digital image correlation measurement errors: methodology and results. Exp. Mech. 49, 353–370 (2009)
13. P. Lava, S. Cooreman, S. Coppieters, M. De Strycker, D. Debruyne, Assessment of measuring errors in DIC using deformation fields generated
by plastic FEA. Opt. Lasers Eng. 47, 747–753 (2009)
14. P. Lava, S. Cooreman, D. Debruyne, Study of systematic errors in strain fields obtained via DIC using heterogeneous deformation generated
by plastic FEA. Opt. Lasers Eng. 48, 457–468 (2010)
15. Y. Wang, P. Lava, S. Coppieters, P. De Strycker, P. Van Houtte, D. Debruyne, Investigation of the uncertainty of DIC under heterogeneous
strain states with numerical test. Strain 48(6), 453–462 (2012)
16. M. Rossi, P. Lava, F. Pierron, D. Debruyne, M. Sasso, Effect of DIC spatial resolution, noise and interpolation error on identification results
with the VFM. Strain 51(3), 206–222 (2015)
17. M. Rossi, M. Badaloni, P. Lava, D. Debruyne, G. Chiappini, M. Sasso, Advanced test simulator to reproduce experiments at small and large
deformations. Conf. Proc. Soc. Exp. Mech. Ser. 3, 27–33 (2014)
18. M. Badaloni, M. Rossi, G. Chiappini, P. Lava, D. Debruyne, D, Impact of experimental uncertainties on the identification of mechanical
material properties using DIC. Exp.Mech. (2015) DOI: 10.1007/s11340-015-0039-8
19. Q. Ma, S. Ma, Experimental investigation of the systematic error on photomechanic methods induced by camera self-heating. Opt. Soc. Am.
21(6), 7686–7698 (2013)
20. A. Hijazi, A. Friedl, C.J. Kahler, Influence of camera’s optical axis non-perpendicularity on measurement accuracy of two-dimensional digital
image correlation. Jourdan J. Mech. Ind. Eng. 4, 373–382 (2011)
21. P. Lava, S. Coppieters, Y. Wang, P. Van Houtte, D. Debruyne, Error estimation in measuring strain fields with DIC on planar sheet metal
specimens with a non-perpendicular camera alignment. Opt. Lasers Eng. 49, 57–65 (2012)
22. B. Pan, L. Yu, D. Wu, High-accuracy 2D digital image correlation measurements using low-cost imaging lenses: implementation of a
generalized compensation method. Meas. Sci. Technol. 25(2), 025001 (2014)
23. M.A. Sutton, J.H. Yan, V. Tiwari, H.W. Schreier, J.-J. Orteu, The effect of out-of-plane motion on 2D and 3D digital image correlation
measurements. Opt. Lasers Eng. 46, 746–757 (2008)

You might also like