Journal of Physics: Conference

Series

PAPER • OPEN ACCESS You may also like

- The Effect of Manufacturing on the Press
Predict the Plastic Deformation of Perforated fit Insertion Force
D I Maxim, L Marsavina and L Rusu
Sheet Metals using Image-based Finite Element - Research Progress of Springback in Multi-
Point Forming of Sheet Metal
Modeling Xinqi Li and Shicheng Hu

- Study on the Anti-friction properties of

To cite this article: Lin Wang 2021 J. Phys.: Conf. Ser. 2012 012062 Chemically Etched surface texture and its
synergistic Anti-friction properties with
nano-lubricating oil in sheet metal deep
drawing process
Tangjie Mei, Dongliang Zhang and Qi
Zhang
View the article online for updates and enhancements.

This content was downloaded from IP address 82.32.17.166 on 27/06/2024 at 10:35

ICMMAP 2021 IOP Publishing
Journal of Physics: Conference Series 2012 (2021) 012062 doi:10.1088/1742-6596/2012/1/012062

Predict the Plastic Deformation of Perforated Sheet Metals

using Image-based Finite Element Modeling

Lin Wang
Department of Civil Engineering, Hefei University, Anhui, 230061, China
Corresponding author’s email: [email protected]

Abstract. In this work, numerical prediction of the plastic deformation behaviors of perforated
sheet metals, specifically AISI 1018 low carbon steel, C260 Cu-Zn brass, under uniaxial tension
was carried out. An image-based mesh generation technique was applied to better represent and
discretize the perforated geometry of sheet metal specimens. Digital image correlation measured
displacement boundary conditions were also implemented into the finite element modeling. The
predicted results of perforated sheet metal under tension as nominal stress-strain curves and
localized plastic strain distribution agree well with the experimental observations. Possible error
sources and uncertainty of this image-based finite element modeling were also discussed.

1. Introduction
Perforated sheet metal is extensively used in industries, like aircraft, automobiles, space vehicles, etc.,
because of its high strength-to-weight ratio and excellent energy absorption capability. For an effective
application, it is essentially necessary to thoroughly understand its mechanical properties under the
influence of such perforations. The equivalent continuum method is traditionally used to analyze the
deformation of the perforated sheets. Based on this method, O’Donnell [1] calculated the apparent elastic
constants of perforated sheets and proposed a criterion of stress limit for structural design. ASME
presented a similar procedure to determine the peak stress multipliers for perforated plates [2].
Nowadays increasing researches have focused on plastic behavior to better apply the perforated sheet
metals. Chen [3], Baik [4], Lee [5] analyzed the plastic deformation behavior of perforated sheets with
circle perforations in a triangular pattern and proposed yield criteria based on the equivalent continuum
concept. Geltmacher [6] numerically examined the flow localization behavior of metal sheet containing
a pair of through-thickness circle perforations under uniaxial and biaxial tension. Reinhardt [9]
developed a fourth-order anisotropic yield criterion to predict the plastic deformation of perforated sheet
metals under biaxial loading. Jia [8, 9] developed numerical methods to predict the strain localization
of metal sheets with randomly distributed circle perforation under uniaxial tension, resulting in good
agreement with experimental results. Duan [10] studied the local deformation of sheet metal with
random perforations from experimental and numerical points, which is significantly affected by the
loading conditions (uniaxial and biaxial tension) and the perforation pattern. Monchiet [11] developed
an anisotropic yield criterion for porous ductile material considering material anisotropy and perforation
shapes. Khatam [12] numerically investigated the plastic yield initiation and strain localization in
perforated sheets in plane stress condition.
Numerical analysis, such as finite element (FE) modeling, was increasingly used for mechanical
investigation of perforated sheet metals as mentioned above. The numerical simulation, if adequately
formulated and accurately realized can fully characterize the mechanical response of the perforated sheet
metals. The accuracy of the FE modeling about the mechanical behavior of perforated sheet metals is
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
ICMMAP 2021 IOP Publishing
Journal of Physics: Conference Series 2012 (2021) 012062 doi:10.1088/1742-6596/2012/1/012062

strongly dependent on material anisotropy, strain hardening, and loading conditions [6]. Due to the cold-
rolling process, the sheet metal usually behaves in an anisotropic manner. Anisotropic yield criterion is
thus suggested to use in FE analysis, and proven to be able to predict the onset of localized necking of
perforated sheet [6], while the von Mises isotropic yield criterion was applied in refs [3, 4, 8-10, 12]. In
turns of the strain hardening behaviors of sheet metals, either the empirical equations [6, 9, 10, 12], or
perfect plastic assumption [9, 12] were adopted in numerical analysis, which might diverge from actual
material behaviors. Homogeneous displacement boundary conditions were usually assumed in
mechanical testing but were found different from some experimental observations [8, 9].
Besides material behaviors, the effectiveness of numerical prediction also relies on the accurate
geometric characterization of perforated sheet metals, as the plastic behavior is a highly localized
phenomenon influenced by the local geometric disorder [6]. For refs [6, 8-10], the FE models of
perforated sheet metals were generated based on perfect perforation patterns. Manufacturing introduced
variations including circle centroids, size, and shape were completely excluded, which may greatly
affect the strain localization [10, 11]. The image-based meshing technique presents a realistic and
accurate process to discretize the perforated sheet metal based on specimen images. This image-based
finite element modeling technique is proven to be able to assess the mechanical behaviors of materials
and structures fully considering their microarchitectures [13-15].
This work presents a mechanical characterization of perforated sheet metals under uniaxial tension
using the image-based FEM method. Specimens with four perforation patterns were first fabricated from
anisotropic sheet metals (AISI 1018 low carbon steel, C260 Cu-Zn brass) and mechanically tested. The
technique of image-based FE modeling was applied to study the plastic deformation of perforated sheet
metals. Geometric characteristics of perforated sheet metals captured by digital images were discretized
into FE models using an image-based mesh generation technique [16]. The effects of meshing methods,
yield criteria, and displacement boundary conditions were discussed. The numerical prediction including
material strength, strain localization was compared with experimental results. Possible uncertainties and
enhancements of this image-based finite element modeling were also discussed.

2. Material and Method

2.1.Sample Preparation
Rectangular blank specimens were sheared from rolled thin sheet metals (C260 Cu-Zn brass, AISI 1018
low carbon steel) with a width of about 25.4mm and a total length of 200mm. The axial direction of
tensile coupons was aligned with the rolling direction of sheet metals. Four perforation patterns within
the central rectangular area of 34.9 mm×25.4 mm were designed, named as regular square array (RGS),
regular triangular array (RGT), random square array (RDS) and random triangular array (RDT) with
distorted middle column perforation. The diameter of circles d equals 2.5mm. The circles located at least
3.05 mm and 2.7 mm, larger than one diameter, away from the specimen edges of RGS and RDS
perforation patterns, respectively. The spacing between columns of RGS and RGT l =5.6mm. The
distance between center lines of perforations 2R is 5.6mm for RGS and respectively 5mm for RGT. The
nominal area porosity of perforation patterns ρa within the interested area (34.9mm×25.4mm) is
calculated as 11.07%. For each of the four patterns of three materials, at least three identical samples
were prepared. Samples were designated using the initial of materials (B: brass and S: steel), following
the perforation patterns (RGS, RDS, RGT, RDT) and the serial number of tested samples with the same
material and perforation pattern (_1, _2, etc). Digital images (1600×1200 pixels) of the perforated region
were recorded by Keyence VHX-500 digital microscope.

2.2.Image-Based Finite Element Modeling

An image-based meshing was applied based on specimen images of resolution 1600×1200 pixels. The
unstructured mesh generator DistMesh2D based on the Delaunay triangulation and force-equilibrium
algorithms [16] were employed to perform the element discretization. An essential implementation of
DistMesh2D is to represent the interested geometry using a signed distance function 𝜙 𝑥, 𝑦 . In

2
ICMMAP 2021 IOP Publishing
Journal of Physics: Conference Series 2012 (2021) 012062 doi:10.1088/1742-6596/2012/1/012062

mathematics, the signed distance function of a set Ω in a metric space determines the closest distance of
a given point x from the boundary of Ω, with a negative sign if x is in Ω. The signed distance function
can be mathematically defined using one or more explicit functions for simple geometries. The function
𝜙 𝑥, 𝑦 𝑥 𝑦 1, for instance, is the signed distance function of a unit circle.
(a) (b)

(c) (d)

Figure 1. Perforation patterns of tensile samples as (a) RGS, (b) RDS, (c) RGT and (d) RDT arrays.

Images are special cases of implicit geometry definitions since the boundaries of objects in the image
are not available in an explicit form. A natural approach for image-based meshing is to identify the
interested objects from the captured image and form a signed distance function implicitly with a level
set to represent the boundaries. An algorithm of Active Contour without Edge [17] was applied to
segment the interested object from the captured image which results in a new binary image with a value
1 or 0 to represent the pixel if inside or outside the object domain. The object boundaries in the binary
image were traced and the interface evolution of closed boundaries or the unsigned distance function
was then calculated by a numerical technique of fast marching method [18]. By comparing this interface
evolution with the segmented binary image, the signed distance function ( x, y) was obtained.
Mesh size function h(x, y) specifying the size of generated elements (the length of element edges) is
another key implementation for image-based meshing. It is necessary for the FE model to have dense
elements to accurately capture the local mechanical phenomenon. A good mesh size function should be
able to resolve the geometry features, including curved boundaries and narrow regions, and also
maintain a proper gradient of mesh sizes among neighboring elements. Definition of mesh size function
was detailed by Persson [19]. The triangular mesh generation is carried on by inputting the signed
distance function 𝜙 𝑥, 𝑦 and mesh size function ℎ 𝑥, 𝑦 into DistMesh2D once ready. The area porosity
of discretized FE model ρa was calculated and compared with the result based on the sample image. Re-
meshing was possible to maintain the same area porosity of the individual sample by adding a small
constant to the signed distance function. An example of image-based meshing of BRDS_3 from the
captured image to mesh generation is shown in figure 2.

3
ICMMAP 2021 IOP Publishing
Journal of Physics: Conference Series 2012 (2021) 012062 doi:10.1088/1742-6596/2012/1/012062

(a) (b)

(c) (d)

(e)

Figure 2. Image-based meshing procedures of selected sample SRDS_2: (a) sample image (the area
within white rectangle is interested), (b) object boundaries traced from image segmentation, (c)
signed distance function, (d) mesh size function, and (e) mesh generation.

Nonlinear numerical analysis of all 24 perforated plates under uniaxial tension using a two-
dimensional plane stress model was performed using commercial FE code ANSYS v12.1. Besides the
generated mesh by the aforementioned image-based meshing technique, the meshing based on the four
perforation designs was also input into ANSYS. Four-node structure element PLANE182 (triangular
degeneration is allowable) was used with full integration and a large strain effect. The quadratic Hill’s
flow function [20] and multi-linear isotropic strain hardening represented by more than 20 discrete data
points were used to characterize rolled sheet metals. The true stress-strain curves of sheet metals along
the rolling direction were obtained from standard uniaxial tension tests. The post-necking strain
hardening behaviors were then calibrated by a local correction factor method [21] and shown in figure
3. The plastic strain ratios R0 brass and steel were found to be 0.77, and 1.55, respectively using digital
image correlation-based strain measurement. The plastic strain ratios R90 calibrated as 1.04 and 1.23
respectively in [22].
The uniaxial tension boundary conditions were applied as follows: at the left edge x=0, ux=0 at all
nodes; at the right edge x=34.9mm, uniform axial displacement ux=d at all nodes with constant increment.
The numerical simulation was terminated once the maximum effective strain of the element becomes

4
ICMMAP 2021 IOP Publishing
Journal of Physics: Conference Series 2012 (2021) 012062 doi:10.1088/1742-6596/2012/1/012062

larger than the input strain hardening behavior. No failure model was considered since the material
response of perforated plates far beyond the peak load was not concerned in this work.

Figure 3. The input stress-strain relation of sheet metals.

2.3.Experiment Procedures
To validate the image-based FE modeling, a set of experimental tests was carried out. The tensile tests
of perforated plates were conducted using a Model 5582 Instron Universal Materials Testing System
(MTS) at a constant cross-head speed of 0.6mm/min for brass, and 1mm/min for steel. Each sample was
stretched until the post-necking force level fell down to 30% of its peak load. Before testing, one surface
from the interested perforated area of each tensile coupon was decorated with fine black-and-white paint
spackles to facilitate the digital image correlation (DIC) based surface strain measurement. Digital
images of the tension coupon with the paint speckles were captured during tensile testing by a CCD
camera with an acquisition rate of around 2 frame/sec. The %elongation of the interested perforated
region over the centered gage length 34.9mm was measured by non-contact DIC-based video
extensometer. The obtained %elongation was then synchronized in time with the applied axial load
recorded by MTS. The nominal engineering stress was defined by dividing the applied axial load with
the initial cross-section area of test coupon assuming with no perforation, and thus the nominal
engineering stress-strain relations were obtained.

3. Results and Discussion

3.1.Experimental Results
The typical force-displacement relation of perforated sheet metal is shown in figure 4, also included are
the captured images at different displacement loadings. The solid and open circles indicate the peak load
and crack initiation points where the crack is detected from captured images. The failure path of samples
is shown in figures 5 and 6 on un-deformed designs. It is assumed that the failure path lies upon line
segments that connect the center of circles, and the first and last line segments of the path are vertical
lines that connect the center of a circle to the edge of the sample [8, 9]. The area porosity of the failure
path is defined as 𝜌 𝑛 ∙ 𝐷/𝐿, where n is the number of circles on the fracture path, D is the diameter
of circle and L is the total length of fracture length. It is straightforward to understand that fracture
occurs on the path of higher area porosity. It is clear to see that for specimens of regular square or
triangular perforation, the failure happens (or would happen) in a single vertical column (for triangular
perforated specimens, the tests terminated before completely fracture). In this way, the area porosity of
the failure path is calculated as 𝜌 =0.38. The failure path of randomized square array perforation is
identical for all specimens except two steel samples. The area porosity is 0.48 for all failure paths of

5
ICMMAP 2021 IOP Publishing
Journal of Physics: Conference Series 2012 (2021) 012062 doi:10.1088/1742-6596/2012/1/012062

randomized square array perforation, making the randomized middle column predominant for ductile
fracture compared with the single vertical column.

Figure 4. Force displacement relation of specimen SRDS_2 with captured images during test.

(a) (b)

(c) (d)

Figure 5. The failure path of perforated brass samples with (a) regular square array, (b) randomized
regular square array, (c) regular triangular array, and (d) randomized regular triangular array. Different
colors and line types represent different samples.

6
ICMMAP 2021 IOP Publishing
Journal of Physics: Conference Series 2012 (2021) 012062 doi:10.1088/1742-6596/2012/1/012062

(a) (b)

(c) (d)

Figure 6. The failure path of perforated steel samples with (a) regular square array, (b) randomized
regular square array, (c) regular triangular array, and (d) randomized regular triangular array. Different
colors and line types represent different samples.

3.2.Numerical Results

3.2.1.Effect of mesh generation methods. A mesh sensitivity study was conducted to analyze the
influence of meshing methods on mechanical response predicted by FE modeling. Figure 7.a compares
the nominal stress-strain relations of sample SRDS_2 using different mesh generation methods. These
methods are image-based meshing using images of perforated sheet metals and perforation design, and
ANSYS-based meshing using perforation design. The numerical results from FE modeling using
specimen image gives the best approximation of experimental results. The numerical results from FE
modeling using either image-based or ANSYS-based mesh generation based on pattern design are
obviously overestimated (figure 7.a). The predicted ultimate strength from these three modelings are
0.6%, 2.6%, and 2.8% higher than corresponding experimental measurements, respectively, at the strain
level of 5.1%. The cross-section area porosity along x (loading) direction ρax of sample SRDS_2 (figure
7.b) indicates to be higher than designed porosity. This is the major reason that FE prediction using
design-based meshing tends to overestimate the ultimate strength. The predicted strain localization
based on different mesh generations shows similar features in locations and adjacent connections,
though small variations in strain magnitude exist among FE analyses.

3.2.2.Effect of the mesh size. To examine the element dependency, three different image-based mesh
generations of sample SRDS_2 were generated based on the digital image with 12776, 20136, 36188
elements, respectively. The nominal stress-strain relations predicted by FE analysis based on three mesh
generations from 2D uniaxial tension simulations were compared in figure 8.a. The results based on fine
mesh discretization give a slightly overestimation of nominal stress-strain relation, while the results
calculated using coarse and medium meshes predicted even higher stress level. The error of predicted
ultimate strength in comparison to experimental ultimate strength is 0.4%, 0.8%, and respectively, 1.3%
higher. The accuracy of FE modeling depends on both the element size and the material properties. The
fine mesh generation was therefore selected, so the combined effect of element size and material
properties is consistent. The image-based mesh generations of all perforated plates were of similar node

7
ICMMAP 2021 IOP Publishing
Journal of Physics: Conference Series 2012 (2021) 012062 doi:10.1088/1742-6596/2012/1/012062

and element numbers. The smallest size of the triangular meshes (the edge of a triangle) is around
0.15mm and each round circle is discretized by at least 50 elements. In addition, three-dimensional (3D)
finite element modeling was performed by offsetting the 2D image-based mesh with 12 increments (12
elements through-thickness) to generate a full representation of samples BRDT_3 and SRDS_3. The
nominal engineering stress-strain relations obtained from 3D FE modeling did not significantly differ
from those of the plane stress models within the interested strain range, see figure 8.b. All of the reported
results, therefore, are from the plane stress cases.
(a) (b)

Figure 7. (a) nominal stress-strain relations of sample SRDS_2 from FE modeling using different
methods, (b) cross-section area porosity along x (loading) direction.

(a) (b)

Figure 8. Nominal stress-strain relations affected by (a) element size of image-based meshing of
sample SRDS_2, (b) 2D and 3D models.

3.2.3.Effect of boundary conditions. One problem during experiments was that as strain localization
progressed, the specimen might rotate [8, 9]. However, the applied uniaxial tension BCs in FE modeling
excluded this rotation effect. To quantify this effect, the actual edge displacement of perforated plates
BRDT_3 on the left and right edges of interested area with gage length 34.9mm were measured via DIC
between image pairs #63 and #503. The measurements indicate that the bottom edge was stretched more
than the top edge (figure 9.a). These actual BCs were then applied in FE modeling node-by-node using
image-based meshing. Only the axial elongation was considered since the relative displacement in the
vertical direction over 34.9mm gage length is about 3% or less of axial deformation. The predicted strain
localization adding uniaxial tension BCs and actual BCs displayed in figures 9.b and c, and compared
with the #503 images (the corresponding image at star symbol). This comparison shows that the
numerical simulation with the actual BCs could give relatively consistent prediction of strain
localization with the experimental results (see the red rectangles in figure 9.b). The global stress-strain
response from FEA using actual BCs matches the experimental results very well, while the results based
on uniaxial tension BCs tends to slightly underestimate the material strength when strain is larger than

8
ICMMAP 2021 IOP Publishing
Journal of Physics: Conference Series 2012 (2021) 012062 doi:10.1088/1742-6596/2012/1/012062

2.5%. The predicted ultimate strength by applying these actual BCs is 0.2% higher than the experiment,
and 0.6% lower if using uniaxial tension BCs at a strain of 3.5%.
(a) (b)

(c) (d)

Figure 9. Influence of boundary conditions: (a) the actual boundary condition measured by DIC, (b)
strain distribution εx of sample BRDT_3 at peak load point by applying (b) uniaxial tension and (c) the
actual BCs, (d) the corresponding specimen image with fracture path.

3.2.4.Effect of yield criterion. The predicted stress-strain relations of sample SRDS_2 using von Mises
yield criterion and quadratic form of Hill yield criterion were compared in figure.10. The stress-strain
relations of steel calibrated by the local correction factor method [21] based on corresponding yield
criteria were set as material inputs. The results indicate that Hill’s model gives a slight overestimation
of nominal stress-strain relation compared with the experiment, while the von Mises criterion tends to
underestimate the mechanical response. The variation in predicted ultimate strength compared with the
experimental result are respectively 0.6% and -0.9% from FE analysis using Hill and von Mises criteria.
The strain localization is similar in location and connection, though the results using the von Mises
criterion gives slightly higher strain magnitude. The Hill yield criterion is found slightly better than the
von Mises criterion to describe the mechanical behavior of perforated sheet metal due to its material and
geometric anisotropy.

Figure 10. Numerical simulation of SRDS_2 using Hill and von Mises yield criteria.

9
ICMMAP 2021 IOP Publishing
Journal of Physics: Conference Series 2012 (2021) 012062 doi:10.1088/1742-6596/2012/1/012062

3.3.Plastic Deformation
Figure 11 compares the experimental and numerical nominal stress-strain relations of all tested samples
(left column), the predicted strain localization of selected 8 (one sample per material and perforation
pattern) at the strain level of star symbol in nominal stress-strain relations (middle column), and the
corresponding captured images during the experiment (right column). It is generally found the predicted
nominal stress-strain relations are in good agreement with the experimental results. The failure path in
the perforated sheet metals is the strain localization of adjacent circles corresponding to the regions of
high magnitude plastic strain [8, 9]. These regions coincide with the actual fracture path of selected
samples (see the solid line in the right column images). As is evident, the image-based FE modeling is
able to capture the major features of strain localization of perforated sheet, including the locations and
connection between each other, and also effectively predict the global mechanical response of perforated
sheet very well until slightly beyond the peak load.
(a)

(b)

(c)

(d)

Figure 11. Nominal stress-strain relations (left), strain localization εx (middle) and corresponding
deformed sample images (right) at the star symbol of samples. (a) BRGS_1, (b) BRDS_3, (c)
BRGT_1, (d) BRDT_3. The solid lines in right columns are the failure path of selected samples, while
the dash and dot lines are for other samples with same the perforation.

10
ICMMAP 2021 IOP Publishing
Journal of Physics: Conference Series 2012 (2021) 012062 doi:10.1088/1742-6596/2012/1/012062

(e)

(f)

(g)

(h)

Figure 12 (Continued). (e) SRGS_1, (f) SRDS_2, (g) SRGT_1 and (h) SRDT_2.

The predicted ultimate strength of each test of image-based FE modeling agrees well with
experimental results. The relative error in predicted ultimate strength of image-based FE modeling based
on sample images and FE modeling using perforation pattern designs varies from -1.12%~ 2.28% and
respectively 1.28%~5.56% in comparison with experiments. This indicates that the ignorance of actual
microarchitecture variation including location, shape and size of circle perforation would results in a
1%~4% error in the prediction of ultimate strength. Similarly, the predicted strain at ultimate strength,
elastic modulus, and yielding stress also show a linear correlation with experimental results.

3.4.Errors and Uncertainties

The level of agreement between numerical and experimental results validates the effectiveness of the
image-based FE modeling in the mechanical determination of perforated sheet metals considering the
microarchitecture features. The differences between experiment and numerical analysis, for instance in
ultimate strength, are -1.12%~2.28%. This is caused by errors or uncertainties in experimental
measurement and image-based FE modeling. Experimental errors or uncertainties include measurement
of axial load, sample geometry, nominal strain, and manufacture-introduced material damage. In the

11
ICMMAP 2021 IOP Publishing
Journal of Physics: Conference Series 2012 (2021) 012062 doi:10.1088/1742-6596/2012/1/012062

numerical modeling aspect, the errors or uncertainties are the material characterizations of sheet metals,
microarchitecture representation of samples, and simplification of computational models.
Based on the measurement precision in axial load, specimen geometry, the error in nominal stress is
calculated at about 0.4%. The measurement error of axial strain by DIC-based video extensometer was
estimated to be better than ±0.005%, and thus is negligible [23]. The effect of residual stress due to
mechanical manufacture, however, is hard to quantify. For plastically deformable materials, residual
stresses may accelerate or delay the initiation of local yielding, which in turn determines the final strain
localization of samples, especially with regular perforation patterns. It was reported that annealing
treatment is able to relieve or even remove the residual stress [8], but not applied in this work due to
lack of annealing condition information.
Material characterization can greatly affect the accuracy of image-based FE modeling. The strain
hardening behaviors of sheet metals were identified to a large strain level by the local correction factor
method, with an error level of about 0.5% [21]. The anisotropic parameters of Hill’s yield criterion in
ANSYS rely on measurements of plastic strain ratio along rolling direction R0, 45° orientation with
respect to rolling direction R45, and transverse direction R90 of sheet metals. In this work, R0 is directly
measured. R0=R45 was initially assumed, and R90 was calibrated in ref [22]. On the other hand, the
yielding behavior of sheet metals may not follow the quadratic form of Hill’s plastic yield criterion.
Non-quadratic form anisotropic yield function has been proved to be able to accurately describe the
material behavior of aluminum [24-25]. Parameter calibration and program implementation are,
however, needed. A better understanding of the mechanical response of sheet metals is thus necessary.
The microarchitecture discretization of perforated sheet metals using image-based meshing
technique also introduces errors or uncertainties into image-based FE modeling. For each strep of image-
based meshing, like image segmentation, edge detection, calculation of signed distance function and
mesh size function, and mesh generation, partial information of samples in the digital image may be lost.
Inaccurate input of each step would then generate new errors in the output. The area porosity of sample
was guaranteed in mesh generation, the variation of local microarchitecture feature is necessary but
difficult to quantify. Meanwhile, modeling strain localization using the FE method introduces an
artificial length scale into modeling once strain localization occurs which in turn makes the FE solution
mesh dependent [11]. Though the strategy to maintain the similar element size in this work can partially
relieve the mesh dependency within interested certain strain range, advanced measures should be applied
to relief or even eliminate the element size effect [26].

4. Conclusion
In this work, perforated sheet metals of AISI 1018 low carbon steel, C260 Cu-Zn brass were fabricated
with four different perforation patterns and tested under uniaxial tension. An image-based FE modeling
technique was employed to numerically investigate the plastic behavior of perforated sheet metals under
tension. The effects of mesh generation methods, element size, displacement boundary conditions, and
yield criterion on the numerical results were assessed by comparing with the experiment measurement.
The predicted nominal stress-strain relations until slightly beyond the ultimate strength as well as the
strain localization obtained from image-based finite element modeling closely matched experimental
results. The predicted ultimate strength well correlated with the corresponding experimental
measurement. This is remarkable given that the applied image-based FEM technique in this work
considered accurate characterization of material behavior and realistic representation of
microarchitecture of perforated sheet metals. It is of great potential to extend the image-based FE
modeling to explore material behavior of porous materials, multi-phase composites of complicated
microarchitectures.

Acknowledge
The financial support from Hefei University (No. 18-19RC01) is greatly appreciated.

12
ICMMAP 2021 IOP Publishing
Journal of Physics: Conference Series 2012 (2021) 012062 doi:10.1088/1742-6596/2012/1/012062

References
[1] O’Donnell, W.J., Langer, B.F. (1962) Design of perforated sheets. J. Eng. Ind. Trans. 84: 1-13.
[2] ASME boiler and pressure vessel code (1998) Section III, Division I, New York, NY.
[3] Chen, F.K. (1993) Analysis of plastic deformation for sheet metals with circular perforations. J.
Mater. Process. Tech. 37: 75-188.
[4] Baik, S.C., Oh, K.H., Lee, D.N. (1997) Plastic behavior of perforated sheets under biaxial stress state.
Int. J. Mech. Sci. 39: 781-793.
[5] Lee, Y-C., Chen, F-K., 2000. Yield criterion for a perforated sheet with a uniform triangular pattern
of round circles and a low ligament ratio. J. Mater. Process. Technol. 103: 353-361.
[6] Geltmachar, A.B., Koss, D.A., Stout, M.G., Matic, P. (1998) Flow localization in sheet specimens
with pairs of circles. Metall. Mater. Trans. A. 9: 775-780.
[7] Reinhardt, W.D. (2001) Yield criteria for the elastic-plastic design of tubesheets with triangular
penetration pattern. J.Press. Vessel. Technol. 123: 118-123.
[8] Jia, S., Raiser, G.F., Povirk, G.L. (2002) Modeling the effects of circle distribution in perforated
aluminum sheets I: representative unit cells. Int. J. Solids Struct. 39: 2517-2532.
[9] Jia, S., Povirk, G.L. (2002) Modeling the effects of circle distribution in perforated aluminum sheets
II: minimum strength failure paths. Int. J. Solids Struct. 39: 2533-2545.
[10] Duan, X., Weck, A., Wilkinson, D.S., Metzger, D.R. (2008) Failure modes of perforated material
under finite deformation. J.Press. Vessel. Technol. 30: 1-6.
[11] Monchiet, V., Cazacu, O., Charkaluk, E., Kondo, D. (2008) Macroscopic yield criteria for plastic
anisotropic materials containing spheroidal voids. Int. J. Plast. 24: 1158-1189.
[12] Khatam, H., Pindera, M. (2011) Plastic deformation modes in perforated sheets and their relation
to yield and limit surfaces. Int. J. Plast. 27: 1537-1559.
[13] Ren, Y, Yang, Z., Sharma, R., Zhang, C., Withers, P.J. (2015) Two-dimensional X-ray CT image
based meso-scale fracture modelling of concrete. Eng. Fract. Mech. 133: 24–39.
[14] Schneidera, Y., Bertrama, A., Böhlke, T., Hartig, C. (2010) Plastic deformation behaviour of Fe-
Cu composites predicted by 3D finite element simulations. Comput. Mater. Sci. 48(3): 456–
465.
[15] Christen, D., Webster, D.J. (2010) Multiscale modeling and nonlinear finite element analysis as
clinical tools for the assessment of fracture risk. Philosophical Transactions A 368: 2653-2668.
[16] Persson, PO., Strang, G., 2004. A Simple Mesh Generator in Matlab. SIAM Rev. 46(2): 329-345.
[17] Chan, T.F., Vese, L.A. (2001) Active countours without edges. Image. Process. 10(2): 266-277.
[18] Hassouna, M.S., Farag, A. (2007) A Multistencils fast marching methods: a highly accurate solution
to the Eikonal equaltion on Cartesian domains. Pattern. Anal. Mach. Intell. 29(9): 1563-1574.
[19] Persson, PO. (2006) Mesh size functions for implicit geometries and PDE-based gradient limiting.
Eng. Comput. 22: 95-109.
[20] Hill, R. (1979) Theoretical plasticity of textured aggregates. Math. Proc. Cambridge 85: 179-191.
[21] Wang, L., Tong, W. (2015) Identification of post-necking strain hardening behavior of thin sheet
metals from image-based surface strain data in uniaxial tension. Int. J. Solids. Struct. 75: 12-
31.
[22] Wang, L. (2021) Hybrid experimental-numerical investigation on plastic deformation and ductile
failure of anisotropic sheet metals. (submitted)
[23] Tong, W. (2011) Subpixel image registration with reduced bias. Opt. Lett. 36(5): 763-765.
[24] Barlat, F., Brem, J.C., Yoon, J.W., Chung, K., Dick, R.E., Choi, S-H., Chu, E. (2003) Plane stress
function for aluminum alloy sheets—part I: theory. Int. J. Plast. 19: 1297-1319.
[25] Korkolis, Y.P., Kyriakides, S. (2008). Inflation and burst of anisotropic aluminum tubes, part II: an
advanced yield function including deformation-induced anisotropy. Int. J. Plast. 24: 1625-
1637.
[26] Brunet, M., Morestin, F., Walter-Leberre, H. (2005) Failure analysis of anisotropic sheet-metals
using a non-local plastic damage model. J. Mater. Process. Technol. 170(1): 457-470.

13