You are on page 1of 10

Journal of Materials Processing Technology 213 (2013) 961970

Contents lists available at SciVerse ScienceDirect

Journal of Materials Processing Technology


journal homepage: www.elsevier.com/locate/jmatprotec

Numerical verication of a biaxial tensile test method using a cruciform specimen


Yasuhiro Hanabusa a , Hideo Takizawa b , Toshihiko Kuwabara c,
Development Department, Universal Can Corporation, 1500 Suganuma, Oyama-cho, Sunto-gun, Shizuoka 410-1392, Japan Central Research Institute, Mitsubishi Materials Corporation, 1975-2 Shimoishitokami, Kitamoto-shi, Saitama 364-0022, Japan c Division of Advanced Mechanical Systems Engineering, Institute of Engineering, Tokyo University of Agriculture and Technology, 2-24-16 Naka-cho, Koganei-shi, Tokyo 184-8588, Japan
b a

a r t i c l e

i n f o

a b s t r a c t
A method of evaluating stress measurement errors in biaxial tensile tests using a cruciform specimen is proposed. The cruciform specimen is assumed to be fabricated from a section of uniformly thick at sheet metal via laser or water-jet cutting and to have a number of parallel slits cut into each of the four arms. Using nite element analyses with the von Mises yield criterion, the optimum geometry of the cruciform specimen and the optimum strain measurement position necessary to minimize the stress measurement error are determined. Additionally, an experimental validation of the FEA is performed using a sheet material that was experimentally conrmed to be nearly isotropic. The following conclusions are drawn: (i) the thickness of the test material should be less than 0.08B (B: side length of the gauge area of the cruciform specimen); (ii) the geometric parameters for the cruciform specimen should be N 7, L B, ws 0.01B, and 0.0034 R/B 0.1 (N: number of slits, L: length of slits cut into the arms, ws : slit width, and R: corner radius at the junction of the arms to the gauge area); and (iii) the strain components in the gauge area should be measured on the centerline of the specimen parallel to the maximum force direction at a distance of approximately 0.35B from the center of the specimen. The stress measurement error is estimated to be less than 2% when the optimum conditions above are satised. 2012 Elsevier B.V. All rights reserved.

Article history: Received 6 October 2012 Received in revised form 18 December 2012 Accepted 19 December 2012 Available online 28 December 2012 Keywords: Sheet metal Finite element method Biaxial tension Isotropy Yield function

1. Introduction Highly accurate material constitutive equations are required in order to perform accurate sheet metal forming simulations using nite element analysis (FEA), and material testing under multiaxial stresses is indispensable for establishing these equations. One of the typical methods for biaxial stress testing of sheet metals utilizes cruciform specimens, as reviewed by Kuwabara (2007) and Hannon and Tiernan (2008). This test method has several advantages, including the ability to measure the elasticplastic deformation behavior of sheet materials for an arbitrary stress ratio and its immunity to the out-of-plane deformations that occur in hydrostatic bulge testing. However, unlike uniaxial tensile testing, the stress distribution in the gauge area of a cruciform specimen is not uniform, which complicates the determination of the biaxial stress components. Therefore, the cruciform specimen geometry and the method used for measuring biaxial stress and strain components are the most important factors for establishing a reliable and accurate biaxial tensile test method for sheet metals.

Corresponding author. E-mail addresses: hanabusa@mmc.co.jp (Y. Hanabusa), takizawa@mmc.co.jp (H. Takizawa), kuwabara@cc.tuat.ac.jp (T. Kuwabara). 0924-0136/$ see front matter 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jmatprotec.2012.12.007

In this study, cruciform specimens are classied into types A and B. Type A has a reduced thickness area (henceforth, referred to as gauge area) for measuring biaxial stress components, while type B has uniform thickness over the whole specimen geometry. Many researchers have investigated biaxial tensile test methods using type A specimens. Pascoe and de Villiers (1967) proposed a cruciform specimen with spherical recesses on both sides of the central region for use in investigating the short-life fatigue of mild and heat-treated steel types under biaxial loading. Shiratori and Ikegami (1968) proposed a cruciform specimen consisting of one cross-shaped sheet sample and eight plates for reinforcing the four arms. This specimen was used to measure the yield loci of a brass sheet that had been prestrained along various loading paths. Hayhurst (1973) developed a biaxial tension creeprupture testing machine and a cruciform specimen with a uniformly thinned central region and slots in the arms. Makinde and Ferron (1988) measured the plane strain and equibiaxial work hardening behavior of aluminum-1050A, (70/30) brass and austenitic stainless steel sheets using a novel and inexpensive biaxial device (jointedarm mechanism) developed by Ferron and Makinde (1988). Boehler et al. (1994) developed a new screw-driven biaxial testing machine that used a specialized off-axis testing device and cruciform specimens with an optimized design to allow almost perfect off-axis biaxial tensile testing of anisotropic sheet materials. Hjelm (1994) successfully measured the yield locus for gray cast iron in the

962

Y. Hanabusa et al. / Journal of Materials Processing Technology 213 (2013) 961970

third quadrant (compressivecompressive) of stress space. Yu et al. (2002) proposed an optimum specimen geometry for measuring limit strains. Green et al. (2004) measured biaxial stress strain curves for a commercial 1145 aluminum alloy sheet deformed up to effective strains of approximately 0.15 along seven different proportional strain paths. The specimen had a sandwich design with the sample sheet bonded by adhesive between two face sheets while leaving the central area exposed on both sides; the specimen geometry was the same as that developed by Makinde et al. (1992). Because the fabrication of type A specimens requires a machining process or sandwich design to reduce the thickness of the gauge area, it is difcult and expensive. Furthermore, the geometrical constraint of the arms on the deformation of the thinned gauge area makes it difcult to accurately determine the biaxial stress components in the gauge area. Type B specimens are easy to fabricate from at sheet metals by laser or water-jet cutting and are, therefore, less expensive. Several authors used type B specimens with no slits cut into the arms. Kreiig and Schindler (1986) determined yield loci of as-received and prestrained (uniaxial and equibiaxial) sheet metal; the proportional elastic limit was used for the denition of yielding. Mller and Phlandt (1996) optimized the cruciform specimen geometry using a FEA and photoelastic tests; the yield point was determined through measurement of the temperature using an infrared thermocouple. However, they pointed out the difculty in dening the effective cross sectional area of the cruciform specimen. Hallfeldt (2002) proposed to optimize an angle between specimen arms depending on the degree of anisotropy of the material. All the above-mentioned authors proposed cruciform specimens with different geometries. However, whatever the geometry of the cruciform specimen is, producing a uniform stress distribution in the gauge area is difcult. This is because the four arms connecting to the gauge area prohibit uniform deformation. As a result, it is difcult to accurately determine the biaxial stress components in the gauge area. Based on the above, from the viewpoints of fabrication cost and stress uniformity in the gauge area, the use of type B cruciform specimens with slits cut into the arms is considered most promising for accurate biaxial tensile testing of sheet metals. Kuwabara et al. (1998) performed biaxial tensile tests of a cold-rolled, ultralow-carbon steel sheet using a cruciform specimen with seven slits cut in each arm. It was found that the measured directions of plastic strain rates were in good agreement with those of the local outward vectors normal to the contours of plastic work constructed in the principal stress space, and that Hills quadratic yield criterion (Hill, 1948) overestimates the ow stresses in the vicinity of equibiaxial tension. Kuwabara et al. (2000) successfully detected the yield vertex and non-normality behavior of the plastic strain rate for ultra-low-carbon steel sheet and aluminum alloy sheet using the abrupt strain path change method proposed by Kuroda and Tvergaard (1999). Kuwabara et al. (2002) performed biaxial tensile experiments of six kinds of steels with different r-values, using the same cruciform specimen as developed by Kuwabara et al. (1998). It was found that the TaylorBishopHill (TBH) model with the full constraints assumption is superior to that with the relaxed constraints assumption in predicting the plastic deformation characteristics of the test materials. Borsutzki et al. (2002) measured a true stressstrain curve for equibiaxial tension of a DC04 steel sheet using a cruciform specimen with six slits in each arm, up to an equivalent plastic strain of 0.09 with the use of a laser extensometer. Naka et al. (2003) investigated the effect of temperature on the yield locus of 5083 aluminum alloy sheet using a cruciform specimen with three slits in each arm. Geiger et al. (2005) recommended cutting six slits into each arm based on FEA results. Kulawinski et al. (2011) applied the partial unloading method to the

measurement of the yield locus of cast TRIP steel using a cruciform specimen with three slits in each arm. However, even though all the above-mentioned authors used type B specimens with slits cut into the arms, stress measurement errors and the optimum specimen geometry were not claried in those studies. Hoferlin et al. (2000) found that the yield loci measured at a 0.2% von Mises equivalent plastic strain for DC06 (cold-rolled, ultra-low-carbon steel) and ZStE 220 BH (bake-hardening steel) were in good agreement with the theoretical predictions based on the TBH model with full constraints using experimentally determined crystallographic textures. They used a type B specimen that had a square metal sheet gauge area with thin metal bars welded to the sides of the gauge area; the geometry of the specimen was optimized using texturebased anisotropic FEA. However, fabrication of such specimens is time consuming and expensive, and the welded zones may impart unfavorable effects on both the uniformity of the stress distribution and the mechanical properties of the gauge area. There have been ongoing efforts to optimize the specimen geometry by quantitatively evaluating stress and strain distributions in the gauge area using FEA. Makinde et al. (1992) optimized the geometry of a cruciform specimen with a circular reduced region and a type B cruciform specimen with seven slots cut into each arm, using a statistical method coupled with a FEA. Demmerle and Boehler (1993) performed a FEA for a type A specimen with slits cut into the arms to estimate the standard deviations of the stresses and strains in the gauge area and to realize shape optimization of biaxial cruciform specimens for isotropic elastic materials. Lin and Ding (1995) performed a FEA for a type A specimen with slits cut into the arms to determine an optimum value of the effective cross sectional area of the gauge area, even though the FEA was performed only for the case of equibiaxial tension. Ikeda and Kuwabara (2002) proposed a planestrain tension specimen with slits cut into the arms and determined the optimum combination of slit number and slit tip hole diameter to minimize the stress measurement error. However, none of these studies discussed the optimum strain measurement position for minimizing stress measurement errors; nor did they quantitatively consider the effects of the geometrical parameters of the cruciform specimens on the stress measurement errors for various stress ratios. The objective of this study is to clarify the optimum strain measurement position and geometrical parameters for the type B specimen geometry proposed by Kuwabara et al. (1998) for minimizing the stress measurement error, on the basis of FEA using the von Mises yield criterion. Additionally, an experimental validation of the FEA is performed using a sheet material that was experimentally conrmed to be nearly isotropic. It should be noted that the biaxial tensile test method using a cruciform specimen proposed in this study has proven to be useful for accurately detecting and modeling the deformation behavior of sheet metals under biaxial tension, and consequently improves the predictive accuracy of the FEA for springback in stretch-bending (Kuwabara et al., 2004), surface deection of an automotive body panel (Moriya et al., 2010), hole expansion of high strength steel sheet with a tensile strength of 590 MPa (Hashimoto et al., 2010) and 780 MPa (Kuwabara et al., 2011), and hydraulic bulge forming of 6000 series aluminum alloy sheet (Yanaga et al., 2012).

2. Evaluation method for stress measurement error In this section, we describe our method for evaluating the accuracy of stress measurements in a biaxial tensile test using a cruciform specimen. FEA is used to determine the optimum geometrical parameters for the cruciform specimen and the optimum position for strain measurement to minimize the stress measurement error.

Y. Hanabusa et al. / Journal of Materials Processing Technology 213 (2013) 961970

963

Fig. 2. Schematic illustration of average stress Fig. 1. Geometry of cruciform specimen.

and local stress

Fig. 1 shows the cruciform specimen geometry investigated in this study. The geometry of the specimen was originally proposed by Kuwabara et al. (1998), and was fabricated from at sheet metal via laser cutting. This specimen consists of a square gauge area (the shaded area in Fig. 1), for which biaxial stress components are determined, and four arms that extend from the gauge area. Parallel slits are cut into each arm to ensure that the stress distribution is kept as uniform as possible in the gauge area, thus minimizing the stress measurement error to the greatest extent possible. The extremity of each arm is held in the grip of a biaxial tensile testing machine. In Fig. 1, B is the width of the arms, and BSx and BSy are the distances between the slit tips along the x- and y-axes (placed along the arm midlines), respectively. It is assumed in this study that BSx = BSy = B. L is the slit length, ws is the slit width, N is the number of slits in each arm, and R is the corner radius at the junction of the arms to the gauge area. The slit tips are semicircular with a radius of ws /2. The initial thickness of the specimen is t0 , which is the same as the as-received sheet sample. In contrast to conventional uniaxial tension specimens, it is difcult to achieve a uniform distribution of biaxial stresses in the gauge area of a cruciform specimen. Therefore, the objective of this study is to determine the optimum strain measurement position and the parameters, L, ws , N and R, which dene the specimen geometry, so that the stress measurement error is minimized. Additionally, the effects of the initial thickness of the specimen t0 and the workhardening exponent n on the stress measurement error are also examined. The present approach to evaluating the stress measurement error is explained using the conceptual illustration shown in Fig. 2. It is assumed that the biaxial strain components in the gauge area are measured at a single point in the gauge area using a sensor such as a strain gauge. In order to determine an accurate stressstrain relationship for a sheet material subjected to biaxial tension, it is crucial to measure L , L ) as accurately as possithe local stress components L ( x y ble, at the position where the local strain components (x , y ) are measured. However, the measurable stress components in the biaxial tenG , G ), determined by dividing sile test are the averages G ( x y the biaxial tensile forces, Fx and Fy , by the respective current crosssectional areas of the gauge area. Consequently, the most important requirement for establishing a highly accurate biaxial stress test method is to make G approach L as closely as possible. It should be

noted that L cannot be measured directly in an actual experiment, but it is possible to estimate the value of L using FEA. In the following section, a method for evaluating the stress measurement errors using FEA is proposed. This method allows users to determine the optimum specimen geometry and the optimum strain measurement position necessary for minimizing the stress measurement error. 2.1. Evaluation of stress measurement error In this section, we dene the stress measurement error es , which is an indicator to show how close G is to L . The relation between L and ( , ) at an integration point in the FEA model follows the x y material constitutive equations used in the FEA. G is calculated using Eq. (1):
G x G y

= =

Fx exp(x ) Bt0 Fy exp(y ) Bt0

(1a) (1b)

in which a uniform deformation in the gauge area and a constant volume condition are assumed. Then, the stress measurement error es is dened as follows: |
G

es

L|

L|

( =

G ij

L )( G ij ij L L kl kl

L) ij

(2)

G = G = G = G = 0 and L = L = where it is assumed that 12 23 31 33 23 31 L = 0 (for the case of plane stress elements). The position at which 33 es becomes a minimum gives the optimum strain measurement position.

2.2. Finite element model and analysis conditions The cruciform specimen geometry shown in Fig. 1 was used as a model specimen for FEA. The clamping area was modeled by a rigid connection of nodes along the edge of the arm. The lateral displacements of the nodes were xed, and the combined tensile forces in the x- and y-directions were applied to the respective connections (see Fig. 5). MSC.Marc2008r1 was used for the analysis. Because of the symmetry of the specimen, only one quarter of the sample was modeled and 2D analyses were performed using plane stress elements, while 3D analyses were also performed using solid elements when evaluating the effect of specimen thickness on es , in which one half

964

Y. Hanabusa et al. / Journal of Materials Processing Technology 213 (2013) 961970

Table 1 Analysis conditions for FEA (underline: standard conditions). Stress ratio: sx : sy Thickness: t0 (mm) Slit length: L (mm) Number of slits: N Slit width: ws (mm) Corner radius: R (mm) 1:0, 4:1, 2:1, 4:3, 1:1 0.6, 1.2, 2.4, 3.6, 4.8 15, 30, 45, innite length (unconstraint along arm edges in transversal direction) 3, 5, 7, 9 0.2, 0.3, 0.5 0.1, 0.5, 1.0, 3.0

Table 2 Work hardening characteristics of material model assumed in FEA (underline: standard condition). n-valuea fs-0 fs-1 fs-2 fs-3
a

0 a 0.0041 0.0038 0.0073 0.0110


n

ca (MPa) 522 505 723 1004

0.209 0.2 0.3 0.4

p) . Swift type ow curve: = c (0 +

volume in the thickness direction of the specimen was modeled. It was assumed that BSx = BSy = B = 30 mm. Table 1 shows the analysis conditions for the FEA. The nominal stress ratios chosen were sx : sy = 1 : 0, 4 : 1, 2 : 1, 4 : 3 and 1 : 1. The effects of the geometrical parameters, t0 , L, ws , N and R, on es were investigated in detail. The underlined values in Table 1 indicate the standard conditions for the FEA. Regarding the material model used in the FEA, isotropy is assumed in both elasticity and plasticity; Youngs modulus was 200 GPa, Poissons ratio was 0.33, and the Von Mises yield criterion was used. Table 2 shows the work hardening characteristics of the material model assumed in the FEA. The effect of the workhardening exponent n on es was also evaluated. The data, fs-0, for the cold-rolled ultralow carbon steel sheet tested by Kuwabara et al. (1998) was used as a standard condition. The parameters for fs-1 through -3 were determined so that they have the same initial yield p = 0.01 as those for fs-0. stress and the ow stress at 3. Results and discussion 3.1. Optimum strain measurement position Fig. 3 shows the contour diagrams for es calculated for the standard conditions at an equivalent plastic strain of approximately

Fig. 4. Distribution of stress components in gauge area. sx : sy : nominal stress ratio applied to specimen.

Fig. 3. Distribution of stress measurement error es in gauge area.

0.01. As the nominal stress ratio sx : sy changes from uniaxial tension (sx : sy = 1 : 0) to equibiaxial tension (sx : sy = 1 : 1), es becomes more uniform and smaller in the gauge area. Fig. 4 shows the stress distributions in the gauge area. For sx : sy = 1 : 1, x is mostly uniform over a wide area from the central area to the vicinity of the slit tips. For sx : sy = 4 : 3, the distribution of y in the vicinity of the slit tips shows nonuniformity over a somewhat wider area than that for sx : sy = 1 : 1, but otherwise, the uniformity of stress distribution is almost identical. Comparing the stress distributions for sx : sy = 2 : 1, 4 : 1 and 1 : 0, it is noted that the degree of stress nonuniformity for y becomes more signicant than that for x as the stress ratio approaches uniaxial tension, sx : sy = 1 : 0. This is because

Y. Hanabusa et al. / Journal of Materials Processing Technology 213 (2013) 961970

965

Stress measurement error es

0.06 0.04 0.02

Plane stress t0 =1.2mm t0 =3.6mm


p

t0 =0.6mm t0 =2.4mm t0 =4.8mm


0.01

fs-0, von Mises 0.00 0.0 0.2 0.4 0.6 0.8 1.0 Relative distance from center of a specimen

(a) sx : s y
Plane stress t0 =1.2mm t0 =3.6mm

1:0
t0 =0.6mm t0 =2.4mm t0 =4.8mm
p

Stress measurement error es

0.06 0.04 0.02

0.01

Fig. 5. Distribution of

for sx : sy = 1 : 0.

s x :s y = 1:0 s x :s y = 2:1 s x :s y = 1:1


Stress measurement error es
0.06 0.04 0.02 0.00 0.0

s x :s y = 4:1 s x :s y = 4:3
Average fs-0, von Mises
p

fs-0, von Mises 0.00 0.4 0.6 0.8 1.0 0.0 0.2 Relative distance from center of a specimen

(b) sx : s y
Plane stress t0 =1.2mm t0 =3.6mm
fs-0, von Mises

2:1
t0 =0.6mm t0 =2.4mm t0 =4.8mm
p

0.01
Stress measurement error es

0.06 0.04 0.02 0.00

0.01

0.2

0.4

0.6

0.8

1.0

Relative distance from center of a specimen


Fig. 6. Effect of stress ratio sx : sy and strain measurement position measurement error es . on stress

0.0 0.2 0.4 0.6 0.8 1.0 Relative distance from center of a specimen

the bending stiffness of the arms parallel to the y-axis becomes larger than that parallel to the x-axis as Fy decreases. As a typical example, Fig. 5 shows the y distribution for sx : sy = 1 : 0. It is clear that the arms parallel to the y-axis are in the elastic range and that the deformation mode is approximately equivalent to that of a beam with both ends xed against rotation but free to move parallel to the x-axis in opposite directions to each other. From Fig. 3ac it is clear that a strain measurement position exists on the x-axis (the axis parallel to the maximum principal stress direction) where es takes a minimum value. Fig. 6 shows the variation of es along the x-axis. The horizontal axis in the gure represents the dimensionless distance along the x-axis from the center of the specimen, using the variable x/(B/2). For all stress ratios, es increases as approaches 1 because of the stress concentration at the slit tip. For sx : sy = 1 : 1 and 4:3, es is less than 1.4% at all strain measurement positions, except for the vicinity of the slit tip. For sx : sy = 2 : 1 and 4:1, es can be suppressed to within 1.3% when the strain is measured at approximately = 0.7. In reality, however, a nite area is necessary when a strain gauge is used for strain measurement. Therefore, it is recommended to place the strain gauge so that its base foil remains inside the area designated by 0.6 0.8.1

(c) sx : s y

1:1
on

Fig. 7. Effect of initial specimen thickness t0 and strain measurement position stress measurement error es .

Fig. 6 suggests that es is greater for sx : sy = 1 : 0 than for other stress ratios. However, there is no need to use a cruciform specimen for a sx : sy = 1 : 0 test, because it can be performed using a standard uniaxial tension specimen. The next section, therefore, focuses on the evaluation of es for stress ratios sx : sy = 2 : 1 and 1:1, which are typical stress states in sheet metal forming processes, and refers to es for sx : sy = 1 : 0 for comparison purposes only. 3.2. Effect of specimen thickness Fig. 7ac shows the effect of the initial specimen thickness t0 on es for sx : sy = 1 : 0, 2 : 1 and 1 : 1. A plane stress analysis using shell elements was performed only for t0 = 0.6 mm along with the 3D analysis using solid elements. No difference is seen between the plane stress and 3D analyses for t0 = 0.6 mm. es increases throughout the gauge area with increasing t0 . This is due to the increase in the stress component z in the thickness direction and shear stress components zx and yz within the gauge area. It is possible to suppress es to a lower value for all t0 by choosing the strain measurement position to be = 0.7. Quantitatively, es is suppressed to less than 1.9% when t0 0.08B (t0 2.4 mm for B = 30 mm).

1 The smallest gauge length for uniaxial strain gauges commercially available in Japan is 0.2 mm (strain limit of 5% at room temperature). The dimensions of the base foil of the strain gauge are 3.3 2.4 mm2 and approximately correspond to the length 0.6 0.8 for the specimen with a standard geometry of B = 30 mm.

966

Y. Hanabusa et al. / Journal of Materials Processing Technology 213 (2013) 961970

Stress measurement error es

0.06 0.04 0.02

N =3

=3N =5
p

N =7
0.01

=5N =9
Stress measurement error es

L =15mm L =45mm
0.06 0.04 0.02 fs-0, von Mises 0.00
p

L =30mm L=
0 .01

fs-0, von Mises 0.00 0.4 0.6 0.8 1.0 0.0 0.2 Relative distance from center of a specimen

(a) sx : s y
Stress measurement error es

1:0
N =7
p

0.4 0.6 0.8 1.0 0.0 0.2 Relative distance from center of a specimen

0.06 0.04 0.02

N =3

=3N =5

=5N =9
Stress measurement error es

(a) sx : s y
L =15mm L =45mm
0.06 0.04 0.02 fs-0, von Mises 0.00
p

1:0
L =30mm L=
0 .01

0.01

fs-0, von Mises 0.00 0.4 0.6 0.8 1.0 0.0 0.2 Relative distance from center of a specimen

(b) sx : s y
Stress measurement error es

2:1
N =7
p

0.4 0.6 0.8 1.0 0.0 0.2 Relative distance from center of a specimen

0.06 0.04

N =3

=3N =5

=5N =9

(b) sx : s y
L =15mm L =45mm
0.06 fs-0, von Mises 0.04 0.02 0.00
p

2:1
L =30mm L=
0 .01

fs-0, von Mises

0.01
Stress measurement error es

0.02 0.00 0.4 0.6 0.8 1.0 0.0 0.2 Relative distance from center of a specimen

(c)

sx : s y

1:1
on stress

Fig. 8. Effect of number of slits N and strain measurement position measurement error es .

0.0 0.2 0.4 0.6 0.8 1.0 Relative distance from center of a specimen

3.3. Effects of the number of slits, slit length and slit width Fig. 8 shows the effect of the number of slits N on es . The reason why es decreases with increasing N is that the bending stiffness of the respective arms between slits decreases, which results in a decrease of the deformation constraint on the gauge area. es is suppressed to less than 1.6% at = 0.7 when N 7. Fig. 9 shows the effect of the slit length L on es . es decreases with increasing L and remains quite small even in the vicinity of = 0.8. As explained above, this is due to a reduction in the bending stiffness of the respective arms with increasing L. es is suppressed to less than 1.3% at = 0.7 when L B. Fig. 10 shows the effect of the slit width ws on es . es decreases with decreasing ws . At = 0.7, holding ws 0.01B (ws 0.3 mm with B = 30 mm) keeps es 1.5%. It should be noted that an increase of N or ws decreases the maximum load that can be applied to the gauge area since the crosssectional area of the arms is reduced. Therefore, it also decreases the maximum plastic strain applicable to the gauge area (see Appendix). 3.4. Effect of corner radius Fig. 11 shows the effect of the corner radius R on es for sx : sy = 2 : 1. The effect of R on es is negligibly small within the investigated range of 0.0034 R/B 0.1 (0.1 R 3.0 mm with B = 30 mm). Similar results were obtained for other stress ratios.

(c) sx : s y

1:1
on stress measure-

Fig. 9. Effect of slit length L and strain measurement position ment error es .

3.5. Effect of work hardening exponent Fig. 12 shows the effect of the work hardening exponent n on es at sx : sy = 2 : 1. n does not affect es for the range of 0.2 n 0.4. Similar results were obtained for other stress ratios. 4. Experimental validation In this section, experimental validation of the FEA described in Sections 2 and 3 is performed using a sheet material that was experimentally conrmed to be nearly isotropic. 4.1. Test material and specimens The test material used was a 1.2-mm-thick dual phase steel sheet with a tensile strength of 590 MPa (JSC590Y). The work hardening characteristics and r-values at 0 , 45 and 90 (transverse direction; TD) to the rolling direction (RD) are listed in Table 3. The test material was experimentally conrmed to have a small degree of anisotropy, and could thus be considered nearly isotropic. Thus, it was deemed an appropriate material for use in validating the FEA performed in Sections 2 and 3, in which a Von Misestype isotropic material model was used. Hereafter, the RD, TD and

Y. Hanabusa et al. / Journal of Materials Processing Technology 213 (2013) 961970

967

Stress measurement error es

0.06 0.04 0.02

ws =0.2mm ws=0.3mm =0.2mm

w =0.3mm s =0.5mm

Table 3 Mechanical properties of the test material (JSC590Y). Tensile direction 0 45 90


a b 0.2

fs-0, von Mises p 0.01

(MPa)

c a (MPa) 1064 1052 1046


p n p

na 0.184 0.181 0.172

0 a 0.0021 0.0023 0.0016

r -valueb 0.81 0.90 1.04

392 394 398

0.00 0.0 0.2 0.4 0.6 0.8 1.0 Relative distance from center of a specimen

Approximated using 1 = c (0 + 1 ) for 1 = 0.0020.092. Measured at uniaxial nominal strain N = 0.1.

(a) sx : s y
Stress measurement error es

1:0
ws =0.5mm =0.5

thickness directions of a specimen are dened as its x-, y- and zaxes, respectively. 4.2. Biaxial tensile test Biaxial tensile tests were performed using the cruciform specimen shown in Fig. 1, with B = L = 60 mm used to determine an appropriate anisotropic yield function that is capable of reproducing the elasticplastic deformation behavior of the test material. The geometry of the specimens was the same as that used by Kuwabara et al. (1998, 2000, 2002, 2004, 2011). The specimens were cut from an as-received large at rolled sheet sample using laser machining to ensure that the arms were parallel to the RD and TD of the sheet. The slits were also fabricated by laser machining. The concept of the contour of plastic work in the stress space (Hill and Hutchinson, 1992; Hill et al., 1994) was used to quantitatively evaluate the work hardening behavior of the test material under biaxial tension. The stressstrain curve obtained from a uniaxial tensile test in the RD was selected as a reference datum for work hardening. The uniaxial true stress 0 , and the plastic work per unit volume W0 , corresponding to a particular value of offset p true plastic strain 0 , were determined. The uniaxial true stress , in the TD and the biaxial true stress components ( x , y ) were 90 then determined at the same plastic work as W0 . The stress points ( 0 , 0), (0, 90 ) and ( x , y ) thus plotted in the principal stress space form a plastic work contour associated with a particular value p p of 0 . When 0 is taken to be sufciently small, the corresponding work contour can be practically viewed as a yield locus. Biaxial tensile forces (Fx , Fy ) were applied to the cruciform specimen using the previously described servo-controlled biaxial tensile testing machine developed by Kuwabara et al. (1998). The tensile true stress components ( x , y ) were in xed proportions during each test: x : y = 1 : 0, 2 : 1, 4 : 3, 1 : 1, 1 : 2 and 0 : 1. Biaxial total strain components (x , y ) were measured at (x, y) = (+21 mm, 0) for x : y = 2 : 1, 4 : 3 and 1 : 1 and at (x, y) = (0, +21 mm) for x : y = 1 : 2, using an optical 3D deformation

0.06 0.04 0.02

ws =0.2mm =0.2

ws=0.3mm =0.3

fs-0, von Mises p 0.01

0.00 0.0 0.2 0.4 0.6 0.8 1.0 Relative distance from center of a specimen

(b) sx : s y
Stress measurement error es

2:1
ws =0.5mm =0.5

0.06 0.04 0.02

ws =0.2mm =0.2

ws=0.3mm =0.3

fs-0, von Mises p 0.01

0.00 0.0 0.2 0.4 0.6 0.8 1.0 Relative distance from center of a specimen

(c) sx : s y

1:1
on stress mea-

Fig. 10. Effect of slit width ws and strain measurement position surement error es .

Stress measurement error es

0.06 0.04 0.02

R =0.1mm =0.1mmR =0.5mm =1.0mmR =3.0mm R =1.5mm =3 0mm

R =1.0mm =0.5mm =1.5mm

fs-0, von Mises

p 0 .01

0.00 0.4 0.6 0.8 1.0 0.0 0.2 Relative distance from center of a specimen
on stress mea-

1.5

Fig. 11. Effect of corner radius R and strain measurement position surement error es for sx : sy = 2 : 1.

1.0 0

Stress measurement error es

0.06 0.04 0.02

n =0.2 (fs-1)
von Mises

n =0.3 (fs-2)
p

n =0.4 (fs-3)
0.01

p 0

= 0.01 Exp. von Mises Hill'48 Yld2000-2d (M = 5)


0.5 x/ 0
p

0.5

0.00 0.0 0.2 0.4 0.6 0.8 1.0 Relative distance from center of a specimen
on

0.0 0.0

1.0

1.5

Fig. 12. Effect of work hardening exponent n and strain measurement position stress measurement error es for sx : sy = 2 : 1.

Fig. 13. Measured stress points forming the contour of plastic work for 0 = 0.01, compared with those calculated using selected yield functions.

968

Y. Hanabusa et al. / Journal of Materials Processing Technology 213 (2013) 961970

135
y

4.3. Results of biaxial tensile tests Fig. 13 shows the measured stress points that form the conp tours of plastic work for 0 = 0.01. All of the stress values of the respective stress points constituting the work contour are normalp ized by 0 corresponding to 0 = 0.01. Also depicted in the gures are the theoretical yield loci based on the Von Mises (1913), Hills quadratic (Hill, 1948) and the Yld2000-2d yield functions with an exponent of M = 5 (Barlat et al., 2003). The unknown parameters of the Hills quadratic yield function were determined using r0 , r45 , r90 and 0 / 0 , while those of the Yld2000-2d yield function were determined using r0 , r45 , r90 and rb and 0 / 0 , 45 / 0 , 90 / 0 and b / 0 , where r and are the r-value and tensile ow stress measured at an angle from the RD, respectively, and where p p rb and b are the ratio of the plastic strain increments, dy /dx , and the ow stress at equibiaxial tension, x : y = 1 : 1, respectively. The values of r0 , r45 and r90 used were the same as those in Table 3, for both the Hill 48 and Yld2000-2d yield functions. The values of 45 / 0 , 45 / 0 , 90 / 0 , b / 0 and rb used to determine the Yld2000-2d yield function correspond to those dening the work contour shown in the gure. The reason why the exponent M = 5 was chosen for the Yld2000-2d yield function was that it gave a smaller standard deviation from the work contour than M = 4 and 6; see Yanaga et al. (2012) for the standard deviation calculation method. It is clear that the Yld2000-2d yield function with M = 5 agrees better with the work contour than other yield functions. In order to validate the normality ow rule for the yield functions used in Fig. 13, the directions of plastic strain rates, , were

Direction of plastic strain rate

90
x

45

p 0

= 0.01 Exp. von Mises Hill'48 Yld2000-2d (M = 5)

-45

15

30

45

60

75

90

Loading direction

Fig. 14. Comparison of the directions of measured plastic strain rates with those of the local outward vectors normal to the yield loci calculated using selected yield functions.

analysis system (GOM, ARAMIS ). True stress components ( x , y ) were determined by dividing (Fx , Fy ) by the current cross-sectional area of the gauge section, which was determined from the meap p sured values of the plastic strain components x , y with an assumption of constant volume. xy was assumed to be zero, as x and y were measured on the centerlines of the specimen. For x : y = 1 : 0 and 0:1, standard uniaxial tensile specimens, JIS 13 B-type (JIS Z2241), were used. The equivalent plastic strain rate was (56) 104 s1 . Details of the biaxial testing apparatus and test method are given in Kuwabara et al. (1998, 2000).

Fig. 15. Contour diagrams for total strains x and y measured for

= 4 : 3 at x = 0.02 (a), compared with those calculated using FEA (b).

Y. Hanabusa et al. / Journal of Materials Processing Technology 213 (2013) 961970 Table 4 p Material parameters for the Yld2000-2d (M = 5) yield function for 0 = 0.01. 1 0.9572 2 0.9905 3 0.9781 4 0.9686 5 1.014 6 0.9715 7 0.9791 8 1.0316

969

of the geometrical parameters of the cruciform specimen on es were also examined in detail. es is estimated to be less than 2% when the conditions (1), (2) and (3) are satised: 1) The thickness t0 of the cruciform specimen is less than 0.08B, where B is the side length of the gauge area. 2) N 7, L B and ws 0.01B, where N is the number of slits in each arm, L is the slit length, and ws is the slit width of the cruciform specimen. The effect of the corner radius R on es is negligibly small when it is in the range of 0.0034 R/B 0.1. 3) The biaxial strain components are measured at the position(s) on the centerline of the specimen, parallel to the maximum force direction, at a distance of approximately 0.35B from the center of the specimen. The validity of the FEA was experimentally veried by comparing the FEA results with those measured from the biaxial tensile tests of a sheet material that was experimentally conrmed to be nearly isotropic. Acknowledgments The authors would like to express their gratitude for the indispensable advice we received from the members of the Committee on Standardization of Biaxial Stress Test Method, which has been supporting the standardization project since 2008 under the auspices of the Osaka Science and Technology Center on commission from the New Energy and Industrial Technology Development Organization (NEDO). Appendix A. For the cruciform specimen investigated in this study, the arms are subjected to uniaxial tension, which means that the test is complete when the nominal stress in the arms reaches the tensile strength of the material. Accordingly, the maximum equivalent p plastic strain max applicable to the gauge area of the cruciform specimen can be estimated using the considere condition for a p maximum load on a strip in tension (Marciniak et al., 2002). max depends primarily on the stress ratio, the work hardening exponent, n, and the anisotropy of the test material, and the slit width ws cut into each arm of the cruciform specimen. This Appendix shows p the effects of n and ws on max . It should be noted here that these calculation results should be viewed only as a reference, because they are numerical analysis solutions based on the simple mechanics of plasticity assuming an isotropic material. p Fig. A.1 shows the effect of the number of slits, N, on max for a cruciform specimen with B = 30 mm and ws = 0.2, 0.3 and 0.5 mm subjected to a true stress ratio of x : y = 2 : 1. For the p standard conditions (N = 7 and ws = 0.2 mm), max = 3.2%, but p decreases to 2.9% when N = 9. When ws = 0.5 mm, max decreases to 1.5% for N = 9. This is because the effective cross-sectional area

800

/ MPa

600

True stress

400
y

=1:0

200

=4:3

Experimental FEA 0 0.00 0.01

True plastic strain


Fig. 16. Biaxial stressstrain curves measured for those calculated using FEA.
x

0.02

= 4 : 3, compared with

measured for all linear stress paths and compared with those of the outward vectors normal to the theoretical yield loci at corresponding stress points. The results are shown in Fig. 14, where is the loading angle of a stress path from the x-axis in the principal stress space, and both and are dened as zero along the x-axis and positive in the anti-clockwise direction. The Yld2000-2d yield function with M = 5 again provides the closest agreement with the measurement. Consequently, we conclude that the Yld2000-2d yield function with M = 5 is an appropriate material model for the test material under linear stress paths. Table 4 shows the material parameters 1 8 for the Yld2000-2d yield function with M = 5. 4.4. Validation of the FEA Fig. 15 shows the contour diagrams for total strains x and y measured for x : y = 4 : 3 at x = 0.02, compared with those calculated using FEA. The FEA was performed using the Yld2000-2d yield function with M = 5, as determined in Section 4.3. The measured and calculated results for both x and y were found to be in good agreement with each other. Fig. 16 shows the biaxial stressstrain curves measured for x : y = 4 : 3, compared with those based on the FEA for the same stress ratio. For the FEA, the calculation procedure used for determining true stress and true plastic strain components was the same as that used in the experiment. The FEA result agreed well with the measurement. From the results shown in Figs. 15 and 16, it can be concluded that the validity of the FEA as described in Sections 2 and 3 has been experimentally veried. This result is consistent with the conclusion obtained by Hanabusa et al. (2010) and indicates that the optimal strain measurement position for an isotropic material, as claried in this study, also holds true for anisotropic sheet materials. 5. Conclusions This paper investigated a method for determining the optimum strain measurement position for minimizing the stress measurement error es in a biaxial tensile test method using the cruciform specimen developed by Kuwabara et al. (1998). The effects

Fig. A.1. Effect of number of slits N and slit width ws on maximum equivalent plastic p strain max applicable to gauge area of cruciform specimen.

970

Y. Hanabusa et al. / Journal of Materials Processing Technology 213 (2013) 961970

(a)

0.12 0.10 0.08

: y =2:1 x

0.06 0.04 0.02 0.00

n =0.30 n =0.25 n =0.20 n =0.10

0.005

0.010

ws / B

0.015

0.020

(b)

0.12 0.10 0.08


n =0.30 n =0.25 n =0.20
x

: y =1:1

0.06 0.04 0.02 0.00

n =0.1

0.005

0.010

ws / B

0.015

0.020

Fig. A.2. Effect of slit width ws and work hardening exponent n on maximum equivp alent plastic strain max applicable to gauge area of cruciform specimen. The number of slits is 7.

of the arms decreases with increasing N, although the stress distribution in the gauge area becomes more uniform. p Fig. A.2 shows the effects of ws and n on max with N = 7 for x : p y = 2 : 1 and 1:1. max decreases with increasing ws , because the effective cross-sectional area of the arms decreases, which results in a decrease in the maximum force transmitted to the gauge area of p the cruciform specimen. max increases with n, because a material with a larger n-value is subject to a higher stress increase rate as the plastic deformation of the arms increases, which in turn causes an increase in the tensile stress acting on the gauge area. References
Barlat, F., Brem, J.C., Yoon, J.W., Chung, K., Dick, R.E., Lege, D.J., Pourboghrat, F., Choi, S.H., Chu, E., 2003. Plane stress yield function for aluminum alloy sheets part 1: theory. International Journal of Plasticity 19, 12971319. Boehler, J.P., Demmerle, S., Koss, S., 1994. A new direct biaxial testing machine for anisotropic materials. Experimental Mechanics 34, 19. Borsutzki, M., Keler, L., Sonne, H-M., 2002. Kennzeichnung des Verfestigungsverhaltens von Werkstoffen mit der Biaxialprfung. In Werkstoffprfung 2002. Proc. DVM-Conference, Bad Nauheim, p. 186. (in German). Demmerle, S., Boehler, J.P., 1993. Optimal design of biaxial tensile cruciform specimens. Journal of the Mechanics and Physics of Solids 41 (1), 143181. Ferron, G., Makinde, A., 1988. Design and development of a biaxial strength testing device. Journal of Testing and Evaluation 16, 253256. Geiger, M., Hubnatter, W., Merklein, M., 2005. Specimen for a novel concept of the biaxial tension test. Journal of Materials Processing Technology 167, 177183. Green, D.E., Neale, K.W., MacEwen, S.R., Makinde, A., Perrin, R., 2004. Experimental investigation of the biaxial behaviour of an aluminum sheet. International Journal of Plasticity 20, 16771706. Hanabusa, Y., Takizawa, H., Kuwabara, T., 2010. Evaluation of accuracy of stress measurements determined in biaxial stress tests with cruciform specimen using numerical method. Steel Research International 81, 13761379. Hannon, A., Tiernan, P., 2008. A review of planar biaxial tensile test systems for sheet metal. Journal of Materials Processing Technology 198, 113.

Hallfeldt, T., 2002. Untersuchungen zur Beschreibung des Tief- und Streckziehverhaltens hherfester Feinbleche ans Stahlwerkstoffen. Fortschr.-Ber. VDI Reihe 2 Nr. 602, Dsseldorf: VDI Verlag. Hashimoto, K., Kuwabara, T., Iizuka, E., Yoon J.W., 2010. Effect of anisotropic yield functions on the accuracy of hole expansion simulations for 590 MPa grade steel sheet. Tetsu-to-Hagan, 96-9 (2010). 557-563. (in Japanese). Hayhurst, D.R., 1973. A biaxial-tension creep-rupture testing machine. Journal of Strain Analysis 8 (2), 119123. Hill, R., 1948. A theory of the yielding and plastic ow of anisotropic metals. Proceedings of the Royal Society of London A 193, 281297. Hill, R., Hutchinson, J.W., 1992. Differential hardening in sheet metal under biaxial loading: a theoretical framework. Journal of Applied Mechanics 59, S1S9. Hill, R., Hecker, S.S., Stout, M.G., 1994. An investigation of plastic ow and differential work hardening in orthotropic brass tubes under uid pressure and axial load. International Journal of Solids and Structures 31, 29993021. Hjelm, H.E., 1994. Yield surface for grey cast iron under biaxial stress. Transactions of the ASME: Journal of Engineering Materials and Technology 116, 148154. Hoferlin, E., Van Bael, A., Van Houtte, P., Steyaert, G., De Mar, C., 2000. The design of a biaxial tensile test and its use for the validation of crystallographic yield loci. Modelling and Simulation in Materials Science and Engineering 8, 423433. Ikeda, S., Kuwabara, T., 2002. Measurement and analysis of work hardening of sheet metals under planestrain tension. In: Yang, D.-Y., Oh, S.I., Huh, H., Kim, Y.H. (Eds.), The Fifth International Conference and Workshop on Numerical Simulation of 3D Sheet Forming Processes (NUMISHEET 2002). Jeju Island, Korea, pp. 97102. Kreiig, R., Schindler, J., 1986. Some experimental results on yield condition in plane stress state. Acta Mechanica 65, 169179. Kulawinski, D., Nagela, K., Henkela, S., Hbnerb, P., Fischerc, H., Kunac, M., Biermanna, H., 2011. Characterization of stressstrain behavior of a cast TRIP steel under different biaxial planar load ratios. Engineering Fracture Mechanics 78, 16841695. Kuroda, M., Tvergaard, V., 1999. Use of abrupt strain path change for determining subsequent yield surface: illustrations of basic idea. Acta Materialia 47, 38793890. Kuwabara, T., Ikeda, S., Kuroda, T., 1998. Measurement and analysis of differential work hardening in cold-rolled steel sheet under biaxial tension. Journal of Materials Processing Technology 8081, 517523. Kuwabara, T., Kuroda, M., Tvergaard, V., Nomura, K., 2000. Use of abrupt strain path change for determining subsequent yield surface: experimental study with metal sheets. Acta Materialia 48 (9), 20712079. Kuwabara, T., Van Bael, A., Iizuka, E., 2002. Measurement and analysis of yield locus and work hardening characteristics of steel sheets with different r-values. Acta Materialia 50 (14), 37173729. Kuwabara, T., Ikeda, S., Asano, Y., 2004. Effect of anisotropic yield functions on the accuracy of springback simulation. In: Ghosh, S. (Ed.), Proc. 8th Int. Conf. on Numerical Methods in Industrial Forming Processes. American Institute of Physics, New York, p. 887. Kuwabara, T., 2007. Advances in experiments on metal sheets and tubes in support of constitutive modeling and forming simulations. International Journal of Plasticity 23 (3), 385419. Kuwabara, T., Hashimoto, K., Iizuka, E., Yoon, J.W., 2011. Effect of anisotropic yield functions on the accuracy of hole expansion simulations. Journal of Materials Processing Technology 211, 475481. Lin, S.B., Ding, J.L., 1995. Experimental study of the plastic yielding of rolled sheet metals with the cruciform plate specimen. International Journal of Plasticity 11 (5), 583604. Makinde, A., Ferron, G., 1988. Strain-hardening characteristics of aluminum-1050A, -(70/30) brass, and austenitic stainless steel under biaxial loading. Journal of Testing and Evaluation 16 (5), 461469. Makinde, A., Thibodeau, L., Neale, K.W., 1992. Development of an apparatus for biaxial testing using cruciform specimens. Experimental Mechanics 32, 138144. Marciniak, Z., Duncan, J.L., Hu, S.J., 2002. Mechanics of Sheet Metal Forming. Butterworth-Heinemann, Oxford. Moriya, T., Kuwabara, T., Kimura, S., Takahashi, S., 2010. Effect of anisotropic yield function on the predictive accuracy of surface deection of automotive outer panels. Steel Research International 81 (9), 13841387. Mller, W., Phlandt, K.J., 1996. New experiments for determining yield loci of sheet metal. Journal of Materials Processing Technology 60, 643648. Naka, T., Nakayama, Y., Uemori, T., Hino, R., Yoshida, F., 2003. Effects of temperature on yield locus for 5083 aluminum alloy sheet. Journal of Materials Processing Technology 140, 494499. Pascoe, K.J., de Villiers, J.W.R., 1967. Low cycle fatigue of steels under biaxial straining. Journal of Strain Analysis 2 (2), 117126. Shiratori, E., Ikegami, K., 1968. Experimental study of the subsequent yield surface by using cross-shaped specimens. Journal of the Mechanics and Physics of Solids 16, 373394. Von Mises, R., 1913. Mechanik der festen Krper im plastisch-deformablen Zustand. Goettinger Nachrichten Math. Phys. Klasse 4, 582592. Yanaga, D., Kuwabara, T., Uema, N., Asano, M., 2012. Material modeling of 6000 series aluminum alloy sheets with different density cube textures and effect on the accuracy of nite element simulation. International Journal of Solids and Structures 49, 34883495. Yu, Y., Wan, M., Wu, X.-D., Zhou, X.-B., 2002. Design of a cruciform biaxial tensile specimen for limit strain analysis by FEM. Journal of Materials Processing Technology 123, 6770.

max

max

You might also like