You are on page 1of 8

Available online at ScienceDirect

J. Mater. Sci. Technol., 2013, 29(11), 1059e1066

Prediction of Failure Modes during Deep Drawing of Metal Sheets with Nickel
Jie Wu1,2), Zengsheng Ma1,2)*, Yichun Zhou1,2)**, Chunsheng Lu3)
1) Key Laboratory of Low Dimensional Materials and Application Technology of Ministry of Education, Xiangtan University,
Xiangtan 411105, China
2) Faculty of Materials, Optoelectronics and Physics, Xiangtan University, Xiangtan 411105, China
3) Department of Mechanical Engineering, Curtin University, Western Australia 6845, Australia
[Manuscript received May 16, 2012, in revised form January 16, 2013, Available online 9 August 2013]

To optimize the process parameters, it is necessary to exactly predict failure modes during deep drawing of
coated metal sheets, where two main failure forms are fracture and wrinkling. In this paper, finite element
simulations based on continuous damage mechanics were used to study the failure behavior during a
cylindrical deep drawing of metal sheets with nickel coating. It is shown that taking the effect of blank holder
force into account, these two failure modes can be predicted. The simulation results are well consistent with
that obtained from experiments.

KEY WORDS: Failure; Deep drawing; Coated metal sheet; Damage

1. Introduction In finite element simulations, a forming limit diagram (FLD) is
widely applied to analyze the fracture phenomena following the
Deep drawing is one of the most popular processes for pioneering work of Keeler[6] and Marciniak and Cuczynski[7].
transforming flat metallic sheet blanks into cup or box shaped Based on major and minor strains, FLD is associated with the
parts in automotive and aerospace industries. During the process strain path; hence, a forming limit stress diagram was sug-
of deep drawing, metallic sheets are subjected to the large irre- gested[8]. Signorelli et al.[9] obtained FLD considering the in-
versible deformation, and two primary failure modes of fracture fluence of strain rate using a rate-dependent polycrystal self-
and wrinkling may appear[1]. Huge design and control efforts consistent plasticity model. Hu et al.[10] proposed a new FLD
have been made to eliminate the occurrence of failure through by introducing the effect of temperature and strain rate. The FLD
the proper design of blank[2], tooling configuration[3], and the provides the foundation of research on the fracture behavior
selection of process parameters[4]. Using the traditional trial-and- during deep drawing of metal sheets. By comparing the strain
error approach, one-fourth or more time spends on the design status corresponding to elements, one can estimate whether or
and control procedure[5]. Fortunately, with development of finite not fracture behaviors appear during forming[11]. However, FLD
element methods and computer technologies, numerical simu- does not have a predictive nature[12]. In particular, it is impos-
lations can be used to model the forming process. In addition to sible to predict when and where a failure can appear, and which
avoiding long and expensive trial and error procedures, a better failure mode may occur in work piece during a forming
understanding on the metal forming process can also be operation.
obtained. There have been many works on the prediction of fracture in
the single layer sheet metal forming by finite element modeling
based on ductile damage. Saanouni[13] predicted the fracture area
in hydro-bulging tests with an ellipsoidal matrix by ductile
Corresponding author. Ph.D.; Tel.: þ86 731 58293586; Fax: þ86 731 damage evolution. Based on Saanouni’s theory, Khelifa and
58293468; E-mail address: (Z. Ma)
Oudjene[12] investigated deep drawing (i.e., Swift’s test) of
Corresponding author. Prof.; Tel.: þ86 731 58293586; Fax: þ86 731
58293468; E-mail address: (Y. Zhou).
aluminum sheets. Fan[14] studied the crack initiation during deep
1005-0302/$ e see front matter Copyright Ó 2013, The editorial office of drawing of square cups with different frames of ductile damage
Journal of Materials Science & Technology. Published by Elsevier evolution. However, there have been few studies on the pre-
Limited. All rights reserved. diction of the failure behavior during deep drawing of a coated metal sheet.

2013. k0 is equal to 1.1. 1. more and more coated and pre-coated metal sheets have been used due to their good wear or corrosion resistance and decorative performance[15]. _ ε ¼ D sinh½k0 ðh  hþ Þ (1) where h ¼ sm =s is the stress triaxiality with hþ ¼ 2/3 and h ¼ 2/3 for isotropic materials in equibiaxial tensile and compressive deformation states. During the forming of a coated metal sheet. This material possesses good corrosion resistance.: J. The metal sheet with nickel coating (MSNC) is a typical type of material for safeguard in engineering. εþ  T and εT correspond to equivalent plastic strains at ductile damage initiation for equibiaxial tensile and compressive deformation. which offer potential for advanced structural engineering applications.18] or zinc phosphate coatings/steel sheets[19].[26] reported an extensive numerical and experimental study of the deep-drawing process leading to initiation and propagation of cracks based on the continuous damage mechanics. Two main failure modes (fracture and wrinkling) were studied under the condition of the blank holder force. Many researchers have used the continuous damage mechanics and the triaxiality dependent failure criteria to predict fracture in sheet metal forming[24e26]. initiation is met when the following condition is satisfied . 2. that is        T sinh k0 h  h εþ T sinh k0 h  h  pl  þ ε þ εpl ¼ f h . 2 Schematic diagram of the model: (a) configures of the tool and related to the complete loss of load-carrying capacity and thus coated metal sheet and (b) their geometric dimensions. Sci. For Fig. and Fe/A1 laminated composite sheets[21]. and coalescence of voids. and excellent plas- ticity. There are different types of coating and substrate systems. 1059e1066 Recently. the good adhesion between electro- deposited nickel coating and substrate is another advantage in practical applications[22]. In this work.1060 J. Finite Element Simulations 2. The criterion for damage fracture. Li et al. Technol. respectively. A set of experiments were performed to verify simulation results and their abilities to predict the failure behavior in a work-piece during deep drawing. with sm ¼ (s1 þ s2 þ s3)/3 and the von Mises equivalent stressps ¼ ½1=2ðs1  s2 Þ2 ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi s s 2 s s þð 2  3 Þ þ ð 3  1 Þ  . 1 Schematic diagram of the failure modes: (a) wrinkling and (b) isotropic materials. results in the progressive degradation of stiffness. nickel coating/steel[17. attractive toughness. Continuous damage is a phenomenological model for predicting the onset of damage due to nucleation. Material failure is Fig. such as hot-dip galvanized/steel or electro-galvanized/steel sheets[16]. Thus. brass/steel two layer sheets[20]. For example.2 1=2 _ pl ε ¼ 2=3_εij ε_ ij is the equivalent plastic strain rate.. it is important to clarify which failure mode occurs for a set of given parameters. 29(11). there is a failure mode like wrinkling[23] or fracture[21] appearing in a single layer sheet. In addition. Continuous damage mechanics In the finite element package ABAQUS. finite element models based on continuous damage mechanics were used to predict failure behaviors during a cylindrical deep drawing of MSNC. and k0 is a material parameter related to anisotropy. growth. Wu et al. The model assumes that the equivalent plastic strain εpl D at the onset of damage is a function of stress triaxiality and strain rate. as shown in Fig. there is a general framework for modeling damage and failure. Mater.

.27 0. (b) finite element meshes of a whole blank. ε E (GPa) n εpl Substrate 209.13 where wD is a state variable that increases monotonically with plastic deformation.0 470.9 264. 6 mm.4 0. ε22. 3 Finite element model and meshes used in simulations: (a) a quarter model. (c) the stress distribution in the thickness direction. Mater. 2013. Wu et al.12 Coating 215.30 0. the incremental Fig. Table 1 Mechanical properties of MSNC. Sci. 4 Simulations at punch displacements of 4 mm. s22 and (d) the true strain distribution in the thickness direction. (b) the damage distribution. J. At each increment. and (c) meshes in plane and thickness directions. 29(11).: J. where substrate and coating are low carbon steel sheet and nickel coating Z dεpl wD ¼   ¼ 1 (2) sy (mm) n _ pl D h. Technol. 8 mm and 10 mm in the case of wrinkling: (a) the damage distribution along a path from center to edge in nickel coating.0 0.. 1059e1066 1061 Fig.

6 mm. The orig- The geometrical method is adopted in our simulation as the inal point of the coordinate system is located at the center of wrinkling criterion. Mx ¼ 0 and Mz ¼ 0 at the original point ux ¼ 0: The dynamic explicit code ABAQUS/Explicit is used to simulate the cylindrical deep drawing of MSNC. The forming tools such as die. 8 mm and 10 mm in the case of success: (a) the damage distribution along a path from center to edge in nickel coating. The gap between die and blank DwD ¼  pl  0 (3) holder for safe products is equal to 0. The crit. The tool and The blank is a low carbon steel sheet of 0.2. 3(a). My ¼ 0 and Mz ¼ 0 on the surface of x ¼ 0. (b) the damage contour. Finite element model ux ¼ 0. 2(a). as shown in εpl D h. 29(11). Mx ¼ 0 and My ¼ 0 on the surface of z ¼ 0. as shown in Fig.1062 J. Both Fig. s22 and (d) the true strain dis- tribution in the thickness direction. 1059e1066 increase of wD is calculated by Fig.5 mm. uz ¼ 0. . (c) the stress distribution in the thickness direction. ε22.: J. Sci. only a quarter of blank is chosen in modeling. rection. 2(b). 2013. For the blank sheet. The die is fixed in of deformed mesh. Mater. which directly measures wrinkle dimensions blank with thickness being the y direction.. Wu et al. Technol. The wrinkle amplitude was measured from the process of deep drawing and punch can move in the y di- the gap between blank holder surface and die surface.3 mm thickness with coated metal sheet in a deep drawing process are illustrated in double-side electrodeposited 3 mm thickness nickel coating. uz ¼ 0. 2. boundary conditions can be ical wrinkle amplitudes for determining the existence of a flange described as wrinkle are chosen to be at 8% nominal sheet thickness[27]. punch and blank holder are considered as rigid bodies and their length param- Dεpl eters are shown in Fig. 5 Simulations at punch displacements of 4 mm. Because of symmetry of the sample. 2(a). ε_ Fig.

3(b) and (c). All the three BHFs are also the same as that adhesion between nickel coating and substrate[22].. normal stress in the thickness direction increases. blank holder. 1. When increasing the BHF. ried out by using oil based lubricant to punch. Experiments were car- (C3D8r) elements of 31. 29(11). 6 mm. 2. w0. (b) the damage distribution. Wu et al. it is shown that the deviation is drawing experiments of cylindrical flat punches were performed very small (w1%) between two results obtained by using the on a RG2000 micromachine-controlled universal tensile machine number of eight-node. The true stressestrain is one of the most important parameters.  s=E s  sy which restrains formation of wrinkles. deep Table 1[28]. as die surfaces as well as to both surfaces of blank. s22 and (d) the true strain distribution in the thickness direction. 1 kN) are applied in a deep drawing process. Fig. 2013. In our simulations. the blank holder force (BHF) material with exponential hardening. The interface between coating and coefficient between tools and blank is the same as that in sim- substrate is assumed to be perfect because of the strong ulations. 6 Simulations at punch displacements of 4 mm. Sci.518 and w50. where sy is the initial yield stress. ε22. . Experimental respectively. 1059e1066 1063 coating and substrate are modeled as an isotropic elasticeplastic During deep drawing of MSNC. respectively.5 kN. linear brick. 3.1[29]. Three constants of BHFs (2 kN. E and n are Young’s modulus and Poisson’s ratio.: J. The friction shown in Fig. Mater. used in finite element modeling. (c) the stress distribution in the thickness direction. and thus fracture may ε ¼    (4) sy =E s=sy 1=n s  sy occur at the cup wall and punch profile. n is the strain hardening exponent. Technol. reduced integration with the tool geometry shown in Fig. J. 8 mm and 10 mm in the case of fracture: (a) the damage distribution along a path from center to edge in nickel coating. and the former is used in simulations of Ni coating metal sheets.000. relationship is given by wrinkles may appear in flange of drawn parts. For a small BHF. Thus. Material properties of nickel coating and low carbon steel substrate used in simulations are listed in To verify the results of finite element simulations.

as shown in Fig. 7 exhibits damage as a function of punch displacement u. Wu et al. respectively.2. and 6. . stress contours in the thickness direction. While the maximum principal strain in fracture is much in wall areas. the displacement of blank in the thickness direction in flange areas is much larger than that in fracture and success. Technol. s22 and true strain distributions in the thickness direc- tion. 10(b) and (c). Damage arrives to 1 in fracture and principal strain in the flange area in the cases of wrinkling. Mater. In wrinkling. There are the same behaviors in the cases of success and fracture. as shown in Fig. The fracture area is located at the punch fillet larger than that in wrinkling and success at the punch fillet radius. 1059e1066 4. Sci. the damage and stress distributions in the thickness di- rection. The failure modes are judged from the distribution in fracture and success. 4(c) and (d). 4(a). two flange area. In the case of success shown in Fig. from center to edge. Fig. where damage is equal to 1. as shown in Fig. Fig. 8 Relationship between the punch displacement and maximum as shown in Figs. Damage is less than 1 in the flange area simulations. as shown in displacement of blank in the thickness direction is discontinuous Fig. 8. where damage firstly arrives at 1 when the punch main failure modes (fracture and wrinkling) were predicted in displacement is 10 mm. Accordingly.. In Figs. 9 Relationship between the punch displacement and maximum cases of wrinkling. s22 and true strain distributions in the thickness direction. success and fracture: (a) in the flange area principal strain at the punch radius in the cases of (a) fracture and (b) at the punch radius. Results and Discussion 4. which refers to the occurrence of failure. 8 and 10 mm. 29(11). where the is much larger than that in fracture and success. Fig. ε22 are shown in success and fracture at different punch 4. 4(b). wrinkling. displacement of blank in the thickness direction is continuous in wall areas. the radius. the distribution of damage at different punch displacements u in nickel coating is shown in Fig. 8 and 10 mm. 7 Relationship between the punch displacement and damage in the Fig. Fracture occurs at the punch fillet radius of displacements of blank in the thickness direction along a path of a cup. 5 and 6.1. Wrinkling takes place in the In consideration of the influence of a blank holder force. 9(a) and (b). This is also seen from the damage contours at different punch displacements of 4. fracture occurs In the flange area.: J. and (b) wrinkling and success. Failure modes displacements of 4. As shown in Fig. 10(b). ε22 are shown in Fig. 6. the values of stress and strain increase. 6. When the punch displacement increases. 10(a).1064 J. 2013. the maximum principal strain in wrinkling when the punch displacement is equal to 8 mm. 5 success and fracture. It is seen that damage increases with the increase of punch dis- placements. Distribution of field variables In the case of wrinkling.

Technol. 12. the punch load increases to another value of 9. Wu et al. 11 Punch loadedisplacement curves of simulation and experiment Fig. . as shown in Fig. the punch load firstly increases to a maximum value of 10 kN when the punch displacement is 8. J. The punch load firstly increases to 9. 10 Deformation shapes with different punch displacements of The punch displacement is equal to 7 mm when fracture occurs.2 kN at the displacement of 16. 13 Punch loadedisplacement curves of simulation and experiment in the case of fracture. the flange area is totally drawn into the die cavity. The sharply decreasing punch load indicates the fracture of MSNC. Finally. 13. leading to decrease of the punch load.8 mm. wrinkling oc- curs in the flange of MSNC. Sci. As shown in Fig. 12 Punch loadedisplacement curves of simulation and experiment in the case of wrinkling. 29(11). which agrees with the experimental value of 6.6 mm.1 kN with the punch displacement.7 mm. the normal stress in the thickness direction of blank suppresses the forma- tion of wrinkles.5 kN. Similarly. But high BHF leads to insufficient metal flow and thus fracture occurs in a deep drawing process. The normal stress in the thickness direction of blank cannot suppress the excessive metal flow. (b) suc. the comparison of deformation shapes between experiments and finite element simulations is based on continuous damage mechanics. When BHF is high.6 mm. 2013. It is found that some parts of MSNC have been drawn into the die cavity.: J. Mater. blank in simulations: (a) fracture with BHF ¼ 2 kN. Then. This is because wrinkling makes blank with nickel coating draw into the die cavity more difficultly.. Fig. Fig. cess with BHF ¼ 1. 1059e1066 1065 Fig. and (c) wrinkling with BHF ¼ 1 kN. as shown in Fig. The actual BHF exerted on blank decreases. The failure modes are judged by using punch loade displacement curves. As predicted. and then decreases with increasing the punch Fig. 11. in the case of success. which is located at the circumference zone contacting the punch radius. 11 shows a comparison of punch loadedisplacement curves obtained by finite element simula- tions and experiments in the wrinkling mode. and then decreases to 5 kN at the displacement of 13.

P. Mater. In the case of success.C. H. Debray. T. failure modes are affected by the blank holder force. Hatta. simulations. S. Int.C. Jiang. Y. Hunan 2005. P. 151 (2004) 88e97. 206 (2008) 431e437. H. Technol. C.G. J. 14 Forming limit diagram of three cases in simulations: fracture. radius. Technol. J.B. The points located above the forming limit curve represent [16] A. . Y. B. Werner. 91 (2002) 5501e The forming limit curve is extracted from our early studies[17. J.H. J. Boyce. Broek. Zhou. 5503.P. Guo.H. Process. K. S. A 501 drawing have been successfully predicted by using finite element (2009) 235e241. 23 mm. Li. J. Wierzbicki. Mater. Signorelli.D. Ferron. [21] H. Altan. Mater. Zhang. Des. Zhou.Q.P. Cao. [3] M. X.C. J. Lin. Mater. Zhou. J. 31 (2010) 1410e 1416. wrin. 125 (2002) 743e750. Santhanam. Turner. This area shows a tendency to wrinkling or fully devel. and the Key Special Program for Science and Technology of Hunan Province. Li. Process. kling and success during finite element simulations is shown in Tools Manuf.L. Wang. Marciniak.1066 J. According to Mater. The strain status in fracture. Technol. Tools Manuf. Technol. Y. F. Guo. J. Lu. (2012) 626e635. Technol. Keeler.C. Mech. Des. 102 (1998) 233e236. Kapinski. Martusiene. Technol.E. Technol. Mater. J. [7] Z. Technol. Y. Mater. Mater. Technol. Plasticity 25 displacement. [8] T. Process. J.Q. Int. [4] Z. Lulea University of Technology. Cai. Fig. G. Long. [28] Y. J. Process. Eng.R. Li. 28 strain. [31] L. [15] M. Asadian-Ardakani. 9 (1967) 613e620. [29] L. 1059e1066 Innovation Group. Surf. Moreira. D. 44 (2004) 487e494. Sheng. plane of principal strains[30]. Process.C. Gelarch. Manolakos. Mori. [2] V. Chun. Luo. Izciler.31].Q. [10] X.Q. 2013. oped wrinkles. Mater. Arwidson. In this paper.Q. J. Stoughton. fracture of blank with nickel coating. T. J. Zhou. 09JJ3003). Y. Process. simulations based on continuous damage mechanics are a Technol. Li. a lot of points Technol. [9] J. 5. 51 (1995) 275e279. J. drawing of coated metal sheets. Perform. J.X. Ind. thesis.M. Crashworthiness 9 For the relative low BHF. the failure modes of nickel coating during deep Technol. Hooputra. Int. [22] L. Dell. Batoz.Q. Mech.P. Process. 42 (1965) 683e691. Y.Y. H. Cuczynski. Int. Bertinetti. S. Bogard. Y. Fract. Process.A. Process. Tsui. simulations. 60 (1996) 197e200. M. Li. Ma. nickel coating sheet.S. Comparison of strain status [11] Y. A.A. success and wrinkling. J. M. Tang. 42 (2000) 1e27. Gavas. wrinkling takes place in the flange of (2004) 449e464. Y. J. 210 (2010) 1738e1747. Provincial Natural Science Foundation of China (No.G. Xiao. T. locate in areas where the minor strain is larger than the major [18] Z. 51172192 and 11102176). Mach. Sci. Appl. 15 (2006) the Natural Science Foundation of Hunan Province for 287e294.P. S. 29(11). Saanouni. L. Sci. Z. Y. Zhou. verify the accuracy of finite element modeling. Morovvati. Sci. Mater. Mater. Burokas. REFERENCES [1] Y. M.L.G. China (No. Ferran. confirmed by experiments. J. the risk of fracture and wrinkling 71e76. occurs along a circumference zone in contact with the punch [26] Y. Phys. the points locate in the safe area. [14] J. G. Fig. Y. C. Numer. Sheet Met. 46 (2006) 1035e1044. J.. Sci. Oudjene. Wu et al. can be evaluated by using the forming limit curve defined in the [13] K. Process. J. Ahmetoglu. Foundation of China (Nos. Ph. 75 (2008) 3545e3559. Khelifa. [12] M. The results show that finite element [27] M. 68 (1997) 71e75. Int. Chan. Mater. J.G. Fract. Su. T. Mech. Mamalis. Mater. Sci. Acknowledgments [30] C. The accuracy of finite element modeling has been 210 (2010) 1858e1869.3.H. Y.M. Mollaei-Dariani. 60 (1996) 291e296. Int. Mater. J. Y. Technol. Fujimoto. Mater. Mater.K. 14. 44 (1995) 247e250. Zhou. Mech. Pegada. Jirathearanat. Takuda. Liao. Gese. N.H. Lee.C. all Technol. M. Zhou.M. M. J. Numerical Simulation of Sheet Metal Forming for This work was supported by the National Natural Science High Strength Steels.P. Eng. MSNC is totally drawn into the die cavity. Eng. Conclusion [20] S. Mater. L. 30 (1990) 1385e1401. Wang. Zhou. fracture [25] T. Technol. A set of experiments were performed to test and [23] M. Altan.J. J. When the punch displacement reaches about (2009) 1e25. CIRP Ann-Manuf. A. Baldoukas. Detraux. Hu. Mech. K. Int. 108 (2000) 78e86. J. L. and in wrinkling. [17] L. Fan.S. Kinzel. Eng. 41 promising method in predicting failure modes during deep (2009) 308e318. [19] V. [24] H. G. which is along a path from the center to edge of blank. M. Mach.: J. Mater. Bikulcius. Methods Eng. G. Coat. [5] J.R.P. Tang. C. J. 119 (1997) 354e365. Zhao. Eng.C. Huang. 2009FJ1002).C. 09JJ7004). T. and for the relative high BHF.W. [6] S. 200 (2008) In a deep drawing procedure. China (No. K. 4. H. 28 (2007) 1641e1646.