You are on page 1of 8

Journal of Materials Processing Technology 159 (2005) 303310

An adaptive simulation approach designed for tube hydroforming processes


A. Aydemir a, , J.H.P. de Vree a , W.A.M. Brekelmans a , M.G.D. Geers a , W.H. Sillekens b , R.J. Werkhoven b
a

Department of Mechanical Engineering, Eindhoven University of Technology, PO Box 513, 5600 MB Eindhoven, The Netherlands b TNO Industrial Technology, Department of Manufacturing Development, PO Box 6235, 5600 HE Eindhoven, The Netherlands Received 15 July 2003; received in revised form 17 May 2004; accepted 20 May 2004

Abstract The efciency of a tube hydroforming process is largely dependent on the process control parameters (i.e. the internal pressure and the axial feeding) since they determine the occurrence of forming limits such as wrinkling and bursting. Therefore, these parameters should be carefully selected. In this paper an adaptive method is presented to obtain a more efcient process control for tube hydroforming processes. This method avoids the onset of wrinkling and bursting via dedicated stability criteria. The wrinkling criterion uses an energy-based indicator inspired on the plastic bifurcation theory. For necking, followed by bursting, a criterion based on the forming limit curve is employed. Applying these two criteria, the process parameters are adjusted during the simulation via a fuzzy knowledge based controller (FKBC). A case study is carried out for the hydroforming of a T-shape part using the designed adaptive system in combination with the nite element method. For the simulations ABAQUS/Explicit is used. 2004 Elsevier B.V. All rights reserved.
Keywords: Tube hydroforming; Process optimization; Adaptive simulation approach

1. Introduction Hydroforming is a manufacturing technology that was developed in the 1960s already. Widespread application, however, did not set in immediately. Today, hydroforming is still considered as a promising technology in forming. This technology, compared to conventional processes, still offers new possibilities with various elds of application. However, the effective use of the hydroforming technology in the manufacturing industry has triggered new challenges in terms of the boundary conditions specied for the forming process and the material properties. A classical example of a hydroforming process is visualized in Fig. 1. The process is controlled by two types of loads: an internal pressure pint and an axial force Fend . The hydroforming process and its limitations corresponding to different loading paths can schematically be represented, as depicted in Fig. 2. The shaded area in this gure is the process window. Any path lying within this window is expected

Corresponding author. Tel.: +31 40 247 58 94; fax: +31 40 244 73 55. E-mail address: a.aydemir@tue.nl (A. Aydemir). 0924-0136/$ see front matter 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2004.05.018

to result in a product without defects due to buckling or fracture. If the axial force is very high while the internal pressure is too low, buckling and wrinkling may occur. If the axial force is too low and the internal pressure is very high, the tube may burst. The virtual process window shown in Fig. 2 is relatively wide, while linear boundaries on both sides are assumed. However, for many processes there is often just a small non-linear corridor for the acceptable loading paths required to achieve a product without failure. Therefore, for a successful application of a hydroforming process, knowledge and control of the process limits is important, in addition to the selection of suitable components, materials and a proper material preparation. The conventional way of designing a hydroforming process starts with the denition of the basic parameters such as the die and tube geometries and the selection of the material. Then, a loading path is estimated to test the feasibility of the planned forming process. Basic information to ensure a stable production process and high-quality parts, as well as the knowledge of possible parameter combinations resulting in good or rejected parts, can be obtained via nite element (FE) simulations or real experiments [13]. Various scenarios can be attempted to obtain the required feeding and

304

A. Aydemir et al. / Journal of Materials Processing Technology 159 (2005) 303310

Fig. 1. Tube expansion process.

pressure histories. Especially for complex processes many trial-and-error cycles will be required to obtain an acceptable result. Therefore, the development of a new hydroforming process often deals with a considerable experimental effort and associated costs. Sometimes, the experiments will indicate that it is not possible to produce the part along any loading path. In that case, the load control parameters have to be changed, after which the whole procedure has to be repeated. Since learning via a trial-and-error method is costly in terms of money and time (especially for experiments but also for non-optimized simulations), an enhanced process simulation may offer better possibilities. The enhancement can be achieved by including a control algorithm in the process simulation to detect and avoid premature failure cases such as buckling, wrinkling or bursting and to adjust the process parameters, i.e. the internal pressure and the axial force, in an adequate way. A ow chart of this type of process design is shown in Fig. 3. This kind of process simulation is called an adaptive process simulation since the process parameters are continuously updated according to the new circumstances during the simulation. Currently, there is no unique method applicable for every type of geometry, to determine the appropriate process parameters. There are some analytical solutions available, but only for very simple geometries. However, several strategies can be used together with a FE analysis to obtain a suitable process plan. Most common used strategies are given below:

Fig. 3. Process design with enhanced simulation capabilities.

The conventional approach: The process parameter curve is estimated based on some simple metal forming equations. The self feeding approach: The process parameter curve is selected from a qualied family of curves. The starting point is to run an initial FE simulation without any forced axial feeding on the tube edges and with zero friction. Then, the simulations are repeated with modied parameters until a satisfactory result is achieved. The optimization approach: The process parameter curve is obtained from a series of simulations with an appropriate goal or criterion to be optimized, for example the thickness distribution. The adaptive approach: The process parameter curve is determined as the result of one single simulation with the help of an on-line control strategy. So far, optimization and adaptive approaches seem to be most promising in tube hydroforming process design, see for example [4,5]. In this paper an adaptive system is proposed to obtain adequate process parameters for hydroforming, with the incorporation of a wrinkle indicator, a necking indicator and a fuzzy knowledge based controller (FKBC). A novel approach is presented towards (adaptive) simulation of hydroforming processes, where the focus lies on the used methodology and not on the outline of a ready to use and fully validated optimization tool for industrial applications. The method will be applied to design the hydroforming process of a T-shaped component as an example. The results show that, in qualitative sense, an agreement with observations in practice has been achieved.

2. Adaptive simulation approach


Fig. 2. Process window allowing different loading paths (adopted from [6]).

An adaptive approach is based on the ability to early detect the onset and growth of defects, or the occurrence of

A. Aydemir et al. / Journal of Materials Processing Technology 159 (2005) 303310

305

Fig. 4. Scheme of the adaptive simulation procedure.

Fig. 5. Strategy for the adaptive nite element simulation.

unwanted situations during the process and promptly react to them. Several applications of this approach can be found in [68]. A simplied scheme of the approach is given in Fig. 4. Since the nite element method is applied using an incremental procedure, at the end of each increment there is a possibility to modify the load incrementations for the subsequent step. In this study, the loads are modied via the rate of the incremental internal pressure () and the rate of the incremental axial force (). The ultimate goal is the selection of a feasible loading path using a minimum number of simulations, or even within a single run. Numerical simulation based on the nite element method with an implicit or explicit integration scheme has become an important tool to simulate hydroforming processes involving a complicated geometry and complex boundary conditions, including friction [9]. When an implicit method is used to predict possible failure in a product, mostly, an initial imperfection, for example a specic mode shape and/or a material imperfection, is assigned to the original model. Unlike the implicit solver, the explicit method can generate failure, for example wrinkling, due to the accumulation of numerical errors without any physical origin. There are several commercial software packages, widely used in research centers as well as in industry, such as AUTOFORM [10], PAMSTAMP [11], etc., which are dedicated to analyze metal forming processes. Usually these packages do not present an extensive exibility for the implementation of user-supplied subroutines. Since the present research aims to pursue a new approach rather than to discuss the possibilities of dedicated commercial software, in this study the simulations are carried out using ABAQUS/Explicit [12]. This selection is motivated by the accessibility of this code and the user-oriented features for own developments. ABAQUS/Explicit can be applied in combination with the subroutines VUMAT and VDLOAD: in the subroutine VUMAT the strain and the stress states are calculated, in the subroutine VDLOAD the loads are modied. In designing an adaptive system for tube hydroforming processes, two important aspects in the whole procedure can be recognised: a reliable wrinkle predictor and an indicator for the probability of necking/fracture. The method proposed in this study readjusts the process parameters based

on some criteria at the end of each incremental solution, in order to avoid the growth of wrinkles and to delay possible bursting. The general strategy of the proposed method is to maximize the axial feeding and to minimize the internal pressure, while preventing the development of irreversible wrinkles and bursting. This ensures that the highest possible wall thickness in the product is obtained. The objective is realized by using the strategy given in Fig. 5. This strategy includes a wrinkle prediction method as well as a method to detect the possibility of necking of the tube material. The deformation state dened by the deformation gradient tensor F is sent to the subroutine VUMAT with some additional information necessary to evaluate the input for the controller. The input value Imin of the controller is the critical value supplied by the wrinkle indicator and the other input parameter d is the critical value of the necking indicator in the product. Then, the controller determines what the values of and will be, based on the algorithms and additional knowledge inserted in the controller. 2.1. Prediction of wrinkling In metal forming the concept of a wrinkle is used to denote short-waved out-of-plane deformations. Whenever the material is in a state of in-plane compression, there is a potential risk for wrinkles. The initiation and growth of wrinkles are inuenced by many factors such as the stress ratio, the mechanical properties, the geometry of the workpiece and the contact conditions. The analysis of the wrinkling initiation and growth considering all these factors is complicated because the associated effects are complex and the wrinkling behavior may show a wide scatter for small deviations of the conditions, as common in instability phenomena. Studying wrinkling, therefore, has been carried out case-by-case for a given process, and a generalised wrinkling criterion that can be used effectively for various processes has not been proposed and accepted yet [13]. There are several geometry-based wrinkle indicators where a wrinkle is detected from predominantly geometrical considerations. Unfortunately, most of them are for a specic product and they cannot be used directly for arbitrary geometries. The most often used procedure for wrinkle

306

A. Aydemir et al. / Journal of Materials Processing Technology 159 (2005) 303310

prediction in metals is based on the plastic bifurcation theory [14]. Although this theory is commonly used, the integration into FE codes in its classic formulation is not trivial. Inspired by the plastic bifurcation theory, a computationally cheap and general-purpose wrinkling detection procedure is given by Nordlund [15]. This wrinkle indicator uses the stress state and the deformation to detect the wrinkle via the local value of the second-order increment of internal work. If the material point is under compression and if additionally the deformation is dominated by rotation then most likely a wrinkle will occur. However, there are some cases where these type of situations are dealt with without the occurrence of any wrinkle. Therefore, this wrinkle indicator fails if a pure global rigid body rotation takes place in the process, which can hardly be expected in a real process, or if a large scale wrinkle starts to evolve [8,16]. The associated indicator quantity Imin has, in spite of its disadvantages, several benecial properties for use in an adaptive FE simulation. For example, it is generally applicable to all loading situations, including cases where contact forces and large elastoplastic deformations are involved. Furthermore, because its value is bounded in the interval of 1 and +1, the classication of a wrinkle is relatively easy compared to other wrinkle indicators. Therefore, the wrinkle indicator dened by Nordlund is used in the present study in an adaptive framework. Note that other wrinkling criteria may be used as well, not inuencing the methodology presented further on. 2.2. Prediction of necking Bursting types of failure are related to the material formability and these phenomena are difcult to quantify since they depend on the material used and the stress and strain conditions imposed on the workpiece. Especially for metals, necking occurs rst and is followed by bursting after a continued forming process. Therefore, if necking can be detected bursting might be prevented. For necking detection several methods may be used [17,18]. The most popular one is based on the actual strain state and the forming limit curve (FLC) using experimental data. FLCs are mostly assembled from the data obtained via several experiments with different linear strain paths. Since formability is strongly dependent on the deformation history, experimental determination of the FLCs is expensive. Therefore, also studies to determine FLCs from analytical expressions [19], or from FE simulations [20] are often considered in practice. In this study, an input term for the controller is designed by calculating the shortest distance d of the strain state of each material point characterised by P to the materials FLC, see Fig. 6. 2.3. The controller Hydroforming is a rather complex deformation process. Therefore, the derivation of a direct analytical relation be-

Fig. 6. A typical forming limit diagram and the denition of d.

tween the input and output variables is impossible for practically realistic congurations. Since a lot of non-linearities are involved in a hydroforming process, even for simple geometries, many assumptions are needed to arrive at an analytical approximation. Additionally, there may be many physical features of the process that are not incorporated in the analytical model used to design the controller, due to the complexity of the process or a lack of understanding of the physics involved. On the other hand, experienced process operators usually possess a number of heuristic design rules which allow them to control a process in a satisfactory manner. As it is extremely difcult to develop techniques which cast a heuristic control policy into analytical expressions, there are other control techniques that circumvent the use of analytical models. A particular expert control technique employing the heuristic knowledge available is known as the so-called fuzzy knowledge based controller. No further details on FKBCs will be presented here, but more information can be found, for example, in [21,22]. In this study, a FKBC is designed for tube hydroforming processes. When the subroutine VUMAT is called, the values of the wrinkle indicator Imin and the necking indicator d are determined. These two values constitute the input to the controller. Then, the controller calculates the rates of the incremental internal pressure and the incremental axial force based on the expert knowledge inserted in it. After that, when the subroutine VDLOAD is called, the load increments will be calculated using the recently determined and . In the following, the designed FKBC for tube hydroforming processes will be outlined. The controller prepares the output in order to modify the loads for the next increment. The deformation state is checked for wrinkles, as well as for necking. In the example presented further on, the knowledge base used is primarily deterministic, since it includes rules relating the deformation state to the output. However, heuristic rules from practice can be added to it without any difculty. 2.3.1. Internal pressure An internal pressure pint is applied to the inner surface of the tube. For the internal pressure the update formula given below is used:

A. Aydemir et al. / Journal of Materials Processing Technology 159 (2005) 303310

307

pnew = pold + pint int int

(1)

where pint is a predened value, pold the value from the int previous increment and is considered as a fuzzy number with 0 1. For the following fuzzy sets are used in the controller: is ZERO (Z), is LOW (L), is HIGH (H). Each fuzzy set is dened by a membership function. A membership function may have different shapes, for example triangular, L-type, etc. In the present implementation only triangular, L- and -types of membership functions are used. Necessary information on the membership functions can be found in almost any book written on fuzzy sets, for example in [21]. 2.3.2. Axial feeding For modeling purposes, the axial feeding is obtained via an axial force Fend applied to the rigid tools (so-called rams) at both ends of the tube, see Fig. 1. Contact conditions are dened such that direct interaction between the die and the rams is avoided. The contact between the tube end and the ram is supposed to be frictionless. The axial force is modied according to:
new old Fend = Fend + Fend

Fig. 7. The rule base for the / controller.

plementation, the FLC, which is needed to determine d, is obtained using the denitions in [19]. The material properties of the hydroformed tube will be given in the next sections. 2.3.5. The rule base The rules to be implemented are usually derived from the experience-based knowledge of the process operator and/or control engineer. This is the rst step to start with. This step helps to provide an initial prototype version of the knowledge base. Consequent tuning of the membership functions and rules is a necessary next step. In this study, the tuning is performed based on FE simulations. The whole set of collected rules is given in Fig. 7. Each cell in Fig. 7 represents a rule. For example, the rule in the upper left cell can be formulated as: IF the wrinkle is very critical (controlled by Imin ) AND the deformation is very close to necking (controlled by d) THEN is zero (Z) and is negative (N). Accordingly, nine rules for both and can be distinguished. The set of rules is complete since none of the cells is empty. The rules ensure that as long as wrinkling is not critical additional material is fed into the die cavity, without increasing the internal pressure. As soon as a critical wrinkle is detected the internal pressure is increased by some value. To determine the output values both the wrinkle indicator and the FLD-distance are taken into account. 3. A case study: hydroforming of T-shapes

(2)

old where Fend is a predened value, Fend the value from the previous increment and 1 1. Accordingly, the axial force is allowed to increase or decrease. Additionally, Fend is limited by a minimum force value to prevent leakage of the internal uid. Here, is considered as a fuzzy number. For the fuzzy sets given below are used in the controller.

is NEGATIVE (N), is ZERO (Z), is POSITIVE (P). 2.3.3. The wrinkle indicator For the wrinkle indicator, three membership functions are dened and used in the controller. Each membership function classies a wrinkle as given below: the wrinkle is very critical (veryCR), the wrinkle is critical (CR), the wrinkle is not critical (notCR). 2.3.4. The necking indicator Associated to the necking indicator d three membership functions are specied, according to: the distance d is very close (veryCL), the distance d is close (CL), the distance d is not close (notCL). In [19], a method is given to obtain the forming limits of tubular hydroforming based on plastic instability. In the im-

The wrinkle indicator, the necking indicator and the controller are implemented into user subroutines additional to the user subroutines VUMAT and VDLOAD. To evaluate the indicators stress and strain states are necessary. In the used version of ABAQUS/Explicit, the user had no direct access to the necessary stress information to evaluate the wrinkle indicator. The only way to obtain the required information was application of the user subroutine VUMAT. Once VUMAT is invoked, the stresses can be calculated via a constitutive law. Therefore, although the assumed material behavior is standard, it appeared to be necessary to implement it inside the subroutine VUMAT. 3.1. Case description As a well-known case study, hydroforming of a T-shaped component, which is one of the most studied parts among

308

A. Aydemir et al. / Journal of Materials Processing Technology 159 (2005) 303310

Fig. 8. An assembled view of the parts involved in the process.

T-branching tubes, is simulated. Different variants of this T-shape can be distinguished. For example, the bulged part may be inclined (Y-shaped, see [4]), bulgings may occur on the top and bottom sides of the tube (cross-joint, see [23]) and the bulging location may not be in the center of the blank (see [24]). In this paper, the simulated part contains a central bulging on the top side. In spite of the simplicity, it contains the basic features of tube hydroforming with two independent loading parameters. The ultimate goal is to manufacture the part with a maximum of material in the die cavity. During the process, the occurrence of any kind of irreversible wrinkles should be prevented while necking is delayed as much as possible. The tubes initial length, external diameter and wall thickness are 190, 42.2 and 1.4 mm, respectively. The nal bulge height is xed at 12 mm by using a stopper, see Fig. 8. The friction coefcient between the tube and the die is assumed to be 0.1. In the FE model, the double symmetry is taken into account, so only a quarter part of the tube is actually analyzed. The tube is modeled using three-dimensional linear brick elements. In the tube thickness direction two layers of elements are used. The tools, i.e. the die and the rams, are modeled by an appropriate set of discrete rigid nite elements. In tube hydroforming simulations, adequate material properties are of prominent importance for the reliability of the results. Principally, the material properties of the original sheet of which the tube is rolled are not applicable anymore as the rolling process will induce some changes: the stressstrain behavior of the tubular blank will be different from the response of the original at sheet. Nevertheless it is important to use a realistic material description in the simulations. Further discussions on this subject can be found, for example, in [2527]. For the present purpose, the material data have been assessed from several test specimens (strips) prepared from the blank tube, based on the performance in standard tensile experiments. The material properties measured for the tube material (steel, St 34) are specied below in the modeling context of a von Mises yield criterion combined with an isotropic power-law hardening description: Youngs modulus Poissons ratio Strength Pre-strain Hardening exponent E = 200 GPa = 0.3 K = 603 MPa 0 = 0.0076 n = 0.226

Fig. 9. Obtained process plan.

The actual ow stress is dened by: f = K(


0

+ )n

(3)

where is the equivalent plastic strain. The FLC is evaluated using the material model and the material parameters specied above, according to the procedure as discussed by Xing and Makinouchi in [19]. 3.2. Results The computationally derived process plan is given in Fig. 9. The process starts with a little axial feeding. Then, almost a linear path is followed until about 16 MPa internal pressure. Then, for a short period, the adaptive system is not able to feed more material into the die cavity and only the internal pressure is increased. Thereafter, the axial feeding starts to increase, and in combination with the internal pressure again following an almost linear path. The increase in the internal pressure results in a strain state which becomes close to necking. Once, the FLC is closely approached, the internal pressure is not allowed to increase anymore. At the moment that the controller detects not to be able to increase the internal pressure or the axial feeding the simulation is stopped. Forming limit curves are commonly used to check whether the strains exceed the materials forming capacity. For the simulated process, the forming limit diagram is shown in Fig. 10. The line with triangular markers is the FLC for the tube material. The dots represent the strain state of each material point at the end of simulation. Fig. 10 demonstrates that at the end of the simulation the FLC is closely approached indeed. The deformed geometry of the tube is shown in Fig. 11, together with the thickness distribution. Some minor thinning occurs in the bulged part of T-shape. In the other regions the loads result in thickening of the tube wall. Visual defects on the product can be observed.

A. Aydemir et al. / Journal of Materials Processing Technology 159 (2005) 303310

309

Fig. 10. The forming limit diagram at the end of the simulation.

Although the process plan has been obtained at the end of one single simulation, generally, in the rst attempt it might be difcult to arrive at an acceptable solution. Therefore, a number of introductory simulations are generally required before the nal simulation run can be performed. This initialization is required because of the fact that the membership functions need to be ne-tuned. This ne-tuning can be realized via experiments, as well as simulations. In the present paper, the latter procedure was pursued. The approach given in this paper has been elaborated for a simple case study. Evidently, validation of the method by performing parallel experiments still needs to be done, which is an issue to consider in the future. Nevertheless, the results show a qualitative agreement with experiences in practice, see for example [7].

Acknowledgements The results presented originate from a Dutch research project on hydro-mechanical forming that is carried out as a co-operation between Eindhoven University of Technology and TNO Industrial Technology. The funding by the Ministry of Education, Culture and Sciences (OC&W) is gratefully acknowledged.

References
[1] F. Dohmann, C.h. Hartl, Tube hydroforming-research and practical application, J. Mater. Process. Technol. 71 (1997) 174186. [2] L. Gao, S. Motsch, M. Strano, Classication and analysis of tube hydroforming processes with respect to adaptive FEM simulations, J. Mater. Process. Technol. 129 (2002) 261267. [3] W.H. Sillekens, P.P.H. Veltmans, M.A. Guiterrez, J.I. Fernandez, Factorial experimentation and simulation of hydroforming processes for tubular parts, in: Proceedings of the Ninth International Conference on Sheet Metal, 2001, pp. 605612. [4] T. Altan, S. Jirathearanat, M. Strano, S.G. Shr, Adaptive FEM process simulation for hydroforming tubes, in: K. Siegert (Ed.), Hydroforming of Tubes, Extrusions and Sheet Metals, vol. 2, 2001, pp. 363384. [5] M. Strano, S. Jirathearanat, S.-G. Shr, T. Altan, Virtual process development in tube hydroforming, J. Mater. Process. Technol. 146 (2004) 130136. [6] E. Doege, R. Kosters, C. Ropers, Determination of optimised control parameters for internal high pressure forming processes with the FEM, in: Proceedings of the Sixth International Conference on Sheet Metal, vol. 2, 1998, pp. 119128. [7] W.H. Sillekens, R.J. Werkhoven, Hydroforming processes for tubular parts: optimisation by means of adaptive and iterative FEM simulation, Int. J. Forming Process. 4 (2001) 377393. [8] M. Strano, S. Jirathearanat, T. Altan, Adaptive FEM simulation for tube hydroforming: a geometry-based approach for wrinkle detection, Ann. CIRP 50/1 (2001) 185190. [9] J. Kim, Y.-H. Kang, H.-H. Choi, S.-M. Hwang, B.-S. Kang, Comparison of implicit and explicit nite-element methods for the hydroforming process of an automobile lower arm, Int. J. Adv. Manuf. Technol. 20 (2002) 407413.

Fig. 11. The deformed geometry and thickness distributions along the top and bottom sides.

4. Conclusions and future work In this study, an adaptive method is designed for tube hydroforming. A criterion for wrinkling, as well as a criterion for necking, is included. Instead of exact statements as wrinkle yes/no, the criteria are formulated in weak terms and rules resulting in a fuzzy knowledge based controller. The applicability of this controller to design a feasible hydroforming process has been demonstrated.

310

A. Aydemir et al. / Journal of Materials Processing Technology 159 (2005) 303310 tional ESAFORM Conference on Material Forming, 1999, pp. 469 472. H.L. Xing, A. Makinouchi, Numerical analysis and design for tubular hydroforming, Int. J. Mech. Sci. 43 (2001) 10091026. E.M. Viatkina, W.A.M. Brekelmans, L.P. Evers, M.G.D. Geers, Forming limit diagrams for sheet deformation processes: a crystal plasticity approach, in: Proceedings of the Fourth International ESAFORM Conference on Material Forming, 2001, pp. 465468. D. Driankov, H. Hellendoorn, M. Reinfrank, An Introduction to Fuzzy Control, Springer-Verlag, 1993. K.M. Passino, S. Yurkovich, Fuzzy Control, Addison-Wesley, Longman Inc., 1998. B.J. Mac Donald, M.S.J. Hashmi, Finite element simulation of bulge forming of a cross-joint from a tubular blank, J. Mater. Process. Technol. 103 (2000) 333342. M. Ahmetoglu, K. Sutter, X.J. Li, T. Altan, Tube hydroforming: current research, J. Mater. Process. Technol. 98 (2000) 224231. T. Altan (Ed.), Research papers on sheet forming, machining and tube hydroforming, J. Mater. Process. Technol. 146 (1) (2004) 1143. M. Ahmetoglu, T. Altan, Tube hydroforming: state-of-the-art and future trends, J. Mater. Process. Technol. 98 (2000) 2533. T. Altan, M. Koc, Y. Aue-u-lan, K. Tibari, Formability and design issues in tube hydroforming, in: K. Siegert (Ed.), Hydroforming of Tubes, Extrusions and Sheet Metals, vol. 1, 1999, pp. 105121.

[10] W. Kubli, J. Reissner, Optimization of sheet-metal forming processes using the special-purpose program AUTOFORM, J. Mater. Process. Technol. 50 (1995) 292305. [11] F. Heislitz, H. Livatyali, M.A. Ahmetoglu, G.L. Kinzel, T. Altan, Simulation of roll forming process with the 3-D FEM code PAM-STAMP, J. Mater. Process. Technol. 59 (1996) 5967. [12] ABAQUS Inc., ABAQUS/Explicit Users Manual, Version 6.3, 2002. [13] J.B. Kim, J.W. Yoon, D.Y. Yang, F. Barlat, Investigation into wrinkling behaviour in the elliptical cup drawing process by nite element analysis using bifurcation theory, J. Mater. Process. Technol. 111 (2001) 170174. [14] R. Hill, A general theory of uniqueness and stability in elasticplastic solids, J. Mech. Phys. Solids 6 (1958) 236249. [15] P. Nordlund, Adaptivity and Wrinkle Indication in Non-Linear Shell Analysis, Doctoral Dissertation, Department of Solid Mechanics, Royal Institute of Technology, Stockholm, Sweden, 1997. [16] A. Aydemir, A Numerical-Experimental Analysis of Tube Hydroforming, MTD Thesis, Eindhoven University of Technology, ISBN 90-444-0279-X, 2003. [17] D. Banabic (Ed.), Formability of Metallic Materials: Plastic Anisotropy, Formability Testing, Forming Limits (Engineering Materials), Springer-Verlag, 2000. [18] M. Dutilly, J.C. Gelin, Design of sheet metal forming processes based on quality functions, in: Proceedings of the Second Interna-

[19] [20]

[21] [22] [23]

[24] [25] [26] [27]

You might also like