導(dǎo)向板沖孔落料彎曲復(fù)合模模具設(shè)計,導(dǎo)向板沖孔落料彎曲復(fù)合模模具設(shè)計,導(dǎo)向,沖孔,彎曲,曲折,復(fù)合,模具設(shè)計 Int J Adv Manuf Technol (2002) 19:587596 Ownership and Copyright Springer-Verlag London Ltd 2002 Influence of Punch Radius and Angle on the Outward Curling Process of Tubes You-Min Huang and Yuung-Ming Huang Department of Mechanical Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan Using the theory of updated Lagrangian formulation, this study adopted the elasto-plastic finite-element method and extended the increment determination method, the r min method, to include the elements yielding, nodal contact with or separation from the tool, maximum strain and limit of rotation increment. The computer code for a finite-element method is established using the modified Coulombs friction law. Conical punches with different radii and angles are used in the forming simulation of hard copper and brass tube ends. The effects of various elements including the half-apex angle of punch (H9251) and its radius (R), the ratio of the thickness of the tube wall to the mean diameter of tube, mechanical properties, and lubrication on the tubes outward curling, are investigated. Simulation findings indicate that when the bending radius at the punch inlet (H9267) satisfies the condition of H9267 H11017 H9267 c , curling is present at the tube end. On the other hand, if H9267 H11084 H9267 c , the tube end experiences flaring. The variable H9267 c is called the critical bend- ing radius. The value of H9267 c increases as the value of H9251 increases. Furthermore, the findings also show that H9267 c is neither correlated with tube material nor lubrication. Keywords: Elasto-plastic; Finite elements; Half-apex angle; Outward-curling process 1. Introduction The process of forming convex edges of a metal tube is often employed for connecting two tube parts, linking and locking a tube part and its components, and connecting fluid pipelines or combining complementary pipelines with reinforcement at the tubes end. It is a common industrial technology related to tube ends. The tube end processes are generally divided into tapper flaring, flange flaring, step flaring, and curl forming. The tube-curling deformation discussed in this paper curls the opening of the tubes end outward in a circle. The simulation Correspondence and offprint requests to: Dr You-Min Huang, Depart- ment of Mechanical Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan. E-mail: ymhuangL50560mail.ntus- t.edu.tw of the tube-curling deformation is a complex and difficult task because the deformation process is highly nonlinear. The nonlinear deformation characteristic is due to: 1. The large displacement, rotation, and deformation during metal deformation. 2. The nonlinear material deformation behaviour when metal material experiences large deformation. 3. The nonlinear boundary condition generated by the friction between the metal and tool interface, and their contact con- ditions. These characteristics made the finite-element method the most widely used of the metal process analyses. To improve the process and increase industrial productivity, this study developed an elasto-plastic finite-element computer code using the selective reduced integration (SRI) simulation method, which employs four integral point elements within the four nodes of a rectangle. The objective is to simulate the tube- curling deformation process. Nadai 1 studied tube-nosing in 1943. The theoretical induc- tion was an extension of the curved shell theory. Nadai assumed the friction coefficient to be constant, and ignored the presence of effective stress variation in the shell. Cruden and Tompson 2 conducted a series of tube-nosing experiments to establish the various limitations of tube-nosing and evaluate the effects of various parameters. Manabe and Nishimura 37 also con- ducted a series of experiments on both conical tube-flaring and tube-nosing processes to investigate the effects of various parameters on the forming load and stressstrain distribution. The parameters studied include the conical punch with different angles, lubrication, material and tube wall thickness. Tang and Kobayashi 8 proposed a rigid-plastic finite-element theory and developed a computer code to simulate the cold-nosing of 105 mm AISI 1018 steel. Huang et al. 9 simulated the cold- nosing process through elasto-plastic and rigid-plastic finite- element methods and compared the simulation results with experimental data to verify the accuracy of the rigid-plastic finite element. Kitazawa et al. 10,11 conducted experiments with conical punches, and copper and brass materials with the assumption that the tube materials were rigid-perfectly plastic. They had explored previously the effect of the arc radius of 588 You-Min Huang and Yuung-Ming Huang the punch, angle, and tube wall thickness on the tube-curling deformation process. Kitazawa 12 used carbon steel, copper, and brass as the tube materials in his experiment of tube- curling deformation. When the flaring energy increment at the front edge of the tube end (H9254W f ) is greater than the energy increment at the curl edge (H9254W c ), the tube end material undergoes outward curling forming because of the absence of unbending. On the other hand, if H9254W c H11084H9254W f , the tube end material is attached to the conical punch surface because of unbending and fails to curl or form flaring. The energy rule was used to induce the deformation energy at the time of tube forming for the purpose of comparison and verification, thus establishing the criterion of outward curling. In this study, the same material constants and dimensions used in other studies 12 are adopted in the simulation of the outward curling process. The findings are compared with the results reported in 12 to verify the accuracy of the elasto- plastic finite-element computer code developed. 2. Description of the Basic Theory 2.1 Stiffness Equation Adopting the updated Lagrangian formulation (ULF) in the framework of the application of incremental deformation for the metal forming process (bulk forming and sheet forming) is the most practical approach for describing the incremental characteristics of the plastic flow rule. The current configuration in ULF at each deformation stage is used as the reference state for evaluating the deformation for a small time interval H9004t such that first-order theory is consistent with the accuracy requirement. The virtual work rate equation of the updated Lagrangian equation is written as Fig. 1. Boundary conditions for the deformation tubes geometry of outward curling at an initial and a particular stage. Units, mm; R, punch radius; H9251, half-apex punch angle; H9267, bending radius at punch inlet. H20885 V (H9270 * ij 2H9268 ik H9280 kj ) H9254H9280 ij -V + H20885 V H9268 jk L ik H9254L ij -V = H20885 S f tH6032 i H9254v i -S (1) where H9270 * ij is the Jaumann rate of Kirchhoff stress, H9268 ij is the Cauchy stress, H9280 ij is the rate of deformation which is the Cartesian coordinate, v i is the velocity, tH6032 i is the rate of the nominal traction, L ij (= H11128v i /H11128X j ) is the velocity gradient, X j is the spatial fixed Cartesian coordinate, V and S f are the material volume and the surface on which the traction is prescribed. The J 2 flow rule H9270 * = E 1+v H9254 ik H9254 jl + v 1 2v H9254 ij H9254 kl 3H9261(E/(1+v) H9268 H11032 ij H9268 H11032 kl 2H9268 2 (0H H11032 + E/(1 + v) H9280 kl (2) is employed to model the elasto-plastic behaviour of sheet metal, where HH11032 is the strain hardening rate, H9268 is the effective stress, E is the Youngs modulus, v is the Poissons ratio, H9268 H11032 ij is the deviatoric part of H9268 ij . H9261 takes 1 for the plastic state and 0 for the elastic state or the unloading. It is assumed that the distribution of the velocity vina discretised element is v = Nd (3) where N is the shape function matrix and d denotes the nodal velocity. The rate of deformation and the velocity gradient are written as H9280 = Bd (4) L = Ed (5) where B and E represent the strain ratevelocity matrix and the velocity gradientvelocity matrix, respectively. Substituting Eqs (4) and (5) into Eq. (1), the elemental stiffness matrix is obtained. As the principle of virtual work rate Eq. and the constitutive relationship are linear Eq. of rates, they can be replaced by increments defined with respect to any monotonously increasing measure, such as the tool-displacement increment. Following the standard procedure of finite elements to form the whole global stiffness matrix, KH9004u = H9004F (6) in which K = H20888 H20855eH20856 H20885 VH20855eH20856 B T (D ep Q) B-V + H20888 H20855eH20856 H20885 VH20855eH20856 E T GE-V H9004F = H20873H20888 H20855eH20856 H20885 SH20855eH20856 N T tH6032-S H20874 H9004t In these equations, K is the global tangent stiffness matrix, D ep is the elemental elasto-plastic constitutive matrix, H9004u denotes the nodal displacement increment, and H9004F denotes the prescribed nodal force increment. Q and G are defined as stress-correction matrices due to the current stress at any stage of deformation. Influence of Punch Radius and Angle on Tube Curling 589 Fig. 2. (a) Deformed geometries, and (b) nodal velocity distributions in the outward curling of tubes at 12 different forming stages. Hard copper tube, n = 0.05; t 0 = 0.8 mm; H9251 = 60; R = 3.4 mm. 2.2 SRI Scheme As the implementation of the full integration (FI) scheme for the quadrilateral element leads to an excessively constraining effect when the material is in the nearly incompressible elasto- plastic situation in the forming process 13, Hughes proposed that the strain ratevelocity matrix be decomposed to the dilation matrix B dil and the deviation matrix B dev , i.e. B = B dil +B dev (7) in which the matrices B dil and B dev are integrated by conven- tional four-point integration. When the material is deformed to the nearly incompressible elasto-plastic state, the dilation matrix B dil must be replaced by the modified dilation matrix B dil that is integrated by the one point integration, i.e. B = B dil +B dev (8) in which B is the modified strain ratevelocity matrix. Substi- tuting Eq. (8) into Eq. (7), the modified strain ratevelocity matrix is B = B+(B dil B dil ) (9) Explicitly, the velocity gradientvelocity matrix E is replaced by the modified velocity gradientvelocity matrix E E = E+(E dil E dil ) (10) 3. Numerical Analysis The analytical model of the tube-curling process is axially symmetric. Thus, only the righthand half of the centre axis is considered. Division of the parts finite elements is automati- cally processed by the computer. Since there is drastic defor- mation from the bending and curling at the tubes end, a finer element division is required for this section in order to derive precise computation results. The lefthand half shown in Fig. 1 denotes the sizes of the part and die at the beginning. The detailed dimensions are given in Table 1. In the local coordi- nates, axis 1 denotes the tangential direction of the contact between tube material and the tool, while axis n denotes the normal direction of the same contact. Constant coordinates (X,Y) and local coordinates (l, n) describe the nodal force, displacement and elements stress and strain. 590 You-Min Huang and Yuung-Ming Huang Fig. 3. As Fig. 2, but R = 3.8 mm. Table 1. Punch angle and radius. Punch apex Punch radius R (mm) angle H9251 (deg.) 60 2.0, 2.4, 2.8, 3.1, 3.4, 3.8, 4.1, 4.4, 4.8, 5.1 65 2.4, 2.8, 3.1, 3.4, 3.8, 4.1, 4.4, 4.8, 5.5, 5.4 70 3.1, 3.4, 3.8, 4.1, 4.4, 4.8, 5.1, 5.4, 5.8, 6.1 75 3.4, 3.8, 4.1, 4.4, 4.8, 5.1, 5.4, 5.8, 6.1, 6.4 80 4.8, 5.4, 5.8, 6.1, 6.4, 6.8, 7.2, 7.6, 8.0, 8.4 85 5.8, 6.1, 6.4, 6.8, 7.4, 9.0, 9.4, 9.8, 10.2, 10.6, 11.0 Table 2 gives the material conditions of our simulation. The exterior radius of the tube remains unchanged at 25.4 mm; but there are three different tube wall thickness values, namely, 0.4, 0.6 and 0.8 mm, used in the experiment and computation. The Poissons ratio and Youngs modulus of the hard copper tube are 0.33 and 110740 MPa, respectively. The Poissons ratio and Youngs modulus of brass are 0.34 and 96500 MPa, respectively. 3.1 Boundary Condition The righthand half of the tube shown in Fig. 1 denotes the deformation shape at a certain stage during the tube-curling Table 2. Mechanical properties of materials used. Materials Outer Heat treatment nF Yield stress diameter in vacuum (Mpa) H9268 0.2 (Mpa) thickness Copper 25.4 0.8 As received 0.09 380 220 500C 1 h 0.53 630 26 As received 0.05 450 280 300C 1 h 0.09 450 260 400C 1 h 0.46 610 50 600C 1 h 0.50 640 42 70/30 Brass 25.4 0.8 As received 0.18 730 280 H9268 = FH9280 n ; H9268 = stress; H9280 = strain. deformation. The boundary conditions include the following three sections: 1. The boundary on the FG and BC sections: H9004f l HS33527 0, H9004f n HS33527 0, H9004v n = H9004v n where H9004f l is the nodal tangential friction forces increment, and H9004f n is the normal force increment. As the materialtool contact area is assumed to involve friction, H9004f l and H9004f n are not equal to zero. H9004v n , which denotes the nodal displacement Influence of Punch Radius and Angle on Tube Curling 591 increment in the normal direction of the profile of the tools, is determined from the prescribed displacement increment of the punch H9004v n . 2. The boundary on the CD, DE, EF, and GA sections: H9004F x = 0, H9004F y = 0 The above condition reflects that the nodes on this boundary are free. 3. The boundary on the AB section: H9004v x = 0, H9004v y = 0 The AB section is the fixed boundary at the bottom end of the tube. The displacement increment of the node at this location along the y-axis direction is set at zero, whereas node B is completely fixed without any movement. As the tube-curling process proceeds, the boundary will be changed. It is thus necessary to examine the normal force H9004f n of the contact nodes along boundary sections FG and BC in each deformation stage. If H9004f n reaches zero, then the nodes will become free and the boundary condition is shifted from (1) to (2). Meanwhile, the free nodes along the GA and EF sections of the tube are also checked in the computation. If the node comes into contact with the punch, the free-boundary condition is changed to the constraint condition (1). 3.2 Treatment of the Elasto-Plastic and Contact Problems The contact condition should remain unchanged within one incremental deformation process, as clearly implied from the interpretation in the former boundary condition. In order to satisfy this requirement, the r-minimum method proposed by Yamada et al. 14 is adopted and extended towards treating the elasto-plastic and contact problems 15. The increment of each loading step is controlled by the smallest value of the following six values. 1. Elasto-plastic state. When the stress of an element is greater than the yielding stress, r 1 is computed by 15 so as to ascertain the stress just as the yielding surface is reached. 2. The maximum strain increment. The r 2 term is obtained by the ratio of the defaulted maximum strain increment H9274 to the principal strain increment dH9280, i.e. r 2 = H9274/dH9280, to limit the incremental step to such a size that the first-order theory is valid within the step. 3. The maximum rotation increment. The r 3 term is calculated by the defaulted maximum rotation increment H9252 to the rotation increment dH9275, i.e. r 3 = H9252/dH9275, to limit the incremen- tal step to such a size that the first-order theory is valid within the step. 4. Penetration condition. When forming proceeds, the free nodes of the tube may penetrate the tools. The ratio r 4 16 is calculated such that the free nodes just come into contact with tools. 5. Separation condition. When forming proceeds, the contact nodes may be separated from the contact surface. The r 5 term 16 is calculated for each contact node, such that the normal component of nodal force becomes zero. 6. Sliding-sticking friction condition. The modified Coulombs friction law provides two alternative contact states, i.e. sliding or sticking states. Such states are checked for each contacting node by the following conditions: v rel(i) l v rel(Il) l H11084 0 (a) if H20841v rel(i) l H20841 H11022 VCRI, then f 1 = H9262f n , the node is in sliding state (b) if H20841v rel(i) l H20841 H11021 VCRI, then f 1 = H9262f n (v rel l /VCRI), the node is in quasi-sticking state. v rel(i) l v rel(I1) l H11021 0. The direction of a sliding node is opposite to the direction of the previous incremental step, making the contact node a sticking node at the next incremental step. The ratio r 6 is then obtained here, r 6 = Tolf/H20841v rel(i) l H20841, which produces the change of friction state from sliding to sticking, where Tolf (= 0.0001) is a small tolerance. The constants of the maximum strain increment H9274 and maximum rotation increment H9252 used here are 0.002 and 0.5, respectively. These constants are proved to be valid in the first- order theory. Furthermore, a small tolerance in the check pro- cedure of the penetration and separation condition is permitted. 3.3 Unloading Process The phenomenon of spring-back after unloading is significant in the tube-forming process. The unloading procedure is executed by assuming that the nodes on the bottom end of the tube are fixed. All of the elements are reset to be elastic. The force of the nodes which come into contact with the tools is reversed in becoming the prescribed force boundary condition on the tube, i.e. F =F. Meanwhile, the verification of the penetration, friction, and separation condition is excluded in the simulation program. 4. Results and Discussion Figure 2 shows the simulation results of curling forming of copper tubes with the punch semi-angle, H9251 = 60, and the arc radius, R = 3.4 mm. Figure 2(a) denotes the geometric shape of the deformation, in which the last shape is the final shape of the part after unloading. Figure 2(b) shows the nodal velocity distribution during deformation. The last diagram shows the nodal velocity distribution after unloading. These figures show that the tube end material enters smoothly along the punch surface at the initial stage. After that, the tube end material gradually bends outward and curls up, resulting in the so- called outward curling forming. Figure 3 shows the simulation result in the case of R = 3.8 mm. The tube end material still shows the so-called flaring forming after passing the arc induc- tion section and conical face forming. Figure 4 shows the result of curling simulation of copper tubes in the case of H9251 = 75 and R = 6.1 mm. In contrast, Fig. 5 shows the simulation result of flaring forming in the case of R = 6.4 mm. Figure 6 shows the strain distributions corresponding to those of Figs 2 and 3 at a punch travel of 12 mm. Figure 7 592 You-Min Huang and Yuung-Ming Huang Fig. 4. As Fig. 2, but H9251 = 75; R = 6.1 mm. shows the strain distributions corresponding to those of Figs 4 and 5 at a punch progress of 14 mm. The deformation is mostly the result of pulling in the circumference direction, bending in the meridian direction and shearing in the thickness direction. The maximum diameter of the tube end is the flaring forming. Thus, its strain value in the circumference direction is far greater than that of outward curling. The reduction in tube wall thickness is also more significant in flaring forming. Thus, under the same progress, fracture of the tube end occurs more easily in flaring forming than in other situations. Figures 8 and 9 show the simulation results for hard copper and brass, respectively. The same tube wall thickness, tube radius, lubrication condition, and half-apex of the punch are used in both simulations. However, different values of the punch arc radius are used, which generate different types of forming, namely, flaring and curling. That is, if the arc radius, H9267, at the punch ent