Journalof ELSEVIER Journal of Materials Processing Technology 59 (1996) 95-105 Materials Processing Technology Investigation of metal flow and preform optimization in flashless forging of a connecting rod T e r u i e T a k e m a s u a, V i c t o r V a z q u e z b'', B r e t t P a i n t e r b, T a y l a n A l t a n b aMechanical Engineering for Intelligent Machinery and Systems, Kyushu University, Japan bERC for Net Shape Manufacturing, The Ohio State University, Columbus, OH 43210, USA Industrial Summary In conventional hot forging of connecting rods, the material wasted to the flash accounts approximately 20 to 40% of the original workpiece In order to reduce the cost of forged products, the forging must be performed in a closed cavity to obtain near-net or net shape parts In flashless forging, the volume distribution of t h e preform must be accurately controlled to avoid overloading the dies and to fill the cavity Additionally, t h e preform must be simple enough to be mass produced This study deals with the design of the optimum preform to forge a connecting rod without flash The initial preform design was obtained from physical modeling experiments The optimization of this preform was found through 3D FEM process simulations The advantage of performing simulations is that no tooling has to be built and the number of experimental tryouts can be significantly reduced A preform optimization methodology was derived for this investigation Introduction If the weight of a connecting rod (see Fig 1) can be reduced while increasing its strength, an automobile's fuel efficiency will be improved Currently, steel connecting rods are used in passenger cars However, some manufacturers have attempted to use alternative lighter materials Recently, various composite materials based cn aluminum have been considered, but not yet successfully adopted, for automotive engines The main reasons are that these materials are not strong enough, or when strong enough, are too expensive Flashless forging offers the possibility of producing aluminum composite connecting rods at competitive costs The design of flashless forging processes is more complex than the design of conventional closed die forging with flash Therefore, in order to accelerate the development of the manufacturing process as well as to reduce t h e development costs, a new design method must be developed and applied The Finite Element Method (FEM) offers the possibility to design the entire manufacturing process on a computer This leads to a "Corresponding author 0924-0136/96/$15.00 © 1996 Elsevier Science S.A All rights reserved PI10924-0136 (96) 02290-X reduction of the cost and time in process and tool design, tool manufacturing, and die try-out In addition, it is possible to iteratively modify t h e process conditions in the simulation to find the best manufacturing conditions for a product Fig 1: Connecting rod 1.1 Forging of Connecting Rods The closed die forging process is often used to manufacture high quality mass production parts like connecting rods, crankshafts, etc., at moderate costs In principle, forging operations are non-steady state 96 7q Takemasu et al / Journal of Materials Processing Technology 59 (1996) 95-105 processes, in which the deformation of the m a t e r i a l takes place under three-dimensional stress and strain conditions The material flow depends mainly on the following [1]: a) Geometry of the cavity b) Geometry of the flash opening c) Initial and intermediate billet geometry d) Percentage of flash e) Heat transfer between the tooling and t h e billet A sketch of a closed-die forging process w i t h flash is shown c~a the left in Fig The upper and lower dies form the closed cavity; the f l a s h originates in the gap between the dies A major advantage of the closed-die forging with flash is that the volume of the preform can vary within a specific range, which makes it easier to continuously manufacture products with the same quality However, a trimming process is necessary to remove the existing flash As shown on the right in Fig flashless forging does not allow the material to leave the cavity and therefore no flash is generated One of the most important advantages of this process is that a significant amount of material can be saved in comparison to forging with flash Furthermore, a trimming operation is not required There are some requirements to get a successful flashless-forging process: a)The volume of the initial preform and t h e volume of the cavity at the end of the process must be the same b) There must be neither a local volume excess nor a shortage, which means that the mass distribution and positioning of the preform must be very exact c) If there is a compensation space in the dies, the real cavity must be filled first 1.2 Research Objectives So far, most FE codes that simulate billet forming processes consider only plane-strain or axisymmetric deformations Since many industrial parts such as connecting rods have very complex geometries, t h e metal flow is three-dimensional and cannot be properly modeled with a two-dimensional approximation This means that a threedimensional simulation of the manufacturing process must be performed to get adequate results The commercial package DEFORM 3D v.2.015] offers t h e possibility of simulating three-dimensional material flow of complex geometries The 3-D FEM simulation of the flashless forging of a connecting rod has already been performed a t the ERC/NSM with DEFORM-3D v.l.0 [2] However, it was not possible to simulate the whole forging process due to the limited remeshing capabilities of DEFORM 3D v.l.0 Therefore, it was not possible to verify that the preform designed from physical modeling experiments [3] was indeed the optimum preform Due to the improved remeshing capabilities of the recently released DEFORM 3D v.2.0, it is now practical to optimize the preform shape through the analysis of the FEM simulation results Forging with Flash Flashless FOrs~ngwith Forging Up Die Flashless Forging Forging Bilh F l a ~ L~~'l ; ' Lower Punch Start of Stroke End of Stroke Fig 2: Closed-die forging with and without flash The objectives of this study are: i) perform the 3-D FEM analysis of an actual connecting rod with DEFORM 3D v.2.0 using the preform defined in the previous studies and find out the problems in the design of the preform ii) optimize the preform design in each region independently, that is, large end section, connecting section, and small end section iii) define a new preform design based cn t h e optimization results and verify the applicability of this optimization method Previous Studies in Preform Optimization Before 3D FEM simulations were practical, physical modeling experiments and 2D FEM simulations were used [3] to define a preform for t h e flashless forging of a connecting rod 3D FEM simulations of the flashless forging of a connecting rod were attempted [2] but were unsuccessful due to limitations in remeshing ] Physical Modeling Experiments In metal forming operations, in order to predict metal flow, die filling, defect occurrence, and T Takeraasu et al / Journal of Materials Processing Technology 59 (1996) 95-105 forming loads, the use of highly deformable model materials represents a valid and powerful tool In the tool design phase, a soft material, either metallic or non-metallic, can be used to carry out several tests by changing the tooling geometry The aim is to optimize material flow and die filling using machinable tools and low cost die materials (i.e acrylic, glass or aluminum) Furthermore, compared to hot forging processes, lower temperatures are typically used in modeling tests Physical modeling experiments were performed for the flashless forging of a connecting rod using plasticine billets and an aluminum tooling [3] The experiments were performed c~ the ERC five ton multi-action press The main objective of t h e plasticine experiments was to find a preform geometry that w o u l d result in complete filling of the die cavity The volume distribution in the connecting rod was obtained by cutting several transverse sections and computing the area of each These values were plotted in Fig as the height versus the length of the connecting rod The area under the curve, indicated by the arrow "A', represents the volume distribution of the piece Based c~ these results an axisymmetric preform was designed The preform suggested in [3] is shown in Fig This preform was modified based cn the physical modeling experiments The final plasticine preform and connecting rod are shown in Figure 2.2 3D FEM Process Simulation of a Connecting Rod 3D FEM analysis of the flashless forging of connecting rods was performed with two different types of preform geometries [2], which were called (1) real geometry and (2) simplified geometry A I i Fig 3: Connecting rod volume distribution versus length • 97 i I ~ Fig 4: Axisymmetric plasticine preform Fig 5: Plasticine preform and connecting rod [3] The shape of the real connecting rod is a modification of a Nissan connecting rod Isometric views of the real and a simplified connecting rod are shown in Fig The simplified geometry of t h e connecting rod was used to verify how simplifications, made in order to accelerate t h e simulation process, affect the simulation results These 3D FEM simulations used brick elements for the billet and rigid surfaces to model the tooling The software used was DEFORM 3D vl.0 The simulations were run isothermally The main limitation of the code used was that it did not h a v e automatic remeshing capabilities Therefore, t h e remeshing had to be performed semi-automatically by approximating the deformed shape of the b i l l e t with surfaces and remeshing the enclosed volume with tetrahedrals which were latter broken into bricks Since this is a a time consuming operation i t was only possible to achieve 75% of the stroke for both geometries The effective strain in t h e deformed connecting rod is shown in Fig Preform Optimization by 3D FEM Simulation In order to verify the applicability of the new FEM code DEFORM 3D (v2.0), a simulation of t h e 98 T Takemasuet al / Journal of Materials Processing Technology 39 (1996) 95-105 real connecting rod was performed using preform-I as defined in previous studies [2,3] A sketch of preform-I is shown in Fig 3.1 F E Simulation o f the Forging Process An upsetting step of the initial preform had to be carried out to start the whole forging process, because the smaller end of the initial preform was too big to fit into the die cavity This operation is performed at hot forging temperature form half of the workpiece is almost 1.9 metric tons, this means that 3.8 metric tons are required for this operation The simulation of the flashless forging was carried out using one upper punch and one die w i t h the upsetted preform in between as shown in Fig 11 Before the upset preform was imported into t h e forging simulation, a remeshing was executed to make meshes finer especially at the smaller end section 1.240 1.13o 1.010 0.896 0.781 ~ 0.665 O.550 0.435 0.319 ~ 0.204 0.089 Fig 7: Effective Strain Distribution S1 sl 3.96 s8 4.70 Fig 6: a) real and b)simplified connecting rod geometry The upsetting process is simulated using one f l a t die (constructed by square shell elements) t h a t moves in the negative Y (down) direction to compress the small end of the connecting rod T e t r a h e d r a l elements were used for the billet in this simulation The simulation was stopped at a stroke of about 1.5 mm The relevant data for this simulation are shown in Table Fig shows the final shape of the preform after upsetting Very little deformation is present in t h e connecting section and the large end of the preform The necessary punch load curve calculated for this operation is shown in Fig 10 The force required to s2 26.77 dl 9.65 $2 s3 12.04 d2 16.76 $3 s4 17.23 d3 7.01 $4 s5 11.61 d4 16.26 $5 s6 6.58 d5 7.06 $6 $7 $8 s7 7.90 Fig 8: Sketch of preform-I of the connecting rod l J~O m ,lt~ *000 Fig 9: Effective strain distribution after upsetting IT Takemasu et al /Journal o f MaterJals Processing Technology 59 (1996) 95-105 i(10 1.864 1.553 1.243 Y a 932 d (N) ~1 311 000 m #~0 t7~ m im ~ i 2e2~ l 3eO~ m 3¢7~ m 4t5~( (ram) Y-Stroke Fig 10: Punch force of preform upsetting process Punch movement Fig 12: Material flow in forging of the connecting rod Fig 11: Simulation model for the forging of the connecting rod Table1: Input data for the upsetting process Simulation Parameter Billet material Punch velocity Stroke Simulation mode Simulation steps (NSTEP) AI 2618 20 mm/s 4.5 mm Isothermal 90 The material flow of the connecting rod forging process is shown in Fig 12 Fig 13 shows the contact condition between the large end portion of the b i l l e t and the tooling at the end of the forging It can be seen from this figure that a relatively large c a v i t y remains at the upper surface of the bigger ring p a r t in the large end section (marked as * in Fig 13) In the connecting or I-beam section, the side wall was initially formed from both ends, gradually proceeded to the middle and combined together finally So the deformation pattern of this section may not be completely in plane strain and does not have enough height at the center even at the end of the stroke +: contact *: no contact Fig 13:Contact condition with the tools in the large end section 99 100 T Takeraasu et al / Journal o f Materials Processing Technology 59 (1996) 95-105 It was concluded from these results that to optimize the initial geometry of the preform we t h e following problems had to solved: a) For the large end section, we need to optimize the preform geometry to fill the cavity completely and uniformly b) For the small end section, control the i n i t i a l volume distribution of this part and transfer the excessive volume to other features of t h e product c) For the connecting I-beam section, we need to optimize the diameter of the preform in order to deform the side wall nearly in plane strain conditions BT1 d2=20 s1=15 =41.5 ,:,T2 d1=22 s1=14 s3=11 3.2 Optimization of the Preform Geometry It seems very difficult to optimize the w h o l e geometry of the preform at once, since the preform shape is relatively complex and has a lot of shape parameters as shown in Fig Hence, the workpiece was divided into three sections: large end section, small end section, and connecting section Each section was optimized independently This optimization procedure was adopted for t h e following reasons: i) Since the connecting section is deformed nearly in plane strain conditions, it is assumed t h a t the deformation of the large end section and that of small end section not strongly interfere with each other ii) The number of shape parameters is reduced and the optimization process becomes easy to handle iii) Simulation time is reduced by working w i t h a smaller model 3.2.1 Large End Optimization There are seven shape parameters in the large end section (see Fig 8) In order to reduce the number of parameters, the diameters dl and d3, the total volume, and the total length were set to be constant The diameters d2, the segment length sl, and t h e length pl=sl+s2+s3 are selected as independent parameters Then the other parameters, that are segment length s2, s3, and s4, were chosen based on these constraints (see Fig 8) Three preform designs for the large end section were selected for the FE simulation model from t h e various combinations of parameters, and are shown in Fig 14 They are named BT0, BT1, and BT2 respectively BT0 is the original preform and is represented by the dotted line in those figures The diameter d2 and the segment length sl of BT1 are both larger than that of BT0 The length of pl of BT2 is shorter than that of BT0 Fig 14 : Sketch of some preforms for BT1 and BT2 The top views of the material flow of the large end section for each preform are shown in Fig 15 Fully 100% of the stroke was achieved in each simulation The shaded area in the top views shows the contacting area between the surface of the b i l l e t and the tools The outer wall of the bigger ring of BT1 and BT2 is deformed almost radially throughout the forging process and has sufficient height at the end of t h e stroke This is compared to BT0, which is sinked in at the beginning of the forging process and remains a little concave even after deformation As for t h e side wall next to the connecting section, the cavity is filled completely in cases BT0 and BT1, while t h a t of case BT2 is not filled at all Comparing m a t e r i a l flow in top views, as the diameter d2 is increased and the length pl is decreased, the die cavity of t h e outer wall of the bigger ring is filled more r a p i d l y and smoothly This is because the material is prevented from flowing to the bigger ring part after the material of the section s3 contacts the die It is seen from these results that in order to deform this large end section successfully, t h e diameter d2, the segment length sl, and the length pl have to be selected correctly 3.2.2 Small End Optimization Since the small end section was completely formed before the end of the stroke in the FE simulation of preform-I, the volume distribution was T, Takemasu et al / Journal o f Materials Processing Technology 59 (1996) 9.5-105 \ BT0 BT1 Fig 15: Material flow of the BT0 - top view (XY plane) varied to optimize the preform geometry There are also seven shape parameters in this section(see Fig 8) The diameters d3 and d5, the segment length s8, and the length p2=s6+s7+s8 are fixed The i n i t i a l volume of this section was controlled by changing the diameter d4 and the segment length s6 Three preforms for the small end section were modeled as shown in Fig 16 They are called TP1, TP2, and TP3 respectively The geometry of t h e small end section of the preform-I is represented by the dotted line The shape parameters of these preforms are compared in Table The volume of TP1 is the smallest TP3 has the same volume as TP2, but the diameter d4 of TP3 is a little larger than that of TP2 Thus the volume distribution of TP3 is gathered to the top end relative to TP2 Table 2: Shape parameters of the small end section TP1 TP2 TP3 101 volume mm 3082 3250 3250 s7 mm 5.70 7.11 6.27 s8 p2 d4 d5 5.0 5.0 5.0 19.0 19.0 19.0 15.5 15.75 16.0 7.0 7.0 7.0 The top view of the deformed billets at the end of the stroke are shown in Fig 17 The small end sections of TP2 and TP3 are completely deformed, while in the case of TP1 a cavity remains at the side wall From these results, it can be concluded that the deformation pattern of this part is not sensitive to BT2 (a) T P (b) T P (c) T P Fig 16: Sketch of FE models of the small end section 102 T Takemasu et al / Journal of Materials Processing Technology 59 ('1996) 95-105 the initial geometry of the preform, since this is formed by the upsetting process before the complete forging process 3.2.3 Connecting I-Beam Section Optimization There are three parameters in the connecting Ibeam section: diameter d3 and segment lengths s4 and s5, as shown in Fig The area of a cross section of the I-beam part calculated by I-DEAS was 42.637 m m z Hence, assuming that the material of this part is deformed under plane strain conditions the initial diameter d3 of the connecting I-beam section was set to 7.37 ram j Underfilling preform design, and was compared with the earlier results The dimensions of the new preform are shown in Table The material flow of the connecting rod forging process with preform-II is shown in Fig 19 As one can see from this figure, the small end section is formed completely The connecting I-beam section flows nearly under plane strain conditions and t h e side wall has enough height after deformation The large end section is deformed almost completely, although a very slight cavity remains between t h e billet and the upper punch at the upper surface of the bigger ring (a) TP1 (b) T P z (c) TP3 Fig 17: Final stage deformed billets of the small end section in top view (XY plane) Fig.18: Sketch of preform-II 3.3 FE Simulation with New Preform Evaluating the results obtained from the optimization method, we proposed a new w h o l e preform design (preform-II), shown in Fig.18 A second 3D FEM simulation was performed with this Fig 19: Material flow of the forging simulation with preform-II 7~ Takemasu et al / Journal of Materials Processing Technology 59 (1996) 95-105 103 range between 0.26 and 2.29 at the end of the stroke The strain distribution of an intermediate simulation step for this model is comparable to Mezger's final results [2] The load predicted for the forging of preform-II is 30% higher than that of preform-I This is because the stroke for preform-II is longer than t h a t of preform-I and a relatively large cavity is observed at the upper surface of the bigger ring p a r t in the large end section of preform-I after deformation .2~5E,01 (b) Preform-II | :::: Fig 20: Deformation condition of the outer wall of the bigger ring at the final stage Table.3: Shape parameters of preform-II, mm sl s2 s3 s4 s5 14.50 6.82 20.18 18.50 13.50 7.19 s6 dl d2 d3 d4 d5 9.00 20.00 7.37 s7 s8 7.11 4.70 15.75 7.00 The deformation condition of the outer wall of the bigger ring of preform-II at the final stage is compared with that of preform-I in Fig 20 The area pointed to by arrow in preform-II has enough height and contacts the upper punch, while t h a t area in preform-I is sinked in and does not contact the upper punch at all The areas pointed to by arrows and in preform-I! are concave because they reflect the geometrical pattern of the upper punch, while such geometrical properties are not observed in similar areas of preform-I The edge areas pointed to by arrows and in preform-II look sharper than that of preform-I This is because t h e die cavity of these areas of preform-II is filled almost completely, while a small cavity can be observed in the same areas of preform-I It was concluded that fair results may be obtained with the design for preform-II Fig 21 shows the effective strain of t h e connecting rod in the final stage The strain values Fig 21: Effective strain of preform-II final stage, a)intermediate step and b) final step Manufacturing the Preform In section 3, a new preform design was suggested and good simulation results were obtained But there is still another important problem: that is, how to manufacture the preform Each manufacturing method has its own advantages and disadvantages concerning tooling cost, lead time, finished shape, and product tolerances In this section, the design and dimensions of the new preform are compared with those of the old preform and the feasibility of producing the new preform by cross rolling is discussed Furthermore, the flashless forging of a connecting rod is also compared with the forging of a connecting rod with flash 1~ Takemasu et al / Journal of Materials Processing Technology 59 (1996) 95-105 104 The principle of operation of cross rolling machines is shown in Fig 22 [4] In this process, a round billet is inserted transversely between two or three rolls, which rotate in the same direction and drive the billet The rolls, which hold replaceable die segments with appropriate impressions, make one revolution while the workpiece rotates several times in the opposite direction Thus, the cross rolling method can form axially symmetrical shafts with complex geometries in one operation FLANK ~GE t.EAD ~'dGLE KNIFE GE REDUCTION AT POS*TI O~ ~ RECTION OF METAL FLOW ON flOLL SURFACE NADIAL ~/ REDUCTION AT ITION Fig 22: Principle of operation of cross rolling machine [4] Conclusions and Future W o r k The cross rolling machines are suitable for automatic production, using bar stock automatically fed to the rolls through an induction heating unit Therefore, cross rolling takes advantage of h i g h productivity (about 900 parts per hour) The design variables of this process are the lead angle of t h e wedge, the flank angle, and the amount of reduction The disadvantages of cross rolling are that selfcontained machines are expensive and the forming rolls are difficult to design This process is also limited to external surfaces The most important problem in using cross rolling for this purpose is t h e desired shape limitations in the preform design At present, cross rolling machines will accept only bars having a maximum diameter of 35 mm The minimum diameter of the product is about 12.5 mm The maximum reduction ratio is 75%, and rolled length of product can be up to 400 mm, depending upon the reduction required The reduction ratio is defined as : R e d u c t i o n R a t i o - Dmax - Dmin x 100 (1) Omax with: Dma x = Dmi n = Fig 23 compares the dimensions and t h e reduction ratio of the new preform with those of t h e old preform The reduction ratio of preform-II is 63.15 % and that of preform-I 58.17 % These are both small enough for the allowable limitation of the reduction ratio for cross rolling The minimum diameters of preform-I and preform-II are much larger than the desired minimum diameter, and t h e total lengths of preform-I and preform-II are smaller than 400 mm So it seems that the cross rolling is applicable for making both i n i t i a l preforms In the flashless forging process the volume of t h e preform must be exactly the same as the finished part and the mass distribution of the preform must be exact to fill up the die cavity correctly Variations in the cross rolling process may affect the required dimensions of the initial preform for t h e flashless forging process In order to verify these points, further investigations of the 3D FEM simulation or physical modeling experiments are needed Especially for the forging process of t h e connecting rods with flash, because this process is not as strict as the flashless forging process in terms of the volume distribution of the initial preform maximum diameter of the preform minimum diameter of the preform The preform design was optimized by dividing the preform into three parts: small end section, connecting I-beam section, and large end section Simulations were performed independently for each section For the large end section, both the i n i t i a l geometry and the volume distribution of the preform was optimized at the same time by changing t h e position and the diameter of the hill section under a condition of constant volume It is concluded from the simulation results that to optimize the geometry of this part is more difficult than the small end section or the connecting I-beam section due to t h e fact that the product shape of this part is very complex and the material flow is strongly influenced by the initial geometry of the preform For the small end section, the initial volume distribution was mainly controlled to fill the die cavity at the end of the punch stroke It is also clear that the deformation pattern of this part is not as sensitive to the initial geometry of the preform as the large end section The deformation pattern of the connecting I-beam section approached plane strain conditions by optimizing the initial diameter of this part The ribs of the I-beam were almost filled at the end of the simulation T Takemasu et al / Journal o f Materials Processing Technology 59 (1996) 95-105 ~I 9o.79 L, Reduction Ratio = 58.17 % (a) Preform-I 92.50 o Reduction Ratio = 63.15 % 105 Eight to ten remeshing steps were needed in the simulation from initial preform to the end of the stroke About 1% volume loss is observed after every remeshing step and the total volume loss reached about 8% The future work related to this project should involve the following aspects: i) The results of the connecting rod forging simulations must be compared with the results from an experimental investigation of the forging process to verify the applicability of the new DEFORM 3D code ii)The simulation of the connecting rod forging process with flash should also be performed to compare the deformation pattern and the advantages and disadvantages with the results of the flashless forging of a connecting rod (b) Preform-II Fig 23: Comparison of the shape parameters of preform-II with that of preform-I Evaluating these results, we proposed a whole new preform design that was called preform-II and performed the FEM simulation of the forging of the connecting rod with it In this simulation, 99.5% reduction ratio was achieved and the small end section was formed completely The connecting Ibeam section was deformed nearly in plane strain conditions and the side wall of this part had enough height after deformation The large end section was formed almost completely and the side wall of the bigger ring part was almost radial A small cavity remained between the billet and the upper punch These results show the validity of the optimization method adopted in this report The load-stroke curve of preform-II is similar to that of preform-I in its trend, but the peak load of preform-II is about 30% higher than that of preform-I References [1] Lange, K (1985) Handbook of Metal Forming McGraw-Hill [2]Mezger, J., Sweeney, K., Altan, T (1994) Investigation of the 3D CODE: Flashless Forging of a Connecting Rod ERC/NSM-B-94-31 [3] Barcellona, A., Long, K., Altan, T (1994) Flashless Forging of a Connecting Rod of an Aluminium Alloy and a Metal Matrix Composite (MMC) Material ERC/NSM-B-94-32 [4] Altan, T., Boulger, F., Becker, J., Akgerman, N., Henning, H (1973) Forging Equipment, Materials, and Practices Metals and Ceramics Information Center [5] SDRC (1990) Finite Element Modeling - User's Guide Structural DynamicsReseach Corporation, Milford, Ohio [6] Scientific Forming Technologies Corporation (1994).DEFORM Version 4.0 User's Manual Columbus, OH