Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Journal of Materials Processing Technology 111 (2001) 29

Recent developments in process simulation for bulk forming processes


S.I. Oha,*, W.T. Wub, K. Arimotob
Department of Mechanical Design and Production Engineering, Seoul National University, Shinrim-dong San 56-1, Seoul 151-742, South Korea b Scientic Forming Technologies Corporation, 5038 Reed Road, Columbus, OH 43220-2514, USA
a

Abstract In this paper, certain critical issues and associated methodologies to develop a three dimensional Finite Element based code for bulk forming simulations are addressed and described. Since heat treatment is a typical post-forging process to improve the micro-structure and mechanical properties, efforts have also been made to develop capabilities for the heat treatment applications. Several forming and heat treatment applications are presented to demonstrate the capability of the Finite Element Modeling technique. # 2001 Elsevier Science B.V. All rights reserved.
Keywords: Finite element method; Bulk forming process; Open die forming

1. Introduction In recent years, due to many successful contributions in the industry applications, nite element method (FEM) based on computer simulation has become extremely common in the design and development of bulk forming processes in many well-known companies. Since the rapid advancement of computing technology and reduction in computer cost, the environment for three-dimensional complex forging process simulation has become realistic. Much effort has been made to make three-dimensional process simulation (open die forming such as cogging, cold, warm and hot forming) practical and easy to employ. Bulk forming, in most cases, is not the only process nor the nal process, but one of many operations required in the manufacture of service components. To understand the quality, reliability or expected performance of a service component, the process simulation result from bulk forming must be propagated to its subsequent process analysis for continuity as well as for a better initial condition denition. Since many forming processes require heat treatment to reduce residual stress and to obtain desire mechanical and/or microstructure properties in the forgings, efforts have also been made to develop capabilities that are critical to the heat-treatment process. In this paper, a brief overview of recent developments, followed by examples, is presented.
* Corresponding author. Tel.: 82-2-880-7151; fax: 82-2-872-0236. E-mail address: sioh@snu.ac.kr (S.I. Oh).

2. Recent developments and considerations The industrial acceptance of metal forming FEM simulation has increased rapidly during recent years [13]. Together with fast improvement in computing speed and reduction in computer cost, the computing environment to exercise 3D forming modeling has become realistic. To develop an accurate, fast, and robust simulation tool for manufacturing processes such as forming and heat-treatment has become a primary objective. To meet this objective, several important issues are being considered, evaluated and developed in DEFORM [4], as described below. 2.1. Material characterization For bulk forming, rigidplastic and rigidvisco-plastic material models (ow stress is a function of strain, strain rate, and temperature) are generally used due to their simplicity and fast convergence. For cases with interest in the residual stress and/or spring-back at room temperature, the elasto-plastic model is a better choice. The elasto-plastic material is generally represented by an initial yielding and strain hardening. For heat-treatment applications, such as annealing, aging, solution treatment, the elasto-viscoplastic model is used to take into account stress relaxation. In this case, the deformation is described by the stress state, temperature, and strain. The grain size is also an important parameter that governs the material behavior as well as the transformation. Effort is being made to model the dynamic,

0924-0136/01/$ see front matter # 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 4 - 0 1 3 6 ( 0 1 ) 0 0 5 0 8 - 8

S.I. Oh et al. / Journal of Materials Processing Technology 111 (2001) 29

meta-dynamic, and static recrystallization and grain growth. For a carbon steel component, the transformation and material data is also determined by the carbon content. The transformational plasticity, latent heat, volume change during the phase transformation are also important to be included in the model. Much efforts has also been made to understand/characterize the damage mechanics [5]. 2.2. Brick element vs. tetrahedral element It is generally recognized that the brick element performs better than the tetrahedral element. For most bulk forming problems, large deformation is expected. When the element is severely distorted, remeshing is necessary to continue the simulation. Since the automatic remeshing of the brick mesh is difcult, the brick element becomes impractical for forming applications that require frequent remeshing. Furthermore, the number of nodes in the brick mesh is generally greater than the total number of elements. In the tetrahedral mesh, it is not uncommon for the element/ node ratio to be 5. To represent a complex shape geometry, the tetrahedral element is generally more economical than the brick (in terms of the total number of equations or nodes). For cases that does not require remeshing, such as cogging (also known as billet conversion), the brick element is still a preferred choice. In heat-treatment applications, if the geometry is so complex that it would take days to generate a brick mesh, the tetrahedral element is the natural candidate. However, further effort should be made in the area of improving the overall performance of the tetrahedral element. 2.3. Automated meshing and remeshing Without a robust remeshing procedure, the 3D forming simulation can never be practical. The ability to adaptively remesh depending upon the solution behavior/error has been an area of interest academically. In the Lagrangian approach, complete remeshing has been one signicant source of losing/gaining volume. Therefore, being able to identify the location that needs remeshing and carrying out local remeshing is crucial to minimize the volume change during remeshing. In many applications such as extrusion with a large extrusion ratio, small elements are needed to represent the localized deformation while large elements are needed in the rigid zone for computational efciency. The ability to transit a small element area to a large element area is also critical for a robust mesh generator. 2.4. Direct solver and iterative solver In solving three-dimensional problem using the implicit method, equation solving generally takes more than 90% CPU time, the required storage for the assembled stiffness matrix increasing quadratically with the total number of equations in the direct solving scheme. In recent years,

sparse matrix solvers such as Spooles [6] have become quite popular in solving 3D problems. It is also known that iterative solver with a proper pre-conditioning method can be highly effective for some types of problems. 2.5. User interface A good graphics user interface is a necessity for any program that deals with three-dimensional problems. Mouse-based dynamic zoom, rotate, etc. to facilitate the visualization is a must. Special facilities are developed in the pre-processor to carry out die positioning, contact boundary condition generation, die stress analysis, etc., and in the post-processor to reveal the under-ll region, die pressure distribution, thermal mechanical history of selected points, etc. Efforts are also made for real time manipulation of the rendered image using hardware. For certain processes, such as cogging (billet conversion), a special pre-processor has been developed to facilitate the simulation of the repetitive bites [7], passes, and heats. 2.6. Computing environment and parallel computing Computer speeds have improved tremendously over the last 5 years making industrial strength simulations feasible on inexpensive hardware. Processor speeds on the Pentium line of chips has increased from 75 to over 500 MHz and the Unix vendors have also enhanced their line of chips. The cost of computers and memory has fallen allowing congurations of 512 MB of RAM or more on inexpensive desktop workstations for running 3D FEM simulations. Computers with multiple processors (especially the Windows NT based PC due to its price) are also becoming very popular and affordable now. The use of MPI has shown linear scaling. Effort has been made in improving code performance by using the parallel computing facility. 3. Examples A series of examples to demonstrate the capabilities of process modeling is presented. Section 3.1 concentrates on metal forming applications, whilst Section 3.2 includes two heat-treatment applications. The integrated forming and heat-treatment simulation of a gear are presented in Section 3.3. 3.1. Metal forming applications 3.1.1. Forward extrusion with a piping defect A cylindrical billet is forward extruded to make the pressure valve. The cross-section of the real part is shown in Fig. 1. It is noted in this gure that there is a piping defect at the center of the part. Although the process is threedimensional, due to the lack of computing power and a fully automated three-dimensional remeshing procedure, the

S.I. Oh et al. / Journal of Materials Processing Technology 111 (2001) 29

Fig. 1. The cross-section of the real pressure valve.

Fig. 3. A cold-formed copper electrical part.

problem was simulated under the plane-strain assumption [8]. To validate the code and remeshing procedure, the actual process is simulated under three-dimensional conditions. The material is AL6062, with a ow stress of s 204:6e0:122 . To take advantage of the symmetric condition, only a quarter of the part is simulated. A local mesh density is dened near to the piping defect for the remeshing. The predicted part geometry at different stages of the extrusion process is shown in Fig. 2. From this gure, it is clear that the defect starts at the center of the part and propagates in the transverse direction where the part is being extruded. 3.1.2. Cold forming of an electrical component A cold-formed copper electrical component with ashes is shown in Fig. 3. The workpiece to make this component is a simple blank sheet. Both forward and backward extrusion were taking place simultaneously. During the process, the upper features (the backward-extruded portion) were formed in the early stage while the lower part of the component (forward extrusion) was formed at the nal stage. The simulation exhibited good correlation with the actual process. 3.1.3. Orbital forging Orbital forging is a very unique process with a rocking die movement that can be used to reduce the axial load requirements for some axisymmetric or near-axisymmetric forging operations. The complex tool movement involves the top die

rotating about an axis that is pivoting about a central axis. The result is a very localized contact between the die and the workpiece. The simulation shows an axisymmetric part made using the orbital forming process. Fig. 4 depicts an orbital forging simulation near to the nal stages of deformation. It is to be noted that there is a slight angle between the die and the top of the workpiece. While orbital forging offers benets, the complex die movement can make it difcult to anticipate the material ow of the workpiece. A defect formed early in the process can be masked later by die contact, thus simulation is very important to the development of a robust process. 3.1.4. Open die forging: cogging The cogging process is used to produce billets by reducing the diameter of cast ingots via numerous hits between simple shaped dies. Being able to predict the thermalmechanical response and the geometric evolution (spread and elongation) during the cogging process, it is essential to optimize the process and to obtain the desired microstructure properties. The workpiece geometry involved in cogging is

Fig. 2. The predicted geometry at different stages of the extrusion process.

Fig. 4. The workpiece (shaded with nite element mesh) and top die (wire frame) are shown in an orbital forging near to the nal stage of deformation.

S.I. Oh et al. / Journal of Materials Processing Technology 111 (2001) 29

Fig. 6. The material ow in this shaped extrusion simulation clearly shows twist in the isometric and top views.

Fig. 5. Contours of temperature during three stages of deformation of a cogging process.

generally simple. The time and effort that are required to carry out the simulation, however, can be very extensive and tedious due to the nature of repetitive bites/passes/heats. Efforts [7] have been made to shorten the simulation time/ staff time. A 6-pass cogging sequence with 46 bites in each pass and heat transfer throughout the process was carried out. The specialized facilities for cogging allowed the simulation to run through numerous forming operations without user intervention. Three representative steps are shown in Fig. 5. 3.1.5. Shaped extrusion In the simulation of a high reduction shape extrusion, DEFORMTM-3D was used to determine if the product would bend during extrusion. Due to the high extrusion ratio, mesh windows were used to generate a very ne mesh in the extrusion zone, an intermediate mesh in the extruded product and a coarse mesh on the undeformed workpiece, resulting in improved computational efciency. Defects such as bending and twisting of the extruded product are the result of a less than optimum die design. Fig. 6 demonstrates a material ow bent from left to right as the shape is extruded. Excessive bend or twist can be very expensive and time consuming to remove. Simulation can be used to optimize die design prior to production trials. 3.1.6. Helical gear extrusion Rotational symmetry conditions can be prescribed in DEFORMTM-3D, which allows a surface to be coupled with a complimentary surface at a prescribed angle from the original surface. In fact, the symmetric surfaces are not required to be planes. Thus, one can simulate cases like a helical gear extrusion, as shown in Fig. 7. If the analysis of the entire part were required, this simulation would become very expensive computationally. The use of rotational symmetry allows for an efcient analysis of this type of process.

Fig. 7. This helical gear extrusion was simulated using a very efcient rotational symmetry algorithm.

3.1.7. Tee forging In hot forging, it is common to form multiple parts in a single plate with ash. The forging of 300 series stainless steel tees was analyzed with and without the bust operation. Without the bust operation, a forging lap was observed, as shown in Fig. 8. This correlated well with actual production

Fig. 8. This image represents the forging (block and nish), immediately prior to lap formation.

S.I. Oh et al. / Journal of Materials Processing Technology 111 (2001) 29

Fig. 9. This simulation graphic is at the nal stage of forging after a bust, block and nish operation.

Fig. 11. Automotive DEFORMTM-3D.

crankshaft

forging

was

simulated

using

experience. The addition of the bust operation eliminated the defect as a result of better volume distribution going into the blocking operation, as shown in Figs. 9 and 10. Process simulation can be used to understand the inuence of volume distribution on both defect formation and product cost. A ash region that is wide in one region and thin in another indicates a less than optimum preform design. Process simulation is routinely used to optimize the volume distribution, resulting in a high quality, low cost process. 3.1.8. Crankshaft This hot-forged automotive crankshaft is a well-known application for three-dimensional simulation. This critical structural component of an internal combustion engine requires proper grain ow and forging practice to produce a satisfactory part. This simulation is challenging because the geometry is critical, the deformation is large and there is a very limited opportunity to take advantage of symmetry. Simulating the thin ash requires a robust 3D system with automatic remeshing and a fast FEM engine. Fig. 11 demonstrates a typical crankshaft forging that can be run on a personal computer or an engineering workstation. 3.2. Heat-treatment applications 3.2.1. Carburized quenching of a rotating body component This example is presented to demonstrate the code's capability [8,9] in simulating multiple heat-treatment

processes including heating, carburizing, and quenching. The material used in the simulation was AISI 5120. The elastic, plastic ow stress, and thermal data for the different phases (austenite, bainite, and martensite) are all functions of temperature and carbon content. The simulation included the transformation kinetics of (1) bainite to austenite (during heat up), (2) austenite to bainite (cool down), and (3) austenite to martensite (cool down). The transformation data are also functions of temperature and carbon content. The initial phase was assumed to be 100% bainite which gradually transformed to austenite during heating. During the carburizing stage, the environment carbon potential was assumed to be 0.9% for 230 min of a boost cycle. After the boost cycle, the environment carbon potential was assigned to reduce to 0.65% for a further 36 min of a diffusion cycle. Fig. 12 shows the carbon content distribution at the end of the carburizing process. Oil quench was simulated as the nal stage of heat treatment. During the quenching, the temperature of the oil is assumed to be room temperature. Fig. 13 shows the volume fraction of martensite at different stages during the quenching process: 8.8, 13.6 and 22.6 s into the quench, as shown by Fig. 13(a)(c), respectively. It is interesting to note that the martensite initially transformed sub-surface. This is due to the fact that the richer carbon content at surface

Fig. 10. Photograph of the actual forging.

Fig. 12. Carbon content after the carburizing process.

S.I. Oh et al. / Journal of Materials Processing Technology 111 (2001) 29

Fig. 15. Residual stress after quenching.

Fig. 13. Volume fraction of martensite during the quenching process.

lowers the martensite transformation temperature. Martensite transformation near to the surface is thus delayed. Fig. 14 illustrates the distortions (magnied) in the rotating body at various stages in the quench process. At the end of quenching, the cone angle of the bottom surface had reduced towards the horizontal. The axial and circumferential components of the residual stress are illustrated in Fig. 15. 3.2.2. Water quenching of a disc with an eccentric hole To demonstrate the use of the system to predict quench cracking, a water-quench simulation was carried out, involving a disc with an eccentric hole (see Fig. 16(b)). The disc was heated to 8108C and cooled by city water to room temperature. The material used was AISI W1 steel, and it

Fig. 16. Simulated stress distribution and quench cracking.

transformed from austenite to martensite during the quench process. The volume fraction of transformed martensite at intermediate stages during the quench simulation is shown in Fig. 17, where (a)(c) illustrates the extent of transformed martensite at 2, 3 and 4.56 s into the quench process. The maximum principle stress on the disc at 4.6 s into the quench process is shown in Fig. 16(a). The location of the highest predicted maximum principal stresses coincided very well with the locations of cracking in the actual component. 3.3. Integrated heat-treatment analysis The forming of a medium carbon, manganese steel bevel gear was analyzed using DEFORMTM-3D on a personal computer [10]. This gear was hot forged with ash. STL les were imported from a CAD system as the geometric input.

Fig. 14. Deformed shapes during quenching.

Fig. 17. Martensite volume fraction.

S.I. Oh et al. / Journal of Materials Processing Technology 111 (2001) 29

transfer, forming-equipment response to the workpiece, die stress and deection, and machining [1113], etc. Coupled with phase transformation, process modeling can now also be applied to the area (e.g. heat treatment and welding) where phase transformation, residual stress and distortion due to thermal effects are critical. To make process modeling a more robust simulation tool for a variety of manufacturing processes, it is expected that future research should be focused on the following:
Fig. 18. Contours of effective strain are shown (with the FEM mesh) on of a hot-forged gear with ash.

Fig. 19. The volume fraction of martensite (dark is higher) in a steel gear after quenching.

One-twentieth of the total volume was simulated using rotational symmetry. Mesh density windows were applied for local mesh renement during simulation. Contours of effective strain (dark is higher strain) at the end of the forging operation are shown in Fig. 18. After deformation, the gear geometry was modied in a CAD system to simulate the ash removal and the drilling of the inside diameter. The gear was austenized by heating to 15608F and cooled in 60 s with a heat-transfer coefcient representative of an oil quench. A TTT diagram for medium carbon, manganese steel determined the diffusion-type transformation from austenite to pearlite/bainite, whereas the Magee equation was used to model the martensitic response. In Fig. 19, the volume fraction of martensite is shown after quenching. The dark regions represent a more complete martensite transformation and the light areas indicate a mixture of bainite and pearlite. 4. Current and future development Process models are now routinely and successfully used in a wide ranges of applications, including material ow, heat

1. Grain size modeling: Grain size is a important factor that will affect the transformation kinetics. It is essential to model the recrystallization kinetics and grain growth that would occur due to the forming processes. 2. Transformation modeling: To save energy for re-heating during heat-treatment, many forging components are quenched at a controlled cooling rate immediately after forging. The initial volume fraction of each of the phases for the quenching analysis should be determined by the forging process. During forming, temperature non-uniformity is expected due to die chilling and deformation heating and certain transformations would take place depending on the material and process window. An accurate material data to represent the transformation under the forming conditions is necessary for the analysis. 3. The thermal boundary condition during the heattreatment application. 4. Fracture model and criteria: A proper fracture model for quench cracking, machining, blanking, shearing, etc., is needed to make the process modeling not only a qualitative but also a quantitative simulation tool. 5. Reliability study of components by considering the manufacturing processes and service conditions. In due time, it should be possible to analyze the entire manufacturing process, from casting through to machining. This will allow designers to include residual stress and grain ow in their product design and application analysis. The benets from this should include reduced product life cycle cost and increased safety margins on critical service components. References
[1] S.I. Oh, W.T. Wu, J.P. Tang, A. Vedhanayagam, Capabilities and applications of FEM code DEFORMTM; the perspective of the developer, J. Mater. Process. Technol. 27 (1991) 3638. [2] J. Walters, Application of the nite element method in forging: an industry perspective, J. Mater. Process. Technol. 27 (1991) 4351. [3] J.P. Tang, W.T. Wu, J. Walters, Recent development and applications of nite element method in metalforming, J. Mater. Process. Technol. 46 (1994) 120121. [4] W.T. Wu, G. Li, A. Arvind, J.P. Tang, Development of a threedimensional nite element method based process simulation tool for the metal forming industry, in: Proceedings of the Third Biennial Joint Conference on Engineering Systems Design and Analysis, Montpellier, France, July 14, 1996.

S.I. Oh et al. / Journal of Materials Processing Technology 111 (2001) 29 [5] Y. Kim, M. Yamanaka, T. Altan, Prediction and elimination of ductile fracture in cold forgings using FEM simulations, ERC Report No. 265. [6] C. Ashcraft, D. Pierce, D. Wah, J. Wu, An object oriented software library for solving sparse linear systems of equations, The Reference Manual for SPOOLES, Release 2.2. [7] A. Arvind, W.T. Wu, G. Li, J.P. Tang, Process modeling of cogging, in: Proceedings of the Sixth ICTP Conference, Nuremberg, Germany, September 1924, 1999. [8] W.T. Wu, S.I. Oh, T. Altan, R.A. Miller, Optimal mesh density determination for the FEM simulation of forming process, in: Proceedings of NUMIFORM'92, Valbonne, France, September 1418, 1992. [9] K. Arimoto, G. Li, A. Arvind, W.T. Wu, The Modeling of Heat Treating Process, ASM 18th Heat & Treating Conference, Oct 1215, 1998, Chicago, IL.

[10] J. Walters, W. Wu, A. Arvind, G. Li, D. Lambert, J. Tang, Recent development of process simulation for industrial application, in: Proceedings of the Third International Precision Forming Technology Conference, Columbus, OH, October 1998. [11] K. Lange, A. Hettig, M. Knoerr, Increasing tool life in cold forming through advanced design and tool manufacturing techniques, J. Mater. Process. Technol. 35 (1992) 495513. [12] E. Taupin, J. Breitling, W.T. Wu, T. Altan, Material fracture and burr formation in blanking results of FEM simulations and comparison with experiments, 1996. [13] T.C. Tszeng, W.T. Wu, J. Tang, An integrated modeling system for heat treating and machining processes, in: Shen, Dawson (Eds.), Simulation of Materials Processing: Theory, Methods and Applications, Balkerma, Rotterdam, 1995, ISBN 90-5410-553-4.

You might also like