CHAPTER 16 Process Modeling in Impression-Die Forging Using Finite-Element Analysis Manas Shirgaokar Gracious Ngaile Gangshu Shen 16.1 Introduction Development of finite-element (FE) process simulation in forging started in the late 1970s. At that time, automatic remeshing was not available, and therefore, a considerable amount of time was needed to complete a simple FE simulation [Ngaile et al., 2002]. However, the development of remeshing methods and the advances in computational technology have made the industrial application of FE simulation practical. Commercial FE simulation software is gaining wide acceptance in the forging industry and is fast becoming an integral part of the forging design and development process. The main objectives of the numerical process design in forging are to [Vasquez et al., 1999]: Develop adequate die design and establish process parameters by: a. Process simulation to assure die fill b. Preventing flow-induced defects such as laps and cold shuts c. Predicting processing limits that should not be exceeded so that internal and surface defects are avoided d. Predicting temperatures so that part properties, friction conditions, and die wear can be controlled Improve part quality and complexity while reducing manufacturing costs by: a. Predicting and improving grain flow and microstructure b. Reducing die tryouts and lead times c. Reducing rejects and improving material yield Predict forging load and energy as well as tool stresses and temperatures so that: a. Premature tool failure can be avoided. b. The appropriate forging machines can be selected for a given application. Process modeling of closed-die forging using finite-element modeling (FEM) has been applied in aerospace forging for a couple of decades [Howson et al., 1989, and Oh, 1982]. The goal of using computer modeling in closed-die forging is rapid development of right-the-first-time processes and to enhance the performance of components through better process understanding and control. In its earlier application, process modeling helped die design engineers to preview the metal flow and possible defect formation in a forging. After the forging simulation is done, the contours of state variables, such as effective strain, effective strain rate, and temperature at any instant of time during a forging, can be generated. The thermomechanical histories of selected individual locations within a forging can also be tracked [Shen et al., 1993]. These functions of process modeling provided an insight into the forging process that was not available in the old days. Integrated with the process modeling, microstructure modeling is a new area that has a bright future [Sellars, 1990, and Shen et al., 2000]. Microstructure modeling allows the right-the-first-time optimum metallurgical features of the forging to be previewed on the computer. Metallurgical aspects of forging, such as
194 / Cold and Hot Forging: Fundamentals and Applications grain size and precipitation, can be predicted with reasonable accuracy using computational tools prior to committing the forging to shop trials. Some of the proven practical applications of process simulation in closed-die forging include: Design of forging sequences in cold, warm, and hot forging, including the prediction of forming forces, die stresses, and preform shapes Prediction and optimization of flash dimensions in hot forging from billet or powder metallurgy preforms Prediction of die stresses, fracture, and die wear; improvement in process variables and die design to reduce die failure Prediction and elimination of failures, surface folds, or fractures as well as internal fractures Investigation of the effect of friction on metal flow Prediction of microstructure and properties, elastic recovery, and residual stresses 16.2 Information Flow in Process Modeling It is a well-known fact that product design activity represents only a small portion, 5 to 15%, of the total production costs of a part. However, decisions made at the design stage determine the overall manufacturing, maintenance, and support costs associated with the specific product. Once the part is designed for a specific process, the following steps lead to a rational process design: 1. Establish a preliminary die design and select process parameters by using experiencebased knowledge. 2. Verify the initial design and process conditions using process modeling. For this purpose it is appropriate to use well-established commercially available computer codes. 3. Modify die design and initial selection of process variables, as needed, based on the results of process simulation. 4. Complete the die design phase and manufacture the dies. 5. Conduct die tryouts on production equipment. 6. Modify die design and process conditions, if necessary, to produce quality parts. Hopefully, at this stage little or no modification will be necessary, since process modeling is expected to be accurate and sufficient to make all the necessary changes before manufacturing the dies. Information flow in process modeling is shown schematically in Fig. 16.1 [Shen et al., 2001]. The input of the geometric parameters, process parameters, and material parameters sets up a unique case of a closed-die forging. The modeling is then performed to provide information on the metal flow and thermomechanical history of the forging, the distribution of the state variables at any stage of the forging, and the equipment response during forging. The histories of the state variables, such as strain, strain rate, temperature, etc., are then input to the microstructure model for microstructural feature prediction. All of the information generated is used for judging the closed-die forging case. The nonsatisfaction in any of these areas will require a new model with a set of modified process parameters until the satisfied results are obtained. Then, the optimum process is selected for shop practice. 16.3 Process Modeling Input Preparing correct input for process modeling is very important. There is a saying in computer modeling: garbage in and garbage out. Sometimes, a time-consuming process modeling is useless because of a small error in input preparation. Process modeling input is discussed in terms of geometric parameters, process parameters, and material parameters [SFTC, 2002]. 16.3.1 Geometric Parameters The starting workpiece geometry and the die geometry need to be defined in a closed-die forging modeling. Depending on its geometrical complexity, a forging process can be simulated either as a two-dimensional, axisymmetric or plane-strain, or a three-dimensional problem. If the process involves multiple stations, the die geometry of each station needs to be provided. A typical starting workpiece geometry for a closed-die forging is a cylinder with or without chamfers. The diameter and the height of the cylinder are defined in the preprocessing stage. A lot of closed-die forgings are axisymmetric, which need a two-dimensional geometry handling. Boundary conditions on specific segments
Process Modeling in Impression-Die Forging Using Finite-Element Analysis / 195 of the workpiece and dies that relate to deformation and heat transfer need to be defined. For example, for an axisymmetric cylinder to be forged in a pair of axisymmetric dies, the nodal velocity in the direction perpendicular to the centerline should be defined as zero, and the heat flux in that direction should also be defined as zero. 16.3.2 Process Parameters The typical process parameters to be considered in a closed-die forging include [SFTC, 2002]: The environment temperature The workpiece temperature The die temperatures The coefficients of heat transfer between the dies and the billet and the billet and the atmosphere The time used to transfer the workpiece from the furnace to the dies The time needed to have the workpiece resting on the bottom die The workpiece and die interface heat-transfer coefficient during free resting The workpiece and die interface heat-transfer coefficient during deformation The workpiece and die interface friction, etc. The die velocity is a very important parameter to be defined in the modeling of a closed-die forging. If a hydraulic press is used, depending on the actual die speed profiles, the die velocity can be defined as a constant or series of velocities that decrease during deformation. The actual die speed recorded from the forging can also be used to define the die velocity profile. If a mechanical press is used, the rpm of the flywheel, the press stroke, and the distance from the bottom dead center when the upper die touches the part need to be defined. If a screw press is used, the total energy, the efficiency, and the ram displacement need to be defined. If a hammer is used, the blow energy, the blow efficiency, the mass of the moving ram and die, the number of blows, and the time interval between blows must be defined. Forgings performed in different machines, with unique velocity versus stroke characteristics, have been simulated successfully using the commercial FE software DEFORM (Scientific Forming Technologies Corp.) [SFTC, 2002]. Fig. 16.1 Flow chart of modeling of closed-die forging [Shen et al., 2001]
196 / Cold and Hot Forging: Fundamentals and Applications 16.3.3 Tool and Workpiece Material Properties In order to accurately predict the metal flow and forming loads, it is necessary to use reliable input data. The stress-strain relation or flow curve is generally obtained from a compression test. However, the test is limited in achievable strains. In order to obtain the flow stress at large strains and strain rates, the torsion test can be used or, alternatively, the compression data is extrapolated with care. In most simulations, the tools are considered rigid; thus, die deformation and stresses are neglected. However, in precision forging operations, the relatively small elastic deformations of the dies may influence the thermal and mechanical loading conditions and the contact stress distribution at the die/workpiece interface. Thus, die stress analysis is a crucial part of process simulation to verify the die design and the forging process parameters. 16.3.4 Interface Conditions (Friction and Heat Transfer) The friction and heat-transfer conditions at the interface between the die and the billet have a significant effect on the metal flow and the loads required to produce the part. In forging simulations, due to the high contact stresses at the interface between the workpiece and the die, the constant shear friction factor gives better results than the coulomb friction coefficient. The most common way to determine the shear friction factor in forging is to perform ring compression tests. From these tests, it is possible to estimate the heat-transfer coefficient, flow stress and friction as a function of temperature, strain rate, strain, and forming pressure, as discussed in Chapter 6, Temperatures and Heat Transfer. Friction factors measured with the ring compression test, however, are not valid for precision forging processes (hot, warm, and cold) where the interface pressure is very high and the surface generation is large. The friction conditions change during the process due to changes in the lubricant and the temperature at the die/ workpiece interface. In such applications, the double cup extrusion test is recommended for estimation of the friction factor, as discussed in Chapter 7, Friction and Lubrication. 16.3.5 Material Parameters The closed-die hot forging modeling is a coupled heat-transfer and deformation simulation. Material parameters that relate to both heat transfer and deformation need to be defined. The material parameters commonly used for heattransfer modeling are the thermal conductivity, heat capacity, and emissivity of the workpiece and die materials. These parameters are usually defined as a function of temperature, The flow stress of the workpiece material is very important for the correct prediction of metal flow behavior. It is usually defined as a function of strain, strain rate, temperature, and possible starting microstructures. The Young s modulus, the Poisson s ratio as a function of temperature, and the thermal expansion of the die materials are important parameters for die stress analysis. 16.4 Characteristics of the Simulation Code 16.4.1 Mesh Generation and Automatic Remeshing In forging processes, the workpiece generally undergoes large plastic deformation, and the relative motion between the deforming material and the die surface is significant. In the simulation of such processes, the starting mesh is well defined and can have the desired mesh density distribution. As the simulation progresses, the mesh tends to get distorted significantly. Hence, it is necessary to generate a new mesh and interpolate the simulation data from the old mesh to the new one to obtain accurate results. Automated mesh generation (AMG) schemes have been incorporated in commercial FE codes for metal forming simulations. In DEFORM, there are two tasks in AMG: 1) determination of optimal mesh density distribution and 2) generation of the FE mesh based on the given density. The mesh density should conform to the geometrical features of the workpiece at each step of deformation [Wu et al., 1992]. In order to maximize the geometric conformity, it is necessary to consider mesh densities that take into account the boundary curvature and local thickness. In DEFORM, two-dimensional (2-D) simulations use quadrilateral elements, whereas three-dimensional (3-D) simulations use tetrahedral elements for meshing and automatic remeshing [Wu et al., 1996]. With this automatic remeshing capability, it is possible to set up a simulation model and run it to the end with very little interaction with the user.
Process Modeling in Impression-Die Forging Using Finite-Element Analysis / 197 16.4.2 Reliability and Computational Time Several FE simulation codes are commercially available for numerical simulation of forging processes, such as DEFORM (2-D and 3-D), FORGE (2-D and 3-D) (Ternion Corp.), Qform (2-D and 3-D), etc. In addition to a reliable FE solver, the accurate and efficient use of metal flow simulations require [Knoerr et al., 1992]: Interactive preprocessing to provide the user with control over the initial geometry, mesh generation, and input data; automatic remeshing to allow the simulation to continue when the distortion of the old mesh is excessive; interactive postprocessing that provides more advanced data analysis, such as point tracking and flow line calculation Appropriate input data describing the thermal and physical properties of die and billet material the heat transfer and friction at the die/workpiece interface under the processing conditions investigated, and the flow behavior of the deforming material at the relatively large strains that occur in practical forging operations Analysis capabilities that are able to perform the process simulation with rigid dies to reduce calculation time and to use contact stresses and temperature distribution estimated with the process simulation using rigid dies to perform elastic-plastic die stress analysis The time required to run a simulation depends on the computer used and the amount of memory and workload the computer has. However, with today s computers, it is possible to run a 2-D simulation in a couple of hours, while a 3-D simulation can take anywhere between a day to a week, depending on the part complexity [Wu et al., 1996]. duces defects in the forging. In real closed-die forging, it is necessary to wait until the forging is finished to see the forged part and the defect, if there is one. The advantage of computer simulation of forging is that the entire forging process is stored in a database file in the computer and can be tracked. Whether there is a defect formed and how it is formed can be previewed before the actual forging. Figure 16.2 shows the lap formation for a rejected process in the design stage. The lap formation can be eliminated by changing the workpiece geometry (the billet or preform), or the die geometry, or both. The computer modeling can again indicate if the corrective measure works or not. 16.5.2 Distribution and History of State Variables The distribution of the state variables, such as the strain, strain rate, and temperature, at any stage of a closed-die forging can be plotted from the database file saved for the forging simulation. The history of these state variables can also be tracked. Figure 16.3(a) shows the effective strain distribution of a closed-die forging forged in an isothermal press. The effective strain has a value of 0.4 to 0.9 in the bore die lock region. The region that is in contact with the upper die has an effective strain value of 0.4 to 0.9, and the region that is in contact with the lower die, a value of 0.7 to 0.9. With an effective strain of 2.0 to 2.8, the bore rim transition region has the largest strain. The effective strain value is approximately 1.5 for both the rim and the midheight of the bore region. From the state variable distribution plot, the state variable at a specific stage of the forging is known. This specific stage, 16.5 Process Modeling Output The process modeling provides extensive information of the forging process. The output of process modeling can be discussed in terms of the metal flow, the distribution and history of state variables, the equipment response during forging, and the microstructure of the forging. 16.5.1 Metal Flow The information on metal flow is very important for die design. Improper metal flow pro- Fig. 16.2 Lap prediction using process modeling tool
198 / Cold and Hot Forging: Fundamentals and Applications shown in Fig. 16.3(a), is the end of the forging. The distribution of the state variables can be plotted for any other stages of forging as well. Figure 16.3(b) shows the effective strain versus time of a material point located at midheight of the bore section of the forging, as shown in Fig. 16.3(a). In this isothermal forging case, a 20 min deformation time was used, as shown in Fig. 16.3 (a) Effective strain distribution and (b) the effective strain history of the center location of a closed-die forging the figure. The final strain value, 1.5, shown in Fig. 16.3(b) is in agreement with the value shown in the distribution plot in Fig. 16.3(a). The history plot of state variables (strain, strain rate, and temperature) provides valuable information on the thermomechanical history of the forging that determines its mechanical properties. 16.5.3 Equipment Response/Hammer Forging Process modeling also provides the information regarding the response of the equipment. Examples of equipment response discussed here are forging load and ram velocity of hammer forging. The information is usually not available in the hammer shop. However, it is useful for understanding the hammer response to a forging process. Figure 16.4 shows the load versus stroke predicted for a hammer forging operation. The figure shows that there are eight blows in the hammer operation. Each ends with a zero load. The stroke in the figure is the stroke of the ram/die. The zero stroke refers to the position of the die, where the first die/workpiece contact occurs during forging. This zero position is the same for all of the eight hammer blows. With the increase in the number of blows, the load increases and the stroke per blow decreases. The last blow of the sequence has the shortest stroke. This behavior is very real for hammer forging operations. During a hammer forging operation, the workpiece increases its contact area with the dies, which increases the forging load. The total available blow energy is fixed for a hammer. With the increase in forging load, the length of Fig. 16.4 Load versus stroke obtained from a hammer forging simulation Fig. 16.5 Ram velocity versus stroke obtained from a hammer forging simulation
Process Modeling in Impression-Die Forging Using Finite-Element Analysis / 199 stroke is reduced. Moreover, the blow efficiency, which is the ratio between the energy used for deformation and the total blow energy, is also Fig. 16.6 Prediction of the distribution of the size (lm) of gamma prime for a Rene 88 experimental forging reduced with the increase in forging load. Thus, a smaller amount of energy is available toward the end of a blow sequence and with the decrease in the stroke per blow. Figure 16.5 gives the ram velocity versus stroke obtained from a simulation of another hammer forging process. There are nine blows for this hammer operation. The velocity of the first blow was smaller than the other eight blows, because a soft blow was used initially to locate the workpiece. In a soft blow, there is only a portion of blow energy applied to the workpiece. Thus, the first blow has a smaller starting ram velocity. After the first blow, full energy was applied to the forging. Thus, the starting ram ve- Fig. 16.7 Comparisons of hot-die forging and mechanical press forging of an experimental part using process modeling Fig. 16.8 Rene 88 experimental part out of forging press [Hardwicke et al., 2000] Fig. 16.9 Predicted model and optically measured grain sizes in the three developmental René 88DT disks with (a) coarse, (b) medium, and (c) fine grains [Hardwicke et al., 2000]
200 / Cold and Hot Forging: Fundamentals and Applications locity for the rest of the blows was the same. There is always an energy loss to surroundings in a hammer blow. Therefore, blow efficiency needs to be factored in for each hammer blow. However, the blow efficiency only has an effect after the ram/die workpiece are in contact. Hence, blow efficiency does not influence the starting velocity of the ram/die. It is factored in during the blow. The decay in ram velocity in each blow is a result of both the energy consumption in deforming the workpiece and the energy lost to the surroundings. 16.5.4 Microstructures in Superalloys Microstructure and property modeling is now the major emphasis in advanced forging process design and improvement, especially in forging aerospace alloys such as nickel and titanium superalloys. The development and utilization of physical metallurgy-based microstructure models and the integration of the models with finiteelement analysis has allowed for microstructure prediction by computer. Two important microstructural features of superalloy forgings are the grain size and the gamma-prime precipitation. The grain size modeling is discussed in detail in Chapter 19, Microstructure Modeling in Superalloy Forging. The prediction of gammaprime distribution is discussed here. Gamma prime is a very important precipitation phase in strengthening superalloys. The size and spacing are two features of interest in gamma-prime precipitation. Figure 16.6 shows the prediction of the distribution of the size of gamma prime of an experimental nickel-base superalloy forging, Rene 88, coupled with a few measurement points. The measurement made is in the range of 0.07 to 0.21 lm. The model predicts a range of 0.08 to 0.14 lm. The fine gamma prime was correctly predicted and the coarser gamma prime was underpredicted, which pointed out the need for further improvement of the gamma-prime model. The microstructure prediction feature is useful for the process development for closeddie forging. 16.6 Examples of Modeling Applications One of the major concerns in the research of manufacturing processes is to find the optimum production conditions in order to reduce production costs and lead-time. In order to optimize a process, the effect of the most important process parameters has to be investigated. Conducting experiments can be a very time-consuming and expensive process. It is possible to reduce the number of necessary experiments by using FEM-based simulation of metal forming processes. Fig. 16.10 Investigation of defects in ring gear forging using FEM [Jenkins et al., 1989]