CIRP Annals - Manufacturing Technology 57 (2008) 305–308
Contents lists available at ScienceDirect
CIRP Annals - Manufacturing Technology journal homepage: http://ees.elsevier.com/cirp/default.asp
Finite element analysis of die wear in hot forging processes B.-A. Behrens Institute of Metal Forming and Metal-Forming Machines, Leibniz Universita¨t Hannover, An der Universita¨t 2, 30823 Garbsen, Germany Submitted by E. v. Finckenstein (1), Dortmund, Germany
A R T I C L E I N F O
A B S T R A C T
Keywords: Forging Wear Finite element method (FEM)
Design optimization of hot forging dies requires an accurate estimation of die wear. The presented paper introduces a ﬁnite element model for wear estimation that includes the process related thermal effects on hardness of the tool material. Fundamental investigations concerning the hardness evolution due to thermal softening of the tool material are presented. To obtain necessary data for model calibration by means of statistical analysis, optical measurements are performed on several industrial forging dies. The introduced model is proved to be applicable in wear estimation of hot forging dies over a large number of operating cycles. ß 2008 CIRP.
2. Fundamental investigations
Efﬁciency of hot forging processes is tightly related to the tool service life, which is mainly limited by wear . In hot forging of steel high thermal loads due to workpiece temperatures up to 1300 8C lead to a thermal softening of conventional tool materials. This is caused by an excess of the tempering temperature associated with a gradual hardness loss of the tool surface layer [2,3]. Hence, the tool wear rate increases over a number of forging cycles. An optimized process design taking into account tool wear is a way to extend tool service life. For a reliable wear prediction, it is necessary to include the decrease of hardness into the calculation via an appropriate approach. So far, there have been several research works dealing with the simulation of tool wear in hot forging taking thermal effects into account as well [4–6]. The authors regard the decrease of hardness due to thermal softening, but the assumptions concerning the present hardness and microstructure within the tool surface layer are not validated with experimental data. Furthermore, the introduced models are only applied to one forging process in each case. There is no investigation concerning versatility to different processes. The main objective of the work presented in this paper is a tool wear model taking into account thermal softening that features an improved versatility concerning multiple processes. To allow a realistic numerical prediction of the tool wear with an improved accuracy, extensive data from tests and industrial processes are used for the development and calibration of the ﬁnite element model. Based on the experimentally determined wear and microstructure evolution in the tool surface layer on a practiceoriented hot forging process, a calculation method is developed, taking the thermal inﬂuences on the wear into account. After that, the wear model is extended in the application on industrial processes, in order to increase the calculation accuracy. An advanced model calibration is done with practical industry data by regarding varying processes and tool materials.
2.1. Forging tests
0007-8506/$ – see front matter ß 2008 CIRP. doi:10.1016/j.cirp.2008.03.087
Experimental wear investigations were carried out to establish a basis for the development of a realistic tool wear model. The used forging process with a rotation-symmetric part geometry is shown schematically in Fig. 1. As most wear is expected at the mandrel, it was replaced after 500, 1000 and 2000 forging cycles, in order to investigate wear progress and hardness within the surface layer. By comparison of the initial and the worn geometry of the mandrel, the wear was determined. 2.2. Modeling of tool wear The wear model according to Ref.  was changed such that the wear depth w can be computed for each forging cycle in the simulation (Eq. (1)). w¼k
Hð#; t t Þ
vrel Dt inc
In this equation sN represents the contact normal stress, vrel refers to the relative velocity between workpiece and tool, and Dt represents the duration of one time increment (inc) within a thermal–mechanical coupled FE simulation with elastic tools. The hardness H is not considered constant, as done by Archard , but depends on temperature W and the process duration tt of the numbers of forging cycles. Due to the complexity of the tribological effects between workpiece and tool surface, not all process-speciﬁc factors are taken into account. Thus, a calibration of the wear model using the wear coefﬁcient k is required to calculate the wear depth quantitatively. By matching the computed and measured wear proﬁles, the wear coefﬁcient k can be determined. For wear determination at every surface node after any number of forging
B.-A. Behrens / CIRP Annals - Manufacturing Technology 57 (2008) 305–308
Fig. 1. Considered forging process. Fig. 3. Comparison of measured und calculated hardness after 2000 forging cycles at mandrel.
cycles, a discrete program which uses the data from FE simulation of one cycle was developed. The surface layer hardness was determined by means of the main tempering curve for the tool material and the tempering parameter P. As the change of the microstructure only takes place in the tool surface layer, the maximum temperature at every surface node was used for determination of the tempering hardness. The tempering parameter P according to Ref.  was determined using the maximum temperature and the process duration, resulting from the cycle duration and the number of forging cycles (Eq. (2)). P ¼ T max ð20 þ lgt t Þ
The standard procedure for the tempering hardness determination using the hot work steel 1.2365 (DIN EN standard) is shown in Fig. 2. Since the calculated tempering hardness exists at room temperature, the hardness was considered at process temperature using a hot hardness curve from literature . 2.3. Results The described wear model was applied and veriﬁed using the forging process shown in Fig. 1. First, the microhardness was measured 20 mm below the mandrel surface using Vickers hardness test with an indentation force of 0.98 N. The Vickers hardness was converted to the Rockwell hardness in order to use consistent units. In Fig. 3 the measured and calculated tempering hardness after 2000 forging are compared. It can be seen, that the hardness decrease from initial value of 53.5 HRC due to tempering
is considered within the simulation. The high hardness of 59 HRC at mandrel radius results form a forming of a so-called white layer during the ﬁrst forging cycles due to the fast cooling-down of the heated surface zone at cold tools . This white layer was nearly completely worn off after 2000 cycles, leaving the tempered layer at the mandrel surface and thus leading to increased wear. Only at small radii with very high thermal loadings a formation of the white layer takes place. Outside of the white layer, the hardness was computed with regard to the tempering effects. The experimentally determined value was assumed for the hardness of the white layer in the mandrel radius area for calculation of wear. At higher number of forging cycles the white layer becomes less important. After 500 forging cycles, tactile measurements at the mandrel radius showed not only abrasive wear but also plastic deformation, caused by the white layer sliding on the thermally softened layer. Therefore, the suitability of the wear model was veriﬁed by wear calculation between 1000 and 2000 forging cycles, since this interval, there was only abrasive wear. The geometry of the mandrel after 1000 forging cycles was used in the simulation. Moreover, the duration of the ﬁrst 1000 cycles was taken into account for the hardness in the mandrel surface layer. The wear coefﬁcient k was deﬁned such that the correct calculation of the maximum wear depth was possible. Fig. 4 shows the measured and computed wear at the mandrel between 1000 and 2000 forging cycles. The curves of measurement and simulation match quite well. This indicates the suitability of the proposed model for wear calculation in hot forging considering thermal softening. 3. Transfer to industrial processes 3.1. Extension of wear model The phenomenological wear model described in chapter 2 takes essential factors inﬂuencing the wear of tools into account, in particular the thermally caused hardness loss of near surface areas
Fig. 2. Procedure for tempering hardness determination. (See Ref. .)
Fig. 4. Comparison of measured und simulated wear proﬁles at mandrel.
B.-A. Behrens / CIRP Annals - Manufacturing Technology 57 (2008) 305–308
Fig. 5. Main tempering curves of the considered hot work steels.
of forging dies. However, due to the complexity of the tribologic conditions within the intermediate layer between workpiece and tool, not all process-speciﬁc and operational effects can be depicted in detail. For an improved and more general quantitative estimation of the local tool wear after large numbers of process cycles, the computational model was additionally calibrated with measured data from industrial processes. As multiple measurements of the tool geometries are not possible in industrial production, the wear behavior was determined by measuring workpiece geometries which were taken from the production process in predeﬁned cycle intervals. This approach offers the possibility to identify tool wear with adequate accuracy, as the workpiece geometry represents the tool shape in bulk metal forming processes. Nevertheless, descaling was carried out carefully with an appropriate grit to avoid modiﬁcation of the workpiece geometries. Furthermore, three components were analyzed for each cycle interval to obtain a statistical estimation. The geometrical range of the considered industrial processes includes two rotationally symmetric and two more complex 3Dgeometries. Tool systems for forward extrusion, hot forging of a wheel hub, bucket teeth of a hydraulic shovel and a driveshaft were analyzed up to 14,000 process cycles in particular cases. The considered die components cover a selection of relevant hot work steels. Material-speciﬁc hot hardness characteristics for the hot work steels 1.2714, 1.2344 and 1.2367 (DIN EN standard) were described analytically via approximation functions and embedded into the wear model with respect to usual heat treatment conditions . In addition, speciﬁc main tempering curves for the considered hot work steels were obtained from technical data sheets of steel manufacturers. The curves were implemented into the computational model by analytical functions, shown in Fig. 5.
Fig. 6. Determination of the wear distribution of an industrial forging punch by optical measuring techniques.
Fig. 7. Systematic parameter calibration by means of statistical evaluation.
3.2. Advanced model calibration The geometrical measurements were carried out with the optical measuring system GOM ATOS that allows an effective 3Ddigitalisation of the forgings with an adequate accuracy as well as acceptable time and effort. For the determination of the wear progress, certain characteristic areas of the geometry were investigated. Based on the wear distributions measured at the workpieces for different cycle numbers, wear proﬁles were evaluated within plane cross-sections, which allow a detailed comparison with the simulation results. Fig. 6 exempliﬁes data collection with optical measuring technology for a tool component for hot forging of bucket teeth of a hydraulic shovel. The forging punch is made of hot work steel 1.2714 (DIN EN standard). Considering the extensive amount of real and simulation results, the systematic calibration of the wear model was realized by means of statistical methods. For input data, local wear depths were each evaluated in discrete positions of the analyzed plane crosssections as shown in Fig. 6 in principle, and the real values were correlated with the respective simulation results. To assess the inﬂuence of relative velocity vrel on the wear in relation to the inﬂuence of the contact normal stress sN between workpiece and tool and the inﬂuence of the temperaturedependent hardness H of the tool material, the computational approach was extended by the model parameter a (Eq. (3)). w¼k
Hð#; t t Þ
vrel Dt inc
The identiﬁcation of the wear coefﬁcient k and the adjustment of the exponent a was done for the considered industrial processes
Fig. 8. Veriﬁcation of the computational model regarding an industrial hot forging process of a driveshaft.
B.-A. Behrens / CIRP Annals - Manufacturing Technology 57 (2008) 305–308
Fig. 9. Comparison of measured and simulated wear proﬁles at certain numbers of process cycles.
and hot work steels within a statistical experiment evaluation. For that purpose, the ranges for the model parameters k and a were speciﬁed iteratively. Herein, optimal parameter values k and a, with respect to the wear depth w as the target variable, were determined by means of factorial two-step plans to match the real data best possible. The identiﬁcation of the parameter sets was carried out for each considered hot work steel according to particular components of the analyzed tool systems. Exemplary, Fig. 7 illustrates the data analysis of one cross-section of the mentioned punch tool for hot forging of bucket teeth.
An effective possibility for the extension of tool service life is an optimized process design taking into account tool wear. The paper introduces a ﬁnite element model for wear estimation with improved versatility which includes the process related thermal effects on hardness of the tool material. Extensive data from experiments and industrial processes are used for the development and calibration of the approach. The considered industrial processes cover a range of relevant hot work steels. The calibrated wear model allows the comparison of different hot work steels regarding their wear resistance within the process design by means of FEA. Thus, it is possible to enhance the process design concerning the die service life and the selection of an adequate die material. The approach is proved to be applicable in wear estimation of hot forging dies over a large number of operating cycles. At present, we are working on the further development of the computational model which will consider the change in tool surface geometry due to wear progress, thermochemical surface treatments and coatings. Acknowledgments The author would like to thank the German Research Foundation (DFG) for the ﬁnancial support of the presented work within the Collaborative Research Center 489 ‘‘Process chain for the production of precision-forged high performance components.’’ Moreover, the author is much obliged to the Industrial Cooperative Research Associations (AiF) and the Research Association of Steel Forming (FSV).
3.3. Veriﬁcation References The modeling approach for die wear calculation was veriﬁed based on an industrial process for the hot forging of a driveshaft, shown in Fig. 8. The model was applied to the upper die of the ﬁnal process stage, made of hot work steel 1.2367 (DIN EN standard), using the appropriate model parameters k and a for this tool material. Identiﬁcation of this parameter set was based on the analysis of two die components of a tool system for hot forging of a wheel hub and one forward extrusion die. Measured data and FEA results of the wear depth were analyzed along the outline of the tool surface that is also represented in Fig. 8. Fig. 9 shows a comparison of measured and simulated wear proﬁles after 4000 and 8000 process cycles exemplary. Regarding characteristic features of the wear proﬁles, the FEA results show a good correlation to the measured data. In areas with maximum wear amount, e.g. the convex radius at the top of the punch, the quantitative estimation of the wear depth is satisfying. The consideration of thermal softening effects on the tool material allows considering the increasing progress of the abrasive wear between 4000 and 8000 process cycles.
 Heinemeyer D (1976) Untersuchungen zur Frage der Haltbarkeit von Schmiedegesenken, Dr.-Ing. Thesis, Hannover.  Bobke T (1991) Randschichtpha¨nomene bei Verschleißvorga¨ngen an Gesenkschmiedewerkzeugen, Dr.-Ing. Thesis, Hannover.  Walter S (1999) Beitrag zu den Werkstoffversagensmechanismen beim Gesenkschmieden, Dr.-Ing. Thesis, Hannover.  Kang JH, Park IW, Jae JS, Kang SS (1999) A Study on a Die Wear Model Considering Thermal Softening. I. Construction of the Wear Model. Journal of Materials Processing Technology 96:53–58.  Kang JH, Park IW, Jae JS, Kang SS (1999) A Study on a Die Wear Model Considering Thermal Softening. II. Application of the Suggested Wear Model. Journal of Materials Processing Technology 94:183–188.  Lee HC, Kim BM, Kim KH (2003) Estimation of Die Service Life in Hot Forging Considering Lubricants and Surface Treatments. Proceedings of the Institution of Mechanical Engineers 217:1011–1022.  Archard JF (1953) Contact and Rubbing of Flat Surfaces. Journal of Applied Physics 981–988.  Liedtke D, Joensson R (2000) Wa¨rmebehandlung: Grundlagen und Anwendungen fu¨r Eisenwerkstoffe, Expert-Verlag, Renningen-Malsheim, 4. durchges. Auﬂ.  Schneider R (1984) Verschleißvorausbestimmung bei Gesenkschmiedewerkzeugen. Drahtwelt 11:299–302.  NN, Boehler Edelstahlhandbuch 2.1 Version 2.1e, Boehler Edelstahl GMBH.  Schruff I (1989) Zusammenstellung der Eigenschaften und Werkstoffkennwerte der Warmarbeitssta¨hle X38CrMoV51, X40CrMo51, X32CrMoV33 und X38CrMoV53. Thyssen Edelstahl Technische Berichte 15(2):70–81.