Journal of Materials Processing Technology 74 (1998) 74 – 82

Study of stress development in axi-symmetric products processed

by radial forging using a 3-D non-linear finite-element method
D.Y. Jang a,*, J.H. Liou b

a
Industrial Engineering Department, Seoul National Polytechnic Uni6ersity, Seoul, South Korea
b
Aircraft Certification Institute, Taipei, Taiwan, ROC

Received 18 October 1996

Abstract

A hollow shafts popular axi-symmetric machine elements, are used frequently under severe service conditions such as for
automobile transmissions, the shafts of aircraft, and turbine shafts. They are usually manufactured through the radial-forging
process. The residual stress in these forged products directly affects the material stability, the resistance to deformation, the
accuracy, and the fatigue life of products. In order to extend the operating life of products and satisfy the required quality of
operation during customer usage, it is necessary to monitor the residual stresses during the forging operations. As a way to study
residual stress formation due to radial forging, a three-dimensional theoretical model was developed using a non-linear
finite-element method program. The workpiece used in the calculations was alloy steel MILS 11595 and was assumed to have
elasto-plastic behavior during forging. The Coulomb friction law was applied to simulate the contact behavior between the
workpiece and the mandrel and between the workpiece and the die. The residual stress distributions in the forged product were
calculated and selected results were compared with published experimental data to verify the effectiveness of the developed model.
© 1998 Elsevier Science S.A.

Keywords: Axi-symmetric products; Stress development; Radial forging

1. Introduction Radial forging is a cost-effective and material-saving

forming process for reducing the cross-sections of rods,
The surface condition of a product can be rated tube and shafts that are made of any metal that is
according to the surface quality, the latter being deter- suited for cold forming. The advantages of radial forg-
mined by properties such as surface roughness, hard- ing are smooth surface finish, considerable material or
ness variation, structural change, and residual stress. weight savings, preferred fiber structure, minimum
Amongst these properties, the residual stress distribu- notch effect, and increased material strength [2]. The
tion is one of the primary aspects of surface quality forged components processed by radial forging will
because of its direct effect on the fatigue life and generally have higher residual stresses and undergo
dimensional stability of a product. A stress field on the some deflections after forging due to the resultant high
surface of a product can cause a crack to propagate residual stresses, thus losing dimensional accuracy.
rapidly, but it can also squeeze a crack shut. If the Residual stresses can also cause stress corrosion for
surface stresses are tensile and tend to pull the material some materials when used in a corrosive environment.
open, they accelerate the crack growth and assist the The residual stress distribution in high-strength metal is
destruction of the object. On the other hand, any small of great importance since its presence, above a particu-
cracks in a compressive stress field opened by corrosive lar level, can shorten the service life of critical compo-
or local deformation do not grow as quickly as they nents used under severe service conditions.
would otherwise [1]. Either experiment or theoretical simulation may be
used to study the residual stress formation due to radial
* Corresponding author. Fax: + 822 9742849; e-mail: dy- forging. However, the experimental approach could be
[email protected] very expensive and time-consuming when the residual

0924-0136/98/$19.00 © 1998 Elsevier Science S.A. All rights reserved.

PII S 0 9 2 4 - 0 1 3 6 ( 9 7 ) 0 0 2 5 2 - 5
 D.Y. Jang, J.H. Liou / Journal of Materials Processing Technology 74 (1998) 74–82 75

stresses are measured by the X-ray or neutron diffrac- been applied to three-dimensional deformation in plate
tion method. During the last decade, the finite-element and edge rolling by More and Osakada [21] and to the
method (FEM) has received considerable attention as analysis of spread in rolling by Li and Kobayashi [22].
being a very powerful computing technique for the Park and Kobayashi [23] solved a three-dimensional
analysis of metal forming processes [3 – 5]. An analytical finite-element problem using the rigid–visco-plastic for-
model using the FEM technique, developed based on mulation for the compression of a simple rectangular
the mechanics of the forging operation, enables the block and a wedge-shape block between two parallel
analysis of the deformed profile, the forming process dies. Side pressing of pipes and the upsetting of rectan-
and the stress distributions including residual stresses, gular prisms were investigated by More et al. [24] for
in a product. By introducing computer simulations at simulations of three-dimensional deformations in metal
the stage of process design, it is possible to verify the forming. Osakada and More [25] analyzed the practical
design for manufacturing and bring the process design limit of applying finite-element analysis in two-dimen-
closer to completion before testing it on the actual sional and three-dimensional metal forming processes
machine. Furthermore, through the stress analysis of a for a 16-bit micro-computer and a super-computer.
product, the post-processing condition can be estimated Tanaka et al. [26] analyzed three-dimensional void-
to relieve residual stresses and decide the optimum crushing during the free forging of a round bar with
operational conditions for forging that will yield favor- opposed flat dies using the general-purpose commercial
able stress conditions in the product. FEM code MARC. With the initial velocity field
Principally, there are three different finite-element guessed, a three-dimensional finite-element analysis of
formulations employed to simulate metal forming: (i) the forging of blocks with continuous curves dies by
the rigid–plastic model; (ii) the elasto-plastic model; considering frictional boundary was investigated by
and (iii) the visco-plastic model. The rigid – plastic Yoon and Yang [27]. Using Bezier curves, Shiau and
model is most commonly used due to its simpler formu- Kobayashi [28] implemented geometrical modelling of
lation and better numerical efficiency [6]. The rigid– the die shape in performing three-dimensional finite-ele-
plastic FEM, which was developed by Lee and ment analysis. Sawamiphakdi et al. [29] investigated the
Kobayashi [7], has been applied to the analysis of residual stresses in drawn wire by an elasto-plastic
various problems such as solid cylinder upsetting, ring formulation, a constant frictional shear stress being
compression, extrusion, and sheet bending [8 – 13]. Since assumed at the die and workpiece interface. Sadok et
the elastic strains are neglected, stresses below the yield al. [30] investigated the influence of the die geometry on
stress are not known, which implied that effects such as the field of strains in the drawing process, concluding
residual stresses and spring-back, which could be im- that a convex die was the most appropriate shape when
portant in connection with metal forming, could not be uniformity of strains and metal flow were taken into
predicted. The application of the elasto-plastic formula- consideration. Liu and Chung [31] predicted the work-
tion to metal forming problems was credited to several ability in conical die extrusion using Coulomb’s friction
researchers: Lee et al. [14], Yamada et al. [15], and law to calculate the friction stresses on the die –work-
Cheng and Kikuchi [16]. They used this model in piece interface. Altan and Knoerr [32] summarized the
uniaxial tension tests, the upsetting of a cylindrical functions of the finite-element code DEFORM, which
block, an indentation problem, and a head-forming was developed by Battelle Columbus Laboratories. This
process. The application of the visco-plastic formula- code added an automatic mesh generation (AMG)
tion to metal-forming was first studied by Zienkiewicz module and was applied successfully to bevel-gears and
et al. [17] and developed further by Onate and elbow-pipe design. They also introduced the computer-
Zienkiewicz [18]. The visco-plastic model, which ac- aided engineering (CAE) techniques in metal-forming
counts for the effect of rate-dependent materials, was research using the FEM. Cheng et al. [33] simulated a
found suitable for modelling hot metal-forming pro- billet upsetting process using the explicit transient non-
cesses, whilst the elasto-plastic formulation was consid- linear finite-element code LS-DYNA3D. They used an
ered to be the appropriate model to represent the elasto-plastic constitutive model with a combination of
material behavior during cold metal-forming processes. Coulomb friction and constant shear friction for the
Oh [19] incorporated a method of discretizing the die contact interface. Domblesky et al. [34] used the rigid–
boundary conditions into a user-oriented computer thermo-visco-plastic FEM to study the temperature
code that was capable of analyzing general plane-strain gradients and strain rate related to the feed rate in
and axi-symmetric problems with arbitrarily shaped radial forging. Their results showed that the strain rate
dies. Sun et al. [20] used simplified three-dimensional and temperature appeared to be influenced by the axial
elements to analyze three-bite flat bar forging. They feed rate to a limited extent.
assumed that the longitudinal velocity components of The only published work conducted to study the
any material point are independent of its co-coordinate residual stresses in the radial forging process was made
in the transverse direction. Similar approaches have by Tszeng [35], who used an axi-symmetric model of
76 D.Y. Jang, J.H. Liou / Journal of Materials Processing Technology 74 (1998) 74–82

Fig. 2. Finite-element model for the simulation of radial forging.

Fig. 1. A schematic diagram showing the arrangement of the forging stresses being in the opposite direction to the relative
dies in the radial forging machine. movement of the workpiece with respect to the die. (c)
The residual stress distributions for each bite-operation
the rigid–plastic FEM code ALPID in the simulation and final bite-operation were calculated. Selected simu-
of the loading process, followed by an elastic program lation results were compared to published experimental
for the unloading process. His results indicates that the data to verify the effectiveness of the developed model
friction factor at the die – workpiece and mandrel– [36].
workpiece interfaces has a more significant effect on the A tool which simulates the manufacturing process
residual stress formation: specifically, smaller friction can provide design engineers with complete information
factors result in lower residual stress levels. Also, on the characteristics of product quality. Designers can
smaller die angles make the residual stresses more then ascertain whether or not a particular manufactur-
compressive. However, since the elastic strains were ing process adversely affects the quality of products.
neglected, the stresses below the yield stress were not They can then decide remedial manufacturing proce-
known, which implied that effects such as the spring- dure, if necessary, in the design stage without requiring
back effect could not be predicted. Furthermore, the an actual manufacturing operation. An integrated man-
two-dimensional model cannot predict the stresses in ufacturing system can also be used for product/process
the u direction, which could be important in the radial optimization of precision manufacturing, which will
forging process. produce products with the desired quality at the maxi-
Therefore, in this study a three-dimensional FEM mum production rate.
model was developed to calculate the residual stresses
in a product resulting from the radial forging process.
The workpiece, alloy steel MILS 11595, was assumed to 2. Formulation of the forging operation and the
have elasto-plastic behavior during the cold forging theoretical model developed using the FEM
process. The contact areas between the workpiece and
the mandrel and between the workpiece and the die 2.1. Formulation
were assumed to follow the Coulomb frictional law.
The simulation was conducted using the ANSYS finite- In the finite-element analysis, the workpiece was as-
element code in the Pittsburgh Super Computer Center, sumed to obey the Von-Mises yield criterion and its
the following subjects being investigated: (a) The defor- associate flow-rule. The deformation process of the
mation shapes at various stages of each bite-operation elasto-plastic materials in the analysis was assumed to
were calculated. Accordingly, distributions of effective
strain and of the stresses in different directions on the Table 2
outer and inner surfaces for each bite operation were Forming geometry and process conditions
displayed. (b) The boundary conditions were investi-
Geometry and process conditions
gated. Frictional stresses were assumed to exist at the
interface between the die and the workpiece, these Geometry of the workpiece
Outside radius of workpiece 33 mm
Table 1 Inside radius of workpiece 7.94 mm
Elements and boundary conditions Mandrel radius 7.94 mm
Outside radius of forged tube 30 mm
Items No. of ele- Type of ele- Boundary con- Reduction in area 18.30%
ments ment ditions
Die configuration
Length of die land 12.7 mm
Die 42 SOLID 45 Rigid
Die inlet angle (a) 6°
Workpiece 300 VISCO 107 Symmetry on
both sides Process conditions
Mandrel 10 SOLID 45 Rigid Friction Rigid Coulomb friction
Contact areas 2418 CONTACT 49 With frictional Friction coefficient 0.05
force Contact stiffness 4.0×106 N mm−1
 D.Y. Jang, J.H. Liou / Journal of Materials Processing Technology 74 (1998) 74–82 77

Table 3
Forging parameters and experimental residual stresses

Die inlet angle Tube diameter Reduction (%) Land length Feed rate (mm Back pressure Peak stresses su /sz
(°) (mm) (mm) rev−1) (MPa) (MPa)

6 33.0 18.3 12.7 1.63 14.5 248/335

Table 4
Comparison between experimental and FEM results

Hoop stress (MPa) Axial stress (MPa) Radial stress (MPa)

Experiment [36] 248 335 —

FEM simulation 232 317 173

be associated with the boundary value problem where where o; ij is the strain-rate and o; v is the volumetric
the stress and strain field solutions satisfy the equi- strain-rate. The final equation can be obtained by re-
librium equations, the constitutive equations in the placing the integral of the first term with the effective
domain, and the prescribed boundary values. stress s̄ and the effective strain-rate ō; . The incompress-
The weak form of the equilibrium equation, neglect- ibility constraint on the admissible velocity field in Eq.

&
ing body force, is expressed by: (3) can be removed by a penalty function constant, K,

& & &

as follows:
sij, jdni dV=0 (1)
6
dp= s̄dō; dV+ K o; vdo; v dV− Tidni dS=0
where sij is the component of stress tensor, dni is an 6 6 SF
(4)
arbitrary variation of velocity, and a comma denotes
partial differentiation. Using the divergence theorem, where s̄=
(3/2)s %ijs %ij, ō; =
(2/3)o; ijo; ij, and K is a large
and the symmetry of the stress tensor, and imposing the positive constant. The effective stress for a specific
essential boundary conditions that dn =0 on S6, where material can be determined by the uniaxial tension or
S6 is the specified velocity field boundary, Eq. (1) compression test as a function of the effective strain

& &
becomes: and the effective strain-rate [37].
Eq. (4) is the final form of the basic equation for the
1
sij (dni, j + dnj,i ) dV − sijni nj dS = 0 (2) finite-element discretization. Once the solution for the
2 6 SF velocity field that satisfies Eq. (4) is obtained, then the
where nj is the unit normal to the surface and SF is the corresponding stresses can be calculated using the flow
surface on which traction is prescribed, namely sijnj = rule and the known mean stress distribution.
Ti on SF. By decomposing the stress tensors into the Von Mises [38] first suggested that the basic behavior
deviatoric component s %ij and the hydrostatic compo- defining the plastic strain increment is related to the
nent sm, Eq. (2) becomes:
& & &
yield surface:
(Q
s %ijdo; ij dV+ smdo; v dV − Tidni dS = 0 (3) do pij = h dF (5)
6 6 SF (sij

Fig. 3. The instantaneous effective-strain distribution at stage 2. Fig. 4. The instantaneous axial-stress distribution at stage 3.
78 D.Y. Jang, J.H. Liou / Journal of Materials Processing Technology 74 (1998) 74–82

Fig. 5. The instantaneous radial-stress distribution at stage 3. Fig. 6. The instantaneous hoop-stress distribution at stage 3.

During an infinitesimal increment of stress, the 2.2. FEM model

changes of strains are assumed to be divisible into
elastic and plastic parts, thus: Deformation in the radial forging process is due to
many continuous short-stroke, high-speed, and radial-
{do} = {do e}+ {do p} (6)
inwards striking dies. The radial hammer dies, usually
but: four, are arranged radially around the workpiece as
{do e} = [De] − 1{ds} (7) shown in Fig. 1. There are one or two chuck heads on
the radial forging machine for holding the workpiece in
(F the proper position and for feeding the dies. In order to
do p =dl (8)
(s obtain round forged workpieces and better finished
with [De] being an elasticity matrix, s is tensile stress surfaces, the workpiece rotates during the interval be-
tween two strokes. The rotating chuck heads stop dur-
! "
and dl is a positive proportionality constant, then:
ing the inwards motion of the forging dies to prevent
(F
{do} =[De] − 1{ds}+ dl (9) the workpiece from twisting. Radial forging is generally
(s
used in the forming of co-axial bimetallic tubing, in
When plastic loading occurs the stresses are on the reducing the wall thickness of tubes, and in generating
yield surface given by the yield function, which, when internal profiles [35].

! " ! "
differentiated, provides: Friction between the workpiece and the die plays an
(F T
(F important role in determining the forming forces and
{dF} = {ds}+ dk =0 (10) strain distributions in the forged parts. It also controls
(s (k
the incidence of failures due to cracking, scoring, wrin-
where Q and h are the scalar functions of the invariant kling, tearing, and geometric variability. Therefore,
of deviatoric stresses and F is the yield function. Q is how to treat the boundary condition in FEM, reflecting
called the plastic potential. From Eqs. (9) and (10):
the lubrication layer between the surface of the work-
{ds} = [Dep]do (11) piece and the die, is a key factor in securing an accurate

! "! "
and:
(F T
(F
[Dep] = [De]− [De] [De]
(s (s
! ! " ! ""
H+
(F T
[De]
(F −1
(12)
(s (s
The elasto-plastic matrix [Dep] takes the place of the
elasticity matrix [De] in the incremental analysis. Eqs.
(11) and (12) are the basic elasto-plastic stress and
strain relationship. For a Von Mises yield surface when

!
the associated flow rule, H is the plastic modulus and F
is given
1 as [39]: 1 1
F = (s1 −s2)2 + (s2 −s3)2 + (s3 −s1)2 +3t 212

"
2 2 2
1/2
+ 3t 223 + 3t 231 −sy (13) Fig. 7. The instantaneous axial-stress distribution at stage 5.
 D.Y. Jang, J.H. Liou / Journal of Materials Processing Technology 74 (1998) 74–82 79

Fig. 8. The instantaneous radial-stress distribution at stage 5.

model of the forming processes. In order to handle the

contact and friction between the workpiece and the die
and the workpiece and the mandrel, a spring-stiffness Fig. 10. The residual-stress distribution on the outer surface of the
relationship between two contact areas was established. workpiece.
The spring will deflect an amount D such that Eq. (14)
is satisfied: essary to develop a suitable representation of the
forging dies, the mandrel, the contact areas, and the
Fn = KsD (14) workpiece. Three different types of elements were used
in modelling the forging process, the detailed informa-
where Ks is the spring stiffness and Fn is the normal tion of the modelling and the boundary conditions
force between the contact surfaces. This method to being given in Table 1.
enforce contact compatibility is called a penalty method Additional assumptions were made to obtain reason-
[40]. The amount of penetration or incompatibility
ably accurate results: (i) rotational feed was neglected;
between the two surfaces is dependent on the stiffness
(ii) the material was assumed to have elasto-plastic
Ks. Friction in the contact surfaces was simulated by
behavior during the manufacturing process; (iii) the
the tangential force of the contact elements. The Cou-
deformation process was assumed as isothermal; and
lomb law states that the sliding force Fs transmitted
(iv) the deformation of the forging dies and the mandrel
between the two bodies cannot exceed a fraction of the
was neglected.
normal force Fn as follows:
In the simulation, the die and the mandrel were
Fs 5 − mFn (15) assumed to be rigid bodies with a high value of elastic
modulus. The workpiece was assumed to be symmetric
where m is the coefficient of friction. Once Fs is ex- on both sides, which latter were composed of r– z
ceeded, the two bodies will slide over each other. In this planes. The die was first moving radially inwards to
study, the rigid Coulomb model was applied since in make contact with the workpiece, then the workpiece
the radial forging process the workpiece slides continu- was moved stroke-by-stroke in the axial direction. For
ously between the die and the mandrel.
Due to the axi-symmetric characteristics of the radial
forging process, only a section of workpiece was simu-
lated. To model manufacturing operations, it was nec-

Fig. 11. The residual-stress distribution on the inner surface of the

Fig. 9. The instantaneous hoop-stress distribution at stage 5. workpiece.
80 D.Y. Jang, J.H. Liou / Journal of Materials Processing Technology 74 (1998) 74–82

performed on the CRAY90 in the Pittsburgh Super

Computer Center (PSC). The ANSYS51 (the maximum
wavefront is 1500) FEM code was used in this study.
The result files were transferred through the file transfer
protocol (FTP). The post-processing was performed on
the VAX computer of the UMC.

3. Results and analysis

The effectiveness of the FEM simulation of the radial

forging operation was verified by comparing simulated
results with the results of Hoffmanner’s [36] experi-
ments, the forging parameters and peak residual
stresses from his work being listed in Table 3.
Fig. 12. The residual-stress distribution with respect to thickness.
In Hoffmanner’s work, the residual stresses in the
workpiece were measured after the dimensional inspec-
each stroke (increment), a new equilibrium position was
tion of each part. The Sach’s boring-out technique was
obtained using the Newton – Raphson iteration method,
utilized with the boring being performed on a 300 A
the new equilibrium position being applied as the refer-
ANOCUT ECM machine using a NaCl (0.12 kg l − 1)
enced position for the next increment. The stroke was
electrolyte. The axial extension (Dz) over 64 mm and
continuous until the workpiece passed through the die.
the diametral expansion (Dr) across the blank diameter
In the simulation of the forging process, the whole
were measured by linear variable differential transducer
process was divided into five stages, each stage includ-
(LVDT), attached to the workpiece with stainless-steel
ing hundreds of strokes, the number of strokes depend-
fixtures. Although this system was calibrated to a sensi-
ing on the contact area. The final stresses in the
tivity of 0.05× 10 − 3 mm, it was usually applied at the
workpiece are the residual stresses due to the forging
gain setting to provide a sensitivity of 0.5× 10 − 3 mm
process. The three-dimensional model used in the simu-
[36].
lation is shown in Fig. 2, whilst the forming geometry
A summary of the results from three-dimensional
and process parameters are listed in Table 2. These
simulations using MILS 11595 material is shown in
conditions were obtained from published experimental Table 4, the forging conditions listed in Table 3 being
data [36]. utilized in the three-dimensional simulation of the forg-
In cold forging, the process is usually assumed to be ing process. As can be seen from Table 4, the peak
under isothermal conditions although the actual process residual stresses from the FEM simulations were 232
is non-isothermal. This assumption can provide good and 317 MPa in the hoop and axial directions, respec-
preliminary results when applied to processes which can tively. Table 4 shows good agreement between experi-
occur so quickly that die chilling effects are small [32]. ments and simulations, the differences being less than
In most radial forging analyses, the effect of die chilling 10%. The difference might be due to the thermal effect,
has been assumed to be negligible due to the relatively which was neglected in this study. It could also be
short time over which the die is in contact with the concluded that the mechanical loads dominated the
workpiece [34]. It has been estimated that tools are in residual stress developments whilst the thermal effect
contact with the workpiece for less than 10 – 15% of was negligible in the cold forging process. Comparison
each forging cycle [41]. between results from FEM simulation and experimental
Due to the limited computing facilities of the Univer- results showed that the elasto-plastic formulation was a
sity of Missouri-Columbia (UMC), calculations were proper way to formulate the cold forging simulation.
Using the proven FEM model as shown in Fig. 2,
calculations of strain and stress distributions, deforma-
tion, and residual stress were performed. The deforma-
tion during forging can be divided into three zones [34]:
sinking, forging, and sizing zones. In the sinking zone,
the workpiece was fed into the die and the mandrel
with only a small change in diameter, the effective-
strain in the sinking zone being approximately 10%, as
shown in Fig. 3. In the forging zone, the workpiece was
forged completely to the die land with most of the
Fig. 13. The effective-strain distribution after the forging process. deformation taking place in this zone. The effective-
 D.Y. Jang, J.H. Liou / Journal of Materials Processing Technology 74 (1998) 74–82 81

strain reached its maximum value of 30%. In the sizing the opposite direction, thus causing tensile stresses at
zone, the workpiece passed the die land and the strain the outer surface of the workpiece. The tensile radial
spring-back occurred due to the elastic or elasto-plastic stresses might have been generated due to the spring-
deformation. back effect of the workpiece in the radial direction.
The stress development in the axial direction for Figs. 10 and 11 show the stress distributions of the
stage 3 is shown in Fig. 4. At the beginning of forging, outer and the inner surfaces along the axial direction,
the outer surface of the sizing zone showed a small from which it can be seen that the outer surface was
tensile stress region. However, the sinking and the dominated by tensile residual stresses with the maxi-
forging zones were dominated by compressive stresses. mum value at the rear part of the workpiece. The axial
As the workpiece continued to feed into the die land, stresses had the minimum stress (− 48 MPa) at the
the tensile stress region increased and the compressive front part of the workpiece and increased along the
region decreased. An interesting result of the simulation axial direction to the maximum stress (300 MPa) at the
was that although the tensile stress region increased rear part of the workpiece. However, the inner-surface
during the forging process, the magnitude of the com- stresses consisted mostly of compressive residual
pressive stresses in the forging zone remained constant stresses, with only small tensile stresses at the front part
at each stage. The stresses in the radial direction at of the workpiece.
stages 3 of the forging process are shown in Fig. 5, Fig. 12 shows the residual stresses of the workpiece
displaying a similar pattern to that of the axial stresses, for different thicknesses. The axial stresses showed large
but the values of the tensile stresses in the radial variations between the inner and outer surfaces, but the
direction were smaller than those in the axial direction. radial and hoop stresses did not show significant
The maximum tensile stress was located at the begin- change with the thickness. The maximum tensile stress
ning of deformation, when the die initially contacted occurred at the outer surface of the workpiece along the
the workpiece. It was found that the radial stresses axial direction.
whilst in the sizing zone were smaller than those when The effective-strain after the forging process is shown
the workpiece had passed through the sizing zone, this in Fig. 13. The maximum strain was found to occur at
being due to the effect of spring-back (elastic recovery) the outer and the inner surfaces at the start of deforma-
after the sizing zone, and causing the existence of the tion, when the die initially contacted the workpiece, and
higher tensile stresses [9]. The radial stresses exhibited decreased radially through the wall to its minimum
radial variations, from compressive stresses at the inner value at the sub-surface. The results also showed that
surface to tensile stresses at the outer region. Fig. 6 the effective-strain was greater at the outer and the
represents the stress distribution of the hoop direction inner surfaces of the workpiece than at its sub-surface.
at forging stage 3, showing tensile stresses at the outer This might be due to the workpiece being subjected to
surface and compressive stresses at the inner surface, intense shear between the contact surfaces of the die
with the maximum tensile stress occurring at the middle and the mandrel.
part of the outer surface.
The residual stress distribution in the axial direction
is shown in Fig. 7. It can be seen that the residual 4. Conclusions
stresses were tensile at the outer surface with a maxi-
mum of 317 MPa and compressive at the inner surface. In the present study, a three-dimensional finite-ele-
The radial residual stresses of the forged product are ment model was developed to simulate the radial forg-
shown in Fig. 8, indicating the tensile stresses at the ing process. The workpiece was assumed to have
outer surface and compressive stresses at the inner elasto-plastic behavior during the cold forging process.
surface. The maximum tensile stress occurred at both In order to show the validity and effectiveness of the
ends of the workpiece with a maximum value of 173 developed finite-element model, the results of this study
MPa. The hoop residual stress distribution of the were compared with published experimental data, the
forged product is shown in Fig. 9, where it can be seen comparison between simulated results and the experi-
that there are tensile stresses at the outer surface of the mental data demonstrating that the elasto-plastic for-
workpiece with a maximum value of 232 MPa. Com- mulation is a proper model for the cold forging process.
paring the axial, the radial, and the hoop residual The results from this paper can be summarized as
stresses, the maximum tensile and compressive stresses follows:
were found in the axial direction. The tensile residual (1) The outer surface of the forged products was
stresses in the axial and hoop directions might have dominated by tensile residual stresses with their maxi-
been caused by the frictional force that always acted in mum values at the rear part of the forged products.
the opposite direction to that of the movement of (2) The inner surface of forged products showed
material, i.e. when the workpiece was fed into the die compressive residual stresses, with only small tensile
land, the frictional force tended to pull the workpiece in residual stresses in the front part of the workpiece.
82 D.Y. Jang, J.H. Liou / Journal of Materials Processing Technology 74 (1998) 74–82

(3) The highest tensile residual stress on the surface [18] E. Onate, O.C. Zienkiewicz, A viscous shell formulation for the
analysis of thin sheet metal forming, Int. J. Mech. Sci. 25 (1983)
of the forged products is in the axial direction.
305.
[19] S.I. Oh, Finite element analysis of metal forming processes with
arbitrarily shaped dies, Int. J. Mech. Sci. 24 (1982) 479.
[20] J. Sun, O. Li, S. Kobayashi, Analysis of spread in flat-tool
Acknowledgements forging by finite element method, in: Proceedings NAMRC,
Madison, WI, 1983, p. 224.
The authors acknowledge the support from the Pitts- [21] K. Mori, K. Osakada, Simulation of three-dimensional deforma-
burgh Super Computer Center (PSC) under Grant No. tion in rolling by finite element method, Int. J. Mech. Sci. 26
(1984) 515.
DDM9400029. The authors appreciate the financial [22] G.J. Li, S. Kobayashi, Analysis of spread in rolling by rigid–
support from the Faculty Development Fund of Seoul plastic finite element method, in: Numerical Analysis of Forming
National Polytechnic University. Processes, Wiley, New York, 1984, p. 77.
[23] J.J. Park, S. Kobayashi, Three dimensional finite element analy-
sis of block forging, Int. J. Mech. Sci. 26 (1984) 165.
[24] K. Mori, K. Nakadoi, M. Fukuda, Simulation of three-dimen-
References sional deformation in metal forming by rigid – plastic finite ele-
ment method, Adv. Tech. Plast. 2 (1984) 1009.
[25] K. Osakada, K. Mori, The use of micro- and supercomputer for
[1] I.C. Noyan, J.B. Cohen, Residual stresses in materials, Am. Sci.
formulation of metal forming processes, Ann. CIRP 34 (1)
79 (1991) 142 – 153.
(1985) 241.
[2] H. Yano, T. Akshi, Application of analytical simulation to
[26] M. Tanaka, S. Ono, M. Tsuneno, T. Iwadate, An analysis of
forged part design, Toyota Tech. Rev. 43 (2) (1994) 20.
void crushing during flat die free forging, Adv. Tech. Plast. 2
[3] J.P. Domblesky, R. Schivpuri, B. Painter, Application of the
(1987) 1035.
finite-element method to the radial forging of large diameter
[27] J.H. Yoon, D.Y. Yang, Rigid – plastic finite element analysis of
tubes, J. Mater. Process. Technol. 49 (1995) 57–74.
three-dimensional forging by considering friction on continuous
[4] J. Tang, W.T. Wu, J. Walters, Recent development and applica-
curved dies with initial guess generation, Int. J. Mech. Sci. 30
tions of finite element method in metal forming, J. Mater.
(1988) 887.
Process. Technol. 46 (1994) 117–126.
[28] Y.C. Shiau, S. Kobayashi, Three-dimensional finite analysis of
[5] G. Pasquinelli, Simulation of metal forming processes by the
open-die forging, Int. J. Numer. Mech. Eng. 25 (1988) 67.
finite element method, Int. J. Plast. 11 (5) (1995) 623–651.
[29] K. Sawamiphakdi, P.K. Kropp, G.D. Lahoti, Investigation of
[6] S. Tjotta, O. Heimlund, Finite element simulations in cold-forg-
residual stresses in drawn wire by the finite element method, J.
ing process design, J. Mater. Process. Technol. 36 (1992) 79.
Eng. Mater. Technol. 112 (1990) 231.
[7] C.H. Lee, S. Kobayashi, New solutions to rigid–plastic deforma- [30] L. Sadok, M. Packo, A. Skolyszewski, M. Ruminski, Influence
tion problems using a matrix method, J. Eng. Ind. 95 (1973) 865. of shape of the die on the field of strains in the drawing process,
[8] S.N. Shah, C.H. Lee, S. Kobayashi, Compression of tall, circu- J. Mater. Process. Technol. 34 (1991) 381.
lar, solid cylinders between parallel flat dies, in: Proceedings of [31] T.S. Liu, N.L. Chung, Extrusion analysis and workability pre-
the International Conference on Production Engineering, Tokyo, diction using finite element method, Comput. Struct. 36 (1990)
1974, p. 295. 369.
[9] H. Matsumoto, S.I. Oh, S. Kobayashi, A note on the matrix [32] T. Altan, M. Knoerr, Application of the 2D finite element
method for rigid – plastic analysis of ring compression, in: Pro- method to simulation of cold-forging processes, J. Mater. Pro-
ceedings of the 18th MTDR Conference, London, 1977, p. 3. cess. Technol. 35 (1992) 275.
[10] C.C. Chen, S. Kobayashi, Rigid–plastic finite-element analysis [33] W.L. Cheng, R.R. Verderber, J.O. Hallquist, Three dimensional
of ring compression, Appl. Numer. Meth. Form. Process. 28 simulation of forging processes, Numer. Meth. Simul. Ind. Metal
(1978) 168. Form. Process. 156 (1992) 95.
[11] C.C. Chen, S.I. Oh, S. Kobayashi, Ductile fracture in axisym- [34] J.P. Domblesky, R. Shivpuri, B. Painter, Application of the finite
metric extrusion and drawing—Part 1. Deformation mechanics element method to the radial forging of large diameter tubes, J.
of extrusion and drawing, J. Eng. Ind. 101 (1979) 23. Mater. Process. Technol. 49 (1995) 57.
[12] C.C. Chen, S. Kobayashi, Rigid–Plastic Finite Element Analysis [35] T.C. Tszeng, Residual stress calculation by finite element method
of Plane Strain Closed-Die Forging, ASM, Metals Park, Ohio, in manufacturing, Ph.D. dissertation, University of California,
1980, pp. 167. Berkeley, CA, 1987.
[13] S.I. Oh, S. Kobayashi, Finite element analysis of plane-strain [36] A.L. Hoffmanner, K.R. Iyer, Residual stress control in precision
sheet bending, Int. J. Mech. Sci. 22 (1980) 538. swaged rifle barrels, NAMRC VI (North America Manufactur-
[14] E.H. Lee, R.L. Mallett, R.M. McMeeking, Stress and deforma- ing Research Conference), Gainesville, FL, 1978, p. 180.
tion analysis of metal forming processes, in: Numerical Mod- [37] S. Kobayashi, S. Oh, T. Altan, Metal Forming and Finite
elling of Manufacturing Processes, ASME, PVP-PB-025, 1977, p. Element Method, Oxford University Press, New York, 1989.
19. [38] L.E. Malvern, Introduction to the Mechanics of a Continuous
[15] Y. Yamada, A.S. Wifi, T. Hirakawa, Analysis of large deforma- Medium, Prentice-Hall, Englewood Cliffs, NJ, 1969.
tion and stress in metal forming processes by finite element [39] O.C. Zienkiewicz, R.L. Taylor, The Finite Element Method,
method, in: H. Lippmann (Ed.), Metal forming Plasticity, IU- McGraw-Hill, New York, 1991.
TAM Symposium, Tutzing, Germany, 1978, pp. 158. [40] L. Jiang, R.J. Rogers, Combined lagrangian multiplier and
[16] J.H. Cheng, N. Kikuchi, An analysis of metal forming processes penalty function finite element technique for elastic impact anal-
using large deformation elastic-plastic formulations, Comput. ysis, Comput. Struct. 30 (1988) 1219.
Meth. Appl. Mech. Eng. 49 (1985) 71. [41] V. Nagpal, G.D. Lahoti, Application of radial forging to cold
[17] O.C. Zienkiewicz, E. Onate, J.C. Heinrich, A general formula- warm forging of cannon tubes, in: Selection of Die and Mandrel
tion for coupled thermal flow of metals using finite elements, Int. Materials, vol. 2, Battelle Columbus Laboratories, Columbus,
J. Numer. Meth. Eng. 17 (1981) 1497. OH, 1980, p. 62.