You are on page 1of 13

Chapter 15

Modeling Techniques in
Forming Processes
W.T. Wu, J.T. Jinn, and C.E. Fischer, Scientific Forming Technologies Corp.

THE OBJECTIVE OF MANUFACTURING This chapter reviews the overall development method. For detailed discussion, refer to text-
is the production of a consistent quality product of modeling techniques for forming processes, books, e.g., Ref 5 to 9. The FEM is discussed in
at a minimal cost. Generally effective goals in- including: detail in a separate section.
clude shortening the lead time in the design Slab Method. As implied in its name, the slab
● Slab method
cycle, reducing tooling cost and machine down- method is based on the assumptions such that the
● Slip-line method
time at the production stage, and developing a deformation of a workpiece can be approximated
● Upper bound method
stable process with a minimal reject rate. The with the deformation of a series of slabs, and the
● Finite element method
ability to predict the performance of a particular shape of slab (either flat or cylindrical shape) does
manufacturing process and to compare it with not change in the course of deformation. Each slab
alternative manufacturing processes at an early having an infinitesimal thickness is sliced so that it
stage in the process design cycle furthers these Modeling Techniques contacts with the tool. These methods assume that
goals by reducing costly trial-and-error design the shear stress due to friction between the work-
iterations using production equipment. The most fundamental calculations used in piece and tool does not affect the stress distribu-
Numerous analytical techniques have been metal forming analysis involve a forming load tion within the slab, but it is considered in the
developed to improve the process designer’s estimate, which is useful in selecting the size of force equilibrium. The stress distribution within
ability to evaluate a process and to predict vari- equipment required to form the product. The each slab is simplified further such that the stress
ous aspects of the metal forming process. Early simplest formula takes the form: component normal to the thickness direction can
methods relied on simple analytical techniques, be varied as a function of the thickness coordinate.
such as the slab method, the slip-line method, –A
P⫽K␴ (Eq 1) The other nontrivial stress components are then re-
the upper bound method, and heuristics to pre- lated to the stress component normal to the thick-
dict forming load, critical ratios, workability where P is the forming load, ␴ – is the mean flow ness direction according to the yield criterion and
limits, and die design features. Complex analyt- stress of a workpiece material under an idealized the flow rule for a rigid-perfectly-plastic material.
ical equations were converted to charts, which state of deformation, A is the planar area of the Then, the differential equilibrium equation ex-
could be applied by the designer. workpiece, and K is an empirically determined pressed in terms of stresses and the thickness co-
As computer technology became prevalent in correction factor for a particular forming process. ordinate is solved with the associated boundary
the engineering and manufacturing industry, The correction factor reflects on the effects of conditions. The slab method has been applied
analytical techniques, such as the upper bound nonuniformity of deformation and friction be- widely to various plane strain or axisymmetric
method, were used to develop specialized com- tween the workpiece and tool. forming problems such as upsetting, extrusion,
puter programs, which could be used to analyze a This estimation has been improved upon by drawing, and rolling.
particular process, such as tube sinking, strip more elaborated approximate solution techniques Figure 1 illustrates a simple example of the
rolling, or extrusion. The finite element method such as the slab method, the slip-line method, the slab method applied to a ring compression prob-
(FEM) for metal forming applications was first upper bound and lower bound methods, Hill’s gen- lem. In fact, the ring compression test has been
introduced in early 1970 (Ref 1). The continuous eral method, and, finally, the finite element meth- used widely as a convenient tool to evaluate the
improvements in computer technology and FEM ods. These methods have been employed to allow friction factor between the tool and workpiece.
finally made an important impact in the metal for the estimation of not only the forming load, but In bulk metal forming industries, the constant
forming industry in the mid-1980s. Due to its also the material flow pattern and the stress distri- shear friction model (based on shear strength of
unique capability in describing complex shapes, bution. The first two methods, i.e., the slab method the workpiece) has been preferred to the well-
boundary conditions, and realistic material ther- and the slip-line method, solve the field equations known Coulomb friction model (based on con-
momechanical response, the development of a in a differential form directly with certain degrees tact pressure). In Fig. 1(a), the inner radius Ri,
general-purpose metal forming analysis software of simplification. On the other hand, the other the outer radius Ro, and the height H character-
has been realized. The method has been used as methods solve the field equations in an integral ize the geometry of the ring. Now consider an
an essential tool for product and process design form derived from the associated energy theorems. infinitesimal element depicted in Fig. 1(b). The
engineers to reduce development time and cost. Except for the FEMs, application is generally lim- equilibrium equation in the r direction can be
Due to the demand from the industry to produce ited to metal forming processes with simple geom- written as follows:
more accurate simulation models, the FEM has etry and idealized materials laws.
continuously evolved from two-dimensional The following sections briefly review some s r Hrdq - (s r + ds r ) H (r + dr ) dq
analysis into the true three-dimensional models classical solution techniques for the slab method, Ê dq ˆ
+ 2s q sinÁ ˜ H dr m 2 mkr dr dq = 0
since the late 1980s and early 1990s (Ref 2–4). the slip-line method, and the upper bound Ë 2¯ (Eq 2)
Chapter 15: Modeling Techniques in Forming Processes / 221

state of stresses at some points. The slip-line


method has also been applied successfully to var-
ious plane-strain forming problems such as in-
dentation, extrusion, drawing, and rolling.
Upper-Bound Method. Unlike the two pre-
viously discussed methods, the upper-bound
method (UBM) is based on the energy principle,
known as the upper-bound theorem. The upper-
bound theorem states that the rate of total en-
ergy associated with any kinematically admissi-
ble velocity field defines an upper bound to the
actual rate of total energy required for the de-
formation. Hence, for a given class of kinemat-
ically admissible velocity fields, the velocity
field that minimizes the rate of total energy is
the lowest upper bound, and therefore is nearest
the actual solution. Here, the kinematically ad-
missible velocity field is used to denote a veloc-
ity field that satisfies the incompressibility re-
quirement for a rigid-plastic material and the
prescribed velocity boundary conditions. How-
ever, the velocity field may be discontinuous on
a finite number of imaginary internal surfaces.
Fig. 1 A ring compression problem. (a) Geometry of ring. (b) Slab element The rate of total energy generally consists of
three terms such that:
where m and k are the shear friction factor and the axial component of stress can be obtained as
the equivalent shear stress, respectively, and the follows:
sign ⫿ is used to express that the direction of
E˙ T = Ú s e˙ dV + Ú k Dv
V SD
t dS + Ú m k Dv
SF
s dS
(Eq 11)
frictional stress changes from outward to inward ␴z ⫽ Y ⫹ ␴r (Eq 8)
at the neutral point as r increases. The neutral
where Y is the yield stress of a material. It is where ␴ – and –e˙ are the equivalent stress and the
point is denoted by the radius rn. Assuming that
␴r ⫽ ␴␪ and neglecting higher order terms, Eq 2 noted that Y ⫽ 3 k for a material complying equivalent strain rate, respectively; ⌬vt is the
can be simplified as: with von Mises yield criterion. The forming load magnitude of velocity discontinuity tangent to
can then be evaluated as: the velocity discontinuity surfaces SD; and ⌬vs is
d␴r ⫽ ⫿ 2␣k dr (Eq 3) the magnitude of sliding velocity on the contact
rn Ro surface SF.
where a =
m
is a constant representing the
ÚRi Ú
P = s z 2 p r dr + s z 2 p r dr
rn Each term in the right-hand side of Eq 11 rep-
H = KSlab s A (Eq 9) resents the rate of plastic deformation energy, the
coining effect, that is, a portion of the total form- rate of energy dissipation associated with internal
ing load required to overcome friction increases velocity discontinuity, and the rate of energy dis-
as the height of ring decreases. It is noted that where KSlab is the correction factor or normal- sipation due to friction between the tool and
the assumption ␴r ⫽ ␴␪ holds when the radial ized forming load defined by: workpiece, respectively. The second term, also
component of velocity u is a function of homo- known as the jump condition, can be omitted
geneous degree one in r so that 1 m when a class of continuous velocity fields is con-
KSlab = 1 + ( Ro - Ri )
2 3 H (Eq 10) sidered. Among various classical solution meth-
du u ods, the upper-bound method has been applied
=
dr r The slip-line method is another simple and most extensively to various two-dimensional or
or powerful classical solution method, although its three-dimensional forming problems because it
application is limited to plane-strain problems delivers a fast and accurate solution as long as
e˙ r = e˙ q (Eq 4) for a rigid-perfectly plastic material (Ref 5, 6). In the trial velocity field can be provided closer to
this method, the equilibrium equations for a the actual velocity field. However, it is not easy
By applying the boundary condition ␴r ⫽ 0 at plane-strain state are first transformed into the to choose a good trial velocity field using a
r ⫽ Ri and r ⫽ Ro, Eq 4 can be integrated in the hyperbolic differential equations expressed in combination of analytic functions for geometri-
form: terms of the mean stress, the maximum/mini- cally complicated problems. In order to relax in
mum shear stress, and the direction of maximum/ such a difficulty, the upper-bound elemental
Ï2a k (r - Ri ) when r £ rn minimum shear stress. The characteristics of the technique (UBET), based on the concept of a
sr = Ì
Ó 2a k ( Ro - r ) when r ≥ rn (Eq 5) hyperbolic differential equations are known as “unit rectangular deforming region” (Ref 10)
the slip-lines. The slip-line field then can be con- has been developed and applied to rather com-
Since the stress must be continuous at rn, the lo- structed by networking two kinds of slip-lines plex forming problems and preform design
cation of neutral point is determined as: representing the maximum and minimum con- applications.
stant shear lines that are orthogonal to each other. In order to compare the characteristics of so-
1 Several useful techniques have been proposed to lutions with different solution methods, i.e., the
rn = ( Ro + Ri )
2 (Eq 6) construct the slip-line field graphically depend- upper-bound and slab methods, the ring com-
ing on the configuration of a problem and the as- pression problem illustrated in Fig. 1 is em-
By applying von Mises yield criterion: sociated boundary conditions. The forming load ployed as an example here. Also, a simple form
can then be obtained by determining integral of the trial velocity field is chosen so that an ex-
– ⫽ Ω␴ ⫺ ␴ Ω ⫽ Y
␴ (Eq 7) constants for particular slip lines from the known plicit form of solution can be obtained. The trial
z r
222 / Process Design and Workability

velocity field used in the present example has and the radial velocity u defined in Eq 12 into Eq some understanding of the three-dimensional
the form (Ref 11): 16, the rate of total energy Ė T can be integrated forming process. Although a considerable
explicitly as: amount of research has been done in develop-
DÊ r2 ˆ ing the FEM for metal forming simulation
u(r ) = Á r- n ˜
2Ë r ¯
E˙ T 1
since the pioneering work (Ref 1) was pre-
= Ê rn4 + 3 Ro4 - rn4 + 3 Ri4 ˆ¯ sented in 1973, rigorous three-dimensional
v=0 2 p k VD 2 Ë
simulation of metal forming problems still re-
w( z ) = - D z (Eq 12)
Ï Ê 2
rn + rn4 + 3 Ro4 ˆ
2¸ mains a challenging task from the standpoint of
rn2 Ô ÊR ˆ Ô
- Ìlog Á 2 ˜ - log Á o ˜ ˝ computational efficiency, solution accuracy,
VD 2 Ô ÁË r + r 4 + 3 R 4 ˜¯ Ë Ri ¯ Ô graphics visualization, mesh generation and au-
where D = , VD is the die velocity, H is the Ó n n i ˛
H tomatic remeshing, and so on. As computer
m Ï4 3 1 3 3 ¸
height of ring, and rn is the radius at a neutral 2
(
+ Ì rn - ( Ri + Ro )rn + Ri + Ro ˝
H Ó3 3
)
˛
technology and FEM advance, wider and more
point. It is noted that this velocity field contains complicated metal forming processes are being
only one unknown, i.e., rn, and it is adequate for (Eq 18) investigated. It is believed that the further de-
investigating the effect of the location of neutral velopment of FEM will be continuously chal-
point on the forming load. The neutral radius rn The unknown coefficient, i.e., the neutral radius lenged by the need from the industry to make
is defined by: rn, is then determined so that the rate of total en- the modeling more accurate, more practical,
ergy attains to its minimum value such that: and more affordable.
u⫽0 at r ⫽ rn Since the two-dimensional FEM implementa-
dE˙ T tion has been discussed elsewhere in detail (Ref
u⬍0 for r ⬍ rn =0 12), this section focuses on the three-dimensional
drn (Eq 19)
u⬎0 for r ⬎ rn (Eq 13) implementation.
After some manipulation, the neutral radius can
Because the radial component of velocity, u, is a be written as:
Preliminary Assumptions
function of r only, the present trial velocity field
cannot reproduce the so-called barreling or In order to narrow down the discussion to the
Ê b2 + 3 + b4 ˆ most practical applications among a variety of
bulging phenomenon due to friction. 1 1 H o o
rn = ( Ri + Ro ) + log Á ˜
All nonzero strain-rate components are then 2 4 m Á b2 + 3 + b4 ˜ (Eq 20)
metal forming simulations, the following pre-
Ë i i ¯ liminary assumptions are first introduced.
written as:
Quasi-Static Analysis. In most metal form-
D Ê r2 ˆ ing processes, dynamic effects can be neglected
e˙ r = Á1 + n2 ˜ rn r
where b i = and b o = n . Equation 20 can then except for high strain-rate processes in which a
2Ë r ¯ Ri Ro realistic deformation mode cannot be obtained
D Ê r2 ˆ be solved for rn numerically with specific values without considering the effect of stress wave
e˙ q = Á1 - n2 ˜ of the ring geometry, i.e., Ri, Ro, and H, and the
2Ë r ¯ propagation and in which the magnitude of ki-
shear friction factor m. netic energy is comparable to that of deforma-
e˙ z = - D (Eq 14) The forming load can be evaluated directly tion energy. In cases where dynamic effects can
by substituting the value of neutral radius into be neglected, the progress of deformation is an-
It is noted that the velocity field satisfies the in- Eq 18 and with use of the relation: alyzed in a manner such that every instantaneous
compressibility condition such that: state of a body in the course of deformation is
E˙ T satisfied with the equilibrium conditions. Such a
P=
ėr ⫹ ė␪ ⫹ ėz ⫽ 0 (Eq 15) VD (Eq 21) method of analysis is generally called the quasi-
static analysis.
Since the present velocity field satisfies the It is noteworthy that even for forming
The correction factor or normalized forming
continuity requirement, the incompressibility processes between medium and low strain-rate
load KUBM can be defined by:
condition, and the prescribed velocity boundary ranges, the explicit method of solution originally
condition, it is proved to belong to a class of designed for dynamic analyses is sometimes
kinematically admissible velocity fields. It is P used for the sake of computational efficiency.
K UBM =
also noted that the second term in Eq 11 can be sA (Eq 22) For the explicit method, however, the size of
omitted because there is no internal velocity time step must be small enough, typically the
discontinuity. order of microseconds, to satisfy the stability
The rate of total energy in Eq 11 can be Finite Element Method criterion of the explicit time marching scheme. It
rewritten as: means that the computational efficiency may not
Metal forming simulation is often classified be achieved because the number of solution
as a class of highly nonlinear continuum me- steps increases so greatly compared with that of
E˙ T = Ú s e˙ dV + Ú mk Dv
V SF
s dS chanics problems because it is accompanied by the implicit method. Thus, the so-called mass
large deformation (geometric nonlinearity), scaling technique (Ref 13), which may be inter-
Ro Ro
s e˙ (2 prH ) dr + 2 nonlinear materials behavior (material nonlin- preted as the introduction of artificial inertia
= Ú Ri Ú
Ri
mk u (2 pr ) dr
earity in both deformation and temperature), term, is usually employed in order to increase
(Eq 16) and frictional contact (nonlinear boundary con- the size of time step. In these cases, the mass
dition). Starting from the mid-1980s, the FEM scaling factor must be selected carefully to get
After substituting the following expression of has shown great success in the axisymmetric the solution with reasonable accuracy and simu-
the equivalent strain-rate: applications such as disk forging, cold forging lation time. In sheet forming applications, the
of cylindrical fasteners, and so forth. The ap- explicit method has been very successful.
proximation of a two-dimensional cross section Rigid-plastic analysis is more advantageous
4
1 Ê rn ˆ in a three-dimensional part, using the plane- for computational efficiency and robustness
ė = D 1 +
3Ër¯ (Eq 17) strain assumption, is an alternative to achieving than elasto-plastic analysis is. This method has
Chapter 15: Modeling Techniques in Forming Processes / 223

been used predominantly for the majority of particle. The motion of a mesh can be chosen that the stress measure in Eq 23 must be the
bulk forming processes where the elastic defor- arbitrarily, but it must preserve the boundary of first kind of Piola-Kirchhoff stress tensor ac-
mation is negligible compared with the plastic a continuum in the course of deformation. When cording to the framework of the referential de-
deformation and the distribution of residual the motion of mesh is set the same as the mo- scription of continuum mechanics. However, it
stresses is not of major concern. By neglecting tion of material, it becomes the same as the can be assumed that the reference configuration
the elastic portion of deformation, the rigid- Lagrangian methods. Several different schemes is updated as frequently as required, so it is
plastic formulation (Ref 14) turns out to be very regarding the selection of mesh motion have therefore not distinguished from the Cauchy
similar to that of fluid flow problems except for been proposed according to the specific area of stress tensor.
the presence of yielding and so it is sometimes applications. The ALE method has been consid- Equation 24 represents the incompressibility
called the flow formulation. The velocity field ered a prominent method, especially for the condition where ui is the velocity vector.
satisfying the equilibrium equations, constitu- steady state metal forming applications. How- Equation 25 represents the constitutive equation
tive equations, and boundary conditions instan- ever, it has disadvantages, such as the increase of based on the so-called J2 flow rule. Here, ėij is
taneously is obtained at each state in the course problem size due to additional variables for the strain-rate tensor or the symmetric part of the
of deformation. Therefore, it is necessary to defining the mesh motion and the requirement of velocity gradient tensor as defined in Equation
adopt an appropriate scheme for updating de- additional computation due to the evaluation of 26. ␴– and –e˙ are the effective stress and the ef-
formed configurations from the velocity field convection terms in updating the state variables. fective strain-rate, respectively, defined by:
obtained at each state.
1 1
Unlike the elasto-plastic formulation, the rigid- Description of the Problem s= [ 3
2 s ¢ijs ¢ij ] 2
and e˙ = [ e˙ ]
2
3 ij
2
(Eq 33)
plastic formulation does not have any ambiguity
related to the choice of the objective stress rates The governing partial differential equations
and the decomposition of deformation gradient and the associated boundary conditions for the Equation 27 represents an implicit form of the
into the elastic and plastic parts (Ref 15, 16). rigid-plastic and rigid-viscoplastic problems can yield criterion as a function of the effective
Updated Lagrangian (UL) Formulation. be written as (Ref 12): strain –e , the effective strain rate –e˙ , and the tem-
There are several different formulations (Ref 17, perature T. Provided the increment of time be-
18) for continuum mechanics problems with large ␴ij,j ⫽ ␴¢ij,j ⫹ ␴m,i ⫽ 0 in ⍀ (Eq 23)
tween any two adjacent referential configura-
deformation and/or large rotation, e.g., total tions is small enough, the effective strain can be
Lagrangian (TL), updated Lagrangian (UL), evaluated approximately such that:
ui,i ⫽ 0 in ⍀ (Eq 24)
Eulerian, and arbitrary Lagrangian Eulerian
–e – –˙
(ALE) methods. In the first two methods, the new ⫽ e old ⫹ e ⌬t (Eq 34)
motion of a continuum is described in terms of 2s
s ¢ij = ˙ e˙ ij in W (Eq 25)
the coordinates of a material particle in an arbi- 3 e where ⌬t denotes the increment of time. Equa-
trarily chosen reference configuration. The TL tion 27 represents the most popular form of the
method uses the initial undeformed configura- e˙ ij = 12 (ui,j + u j,i ) in W (Eq 26) rigid-plastic constitutive equations, but more
tion as a fixed reference configuration, but the complicated forms of constitutive equations can,
UL method uses the most updated configuration ␴ – (e–, –e˙ , T )
– ⫽␴ in ⍀ (Eq 27) of course, be used within this structure of for-
during the progress of deformation as the refer- mulation such as the constitutive equations hav-
ence configuration. Both methods are amenable ui ⫽ ûi on ⌫u (Eq 28)
ing internal state variables with the associated
to solid mechanics problems in which the con- evolution equations.
figuration of a boundary is not fixed in space and Equations 28 and 29 are the prescribed veloc-
changes in the course of deformation. Compared ␴ijnj ⫽ t̂ i on ⌫t (Eq 29) ity and traction boundary conditions, respectively.
with the TL method, the UL method can use the The hat symbol on ui and ti is used to denote pre-
simplified kinematics of a continuum assuming Friction & contact conditions on ⌫c (Eq 30) scribed values, and the vector ni denotes the out-
that the reference configuration is updated con- ward normal vector to the body at a point on the
tinuously within a small amount of increment of Here, the open domain ⍀ and its associated boundary. The friction and contact conditions are
deformation according to the desired degree of boundary ⌫ represent the current configuration discussed in detail in the subsequent section.
solution accuracy. Thus, the UL method has of a body according to the UL formalism. The
been used widely in most metal forming simula- subscripts u, t, and c in ⌫ are used to denote
tion. It is worthwhile to mention that this method three different types of the boundary associated Friction and Contact Conditions
generally requires frequent remeshing (or rezon- with the boundary conditions: the prescribed ve-
locity, the prescribed traction, and the frictional The friction and contact conditions in this sec-
ing) when mesh is severely distorted in the
contact conditions, respectively. Indices i, j, and tion are described between any two boundaries
process of large deformation. Furthermore, it may
k are used to denote the components of a tensor whether they belong to the same body (i.e., self-
not be an effective method for a certain class of
and a comma denotes the spatial derivative with contact), rigid and deformable bodies, or two or
forming problems, such as extrusion, rolling,
respect to the current configuration. more deformable bodies. Although several differ-
and machining, in which the steady state solu-
In Eq 23, i.e., the equilibrium equation, ␴ij , is ent methods (Ref 19–21) have been presented in
tion is of major concern.
the stress tensor, and ␴⬘ij and ␴m denote its devi- describing the macroscopic contact and friction
In the Eulerian method, the motion of a con-
atoric and volumetric components, respectively, phenomena within the framework of continuum
tinuum is described in terms of the spatial coor-
such that: mechanics, the method of pointwise description
dinates of a material particle in the current con-
of friction and contact conditions has been used
figuration. Therefore, this method is particularly
␴⬘ij ⫽ ␴ij ⫺ ␴m␦ij (Eq 31) in most practical applications. The contact condi-
suitable for well-constrained fluid flow prob-
tion and friction can be summarized as:
lems in which the domain of interest is fixed in
the space. However, there are inherent difficul- and ● Contact condition (non-penetration condi-
ties in applying this method to the moving tion):
boundary or free surface problems. The ALE s m = 13 s kk (Eq 32) a. Any material particle of a given body
method has been developed to remove these cannot penetrate into another
difficulties by introducing the concept of mesh where ␦ij is the Kronecker delta, and the re- b. The normal component of contact traction
motion independent of the motion of material peated index denotes the summation. It is noted must be compressive for each body
224 / Process Design and Workability

c. A pair of contact points can separate only classical theory of plasticity. The slip function is so-called weak formulation or the principle of
when the contact traction vanishes or be- defined by: virtual work, although certain mathematical fea-
comes tensile tures of variational problems with the quadratic
ƒ(pi) ⬅ Ω(pt)iΩ ⫹ ␮pn (Eq 40) functional such as the existence and uniqueness
● Friction condition:
of solutions and the stability and accuracy of fi-
a. The magnitude of the tangential compo-
and nite element solutions cannot be stated. Here,
nent of contact traction must be less
the weak formulation is used to accommodate a
than or equal to that of the normal com-
ƒ(pi) ⬅ Ω(pt)iΩ ⫺ mk (Eq 41) broad class of plastic constitutive models to the
ponent multiplied by a coefficient of
same framework of formulation.
friction.
corresponding to the Coulomb’s law of friction The constraint conditions such as the incom-
b. The instantaneous relative motion in the
tangential direction for a pair of contact and the shear friction law, respectively, where ␮ pressibility condition and the contact condition
is the coefficient of friction, k is the shear yield can generally be incorporated into the variational
points can take place when the equality
stress of the weaker material, and m is a constant formulation by using one of two techniques: the
in (a) above holds.
friction factor. penalty method or the Lagrange multiplier
c. The tangential relative motion must be
It is noted that the frictional dissipation func- method. The penalty method has the advantage
along the same line as the tangential
tional associated with the friction law in Eq 38 of simple implementation, but it has a drawback
component of contact traction but in the
or 39 has a non-differentiable form with respect such that it can result in an overconstrained prob-
opposite direction.
to its primary variables, i.e., the velocities of lem or an underconstrained problem depending
In most metal forming applications, two dif- each body. In the area of metal forming simula- on the choice of the penalty parameter. The over-
ferent types of friction laws have been widely tion, the following form of regularized friction constraint means the volumetric locking or the
used: the Coulomb’s law of friction and the laws has been widely used. For any (ż t)i: locking of contact surfaces, and the undercon-
shear friction law. Statement (a) under “Friction straint means the inaccuracy of solution in the
condition” represents the Coulomb’s law of fric- ÏÔ 2 sense of incompressibility or nonpenetration. On
Ê ( z˙t ) i ˆ ¸Ô ( z˙t ) i
tion, but it also represents the shear friction law (p t ) i = m p n Ì tan -1 Á ˜ ˝ ( z˙ ) the other hand, the Lagrange multiplier method
by replacing the normal component of contact ÔÓ p Ë z˙o ¯ Ô˛ t i (Eq 42) can avoid the drawback of the penalty method,
traction by the shear yield stress of the weaker but it has a disadvantage concerning the increase
material. Also, the coefficient of friction is gen- and of problem size because the Lagrange multipliers
erally referred to as the constant factor in the are treated as additional solution variables such
shear friction law. as the velocity of material particles. The
This statement of contact and friction condi- ÔÏ 2 Ê ( z˙t ) i ˆ Ô¸ ( z˙t ) i Lagrange multiplier can be interpreted as the hy-
(p t ) i = - m k Ì tan -1 Á ˜˝
tions can be expressed in a mathematical form as: p Ë z˙o ¯ Ô˛ ( z˙t ) i (Eq 43) drostatic stress for the incompressibility con-
ÓÔ
straint and the normal contact traction for the
(p ) nonpenetration condition.
z˙i = l˙ n I (p n )ni + l t I ( f ) t i where żo is a positive constant. These regular-
(p t ) i (Eq 35) In areas of metal forming simulation, it is pop-
ized friction models approach the original fric- ular to use the Lagrange multiplier method for
tion laws asymptotically as żo approaches zero. the incompressibility condition and the penalty
where ż is the relative velocity between two However, a very small value of żo can make a method for the contact condition. However, the
points in contact at an instant as: difficulty in convergence, and a larger value may penalty method has also been used successfully
give a solution deviated from the original fric- with a certain class of finite elements with the
ż i ⫽ uai ⫺ ubi (Eq 36) tion laws. selective reduced integration scheme.
where the superscripts a and b denote the corre- Mixed Variational Formulation Finite Element Formulation
sponding contact surfaces chosen arbitrarily, ni
is the outward normal vector with respect to the Boundary value problems (BVP) in contin- The finite element method can be distin-
contact surface “a,” p is the contact traction vec- uum mechanics can be expressed in two different guished from other approximate methods by the
tor on the contact surface “a” as: ways: the partial differential equations (PDE) way it constructs the trial solution (e.g., kine-
with the associated boundary conditions and the matically admissible velocity field) with a finite
pi ⫽ pai ⫽ ⫺pbi (Eq 37) variational equations with the appropriate func- number of piecewise continuous trial functions
tion space. The solutions of these two different (polynomial functions in most applications)
where the subscripts n and t denote the normal forms of BVP are referred to as the strong solu- (Ref 22–24).
and tangential components, respectively, as: tion and the weak solution, respectively. The For a particular class of metal forming
term weak is used in the sense that the require- processes, it is always important to select an ap-
ż i ⫽ ż nni ⫹ (ż t)i and pi ⫽ pnni ⫹ (pt)i (Eq 38) ment of continuity (differentiability) of solution propriate type of element (i.e., the order of poly-
is weakened in the variational form of BVP. If a nomials, the geometric shape, the rule of numer-
Continuing with the description of terms in Eq strong solution requires the existence of a second ical integration, and so on). Some issues related
35, the symbol Ω⭈Ω denotes the absolute magni- derivative, the corresponding weak solution re- to the selection of element are discussed subse-
tude of a vector, and l̇n and l̇t are non-positive quires only the existence of a first derivative in quently. From the standpoint of polynomial
constants. I(g) is an indicator defined for any the sense of distribution; that is, the first deriva- order, linear elements are generally preferred to
scalar-valued function g as: tive needs to be continuous within each of a fi- quadratic or higher-order elements for most metal
nite number of subdomains but not necessarily forming applications in which the friction and
if g = 0 across the interboundary between subdomains. contact conditions are always present. With use
È1,
I ( g) = Í The variational form of BVP can be obtained of the friction and contact conditions described
Î0, if g π 0 (Eq 39)
only when the quadratic form of functional ex- previously, the so-called node-to-segment contact
ists so that the set of Euler equations, obtained situation cannot be avoided because these con-
In order to describe the friction laws effectively, by the vanishing of the first variation of the func- straints need to be imposed point-wise. The node-
a slip function f is introduced for each pair of tional, is identical to the original PDE. However, to-segment contact can be treated in a simpler
contact points similar to the yield function in the the same form can be obtained by using the manner with linear elements rather than higher-
Chapter 15: Modeling Techniques in Forming Processes / 225

order elements. Also, linear elements are gener- Some disadvantages of CST elements are MINI element, the shape function for the veloc-
ally easier to use without a-priori knowledge listed as follows. The CST elements are apt to ity is enriched with bubble terms associated with
of solution than higher-order elements with the have directionality in the mesh topology and an additional bubble node, although the shape
same degrees of freedom for a given problem. thus in the solution because the mesh topology function for the mean stress remains as a linear
Next, from the standpoint of geometric shape, directly reflects on the form of trial solutions. function interpolated with the values at four ver-
there are two different kinds of linear elements: Here, the directionality implies that triangular tex nodes. The bubble node is located at the cen-
triangular or quadrilateral for two-dimensional and tetrahedral elements are globally biased in troid of tetrahedron and has only the velocity de-
elements and tetrahedral or hexahedral for three- a certain direction. Also, the CST elements grees of freedom. Here, the word bubble means
dimensional elements. Shape functions for trian- may show more tendency of volumetric locking that its value always vanishes along the bound-
gular and tetrahedral elements contain polyno- due to the incompressibility constraint condition ary of the element. This element does not have
mial terms such as (1, ␰, ␩) and (1, ␰, ␩, ␨), than quadrilateral and hexahedral elements. The the volumetric locking problem. Moreover, the
respectively. On the other hand, those for quadri- term volumetric locking means that each ele- total number of equations can be maintained
lateral and hexahedral elements contain polyno- ment endowed with a linear velocity field cannot as the same as that with the tetrahedral elements
mial terms such as (1, ␰, ␩, ␰␩) and (1, ␰, ␩, ␨, ␰␩, deform properly when the same degree of satis- because the velocity components at the bubble
␩␨, ␨␰, ␰␩␨), respectively. Triangular and tetrahe- faction is required elementwise for both the de- nodes can be eliminated at the element level by
dral elements are known as constant stress/strain formation and the constraint condition. In order means of the static condensation.
elements (CST) because all derivatives with re- to avoid such a locking phenomenon, the degree Although the bubble node is introduced mainly
spect to any component of coordinates vanish. In of satisfaction for the constraint condition needs for a systematic stabilization of the ill-posed fi-
the rigid-plastic formulation, vanishing deriva- to be relaxed in a certain manner such as with nite element system, a considerable amount of
tives imply constant strain rates and, therefore, the selective reduced integration scheme or with computational effort is additionally required for
constant stresses, within an element. On the other the mixed formulation as described previously. the static condensation and recovery of the ve-
hand, the quadrilateral and hexahedral elements Another important aspect to be considered in de- locity of the bubble node. However, such addi-
have polynomial terms in the product form of termining the element shape, especially for large tional efforts can be minimized with a few mod-
coordinates, and thus, the velocity gradient in deformation problems, may be whether an effi- ifications of the standard MINI element. In fact,
one component of coordinates is linear with re- cient and robust method for automatic remesh- it turns out that the derivation of this modified
spect to the other components of coordinates. ing is available with the shape of element. In MINI element can be conceived as a systematic
Strain rates and stresses are linear within an ele- areas of metal forming simulation, quadrilateral stabilization of the standard mixed tetrahedral
ment accordingly. Both kinds of linear elements elements are preferred to triangular elements for element because the effect of the bubble node
have been used widely by weighing the pros and the two-dimensional analysis, but tetrahedral appears on only the block diagonal terms associ-
cons for a particular application. elements have been used more than hexahedral ated with the mean stress.
Generally speaking, triangular and tetrahedral elements for the three-dimensional analysis be-
elements have more flexibility in filling meshes cause of the versatility in automatic remeshing. Heat Transfer
into any complicated shape than quadrilateral As mentioned previously, linear tetrahedral
and hexahedral elements. It is noted that triangu- elements cannot be used for metal forming sim-
lar and tetrahedral elements in the group of CST ulation without an appropriate resolution of the In the metal forming industry, heating is fre-
elements must be distinguished from the so- volumetric locking problem. The most common quently used to increase material workability and
called degenerated elements whose strain rates approach to enforce incompressibility includes control forming loads. During the forming
and stresses are not constant within an element. the penalty method and the Lagrange multiplier process, heat is generated from plastic and fric-
Degenerated elements can be obtained from as discussed previously. These methods, how- tion work. Heat is also lost through contact with
coalescing adjacent nodes into the same node, for ever, are limited to the quadrilateral element in colder dies and through convection and radiation
instance, mapping from the four-node quadrilat- two-dimensional and hexahedral element in three- with the environment. Since materials properties
eral parent element into a three-node triangular dimensional. Furthermore, the resulting matrix vary considerably with temperature, accurate
shape or mapping from the eight-node hexahe- is ill conditioned for a conjugate-gradient solver temperature prediction is required. For conven-
dral parent element into a four-node tetrahedral due to large penalty values for incompressibility. ience, the deformation analysis and thermal
shape. The performance of degenerated elements The mixed formulation uses the Lagrange multi- analysis can be loosely coupled in such a way that
is not as good as that of the CST elements or the plier method to achieve incompressibility condi- plastic work and friction are considered as heat
original quadrilateral and hexahedral elements. tion. Generally speaking, the order of polynomi- source in the thermal analysis while the updated
Also, a numerical integration scheme for the als for the velocity shape function needs to be temperature field is used to determine the flow
CST elements in evaluating the stiffness matrix is higher than that for the mean stress shape func- stress behavior during the deformation analysis.
unnecessary because they have constant values tion because the mean stress is originally a dual- The governing equation for heat transfer can
of strain rate and stresses within an element. In ity variable paired with the volumetric strain rate be expressed as:
other words, the integration is always exact since that is the trace of the velocity gradient. For in-
the integrand is constant with respect to coordi- stance, when the linear shape function is used k1 Tii ⫹ q̇ ⫺ ␳CṪ ⫽ 0 (Eq 44)
nates. On the other hand, for quadrilateral and for both variables, the volumetric strain rate be-
hexahedral elements, a numerical integration rule comes constant within an element, but the mean where ␳ is the density, C is the specific heat ca-
based mostly on the Gaussian quadrature for- stress is linear. It is obvious that a linear function pacity, T is the temperature, t is time, k is the
mula is required in evaluating the stiffness ma- cannot be matched with a constant value in a thermal conductivity, and q̇ is a heat generation
trix. The numerical integration is exact when a general case and a mesh system may tend to lock term. Heat generation in metal forming is due to
real element keeps the same shape as the parent as a result. When the quadratic shape function is work of plastic deformation and friction. Heat
element, i.e., a rectangular shape or a brick shape, used for the velocity in order to make the well- generation due to plastic deformation is given by
although it is not possible except for problems posed problem, a considerable increase of prob-
with a very simple geometry. In such cases, the lem size cannot be avoided.
determinant of the Jacobian matrix mapping be- The MINI element is effectively designed for
q˙ pw = k Ú s e˙ dV (Eq 45)

tween the domains of the parent element and a the three-dimensional simulation of incom-
real element is constant. Errors originated from pressible problems such as fluid flow, rubber where k is a deformation efficiency term, repre-
the numerical integration increase as the shape of elasticity, and rigid-plasticity problems under senting the fraction of the work of deformation
element is apart from that of the parent element. the framework of the mixed formulation. In the converted to heat.
226 / Process Design and Workability

The boundary condition for the tool-work- compression, and three methods (slab, UBM, is based on a rectangular workpiece shape. The
piece contact surface includes friction heating and FEM) are used to study the process. The radial velocity is a function of the radius and the
and heat exchange via temperature difference of second example is the cold forging of an elec- neutral radius, the axial velocity is a function of
two objects. trode, the third example is the hot forging of a the height position, and the neutral radius is a
crankshaft, and the last example considers mate- function of friction factor and shear stress.
rial cutting. Only FEM is used in the last three This simplification/assumption introduces
q̇1 = Ú hfS1
s1 us1 dS1 + Ú HDTdS
S1
1 (Eq 46) examples due to the complexity of the processes. error when the ID and OD surfaces of workpiece
start to bulge during the compression. As the
where us is the sliding velocity, fs is the friction Ring Compression workpiece height increases, it is known that the
stress, ␩ is the percentage of friction energy ab- unstable flow (or buckling mode) will gradually
sorbed by the object, H is the lubricant heat The ring compression test is one of the meth- dominate the deformation. In such situations, the
transfer coefficient, and ⌬T is the temperature ods used to determine the friction factor. The assumed UBM velocity field is moving away
difference between two objects. process is generally assumed to be two-dimen- from the true velocity field. To demonstrate the
The boundary condition of the free surface in- sional or axisymmetric. Due to the simplicity of buckling mode, another ring compression exam-
cludes convection heat and radiation heat from/to the workpiece and die geometry, ring compres- ple with OD, ID, and height ratio of 6:3:6 was
the environment: sion is used as the first example to demonstrate carried out. In this case, the flow stress is assumed
and compare the capability of slab, UBM, and to be in the form of ␴ ⫽ 10e0.1. A friction factor
FEM. of 0.2 was used. The predicted FEM mesh and
q˙2 = Ú hc (T - T• ) dS2 + Ú se(T 4 - T•4 )dS2 (Eq 47) Two-ring test FEM simulations with outer di- various deformation stages are shown in Fig. 4.
S2 S2
ameter (OD), inner diameter (ID) and height ra- Figure 5 shows the effect of friction on the lo-
tios of 6:3:2 and 6:3:0.5 were carried out with a cation of a neutral point with three different as-
where hc is the convection heat transfer coeffi- pect ratios of a ring: i.e., 6:3:0.5, 6:3:1, and 6:3:2,
cient, T⬁ is the environment temperature, ␴ is constant friction factor of 0.4. A constant flow
stress of 70 MPa (10 ksi) was used in the simu- in terms of the OD, the ID, and the H, in se-
the Stefan-Boltzmann radiation constant, and e quence. From this figure, it is noted that the
is the emissivity of the surface. It is also noted lation. The predicted shapes using the FEM and
the associated neutral line within the workpiece upper-bound solutions approach to the mean ra-
that the inclusion of the view factor for radiation dius of ring asymptotically as the magnitude of
heat calculation (which is not addressed here) at different stages of deformation in both cases
are shown in Fig. 2 and 3, respectively. In the shear friction factor increases. It is because, ex-
can improve the accuracy of the thermal model, cept for point sticking, neither local nor global
especially for the hot forging condition. case with the ratio 6:3:2, the neutral line within
the workpiece is a function of height. In the case sticking phenomenon can be represented with
By substituting Eq 45 into Eq 44 and intro- the present velocity field. It is also shown that the
ducing a small, arbitrary temperature variation with 6:3:0.5, the neutral line remains to be a ver-
tical line. Furthermore, the ID and OD surface upper-bound solutions correlate better with the
␦T, and applying the divergence theorem, Eq 44 FEM solutions as the height of ring reduces.
can be written in the form: bulged due to the friction force on the two con-
tacting surfaces.
Ti dTj dV + rCT˙ dT dV To simplify the mathematical derivation, the
Úk
V
1 Ú
V
assumed velocity field (Eq 12, 13) in the UBM
+ ks ij e˙ ij dT dV - q n dT dS = 0
Ú Ú Sq
V (Eq 48)

where qn is the heat flux across the boundary. It


includes the convection heat and radiation heat
to the environment for the free surface and fric-
tion heat and heat gain or loss to the contacting (a) (b) (c)
surface. The temperature distribution function
Fig. 2 The effect of friction on the location of the neu-
can be expressed through nodal temperatures tral point for a thick ring compression test. (a) (a) (b) (c) (d)
and shape functions (Ref 12). 2%. (b) 10%. (c) 20%. The cross section of the ring is
After discretization, Eq 48 can be further ex- shown with the curved line denoting the neutral point Fig. 4 A two-dimensional buckling simulation. (a) 1%.
over the height. (b) 20%. (c) 35%. (d) 50%
pressed in matrix form (Ref 24) as:

CṪ ⫹ KcT ⫽ Q (Eq 49)

where C is the heat capacity matrix, Kc is the (a)


heat conduction matrix, T is the node point tem-
perature vector, and Ṫ is a vector containing the
time rate of change of temperature of node
points. The heat flux vector, Q, for metal form-
ing simulations considers plastic work of defor-
mation, heat generation due to sliding contact (b)
friction, and heat flux due to lubricant conduc-
tion, convection, and radiation.

(c)
Example Simulations
Fig. 3 The effect of friction on the location of the neu-
tral point in a thin ring compression test. (a) 4%.
To demonstrate the capability of the modeling
techniques, four examples are included in the
(b) 20%. (c) 40%. The cross section of the ring is shown Fig. 5 The effect of friction on the location of the neu-
with the curved line denoting the neutral point over the tral point. UBM, upper-bound method; FEM,
following section. The first example is the ring height. finite-element method
Chapter 15: Modeling Techniques in Forming Processes / 227

Fig. 7 Progression of electrode forging

mensions, and the final forming was modeled in ternal fold is successfully predicted in the
three dimensions. The shearing and squaring op- zoomed area of the figure. Due to the axisym-
erations were not modeled in this study. metric assumption, the internal fold is not seen
Fig. 6 The variation of forming load with the magnitude Two laps were developed during the back- in the simulation results. Further study, consid-
of friction factor. UBM, upper-bound method ward extrusion as shown in Fig. 8(a) and (b). ering tooling eccentricity, will be carried out in
One lap is seen on the outside wall (Fig. 8b) and the near future. Due to the symmetry condition,
The normalized forming loads obtained using extends the entire circumference. The other lap 1/12 of the workpiece was modeled in the simu-
the UBM are compared with those obtained occurred only partially on the inside wall (Fig. lation to reduce the simulation time.
using the slab method presented in Eq 10 as well 8a). It is believed that eccentricity of the forming The predicted geometry at early stage of the
as the corresponding FEM solutions. It is note- process caused the internal lap. From visual in- tooth forming is shown in Fig. 10. It is clearly
worthy that the forming load estimated using spection, the lap occurred in every tooth in the seen that the material is pushed both upward and
both the UBM and the FEM based on the veloc- final part as shown in Fig. 8(c). Figure 8(c) also downward through the contact of the punch.
ity formulation can be used as a measure of the shows a change in the texture at the inner wall of Folds occurred on the wall above the teeth and
upper limit of the actual load. As shown in Fig. 6, the electrode, above the teeth. Figure 8(d) shows below the teeth. It is also noted that the partially
the forming load estimated using the FEM is a lap that occurred at the bottom of the cavity, formed tooth is seen as a concave shape. As the
smaller than that using the UBM, and thus it can which is seen as a round pattern pointed to by punch moved farther down, the fold began to
be said that the present UBM solution becomes the arrow. Due to the size of the part, it is very wrap around the top of the teeth (Fig. 11). Figure
less accurate than the FEM solution as the mag- difficult to visualize the fold, other than the one 12 shows that this material peeled down into the
nitude of friction factor increases. On the other on the tooth tip. bottom of the cavity as shown in Fig. 8(d). Also,
hand, the load estimated using the slab method The strain distribution and predicted geome- the fold above the teeth is shown to smear,
can be used as a measure of lower limit when the try at the end of each operation through extru- which correlates well to Fig. 8(c) as the cause of
stress field used in the slab method satisfies all sion are shown in Fig. 9. It is seen that the ex- the uneven texture. The final modeled part is
requirements of the kinetically admissible stress seen in Fig. 13. As shown in the real part, the
field: the equilibrium condition, the yield crite- tooth tip is the last area to fill. A clear fold is pre-
(a) (b)
rion, and the prescribed traction boundary con- dicted in the same area.
dition. As shown in Fig. 6, however, the present
slab method does not provide a lower-bound so-
lution because the stress field does not take into
account the shear stress due to friction, and
hence, it violates the Cauchy stress law on the
traction boundary. (c)
(d)
Electrode Cold Forging Process
Folding, also known as lapping, is a severe
forging defect that process designers would Fig. 10 Early stage of tooth forming. The material flow
like to predict and avoid in the manufacturing can be seen as driven up and down in forming
process. There are many challenges to model- Fig. 8 Arrows highlight the defects seen in the forming the teeth.
of the electrode.
ing a fold properly, especially in the three-
dimensional UL method. The methodologies,
to mesh a workpiece with very thin fold and to
account for self-contact condition in the FEM
formulation, are important for a robust and ac-
curate FEM model.
This section demonstrates the modeling of the
forming of a copper electrode (Ref 25). The ac-
tual progression is shown in Fig. 7. The work-
piece undergoes a total of five operations: shear-
ing, squaring, pancaking, backward extrusion,
and final forming. The pancaking and backward Fig. 9 Forming results after squaring and backward Fig. 11 As forming of the teeth continues, note the
extrusion operations were modeled in two di- extrusion. smearing of the material above the teeth.
228 / Process Design and Workability

Fig. 14 Outline of forging process of crankshaft forging

to the temperature distribution inside the work-


piece. Being able to describe the flow stress be-
havior within the process window is extremely
Fig. 12 The material peeling down as the teeth are important for an accurate simulation. Further-
formed
more, the flow stress is generally sensitive to
the strain rate at a hot forging condition; ram
speed will also play an important role, not only
to predict accurately in flow pattern, but also in Fig. 15 Crankshaft forging after buster stage
the prediction of the load. Interface properties
such as friction and heat transfer coefficient are
also important variables that will influence the
heat loss rate of the workpiece, material flow,
and forging load. Therefore, an accurate model
should take into account the coupling between
the process variables, for example, ram speed,
friction factor, heat transfer coefficient, and the (a)
Fig. 13 The final part shape material data (e.g., flow stress representation
and thermal data).
Toward this goal, a parametric study was first
Due to the small part size, it is very difficult to carried out to evaluate these critical process vari-
visualize the folds as well as understand the ables to assure the accuracy of the model. The
overall folding development. The simulation variables under study included the flow stress,
has, however, clearly revealed the complete his- friction factor, heat transfer coefficient, and bil-
tory of the fold development. (b)
let temperature.
The materials properties for AISI-1045 and Fig. 16 Comparison of buster results for (a) high heat
transfer and (b) low heat transfer
Crankshaft Hot Forging Process AISI-1055 were selected for the simulation.
There are three stages to forge the crankshaft:
This forging process and the corresponding busting, blocking, and finishing. profile reduces as the heat transfer coefficient
simulation (Ref 26) were carried out at the The workpiece is first heated to 1200°C increases. This is due to the fact that heat loss
Hyundai Motor Company. The objective of the (2192 °F). All dies are heated to 200°C (392 from the die-workpiece interface is faster using
numerical simulation was to evaluate the forma- °F). As shown in Fig. 14, after 10 s of air trans- a higher lubricant heat transfer coefficient. The
bility and to predict the forging load for both a fer, the workpiece is placed in the buster dies difference in temperature distribution influ-
crankshaft and a connecting rod. In this section, for the first forging operation. After 1.5 s air ences the material flow in the die cavity and
only the crankshaft simulation is presented. transfer, the workpiece is removed from buster flash.
Specific interests included: and placed in the blocker for the second forg- At the end of the blocker operation, the pre-
ing process. After another 1.5 s air transfer, the dicted shape and the actual shape are in excel-
● The flow pattern and potential defects such
workpiece is removed from blocker and placed lent agreement as shown in Fig. 17. It is also
as under fill or folding
in the finisher for the final forging process. A noted that the folding defects (also known as
● The forging load for tool stress analysis and
6500 ton mechanical press is used in the forg- laps) are successfully predicted by the simula-
press selection
ing and simulation. To accurately model the tion as shown in Fig. 18 and 19. The prediction
● Stress and strain distribution for possible
temperature evolution during the entire form- of the lap shape and the propagation pattern are
evaluation of microstructure
ing processes, the simulation was carried out well correlated with the corresponding experi-
● Improvement and optimization of the exist-
in a nonisothermal manner, including heat mental results.
ing forging processes
transfer analysis to account for the air transfer To validate the model further, the flash thick-
An accurate hot forging model involves many time. ness and the corresponding forging load are
material variables and process variables. It re- During the parametric study, it was found that measured and compared with the numerical pre-
quires proper characterization of the material an accurate lubricant heat transfer coefficient diction with excellent agreement as shown in
such as flow stress behavior and thermal data as was very critical to obtain a similar flash profile Fig. 20.
well as interface properties such as the friction as the actual part. The actual workpiece at the
factor and the lubricant heat transfer coefficient. end of busting is shown in Fig. 15. The pre- Cutting Modeling
In the case of a hot forging process, the thermal dicted shapes based on two lubricant heat trans-
data will affect the accuracy of the temperature fer coefficients are given in Fig. 16. From these Machinability is of primary interest in the
prediction. The deformation pattern is sensitive figures, it is noted that the waviness of the flash materials cutting process. It is affected by many
Chapter 15: Modeling Techniques in Forming Processes / 229

(a)

Fig. 20 Results of actual and simulated forming loads

(b)
Fig. 17 Comparison of (a) actual to (b) simulated
blocker result

Fig. 21 Initial and final stages of Lagrangian analysis of chip formation

Fig. 18 Actual lap on crankshaft forging lation for transient analysis and Eulerian formu- updated so that it satisfies the free surface con-
lation for steady state has been developed (Ref dition and contact condition with the given ve-
27). In this hybrid procedure, the transient UL locity field. The first part is based on the stan-
method is first used to predict the initial chip for- dard flow formulation of plastic materials. For
mation as shown in Fig. 21. The solution at the the sake of computational effectiveness, the sec-
end of the transient run is then used as an initial ond part is divided into two levels—that is, the
guess for the steady state analysis. determination of the position of surface nodes
In most solid mechanics problems, the do- and internal nodes.
main, i.e., the steady state configuration of a Employing surface elements that have both
deformable body, is unknown a priori and must membrane and edge bending stiffness then
be determined as a part of the solution. From solves the position of surface nodes. The free
the standpoint of the FEM, each node has de- surface condition is also embedded into the ele-
grees of freedom for both the velocity and the mental equations so that each free surface ele-
position. Thus, the velocity field needs to be ment is positioned to be parallel to the velocity
satisfied for the equilibrium equations, the con- at the center of element. Since the resulting set
stitutive equations and the boundary conditions, of equations is nonlinear due to the geometric
and the configuration needs to be satisfied for nonlinearity, the Newton-Raphson iterative
Fig. 19 Simulated lap on crankshaft forging
the free surface condition and the contact con- method is used. Once the position of surface
dition. Here, the free surface condition implies nodes is obtained, employing incompressible
factors, among which are materials behaviors, that the velocity on the free surface must be elastic solid elements solves for the position of
insert shape, and cutting condition. A good un- tangent to the surface and there is no traction internal nodes.
derstanding of the interactions among the chip on the surface. State variables such as stresses and strains are
flow, heat generation, residual stress, tool stress, A great number of different solution methods updated in two different ways: simple interpola-
and tool wear is crucial in order to optimize the have been presented, including ALE methods, tion scheme of existing solutions and streamline
design of the process and tooling (Ref 27). pseudo-solid domain mapping methods, adaptive tracking scheme, depending on whether the up-
To understand the thermal-mechanical re- h-p FE methods, streamline tracking methods, dated Lagrangian solutions are available. A re-
sponse on an insert (tool wear) during a long and methods of splines. In this current approach, sult of steady state thermomechanical machining
period of cutting, the conventional updated two sets of coupled governing equations are simulation is presented in Fig. 22. The steady
Lagrangian transient approach, especially for solved for both the velocity and the position it- state insert temperature is shown in Fig. 23. The
three dimensional analysis, is not efficient since eratively. First, the Eulerian velocity solution is work to validate the cutting force, temperature,
it requires enormous CPU time to reach a steady obtained based on a given configuration. Next, and wear using the model is being carried out
state. A hybrid procedure using both UL formu- the new configuration, i.e., the nodal position, is (Ref 27 and 28).
230 / Process Design and Workability

plastic deformation using rigid-plastic and


rigid-viscoplastic formulation. For processes
in which spring back or residual stress is of
interest, an elasto-plastic formulation should
be used. At high temperatures, materials
can be highly rate sensitive, in which case
an elasto-viscoplastic formulation should be
considered.
● Microstructure modeling. Currently, most
modeling considers isotropic behavior.
Recrystallization, texture, damage, phase
transformation, grain size effect, precipitate
size, and distribution could potentially af-
fect the deformation behavior. Understand-
(a) (b)
ing the microstructure properties evolution
Fig. 22 Chip shape before and after Eulerian calculation during the thermal mechanical process and
its associated effect with respect to the me-
● The forging load can be predicted for equip-
chanical properties, such as flow stress,
ment selection.
Young’s modulus, and creep behavior, is
The solution accuracy depends largely on the va- very crucial to the model accuracy since
lidity of the assumptions that are made for the these properties are generally the input data
simulation. Minimizing the time and cost of pro- to the FEM analysis.
duction trials and design iterations continues to
be the goal of FEM development. Process Conditions, Including the Friction and
Toward this goal, ongoing research is focused the Lubricant Heat Transfer Coefficient. During
on the following areas. the forming process, the surface of the workpiece
Computational Efficiency. Three-dimen- is severely deformed/stretched, wherein the lu-
sional FEM transient metal forming analysis is bricant film thickness and film temperature are
very computationally intensive. Although the continuously changing in time. Consequently,
computing power has been improving dramati- this will affect the material flow behavior. Proper
cally every year, the size of the FEM model for characterization of the lubricant behavior is im-
simulation has also been increasing quickly to portant for the accuracy of the model.
improve the model accuracy. Methodologies to Integrated FEM for Complete Manufac-
improve the computing speed, such as parallel turing Processes. Most discrete mechanical
Fig. 23 Insert temperature after cutting
components go through a manufacturing cycle,
computing and efficient contact searching algo-
rithms, are under continuous exploration and de- which typically includes ingot breakdown, forg-
velopment. Steady state or quasi-steady state ing, trimming, heat treating, welding, machin-
Current and Future Works methods are being investigated for certain classes ing, and assembling and installation. A complete
of forming processes, such as extrusion, rolling, material flow and thermal mechanical history is
and ring rolling. important to model the behavior properly under
Modeling techniques for metal forming the service condition. It is also necessary to have
Robustness and Ease of Use. The required
have continuously evolved throughout the a better understanding of the material evolution
geometry data is generated by a computer-aided
years. Because of the unique capability of FEMs and modeling throughout the entire manufactur-
design (CAD) system. Generally speaking, the
to describe the complex geometry and bound- ing process.
data may not meet the requirement for com-
ary condition of the forming process, the Optimization. The FEM provides a predic-
puter-aided engineering (CAE) analysis. An
method has proven itself by numerous success tion of the result of a proposed manufacturing
error-tolerant system is necessary for a robust
stories experienced by the industry or reported process but still relies on an experienced de-
CAE tool. In addition, realistic graphics presen-
by research organizations, such as reducing signer to interpret the results of the analysis and
tation is also important for post-processing the
the trials and errors on the shop floor, and modify the process based on prior knowledge
simulation result. To facilitate the use of the
shortening the lead time to develop a new and experience. Some recent research efforts have
FEM model, special purpose preprocessors
product, and so on. sought to use computational resources to en-
aimed for specific manufacturing process are
Specifically, the FEM has been used by the hance and optimize process designs based on a
being developed.
forming industry primarily for the following starting design, and improvement of the design
Accuracy. The FEM approach is an error
purposes: is based on the sensitivity analysis of the design
optimization process. Generally speaking, the
● The forming flow can be analyzed to evalu- larger the model is, the better the solution accu- variables. Because the amount of simulations are
ate the die/model design and to avoid flow- racy is expected. Certain assumptions, due to the generally large for the sensitivity analysis during
related defects such as folds, suck-in de- lack of understanding of many real phenomena, optimization, these techniques are still some-
fects, flow-through defects, under fill, and may introduce inevitable error. Specifically, they what limited to two-dimensional analysis but are
ductile fracture. are as follows: emerging and evolving quickly.
● The tool stress can be analyzed to improve Material Characterization: With advances in the FEM, more manufactur-
the tool life. ing processes/problems with greater complexity
● The process window control (e.g. press ● Flow stress data should cover the process are being studied and investigated. In addition to
speed, forging temperature, lubricant, and window of the actual forming process, in- the development of an integrated manufacturing
heat treatment procedure) can be modeled cluding temperature, strain, and strain rate. software, utilizing optimization techniques to
to ensure that the resulting microstructure The extrapolation of the data range usually systematically achieve design objectives is also
properties (such as grain size) meet the results in unexpected error. the focus of current and future work. Due to the
requirement. ● Elastic response has been neglected for large required computing resource in optimization and
Chapter 15: Modeling Techniques in Forming Processes / 231

to minimize the user’s time, the speed and the Kobayashi, Mechanics of Plastic Deforma- Int. J. Solids Struct., Vol 6, 1970, p. 1069–
accuracy, the robustness of the FEM procedure, tion in Metal Processing, Macmillan, New 1086
and computer graphics in three dimensions will York, 1965 18. J. Wang and M.S. Gadala, Formulation and
continuously be challenged. 6. W. Johnson and P.B. Mellor, Engineering Survey of ALE Method in Nonlinear Solid
Plasticity, Van Nostrand and Reinhold, Mechanics, Finite Elem. Anal. Des., Vol 24,
London, 1973 1997, p 253–269
ACKNOWLEDGMENTS 7. B. Avitzur, Metal Forming: Processes and 19. J.T. Oden and J.A.C. Martins, Models and
Analysis, McGraw-Hill, New York, 1968 Computational Methods for Dynamic Fric-
The authors would like to thank Professor
8. R.A.C. Slater, Engineering Plasticity: tion Phenomena, Comp. Meth. Appl. Mech.
Soo-Ik Oh, Seoul National Unversity, for pro-
Theory and Application to Metal Forming Eng., Vol 52, 1985, p 527–634
viding the upper bound method of ring com-
Processes, Macmillan, London and Basing- 20. A. Curnier, A Theory of Friction, Int. J.
pression, and Mr. D.H. Seo, Hyundai Motor
stoke, 1977 Solids Struct., Vol 20, 1984, p. 637–647
Company, for their crankshaft forging example.
9. T. Altan, S.I. Oh, and H.L. Gegel, Metal 21. A. Klarbring, General Contact Boundary
The authors would also like to thank Mr. Tony
Forming: Fundamentals and Applications, Conditions and The Analysis of Frictional
Perez and Dr. Daqing Jin, both of the Timken
American Society for Metals, Metals Park, Systems, Int. J. Solids Struct., Vol 22, 1986,
Company, for their collaborate effort on material
Ohio, 1983 p 1377–1398
cutting modeling under the DOE funded CTMP
10. H. Kudo, Some Analytical and Experimental 22. G. Strang and G.J. Fix, An Analysis of The
project.
Studies of Axi-Symmetric Cold Forging and Finite Element Method, Prentice-Hall,
Extrusion, Parts I & II, Int. J. Mech. Sci., Vol Englewood Cliffs, New Jersey, 1973
REFERENCES 2, 1960, p 102–127; Vol 3, 1961, p 91–117 23. J.T. Oden and J.N. Reddy, An Introduction
11. S.I. Oh, “Ring Compression Modeling to the Mathematical Theory of Finite Ele-
1. C.H. Lee and S. Kobayashi, New Solutions Using Upper Bound Method,” Private Com- ments, John Wiley & Sons, New York, 1976
to Rigid-Plastic Deformation Problems munication, 2002 24. O.C. Zienkiewicz and R.L. Taylor, Chapter
Using a Matrix Method, J. Eng. Ind. (Trans. 12. S. Kobayashi, S.I. Oh, and T. Altan, Metal 10, The Finite Element Method, 4th ed. Vol
ASME), Vol 95, 1973, p 865 Forming and The Finite Element Method, 2, McGraw-Hill Book Company
2. J. J. Park and S.I. Oh, Application of Three Oxford University Press, New York, 1989 25. M. Foster, Scientific Forming Technologies
Dimensional Finite Analysis to Metal 13. W.J. Chung, J.W. Cho, and T. Belytschko, Corporation Simulation Database Archive
Forming Process, Proc. NAMRC XV, On the Dynamic Effects of Explicit FEM 26. D.H. Seo, Numerical Simulation of Crank
(Bethlehem, PA), 1987, p 296 in Sheet Metal Forming Analysis, Eng. Shaft Forging Process, DEFORM Korean
3. T. Coupez, N. Soyris, and J.L. Chenot, 3-D Comput., Vol 15, 1998, p 750–776 User’s Group Meeting, 29 Oct 2002,
Finite Element Modeling of the Forging 14. R. Hill, The Mathematical Theory of Plastic- Changwon, Korea
Process with Automatic Remeshing, J. ity, Oxford University Press, London, 1971 27. Y.C. Yen, A. Jain, C. Avanachand, W.T. Wu,
Mater. Process. Technol., Vol 27, 1991, 15. L.E. Malvern, Introduction to The Mechanics and T. Altan, “Computer Simulation of
p 119–133 of a Continuous Medium, Prentice-Hall, Orthogonal Cutting Using a Tool with
4. G. Li, W.T. Wu, and J.P. Tang, DEFORM- Englewood Cliffs, New Jersey, 1969 Multiple Coatings,” submitted to 2003
3D: A General Purpose 3-D Finite Element 16. A.E. Green and P.M. Naghdi, A General College International pour l’Etude Scienti-
Code for the Analysis of Metal Forming Theory of Elastic-Plastic Continuum, fique des Techniques de Production Meca-
Processes, Metal Forming Process Simula- Arch. Rational Mech. Anal., Vol 18, 1965, nique (CIRP)
tion in Industry, 28–30, Sept 1994, (Baden- p 251–281 28. C. Pavanachand, J.T. Jinn, A. Perez, D.Q.
Baden, Germany), Internationale Kongress 17. H.D. Hibbit, P.V. Marcal, and J.R. Rice, Jin, et al., “Machining Modeling,” SFTC,
und Tagungsorganisation Finite Element Formulation for Problems Timken, DOE Project Final Report, under
5. E.G. Thomsen, C.T. Yang, and S. of Large Strain and Large Displacements, preparation, 2003
ASM International is the society for materials engineers and scientists,
a worldwide network dedicated to advancing industry, technology, and
applications of metals and materials.

ASM International, Materials Park, Ohio, USA


www.asminternational.org

This publication is copyright © ASM International®. All rights reserved.

Publication title Product code


Handbook of Workability and Process Design 06701G

To order products from ASM International:

Online Visit www.asminternational.org/bookstore

Telephone 1-800-336-5152 (US) or 1-440-338-5151 (Outside US)

Fax 1-440-338-4634

Customer Service, ASM International


Mail
9639 Kinsman Rd, Materials Park, Ohio 44073, USA

Email Cust-Srv@asminternational.org

American Technical Publishers Ltd.


27-29 Knowl Piece, Wilbury Way, Hitchin Hertfordshire SG4 0SX, United
In Europe Kingdom
Telephone: 01462 437933 (account holders), 01462 431525 (credit card)
www.ameritech.co.uk
Neutrino Inc.
In Japan Takahashi Bldg., 44-3 Fuda 1-chome, Chofu-Shi, Tokyo 182 Japan
Telephone: 81 (0) 424 84 5550

Terms of Use. This publication is being made available in PDF format as a benefit to members and customers of
ASM International. You may download and print a copy of this publication for your personal use only. Other use
and distribution is prohibited without the express written permission of ASM International.

No warranties, express or implied, including, without limitation, warranties of merchantability or fitness for a
particular purpose, are given in connection with this publication. Although this information is believed to be
accurate by ASM, ASM cannot guarantee that favorable results will be obtained from the use of this publication
alone. This publication is intended for use by persons having technical skill, at their sole discretion and risk. Since
the conditions of product or material use are outside of ASM's control, ASM assumes no liability or obligation in
connection with any use of this information. As with any material, evaluation of the material under end-use
conditions prior to specification is essential. Therefore, specific testing under actual conditions is recommended.

Nothing contained in this publication shall be construed as a grant of any right of manufacture, sale, use, or
reproduction, in connection with any method, process, apparatus, product, composition, or system, whether or not
covered by letters patent, copyright, or trademark, and nothing contained in this publication shall be construed as a
defense against any alleged infringement of letters patent, copyright, or trademark, or as a defense against liability
for such infringement.

You might also like