You are on page 1of 8

Journal of Materials Processing Technology 72 (1997) 94 101

An upper-bound analysis of metal forming processes by nodal


velocity fields using a shape function
J.W. Park, Y.H. Kim *, W.B. Bae
Engineering Research Center for Net-Shape and Die Manufacturing, Pusan National Uni6ersity, San 30, Jangjeon-dong, Kumjeong-gu,
Pusan 600 -735, South Korea
Received 10 May 1996

Abstract
In this paper, a new method of constructing a velocity field is proposed to solve metal forming problems by the upper-bound
elemental technique (UBET). The velocity fields can be composed of nodal points using a shape function. The forging load and
the deformed profile are obtained by minimizing the total energy-consumption rate, which is a function of unknown velocities at
each nodal point. The velocity and streamline distribution can be investigated at each time step. As an example, these velocity
fields are applied to analyze the axisymmetric cylinder upsetting problem. The effectiveness of the proposed method is verified by
comparing the results of this method with those of the finite-element method (FEM) and of experiment with fairly good agreement
being found between them. 1997 Elsevier Science S.A.
Keywords: Upper-bound elemental technique; Nodal velocity fields; Shape function

1. Introduction
There are several numerical methods for analysing
metal forming processes, such as the upper-bound elemental technique (UBET) and the finite element
method (FEM), which were developed on the basis of
the upper-bound theorem established by Prager and
Hodge [1]. Although the FEM is recognized as one of
the strongest methods because of the rapid development of computer technology, it still has some demerits
when applied to actual forming fields and requires
further study for development. On the other hand,
UBET has some merits over FEM in some respects, but
it cannot be used generally. Therefore, many researchers such as Kudo [2,3], Kobayashi [4], Bramley et
al. [57], Kiuchi et al. [8,9] and Kim [10] have made
efforts to extend its range and develop it further. Because there was not a clear breakthrough to the problems mentioned above the study of UBET is in decline.
One of its biggest problems is the limitation which is
brought about by applying velocity fields using rightangled tetragons or right-angled triangles. To overcome

* Corresponding author. Fax: +82 51 5147640.


0924-0136/97/$17.00 1997 Elsevier Science S.A. All rights reserved.
PII S 0 9 2 4 - 0 1 3 6 ( 9 7 ) 0 0 1 3 5 - 0

such problems, velocity field formulation using a steam


function was proposed by the Shimizu [11] and generalized by Lin [12,13] using a shape function that is used
in the FEM. Velocity fields by stream function are an
important advancement because the analysis can be
performed using more general forms of velocity fields.
The configuration of the element boundary can even
be optimized to the direction that minimizes the energy
rate. Also, owing to the characteristics of a stream
function, these velocity fields automatically satisfy the
volume-constancy condition, which was a great
difficulty in conventional UBET velocity fields. However, there still remain some limitations in stream function velocity fields.
Because the stream function is derived from the
incompressibility condition of two-dimensional flow, it
is difficult to be applied directly to solve three-dimensional problems. Thus Jiang [14] tried to construct more
generalized velocity fields. Although Jiang tried to solve
the problem using numerical analysis as did Shimizu, he
failed to generalize the velocity fields themselves.
Accordingly, in this paper the authors propose fields
which are constructed by interpolating the nodal velocities using a shape function. To verify the validity of the

J.W. Park et al. / Journal of Materials Processing Technology 72 (1997) 94101

95

proposed method, it is applied to analyze the cylinderupsetting problem, the result being compared with
those of the FEM and of experiment. In addition, the
bulging profile according to different friction conditions
the velocity distribution and the pattern of the streamlines are investigated.

2. Theoretical analysis

2.1. Kinematically admissible 6elocity fields


The velocity fields are constructed by the shape function and the velocity components of each nodal point.
If the given shape is a triangle, a triangular element
shape function is used and if it is a rectangle, a rectangular element shape function is used. For the simple
upsetting problem, the numbering scheme of the eight
node quadratic elements used is shown in Fig. 1. At
each nodal point, the velocity components are (u; i, u; i ).
The velocity fields are as follows:
n

U: R = % NkU: Rk
k=1
n

U: Z = % NkU: Zk
k=1
n

U: u = % NkU: uk

(1)

k=1

Fig. 1. Boundary conditions for upset forging: (a) Both upper die and
lower die moving; (b) upper die only moving.

tion from the initial situation. Thus it is somewhat


difficult to construct velocity fields that are kinematically admissible but only if they are constructed is it
easy to calculate the total energy consumption rate and
the load required for the forming of materials. However, there is no assurance that the velocity fields proposed in this paper (Eq. (1)) satisfy the
volume-constancy condition. Therefore, the volumeconstancy condition must be taken into consideration
during the optimization process of the total energy
consumption rate as a constraint condition.
Fig. 2 shows the difference of algorithms between the
conventional UBET and the method proposed in this
paper. Generally, the volume-constancy condition can

where Nk denotes the shape function.

2.2. Strain rates


The strain rates obtained from velocity fields are as
follows:
o; r =

n
(U: r
(Nk
= %
U: Rk
(r
k = 1 (r

o; u =

n
U: r 1 (U: u
N
+
= % k U: Rk
r
r (u
k=1 r

o; z =

n
(U: z
(Nk
= %
U: zk
(z
k = 1 (z

o; rz =

 

1 (U: r (U: z
1
+
=
(r
2 (z
2

n
(Nk
(Nk
U: Rk + %
U: zk
k = 1 (z
k = 1 (r
n


(2)

where n is the number of nodal points.

2.3. Constraint condition to the 6elocity fields


To be accepted as being kinematically admissible, the
velocity fields must satisfy the volume-constancy, or
incompressibility, condition. In the conventional UBET
or upper-bound analysis, the velocity fields are constructed on the grounds of the volume constancy condi-

Fig. 2. Difference between the UBET and that presented in this


paper.

96

J.W. Park et al. / Journal of Materials Processing Technology 72 (1997) 94101


Table 1
Variable value of nodal points

Fig. 3. Geometric expression of dQ= V dA.

be satisfied by considering the relationship of the incoming flow rate with the outgoing flow rate in planestrain or axisymmetric problems. The geometric
relationship of the flow rate is presented in Fig. 3.

Nodal points

r-direction velocity

z-direction velocity

1
2
3
4
5
6
7
8

O
O

O
O
O
O
O
O
O
O

1
2
3
4
5
6
7
8

O
O

O
O
O
O
O
O
O

2.4. Velocity discontinuity between adjacent elements


Fig. 4 shows a typical form of velocity discontinuity
defined by the nodal pairs (1, 2) (3, 4) and (5, 6). The
deviations of the normal and tangential velocities at
each pair (i, j ) are given respectively by:
U ijt =cos u(Uj Ui )+ sin u(Vj Vi )
U ijn =sin u(Uj Ui )+ cos u(Vj Vi )

(3)

where (i, j ) represents the values (1, 2), (3, 4) and (5, 6)
and u is the angle of the discontinuity to the x-axis or
r-axis. The condition of normal velocity continuity
between adjacent elements can be satisfied by considering U ijn = 0 as the constraint condition.

, unknown variable; O, known variable.

2.5. Upper-bound theorem


The upper-bound theorem formulated by Prager and
Hodge is to minimizing the energy consumption rate,
formulation (Eq. (4)), using assumed kinematically admissible velocity fields:
J*=

&

Fig. 4. Quadratic velocity discontinuity.

s o; dV+

&

D6 dS +

&

D6 dS

(4)

The actual energy rate that should be provided to


change the shape of material shape is nearly equal to,
or less than, the energy rate calculated by means of Eq.
(4).

J.W. Park et al. / Journal of Materials Processing Technology 72 (1997) 94101

97

2.6. Optimization by the FTM


Generally FTM [15] is used in optimizing non-linear
multi-variable functions having constraint conditions:
Minimize:

y= f(x), x

subject to:

hi (x)= 0, yi (x)] 0

where f(x) is the object function to be optimized,


x=(xi, , xn ) is a column vector having n variables in
n-dimension space, hi (x) is the equal constraint conditions and gi (x) is the unequal constraint conditions.
In analyzing the upsetting shown in Fig. 1, the
known velocity and unknown velocity of each node are
input into the program as boundary conditions. Table 1
shows the details.

3. Results and discussion


Fig. 5. Variation of forging loads according to stroke for the same
friction conditions of the upper and lower die.

The first term of Eq. (4) stands for the internal


energy consumption rate, the second term means the
shear energy consumption rate occurring along the
velocity discontinuity plane and the last term is the
frictional energy consumption rate needed to overcome
the friction between the material and the tool.
In order to minimize formulation (Eq. (4)), the flexible tolerance method (FTM) is used, which can minimize non-linear multi-variable functions having
constraint conditions. From the optimized energy consumption rate, the forming load can be found.

Fig. 6. Minimization processing by FTM.

The energy-consumption-rate formulation derived


from the velocity field proposed in this paper can be
applied to any cases of upsetting, irrespective of
whether or not the friction conditions of the upper part
of the billet and the lower part of it are the same. Fig.
5 shows the load-stroke curve of the FEM, of experiment and this simulation, when the friction conditions
are the same. As shown in this figure, good agreement
between the three is obtained.
Fig. 6 shows the iteration number and process
through which the object function is optimized when
the object function is optimized by the FEM in the case
of the load results shown in Fig. 5. As the iteration
number increases, the convergence to the optimal solution is more easily obtained. Each step has a different
iteration number to arrive at the optimal solution.
Fig. 7 shows the bulging profiles at each step of
different ram movement, for the same frictional conditions, where: (a) Shows the results when both the upper
and lower die move; and (b) shows those results when
only the upper die moves. As is expected, the same
results are obtained. The velocity distribution and the
streamlines for 30% reduction are shown in Figs. 8 and
9, respectively.
It is also an interesting investigation to check the
bulging profile when the frictional conditions of the
upper and lower part of the billet are different, the
results being shown Fig. 10. From this figure it can be
found that the outward flow is more easily achieved
smaller levels of friction than for larger levels. Although
it is not dealt with in this paper, when the friction is
large the folding phenomenon occurs easily.
The velocity and streamline distributions at 30 and
33.3% height reduction for Fig. 10 are shown in Figs.
11 and 12. As can be seen from these figures, the case
of both dies moving and only the upper die moving

98

J.W. Park et al. / Journal of Materials Processing Technology 72 (1997) 94101

Fig. 7. Bulging profile for the same friction conditions of the upper and lower die: (a) Both upper die and lower die moving; (b) upper die only
moving.

Fig. 8. Velocity distributions at 30% reduction for the same friction conditions of the upper and lower die: (a) Both upper die and lower die
moving; (b) upper die only moving.

have the same physical results, but the flow characteristics of the material are different. For the flow results for the case in which both the upper and lower
die move and the interface has different frictional
conditions, the flow of the material inclines towards
the interface having the greater friction. In the case
where only the upper die is moving, the magnitude of
the velocity vector becomes smaller from the upper to
the lower die.
The results show that the necessary information for
every step can be obtained if so the boundary condition and the constraint condition are set exactly.

4. Conclusions
Kinematically admissible velocity fields were constructed by using a shape function. From the velocity
fields, the upper-bound formulation was derived and the
minimized energy consumption rate which is a function
of unknown velocity components at the nodal points,
was found. The velocities thus obtained gave the deformed shape after a particular time. The forming load
calculated from the minimized energy rate. Both the
forming load and deformed shape are in good agreement
with the results of FEM simulation and of experiment.

J.W. Park et al. / Journal of Materials Processing Technology 72 (1997) 94101

99

Fig. 9. Streamline distributions for the same friction conditions of the upper and lower die: (a) Both upper die and lower die moving; (b) upper
die only moving.

Fig. 10. Burging profile for different friction conditions of the upper and lower die: (a) Both upper die and lower die moving; (b) upper die only
moving.

The main advantages of the method suggested are as


follows. First of all, it is easy to construct velocity fields
that are good enough to be used in the upper-bound
formulation and there is no need to assume a right-angled triangle, a rectangle, a fan shape, or a trapezoid
which is the case with conventional UBET. If the
number of nodal points is increased, a more exact
solution can be approached. Even if the velocity fields
by the shape function do not satisfy the volume constancy condition by themselves, they can be made to

satisfy it by adding a constraint condition during the


optimization procedure.

Acknowledgements
This study was supported by the overseas education and
research program of the Engineering Research Center for
Net-Shape and Die Manufacturing, to which the authors
of this study would like to express their gratitude.

100

J.W. Park et al. / Journal of Materials Processing Technology 72 (1997) 94101

Fig. 11. Velocity distributions at 30 and 33.3% reduction for different friction conditions of the upper and lower die: (a) Both upper die and lower
die moving; (b) upper die only moving.

Fig. 12. Streamline distributions for different friction conditions of the upper and lower die: (a) Both upper and die and lower die moving; (b)
upper die only moving.

References
[1] W. Prager, P.G. Hodge, Theory of Perfectly Plastic Solids,
Chapman and Hall, London, 1951.
[2] H. Kudo, An upper-bound approach to plane-strain forging
and extrusion-I, II. Int. J. Mech. Sci. 1 (1960) 5783, 229
252.
[3] H. Kudo, Some analytical and experimental studies of axisymmetric cold forging and extrusion-I, II. Int. J. Mech. Sci. 2
(1960) 102 127; 3 (1960) 91117.
[4] S. Kobayashi, Upper-bound solutions of axisymmetric forming
problems-I, II. J. Eng. Ind. Trans. ASME, (1964) 112 126,
326 332.
[5] R.P. McDermott, A.N. Bramley, Forging analysisA new approach, in: 2nd NAMRC Conference, 1974, pp. 3547.

[6] A.S. Cramphorn, A.N. Bramley, R.P. McDermott, UBET related


developments in forging analysis, in: 4th NAMRC Conference,
1976, pp. 80 86.
[7] A.S. Cramphorn, A.N. Bramley, computer aided forging design
with UBET, in: Proceedings 18th International MTDR Conference, Pergamon, Oxford, 1976, pp. 717 724.
[8] M. Kiuchi, S. Shigeta, Application of upper-bound elemental
technique (UBET) to axisymmetric forging process, JSTP 251
(1981) 1208 1214.
[9] M. Kiuchi, A. Karato, Application of (UBET) to non-axisymmetric forging process, Adv. Technol. Plasticity 2 (1984) 967972.
[10] H.Y. Kim, Preform design in closed-die forging by UBET, PhD
Thesis, Seoul National University, 1990.
[11] T. Shimizu, K. Ohuchi, T. Sano, An application of stram function
to UBET, Numerical Meth. Ind. Forming Proc. 2 (1992) 651656.

J.W. Park et al. / Journal of Materials Processing Technology 72 (1997) 94101


[12] Y.T. Lin, J.P. Wang, A new upper-bound elemental technique
approach to axisymmetric metal forming process, Int. J. Mach.
Tools Manufact. 33 (2) (1993) 135151.
[13] J.P. Wang, Y.T. Lin, The load analysis of the plane-strain
forging process using the upper-bound stream-function elemental
technique, J. Mater. Proc. Technol. 47 (1995) 345359.

101

[14] Jiang Qin, An Upper-bound approach to plane-strain problems


using a general triangular element, J. Maters. Proc. Technol. 40
(1994) 263 270.
[15] D.M. Himmelblau, Applied Nonlinear Programming, McGrawHill, New York, 1972, pp. 341 365.

You might also like