Prediction of Multipreform Shapes in Warm Forming with Experimental Verification
Prediction of Multipreform Shapes in Warm Forming with Experimental Verification
T. F. KONG
L. C. CHAN
This study uses a computer-aided simulation approach to predict the multipreform shapes of warm-forming intricate components. Nearly 100% of the scraps of primary hollow preforms are used to make secondary hollow preforms. This study simultaneously fabricates the AISI 316L stainless steel watch bezel by using scraps from the corresponding watch case. The appropriate preforms are designed with the aid of computer simulation such that die filling is completed, flash is reduced, and forming load is decreased. The specimens were prepared by custom-made tooling to verify the simulation results. Furthermore, the forming facilities are specially configured to carry out the physical experiments. Engineers eventually gain a better understanding of the warm-forming process using computer simulation. Moreover, they are able to design accurate preforms and fully utilize the material, which leads to a 50% improvement of the material utilization rate. The full material utilization also saves 40% and 20% of the total production cost and time, respectively.
Most metal-formed components are hollow
objects (e.g., watch cases, watch bezels, and spur
gears). Forming hollow components using solid
blanks or preforms in one single operation involves
a high degree of deformation. This deformation
results in increased forming loads and shortened
die life, which make this process unwise and
impractical. Reasonably, the hollow preform should
be employed to form the hollow component. The
forming load is reduced if the hollow preform is
correctly designed. Moreover, its scrap from
the center hole is reused to produce other
A generic procedure has been developed for
integrating computer-aided design (CAD) and
computer-aided engineering into the bulk-forming
design.110 Engineers effectively minimize the
number of trial runs using these tools and software.
As a result, cost, time, and resources are
significantly reduced. Most of the previous studies have
investigated forming objects or problems typically
related to rib-web parts including H-shaped
parts,1117 connecting rods,1719 and disks.2022
Moreover, generalizing blank or preform design for
various bulk-forming types follows very few rules
and guidelines. Only the empirical rules of Lange
and Meyer-Nolkemper23 cover most of the
axisymmetric and nonaxisymmetric formed
components. Accordingly, Lange and Meyer-Nolkemper23
suggested designing the blank or preform with an
outer profile nearly the same as the largest slice at
the cross-section of the formed component.
However, the literature mainly focuses on forming solid
components and preforms without a center hole
(i.e., nonhollow objects). The preforms in these
cases should be designed in solid objects. Hence, no
scrap is used as material for forming another
One blank metal is able to produce at least a pair
of hollow components. The scraps are reused to
produce subsequent component if the scrap volume
produced during the primary component formation
is larger than that of the secondary component. The
simple concept is shown in Fig. 1 as a flowchart
highlighting how the secondary hollow preform is
subsequently produced. However, concept
implementation is very difficult because the preforms
have a large range of possible configurations that
depend on the desired shapes of the final formed
Apart from the preform design, material
temperature is another factor highly influencing the
effectiveness of bulk forming. The formability of most
metals under room temperature is relatively lower
than that under elevated temperatures. A large
number of dies and multistage operations are
involved in the manufacture of metallic components
using cold forming. The process is improved by
employing warm forming with an operation carried
out above room temperature and below the
workpiece recrystallization temperature. The main
advantage of this method is that it practically
attains a close tolerance similar to cold forming and
achieves nearly the same formability with that of
hot forming.24,25 Under suitable conditions, warm
forming effectively forms preheated metals into
desired shapes in one single stroke. The cost of heating
a workpiece up to its hot-forming temperatures is
very expensive. The cost of the overall
manufacturing process is reduced if the operation
temperature is decreased.26
This article provides a solution based on the
computer simulation methodology to demonstrate a
cost-saving and effective approach. The
methodology aids the design of both primary and secondary
hollow preforms for warm forming intricate
components. The primary objective of this study is to
determine the multipreform shapes to ensure that
the metal can completely fill up the die cavity with
minimum material loss into flash. An analytical
approach using CAD and numerical simulation is
proposed to predict the multipreform shapes
accurately. This study uses the AISI 316L stainless steel
watch bezel produced from the scrap of the
corresponding watch case. Moreover, a finite-element
(FE) simulation software (i.e., DEFORM-3D;
Scientific Forming Technologies Corporation,
Columbus, OH) is used to predict the die filling of the
warm-forming processes to achieve a significant and
efficient improvement of the material utilization
rate (i.e., over 50%). The scrap taken from the
center hole of a primary hollow preform used to form
the watch case is reused to produce the secondary
hollow preform for watch bezel. Custom-made
warm-forming dies and a mechanical press are
employed to produce the specimens. These
specimens are used to verify and compare the results of
the simulations and experiments. The proposed
methodology allows engineers to use the material
fully by gaining a better understanding of the
Configurations and Material of Formed
This study used a circular nonaxisymmetric
watch case as the primary formed component. The
secondary formed component was an axisymmetric
watch bezel. The NX (Siemens PLM Software,
Plano, TX) CAD software was employed to construct 3D
models. Model shapes and dimensions are presented
in Fig. 2. The watch case was a good example for
warm forming because of its intricate shape and
compact size. The volume of the primary (i.e., VcI)
and secondary formed components (i.e., VcII) were
3,753.86 mm3 and 2,749.69 mm3, respectively. For
these types of compact-size components, a hollow
preform was initially produced from a thick sheet
metal, followed by warm forming using closed-die in
one single stroke. The parting lines of the
warmforming dies were placed at the areas with largest
cross-sections, center holes, and around the entire
perimeters. Flashes with different thicknesses were
extruded at parting lines after the die cavities were
completely filled up. Other traditional processes
(e.g., trimming of flashes and machining) were then
applied to finish the components construction.
The formed material was an austenitic stainless
steel (i.e., AISI 316L) with excellent strength and
corrosion resistance. The flow stress r data used in
the FE simulations were obtained from a handbook
published by the ASM International.27 The data
covered strain e = 0.1, 0.2, 0.3, 0.4, and 0.5;
temperature T = 600 C, 700 C, 800 C, 900 C, and
1000 C; and strain rate e_ = 0.001, 0.01, 0.1, 1, 10,
and 100/s. The approximate flow stress r and strain
e relationship was induced by an exponential
equation Eq. 1 for conditions that exceeded the bounds of
the preceding e, T, or e_ values:27
where K is the strain coefficient and n is the
strainhardening exponent obtained based on further
DEFORM-3D computation (Table I).
Experimental Setup for Warm Forming
The warm-forming dies were mounted inside a
mechanical press with a maximum load capacity of
300 tons. The preforms were heated up to its
warmforming temperature using a high radio-frequency
(RF) induction heater. The setup of the
warmforming equipment is illustrated in Fig. 3. The
warm-forming dies were made according to the final
shapes of the formed component because this
process was a single-stroke forming operation. The
operation included top and bottom dies, punch, and
ejector. The configuration was designed for a
backward-extrusion process, where the metal flows in
the opposite direction of the punch during the
forming process. A heat-treated hot work tool steel
W302 equivalent to AISI H13 had good temper
resistance and superior strength under
warmforming conditions. Hence, it was used as the
tooling material. An allowance of 0.8% was added to
each die impression to compensate for the thermal
expansion of 1.926 9 10 6/ C at around 870 C28
during the warm-forming process and the shrinkage
Fig. 1. The concept of using the scrap taken from the center hole of the primary hollow preform to make subsequent hollow preform for warm
forming the secondary formed component.
Fig. 2. The overall configurations of the formed components.
Table I. The K and n values for AISI 316L stainless steel at the working ranges of temperatures and strain
rates used in this study
Fig. 3. The setup of warm-forming equipment.
after workpiece cooling. A lubricant W-400 with an
approximate 0.25 coefficient of friction was spread
over the tooling surfaces.
Hollow Preform Design
Lange and Meyer-Nolkemper23 established three
rules: (I) the cross-sectional area for each preform
slice along the height axis should be equal to that of
the formed component augmented by the area
necessary for flash, (II) all the concave radii of the
preform should be larger than the radii of the
formed component, and (III) the dimensions in the
forming direction of the preform should be larger
than that of the formed component so that metal
flow is mostly of upsetting rather than extrusion
type. In this manner, the material was laterally
squeezed toward the die cavity with little shear at
the tool-workpiece interface. This condition
minimized the friction and the forming load, thereby
reducing the wear along the die surface.
The preceding rules also guide the design of the
hollow preform outer profile. We provided an the
outer profile, which was 0.2 mm smaller than the
outer profile of the formed component, to ensure
that the preform was properly located inside the
bottom die (Fig. 4). The hollow preforms were
assumed to be fabricated from raw, thick sheet metals
using either the blanking or the fine-blanking
process. The blanked preform had a uniform
crosssection along the height axis. Hence, the geometric
variables affecting the metal flow during the
forming process were the center-hole diameters dI and
dII for the primary and secondary hollow preforms,
respectively, and the material thickness tI and tII for
the primary and secondary hollow preforms,
Determination of Center Hole Diameters
The preform geometries were modeled by NX. The
punch and ejector designs were highly dependent on
Fig. 4. The configurations of (a) the primary hollow preform and (b) the secondary hollow preform.
Fig. 5. The geometric variables of the protrusions of the punches and ejectors for warm forming the hollow components.
dI and dII. The punch and ejector configurations of
the primary and secondary formed components for
simulation are illustrated in Fig. 5. The punch and
ejector geometries followed the inner contours of the
formed components. The protrusions were tapered,
and the corners were rounded and smooth with the
The end-face diameters of the punches and the
ejectors included the larger end-face punch
diameter d1p, smaller end-face punch diameter d2p, larger
end-face ejector diameter d1e, and smaller end-face
ejector diameter d2e. The d1p, d2p, d1e, and d2e
values were determined using Eqs. 25:
2Hp tan hp
2He tan he
2Hp tan hp (4)
2He tan he (5)
where Dp and De are the punch and ejector
diameters, respectively; Hp and He are the punch and
ejector protrusion heights, respectively; hp and he
are the punch and ejector draft angles; and Rp and
Re are the punch and ejector corner radii.
This study used the simulation software to test
five different primary hollow preform designs (i.e.,
P1, P2, P3, P4, and P5). dI of the P1, P2, P3, and P4
preforms were equal to d1p, d2p, d1e, and d2e of their
tooling. P5 was a nonhollow, solid metal block used
to compare the required forming loads between the
hollow and nonhollow preforms.
The punch and ejector geometries for forming the
secondary formed component were determined in
the same manner with that of the primary formed
component. The simulation software tested two
different designs (i.e., PA and PB). dII of the two
designs were then set to d1p and d2p of their tooling.
Determination of Material Thicknesses
The CAD software was then employed to
determine the cross-sectional areas AI and AII for the
primary and secondary hollow preforms. Assuming
the volume of the preform equals the volume of the
formed component plus flash, tI and tII were
calculated by Eqs. 6 and 7:
where VcI and VcII are the volumes of the primary
and secondary hollow preforms, respectively; and
VfI and VfII are the flash volumes of the primary and
secondary formed components, respectively.
Minimizing the amount of metal loss into flash
was one of the most significant preform design
requirements. Accordingly, Biswas et al.13
successfully reduced the excess material (i.e., flash) to 6%
and 8.75% of their formed components using CAD.
The excess material of the secondary formed
component (i.e., watch bezel) in this study was
relatively small and VfII was 5% of VcII. The primary
formed component was a nonaxisymmetric object
with many intricate features. Hence, forming was
more difficult than that of the secondary formed
component and the rib-web components
demonstrated in the previous study.13 Correspondingly,
VfI for P1, P2, P3, and P4 was theoretically set to
10% of VcI. P5 did not produce internal flash. Hence,
VfI for P5 was set to 5% of VcI.
The primary hollow preform was blanked from
the raw thick metal/plate supplied with thicknesses
in steps of 0.5 mm (e.g., 6.0 mm, 6.5 mm, 7.0 mm,
and 7.5 mm). Therefore, tI calculated by Eq. 6 was
rounded up to the actual thickness t of the sheet/
plate. The values of all the geometric variables for
the primary and secondary hollow preforms are
listed in Tables II and III. The flash was around
10.9% to 14.4% of VcI, which was better than the
average value of 25% as in the case of conventional
Applicable Preform Design Solution
An appropriate primary hollow preform design
should fulfill two criteria: (I) It should produce
defect-free formed component with minimum flash
and (II) the scrap should have sufficient volume to
produce the secondary formed component.
Scrap of P1
Scrap of P2
Scrap of P3
Scrap of P4
Fig. 6. The approach to designing the primary and secondary hollow preforms.
The secondary formed component was not
produced in one single operation because its shape
significantly differed from the scrap. An
intermediate preform was produced from the scrap instead.
The volume and geometry of intermediate preform
were determined after the dII and tII values were
found. The scrap volume Vs had to be larger than
the volume of the intermediate preform plus flash.
Consequently, Vs of P1, P2, P3, and P4 were
determined in this manner.
The primary hollow preform was produced either
by the blanking or the fine blanking process. The
scrap diameter ds was different from dI because of
the gap between the punch and the die. ds was be
determined using Eq. 8:
where ct is the clearance percentage equal to 5% and
0.5% for blanking and fine-blanking processes,
respectively.29 The scraps of P1, P2, P3, and P4
were cylindrical billets. Hence, their cross-sectional
areas As and volumes Vs were simply calculated.
The calculation results are tabulated in Table IV.
The applicable solution with minimum t was
usually considered as the most suitable among the
multiple design solutions satisfying the
core-forming requirements (i.e., complete die filling and
defect-free forming) because the thinner sheet metal
reduced material cost. The forming load also
occasionally played an important role because it reduced
the die wear. The whole design procedure is
summarized in Fig. 6.
The FE simulation software (i.e., DEFORM-3D)
was used to simulate the warm-forming processes
under the conditions listed in Table V. All the
geometric models used in this study were quartered to
improve the accuracy for metal-flow prediction,
speed up computations, and increase the number of
meshes for each volume. All the tool components
were assumed as rigid-body objects. The preforms
were modeled using rigid viscoplastic formulation,
which only considered plastic deformation because
elastic deformation was not significant in this bulk
forming problem. Furthermore, elastic deformation
Fig. 6. continued.
was neglected to simplify and accelerate
convergence in the iteration. The metal deformation
behaviors were modeled using the flow stress data
of the AISI 316L stainless steel published in the
metal handbook.27 The thermal properties including
thermal conductivity and heat capacity were taken
from the literature edited by the Airco Company
International.28 An emissivity of 0.5 was taken to
represent the surface quality of the raw thick sheet
Deformation solver and
Workpiece-tool shear friction factor
Coefficient of heat transfer
Ram speed of top die and punch
Initial temperature of preform
metal. This value was verified using an infra-red
thermometer and a K-type thermocouple.
RESULTS AND DISCUSSIONS
The watch bezel used in this study was produced
by reusing the scrap obtained during the watch case
fabrication. The work flow for this process is
illustrated in Fig. 7. The FE simulation software
facilitated the relevant tooling design.
Acquisition of Secondary Formed Components
Through Warm Forming
The simulation results of the warm-forming
process, which produced the secondary formed
components from two different hollow preforms (i.e., PA
Fig. 7. The main flow illustrates overall production of primary and secondary formed components through simulation process.
Fig. 8. The simulation results of warm forming for producing secondary formed components from (a) PA, and (b) PB.
and PB), are shown in Fig. 8. The main difference
between the two was the die filling of the boss at top
of the component (i.e., the die cavity bottom). The
metal in the case of PB completely filled up the
groove in the bottom die to form the entire boss.
However, the metal in the case of PA only partially
filled up that region. A shortcoming of internal
flashes was also observed, which had no effect on the
final product because these internal flashes were
trimmed off after the forming process. Therefore,
PB was the most appropriate design.
Fig. 9. The configuration of the intermediate preform.
Warm-Forming Scraps for Intermediate Preform Production
The PB design indicated that the dII and tII values
were 24.1 mm and 5.4 mm, respectively. The
intermediate preforms illustrated in Fig. 9 were
designed based on these two values. The total volume
for the preform with 2.5% flash was 3617.8 mm3. A
comparison of the data shown in Table V showed that
only the scraps of P2 and P4 had sufficient volumes.
Hence, P1 and P3 were rejected.
The simulation results of warm-forming the
intermediate preforms from the scraps of P1, P2, P3,
and P4 are shown in Fig. 10. Many regions of the die
cavities in the scraps of P1 and P3 were not filled up
because of the insufficient metal volume. By
contrast, the metal completely filled up the die cavities
in the scraps of P2 and P4. The amount of flash in
P4 was higher because its scrap was larger in size.
The flashes and excess materials at the bottom of
the intermediate preforms were trimmed off to
produce the secondary hollow preforms. Moreover,
the scraps of P2 and P4 were used to fabricate the
secondary formed component.
Prediction of Primary Hollow Preform Design Using
The simulation results of warm forming from five
different hollow preforms (i.e., P1, P2, P3, P4, and
P5) are presented in Fig. 11. Accordingly, the
forming was successful under all five conditions.
The die cavity was completely filled up.
Furthermore, the excess material overflowed as flash
toward the parting line. The flash formations for P2,
P3, and P4 were similar even though their dI and t
values were different. In addition, the amounts of
internal and external flashes were nearly the same.
The tapered punch extruded the central material of
these three hollow preforms outward and
Fig. 10. The simulation results of warm forming for producing intermediate preforms from scraps of (a) P1, (b) P2, (c) P3, and (d) P4.
Fig. 11. The simulation results of warm forming for producing primary formed components from (a) P1, (b) P2, (c) P3, (d) P4, and (e) P5.
quently compressed them. An excessive amount of
internal flashes was observed in P1. The dI of P1
was set to d1p (i.e., 23.80 mm), which was much
smaller than d2p (i.e., 26.60 mm), d1e (i.e.,
25.50 mm), and d2e (i.e., 27.50 mm) for P2, P3 and
P4, respectively. A larger amount of the central
material in P1 was extruded, which caused the
excessive amount of flashes along the center hole
contours of the formed component.
P2 and P4 fulfilled the objective of this study
because the scraps produced in forming the
primary formed component of these designs were used
to produce the secondary formed component. P2
was the most appropriate design because of its
smaller t and material cost (i.e., 7%) lower than
that of P4.
However, the best choice was P1 if the feasibility
of using the scrap to form the secondary hollow
preform was not considered. P1 was suitable
because its t value was the smallest among the
primary hollow preforms. Although t of the metal block
for P5 was thinner than that of P1, the required
forming load of P5 was much higher. The
loadstroke curves in Fig. 12 demonstrate that the load is
more than double the corresponding loads for the
Fig. 12. The load-stroke curves of warm forming the primary formed
components from P1, P2, P3, P4, and P5.
hollow preforms. Forming the hollow metallic
components using the hollow preforms was more
economical because die life increased with decreasing
Verification by Experiments
An experiment using custom-made tooling and a
mechanical press specially modified for warm
forming was carried out to compare and verify the
results obtained from the simulation software. The
AISI 316L stainless steel specimens were produced
with the approximate dimensions of the best
solution (i.e., P2 and its scrap). The t values of the
specimens were around 7.2 mm to 7.5 mm because
of the die roll effect of blanking. dI and ds, which
were 26.6 mm and 25.9 mm, were similar to the
parameters used in the simulation. The specimens
were processed following the procedures in Fig. 13.
Accordingly, the process conditions of warm forming
were compatible with the simulations.
No forming defect was nearly observed on the
formed components. Moreover, the accuracy of their
critical dimensions (e.g., total lengths, widths,
heights, and flash thicknesses) was within
0.1 mm. The components appearances were very
similar to those predicted by the simulation
software. The only discrepancy found was in the
formation of an external flash near the lugs of the
watch case. The formation was due to the die-roll
Fig. 13. The verification by physical experiments.
Volume of the blank(s) required
Volume of the primary formed component
Volume of the secondary formed component
Volume of wastes
Rate of material utilization
Improvement of material utilization
Total production cost
Overall production rate
effect, which resulted in an uneven t along the
specimen edge. This error was acceptable because of
the sufficient design margin between the formed
components and the final machined product.
The metal might not be able to fill up some
intricate regions of the die, especially at the outer
region of the formed component where the
simulation showed only a small amount of flash, if the total
die roll of the raw blank was greater than 0.5 mm.
This error could cause a great discrepancy between
the simulation and the actual experiment.
Approximately 40% of the total production cost
was saved using the methodology reported in this
article. In addition, the overall production rate
increased by around 20%. By contrast, the
conventional method required two separate raw blanks
with volumes of 6964.6 mm3 and 5598.3 mm3 for
the watch case and bezel, respectively. These values
led to a total volume of 12562.9 mm3, which was
much higher than the 8125.3 mm3 volume
presented in this study. The comparison of the
conventional and the proposed approaches is presented
in Table VI. The greatest benefit achieved was the
50% improvement in material utilization.
An analytical approach in designing primary and
secondary hollow preforms for warm-forming
intricate components was successfully developed using
computer simulation. The AISI 316L stainless steel
watch case and its bezel were fabricated and used as
examples. The scrap of the primary hollow preform
center hole was reused through careful design to
produce the secondary hollow preform. The reuse
resulted in more than 50% improvement in the
material utilization rate. The center-hole diameters
(i.e., dI and dII) and material thickness (i.e., tI and tII)
were significant geometric variables influencing the
metal flow, die filling, and flash formation during the
warm-forming process. Their values were
successfully determined with the aid of computer simulation.
The proposed methodology was intensively
verified through experimental work. The appearances of
the formed components were very similar to those
predicted by numerical simulations. This approach
was satisfactory and able to improve greatly the
efficiency of the warm-forming process design.
Approximately 870 pieces per day
Increased by 54.6%
Approximately 1,050 pieces per day
The work described in this article was supported
by grants from the Research Grant Council of the
Hong Kong Special Administrative Region, P.R.
China (Project No. PolyU 511511).
1. H.A. Kuhn and L.B. Ferguson , JOM 36 , 69 ( 1984 ).
2. A.G.K. Jinka , JOM 47 , 42 ( 1995 ).
3. B.P.P.A. Gouveia , J.M.C. Rodrigues, and P.A.F. Martins , J. Mater . Process. Technol. 73 , 281 ( 1998 ).
4. H.C. Altmann and W.J. Slagter ( Paper presented at NUMIFORM-Proc. of 7th International Conference on Numerical Methods in Industrial Forming Processes , Toyohashi, Japan, 2001 ).
5. S.I. Oh , W.T. Wu , and K. Arimoto , J. Mater . Process. Technol. 111 , 2 ( 2001 ).
6. G. Li , J.T. Jinn , W.T. Wu , and S.I. Oh , J. Mater . Process. Tech. 113 , 40 ( 2001 ).
7. M.L. Alves , J.M.C. Rodrigues, and P.A.F. Martins , Model. Simul. Mater. Sci. 11 , 803 ( 2003 ).
8. G.E. Dieter , H.A. Kuhn , and S.L. Semiatin, eds., Handbook of Workability and Process Design (ASM International: Materials Park , OH, 2003 ).
9. T.J. Shin , Y.H. Lee , J.T. Yeom , S.H. Chung , S.S. Hong , I.O. Shim , N.K. Park , C.S. Lee , and S.M. Hwang , Comput. Method Appl. M . 194 , 3828 ( 2005 ).
10. L.H. Lang , A.J. Xu , and F. Li , JOM 64 , 309 ( 2012 ).
11. J.J. Park , N. Rebelo , and S. Kobayashi , Int. J. Mach . Tool . D. R. 23 , 71 ( 1983 ).
12. S.I. Oh and S.M. Yoon , CIRP Ann. Manuf. Technol . 43 , 245 ( 1994 ).
13. S.K. Biswas and W.A. Knight , Int. J. Mach . Tool . D. R. 15 , 179 ( 1975 ).
14. S. Sheng and L.Y. Guo , J. Mater . Process. Technol. 34 , 349 ( 1992 ).
15. C.S. Han , R.V. Grandhi , and R. Srinivasan , AIAA J . 31 , 774 ( 1993 ).
16. R.Y. Lapovok and P.F. Thomson , Int. J. Mach . Tool. Manuf. 35 , 1537 ( 1995 ).
17. H.Y. Kim and D.W. Kim , J. Mater . Process. Technol. 41 , 83 ( 1994 ).
18. T. Takemasu , V. Vazquez , B. Painter , and T. Altan , J. Mater . Process. Technol. 59 , 95 ( 1996 ).
19. V. Vazquez and T. Altan , J. Mater . Process. Technol. 98 , 81 ( 2000 ).
20. S.M. Hwang and S. Kobayashi , Int. J. Mach . Tool . D. R. 26 , 231 ( 1986 ).
21. G. Zhao , E. Wright , and R.V. Grandhi , Int. J. Mach . Tool. Manuf. 35 , 1225 ( 1995 ).
22. G. Zhao , Z. Zhao , T. Wang , and R.V. Grandhi , J. Mater . Process. Technol. 84 , 193 ( 1998 ).
23. K. Lange and H. Meyer-Nolkemper , Close-Die Forming (in German) (Berlin: Springer, 1977 ).
24. S. Fujikawa , H. Yoshioka , and S. Shimanmura , J. Mater . Process. Technol. 35 , 317 ( 1992 ).
25. M. Hirschvogel and H.V. Dommelen , J. Mater . Process. Technol. 35 , 343 ( 1992 ).
26. G.E. Totten , K. Funatani , and L. Xie, eds., Handbook of Metallurgical Process Design (New York: Marcel Dekker , 2004 ).
27. Y.V.R.K. Prasad and S. Sasidhara , Hot Working Guide: A Compendium of Processing Maps (Materials Park, OH: ASM International , 1997 ).
28. Allegheny Ludlum Steel Corporation, Stainless Steel Fabrication (Pittsburgh: Allegheny Ludlum Steel Corporation , 1957 ). https://books.google. com.hk/books?id=TOJTAAAAMA AJ&q=%22Stainless+Steel+Fabrication%22+1959&dq=%22 Stainless+Steel+Fabrication%22+ 1959 & hl=zh-TW &sa=X& ei=0oqOVNHpK-OxmAXIsICIBg&ved=0CB0Q6AEwAA.
29. Schuler GmbH , Metal Forming Handbook ( New York : Springer, 1998 ).