1 s2.0 S089069550300172X Main - Flowforming
1 s2.0 S089069550300172X Main - Flowforming
1 s2.0 S089069550300172X Main - Flowforming
Abstract
In the last two decades or so, spinning and flow forming have gradually matured as metal forming processes for the production
of engineering components in small to medium batch quantities. Combined spinning and flow forming techniques are being utilised
increasingly due to the great flexibility provided for producing complicated parts nearer to net shape, enabling customers to optimise
designs and reduce weight and cost, all of which are vital, especially in automotive industries.
In this paper, process details of spinning and flow forming are introduced. The state of the art is described and developments
in terms of research and industrial applications are reviewed. Also, the direction of research and development for future industrial
applications are indicated.
2003 Elsevier Ltd. All rights reserved.
0890-6955/$ - see front matter 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0890-6955(03)00172-X
1420 C.C. Wong et al. / International Journal of Machine Tools & Manufacture 43 (2003) 14191435
Fig. 10. Development of work zone and material stress during conventional spinning [3].
Fig. 12. (a) Comparison of measured strains from multi-pass oper- 4.2.2. Roller path
ation and calculated strains [8]. (b) Comparison of measured strains The roller path is particularly important in affecting
from single-pass operation and calculated strains[8].
the quality of a spun part. Different roller paths such as
linear, concave, convex, involute and quadratic relative
the mandrel, roller and tailstock were modelled as rigid to the workpiece have an influence on the deformation
surfaces. In addition, the friction coefficient for the roller of the blank. The tendency to buckle and cause wrinkles
is assumed to be 0 to represent the very low friction as well as cracking can be avoided by introducing the
rolling contact with the disk. The roller forces obtained correct roller path.
from their FE model were compared to the experimental A concave roller path is the most widely used one in
forces obtained by Wang et al. [6]. The axial force conventional spinning. The thinning rate in designing a
obtained from FE analysis is about 200 N for a period roller path for the first-pass should be taken into account
of only 0.75 s and Wang et al. [6] obtained about 300 as it plays a decisive role in the final wall thickness [7].
N for a process with three roller pass lasting over 100 Liu et al. [12] established an elasto-plastic FEM
s. Although comparison showed reasonably close agree- model to analyse the stress and strain distribution of the
ments, the force output from the FE analysis is only for first-pass of conventional spinning with different roller
a short time, whereas the experimental forces are the paths, namely linear, involute and quadratic, to convert
results of a process with three spinning passes. the shape of the blank to that of the mandrel. They
1426 C.C. Wong et al. / International Journal of Machine Tools & Manufacture 43 (2003) 14191435
reported that both the radial and the tangential stress and
strain are the smallest for the involute curve. They con-
cluded that a comparison of the distribution of stresses
and strains under the three different paths could provide
a theoretical basis for selecting a suitable roller path in
conventional spinning.
strain) which assumed simple shear deformation under cess variables on shear forming forces and surface finish
the roller, to derive the tangential force. He commented using regression analysis. In their study, experimental
that the tangential force formulated cannot be considered results of axial, radial and tangential forces obtained for
conclusive because flow stress of the material used to different values of feed rate, roller nose radius, mandrel
calculate the tangential force was obtained by conven- rotational speed and the initial thickness of workpiece
tional tensile test data, where the high strain rate in spin- were fed into the Statistics Analysis System
ning was not considered. Avitzur and Yang [16] also (SAS/STAT) software to determine a set of regression
attempted to calculate the tangential force by assuming equations. The results of this analysis are shown in
shear deformation. However, their formulations are quite Fig. 19.
complex and the aid of a computer is necessary to deter- In terms of experimental results, both Slater and Chen
mine the solution. Kobayashi et al. [17] re-evaluated the et al. reported a decrease in tangential force as mandrel
work done by Kalpakcioglu [14] and Avitzur and Yang rotational speed increase. For both axial and radial force,
[16] by formulating a theoretical expression for the tan- Slater observed a decrease in values as mandrel speed
gential force, including the strain due to bending in increased but Chen et al. reported a gradual increase.
addition to the shear strain. From Figs. 19 and 20, a very important point to note is
Slater [20] attempted to give an approximate upper- that experimental results, in terms of feed rate, mandrel
bound estimate for the dimensionless tangential compo- speed and tangential force results obtained by both the
nent force assuming a plane strain and axisymmetric researchers indicate that an optimum mandrel rotational
deformation and compared the theoretical predictions speed exists for a given cone angle (inclined angle of
with experimental results, as shown in Fig. 18. He the mandrel) and feed rate.
reported that the upper bound estimates and the experi- For surface finish, Slater concluded that the surface
mental results have a similar characteristic shape which roughness attained a minimum at a particular mandrel
are concave upwards. rotational speed for a given cone angle and constant
Recently, Chen [19] investigated the influence of pro- feed rate.
From the regression analysis, Chen et al. confirmed
that slower feeds, when combined with a larger roller
nose, result in reduced outer surface roughness of the
spun cone.
5.3. Spinnability
Fig. 18. Comparison of the approximate upper bound estimates with Fig. 19. Effects of mandrel speed on the tangential force at various
experimental values of dimensionless tangential component force [20]. roller feed by Chen [19].
C.C. Wong et al. / International Journal of Machine Tools & Manufacture 43 (2003) 14191435 1429
Kawai et al. [24] reported that an optimum combi- process, as shown in Fig. 24, the metal is displaced axi-
nation of the semi-cone angle of the product and feed ally along a mandrel, while the internal diameter remains
rate would prevent the occurrence of wall fracture and constant. It is usually employed to produce cylindrical
flange wrinkles. Moreover, both Kawai et al. [24] and components [26]. Most modern flow forming machines
Kang et al. [7] concluded that a minimum flange width, employ two or three rollers and their design is more
w, as shown in Fig. 22, is required to prevent the diame- complex compared to that of spinning and shear for-
tral shrinkage of the product and to realise the sine law ming machines.
in die-less shear forming. The starting blank can be in the form of a sleeve or
cup. Blanks can be produced by spinning, deep drawing
5.5. Reducing/closing thick-wall tubes or forging plus machining to improve the dimensional
accuracy. Advantages such as an increase in hardness
Thick-wall cylinders, e.g. gas cylinders, at one time
were produced from seamless tubes where the closed-
end was hot forged welded on a press. In this process,
pieces of extruded closed-end tube are fed individually
from a magazine and into a mandrel where open tube-
end to be formed is heated to a temperature of 9001000
C. by an induction coil [5,25], as shown in Fig. 23. The
spinning roller moves in a combination of axial and rad-
ial direction (controlled by CNC) to obtain the desired
reduction.
6. Flow forming
direction of the roller towards the unsupported end of Xu et al. analysed the deformation mechanism and
the mandrel. Backward flow forming is especially suit- concluded that for the same process conditions, there is
able when the original ductility of the blank is too low no obvious difference in stress and strain rate on the
to accommodate tensile stresses, such as cast and surface between forward and backward tube spinning.
welded parts. In most cases, flow forming is carried out by
The forward method is normally preferred because in employing more than one roller. Most modern machines
the backward method, worked material is required to employ the three-roller configuration mainly to achieve a
flow over the length of the mandrel, making the material better balance of loads for flow forming precision parts.
more susceptible to distortion like bell-mouthing at the Normally, the three rollers are spaced circumferentially
free end of the blank and loss of straightness [28]. More- at 120 apart, providing a uniform load distribution to
over, backward flow forming is normally prone to non- prevent the mandrel from being deflected from the centre
uniform dimension across the length of the product [29]. line. Furthermore, the rollers can be offset or staggered
Forward flow forming is usually less productive as [32,33] at a particular distance in the axial and radial
compared to the backward method as the roller must tra- direction to improve dimensional accuracy and surface
vel over the entire elongated length of the blank. In finish.
addition, workpiece length with forward method is
restricted by slide stroke and mandrel length. 6.2. Power, forces and process variables
Both Xu et al. [30] and Li et al. [31] adopted finite
element simulation to study the deformation character- There are several scientific papers and articles on flow
istics and the axial displacement distribution, respect- forming/flow forming of tubes with regard to the devel-
ively, in flow forming of tubes. For the backward opment of the theoretical methods for forces and power
method, Li et al. observed that material was displaced [28,3438]. In addition, many researchers have investi-
in the same direction (negative) as the axial feed (Fig. gated the effects of process variables on different
27a) around the undeformed tube while it flows in the components of force, power, surface finish and mechan-
opposite direction to the feed of the roller (positive) ical properties. These aspects of the investigations are
primarily in the deformation zone, thus elongating the described below.
spun parts, as shown in Fig. 27b,c. In the spun section,
metal flow is mainly in the negative direction due to the 6.2.1. Power and forces
push of local plastic deformation (Fig. 27d,e). Over the For the past few decades, several researchers have
thickness, the tangential displacement is larger in the undertaken theoretical analyses of power and force in
external layers than it is in the internal layers. tube spinning.
Fig. 27. Displacement distribution in the axial direction: (a) 1-1, (b) 2-2, (c) 3-3, (d) 4-4, (e) 5-5 [31].
C.C. Wong et al. / International Journal of Machine Tools & Manufacture 43 (2003) 14191435 1433
In 1972, Mohan and Misra [34] developed a theoreti- fish skins) on the outer surface of the formed blank
cal analysis of the plastic flow mechanism of tubes, [27]. From Fig. 28, it can be seen that by increasing the
involving the use of grid-line analysis for each individual attack angle, shown in Fig. 26, the ratio will increase
working condition and constant. The effective strains and because of interfacial friction, the higher the ratio,
and roller forces were then calculated using the plastic the higher is the tendency for the material to flow in the
work deformation by assuming that the strain path dur- axial direction. Under this condition, the circularity of
ing deformation is linear and the strain components in the tube is well kept, whereas a small ratio leads to geo-
three principal directions can be evaluated from the total metrical inaccuracies [27,39].
displacement after deformation. They reported that the On the other hand, very large circumferential to axial
values of axial, radial and tangential force show good contact ratio will cause the metal to flow at an angle
agreement with the experimental results for commer- smaller than the attack angle, leading to wave like sur-
cially pure copper. face and differences in thickness. Similarly, too large an
Later, Hayama and Kudo [35] attempted an analysis attack angle will increase the required power and
to estimate the working forces and the diameter accuracy decrease the efficiency of the forming process. There-
by using the energy method. They reported that the fore, an optimum balance between the percentage
agreement between the calculated and the experimental reduction and attack angle is necessary [27].
values is good over a wide range of conditions such as Using the basis of minimising the resultant spinning
angle of the roller, feed rate and the reduction in thick- force, Ma [40] concluded that the optimal angle of attack
ness. decreases with increasing roller diameter and friction
Wang et al. [36] adopted a plain strain model and slip- factor, but increases with larger feed rate, reduction and
line field method to solve for the forces involved in three initial thickness of the tube wall.
dimensional flow forming. They reported that the results Generally, in order to achieve suitable flow in flow
correlate well with those of Hayama and Kudo [35]. forming and achieve the desired surface finish, a
In 1990, Singhal et al. [28] proposed a generalised compromise has to be established between the feed rate,
expression for the power required in tube spinning, by thickness of blank and roller profile [39]. If too low a
assuming no build up of material, so that it is applicable feed rate is used for a particular blank thickness, the
to both forward and backward tube spinning. In addition, material will tend to flow in the radial direction, increas-
they assumed a constant friction factor between the roller ing the internal diameter of the blank. On the other hand,
and material and also no diametral growth, as hard when high feed rates are employed, defects such as non-
materials were used. uniform thickness, reduction in diameter as well as a
Park et al. [37] adopted the upper bound stream func-
tion method to calculate the total power consumption
required in deformation and the related tangential force.
Trapezoidal and spherical velocity fields using the
stream function are suggested and their results suggested
that the trapezoidal velocity field has a better correlation
to the experimental results.
Recently, Paunoiu et al. [38] developed a generalised
theoretical model using the upper bound method for
force calculation. In their method, the characteristic con-
tact zone, considering the deformation with flat roller
nose and, respectively, with roller nose radius, is divided
using different triangular velocity fields. In this way, the
number of velocity fields could be chosen in such a way
that the more accurate values of forces could be attained.
They proposed that the method could be extended by
considering more than one-stream line of velocity and
could be applied for analysing other manufacturing pro-
cesses. Based on their theoretical results, they concluded
that flat roller nose will produce higher force as com-
pared to roller with nose radius.
Fig. 29. Production of one-piece alloy wheel, starting from rolled plate. [47].
rough surface will result [41]. If the feed rate exceeds a predominantly dominated by welded steel structures and
certain limit, it can spoil the surface finish and produce the search of a light-weight alloy to substitute for steel
thread-like serrations on the tube [26,41]. This is because wheels has led to a number of designs from North Amer-
with very large feed rates, the material tends to flow ica and Japan. The interest in Europe on a one-piece
underneath the roller in the opposite direction to the wheel saw the development of a combination of splitting
roller axial movement. [47,48], spinning and flow forming to produce the
Recently, Yao and Murata [42] investigated exper- internal nave and rim of an alloy wheel from rolled plate
imentally the effects of feed rate on thickness strain, rad- as shown in Fig. 29. Typically, the material is an alu-
ial force, diameter accuracy and surface roughness for minium alloy, AIMg2Mn0.8 (A5351).
the flow forming of aluminium tube. They concluded
that the increase in feed rate will increase spinning force,
thickness strain and surface roughness. In addition, 8. Conclusions
diametral accuracy will decrease with the increment of
feed rate. In this paper, the principles and developments of spin-
Over the years, tube spinnability (maximum ning and flow forming have been reviewed. It can be
reduction) has also been an area of investigation seen that although spinning and flow forming can be a
[26,43,44] in flow forming. The earliest investigation very complicated process in terms of deformation
was carried out by Kalpakcioglu [43]. He reported that characteristics, they have a great potential in the devel-
complete similarity was observed for cone [21,23] and opment, for the manufacture of complex shapes which
tube spinnability for the maximum reduction with tensile are being required in increasing numbers by global
reduction of area. He concluded that among the process manufacturing industries.
variables considered, nose radius, roller angle and feed, Future prospects for these techniques should see
only feed rate had a significant influence on spinnability. research directed to the manufacturing of parts with even
Xu and Feng [45] studied the spinnability of steel cast- greater complex geometries with higher accuracy and
ings and concluded that the main factor in affecting the improved performance. Of particular interest are single
spinnability is the tangential tensile strain that causes components consisting of different cross sections with
diametral growth during the forming process. flanges, sidewalls and undercuts.
[7] D.C. Kang, X.C. Gao, X.F. Meng, Z.H. Wang, Study on the [27] M. Gur, J. Tirosh, Plastic flow instability under compressive load-
deformation mode of conventional spinning of plates, Journal of ing during shear spinning process, Transactions of the ASME,
Materials Processing Technology 91 (1999) 226230. Journal of Engineering for Industry 104 (1982) 1722.
[8] E. Quigley, J. Monaghan, Metal forming: an analysis of spinning [28] P.R. Singhal, P.K. Saxena, R. Prakash, Estimation of power in the
processes, Journal of Materials Processing Technology 103 shear spinning of long tubes in hard-to-work materials, Journal of
(2000) 114119. Materials Processing Technology 23 (1990) 2940.
[9] S. Kobayashi, Instability in the conventional spinning of cones, [29] K. Xue, Z. Wang, Y. Lu, K. Li, Elasto plastic FEM analysis and
Transactions of the ASME, Journal of Engineering for Industry experimental study of diametral growth in tube spinning, Journal
85 (1963) 4448. of Materials Processing Technology 69 (1997) 172175.
[10] E. Quigley, J. Monaghan, The Finite Element of conventional [30] Y. Xu, S.H. Zhang, P. Li, K. Yang, D.B. Shan, Y. Lu, 3D rigid-
spinning using multi-domain model, Journal of Materials Pro- plastic FEM numerical simulation of tube spinning, Journal of
cessing Technology 121 (2002) 16. Materials Processing Technology 113 (2001) 710713.
[11] E. Quigley, J. Monaghan, Enhanced finite element models of [31] K. Li, N. Hao, Y. Lu, K. Xue, Research on the distribution of
metal spinning, Journal of Materials Processing Technology 121 the displacement in backward tube spinning, Journal of Materials
(2002) 4349. Processing Technology 79 (1998) 185188.
[12] J.H. Liu, H. Yang, Y.Q. Li, A study of stress and strain distri- [32] K. Xue, Y. Lu, X. Zhao, A study of matching relationships
butions of first-pass conventional spinning under different roller amongst technical parameters in stagger spinning, Journal of
traces, Journal of Materials Processing Technology 129 (2002) Materials Processing Technology 69 (1997) 167171.
326329. [33] K. Xue, Y. Lu, X. Zhao, The disposal of key problems in the
[13] C.L. Packham, Metal spinning and shear and flow forming, Sheet FEM analysis of tube stagger spinning, Journal of Materials Pro-
Metal Industries (1997) 382389. cessing Technology 69 (1997) 176179.
[14] S. Kalpakcioglu, On the mechanics of shear spinning, Trans- [34] T.R. Mohan, R. Misra, Studies on power spinning of tubes, Inter-
actions of the ASME, Journal of Engineering for Industry 83 national Journal of Production Research 10 (1970) 351364.
(1961) 125130. [35] M. Hayama, H. Kudo, Analysis of diametrical growth and work-
[15] B. Avitzur, Handbook of Metal-Forming Processes, John Wiley ing forces in tube spinning, Bulletin of Japan Society of Mechan-
and Sons, Inc., Canada, 1983. ical Engineers 22 (1979) 776784.
[16] B. Avitzur, C.T. Yang, Analysis of power spinning of cones, [36] T. Wang, Z.R. Wang, Q. Wang, Y. Zhao, S. Wang, The slipline
Transactions of the ASME, Journal of Engineering for Industry fields of thickness-reduction spinning and the engineering calcu-
82 (1960) 231245. lation of the spinning forces, in: Proceedings of the Fourth Inter-
[17] S. Kobayashi, I.K. Hall, E.G. Thomsen, A theory of shear spin- national Conference of Rotary Forming, October, 135139, 1989,
ning of cones, Transactions of the ASME, Journal of Engineering pp. 8993.
for Industry 81 (1961) 485495. [37] J.-W. Park, Y.-H. Kim, W.-B. Bae, Analysis of tube spinning
[18] M.D. Chen, R.Q. Hsu, K.H. Fuh, Analysis of internal dieless processes by the upper bound stream function method, Journal
spinning of cones, in: Proceedings of the Fourth International of Materials Processing Technology 66 (1997) 195203.
Conference of Rotary Forming, Oct 1721, 1989, pp. 114119. [38] V. Paunoiu, D. Nicoara, M. Teodorescu, A general upper bound
[19] M.D. Chen, Forecast of shear spinning force and surface rough- method for forces calculation in tube spinning process, in: Pro-
ness of spun cones by employing regression analysis, Inter- ceedings of the Sixth ICTP, Sept 1924, Advanced Technology
national Journal of machine tools and manufacture 41 (2001) of Plasticity, vol. 1, 1999.
17211734. [39] M. Jahazi, G. Ebrahimi, The influence of flow-forming para-
[20] R.A.C Slater, Spin forging of sheet metal cones having various meters and microstructure on the quality of a D6ac steel, Journal
cone angles from 70/30 brass and commercially pure aluminum, of Materials Processing Technology 103 (2000) 362366.
in: Symposium Tutzing/Germany, Aug 28Sep3, 1978, Metal [40] Z.E. Ma, Optimal angle of attack in tube spinning, Journal of
forming plasticity, Springer-Verlag, Heidelberg, Berlin, New Materials Processing Technology 37 (1993) 217280.
York, 1979. [41] P. Rajnish, R.P. Singhal, Shear spinning of technology for the
[21] R.L. Kegg, A new test method for determination of spinnability manufacture of long thin wall tubes of small bore, Journal of
of metals, Transactions of the ASME, Journal of Engineering for Materials Processing Technology 54 (1995) 186192.
Industry 83 (1961) 119124. [42] J. Yao, M. Murata, An experimental study on paraxial spinning
[22] S. Cui, Discussion on the causes of scaling cracks on the outside of one tube end, Journal of Materials Processing Technology
surface of shear-spun steel conical workpiece, in: Proceedings of 6066 (2002) 16.
the Fourth International Conference of Rotary Forming, Oct 17 [43] H. Xu, W. Feng, Maximum reduction in power spinning of tubes,
21, 1989, pp. 128130. Transactions of the ASME, Journal of Engineering for Industry
[23] S. Kalpakcioglu, A study of shear-spinnability of metals, Trans- 86 (1964) 4554.
actions of the ASME, Journal of Engineering for Industry 83 [44] X. Honglie, F. Wengang, A research on the spinnability of normal
(1961) 478483. steel castings, in: Proceedings of the Fourth International Confer-
[24] K. Kawai, L.N. Yang, H. Kudo, A flexible shear spinning of ence of Rotary Forming, Oct 1721, 1989, pp. 8993.
truncated conical shells with a general-purpose mandrel, Journal [45] D. Xu, Thickness reduction spinning of the thin-walled cylinder
of Materials Processing Technology 113 (2001) 2833. with inner stiffness, in: Proceedings of the Fourth International
[25] Y. Han, Z. Si, S. Wan, W. Wang, Q. Guo, J. Ma, Hot spinning Conference of Rotary Forming, Oct 1721, 1989, pp. 97102.
technology for thickening the bottom from pipe billets, in: Pro- [46] D. Pollit, Metal spinning in the automotive industry, Sheet Metal
ceedings of the Fourth International Conference of Rotary For- Industries 72 (4) (1995).
ming, Oct 1721, 1989, pp. 109113. [47] S. Dieter, H. Stefan, Tooling and process control for splitting of
[26] S.C. Chang, C.A. Huang, S.Y. Yu, Y. Chang, W.C. Han, T.S. disc blanks, Journal of Materials Processing Technology 98
Shieh, H.C. Chung, H.T. Yao, G.D. Shyu, H.Y. Hou, C.C. Wang, (2000) 6569.
W.S. Wang, Tube spinnability of AA2024 and 7075 aluminum, [48] S. Hauk, V.H. Vazquez, T. Atlan, Finite element simulation of
Journal of Materials Processing Technology 8081 (1998) 676 the Flow-Splitting-Process, Journal of Materials Processing Tech-
682. nology 98 (2000) 7080.