DOI: 10.1002/srin.201000109 steel research int. 81 (2010) No.

12

Deformation Analysis for the Rotary Draw Bending Process of Circular Tubes:
Stress Distribution and Wall Thinning

E. Daxin*, Yafei Liu, and Huaibei Feng

School of Materials Science and Engineering, Beijing Institute of Technology, Beijing 100081, China
* Corresponding author; e-mail: [email protected]

Rotary draw bending is a CNC tube bending process widely used in industry. Compared with pure bending, rotary draw bending involves a much
more complex loading condition. In addition to the bending moment, an axial force is generated in the process even if no drawing force is
externally applied. Taking this axial force into account, this paper presents an analysis of tube bending. The stress distribution and wall thinning
are calculated. Calculations are compared with experimental and numerical results and a good agreement is shown.

Keywords: Metal forming, Tube bending, Stress, Cross section, Wall thinning

Submitted on 4 May 2010, accepted on 9 August 2010

Introduction Analysis

The application of tubes in aerospace, shipbuilding Basic assumptions. The rotary draw bending of a tube is
and automotive industries is becoming increasingly wide shown in Figure 1. To simplify the analysis, four basic
because of the advantage of tubular parts: they are assumptions are employed:
lightweight, material-saving and have an excellent cushion (a) An ideal rigid-plastic material model is used. Elastic
performance. However, the tube bending technology has deformation and strain hardening are not considered.
been limited for a long time by a series of forming defects, (b) A plane strain state is assumed. The circumferential
including bending springback [1–3], wrinkling [4], ovaliza- strain, denoted as ew, is considered to be zero.
tion [5–10] and wall thinning [10–12]. (c) The direction of axial force (Nu) is the tangential direction
The prediction of wall thinning is crucial in the pipeline of the bending wheel.
industry. Wall thinning undermines the pressure-bearing (d) The deformation of the material follows the condition of
capacity of a bent tube. Therefore, the bending process incompressibility.
must be suitably controlled so that the wall thinning will not
exceed a certain limit; otherwise, the pipeline may fail in
Analysis of stress distribution. The stress equilibrium
use.
equation under the plane strain assumption could be given as
Investigation into the stress distribution is indispensible
for improving the tube bending process. Especially, it is ds r s r
s u
closely related to the formability of tubes. Strain-based þ ¼0 (1)
dr r
forming limit diagrams (FLD) [13, 14] have been used for
predicting the forming limit in metal forming. More where sr and su and are the principal stresses in the radial
recently, the forming limit stress diagram (FLSD) [15], and tangential directions, respectively; r denotes the
which is considered path-independent, has been widely bending radius at arbitrary point in the tube wall.
accepted as a more powerful and accurate approach than
FLD. Since FLSD is based on stress instead of strain,
knowledge of the stress distribution is believed to be of
great importance for future studies into the forming limit of
tube bending.
The tube bending deformation is a complicated elastic-
plastic problem. The plastic deformation is much larger
than the elastic deformation and it could serve as an
approximation of the total deformation. When discussing
the elastic-insensitive materials with low strain hardening,
it is feasible to use an ideal rigid-plastic model to analyse
the plastic flow and stress distribution. On this basis, the
stress distribution and wall thinning are investigated in
this paper. Figure 1. Schematic diagram of rotary draw bending.

1084 ß 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.steelresearch-journal.com
Metal Forming steel research int. 81 (2010) No. 12

Based on the Basic Assumption (b), there is sw ¼ (su þ sr)

/ 2. Thus, the von Mises stress can be written as
pffiffiffi
3  
s¼ s u
s r  (2)
2
Since su > 0 and sr < 0, sw is the intermediate principal
stress. In this case, the Tresca and von Mises yield criteria
are the same: s ¼ s y . Therefore, Eq. (2) becomes

2
s u
s r ¼  pffiffiffi s y (3)
3
The right side is positive for the tangential tensile region
and negative for the compressive region.
Through substituting this equation into Eq. (1) and
integration, we obtain

2
s r ¼  pffiffiffi s y lnr þ C (4)
3
where C is an integral constant.
At the surface of the outer convex side (i.e. r ¼ Ro), there
is sr(Ro) ¼ 0. At the surface of the inner concave side (i.e.
r ¼ Ri), there is sr(Ri) ¼ sbw. sbw is the pressure between the
bending wheel and the tube surface. Thus, the integration
constants for the outer and inner sides could be determined:

2
Ci ¼ s bw þ pffiffiffi s y lnRi (5) Figure 2. Geometry (a) and stress equilibrium (b) of tube bending with
3
axial force.

2 As Figure 3 shows, the contact force of the bending wheel

Co ¼
pffiffiffi s y lnRo (6)
3 could be calculated as

The radial stress is supposed to be continuous at the neutral Z p

2 R2
surface. In other words, sro equals sri at r ¼ R. Based on Fr ¼ s bw sin’  dsðRi daÞ ¼ pffiffiffi s y dRi daln (11)
0 3 R o Ri
this we can obtain

2 R2 Considering the force equilibrium, the relation between the

s bw ¼ pffiffiffi s y ln (7) axial force, Nu, and the contact force, Fr, could be obtained:
3 Ro Ri

Since R2  Ro Ri (this is based on the arithmetic and da

Fr ¼ 2Nu sin (12)
geometric means equality), sbw > 0. The stresses at the 2
inner concave side and the outer convex side are

2 R2
s ri ¼ pffiffiffi s y ln (8)
3 rRo

2 r
s ro ¼ pffiffiffi s y ln (9)
3 Ro

As Figure 2 shows, the tube is subject to the bending

moment M and the axial force Nu during bending. Assuming
that the tangential bending stress resulted from the axial
force is uniform in the cross section, there is
h i
Figure 3. Contact force between the bending wheel and the inner side
Nu ¼ s uN  p r2
ðr
tÞ2 (10)
of the bent tube.

www.steelresearch-journal.com ß 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1085
steel research int. 81 (2010) No. 12 Metal Forming

It is important to note that the deformation in rotary draw bending wheel. Thus, the outer tube radius of this side does
bending is point-by-point in nature. That is to say, at an not change. In the inner concave side, the compressive
arbitrary moment the bending angle produced is very small. tangential stress results in an increase in wall thickness. In
Thus, we can safely assume that sin da da
2  2 . Substituting contrast, the outer convex side could deform freely, which
Eqs. (10) and (11) into Eq. (12), the tangential bending means the tube wall could move inward at the same time of
stress resulting from the axial force could be expressed as becoming thinner [16]. The distance between the outmost
point of the cross section and the neutral surface after the
4 rRi R2 inward movement is denoted as ro here. Based on Basic
s uN ¼ pffiffiffi s y ln (13)
3 ptð2r
tÞ Ro Ri Assumptions (b) and (d), we can combine Eqs. (19) and (20)
Since the tube wall in the inner concave side is constrained and obtain
by the bending wheel, the outer tube diameter remains the R
original value (i.e. d) during bending. Therefore, t0 ¼ t  (21)
R þ ro
1 2R=d
1 R2 Since the outer radius at the outmost point, ro, can be easily
s uN ¼ pffiffiffi s y ln (13a)
3 pt=d ð1
t=d Þ Ro Ri measured, t’ can be calculated from Eq. (21).
In the pipeline industry, the cross sectional distortion of
Substituting this equation into Eq. (10), we obtain
the bent tube is usually characterized by the change rates of
h i 4 R2 the lengths of major and minor axes, i.e. wmax and wmin. They
Nu ¼ s uN p r2
ðr
tÞ2 ¼ pffiffiffi s y rRi ln (14) are calculated as
3 Ro Ri
0 0
Based on Eqs. (3),(8),(9) and (13a), the tangential stress dmax
d d
dmin
’max ¼ ; ’min ¼ (22)
distribution in the tensile and compressive regions could be d d
expressed as, respectively The wall thinning could be characterized by the wall-
  thinning ratio, which is expressed as
2 R2 2R=d
1 R2
p ffiffi
ffi
s ui ¼
s y 1
ln þ ln (15) 0
3 rRo 2pt=d ð1
t=d Þ Ro Ri t
tmin
nj ¼ (23)
  t
2 r 2R=d
1 R2
s uo ¼ pffiffiffi s y 1þ ln þ ln (16) where t’min is the minimum wall thickness after bending.
3 Ro 2pt=d ð1
t=d Þ Ro Ri The wall of the bent tube is thinnest at the outmost point.
Furthermore, based on sw ¼ (su þ sr) / 2, the intermediate Substituting r ¼ ro into Eq. (21), we obtain
principal stress could be obtained as 1
  nj ¼ 1
(24)
1 R2 2R=d
1 R2 1 þ 1
’ min
s ’i ¼ pffiffiffi s y 2ln
ln
1 (17) 2R=d
3 rRo 2pt=d ð1
t=d Þ Ro Ri
  Results and Discussion
1 r 2R=d
1 R2
s ’o ¼ pffiffiffi s y 1þ 2ln þ ln (18)
3 Ro 2pt=d ð1
t=d0 Þ Ro Ri
Stress distribution. Figure 4 shows the average value
of the tangential stress at the innermost point and that at the
Analysis of wall thinning. The tangential strain in outermost point, which are calculated using Eq. (15) and
rotary draw bending could be calculated due to finite strain Eq. (16), respectively. It is indicated that as R/d increases,
theory by su decreases slowly when R/d  2 while decreases
sharply when R/d < 2. This transition coincides with the
R  rsin’
eu ¼ ln (19) bending limit observed in experiments. Actually, the
R
where w indicates the circumferential position in the cross
section.
To simplify the calculation, the shear deformation of the
cross section is neglected while the change in wall thickness
is considered:
t0
er  ln (20)
t
where t and t’ denote the wall thickness before and that after
bending, respectively.
Similarly to the analysis of stress distribution, the analysis
of cross sectional distortion is based on the assumption that
the tube wall in the inner concave side is constrained by the Figure 4. Calculated tangential stress of various R/d and t/d values.

1086 ß 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.steelresearch-journal.com
Metal Forming steel research int. 81 (2010) No. 12

minimum R/d is between 2 and 2.5 for small-diameter-thin-

walled tubes [17].
Considering the difficulty in measuring stress, finite
element simulations have been carried out and one of the
results is presented in Figure 5. The numerical results of
the outer convex side agree well with the analytical results.
However, the absolute values of the simulated stress of the
inner concave side are greater than the corresponding
analytical values. This discrepancy might be resulted from
neglecting the mechanical parameters of the tube material
and the shift of the neutral surface during bending.

Wall thinning. Eq. (24) indicates that wall thinning Figure 6. Correlation between wall thinning and bending radius ratio.
depends on not only R/d but also the displacement of the
outer convex side. The wall-thinning ratio decreases with an
increase of wmin. In other words, the increase in the change of experimental data. For R/d > 2.5, the measured wall-
the length of minor axis lessens the deformation in the thinning ratios are all below 0.1 (i.e. the limit for high-
thickness direction and thus reduces the wall-thinning ratio. pressure tubes). The nj values of T2 tubes are slightly smaller
Since the transmission detection of wall thickness is both than those of 1Cr18Ni9Ti tubes, which is because the
expensive and time-consuming, nj was obtained from Eq. material parameters are not involved in the analysis [18].
(24), in which the wmin value could be calculated based on
simple measurement. Actual detection shows that the Conclusions
calculation of Eq. (24) is higher than the measured value
and the equation could be corrected as In this study, the stress distribution and wall thinning in
rotary draw bending are investigated analytically based on
1 the ideal rigid-plastic materials model. Since the axial force
nj ¼ 1
(25)
1
b’min is incorporated in the deformation condition, the analysis is
1þ
2R=d more practical for the CNC bending process widely used in
industry. The following conclusions could be drawn.
where b is the correction factor. Extensive repeated tests (1) The analytical expression of the stress distribution is
show that b ¼ 2 could yield a satisfactory correction. derived. Equations for the determination of the axial
To elucidate the nj-wmin- R/d correlation, the analytical force and the contact force are also given. In the outer
and experimental results are presented in Figure 6. The two convex side, the analytical values of stress agree well (the
horizontal lines are the wall-thinning limits for high- and error is within 5%) with the simulations. In the inner
medium-pressure tubes widely used in the pressure pipeline concave side, the analytical values are by about 27%
industry. smaller than the numerical values. This discrepancy is
As the figure shows, all of the measured wall-thinning believed to be a result of neglecting the displacement of
ratios are smaller than the corresponding analytical values. the stress neutral surface and strain neutral surface.
Despite this quantitative discrepancy, the trends of the (2) The tangential bending stress decreases with the increase
analytical curves agree basically with those of the of R/d. It decreases sharply in the range R/d < 2  2.5.
This coincides with the fact that the R/d ¼ 2  2.5 is the
forming limit for small-diameter-thin-walled tubes.
Furthermore, the analytical results indicate that the
tangential bending stress decreases as t/d increases when
R/d  3.
(3) Wall thinning is a function of the bending radius ratio and
the displacement of the outmost point of the tube wall.
The wall thinning ratio increases with an increase of the
bending radius ratio (R/d) or the flattening rate (wmin).
Actually, the increase in R/d lessens the bending
deformation while the increase in wmin lessens the radial
deformation, both of which will result in a decrease in
wall thinning.

Nomenclature

Figure 5. FEM result of tangential bending stress (in brackets are the er, eu, ew principal strains in the radial, tangential and
analytical values). circumferential directions, referred to as

www.steelresearch-journal.com ß 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1087
steel research int. 81 (2010) No. 12 Metal Forming

radial strain, tangential strain and circum- Fr contact force of the bending wheel
ferential strain, respectively M bending moment
sr, su, sw principal stresses in the radial, tangential and
circumferential directions, referred to as
radial stress, tangential stress and circum-
ferential stress, respectively References
sri, sui, swi radial, tangential and circumferential stresses
in the inner (compressive) side of the tube [1] D. E, Y. Liu: Springback and time-dependent springback of
1Cr18Ni9Ti stainless steel tubes under bending, Materials and
wall, respectively Design, 31 (2010), 1259–1261.
sro, suo, swo radial, tangential and circumferential stresses [2] D. E, M. Chen: Numerical solution of thin-walled tube bending
in the outer (tensile) side of the tube wall, springback with exponential hardening law, Steel Research Inter-
respectively national, 81 (2010), 286–291.
[3] Z. Q. Jiang, H. Yang, M. Zhan, X. D. Xu, G. J. Li: Coupling effects of
s von Mises stress material properties and the bending angle on the springback angle
sy yield strength of a titanium alloy tube during numerically controlled bending,
suN tangential bending stress resulted from axial Materials and Design, 31 (2010), 2001–2010.
[4] G. Y. Zhao, Y. L. Liu, C. S. Dong, H. Yang, X. G. Fan: Analysis of
force wrinkling limit of rotary-draw bending process for thin-walled
sbw pressure between bending wheel and tube rectangular tube, Journal of Materials Processing Technology,
surface 210 (2010), 1224–1231.
[5] J. E. Miller, S. Kyriakides, A. H. Bastard: On bend–stretch forming
R bending radius (curvature radius of the of aluminum extruded tubes I: experiments, International Jour-
neutral surface) nal of Mechanical Sciences, 43 (2001), 1283–1317.
r curvature radius of arbitrary fiber layer in the [6] L. G. Brazier: On the effect of flexure of thin cylindrical shells and
tube wall other thin sections, Proceedings of the Royal Society A, 116
(1927), 104–114.
R/d bending radius ratio [7] J. E. Miller, S. Kyriakides, E. Corona: On bend–stretch forming of
t/d wall thickness ratio aluminum extruded tubes II: analysis, International Journal of
Ro curvature radius of the bent tube at the Mechanical Sciences, 43 (2001), 1319–1338.
[8] S. Kyriakides, E. Corona, J. E. Miller: Effect of yield surface
outmost point evolution on bending induced cross sectional deformation of
Ri curvature radius of the bent tube at the thin–walled sections, International Journal of Plasticity, 20
innermost point (2004), 607–618.
original outer radius of the cross section [9] J. Wang, A. Agarwal: Tube bending under axial force and internal
r
pressure, Journal of Manufacturing Science and Engineering
ro outer radius of the distorted cross section at (Trans. ASME), 128 (2006), 598–605.
the outmost point [10] N. C. Tang: Plastic–deformation analysis in tube bending, Inter-
ri outer radius of the distorted cross section at national Journal of Pressure Vessels and Piping, 77 (2000), 751–
759.
the innermost point [11] C. K. Oh, Y. J. Kim, C. Y. Park: Effects of local wall thinning on
t original wall thickness of the tube net-section limit loads for pipes under combined pressure and
t’ wall thickness after bending bending, Nuclear Engineering and Design, 239 (2009), 261–273.
[12] H. Lee, C. J. van Tyne, D. Field: Finite element bending analysis of
t’min wall thickness after bending at the thinnest oval tubes using rotary draw bender for hydroforming appli-
point cations, Journal of Materials Processing Technology, 168 (2005),
d original outer diameter of the cross section 327–335.
length of major axis of the distorted cross [13] S. P. Keeler, W. A. Backhofen: Plastic instability and fracture in
d’max
sheet stretched over rigid punches, ASM Transactions Quarterly,
section 56 (1964), 25–48.
d’min length of minor axis of the distorted cross [14] G. M. Goodwin: Application of strain analysis to sheet metal forming
section in the press shop, SAE paper, No. 680093, 1968.
[15] T. B. Stoughton: A general forming limit criterion for sheet metal
wmax change rate of the length of major axis of the forming, International Journal of Mechanical Sciences, 42 (2000),
distorted cross section 1–27.
wmin flattening rate (change rate of the length of [16] D. E, R. Ning, Y. Li: Experimental study on the wall thickness
minor axis of the distorted cross section) thinning of the curved tube affected by material property.
Materials Science and Technology, 2 (2008), 200–203.
nj wall-thinning ratio [17] D. E, R. Ning: The study on the minimum relative bending radius
a infinitesimal angle in the tangential direction of tube coreless bending, Chinese Journal of Mechanical Engin-
b correction factor (in the expression of nj) eering, 5 (2007), 219–222.
[18] M. Murata, T. Kuboki, K. Takahashi, M. Goodarzi, Y. Jin: Effect of
C, Ci, Co integral constants hardening exponent on tube bending, Journal of Materials Pro-
Nu axial force (in the tangential direction) cessing Technology, 201 (2008), 189–192.

1088 ß 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.steelresearch-journal.com