JOURNAL OF THEORETICAL

AND APPLIED MECHANICS

52, 4, pp. 1093-1106, Warsaw 2014

GENERALIZED MODEL OF STRAINS DURING BENDING OF METAL

TUBES IN BENDING MACHINES

Zdzisław Śloderbach
Opole University of Technology, Faculty of Engineering and Logistics, Opole, Poland
e-mail: z.sloderbach@po.opole.pl

According to the postulate concerning a local change of the “actual active radius” with
a bending angle in the bent zone, a generalized model of strain during bending of ametal
tube has been derived. The considered tubes should be subjected to bending in tube bending
machines by the method of wrapping on a rotating template and with the use of a lubricated
steel mandrel. The model is represented by three components of strain in the analytical form,
including displacement of the neutral axis. Generalization of the model of metal tubes in the
existing papers consists in including the displacement of the neutral axis and the possibility
of determination of strains at each point along the thickness of the wall of the bent tube
in the bending zone. The derived scheme of strain satisfies initial and boundary kinematic
conditions of the bending process, conditions of continuity and inseparability of strains. The
obtained analytical expressions can be classified as acceptable from the kinematic point of
view.
Keywords: tubes bending, deformations, bending angle, thickness of the wall, neutral axis

1. Introduction

Tube bending (see e.g. Beskin, 1945; Boyle, 1971; Franz, 1961, 1969; Gruner, 1960; Grunow,
1985; Korzemski, 1968, 1971; Li et al., 2006; Pesak, 1953; Śloderbach et al., 1999, 2000, 2002,
2012; Tang, 2000; Wick et al., 2001; Yang and Lin 2004; Zdankiewicz, 1970, 1998; Zhang et al.,
2011; Zhiqiang et al., 2011) as a technological problem appeared in the end of the 19th century
when production of tubes started to an industrial scale. Tubes were delivered mainly to industry
of steam engines and boilers, gas engineering, power engineering, civil engineering. At present,
tubes and elbows are purchased by almost all branches of the industry, and tube bending is a
typical activity in many technological processes in the metal industry. Production of tubes and
elbows is increasing more quickly than production of steel because tubes and elbows are made
also of other materials, e.g. plastics.
At present, tube bending in tube bending machines using the method of wrapping on a rota-
ting template with a lubricated mandrel is the most widely used. Such bending always leads to
formation of thinner walls in the layers subjected to elongation, thickening and wrinkling in the
layers subjected to compression, and deformation (ovalization) of the cross section. Such unfa-
vorable phenomena should be included into the limits of tolerance given in European standards
and recommendations (European Standard, 1993; Zdankiewicz, 1998), as well as regulations of
UDT (Polish Office of Technical Inspection, UDT Conditions, 2003). The acceptable ovalization
of the cross section according to the European Standard (EN 448, 1993) is up to 6%. In this
paper, the author considers only cold bending of metal tubes of the assumed technological wall
thickness s∗ ¬ 0.10 and maximum dext = 160 mm (Śloderbach, 2012), where s∗ = g0 /dext , g0
and dext are the initial thickness and external diameter of the bent tube, respectively. In regu-
lations of UDT (UDT Conditions, 2003), the pressure tubes are assumed as thin-walled, when
s∗w ¬ 0.05, where s∗w = g0 /dint , dint = dext −2g0 , then s∗w = s∗ /(1−2s∗ ). The above assumptions
1094 Z. Śloderbach

follow from the practical results obtained at the Research and Development Institute for Power
Plant Maintenance in Wrocław, Poland (it stopped to exist in the 2006).
In literature, there are no analytical expressions for strains depending on an actual value of
the bending angle, angles determining the position of each point (particle) in the bending zone,
displacement of the neutral axis, bending radius, dimensions of the cross section of the bent tube
and suitable technological-material coefficients k and λi dependent on technological parameters
and tube materials. Tube bending always leads to reduction of the wall in elongated layers and
an increase in tube thickness in layers subjected to compression, ovalization and formation of
corrugation which distort the cross section.
In this paper, the author tries to describe the strain state, understanding the tube bending
as a three-dimensional and heterogeneous aproblem including the influence of displacement of
the neutral axis. Such influence is understood in the following way: the forming field of plastic
strains generates displacement of the neutral axis (in this method, in the direction toward the
layers subjected to compression), and the magnitude of the neutral axis displacement affectts
the distribution of these strains.
From experiments, technological tests and industrial practice (Wick et al., 2001), it appears
that this displacement is only about 5% of the external diameter of the bent tube for thick-
walled tubes s∗ ­ 0.10 bent at relatively large bending radii R > 2dext , and about 25% for
very thin-walled tubes s∗ ¬ 0.01 bent at very small bending radii R ¬ dext . Thus, such a
high displacement of the neutral axis of plastic bending should not be disregarded in analytical
description of the strain state. Generalization of the strain description while bending metal tubes
in bending machines related to the data from other papers (Śloderbach, 1999, 2000) consists in
taking into account the neutral axis displacement.
It is assumed that strains in the tube bending process are identified with plastic strains.
Thus, it appears that plastic strains are of the order of some tens of percent (even to 50%), and
the maximum elastic strains are equal to decimal parts of the percent, and they are neglected
in the description.

2. Geometric-analytical description of the bending process

Fig. 1. Geometrical and dimensional quantities pertaining to tube-bending processes

dext , dint – external and internal diameter of a bent tube,

g0 – initial thickness of the bent tube,
 Generalized model of strains during bending of metal tubes... 1095

gi – actual thickness of the bend in the bending zone (i = 1 for elongated, i = 2 for
compressed layers),
rext , rint – external and internal radius of the bent pipe,
R – bending radius,
R0 – radius of the neutral surface following the bending,
Ri – larger actual radius of the bend associated with longitudinal strain,
y0 – displacement of the neutral surface with respect to the initial position,
α – actual angle of the bending zone determined at the principal bending plane and at
planes parallel to it, α ∈ h0◦ , αb /2i, where αb is the active bending angle measured
over the bending zone, αb ∈ h0◦ , 180◦ i,
α0 – angle of bend (the angle by which a template or a former is rotated); in theory
for spirals α0 ∈ h0◦ , ∞) but for the analyzed method α0 ∈ h0◦ , 180◦ i. Obviously,
within the bending zone the two angles are equal α0 = αb . When the plateau zone
is formed, then α0 = αb + αpl , where αpl is the angle of the plateau zone (Franz,
1961; Śloderbach, 2002). Hence it follow that α0 ¬ αb , becouse αpl ¬ 0, see Franz
(1961),
β – actual angle determined at the planes perpendicular to the bending plane,
β ∈ h0◦ , 90◦ i.

3. The basic assumptions

It is assumed that the tube material is an incompressible rigid-plastic (with isothropic harde-
ning) continuous medium satisfying the condition of plasticity (M-H-H) and the Levy-Mises
flow laws. Its properties while bending are described by two technological-material coefficients k
and λi . Thus, the constant plastic volume of the material is assumed before and after bending.
The neutral axis of plastic bending is determined by the radius R0 . The axis separating the
elongated and compressed layers in the bending zone is a line of stress discontinuity (Hill, 1986;
Johnson and Mellor, 1975; Marciniak, 1971; Mendelson, 1988; Olszak et al., 1985; Szczepinski,
1973; Śloderbach, 1999, 2002; Tang, 2000). It is also assumed that in the tip points of the elon-
gated layers (environment of the mandrel), the tube bending process in bending machines (tube
bending by wrapping on a rotational template and with the use of a lubricated mandrel) is a
complex process of heterogeneous curvilinear elongation (biaxial stretch drawing) under a three
dimensional stress state during bending of thick-walled tubes and a plane state for thin-walled
tubes. In the case of the compressed layers, it is a composition of two processes: heterogeneous
curvilinear compression and unfree upsetting. Introduction of the technological-material coeffi-
cient k allows one to include also the influence of mandrel friction with the tube wall on the
strain distribution.
In real tube bending processes in bending machines, in the compressed layers, there are no
such strain states as those occurring under free compression and upsetting. Tube bending is
not a case of free (unbounded) bending but it is forced by the structure of working tools of the
bending machine (matrix, mandrel, flatter, strip and its pressure force), and the lubrication is
of great importance, too.

4. Procedure of strain component derivation

Derivation of the generalized expressions for three components of the strain state in the consi-
dered tube bending is going to be realized at three stages:
1. Derivation of the expressions for principal strains in relative and next in logarithmic me-
asures while bending when the influence of the neutral axis displacement on the strain
distribution is not taken into account.
1096 Z. Śloderbach

2. Derivation of the expression for the neutral axis displacement in plastic bending y0 .
3. Formal derivation (according to Fig. 4) of the expressions for strain components in order
to describe the deformation state for the problem including the neutral axis displacement
by introduction of the derived relationship for y0 into the equations.

4.1. Derivation of the expression with no neutral axis displacement

The basic aim is analytical derivation of the plastic strain components, i.e. the longitudinal
(along the tube axis), circumferential and radial (along the thickness) components in measures
of relative and logarithmic strains during tube bending in bending machines by wrapping on a
rotating template using a mandrel or with no mandrel, on the assumption that dint ∼ = const .
It concerns thin-walled and thick-walled metallic tubes subjected to cold bending in bending
machines.
An important problem is to determine a form for a change of the “actual active radius” r ∗(N )
in the bending zone (determined at each N -th point along thickness, and for total thickness of
(N )
the bent tube, so as r ∗ = r ∗(N ) for g0 = g0 ), depending on a change of the bending angle αb .
That increase related to length of the bending radius R defines an increment of the relative
(N )
longitudinal strain dε1 . The local “actual active radii r ∗(N ) and r ∗ depend on the actual
value of the angles α and β determining positions of the points in the bending zone, and on
the actual position of the external point in the considered N -th layer included into the wall of
the bent tube (see Fig. 2), also on total dimensions of the cross section in the elongated and
compressed layers.

Fig. 2. A concept of the division of the transversal section of a thick walled pipe for the analytical
and FEM method

In the analytical method, a possibile number of the analyzed points along the thickness of the
bent tube wall is N ∈ h1, n), when n → ∞. In this method, a division along the wall thickness
(N )
is laminar into the N -th number of layers gl0 in thickness, the thickness of which is measured
from the internal tube surface to the external point of the N -th layer (see Figs. 1 and 2), where
(k) (k) (k)
g0 = (dext − dint )/2 = rext − rint .
Such division results from the fact that tube dimensions are given as l × dext × g0 , where l is
(1) (N )
the tube length. When N = 1, then gl0 = g0 , and when N → ∞, then gl0 → 0. In the case
 Generalized model of strains during bending of metal tubes... 1097

of the finite element method (FEM) applied for the thick walled tubes, the division is annular
into a finite number of cylinders (rings) n of equal thickness. From Fig. 3 it also appears that
(1) (1) (1)
when the wall thickness is considered as one layer for N = 1, then gl0 = g0 = g0 , dint = dint
(1) (k) (k)
and dext = dext , where dext = dint + 2g0 and dl int , dext are internal and external diameter of
the k-th layer, respectively, dext , dint are external and internal diameter of the tube for bending,
(k) (1) (k) (n)
respectively, g0 – thickness of the k-th layer measured from dint , gl0 , gl0 , gi0 – thickness of
the 1-st (first), k-th (currently) and n-th (last) layer.
The value of the relative strain ε1 , (i.e. the strain corresponding to the experimental results)
obtained after previous integration of the expression for dε1 within the limits of the angle
h2α, αb i depends on the actual value of the bending angle αb , actual values of the point position
angles α, β in the bending zone, actual dimensions of the bent tube, the bending radius R, and
suitable technological-material coefficients k and λi . Let us describe the strain state assuming
the following postulate.
Postulate A
(N )
The increment of the longitudinal component (along the tube axis) dε1 of the strain state,
according to the notations in Figs. 1 and 2 (tube bending in bending machines by using a
mandrel or with no mandrel but keeping almost the constant internal diameter dint ≈ const ),
for the external point of every N -th layer included into the tube wall is directly connected with
the increment of “the local active actual radius” r ∗(N ) , and the increments of circumferential
(N ) (N )
and radial strains dε2 and dε3 which are presented in the following way
|dr∗(N ) | 
(N ) (N ) 
dε1 =± dε1 = dε1  (N)
|R| g0 =g0
(N ) (N ) (4.1)
(N ) |dri | (N ) dgi
dε2 = (N )
dε3 = (N )
|rext | g0
(N ) (N ) (N )
where dr∗ = dr∗(N ) = (sαb ×dαb /2) cos β for g0 = g0 , because when N = h1i, then g0 = g0
and the sign (+) is related to the elongated layers, and (−) to the compressed ones.
For the division of a thick-walled tube into a finite number of cylinders (rings) for the FEM
method the increment is
 (N )
(N ) |dr∗(N ) | (N )  (N ) |dri |
dε1 =± dε1 = dε1  (N)
dε2 = (N )
|R| g0 =g0 |rext |
(N )
(N ) dgli n
P (N )
dε3 = (N )
gi = gli
gl0 N =1

where gi is the running thickness of the bend within the bending and bend zone.
(N )
From properties of the vector product, it results for g0 = g0 that
cosβ (N ) cosβ  (N ) αb 
 cosβ  (N )  αb 

dε1 = ± |sαb × dαb | = ± ri sin × dαb  = ± ri sin × dαb  (4.2)
2|R| 2|R| 2 2|R| 2
(N )
for g0 = g0 and
αb 1 αb 
(N ) (N ) (N ) 
sαb = sαb = ri sin so dε1 = |ri | cos β sin |dαb |
2 2|R| 2
(N )
Then for g0 = g0
α  (N ) (N )
1 (N ) b dri dgi
dε1 = ± ri cos β sin dαb dε2 = dε3 = (4.3)
R 2 (N )
rext g0
(N )
1098 Z. Śloderbach

(N )
where ri is the local actual radius related to the external point of the N -th layer included
(N ) (N ) (N )
into the tube wall ri = rint + gi and ri ∈ hrint , ri i.
When N = 1 and when the averaged strains are related to the central layer, then
(N )
ri ≡ rim = rint + gi /2, where rim is the averaged small active bending radius related to
(N )
a half of the actual thickness of the bent tube wall, rext is the local external radius related
(N )
to the external point of the N -th layer included into the tube wall, rext ∈ (rint , rext i and
(N ) (N ) (N )
rext = rint + g0 . When N = 1, then rext = rext , where gi is the local actual thickness of
(N ) (N )
the wall of the whole section, g0 and gi are calculation initial thicknesses of the considered
(N )
N -th layer of the bent tube wall, such that g0 ∈ h0, g0 i, and the actual local wall thickness
(N )
for the external point of the N -th layer of the tube section such that gi ∈ h0, gi i.
After integration of expression (4.3)1 in the range of change of the angle ξ from the actual
value of the angle 2α up to αb , i.e. in the range ξ ∈ h2α, αb i, are obtains
( " Zαb (N )
# )
h1  αb i cos β ξ dri 
(N ) 
ε1 = ± ri cos β cos α − cos  (N) + cos dξ  (4.4)
R 2 g0 =g0 R 2 dξ  (N)
2α g0 =g0

As it results from expression (4.4), the longitudinal component of the plastic strain ε1
(expressed in measures of relative strains) contains two terms. The second term estimated from
calculations is negligibly low because its maximum value equals to only some percent as com-
pared with the value of the first term. Thus, this term is disregarded in further considerations.
(N )
Thus for g0 = g0

1 (N ) αb 
ε1 ∼
= ± ri cos β cos α − cos (4.5)
R 2
The expression determining the circumferential and radial components has been obtained
on the assumption that during bending of the tube material particles are moving along the
elbow radius to the center of the bent tube in the elongated layers and from the center in
the compressed layers. The actual and averaged local components of the strain state take the
(N )
following form for g0 = g0

 (N ) (N )
(N ) 1 (N ) αb  (N ) d − dext
ε1 ∼
= ε1 = ± di cos β cos α − cos ε2 ∼
= ε2 = i
2R 2 (N )
dext
(N ) (N )
(4.6)
(N ) gi − g0
ε3 ∼
= ε3 = (N )
g0

The strains should be measurable during experiments performed in order to verify analytical
or numerical calculations. Thus, transformation of Eqs. (4.6) for determination of strains for
the total thickness of the elbow wall (it concerns especially thin-walled tubes) measured on the
external surfaces (external measuring quantities after bending), or measurements of the initial
thickness of the tube which is going to be bent requires the averaging and replacement of the
(N ) (N ) (N ) (N )
values dext from (4.6) by dext , and g0 = g0 , because when N = 1 then g0 = g0 , dext = dext ,
(N )
di = di . Now, it is necessary to derive measures of the logarithmic strains, useful in plastic
work technologies in the case of large strains (Franz, 1961)
h 1  αb i
ϕ1 = ln 1 ± di cos β cos α − cos
2R 2
(4.7)
di dint + 2gi gi
ϕ2 = ln = ln ϕ3 = ln
dext dext g0
 Generalized model of strains during bending of metal tubes... 1099

The expressions for component strains (4.1)-(4.7) describe deformation in bent tubes made
of plastic and incompressible continuous media (Śloderbach, 1999; 2002). Real materials undergo
deformation in another way (especially in the compressed layers), and tube bending in bending
machines is not unbounded upsetting in those layers. There are boundary limitations and forcing
for displacements (especially in the compressed layers of the bent tube) resulting from the
structure of the bending machine and its rigidity. There are also forces of external friction
of the tube with the bending machine tools, and internal friction in the bent tube materials,
and others. From the tests described in literature (Franz, 1961; Gruner, 1960; Grunow, 1985;
Korzemski, 1968, 1971; Wick et al., 2001; Yang and Lin, 2004; Zhang et al., 2011; Zhiqiang
et al., 2011) and the tests performed by the author (Śloderbach, 1999, 2000, 2002) it appears
that expressions (4.1)-(4.7) should be modified. Only the longitudinal component (along the
tube axis) should be modified because values of the circumferential and radial components
(along thickness) result directly from the longitudinal component and the condition of plastic
incompressibility of the material. The longitudinal component expressed in logarithmic measures
of strains (4.7) is modified because the results of experimental measurements are defined in such
measures and determined on the external surfaces of the bent tube (Franz, 1961; Gruner, 1960;
Grunow, 1985; Korzemski, 1968, 1971; Śloderbach, 1999, 2000, 2002).
The modification of expressions (4.7) consists in introduction – according to the experimental
data – of two parameters k and λi (technological-material parameters of the tube bending
process). Thus
h 1  kαb i
ϕ1 = λi ln 1 ± di cos β cos(kα) − cos
2R 2
(4.8)
di dint + 2gi gi
ϕ2 = ln = ln ϕ3 = ln
dext dext g0

where gi is the local actual thickness of the bent elbow wall, di – local actual external diameter
of the elbow in the bending zone: di = 2ri , di = dint + 2gi , λi – technological-material coefficient
of strain distribution in the elongated layers i = 1 and compressed layers i = 2, defined from
experiments, so as λ1 ∼ = 1 and λ2 ∈ h0, 1i. In the most cases of known tests on the tube
bending process using the method considered in this paper it can be assumed that λ2 ≈ 0.5,
k – technological-material coefficient dependent on the bent tube material and the applied
bending technology determining the bending zone range in the bent zone. This coefficient is
defined during experiments, theoretically k ∈ h1, ∞). It seems that in the case of most of
metallic materials it is sufficient when k ∈ h1, 6i. From the recognized tests and calculations it
even appears that k ∈ h1, 3i (see e.g., Franz, 1961; Gruner, 1960; Grunow, 1985; Korzemski,
1968, 1971; Śloderbach, 1999, 2000, 2002). In the case of more ductile, soft, plastic materials bent
at elevated temperatures (hot, semi-hot or preheated bending) or bent with a greater radius R
and with a more fitted expanding mandrel (segment with an adjusted external diameter) with
rich lubrication of the mandrel and the tube interior the coefficient k is lower (tends to unity,
k → 1). Thus, it appears that the coefficient k allows one to include (indirectly and in part)
some effects of friction between the mandrel and the bent tube wall. For elbows bent to 180◦ ,
the coefficient k expresses the ratio of the bending angle α0 to the real value of the bending
angle αb , i.e. k = α0 /αb . When the bending angle is α0 = kαb = 180◦ , for example as in
Franz (1961), Korzemski (1971), Śloderbach (1999, 2002), then k = 180◦ /αb . If α0 = 90◦ , then
2α0 = kαb = 180◦ , when α0 = 60◦ , then 3α0 = kαb = 180◦ , etc.
From the known tests on bending using the method of wrapping and the template and
mandrel, it appears that the coefficient k decreases when the mandrel is well chosen, put forward
and lubricated, and if the applied metals and their alloys are soft and very soft.
The transformed and adapted for analytical calculations of strains in external points of
every N -th layer included into the bent tube wall (especially, it concerns the division into a
1100 Z. Śloderbach

finite number of cylinders (rings) for thick-walled tubes and the FEM method) in the elongated
and compressed layers expressions (4.8) take the form
h 1 (N )  kαb i
(N )
ϕ1 = λi ln 1 ± di cos β cos(kα) − cos
2R 2
di dint + 2gi gi n
X (4.9)
(N ) (N ) (N )
ϕ2 = ln (N )
= ln ϕ3 = ln gi = gli
dext dext g0 N =1

From the assumptions and derived expressions (4.1)-(4.9), it appears that the bending an-
gle αb is the basic parameter determining the advancing of the bending process.

4.2. Derivation of the expression for displacement of the neutral axis

The aim is to derive an extended (for transient zones and for unfree bending) expression
determining displacement of the neutral axis in plastic bending. In the paper by Tang (2000),
the autor derived the following approximate expression for the displacement of the neutral axis,
see Fig. 3
0.42
y0 = rm (4.10)
re

Fig. 3. Schematic picture of the elbow cross-section and its characteristic parameters

The extended expression determining the displacement of the neutral axis resulting from
Section 4.1 in this paper (with no derivation), valid for transient zones and not unbounded
bending, is
0.42  kαb 
y0 ∼
= λ0 rm cos(kα) − cos (4.11)
re 2
where λ0 is the correction coefficient of displacement of the neutral layer, λ0 ∈ h0, 1i, re – relative
radius of bending, re = R/dext , rm – mean radius of the bent tube, rm = rint + g0 /2.
The coefficient λ0 determines characteristic technological-material parameters of the tube
bending process, such as kind of the mandrel, tube material, shape of the template and the
flatter, strip pressure, clearances, forces of friction between the bent tube and the bending
machine rigidity of the bending machine, kind of bending (cold, hot, self-hot, with preheating).
From Eqs. (4.10) and (4.11), it appears that for very small bending radii R ∈ h0.5dext , dext i and
more thin-walled tubes s∗w ≪ 0.05, the maximum displacement of the neutral axis can be equal
to ∼ 25% of the diameter of the tube which is going to be bent. Greater displacements of the
neutral axis can be caused by another bending technology because in the case of the considered
 Generalized model of strains during bending of metal tubes... 1101

ranges re and s∗w tubes are often bent with the use of a force which is opposite to the force
rotating the template so as to obtain a suitable stress distribution in the cross section. From
extended Eq. (4.11), it also appears that the displacement of the neutral axis is influenced not
only by the bending radius and the tube thickness (thin-walled) (see Franz, 1961; Hill, 1986;
Korzemski, 1968, 1971; Śloderbach, 1999, 2002; Wick et al., 2001, Yang and Lin, 2004), but
by suitable technology, bending parameters and the tube material as well. From Eq. (4.11), it
also appears that there are three additional parameters determining the displacement of the
neutral axis and its position in the bending zone: the bending angle and the angle determining
the position of the point in the bending zone and the coefficient k. Thus, if cos(kα) = 1 and
cos(kαb /2) = 0 then y0 = y0 max ∼= λ0 (0.42/re)rm , see Eq. (4.11).
The introduced limitations concerning the tube bending parameters cause that, for example,
the maximum displacement (for instance for R = dext , s∗w = 0.03 and λ0 = 0.5) relative to the
external diameter of the bent tube of the neutral axis is y0 /dext ≈ 10%. However, for some ranges
R and s∗w and bending technologies and tube materials, relationships (4.8) or (4.9) which do not
include the displacement of the neutral axis y0 can be applied in the strain description. Thus,
they are used for precise description of fundamental experiments presented by Franz (1961).
The estimated maximum value y0 can be in practice even lower owing to a suitable selection
and set-up of tooling of the bending machine, removal of clearances, a more plastic material
for the bent tube, application of bending at elevated temperatures, increase of rigidity of the
bending machine and so on. In the compressed layers, the effects resulting from not unbounded
upsetting may be less, they will be more intense along the perimeter of the displacement of the
bent tube material to the sides, upward and along the bent axis. These can cause lower values
of the coefficient λ0 .

4.3. Relationshisps including displacement of the neutral axis

The derived relationships for the strain state determination which describe the problem of
the displacement of the neutral axis in plastic bending (according to Figs. 2-4) and according
to Sections 4.1 and 4.2 by suitable substitution of expression (4.11) determining y0 to modified
Eqs. (4.8) and (4.9) are as follows:
— for elongated layers
h di cos β + 2y0  kαb i
ϕ1 ∼
= λi ln 1 + cos(kα) − cos
2(R − y0 ) 2
(4.12)
d g
ϕ2 ∼ ϕ3 ∼
i i
= ln = ln
dext g0
— for compressed layers
h di cos β − 2y0  kαb i
ϕ1 ∼
= λi ln 1 − cos(kα) − cos
2(R − y0 ) 2
(4.13)
di gi
ϕ2 ∼
= ln ϕ3 ∼= ln
dext g0

The system of equations (4.12), (4.13) is the searched set of expressions for description of
the strain state in the bending process in bending machines (thin-walled metal tubes) with
the method of wrapping on a rotating template and using the mandrel, or without a mandrel,
but when dint ≈ const is kept while bending. The derived equations include the effect of the
displacement of the neutral axis in plastic bending on the deformation field. If in equations (4.12)
and (4.13) we substitute y0 = 0, then the considered problem does not include the influence
of the displacement of the neutral axis in plastic bending on the strain distribution. When,
for example R = dext , s∗w = 0.03 and λ0 = 0.5, then y0 max /dext ≈ 10%, and the calculated
1102 Z. Śloderbach

increments of longitudinal and equivalent strains including the effect of the displacement of the
axis are by ∼ 20% greater as compared with the values obtained for the case which does not
include the axis displacement y0 . It means that for some certain values of the bending radius
R < 1.5dext and for thin-walled tubes s∗w ¬ 0.03, the displacements of the neutral axis should
not be neglected in the analytical description of the strain state. The estimated maximum
value of y0 /dext can be lower when, for example, R ­ 1.5dext and 0.03 ¬ s∗w < 0.125, then
y0 max /dext ≈ 6% and less. This is an additional reason why relationship (4.8) can be used for
analysis and description of experimental results given in the fundamental paper by Franz (1961),
not including the displacement of the neutral axis. Relationships (4.8) used in Śloderbach (1999,
2000) do not include the displacement of the neutral axis y0 , because the bent tube tested by
Franz (1961) was thick-walled, in which s∗w ≈ 0.127 (then s∗ ≈ 0.1), and it was bent at the radius
R∼ = 1.73dext (so R > 1.5dext ). Then for λ0 = 0.5, y0 max ≈ 2.43 mm and y0 max /dext ≈ 5.5%.
Expressions (4.12) and (4.13) can be written in a more compact form as
h di cos β ± 2y0  kαb i
ϕ1 ∼
= λi ln 1 ± cos(kα) − cos
2(R − y0 ) 2
(4.14)
d g
ϕ2 ∼ ϕ3 ∼
i i
= ln = ln
dext g0

According to the assumptions that the derived expressions for strain components in tube
bending processes are identified with plastic strains, we obtain

ϕ1 = ϕp1 ϕ2 = ϕp2 ϕ3 = ϕp3 ϕ(i) = ϕp(i)

(4.15)
ε1 = εp1 ε2 = εp2 ε3 = εp3 ε(i) = εp(i)

5. Initial and boundary conditions for the model of strains

Expressions (4.12)-(4.14) satisfy the following boundary and initial conditions

a) when α = αb /2 = 0, we have the beginning of the bending process (no bending),
b) when α = αb /2 6= 0 – beginning and end of the bending zone,
c) when β1 = 90◦ + β0 and β2 = 90◦ − β0 – position of the layer of zero elongations (the
neutral axis of plastic bending) defined by the radius R0 in the bent zone. Then

Ri = R0 ri = rext gi = g0

and

ϕ1 = ϕ2 = ϕ3 = ε1 = ε2 = ε3 = 0 and also ϕ(i) = ε(i) = 0

d) when kα = β1 = β2 = 0◦ – tip points of the bending zone, and kαb ∈ (0◦ , 180◦ ), then
 kαb 
Ri = R − y0 ± (rint + gi ± y0 ) 1 − cos ri = rint + gi
2

e) when kα = β1 = β2 = 0◦ and kαb = 180◦ , then Ri , and ri , gi reach their extreme values,
i.e. the maximum and minimum (in elongated layers) or the minimum and maximum (in
compressed layers), respectively. It is the condition of initiation of the maximum stra-
ins at that point, and formation of the plateau zone (see, Franz, 1961; Korzemski 1971;
Śloderbach, 1999, 2000, 2002).
 Generalized model of strains during bending of metal tubes... 1103

– Zone of elongated layers

g1 = g1 min R1 = R − y0 + (rint + gmin + y0 ) r1 = rint + g1 min

– Zone of compressed layers

g2 = g2 max R2 = R − y0 − (rint + gmax − y0 ) r2 = rint + g2 max

In this case, the main components of logarithmic and relative strains and their intensities
also reach the extreme values which differ in the zones of elongated and compressed layers,
respectively.
f) when di = dint = dext (internal surface of the bent tube), then – according to expressions
(4.12)-(4.14) – it appears that ϕ2 = 0 ⇒ ε2 = 0.
Note: Derived expressions (4.12)-(4.14) have a physical sense when the conditions for R > y0 or
for R > y0 max are satisfied, and in practice they are always satisfied.

6. Exemplary calculations

This Section presents exemplary calculations of variation of longitudinal strains and wall thick-
ness, including the displacement of the neutral axis of in plastic bending related to the external
diameter of the bent tube from the range from 0% up to the maximum value of 25%. Simulation
calculations have been performed for a metallic (steel) tube of dimensions ∅44.5 mm × 4.5 mm
and s∗w ≈ 0.127 in the main bending plane in elongated layers α = β1 = 0◦ for the bending angle
α0 = kαb = 180◦ . For calculations, the following values of the technological-material coefficients
have been assumed: k ∼ = 3 and λ1 ∼ = 1 and the bending radius R ≈ 77 mm (R ≈ 1.73dext ). The
data for calculations have been taken of Franz (1961). The calculations have been performed
using expressions (4.14) and the condition of plastic incompressibility of the bent tube material.
The calculation results are presented in Figs. 4 and 5.

Fig. 4. Exemplary calculations of longitudinal deformations, depending on the neutral axis

displacement y0 , where dz ≡ dext

7. Conclusions

• The paper contains derivation of the generalized relationships for logarithmic and relative
measures of strains: longitudinal (along the tube axis), circumferential and along thickness
(radial) during bending of thin- and thick-walled metallic tubes in bending machines.
Generalization of the strain description as compared to the previous papers by the author
1104 Z. Śloderbach

Fig. 5. Exemplary calculations of the wall thickness distribution as depending on the neutral axis
displacement y0 , where dz ≡ dext

(Śloderbach, 1999; Śloderbach and Rechul, 2000) consists in including the displacement of
the neutral axis. The strains can be defined in the main bending plane and each parallel
or perpendicular plane, i.e. in all points of the bending zone. The derived relationships
describing the measures of logarithmic and relative strains depend on the bending radius R,
geometrical dimensions of the tube, bending angle αb , angular coordinates α and β,
which describe the bending zone in the range of the bending angle kαb ∈ h0◦ , 180◦ i,
displacement of the neutral axis y0 and two technological-material coefficients k and λi .
The results of exemplary calculations of longitudinal strains and wall thickness distribution
for the elongated layers performed for the bending angle α0 = kαb = 180◦ including the
displacement of the neutral axis related to the external diameter from the range from 0%
up to the maximum value of 25% are shown in form of suitable curves. The calculations
include the value of the coefficient of the bending zone range k = 3 and the coefficient of
strain distribution in the elongated layers λ1 = 1. From the obtained graphs, it appears
that there is a certain proportionality between the values of displacement of the neutral
axis expressed in % and the relative increment of the longitudinal strain as well as reduction
of the elbow wall thickness.
• In future tests an explicit (analytical) form of k depending on the coefficient of friction
and suitable technological-material parameters of bending can be searched. When the
coefficient of friction between the mandrel and the internal wall of the bent tube tends
to infinity, then the coefficient k tends to infinity, too. It means no bending because the
angular range of the bending zone tends to zero αb → 0◦ . Then, also the angle of bend
zone tends to zero α0 → 0◦ .
• In the considerations on the problem of taking into accout the displacement of the neutral
axis in plastic bending, some simplifications can be introduced, for example the neglecting
of y0 in the first or second term in the numerator in expressions (4.12)1 and (4.13)1 or the
numerators or denominators of expressions (4.12)1 and (4.13)1 in order to obtain the best
conformity of the calculated quantities with the experimental data or those taken from
literature. Such simplifications should be dependent on the bending parameters occurring
in Eqs. (4.10) and (4.11).
• Śloderbach (1999) presented the results of the analytical and numerical calculations based
on derived relationships (4.12)-(4.14). The results of calculations coincide with the experi-
mental data of Franz (1961) for a tube of dimensions ∅44.5 mm × 4.5 mm and s∗w ≈ 0.127,
the bending radius R ≈ 77 mm (R ≈ 1.73dext ). Corresponding calculations results have
been shown in form of appropriate graphs.
 Generalized model of strains during bending of metal tubes... 1105

References

1. Beskin L., 1945, Bending of thin curved tubes, Journal of Applied Mechanics, Transactions of the
ASME, 12
2. Boyle M., 1971, Bending thin wall stainless tubing, Machinery, 77, 71
3. European Standard, 1993, Energetyka, Wytwarzanie, 448
4. Franz W.D., 1961, Das Kalt-Biegen von Rohren, Springer-Verlag, Berlin
5. Franz W.D., 1969, Numerisch gesteuerte Rohrkaltbiegemaschinen, Werkstatt und Betrieb, 9, 69,
129-145
6. Gruner P., 1960, Über Rohrbiegeverfahren, Maschinenmarkt, 30/31, 120-129
7. Grunow O., 1985, Praktisches Rohrbiegen, Springer-Verlag, Berlin
8. Hill R., 1986, Mathematical Theory of Plasticity, At the Clarendon Press, London-Oxford
9. Johnson W., Mellor P.B., 1975, Engineering Plasticity, van Nostrand Reinhold Company,
London
10. Korzemski J.W., 1968, Thin-walled pipe bending with use of mandrels, Mechanik, 4, 68, 207-210
11. Korzemski J.W., 1971, Bending of Thin-Walled Pipes (in Polish), WNT, Warszawa
12. Li H., Yang H., Zhan M., Gu R.J., 2006, A new method to accurately obtain wrinkling limit
diagram in NC bending process of thin-walled tube with large diameter under different loading
paths, Journal of Materials Processing Technology, 177, 192-196
13. Marciniak Z., 1971, Limit Deformations in Sheet Metal Stamping (in Polish), WNT, Warszawa
14. Mendelson A., 1988, Plasticity-Theory and Applications, Mc Millan Company, New York
15. Olszak W., Perzyna P., Sawczuk A., 1985, Theory of Plasticity (in Polish), PWN, Warszawa
16. Pesak F., 1953, Bending thin wall tubing, Machinery, 60, 147-151
17. Szczepiński W., 1973, Theory of Plastic Working of Metals, PWN, Warszawa
18. Śloderbach Z., 1999, A model for strain geometry evaluation in pipe bending processes, Engi-
neering Transactions, 47, 1, 3-20
19. Śloderbach Z., 2002, Some Problems of Mechanics in Pipeline Bending Processes (in Polish),
Publishing of Wrocław University of Technology, Wrocław, ISBN 83-7085-665-9
20. Śloderbach Z., 2012, Application of the new flexible multi-segment mandrel for bending metallic
tubes on machine benders (in Polish), Energetyka, Zeszyt Tematyczny nr XXIV, 97-99
21. Śloderbach Z., Rechul Z., 2000, Effect of strain hardening and normal anisotropy on admissible
values of strain and stress in pipe-bending processes, Journal of Theoretical and Applied Mechanics,
38, 4, 843-859
22. Tang N.C., 2000, Plastic-deformation analysis in tube bending, International Journal of Pressure
Vessels and Piping, 77, 12, 751-759
23. UDT Conditions, WUDT-UC-WO-O/02:10, 2003, Pressure Installations. General Requirements.
Strength Calculations (in Polish), Issue I, Warszawa
24. Wick Ch., Benedict J.T., Veilleux R.F., 2001, Tool and Manufacturing Engineers Handbook.
A Reference Book for Manufacturing Engineers, Managers and Technicians, Volume II, Forming,
Fourth Edition, Society of Manufacturing Engineers, One SME Drive, Dearbon, Michigan, USA
25. Yang H., Lin Y., 2004, Wrinkling analysis for forming limit of tube bending processes, Journal
of Materials Processing Technology, 152, 363-369
26. Zdankiewicz M., 1970, Investigations of the cold bending processes of tubes (in Polish), Mechanik,
7/70, Warszawa
1106 Z. Śloderbach

27. Zdankiewicz M., 1998, European directive concern in pressure installations. Requirements for
the manufacturing (in Polish), Dozór Techniczny, 98, 2, 25-33, 48
28. Zhang Z., Yang H., Li H., Ren N., Tian Y., 2011, Bending behaviors of large diameter thin-
walled CP-Ti tube in rotary draw bending, Progress in Natural Science: Materials International,
21, 401-412
29. Zhiqiang J., Mei Z., He Y., Xudong X., Guangjun L., 2011, Deformation behavior of
medium-strength TA18 high-pressure tubes during nc bending with different bending radii, Chinese
Journal of Aeronautics, 24, 657-664

Manuscript received November 26, 2013; accepted for print June 10, 2014