Академический Документы
Профессиональный Документы
Культура Документы
Materials
Processing
Technology
ELSEVIER Journal of Materials Processing Technology 48 (1995) 159-172
The continuous four-roll bending of sheet is an efficient process, it produces large and medium
diameter tube sections which are then welded to form a single seamed tube. A particular advantage of
the process is to minimise the remaining straight section of a finished bent sheet, thus achieving better
circularity of the seamed tube. This improvement in the bending process reduces waste material and
the subsequent cost of machining. Unfortunately, the understanding of the bending operational
mechanisms is still in its infancy, and its operation, at the present stage, still depends heavily upon the
skill of operators and "trial and error" methods. This restricts the full exploration of the capacity and the
flexibility of the four-roll bender. Thus, the present research was initiated in order to optimise this
bending process and to remove its dependency upon skilled labour. This paper discusses the principle
mechanisms of the bending process and some preliminary test results in bending HP30 aiuminium
plates of 8mm, 10mm and 12mm thickness for (i) single pass bending and (ii) multipass bending, using
a four-roll model bender. The test results relate the force on the rollers to the inner bend radius of the
bent plate.
continuous four-roll plate bending [5] was 10.09mm, using a specific roller geometry (see
initiated at Aston University, Birmingham, section 3.2.3. Geometrical set-up of the
England. Some preliminary results from bending experimental rig) is reported, and some related
aluminium plates, with thicknesses of 8mm and bending mechanisms are discussed.
/ I
I
I
\
/ \
I
topIroLl
/
plate /
thickness
m I
/
/I':::::::L
I
roll
bottom ~oolt
pre-be nding
active side
roll I ,\
I
L _X ° Xo
I-
3
y
Pro -bend ing o p e r a t i o n s
~dg~ s e t t i n g : 1
bending : Z
Steady-continuous operations:3, 6 . 5 and 6
Side-~ '
r °l i-(~- ~,,~ ---:
o0.S
0.0.~
/ , J
\
'
holding the connecting blocks of the bottom roll permitted access to the end drive to the side
in position and guiding the vertical movement of rolls and also allowed free movement of the
the blocks. There was also a rectangular cut bottom and side rolls. It subsequently provided
section 525mm wide by 180mm high at the top a rigid framework for the model bender unit and
mid-span edge of each end plate. This accommodated levelling of the roll centres
164 M. Hua et aL / Journal of Materials Processing Technology 48 (1995) 159-172
\
,.~S, ,><
/ / / /
/ /
..'-~.
/ / \\ -
/ /
/
m,.." /
@
J
I ,
:$I.-
"I" -t!
i i I
. . . . . . . h~_- -- 1 . . . . . . . .
..... ! ........ t ........ i"
i i¸ li "'~ I
LA
I i} End v i : w
were connected in parallel, and a single d.c. achieve the necessary self-amplification.
Farnell L30B stabiliser power supply was used All the signals were traced by:
to provide a constant bridge voltage. RS (i) a Southern Instruments u.v. chart recorder,
potentiometers were introduced into the circuit (ii) a SE u.v. chart recorder, and (iii) a Solartron
for bridge balancing. The signal from the top A200 d.v.m, with an accuracy of 1 mv.
roll axial Ioadcells was supplied by a Farnell The power stabilisers used were:
d.c. stabiliser. Each bridge circuit for (i) a Farnell L30BT for the CIL strain gauge
thetorquemeters had eight gauges in order to amplifier circuitries, (ii) a Farnell L30B for the
M. Hua et aL / Journal of Materials Processing Technology 48 (1995) 159-172 165
within the allowable envelopes. side rolls and the bottom roll
The rolls consisted of: The vertical translational movement of both
(i) a 90mm dia. top roll, (ii) a 90mm dia. bottom bottom pinch and side rolls was provided by
roll and (iii) two 75mm dia. side rolls. hydraulic circuitry. Each hydraulic cylinder was
All the four rolls had a full roll length of 310mm powered by an Enerpac hydraulic hand pump.
and were made from Uddeholm tool steel(UHB By suitable control of the installed valves the
Arne). The top surfaces were heat-treated to a corresponding side roll No.1 or No.2 could be
hardness of 58/62 HRC and to a depth of 5mm. activated. Levelling of the side roll during the
The mounting of the rolls gave an estimated bending process could be achieved by
deflection of 0.36mm for the top roll, 0.52mm adjusting the corresponding pair of valves,
for the bottom roll at the full load of 500kN, and while the levelling of the bottom pinch roll was
0.43mm for each side roll at the full load of controlled by the opening of the circuit valves.
120kN. Torrington spherical bearings were
used for the bottom pinch and the side rolls. 3.1.3. Motor drives for the model bender
The following is a numbered listing of the The top roll drive line was designed to
essential components of the bending unit transmit a torque not less than 2.45kNm at a
assembly as depicted in Figures 4 (i) and (ii)): minimum speed of 4rev/min. It included a 5.5hp
1 End supporting plates, 2 Top cross beam, 3 Kopp Variator, a HT6R1 gear box, a Desch
Top roll Ioadcells, 4 Holding strips for the end Conax RTW2 safety clutch and a propshaft
supporting plates, 5 Cross holding strips, 6 which was connected to a torquemeter at the
Bearing housing of the top roll, 7 Bottom top roll (Figure 5). Each side roll was also
holding plates for the bending unit, 8 Locking driven by a Kopp Variator, through a propshaft
pins for the top roll Ioadcells, 9 Clamping plates and a torquemeter. The latter transmitted a
for the top beam Ioadcells, 10 The top roll, 11 minimum torque of 1.23kNm at 4rev/min.
The bottom pinch roll, 12 Bearing housing of
the bottom roll, 13 Connecting block of the 3.2 Instrumentation and Calibrations
bottom roll, 14 Base block of the bottom roll 3.2.1. Sensing components and circuitries
hydraulic cylinder, 15 Hydraulic cylinder The axial and bending loads for each roll
retaining blocks for the bottom roll, 16 Locking were measured by Ioadcells, whilst
rings for the bottom roll spherical torquemeters were used to sense the torque
bearings(inner raceway), 17 Torrington 20SF32 induced between the roll and propshaft. The
bearing housing locking rings, 18 The side roll, layout of the sensing components in the model
19 Bearing housing of the side roll, 20 Top bender are shown in Figure 5. The four top load
clamping block of the side roll hydraulic cells (indicated as 1, 2, 3 and 4) are located at
cylinder, 21 Locking plate for the side roll top each corner of the bending unit. The two
clamping block, 22 Connecting blocks for the Ioadcells (A and B) were located to support the
side rolls, 23 Base-block for the side roll bearing housing at either end of the bottom roll,
hydraulic cylinder, 24 Bottom clamping block and the pair of Ioadcells (D and E) and (F and
for the side roll hydraulic cylinder, 25 Locking G) were positioned to support the bearing
rods for the bottom clamping block(side roll housing at either end of side roll 1 and side roll
hydraulic cylinder), 26 Locking plates for the 2 respectively. T t, and Tql and Tq2 are the
side roll bottom clamping block, 27 Torrington torquemeters for the top roll and the two side
17SF28 bearing housing locking ring, 28 rolls, respectively. Additionally, a Servo-
Locking ring for the side roll spherical bearing Tektachometer mounted at the free end of the
(inner raceway, free end), 29 Loadcell for the top roll monitored the roll rotational speed.
bottom pinch roll and the side rolls, 30 Glacier The strain gauge bridge circuitries for the
MB100DU bearing locking ring, 31 Locking ring Ioadcells and torquemeters:
for the side roll spherical bearing (inner The circuitries were designed to eliminate any
raceway, driven end), 32 Loadcell cover for the possible spurious signals. All the strain gauges
bottom roll, 33 Bush bearings for the side roll in each full bridge circuit were 100W resistive
connecting block. filament type and each element had a safe
maximum current of 20mA. All six bending
3.1.2. The hydraulic drive and circuitry of the bridges for the bottom and side roll Ioadcells
166 M. Hua et aL / Journal of Materials Processing Technology 48 (1995) 159-172
.
/
i /0
]
7"
G-...II'I I z~
/ [ n,~rl~ac___l!'fd_r__a.uli_c_.S1}indQr R(~.I01 0
four top roll axial strain gauge bridges, (iii) a calibration. Results obtained were analysed
Farnell L30B for the four top roll bending strain with the aid of a linear regression statistical
gauge bridges, and (iv) a Farnell L30B for the software package. The calibration equations
six bending strain gauge bridges of the bottom and their corresponding coefficient of
pinch and the side roll Ioadcells. determination, the coefficient of correlation, and
the standard error of estimation were recorded.
3.2.2 Calibrations For this particular experiment, the calibrated
Calibration for the Ioadcells were performed range for the Ioadcells was:
using a 1 MN Avery compression testing
machine. The torquemeters were calibrated (a) Axial load:
using an Avery torsion testing machine. The 0- 100 kN for the top and
tachometer was calibrated in situ by noting the bottom roll,
time taken for every twenty revolutions of the 0- 50 kN for the side rolls.
top roll propshaft. During calibration, particular
attention was paid to the proper alignment of (b) Bending load:
load transmission to attenuate any possibly 0- 60 kN for the bottom
induced effects. Repeatability of the calibration and side rolls,
was ensured by performing at least twice, each 0- 30 kN for the top roll.
M. Hua et aL / Journal of Materials Processing Technology 48 (1995) 159-172 167
© ®
I
I
o
I
I
I
I
-I
Motor Motor
t-
o Oear-box
>1
L_
,,ml
~t - II
'-I
I .~ Motor
I
Y
C~n
_.Po.sitional powering centre
ii
0
\,
\_ plane
Pre-aciive I '
-Xbn
I I
i
' 1_ X° _l
I-
Figure 6 Geometrical conaitions for the
single/first pass pre-bend mode
and at the initiation of the
continuous bending mode.
M. Hua et aL / Journal of Materials Processing Technology 48 (1995) 159-172 169
YI .!o
! -b°
/
pre-inactive side roll.
plato /
I /
-1-
I pre-inactiv~ f ~ " °
I-._ t "
side roll /
s,
I Xbn
x-~cn"~tnL~
- Jl
I~',,n _-J
Figure 7 Geometrical condition for the
single/first pass steady continuous
bending mode with the pre-inactive
side roll operating.
170 M. Hua et aL / Journal of Materials Processing Technology 48 (1995) 159-172
n n
= ~J{[T.Fa] 2 + [7_.Fb]
2} (1) Wd = [2NTm/6OVc][La/2 + Ls
i=1 i=1 + Ld/2] (4)
n = 2 for the bottom pinch roll and for each of where, V c is the chart speed of the u.v.
the side rolls, recorder, La, Ls and Ld are the length of the
n = 4 for the top roll, u.v. traces during acceleration, steady state
where Fa and F b are the measured axial and driving and deceleration, respectively (these
bending load, respectively, on each Ioadcell. lengths are in relation to the torque on the top
roll during bending).
Torque on top roll - the measured torques
were (i) the mean torque T m and (ii) the 4. Graphical Presentation of the
maximum torque T m a x . In determining the Experimental Results and Discussions
mean torque it was noticed that its trace on the
u.v. chart was oscillating within a small band 4.1. Properties of the HP30 Aluminium Plate
width. The mean torque Tm was thus evaluated Specimens
from the difference on the u.v. chart between Uniaxial tensile and indentation tests
the mean chart value Tr of the oscillation and indicated that the anisotropy of the HP30
the initially set reference chart value To as the aluminium plate specimens was negligible [5].
However, both tests showed various degrees of
continuously steady state was reached, i.e.
hardening, tempering and a n n e a l i n g for
different thicknesses of the specimens. In
Tm = a(Tr - To ) + b (2) particular, it was observed that the material
hardening constant for thinner specimens was
where a, b are calibration constants; T r , To higher than that of the thicker ones. This
are measured chart widths of the trace in mm. implied that the thinner plates were more
The m a x i m u m torque was defined as the susceptible to work hardening.
highest torque experience in each continuous
bending mode and was evaluated by replacing 4.2. Discussion on the Experimental Results
the reference quantities of Tm and Tr by Tmax in Bending HP30 Aluminium Plates
and Ti respectively, in the calibration equation Since there was a collated total of sixteen
(2). In this case Ti was the chart width u.v. galvanometer signals, some of the output
measurement in mm at the transitional state. signals were enmeshed together and were
difficult to separate, and therefore only those
Rotational speed of the top roll - the top roll results which were distinguishable from the u.v.
speed was evaluated from the difference of the chart are listed. The respective measurements
u.v. chart readings of the corresponding trace were subsequently presented for the following
at the instant of steady rotation and its initial two operative conditions:
datum by the following equation: (i) the pre-bending active (pre-active) side
roll operation and
N = c((or-~o) + d (3) (ii) the pre-bending inactive (pre-inactive)
side roll operation.
where N is the rotational speed in rev/min, (Or, All the data were taken from the u.v. chart at
(0o are the measurements (in mm) from the the steady equilibrium operating conditions.
chart and denotes the steady rotation and the The corresponding data for the m a x i m u m
initial datum respectively, and c, d are the torque were presented on the basis that the top
calibration constants. roll drive experienced and required torque,
otherwise the continuous bending operation
Experimental power - the total work done would subsequently fail.
Wd to complete a bend in the continuous roller
bending process can be evaluated from the 4.3. General Observation from the Preliminary
traces of the u.v. chart using the following Tests.
equation: Preliminary results of the bending force on
172 M. Hua et aL / Journal of Materials Processing Technology 48 (1995) 159-172
the rollers for some initial single and multipass resulted from a changing deformation pattern
bending indicated that the measured force on and bender/plate geometry, as the bending
the top roll was lower than that on the bottom process proceeded.
roll. This contradicts the expected results since,
generally, it was expected that the top roll 6. Acknowledgments
would exert a higher force than the bottom roll.
Initially, it was thought that these results were The financial support of S.E.R.C. and the
due to the fact that the top roll Ioadcells were facilities provided by the Department of
mounted in such a way as to give a 'portal Mechanical and Electrical Engineering, Aston
frame' structure within the bending unit of the University, Birmingham, England are gratefully
model bender. The additive vertical force appreciated.
components from the operative side roll and the
bottom roll would thus generate twisting 7. References
moments in the top beam and the Ioadcell
interfaces. Consequently, the induced twisting [1] J.T. Winshop, American Machinist,127,2,
moments would oppose and cancel the (1983) 105.
expected bending signal, thus affecting the [2] S. Roggendorff and J. Haeusler, Welding
accuracy of the measured load on the top roll. and Metal Fabrication, Jul./Aug. (1979)
In order to establish this assumption, an 353.
estimation of the magnitude of induced twisting [3] M.B. Bassett and W. Johnson, J. Strain
moments was carried out with a load of 250 kN Analysis, 1,5 (1966) 398.
applied vertically at the mid-span of each top [4] NE. Hanson and O. Jannerup, A.S.M.E.
beam. This, in total, gave a maximum of Paper No. 78WA/Prod. (1979) 6,
induced bending force on the top Ioadcells [5] M. Hua, Ph.D. Thesis, University of
within an estimated 12% of the total applied Aston, England, (1986).
load. Observation indicated that the top roll
bending force was 34% lower than that of the
bottom roll at rff = 711mm, for single pass
bending of 8 mm thick x 80 mm wide HP30
aluminium plates. This gave at least 20%
difference from the estimated value of 12%
(taking account of the side roll bending force,
the difference would be even higher than 20%).
Consequently, this analysis seems to suggest
that the observed characteristics for the top roll
could be an inherent bending mechanism of the
process, however an understanding of the
cause and the nature of this mechanism would
need further investigation.
5. Conclusions