A Tool Deflection Compensation

System for End Milling Accuracy

M. Y. Yang
Professor.
Improvement
Mem. ASME
In an effort to reduce machining surface errors due to tool deflection in the end
milling process, methods regulating cutting forces have been implemented with on-
J. G. Choi line feed rate controls. Such schemes are able to improve the parts dimensional
Graduate Student. accuracy, but unfortunately they can exhibit undesirable aspects in which the allevia-
tion of the cutting conditions deteriorates the productivity. In addition the frequent
changes of the feed rate would spoil the surface quality. As a new approach to
Department of Mechanical Engineering,
Korea Advanced Institute of Science
achieve the precision machining efficiently, this paper introduces a tool deflection
and Technology compensation system. This compensation system is a computer controlled special tool
Taejon, 305-701, Republic of Korea adapter which is capable of measuring the cutting forces and minutely adjusting the
position of the tool without interfacing with the NC controller of the milling machine.
Such a system allows for on-line estimation of the tool deflections and reduction
of the surface errors. Experimental investigations for typical shaped workpieces
representing various end milling situations are performed to verify the ability of the
system to suppress the surface errors due to tool deflections in more productive
machining condition.

1 Introduction 2 Fundamentals of Tool Deflection Compensation

In the metal removal process, end milling is still one of most
common procedures used today. In recent years, due to the need
in improving the quality of parts, there has been a push toward
eliminating the machining errors commonly generated in the
end milling process. The machining errors that generally occur
in the end milling process can originate from a number of
sources, such as tool deflection, tool wear, tool run-out, and
chatter vibration. Of these, tool deflection is the biggest problem
that stands in the way of precision machining (Sutherland et
al, 1986; Matsubara et al, 1987, 1991).
A number of solutions have been suggested in reducing the
machining error caused by the tool deflection. One of the most
common methods of reducing this machining error is to adjust
the feed rate, which can be set by a constrained cutting force
or machining error (Sim and Yang, 1993; Yazar et al., 1994;
Tarng and Cheng, 1993; Altintas, 1994). A disadvantage of
this method is that, by reducing the feed rate, the tool may be
operating at a level far below its potential thus making this
method inefficient. Also frequent changes of the feed rate may
result in a lower surface quality.
In contrast to this approach, Watanabe and Iwai (1983) sug-
gested an adaptive control system in which the tool position is
shifted as a means of reducing the surface error. The shifting
of the tool is achieved by a specially designed NC controller
and is effective in reducing the surface error.
In this paper, a new tool deflection compensation system that
can evaluate the machining errors and reduce mem is presented.
First the machining errors are estimated from the cutting forces
measured by a devised built-in sensor. Then to reduce the ma-
chining error, tool deflection is compensated by tool tilting.
This proposed system operates on line and works independent
of the NC controller of the milling machine. By using this
system, it is possible to obtain improved machining accuracy
and higher efficiency simultaneously while maximizing the cut-
ting capability of the end mill.
Contributed by the Manufacturing Engineering Division for publication in the
JOURNAL OF MANUFACTURING SCIENCE AND ENGINEERING. Manuscript received
June 1995; revised Nov. 1996. Associate Technical Editor: M. Elbestawi.
System
2.1 Surface Error by Tool Deflection. In the end milling
process, the lateral cutting force exerted on the tool causes the
tool to deflect because of the lack of stiffness and the cantilever
form of the tool. The instantaneous tool deflection at a given
angular position is affected by not only the magnitude of the
instantaneous cutting force, but its distribution along the tool
axis. The cutting force in the end milling can be analyzed by
considering the chip load on the cutting edges, with the uncut
chip thickness being expressed in Eq. (1).
tc = f • sin a (1)
Figure 1 shows that, while an end mill with equally spaced
helical flutes rotates, the chip loads change according to its
angular position. The variation of the chip loads induces the
change of the magnitude and distribution of the cutting force
and thus changes the force center, which is defined to be the
location of application of the total point force necessary to
produce the same bending moment about the fixed end of the
tool as the distributed force. Accordingly, the tool deflection at
each angular position also varies along with rotation of the tool.
On the other hand, when cutting the surface of a given axial
depth of cut {da) with an end mill of helical flutes, in order to
produce a surface profile the tool needs to be rotated by the
angle given by Eq. (2), in which the surface profile is the cross
section perpendicular to the feed motion of the tool. During the
rotation of the end mill, a flute sweeps from the bottom to the
top of the machining surface and produces a new surface profile
corresponding to the given axial depth. Since the tool deflection
is continuously changing during the rotation, the surface profile
does not coincide with an deflected tool shape, i.e. an instanta-
neous tool deflection generates only one point on the surface
profile.
• tan ah (2)
R
As an example of the machined surfaces, Fig. 2 represents
the surface profiles generated with the three kinds of radial

222 / Vol. 120, MAY 1998 Transactions of the ASME

Copyright © 1998 by ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27323/ on 05/02/2017 Terms of Use: http://www.asme.o

 TOP VIEW
OF A CROSS SECTION EXPANDED
OF AN END MILL TOOL

x. SURFACE NORMAL
" ^ RESULTANT FORCES[N]

«55. *
^
-&

Fig. 1 Chip load distributions and cutting force changes in the cutting with an end mill of 6 mm diameter
having 4 flutes of 30 degree helix angle in f = 0.03 mm/flute, d„ = 10 mm, d r = 2 mm and down cut

depth of cuts with a constant axial depth of cut. This figure

describes that though the waved shapes of these surface profiles
D dr=2.0mm change according to the variation of radial depth of cut, the
A dr = 1.0 mm surface profiles include each shifted component, as compared
with the objective surfaces that would be produced by a com-
O dr =0.5 mm
E pletely rigid machining system. This implies that the surface
error can be reduced by adjusting the relative positions of the
tool with respect to the workpieces. Here, as a means for posi-
CL tioning the tool, a tool tilting method is introduced.
UJ
o
2.2 Tool Deflection Compensation. A small displace-
i
<: ment at the end of the tool can be generated with a small angular
x displacement of the tool with respect to the center point, which
Y^ « - ® <c is separated adequately from the end of the tool. Consider the
case that the distance from the tilting center to the end of tool
is 100 mm, the tilting angle needed for the end of the tool to
traverse 180 /urn becomes approximately 0.1 degree. This value
^ »•! -: o in the traverse movement of the tool, as can be seen in Fig. 2,
o o o
nearly corresponds to the surface error in the case of a radial
ERROR [mm] depth of cut of 2 mm and an axial depth of cut of 10 mm. Here,
the difference of the radial depth of cut at the top and the bottom
Fig. 2 Machined surface profiles in three radial depth of cuts with an of the surface by the inclination of tool is 18 fim, which is
end mill of 6 mm diameter having 4 flutes of 30 degree helix angle in
0.12 mm/rev feed rate, 600 rpm, down cut and AI2014-T6 work material exactly one tenth of the movement of its end. As noticed in the
cutting example shown in Fig. 2, in the cutting condition where

Nomenclature
a, b = empirical constants in the
minimum surface error equa-
tion
Ax, Ay = amount of adjustment of the
tool position in X and Y direc-
tion
da = axial depth of cut
dr = radial depth of cut
dF,, dFr = elemental tangential and ra-
dial forces on the end mill
dFx, dFy = elemental forces applied to
the end mill in X and Y direc-
tion
Ey,mi„ = surface error estimated in Y
direction
e = control objective error in the
control loop
/ = feed rate [mm/flute]
/, = sampling frequency for dig-
itizing of the sensor signal
Fx,m, FyM = mean cutting forces during
one revolution in X and Y di-
rection
Kp, Ki = proportional and integral
control gains
R = radius of the end mill
tc = instantaneous uncut chip
thickness
u = controller output to drive the
actuator
a = cutting edge immersion
angle measured from the
tool exit point
aen = tool engagement angle mea-
sured from the tool exit point
ah = helix angle of the end mill
/3 = tool rotation angle for
sweeping from the bottom
to the top of the machined
surface by a flute
AFX, AFy = fluctuating cutting force
components during one rev-
olution in X and Y direction
6 = apparent surface error
which means the surface er-
ror by the tool deflection
without any correction
(j> — tool rotation angle mea-
sured from the point in
which the machined surface
coincides with the bottom
of a reference flute

Journal of Manufacturing Science and Engineering MAY 1998, Vol. 120 / 223

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27323/ on 05/02/2017 Terms of Use: http://www.asme.o

 Stepping Motor
Cam & Com
Follower
Compliance
I:
i
16 Strain 2-Axis
Gauges Forces
X Strain
Amplifier
2-Axes
Tilting
A/0
Converter
C=tP
,,
Tool
Deflection Proximity Sensor
fy Workpiece
Fig. 3 Configuration for the tool deflection compensation system
the error at the end of the tool is generated as much as 180 fiva,
the error at the bottom of the machined surface is more than
18 ,um that of the top. Furthermore, in a cutting condition, the
smaller the difference of the errors in the two positions, the less
tilting is needed. Therefore, adjusting the position of the tool
by tilting is very effective in controlling the surface errors.
2.3 System Configuration. In order to implement this
scheme in reducing machining errors through the adjustment of
tool position, the tool deflection compensation system as shown
in Fig. 3 was composed. This system is a computer integrated
tool adapter including a force sensor and a tool tilting mecha-
nism. The systematic procedure for the compensation of the
tool deflection is as follows. The cutting forces generated in the
end milling process are measured by the biaxial force sensor
enclosing the tool holder. The measured signals are then fed
into the computer after a low pass filtering and A/D conversion.
The angular positions of a flute on the end mill are monitored
by a stationary proximity sensor located near the rotating trace
of the protrusion on the tool holder. This information enables
a point on the cutting force signal to synchronize with the
angular position of a flute. With the aid of this information, the
computer can now estimate the tool deflection and the machin-
ing error. Using these results, it can control the tool position
and thus minimize the expected surface error by sending out a
tilting command through the interfacing board. This tilting ac-
tion by the motor allows the tool to have a different radial depth
of cut than before, thus a different tool deflection. The system
then makes a loop, and again the process is repeated.
2.4 Control Algorithm. As described in the preceding
section, though the tool deflection changes continuously during
the sweeping of the machined surface by a flute and the surface
error profiles have various shapes according to the given cutting
conditions, these surface errors could be improved with the
revised tool positions corresponding to the errors. Based on
this concept, a control plan for the compensation system was
established, in which a control action is taken every couple of
turns of the end mill. For example, at feed rate of 72 mm/min
and rotating speed of 600 rpm, the tool moves about 0.2-0.3
mm along the feed direction. This method is expected to adapt
to the variation of the cutting loads according to the shape of
the workpiece and prevent the machining surface from being
damaged with disturbances of the system due to the uneven
cutting forces.
The block diagram of the control algorithm for the compensa-
tion system is shown in Fig. 4. The objective of this control
224 / Vol. 120, MAY 1998
Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27323/ on 05/02/2017 Terms of Use: http://www.asme.o
Pulse
Generator
Force and Surface
Error Relationship
Estimation of Decision for
Surface Error Tilting Action
COMPUTER
flow is to achieve zero error value, e(z), in the presence of tool
deflection. Here, a digital PI controller as shown in the following
Eq. (3) was implemented on the computer:
Ki
u(z) = | Kp + •e(z) (3)
1
where u{z) is the controller output for driving the tilting actua-
tor, e{z) is the deviation indicating the surface error, and Kp
and Ki are the proportional and integral gains, respectively. In
the control flow, to estimate the deviation e(z) from the refer-
ence of zero, a suitable value about the surface error is needed.
As a suitable value, the minimum surface error, which is defined
to be the minimum deviation from the surface to be produced
without tool deflection, is selected. This is because the compen-
sation relying on this value could be expected to be effective
for obtaining a satisfactory machined surface without any over-
cut due to excessive compensation.
2.5 Estimation of Minimum Surface Error. In this kind
of control problem, i.e. geometrical adaptive control, in order
to avoid the difficulty due to the direct sensing of the dimen-
sional error, the estimation of the control parameter using a
system model is frequently used. In the end milling, it is difficult
to know the exact cutting size, like radial depth and axial depth
of cuts, prior to the real cutting. This is due to the lack of the
information about the detailed shape of the raw workpiece. Thus
it is not possible to calculate the machining surface error in a
manner using the mechanistic cutting model without knowing
the cutting size (Sutherland and Devor, 1986; Smith and Tlusty,
1991). Though the estimation of depth of cuts based upon some
measured cutting parameters has been studied by Altintas and
Yellowley (1987), Tarn and Tomizuka (1989), etc., it is still
not enough to implement in this system, and also the estimation
dr
- CUTTING Force
CONTROLLER Z.0.H PROCESS
+
ERROR ESTIMATION
ALGORITHM
Fig. 4 System block diagram
Transactions of the ASME
 Fig. 5 Estimation of the minimum surface errors by cutting forces in a computer simulation. Applied cutting
conditions are the same with those in Fig. 2.

is time consuming process. Therefore as an alternative, the sys- of cut. The experiments were performed on a vertical type CNC
tem predicts the control parameter, that is, the minimum surface milling machine equipped with a Heidenhain TNC 155 control
error, based on the cutting force which is relatively easy to unit. High speed steel end mills of 6 mm diameter with 4 flutes
measure compared with measuring the error directly. Figure 5 of 30 degree helix angle were used. The end mills in all tests
shows the result of the computer simulation representing the were held by a collet, and the effective tool lengths, unsup-
relationship between the errors and cutting forces, where the ported, were 30 mm. Coolant was not used. The work material
mechanistic cutting force model by Kline et al. (1982) and the for the tests was A12014-T6. Following the machining, for the
tool deflection evaluation algorithm evolved from Budak and measurements of the surface errors, a Heidenhain MT12 digital
Altintas (1994) were applied. In part (a) of Fig. 5, the mean prove having the resolution of 0.1 /im and a probe of 0.5 mm
cutting forces, Fym, linearly increase according to the increment in diameter was used with a computer interfaced data acquisi-
of depth of cuts, while part {b) shows that the minimum surface tion. From preceding experiments, the empirical constants a and
error, EyMn, does not increase like Fyjn. This is due to the large b were determined as follows:
fluctuation of the cutting force, as shown in part ( c ) , compelling
the minimum surface errors to be further from their average a = 1.05 X 10"
values, especially in the region of a small cutting load. Thus,
it is considered that the fluctuation of the cutting force becomes b = 0.45 X 10" (5)
a correction factor in the estimation of the minimum surface Figures 7 and 8 are the results of the tests, in which the
error from the mean cutting force. Based upon this, an empirical top figures present the measured cutting forces as well as the
equation is suggested to evaluate the minimum surface error estimated minimum surface errors, and the bottom figures com-
from the measured cutting force. This is shown in following pare the estimated minimum surface errors with the measured
Eq. (4). ones. Figure 7 shows that the mean cutting forces linearly in-
crease according to increment of the radial depth of cuts. The
F • — n'F — h' AF (4) estimated minimum surface errors also increase in the same
manner as the cutting forces, but it shows a rapid increase in
Here, the empirical constants a and b are affected by mechan-
the area of the end of the cutting. This can also be noticed in
ical properties of the workpiece, the tool and the feed rate, but
Fig. 5. The measurement of the machined surface shows that
all of them are factors that can be fixed in the operating the
these are in good agreement with the estimated results. In Fig.
system, and be valid in the change of cutting condition in dr
8, while the mean cutting forces also increase in proportion to
and da. Accordingly, if these are fixed, the empirical constants
the axial depth of cuts, the estimated minimum surface errors
can be determined by the least square fitting method using
increase more rapidly in the region of the relatively small axial
the force and error data obtained from cutting experiments in
depth of cuts, and then nearly level. In this leveled region,
ordinary cutting conditions. Figure (d) shows the result of the
estimation by Eq. (4) and implies this is a good approximation
in comparing it with (b). It is also important to note that the
linear form of this equation suggests a good feature, in which,
even in the case that the normal direction of the machining
surface does not coincide with the principal direction of the
measured force, the 2-dimensional compensation using this
equation is capable of being applied effectively to compensate
the tool deflection in any lateral direction.
In order to prove the validity of this empirical equation, ex-
periments have been executed for the specimens shown in Fig.
6. The specimen (a) is the case where the radial depth of cut
is linearly increased at a constant axial depth of cut. In specimen Fig. 6 Specimens for estimation tests of the minimum surface errors
(£>), the axial depth of cut is changed at a constant radial depth (AI2014-T6)

Journal of Manufacturing Science and Engineering MAY 1998, Vol. 120 / 225

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27323/ on 05/02/2017 Terms of Use: http://www.asme.o


0 10 30 40 50 60
CUTTING LENGTH [mm]
CUTTING LENGTH [mml
Fig. 7 Estimation of the minimum surface errors using the measured
cutting forces for specimen (a) in Fig. 6
although mean cutting forces are increased continuously with
the increasing of the axial depth, due to the rapid increase of
fluctuation of the forces, which is a negative factor in estimation
of the minimum surface error in Eq. (4), the errors do not
increase any more. The comparison between the measured and
the estimated surface errors confirms accuracy of the equation.
Therefore, the suggested equation could be applied to predict
the minimum surface error depending on the measured forces.
3 System Hardware Design
3.1 Force Sensor. The force sensor used to measure the
lateral force acting on the tool was devised as an adapter built-
in type for compactness and easy application. The sensor was
based on the principle of the deformation of the parallel plate
and the application of strain gauges since this type has flexibility
in design and is superior in long term stability. Figure 9 depicts
the compliance of the sensor, in which the basic structure is
capable of detecting bi-axial lateral forces without the coupling
of the signal, which would be caused by bendings, torsions and
thermal deformations. With this design, the tool holder is set
up through the hole of the compliance, and the forces acting
on the tool are transferred to the compliance via the pre-loaded
angular contact ball bearings. Therefore, this structure allows
the measurement of the forces on the rotating parts at a station-
CUTTING LENGTH [mm]
£
a.
CUTTING LENGTH [mm]
Fig. 8 Estimation of the minimum surface errors using the measured
cutting forces for specimen (b) in Fig. 6
226 / Vol. 120, MAY 1998
Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27323/ on 05/02/2017 Terms of Use: http://www.asme.o
Angular Contact
Ball Bearing
Compliance
Strain Gauge
Angular Contact
Ball Bearing
Fig. 9 Compliance for adapter built-in type force sensor
ary point. To measure the deformation of the compliance due
to the lateral cutting forces, a total of 16 strain gauges of which
there are 4 to every 4 wings of the compliance were applied
with the Wheatstone bridge circuit, two to every arm.
The results of calibration for the completed sensor are shown
in Fig. 10. The primary output of the sensor, which means the
output from the principal sensible direction coinciding with the
acting force in the design view point, has a linear relation with
applied load. The secondary output, which is perpendicular to
the direction of the acting force, were minimal. Also, to investi-
gate the dynamic characteristic of the devised sensor, an impulse
response from the sensor system including the compliance, the
tool holder and a tool is measured in the circumferential direc-
tion of the tool holder with an acceleration pickup. As the result
of the test, it is revealed that the compliance of the measurement
system has its first natural frequencies of 574 Hz and 612 Hz
in the two principal directions, respectively. Considering a flute
passing frequency of 20-80 Hz at an ordinary tool rotating
speed of 300-1200 rpm, the dynamic characteristics of the
devised sensor are expected to be sufficient for measuring the
cutting force. Also, to examine its performance in real cutting,
a test comparing directly it with a commercial sensor had been
carried out. The commercial sensor used in this comparative
tests was a 3-axes Kistler model 9257B piezoelectric dynamom-
eter with a model 5007 charge amplifier. Figure 11 shows that
there is no significant difference between the devised and the
commercial sensor. Hence, a linear equation could be used for
converting the sensor signal to the cutting force.
3.2 Tool Tilting Device. In the previous section, as a
means for adjusting the tool position, the tilting method, which
makes an angular displacement to the tool instead of a direct
translation, was suggested. Here, a tilting mechanism to imple-
Transactions of the ASME
 ° Yprimocy
° Xsecondary

y
0 200 -100 0 100 200
LOAD [Ml L0AD[N]

Fig. 10 Static calibration of the devised sensor (The plots for the unloading tests only
slightly differ from the loaded test. To avoid confusion they have been omitted.)

.Ydevised

Ycommercial

X commercial

1000
NO. OF SAMPLE
Fig. 11 Dynamic comparison of the devised sensor to a commercial dynamometer {fs •-
13.6 KHz, end mill of 2 flutes at 1200 rpm)

ment this idea was devised. Figure 12 represents the detailed the initial experiments. Three types of specimens that represent
drawing of the tilting mechanism. Referring to this figure, the the shapes of workpieces commonly encountered in the end
tilting action of the tool is achieved as followings. The rotation milling were prepared as shown in Fig. 14. In the figure, speci-
of the stepping motor is converted into the linear motion of the men ( a ) , with a linearly increased radial depth of cut at a
tilting rod by the cam drive mechanism. At the same time, the constant axial depth, was used to test the adaptability of the
driving force is amplified enough to operate against the cutting
force, and also, the movement becomes more precise to obtain
the required high resolution. The linear displacement of the SCHEMATIC DIAGRAM
tilting rod is transformed into the angular displacement through
the spherical bearing. Then the tool holder, which is located in Tilting Com
the center of the spherical bearing and is connected with the
spindle of the machine tools using a flexible coupling, obtains Roller Follower
the tilting motion for the proper adjustment of the tool position.
Since two sets of the tilting driver are actually located perpen- Holder Shaft h
dicular to each other in the circumference of the spherical bear-
ing, the angular motions as above in orthogonal direction make
the system capable of adjusting the tool position in any direc-
tion. By the aid of this mechanism, in the case of using a tool Force Sensor
Compliance
having an effective length of 30 mm, the resolution for the one
step (0.9 degree) of the stepping motor is 4 yum, and the range
of moving to one side is 0.8 mm. Tilting Rod
Spherical Bearing
End mi
4 Compensation Experiments Spherical Joint
The tool deflection compensation system was completed as Guide Pin
a computer integrated tool adapter based on the driving compo-
nents and the operating algorithms explained earlier. In order 11/ Tool Holder
to verify the ability of the system for suppressing the generation
of surface errors by the tool deflection, experiments using this
system have been carried out. Figure 13 shows a typical machin- MOTION TRANSFER ROUTE
ing situation using the system. Stepping Motor - > Tilting Cam - > Roller Follower
In these experiments, the tool and equipment sets in the test - > Tilting Lever - > Spherical Bearing - > Compliance
of Sec. 2.5 were used. A feed rate of 72 mm/min and spindle - > Bearing - > Tool Holder - > Endmill
speed of 600 rpm were applied. The control gains of Kp and £,
appearing in Eq. (3) were set to 0.15 and 1.0 respectively in Fig. 12 Tool tilting mechanism

Journal of Manufacturing Science and Engineering MAY 1998, Vol. 120 / 227

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27323/ on 05/02/2017 Terms of Use: http://www.asme.o

 Fig. 13 The computer integrated tool adapter and a typical machining Fig. 14 Specimens for tool deflection compensation tests
using the system

machined surfaces reveals that the errors over the circumference

system for the gradual increments of the cutting load. Specimen
(b), with a stepped radial depth of cut while maintaining an
axial depth cut at a constant depth, was prepared to check the
behavior of the system for the stationary and the abrupt changes
of the cutting load. Specimen (c), which is a cylinder block,
was subjected to a cutting load of a constant magnitude while
its direction continuously changed. This specimen was selected
to examine how well the system could handle the changes in
all directions of the tool deflections.
Figures 15, 16 and 17 show the results of the experiments
for specimens (a), (b) and (c) in Fig. 14, respectively. In these,
the top figures represent the measured cutting forces and the
system outputs for the compensation of the generated tool de-
flections. The middle figures in (a) and (b) depict the 3-dimen-
sional error map of the measured errors respectively. The lower
figures describe the measured minimum surface errors, in which
the circle connected lines indicate the case where the compensa-
tion system was used and the triangle connected lines corre-
sponds to the case where it was not used. These surface errors
are the minimum values extracted from the measurement in the
tool axial direction for the surface profiles at the reading points.
In Fig. 15, change of the cutting force due to the gradual
increment of the cutting load lets the system predict the occur-
rences of the more machining errors and thus, the system sends
larger compensating signals to the tilting actuator. In the case
where the system was not used, the surface errors enlarge to
140 ^m according to the increment of the radial depth of cut,
of the machined workpiece, in the case of not using the system,
exhibit about 80 /um in the entire machined surfaces, while in
the case of using the system are limited within 20 [im by the
compensations of the tool deflections. This might appear fre-
quently in the machining of plate cams, in these cases the sug-
gested system can be applied successfully.
5 Conclusions
This paper has presented a tool deflection compensation sys-
tem with the ability to measure the cutting force, estimate the
surface error and adjust the tool position in on-line to reduce
the machining error due to the tool deflection in the end milling
process.
In the implementation of this task, the built-in type force
sensor and the tilting device were devised along with some
algorithms to operate the system. To verify the performance of
150-t r-O.if
-) ! ] , ! , , j

0 10 20 30 40 50 60 70 80
while in the case where the system was used they are maintained
within 30 /mi. The results of the second test shown in Fig. 16 CUTTING LENGTH [mm]

indicate that while the cutting loads have changed in stepwise,

the mean cutting forces increase, but the fluctuating forces are
almost level in the region where the radial depth of cuts are
greater than 1.0 mm. On the other hand, there appear less ma-
chining error in the area where the cutting load are heavier.
However, it should be pointed out that the value displayed in
these figures is the minimum surface error in every measuring
section, so the small value does not always mean a better surface
accuracy. This can be shown in the 3-dimensional view found
in the middle figures. But, even here, it is also shown that the
surface error are effectively suppressed by using the compensa-
tion system. In the result of the circular cutting test shown in
Fig. 17, the cutting loads remain constant, while the locations of
the machining surface with respect to the tool are continuously (XL —0.1 —J 1 1 | 1 1 1 1
changed. In conforming with that, the two directional cutting Cn 0 10 20 30 40 50 60 70
forces in each other orthogonal, which were measured relying CUTTING LENGTH [mm]
upon a fixed coordinate system, show the curves of a sinusoidal
form. The compensating command by the system also trace Fig. 15 Tool deflection compensation tests and machined surface er-
rors in the case of using vs. nonusing of the system for Specimen (a) in
out the similar patterns with them. The measurement for the Fig. 14

228 / Vol. 120, MAY 1998 Transactions of the ASME

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27323/ on 05/02/2017 Terms of Use: http://www.asme.o

 cuts, it is still possible to obtain good approximations for the
til
application of the compensation system.
/I, o.i & (2) The force sensor fixed on the housing of the tool adapter
—Jr shows good performance in the measurements of the cutting
forces activated on the rotating tools.
—A'
100 r _ 1
(3) The proposed compensation system adapts properly to the
r cutting environments. The system proves to suppress the surface
errors in the more productive cutting conditions without any
1 1 1 alleviation of the cutting load.
20 30 40 50
CUTTING LENGTH [mm] References
Altintas, Y., 1994, "Direct Adaptive Control of End Milling Process," Int.
Mach. Tools Manufact., Vol. 34, No. 4, pp. 461-472.
Altintas, Y., and Yellowley, I., 1987, "The Identification of Radial Width and
Axial Depth of Cut in Peripheral Milling," Int. Mach. Tools Manufact., Vol. 27,
No. 3, pp. 367-381.
Armarego, E. J. A„ and Deshpande, N. P., 1991, "Computerized End-Milling
Forces Predictions with Cutting Models Allowing for Eccentricity and Cutter
Deflections," Annals of the CIRP, Vol. 40, No. 1, pp. 25-29.
Ber, A., Rotberg, J., and Zombach, S., 1988, " A Method for Cutting Force
Evaluation of End Mills," Annals of the CIRP, Vol. 37, No. 1, pp. 37-40.
Budak, E., and Altintas, Y., 1994,' 'Peripheral Milling Conditions for Improved
Dimensional Accuracy," Int. Mach. Tools Manufact., Vol. 34, No. 7, pp. 9 0 7 -
918.
Elbestawi, M. A., Mohamed, Y., and Liu, L., 1990, "Application of Some
Parameter Adaptive Control Algorithms in Machining," ASME Journal of Dy-
namic Systems, Measurement, and Control, December, Vol. 112, pp. 611-617.
Kline, W. A., Devor, R. E., and Lindberg, J. R., 1982, "The Prediction of
£-0.1- Cutting Forces in End Milling with Application to Cornering Cuts," Int. J. Mach.
20 30 40 50 Tool Des. Res., Vol. 22, No. 1, pp. 7-22.
CUTTING LENGTH [mm] Kline, W. A., Devor, R. E., and Shareef, I. A., 1982,' 'The Prediction of Surface
Accuracy in End Milling," ASME JOURNAL OP ENGINEERING FOR INDUSTRY,
Fig. 16 Tool deflection compensation tests and machined surface er- August, Vol. 104, pp. 272-278.
rors in the case of using vs. nonusing of the system for specimen (b) in Kolarits, F. M., and DeVries, W. R„ 1991, " A Mechanistic Dynamic Model
Fig. 14 of End Milling for Process Controller Simulation," ASME JOURNAL OF ENGI-
NEERING FOR INDUSTRY, May, Vol. 113, pp. 176-183.
Liang, S. Y., and Perry, S. A., 1994, "In-Process Compensation for Milling
Cutter Runout via Chip Load Manipulation," ASME JOURNAL OF ENGINEERING
the system, experiments were carried out with the specially FOR INDUSTRY, May, Vol. 116, pp. 153-160.
prepared workpieces. Based upon this study, the following con- Matsubara, T., Yamamoto, H„ and Mizumoto, H., 1987, "Study on Accuracy
in End Mill Operations (1st Report)—Stiffness of End Mill and Machining
clusions can be obtained: Accuracy in Side Cutting," Bull. Japan Soc. ofPrec. Engg., Vol. 21, No. 2, June,
pp. 95-100.
(1) By using the mean and thefluctuationvalue of the mea- Matsubara, T., Yamamoto, H„ and Mizumoto, H„ 1987, "Study on Accuracy
sured cutting force, it is possible to estimate the minimum sur- in End Mill Operations (2nd Report)—Stiffness of End Mill and Machining
face error by tool deflection. Without the information for the Accuracy in Side Cutting," Bull. Japan Soc. ofPrec. Engg., Vol. 25, No. 4, Dec,
cutting conditions, such as the radial and the axial depth of pp. 291-296.
Milner, D. A., 1974, "Controller System Design for Feedrate Control by De-
flection Sensing of a Machining Process," Int. J. Mach. Tool Des. Res., Vol. 14,
pp. 187-197.
Oraby, S. E., and Hayhurst, D. R., 1990, "High-Capacity Compact Three-Com-
ponent Cutting Force Dynamometer," Int. Mach. Tools Manufact., Vol. 30, No.
4, pp. 549-559.
Sim, C. G., and Yang, M. Y., 1993, "The Prediction of the Cutting Force in
Ball-End Milling with a Flexible Cutter," Int. Mach. Tools Manufact., Vol. 33,
No. 2, pp. 267-284.
Smith, S., and Tlusty, J., 1991, "An Overview of Modeling and Simulation of
the Milling Process," ASME JOURNAL OF ENGINEERING FOR INDUSTRY, February,
Vol. 113, pp. 169-175.
Sutherland, J. W., and Devor, R. E., 1986, "An Improved Method for Cutting
Force and Surface Error Prediction in Flexible End Milling Systems," ASME
90' 180' 270' JOURNAL OP ENGINEERING FOR INDUSTRY, November, Vol. 108, pp. 269-279.
MACHINING POSITION [degree] Tani, Y., Hatamura, Y„ and Nagao, T., 1983, "Development of Small Three-
component Dynamometer for Cutting Force Measurement," Bulletin of the JSME,
Vol. 26, No. 214, April, pp. 650-658.
WITH THE COMPENSATION SYSTEM
Tarn, J. H., and Tomizuka, M., 1989, "On-Line Monitoring of Tool and Cutting
WITHOUT THE SYSTEM
Conditions in Milling," ASME JOURNAL OF ENGINEERING FOR INDUSTRY, August,
Vol. I l l , pp. 207-212.
Tarng, Y. S„ and Cheng, S. T., 1993, "Fuzzy Control of Feed Rate on End
Milling Operations," Int. Mach. Tools Manufact., Vol. 33, No. 4, pp. 643-650.
Watanabe, T., and Iwai, S., 1983, "A Control System to Improve the Accuracy
SURFACE ERR0R[mm] of Finished Surfaces in Milling," ASME Journal of Dynamic Systems, Measure-
ment, and Control, September, Vol. 105, pp. 192-199.
Yang, M. Y., and Park, H. D., 1991, "The Prediction of Cutting Force in Ball-
End Milling," Int. Mach. Tools Manufact., Vol. 31, No. 1, pp. 45-54.
Yang, M. Y., and Sim, C. G„ 1993, "Reduction of Machining Errors by Adjust-
ment of Feedrates in the Ball-End Mill Process," Int. J. Prod., Vol. 31, No. 3,
pp. 665-689.
Yazar, Z., Koch, K. F., Merrick, T., and Altan, T., 1994, "Feed Rate Optimiza-
Fig. 17 Tool deflection compensation tests and machined surface er- tion Based on Cutting Force Calculations in 3-Axis Milling of Dies and Molds
rors in the case of using vs. nonusing of the system for specimen (c) in with Sculptured Surfaces," Int. Mach Tools Manufact., Vol. 34, No. 3, pp. 3 6 5 -
Fig. 14 377.

Journal of Manufacturing Science and Engineering MAY 1998, Vol. 120 / 229

Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jmsefk/27323/ on 05/02/2017 Terms of Use: http://www.asme.o