Int. I. Mach. Tools Manufact. Vol. 35. No. 1. pp. 91-108.

1995
Copyright @ 1994 Elsevier Science Ltd
Pergamon Printed in Great Britain. All rights reserved
0690-6955/95$7.00 + .oo

A STUDY OF CUTTING PROCESS STABILITY OF A BORING

BAR WITH ACTIVE DYNAMIC ABSORBER
SANJIVG. TEWANI,~ KEITH E. ROUCH$ and BRUCE L. WALCOTT~
(Received 27 August 1992; in final form 15 November 1993)

Abstract-In this paper, the use of an active dynamic absorber to suppress machine tool chatter in a boring
bar is studied. The vibrations of the system are reduced by moving an absorber mass using an active device
such as an piezoelectric actuator, to generate an inertial force that counteracts the disturbance acting on
the main system.
An equivalent lumped mass model of a boring bar with active dynamic absorber is considered. A cutting
process model that considers the dynamic variation of shear and friction angle, that causes self-excited
chatter during the cutting process, is applied to the lumped mass model. The theory of regenerative chatter
is also applied to the model. Stability boundaries have been calculated for maximum permissible width of
cut as a function of cutting speed. A comparison of the boundaries for chatter-free cutting operation of a
plain boring bar, a boring bar with passive tuned dynamic absorber and a boring bar with active dynamic
absorber is provided in this paper. The comparison shows that a substantial increase in the maximum
permissible width of cut for stable cutting operation, over a range of cutting speeds, is obtained for a boring
bar equipped with an active dynamic absorber.

NOMENCLATURE
AJ system matrices
c force ratio
modulus of elasticity
force acting on the system
actuator force
cutting force
thrust force
transfer function of the system (displacement/force)
moment of inertia
K state variable feedback row vector (Kalman gains)
KP stiffness of the piezoelectric pusher
KS shear stress
K, equivalent stiffness of a cantilever beam
& stiffness of the absorber spring
L length of the cantilever beam
M main mass of the system
N number of discs in the piezoelectric pusher
Q cutting constant
X state variable vector
V applied voltage
cl damping in the main system
c2 damping in the absorber system
d diameter of the boring bar
d 33 expansion coefficient of the piezoelectric material
e, f,g,h cutting constants
m absorber mass
P chip thickness ratio
s instantaneous depth of cut
w width of cut
x displacement of the main mass
displacement of the absorber mass
Deflection of the tip of a cantilever beam
Expansion of the piezoelectric pusher
rake angle
friction angle
shear angle

tRoush Anatrol Inc., Cininnati, Ohio, U.S.A.

SCollege of Engineering, University of Kentucky, Lexington, KY 40506, U.S.A.

91
92 S. G. TEWANIet al.

P density per unit length

7 ultimate shear strength
% natural frequency

1. INTRODUCTION

MACHINE TOOL chatter is an undesirable phenomenon. Its adverse effects include unde-
sirable noise, poor surface finish of the workpiece, reduced dimensional accuracy and
reduced machine tool life. There are several factors that are responsible for machine
tool chatter. The primary causes of chatter can be categorized as forced, self-excited
and regenerative vibrations. Forced vibrations can result from an imbalance of rotating
members of the machine tool or impacts due to a multitooth cutter.
Self-excited and regenerative chatter are dependent on the relative motion between
the tool and the workpiece. Self-excited chatter can be due to the variation of the
friction force on tool rake due to the dynamic variation of shear and friction angle.
The friction force can be a strong function of the relative motion of the tool and the
workpiece. Regenerative chatter develops when the cutting pattern in the previous tool
pass affects the chip thickness and, therefore, the cutting force during the current tool
pass. For certain conditions of cutting speed, overlap between the two successive tool
passes, tool geometry and the transfer function characteristics of the machine tool, the
cutting process can become unstable and develop machine tool chatter.
Machine tool chatter has been studied extensively by a number of renowned
researchers. Tobias [l] studied the instability in the metal cutting process due to the
regenerative effect. Tlusty and Spacek [2] showed that chatter can occur due to mode
coupling effect. The extensive research performed in this area has been compiled by
Tlusty [3]. Merritt [4] forwarded the theory that regenerative chatter can be represented
as a feedback loop. The cutting process considered by Merritt was based on the theory
of the steady orthogonal cutting process given by Merchant [5] which assumes that the
cutting force is directly proportional to the depth of cut.
Under actual cutting conditions, the steady state cutting theory does not hold. The
theory of steady state cutting process assumes that the shear plane is inclined at a
constant angle. This is not true under actual cutting conditions as shown experimentally
by Knight [6]. Nigm et al. [7] used a dimensional analysis of orthogonal cutting under
steady state cutting conditions to model the dynamic cutting process. The model takes
into consideration the angular oscillation of the shear plane in response to the variations
in the cutting parameters. Equations have been obtained which give the dynamic cutting
coefficients in an explicit form, in terms of material constants and dynamic factors
determined from some cutting tests [7].
In this work, the dynamic modelling of the cutting process by Nigm et al. [7] formed
a basis for modelling the dynamic cutting forces. Wu and Liu [8] have used this theory
to develop a dynamic cutting force model and verified the theory experimentally [9].
Wu [lo] has further used the same model to obtain the governing equations for the
oscillation of the shear angle during orthogonal cutting. Comparison between the
experimental dynamic variation of shear angle and the theoretical simulation of the
cutting process has been shown to be in good agreement. Wu [ 111 also extended the
theory to include thermal effects during the cutting process.
Minis et al. [12] used the dynamic cutting force model given by Wu and Liu [8]
to obtain cutting transfer functions. These cutting transfer functions were obtained
experimentally for a wide range of cutting conditions. These transfer functions have
been used by Minis et al. [12] to predict chatter in turning operation.
Because of the detrimental effects of chatter, substantial research has been reported
on methods to avoid it. In a production environment, a system with reduced chatter
potential allows greater depth/width of cuts with stable operation of the machine tool.
Studies have indicated that an increase in the dynamic stiffness of the machine tool
can increase the stability limits of a cutting process [4]. To improve the dynamic
stability of the machine tools using passive techniques, creative ideas have been used
by a number of researchers to redesign the cutting tool. Opitz et al. [13] designed a
 Cutting Process Stability of a Boring Bar 93

milling cutter with irregular tooth pitch cutters, and showed that the cutting stability
can be improved in some speed ranges.
In the case of boring bars, owing to their large length to diameter ratios and the
associated low dynamic stiffness at resonance, chatter can easily develop during cutting
operation. To improve their performance, boring bars with passive dynamic absorbers
were introduced by several manufacturers [14]. It has been found that these boring
bars are often superior to the conventional boring bars, but the increase in performance
is not always sufficient. A number of researchers applied active control to these systems.
Klein and Nachtigal [15, 161 have presented a theoretical and experimental basis for
active control of a boring bar. Experiments were performed on a lathe equipped with
a pivoted boring bar controlled by an electro-hydraulic servo system. Glaser and
Nachtigal [17] developed a hydraulic chambered actively controlled boring bar.
Tellbuescher [18] has looked into using piezoelectric pushers to actively control the
boring bar against a rigid reference. The concept of using an active dynamic absorber
in a boring bar was introduced by Tewani et al. [ 191 and Stephens [20]. The frequency
response of a boring bar with such an active dynamic absorber, as well as the stability
analysis by Stephens [20] show that the use of active dynamic absorbers in boring bars
could be promising.
In this paper, the stability analysis of a boring bar with active dynamic absorber is
obtained using the model for cutting dynamics and the theory for stability analysis
given by Nigm et al. [21]. The maximum allowable width of cut is calculated as a
function of the cutting speed. The threshold of stability is plotted for a typical boring
bar under no control. This is then compared with the corresponding limits obtained
for a boring bar with a passive absorber and with an active absorber. Two different
schemes are used for the active feedback control. The recommendations of feedback
control made by Stephens [20] are used first. Then the theory of optimal control is
used to obtain the feedback parameters. The stability threshold under these conditions
of active feedback control are compared.
The authors have designed and built a boring bar with a secondary mass inside a
cavity near the tool end of the bar. A commercial piezoelectric actuator was used to
move the secondary mass, in conjunction with a mechanical amplification [22]. The
details of this experimental work [23] are beyond the scope of this paper, although
some results are presented as an example. The assumptions in this paper are based
on the dimensional and physical characteristics of this experimental device.
2. THEORY
2.1. Modelling of the boring bar
For the conceptual evaluation, an equivalent lumped mass model of a typical 2 inch
diameter high speed carbon steel boring bar is used (Fig. 1). This simplifies the analysis
and is a reasonable approximation for vibration of the boring bar in the first bending
mode. Prior work, comparing the lumped mass model with a finite element model,
confirmed the adequacy of the lumped mass model in the frequency range of interest
[24]. Comparison with the experimental data on boring bars also confirms the adequacy
of the lumped mass model. For this case study, a length to diameter ratio of 1O:l is
taken for the boring bar. Assuming a boring bar as a clamped beam, the deflection
of the free end of the beam, for a point force applied at the same point is given by:

FIG. 1. Equivalent lumped mass model of a typical boring bar

94 S. G. TEWANI et al.

where F is the point force, L is the length of the beam, E is the modulus of elasticity
and I is the moment of inertia. Since the tool tip is attached at the free end of the
boring bar, the equivalent stiffness of the boring bar can be given as:

K
1
z3E
L3 .
During a machining operation, the first mode of vibration of the boring bar is most
likely to be excited. The first natural frequency of a clamped beam is given by the
equation:

(3)

where p is the mass per unit length of the beam. The equivalent lumped mass of the
cantilevered beam based on the first natural frequency of the system is then found by
using the equation:

(4)

A schematic of the idealized model of the boring bar with passive dynamic absorber,
as used in this study, is shown in Fig. 2. A typical frequency response of a system with
passive dynamic absorber is shown in Fig. 3. For a particular combination of absorber
stiffness and damping factor, the absorber can be tuned such that the amplitude of
vibration of the main mass can be minimized. The theory of an optimized passive
dynamic absorber is given by Den Hartog [25]. A main mass to absorber mass ratio
of 1O:l is assumed for the test case and the tuned values of the absorber stiffness and
damping factor are obtained. The mass ratio of 10: 1 is easily attained in the experimental
work of the authors.
The schematic of the system with active dynamic absorber is shown in Fig. 4. An
active force-generating actuator is used in conjunction with passive elements (spring
and damper). Prior work [19] showed that when the actuator generates a force pro-
portional to the acceleration and velocity of the main mass (a boring bar in this case),
the frequency response shows a substantial reduction in the amplitude of the main
mass. Stephens [20] studied the range of feedback parameters necessary for stable
operation of the system. In his study, Stephens also optimized the passive parameters
such as the absorber spring stiffness and damping factor such that the amplitude of

FIG. 2. Schematic of a boring bar with dynamic absorber.

 Cutting Process Stability of a Boring Bar 95

FIG. 3. Typical frequency response of a system with passive dynamic absorber

52
M

ACW
K2 c2

FIG. 4. Schematic of a system with active dynamic absorber.

vibration of the main mass is minimized. The feedback control algorithm proposed by
Stephens [20] is one of the algorithms used in this study. For this feedback control
algorithm, these optimized parameters given by Stephens [20] are used in this paper.
The dynamical equations of system shown in Fig. 4 are given as:

M_? = - K,x - c,k - K2(x - y) - c&t - j) - F,,, + F,

mji = K,(x - y) + c*(i - j) + F,,, (5)
where F,,, is the force given by the actuator. From equation (5), the transfer function
G, for the displacement of the main mass can be obtained, such that

$= G,. (6)

Tewani ef al. [19] and Stephens [20] proposed a piezoelectric pusher to serve as a
force generating actuator in the active dynamic absorber. Piezoelectric pushers provide
a displacement directly proportional to the input voltage. Owing to the small size and
fast response characteristics, piezoelectric pushers are suitable to use in such appli-
cations. A commercially available piezoelectric pusher consists of a stack of ceramic
discs arranged on top of one another and connected in parallel such that same voltage
96 S. G. TEWANI et al.

can be applied to each disc. The piezoelectric discs expand in response to the applied
voltage. A typical piezoelectric pusher can be represented as shown in Fig. 5. Owing
to the elastic constant of the ceramic discs, a piezoelectric pusher has some stiffness
associated with it. This stiffness is in series with the extension of the pusher owing to
the applied voltage. The expansion of a piezoelectric pusher is given by:

A’ = Nd,3V, (7)

where N is the number of ceramic discs, d33 is the strain constant of the pusher and
V is the applied voltage. From equilibrium of the actuator, the force acting on the
actuator is equal to

Fact = K&t - y) + K,Nd33V.

If a piezoelectric pusher is considered as a force generating device in the active dynamic

absorber, the applicable dynamic equations for the system are given as:

Mi = -K,x - cIk + &(y - x) - cz(i - j)

+ K,(y - x) - K,Nd,,V + F

my = c&i - j) - K,(y - x) - K,(y - x) + K&i&‘. (9)

Defining the state variables as

x1 =x; x* =i; x3 =y; x4 =j, (10)

the dynamic equations given by equation (9) can be transformed into a state-variable
equation

~=Ax+ BV+ F. (11)

For calculating the feedback voltage required for controlling the vibrations of the
boring bar, two approaches have been used. First, the approach of proportional feed-
back proposed by Tewani et al. [19] and Stephens [20] is used in which the feedback
voltage is assumed to be of the form

V= ti + bk, (12)

where a and b are constants. The values of a and b are calculated to ensure that the
closed loop system is stable and a minimum in the response of the closed loop system

FIG. 5. Schematic of a typical piezoelectric pusher.

 Cutting Process Stability of a Boring Bar 97

is obtained. In the second approach, the theory of optimal control [26, 271; is applied
to the system. The state variable equations obtained for the system (equation (11))
are posed as a Linear Quadratic Regulator (LQR) problem to obtain the feedback
voltage as

V= -KX, (13)

where K is the Kalman gain matrix. Having found the Kalman gain matrix, equation
(13) can be substituted into equation (9) to obtain the transfer function of the amplitude
of vibration of the boring bar, x vs the force F acting on the boring bar (as given by
equation (6)).

2.2. Modelling of the cutting process

A schematic representation of orthogonal cutting process with continuous chip forma-
tion without built-up edge forming on the tool is shown in Fig. 6. According to the
conventional thin shear plane model given by Merchant [5] the thrust force F, and the
cutting force F, given by the equations:

sw sin( l3 - CY)
Ft = (14)
sin+ cos($ + l3 - 0~)
SWTsin@ - 0~)
F, = (15)
sin+ cos($ + p - 0~) ’

where

s = (x0 - x)

is the instantaneous depth of cut, IXis the rake angle, p is the friction angle, $ is the
shear angle, r is the ultimate shear strength of the material being machined, w is the
width of cut and (x0 - X) is the instantaneous depth of cut. From the geometry of the
cutting process, and using the assumption that there is no built-up edge formation on
the cutting tool, the ratio of the depth of cut to chip thickness is given by

sin+
p = cos($ - OL)’ (16)

FIG. 6. Schematic of orthogonal cutting process with continuous chip formation and no built-up edge.
98 S. G. TEWANI et al.

where p is called the chip thickness ratio. From equation (16) the shear angle is found
to be given by

4 = tan -‘( 1 !;:;J (17)

From dimensional analysis in orthogonal cutting done by Nigm et al. [7], the chip
thickness ratio, p and the force ratio C are given by the equations:

p = esinol + (1 - fsina)
[
g+ ~
(I&k )I
In ‘5
(

[ i )I.
SV”
C=+sincx+(l+fi &-(1(&i In z (18)

These equations are valid in the absence of a built-up edge. The coefficients in the
above expressions are characteristic of the workpiece material and are determined
experimentally.
If the instantaneous values of the depth of cut, the shear angle and the force ratio
are denoted by Si, 4i and Ci, respectively, then the instantaneous force components
acting on the tool during the cutting process are given by

si
Fci = W?
SiIl$i( COSd&- C&l&)

ciri
F,i = WT . (19)
Slll+i( CO&t& - C$iIl+i) ’

Fci and Fti are the instantaneous cutting forces along and perpendicular to the instan-
taneous direction of motion of the cutting tool. Using the theory of partial differen-
tiation, the incremental variation of the cutting forces as a function of increments in
the depth of cut, shear angle, and force ratio can be obtained. Using only the first
order terms, the equations for cutting force variation are obtained as:

dF, = w+(& + e,d4 + e,dC)

dF, = ti( Cds + Ceid4 + e,cot4 + dC), (20)

where

’ = sin4(cos4 - Csin4)

sin4 + Ccos4
(
e1 = cos4 - Csin4
- cot4 s
1
sin4
e2 = ( cos4 - Csin4 1” (21)

The instantaneous directions of the cutting forces may be different from the direction
of the global coordinates as shown in Fig. 7. The angle between the global coordinate
system and the instantaneous coordinate system attached to the cutting tool may be
assumed to be small. Denoting the small angle between the two coordinate systems
by dJI,, the incremental force components in the directions of global coordinate system
are given by:
 Cutting Process Stability of a Boring Bar 99

FIG. 7. Schematic of instantaneous cutting forces.

dF, = ti(ds + eId+ + e,dC - Csd$r)

dF, = wi(Cd.s + CeIdd, + e,cot$ + sd+,). (22)

The independent variables determining the cutting forces are the depth of cut S, the
rake angle cx and the cutting speed v. The incremental changes in the force ratio C
and the shear angle $ can be found as a function of the incremental changes of the
above-mentioned independent variables. Therefore,

dC = ml& + m,da + m,dv (23)

d$ = n,ds + n2dct + n3dv + n,d$. (24)

The fourth term in equation (24) becomes important owing to the dynamics of the
cutting process. For the case of “wave cutting”, the instantaneous direction of cutting
operation is at an angle of d$, with the direction against which the rake angle of the
tool is measured. The effect due to this phenomena is that the effective rake angle is
increased. For the case of the “wave removing” process, the opposite phenomena
occurs and the effective rake angle changes by d&. During the actual cutting dynamics,
a combination of the above two procedures occurs. The effective change in the rake
angle is, therefore, given by

d$ = de, + d&. (25)

The change in the effective rake angle adds on to the expression of the change in the
shear angle. If the motion of the lower chip boundary is denoted by x1, and the motion
of the upper chip boundary is denoted by x2, then the variation of the cutting parameters
are given by:
100 S. G. TEWANI et al.

(27)

(28)

d,,+ (29)

In equation (26), r denotes the geometric lead of the free end of the shear plane with
respect to its end on the cutting edge of the tool and is expressed as:

E = scat+. (30)

Substituting equations (23)-(29) into equation (22), the equations for force variations
are obtained as

dF, = wi T,(xi - x,) + +i + T$ (k, - &) , (31)

I

where

Ci = 1 + niei + mle2
C2 = e + (nie + n2)e1 + (mie + m2)e2 - Cs
C3 = n2el
Ti = (1 + niei)C + m,e2cot+
T2 = [E + (nie + n,)e,]C + (mie + m2)ezcot+ + s
T3 = n3elC

and

64 84 64
nl=--; n2=g,; n,=G
6s

6C ik
m 2.
ml=-;
6s 2 6a’ m3=G’

In deriving equation (31), only the first order terms are considered. Higher order
terms are neglected based on the assumption that xi and x2 are small. Assuming that
x1 and x2 are harmonic with frequency o, the incremental force components are given
by:

dF, = ti Ci(xi - x2) + j ; [C,xi + C3(x1 - x2)]

( 0 1

dF, = wi T,(x, - x2) + j ; [ T2x1 + T3(x1 - x2)] (32)

1 0 1

The coefficients e, f, g, h, e’, f, g’, h’, n, k and k’ are given by Nigm et al. [21] for
 Cutting Process Stability of a Boring Bar 101

a typical tool-workpiece material combination (carbide cutting tool-mild steel). It must

be noted that the forces in the two directions are functions of the motion of the cutting
tool in the direction normal to the machined surface only.
In this paper, the lumped mass model of the boring bar is assumed to have its single
degree of freedom in the direction normal to the surface being machined (as shown
in Fig. 6). The absorber mass is allowed to move in the same direction. From the
cutting process model, it must be noted that the force in the tangential direction of
the surface being machined is proportional to the motion of the boring bar in the
normal direction. If the lumped mass model of the boring bar is assumed to have two
degrees of freedom (in the normal and tangential direction of surface being machined),
and there is no other cross-coupling between these two degrees of freedom (besides
the one due to cutting process), then the motion of the bar in the tangential direction
can be reduced by reducing its motion in the normal direction.

2.3. Stability analysis

For the one-dimensional model of the boring bar with active dynamic absorber, the
transfer function for the displacement of boring bar vs the force acting on the boring
bar was expressed by equation (6). If the displacement of the boring bar x is assumed
to be normal to the machined surface, then from equation (6)

x = -G,dF, (33)

The negative sign in the above equation has been introduced due to the positive
direction of the thrust force dF, being opposed to that of normal displacement of the
boring bar. Assuming that at the threshold of stability, the displacement of the boring
bar x(t) is a simple harmonic function of time, the transfer function G, can be obtained
as

G,,(w) = P + iQ,

where P and Q are functions of the frequency w. Assuming that

q(t) = A&“’ (34)

and the workpiece is moving at N revolutions per second, the change of depth of cut
due to previous tool pass is given by:

(35)

where p, denotes the fraction of the overlap between the previous and present tool
passes. Substituting equation (34) into equation (35), the change in the depth of cut
due to “wave removing” is given by the equation

X2(t) = ~,ei(~~-~) (36)

where

(37)
102 S. G. TEWANIet al.

and R is the radius of the workpiece. Substituting equations (34)-(37) into equations
(32) and (33) we get:

A& = -(P + jQ)wi [ T2 + T3( 1 - t.&“)] Aod”‘. (38)

Equating the real and imaginary parts of the above equations, the following two
equations are obtained

. *
AI(l - txos9) + AzksmO + LTP=O
27FR 2

-A*(1 - pcosti) + A,tJ,sinb + (40)

where

AI = TIQ + $3’
AZ= TIP- &GQ.

The chatter frequencies are found by finding the roots of equation (39). The correspond-
ing limiting widths of cut are found from equation (40).

3. RESULTS

A 2 inch diameter boring bar with a length to diameter ratio of 1O:l was considered
for the test case. The equivalent stiffness and mass of the boring bar were calculated
based on the first cantilever mode of vibration and were found to be equal to 1.55e6
[N/m] and 2.0 [kg], respectively. A structural damping factor of 1% is assumed.
Assuming a main mass to absorber mass ratio of lO:l, and using the theory of optimiz-
ation on the absorber spring and damping given by Stephens [20], the stiffness and
damping coefficient of the passive absorber are calculated to be 14526.0 [N/m] and
32.4 [N s/m], respectively.
A commercially available piezoelectric pusher is used as an active control element.
The stiffness and the electric strain constant of the pusher are found from the data
provided with the actuator. These constants are used in the state-variable model of
the system as described by equation (9). A computer program was written using an
algorithm given by Meirovitch [ 171 to solve the steady state Ricatti equation and obtain
the Kalman gain factors for state-variable feedback control. For the state variables
defined by equation (lo), the Kalman gain factors were calculated to be

K = [ -2.22~~ + 9 2.54e + 5 -3513 272051.

A computer program was written which considers the cutting dynamics and calculates
the width of cut at the threshold of stability. The program is used to solve equation
(39) for the given system parameters. Equation (40) is then solved to calculate the
width of cut. The parameters that needed to be defined were the rake angle, the
chosen depth of cut and the radius of the workpiece. A typical value of 5” is used for
the rake angle. The desired depth of cut (feedrate) was chosen to be equal to 0.05
inch per revolution and the radius of workpiece was taken to be 2.9 inches. The width
of cut was obtained for four different cases:
l plain boring bar
 Cutting Process Stability of a Boring Bar 103

l boring bar with a passive dynamic absorber

l boring bar with active dynamic absorber using the control algorithm proposed by
Stephens
l boring bar with control algorithm using Kalman gain factors obtained from solving
steady state Ricatti equation. (This control algorithm will be referred to as sub-
optimal control in the rest of this paper.)

A comparison of the maximum possible width of cut as a function of the spindle speed
at the threshold of stability of machining operation is shown in Fig. 8. It is seen that
the irnl- Jvement in the width of cut for using a passive dynamic absorber inside the
boring bar when compared with a plain boring bar translates into an increase in the
permissible length to diameter ratio (L/d) of the boring bar by about 21%. When
acceleration and velocity feedback proposed by Stephens is used, a more substantial
increase in the permissible width of cut is observed. In the range of spindle speeds
considered, the allowable L/d ratio of the boring bar can be increased by at least 41%
above the case of boring bar with passive dynamic absorber. The threshold of stability
limits are further increased when a sub-optimal control algorithm is used with the
boring bar. This translates into a further increase in the L/d ratio by 25% over the
case of the boring bar with control algorithm proposed by Stephens [20]. The overall
improvement in the L/d ratio when compared with using a plain boring bar is about
210%.
The effect of higher structural damping in the boring bar was evaluated, by using
10% critical damping of the boring bar. The permissible widths of cut for stable
machining operation are obtained for the four cases considered earlier. For the case
of a plain boring bar, the increase in the limit of stability is substantial. The comparison
is shown in Fig. 9. A similar observation is made for the case of a boring bar with
passive dynamic absorber. This is shown in Fig. 10. For the case of boring bar with active
dynamic absorber using either the acceleration-velocity feedback control technique or
sub-optimal control, the increase in ,tability limits are barely noticeable. These compari-
sons are made in Figs 11 and 12. It can, therefore, be implied that the structural
damping is effective only in the case of a plain boring bar or a boring bar with passive

28

24
Active dynamic absorber
(Sub-optimal control strategy)
/

Active dynamic absorber

(Stcphens[l Z]control strategy)

6
t ,- Toned passive absorber

20 22 24 26 28 30 32 34 36 3x 40

Spindle speed (rcvlscc)

FIG. 8. Comparison of the maximum possible stable width of cut at threshold of stability of machining
operation.

m 351-H
104 S. G. TEWANI et al.

\
.
\ _‘.\ 10% structural drmpiq
\
\
\
\
.
4 / , e4. .
*W-d
*.
*.
*-
-I --__
1 C structural dmtping

I I I I I I I I I I
0
20 22 24 26 28 30 32 34 36 38 40

Spindle speed (revlsec)

FIG. 9. Cutting process stability limits of a plain boring bar as a function of its structural damping.

12

‘1 \
\
10 1
F \

I I I I I I I I I I
01
20 22 24 26 28 30 32 34 36 38 40

Spindle speed (revlsec)

FIG. 10. Cutting process stability limits of a boring bar with passive dynamic absorber, as a function of its
structural damping.

dynamic absorber. However, it is very difficult to obtain these high damping levels in
structural materials and such an approach of adding damping is not practical. It also
confirms that the effectiveness of the active techniques is not greatly affected by the
degree of damping in the main system.
A number of cutting tests have been performed using a commercial boring bar
equipped with an active dynamic absorber. The cutting tests were performed with and
 Cutting Process Stability of a Boring Bar 105

10% structural damping

1% structural damping
11 -

10 -

9-

8-

71 I I I I I I I I I I
20 22 24 26 28 30 32 34 36 38 40

Spindle speed (rev/set)

FIG. 11. Cutting process stability limits of a boring bar with active dynamic absorber using acceleration and
velocity feedback control as a function of its structural damping.

,lO% structural damping

/
1% rtructaraJ damping

la
20 22 24 20 28 30 32 34 32 32 a

spindle speed (rev!sec)

FIG. 12. Cutting process stability limits of a boring bar with active dynamic absorber using optimal control
as a function of its structural damping.

without the active control device operational. The tests were performed for different
length to diameter ratios (ranging from 6 to 10) and under different cutting conditions
by changing the speed of the workpiece (between 235 and 600 rpm) and the depth of
cut (between 0.08 and 0.01 inches). The feedrate of the cutting tool was maintained
at 0.009 inches per revolution. A typical result obtained for these tests is shown in
106 S. G. TEWANI et al.

Fig. 13, which shows the surface finish of the workpiece obtained under the same
cutting conditions of depth of cut, feedrate and length to diameter ratio (LID = 6) of
the boring bar, but with and without the feedback control operational. Figure 14 shows
the acceleration signal under the two conditions. From these two results, it can be
concluded that a substantial improvement in the stability of the cutting process with
the boring bar can be obtained with the active control system. The details of the
experiment are beyond the scope of this paper, but have been presented in another
related paper by Tewani et al. [23].

4. CONCLUSIONS

Cutting stability of a boring bar with a vibration suppression device has been con-
sidered using the theory of cutting given by Nigm et al. [7]. An active dynamic absorber
[19, 201 is considered as a vibration control device to be used with a boring bar. An
equivalent lumped mass model of the boring bar with dynamic absorber is obtained.
Two different control strategies are considered for the active dynamic absorber. The
first strategy involved the use of acceleration and velocity feedback proposed by
Stephens. The second strategy involved obtaining the state-variable model of the system
and calculating the Kalman gains for sub-optimal control of the system.
Stability charts were calculated for a typical boring bar equipped with the active
dynamic absorber for each control strategy. For the sake of comparison, the correspond-
ing stability charts for a plain boring bar and a boring bar with tuned passive dynamic
absorber are calculated. The results show that the optimal control strategy applied to
a boring bar with active dynamic absorber substantially increases the stability limits when
compared with a plain boring bar. The acceleration and velocity feedback proposed by
Stephens also show a substantial improvement in stability limits, though not as much
as that obtained with sub-optimal control.

FIG. 13. Surface of the workpiece obtained from the cutting operation performed with plain boring bar, bar
with passive dynamic absorber, and bar with active dynamic absorber, under same cutting conditions, for
same length to diameter ratio of the boring bar.
 Cutting Process Stability of a Boring Bar 107

Waveform of channel-Y in g
24
RMS = 6.132
18

12

--_0.0. 011 0:2 0:3 0:4 0.5

Time (set)

Waveform of channel-Y in g
24 T
RMS = 1.455
18 ..

12 . .
E
t 6
2
3 0.
>
5 -6 .

-12 ..

-18 .

-24 4 c
0.0 0.1 0.2 0.3 0.4 0.5
Time (set)

FIG. 14. Acceleration signals (in time domain) from the accelerometer located at the tip of the boring bar,
obtained during cutting operations with no feedback control and with feedback control, respectively.

Acknowledgemenl-This research has been performed with the support of the Center of Robotics and
Manufacturing Systems at the University of Kentucky and Kennametal Inc. at Latrobe, Pennsylvania.

REFERENCES

[II S. A. TOBIAS, Machine Tool Vibration. Blackie and Son (1965).

PI .I. TLUSTY and L. SPACEK, Self-excited vibrations in machine tools, Czech. Nakludatolstvi. VSAV, Prague
(1954).
J. TLUSTY, Analysis of the state of research in cutting dynamics, Ann. CIRP 27, 583 (1978).
1:; H. E. MERRIIT, Theory of self-excited machine tool chatter, Trans. ASME, J. Engng Ind., Series B.
87(4), 447 (1965).
[51 M. E. MERCHANT. Basic mechanics of the metal cutting process, J. a&. Mech., Trans. ASME 66,
Al68 (1944).
(61 W. A. KNIGHT, Application of the universal machinability chart to the prediction of machine tool
stability, ht. J. Mach. Tool Des. Res. 8, 1 (1968).
[71 M. M. NIGM, M. M. SADEK and S. A. TOBIAS, Dimensional analysis of the steady state orthogonal
cutting process, Int. J. Mach. Tool Des. Res. 17, 1 (1977).
PI D. W. WV and C. R. Lru, An analytical model of cutting dynamics-Part 1. Model building, Trans.
ASME, J. Engng Ind. 107, 107 (1985).
[91D. W. WV and C. R. LIU, An analytical model of cutting dynamics-Part 2. Verification, Trans.
ASME, J. Engng Ind. 107, 112 (1985).
[W D. W. Wu, Governing equations of the shear angle oscillation in dynamic orthogonal cutting, Trans.
ASME, J. Engng Ind. 108, 280 (1986).
1111D. W. Wu, Comprehensive dynamic cutting force model and its application to wave-removing process,
Trans. ASME, J. Engng Ind. 110, 153 (1988).
[=I I. E. MINIS, E. B. MAGRAB and I. 0. PANDELIDIS, Improved methods for the prediction of chatter in
turning-Part 2. Determination of cutting process parameters, Trans. ASME, J. Engng Ind. 112, 21
(1990).
108 S. G. TEWANIet al.

[13] H. OPITZ, E. U. DREGGERand H. ROESE, Improvement in the dynamic stability of the milling process
by irregular tooth pitch, Proc. 7th Int. M. T.D.R. Conf., Birmingham, p. 213 (1966).
[ 141 Kennametal Manual for Metalcutting Tools, Kennametal Inc., Latrobe, Pennsylvania.
[ 151 R. G. KLEINand C. L. NACHTIGAL, A theoretical basis for the active control of a boring bar operation,
1. Dvn. Svst. Meas. Control, Trans. ASME 97, 172-178 (1975).
[16] R. 6. KLEIN and C. L. NACH~GAL,The application of active control to improve boring bar performance,
J. Dvn. Svst. Meas. Control. Trans. ASME, 97, 179-183 (1975).
[ 171 D. J: GM>ER and C. L. NA~HTIGAL,Development of a hydraulic chambered, actively controlled boring
bar, Trans. ASME, J. Engng Ind. 101, 362-368 (1979).
I181 E. TELLBUESCHER, Piezoelectriche Aktuatoren als Erregereinheit und Stellglied an Werkzeugmachinen,
Jendryschih, Juergen, VDZ Z 127(6), 183-188 (1985).-
1191 S. G. TEWANI. L. S. STEPHENSand K. E. ROUCH, Control of machining chatter
L 1
by use of active
elements, Proc. Symp. Adv. Manufacr. p. 31. University of Kentucky (1988). _
[20] L. S. STEPHENS,An active dynamic absorber for the suppression of machining chatter, M.S. thesis,
Universitv of Kentucky (1989).
[21] M. M. &GM, M. M. -S~DEKand S. A. TOBIAS,Determination of dynamic cutting coefficients from
steadv state cutting data. Znt. J. Mach. Tool Des. Res. 17, 19 (1977).
[22] K. E: ROUCH,S. ?;. TE~ANI, B. L. WALCOTT, T. R. MASSA,R. W.‘.%EPHENSONand L. S. STEPHENS,
Active vibration control device, U.S. Patent 5,170,103 (1992).
[23] S. G. TEWANI,T. C. SWITZER,B. L. WALCOTT,K. E. ROUCHand T. R. MASSA, Active control of
maching tool chatter for a boring bar: experimental results, Proc. 14th Biennial Conf. Vibr. Noise,
Albuquerque, New Mexico (September 1993).
[24] S. G. TEWANI,K. E. ROUCHand B. L. WALCOXT, A finite element analysis of a boring bar using active
dynamic absorber for active vibration control, Proc. ASME Winfer Annual Meeting, Atlanta, Georgia,
p. 173 (1991).
[25] .I. P. DEN HARTOG,Mechanical Vibrations, 3rd edition, p. 123. McGraw-Hill, New York (1947).
[26] F. L. LEWIS, Optimal Conrrol, 1st edition, p. 227. John Wiley, New York (1986).
[27] S. G. TEWANI,B. L. WALCOTTand K. E. ROUCH, Active optimal vibration control using dynamic
absorber, Proc. IEEE Int. Conf. Robotics Automation, Sacremento, California, p. 1182 (1991).
[28] L. MEIROVITCH, Dynamics and Conrrol of Structures, 1st edition, p. 195. John Wiley, New York (1990).