ARTICLE IN PRESS

International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473

www.elsevier.com/locate/ijmactool

Generalized modeling of drilling vibrations. Part I: Time domain

model of drilling kinematics, dynamics and hole formation
Jochem C. Roukema, Yusuf Altintas
Manufacturing Automation Laboratory, Department of Mechanical Engineering, The University of British Columbia,
6250 Applied Science Lane, Vancouver, BC, Canada V6T1Z4
Received 18 September 2006; accepted 4 October 2006
Available online 5 December 2006

Abstract

This two part paper presents a comprehensive exercise in modeling dynamics, kinematics and stability in drilling operations. While
Part II focuses on the chatter stability of drilling in frequency domain, Part I presents a three-dimensional (3D) dynamic model of drilling
which considers rigid body motion, and torsional–axial and lateral vibrations in drilling, and resulting hole formation. The model is used
to investigate: (a) the mechanism of whirling vibrations, which occur due to lateral drill deflections; (b) lateral chatter vibrations; and (c)
combined lateral and torsional–axial vibrations. Mechanistic cutting force models are used to accurately predict lateral forces, torque
and thrust as functions of feedrate, radial depth of cut, drill geometry and vibrations. Grinding errors reflected on the drill geometry are
considered in the model. A 3D workpiece, consisting of a cylindrical hole wall and a hole bottom surface, is fed to the rotating drill while
the structural vibrations are excited by the cutting forces. The mechanism of whirling vibrations is explained, and the hole wall formation
during whirling vibrations is investigated by imposing commonly observed whirling motion on the drill. The time domain model is used
to predict the cutting forces and frequency content as well as the shape of the hole wall, and how it depends on the amplitude and
frequency of the whirling vibration. The model is also used to predict regenerative, lateral chatter vibrations. The influence of pilot hole
size, spindle speed and torsional–axial chatter on lateral vibrations is observed from experimental cutting forces, frequency spectra and
shows good similarity with simulation results. The effect of the drill–hole surface contact during drilling is discussed by observing the
discrepancies between the numerical model of the drilling process and experimental measurements.
r 2006 Elsevier Ltd. All rights reserved.

Keywords: Drilling; Lateral vibration; Chatter; Time domain simulation; Torsional; Axial

1. Introduction bottom and a smooth surface, generated by lateral whirling

Tool vibrations during drilling can cause errors in hole
size and shape that may be unacceptable. Fig. 1 shows four
different hole types generated by a regular twist drill. The
hole shown in Fig. 1a was drilled with a very short drill bit,
is perfectly round, and has a smooth surface without drill
vibration marks. Fig. 1b shows a sunray pattern obtained
from an unstable cut (chatter), due to coupled vibrations in
axial and torsional directions [1]. The hole is round as in
Fig. 1a. Fig. 1c shows a three-sided polygon shape at the
Corresponding author. Tel.: +1 604 822 5622; fax: +1 604 822 2403.
E-mail addresses: roukema@interchange.ubc.ca (J.C. Roukema),
altintas@mech.ubc.ca (Y. Altintas).
vibrations.
Fig. 1d shows a three-sided hole with sunray pattern at
the bottom, left by a drill undergoing torsional–axial
chatter as well as lateral whirling vibrations. The holes
shown in Figs. 1b–d were drilled with a slender drill bit.
These photographs show that torsional–axial and lateral
whirling vibrations can occur independently or at the same
time. Detailed understanding of the mechanisms that cause
these drill vibrations will allow for improvements in speed
and precision of drilling operations. Accurate prediction of
the cutting force system, the tool dynamic properties, the
chip generation and hole formation mechanisms are
essential for realistic numerical simulation of the drilling
process.

0890-6955/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijmachtools.2006.10.005
 ARTICLE IN PRESS
1456 J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473

Nomenclature Lf flute length measured from drill tip (mm)

L1 drill dimension (mm)
b1, b2 radial depth of cut flute 1, flute 2 (mm) M mass matrix of drill bit (kg)
Db elemental width of cut (mm) m number of elements along the cutting edge (—)
bt back taper of drill (reduction of flute diameter) N number of teeth on drill/reamer (—)
(mm) Ng number of grid points along the hole circum-
C damping matrix of drill bit (N s/m) ference (—)
D drill diameter (mm) Nr number of revolutions passed in simulation (—)
e amplitude of imposed circular whirling motion r radial distance from drill axis to midpoint of
(mm) element on flute (mm)
Fx, Fy, Fz cutting forces acting on tool tip (N) rr radial drill runout (in direction of cutting lips)
F ti ; F ri ; F zi tangential, radial and axial force acting on (mm)
drill flute i (N) rt tangential drill runout (perpendicular to cutting
Fwi contact force between drill and hole wall (N) lips) (mm)
fc chatter frequency (Hz) rz axial drill runout (lip height difference) (mm)
fr feedrate (mm/rev) Dt time step (s)
Dfr feedrate increment (mm) t1(i, k) chip height for cutting edge point i on edge 1 at
fs spindle frequency (Hz) timestep k (mm)
hf chip thickness (mm) Tc cutting torque acting on tool tip (N m)
fw whirling frequency in fixed coordinate system 2W web thickness of drill bit (mm)
r (Hz) Wx,1,Wy,1,Wz,1 workpiece surface coordinate matrices
fw whirling frequency in coordinate system rotat- for surface generated by flute 1 (mm)
ing with the tool (Hz) xc, yc, zc lateral and axial tool tip deflections (mm)
hlip distance from drill tip to start of cutting flutes xi,1, yi,1 coordinates of intersection points of cutting
(mm) edge with grid circles (mm)
K stiffness matrix of drill bit (N/m) xfi,1, yfi,1 coordinates of intersection point of flute with
k time step (—) wall grid circle (mm)
kcz specific thrust force (N/mm2) xp,1, yp,1 coordinates of the peripheral point of drill
kcp specific torque force (N/mm2) flute 1 (mm)
kct specific tangential force (N/mm2) xp,2, yp,2 coordinates of the peripheral point of drill
kcr specific radial force (N/mm2) flute 2 (mm)
ki stiffness (N/m) ~
xt;1 ; ~
yt;1 ;~zt;1 coordinates of tool tip points on cutting
km number of modes included in vibration direc- edge 1 (mm)
tion (—) zd height of whirling grid layer in global coordi-
kZFz direct axial stiffness of drill bit (axial due to nate system (mm)
thrust) (N/m) zi,1(i,k) height of cutting edge at intersection points
kZTc cross axial stiffness of drill bit (axial due to with grid (mm)
torque) (Nm/m) q1 ; q2 ; q_ 1 ; q_ 2 state space variables (m, m/s)
kyTc direct torsional stiffness of drill bit (torsional b0 drill bit helix angle at drill periphery (deg)
due to torque) (N m/rad) zi damping ratio (—)
kyFz cross torsional stiffness of drill bit (torsional yc torsional tool tip deflection (rad)
due to thrust) (N/rad) 2kt drill tip angle (deg)
k0 f,1, kf,2 estimate location of cutting edge 1,2 with wkt tip angle grinding error (deg)
respect to workpiece grid (—) F(s) transfer function (N/m)
kf1, kf2 discrete location of cutting edge 1,2 with respect o0 web angle at drill periphery (rad)
to workpiece grid (—) oni natural frequency mode i (Hz)
L drill length (mm) O angular speed of the tool (rad/s)

2. Literature review geometry with hole accuracy [2]. Galloway attributed

There has been significant research effort in modeling the
drilling forces as a function of tool geometry, grinding
errors, material properties, structural dynamics of the drill,
and initial penetration of the chisel edge into the material
since Galloway’s experimental results relating the tool
experimentally observed hole accuracy (oversize) to the
drill point symmetry. A lip height error rz on a drill with
tip angle 2kt caused a hole roundness error of rz tan kt,
as the drill would rotate about an axis displaced 0.5rz tan kt
from the spindle axis, creating equal chip areas on
each flute.
 ARTICLE IN PRESS
J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473
Fig. 1. Hole shapes resulting from drilling blind holes in full material: (a)
stable cut, no visible vibrations; (b) sunray pattern due to unstable
torsional–axial chatter vibration; (c) trigon caused by whirling vibrations;
and (d) surface resulting from combined torsional–axial chatter and
whirling vibrations.
Lee et al. proposed a kinematic model of lobing motion,
which results in odd numbered polygonal shapes if the
proper backward whirling motion is imposed. Lin et al.
found in experiments that the tool whirling frequency was
slightly smaller than an odd multiple of the rotation speed,
which allowed for chip thickness variation and regenerative
lateral chatter. Zelentsov observed that deviation of the
drill rotation axis from the spindle center resulted in both
hole enlargement and lobing (polygonality) of the hole
being drilled. The models in [3–5] did not explain the forces
that cause the whirling motion.
Reinhall and Storti [6] modeled the drill as an impacting
rotating rod, and found that the periodic motion could be
stabilized by the chip removal process. Tekinalp and Ulsoy
[7,8] used finite-element solutions to analyze the effects of
geometric and process parameters on the drill bit’s
transverse vibrations and frequency, and boundary condi-
tions at the drill tip. In [9] they included the effects of
rotary inertia and gyroscopic moments. Rincon and Ulsoy
[10] calculated the fluctuations in torque and thrust as a
result of imposed lateral vibrations, but did not include the
regenerative effects of the cutting process.
Ema et al. [11,12] experimentally studied the influence of
the chisel edge and drill geometrical parameters on drill
whirling vibrations. They concluded that the chisel edge
acts as a vibration damper, but that the drill flank can also
dampen whirling motion, depending on its geometry. In
[13] Ema et al. proposed a numerical simulation which
1457
considered regenerative forces as well as rubbing forces on
the clearance faces, and they predicted whirling motion
consistent with experiments.
In [14,15] Bayly et al. developed a quasi-static model of
reaming in which they explained the formation of reamed
holes with N+1 or N1 sides for a reamer with N teeth.
Bayly et al. neglected the mass and damping terms in the
equations of motion on the assumption of low spindle
speeds, but included the regenerative forces. In [16] Bayly
et al. found backward whirling modes in drilling at odd
integers of the spindle frequency by modeling regenerative
cutting and rubbing forces. These modes would become
unstable as rubbing and cutting forces increased relative to
drill stiffness, and showed qualitative agreement with
experimental measurements. Pirtini and Lazoglu [17]
predicted oval hole shapes as the result of lip height error
on short drill bits. The predicted hole shape would have
been round as found in experiment, if the unequal tool
stiffnesses had rotated with the tool. Gupta et al. used a
mechanistic cutting force model that takes drill alignment
errors into account to predict the tool deflection resulting
from radial forces. Their model did not take regenerative
forces into account. Gupta et al. [18] imposed the
wandering motion of Lee et al. [3] to generate three-sided
hole shapes, the amplitude of the wandering motion being
dependent on the drill deflection due the lateral forces.
After the full engagement of the cutting lips, the wandering
motion was switched off, and the hole gradually became
round due to interaction with the wall surface. Forces
arising from wall contact were modeled by Jalisi et al. [24]
The prediction of the average diameter along the hole by
Gupta et al. [19] agreed well with experiment.
Bayly et al. [20] derived a bending chatter stability
solution for drilling holes with large pilot holes, in which
the drill tip was free to move. Arvajeh and Ismail [21,22]
modeled bending chatter stability in drilling by assuming
the drill tip is pin supported, as proposed by Ema et al. [23]
providing for a chip thickness regeneration mechanism.
They showed with experiments that a drill with length to
diameter ratio of 20 (L/D ¼ 20) would vibrate in torsio-
nal–axial mode at certain speeds, but in bending at other
speeds.
Although considerable research has been conducted in
the area of lateral drill vibrations, the initiation of whirling
vibrations and the interaction between lateral drilling
forces and tool deflections remain unexplained. A variety
of lateral vibration models for drilling have been devel-
oped, but none of these take the torsional–axial vibrations
into account, although these are predominant in drilling
operations [1].
This paper presents a model that allows for regenerative
vibrations in lateral, axial and torsional directions. The
model covers the drilling of piloted holes, and tool grinding
errors and misalignment are included. 3D cutting force
prediction is presented and experimentally verified. The
numerical treatment of lateral vibrations and surface
generation on both bottom hole and wall surfaces are
 ARTICLE IN PRESS
1458 J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473
explained. The physics of lateral, whirling vibrations in
drilling are discussed, and using the proposed model the
cutting forces and hole profile generation due to commonly
observed whirling vibrations are investigated. The model is
also used to predict lateral chatter, and combined lateral
and torsional–axial vibrations. The interaction between
lateral and torsional–axial vibrations and forces is studied
in both simulation and experiment.
3. 3D dynamic model of the drilling process
The drill penetrates into the work material as it rotates,
which defines its rigid body motion. The cutting forces
excite the flexible drill body, and cause lateral, torsional
and axial vibrations. This paper presents a complete
kinematic model which combines rigid body motion and
dynamic displacements of the drill as both mechanisms are
affected by the drill geometry, grinding errors and the
cutting process.
3.1. Cutting force model
The drill experiences tangential, radial, axial forces and
torque due to the cutting process. Since the cutting edge
geometry, such as rake and clearance angles of the drill
vary along the lips, the cutting force coefficients vary as
well, both as a function of edge location from the drill
center and chip thickness. There has been significant efforts
in modeling the prediction of cutting forces and torque in
drilling either using orthogonal to oblique transformation
by Armarego et al. [25] or mechanistic approaches,
Chandrasekharan et al. [26]. The oblique transformation
models require a material data base which consists of shear
angle, shear stress, and average friction constants measured
in orthogonal cutting tests, which are transformed into 3D
oblique cutting models using the tool geometry. The
mechanistic approach requires experimental calibration of
cutting force constants as a function of chip thickness and
cutting edge geometry. The details of such approaches can
be found in metal cutting text books and past articles as
cited. The authors presented a detailed model of both
methods applied to drilling in the torsional–axial dynamic
of drilling in [27] hence the information is not repeated
here.
The tangential force Ft is perpendicular to the cutting
edge and the radial force Fr acts along the flute, and the
axial force (Fz) is aligned with the drill axis (see Fig. 2). The
cutting edge width is divided into m small differential
segments (Db), and the tangential, radial, axial forces and
torque Tc for each flute (i) are expressed as a function of
corresponding chip thickness (h) and width (Db):
P P
F ti ¼ kct ðr; hÞ Db h; F ri ¼ kcr ðr; hÞ Db h;
elements elements
P P
F zi ¼ kcz ðr; hÞ Db h; T c ¼ kcp ðr; hÞ Db hr:
elements elements
(1)
Y
Ft2
Fr2
Fz2
2
X
Ω 1
Fz1 Ωt
Fr1
Ft1
Fig. 2. Radial and tangential forces acting on drill flutes.
The tangential kct(r, h), radial kcr(r, h), thrust kcz(r, h),
cutting force and torque kcp(r, h), coefficients are experi-
mentally calibrated as a function of chip thickness (h) and
radial location (r) of the edge segment from the drill center
and chip thickness of the differential element. The
identified cutting force coefficients are given in Table 1.
The total forces on the flute are evaluated by digital
summation of elemental forces distributed along the
cutting lip which is in cut.
Drill radial runout rr is defined as the deviation from the
spindle axis in the direction of the cutting lips, and
tangential runout rt as deviation perpendicular to the
cutting lips (see Fig. 3a). The difference in lip height rz is
called axial runout (Fig. 3b). These runout values change
the locations of the cutting edges in 3D space, and
hence the value of (r). As a result of axial runout, the
Z-coordinates of one cutting flute are increased by the
runout amount. A lip angle grinding error (Fig. 3c) is
similar to a lip height error, but the lip height difference
varies linearly along the cutting edge. Drill bending
vibrations lead to unequal chips being cut by the flutes,
resulting in force unbalances in lateral directions. The
runouts and grinding errors are all integrated into the
numerical time domain model of the drilling process as
explained in the following sections (Fig. 4).
The total tangential and radial forces for lip i (i ¼ 1, 2),
which rotate with the tool at spindle speed O (rad/s), are
found by summing the product of specific force and uncut
chip area (Db  h) over the m elements along the cutting
lips. The lateral forces (Fx, Fy) in the stationary global
coordinate system are obtained by projecting the resultants
of the radial and tangential forces (Fr, Ft) which rotate with
the tool (see Fig. 2):
( ) " #( )
F x ðtÞ SinðOtÞ cosðOtÞ F t1  F t2
¼ . (2)
F y ðtÞ cosðOtÞ SinðOtÞ F r1 þ F r2
The predicted and experimentally measured forces with
one and two fluted drills with identical cutting edge
geometries are given in Fig. 5, and demonstrate the validity
of the force model used in the following dynamic analysis
 ARTICLE IN PRESS
J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473 1459

Table 1
Experimentally identified cutting force equations for a drill with diameter ¼ 16 mm in AL7050-T7451; Guehring ]217 drill geometry, TiAlN coating

kct ðr; hÞ ¼ 1449  ð1  2:77h þ 3:85h2 Þð1 þ 0:228r  4:78  102 r2 þ 2:63  103 r3 Þ
kcr ðr; hÞ ¼ 523:1  ð1  1:49h þ 37:7h2 Þð1 þ 0:242r  4:86  102 r2 þ 2:67  103 r3 Þ
kcz ðr; hÞ ¼ 2028  ð1  8:21h þ 21:8h2 Þð1  0:166r  4:66  103 r2 þ 1:68  103 r3 Þ
kcp ðr; hÞ ¼ 5228  ð1  7:24h þ 17:8h2 Þð1  0:270r  3:51  102 r2 þ 1:67  103 r3 Þ

a a
500
0.6mm/rev

Radial Force [N]

400

rr 300

200 0.1mm/rev

100

0
rt
1900
0.6mm/rev Experiment f=0.1mm/rev

Tangential Force [N]

spindle center Experiment f=0.6mm/rev
1425 Mechanistic model

b c 950

475 0.1mm/rev

rz Δκt 0

b
800 0.6mm/rev

Fig. 3. Definition of: (a) drill runout, (b) lip height and (c) lip angle errors.
Thrust [N]

600

400 0.1mm/rev
of the drilling process. Further details of the cutting force 200
model can be found in [27].
0
2000
0.6mm/rev
3.2. Structural dynamic model of the drill
Torque [Ncm]

1500

The general equations of motion for the dynamic drilling 1000

system can be formulated in the stationary frame as
500 0.1mm/rev
follows:
8 9 8 9 8 9 8 9 0
>
> x€ c ðtÞ >
> >
> x_ c ðtÞ >
> >
> xc ðtÞ >
> >
> F x ðtÞ >
> 0 2 4 6 8 10 12 14 16
>
> >
> >
> >
> >
> >
> > > >
>
< y€ c ðtÞ = < y_ c ðtÞ = < yc ðtÞ = < F y ðtÞ =
Pilot hole diameter [mm]

½M  þ ½C  þ ½K  ¼ , Fig. 4. Comparison of measured and predicted drilling: (a) lateral forces

> z€c ðtÞ >
> > > z_c ðtÞ >
> > > zc ðtÞ >
> > > > F z ðtÞ >
>
>
> > > > > > > >
: y€ c ðtÞ >
; : y_ c ðtÞ >
> ; : yc ðtÞ >
> ; > : T c ðtÞ >
; on a single fluted 16 mm drill bit; (b) thrust and torque on 16 mm drill with
two flutes [27], using the mechanistic model in Table 1; Tool: Guehring
16 mm HSS twist drill, TiAIN-coated, geometry #217; workpiece material:
(3)
AL7050-T74515.
where (xc, yc) denote the lateral, (zc) the axial deflections of
the drill in the global coordinate system, as illustrated in
Fig. 5. evaluated from the cutting force model. The dynamic
(yc) is the torsional deflection of the drill bit itself, with properties for the drill used in this research are summarized
respect to the rigid body spindle motion. The rotation in Table 2, detailed explanation on how these were
speed of the tool is O in (rad/s). The matrices [M], [C] and obtained can be found in [1]. The drill’s frequency response
[K] contain the lumped mass, damping and stiffness function was measured with impact tests at its free end
characteristics reflected at the drill tip, respectively. The when it was mounted on the spindle. The torsional and
cutting loads acting on the drill include two lateral forces axial modes are coupled due to the twisted shape of the
(Fx, Fy), thrust force (Fz), and torque (Tc), which are drill, and as a result they have the same natural frequency.
 ARTICLE IN PRESS
1460 J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473

natural modes of the drill bit in that direction:

Y
cutting lip chisel edge X
km
kz
ky qðtÞ ¼ DqðtÞ. (6)
i¼1
X
Ω
The cutting force system (Fx, Fy, Fz and Tc) is coupled
Z kθ through the shared chip thickness h. Since the tool
kx
deflection in any direction changes the chip thickness
distribution, it changes the cutting forces in all directions,
b X
which in turn affects the vibrations in all four directions.
zc Y
The fundamental coupling term is the dynamic chip
thickness distribution along the cutting edges, which varies
with time as a function of vibrations and tool geometry
deformed drill geometry
errors.
θc
yc 4. Time domain simulation model
xc

The fundamental contribution of this paper is to

Fig. 5. Dynamic model of drill bit: (a) view of drill form shank; and (b) integrate cutting forces, rigid body motion and vibrations
drill bit deflections.
of the drill into a general numerical model which allows
virtual testing of various drilling scenarios in time domain.
While only torsional and axial motions were presented in
Table 2 [1], vibrations in one rotational and three translational
Dynamic properties of drill bit; length ¼ 176 mm; diameter ¼ 16 mm (L/D directions are included here along with drill grinding errors
ratio ¼ 11) [1] and tool misalignment. The numerical simulation starts by
Mode Frequency (Hz) Stiffness Damping, z (%) creating a grid map of the flat surface where the hole will be
drilled. The grid mesh starts from the boundary of the
XX 363 1.187  105 (N/m) 0.32 circular pilot hole. The drill geometry contains twisted,
YY 339 1.053  105 (N/m) 0.44
helical flutes with a tapered shaft diameter, lips with
kZF z 3357 5.277  107 (N/m) 0.29
kyT c 3357 3.889  102 (N m/rad) 0.29 varying rake and clearance angles, grinding errors and
kZT c 3357 2.168  105 (N m/m) 0.29 runouts. In short, the 3D geometric model of the drill with
kyF z 3357 2.462  105 (N/rad) 0.29 dimensional errors are mathematically modeled.
The drilling process starts by aligning the spindle axis
with the center of the pilot hole, and the drill tip is
positioned at the zero height of the workpiece surface. As
The dynamic deflections of the drill bit are determined the drill rotates and penetrates into material, the cutting
from the structural model that is excited by the cutting edge and work material intersections are calculated at
forces. The general transfer function of the drill structure in discrete time intervals. The coordinates of grid points are
the direction (q) at the tip of the drill is obtained in the updated at each discrete time interval as the material is
Laplace domain as removed. The location of the cutting edge points are
evaluated by considering both rigid body or drilling
qðsÞ X
km
o2ni =ki
FðsÞ ¼ ¼ ; motion, and the three lateral and torsional vibrations. As
F q ðsÞ i¼1
s2 þ 2zi oni s þ o2ni a result, the chip thickness is calculated by subtracting the
ðq ¼ xc ; yc ; zc ; yc ; ðF q ; F x ; F y ; F z ; T c ÞÞ, ð4Þ present location of the cutting edge coordinates from the
previously generated hole surface, hence the true dynamic
where (km) is the total number of modes in that direction of chip load distribution as well as the time varying hole
the system. (oni), (ki) and zi are the natural frequency, surface are predicted. The dynamic chip thickness is used in
modal stiffness and damping, respectively. evaluating the cutting force expressions in Table 1, which
The transfer function is converted into observable state- are applied to the dynamics of the drilling system given in
space canonical form for each mode as Eq. (3). The displacements are solved from Eq. (4) using
( ) " #( ) " 2 # the structural modal parameters of the drill in discrete time
q_ 1i ðtÞ 0 o2ni q1i ðtÞ oni =ki
¼ þ fF q ðtÞg intervals by applying a fourth-order Runge–Kutta digital
q_ 2i ðtÞ 1 2zi oni q2i ðtÞ 0 integration method. Besides the tool geometry errors, the
model allows for nonlinear variation of the cutting forces
qðtÞ ¼ q2 ðtÞ. ð5Þ
along the lips, chip size effects, and the tool loosing the
The total displacement (q) in any four direction is contact with the material due to excessive vibrations.
evaluated by summing the vibrations contributed by all Although the model represents a comprehensive, combined
 ARTICLE IN PRESS
J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473 1461

rigid body and structural dynamic model of the drill, the

numerical model complexity brings lengthy computation points for
times. previous interpolation previous surface
tooth path current surface
At the bottom of the hole, each drill tooth generates a interpolated point
surface that consists of points arranged in evenly spaced
tooth
circles. The hole is flat in the beginning. The drill starts
rotating while the workpiece surface is fed to the drill. The
hole diameter gradually increases until it becomes steady chip height t
tooth, 1 timestep
state and equal to the hole diameter. For each integration earlier
time step, two modifications are made to update the
new point on
surface. First, the rigid body feeding motion in positive current surface current tooth path
Z-direction is accounted for by adding a feed increment to
every entry of matrices (Wz, 1) and (Wz, 2), which contains b
the Z-coordinates of the workpiece surfaces. The feed tooth, 1 timestep
increment equals earlier
new point on tooth
current surface
f
Df r ¼ r , (7) previous tooth path
Ng

where fr is the feedrate in (mm/rev). Secondly, the cutting

action of the drill is accounted for on matrices (Wx, 1),
(Wy, 1), (Wz, 1) and (Wx, 2), (Wy, 2), (Wz, 2). The cutting edge
points will not necessarily coincide with points on the
initially equally spaced workpiece grid once lateral and/or
torsional deflections develop due to tool loading. The
simulation finds the points on the previously generated
surface ((Wx, 2), (Wy, 2), (Wz, 2) for flute 1) that are close to
the current cutting edge position: one point ahead of the
cutting edge, one behind. The index of these points in the
surface matrix is estimated based on the time step (k):
k0 f ; 1 ¼ k  N r N g fluteð1Þ
where (Nr) is the integer number of revolutions that has
passed in the simulation. The surface matrix index for
interpolation, (kf, 1), is exactly identified through iteration.
The height of the workpiece at the location of the cutting
edge is then calculated by linear interpolation of the height
of the two workpiece points. For each point on the cutting
edge ~ zt;1 ði; kÞ, the Z-coordinate is compared with the
corresponding interpolated Z-coordinate of the workpiece
surface point, Wz,20 (i,kf,1), at the exact location of the
cutting edge point. If ~ zt;1 ði; kÞoW 0 z;2 ði; kf ;1 Þ, tooth 1 is in
cut. A new point is generated on workpiece surface (Wz,1)
using the current coordinates of the cutting edge point
(xt,1(i,k), yt,1(i,k), zt,1(i,k)) and the chip height t1(i,k) is
determined:
W x;1 ði; kf ;1 Þ ¼ xt;1 ði; kÞ; W y;1 ði; kf ;1 Þ ¼ yt;1 ði; kÞ,
W z;1 ði; kf ;1 Þ ¼ zt;1 ði; kÞ,
t1 ði; kÞ ¼ W 0 z;2 ði; kf ;1 Þ  zt;1 ði; kÞ.
The workpiece updating procedure is shown schemati-
zt;1 ði; kÞ4W 0 z;2 ði; kf ;1 Þ, the tooth is out of
cally in Fig. 6. If ~
cut, due to excessive vibrations or because it has not
engaged with the workpiece yet. The Z-coordinate of the
new workpiece point is obtained from the interpolated
height of the previous surface, and the chip height is set to
current tooth path
Fig. 6. Exact kinematics approach for surface updating: (a) tooth in cut,
regular cutting and (b) tooth out of cut, due to excessive vibrations.
zero:
W x;1 ði; kf ;1 Þ ¼ xt;1 ði; kÞ; W y;1 ði; kf ;1 Þ ¼ yt;1 ði; kÞ;
(10)
W z;1 ði; kf ;1 Þ ¼ W 0 z;2 ði; kf ;1 Þ; t1 ði; kÞ ¼ 0:
(8)
This case is illustrated in Fig. 6(b). The same procedure
is followed for flute 2. The two modifications to the surface
vector (feed motion, cutting action) completely define any
change to the workpiece surface.
The chip thickness is evaluated at the (m+1) cutting
edge points, for which the chip height t1(i,k) is determined.
The chip thickness of element (j) is then determined by the
average of the chip heights on the sides of the element
h1 ðj; kÞ ¼ ft1 ðj; kÞ þ t1 j þ 1; kÞg=2 ðj ¼ 1 . . . mÞ. (11)
The elemental chip thickness distribution is used in
Eq. (1) to calculate the cutting forces acting on the drill bit
system. Eq. (1) also requires the width of cut distribution,
Db, for each flute. At time step k, the intersection points for
tooth 1 (xi,1(i,k), yi,1(i,k)) and tooth 2 with the grid circles
are determined as illustrated in Fig. 7. Due to lateral
deflections of the drill bit, the distance between any two
adjacent intersection points ( ¼ elemental width of cut) on
ð9Þ a cutting edge varies over time (e.g. db1,1 and db2,1). The
peripheral points of the drill move in and out of the largest
grid circle, and the width of cut of each peripheral element
equals the distance between the drill peripheral point and
the intersection on the second largest grid circle (db1,7 and
db2,7). Using this approach, the total width of cut for each
cutting edge is continuously variable. For each of the
intersection points with the grid circles, the height of the
 ARTICLE IN PRESS
1462 J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473

db2,7
cutting edge peripheral point
intersection point
a b
db2,1

b2

xi,1(2,k), yi,1(2,k)

db1,1
db1,7
grid circles b1

Fig. 9. Example of hole wall and bottom surfaces: (a) start of simulation
Fig. 7. Intersection points of cutting edges with grid circles; number of and (b) end of simulation.
elements along cutting edge m ¼ 7.

toolholder

D-bt

L
β0 Lf

surface being cut by flute 2 surface being cut by flute 1

Δx(zd)
intersection points of cutting edge with grid circles

Fig. 8. Tooth surfaces used for chip thickness calculation.

hlip

cutting edge at that location, zi,1(i,k), is calculated based on
the tool geometry and tool vibrations, as explained above.
As the drill tip can vibrate freely in the XY-plane, the
cutter can move backwards as well. In the simulation, the
drill cannot cut while moving backwards, as the relief angle
acts as a highly negative rake angle. Regardless of whether
the tool moves forwards or backwards, the chip thickness is
taken as the distance between the cutting edge location
and the surface left by the previous tooth, as illustrated in
Fig. 8. Material removal is not allowed when the tool
moves back, and the new point generated by the flute in
this situation will simply lie on top of the recently
generated surface.
4.1. Hole wall formation
The hole wall surface is modeled by a number of circles
with an evenly spread, fixed number of points that form a
radial Z-buffer. The angular locations of the grid points are
fixed, only the radial location of each point can change.
These circles provide cross sections of the machined hole at
Fx
Δx
D Z
Fig. 10. Interpolation of tool deflection along the flute length.
axial increments, that can be compared with measurements
from a hole measurement instrument. Fig. 9 provides a 3D
view of the finished bottom workpiece and hole wall
surfaces, at the beginning of the simulation (a), and at the
end of the simulation (b). In the hole wall formation
process, forces arising from the interaction of the flutes
with the wall are neglected for simplicity. The cylindrical
wall surface is fed towards the drill simultaneously with the
bottom hole surface grid.
The drill deflections calculated in Eq. (4) govern the
location of the flute in 3D space, as illustrated in Fig. 10.
The tool deflections xc(t), yc(t) are interpolated by
assuming the tool deflects as a simply supported beam,
clamped in the tool holder. The drill has a back taper which
 ARTICLE IN PRESS
J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473 1463

flute point on wall grid circle, current timestep ðL  zd Þ2 ð2L þ zd Þ

flute point on wall grid circle, previous timestep þ yc ðkDtÞ
2L3
L  zd
current timestep þ ðrr sinðOk DtÞ þ rt cosðOkDtÞ. ð12Þ
L  hlip

previous timestep 5. Simulation of drilling cases

The proposed comprehensive model allows the simula-

wall grid initial point tion of the drilling process with runouts, misalignment
wall grid updated point errors, with various drill geometries and the structural
intended hole profile
actual hole profile
vibrations in three lateral and the torsional directions.
Some of the scenarios, which illustrate the source of hole
Fig. 11. Mechanism of wall surface updating; points on the radial grid are errors and vibration mechanisms, are presented in the
pushed outwards. following.

reduces the fluted diameter by an amount bt ¼ 52 mm along

flute length Lf ¼ 149 mm, to provide clearance as the drill 5.1. Drill misalignment and grinding errors
penetrates the work piece. The back taper is included as a
linear diameter reduction. Finally, the drill runout rr, rt are Fig. 12 shows the simulated chip thickness, thrust and
included, which are assumed to affect the coordinates of lateral forces for a rigid drill with lip height error of
the intersection points linearly, which means that the rz ¼ 10 mm, which leads to chip unbalance between the
runout at the tool holder is zero. flutes, i.e. one flute cutting 10 mm more, the other flute
At each time step, the (x, y) location of the flute on each cutting 10 mm less. The corresponding unbalance force
layer of the grid is determined. Because the flute location and grows during engagement, but remains constant after full
grid points do not coincide, the flute radius at the grid points engagement (after 0.34 s in this example). The force rotates
is interpolated based on the flute intersection point at the with the tool, and can be identified from a single peak in
current and the previous time steps. If the flute radius at a the Fourier spectrum at the spindle frequency fs. In case
point is greater than the workpiece radius at that location, one tooth has a different lip angle, the elements from both
the radial Z-buffer point is updated with the flute radius lips close to the chisel edge cut almost the same chip
(flute in cut), otherwise it is left unchanged. Fig. 11 illustrates thickness, but the elements close to the drill periphery show
this process, where one grid point is passed between the a larger difference in chip thickness.
consecutive time steps. The input for calculating the flute The same drilling process is simulated with flexibility in
intersection points is the height of the whirling grid layer, zd, X and Y directions with 300% damping to obtain a quasi-
which is measured from z ¼ 0. The coordinates of the static response. The torsional and axial modes are rigid.
intersection point depend on the helix angle b0 ¼ 301, back The results in Fig. 13 show that as the tool engages, the
taper bt, tool runout and tool vibrations. hlip is the distance drill quickly deflects and the chip areas on each flute
from z ¼ 0 (drill tip) to the start of the flutes. The web angle become almost identical, resulting in an lateral unbalance
at the periphery of the drill is o0 ¼ sin1(2W/D) and 2W is force of only 1 N (the force history is not shown but is
the web thickness of the drill. The intersection points of the similar in shape to the deflection history). This force is
flute with the wall grid circle at height zd and time step k are much smaller than the 92 N for a rigid tool. The steady-
analytically expressed by state tool motion is circular in the direction of tool
rotation, with an amplitude of about 8 mm, and this is not a
 
1 zd  hlip whirling motion. The simulation predicts a 15 mm hole
xfi;1 ðk; zd Þ ¼ D  bt oversize. The detail graphs of the chip thickness show that
2 Lf
  during two revolutions of the tool the chip thickness varies
2 tan b0
 cos ðzd  hlip Þ  Ok Dt  o0 four times—twice per spindle revolution, due to alternating
D
excitation of the X and Y directions, which have different
ðL  zd Þ2 ð2L þ zd Þ stiffnesses. Galloway established an empirical relationship
þ xc ðkDtÞ
2L3 for the hole oversize and deflected axis position due to lip
L  zd height error rz. Half the tip angle is kt ¼ 591 for the drill
þ ðrr cosðOk DtÞ þ rt sinðOkDtÞÞ
L  hlip bit. According to Galloway [2] the tool deflection should be
  0.5rz tan kt ¼ 8.3 mm and the hole oversize rz tan kt ¼
1 zd  hlip
yfi;1 ðk; zd Þ ¼ D  bt 16.6 mm. These numbers agree favorably with the simula-
2 Lf
  tion presented here. In this expression, the tool deflection
2 tan b0
 sin ðzd  hlip Þ  Ok Dt  o0 and hole oversize depend on the tool grinding error, and
D not on the lateral tool stiffness, which is typically very low.
 ARTICLE IN PRESS
1464 J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473

x 104
6
100

Lateral force [N]

5 1.0fs
50 4
0 3

-50 2
1
-100
0
0 0.2 0.4 0.6 0 2fs 4fs 6fs 8fs
Time [s] Frequency

0.18
Chip thickness [mm]

0.15
0.12 flute 1 flute 2

element at
element at
periphery
pilot hole

0 0.2 0.4 0.6 0 0.2 0.4 0.6

Time [s] Time [s]

Fig. 12. Simulated chip thickness, thrust force and lateral forces and Fourier spectra for a drill with lip height error rz ¼ 10 mm; conditions: 2400 rpm,
fr ¼ 0.30 mm/rev, pilot hole diameter ¼ 4 mm, hole depth ¼ 15 mm.

5.2. Whirling vibrations minus sign, the whirl is backwards-counterclockwise. A

It has been experimentally observed by several authors
[3,5,6] that drills whirl at odd integer multiples of the drill
spindle frequency fs, as measured in the frame rotating with
the drill. This whirling motion consists of the deflected drill
center tracing a circular or elliptical path in the opposite
direction of the drill rotation. The whirling frequency in the
fixed frame f fw can be obtained from f fw ¼ f rw  f s ; where
f fw is the whirling frequency in the rotating frame, and the
minus sign applies to backward (regressive) whirling modes
[15, p. 390]. The physics of whirling vibrations in drilling
can be explained as follows. A tool deflection in the
negative X direction on a drill with identical flutes results in
an increased cutting force on flute 2, F2+DF, and a
decreased cutting force on flute 1, F1DF (see Fig. 14(a)).
These two forces can be replaced by a single unbalance
force through the drill center, 2DF and the torque
Tc+2R DF (Fig. 14(b)). This force initiates a counter-
clockwise motion-backward whirling. When the drill bit
contacts the wall of the hole being cut, a contact force Fwi
will arise (Fig. 14(c)), which promotes a backward whirling
motion as well. If we assume the drill is executing a circular
whirling motion in the fixed frame, the tool center position
is defined by
 
xc ðtÞ ¼ e sin f fw t
 ; f fw t ¼ 2nf s ;
yc ðtÞ ¼ e cos f fw t
where e is the amplitude of the whirling motion, fs is the
spindle frequency, and n is a positive integer. Due to the
backward whirling motion as in Eq. (13) can be seen as a
deflected drill motion restrained by the round shape of the
hole.
There is no accurate mechanics model to predict the
initiation of whirling in drilling, since it requires the
modeling of contact mechanics between the clearance face
of the tool and wavy hole surface which is an unresolved
problem. However, the mechanism of whirling is illustrated
by imposing a whirling motion in the proposed drilling
simulation algorithm. The intended hole has a diameter of
16 mm, and a whirling amplitude of e ¼ 500 mm is imposed
for visualization.
The location of the drill peripheral points are governed
by
 
xp;1 ðtÞ ¼ L1 cosðOtÞ  W sinðOtÞ þ e sin f fw t ;
 
xp;2 ðtÞ ¼ L1 cosðOtÞ þ W sinðOtÞ þ e sin f fw t ;
  (14)
yp;1 ðtÞ ¼ L1 sinðOtÞ  W cosðOtÞ þ e cos f fw t ;
 
yp;2 ðtÞ ¼ L1 sinðOtÞ þ W cosðOtÞ þ e cos f fw t ;
where L1 ¼ D/2  cos{sin1(2W/D)} depends on drill
diameter D ¼ 16.00 mm and web thickness 2W ¼ 2.12 mm
(see Fig. 14d).
Figure shows the traces of the drill peripheral points for
backward whirls at two, four and six times the spindle
n ¼ 1; 2; . . . , (13) frequency. Three, five and seven sided hole shapes can
clearly be seen. After half a tool revolution (1801), the
cutting edge peripheries coincide with the profile generated
by the previous flute—the tool center is at the same
location as it was at 01 tool rotation.
 ARTICLE IN PRESS
J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473 1465

a a b
300 Y Y
X,Y Deflection [μm]

10
F2+ΔF Ω R Ω
2ΔF
0
1 X 1 X
-10 2 2
0
0 0.1 0.2 0.3 0.4 0.5 0 40 200 400
Tc+2RΔF
Time [s] Frequency [Hz] F1-ΔF

b
10 c
Y
Ω
Y-axis [micron]

5 Fwi

0
1 X

-5 2

-10
-10 -5 0 5 10
X-axis [micron]
d
c
D
Chip thickness 1 [μm]

200
151
150 element at
W
periphery
100 flute 1 150
element at
pilot hole L1
50
149 Fig. 14. (a)–(c) Physics of whirling vibrations in drilling, drill seen from
0
the shank side; and (d) detailed drill geometry.
Chip thickness 2 [μm]

200
151
150
100 flute 2 150
50
149
0
0 0.1 0.2 0.3 0.4 0.5
Time [s]
Fig. 13. Simulation for drill with 10 mm lip height error on flute 1, and
lateral flexibility with a damping of 300%: (a) tool deflection and Fourier
spectrum; (b) tool trajectory; (c) chip thickness for each flute, and details;
Conditions: 2400 rpm, fr ¼ 0.30 mm/rev, pilot hole diameter ¼ 4 mm, hole
depth ¼ 6 mm.
5.2.1. Cutting forces during whirling vibrations
Commonly observed whirling motions were imposed on
the drill bit to investigate the chip thickness, thrust force
and lateral forces as a function of time. The simulation
results are summarized in Table 3, and shown in Fig. 16.
A backward whirling motion at f rw ¼ 3:00f s (in the
rotating frame) is achieved by imposing a circular tool
vibration (Eq. (13)) in the fixed frame using n ¼ 1 and
whirling amplitude e ¼ 100 mm. In Fig. 16a, the elemental
element at
chip thickness fluctuates during one drill revolution, but
pilot hole remains constant in the following as the wave left on the
surface is in phase with the tool motion. This phenomenon
element at only occurs when the whirling frequency in the rotating
periphery frame is an odd integer of the spindle frequency fs. The
width of cut on each flute (b1, b2, Fig. 7) varies from the
0.45 0.50 very beginning at 3.00fs, and as a result, the tangential and
Time [s]
radial forces in the frame rotating with the tool vary with
the same frequency, 3.00fs. As shown in Fig. 15a the
resulting hole shape is three sided. The spectrum of the
lateral forces in the stationary frame shows peaks at 2fs and
at 4fs. The occurrence of two peaks, rather than one, can be
explained using the common trigonometric product-to-sum
identity:
sinð3f s 2ptÞsinðf s 2ptÞ
1
¼ fcosð6pf s t  2pf s tÞ  cosð6pf s t þ 2pf s tÞg
2
1
¼ fcosð2f s 2ptÞ  cosð4f s 2ptÞg, ð15Þ
2
where sin(fs 2pt) represents term sin(Ot) in Eq. (2pfs ¼ O,
the speed of the frame rotating with the tool), and
sin(3fs 2pt) is the force measured in the frame rotating
 ARTICLE IN PRESS
1466 J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473

Table 3
Summary of simulation results of imposed whirling motions with 100 mm amplitude

Imposed whirling frequency in rotating frame (fs ¼ spindle frequency) 3.00fs 2.90fs 4.90fs 6.90fs

Details in figure 15a, 16a 16b, 17 16c 16d

Peaks in chip thickness spectrum (h1) None 2.90fs 4.90fs 6.90fs
Peaks in width of cut spectrum (b1, b2) 3.00fs 2.90fs 4.90fs 6.90fs
Peaks in thrust force spectrum (Fz) None 5.80fs 9.80fs 13.80fs
Peaks in lateral force spectrum (Fx, Fy) 2.00, 4.00 1.90, 3.90 3.90, 5.90 5.90, 7.90

2.90fs+fs ¼ 3.90fs. The chip thickness in Fig. 16b varies

throughout the cut at 2.90fs. The lateral forces fluctuate
fs fs fs strongly compared to Fig. 16a, and grow until the drill has
2 2
fully engaged with the workpiece. The thrust force
2
1 1 1 fluctuates with a small amplitude at 5.80fs because the
f f drill has two flutes. The coupling between lateral vibrations
f w=2fs f
f w=4fs f w= 6 f s and torque and thrust is very weak. In Fig. 16b, lateral
0° intended tool axis forces in the order of 200 N are generated for a 100 mm
trajectory
hole profile whirling vibration while the trust force only fluctuates by
xp,2 ,yp,2 less than 0.1% of its 465 N steady-state average. Figs. 16c
and d show the simulation results for f yw ¼ 4:90f s and
f yw ¼ 6:90f s , respectively. Similar to Fig. 16b the chip
thickness spectra show a peak at f yw , the thrust spectra
show a peak at 2f yw , and the lateral force spectra show
36° xp,1,y p,1 peaks at f yw  f s and f yw þ f s .

5.2.2. Hole wall formation during whirling vibrations

36°
36°
Fig. 15. Traces of the cutting edges during backward whirling motion at
exact integers of the spindle frequency fs at four different tool rotations:
(a) fw ¼ 2fs; (b) fw ¼ 4fs; (c) fw ¼ 6fs; drill diameter ¼ 16 mm, fixed
whirling amplitude ¼ 500 mm (for visualization).
with the tool (at 3fs). The force functions on the right-hand
side represent the forces in the stationary frame. Fig. 17a
shows the tangential and radial forces in the frame rotating
with the drill for the case of Fig. 16b (backward whirling
frequency f rw ¼ 2:90f s ).
The Fourier spectrum of the tangential and radial forces
shows peaks only at 2.90fs. In Fig. 17b (and Fig. 16b) the
lateral forces in the stationary frame are shown, and the
spectrum shows two peaks. Using Eq. (15) the first
peak occurs at 2.90fsfs ¼ 1.90fs and the second at
Fig. 15 showed the trace of the cutting edges during
whirling vibrations. In reality, the flutes may cut the
previously generated surface again, depending on tool
geometry and tool deflections. This process is simulated in
the proposed time domain model. Fig. 18 shows the
simulated cross sections of the hole wall at the top and
bottom of the hole, including the average diameter of the
hole. It can be seen that if the whirling frequency in the
rotating frame is an odd integer of the spindle frequency
(see Eqs. (13) and (14)), the odd sided hole shape is
maintained at the top of the hole, despite being cut again
by parts of the flute away from the tool tip. The shape of
the cross section resulting from repeated cutting action
by the flutes, depends on the drill back taper, the whir-
ling frequency and the amplitude of the whirling vibration.
Fig. 19 shows cross sections for whirling at 7fs for the
amplitudes 1, 2, 5, 10, 20 and 50 mm. For amplitudes up to
5 mm (Figs. 19a–d), the seven-sided shape is very visible at
every level in the hole. For amplitudes 10, 20 and 50 mm the
profile at the top of the hole becomes more round, followed
by the middle section and finally the bottom of the hole.
5.3. Simulation of chatter vibrations
The time domain algorithm allows the evaluation of
regenerative chip thickness by considering present and past
vibrations in lateral, axial and torsional directions. It is
possible to test the vibrations individually, or as a complete
coupled system.
 ARTICLE IN PRESS
J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473 1467

backward whirl at 3.00fs backward whirl at 2.90fs backward whirl at 4.90fs backward whirl at 6.90fs
0.20 0.20 0.20 0.20
h1 [mm]

0.10 0.15 0.10 0.15 0.10 0.15 0.10 0.15

0 0 0 0
0.01 0.05 0 0.01 0.05 0.01 0.05 0 0.01 0.05
0 0
0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0
Time [s] Time [s] Time [s] Time [s]
40 40 40 40
FFT h1 [mm]

2.9fs 4.9fs 6.9fs

20 20 20 20

0 0 0 0
0 2fs 4fs 6fs 8fs 0 2fs 4fs 6fs 8fs 0 2fs 4fs 6fs 8fs 0 2fs 4fs 6fs 8fs
Frequency Frequency Frequency Frequency
6.37 6.37 6.37 6.37
b1
b1,b2 [mm]

b2

6.10 6.10 6.10 6.10

0.96 0.98 1 0.96 0.98 1 0.96 0.98 1 0.96 0.98 1
Time [s] Time [s] Time [s] Time [s]
FFT b1,b2 [mm]

250 250 250 250

3.0fs 2.9fs 4.9fs 6.9fs

0 0 0 0
0 2fs 4fs 6fs 8fs 0 2fs 4fs 6fs 8fs 0 2fs 4fs 6fs 8fs 0 2fs 4fs 6fs 8fs
Frequency Frequency Frequency Frequency
500 500 500 500
466.0 466.6
Fz [N]

465.5 465.25

464.6 464.75 464.5 464.4

0.42 0.46 0.42 0.46 0 0.42 0.46 0.42 0.46
0 0 0
0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0
Time [s] Time [s] Time [s] Time [s]
x103 x103 x103 x103
3 3 3 3
FFT Fz [N]

13.8fs
2 2 5.8fs
2 9.8fs 2

1 1 1 1
0 0 0
0 4fs 8fs 12fs 16fs 0 4fs 8fs 12fs 16fs 0 4fs 8fs 12fs 16fs 0 4fs 8fs 12fs 16fs
Frequency Frequency Frequency Frequency
200 200 200 200
Fx, Fy [N]

0 0 0 0

-200 -200 -200 -200

0 0.5 1.0 0 0.5 1.0 0 0.5 1.0 0 0.5 1.0
Time [s] Time [s] Time [s] Time [s]
x104 x104 x104 x104
FFT Fx, Fy [N]

4 2.0f 4.0fs 1.9fs 3.9fs 3.9fs 5.9fs 5.9fs 7.9fs

s
10 10 10
2 5 5 5
0 0 0 0
0 2fs 4fs 6fs 8fs 0 2fs 4fs 6fs 8fs 0 2fs 4fs 6fs 8fs 0 2fs 4fs 6fs 8fs
Frequency Frequency Frequency Frequency

Fig. 16. Simulated chip thickness, thrust force, lateral forces and Fourier spectra for a drill with imposed backward whirling motion with an amplitude of
100 mm at: (a) 3.00fs; (b) 2.90fs; (c) 4.90fs; (d) 6.90fs (frequency in rotating frame); cutting conditions for all cases: 2400 rpm, fr ¼ 0.30 mm/rev, pilot hole
diameter ¼ 4 mm, hole depth ¼ 12 mm.
 ARTICLE IN PRESS
1468 J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473

Rotating frame Stationary frame

intended hole profile (=tool diameter) simulated hole profile
200
Ft Fx
Fr Fy

Fx, Fy [N]
100
Ft, Fr [N]

0
top top top
-100
16.1207 [mm] 16.1885 [mm] 16.1551 [mm]
-200
0 0.5 0 0.5
Time [s] Time [s]

x105 x104
2.5 15
2.0 bottom bottom bottom
16.0203 [mm] 16.1539 [mm] 16.0363 [mm]
FFT [N]

1.5 2.9fs 10 1.9fs 3.9fs

1.0
5
0.5
0 0 3.00fs 2.90fs 5.00fs
0 fs 2fs 3fs 4fs 5fs 0 fs 2fs 3fs 4fs 5fs
Frequency Frequency
d e f
Fig. 17. (a) Tangential and radial force Ft, Fr in the frame rotating with
the drill bit, and Fourier spectrum; (b) lateral forces Fx, Fy acting on the
drill bit in the stationary frame, and Fourier spectrum.
top top top
16.1878 [mm] 16.1806 [mm] 16.2018 [mm]

5.3.1. Lateral chatter vibrations

The torsional and axial modes are kept rigid in the
simulation in order to investigate lateral, regenerative
chatter vibrations. Fig. 20 presents the simulation result for
a drill with 3 mm radial runout, and identical dynamics in
X, Y directions (339 Hz) with 5% damping ratio and
2  107 N/m stiffness. A higher stiffness and damping are
used to consider the wall contact stiffness and friction
during actual drilling as well as preventing the numerical
instability in digitizing the hole surface. In an actual
drilling process, the drill is restrained from large deflections
due to the hole wall, where interference with the just cut
wall also create process damping. As a result, the drill
will show a stiffer and more damped vibration than a
freely vibrating drill with the structural stiffness and
damping listed in Table 2. The drill is given a small
runout, because in simulation the lateral forces on a perfect
drill would be zero, and it would not deflect. Lateral
chatter occurs at 362 Hz in the lateral deflection spectrum
(Fig. 20a), leaving 9 waves on the surface. There is a
small peak at spindle frequency (fs ¼ 40 Hz) due to the
radial runout. The time domain model does not predict
lateral whirling motions (circular motion of the tool
center in the direction opposite to tool rotation), possibly
due to the absence of hole wall contact forces, which are
difficult to model. These can be included in the model at a
later stage.
5.3.2. Combined lateral and torsional–axial chatter
vibrations
Fig. 21 shows the simulated tool vibrations for a drill
with 3 mm radial runout and which is flexible in torsional,
bottom bottom bottom
16.1414 [mm] 16.1663 [mm] 16.1311 [mm]
4.90fs 7.00fs 6.90fs
Fig. 18. Simulated cross sections of hole wall for six different whirling
frequencies (defined in the rotating frame), at top and bottom of hole: (a)
3.00fs (refer to Fig. 16a); (b) 2.90fs (refer to Fig. 16b); (c) 5.00fs; (d) 4.90fs
(refer to Fig. 16c); (e) 7.00fs; (f) 6.90fs (refer to Fig. 16d); hole
depth ¼ 15 mm.
axial and lateral directions. The drill has identical dynamics
in X, Y directions (339 Hz) with a 5% damping ratio and
2.5  107 N/m stiffness. In torsional–axial directions the
stiffnesses from table are used with 3% damping.
Torsional–axial chatter at 2983 Hz and lateral chatter at
362 Hz develop quickly after full engagement at 0.30 s
(Figs. 21c and d). In the axial frequency spectrum
(Fig. 21c) there is a clear peak at 3kHz—which leaves
about 75 waves on the surface as can be seen in Fig. 21a.
The lateral chatter at 360 Hz (Fig. 21d) leaves about 9
waves on the surface, as is very clear from Fig. 21b, but not
from Fig. 21a. The lateral spectrum also contains a peak at
spindle frequency (fs ¼ 40 Hz), due to the radial runout.
This result clearly shows the capability of the model to
handle combined lateral, torsional and axial vibrations.
 ARTICLE IN PRESS
J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473 1469

intended hole profile (=tool diameter) simulated hole profile

top top top

16.0006 [mm] 16.0021 [mm] 16.0073 [mm]

middle middle middle

16.0007 [mm] 16.0022 [mm] 16.0069 [mm]

b
100

X,Y deflection [μm]

150

bottom bottom bottom

16.0003 [mm] 16.0009 [mm] 16.0025 [mm]
0 50

-150 0
1μm 2μm 5μm
0 0.1 0.2 0.3 0 1 2 3 4 5
Time [s] Frequency [kHz]
d e f
Fig. 20. Simulation of lateral chatter in drilling: (a) bottom workpiece
surface; (b) lateral tool deflection and Fourier spectrum; conditions:
2400 rpm, fr ¼ 0.30 mm/rev, pilot hole diameter ¼ 4 mm;
top top top
16.0163 [mm] 16.0343 [mm] 16.0883 [mm]

torsional–axial chatter (Sections 5.1 and 5.3) the trends in

force signatures and the frequency content are compared to
obtain a better understanding of the drilling process. The
depth of cut (depth of cut ¼ drill radiuspilot hole radius)
middle middle middle
has been varied to investigate its influence on the different
16.0151 [mm] 16.0325 [mm] 16.0887 [mm]
types of vibrations.
Cutting tests were conducted on a Mori Seiki SH403
horizontal machining center, using a Sandvik B100 collet
chuck (ER32 collet, HSK63A interface) and a Guehring
16 mm HSS drill bit with 176 mm stick-out (11xD). Figs. 22
bottom bottom bottom and 23 show time domain measurement data of thrust Fz
16.0053 [mm] 16.0251 [mm] 16.0850 [mm] and lateral forces Fx, Fy and two Fourier spectra of each
force. The first Fourier spectrum shows frequency compo-
nents up to 5000 Hz, the second spectrum shows the
10μm 20μm 50μm
frequency content up to ten times the spindle frequency fs.
Drilling tests were conducted at four depths of cut: full
hole, 4 mm pilot hole, 8 mm pilot hole and 12 mm pilot
Fig. 19. Whirling cross sections as a function of whirling amplitude; hole. For each depth of cut, a low and high spindle speed
Whirling frequency is 7.00fs in the rotating frame: (a)1 mm; (b)2 mm;
(c)5 mm; (d)10 mm; (e)20 mm; (f)50 mm; hole depth ¼ 15 mm.
were used. The feedrate was 0.30 mm/rev in all cases.
The low spindle speed range was chosen such that no
torsional–axial chatter occurred. The high spindle speed
was chosen such that torsional–axial chatter just
6. Experimental results started to occur. For this purpose, the spindle speed was
increased in steps of 200 rpm until the speed was found that
Based on the simulation results of imposed whirling resulted in torsional–axial chatter. No lateral chatter has
vibrations (Section 5.2), lip height error, lateral and been identified from the experiments, which covered
 ARTICLE IN PRESS
1470 J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473

b frequency may be slightly lower, as in Fig. 16b. The
Cross-sectional surface profile at pilot hole
c When a 4 mm pilot hole is used at 400 rpm (Fig. 22c),
0 6
Axial deflection [μm]
(see Fig. 12). In the case of a full hole, this peak is smallest.
At the higher spindle speeds, both thrust and lateral force
spectra show strong peaks near the torsional–axial chatter
frequency fc, around 3300 Hz, leaving a sunray pattern at
the bottom of the hole, as can be seen in the photographs
of the experiments, and was accurately predicted before [1].
The relative size of the peak at fs does not appear to be
affected by spindle speed or torsional–axial chatter.
The full hole, 8 and 12 mm pilot hole results (Figs. 22a, b
and 23 ) show a strong peak at 2fs and a smaller one at 4fs
in the lateral force spectra. These peaks indicate that the
drill whirls at 3fs in the rotating frame (see Fig. 16a),
leaving a three sided hole—as can be seen in the
photographs (Figs. 22a and b). The peaks in the thrust
force spectrum close to 6fs indicate that the whirling
whirling motion is not affected by spindle speed or
torsional–axial chatter either. For a full hole at 400 rpm,
the lateral forces slowly develop after full engagement and
are sustained. At 2400 rpm, the lateral forces quickly
become much larger once the torsional–axial chatter
develops from 1.0 s, which can be seen from the thrust
time domain data (similar to Figs. 22a and b)—and this
shows the coupling between the lateral forces and the
vibrations in torsional–axial directions.
strong peaks occur at 6fs and 8fs but also at 2fs and 4fs in

the lateral force spectrum, indicating that there are

whirling contributions in the rotating frame at 3fs, 7fs
-20 and possibly 5fs. At 2400 rpm (Fig. 22d), the peaks at 2fs
and 4fs are relatively stronger. Due to the pilot hole, no
polygon can be distinguished in the chisel edge region. The
-40
0 lateral force spectra also shows small peaks around
0 0.1 0.2 0.3 0 1 2 3 4 5
1430 Hz, however, no clear lateral chatter could be
Time [s] Frequency [kHz]
observed—the surface does not show marks that indicate
d this frequency. The small pilot hole size promotes higher
25
frequency whirls. When cutting a full hole, lateral
25
X,Y deflection [μm]

deflections are inhibited by the chisel edge. When using a

small pilot hole, the drill can deflect more easily, but the
0
cutting lips quickly interfere with the drilled surface close
to the pilot hole, not just the hole wall. For larger pilot
holes, the drill will mostly interfere with the hole wall.
-25 0
0 0.1 0.2 0.3 0 1 2 3 4 5 7. Conclusions
Time [s] Frequency [kHz]
This paper presents a 3D time domain simulation model
Fig. 21. Simulation for drill with flexibility in lateral and torsional–axial of the drilling process by considering rigid body kinematics
directions: (a) bottom workpiece surface; (b) cross-section surface profile
and two lateral, axial and torsional vibrations of the drill.
at pilot hole; (c) axial tool deflection and Fourier spectrum; and (d) lateral
tool deflection and Fourier spectrum; conditions: 2400 rpm, fr ¼ 0.30 mm/ Hole oversize due to grinding errors on the tool is predicted
rev, pilot hole diameter ¼ 4 mm. well and matches with results found in literature. The
physics of backward whirling in drilling are explained and
the cutting edge traces during common whirling vibrations
a 200–2600 rpm spindle speed range and pilot hole sizes of are shown. The time domain simulation model considers
0, 4, 6, 8, 10 and 12 mm. true kinematics and dynamics of the drill with nonlinea-
The lateral force spectra in Figs. 22 and 23 each contain rities, which is helpful to analyze the influence of drill
energy at spindle frequency fs, which can be attributed geometry parameter, grinding errors and structural dy-
to runout, lip height differences or lip grinding errors namic parameters on the hole quality. In addition, the time
 ARTICLE IN PRESS
J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473 1471

full hole, 400rpm full hole, 2400rpm 4mm pilot hole, 400rpm 4mm pilot hole, 2400rpm
1000
2000
Fz [N]

500
1000
0 0
2 4 6 8 0.8 1.0 1.2 1.4 1.6 1.8 2 4 6 8 1.0 1.5 2.0
Time [s] Time [s] Time [s] Time [s]
x106 x106 x105 x105
fc fc
FFT Fz [N]

5 5
10 10

0 0 0 0
0 5 0 5 0 5 0 5
Frequency [kHz] Frequency [kHz] Frequency [kHz] Frequency [kHz]
x106 x105 x106 x105
3 2 3
2f s 4f s 6f s 8f s 5 2f s 4f s 6f s 8f s 2f s 4f s 6f s 8f s 2f s 4f s 6f s 8f s
FFT Fz [N]

2 2
1
1 1
0 0 0 0
0 20 40 60 0 200 400 0 20 40 60 0 200 400
Frequency [kHz] Frequency [kHz] Frequency [kHz] Frequency [kHz]
400
200
Fx, Fy [N]

200
0 0
-200 -200
-400
2 4 6 8 0.8 1.0 1.2 1.4 1.6 1.8 2 4 6 8 1.0 1.5 2.0
Time [s] Time [s] Time [s] Time [s]
x106 x106 x105 x105
2
FFT Fx, Fy [N]

2 fc fc
10 10
1 1 5 5

0 0 0 0
0 5 0 5 0 5 0 5
Frequency [kHz] Frequency [kHz] Frequency [kHz] Frequency [kHz]
x106 x105 x106 x104
5 5
FFT Fx, Fy [N]

2 2fs 4fs 6fs 8fs 2fs 4fs 6fs 8fs fs 2fs 4fs 6fs 8fs fs 2fs 4fs 6fs 8fs
10
1
5
fs fs
0 0 0 0
0 20 40 60 0 200 400 0 20 40 60 0 200 400
Frequency [kHz] Frequency [kHz] Frequency [kHz] Frequency [kHz]

Fig. 22. Details of 4 cutting experiments with 16 mm drill bit: 8 mm piloted at 400 rpm (a) and 1000 rpm (b), 12 mm piloted at 400 rpm (c) and 800 rpm (d);
thrust vs. time, FFT from 0 to 5000 Hz and FFt from 0 to 10fs; Lateral forces vs. time, FFT from 0 to 5000 Hz and FFT from 0 to 10fs; fr ¼ 0.30 mm/rev,
hole depth ¼ 15 mm.
 ARTICLE IN PRESS
1472 J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473

8mm pilot hole, 400rpm 8mm pilot hole, 1000rpm 12mm pilot hole, 400rpm 12mm pilot hole, 800rpm
600 300
Fz [N]

400 200
200 100

0 0
2 4 6 8 1 2 3 2 4 6 2 3 4
Time [s] Time [s] Time [s] Time [s]

x105 x105 x105 x105

FFT Fz [N]

fc 5 5 fc
5 5

0 0 0 0
0 5 0 5 0 5 0 5
Frequency [kHz] Frequency [kHz] Frequency [kHz] Frequency [kHz]

x105 x105 x105 x105

2 5 3
FFT Fz [N]

2fs 4fs 6fs 8fs 2fs 4fs 6fs 8fs 5 2fs 4fs 6fs 8fs 2fs 4fs 6fs 8fs
fs 2
1
1
0 0 0 0
0 20 40 60 0 50 100 150 0 20 40 60 0 50 100
Frequency [Hz] Frequency [Hz] Frequency [Hz] Frequency [Hz]

200
100
Fx, Fy [N]

0 0

-100
-200
2 4 6 8 1 2 3 2 4 6 2 3 4
Time [s] Time [s] Time [s] Time [s]

x105 x105 x105 x105

5 5 5
FFT Fx, Fy [N]

5
fc fc

0 0 0 0
0 5 0 5 0 5 0 5
Frequency [kHz] Frequency [kHz] Frequency [kHz] Frequency [kHz]

x105 x105 x105 x105

4 3
FFT Fx, Fy [N]

f s 2f s 4f s 6f s 8f s f s 2f s 4f s 6f s 8f s f s 2f s 4f s 6f s 8f s f s 2f s 4f s 6f s 8f s
5 5 2
2
1
0 0 0
0 20 40 60 0 50 100 150 0 20 40 60 0 50 100
Frequency [Hz] Frequency [Hz] Frequency [Hz] Frequency [Hz]

Fig. 23. Details of 4 cutting experiments with 16 mm drill bit: 8 mm piloted at 400 rpm (a) and 2400 rpm (b), 4 mm piloted at 400 rpm (c) and 2400 rpm (d);
thrust vs. time, FFT from 0 to 5000 Hz and FFt from 0 to 10fs; Lateral forces vs. time, FFT from 0 to 5000 Hz and FFT from 0 to 10fs; fr ¼ 0.30 mm/rev,
hole depth ¼ 15 mm.
 ARTICLE IN PRESS
J.C. Roukema, Y. Altintas / International Journal of Machine Tools & Manufacture 47 (2007) 1455–1473
domain simulations are basis to validate the linear
chatter stability solutions as presented in the Part II of
the article [29].
The simulated forces show great similarity with experi-
mental cutting forces. Coupling between vibrations in
lateral and torsional–axial directions proves to be small in
the simulation model and is absent in the experimental
results. Both lateral and torsional–axial chatter vibrations
can be predicted using the model, but only torsional–axial
chatter is observed to be dominant in actual drilling
experiments. Lateral chatter can become important for
drills with larger length to diameter ratios.
The accurate prediction of chatter vibrations in drilling
still remains to be a research challenge. The stiffness and
damping of the drill modes change significantly in
comparison to the experimentally measured frequency
response function of the drill at its free tip when it was
mounted on the spindle. The clearance face of the drill lips
rubs against the wavy surface of the hole bottom; the side
flutes of the drill are prevented to vibrate freely due to hole
wall contact; the drill tip and wall contact boundaries
change continuously as the drill penetrates into the
material further. The complex drill–hole surface contact
and friction causes time varying contact stiffness and
process damping which is a challenging future research in
order to improve prediction of drilling dynamics.
Acknowledgments
This research is sponsored by NSERC and Pratt &
Whitney Canada. Guehring and Sandvik companies
provided the cutting tools and tool holders, respectively.
The Mori Seiki SH403 machining center was donated
through MTTRF by Mori Seiki, Japan.
References
[1] J.C. Roukema, Y. Altintas, Time domain simulation of torsional–ax-
ial vibrations in drilling, International Journal of Machine Tools &
Manufacture 46 (2006) 2073–2085.
[2] D.F. Galloway, Some experiments on the influence of various factors
on drill performance, Transactions of ASME 79 (1957) 191–231.
[3] S.J. Lee, K.F. Eman, S.M. Wu, An analysis of the drill wandering
motion, ASME Journal of Engineering for Industry 109 (4) (1986)
297–305.
[4] C. Lin, M.N. Jalisi, K.F. Ehmann, Experimental analysis of initial
penetration in drilling, in: D.A. Stephenson, R. Stevenson (Eds.),
TMS & ASME: Materials Issues in Machining—II and the Physics of
Machining Processes—II, 1994, pp. 383–407.
[5] V.V. Zelentsov, The lobing of drilled holes, Soviet Engineering
Research 1 (10) (1981) 52–54.
[6] P.G. Reinhall, D.W. Storti, Modeling and analysis of the dynamics of
a drill penetrating a thin plate, ASME Journal of Applied Mechanics
53 (1986) 690–694.
[7] O. Tekinalp, A.G. Ulsoy, Modeling and finite element analysis of
drill bit vibrations, ASME Journal of Vibration, Acoustics, Stress,
and Reliability in Design 111 (2) (1989) 148–155.
[8] O.T. Tekinalp, A.G. Ulsoy, Effects of geometric and process
parameters on drill transverse vibrations, ASME Journal of
Engineering for Industries 112 (2) (1990) 189–194.
1473
[9] D.M. Rincon, A.G. Ulsoy, Complex geometry, rotary inertia and
gyroscopic moment effects on drill vibrations, Journal of Sound and
Vibration 188 (5) (1995) 701–715.
[10] D.M. Rincon, A.G. Ulsoy, Effects of drill vibrations on cutting forces
and torque, CIRP Annals 43 (1) (1994) 59–62.
[11] S. Ema, H. Fujii, E. Marui, Whirling vibration in drilling Part 1:
cause of vibration and role of chisel edge, ASME Journal of
Engineering for Industries 108 (3) (1986) 157–162.
[12] S. Ema, H. Fujii, E. Marui, Whirling vibration in drilling Part 2:
influence of drill geometry, particularly of the drill flank, on the
initiation of the vibration, ASME Journal of Engineering for
Industries 108 (3) (1986) 163–168.
[13] S. Ema, H. Fujii, E. Marui, Whirling vibration in drilling Part 3:
vibration analysis in drilling workpiece with a pilot hole, ASME
Journal of Engineering for Industries 110 (4) (1988) 315–321.
[14] P.V. Bayly, J.E. Halley, K.A. Young, Tool oscillation and the
formation of lobed holes in a quasi-static model of reaming, DETC/
VIB8061, in: ASME Design Engineering Technical Conference, Las
Vegas, NV, 1999.
[15] P.V. Bayly, K.A. Young, J.E. Halley, S.G. Calvert, Analysis of tool
oscillation and hole roundness error in a quasi-static model of
reaming, ASME Journal of Manufacturing Science and Engineering
123 (3) (2001) 387–396.
[16] P.V. Bayly, M.T. Lamar, S.G. Calvert, Low-frequency regenerative
vibration and the formation of lobed holes in drilling, ASME Journal
of Manufacturing Science and Engineering 124 (2) (2002) 275–285.
[17] M. Pirtini, I. Lazoglu, Forces and hole quality in drilling,
International Journal of Machine Tools & Manufacture 45 (11)
(2005) 1271–1281.
[18] K. Gupta, O. Burak Ozdoganlar, S.G. Kapoor, R.E. DeVor,
Modeling and prediction of hole profile in drilling, Part 1: modeling
drill dynamics in the presence of drill alignment errors, ASME
Journal of Manufacturing Science and Engineering 125 (1) (2003)
6–13.
[19] K. Gupta, O. Burak Ozdoganlar, S.G. Kapoor, R.E. DeVor,
Modeling and prediction of hole Profile in drilling, Part 2: modeling
hole profile, ASME Journal of Manufacturing Science and Engineer-
ing 125 (1) (2003) 14–20.
[20] P.V. Bayly, S.A. Metzler, K.A. Young, J.E. Halley, Analysis and
simulation of radial chatter in drilling and reaming, DETC/VIB8059,
in: ASME Design Engineering Technical Conference, Las Vegas, NV,
1999.
[21] T. Arvajeh, F. Ismail, Machining stability in high-speed drilling—
Part 1: modeling vibration stability in bending, International Journal
of Machine Tools & Manufacture 46 (2006) 1563–1572.
[22] T. Arvajeh, F. Ismail, Machining stability in high-speed drilling—
Part 2: time domain simulation of a bending-torsional model and
experimental validations, International Journal of Machine Tools &
Manufacture 46 (2006) 1573–1581.
[23] S. Ema, H. Fujii, E. Marui, Chatter vibration in drilling, ASME
Journal of Engineering for Industry 110 (4) (1988) 309–314.
[24] M.N. Jalisi, C. Lin, K.F. Ehmann, Modeling and experimental
analysis of margin forces in micro-drilling, ASME, Manufacturing
processes and materials challenges in micro-electronic packaging,
AMD 131/EEP 1(1991)25–33.
[25] E.J.A. Armarego, S. Wiriyacosol, G. Lorenz, Thrust and torque
prediction in drilling from a cutting mechanics approach, CIRP
Annals 28 (1) (1979) 87–92.
[26] V. Chandrasekharan, S.G. Kapoor, R.E. DeVor, A mechanistic
model to predict the cutting force system for arbitrary drill point
geometry, ASME Journal of Manufacturing Science and Engineering
120 (3) (1998) 563–570.
[27] J.C. Roukema, Y. Altintas, Kinematic model of dynamic drilling
process, IMECE2004-59340, in: ASME International Mechanical
Engineering Congress 2004, Anaheim, CA.
[29] J.C. Roukema, Y. Altintas, Generalized modeling of drilling
vibrations: Part II—chatter stability in frequency domain, Interna-
tional Journal of Machine Tools & Manufacture.