5th International & 26th All India Manufacturing Technology, Design and Research Conference (AIMTDR 2014) December

12th–14th, 2014, IIT

Guwahati, Assam, India

Determining Cutting Force Coefficient from Instantaneous Cutting

Force in Ball-End Milling

Mithilesh Kumar Dikshit1*, Asit Baran Puri2, Atanu Maity3, Amit Jyoti Banarjee4
1*
Department of Mechanical Engineering, NIT Durgapur, Durgapur - 713209,
E-mail: dixit.mithilesh@gmail.com
2
Department of Mechanical Engineering, NIT Durgapur, Durgapur - 713209,
E-mail: abpuri2000@yahoo.com
3
Advance Design & Optimization, CSIR-CMERI, Durgapur - 713209,
E-mail: maity@cmeri.res.in
4
Manufacturing Technology Group, CSIR-CMERI, Durgapur - 713209,
E-mail: ajbanerjee@cmeri.res.in

Abstract
Specific cutting force coefficients play an important role in study of cutting forces in ball end milling. Due to the
complicated geometry of ball end milling cutter the effective diameter of the tool (with active cutting edges) varies
along the axis of the tool during machining which leads to the discrepancies in the prediction of cutting forces. In
this paper, an improved method of identification of specific cutting force coefficient is proposed for ball end milling
cutter. The cutter is discretized into finite number of axial discs along the axis of the cutter. A semi-mechanistic
force model is adopted to relate the cutting forces in each disc and specific cutting force coefficients are calculated
by inverse method and a fourth order polynomial fitting has been obtained. Several experiments were carried out at
different feed and depth of cut to calibrate the proposed identification method and the same is used for predicting the
cutting forces. The shearing force coefficients have larger value at lower depth of cut and they decrease with
increase in depth of cut. Edge force coefficients do not vary with the depth of cut significantly.
Keywords: Ball end milling, cutting forces, force coefficients, polynomial fitting

1 Introduction cutting forces efficiently and accurately in ball end

Ball-end milling is a versatile milling process and one
of the most commonly used milling process in
automobile, aerospace and biomedical engineering
sectors and especially in die and mould making
industries. The quality and precision of the final
components are highly influenced by the cutting forces.
The precise prediction of cutting forces helps in
improving the machining performances. An accurate
prediction of the cutting forces depends on the
modelling of cutting forces along with cutting force
coefficients. Cutting force model is an important
process model that predicts the cutting forces
components in tangential, radial and axial directions.
The cutting force predictions are mainly categorised
into three methods: analytical, mechanistic and
numerical methods. From the literature, it was found
that the recent works for the force modelling in ball end
milling have focused on mechanistic approach [Lee et
al. (1996), Budak et al. (1996)]. In this approach cutting
forces are assumed to be proportional to the chip cross-
sectional area and cutting force coefficients are the
proportionality constant between the cutting forces and
chip cross sectional area. Considerable amount of
literatures has been found to estimate the cutting force
coefficients which are the key factors for predicting the
milling.
However, in most of the research works two major
approaches for estimation of cutting force coefficients
have been proposed: (i) orthogonal to oblique cutting
transformation and (ii) estimation of cutting force
coefficients through regression analysis. In orthogonal
to oblique cutting transformation approach, the cutting
force coefficients are the function of shear stress, shear
angle and friction angle. The coefficients are determined
through the orthogonal turning experiments which are
then transformed into oblique cutting edge through the
transformation matrix for prediction of cutting forces in
helical end milling forces [Armarego and Deshpande
(1991), Budak, E. et al. (1996)]. Altintas and Lee (1998)
proposed a mechanics of ball end milling using
orthogonal to oblique transformation considering
shearing and ploughing effect of the ball end milling
cutter. It is found that a large number of experiments
were required to build orthogonal database which is a
time consuming process and large number of
computation would be required.
Another approach is known as mechanistic approach
which is a direct calibration approach. In mechanistic
approach, the cutting force coefficients are calculated
directly from the experiments. Feng and Menq (1994)
developed a 2D cutting force model in tangential and
74-1
Determining Cutting Force Coefficient from Instantaneous Cutting Force in Ball-End Milling

radial direction, and established an empirical relation Where ‘θ’ is the spindle rotation angle, ϕp is the pitch
using the numerical polynomial fit. Lamikiz A et al. angle defined by ϕp = 2π/N and δi (z) is radial lag angle
(2004) studied the milling of complex surfaces and varying between 0 and δ0, and varies with the local helix
proposed a semi-mechanistic model for cutting forces. angle ψi for i-th element. The value of δ0 is constant for
The model is only suitable for a specific cutting a specific tool and depends on helix angle ψ and tool
condition only. Gradisek J et al. (2004) have shown the radius R. The local radius R(z) of each disc can be given
relationship between cutting force coefficients and axial by:
depth of cut for down and up ball end milling. Kim et al. 2 2
(2000) proposed that the cutting force coefficients in R( z ) = R − ( R − z ) (2)
ball end milling are the functions of cutting edge The axial immersion of the cutting edge P can be given
position angle. Most of the researches mentioned above, by:
focus on determining the cutting force coefficients for a
particular machining condition and a certain work-tool R 
κ = arc sin  z  (3)
combination. In the present study, a mechanistic force  R 
model has been developed considering both the effect of Where κ is the axial immersion angle of the cutting edge
shearing mechanism due to the chip generation process on the disc measured with the Z-axis from the centre of
on the tool’s rake face and the effects of ploughing the ball part. As shown in the Fig. 1, the position of
mechanisms on the flank face based on Lee and Altintas each discrete element i of the cutting edge j can be
(1996). The cutting force coefficients were calculated given as the function of angular position of the
by a comparatively new method using the instantaneous elemental cutting edge and the local radius as given
cutting force values. A fourth order polynomial fitting below.
was adopted for determining the cutting force
coefficients. 
2 Geometry of ball-end milling cutter
 x = R sin β
 ji zi ji ( )

The geometry of ball end milling cutter is shown in the  y ji = Rzi cos β ji ( ) (4)
Fig. 1. The geometry of the cutter may be divided into 
two parts: a cylindrical part with constant helix angle ψ, z = Rδi
and a hemispherical part of radius R equal to the radius  ji tan (ψ )
of cylindrical part. The cutting edge lie on the
In vector form the position of each cutting edge
hemispherical part. The ball part of the ball-end milling
element can be given by the following formula:
cutter is discretised into a finite number of discs along

the axis of the cutter. The cutting edge discretization r
ji (
β ) = x jiiˆ + y ji ˆj + z ji kˆ (5)
allows the simplification of the cutting edge as a
      Rδ ˆ
sequence of the linear cutting segments. Let us consider = R  δ  sin  β  iˆ + cos  β  ˆj  + i k
an elemental cutting edge P of a disc. The position of i  i    ji   ji   tan (ψ )
elemental cutting edge P can be characterized by The radius of each discrete disc can be calculated as the
elevation (z) from the tip of the tool, radial distance (Rz) function of lag angle using Eq. (2) and (3) as:
from the cutter axis, axial immersion angle (κ), and 2
radial immersion angle (β) as shown in the Fig. 1. The  δi 
position of the ith element on the jth cutting edge with ( )
Ri δ i = R 1 −  − 1 (6)
elevation z may be given by:  tan (ψ ) 
After defining the position of each elemental cutting
edge its edge length can be given by:
a b 2 R
2


ds ji = dr ji =  Ri (δ i ) + Ri2 (δ i ) + 2 dδ (7)
tan (ψ )
It is necessary to find the uncut chip thickness at any
cutter point in the engagement region to determine the
differential cutting forces. Once the position of the each
discrete cutting edge element is defined, it is necessary
to calculate the uncut chip thickness based on the
analysis of the cutter geometry modelling as shown in
Fig. 1. (a) Front view of discretized cutter, (b) the Fig. 2. The chip thickness formulation proposed by
top view and position angles of cutting edge Martelotti for the straight end mills has been modified,
since the chip thickness varies as the depth of cut and
β ji ( z ) = θ + ( j − 1) φ p − δ i ( z ) (1) radial immersion of the cutter changes. The uncut chip
thickness is expressed on the basis of radial and axial
immersion angles as shown in Eq. (8).
74-2
5th International & 26th All India Manufacturing Technology, Design and Research Conference (AIMTDR 2014) December 12th–14th, 2014, IIT
Guwahati, Assam, India

( ) ( )
t nj β j , κ = ft sin β j sin (κ ) .ε ji (8)
Where ft is the feed per tooth and it is determined from
feed rate and εji is the factor which shows the
engagement/disengagement of the cutter with workpiece
at a particular instant of rotation.
a elemental chip width in each cutting edge element. Six
b A semi-mechanistic force model is adopted and it is
Where dFtj, dFrj, dFaj are the tangential, radial and axial
force components acting on the j-th cutting edge, Kte,
Kre, Kae (N/mm) are the edge force coefficients or
specific edge coefficients, Ktc, Krc, Kac (N/mm2) are the
shear specific coefficients, dS (mm) is the length of
each discrete element, tnj (mm) is the uncut chip
thickness (tn) for j-th cutting edge and db (mm) is the
specific cutting force coefficients needed to be
identified through the experiments for a particular
cutter-workpice material under specific cutting
conditions. Thus, the three elemental cutting force
components dFt, dFr dFa acting on the infinitesimal
cutting edge, are determined in an orthogonal system of
cutting. The actual forces in oblique cutting are
determined further through coordinate transformation of
the above forces. Finally, these three cutting forces Ft,
Fr and Fa (as mentioned in equation 9) are evaluated
through numerical integration.
4 Specific cutting force coefficients
assumed that the average cutting force from the
experiments are the input factors to derive the specific
cutting force coefficients. The analytical equation for
average cutting force is

1 φex z2
F ( t , r ,a ) = ∫ ∫ dF (φ , z ) d φ (10)
φ p φst z1 ( t ,r ,a )
Where ϕst and ϕex are the start and exit radial
immersion angles, respectively. The axial integration
limits for each cutting tooth can be determined from
cutter-workpiece engagement area. In a slot milling ϕst
= 0 and ϕex = π. The instantaneous force at radial
immersion angle β can be given as, Gradisek J et al.
Fig. 2. (a) Idealized chip thickness (b) Actual chip (2004),
thickness and its projection in ball end milling
 Ft ( β )   A1   B1 
3 Mechanistic force model  F ( β ) = ft
T1   A2  + T2   B2  (11)
 r  2    
In this paper cutting forces are modelled through the  Fa ( β )  A3   B3 
two fundamental factors namely shearing effect taking
place in the shear zone and edge effect induced due to Where A’s and B’s represent the influence of cutter
the ploughing at the cutting edge. Due to the geometry on the average cutting force (due to shearing
complicated geometry of ball-end milling cutter, the action) and average edge force (due to ploughing)
cutting forces vary in different discs. For each axially respectively. They are called the geometric constants
discretized disc, the cutting force is regarded as an and are determined by the cutter geometry and can be
independent element. The total force acting on the cutter given by:
z2 z2 z2
may be evaluated through the numerical integration of
A1 = ∫ dz , A2 = ∫ sin κ ( z ) dz, A3 = ∫ cos κ ( z ) dz
the forces acting on each discretized disc along the axis z1 z1 z1
of the tool. As per the mechanistic method, the cutting z2 z2 z2
force components may be expressed as under Lee P, B1 = ∫ dS ( z ), B2 = ∫ sin κ ( z ) dS ( z ), B3 = ∫ cos κ ( z ) dS ( z ) (12)
Altintas Y (1996): z1 z1 z1

dFtj (θ , z ) = K te dS + K tc tnj ( β , θ , κ ) db

dFrj (θ , z ) = K re dS + K rc tnj ( β , θ , κ ) db (9)
dFaj (θ , z ) = K ae dS + K ac tnj ( β , θ , κ ) db

74-3
Determining Cutting Force Coefficient from Instantaneous Cutting Force in Ball-End Milling

 − K tc sin ( 2 β ) −2 K rc sin
2
(β ) −2 K ac sin
2
( β ) discrete cutting depths in experiments are {x1, x2, x3, x4,
  x5} with feed rates of { f1, f2, f3, f4, f5} and {F1, F2, F3,
[T1 ] = 2 Ktc sin 2 ( β ) − K rc sin ( 2 β ) − K ac sin ( 2 β )
 F4, F5} is the corresponding cutting forces. While F in
 0 −2 K ac sin ( β ) 2 K rc sin ( β )  Eq. (13) at different depths should be changed into
  F ( t ,r ,a ) = F ( t ,r ,a ), N − F ( t ,r ,a ), N −1 (17)
− K te cos ( β ) ( )
− K re sin β ( )
− K ae sin β
[T2 ] =  Kte sin ( β ) − K re cos ( β ) − K ae cos ( β ) N is the current cutting depth and N-1 refers to the
  previous cutting depth. Since the cutting force
 0 − K ae K re
 components are in three axis, each element of {F1, F2,
The average force from the instantaneous forces are F3, F4, F5} has three components as well. The force on
given by: each disc is recalculated based on the forces on the
current and adjacent layers. Taking one force
 Ft  K  K 
components in tangential direction as an example, the
 F  = ft [T ]  K tc  + 1 [T ]  K te  (13)
cutting forces adopted in different discs are:
 r  φ p 3  rc  φ p 4  re   Ft  F
 Fa   K ac   K ae   1  1 
Where matrices T3 and T4 depends on the cutter  Ft2 − Ft1   F2 
geometry, radial immersion angle and cutting depth.    
The matrices would be obtained by two independent  3 2  =  F3 
Ft − Ft (18)
integration related to start and exit radial immersion    
angle and axial depth of cut as given below.   F 
 tN t N −1   N 
F − F


C3 A1 ( C2 − C1 ) A2 ( C2 − C1 ) A3 
T3  = − ( C2 − C1 ) A1 Where {F1, F2, F3,…FN} are input element of Eq. (18)
C3 A2 C3 A3 
  which are considered as the cutting forces in the {1, 2,
0 −C5 A3 C5 A2
  3,…N}th disc. Cutting force coefficients are the
−C4 B1 C5 B2 C5 B3  differential value corresponding to the each disc. The
T4  =  −C5 B1 −C4 B2 −C4 B3  coefficients are fitted with experimental data through 4th
order polynomial fitting. The detailed expression for
 0 2C1B3 −2C1B2 
any cutting force coefficient may be expressed as:
4 3 2
Where c’s are the immersion constants and depends on K t,r,a = P1 z + P2 z + P3 z + P4 z + P5
( ) (19)
the immersion angles and can be given by:
P1, P2, P3, P4, P5 are the fitted coefficients and z (mm)
1 β 1 β 1 β
C1 = β βex ; C2 = sin 2 β βex ; C3 = cos 2 β βex refers to the distance along the axial direction.
2 st 4 st 4 st
β βex
C4 = sin β βex ; C5 = cos β βst (14 )
st
For slot milling the value for the immersion constants
are:
π
C1 = ; C2 = C3 = C 4 = 0 ; C5 = −2 (15)
2
From the Eq. (14) we observed that the average cutting
force has action called shearing action and edge action
per cutter tooth is a linear function of feed and can be
written as:
F ( t ,r ,a ) = F s ( t ,r , a ) f t + F e ( t , r , a ) (16)
Fig. 3 Experimental setup
Forces due to shearing action and edge action can be
obtained experimentally from milling test at different 5 Simulation and experimental results
feed rates keeping other parameters viz. axial and radial
depth of cut and cutting speed. There are two unknowns In order to determine the specific cutting force
and one equation as shown in Eq. (16), therefore linear coefficients a series of slot milling test were performed
interpolation method, George W. Collins (2003) is on Mikron-VCP 710, 3-axis vertical milling machining
adopted to calculate the specific cutting force center. The cutter was two fluted solid carbide ball end
coefficients at different feed rates keeping cutting speed, mill cutter with mono layer of TiAlN coating from
radial and axial depth of cut constant. Suppose the CoroMill Plura series of Sandvik with 10 mm diameter,

74-4
5th International & 26th All India Manufacturing Technology, Design and Research Conference (AIMTDR 2014) December 12th–14th, 2014, IIT
Guwahati, Assam, India

72 mm projection length and 30 degrees helix angle. It is found that the shearing coefficients vary largely
The workpiece material was aluminium block (Al2014- with the cutting depths, and variations of edge
T6) of size 160 × 100 × 100 mm. A 3-component coefficients with cutting depths are relatively small. As
dynamometer (Make: Kistler, model 9257B) was used shown in the Fig. 4 (a), the specific shear cutting force
for force data acquisition. The dynamometer has been coefficient fluctuates very much with depth of cuts. The
fixed to the machining centre using the fixtures along value of specific cutting coefficients are very large at
with the aluminium block. The cutting forces are sensed smaller depth of cut. Also, as is evident from Fig. 4 (c),
by the piezoelectric transducers within the dynamometer the specific cutting edge coefficients have small
and an electric charge is formed due to the dynamic fluctuation range compared to former. This
forces generated in the cutting edge. The charge phenomenon can be justified by material removal
obtained from the dynamometer is the sum of all mechanism on the rake and flank contact with the work
dynamic differential forces acting on both flutes. A material. The verification result is shown in the
multichannel charge amplifier of type 5070 is used for
converting this electric charge to voltage output. The 6
sensitivity values for the three channels (x, y and z) in 4
the amplifier were -7.87, -7.90 and -3.69 pC/N

coefficients (N/mm)
respectively. Data has been collected for full 2

Edge force
engagement region and sampling frequency rate of 1200 0
Hz for all tests. An inbuilt software called dymoware
-2
was used to display and record the measured data. The
experimental setup is shown in Fig. 3. The ball end -4
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
milling cutter was divided into five segments axially Cutting depth (mm)
from 0.00-0.2, 0.2-0.6, 0.6-1, 1.1.4 and 1.4-1.8 [mm]
Fig. 4 (d) Fitting of Specific edge
and a series of feed rate with 0.02, 0.07, 0.12, 0.17 and
0.22 [mm/tooth] were applied. Thus, six specific cutting cutting force coefficient
force coefficients were calculated as discussed in the
section 4. The shear and edge specific cutting force Fig. 5 (a-c). for the cutting condition, feed of 0.07
coefficients are shown in the Fig. 4. mm/tooth, depth of cut 0.6 mm, radial depth of cut 0.7,
3000 cutting speed 150 m/min and spindle rotation 10052
coefficients (N/mm2)

Ktc Krc Kac

rpm based on the proposed specific cutting force
2000
Shearing force

identification procedure. The predicted results were

1000 compared with the measured cutting forces obtained by
0 dynamometer. The compared region is shown with the
-1000
circle for one time period of the rotation of the ball end
milling cutter as shown in the Fig. 5 (a-c). It was found
-2000
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 that the predicted results is in well agreement with the
Depth of cut (mm) measured cutting forces. There is small deviation in the
predicted cutting force components to the measured
Fig. 4(a): Experimental data of specific ones and is below 7% which is quite satisfactory. These
shearing coefficients deviations can be caused my machine tool vibration at
higher spindle rotation and due to force measuring
3000 device during the experiments.
coefficients (N/mm2)

Ktc Krc Kac

2000
Shearing force

1000

-1000

-2000
0 0.5 1 1.5 2
Depth of cut (mm)

Fig. 4(b): Fitting of specific shearing

coefficients
6
Kte Kre Kae
coefficients (N/mm)

4
Edge force

-2

-4
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 74-5
Depth of cut (mm)

Fig. 4(c): Experimental data of

specific edge coefficients
Determining Cutting Force Coefficient from Instantaneous Cutting Force in Ball-End Milling

Fig. 5 comparison of experimental and predicted

cutting forces for ball end milling cutter, f=0.07
mm/tooth, Ap=0.6 mm, Ae=0.7 mm, Vc= 150
m/min, N=10052 rpm. (a) Tangential cutting
force (Fy), (b) Axial cutting force Fz and (c)
Radial cutting force.

6 Conclusions

An improved identification procedure is proposed to

determine specific cutting coefficients for ball end
milling cutter. A semi-mechanistic force model was
adopted to derive the expression of cutting force
components for axially discretized ball end milling
cutter total force acting on the cutter was obtained by
the numerical integration of all the elements along the
tool axis. The specific cutting force coefficients were
obtained by the inverse method and fitted in 4th order
polynomial and finally the specific force coefficients
identification approach was validated with experimental
results. The force coefficient identification process
discussed in the present paper is fast and more accurate
compare to conventional identification method. The
effectiveness of the identification process was
demonstrated through experimental and analytical
derivations and forces obtained by identification process
and experiments are in well agreement and it can be
further use in the study of tool inclination and dynamics
of ball end milling.

References
Altintas, Y. Lee, P. (1998) Mechanics and dynamics of
ball-end milling. ASME Journal of Manufacturing
Science and Engineering, 120: 684 692.
74-6
5th International & 26th All India Manufacturing Technology, Design and Research Conference (AIMTDR 2014) December 12th–14th, 2014, IIT
Guwahati, Assam, India

Lamikiz, A, L. N. Lo’pez de Lacalle, J. A. Sa’nchez, M.

Salgado, A. (2004) Cutting force estimation in
sculptured surface milling. International Journal of
Machine Tools & Manufacture, 44: 1511–1526.
Armarego, E.J.A.Deshpande, N.P. (1991) Computerized
end-milling force predictions with cutting models
allowing for eccentricity and cutter deflections. CIRP
Annals,40: 25–29.
Budak, E. Altintas, Y. Armarego, E.J.A. (1996)
Prediction of milling force coefficients from orthogonal
cutting data. Transactions of ASME Journal of
Engineering Industry, 118: 216–224.
Feng, H.S.; Menq, C.H. (1994) The prediction of cutting
forces in the ball-end milling process I: Model
formulation and model building process. International
Journal of Machine Tools & Manufacture, 34: 697-710.
George W. Collins, II, Fundamental numerical methods
and data analysis. Case Western Reserve University,
Cleveland, USA (2003).
Gradisek J, Kalveram M, Weinert K (2004) Mechanistic
identification of specific force coefficients for general
end mill. International Journal of Machine Tools &
Manufacture, 44: 401-414.
Kim GM, Cho PJ, Chu CN (2000) cutting force
prediction of sculptured surface ball-end milling using
Z-map. International Journal of Machine Tools &
Manufacture, 40: 277–291.
Lee P, Altintas Y, Predictions of ball-end milling forces
from orthogonal cutting data. Int J Mach Tools Manuf
1996; 36 (9):1059–72.

74-7