micromachines

Article
Tool Run-Out Measurement in Micro Milling
Aldo Attanasio ID

Department of Mechanical and Industrial Engineering, University of Brescia, Via Branze 38, 25123 Brescia, Italy;
aldo.attanasio@unibs.it; Tel.: +39-331-613-3181

Academic Editor: Nam-Trung Nguyen

Received: 29 May 2017; Accepted: 12 July 2017; Published: 24 July 2017

Abstract: The interest in micro manufacturing processes is increasing because of the need for
components characterized by small dimensions and micro features. As a result, researchers are
studying the limitations and advantages of these processes. This paper deals with tool run-out
measurement in micro milling. Among the effects of the scale reduction from macro to micro, tool
run-out plays an important role, affecting cutting force, tool life, and the surface integrity of the
produced part. The aim of this research is to develop an easy and reliable method to measure tool
run-out in micro milling. This measuring strategy, from an Industry 4.0 perspective, can be integrated
into an adaptive model for controlling cutting force, with the aim of improving the production
quality and the process stability, while at the same time reducing tool wear and machining costs.
The proposed procedure deduces tool run-out from the actual tool diameter, the channel width,
and the cutting edge’s phase, which is estimated by analyzing the cutting force signal. In order to
automate the cutting edge phase measurement, the suitability of two functions approximating the
force signal was evaluated. The developed procedure was tested on data from experimental tests.
A Ti6Al4V sample was machined using two coated micro end mill flutes made by SECO setting
different run-out values. The results showed that the developed procedure can be used for tool
run-out estimation.

Keywords: micro milling; tool run-out; experimental measuring; experimental tests; Ti6Al4V alloy;
Industry 4.0

1. Introduction
Micro manufacturing processes are becoming fundamental in several industrial fields. The
need for having products characterized by very small dimensions and features is continuously
increasing in biomedical, mechanical, automotive and aerospace applications. At the same time,
many researchers are focusing their attention on manufacturing processes that are able to realize these
micro components. As reported in [1], several different definitions of “micro machining” can be found
in the literature. Dornfeld et al. [2] referred to their modified version of the Taniguchi graph [3], which
shows the development of the manufacturing capability in terms of achievable machining accuracy,
when they defined micro machining as a process in which the accuracy is lower than 1 microns. In
Masuzawa et al. [4], Masuzawa et al. state that, although micromachining literally is a machining
process with dimensions between 1 µm and 999 µm, it is possible to consider as micro machining
processes all the processes utilized for realizing features that cannot be achieved by conventional ones.
A similar definition is given in [5], where the production of micro parts is defined as “the production of
parts or structures with at least two dimensions in the sub-millimeter”. Differently, Alting et al. [6] focused
their attention on the definition of micro engineering as all the processes dealing with developing and
manufacturing components with functional characteristics of at least one dimension of the order of
millimeters. It is evident that all these definitions refer to different process issues, including feature
dimension, process accuracy, surface roughness, feature feasibility, and so on. A definition covering all

Micromachines 2017, 8, 221; doi:10.3390/mi8070221 www.mdpi.com/journal/micromachines

Micromachines 2017, 8, 221 2 of 16

the issues characterizing a micro manufacturing process is hard or even impossible to give. For this
reason, a general definition of micro machining has been assumed in this paper. This definition
considers micromachining to be all the manufacturing processes that allow dimensions to tenths of a
millimetre through using tools with sub-millimetre dimensions.
Of course, this definition covers a wide range of manufacturing processes. Micro cutting or micro
mechanical machining processes are amongst them, and have many advantages. They are cost effective,
flexible, efficient, and guarantee the possibility of realizing complex shapes with high removal rates,
high accuracy, good roughness and low costs with respect to other non-conventional processes.
The study of micro cutting operations involves phenomena that can be neglected at the macro
scale. These phenomena become important at the micro scale since the tool dimension, the uncut chip
thickness and the material grain size are of the same order of magnitude. Consequently, a variation of
characteristics from a general behaviour (i.e., size effect) can be observed when reducing the size of the
workpiece, tool or process parameters [7]. Under these working conditions, one of the main aspects
to be considered is the identification of the minimum chip thickness value under which the removal
process is dominated by a ploughing regime instead of a shearing one. The ploughing regime is an
undesired regime, since the material is not removed (i.e., the chip does not form); instead, it is just
plastically deformed under the tool edge. It was demonstrated that the minimum chip thickness mainly
depends on the tool material and its geometry [8], and the workpiece material and its microstructure [9].
The size effect also strongly affects the specific shear energy (SSE) needed for material removal. In fact,
as reported in [8], SSE increases as the uncut chip thickness decreases.
When approaching micro milling processes [10–12], it is essential to understand how process
parameters (feed per tooth, cutting velocity, and depth of cut), tool geometry (tool edge radius, helix
angle, etc.) and the tool–machine pairs (tool–spindle run-out, machine stiffness, etc.) affect the process
and the part quality. For this purpose, the development of analytical models suitable for describing
cutting operations on a micro scale, is very helpful. Therefore, many papers in the literature are focused
on modelling different issues of micro milling operations.
Yoon et al. [13] and Kim et al. [14] investigated the influence of tool deflection and the radial
depth of cut on chip formation in micro milling. These works showed that the axial component of the
cutting force can be neglected when the axial depth of the cut is limited to few microns.
Garzòn et al. [15] analyzed the problems related to cutting force measurement during micro
cutting operations. It was demonstrated that the dynamometers used in micro cutting operations
must be characterized by high sensitivity and bandwidth because of the high spindle rotation regime.
Zhu et al. [16] focused their attention on force signal and showed how to manage a micro milling force
signal affected by strong noise.
An analytical model for estimating unbalanced cutting forces due to tool run-out is proposed
by Bissacco et al. [17]. In this model, different force profiles for each cutting edge are considered as a
function of tool run-out.
In their model, Park et al. [18] considered the effects of tool deflection and tool run-out on the
expression of tool trajectories. They suggested that the main phenomena characterizing the ploughing
regime are ploughing and the elastic recovery of workpiece material, and that beyond a critical uncut
chip thickness value, if elasto-plastic effects are negligible, then the cutting regime can be considered
shearing dominant.
For studying tool deflection, Mamedov et al. [19] developed a Finite Element (FE) model
including tool run-out, where the tool is considered as a cantilever beam with effective diameter
in the fluted portion.
Bao et al. [20–22] developed an analytical model for micro-end milling operations, including tool
run-out, while Afazov et al. [23] introduced a new approach for evaluating micro-milling cutting forces
that considered ploughing regime and the effects of tool run-out.
An analytical model for predicting the three-dimensional cutting forces of micro-end milling
processes, based on the trochoidal trajectory of the cutting edge and tool run-out, was developed by
Micromachines 2017, 8, 221 3 of 16

Li, et al. [24]. This model is based on six optimized coefficients, which minimizes the errors between the
model and the experimental force values through the least square methods. They studied the influence
of feed per tooth and tool run-out on the uncut chip thickness of each cutting edge. The limitation of
this model is that the six coefficients are not constant; they are optimized at every spindle rotation.
Consequently, this procedure requires high computational time. Attanasio et al. [25] proposed a
procedure based on the particles swarm optimization strategy for calibrating the coefficients of an
Micromachines 2017, 8, 221 3 of 16
analytical model of cutting force in micro milling, including tool run-out.
Otherthe works
model[26–32]
and the dealing withforce
experimental thevalues
development
through the ofleast
micro-milling
square methods. cutting
They force models
studied the can be
found in the influence of feedwhich
literature, per tooth and tool run-out
demonstrates theon high
the uncut
level chipof thickness
interestofineach thiscutting
topic.edge.
All The
these papers
limitation of this model is that the six coefficients are not constant; they are optimized at every
agree on the need for these models to include the influence of cutting regimes (ploughing and shearing),
spindle rotation. Consequently, this procedure requires high computational time. Attanasio et al.
the trajectory of the cutting
[25] proposed edges,
a procedure tool
based onrun-out,
the particlestool deflection,
swarm optimizationelastic recovery,
strategy and entry
for calibrating the and exit
angles. Amongst allofthese
coefficients influencing
an analytical model of factors,
cutting tool
force run-out plays an
in micro milling, important
including role. This is because in
tool run-out.
micro milling,Other worksbetween
the ratio [26–32] dealing with the development
tool run-out and feed per of micro-milling
tooth is very cutting
high force models
[32]. can becases, tool
In some
found in the literature, which demonstrates the high level of interest in this topic. All these papers agree
run-out is so high that just one flute cuts the material, generating an asymmetric cutting [27].
on the need for these models to include the influence of cutting regimes (ploughing and shearing), the
The objective ofthe
trajectory of this research
cutting edges,istool
to run-out,
develop andeflection,
tool easy and reliable
elastic method
recovery, to measure
and entry tool run-out in
and exit angles.
micro milling.
Amongst Theallproposed strategy
these influencing starts
factors, toolwith
run-out realizing
plays an channels on aThis
important role. sample. During
is because in microthe process,
the cuttingmilling,
forcethe ratio between
is acquired attool run-out
a high and feed per
sampling tooth
rate inisorder
very high [32]. Inasome
to have cases, tool run-out
high-resolution is
signal. After
so high that just one flute cuts the material, generating an asymmetric cutting [27].
that, the channel width is measured with a microscope. A geometrical model then utilizes these data
The objective of this research is to develop an easy and reliable method to measure tool run-out
to provideinthe toolmilling.
micro run-out Thevalue.
proposed strategy starts with realizing channels on a sample. During the
In order to study
process, the limits
the cutting force isofacquired
this method,
at a higha sampling
sensitivity rate analysis
in order toofhavethe ainfluence of the spindle
high-resolution
signal. After that, the channel width is measured with a
speed on the result accuracy was performed, which showed the need to use a high-signal microscope. A geometrical model then sampling
utilizes these data to provide the tool run-out value.
rate. Moreover, with the aim of automating the proposed strategy, the suitability of two approximating
In order to study the limits of this method, a sensitivity analysis of the influence of the spindle
functions,speed
namely theresult
on the Fourier series
accuracy was and the sum
performed, whichofshowed
sine, was estimated.
the need The developed
to use a high-signal samplingprocedure
was testedrate.on data coming
Moreover, withfrom
the experimental
aim of automating tests,thewhich consisted
proposed strategy,of realizing microchannels
the suitability of two with
800 µm width approximating
by usingfunctions, namely the
coated tungsten Fouriermicro
carbide series mills
and the onsum of sine,
samples was of
made estimated.
Ti6Al4V Thealloy.
developed procedure was tested on data coming from experimental tests, which consisted of
realizing microchannels with 800 μm width by using coated tungsten carbide micro mills on
2. Tool Run-Out in Micro Milling
samples made of Ti6Al4V alloy.
Tool run-out is a phenomenon that occurs due to the sum of the geometrical displacements of
2. Tool Run-Out in Micro Milling
the spindle axis, tool-holder axis and tool axis from the theoretical rotation axis. This sum produces a
Tool run-out is a phenomenon that occurs due to the sum of the geometrical displacements of
deviation between the theoretical cutting edge’s trajectory and the actual one.
the spindle axis, tool-holder axis and tool axis from the theoretical rotation axis. This sum produces a
As described in [28],the
deviation between from a geometric
theoretical pointtrajectory
cutting edge’s of view, tool
and run-out
the actual one.can be summarized by four
parameters (see As Figure
described1).
in Two parameters
[28], from a geometric define the
point of axistool
view, offset, namely,
run-out offset distance
can be summarized by fouror run-out
parameters (see Figure 1). Two parameters define the axis offset, namely, offset distance
length (r0 ) and offset or run-out angle (γ0 ); the other two parameters define the axis tilt, namely, tiltor run-out
length (r0) and offset or run-out angle (γ0); the other two parameters define the axis tilt, namely, tilt
angle (τ) and the location angle of the tilt angle (ϕ).
angle (τ) and the location angle of the tilt angle (ϕ).

Figure 1. Tool run-out and its geometric parameters.

Figure 1. Tool run-out and its geometric parameters.
Micromachines 2017, 8, 221 4 of 16
Micromachines 2017, 8, 221 4 of 16

Typically
Typically in conventional
in conventional milling,
milling, sincesince the mill
the mill diameter
diameter is large
is very very large
compared compared
to tooltorun-out,
tool
run-out, this error has a low influence on the cutting process. In micro milling,
this error has a low influence on the cutting process. In micro milling, because of the limited tool because of the limited
tool dimension, the influence of tool run-out on the process cannot be neglected. In particular, tool
dimension, the influence of tool run-out on the process cannot be neglected. In particular, tool run-out
run-out causes unbalanced chip thickness between the mill teeth. Consequently, different cutting
causes unbalanced chip thickness between the mill teeth. Consequently, different cutting forces arise on
forces arise on each cutting edge. This unbalanced load on the teeth generates unavoidable
each cutting edge. This unbalanced load on the teeth generates unavoidable vibrations that affect the
vibrations that affect the process stability. Therefore, minimizing tool run-out in micro milling is
process stability. Therefore, minimizing tool run-out in micro milling is essential in order to increase
essential in order to increase the final quality of the surface finish, avoid accelerated tool wear or
theeven
finaltool
quality of theand
breakage, surface finish,
avoid avoid accelerated
the inception of undesired toolvibrations
wear or even(i.e., tool breakage,
increase and avoid
the stability of thethe
inception
process). of undesired vibrations (i.e., increase the stability of the process).
However,
However, thethe
tooltoolrun-out
run-outparameters
parameters are are not
not easy to measure
easy to measurein inmicro
micromilling.
milling.InIn the
the literature,
literature,
different approaches can be found. These strategies often measure tool
different approaches can be found. These strategies often measure tool run-out on the tool shank. run-out on the tool shank.
However,
However, as aspointed
pointedout outininLiLietetal.
al.[32],
[32],tool
toolrun-out
run-out cannot
cannot bebe estimated
estimatedatatthe thetool
toolshank
shank inin micro
micro
milling
millingbecause
because ofofthe
theadditional
additionalrun-out
run-out duedue toto the
the tapered areaof
tapered area ofthe
thetool
toolthat
thatconnects
connects the
the cutting
cutting
edges
edgesto the tooltool
to the shank.
shank.In particular,
In particular,dialdial
indicators
indicators[23,27,28], microscopes
[23,27,28], microscopes [29], capacitive
[29], capacitivesensors
sensors[30],
and laser
[30], andsensors [31] have
laser sensors [31]been
have utilized for this
been utilized for purpose. Other
this purpose. strategies
Other strategies defined
definedthethevalues
valuesofofthe
the run-out
run-out parameters
parameters by optimization
by optimization algorithms
algorithms [25,32,33].
[25,32,33]. As aAs a general
general remark,
remark, when when applying
applying these
these methods, the cutting force signals coming from experimental data
methods, the cutting force signals coming from experimental data are utilized for setting the best are utilized for setting the
best run-out
run-out valuesvalues that minimize
that minimize an objective
an objective function.
function. Applicationsof
Applications of this
this strategy
strategyfor fortool
toolrun-out
run-out
estimations
estimations cancan
be be
foundfound in in conventional
conventional milling
milling [34–37],which
[34–37], whichdemonstrates
demonstratesthere thereisisalso
alsointerest
intereston
toolonrun-out
tool run-out
at the at
macrothe macro scale.
scale. All thisAll this research
research aims toaims to provide
provide a reliable
a reliable tool that tool
is that
able is
toable to
estimate
estimate tool run-out during micro or macro milling. From an Industry
tool run-out during micro or macro milling. From an Industry 4.0 perspective, these tools can be 4.0 perspective, these tools
can be integrated into an adaptive model designed for cutting force control [38], with the aim of
integrated into an adaptive model designed for cutting force control [38], with the aim of improving
improving the production quality and the process stability, while also reducing tool wear and
the production quality and the process stability, while also reducing tool wear and machining costs.
machining costs.
3. Geometrical Model of Run-Out
3. Geometrical Model of Run-Out
The present geometrical model of tool run-out neglects tilt angle and the location angle of tilt angle
The present geometrical model of tool run-out neglects tilt angle and the location angle of tilt
(Figure 1). This choice was made after considering that when performing micro-machining operations,
angle (Figure 1). This choice was made after considering that when performing micro-machining
the use of suitable machine tools—namely nano-precision, ultra-precision or high-precision machine
operations, the use of suitable machine tools—namely nano-precision, ultra-precision or
tools, depending on the accuracy level—is strongly suggested. These types of machine tools typically
high-precision machine tools, depending on the accuracy level—is strongly suggested. These types
guarantee tilt tools
of machine angles lower guarantee
typically than a hundredth
tilt anglesoflower
degree.
thanItaishundredth
evident that when using
of degree. standard
It is evident that or
in-house machine tools [28], this simplification is not acceptable. Figure 2 schematizes
when using standard or in-house machine tools [28], this simplification is not acceptable. Figure 2the considered
configuration;
schematizes run-out length configuration;
the considered (r0 ) and run-out anglelength
run-out (γ0 ) are
(r0)reported.
and run-out angle (γ0) are reported.

Figure 2. Micro end mill configuration with tool run-out.

Figure 2. Micro end mill configuration with tool run-out.
Micromachines 2017, 8, 221 5 of 16

In a previous work [25], the author presented a geometrical model of the cutting edge trajectories
of a two-flute micro end mill that considered tool run-out.
Three effects are due to tool run-out:

1. The radius of the first cutting edge (rCE1 in Figure 2) increases according to Equation (1);
v
d·r0 ·(cos γ − 1)

d u
rCE1 = OA = + r0 ·u
t1 + 2 (1)
2

d
2 + r0

2. The radius of the second cutting edge (rCE2 in Figure 2) decreases according to Equation (2);
v
d·r0 ·(cos γ − 1)

d u
rCE2 = OB = − r0 ·u
t1 − 2 (2)
2

d
2 − r0

3. The phase between the first and second cutting edge (referred to as cutting edge phase in the
following) changes from π to α.

Concerning the cutting edge phase (α), this parameter can be estimated as the sum (Equation (3))
between α1 and α2 angles (Figure 2), i.e., the angle between the radius of the first cutting edge and the
run-out length, and the angle between the radius of the second cutting edge and the run-out length.
These angles can be derived by Equations (4) and (5).

α = AOB
[ = α1 + α2 (3)

where " #
d
\0 r0 + 2 · cos γ
α1 = AOO = arcos (4)
rCE1
" #
\0 r0 − d2 · cos γ
α2 = O OB = arcos (5)
rCE2

As demonstrated in [25], the trajectories of the first cutting edge (A in Figure 2) and the second
cutting edge (B in Figure 2) can be expressed by Equations (6)–(9).

f
xCE1 = rCE1 · sin(ωt) + ·t (6)
60

yCE1 = rCE1 · cos(ωt) (7)

f
xCE2 = rCE2 · sin(ωt + α) + ·t (8)
60
yCE2 = rCE2 · cos(ωt + α) (9)

where f is the tool feed defined by Equation (10).

f = fz ·z·n (10)

All these equations depend on tool run-out parameters (r0 and γ0 ); for this reason, their
computation is fundamental.
In the following, it will be demonstrated how it is possible to estimate the values of tool run-out
parameters through realizing microchannels and measuring the actual tool diameter (i.e., AB), the
rotational radius of the first cutting edge (i.e., rCE1 = OA), and the cutting edge phase (i.e., α).
Micromachines 2017, 8, 221 6 of 16

Referring to the AOB triangle of Figure 2 and applying the sine law, it is possible to compute the
δ and β angles (Equations (11) and (12)).
 
AB OA OA OA
= → sin δ = · sin α → δ = arc sin · sin α (11)
sin α sin δ AB AB

β = π−α−δ (12)
Micromachines 2017, 8, 221 6 of 16
Then, applying the law of cosines (Equation (13)), it is possible to calculate the rotational radius
of the second cutting edge rCE2 ̅̅̅̅
(OB).
̅̅̅̅̅ ̅̅̅̅̅ ̅̅̅̅̅
AB OA OA OA
= → sin δ = ∙ sin α → δ = arc sin ( ∙ sin α) (11)
sin α sin δ q ̅̅̅̅AB ̅̅̅̅
AB
2 2
rCE2 = OB = OA + AB − 2·OA·AB· cos β (13)
β=π-α-δ (12)
Micromachines 2017, 8, 221 6 of 16
Then, applying 0the law of cosines (Equation (13)), it is possible to calculate the rotational radius
Now, referring to BOO triangle̅̅̅̅ of Figure 2, the tool run-out length r0 (OO0 ) can be obtained by
of the second cutting edge rCE2 (OB ).
̅̅̅̅
utilizing the law of cosines (Equation
AB ̅̅̅̅̅
OA (14)). ̅̅̅̅̅
OA ̅̅̅̅̅
OA
= → sin δ = 2 ∙ sin2α → δ = arc sin ( ∙ sin α) (11)
sin α rCE2δ = ̅̅̅̅
sin OA̅̅̅̅+ ̅̅̅̅
OB = √̅̅̅̅̅
AB ̅̅̅̅̅∙AB
AB − 2∙OA ̅̅̅̅
̅̅̅̅∙ cos
ABβ (13)
q
2 β=π-α-δ 2
0 0 tool ̅̅̅̅̅̅) can be obtained
Now, referringr0to=BOO′ = of OB
OOtriangle Figure+2, O
theB − run-out ·O0 Br·0 cos
2·OBlength (OO′ δ (12) by (14)
utilizingThen,
the law of cosines
applying (Equation
the law (14)).
of cosines (Equation (13)), it is possible to calculate the rotational radius
̅̅̅̅).
of the second cutting edge rCE2 (OB
While the tool run-out angle (γ0r )=OO'
is computed
̅̅̅̅̅̅ = √OB ̅̅̅̅̅2using
̅̅̅̅2 +O'B ̅̅̅̅the
− 2∙OB ̅̅̅̅̅sine
∙O′B ∙ cos δ
law (Equation (15)).(14)
0
̅̅̅̅ = √OA
rCE2 = OB ̅̅̅̅̅2 + ̅̅̅̅ 2
̅̅̅̅̅∙AB
AB − 2∙OA ̅̅̅̅∙ cos β (13)
OO 0
While the tool run-out angle (γ0 ) is computed using the sine law (Equation
OB OB (15)). 
BOO′ ̅̅̅̅̅̅) can OB
=
Now, referring
̅̅̅̅̅̅
OO'
→ ̅̅̅̅
to sin γ0 =
OB
triangle of Figure
̅̅̅̅
OB0
· sin δ → γ0 = ̅̅̅̅
2, the tool run-out length arc sin
r
OB
0 (OO′ be · sin δ
obtained by (15)
sin δ thesin lawγ 0
utilizing of0 cosines
= (Equation γ OO
→ sin(14)).= ∙ sin δ → γ = arc sin ( ∙ sin δ)OO (15)
sin δ sin γ0 0 ̅̅̅̅̅̅
OO' 0 ̅̅̅̅̅̅
OO'
The flow chart of Figure ̅̅̅̅̅̅ = √the
̅̅̅̅ ̅̅̅̅̅ − of2
̅̅̅̅ ̅̅̅̅̅proposed
2 (14)
The flow chart of 3 summarizes
Figure r0 =OO'
3 summarizes thesteps
OB +O'B
steps of2∙OB
thethe∙O′B ∙ cos δ
proposed procedure.
procedure.
While the tool run-out angle (γ0 ) is computed using the sine law (Equation (15)).
̅̅̅̅̅̅
OO' ̅̅̅̅
OB ̅̅̅̅
OB ̅̅̅̅
OB
= → sin γ0 = ∙ sin δ → γ0 = arc sin ( ∙ sin δ) (15)
sin δ sin γ0 ̅̅̅̅̅̅
OO' ̅̅̅̅̅̅
OO'

The flow chart of Figure 3 summarizes the steps of the proposed procedure.

Figure 3. Procedure for computing tool run-out parameters (r0 and γ0).
Figure 3. Procedure for computing tool run-out parameters (r0 and γ0 ).
The measurement of the actual tool diameter (d) and the rotational radius of the first cutting edge
(rCE1) are quite easy. The actual tool diameter can be directly measured from the tool. When realizing
The measurement of the actual tool diameter (d) and the rotational radius of the first cutting edge
channels, the rotational radius of the first cutting edge can be measured from the channel width, which
Figure 3. Procedure for computing tool run-out parameters (r0 and γ0).
(rCE1 ) are quite easy. The
corresponds to two actual
times tool diameter
the rotational can
radius of be directly
the first cuttingmeasured
edge, as shown from the tool.
in Figure When realizing
4. Several
instruments
channels, the rotational are available
radiusofof
The measurement to get
thethe these
actualfirstkinds of
cutting
tool diameter measurements,
(d) edge
and the can including:
be radius
rotational measuredmicroscopes,
from
of the first coordinate
cuttingthe
edgechannel width,
vision-measuring machines,
(rCE1) are quite easy. interferometers,
The actual tool diametercapacitive sensors,
can be directly and laser
measured fromsensors.
the tool. When realizing
which corresponds to two times the rotational radius of the first cutting edge, as shown in Figure 4.
channels, the rotational radius of the first cutting edge can be measured from the channel width, which
Several instruments areto available
corresponds to get radius
two times the rotational theseof kinds of measurements,
the first cutting including:
edge, as shown in Figure 4. Several microscopes,
instruments are available to get these kinds of measurements, including:
coordinate vision-measuring machines, interferometers, capacitive sensors, and laser sensors. microscopes, coordinate
vision-measuring machines, interferometers, capacitive sensors, and laser sensors.

Figure 4. Experimental measurement of the rotational radius of the first cutting edge (rCE1) from
channel width.

Figure 4. Experimental measurement of the rotational radius of the first cutting edge (rCE1) from
Figure 4. Experimental
channel width.measurement of the rotational radius of the first cutting edge (rCE1 ) from

channel width.
Micromachines 2017, 8, 221 7 of 16
Micromachines 2017, 8, 221 7 of 16

However,
However, measuring
measuring the the cutting
cutting edge
edge phase
phase (α)
(α) isis difficult.
difficult. Microscopes
Microscopes can can be
be used
used for
for this
this
purpose [29], but this kind of approach cannot be used in the industry
purpose [29], but this kind of approach cannot be used in the industry due to the high costs of the due to the high costs of
the measuring
measuring systemsystem (microscopes,
(microscopes, customized
customized fixtures,
fixtures, etc.),the
etc.), theneed
needofof stopping
stopping thethe cutting
cutting
operation whenever tool run-out needs to be measured, and, above all,
operation whenever tool run-out needs to be measured, and, above all, the impossibility of the impossibility of automating
this measurement.
automating this measurement.
For
For these reasons,
these reasons,an anindirect
indirectmeasuring
measuringofofthe thecutting
cuttingedge
edgephasephasefromfrom the cutting
the cuttingforce signal
force signalis
proposed. The change of cutting edge phase from π to α causes a different
is proposed. The change of cutting edge phase from π to α causes a different cutting time between cutting time between teeth.
Referring to Figure
teeth. Referring 2, the cutting
to Figure time oftime
2, the cutting the first cutting
of the edge (TCE1
first cutting edge ) is(Thigher
CE1) is than
higherthat of the
than thatsecond
of the
cutting edge (T ). Figure 5 shows the typical curve representing the cutting
second cutting edge (TCE2). Figure 5 shows the typical curve representing the cutting force signal
CE2 force signal (one round)
in milling
(one round) when using awhen
in milling two-flute
usingmicro end mill
a two-flute in theend
micro presence
mill inofthe tool run-out.ofThe
presence tooldifferent
run-out.loadThe
and cutting
different time
load and between
cuttingthetimefirst and thethe
between second cutting
first and edge arecutting
the second evident. Analyzing
edge this signal,
are evident. Analyzingit is
possible to measure the cutting time of each cutting edge.
this signal, it is possible to measure the cutting time of each cutting edge.

Figure
Figure 5.
5. Experimental
Experimental signal
signal of
of cutting
cutting force for aa two-flute
force for two-flute micro
micro end
end mill
mill with
with run-out.
run-out.

The cutting time of each tooth depends on the cutting edge phase (α), and can be estimated by
usingThe cutting (16)
Equations timeand
of each
(17):tooth depends on the cutting edge phase (α), and can be estimated by
using Equations (16) and (17):
2π
2π − − αα
TCE1== T∙
TCE1 T· (16)
(16)
2π2π
αα
TCE2== T∙
TCE2 T· (17)
(17)
2π2π
where
where
T = TCE1 + TCE2 (18)
T = TCE1 + TCE2 (18)
is the cutting period.
is the cutting period.
Then, the cutting edge phase (α) can be derived by Equation (19).
Then, the cutting edge phase (α) can be derived by Equation (19).
TTCE2
2π· CE2
α α== 2π∙ (19)
(19)
TT
4. Experimental Set-Up
4. Experimental Set-Up
4.1. Measuring Systems and Sensitivity Analysis
4.1. Measuring Systems and Sensitivity Analysis
Since the tool run-out parameters are indirectly measured, the accuracy of this measurement is
Since
relative the accuracy
to the tool run-out
of allparameters are indirectly
the measuring measured,
instruments involvedthe
in accuracy of thisprocedure.
the suggested measurement is
relative to the
In this accuracya of
research, all the measuring
confocal instruments
digital microscope involved
(Hirox in thewas
RH-2000) suggested
used forprocedure.
measuring the
actualIntool
thisdiameter
research,(OA)
a confocal digital
of the mill. microscope
This microscope(Hirox RH-2000)
guarantees was usedaccuracy
a measuring for measuring the
of 0.8 µm.
actual tool diameter ̅̅̅̅̅
(OA ) of the mill. This microscope guarantees a measuring accuracy
The channel width was measured using a fast-scanning autofocus 3D laser probe for surface texture of 0.8 μm.
The channel width was measured using a fast-scanning autofocus 3D laser probe for surface texture
measurement (Mitaka PF-60), characterized by an accuracy of 0.1 μm. As far as the cutting force is
Micromachines 2017, 8, 221 8 of 16

measurement
Micromachines(Mitaka
2017, 8, 221PF-60), characterized by an accuracy of 0.1 µm. As far as the cutting 8 offorce
16 is
concerned, a Kistler load cell (code 9317C) was used. Three Kistler charge amplifiers (code 5015A)
concerned,
amplified a Kistler
the signals load cell
coming from (code
the 9317C) waswhich
load cell, used. gave
Threethree
Kistler chargeoutput
voltage amplifiers (codeproportional
signals 5015A)
amplified the signals coming from the load cell, which gave
to the cutting force components along the machine tool axis. Then, the amplified force signals three voltage output signalswere
proportional to the cutting force components along the machine tool axis. Then, the amplified force
acquired by a National Instrument cDAQ-9174 equipped with a National Instruments 9205 board.
signals were acquired by a National Instrument cDAQ-9174 equipped with a National Instruments
This system guarantees an accuracy of the force measurement to 0.1 N.
9205 board. This system guarantees an accuracy of the force measurement to 0.1 N.
It is important to highlight that, instead of the force measurement accuracy, the sampling rate
It is important to highlight that, instead of the force measurement accuracy, the sampling rate of
of thethe
cutting
cutting force
forcesignal
signal(i.e.,
(i.e., the
the interval timebetween
interval time betweentwo two samples
samples or the
or the number
number of samples
of samples in a in a
period) affects
period) the the
affects measurement
measurement accuracy
accuracyof ofthe
the cutting edgephase
cutting edge phase(α).
(α). This
This is because
is because the cutting
the cutting
edge edge
phase (see(see
phase Equation
Equation (19))
(19)) is isa afunction
functionofof the
the ratio betweencutting
ratio between cutting edge
edge timetime
andand the cutting
the cutting
period.
period. So, a So, a higher
higher accuracyaccuracy
can be can be obtained
obtained by setting
by setting a higha sampling
high sampling
rate. rate. The limit
The limit of theofsampling
the
sampling on
rate depends ratethe
depends
acquisitionon thesystem,
acquisitionbut system, but other
other aspects aspects
have to behave to be considered
considered when the
when selecting
selecting
sampling rate.the A sampling
high sampling rate. A rate
highgenerates
sampling rate
largegenerates
storing large storing
files (up files (up
to GBs); to GBs); moreover,
moreover, the higher the
the higher the sampling rate, the higher the signal noise. In this research, the sampling rate was
sampling rate, the higher the signal noise. In this research, the sampling rate was limited to 50 kHz; an
limited to 50 kHz; an acceptable file size (less than 100 Mb) and signal noise were obtained in this
acceptable file size (less than 100 Mb) and signal noise were obtained in this manner.
manner.
Before realizing
Before the cutting
realizing the cuttingtests,tests,
an analysis of the
an analysis of influence of the
the influence of cutting speed
the cutting on the
speed on accuracy
the
of theaccuracy
proposed of the proposed procedure was done with the aim of defining the spindle speed during
procedure was done with the aim of defining the spindle speed to use to use the
experimental
during thetests. Figure 6tests.
experimental schematizes how the sensitivity
Figure 6 schematizes analysis was
how the sensitivity conducted.
analysis was conducted.

Figure 6. Flow chart of the sensitivity analysis of spindle speed influence on procedure accuracy.
Figure 6. Flow chart of the sensitivity analysis of spindle speed influence on procedure accuracy.
The first step defines a process configuration comparable to the actual process. Thus, it was
The first step
supposed to havedefines
a toola with
process configuration
a diameter of 800 μmcomparable to the
characterized by actual process.
a run-out length Thus, it was
of three
supposed to have
microns (rh), a arun-out
tool with a diameter
angle of 800
of 20 degrees andcharacterized
(γh),µm byofa 50
a sampling rate run-out
kHz tolength
acquireofthe
three microns
cutting
(rh ), a run-out angle of 20 degrees (γh ), and a sampling rate of 50 kHz to acquire the cutting force.
In this sampling rate, the interval time between two consecutive samples (∆t) is equal to 0.00002 s.
Micromachines 2017,
Micromachines 2017, 8,
8, 221
221 99 of
of 16
16

force. In this sampling rate, the interval time between two consecutive samples (Δt) is equal to
Then,
0.00002 thes. analysis
Then, theiteratively
analysis changes
iteratively thechanges
spindlethespeed, increasing
spindle speed, its value from
increasing its 500
value rounds
from per 500
minute per iteration up to 50,000 rounds per minute, which corresponds
rounds per minute per iteration up to 50,000 rounds per minute, which corresponds to the to the maximum spindle
speed
maximum of thespindle
machine toolof
speed used
the in this research.
machine tool usedForineach
thisspindle
research. speed (i.e., for
For each each speed
spindle iteration),
(i.e., the
for
correct cutting period
each iteration), (T) and
the correct the period
cutting cutting(T)
time
andofthe
each cutting
cutting timeedge (TCE1
of each and Tedge
cutting CE2 ) are
(T calculated.
CE1 and T CE2)
The following step adds and subtracts the ∆t value to these time intervals, defining
are calculated. The following step adds and subtracts the Δt value to these time intervals, defining the maximum
error achievable
the maximum in time
error estimation.
achievable in timeThe geometrical
estimation. Themodel is then model
geometrical appliedisusing these wrong
then applied usingtimes
these
for calculating
wrong times for the calculating
run-out parameters
the run-out(re and γe ). Finally,
parameters (re the
andmaximum errors
γe). Finally, theare estimatederrors
maximum applying are
Equations (20) and (21).
estimated applying Equations (20) and (21).
error r0 = re − rh (20)
error r0 = re − rh (20)
error γ0 = γe − γh (21)
error γ0 = γe − γh (21)
The results of the sensitivity analysis are reported in Figure 7. The maximum measuring errors of
The results of the sensitivity analysis are reported in Figure 7. The maximum measuring errors
tool run-out length and angle decrease as the spindle speed decreases. In particular, the run-out length
of tool run-out length and angle decrease as the spindle speed decreases. In particular, the run-out
error shows a linear trend; while the run-out angle error is quite constant for spindle speeds higher
length error shows a linear trend; while the run-out angle error is quite constant for spindle speeds
than 10,000 rpm, it shows a rapid drop for spindle speed lower than 10,000 rpm.
higher than 10,000 rpm, it shows a rapid drop for spindle speed lower than 10,000 rpm.

Figure 7. Errors in estimating the tool run-out parameters as a function of the spindle speed (d = 800 μm;
Figure 7. Errors in estimating the tool run-out parameters as a function of the spindle speed (d = 800 µm;
r0 = 3 μm; γ0 = 20°; sampling rate = 50 kHz).
r0 = 3 µm; γ0 = 20◦ ; sampling rate = 50 kHz).

In conclusion, the sensitivity analysis suggests that spindle speeds lower than 5000 rpm
In conclusion,
guarantee the lowerthe sensitivity
run-out analysis
lengths suggests
and angle that spindle speeds lower than 5000 rpm guarantee
errors.
the lower run-out lengths and angle errors.
Taking these results into account, the spindle speed was set at 4166 rpm. In this manner, 720
Taking
samples per these
periodresults into account,
(i.e., a sample thedegree
every 0.5 spindleof speed was setwere
tool rotation) at 4166 rpm. This
acquired. In this manner,
choice limits
720 samples per period (i.e., a sample every 0.5 degree of tool rotation) were
the maximum error of run-out length and run-out angle parameters. Because of the small toolacquired. This choice
limits the maximum
dimensions, error
this value ofofspindle
run-outspeed
lengthgives
and run-out angle parameters.
a low cutting speed (10.5Because
m/min).ofHowever,
the small tool
this
dimensions, this value of spindle speed gives a
cutting speed is acceptable for cutting Ti6Al4V alloy. low cutting speed (10.5 m/min). However, this cutting
speed is acceptable for cutting Ti6Al4V alloy.
4.2. Experimental Test
4.2. Experimental Test
The experimental tests consisted of milling under the same cutting conditions: micro channels
The experimental tests consisted of milling under the same cutting conditions: micro channels
on a Ti6Al4V alloy sample. The sample material was heat treated [11,12] in order to obtain a fully or
on a Ti6Al4V alloy sample. The sample material was heat treated [11,12] in order to obtain a fully or
lamellar microstructure. Figure 8a shows a micrograph of the sample after 20 s of etching with Kroll.
lamellar microstructure. Figure 8a shows a micrograph of the sample after 20 s of etching with Kroll.
Tables 1 and 2 summarize the chemical composition and the mechanical properties of the sample
Tables 1 and 2 summarize the chemical composition and the mechanical properties of the sample
material, respectively.
material, respectively.
Micromachines 2017, 8, 221 10 of 16
Micromachines 2017, 8, 221 10 of 16

(a) (b)
Figure 8. (a) Widmanstätten microstructure of Ti6Al4V sample; (b) Micro end mill
Figure 8. (a) Widmanstätten microstructure of Ti6Al4V sample; (b) Micro end mill
(SECO-JM905L008-MEGA-T).
(SECO-JM905L008-MEGA-T).

Table 1. Chemical
Table 1. Chemical composition
composition of Ti6Al4V (percentage
of Ti6Al4V (percentage in
in weight).
weight).

Ti Al V Fe O
90% Ti 6% Al V
4% Fe
<0.25% O <0.20%
90% 6% 4% <0.25% <0.20%
Table 2. Mechanical and physical properties of Ti6Al4V.
Table 2. Mechanical and physical properties of Ti6Al4V.
Parameter Value
Young’s
Parameter Modulus Value GPa
113.8
Yield
Young’s Modulus
Stress 880 N/mm2
113.8 GPa
Tensile
Yield Stress Stress 950
880 N/mm N/mm2 2

Elongation
Tensile Stress at break 950 N/mm14%2
Elongation at break
Reduction at break 14%36%
Reduction at break 36%
Hardness 36 HRC
Hardness 36 HRC
NotchNotch impact
impact 17 17
J J
Poisson’s Coefficient
Poisson’s Coefficient 0.342
0.342
3 3
Density
Density 4.430
4.430 kg/dm
kg/dm
Thermal Expansion
Thermal Expansion 8.68.6
µm/m ·K K
μm/m·
Thermal conductivity 6.7 W/m·K
Thermal conductivity 6.7 W/m·K

The tests
The testswere
wereperformed
performedonon a five-axis Nano
a five-axis Precision
Nano Machining
Precision Centre
Machining KERNKERN
Centre Pyramid Nano
Pyramid
equipped
Nano with awith
equipped Heidenhain iTCN 530
a Heidenhain iTCNnumeric control.
530 numeric A coated
control. two-flute
A coated tungsten
two-flute carbidecarbide
tungsten micro
end millend
micro (Figure 8b) (Figure
mill with a 8008b)
µm diameter
with a made 800 byμmSECO (specification
diameter made SECO-JM905L008-MEGA-T)
by SECO (specification
was used. The PVD coating was
SECO-JM905L008-MEGA-T) was nitride
used. Theof titanium and aluminium,
PVD coating was nitride a ofsuitable
titaniummaterial for cutting
and aluminium, a
titanium alloys. A confocal digital microscope (Hirox RH-2000) was used to measure
suitable material for cutting titanium alloys. A confocal digital microscope (Hirox RH-2000) was the cutting edge
radius,
used to obtaining
measure thea value of 4edge
cutting µm. radius, obtaining a value of 4 μm.
The realized
realized channels
channelsconsisted
consistedofofrectangular
rectangularslots
slotswith
witha depth
a depth of of
100100
µm, a width
μm, equal
a width to the
equal to
tool diameter, and a length of 12 mm.
the tool diameter, and a length of 12 mm.
Table 3 summarizes
summarizes the cutting parameters utilized during the tests. According to the tool data
sheet, a feed per tooth higher than the cutting edge radius radius was
was selected
selected toto avoid
avoid ploughing
ploughing regime.
regime.

3. Cutting
Table 3.
Table Cutting condition
condition utilized
utilized during
during the
the experimental
experimental tests.
tests.

Parameter
Parameter ValueValue
CuttingCutting
speed speed 10.510.5
m/min
m/min
SpindleSpindle
speed speed 41664166
rpmrpm
Feed per tooth (fz ) 10 µm/tooth
Feed per tooth (fz) 10 μm/tooth
Axial depth of cut (ap ) 100 µm
Axial depth
Radial depth of cut (aof
e)
cut (ap) 100
Toolμmdiameter
Radial depth of cut (ae)
Lubrication Tool diameter
Dry
Lubrication Dry
Micromachines 2017, 8, 221 11 of 16

An easy strategy
Micromachines was used to have different run-out conditions under the same cutting conditions.
2017, 8, 221 11 of 16
The same tool was used for realizing all the tests. At the end of each cut, the tool holder was unmounted
An easytostrategy
from the spindle check thewastool
usedwear to and
have ensure
different run-out
that the toolconditions
wear didunder the same
not affect cuttingforce
the cutting
conditions. The same tool was used for realizing all the tests. At the end of each cut, the tool holder
signal. Due to the low cutting length, no evidence of tool wear was highlighted during these control
was unmounted from the spindle to check the tool wear and ensure that the tool wear did not affect
phases. In particular, the cutting edge radius was monitored to guarantee the shear-cutting regime
the cutting force signal. Due to the low cutting length, no evidence of tool wear was highlighted
during the experiments.
during these control In fact, an
phases. excessive increase
In particular, the cutting of edge
the cutting edgemonitored
radius was radius cantochange
guaranteethethe
cutting
regime from shearing to ploughing. The conducted measures confirmed that the cutting
shear-cutting regime during the experiments. In fact, an excessive increase of the cutting edge radius edge radius
was constant
can change (4 the during
µm)cutting all the
regime fromtests. Then,tothe
shearing tool holder,
ploughing. The rotated ninety
conducted degrees,
measures was mounted
confirmed that
on thethe
spindle
cuttingandedgearadius
new channel was (4
was constant produced.
μm) during Inall
this
themanner, thethe
tests. Then, tooltool
run-out
holder,randomly changes
rotated ninety
degrees,
for each test. was
Four mounted
channels on were
the spindle
madeand a new channel
applying was produced.
this procedure, In this
obtaining manner,
four the tool
different run-out
run-out randomly changes for each test. Four channels were made applying this
conditions (i.e., Test 0, Test 90, Test 180, and Test 270). Figure 9 shows the cutting force signal measured procedure,
alongobtaining
the Y-axis fourfordifferent run-out
each test; conditions
the higher (i.e., Testis0,noticeable
tool run-out Test 90, Test
for180,
Testand
180Test 270).9c),
(Figure Figure
the 9lower
shows the cutting force signal measured along the Y-axis for each test; the higher tool run-out is
for Test 270 (Figure 9d).
noticeable for Test 180 (Figure 9c), the lower for Test 270 (Figure 9d).

(a) (b)

(c) (d)
Figure 9. Cutting force component along Y-axis. One round. (a) Test 0; (b) Test 90; (c) Test 180; (d)
Figure 9. Cutting force component along Y-axis. One round. (a) Test 0; (b) Test 90; (c) Test 180;
Test 270.
(d) Test 270.
4.3. Fitting Functions Analysis
4.3. Fitting Functions Analysis
In order to automate the procedure for the calculation of the cutting edge phase (α), different
In order tofitting
functions the cutting
automate force signals
the procedure werecalculation
for the tested. Matlab ® environment was used for this
of the cutting edge phase (α), different
purpose.
functions Since
fitting thethe force signal
cutting is used to
force signals measure
were theMatlab
tested. cutting edge’s
® time and was
environment the cutting periods,
used for the
this purpose.
fitting function analysis was applied to the force component signal along
Since the force signal is used to measure the cutting edge’s time and the cutting periods, the fittingthe Y-axis. The same
results can be indifferently obtained considering the force component along X or Z directions.
function analysis was applied to the force component signal along the Y-axis. The same results can be
In particular, for each test, a portion of the signal of the experimental cutting force
indifferently obtained considering the force component along X or Z directions.
corresponding to twenty rounds was extracted. Then, the coefficients of two fitting functions,
In particular,
namely Fourier for eachand
series test,
thea sum
portion of the
of sine signal
model, bothoflimited
the experimental cutting force
to eight components, were corresponding
calculated
to twenty rounds was extracted. Then, the coefficients of two fitting functions,
through the curve-fitting application of Matlab . The same analysis was repeated on data
® namely Fourier
filteredseries
and the sumCentral
using of sineMoving
model,Average
both limited (CMA) to eight
filter, components,
with a subset were calculated
size equal to five.through the curve-fitting
In this manner, the
application
influence ofofMatlab ® . The
the noise of thesame
cuttinganalysis was on
force signal repeated on dataoffiltered
the estimation using
the cutting Central
edge’s Moving
time and
cutting
Average (CMA)period should
filter, withbe reduced.
a subset The
size graphs
equal of Figure
to five. In this10manner,
report the
thesuperimposition
influence of thebetween
noise of the
experimental data (raw and filtered) and the sum of sine model fitting for Test
cutting force signal on the estimation of the cutting edge’s time and cutting period should be reduced. 90.
Micromachines 2017, 8, 221 12 of 16

The graphs of Figure 10 report the superimposition between experimental data (raw and filtered) and
the sum of sine model fitting for Test 90.
Micromachines 2017, 8, 221 12 of 16
Micromachines 2017, 8, 221 12 of 16

(a)

(a)

(b)
Figure 10. Sum of sine fitting function obtained from rough (a) and filtered (b) experimental cutting
Figure 10. Sum
force of sine fitting
components function
along the obtained
Y-axis (Test 90). from rough (a) and filtered (b) experimental cutting
(b)
force components along the Y-axis (Test 90).
Figure
Once 10.the
Sum of sine
fitting fitting
curves function obtained
coefficients from rough
were defined, (a) and
a Matlab filtered
® script (b) experimental
allowed cutting
the calculation of the
force components
cutting time of the along the Y-axis
first cutting (Test
edge 90).), the cutting time of ®
(TCE1 the second cutting edge (TCE2), the
Once the fitting curves coefficients were defined, a Matlab script allowed the calculation of the
cutting period (T), and, finally, the cutting edge phase (α).
cutting time
Onceofthe
thefitting
first curves
cutting edge (TCE1
coefficients ), the
were cutting
defined, time ®ofscript
a Matlab the second
allowed cutting edge (T
the calculation the), the
ofCE2
cutting period
cutting (T),
time
5. Results of and,
andthe finally,
first the edge
cutting
Discussion cutting edge
(TCE1 ), thephase (α).
cutting time of the second cutting edge (TCE2), the
cutting As
period (T), and, finally, the cutting edge phase (α).
mentioned previously, the tool diameter was measured by using a Hirox RH-2000 confocal
5. Results and Discussion
microscope (Figure 11a); while the width of the four channels was estimated as an average value of
5. Results and Discussion
As measurements, using a Mitaka
mentioned previously, PF-60diameter
the tool laser probe
wasin measured
four different
bypositions
using a along
Hiroxthe channels’
RH-2000 confocal
length.
As An example
mentioned of output
previously, of
the the
tool microchannel
diameter was profile given
measured byby the
using Mitaka
a Hirox PF-60 measuring
RH-2000 confocal
microscope (Figure 11a); while the width of the four channels was estimated as an average value of
system is (Figure
microscope shown in Figure 11b.the
Table 4 summarizes the measuring
wasresults.
measurements, using a 11a);
Mitakawhile
PF-60 width of the in
laser probe four channels
four different estimatedalong
positions as an average value of
the channels’ length.
measurements, using a Mitaka PF-60 laser probe in four different positions along the channels’
An example of output of the microchannel profile given by the Mitaka PF-60 measuring system is
length. An example of output of the microchannel profile given by the Mitaka PF-60 measuring
shown in Figure 11b. Table 4 summarizes the measuring results.
system is shown in Figure 11b. Table 4 summarizes the measuring results.

(a) (b)
Figure 11. (a) Tool diameter measuring (Hirox RH-2000); (b) microchannel profile (Mitaka PF-60).

(a) (b)
Figure 11. (a) Tool diameter measuring (Hirox RH-2000); (b) microchannel profile (Mitaka PF-60).
Figure 11. (a) Tool diameter measuring (Hirox RH-2000); (b) microchannel profile (Mitaka PF-60).
Micromachines 2017, 8, 221 13 of 16

Table 4. Measured tool diameter and width of channels.

Parameter Value
Tool diameter 802.2 µm
Channels width
Test 0 807.2
Test 90 806.2
Test 180 806.5
Test 260 805.5

Table 5 reports the R-square goodness-of-fit statistic, the cutting time of the first cutting edge
(TCE1 ), the cutting time of the second cutting edge (TCE2 ), the cutting period (T), and the cutting edge
phase (α) obtained from the different fitting functions.
The high value of the R-square goodness-of-fit statistics confirms the suitability of all the selected
fitting functions for correctly representing the cutting force signal. It is possible to use either the Fourier
series or the sum of sine model indifferently, and to process raw or filtered data. The Fourier series
applied to raw data is the preferred solution, due to the lower computational time.
In Table 5, the tool run-out parameter values (r0 and γ0 ) that were obtained applying the proposed
procedure are also reported. As a general remark, it is possible to state that for each test, the fitting
functions give very close run-out parameters. In agreement with the measured cutting force signals
(see Figure 9), the higher run-out parameters were calculated for Test 180, and the lower for Test 270.

Table 5. Fitting functions results: TCE1 , TCE2 , T, α, r0 , γ0 , SSE and R2 .

Test 0 Raw Data Filtered Data

Parameter Fourier Series Sum of Sine Fourier Series Sum of Sine
R-square 0.9645 0.9502 0.9708 0.9565
TCE1 [s] 0.007347 0.007371 0.007343 0.007373
TCE2 [s] 0.007153 0.007129 0.007157 0.007127
T [s] 0.01450 0.01450 0.01450 0.01450
α [◦ ] 182.4 183.0 182.3 183.0
r0 [µm] 8.77 10.78 8.43 10.95
γ0 [◦ ] −74.0 −77.3 −73.3 −77.6
Test 90 Raw Data Filtered Data
Parameter Fourier Series Sum of Sine Fourier Series Sum of Sine
R-square 0.9781 0.9732 0.9825 0.9776
TCE1 [s] 0.007334 0.007300 0.007295 0.007300
TCE2 [s] 0.007166 0.007200 0.007205 0.007200
T [s] 0.01450 0.01450 0.01450 0.01450
α [◦ ] 182.1 181.2 181.1 181.2
r0 [µm] 7.55 4.77 4.38 4.77
γ0 [◦ ] −75.2 −65.5 −63.1 −65.5
Test 180 Raw Data Filtered Data
Parameter Fourier Series Sum of Sines Fourier Series Sum of Sines
R-square 0.9740 0.9628 0.9787 0.9674
TCE1 [s] 0.007541 0.007531 0.007546 0.007531
TCE2 [s] 0.006959 0.006969 0.006954 0.006969
T [s] 0.01450 0.01450 0.01450 0.01450
α [◦ ] 187.2 186.9 187.3 186.9
r0 [µm] 25.36 24.49 25.79 24.49
γ0 [◦ ] −86.9 −86.7 −87.0 −86.7
Micromachines 2017, 8, 221 14 of 16

Table 5. Cont.

Test 270 Raw Data Filtered Data

Parameter Fourier Series Sum of Sines Fourier Series Sum of Sines
R-square 0.9737 0.9554 0.9780 0.9708
TCE1 [s] 0.007308 0.007252 0.007266 0.007256
TCE2 [s] 0.007192 0.007248 0.007234 0.007244
T [s] 0.01450 0.01450 0.01450 0.01450
α [◦ ] 181.4 180.1 180.4 180.2
r0 [µm] 5.29 1.66 2.16 1.73
γ0 [◦ ] −72.2 −6.0 −40.2 −18.0

6. Conclusions
The purpose of this activity has been to design and develop a procedure that was able to measure
the tool run-out length and angle parameters, and thus overcome the limits of the actual measuring
strategies based on the use of microscopes, laser or interferometers. A geometric model involving
the tool diameter, the channel width and the cutting edge phase, was introduced. For the automatic
estimation of the cutting edge phase, the cutting edge force signal was processed through a procedure
based on the curve-fitting application of Matlab® . Two fitting functions were tested by processing
raw and data filtered through the Central Moving Average filter. The limitations of this procedure
relate to the accuracy of the measuring systems used for estimating the tool diameter and the channel
width. For this reason, measuring instruments with high accuracy were used. The results obtained
applying this procedure on experimental data demonstrated the possibility of calculating the run-out
parameters with very good accuracy.
From an Industry 4.0 perspective, this procedure can be easily implemented in a cyber–physical
system that, through changing the cutting parameters as a function of the measured tool run-out, will
improve the quality of the micro-machined surfaces (i.e., surface roughness, geometrical accuracy,
and surface integrity) and reduce the process costs. For this reason, further research will be focused
on studying the influence of new materials, new tools dimensions and geometries, and new process
parameters on this procedure. Moreover, research aimed to improve the accuracy of the measuring
chain will be performed, which will make it possible to test higher cutting speeds.

Conflicts of Interest: The author declares no conflict of interest.

References
1. Dornfeld, D.; Min, S.; Takeuchi, Y. Recent advances in mechanical micromachining. CIRP Ann. Manuf. Technol.
2006, 55, 745–768. [CrossRef]
2. Dornfeld, D.; Denkena, B.; Byrne, G. Advancing cutting technology. CIRP Ann. Manuf. Technol. 2003, 52,
483–507. [CrossRef]
3. Taniguchi, N. Current status in, and future trends of, ultraprecision machining and ultrafine materials
processing. CIRP Ann. Manuf. Technol. 1983, 32, 573–582. [CrossRef]
4. Masuzawa, T.; Toenshoff, H.K. Three-dimensional micromachining by machine tools. CIRP Ann.
Manuf. Technol. 1997, 46, 621–628. [CrossRef]
5. Geiger, M.; Kleiner, M.; Eckstein, R.; Tiesler, N.; Engel, U. Microforming. CIRP Ann. Manuf. Technol. 2001, 50,
445–462. [CrossRef]
6. Alting, L.; Kimura, F.; Hansen, H.N.; Bissacco, G. Micro engineering. CIRP Ann. Manuf. Technol. 2003, 52,
635–657. [CrossRef]
7. Mian, A.J. Size Effect in Micromachining. Ph.D. Thesis, The University of Manchester, Faculty of Engineering
and Physical Sciences, School of Mechanical, Aerospace and Civil Engineering, Manchester, UK, 2011.
8. Lai, X.; Li, H.; Li, C.; Lin, Z.; Ni, J. Modelling and analysis of micro scale milling considering size effect,
micro cutter edge radius and minimum chip thickness. Int. J. Mach. Tools Manuf. 2008, 48, 1–14. [CrossRef]
Micromachines 2017, 8, 221 15 of 16

9. Komatsu, T.; Yoshino, T.; Matsumura, T.; Torizuka, S. Effect of crystal grain size in stainless steel on cutting
process in micromilling. Procedia CIRP 2012, 1, 150–155. [CrossRef]
10. Monroy, K.; Attanasio, A.; Ceretti, E.; Siller, H.R.; Hendrichs, N.; Giardini, C. Evaluation of superficial and
dimensional quality features in metallic micro-channels manufactured by micro-end-milling. Materials 2013,
6, 1434–1451. [CrossRef]
11. Attanasio, A.; Gelfi, M.; Pola, A.; Ceretti, E.; Giardini, C. Influence of material microstructures in micromilling
of Ti6Al4V alloy. Materials 2013, 6, 4268–4283. [CrossRef]
12. Gelfi, M.; Attanasio, A.; Ceretti, E.; Garbellini, A.; Pola, A. Micromilling of lamellar Ti6Al4V: Cutting force
analysis. Mater. Manuf. Process. 2016, 31, 919–925. [CrossRef]
13. Yoon, M.C.; Kim, Y.G. Cutting dynamic force modelling of endmilling operation. J. Mater. Process. Technol.
2004, 155–156, 1383–1389. [CrossRef]
14. Kim, G.; Mayor, J.R.; Ni, J. A static model of chip formation in microscale milling. J. Manuf. Sci. Eng. 2005,
126, 710–718. [CrossRef]
15. Garzòn, M.; Adams, O.; Veselovac, D.; Blattner, M.; Thiel, R.; Kirchheim, A. High speed micro machining
processes analysis for the precision manufacturing. Procedia CIRP 2012, 1, 609–614. [CrossRef]
16. Zhu, K.; Vogel-Heuser, B. Sparse representation and its applications in micro-milling condition monitoring:
Noise separation and tool condition monitoring. Int. J. Adv. Manuf. Technol. 2014, 70, 185–199. [CrossRef]
17. Bissacco, G.; Hansen, H.N.; Slunsky, J. Modelling the cutting edge radius size effect for force prediction in
micro milling. CIRP Ann. Manuf. Technol. 2008, 57, 113–116. [CrossRef]
18. Park, S.S.; Malekian, M. Mechanistic modeling and accurate measurement of micro end milling forces.
CIRP Ann. Manuf. Technol. 2009, 58, 49–52. [CrossRef]
19. Mamedov, A.; Ehsan Layegh, K.S.; Lazoglu, I. Machining forces and tool deflections in micro milling.
Procedia CIRP 2013, 8, 147–151. [CrossRef]
20. Bao, W.Y.; Tansel, I.N. Modelling micro-end-milling operations. Part I: Analytical cutting force model. Int. J.
Mach. Tools Manuf. 2000, 40, 2155–2173. [CrossRef]
21. Bao, W.Y.; Tansel, I.N. Modelling micro-end-milling operations. Part II: Tool run-out. Int. J. Mach. Tools Manuf.
2000, 40, 2175–2192. [CrossRef]
22. Bao, W.Y.; Tansel, I.N. Modelling micro-end-milling operations. Part III: Influence of tool wear. Int. J. Mach.
Tools Manuf. 2000, 40, 2193–2211. [CrossRef]
23. Afazov, S.M.; Ratchev, S.M.; Segal, J. Modelling and simulation of micro-milling cutting forces. J. Mater.
Process. Technol. 2010, 210, 2154–2162. [CrossRef]
24. Li, C.; Lai, X.; Li, H.; Ni, J. Modelling of three-dimensional cutting forces in micro-end-milling.
J. Micromech. Microeng. 2007, 17, 671–678. [CrossRef]
25. Attanasio, A.; Garbellini, A.; Ceretti, E.; Giardini, C. Force modelling in micromilling of channels.
Int. J. Nanomanuf. 2015, 11, 275–296. [CrossRef]
26. Malekian, M.; Park, S.S.; Jun, M.B.G. Modeling of dynamic micro-milling cutting forces. Int. J. Mach.
Tools Manuf. 2009, 49, 586–598. [CrossRef]
27. Rodríguez, P.; Labarga, J.E. A new model for the prediction of cutting forces in micro-end-milling operations.
J. Mater. Process. Technol. 2013, 213, 261–268. [CrossRef]
28. Zhang, X.; Ehmann, K.F.; Yu, T.; Wang, W. Cutting forces in micro-end-milling processes. Int. J. Mach.
Tools Manuf. 2016, 107, 21–40. [CrossRef]
29. Jin, X.; Altintas, Y. Prediction of micro-milling forces with finite element method. J. Mater. Process. Technol.
2012, 212, 542–552. [CrossRef]
30. Jun, M.B.G.; Goo, C.; Malekian, M.; Park, S. A new mechanistic approach for micro end milling force
modeling. J. Manuf. Sci. Eng. 2012, 134. [CrossRef]
31. Jing, X.; Tian, Y.; Yuan, Y.; Wang, F. A runout measuring method using modeling and simulation cutting
force in micro end-milling. Int. J. Adv. Manuf. Technol. 2017. [CrossRef]
32. Li, K.; Zhu, K.; Mei, T. A generic instantaneous undeformed chip thickness model for the cutting force
modeling in micromilling. Int. J. Mach. Tools Manuf. 2016, 105, 23–31. [CrossRef]
33. Lu, X.; Jia, Z.; Wang, X.; Li, G.; Ren, Z. Three-dimensional dynamic cutting forces prediction model during
micro-milling nickel-based superalloy. Int. J. Adv. Manuf. Technol. 2015, 81, 2067–2086. [CrossRef]
34. Diez, E.; Perez, H.; Guzman, M.; Vizan, A. An improved methodology for the experimental evaluation of
tool runout in peripheral milling. Int. J. Adv. Manuf. Technol. 2013, 65, 283–293. [CrossRef]
Micromachines 2017, 8, 221 16 of 16

35. Zhang, D.; Mo, R.; Chang, Z.; Sun, H.; Li, C. A study of computing accuracy of calibrating cutting force
coefficient and run-out parameters in flat-end milling. Int. J. Adv. Manuf. Technol. 2016, 84, 621–630.
[CrossRef]
36. Wan, M.; Zhang, W.-H.; Dang, J.-W.; Yang, Y. New procedures for calibration of instantaneous cutting force
coefficients and cutter runout parameters in peripheral milling. Int. J. Mach. Tools Manuf. 2009, 49, 1144–1151.
[CrossRef]
37. Wan, M.; Zhang, W.H.; Qin, G.H.; Tan, G. Efficient calibration of instantaneous cutting force coefficients and
runout parameters for general end mills. Int. J. Mach. Tools Manuf. 2007, 47, 1767–1776. [CrossRef]
38. Petruck, H.; Schlick, C.M.; Stemmler, S.; Abel, D.; Adams, O.; Klocke, F. Online parameter estimation for
model-based force control in milling processes. IFAC-PapersOnLine 2016, 49, 634–639. [CrossRef]

© 2017 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).