Contents lists available at ScienceDirect

Mechanical Systems and Signal Processing

Mechanical Systems and Signal Processing 52-53 (2015) 559–576

http://d0888-32

n CorrE-m

journal homepage: www.elsevier.com/locate/ymssp

Development and testing of an integrated rotatingdynamometer on tool holder for milling process

Muhammad Rizal a,n, Jaharah A. Ghani b, Mohd Zaki Nuawi b,Che Hassan Che Haron b

a Department of Mechanical Engineering, Faculty of Engineering, Syiah Kuala University (UNSYIAH), 23111 Darussalam,Banda Aceh, Indonesiab Department of Mechanical and Materials Engineering, Faculty of Engineering and Built Environment, Universiti Kebangsaan Malaysia,43600 Bangi, Malaysia

a r t i c l e i n f o

Article history:Received 27 June 2013Received in revised form24 July 2014Accepted 27 July 2014Available online 18 August 2014

Keywords:The rotating dynamometerCross beam type sensorStrain gaugeMilling processDrilling processCutting force measurement

x.doi.org/10.1016/j.ymssp.2014.07.01770/& 2014 Elsevier Ltd. All rights reserved.

esponding author. Tel.: þ62 603 8921 6505ail address: muhrizal@unsyiah.ac.id (M. Riza

a b s t r a c t

The cutting force provides significant information to help understand the machiningprocess, optimization, tool condition monitoring, tool design and others. Hence, variousmethods of measuring the cutting force have been proposed by many researchers. In thiswork, an innovative integrated rotating dynamometer and tool holder is designed,constructed and tested that can fulfil the requirement to measure the cutting force in awireless environment system. The device consists of a strain gauge based sensor that ismounted on a newly designed force sensing element which is then placed in the rotatingtool holder. The force sensing element is designed in the form of a symmetrical crossbeam type with four arms, shaped as a rectangular parallelepiped. This device is intendedto be used in a rotating spindle such as in milling and drilling processes. A conditioningsystem and an inductive telemetry transmitter unit are incorporated into a modified toolholder in order to collect and transmit the cutting force signal to the data acquisitionsystem. The rotating dynamometer has been subjected to a series of tests to determine itsstatic and dynamic characteristics. Thus, it is tested experimentally by conducting cuttingtests up to cutting speed 550 m/min with a single-tool insert. The results show it issuitable and reliable to measure the cutting force in milling processes.

& 2014 Elsevier Ltd. All rights reserved.

1. Introduction

The cutting force measurement is an essential requirement in the machining process. One of the most importantmachining process variables related to the cutting performance is the cutting force that is generated by the cutting tool as itcuts and shears the workpiece. It is also used as an important indicator in designing a machine tool, and for cutting processoptimization [1], investigation of the fundamental study of cutting tools performance [2], prediction of surface roughness[3], tool wear monitoring [4], prediction of chattering [5] and others.

Commonly, table dynamometers are used to measure cutting force in the milling and drilling process, where a workpieceis mounted on top of the dynamometer which is clamped to a machine tool table. The principle of the commercialdynamometer is pressure detection using piezoelectric materials that are used in dynamometer construction as the mainelement and are converted to a proportional electric charge. A table dynamometer based on a strain gauge has also been

; fax: þ62 603 8925 9659.l).

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576560

developed by Korkut [6]. He developed a three-force component analogue dynamometer which consists of four elasticoctagonal rings, on which strain gauges were mounted, clamped between the upper and lower plates, forming a platform.There were differences from the table dynamometer designed by Yaldiz et al. [7], who developed a table dynamometer tomeasure three perpendicular cutting force components and torque. The system combined the strain gauge and piezoelectricaccelerometer to measure static and dynamic cutting forces.

Nowadays, flexibility and reconfigurability are the most significant challenges for the machining process. In this regard,the application of the sensors system must have a sufficiently broad operating range to allow for various cutting tool sizesand workpiece configurations. Therefore, there has been interest in developing a rotating force-sensing system built into themachine tool structure in order to allow for efficient reconfigurability. A spindle-integrated force sensor using a piezoelectricring has been proposed for milling and drilling processes by Scheer et al. [8], Park et al. [9] and Byrne and O’Donnell [10].They placed a flange piezoelectric force ring into the spindle flange and the spindle suspension, and data were transmittedfrom the rotating part of the sensor to a stator via telemetry. Also, Totis et al. [11] proposed a rotating dynamometer using3D piezoelectric for measuring triaxial cutting force components in face milling. The sensors were clamped between themodular cartridge and the cutter body by means of a preloading screw.

However, commercial table dynamometers based on piezoelectric are commonly used for fundamentals study since theyprovide highly accurate measurement of cutting forces. However, they have a limited use in laboratory settings due tolimited workpiece geometry and dimensions, and it is also difficult to use piezoelectric sensors to measure static forces overa long period without drift since their dynamic response is influenced by the mass and geometry of the workpiece. They arealso not suitable instruments for industrial use due to their lack of overload protection and their high costs [12].Alternatively, forces can be estimated from the elastic deformation that can be measured by a strain gauge. This is a sensorwhich produces an output voltage proportional to the elastic deformation and is also small in size and mass, low in cost,easily attached, and highly sensitive to strain. In the past, not much work has been reported on rotating dynamometersusing a strain gauge as a transducer. There are several uses of a strain gauge on a rotating spindle, such as those reported byAdolfsson and Stahl [13], who have built equipment for measuring cutting force components at each cutting edge for facemilling, similarly to Smith [14], and Suprock [15], who proposed a sensor-integrated spindle for torque measurement.

This present study makes a further contribution in addressing the issues, dealing with the design and construction of anintegrated rotating dynamometer and tool holder. The force sensing element used in this system is strain gauge-based andof a cross beam type, with a data transmitter using inductive telemetry. It is capable of measuring the main cutting force, Fc,the thrust force, Ft and the perpendicular cutting force, FcN, in milling and also in drilling operations. The advantage of thisrotating dynamometer is its flexibility as it can be assembled with a variety of cutting tools size and geometries.

2. Force component in milling operation

Fig. 1 shows the cutting force generated when the milling cutter cuts the workpiece material. The direction of the activeforce (Fa) changes with the entry angle φ. The components of the active force have two directions inclusive of direction ofcutting speed (vc) and feed rate (vf). The components cutting force (Fc) and perpendicular cutting force (FcN) are related to aco-rotating system of coordinates, with their directions parallel to the cutting speed and toward the centre of the spindle.The components feed force (Ff) and perpendicular feed force (FfN) are related to a fixed system of coordinates. The forces thathave a relationship with tool-specific components are cutting force and perpendicular cutting force. The thrust force (Ft) alsohas a direct relationship with the tool system because the workpiece is against the cutting tool and its direction is always inthe vertical axis. For converting the active force from the fixed system of coordinates into a co-rotating system, the followingequations apply [16].

Fc ¼ Ff cos φþFfN sin φ ð1Þ

Fig. 1. Components of the cutting force in face milling.

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576 561

FcN ¼ Ff cos φþFfN cos φ ð2Þ

Fx ¼ Ff ð3Þ

Fy ¼ FfN ð4Þ

Fz ¼ Ft ð5Þ

3. Milling's rotating dynamometer

3.1. Design and model of force sensing element

In order to design the structure of the integrated rotating dynamometer and rotating tool holder, important factors ofgeometry, size, stiffness, stability and accuracy of measurement were considered to ensure their adaptability with thewireless system. Fig. 2 illustrates the geometrical design of the force sensing element. The force sensing element of therotating dynamometer is designed in the form of a cross beam type. It consists of a central shaft that provides a connectionto the tool holder, cross and compliant beams as the sensing element and the bottom ring as a distributor of forces to thesensing elements. The cross beam has four symmetric horizontal beams, as well as the compliant beam. The sizes of thesensing element were determined, including height of compliant beams (h) 16 mm, horizontal length of cross beams (l)17 mm, sectional dimension of beams 8 mm in thickness (t), 8 mm in width (b), 6 mm in width (b1) and width at the basenear the shaft is 9.8 mm.

The properties of the force sensing element are defined by its material and design. There are several factors for selectingthe material of the force sensing element, including environmental concerns, the magnitude of the force, mechanicalintegration, rigidity, high natural frequency and corrosion resistance [6]. In this work, stainless steel grade 304 was chosenbecause it satisfies the above criteria. The mechanical properties of the force sensing element material are summarizedin Table 1.

In designing a force sensing element that is capable of measuring the cutting force and integrated in the spindle toolholder in the milling and drilling operation, the maximum exerted force, which is the cutting force (Fc) in a rotatingdirection, is assumed as 2000 N. The thrust force (Ft) due to spindle pressure on the workpiece is assumed as approximately3000 N in the direction of the z-axis. Because the milling processes have a translation movement due to the feed rate, aperpendicular cutting force, (FcN), will occur along the motion and a maximum value of approximately 2000 N is assumed.Fig. 3 illustrates the directions of the exerted forces that are applied to the force sensing element.

Fig. 2. Model of force sensing element (a) perspective view and (b) dimensions.

Table 1Properties of stainless steel grade 304.

Properties Values

Density 8000 kg/m3

Poisson ratio 0.3Modulus of elasticity 193,000 N/mm2

Tensile strength 515 N/mm2

Yield strength 205 N/mm2

Hardness 201 HB

Fig. 3. Directions of applied force on the force sensing element.

Fig. 4. Schematic diagram of a cantilever beam.

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576562

3.2. Theoretical analysis of the static properties

The geometry of the force sensing element in this study is a cross beam that is identical in shape to the cantilever beams.Deformation of the cantilever beam caused by the force applied to it causes a strain value change in the surface, asillustrated in Fig. 4.

The stress–strain relationship of elastic deformations of the cantilever beam will be governed by Hooke's law [17]:

σ ¼ Eε ð6Þwhere σ is the stress, E is the modulus of elasticity, and ε is the strain. For a beam subjected to a moment force, themaximum stress is defined as

σc ¼McI

ð7Þ

I¼ bt3

12; c¼ t

2ð8Þ

where M is the moment (FL), c is the distance from the neutral axis to the edge of the beam, I is the moment of inertia, b isthe width of the cantilever beam and t is the thickness of the cantilever beam. By substituting the following equations intoHooke's law in Eq. (6), the relationship between the strain and applied force can be attained as

ε¼ σE¼ 6FLEbt2

ð9Þ

When the main cutting force (Fc) is applied in a direction parallel to the cutting speed as described in Fig. 1, the bottomring of the force sensing element will transmit the force or moment to the cross beams through the compliant beam.Because the cross beams are symmetrical, they will be deformed uniformly into four horizontal cross beams and can beexpressed as

F ¼ Fc4

ð10Þ

The moment equilibrium condition is at the central shaft, so the length of moment is the resultant distance from thecross beam to the bottom ring, and can be written as

L¼ ðl2þh2Þ1=2 ð11ÞThe strain rate that occurs on the surface of the cross beam due to the cutting force action can be simplified by

substituting Eqs. (10) and (11) into Eq. (9), which can be written as

εFc ¼3Fc

2Ebt2ðl2þh2Þ1=2 ð12Þ

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576 563

When the thrust force and perpendicular cutting force are applied to the bottom ring of the force transducer, themoment equilibrium is still at the central shaft, but the length of the moment is the horizontal and vertical length of thecross beam, and the strain rate due to thrust and perpendicular force can be expressed as

εFt ¼3Ftl2Ebt2

ð13Þ

εFcN ¼3FcNh2Ebw2 ð14Þ

By using Eqs. (12)–(14) and determined variables of the force sensing element, the elastic strains can be obtained asfollows:

εFc ¼3Fc

2Ebt2ðl2þh2Þ1=2

εFc ¼3 � 2000

2 � 193;000 � 8 � 9:82ð172þ162Þ1=2

εFc ¼ 4:72� 10�4

εFt ¼3Ftl2Ebt2

εFt ¼3 � 3000� 17

2 � 193;000� 9:8� 82

εFt ¼ 6:32� 10�4

εFcN ¼3FcNh2Ebw2

εFcN ¼3� 2000� 16

2� 193;000� 8� 62

εFcN ¼ 8:64� 10�4

The stress occurring on the cross beam of the force sensing element caused by the main cutting forces, thrust force andperpendicular cutting force can be obtained by using Eq. (6), that are 91.1 MPa, 121.9 MPa and 166.7 MPa. The value of themaximum stress on the force sensing element is smaller than the yield strength and tensile strength of this material. Itmeans that the force sensing element is in safe margin.

3.3. Theoretical analysis of the dynamic properties

The dynamic responses of a force sensing element subjected to the forces exerted during the machining process areimportant information in the design and development process. The vibration frequency of the machine tool occurswhen a cyclic exciting force is applied to an elastic structure like a force sensing element. It has a great impact on themachining process when one or more of the frequencies of the cyclic shock and varying cutting force are equal or closeto one or more natural frequencies of the force sensing element [18]. In order to ensure stability, the force sensingelement should have a natural frequency at least four times the vibration frequency of the machine tool [19]. Thevibration frequency (fe, Hz) in the machining process is related to the spindle speed (n, rpm) of the machine tool, whichcan be expressed as follows [19]:

f e ¼n60

ðHzÞ ð15Þ

This dynamometer is designed to be applied with a spindle speed of 5000 rpm, so the machine tool vibration frequencyis

f e ¼500060

¼ 83 Hz

Therefore, the natural frequency of the force sensing element should be given by

f Z4� 83 Hz¼ 333:33 Hz

In order to determine the natural frequency of the force sensing element, the stiffness of the structure material should bedetermined based on the force displacement relationship. The stiffness can often be described by a simple spring model. A

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576564

spring has the characteristic that force is the function of deformation, which is shown as

F ¼ kðxÞ ð16Þwhere F¼ is the excited force (N), k is the spring constant or stiffness (N/m) and x¼δ is the displacement (m). In this work,the stiffness of the designed force sensing element depends on the directions of the applied forces on the structure, whichcan be described by the cantilever beam and its free body diagram as shown in Fig. 5.

The displacement that occurs at point B in the direction of Fc is given by

δ¼ ∂U∂F

ð17Þ

Element BC in the model is in bending only, so it can be expressed by

δBC ¼∂U∂F

¼Z

1EI

M∂M∂F

� �dy¼ 1

EI

Z h

0�Fyð Þ �yð Þdy¼ Fh3

3EIBCð18Þ

Element CD in the model is in bending and torsion. The torsion is constant so the deflection due to torsion can be writtenas

δCD ðtorsionÞ ¼∂U∂F

¼ T∂T∂F

� �lGJ

¼ ðFhÞðhÞ lGJ

¼ Fh2lGJ

ð19Þ

For the bending in CD,

δCD ðbendingÞ ¼∂U∂F

¼Z

1EI

M∂M∂F

� �dx¼ 1

EI

Z l

0ð�FxÞð�xÞdx¼ Fl3

3EICDð20Þ

By adding Eqs. (18)–(20) and noting that J¼2I, G¼E/[2(1þv)], the displacement at point B is δ (mm) or can be denoted asδFc and be obtained as follows:

δFc ¼∂U∂Fi

¼ Fch3

3EIBCþFch

2lð1þvÞEICD

þ Fcl3

3EICDð21Þ

δFc ¼Fc3E

h3

IBCþ3h2lð1þvÞ

ICDþ l3

ICD

" #ð22Þ

By substituting Eq. (16) into Eq. (22) and inputting all the values of the force sensing element dimensions, it becomes:

kFc ¼FcδFc

¼ 2000

4:39� 10�2 ¼ 45;519 N=mm

Similarly to the displacement due to the thrust force and perpendicular cutting force, the stiffness beam in the thrustforce direction is affected by compression in the element BC and bending in the element CD and it can be written as

δFt ¼∂U∂Fi

¼Z

1AE

F∂F∂Fi

� �dxþ

Z1EI

M∂M∂F

� �dy ð23Þ

δFt ¼Fth

3

EABCþ Ftl

3

3EICDð24Þ

kFt ¼FtδFc

¼ 3000

1:65� 10�2 ¼ 181;649 N=mm

For the displacement due to the perpendicular cutting force, its direction is parallel with axis x, so the amount ofdisplacement can be obtained by adding displacement element BC due to the force bending and displacement element CDdue to moment bending, and also displacement on cross beam due to cutting force which is obtained in eq. (22), it can be

Fig. 5. Force action on the beam model and its free body diagram.

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576 565

written as

δFcN ¼ 2FcNh

2lEICD

þFcNh3

3EIBCþFcNh

3

3EIBCþFcNh

2lð1þvÞEICD

þ FcNl3

3EICD

" #ð25Þ

kFcN ¼FcNδFcN

¼ 2000

1:9� 10�1 ¼ 10;475 N=mm

The natural frequency of the force sensing element, which is assumed to be a small mass supported by ring elements, canbe obtained from the following relation [19]:

f ¼ 12π

ffiffiffiffiffiffiffiffiffik=m

qð26Þ

where k is the force sensing element stiffness (N/m); m is the force sensing element mass (kg) and f is the natural frequencyof the force sensing element (Hz).

From the design and construction of the force sensing element, we found the weight of the force sensing element and itscomponent of instrumentation to be 1.2 kg. Placing all the stiffness values of the sensing element based on each direction,the natural frequencies of the sensing element are calculated as

f Fc ¼12π

ffiffiffiffiffiffiffiffiffiffiffiffiffikFc=m

q¼ 12� 3:14

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi45519� 1000

1:2

r¼ 980:7 Hz

f Ft ¼12π

ffiffiffiffiffiffiffiffiffiffiffiffiffikFt=m

q¼ 12� 3:14

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi181649� 1000

1:2

r¼ 1959:2 Hz

f FcN ¼12π

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffikFcN=m

q¼ 12� 3:14

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi10475� 1000

1:2

r¼ 470:5 Hz

It was previously stated that fZ4fe. It can be observed that all the values of the natural frequency of the force sensingelement are still above the machine tool frequency. Therefore, the force sensing element that has been designed meets thesafety factor requirements.

3.4. Strain gauges arrangement

The force sensing elements based on strain gauges are referred to as spring element. When an external load is applied tothe force sensing element, the stress and strain within the surface of the material will change. The strain gauge will convertthese changes of strain into voltage signals that represent the applied force. When the strain gauges are mounted on theforce sensing element, they will undergo changes in the strain value or resistance that reflect the change of stress due to theapplied load or forces. This change is normally expressed in terms of an empirically determined parameter called the gaugefactor, GF. It is also a fundamental parameter for strain sensitivity, and can be expressed as [17]

GF ¼ΔR=RΔL=L

¼ΔR=Rε

ð27Þ

where R is the original resistance of the strain gauge, L is the original length, and ε is the strain detected by the strain gauge.In order to achieve a maximum sensitivity, the location of the point where the strain gauges are mounted is critical.

There are three orientations of strain gauge locations for measuring the main cutting force, thrust force and perpendicularcutting force. Fig. 6 shows the mounted location of the strain gauges on the force sensing element for three directions offorce components. For each channel, a full Wheatstone bridge circuit is used by constructing the full bridge circuit as shownin Fig. 7.

The arrangement strain gauge for detecting the main cutting force Fc is described in Fig. 7(a). The strain gauge R1, R3,R5 and R7 are subjected to tensile stress while R2, R4, R6 and R8 are subjected to compressive stress, respectively. Thethrust force Ft is detected by R9, R10, R11 and R12 in which compressive stress occurs to these strain gauges as shown inFig. 7(b), while R13, R14, R15 and R16 are subjected to tensile stress. The perpendicular cutting force FcN is supported byfour beams with strain gauge arrangement as shown in Fig. 7(c). The compressive stress occurs on strain gauge R17, R18,R23 and R24, while strain gauge R19, R20, R21 and R22 are subjected to tensile stress. All strain gauges used in thepresent study are general purpose linear strain gauge (SGD-3/350-LY11) bought from Omega Company with thenominal resistance of 350 Ω. The gauge factor of this gauge is 2.14, the length of the gauge is 3 mm, and the width of thegage is 3.2 mm. In order to convert the rated strain in theoretical analysis (mm/m) and the rated output value in voltagesignal (mV/V), the equation is as follows:

Vo

Vi¼ 14GFε ð28Þ

Fig. 7. Strain gauge arrangement on Wheatstone bridges circuit: (a) channel on Fc; (b) channel on Ft; and (c) channel on FcN.

Fig. 6. Location of the strain gauge on the force transducer.

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576566

where Vi is the input voltage of the Wheatstone bridge (fixed 4 V), Vo is the output voltage of a Wheatstone bridge, GF isthe factor of strain gauge, and ε is the rated strain from theoretical analysis. By using Eq. (28), the rated strains in theoryare changed into rated output of electrical signal in mV/V.

3.5. Theoretical and numerical analysis of cross sensitivities

The dynamometer was developed by a single force sensing element to measure the cutting force in three channelssimultaneously. Thus, the structure deformation on the force sensing element will affects the sensitivity in each channel.The cutting force signals collected from a three-axis dynamometer in the directions other than the direction of forceapplication are known as cross sensitivity. However, it is usually impossible to get accurate results of cross sensitivity,because the cross sensitivity is caused by multi-external factors. This study attempts to analyse cross-sensitivity byconsideration of geometry and load factor on the force sensing element. Therefore, cross sensitivity can be defined as theratio of apparent load measured on the primary axis to actual load applied on the secondary axis as given in the followingequation [20].

CS¼ Apperant Primary LoadApplied Secondary Load

ð29Þ

The structure deformation due to the moment of a force about an axis will affect in all directions as modelled in Fig. 8. Inthis study, the cross sensitivity analysis took one arm of the cross beam since the force sensing element is symmetry. Thepoints on the beam refer to the placement of strain gauge locations. When force Fc is applied at the end of beam BC, thepoint M in the z-axis is subjected by moment bending, but point N and O will be subjected by the moment torsion. The shearstrain that occurs in point N and O can be calculated as follows.

γN ¼ 12Fchð1þvÞEbCDðbCD2þtCD2Þ

γO ¼ 12Fclð1þvÞEbBCðbBC2þtBC2Þ

ð30Þ

According to Fig. 6 that all strain gauges were mounted in symmetrical and using a linear gauge. So, shear strains fromEq. (31) are detected by gauge at point N and O would be considered as misalignment error with respect to the intended axis

Fig. 8. A single arm model are subjected to three axis of force.

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576 567

of the main cutting force measurement. Therefore, the strain measured misaligned by an angle θ from direction of main thenormal strain. Thus, the above equation can be written as follows:

εN ¼ 6Fchð1þvÞEbCDðbCD2þtCD2Þ

ð1�νÞþð1þνÞcos 2θ� �εO ¼ 6Fclð1þvÞ

EbBCðbBC2þtBC2Þð1�vÞþð1þvÞcos 2θ� � ð31Þ

When force Ft is applied at the bottom of beam BC, the point M in the neutral axis of beam CD, so the strain in this pointis close to zero. But, beam BC in y-axis is subjected by axial stress. So, the axial strain that occurs in point O can be calculatedby the following equations:

εM ¼ 0 εO ¼ FtEABC

ð32Þ

Likewise force FcN if applied in the end of beam BC in x-axis direction, the pointM and N are subjected by axial stress. Thenormal strain that occurs in point N and O can be calculated by the following equations:

εM ¼ FcNEACD

εN ¼ FcNEACD

ð33Þ

Generally, the force sensing element can be considered as a linear system, so, the relationship can be described asfollowing:

S¼ C½ �F ð34Þ

SFcSFtSFcN

264

375¼

C11 C12 C13

C21 C22 C23

C31 C32 C33

264

375

FcFtFcN

264

375 ð35Þ

where F is the vector of the input component force value [Fc Ft FcN], [C] is an strain compliance matrix and S is the outputvector of voltage value. The output voltage values of the force sensor are obtained by loading the known single dimensionforce. When main cutting force Fc is imposed at tangential direction of sensing element, output voltage value SFc, SFt and SFcNare generated by all directions force. Simultaneously, when applying force in other direction, output voltage value also canbe obtained by all channels. So, by using the results from analytical model and Eqs. (28–33), the sensitivity matrix intheoretical analysis can be obtained as

CTheory ¼0:5051 0:0119 0:0248

0 0:4508 0:03850:0385 0:0385 0:9247

264

375

According to above matrix, the values of other two directions are close to zero, when applied single-axis force on onedirection. It is obvious that the matrix [C] is approximately a diagonal matrix. The cross sensitivity errors from thetheoretical analysis was described in Table 2. It is obvious that the cross sensitivity error of FcN component showed the largevalue 8.54%, while other components was less than 5.0%.

In order to quantify the strain distribution and cross sensitivity using numerical calculation, ANSYS was used to performthe static analysis of the force sensing element subjected to three directions of force. Fig. 9(a) shows the finite element meshof force sensing element. The model was covered with isoparametric triangular meshes, and it generated the optimummeshconsisted of 61,925 nodes and 35,009 elements. Because the model is circularly symmetric, the central shaft of sensingelement was selected to fix, and applied forces were uniformly distributed in whole plane of the bottom ring. The maincutting force (Fc) that is rotating direction was simulated, assuming that the maximum cutting force was 2 kN. The thrustforce (Ft) due to spindle pressure to workpiece is determined approximately 3 kN with direction to y-axis. The perpendicular

Fig. 9. Finite element analysis (a) mesh model; (b) strain distribution due to force Fc; (c) strain distribution due to force Ft; and (d) strain distribution due toforce FcN.

Table 2Cross sensitivity error rotating dynamometer based on theoretical analysis.

Force orientation Cross sensitivity error (%)

Fc Ft FcN

Fc – 2.35 4.91Ft 0 – 8.54FcN 3.82 3.82 –

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576568

cutting force (FcN) will occur along the motion and it is assumed a maximum value of approximately 2 kN. A deformationanalysis was performed, and normal elastic strain distributions occurred on the surface of the cross beams due to the maincutting force, the thrust force and the perpendicular cutting force are shown in Fig. 4(b, c and d), respectively. By usingEq. (28), the sensitivity matrix in based on FEM analysis can be obtained as

CFEM ¼0:4911 0:0024 0:00850:0064 0:4279 0:00280:0381 0:0407 0:7939

264

375

Table 3 shows the calculation results of the cross sensitivity errors from FEM analysis. When the force was simulated inFc-direction, the cross sensitivity error for Ft and FcN-directions was calculated as 0.49% and 1.72%. While the simulation wasbeing carried out on Ft and FcN-directions, respectively, the cross sensitivity error was 1.48% and 0.66%, and 2.88% and 3.32%.

3.6. Construction and fabrication of the rotating dynamometer

The construction of the model rotating dynamometer integrated in the tool holder spindle of the milling process isillustrated in Fig. 10. The components of the complete system comprise a standard tool holder spindle, force sensingelement, bottom shaft, the tool's modular of face milling, cover, top cover plate, bottom cover plate and mounting of thestrain gauge module. The force sensing element is attached to the bottom of the tool holder spindle. The tool's modular is

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576 569

joined to the sensing element through the bottom shaft. The cover is mounted outside the top and bottom plate, and thenjoined to the shaft of the tool holder spindle. The cover also serves for wrapping the coil of the telemetry transmitter system.The forces that occur in the cutting tool will be transmitted through the bottom shaft and will deform the sensing elementand then be detected by the strain gauge.

The force sensing element has been built using a CNC machining process on a stainless steel bar with diameter 100 mm.The sensing element prototype is shown in Fig. 11, together with its components to support the measurement system usingthe telemetry system. Fig. 12 shows the photograph of the complete developed rotating dynamometer.

The signals coming from the force sensing element in the rotating dynamometer are amplified by an acquisition modulefor the strain gauge (MT23-STG) and then the analogue signals are converted to a digital output. By using a transmittermodule (MT32-IND-Tx-45 MHz-2560k), the signals transmit within a frequency of 45 MHz and a transmission rate up to2560 kb/s. The sensor and transmitter modules are mounted in the space that is integrated with the standard tool holder. Atelemetry receiver (MT32-DEC8) and data logger (DT9836) were used to collect the signals. The acquisition, visualizationand processing of the collected signals was performed via MATLAB software.

4. Testing of rotating dynamometer

4.1. Static calibration test

Calibration test is a process to determine the relationship between the input and output data. Static calibration was doneto investigate the performance of the force transducer after design and construction using a servohydraulic testing system(Instron 8874). The loading force for detection of the main cutting force Fc is in the tangential direction the force sensingelement which is converted into a torque. The loading of the perpendicular cutting force FcN is in the horizontal directionand perpendicular to shaft centre, while to detect the thrust force Ft is in the vertical direction. According to design analysisof force sensing element, the calibration loads are based on full scale output (FSO) that are 2000 N for Fc, 3000 N for Ft and2000 N for FcN with an incremental step of 100 N. The output voltage values in millivolts were averaged and recorded foreach load interval. Then, calibration curves were obtained to convert the output voltage readings into force values. Figs. 13–

Fig. 10. Model structure of an integrated rotating dynamometer and tool holder.

Table 3Cross sensitivity error rotating dynamometer based on FEM analysis.

Force orientation Cross sensitivity error (%)

Fc Ft FcN

Fc – 0.49 1.72Ft 1.48 – 0.66FcN 2.88 3.32 –

Fig. 12. Photograph of a completely rotating dynamometer.

Fig. 11. Photograph of the fabricated prototype (a) force sensing element and (b) supporting components of the measurement system.

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576570

15 show the calibration curve for the main cutting force, the thrust force and the perpendicular cutting force, respectively.The measurements were repeated five times to verify the consistency and the average values were recorded in curves. Thecalibration matrix from the experiemental results was obtained as follows:

CExp ¼0:4981 0:0003 0:00780:0069 0:4145 �0:00020:0289 �0:0338 0:8356

264

375

According to the calibration curve and matrix, it is obviously that the rotating dynamometer's sensitivities were obtainedabout 4.98�10�4 mV/N, 4.23�10�4 mV/N and 8.53�10�4 mV/N. Table 4 shows cross sensitivity errors for the Fc, Ft andFcN directions. The maximum error shown in the FcN component was about 4.05% in the Ft direction, while for othercomponents it did not exceed 2%.

Fig. 16 shows the comparison of cross sensitivities from the theoretical, numerical and experimental studies. It can beobserved that results of analytical study obtain the highest cross sensitivity errors. It is caused by many factors includingmodelling a single beam on an analysis and its boundary condition. However, the numerical and experimental results arevery well correlated. If the cutting force components are compared, it is seen that in the Fc component shows a goodagreement between the theoretical, numerical, and experimental results. When comparing the overall results of the analysisof the cross sensitivity errors, it appears that the error does not exceed a maximum of 8.5% and an experimental basis was

Fig. 14. The force transducer calibration curve in Ft direction.

Fig. 13. The force transducer calibration curve in the Fc direction.

Fig. 15. The force transducer calibration curve in FcN direction.

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576 571

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576572

only about 4.05%. It means that the rotating dynamometer is acceptable to use for measurement of cutting force and theresult is in agreement with Jun et al. [21] which found the spindle-based force sensor has a cross sensitivity error maximumof 12%.

4.2. Dynamic calibration test

When the rotating dynamometer was mounted on the machine, the cutting forces that occurred were not static, as thesystem response of dynamic excitations should have been taken into consideration. The dynamic response was affected bythe natural frequencies of the rotating dynamometer. The natural frequency must be higher than the frequency of excitingvibration during the machining process in order to ensure that the recorded cutting force signal is not influenced by thedynamic response of the rotating dynamometer [19]. The natural frequencies of the rotating dynamometer weredetermined by a frequency response function that was obtained by means of experimental modal analysis [22]. Therotating dynamometer was excited by using a modal impact hammer Endevco type 3012, and an accelerometer (Endevco751-100) was connected to the component of the dynamometer. The signals were acquired by a pulse analyser and modalanalysis was performed in order to derive the frequency response function of the rotating dynamometer for the three

Fig. 16. Comparison between the theoretical, numerical and experimental results for cross sensitivity errors.

Table 4Cross sensitivity error rotating dynamometer based on experimental analysis.

Force orientation Cross sensitivity error (%)

Fc Ft FcN

Fc – 0.92 1.57Ft 1.61 – 0.19FcN 3.46 4.05 –

Fig. 17. Frequency response for main cutting force (Fc) direction.

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576 573

directions of force measurement. Figs. 17–19 show the frequency response functions for the main cutting force direction,thrust force direction and perpendicular cutting force direction. It is observed that the natural frequencies of the rotatingdynamometer are approximately 1050, 2079 and 450.9 Hz.

In order to analyse the interference error of dynamic characteristics from the theory and testing analysis, Table 5 showsthe results of comparison the natural frequency between theory calculation and testing analysis. The natural frequenciesfrom theory analysis are obtained that are 980.7, 1959.2 and 470.5 Hz. If these results are compared to that from testinganalysis, and their range of error is less than 7.1%. It means that the theoretical calculation and testing have a goodagreement.

4.3. Thermal effects test

In the milling process, the rubbing between cutting tool and workpiece may increase the temperature. This energy willpropagate to the tool holder and causes temperature fluctuations around the force sensing element which is attached in themiddle of tool holder. However, the fluctuations of temperature in the tool holder are a few degrees, but immanent andcauses the deformation of structure of machine tool. When milling under dry cutting condition, the temperature increases

Fig. 19. Frequency response for perpendicular cutting force (FcN) direction.

Table 5Comparison of natural frequency in theory and testing analysis.

Direction orientation Analysis Natural frequency (Hz) Error (%)

The main cutting force, Fc Theory 980.7 7.1Testing 1050

The thrust force, Ft Theory 1959.2 6.1Testing 2079

The perpendicular cutting force, FcN Theory 470.5 4.2Testing 450.9

Fig. 18. Frequency response for thrust force (Ft) direction.

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576574

and sensing element enlarges. Conversely, when using the coolant or cryogenic fluids, leads to the temperature drop andsensing element shrinks. These changes cause the sensor readings disturbed and can be interpreted as falsely force.Relationship of temperature fluctuates toward deformation of structure can be expressed as

ε¼ αðΔTÞ ð36Þwhere α is the coefficient of thermal expansion and ΔT is the temperature change, in degrees. Since the force sensingelement is made out of stainless steel, which has a coefficient of thermal expansion is 17.3�10�6 1C�1, an increase intemperature of 1 1C theoretically expands the structure approximately 17.3 mm/m. By using Eq. (30), the strain occurredwhen temperature increase about 5 1C is 8.65�10�5 mm/mm. According to theoretical analysis, the stiffness of forcesensing element in direction Fc, Ft, and FcN are 45,519 N/mm, 181,469 N/mm and 10,475 N/mm, respectively. So, force offsetchanges due to increase of temperature can be obtained about 3.94 N, 15.71 N and 0.91 N. It is a small change of static forceoffset and does not affect the dynamic force measurement.

An experimental test was also conducted to examine the thermal effects of force sensing element. A controlled heater hasbeen used to warm up the tool tip from 25 1C to 30 1C. Two thermocouples have been placed at the bottom and the middleof force sensing element. Force and temperature signals were simultaneously recorded with a sampling rate of 100 Hz for15 min. Fig. 20 shows the static force offset caused by expansion of force sensing element whenwarming it until 30 1C. It canbe seen that static force quantities show a linear correlation. When the temperature rises up to 30 1C, the static force offsetof Fc, Ft, and FcN changed to 4.64 N, 14.45 N and 1.37 N, respectively. These readings compared to the theoretical analysisshowed almost the same results.

4.4. Cutting test

The cutting test is important in dynamometer development to evaluate the performance of cutting force measurement inreal machining operation. Several cutting tests were carried out by end milling AISI P20 tool steel using a single insert

Fig. 20. The changes of static force offset due to temperature increase.

Fig. 21. Plots of cutting force signals in time and frequency domain at a cutting speed of 200 m/min.

Fig. 22. Plots of cutting force signals in time and frequency domain at a cutting speed of 375 m/min.

Fig. 23. Plots of cutting force signals in time and frequency domain at a cutting speed of 550 m/min.

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576 575

coated tungsten carbide (Sumitomo AXMT170504PEER-G) with coated grade ACP200 which is placed on modular of cuttingtool with diameter 40 mm. These experiments were performed using a DMC 635 V eco CNC milling machine under drycutting condition with radial depth of cut (ae) of 0.4 mm, axial depth of cut (ap) of 1 mm, feed rate (fz) of 0.2 mm/tooth andvaried cutting speed (vc) that are 200 m/min, 375 m/min and 550 m/min. The cutting forces measured using a wirelesstelemetry system at a sampling rate of 5 kHz. The acquired cutting force signals are given in Figs. 21–23. These figures showthat the top graph is plot of dynamic cutting forces in time domain. The bottom graph of each of these figures is plot thechange of tool passing frequency throughout the machining process.

In Fig. 21 shows the results of measured cutting forces for a spindle speed of 1592 rpm (vc¼200 m/min), generated thetool passing frequency is 27 Hz. When the spindle speed is increase to 2986 rpm (vc¼375 m/min), the tool passingfrequency is 50.5 Hz, see Fig. 22. At 4379 rpm (vc¼550 m/min), the tool passing frequency is also increases to 73.7 Hz, seeFig. 23. However, these frequencies are lower than the low natural frequency of 470.5 Hz, it indicates that this dynamometeris safe to use in operations below 5000 rpm. In time domain results, it is apparent from the graphs that the main cutting

M. Rizal et al. / Mechanical Systems and Signal Processing 52-53 (2015) 559–576576

force, Fc is higher than the thrust force, Ft and the perpendicular cutting force, FcN. This is caused by the radial depth of cut issmall which resulted in the contact area on the direction of thrust and perpendicular cutting forces becomes small. Whilethe axial depth of cut of 1 mm, making a large contact area in the direction of the main cutting force, so that the maincutting force is becoming higher. By analysing the measured cutting force signals, it can be evaluated that the influences ofcutting speed, entrance and exit conditions of cutting tool from the workpiece and instantaneous cutting force readings areobviously visible in the signals.

5. Conclusion

In this work, an innovative strain gauge-based rotating dynamometer for the milling and drilling process was developed. Thedevice system is capable of measuring the main cutting force that has a direction parallel with the cutting speed of the spindle, thethrust force and also the perpendicular cutting force. The cutting tools are interchangeable and the device is compatible withdifferent standard modules, so supporting the system that provides a flexible and reconfigurable machining process. Thisdynamometer was designed to measure cutting forces up to 3000 N and the rotation up to 5000 rpm. The characteristics tests weredone in order to evaluate the performance of the developed rotating dynamometer including static, dynamic, thermal effects andreal machining test. The results of showed that sensitivity approximately in the range of 4.23�10�4–8.53�10�4 mV/N and low ofcross sensitivity errors is below 4.05%. The results of dynamic analysis and testing show that the natural frequencies of the rotatingdynamometer in all the force orientations are approximately 1050, 2079 and 450.9 Hz. This means that its rigidity and dynamicrange are suitable for the machining process. The thermal effects analysis also confirmed that temperature fluctuations given asmall change in the static force offset, but not affect to the dynamic cutting force. Thus, the cutting tests indicated that the effects ofspindle rotation and cutting parameters on the component of cutting force are clearly visible in the signals. The developed rotatingdynamometer could be used to study the dynamics of the cutting process, optimizations, machine tool design and also toolcondition monitoring systems.

Acknowledgements

The authors would like to thank the Government of Malaysia (MOSTI) and Universiti Kebangsaan Malaysia for theirfinancial support under Grant 03-01-02-SF0843.

References

[1] F. Cus, M. Milfelner, J. Balic, An intelligent system for monitoring and optimization of ball-end milling process, J. Mater. Process. Technol. 175 (2006)90–97.

[2] J.A. Ghani, I.A. Choudhury, H.H. Masjuki, Performance of P10 TiN coated carbide tools when end milling AISI H13 tool steel at high cutting speed,J. Mater. Process. Technol. 153–154 (2004) 1062–1066.

[3] B. Huang, J.C. Chen, An in-process neural network-based surface roughness prediction (INN-SRP) system using a dynamometer in end millingoperations, Int. J. Adv. Manuf. Technol. 21 (2003) 339–347.

[4] Z. Uros, C. Franc, K. Edi, Adaptive network based inference system for estimation of flank wear in end-milling, J. Mater. Process. Technol. 209 (2009)1504–1511.

[5] R.P.H. Faassen, N. v.d. Wouw, J.A.J. Oosterling, H. Nijmeijer, Prediction of regenerative chatter by modelling and analysis of high-speed milling, Int.J. Mach. Tools Manuf. 43 (2003) 1437–1446.

[6] I. Korkut, A dynamometer design and its construction for milling operation, Mater. Des. 24 (2003) 631–637.[7] S. Yaldız, F. Unsacar, H. Saglam, H. Isık, Design, development and testing of a four-component milling dynamometer for the measurement of cutting

force and torque, Mech. Syst. Signal Process. 21 (2007) 1499–1511.[8] C. Scheer, P. Hoffmann, A. Kirchheim, G. Schaffner, Spindle-Integrated Force Sensors for Monitoring Drilling and Milling Processes, in: Proceedings of

the 9th International Conference for Sensors, Transducers & Systems, Nürnberg, Germany, 1999.[9] S.S. Park, High Frequency Bandwidth Cutting Force Measurements in Milling Using the Spindle Force Sensor System (Ph.D. thesis), University of British

Columbia, Vancouver, Canada, 2004.[10] G. Byrne, G.E. O'Donnell, An integrated force sensor solution for process monitoring of drilling operations, Ann. CIRP 56 (2007) 89–92.[11] G. Totis, G. Wirtz, M. Sortino, D. Veselovac, E. Kuljanic, F. Klocke, Development of a dynamometer for measuring individual cutting edge forces in face

milling, Mech. Syst. Signal Process. 24 (2010) 1844–1857.[12] G. Byrne, D. Dornfeld, I. Inasaki, G. Ketteler, W. Konig, R. Teti, Tool condition monitoring (TCM) – the status of research and industrial application, Ann.

CIRP 44 (1995) 541–567.[13] C. Adolfsson, J.-E. Stahl, Cutting force model for multi-toothed cutting processes and force measuring equipment for face milling, Int, J. Mach. Tools

Manuf. 35 (1995) 1715–1728.[14] D.A. Smith, S. Smith, J. Tlusty, High performance milling torque sensor, J. Manuf. Sci. Eng. 120 (1998) 504–514.[15] C.A. Suprock, J.S. Nichols, A low cost wireless high bandwidth transmitter for sensor-integrated metal cutting tools and process monitoring, Int.

J. Mech. Manuf. Syst. (2009) 441–454.[16] K.-H. Grote, E.K. Antonsson (Eds.), Springer Handbook of Mechanical Engineering, Springer, New York, 2009.[17] R.S. Figliola, D.E. Beasley, Theory and Design for Mechanical Measurements, Third ed. John Wiley & Sons, Inc., New York, 2000.[18] X. Liu, Machining dynamics in milling processes, in: K. Cheng (Ed.), Machining Dynamics: Fundamentals, Applications and Practices Springer, 2009,

pp. 167–231.[19] M.C. Shaw, Metal Cutting Principle, 2nd ed. Oxford University Press, New York, 2005.[20] N.B. McLaughlin, Y. Chen, Effect of strain gage misalignment on cross sensitivity of extended ring (ER) transducers, Can. Biosyst. Eng. 54 (2012)

223–231.[21] M.B. Jun, O.B. Ozdoganlar, R.E. DeVor, S.G. Kapoor, A. Kirchheim, G. Schaffner, Evaluation of a spindle-based force sensor for monitoring and fault

diagnosis of machining operations, Int. J. Mach. Tools Manuf. 42 (2002) 741–751.[22] D.J. Ewins, Modal Testing, Theory, Practice, and Application, Second ed. Research Studies Press, England, 2000.