Gazi Universitv Journal of ScienceGUJSci25(3):803-814 (2012)

ORIGINAL ARTICLE

Investigation of the effects of different chip breaker forms

on the cutting forces using artificial neural networks

Hüseyin GÜRBÜZ''*, Abdullah KURT^ Ulvi

'Hacettepe University, Faculty of Engineering, 06800 Beytepe, Ankara/TURKEY

^zi University, Technology Faculty, 06500 Teknikokullar, Ankara/TURKEY

Received: 06.01.2012 Resubmitted: 30.04.2012 Accepted: 03.05.2012

ABSTRACT

This paper presents a new approach based on artificial neural networks (ANNs) to detennine the effects ofdifferent chip breaker forms on cutting forces such as principal cutting force, feed force and passive force, in themachining of AISI 1050. The backpropagation learning algorithm and fenni transfer function were used in thenetwork. Tbe best fitting training data set was obtained witb nine neurons in the hidden layer, which made itpossible to predict cutting forces with an accuracy whicb is at least as good as tbat of the experimental error,over the whole experimental range. After training, it was found that tbe R" values are 0.9829, 0.9667 and 0.9492for Fc, Ff and Fp. respectively. The average error is %0.145. As seen from the results of mathematical modeling,the calculated cutting forces are obviously within acceptable uncertainties.

Keywords: Cutting forces. Chip breaker form. Artificial neural networks

1. INTRODUCTION

A chip breaker is the small step or groove formed bygrinding or the separate part connected to tool holders tobreak off the chip into short pieces. The obstructiondistance behind the cutting edge controls the chip form.

With broader applications of flexible manufacturingsystems, computer integrated manufacturing systems,etc., modem technologies using indexable tool insertswith new-style chip breakers are employed widely at thepresent time. The reliability of machining operations is anessential requirement for modem automatic

manufacturing systems. For tuming operations, in whichunbroken chips are the major obstacles for theautomation, reliability considers chip control as a majoraspect.The initial research on the mechanics of chip in metalcutting was carried out in the second half of the 1900's[1,2]. Different studies were perfomied by Karahasan [3]in order to evaluate the effect of chip breakingmechanism on cutting. Karahasan used chip breakerforms and manifested the geometrical properties of anoptimum chip breaker form to obtain an acceptable chip

Corresponding author:E-mail: huseyin.gurbuz@hacettepe.edu.tr

804 GUJSci. 25(3):803-814 (2012)/Huseyin GURBÜZ'', AbdullahKURf,

geometry. Mesquita and Barata [4] developed a methodfor predicting cutting forces. In their study, the modelconsiders the indentation or ploughing effect and thepresence of a parallel groove type chip breaker. Thetechnique is based on the measurement of the chipbreaker geometry and the calculation of the effective siderake angle. The indentation effect, the dynamic shearstress, and the cutting forces were determined fromexperiments. The proposed model was applied to themachining of martensitic stainless steels using coatedcarbide tools. The chip breaking performances of theasymmetric groove type (AGT) chip breaker and that ofthe symmetric groove type (SGT) chip breaker werecompared with each other in detail by Fang [5]. Theexperimental result shows that it is practical to substitutethe AGT chip breaker for the SGT breaker and toinvestigate the rules of chip breaking during machiningprocesses. The influence of the geometrical parameters ofthe chip groove on the chip breaking performance of thetool insert was investigated through numerous cuttingexperiments using new style chip breakers. Twomathematical models were suggested to predict the chipbreakability for the new style chip breakers usingmultiple linear regressions. The theoretical predictionswere in line with the experimental results under the givencutting conditions. Kim and Kweun [6] modeled the chipfiow formation process using different insert geometriesand investigated the important characteristic parametersin chip control. The study focuses on the chip breakerdesign and the experimental cutting of mild steel with thechip breaker. The characteristics of chip breakage formild steel with respect to the cutting speed, depth of cutand feed rate were analyzed using the experimentalresults. Das et al. [7] showed that the breaking strain inthe chip is the most important factor on which chipbreaking depends and a method was suggested fordetermining the chip breaker distance for any given feedrate and the chip breaker height for effective chipbreaking. The method also showed that the chip breakingcriterion is based neither on the specific cutting energynor on the material damage, which can be considered asthe adequate criterion for the chip breaking. Mahasharand Murugan [8] performed an experimental work thatdeal with the infiuence of two design parameters; thewidth of chip breaker and the angle of chip breaker of aclamped on chip breaker on effective chip breaking.Artificial neural networks (ANNs) have already beenapplied to various fields [9-13]. Cardi et al. [14] usedANNs for the detection of chatter vibration in turning;and also Jamali et al. [15] and Nalbant et al. [16] modeledthe explosive cutting process by using ANNs.Soleimanimehr et al. [17] used ANNs for the predictionof machining force and the surface roughness inultrasonic vibration-assisted turning and Szecsi [18] usedANNs to model the cutting force components. Kim et al.[19] evaluated the performance of commercial chipbreakers, using a neural network that was trained throughthe backpropagation algorithm. Important element forms(depth of cut, land, breadth, and radius) that infiuencedthe chip formation directly were chosen among thecommercial chip breakers and were employed as inputvalues of the neural network. As a result, Kim et al. [19]developed the performance evaluation method andapplied it to commercial tools, which resulted in asignificant performance. The chip breaking performanceof the tool insert is regarded as one of the most important

factors that maintain the continuity of the machiningprocess. Consequently, it is necessary to investigate therules of chip breaking systematically andcomprehensively when new style chip formers are used.Because of this reason, this subject is interesting formany researchers [20-24].

Cutting forces that occurs during the metal cutting isaffected directly from the cutting tool geometry (cuttingedge form, nose radius, rake angle, clearance angle andchip breaker form), the cutting tool performance and thecost per product. These forces cause wear, crack, fractureand various deformations on the cutting tool; and they arethe basic criteria for machinability. These formationscause the cutting tool function not to be performed. Inthis condition, the cutting tool is renewed and the costincreases. This is the essential step for the optimumanalysis of this process. The measurement of cuttingforces is a crucial step in metal cutting for the workbenchdesign, no vibration/rigid machine production and thepower required. This paper studies the effects of differentchip breaker forms and the cutting parameters such ascutting speed, feed rate and depth of cut on the cuttingforce components (the principal cutting force, feed forceand the passive/radial force) during the machining ofAISI 1050. The cutting forces are measuredexperimentally [25-26].

The progress in neurobiology allowed researchers tobuild mathematical models based on neurons to simulatethe neural behavior. ANN approach has been a wellknown type of evolutionary computation method in thelast decades. ANNs were adapted to a large number ofapplications in various fields [27-30]. In the field ofprocess engineering, ANNs is a good alternative to theconventional empirical modeling, based on polynomialand linear regressions.

This paper proposes a new approach, to determine theeffects of different chip breaker forms on the cuttingforces in machining of AISI 1050 with empiricalequations based on ANNs. The empirical equations areobtained by ANN using a limited amount experimentaldata, which is obtained fVom experimental studies. Theseequations have a high accuracy.

2. Material and Methods

2.L Experimental study

Cutting tools and chip breakers: The SNMG 120408Rinserts and the PSBNR 2525M12 tool holder that has 75°approaching angle according to the ISO 3685 were usedin the experiments. Five groups of chip breaker formssuch as STD, MS, GH, SA and MA coated inserts [31]were used in the experiments. Tliese inserts areMitsubishi grade UC6010 corresponding to the ISO gradeP15. Fig. 1 shows the cutting tools. Shape elements of thechip breaker tliat is sold by Mitsubishi are illustrated inTable 1 and Fig. 1. Although the chip breakability can bedetermined by diverse elements, in this study, the chipbreakability was evaluated with the dimension ofelements such as lengths and angles, which are the mostimportant elements tliat are affected during the chipbreaking.

GUJSci, 25(3):803-814 (2012)7Hüseyin GÜRBÜZ'', Abdullah KURf, Ulvi 805

Fig. 1. The cutting tools used for the tests and their chip breaker forms [31].

Table 1. Chip breaker could be determitied shape eletnetits.

Chip breaker Type

STDMSGHSAMA

Shape elements (Lengths ({) and Angels (a) for determined chip breaker

ai

0°15°

0°10°6°

Í1

0.250.500.32

0.300.2

15°25°

18°25°22°

Í2

1.050.95

2.480.570.9

15°25°

25°35°30°

Cutting parameters and the machine tool: Machiningtests were carried out by using five levels of cuttingspeeds such as 150, 200, 250, 300, 350 m/min, threelevels of feed rate such as 0.15, 0.25, 0.35 mm/rev and

Table 2. Test parameters.

two levels of depth of cut such as 1.6, 2.5 mm. Thecutting parameters that are used in the experiments areshown in Table 2. JOHNFORD T35 CNC Lathe was usedin the tests.

Cutting speed., V (m/min)

150,200,250,300,350

Feed rate, f (mm/rev)

0.15,0.25,0.35

Depth of cut, a (mm)

1.6,2.5

The Workpiece material and the cutting forcemeasurement: AISI 1050 (DfN 1.1210) workpiecematerial, which is the most widely used material in themanufacturing industry, was used in the tests. The

chemical composition of AISI 1050 is shown in Table 3.Cutting forces were measured using the KISTLER 9257Bdynamometer.

Table 3.

% C

0.430

The chemical

%Si

0.212

composition of AISI

%Mn

0.730

%P

0.0197

1050.

%S

0.0390

%Cr

0.0776

%Mo

0.00752 0

%Ni0972

%^

0.01 10

%Co

0.00603

%Cu

0.297

%Fe

98.06

806 GUJSci. 25(3):803-814 (2012)/Hüseyin GÜRBÜZ''. Abdullah KURl^. Uhi SEKER'

The Test method: 30 tests were carried out for each chipbreaker form. A total of 150 experiments were performedwith a dry condition. New (unused) inserts were used ineach test. The measurement of cutting forces was carriedout during the machining. The unvarying/stable cuttingforce values were determined after all the measurementswere performed and the cutting force values for each testwere decided.

2.2. Mathematical Modeling: Artiflcial NeuralNetwork

ANNs are used widely in many application areas such asmathematics, engineering, medicine, economics, etc.ANNs are trained to overcome the limitations of theconventional approaches to solve complex problems.ANN learns from given examples, by constructing aninput-output mapping in order to perform predictions[28]. ANNs are practical because of the followingreasons [28]:

1. ANNs provide a weighted connection and massivelyparallel processing with fault tolerance, so that they canautomatically learn from experience. This is calledinternal representation.

2. ANNs have the generalization capability to leamcomplex patterns of inputs and provide meaningfulsolutions to problems even when input data containerrors, data are incomplete, or input data are notpresented during training. In other words, they have theability to integrate information from multiple sources andincorporate new features without degrading priorlearning.

4. ANNs can determine how much weight each datasource should have in the classification, which remains asa problem in statistical methods. The non-linear leamingand smooth interpolation capabilities give the neuralnetwork an edge over standard computers and rule-basedsystems for solving various problems.

Most applications use a feed-forward neural net\vork withthe backpropagation training algorithm. The mainadvantage is that they possess an inherent generalizationability. This means that they can identify and respond topatterns which are similar but not identical to the oneswith which they have been trained. A leaming algorithmis defined as a procedure that consists of adjustingweights and biases in a network, to minimize an errorfunction between the network outputs, for a given set ofinputs, and the correct outputs. There are differentlearning algorithms. A popular algorithm is thebackpropagation algorithm, which has different versions.Backpropagation training algorithms, gradient descentand gradient descent with momentum are too slow forpractical problems, because they require small learningrates for a stable leaming.

The absolute fraction of variance (R )̂ defines theabsolute percentage of error, and it is given as follows:

R - = l - (1)

3. ANNs are distribtition free, because no priorknowledge is needed about the statistical distribution ofmethods that require modeling of the data. Neuralnetworks can avoid some of the shortcomings ofstatistical and empirical techniques.classes in datasources in order to apply the method for classification.This is an advantage over most statistical and empiricaltechniques.

The ANN structure is shown in Fig. 2. The Levenberg-Marquardt (LM) algorithm is used in the study. Inputsand outputs are normalized so that they fall in the range(0, 1). Neurons in the input layer do not have a transferfunction. Fermi transfer function is used. Products andbiases are simply summed, which is followed bytransforming through a transfer function to generate aresult, and as a result, the output is obtained.

Cut

ting

Spee

d (C

S)

Feed

rat

e(F

R)

Dep

th o

fcu

t (D

C)

a S S -

Figure 2. ANN structure.

GUJSci. 25(3):803-8¡4 (2012)/Hüseyin GURBÜZ'', AbdullahKURf. 807

Five different chip breaker forms are used in the inputlayer of the network. The cutting forces are in the outputlayer for all models. The hidden layers have nineneurons. The basic parameters, such as cutting speed,feed rate and depth of cut are considered as an input tothe ANN,

3. Results and Discussion3.L Experimental Results

The values of cutting force components (principal cuttingforce, feed force and the passive force; FQ, Ff, Fp,respectively), which are measured by the cutting testsaccording to different chip breaker forms and variationsin the cutting parameters (cutting speed, feed rate anddepth of cut) are shown in Fig. 3 to 5. The test results

- • -STD ^A^MS H

show that the highest cutting force values are measuredon the principal cutting force, feed force and the passiveforce, respectively. When the graphs in Fig, 3 to 5 areexamined, it is seen that the values of the cutting forcecomponents {FQ, Ff and Fp) increase with the increase ofthe depth of cut and the feed rate and decrease with theincrease of the cutting speed for all the chip breakertypes. This situation is in agreement with the otherresearch [31-34]. Decreasing cutting forces can beexplained by increasing energy spent as the cutting speedincreases. Almost all of the energy is transformed intotemperature. This temperature facilitates the chipformation during machining. According to the followingKienzle's equation; "F(=A.k¡' cutting forces increasedepending on the increase of chip cross-section (A),which is the product of feed rate and depth of cut [35].

GH •SA MA

2500-,

2000

1500

'̂ .

1000

0,15 150

a) a=l,6 mm b) a=2,5 mm

Fig, 3. Variation of primary forces (Fc) depending on chip breaker forms.

3 5 0

0,15 150

a) a=l,6 mm0,15 150

b) a=2,5 mm

Fig, 4. Variation of feed forces (Ff) depending on chip breaker forms.

808 GUJSci. 25(3):H03-H14 (2012)/Hüseyin GÜRBÜZ'', Abdullah KURf, Ulvi

STD M S - ^ G H - • - S A - U

400

350 350

Ü.I5 ISO

a) a=l,6 mm b) a=2,5 mmFig. 5. Variation of passive forces (Fp) depending on chip breaker forms.

Between cutting force components, the principal cuttingforce has the greatest impact known. From this point ifthe chip breaker forms are evaluated, the values ofhighest principal cutting forces on the chip breaker formsSA and GH, the lowest principal cutting forces values onthe chip breaker forms STD, MS and MA can beobserved. This situation occurred since the manufacturerrecommended the light cutting condition for the chipbreaker form SA, semi-heavy cutting condition for the

chip breaker form GH and medium cutting condition forchip breaker forms STD, MS and MA. In cuttingconditions in which the depths of cut is 2.5 mm and feedrate is 0.35 mm/rev, the highest cutting force componentvalues were measured with the depths of cut is 1.6 mmand the feed rate is 0.15 mm/rev, the lowest cutting forcecomponent values were measured for all chip breakerforms.

3.2. Results of the Mathematical Analysis

The new formulations, which are dependent on the maincutting parameters for the outputs, are given withEqs. 2-4 . The equations are used for the estimation of

the cutting forces, using different main cuttingparameters:

E =1

c . -4(-0.122732f+0.493827F2+0.521887F3-0.404305F4+0.333055^-0.453527^+0.940058^+0.743384^-0.5)1 ~r C

1

F =

-I -4( 0.35 I 140f+0.057244^2+0.314S57F,-0.129994^4+0.4177915-0.487852^-1.0339405+0.5145371-0.5)

1

1 + e-4(0.0786951i+0.421216Fj+0.602033',-0.258552F4+0.796264Ç-0.7961 55If,-2.0684091^+0.3260631-0.5)

(2)

(3)

(4)

where; F; (i=l,2,...,8) can be calculated by the fermifunction according to Eq.5. The formulation for theprediction of the cutting forces (Eqs. 2 to 4), which are

dependent on the cutting parameters, are given with Eq.6. In Eq. 5, Ej (i=I,2,...,9) is given with Eq. 6, which isbased on the main cutting parameters.

F =1

1 + e-4(E¡-0.5)(5)

Ei = Cu * ai + 1, + . + C4Í * I2 + + Csi * C S + C 7 Í * F R + C 8 Í * D C (6)

GUJSci, 25(3):803-814 (2012)/Huseyin GÜRBÜZ'', Abdullah KURf, 809

The constants (Cy) in Eq.6 are given in Table 4.

Table 4. Constants in Eq.6

CH

C2i

C3i

C4i

Csi

c«

Cvi

E,

-0.476620

0.929563

-1.027395

-7.131836

1.118563

-0.132383

1.288653

E2

-0.559270

0.716396

0.045802

-0.357313

-0.007316

-0.158886

1.393166

E3

-0.587898

-2.136200

-0.713339

0.615304

1.284926

-0.039975

-0.321403

E4

0.175976

1.483085

-1.468852

-0.165433

-0.141403

-1.271354

0.173557

E5

-1.210561

-0.880276

-0.715532

-8.204341

-0.157377

-0.119264

1.497301

E6

-0.796399

-2.931872

-0.231816

-1.548001

-2.147223

-0.389275

0.755749

E7

1.825169

-1.962083

0.972018

1.555212

-2.259936

0.062223

-0.484406

Eg

-2.724624

2.351739

-1.880695

-1.902957

2.506062

-0.927386

0.465857

E9

1.431483

-1.958817

0.048784

0.257463

0.190066

0.111294

-3.047957

To be used in training data. Figs. 6-8 present the ANNperformance of the determination of cutting forces foreach force, respectively. In general perspective,according to the results obtained, the deviation of cutting

forces between the measurement and the prediction of aANN is negligible in the range of ±0.4% (Figs. 9-13) fordifferent chip breaker forms.

Fig.6. Comparison of the j data and

1400 1

^ 1000

" 8003

« 600

g 400

w 200

0

R- = 0,9667

200 400 600 800 1000 1200

Fig.7. Comparison of the Ff.measured data and Ff.̂

^ 600 •

— 500 -"O

^ 400 •

« 300-

Z 200 -

^ 100 -

R- = 0.9492 ^ ^

0 100 200 300 400 500 600

Fig. 8. Comparison of the Fp.mcasiired data and

810 GUJ Sei. 25(3):S03S14 (2012)/Hüseyin GÜRBÜZ'', Abdullah KURf,

0.15 ISO V

Fig. 9. The performance of ANN (chip breaker forms STD).

-MS(a=l,6)-e~ MS(a=l,6)ANN " ^ MS(a=2,5) -A- MS(a=2,5)

Fig. 10. The performance of ANN (chip breaker forms MS).

GUJSci. 25(3):803-814 (2012)/Hüseyin GÜRBÜZ''. Abdullah KURf. 811

-GH(a=l,6)-€>- GH(a=l,6)ANN - ^ GH(a=2,5) -A- GH(a=2,5)

U.15 150 V '

Fig. 11. The performance of ANN (chip breaker forms GH).

-SA (a=l,6) -•©- SA (a=l,6)ANN " ^ SA (a=2,5) -A- SA (a=2,5)

0.15 150

Fig. 12. The performance of ANN (chip breaker forms SA).

812 GUJSci. 25(3);803-814 (2012)/Hüseyin GÜRBUZ''. Abdullah KURf,

MA(a=l,6)-O~MA(a=l,6)ANN -^ MA(a=2,5)

Fig. 13, The performance of ANN (chip breaker forms MA),

4. Conclusions

From the present work, the following conclusions can bedrawn:

> The test results show that the highest cutting forcesvalues are measured on the principal cutting force(Fc), the feed force (Ff) and the passive force (Fp),respectively,

> Generally, the cutting force component (Fc, Ff andFp) values increase when depth of cut and feed rateincreases and decrease when cutting speed increasesfor all the chip breaker types,

> The highest values of the principal cutting forces onthe chip breaker forms SA and GH, the lowestvalues of the principal cutting forces on the chipbreaker forms STD, MS and MA are observed,

> In cutting conditions in which the depth of cut is 2,5mm and feed rate is 0,35 mm/rev, the highest cuttingforce component values and in cutting conditions inwhich the depth of cut is 1,6 mm and feed rate is0,15 mm/rev, the lowest cutting force componentvalues were measured for all chip breaker forms.

> The results of the validation and the comparativestudy indicate that the ANN based estimationtechnique for the eutting force component isappropriate.

> Developed chip breaker forms tested with ANN wereapplied to a commercial cutting tool with differentchip breaker forms and in various eutting conditionsto predict the cutting force components.

Acknowledgements

The authors thank to Gazi University for providingfinancial support for the project (Project Code: 07/2005—01). Also, the author would like to thank Prof. Dr. AdnanSÖZEN for critically reviewing the original manuscriptand application of ANN,

References

[1] Cook, N,H., Jehaveri, P,, "The mechanism of chipcurl and its importance in metal cutting", Trans.,85(B): 374-380 (1963).

[2] Spaans, C, Geel, P,F,H,J,, "Breaking mechanisms incutting with a chip breaker", Ann. CIRP,\%: 87-92(1966).

[3] Karahasan, Z,0, The Infiuences of tool geometryand chip breaker form of cutting on toolperformance. MSc. Thesis (in Turkish), YildizTechnical University, Science and Technology,Istanbul, Turkey, pp. 153-161 (1995),

[4] Mesquita, R.M.D., Barata Marques. M,J,M,, "Effectof chip-breaker geometries on cutting forces". J.Mater. Pro. Tech., 31: 317-325 (1992),

[5] Fang, N,, "Influence of the geometrical parametersof the chip groove on chip breaking performanceusing new-sty le chip formers", J. Mater. Pro. Tech.,74: 268-275 (1998).

GUJSci, 25(3):803-814 (2012)/Hüseyin GÜRBÜZ'', Abdullah KURf, 813

[6] Kim, J.D., Kweun, O.B., "A chip-breaking systemfor mild steel in turning", Int J. Mach. Tools andManuf., 3,1: 601-611 (1991).

[7] Das. N.S.. Chawla B.S.. Biswas C.K., "An analysisof strain in chip breaking using slip-line field theorywith adhesion fi-iction at chip/tool interface".Journal of Materials Processing Technology, 170:509-515(2005).

[8] Mahashar, A. J. Murugan M., "Influence of chipbreaker location and angle on chip form in tuminglow carbon steel". International Journal ofmachining and machinability of materials A., 5(4):452-475 (2009).

[9] Zuperl. U., Cus. F.. Mursec. B.. and Ploj. T.. "Ahybrid analytical-neural network approach to thedetermination of optimal cutting conditions".Journal of Materials Processing Technology, 157-158:82-90(2004).

[10] Ezugwua, E. O., Fadarea, D. A., Bonney, J., DaSilva, R. B. and Sales, W. F., "Modelling thecorrelation between cutting and process parametersin high-speed machining of lnconel 718 alloy usingan artificial neural network". International Journalof Machine Tools & Manufacture,-,A5\ 1375-1385(2005).

[11] Karayel, D., "Prediction and control of surfaceroughness in CNC lathe using artificial neuralnetwork". Journal of Materials ProcessingTechnology, 209: 3125-3137 (2009).

[12] Paulo Davim, J., Gaitondeb, V.N., and Kamik, S.R.,"Investigations into the effect of cutting conditionson surface roughness in tuming of free machiningsteel hy ANN models". Journal of MaterialsProcessing Technology, 2 0 5: 16-23 (2008).

[13] Kurt, A., "Modelling of the cutting tool stresses inmachining of lnconel 718 using artificial neuralnetworks". Expert Systems with Applications, 36:9645-9657 (2009).

[14] Adam, A. Cardi, Hiram A. Firpi, Matthew T.Bement, and Steven Y. L., "Workpiece dynamicanalysis and prediction during chatter of tumingprocess". Mechanical Systems and SignalProcessing, 22: 1481-1494(2008).

[15] Jamali, A., Nariman-zadeh, N., Darvizeh, A.,Masoumi, A., and Hamrang, S., "Multi-objectiveevolutionary optimization of polynomial neuralnetworks for modelling and prediction of explosivecutting process". Engineering Applications ofArtificial Intelligence, 22: 676-687 (2009).

[16] Nalbant. M.. Gökkaya, H., Tokta?, L. Sur, G., "Theexperimental investigation of the effects ofuncoated, PVD-and CVD-coated cemented carbideinserts and cutting parameters on surface roughnessin CNC tuming and its prediction using artificialneural Networks", Robotics and Computer-Integrated Manufacturing, 25: 211-223 (2009).

[17] Soleimanimehr, H., Nategh, M. J. and Amini, S.,"Prediction of machining force and surfaceroughness in ultrasonic vibration-assisted tumingusing neural networks". Advanced MaterialsResearch, 83-86: 326-.334 (2010).

[18] Szecsi, T., "Cutting force modeling using artificialneural networks". Journal of Materials ProcessingTechnology, 92-93: 344-349 (1999).

[19] Kim H.G., Sim J.H., Kweon H.J., "Perfonnanceevaluation of chip breaker utilizing neural network".Journal of Materials Processing Technology A.209(2): 647-656 (2009).

[20] Jiang, C.Y., Zhang, Y.Z. and Chi, Z.J.,"Experimental research of the chip flow directionand its application to the chip control". Ann. CIRP,33:81-84(1984).

[21] Komanduri, R. and Brown. R.H.. "On the mechanicsof chip segmentation in machining". Transactionsof American Society of Mechanical EngineersJournal of Engineering for Industry, 103: 33-51(1981).

[22] Jawahir, l.S. and Luttervelt, C.A.V. "RecentDevelopments in Chip Control Research andapplications". Annals of the CIRP, 42(2): 659-693(1993).

[23] Jawahir, I.S., Qureshi, N. and Arsecularatne, J.A.,"On the interrelationships of some machinabilityparameters in finish tuming with cermet chipforming tool inserts", Int J. Mach. Tools Manuf.,32: 709-723 (1992).

[24] Fang, X.D., Fei, J. and Jawahir, I.S., "A HybridAlgorithm for Predicting Chip Form/ChipBreakability in Machining", Int. J. Mach. ToolsManuf, 36(10): 1093-1107 (1996).

[25] Gürbüz, IL, The Effect on The Cutting Tool Stressesof The Different Chip Breaker Form in Tuming,MSc. Thesis (in Turkish), Gazi University, Instituteof Science and Technology, Ankara, Turkey, (2006).

[26] Gürbüz, H., Kurt, A., Korkut, L, §eker, U., "Theexperimental investigation of the effects of differentchip breaker forms on the cutting forces". AdvancedMaterials Research. 23: 191-194 (2007).

[27] Kalogirou. S.A., "Artificial intelligence for themodeling and control of combustion processes: areviev\". Progress in Energy and CombustionScience, 29: 515-566 (2003).

[28] Sozen, A., "Future projection of the energydependency of Turkey using artificial neuralnetwork". Energy Policy, 37(11): 4827-4833 (2009).

[29] Karata?, C, Sozen, A., Dulek, E., "Modelling ofresidual stresses in the shot peened material C-l020by artificial neural network". Expert Systems withApplications, 36(2): 3514-3521 (2009).

[30] Sozen. A.. Arcaklioglu, E.. Menlik, T., Ozalp. M.."Determination of thermodynamic properties of analtemative refrigerant (R407c) using artificial neuralnetwork". Expert Systems with Applications, 36(3):4346-4356 (2009).

[31] Mitsubishi Carbide, Tuming Tools, Rotating Tools,Tooling Solutions, General Catalogue, (2005).

[32] Chen, W., "Cutting Forces and Surface Finish WhenMachining Medium Hardness Steel Using CBNTools". International Journal of Machine Tools &Manufacture, 40: 455^66 (2000).

814 GUJSci, 25(3):803-814 (2012)/Huseyin GÜRBUZ'-', Abdullah KURf,

[33] Trent, E.M., Metal Cutting, Butterworths Press, 1- [35] Seker, U., Lecture notes in Tool design (in Turkish),171, London, (1989). Gazi University, Technical Education Faculty,

Ankara, Turkey, pp. 33-44, pp.47-72 (1997).[34] Zhao. J.. Ai X. and Li, Z..-'Finite element analysis ^^^ ^^

of cutting forces in high speed machining".Materials Science Forum, 532-533: 753-756(2006).

Copyright of Gazi University Journal of Science is the property of Gazi University Journal of Science and its

content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's

express written permission. However, users may print, download, or email articles for individual use.