Prediction of Optimal Cutting Parameters for High Speed Dry Turning of Inconel 718 Using Gonns

7/30/2019 Prediction of Optimal Cutting Parameters for High Speed Dry Turning of Inconel 718 Using Gonns

1/12

International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976

6340(Print), ISSN 0976 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) IAEME

294

PREDICTION OF OPTIMAL CUTTING PARAMETERS FOR

HIGH SPEED DRY TURNING OF INCONEL 718

USING GONNS

1

Satyanarayana.B,2

Ranga Janardhana.G,3

Hanumantha Rao.D, and4Kalyan.R.R.

1,4Department of Mechanical Engineering, VNR Vignana Jyothi Institute of Engineering &

Technology,Hyderabad, India,

E-mail: [email protected] Principal, College of Engineering, J N T U, Vizayanagaram, India,

E-mail: [email protected]

Principal, Matrusri Engineering College, Hyderabad, India,E-mail: [email protected]

ABSTRACT

Inconel 718, a Fe-Ni based super alloy widely used in the aircraft industry for its capacityto keep mechanical properties at high temperatures. Characteristics like lower thermalconductivity, work hardening make Inconel 718 a difficult-to-cut material. TheGenetically Optimized Neural Network System (GONNS) is proposed for the predictionof optimal cutting conditions, in high speed dry turning of Inconel 718 using three typesof tungsten carbide tools, to obtain the best results than other mathematical models.GONNS uses Back propagation (BP) type Neural Networks (NN), which is GeneticallyOptimized, to establish the relationship between input and output parameters. Optimalcutting conditions are then obtained from the generated NN model. GONNS was used in

two case studies for prediction of optimal cutting parameters i) To keep the cutting forcesminimum ii) To obtain the best possible surface roughness. Sensitivity test was alsoconducted to study the influence of cutting parameters.

Keywords: Inconel 718, High speed turning, Neural Networks, Genetic Algorithm

INTERNATIONAL JOURNAL OF MECHANICAL ENGINEERINGAND TECHNOLOGY (IJMET)

ISSN 0976 6340 (Print)

ISSN 0976 6359 (Online)

Volume 3, Issue 3, September - December (2012), pp. 294-305

IAEME: www.iaeme.com/ijmet.asp

Journal Impact Factor (2012): 3.8071 (Calculated by GISI)www.jifactor.com

IJMET

I A E M E


2/12


3/12

International Journal of Mech

6340(Print), ISSN 0976 6359(O

Figure 1 CoNetw

One of the most promisimerging of EA and ANN wilsignificantly better intelligent sbased neural network (GA-NN)will acquire the optimal weighminimal errors

2.2 Steps in hybrid model (GA

The NN to be geneticallSquared Error of weights generinvolved in the process areas fol1.Collection of experimentalInconel 718.

2.Providing these input and outfashion.

3.Tag the data in software as iassist the software while develop

4.Create a custom network of re5.Further, train and test the netother operations will be done. T

6.Selection of optimal parametein GA using MATLAB to find o

7.Sensitivity Analysis: This isparameters on outputs.

3. EXPERIMENTAL WORK

3.1 Machining parameters

From survey, the tool is

38% of the tools are used up tospeeds are far below the optim

roughness and cutting force incut. These cutting conditions

important parameters, that effec

cut, and spindle speed. Hence texperiments were carried out to

on Inconel 718 from hand book

anical Engineering and Technology (IJMET

line) Volume 3, Issue 3, Sep- Dec (2012) IAE

296

mbination of Genetic Algorithm and Neuralork (in part adapted) from [10, p. 29]

g techniques is the adaptation of network trainigain adaptability to dynamic environment

stems than relying on ANN, PSO, or GA. In tmodel (Fig.1) was framed to optimize the turnis using GA which leads to a high performan

NN) developed for optimization

optimized during training and testing, whichtion is optimized resulting in less error percelows.ata set of inputs and outputs (in excel shee

ut pool to software tool (Nuero Solutions 5.1)

nputs, outputs, training, testing and productioing model in step 5)

quired structure and save it.

ork. The data tagged as train will be used for tis will generate the required model.

rs: After developing NN system, end relationt the best optimal parameters for machining.

analogues to ANOVA which shows the infl

used at the rated cutting speed only 58% of th

their full tool life capability [12] showing thatal economic speeds. Several factors influence

urning operation such as spindle speed, feed ran be setup in advance to have best result

t surface roughness and cutting force are cutti

hese parameters are chosen as cutting paramefix the limits of these parameters based on th

and literature.

), ISSN 0976

E

ng with EA. Thend will lead to

his paper, a GA-g process. ANN

ce network with

eans that Meantages. The steps

s) in machining

[11] in a jumbled

. (Tagging is to

raining; similarly

an be optimized

uence of cutting

e time, and only

selected cuttinghe final surface

te, and depth of[12].The most

g feed, depth of

ers. Preliminarymachining data


4/12


6340(Print), ISSN 0976 6359(O

Table 1 Parameters and their lim

Tool Material

Uncoated WC

TiN/TiAlN PVD coatedTiN/Al2O3/Ti(CN) CVD coated

3.2 Work piece and Tools

The billets used were I

0.66 %, Ti = 0.96% balance

strength is 1100 Mpa. Conside

grade, CVD coated tool of Tselected. The tool signature of t

Table 2 Tool Nomenclature

Rakeangle()o

Clearance

angle()

-

3.3 Machining

Turning operations wereSpeed range of 45-2000 RPM

dynamometer and a high qualiTurning trials were carried outL27 Orthogonal Array [13].

Figure 2 Inconel 718 billet Fig

4. MACHINING PERFORM

It has been recognized toptimization strategies for selinformation dealing with test me



297

its.

Cutting

Speed(m/min)

Feed (mm/rev) DOC

(mm)

50,60,70 0.103,0.137,0.164 0.5,0.75,0.150,60,70 0.103,0.137,0.164 0.5,0.75,0.150,60,70 0.103,0.137,0.164 0.5,0.75,0.1

conel 718 rods (Ni = 54.48 %, Cr = 17.5%,

re Fe and other) with 30 mm diameter who

ring economical advantages, Un-Coated carbi

4000 grade and PVD coated tool of TS 2e same is presented in Table 2.

Inclination

angle ()

Approach

angle ()

Included

angle()

NoseRadius(r)mm

- 75 .

carried out on a conventional lathe of bed wid. This lathe is provided with three compon

ty feed mechanism which maintains the setn super alloy Inconel 718 material in dry con

re 3 Experimental Set-up

NCE MEASURE

at the reliable quantitative predictions are esscting cutting conditions. Unfortunately the

thodology and data evaluation in metal cutting

), ISSN 0976

E

b = 4.9%, Al =

e Tensile Yield

e tool of H13A

000 grade were

h 242 mm and ant strain gauge

feed accurately.itions following

ential to develope is a lack ofxperiments [14].


5/12


6340(Print), ISSN 0976 6359(O

Figure 4 Surface Roughness

Tester

Roughness plays an importenvironment. Irregularities in thTherefore appropriate cutting cHence, surface roughness is chSurface Roughness tester (Fig.4)

Knowledge of the cutting fotool and Calculation of the machconditions to control cutting forcis to change the cutting conditi

dynamometer shown in the figur

5. EXPERIMENTAL RESULTS

5.1 Surface Roughness results

Table 3 Experimental layout usiuncoated tool

Test No. Speed Feed DO [m/min] [mm/rev] [mm

1 50 0.103 0.52 50 0.103 0.753 50 0.103 14 50 0.137 0.55 50 0.137 0.756 50 0.137 17 50 0.164 0.58 50 0.164 0.759 50 0.164 110 60 0.103 0.511 60 0.103 0.7512 60 0.103 113 60 0.137 0.514 60 0.137 0.7515 60 0.137 116 60 0.164 0.517 60 0.164 0.7518 60 0.164 119 70 0.103 0.520 70 0.103 0.7521 70 0.103 122 70 0.137 0.523 70 0.137 0.7524 70 0.137 125 70 0.164 0.526 70 0.164 0.7527 70 0.164 1



298

Figure 5 Three component

strain gauge dynamometer

nt role in determining how a real object willsurface may form nucleation sites for cracks o

onditions are to be established to reach thesen in our study and was measured by Mitut

.ces is essential for Proper design of the cuttingine tool power. Thus need for a good predictivee has been identified. The simplest way to contns. The main cutting force has been noted u

e 5.

g standard L27 orthogonal array with experime

SR] [m]

0.7340.7070.6790.9290.90160.8741.0831.0561.0280.8300.8020.7751.0250.9970.9691.1791.1521.1240.9260.8980.8711.1201.0931.0651.2751.2471.220

), ISSN 0976

E

interact with itsr corrosion. [15].required quality.oyo make SJ201

tools and cuttingmodel of cuttingrol cutting forcesing strain gauge

ntal results for


6/12



299

Table 4 Experimental layout using standard L27 orthogonal array with experimental results forcoated tool (PVD)

Test No. Speed Feed DOC SR[m/min] [mm/rev] [mm] [m]

1 50 0.103 0.5 0.6412 50 0.103 0.75 0.6143 50 0.103 1 0.5864 50 0.137 0.5 0.8365 50 0.137 0.75 0.8086 50 0.137 1 0.7817 50 0.164 0.5 0.9908 50 0.164 0.75 0.9639 50 0.164 1 0.935

10 60 0.103 0.5 0.73711 60 0.103 0.75 0.70912 60 0.103 1 0.68213 60 0.137 0.5 0.93214 60 0.137 0.75 0.90415 60 0.137 1 0.87616 60 0.164 0.5 1.08617 60 0.164 0.75 1.05918 60 0.164 1 1.03119 70 0.103 0.5 0.83320 70 0.103 0.75 0.80521 70 0.103 1 0.77822 70 0.137 0.5 1.02723 70 0.137 0.75 1.00124 70 0.137 1 0.97225 70 0.164 0.5 1.18226 70 0.164 0.75 1.15427 70 0.164 1 1.127

Table 5 Experimental layout using standard L27 orthogonal array with experimental results forcoated tool (CVD)

Test No. Speed Feed DOC SR[m/min] [mm/rev] [mm] [m]

1 50 0.103 0.5 0.7082 50 0.103 0.75 0.6813 50 0.103 1 0.6534 50 0.137 0.5 0.9035 50 0.137 0.75 0.8756 50 0.137 1 0.8487 50 0.164 0.5 1.0578 50 0.164 0.75 1.0309 50 0.164 1 1.002

10 60 0.103 0.5 0.80411 60 0.103 0.75 0.77612 60 0.103 1 0.74913 60 0.137 0.5 0.99914 60 0.137 0.75 0.97115 60 0.137 1 0.94316

60

0.164

0.5

1.153

17 60 0.164 0.75 1.12618 60 0.164 1 1.09819 70 0.103 0.5 0.90020 70 0.103 0.75 0.87221 70 0.103 1 0.84522 70 0.137 0.5 1.09423 70 0.137 0.75 1.06724 70 0.137 1 1.0397425 70 0.164 0.5 1.2426 70 0.164 0.75 1.2227 70 0.164 1 1.194


7/12



300

5.2 Cutting Force resultsTable 6 Experimental layout using standard L27 orthogonal array with experimental results foruncoated tool

Test No. Speed Feed DOC CF[m/min] [mm/rev] [mm] [N]

1 50 0.103 0.5 742 50 0.103 0.75 3483 50 0.103 1 6224 50 0.137 0.5 2355 50 0.137 0.75 5096 50 0.137 1 7837 50 0.164 0.5 3648 50 0.164 0.75 6379 50 0.164 1 911

10 60 0.103 0.5 6011 60 0.103 0.75 21312 60 0.103 1 48713 60 0.137 0.5 10014 60 0.137 0.75 37415 60 0.137 1 64816 60 0.164 0.5 22917 60 0.164 0.75 50218 60 0.164 1 77619 70 0.103 0.5 19520 70 0.103 0.75 7821 70 0.103 1 35222 70 0.137 0.5 3423 70 0.137 0.75 23924 70 0.137 1 51325 70 0.164 0.5 9426 70 0.164 0.75 36727 70 0.164 1 641

Table 7 Experimental layout using standard L27 orthogonal array with experimental results forcoated tool (PVD)


1 50 0.103 0.5 612 50 0.103 0.75 3353 50 0.103 1 604 50 0.137 0.5 2225 50 0.137 0.75 4966 50 0.137 1 7707 50 0.164 0.5 3508 50 0.164 0.75 6249 50 0.164 1 898

10 60 0.103 0.5 7311 60 0.103 0.75 20012 60 0.103 1 47413 60 0.137 0.5 8714 60 0.137 0.75 36115 60 0.137 1 63516 60 0.164 0.5 21517 60 0.164 0.75 489

18 60 0.164 1 76319 70 0.103 0.5 20820 70 0.103 0.75 6521 70 0.103 1 33922 70 0.137 0.5 4723 70 0.137 0.75 22624 70 0.137 1 50025 70 0.164 0.5 8026 70 0.164 0.75 35427 70 0.164 1 628


8/12



301

Table 8 Experimental layout using standard L27 orthogonal array with experimental results for coated tool (CVD)


1 50 0.103 0.5 1022 50 0.103 0.75 376

3 50 0.103 1 6504 50 0.137 0.5 2635 50 0.137 0.75 5376 50 0.137 1 8117 50 0.164 0.5 3918 50 0.164 0.75 6659 50 0.164 1 939

10 60 0.103 0.5 3211 60 0.103 0.75 24112 60 0.103 1 51513 60 0.137 0.5 12814 60 0.137 0.75 40215 60 0.137 1 67616 60 0.164 0.5 25617 60 0.164 0.75 53018 60 0.164 1 80419 70 0.103 0.5 16720 70 0.103 0.75 10621 70 0.103 1 38022 70 0.137 0.5 623 70 0.137 0.75 26724 70 0.137 1 54125 70 0.164 0.5 12126 70 0.164 0.75 39527 70 0.164 1 663

6. SIMULATION RESULTS

The effects of cutting parameters on the cutting force and surface roughness wereevaluated by applying SENSITIVITY analysis in Nuero solutions. A single-factor experimentwas conducted to explore the influence of feed on Surface Roughness and cutting speed on

cutting force.

Fig 6 Fig 7 Fig 8

Fig 9 Fig 10 Fig 11

Figures presenting Sensitiveness results for Surface Roughness (6, 7, 8) and cutting force (9, 10,11), showing the parameter with high influence on output.


9/12



302

Fig 12 Fig 13 Fig 14

Fig 15 Fig 16 Fig 17

Figure 12, 13, 14, 15, 16, 17 are graphs of output with respect to high influence parameter(Speed for Surface Roughness and DOC in case of Cutting Force).

Table 9 Percentage of influence of Each Parameter on Output

Tool Parameter Cutting force Surface

Roughness

Speed 16.6 11.5

Uncoated tool Feed 28.2 63.6

DOC 55.2 24.9

Speed 17.9 7.6

CVD tool Feed 30.9 65.2

DOC 51.2 27.2

Speed 17.5 10.2

PVD tool Feed 28.5 66.2

DOC 54 23.6

6.1 Conclusions on Effect of Parameters

Feed has more effect on surface roughness than depth of cut and speed. During turningsometimes a higher value of roughness due to the presence of hard carbide particles presentin the matrix has been identified.

DOC has more effect on Cutting Force than feed and speed .Increasing the cutting speed toobtain smaller values of cutting forces is the most frequent method used in the literature.Also, higher feed rates were observed to be causing higher cutting forces.


10/12



303

A direct relation was determined between the tested feed rates and average surface roughnessand also between DOC and Cutting force.

Surface roughness increases with an increase in the feed rate and depth of cut and a decreasein cutting speed.

7 .OPTIMAL PARAMETERS

In order to obtain the optimal cutting conditions, the developed relation (mathematicalform) has been feed to GA in MATLAB optimal Tool. Final Point at the left most corner ofoptim toll is the required optimum point.

Table 10 Optimal Cutting Conditions for Lower SR

Surface Roughness

SPEED FEED DOC

Uncoated 68.379 0.103 0.5

CVD 68.562 0.103 0.5

PVD 68.661 0.1031 0.502

Table 11 Optimal Cutting Conditions for minimal CF

Cutting Forces

SPEED FEED DOCUncoated 69.215 0.103 0.5

CVD 69.99 0.1042 0.5004

PVD 69.998 0.1033 0.5001

From the Tables 10 and 11 the Surface roughness and Cutting force has been found to beminimum at high speed, low feed and depth of cut, which agreements with metal cutting theory

8. CONFIRMATION TESTS

Confirmation experiment was conducted using optimal conditions and the results are

presented in table 12.


11/12



304

Table 12 Machining Results at Optimum conditions

Tool Value Cutting force Surface

Roughness

Exp. Value 154 0.823

Uncoated tool NN Value 157.512 0.8362

% Error 2.3 1.7

Exp. Value 177 0.76

CVD tool NN Value 180.954 0.779

% Error 2.2 2.6

Exp. Value 142 0.786

PVD tool NN Value 144.55 0.802

% Error 1.8 2.1

9. CONCLUSIONS

In this study, experimentation was done on Super alloy Inconel 718 to aquire training

data NNs. Results from NN analysis are like this The optimum set of control factors indicates that at high cutting speed, and at low feed and

depth of cut the cutting force & Surface Roughness is optimum in the selected range.

For any required surface roughness, the developed model can predict the optimal values ofspeed, feed and depth of cut in any selected range.

From the experimental results, tool with PVD coating have shown good performance thantool with CVD coating. Hence, PVD coated tools may be preferable for high speed dryturning of Inconel 718.

REFERENCES

1. E.O. Ezugwu, Advances in the machining of nickel and titanium base super alloys, Keynotepaper presented at the Japan Society for Precision Engineering Conference 2004, pp. 140.2. Sims C. T, Stoloff N. S and Hagel W. C. (1987) Superalloys II- High temperatureMaterials for Aerospace and Industrial power New York: Willey.3. Choudary I. A and M. A. EI-Baradie (1997) Machining nickel base super alloys: Inconel718 Proceedings of Institution of Mechanical Engineers, part B 212, pp 195-206.4. R.S. Pawade, S.S. Joshi, P.K. Brahmankar, Rahman, Some investigations of high-speed

turned Inconel 718, in: V.S. Raja, Kuppan (Eds.), Proc.eedings of the of the 21st AIMTDR


12/12



305

Conference, VIT, Vellore, 2004.

5. D. Dudzinski a, A. Devillez a, A. Moufki a, Larrouque`re b, V. Zerrouki b, J. Vigneau,

ISGMP, University.6. Zuperl U, Cus F, Mursec B, Ploj T (2004), A hybrid analytical neural network approach to thedetermination of optimal cutting conditions. J Mater Process Technol 157 158:82907. Yang WH, Tarng YS (1998) Design optimization of cutting parameters for turning operationsbased on the Taguchi method. J Mater Process Technol 84:12212998. E.O. Ezugwu, K.A. Olajire, J. Bonney, Modelling of tool wear based on component forces,Tribol. Lett. 11 (1) (2001).9. E.O. Ezugwu, S.J. Arthur, E.L. Hines, Tool-wear prediction using artificial neural networks, J.Mater. Process. Technol. 49 (1995) 255264.10. Ribeiro Filho, Jose L.; Treleaven, Philip C.; Alippi, Cesare: Genetic Algorithm ProgrammingEnvironments. III: Computer (1994) June, p. 28 - 43.11. http://www.neurosolutions.com/12 Valery Marinov, Manufacturing Technology, Kendall/Hunt Pub. Co, 2

ndedition.

13. Phillip J. Ross, Taguchi Techniques for Quality Engineering, McGraw-Hill, 1996.

14. D. G.Thakur, B. Ramamoorthy, L. Vijayaraghavan, Effect of High Speed CuttingParametersWC.15. R. A. Smith, 2007 Jordan Journal of Mechanical and Industrial Engineering. - Volume 1,Number 1 (ISSN 1995-6665)

Prediction of Optimal Cutting Parameters for High Speed Dry Turning of Inconel 718 Using Gonns

Documents

Transcript of Prediction of Optimal Cutting Parameters for High Speed Dry Turning of Inconel 718 Using Gonns