Investigation of machining rate and roughness for rotary ...pdfs.semanticscholar.org/76ca/dbd77c0960c4e26e29b4a1050c9681f0c411.pdfInvestigation of machining rate and roughness for

68 Int. J. Manufacturing Research, Vol. 13, No. 1, 2018

Copyright © 2018 Inderscience Enterprises Ltd.

Investigation of machining rate and roughness for rotary ultrasonic drilling of Inconel 718 alloy with slotted diamond metal bonded tool

Dipesh Popli* and Meenu Gupta Department of Mechanical Engineering, National Institute of Technology Kurukshetra, Kurukshetra-136119, Haryana, India Email: [email protected] *Corresponding author

Abstract: Super alloy has extensive engineering applications in nuclear industries owing to its outstanding performance characteristics at elevated temperature. More than 30% of consumption in nuclear industries is super alloy. A number of holes are required to be drilled into super alloys for their final stage assembly. During conventional machining process, excessive heat is generated, which is the cause of heat treated zone that further affect the life and performance of the product. To overcome this difficulty, a rotary ultrasonic machining method came into existence to machine super alloys by hybrid cutting action (vibratory and rotary) of diamond impregnated core drill. This study focuses on the machining characteristics of super alloy by RUM process. The empirical modelling of process parameters of RUM is carried out for super alloy (Inconel 718) using an experimental design approach called response surface methodology. Material removal efficiency of RUM and surface topography of the machined material are studied. The results reported that for quality and productivity aspect, the feed rate in RUM is found a most critical factor.

[Received 5 April 2017; Revised 10 June 2017; Accepted 26 June 2017]

Keywords: rotary ultrasonic machining; RUM; ultrasonic; machining; roughness; analysis of variance; ANOVA.

Reference to this paper should be made as follows: Popli, D. and Gupta, M. (2018) ‘Investigation of machining rate and roughness for rotary ultrasonic drilling of Inconel 718 alloy with slotted diamond metal bonded tool’, Int. J. Manufacturing Research, Vol. 13, No. 1, pp.68–95.

Biographical notes: Dipesh Popli is a PhD Scholar in the Department of Mechanical Engineering at National Institute of Technology (NIT), Kurukshetra, Haryana, India and working in the area of production engineering. He obtained his BTech in Mechanical Engineering in 2006, MTech in the Mechanical Engineering in 2010 from the National Institute of Technology, Kurukshetra, India. His research interest includes machining, optimisation, etc. He has near about seven years of experience in industry, teaching and research.

Meenu Gupta is presently working as Associate Professor in the Department of Mechanical Engineering, National Institute of Technology (NIT), Kurukshetra, Haryana, India. NIT is an institution of national importance. She has more than 25 years of experience in teaching and research. Her current areas of research include machine vision, image processing, optimisation, and modelling.

Investigation of machining rate and roughness for rotary ultrasonic drilling 69

1 Introduction

The super alloys are broadly used in modern industries due to excellent corrosion resistance, chemical stability, and temperature strength (A.S.M. International and A. Rights, 2000). The uses of these alloys can be widely seen in the nuclear and aerospace industries. Beside it, the application of such super alloys has been extended into various engineering applications like manufacturing of moulds for pressure die casting, punching die, nuclear furnaces, gas processing and chemical processing. The processing of super alloys requires high cost that directly affects the characteristics and performance of the material at the end user. In composition, Nickel-base alloys usually contain more than 50% nickel, whereas, in the nickel-iron-base alloy, nickel is the main (less than 50%) component of the alloy. In addition, some minor elements (i.e., silicon, phosphorus, sulphur, oxygen and nitrogen) are accommodated to small level in the formation of the product.

Many wrought nickel-base super alloys contain 10%–20% Cr, near about 8% Al and Ti, 5%–15% Co, and a little amount of boron, zirconium, magnesium, carbon and other additives are added to improve surface stability (Nalbant et al., 2007). Amongst the commercially utilised super alloys, nickel-based super alloys are considered as the most leading alloys in manufacturing industries, accounting for around 45% of wrought nickel-based alloy products and 25% of cast nickel-based products.

1.1 Machining of nickel-based super alloys

Ni-based super alloys are sophisticated structured materials that make it responsible for its poor machinability. It consists of an austenitic matrix like stainless steels and it work hardens quickly due to excessive heat generation during machining. Moreover, localisation of shear in the chip produces an abrasive saw-toothed edge which makes swarf handling difficult. These alloys likewise have the probability to weld with the tool materials at the high temperature generated during machining. The tendency of built up edge during machining and the occurrence of hard abrasive carbides in their microstructure also deters machinability of such materials. These characteristics are generally responsible for the development of high stresses and high temperature (more than 1,000ºC) generation in the cutting zone, resulting in accelerated flank wear, cratering and notching (Khidhir and Mohamed, 2010). The nickel-base super alloys harden by the precipitation of a γ′ phase of the type Ni3 (Ti, Al). Increase in the quantity of γ′ phase that is obtained by raising the amount of titanium and aluminium escalates the rate of tool wear (Dreshfiezd, 1970). Amongst all the difficulties, the cubic boron nitride (CBN) is one of the solutions for the processing of super alloys (Zhou et al., 2012). It is one of the hardest materials available after diamond but does not occur in nature. The synthesis of polycrystalline CBN is composed of more than 50% CBN combined with ceramic binders such as titanium nitride and titanium carbide. High amount of CBN content is desirable for cutting of such kind of tough materials like super alloys. Higher CBN content generally offers enhancement in the chipping resistance. It has been reported that while machining of Inconel 718, the performance of the tools with high CBN content is better because of their high hardness (Bhatt et al., 2010). The CBN enabled tools are used to machine nickel or cobalt-based alloys containing hardness equal to or greater than 340 HV.

70 D. Popli and M. Gupta

A conventional diamond cutting method aided with ultrasonic vibration has come into existence in the form of hybrid machining method called rotary ultrasonic machining (RUM) in which a high frequency oscillating and rotary metal bonded diamond core drill tool enables it to machine hard and complex materials (i.e., ceramic, super alloys, titanium, etc.). In addition, no heat treated/affected zone is created after machining with RUM process (Singh and Singhal, 2016a; Abdo et al., 2013; Feng and Xu, 2012). The effective trial of RUM process as an ultrasonic milling was carried out by Pei et al. in 1995. The application of this hybrid method was extended to the drilling, grinding and face milling of ceramics with preliminary experiments. The ranges of the parameters employed during the experiment were: 1,000 and 3,000 rpm for tool rotation, 0.023 and 0.033 mm for the amplitude of tool, 0.1, 0.35 and 1 mm for cutting depth and 0.381, 0.762, 1.524, 3.048, 6.096 mm/min for feed rate. Zhang et al. (2011) studied the ultrasonic process with variable spindle speeds (1,000 up to 3,000 rpm) and observed that there is not any influence of tool speed on the MRR. Hu et al. (2002) conducted ultrasonic drilling process for zirconia ceramic by varying the RUM power rating (percentage of 500 Watts) from 40% to 70% in step size of 10% and finally achieved best MRR at a power rating of 70%. Lauwers et al. (2010) examined the distinct cutting depth (10–50 µm) and different feed rate (100–500 mm/min) during the ultrasonic grinding process. The outcome demonstrated that a feed rate of 100 mm/min resulted in a minimum cutting force while machining of ZrO2. Li et al. (2005) investigated the viability of RUM on two varieties of ceramic matrix composites (CMC). It was found that with diamond drilling procedure, the cutting force can be lowered substantially and the edge quality of hole at entrance and exit was made almost free of fibre pull-out, under appropriate machining parameters. Kataria et al. (2015) investigated the hole quality and machining characteristics of tungsten carbide composite by utilising a sinker type ultrasonic machine. Power rating, diamond abrasive grit size, and tool material were found to be the most significant parameters for machining characteristics namely; MRR and TWR. The highest MRR was obtained at a combination of high power rating and coarse grit size. Singh and Singhal (2016a, 2016b, 2016c) investigated the machining characteristics of Al2O3 and quartz ceramic by employing an experimental design approach through response surface methodology (RSM) and concluded that higher feed rate gives the best solution with respect to MRR. Increased spindle speed leads to an increase in the volume of indentation proportionally, which further caused higher MRR. Furthermore, the reference Churi et al. (2006) and Zou et al. (2013) studied the effects of machining variables of RUM for titanium and its alloy to analyse its effects on MRR, cutting force, roughness inside the hole surface and cutting temperatures. Lv et al. (2013) worked on RUM and concluded that subsurface damage induced in RUM was discriminated as grinding, chipping and cracking on the BK7 glass. Feng et al. (2012) investigated the hole quality, roughness, TWR, MRR, thrust force, at distinct levels of parameters such as RPM, ultrasonic power and feed rate of RUM.

Much research has been carried out on RUM under different working environment. It has been found out that till today, 40% literature is on RUM development, 58% on monitoring and control of processes and only 2% work has been done on the optimisation of the process variables of RUM (Jiao et al., 2005; Zhang et al., 2013; Churi et al., 2007). In addition, hardly any study reported on RUM for nickel alloys which offer low amplitude and high-frequency vibration; 28 µm and 20,000 Hz, respectively. To improve the productivity and quality of the product, a performance analysis must be carried out


for such kind of complex materials (i.e., Inconel 718) on the RUM. The main objective of the present research is to study/investigate the effect of different process parameters of RUM and find out significant or non-significant terms using RSM.

2 Materials and methods

In the present work, RUM is used to carry out the necessary experiments, made by Sonic-Mill Series 10 (Sonic-Mill, Albuquerque, NM, United States). In the machining setup, a high-frequency (20 kHz) electrical energy is converted from a normal power supply (50Hz). A piezoelectric transducer converts electrical energy into mechanical activity. The amplified motion causes the diamond tool attached to the spindle to oscillate along feed direction. The frequency of vibration is 20 kHz. The power rating can be adjusted by adjusting the power supply. An electric motor is attached to the top of the ultrasonic spindle supplies rotating motion to the tool. The variable speeds are adjusted by the speed controller attached to the control panel. The coolant flow can be adjusted by the valve mounted on the pressure gauge. Diluted water-soluble coolant or cutting oil (Mobilmet S-122, Mobil Oil Corporation, Fairfax, VA, United States) is used with the ratio of oil to water 1:20. Figure 1 illustrates the RUM process.

Figure 1 Illustration of rum process (see online version for colours)

The Inconel 718, a Ni-based super alloy is selected for the machining on RUM. The size of the workpiece is decided on the basis of the working conditions. A square sheet of 50 × 50 × 5 mm thickness is used. So that it can be easily held to the fixture. The mechanical properties and chemical composition of Inconel 718 are demonstrated briefly in Table 1.


Table 1 Chemical composition and other mechanical properties of Inconel 718

Chemical composition (by weight %) of material Element Ni Fe Cr Nb Mn C Co Al Si Ti Mo Other Weight (%) 51.05 19.43 18.7 5.7 0.07 0.04 0.2 0.56 0.08 1.01 3.1 0.06 Yield strength 1035 MPa Ultimate strength 1240 MPa Hardness 97 HRB Density 8.19 g/cm3

In the present research work, a slotted metal bonded diamond core drill tool is utilised for the machining of selected material. It has wide applications in machining of brittle and hard materials. In metal bonded tool, diamond abrasives are mixed with metal powder to make a bond between abrasives and metal. The diamond segments only take part in material removal. With the passage of time, the metallic powder wears down and exposes new diamond crystals at the tool surface, thereby maintaining a fresh abrasive on the cutting surface.

The dimension and profile of the tool play a critical role in the RUM process. For the present investigation, the outer (OD) and inner (ID) diameters of the diamond drill are selected as 8 mm and 6 mm respectively. Figure 2 shows the profile of tool used for the present research work.

Figure 2 Metal bonded tool and its profile (see online version for colours)


2.1 Experimentation

A human process planner can select machining process parameters using his own knowledge or experience. However, every time these parameters do not give the optimal outcomes. Studying the effects of experiment’s parameters needs many trials and a lot of time. Several design-of-experiments (DOE) approaches are broadly used to overcome this kind of problems. To find the optimum conditions of parameters for RUM, RSM is used design the layout of the experiments. RSM is a collection of mathematical and statistical techniques useful for analysing problems in which several independent variables influence a dependent variable or response, and the goal is to optimise this response. The advantages of using the RSM method are that many more factors can be optimised simultaneously and quantitative information can be obtained by only a few experimental trials. Table 3 shows the experimental plan to conduct the experiments. Total 21 experiments are decided to perform. The range of the four selected variables as input parameters are shown in Table 2. For four variables (n = 4) and two levels [low (–) and high (+)], the total number of experiments 21 is determined by the expression: 2n – 1(24 – 1 = 8: factor points) + 2 × n(2 × 4: 8axial points) + 5 (centre points), as shown in Table 3. For the present case, four RUM parameters such as tool rotation, feed rate, ultrasonic power and diamond abrasive grit size are selected. A full second-order polynomial model is obtained by multiple regression technique for four factors by using analysis of variance (ANOVA). In developing the regression equation developed by Box and Hunter (1957). The test factors are coded according to the following equation (Arnold, 2006):

0i ii

i

X XxX−

=Δ

(1)

where xi is the coded value of the ith independent variable, Xi the natural value of the ith independent variable, Xi0 the natural value of the ith independent variable at the centre point, and ΔXi is the step change value.

12

01 1 2 1

k k k k

i i ij i j ii ii i j i

Y b b x b x x b x e−

= = = =

= + + + +∑ ∑∑ ∑ (2)

where Y is the predicted response, b0 and the offset term, bi the linear effect, bij represents square effect and bii is the interaction effect. x represents the significant independent and the mathematical relationship of the response Y on these variables can be approximated by second order equation as shown below (Lorenzen and Anderson, 1993):

2 20 1 1 2 2 3 3 4 4 11 221 2

2 233 44 13 1 3 23 2 333 44

Y b b x b x b x b x b x b xb x b x b x x b x x

= + + + + + +

+ + + + (3)

where Y is the predicted response variable, b0 is the constant, b1, b2, b3 and b4 are the linear coefficients, b13 and b23 are the cross-product coefficients, and b11, b22, b33 and b44 are the quadratic coefficients. Some other parameters are kept constant during the machining namely; tool diameter 8 mm, the frequency of vibration 21 kHz, coolant pressure 300 kPa.


Table 2 Factors and their levels

Sr. no. Symbol Input factors Level

Units Low Centre High

1 A Spindle speed (RPM) 4,200 4,600 5,000 5,400 5,800 RPM 2 B Feed rate (mm/sec) 0.01 0.0125 0.0150 0.0175 0.02 mm/sec 3 B Power rating (%) 55 60 65 70 75 % 4 D Diamond abrasive grit

size (mesh) 80 100 120 140 160 mesh

Table 3 Design of experiment and their results

Std Run

Factor 1 Factor 2 Factor 3 Factor 4 Response 1 Response 2 Tool

rotation Feed rate Ultrasonic power

Diamond abrasive grit size MR Ra

RPM mm/sec % of 500 Watts Mesh mm3/sec µm

15 1 5,000 0.02 65 120 1.00204 1.398 18 2 4,600 0.0175 60 140 0.807896 1.311 6 3 5,000 0.015 55 120 0.711062 1.019 21 4 5,000 0.015 65 160 0.756865 1.131 14 5 5,800 0.015 65 120 0.704554 0.912 11 6 5,000 0.015 65 120 0.70473 1.016 2 7 4,600 0.0125 60 100 0.607752 1.033 8 8 4,200 0.015 65 120 0.71027 1.161 10 9 5,000 0.015 65 120 0.710414 1.041 17 10 5,400 0.0125 60 140 0.543413 0.603 5 11 5,400 0.0175 70 100 0.89124 1.136 19 12 5,400 0.0125 70 140 0.545573 0.902 3 13 5,400 0.0175 60 100 0.879512 1.159 16 14 5,000 0.015 75 120 0.689029 1.005 9 15 5,000 0.015 65 120 0.72035 1.001 20 16 4,600 0.0175 70 140 0.82261 1.196 12 17 5,000 0.015 65 120 0.70145 0.999 1 18 5,000 0.015 65 80 0.731171 1.233 7 19 5,000 0.01 65 120 0.4382 0.711 4 20 4,600 0.0125 70 100 0.620726 0.98 13 21 5,000 0.015 65 120 0.711422 0.996

The machining rate (MR) and roughness (Ra) are selected as the response variables for this investigation. MR is obtained from the weight measurement (before and after each experiment) of the workpiece. An electronic weighing scale (±0.0002 g) is used for measuring the weight of workpiece after each trial. The measurements are repeated two times. The response parameter MR is calculated by using the equations (4) and (5). Ra is measured by using roughness measuring machine (Surfcom, Flex) as shown in Figure 3.


volume density mass= × (4)

volume of material removed from workpiecetime of

MachiningRatemachining

= (5)

Figure 3 Roughness metre (see online version for colours)

Source: Surfcom

3 Results and discussions

The central composite design (CCD) is used to build up an empirical model to inter-relate RUM process parameters. 21 experiments are carried out according to CCD. To predict the response and analyse the process parameters producing an optimum value of variables, empirical models are developed. For the development of empirical models, the responses taken are

1 machining rate (MR)

2 roughness (Ra).

Using statistical analysis tool, control of each parameter on each performance feature is discussed. The experiments are carried out with one replication. The average values of MR and Ra are shown in Table 3. For the analysis and checking the fitness value of the model, ANOVA is performed. The model adequacy checking incorporates a test

1 for the importance of empirical model

2 lack of fitness test.

The following sections analyse the results obtained for MR and Ra.


3.1 Analysis of MR

The experimental results for MR are demonstrated in Table 3. These results are analysed using design-expert package. Three different tests, lack-of-fit test, the sequential model sum of squares, and model summary statistics are performed to check the adequacy of the model. The quadratic model is found to be adequate as shown by the fit summary of the tests. Table 4 revealed that there are a lot of insignificant model terms available in the model. Hence, to improve the adequacy of the model, the model reduction is performed by the backward elimination process that eliminates the unimportant terms in order to fit the quadratic model for sustaining the hierarchy of the model. Table 4 ANOVA for MR

Source Sum of squares df Mean

square F-value p-value prob > F

Model 0.318362 14 0.02274 193.3997 8.99E-07 significant A – tool rotation 1.63E-05 1 1.63E-05 0.138936 0.722153 B – feed rate 0.158958 1 0.158958 1351.897 2.7E-08 C – ultrasonic power

3.88E-07 1 3.88E-07 0.003296 0.956085

D – diamond abrasive grit size

0.00033 1 0.00033 2.807342 0.144852

AB 0.006853 1 0.006853 58.28119 0.000264 AC 2.38E-05 1 2.38E-05 0.202456 0.668545 AD 0.00012 1 0.00012 1.02375 0.35068 BC 1.6E-05 1 1.6E-05 0.135939 0.725012 BD 9.28E-06 1 9.28E-06 0.078934 0.788184 CD 7.66E-06 1 7.66E-06 0.065144 0.807067 A2 2.25E-05 1 2.25E-05 0.19168 0.676843 B2 0.000125 1 0.000125 1.060881 0.342733 C2 0.000195 1 0.000195 1.660648 0.244971 D2 0.001689 1 0.001689 14.36612 0.009071 Residual 0.000705 6 0.000118 Lack of fit 0.000496 2 0.000248 4.72991 0.088316 Not significant Pure error 0.00021 4 5.24E-05 Cor. total 0.319068 20

Table 5 shows the results of pooled ANOVA after implementing the backward elimination. Figure 4(a) shows the normal probability plot of the residuals which reveals that more than 90% of residuals are declining within ±3 sigma limit and residuals are settled by the straight line (Torres et al., 2015). In additional, it can be seen from Figure 4(b) that the experimental values are chasing the predicted ones calculated from the obtained model. The F-value of the model is obtained as the mean square of the model divided by the residual mean square. The F-value test is carried out by analysing a


comparison of the model variance with the residual variance. Once the value of the variances is just about same, the proportion will be close to unity and which is not as likely that model has a significant influence on the response. A specific source of deviation is significant if the computed F-value at a particular confidence level is larger than the tabulated F-value at the same level of confidence. The model has F-value of 556.61 and P-value significantly less than 0.01. It depicts that the model is significant for machining rate. Table 5 Pooled ANOVA for MR



Model 0.318007 7 0.04543 556.6073 4.52E-15 significant A – tool rotation 7.13E-06 1 7.13E-06 0.087311 0.77229 B – feed rate 0.305664 1 0.305664 3745.019 2.19E-17 C – ultrasonic power

3.88E-07 1 3.88E-07 0.004748 0.946115


0.00033 1 0.00033 4.044304 0.06553

AB 0.006853 1 0.006853 83.96087 4.91E-07 C2 0.000247 1 0.000247 3.030272 0.105323 D2 0.001729 1 0.001729 21.18606 0.000495 Residual 0.001061 13 8.16E-05 Lack of fit 0.000851 9 9.46E-05 1.804819 0.298475 Not significant Pure error 0.00021 4 5.24E-05 Cor. total 0.319068 20

Figure 4 (a) Normal probability plot (b) Predicted vs. actual response for MR (see online version for colours)

(a) (b)


The B is found the most significant model terms followed AB and D for machining rate. The percentages of contributions can be computed by dividing the ‘individual’ sum of squares to the ‘model’ sum of squares. Figure 5 shows the percentage contribution of all significant terms. Table 6 R2 values for MR

Std. dev. 9.034E-003 R-squared 0.9967 Mean 0.71 Adj. R-squared 0.9949 C.V. % 1.26 Pred. R-squared 0.9861 PRESS 4.441E-003 Adeq. precision 99.150

Higher than 0.1 P-value reveal the insignificance of the model term. The lack-of-fit values of 1.80 show that it is insignificant relative to pure error. There is only 29.85% possibility that a lack-of-fit value may occur due to noise. Therefore, the built up model can be established. From Table 6, it is revealed that the R2 value for machining rate is found to be 0.9967. This high value of R2 may propose the model to be adequate. As far as other R2 statistics are concerned, there is a close agreement between pred. R2 (0.9861) and adj. R2 (0.9949). Furthermore, the adeq precision measures the signal-to-noise ratio. The adeq precision found for the model is 99.15 Usually, a value greater than four is desirable (Montgomery, 2008). Thus, the model can be utilised to find predicted values of the machining rate. From the obtained model for machining rate, the values of R2 (0.9967), and adeq precision (99.15) shows significance for fitting and predicting the experimental results. The equation (6) defines the regression model for machining rate for RUM process.

2

2.73516145 0.000622529 151.66690 0.015655503 0.0044647 0.04139075 0.0001206664 0.

Machining Rate Tool RotationFeed Rate Utrasocic PowerDiamond Abrasive Grit Size Tool RotationFeed Rate Ultrasonic Power

= + − × −× + × −× + ×

× − ×

+ 2000019941 Diamond Abrasive Grit Size×

(6)

3.2 Analysis of Ra

The investigation results shown in Table 3 are analysed for Ra. On the basis of fitted value, the adequacy of the quadratic model is proposed by design expert package. ANOVA is performed in order to analyse the obtained results of the quadratic model. From Table 7, it is found that there is a lot of non-significant model terms present in the model. Table 8 shows the results of pooled ANOVA. A backward elimination method is used to eliminate the non-significant terms from the quadratic model. From Figure 6 it can be observed that more than 90% of residuals are falling within ±3σ limits. It can also be seen from Figure 6(b) that the experimental values are chasing the predicted ones calculated from the obtained model for Ra.


Table 7 ANOVA for Ra



Model 0.648196 14 0.0463 43.03316 7.7E-05 significant A – tool rotation 0.031001 1 0.031001 28.81333 0.001715 B – feed rate 0.235985 1 0.235985 219.3351 5.96E-06 C – ultrasonic power

0.0004 1 0.0004 0.371779 0.564398

D – diamond abrasive grit sise

0.005202 1 0.005202 4.834984 0.070212

AB 0.000529 1 0.000529 0.491678 0.509447 AC 0.024642 1 0.024642 22.90344 0.003045 AD 0.000506 1 0.000506 0.470533 0.518355 BC 0.018432 1 0.018432 17.13157 0.006085 BD 0.00308 1 0.00308 2.86293 0.141591 CD 0.00845 1 0.00845 7.853829 0.031063 A2 0.000166 1 0.000166 0.153925 0.70837 B2 0.001254 1 0.001254 1.165601 0.321783 C2 0.000317 1 0.000317 0.295018 0.606596 D2 0.038064 1 0.038064 35.37855 0.00101 Residual 0.006455 6 0.001076 Lack of fit 0.005062 2 0.002531 7.267083 0.046577 significant Pure error 0.001393 4 0.000348 Cor. total 0.654652 20

Figure 5 Percentage contribution of RUM parameters (see online version for colours)


Figure 6 (a) Normal probability plot (b) Predicted vs. actual response for Ra (see online version for colours)

(a) (b)

Table 8 Pooled ANOVA for Ra


square F-value p-value Prob > F

Model 0.65 9 0.072 82.76 < 0.0001 significant A – tool rotation 0.031 1 0.031 35.79 < 0.0001 B – feed rate 0.44 1 0.44 509.82 < 0.0001 C – ultrasonic power

4.000E-004 1 4.000E-004 0.46 0.5108


0.016 1 0.016 18.04 0.0014

AC 0.025 1 0.025 28.45 0.0002 BC 0.018 1 0.018 21.28 0.0007 BD 3.080E-003 1 3.080E-003 3.56 0.0860 CD 8.450E-003 1 8.450E-003 9.76 0.0097 D2 0.040 1 0.040 46.43 < 0.0001 Residual 9.527E-003 11 8.661E-004 Lack of fit 8.134E-003 7 1.162E-003 3.34 0.1307 not significant Pure error 1.393E-003 4 3.483E-004 Cor. total 0.65 20

The conformity for the reliability of ANOVA results is shown in Table 8. The significance of the model is shown in this table with the 82.76 F-value. The resulted P-value less than 0.0001 points to the significance of the model terms. In the present study, the model terms A, is found the most significant term of RUM followed by A, D, A, C, AC, BC, CD, and BD for Ra. The lack-of-fit value obtained from the model is 19.03


which show that its value is insignificant relative to pure error. There is only 13.07% chance that a lack-of-fit could occur due to noise. Insignificant lack of fit is fine since it allows the model to fit the actual or experimental values. Table 9, shows that the R2 value for Ra is 0.9854. It explains that there is less variation between predicted value and experimental value of Ra. As far as other R2 statistics are concerned, there is a good agreement in the value of pred R2 (0.9986), and Adj R2 (0.9735). Furthermore, the adeq precision of 35.581 measures the signal-to-noise ratio. Usually, a value greater than four is desirable (Montgomery, 2008). Thus, the developed model can be utilised to navigate the design space and predict the values of the Ra. From the obtained model for Ra, the values of R2 (0.9854), and adeq precision (35.581) shows significance for fitting and predicting the experimental results. The equation (7) defines the regression model for roughness:

11.09754 0.0019593 249.450.11915 0.0538289

0.000027753.84

0.5549

Roughness Tool RotationFeed Rate Ultrasonic PowerDiamond Abrasive Grit Size Tool RotationUltrasonic Power Feed Rate UltrasonicPower

Feed Ra

= + − × +× − × −× + ×× − × ×+ ×

2

0.0003240.00009506

te Diamond Abrasive Grit SizeUltrasonic Power Diamond Abrasive Grit SizeDiamond Abrasive Grit Size

× +× × +

×

(7)

Table 9 R2 values for Ra

Std. dev. 0.029 R-squared 0.9854 Mean 1.04 Adj. R-squared 0.9735 C.V. % 2.82 Pred. R-squared 0.9096 PRESS 0.059 Adeq. precision 35.581

3.3 Effect of significant process parameters on MR

From the ANOVA, (Table 5) results demonstrate that the variable feed rate (B) keep highly significant model term, second order terms of tool rotation (C2), (D2) and mutual terms AB are found as considerable effects on the MR. The MR is found to increase (0.57432 mm3/sec–0.84478mm3/sec) sharply as penetration rate of tool incremented from 0.0125 mm/s to 0.0175 mm/s. A little decrement of found in MR (0.71163 mm3/sec–0.71038 mm3/sec) as tool rotation incremented from 4,600 rpm to 5,400 rpm. From the ANOVA results, it is also clear that ultrasonic power of the tool shows a square model term and for the reason that a curvature is obtained in Figure 7. From ultrasonic power 55% to 65%, MR is increased (0.69923 mm3/sec–0.71099 mm3/sec) and then decreased (0.71099 mm3/sec–0.69865 mm3/sec) up to 75%. Big diamond abrasive size means lower the value of mesh value. Figure 7 depicts that higher machining rate can be achieved when big abrasive size of diamonds are used in diamond core drill tool. Like ultrasonic power abrasive size of diamond also shows a square model term and a curvature is obtained in Figure 7. The lower is the mesh size, faster the removal of debris and swarf from the machining surface. From abrasive grit size 80 mesh to 120 mesh, MR is


decreased (0.73005 mm3/sec–0.71099 mm3/sec) and then increased (0.71099 mm3/sec–0.75574 mm3/sec) up to 160 mesh.

Figure 7 Effect of parameters on MR, (a) tool rotation (RPM) (b) feed rate (mm/sec) (c) ultrasonic power (% of 500 watts) (d) diamond abrasive grit size (mesh) (see online version for colours)

(a) (b)

(c) (d)

In the interaction plot, the simultaneously two parameters are varied remaining two process parameters kept constant. This type of plot is called 3D surface plot. The interactions plots are plotted in Figure 8 tool rotation is an insignificant factor in the RUM of Inconel 718. Although, a combined effect of feed rate and rotation speed play a vital role in cutting.

The combined effect of feed rate and spindle speed on MR (Figure 8) shows that MR goes to the highest value 0.98641 mm3/sec at a maximum value of tool rotation (5,400) and higher value of feed rate (0.0175). A possible reason for happening it may be the increase in tool rotation speed and feed rate, it enhances the contact duration between workpiece and tool. It may cause the higher metal removal from the surface of the workpiece. While it arrives at its minimum level (0.43452) when the tool rotation


(5,400), and feed rate are at minimum (0.01) level. This is due to the fact that it diminished the contact time between workpiece and tool. The higher tool rotation and feed rate both factors are responsible for frequent metal removal in RUM.

Figure 8 Two variables interactive plot for MR (see online version for colours)

3.4 Effect of significant process parameters on Ra

In the RUM of Inconel 718, main effects of variables namely; tool rotation (A), feed rate (B), diamond abrasive grit size of tool (D), interaction terms (AC), (BC), (BD), (CD), second order term of feed rate (B2) are found to have significant effects on the machining surface. Tool rotation (A), feed rate and diamond abrasive grit size of the tool is observed to contribute approximately 90% of the overall disparity in the response data. Figure 9 shows the effects of all factors on the roughness of the machining surface. The Ra is found to increase steeply as feed rate is incremented from 0.01 mm/s to 0.02 mm/s. Higher Ra has revealed as feed rate of diamond impregnated core drill tool increases. The increment in feed rate causes into the increased indentation depth of the diamond abrasives which further promotes the rough cutting of work surface (chipped out and dipper groove region), hence increased Ra is observed. However, the increment in tool rotation resultant with continuing decreases in Ra. From the ANOVA (Table 8) it is noted that abrasive grit size of the tool shows a square model term and because of this, a curvature is plotted against the roughness. From abrasive grit size 80 mesh to 120 mesh, surface roughness is decreased (1.2305 micron–1.0159 micron) and then increased (1.0159 micron–1.1037 micron) up to 160 mesh.

Figure 10 illustrated two factors interactive effect plots while machining Inconel 718 with the RUM process. Figure 10(a) illustrates the combined effects of ultrasonic power and tool rotation on Ra. The Ra is observed to fall (1.1391 micron–0.7917 micron) as tool rotation level improved from 4,200 rpm to 5,800 rpm. As tool rotation enhances from 4,200 rpm to 5,800 rpm, the chances of enhancing the grinding action which further promotes the fineness in the machining surfaces. Furthermore, increase in ultrasonic power with tool rotation resulted into enhancing the indentation length of abrasives which


heals the sharp regions of the work surface. A local rise in temperature of the machining region which makes work surface softer and enhances the propensity of plastic flow of material, thus reduced Ra is observed. The very slow increment (1.010 micron– 1.021 micron) in Ra is observed as ultrasonic power incremented from 55% to 75%.

Figure 9 Effect of parameters on Ra, (a) tool rotation (RPM) (b) feed rate (mm/sec) (c) ultrasonic power (% of 500 Watts) (d) diamond abrasive grit size (mesh) (see online version for colours)

(a) (b)

(c) (d)

The combined influence of feed rate and ultrasonic power on Ra is represented in Figure 10(b). Higher Ra is revealed as feed rate of diamond impregnated core drill tool increases. The increment in feed rate causes into the increased indentation depth of the diamond abrasives which further promotes the rough cutting of work surface (chipped out and dipper groove region), hence increase in surface roughness is observed. The Ra is reduced at higher levels of ultrasonic power which may be due to the grinding length of the abrasive tool.

The combined influence of feed rate and abrasive grit size of diamonds on Ra is represented in Figure 10(c). Higher Ra is revealed as feed rate of diamond impregnated core drill tool increases. With increase in feed rate, Ra increases steeply, whereas Ra is


reduced at higher levels of abrasive grit size which may be due to the conventional grinding action of the abrasive tool.

The combined influence of ultrasonic power and abrasive grit size on Ra is described in Figure 10(d). With increase in ultrasonic power, Ra decreases steeply, whereas Ra first decreases as abrasive grit size decrease from 80 mesh to 120 mesh, and then increases from 120 mesh to 160 mesh. Figure 11 shows a pictorial view of machined workpiece ‘Inconel 718’.

Figure 10 Two variables interactive plots for Ra (see online version for colours)

(a) (b)

(c) (d)

Figure 11 Pictorial view of machined ‘Inconel 718’ workpiece (see online version for colours)


3.5 Micro structure analysis of machined surface

To study the microstructure of workpiece before and after machining with RUM, scanning electron microscopy test is executed on the work samples. Figure 12 shows the microstructure of Inconel 718 prior to RUM process. Before performing the SEM analysis, the machined specimens/workpiece are cleaned with the acetone solution. Prior to RUM it can be observed that the grain size is almost uniform throughout the surface.

Figure 12 Microstructure of Inconel 718 prior to RUM process

Figure 13 Microstructure referred to experimental run 5 (see online version for colours)




Figures 13, 14 and 15 demonstrated the microstructure of machined surface processed under the experimental conditions corresponding to experiment no. 5, 11 and 19 respectively. Figure 13 depicts the SEM image of the machined surface corresponding to the experimental no. 5 at 1,000 X. The parametric setting for experimental run 5 was having a combination of high level of feed rate and tool rotation. In RUM process, the diamond abrasive tool vibrating at ultrasonic frequency hammered the machining surface and because of this the initiation and propagation of micro-cracks take place. The cracks propagation length is longer as the feed rate is high, so material removal volume of crack propagation is also larger, and the surface morphology change intensity is gentle. While the cracks propagation length is shorter because of low feed rate, so material removal volume of crack propagation is tiny, and the detail of surface morphology is clear (Figure 14). The deep indentations mark present over the surface, further promote the removal of material from machining surface. Higher tool rotation is showing great influence over the machined work surface (Singh and Singhal, 2016a; Pi et al., 2013). This removal of material in the form of bigger chunks is due to the higher feed rates that cause the larger indentation of abrasives into the work surface.

Figure 15 depicts the SEM image of the machined surface corresponding to the experimental no. 11 at 1,500 X. This microstructure reveals that, at the parametric setting of moderate level of higher tool rotation, feed rate, ultrasonic power, and abrasive diamond grit size, more regions of plastic deformation can be observed. Higher contact of the diamond abrasives cause the shatter of the work material, and hence results in a regime layer spread over the machined surface.

In addition to this, the machined surface has also been observed to be free from any kind of thermal effects. Figure 16 exemplified the EDX image of workpiece surface after RUM process which ensures that the composition of the machined surface still remains same to its parental material’s properties. The material is also coupled with the superior fracture toughness which does not support the propagation of micro cracks. In machining of Inconel 718, crack propagation is rarely observed because the machining work surface is pressed by the tool abrasive. Figures 17, 18 and 19 demonstrated the two-dimensional waviness and roughness profile of the machined surfaces corresponding to experiment no. 5, 11 and 19, respectively. A small region is extracted from the machined surface to analyse the surface topography. Profile and waviness in Figures 17, 18 and 19 show the trend of the average roughness and volume of the material removal over the entire machined surface corresponding to experiment no. 5, 11 and 19, respectively.


Experimental run 5 (Figure 17) shows higher machining rate over the other experimental runs and experimental run 19 (Figure 19) shows lower surface roughness over others experimental runs.

Figure 16 EDX analysis of Inconel 718 (see online version for colours)

Figure 17 Two-dimensional waviness and profile of the machined surface (see online version for colours)

Note: Experiment no. 5







4 Optimisation

Increasing the value of machining rate decreased the integrity of the machined surface. The setting of RUM at high value causes high machining rate but result in high irregularities on the work surface. Therefore, it is essential to obtain the optimal parameters setting to achieve higher machining rate and minimum roughness. The optimal RUM parameters setting for MR and Ra can be obtained using desirability function. Derringer and Suich (1980) proposed a multi-objective optimisation technique called desirability function. This common technique is to first convert each of the response yi(x) into an individual desirability function (di) and vary over the range zero to one. In the present investigation, statistical tool is utilised to find the optimal values of the response variables simultaneously. Derringer and Suich (1980) proposed the few types of desirability function. Some of them as:

1 For the smaller-the-better type

*

*

*

*

1

,

0

i i

t

i i

i i i i

i i

i i

y y

y yd y y yy y

y y

′′≤

⎡ ⎤− ′′= < <⎢ ⎥′′−⎣ ⎦≥

⎧⎪⎪⎨⎪⎪⎩

2 For the ‘larger-the-better’ type

*

**'

*,

0

1

i i

ti i

i i i ii i

i i

y y

y yd y y yy y

y y

≤⎧⎪⎪⎡ ⎤− ′= < <⎨⎢ ⎥−⎪⎣ ⎦⎪ ′≥⎩

Overall desirability function of the multi-response system may be measured by combining the individual desirability functions. It is represented as

( )1 21 2, , ,w w wn

nD d d d= … where wj(0 < wj < 1) is the weight value given for the importance

of jth response variable and 1

0

1.jj

w=

=∑ The parameters settings corresponding to

maximum overall desirability value considered to be the optimal setting (Kumar et al., 2016).

4.1 For single response optimisation

The optimisation of single response has been performed if productivity (MR) dominates the quality (Ra) or vice versa. For the process variables to be optimised, the goals and range are illustrated in Table 10. The goal is set to ‘in range’ for all the process parameters and responses, which means that parameters will be optimised for searching the space within the design limit i.e., lower and upper limit. All the variables are assigned with equal weights and same importance. The machining solution possesses overall


desirability value which must be closer to one as optimised solution. For MR as the single response optimisation: tool rotation – 5,200 rpm feed rate – 0.0198 mm/s, ultrasonic power – 65%, and diamond abrasive diamond grit size – 80 mesh are the best combination for optimum recipe. For Ra as the single response optimisation: tool rotation – 4,900 rpm, feed rate – 0.0101 mm/s, ultrasonic power – 60% and diamond abrasive diamond grit size – 140 mesh are the best combination for optimum recipe. At these predicted optimum conditions, the confirmatory experimental runs have been conducted. The predicted values and average of confirmatory experimental results for MR and Ra are also tabulated in Table 11. Table 10 Factors with their constraints

Name Goal Lower limit

Upper limit

Lower weight

Upper weight Importance

A Tool rotation in range 4,200 5,800 1 1 3 B Feed rate in range 0.01 0.02 1 1 3 C Ultrasonic power in range 55 75 1 1 3 D Abrasive diamond

grit size in range 80 160 1 1 3

Machining rate maximise 0.4382 1.00204 1 1 3 Roughness minimise 0.603 1.398 1 1 3

Figure 20 Corresponding values of all factors with desirability (see online version for colours)


Table 11 The optimised settings of process variables

Opt

imis

ed c

ondi

tion

S. n

o.

Opt

imis

atio

n Re

spon

se

Tool

rota

tion

(RPM

) Fe

ed ra

te

(mm

/sec

) U

ltras

onic

po

wer (

%)

Dia

mon

d ab

rasi

ve

grit

size

(mes

h)

Pred

icte

d C

onfir

mat

ory

expe

rim

ents

Mac

hini

ng ra

te

(mm

3 /sec

) 5,

200

0.01

98

65

80

1.02

617

0.98

614

1 Si

ngle

resp

onse

op

timisa

tion

Rou

ghne

ss (µ

m)

4,90

0 0.

0101

60

14

0 0.

4717

0.

4920

0.

8684

9 0.

8345

2 2

Mul

ti re

spon

se

optim

isatio

n M

achi

ning

rate

and

ro

ughn

ess

5,80

0 0.

0167

55

14

0 0.

8002

0.

7763


4.2 For multi-response optimisation

The responses MR and Ra are highly conflicting in nature. The aim of the present investigation is to hit upon the optimal parameter settings and attain the maximum overall desirability for superior RUM machining rate and surface roughness. The ranges of input parameters namely tool rotation, feed rate, power rating and abrasive diamond grit size for machining rate and surface roughness are given with their constraint in Table 10. Table 11 shows the grouping of RUM process parameters that offer the maximum desirability. Figure 20 shows the plots offering overall desirability for MR and Ra. The overall desirability index of 0.76 is attained at optimised condition, as shown in Figure 20. As the higher value of desirability, optimal recipe of RUM parameters for multi-response is as: tool rotation 5,800: feed rate: 0.0167, ultrasonic power: 55, and mesh size: 140. Experimental values resulted in the corresponding setting to optimal response for MR and Ra are 0.83452 mm3/sec and 0.7763 micron that is found much closer to the predicted values.

5 Conclusions

This research is aimed at exploring the use of the rotary ultrasonic machine for cost-effective machining of super alloy and establishing optimised process settings for MR and Ra through a central composite technique for designing the experiments. The following conclusions have been drawn from this research:

• With the assistance of ultrasonic vibration, a better surface quality can be expected for machining of super alloys.

• Using ANOVA, quadratic model is found significant for both MR and Ra. Increasing feed rate increases the indentation rate of tool that results in a high value of MR but it adversely affects the Ra. Increasing mesh size decreases the MR but improves the surface finish as fine grinding. The factor tool rotation is no very significant in the case of machining rate but improves the Ra due to the rotary grinding action RUM. Similarly, ultrasonic power rating also is showing the same effect as tool rotation.

• During RUM, as the penetration depth tool increases (at higher feed rate), the material is removed in the form bigger chunks and developed inter granular cracks further causes the grains removed out quite easily.

• Developed empirical models for roughness and machining rate will provide guidelines for the prediction of Ra and MR in advance. Additive test results show that models are quite fit with the experimental test results.

• The best possible parametric combination for MR was devised as tool rotation – 5,200 rpm, feed rate – 0.0198 mm/sec, ultrasonic power – 65%, and diamond abrasive grit size – 80 mesh. The best parametric combination for Ra was obtained as tool rotation – 4,900 rpm feed rate – 0.0101 mm/sec, ultrasonic power – 60% and diamond abrasive grit size – 120 mesh.


• The optimum parametric combination for MR and Ra is formulated as are as tool rotation – 5,800 rpm, feed rate – 0.0167 mm/s, ultrasonic power – 55%, and diamond abrasive grit size – 140 mesh. The maximum value of MR, i.e., 0.86849 mm3/sec and a minimum value of Ra, i.e., 0.8002 µm indicate that the Inconel 718 can be effectively machined by rotary ultrasonic machine despite the toughness of the material. The RUM can also be utilised for the machining of other tough aerospace materials.

References A.S.M. International and A. Rights (2000) ASM Specialty Handbook: Nickel, Cobalt and their

Alloys. Abdo, B.M.A., Darwish, S.M., Al-Ahmari, A.M. and El-Tamimi, A.M. (2013) ‘Optimization of

process parameters of rotary ultrasonic machining based on Taguchi’s method’, Adv. Mater. Res., Vol. 748, pp.273–280.

Arnold, S.F. (2006) ‘Design of experiments with MINITAB’, Vol. 60, No. 2. Bhatt, A., Attia, H., Vargas, R. and Thomson, V. (2010) ‘Wear mechanisms of WC coated and

uncoated tools in finish turning of Inconel 718’, Tribol. Int., Vol. 43, Nos. 5–6, pp.1113–1121. Box, G. and Hunter, J. (1957) ‘Multi-factor experimental designs for exploring response surfaces’,

Ann. Math. Stat, Vol. 28, No. 1, pp.195–241. Churi, N.J., Pei, Z.J. and Treadwell, C. (2006) ‘Rotary ultrasonic machining of titanium alloy:

effects of machining variables’, Mach. Sci. Technol., September, Vol. 10, No. 3, pp.301–321. Churi, N.J., Pei, Z.J., Shorter, D.C. and Treadwell, C. (2007) ‘Rotary ultrasonic machining of

silicon carbide: designed experiments’, Int. J. Manuf. Technol. Manag., Vol. 12, Nos. 1/2/3, pp.284–298.

Derringer, G. and Suich, R. (1980) ‘Simultaneous optimization of several response variables’, J. Qual. Technol., Vol. 12, Nos. 4, pp.214–219.

Dreshfiezd, R.L. (1970) Estimation of Gamma Phase Composition in Nickel-Base Superalloys (Based on Geometric Analysis of a Four-Component Phase Diagram), NASA Tech. Note.

Feng, D.J. and Xu, Z.G. (2012) ‘Research on tool wear and compensation method of ultrasonic milling’, Appl. Mech. Mater., Vols. 217–219, pp.2088–2092.

Feng, Q., Cong, W.L., Pei, Z.J. and Ren, C.Z. (2012) ‘Rotary ultrasonic machining of carbon fiber-reinforced polymer: feasibility study’, Mach. Sci. Technol., July, Vol. 16, No. 3, pp.380–398.

Hu, P., Zhang, J.M., Pei, Z.J. and Treadwell, C. (2002) ‘Modeling of material removal rate in rotary ultrasonic machining: designed experiments’, J. Mater. Process. Technol., October, Vol. 129, Nos. 1–3, pp.339–344.

Jiao, Y., Liu, W.J., Pei, Z.J., Xin, X.J. and Treadwell, C. (2005) ‘Study on edge chipping in rotary ultrasonic machining of ceramics: an integration of designed experiments and finite element method analysis’, J. Manuf. Sci. Eng., Vol. 127, No. 4, p.752.

Kataria, R., Kumar, J. and Pabla, B.S. (2015) ‘Experimental investigation into the hole quality in ultrasonic machining of WC-Co composite’, Mater. Manuf. Process., Vol. 30, No. 7, pp.921–933.

Khidhir, B.A. and Mohamed, B. (2010) ‘Machining of nickel based alloys using different cemented carbide tools’, J. Eng. Sci. Technol., Vol. 5, No. 3, pp.264–271.

Kumar, V., Kumar, K., Vikas, J. and Neeraj, K. (2016) ‘WEDM of nickel based aerospace alloy: optimization of process parameters and modelling’, Int. J. Interact. Des. Manuf.

Lauwers, B., Bleicher, F. and Haaf, P. (2010) ‘Investigation of the process-material interaction in ultrasonic assisted grinding of ZrO 2 based ceramic materials’, Proc. 4th CIRP Int. Conf. High Perform. Cut.


Li, Z.C., Jiao, Y., Deines, T.W., Pei, Z.J. and Treadwell, C. (2005) ‘Rotary ultrasonic machining of ceramic matrix composites: feasibility study and designed experiments’, Int. J. Mach. Tools Manuf., October, Vol. 45, Nos. 12–13, pp.1402–1411.

Lorenzen, T. and Anderson, V. (Eds.) (1993) Design of Experiments (Statistics: a Series of Textbooks and Monographs), CRC Press.

Lv, D., Tang, Y., Wang, H. and Huang, Y. (2013) ‘Experimental investigations on subsurface damage in rotary ultrasonic machining of glass BK7’, Mach. Sci. Technol., Vol. 17, No. 3, pp.443–463.

Montgomery, D.C. (2008) Design and Analysis of Experiments, John Wiley & Sons, Inc., Arozona State University.

Nalbant, M., Altın, A. and Gokkaya, H. (2007) ‘The effect of cutting speed and cutting tool geometry on machinability properties of nickel-base Inconel 718 super alloys’, Mater. Des., Vol. 28, No. 4, pp.1334–1338.

Pei, Z.J., Ferreira, P.M., Kapoor, S.G. and Haselkorn, M. (1995) ‘Rotary ultrasonic machining for face milling of ceramics’, Int. J. Mach. Tools Manuf., Vol. 35, No. 7, pp.1033–1046.

Pi, J., Yang, G. and Ye, F. (2013) ‘Study on impact stress characteristics in ultrasonic cutting and brittle material removal mechanism’, Adv. Mater. Res., Vols. 753–755, pp.287–297.

Singh, R.P. and Singhal, S. (2016a) ‘Experimental investigation of machining characteristics in rotary ultrasonic machining of quartz ceramic’, Journal Mater. Des. Appl., pp.1–20.

Singh, R.P. and Singhal, S. (2016b) ‘Experimental study on rotary ultrasonic machining of alumina ceramic: microstructure analysis and multi-response optimization’, Proc. Inst. Mech. Eng. Part L J. Mater. Des. Appl., pp.1–20.

Singh, R.P. and Singhal, S. (2016c) ‘Investigation of machining characteristics in rotary ultrasonic machining of alumina ceramic’, Mater. Manuf. Process., May, Vol. 32, No. 3, pp.309–326.

Torres, A., Luis, C.J. and Puertas, I. (2015) ‘Analysis of the influence of EDM parameters on surface finish, material removal rate, and electrode wear of an INCONEL 600 alloy’, Int. J. Adv. Manuf. Technol., March, Vol. 80, Nos. 1–4, pp.123–140.

Zhang, C.L., Feng, P.F., Pei, Z.J. and Cong, W.L. (2013) ‘Rotary ultrasonic machining of sapphire: feasibility study and designed experiments’, Key Eng. Mater., Vols. 589–590, pp.523–528.

Zhang, C.L., Feng, P.F., Wu, Z.J. and Yu, D.W. (2011) ‘An experimental study on processing performance of rotary ultrasonic drilling of K9 glass’, Adv. Mater. Res., Vols. 230–232, pp.221–225.

Zhou, J., Bushlya, V., Avdovic, P. and Sthl, J.E. (2012) ‘Study of surface quality in high speed turning of Inconel 718 with uncoated and coated CBN tools’, Int. J. Adv. Manuf. Technol., Vol. 58, Nos. 1–4, pp.141–151.

Zou, X. et al. (2013) ‘Cutting temperature in rotary ultrasonic machining of titanium: experimental study using novel Fabry-Perot fibre optic sensors’, Int. J. Manuf. Res., Vol. 8, No. 3, pp.250–261.

Documents

Investigation of machining rate and roughness for rotary ...pdfs.semanticscholar.org/76ca/dbd77c0960c4e26e29b4a1050c9681f0c411.pdfInvestigation of machining rate and roughness for