Strojniski vestnik - Journal of Mechanical Engineering 56(2010)6, 391-401 UDC 669.14:621.7.015: 621.9.02 Paper received: 01.04.2009 Paper accepted: 29.04.2010 The Optimization of Machining Parameters Using the Taguchi Method for Surface Roughness of AISI 8660 Hardened Alloy Steel Ali Riza Motorcu Canakkale Vocational College, Canakkale Onsekiz Mart University, Turkey The surface roughness in the turning of AISI 8660 hardened alloy steels by ceramic based cutting tools was investigated in terms of main cutting parameters such as cutting speed, feed rate, depth of cut in addition to tool's nose radius, using a statistical approach. Machining tests were carried out with PVD-coated ceramic cutting tools under different conditions. An orthogonal design, signal-to-noise ratio and analysis of variance were employed to find out the effective cutting parameters and nose radius on the surface roughness. The obtained results indicate that the feed rate was found to be the dominant factor among controllable factors on the surface roughness, followed by depth of cut and tool's nose radius. However, the cutting speed showed an insignificant effect. Furthermore, the interaction of feed rate/depth of cut was found to be significant on the surface finish due to surface hardening of steel. Optimal testing parameters for surface roughness could be calculated. Moreover, the second order regression model also shows that the predicted values were very close to the experimental one for surface roughness. © 2010 Journal of Mechanical Engineering. All rights reserved. Keywords: ceramic cutting tools, surface roughness model, Taguchi method, hardened alloy steel 0 INTRODUCTION Hard turning is a process, in which materials in the hardened state (50 to 70 HRC) are machined with single point cutting tools. This has become possible with the availability of the new cutting tool materials (cubic boron nitride and ceramics). A large number of operations are required to produce the finished product and if some of the operations can be combined, or eliminated, or can be substituted by the new process, product cycle time can be reduced and productivity can be improved. The traditional method of machining the hardened materials includes rough turning, heat treatment followed by the grinding process. Hard turning eliminates a series of operations required to produce the component and thereby reducing the cycle time and hence resulting in productivity improvement [1] and [2]. The advantages of hard tuning are higher productivity, reduced set up times, surface finish closer to grinding and the ability to machine complex parts. Various work materials which can be machined by the hard turning process include high speed steels, die steels, bearing steels, alloy steels, case hardened steels, white cast iron and alloy cast iron. Rigid machine tools with adequate power, very hard and tough tool materials with appropriate tool geometry, tool holders with high stiffness and appropriate cutting conditions are some of the prerequisites for hard turning. This paper deals with the hard turning of hardened alloy steel (AISI 8660) with physical vapour deposition (PVD) coated ceramic tools. The ceramic cutting tools including aluminium oxide, mixed alumina, silicon nitride based tools, and coated alumina based ceramic tools were developed in the past few decades for cutting various hard materials. However, some of the new developments in cutting tools have not been successful in improving the machinability of hardened steels since a wide range of hardened steels has been used by the manufacturing industry [3] to [5]. For example, AISI H13 series (45 to 56 HRC) are used for forging, extrusion and die casting while AISI E52100 (62 to 64 HRC) steels are the most commonly used steel materials for rolling bearings. The hardened workpiece materials are either cut near-net-shape in the annealed condition, heat-treated or ground to final dimensions and surface finish. Manufacturing costs could, therefore, be high and lead times excessive. *Corr. Author's Address: Canakkale Vocational College, COMU, 17100, Canakkale, Turkey, armotorcu@comu.edu.tr On the other hand, in metal cutting processes, the desired cutting parameters are determined either by experience or by using a handbook which does not ensure the selected parameters to be optimal. To determine the optimal cutting conditions, reliable mathematical models have to be formulated to associate the cutting parameters with cutting performance in terms of statistical approach. In literature, Response Surface Methodology (RSM) has been used by some researchers for the analysis and prediction of tool life or surface roughness [2] and [5] to [13]. Moreover, some works on machining of carbon or alloy steel have given to a full or fractional factorial design [14] to [16]. However, few of the existing researchers have applied a Taguchi approach to cross examine the impact of individual factors and factor interactions although the Taguchi method is relatively simple and can be used for optimizing different production stages with few experimental runs [17] to [26]. The aim of the present study is, therefore, to investigate the surface roughness in turning of the AISI 8860 steel (50 HRC) with the aid of a Taguchi design of experiment, using PVD-coated ceramic cutting tools under various cutting conditions. In addition, an analysis of variance is employed to find out effective cutting parameters on surface finish. 1 EXPERIMENTAL PROCEDURE The machine used for the turning tests was a Johnford TC35 Industrial type of computer numeric control (CNC) lathe machine. The lathe equipped with variable spindle speed from 50 to 3500 rpm, and a 10 KW motor drive was used for the tests. The insert was coated using a PVD method. The coating substance took place on the mixed ceramic substrate and PVD-TiN coated mixed ceramic with a matrix of Al2O3 (70%):TiC (30%) +TiN, which is called KY4400 grade [27]. The insert types were SNGA 120408 and SNGA 120412 (KY4400). The cutting tool's types used in the experiments are listed in Table 1. All tools are commercially available inserts according to the ISO code and the cutting tools were supplied by Kennametal Inc. for the machining tests. The material used throughout this work was an AISI 8660 steel. AISI 8660 is a high carbon, chromium-nickel-molybdenum alloy steel with high hardness and strength and is suitable for springs and axle shafts. Table 2 shows the chemical composition (wt. %) of AISI 8660 steel [28]. The work pieces were in the form of cylinders of 52 mm diameter and 220 mm length. The standard heat treatment process to specimens was applied under water condition and the average hardness measured was about 50 HRC. These bars are machined under dry condition. The work material bars were trued, centered and cleaned by removing a 0.3 mm depth of cut from the outside surface, prior to the actual machining tests. The surface roughness of the work piece was measured by a stylus instrument. The equipment used for measuring the surface roughness was a surface roughness tester, MAHR Perthometer-M1 type of portable. The surface roughness measures used in this paper is the arithmetic mean deviation of the surface roughness of profile, Ra. In collecting the surface roughness data of the shaft with the surface profilometer, three measurements were taken along the shaft axis for each sample with the measurements being about 120o apart. Table 1. The cutting tool's type used in the experiments Types of cutting tools Tool designation Chemical composition of coating materials Cutting fluids Coated ceramic tools (KY4400) SNGA 120408 Ti (C,N)+TiC+Al203+TiN Dry Coated ceramic tools (KY4400) SNGA 120412 Ti (C,N)+TiC+Al203+TiN Dry Table 2. Chemical composition of AISI 8660 steel [28] Chemical composition of AISI 8660 alloy (wt.%) C Mn Si S Cr Ni Mo Balance 0.61 0.93 0.28 0.030 0.51 0.52 0.36 96.74 2 METHODOLOGY 2.1 Application of the Taguchi Method One method presented in this study is an experimental design process called the Taguchi design method. Taguchi design is a set of methodologies by which the inherent variability of materials and manufacturing processes has been taken into account at the design stage. The application of this technique had become widespread in many US and European industries after the 1980s. The beauty of the Taguchi design is that multiple factors can be considered at once. Moreover, it seeks nominal design points that are insensitive to variations in production and user environments to improve the yield in manufacturing and the reliability in the performance of a product. Therefore, not only controlled factors can be considered, but noise factors as well. Although similar to the design of experiment (DoE), the Taguchi design only conducts the balanced (orthogonal) experimental combinations, which makes the Taguchi design even more effective than a fractional factorial design [24]. The philosophy of Taguchi is broadly applicable. He proposed that engineering optimization of a process or product should be carried out in a three-step approach, i.e., system design, parameter design, and tolerance design (Figure 1) [26]. In system design, the engineer applies scientific and engineering knowledge to produce a basic functional prototype design, which includes the product design stage and the process design stage. In the product design stage, the selection of materials, components or tentative product parameter values are included. As to the process design stage, the analysis of processing sequences, the selections of production equipment or tentative process parameter values are involved. Since system design is an initial functional design, it may be far from optimum in terms of quality and cost. The objective of the parameter design is to optimize the settings of the process parameter values for improving performance characteristics and to identify the product parameter values under the optimal process parameter values [26]. Parameter design System design Tolerance design Determine the results of parameter design by tightening the tolerance of the significant factors Fig. 1: Taguchi design procedure [24] In addition, it is expected that the optimal process parameter values obtained from the parameter design are insensitive to the variation of environmental conditions and other noise factors. Therefore, the parameter design is the key step in the Taguchi method in achieving high quality without increasing the costs [26]. The classical parameter design developed by Fisher is complex and not easy to use. In particular, a large number of experiments have to be carried out when the number of the process parameters increases. To solve this task, the Taguchi method uses a special design of orthogonal arrays to study the entire parameter space with a small number of experiments only. A loss function is then defined to calculate the deviation between the experimental value and the desired value. Taguchi recommends the use of the loss function to measure the performance characteristic deviating from the desired value. The value of the loss function is further transformed into a signal-to-noise (S/N) ratio j. Usually, there are three categories of the performance characteristic in the analysis of the S/N ratio, that is, the lower-the-better, the higher-the-better, and the nominal-the-better. The S/N ratio for each level of process parameters is computed based on the S/N analysis [26]. Regardless of the category of the performance characteristic, the larger S/N ratio corresponds to the better performance characteristic. Therefore, the optimal level of the process parameters is the level with the highest S/N ratio r . Furthermore, a statistical analysis of variance (ANOVA) is performed to see which process parameters are statistically significant. With S/N and ANOVA analyses, the optimal combination of the process parameters can be predicted [26]. Finally, a confirmation experiment is conducted to verify the optimal process parameters obtained from the parameter design. Nominal-is-the-best: S/Nt = 10• logj-^j . (1) Larger-is-the-better (maximize): S / nl ="10 • log [n | yr). (2) Smaller-is-the-better (minimize): S / ns =-10 • log 1 y2 j, (3) where y, is the average of observed data, sy2 is the variance of y, n is the number of observations and y is the observed data. Notice that these S/N ratios are expressed on a decibel scale. We would use S/NT if the objective is to reduce variability around a specific target, S/NL if the system is optimized when the response is as large as possible, and S/NS if the system is optimized when the response is as small as possible. Factor levels that maximize the appropriate S/N ratio are optimal. The goal of this research was to produce minimum surface roughness (Ra) in a turning operation. Smaller Ra values represent better or improved surface roughness. Therefore, a smaller-the-better quality characteristic was implemented and introduced in this study [26]. The Taguchi method, which is a powerful tool in the design of an experiment, is used to optimize the turning parameters for effective machining of AISI 8660 hardened alloy steel [23]. This method recommends the use of S/N ratio to measure the quality characteristics deviating from the desired values. To obtain optimal testing parameters, the-lower-the-better quality characteristic for machining the AISI 8660 steel was taken due to the measurement of the surface finish. The S/N ratio for each level of testing parameters was computed based on the S/N analysis. This design is sufficient to investigate the four main effects and the influence of their interactions on the surface roughness. With S/N ratio analysis, the optimal combination of the testing parameters could be determined. The control parameters were cutting speed (V), feed rate (f), depth of cut (d) and tool's nose radius (r, mm). Two levels were specified for each of the factors as indicated in Table 3. The orthogonal array chosen was L16, which has 16 rows corresponding to the number of parameter combinations (15 degrees of freedom), with 15 columns at two levels as shown in Table 4 [17]. The first column was assigned to the cutting speed (V), the second column to the feed rate (f), the fourth column to the depth of cut (d), the eighth column to the tool's nose radius (r) and the remaining columns to the interactions. Table 3. Assignment of the levels to the factors Control Factors Unit Levels Degrees of freedom Level 1 Level 2 (DoF) Cutting speed, V m min-1 150 210 1 Feed rate, f mm rev-1 0.11 0.24 1 Dept of cut, d mm 0.3 1.0 1 Tool's nose radius, r mm 0.8 1.2 1 Table 4. Orthogonal array L16 (215) of Taguchi [17] Test No Column No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 3 1 1 1 2 2 2 2 1 1 1 1 2 2 2 2 4 1 1 1 2 2 2 2 2 2 2 2 1 1 1 1 5 1 2 2 1 1 2 2 1 1 2 2 1 1 2 2 6 1 2 2 1 1 2 2 2 2 1 1 2 2 1 1 7 1 2 2 2 2 1 1 1 1 2 2 2 2 1 1 8 1 2 2 2 2 1 1 2 2 1 1 1 1 2 2 9 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 10 2 1 2 1 2 1 2 2 1 2 1 2 1 2 1 11 2 1 2 2 1 2 1 1 2 1 2 2 1 2 1 12 2 1 2 2 1 2 1 2 1 2 1 1 2 1 2 13 2 2 1 1 2 2 1 1 2 2 1 1 2 2 1 14 2 2 1 1 2 2 1 2 1 1 2 2 1 1 2 15 2 2 1 2 1 1 2 1 2 2 1 2 1 1 2 16 2 2 1 2 1 1 2 2 1 1 2 1 2 2 1 Table 4. Table 5 shows the sets of experiments of L16 (215) orthogonal array with experimental results of surface roughness height Ra along with their arithmetic average values and S/N ratio (dB) [23]. According to the Taguchi quality design concept an L16 (215) orthogonal array has been used to determine the S/N ratio (dB), ANOVA and 'F' test values for indicating the most significant parameters affecting the machining performance criteria, i.e. surface roughness, Ra. The main purpose of the ANOVA is to investigate the design parameters and to indicate which parameters significantly affect the quality characteristic. This analysis helps to find out the relative contribution of machining parameter in controlling the response of turning operation. The optimal parametric setting value will directly influence the objective function for determining the minimum cost at the optimal policy [23]. 2.2 Mathematical Models The experimental results would be used to build first-order and second-order models by the multiple regression method. The purpose of developing the mathematical models is to understand the combined effect of involved parameters and to facilitate the optimization of the machining process [29] and [30]. The following relationship is commonly used for representing the mathematical models: Y = O(V, f, d, r) + s, (4) where Y is the turning response, O is the response function, and ' V is the error which is normally distributed with zero mean according to the observed response. The relationship between surface roughness and other independent variables is modelled as follows: Table 5. Experimental results and S/N ratio of Ra Experiment number Column Coded level Actual setting values Test result of Ra [|im] Average Ra [|m] S/N ratio [dB] for Ra 1 2 4 8 1 1 1 1 1 150 0.11 0.3 0.8 1.377 1.460 1.441 1.4260 -3.0851 2 1 1 1 2 150 0.11 0.3 1.2 0.972 0.997 0.983 0.9840 0.1396 3 1 1 2 1 150 0.11 1.0 0.8 1.410 1.398 1.418 1.4087 -2.9763 4 1 1 2 2 150 0.11 1.0 1.2 1.144 1.207 1.200 1.1837 -1.4671 5 1 2 1 1 150 0.24 0.3 0.8 2.701 2.865 2.701 2.7557 -8.8080 6 1 2 1 2 150 0.24 0.3 1.2 1.923 1.891 1.817 1.8770 -5.4717 7 1 2 2 1 150 0.24 1.0 0.8 3.402 3.939 3.523 3.6213 -11.1949 8 1 2 2 2 150 0.24 1.0 1.2 3.457 3.805 3.489 3.5837 -11.0949 9 2 1 1 1 210 0.11 0.3 0.8 1.390 1.564 1.602 1.5187 -3.453 10 2 1 1 2 210 0.11 0.3 1.2 0.944 1.156 0.968 1.0227 -0.2318 11 2 1 2 1 210 0.11 1.0 0.8 1.301 1.389 1.404 1.3647 -2.7053 12 2 1 2 2 210 0.11 1.0 1.2 0.980 1.026 1.202 1.0693 -0.6169 13 2 2 1 1 210 0.24 0.3 0.8 2.590 2.764 2.684 2.6793 -8.5636 14 2 2 1 2 210 0.24 0.3 1.2 1.848 1.795 1.812 1.8183 -5.1941 15 2 2 2 1 210 0.24 1.0 0.8 4.349 4.060 4.708 4.3723 -12.8302 16 2 2 2 2 210 0.24 1.0 1.2 3.157 3.349 3.745 3.4170 -10.6951 Ra = C-V" ■ fm ■ dp ■ rs, (5) where C is a constant and n, m, p and s are the exponents. The above function can be represented in linear mathematical form as follows: lnR = lnC + n.lnV+m.ln f + p.lnd + a J F . (6) +s.ln r + lns The constants and exponents C, n, m, p and s can be determined by the method of least squares. The first-order linear model, developed from the equation, can be represented as follows: Y = y — s = b0 x0 + bl xl + b2 x2 + b3 x3 + b4 x4, (7) where Yi is the estimated response based on first order equation, and y is the measured surface roughness on a logarithmic scale, x0 = i (dummy variable), xi, x2, x3 and x4 are logarithmic transformations of cutting speed, feed rate, depth of cut and nose radius respectively, s is the experimental error and 'b' values are the estimates of corresponding parameters. If this model is not sufficient to represent the process, then the second order model will be developed [29] and [30]. The general second-order model is as follows: Y2 = Y — s = b0 x0 + b x + b2 x2 + b3 x3 + b4 x4 + +b12 x,x2 + b23 x2 x + b14 x,x4 + b24 x2 x4 + (8) +b13 xi x3 + b34 x3 x4 , where Y2 is the estimated response based on second order equation. The parameters, i.e. b0, bi, b2, b3, b4, bi2, b23, bi4 are to be estimated by the method of least squares [29] and [30]. 3 RESULTS AND DISCUSSION 3.1 The Taguchi Method Evaluation Results The experimental results for Ra illustrated in Table 4 are analysed with the Minitab 15.0 software. This table shows an experimental layout with calculated S/N ratios for cutting tests of the AISI 8660 steel. An analysis of the influence of each control factor on the surface finish was performed with mean response table. The response table of the cutting process is presented in Table 6. The influence of interaction between control factors was also analysed in the response table. The control factor with the strongest influence was determined by differences values. The higher the difference was, the more influential the control factor or an interaction of two controls was. The strongest influence was found by f and d, respectively. The analysis of interactions gives additional information about the nature of the process under consideration. This table showed the analysis of all interactions obtained by calculating all four combinations of interaction of two control factors. The main effects and their interaction plots of the surface roughness of the alloyed steels for S/N ratios are shown in Fig. 2. Optimal testing conditions of these control factors can be easily determined from this graph. A response graph showed the change of the S/N ratio when the setting of the control factor was changed from one level to the other. The best surface finish value was at the higher S/N values in the response graphs. It could be seen in Fig.2 that the optimum testing conditions for the tested samples became V1-f1-d1-r2 for main control factors in this study. Based on the S/N ratio, the optimal testing parameters for surface roughness were cutting speed at levell, feed rate at level 1, depth of cut at level 1, and the tool's nose radius at level 2 (Table 7). The analysis of variance (ANOVA) was used to investigate which design parameters significantly affect the surface quality. Examination of the calculated values of variance ratio (F), which is the variance of the factor divided by the error variance for all control factors showed a much higher influence of factor f, d, and a high influence of factor r on the surface roughness of the alloy steel (Table 8). The F value of each design parameters was calculated. The change of the cutting speed in the range given Table 8 had no significant effect on the roughness of machining the steel because of the lower F-value. In addition, the interaction between the f x d showed a much higher influence while the interactions of V5f, V5d, V5r, f5r and d5r showed no significant influence on the surface roughness of tested materials. The F table 0.05 (1, 5) equals to 6.60 at 95% confidence level. Thus, based on the level of confidence (95 %), the main factors of f, d, r and interaction of f5d were significant while other factors such as V and Vf V5d, Vr and fr were pooled as well. Table 6. Response table for means of surface roughness in machining the steel Factors and their interactions V f d r Vf Level 1 2.10491 1.24709 1.76013 2.39322 2.16109 Level 2 2.15769 3.01550 2.50247 1.86938 2.10150 Difference 0.05278 1.76841 0.74234 0.52384 0.05959 Rank 10 1 2 4 8 Factors and their interactions V5d V5r fd fr d5r Level 1 2.15819 2.06728 2.49312 2.05166 2.20406 Level 2 2.10441 2.19531 1.76947 2.21094 2.05853 Difference 0.05378 0.12803 0.72366 2.21094 0.14553 Rank 9 7 3 5 6 3.1-2.7i 2.3 1.91.5- V Vxf Vxd fxd Vxr fxr dxr u Ö ■a 8