1 Introduction

Machining processes are one of the important manufacturing activities contributing to the growth of global economy [1]. Inconel 718 (nickel-based super alloy) is having diverse applications in marine, aerospace, petroleum, food processing industries, etc. due to its properties such as high strength, high hot hardness, high impact strength, and corrosion resistance [2,3,4]. These properties along with work hardening tendency, poor thermal conductivity (11.2 W/mK) [5, 6] and tendency to form a built up edge pose difficulties in machining. Machining of Inconel is associated with high value of cutting forces resulting in higher vibrations and poor surface finish of the work piece. Also, low thermal conductivity of the material causes more heat transfer to the tool resulting in rise of tool temperature [7] and accelerated wear of the tool. The selection of the cutting parameters and cutting fluid is an inseparable part in any machining process [8] and it is much more important in machining of difficult-to-cut material such as Inconel 718.

Surface roughness investigation plays an important role in controlling friction, wear and fatigue of the machined parts [9] and thus it is an important response to be measured. Surface finish in machining depends upon the machine tool, cutting process, cutting parameters, cutting fluid, tool and work piece material [10]. Turning is fundamental manufacturing operation and it is imperative that the selection of appropriate cutting tool, cutting fluid and the method of application of cutting fluid plays a crucial role in turning. The investigation of cutting forces is critical as it affects the surface finish of the work piece and wear of the tool.

Selection of cutting fluid and the method of its application is one of the important parameters that affect the quality of work part. Selecting right kind of cutting fluid ensures improved quality of the surface by avoiding the overheating of the cutting zone, arresting the heat transfer to adjacent machining zone and conveying the chips away from the machining area [11]. Water-based flood cooling is a cheap, popular and prevalent method in machining. However, major disadvantage of water-based fluid is that it can propagate the growth of microorganisms in all spaces of machine tool. Antibacterial agents which are added to fluids to reduce the growth of microorganisms get in contact with high temperature machined surface and liberate gases that are harmful to the machinist’s health. Second, the repeated use of the cutting fluid alters its chemical and physical properties which may be detrimental to machining process [11, 12]. The usage of water-based fluid is also marred by low viscosity index, inadequate thermal stability and cooling capacity, thereby rendering them ineffective for the machining of Inconel.

Biodegradable vegetable oils are popular choice over the water-based fluids as they retain their properties at elevated temperature. Davoodi and Tazehkandi [13] used biodegradable vegetable oil as a cutting fluid in turning of Inconel 738. They developed empirical models using analysis of variance (ANOVA) and optimized the cutting parameters using response surface methodology (RSM). However, these cutting fluids are still not popular on account of their high cost. The quantity of the vegetable oil used for the machining can be reduced using it with minimum quantity lubrication (MQL) approach [11]. In view of all the issues discussed above, it is very important to formulate a new approach which is sustainable and improves the machining performance [14]. The use of MQL, cryogenic cooling, cryogenic treatment (cryo treatment) of tools, tool coatings, plasma heating of work piece, etc. are some of the techniques used for improving the machining of nickel alloys. Some works reporting use of these techniques are discussed in the following paragraphs.

One of the popular cryogenic techniques is use of cryogen such as liquid Nitrogen for cooling. Pusavec et al. [1, 15] have developed performance-based models for estimating tool wear, cutting forces and surface roughness using ANOVA for machining of Inconel 718. They evaluated the combined effect of cryogenic cooling with MQL and optimized the machining parameters. It was found that cryogenic cooling enhanced the quality of the machined surface compared to other conditions. Kaynak [16] had also used cryogenic cooling for turning of Inconel 718 and compared the results with those of dry and MQL conditions. MQL conditions demonstrated significant reduction in cutting forces compared to dry and cryogenic cooling at low speeds of about 60 m/min.

Cryogenic treatment of tools is also used for improving machining. It alters the metallurgical properties of tools material [17, 18] that result in improved mechanical properties. Gill et al. [19] have used cryogenically treated (cryo-treated) tungsten carbide inserts in turning of steels. They reported better wear resistance, tool life, cutting forces and surface roughness while using cryo-treated inserts compared to untreated inserts. Gill et al. [20] also reported better performance of cryo-treated tool in wet condition compared to dry machining. This is quite natural as cutting fluid provides cooling and lubrication resulting in reduced cutting forces, temperature, tool wear and surface roughness. Thakur et al. [6] have proposed a regression equation for tool life and degree of work hardening for machining of Inconel 718 using cryo-treated tools. They observed minimum work hardening, tool wear and cutting forces while using cryo-treated tool. Deshpande et al. [21] have established regression models using RSM for estimating surface roughness with untreated and cryogenically treated inserts in turning of Inconel 718.

Another method which is attempted by some researchers, to improve machining, is hybrid machining. In this method, work piece is softened by laser or plasma heating and at the same time cryogenic cooling by liquid nitrogen (LN2) is also used at the chip tool interface to carry away the heat produced in machining region. Researchers have reported reduction in cutting forces, tool wear and surface roughness in hybrid machining of Inconel 718 compared to conventional machining. As reported by Feyzi and Safavi [22], there was no change in the hardness of the machined surface in hybrid machining. Few researchers have attempted dry machining of nickel-based alloys using integrated optimization techniques discussed in the following paragraphs.

Pawade and Joshi [23] had performed dry turning operation on Inconel 718 using CBN insert. The optimum cutting parameters were found out using Taguchi grey relation analysis for minimizing cutting forces and surface roughness. Homami et al. [24] performed dry turning of Inconel 718 with coated insert. Cutting speed, feed rate, nose radius and approach angle were selected as cutting parameters. ANOVA and genetic algorithm approach were used to find the optimum cutting condition for achieving minimum surface roughness and flank wear. For achieving good surface quality of Inconel 825 (nickel-based alloy) in minimum time, Tamang and Chandrasekaran [25] combined ANN model with particle swarm optimization (PSO) and optimized machining parameters in dry turning of Inconel 825 using polycrystalline diamond (PCD) inserts.

From the various works discussed above, it is seen that dry machining with uncoated tool is not suitable for Inconel 718. The absence of coolant and lower thermal conductivity of Inconel 718 results in increased heat transfer to the tool, which results in rapid tool wear. The excessive friction and higher contact between the tools and chip edge result in formation of the dull tool edge which further increases the cutting force. This is a vicious circle as blunt cutting edge increases the cutting force which in turn increases the wear. Cryogenic cooling can have adverse effect on cutting force and surface finish due to excess cooling [26, 27]. Also, the hybrid machining of nickel alloys is not economical. It is also found from literature that MQL provide better results at low cutting speed of about 60 m/min as MQL helps in heat dissipation. However, at high speed, MQL does not give appreciable results in the machining for Inconel 718 [16]. The cryo-treated tool can bear high temperature without failing by plastic deformation [18]. The literature review reveals that use of cryogenic techniques enhance the machinability of various materials [19, 20, 28,29,30]. Therefore, technique of cryogenic tool treatment can also be tried for material such as Inconel 718.

From the various reported works, it is observed that there is hardly any work on machining of nickel-based super alloy using cryogenic treatment of tools and MQL approach. Thakur et al. [6] have reported the use of cryo-treated inserts for machining of Inconel 718 and this is limited to dry condition only. It is also seen that the cryogenic cycle to be used for tool treatment and the appropriate parameters required for machining of nickel alloys are required to be examined. Also, use of modelling techniques in such investigations would be beneficial for reducing the costly and time-consuming experimentation [31]. Further, using MQL with cryo-treated inserts is cheaper than using cryogenic cooling with liquid nitrogen.

In this context, the present work deals with the novel combination of use of cryogenically treated tools with MQL condition in machining of Inconel 718. The effect of cutting parameters on responses such as cutting force, tool vibration, cutting temperature, surface roughness and tool wear is determined using different machining conditions to optimize the cutting parameters for improving the machining of Inconel 718.

2 Experimentation

2.1 Work piece and tool material

In this experimental study, bars of diameter of 22 mm and length of 150 mm of Inconel 718 having hardness 46 HRC were selected for machining. Figure 1 shows the microstructure and EDX plot with chemical composition of Inconel 718, developed by scanning electron microscope (SEM). The microstructure consists of gamma (γ), gamma prime (γ′) phase, carbides and metal precipitates. Gamma phase has a continuous matrix in a face-centered cubic structure. Nickel-based phase (γ phase) consists of a high percentage of disordered elements in solid-solution which is soft and ductile in nature. Gamma prime is the main strengthening and a coherently precipitating phase in Inconel 718 having a regular crystal structure, has long time stability and is brittle in nature except at very high temperature. Along with these phases, there are various inter-metallic phases of carbide and other metals. The microstructure of the nickel is strengthened by these phases [32].

Fig. 1
figure 1

Inconel 718 a Microstructure and b EDX plot

Full size image

Uncoated tungsten carbide inserts (Kennametal make) having ISO designation TNMG 160408 with double chip breaker geometry and nose radius of 0.8 mm were utilized for experiments. Each insert was mounted on left hand tool holder with ISO designation MTJNL2525-M16 and new cutting edge was used for every experiment. The selection of the tools, cryo treatment cycle and cutting conditions were based on the tool manufacturer’s manual and references from the literature [1, 6, 33]. The details pertaining to the cutting conditions are as follows: cutting speed = 9.5–110.45 m/min, feed rate = 0.06–0.23 mm/rev and depth of cut = 0.32–1.17 mm at constant cutting length of 60 mm.

2.2 Machining conditions in turning of Inconel 718

Following are the conditions used for machining of Inconel 718 alloy.

  1. 1.

    Dry cutting with untreated carbide inserts (DRY).

  2. 2.

    Dry cutting with cryogenically treated inserts (DRY-CT).

  3. 3.

    Minimum quantity lubrication with untreated inserts (MQL).

  4. 4.

    Minimum quantity lubrication with cryogenically treated inserts (MQL-CT).

For conditions 3 and 4, cutting fluid used is BioCut 3600. The pure cutting oil is sprayed directly at cutting region with the use of minimum quantity lubrication (MQL), and single nozzle spraying system (Make: DROPCO). The coolant pressure is kept at 4 bars with flow rate of 80 mlph. Uncoated tungsten carbide inserts were treated using deep cryogenic cycle as discussed in the next section.

2.3 Cryogenic treatment for carbide inserts

From the various cryo treatment cycles used by different researchers, it is seen that treatment temperatures vary significantly from − 70 to − 196 °C. Also, the overall cycle time of treatment has lot of variations. The increase in the tool life, wear resistance and improved surface quality was recorded using cryo-treated inserts in different operating conditions [20, 34,35,36]. It is also revealed that use of deep cryogenic treatment cycle (− 196 °C) can be beneficial for better wear resistance of carbide inserts and further use of tempering cycle (+196 °C) removes the stresses developed in the cryogenic cycle. Hence, cooling rate of cryo treatment cycle was employed as 0.5 °C/min to prevent thermal shocks from swift cooling resulting in possible damage to carbide inserts. [19, 20, 28,29,30].

The schematic of the cryogenic system and details of cryogenic double tempering cycle used for treatment of tool inserts are shown in Fig. 2. In the cooling cycle, temperature was controlled with the computerized temperature controller at the rate of 0.5 °C/min which required 8 h to obtain − 196 °C (A–B). In the soaking cycle, deep cryogenic temperature (DCT) was maintained for 24 h at the coolest temperature (− 196 °C) from B to C. Then, it was allowed to achieve the room temperature in 8 h which is known as the warming temperature cycle (C–D). The entire duration of the cryogenic cycle was about 40 h (A–D).

Fig. 2
figure 2

Schematic of the cryogenic system (left) and deep cryogenic treatment cycle (right)

Full size image

Once the cryogenic cycle was completed, the material was tempered. In tempering cycle, it was heated to +196 °C in 1 h (D–E). In holding cycle, material was maintained at the high temperature for 2 h (E–F) and then allowed to cool to room temperature in the subsequent cooling cycle (F–G). The entire length of tempering cycle (D–G) is 4 h and it was repeated (G–J) for releasing the stresses developed in the cryogenic cycle. The total duration of the cryogenic double tempering cycle is 48 h (A–J). For the present work, the cryo treatment of inserts was performed at KRYO SPACE, India.

Figures 3 and 4 show the microstructure with EDX report of the untreated and treated tungsten carbide tools, obtained by SEM. The microstructure mainly comprises of the three distinct phases, alpha (α), beta (β) and eta (η), and it is similar to that reported by researchers—Gallagher et al., Gill et al., Reddy et al. and SreeramaReddy et al. [17, 18, 28, 37]. The effects of cryo treatment on microstructure of inserts are compared using EDX report and Vickers hardness values.

Fig. 3
figure 3

Untreated carbide insert a Microstructure, b EDX plot

Full size image
Fig. 4
figure 4

Deep cryogenic treated carbide insert a Microstructure, b EDX plot

Full size image

For untreated inserts, rough discontinuous transfer film is present in matrix (α phase) with 89.52% of tungsten and carbide along with carbon atom (W + C). White spots (β phase) are present in the matrix with 2.46% of cobalt binder (Co) as shown in Fig. 3. After cryogenic treatment, the α phase is evenly scattered in the β phase and shows smooth and homogenous distribution as in Fig. 4a. Here, the percentage of W + C is 96.26; thus, there is 7.52% improvement compared to untreated insert as seen through EDX plot shown in Fig. 4b. Cryo treatment also reduces the cobalt binder (Co) by 25.2%. The improvement in α and β phases help to improve the overall thermal conductivity and wear resistance thus enabling better heat dissipation compared to untreated tool. Enhanced heat transfer reduces the issue of the heat build up on the tool surface and results in reduced wear. Hence, cryo-treated tool can survive at high temperature without deteriorating by plastic deformation [18].

The η phase consists of the carbides of tungsten accompanied by at least one metal binder. This phase may appear as grey spots and scatter throughout the entire bulk of the material. The change of η phase carbides after cryo treatment was reported by Gill et al. [18]. However, in the present investigation, this phase is not distinguished clearly through SEM report. Additionally, Vickers hardness (30 kgf hardness) test is performed to confirm any improvement with these phases. Hardness measurement tests are repeated for several points. The measured values are found close to each other. Hence, the mean value of hardness in Vickers Hardness Number (VHN) is reported. These values are obtained as 1474 VHN and 1576 VHN for untreated and cryo-treated inserts, respectively. The standard deviation is evaluated from mean values and these values are found as 14.84 and 18.23 for untreated and cryo-treated inserts, respectively. The hardness of carbide insert was increased by 7% after the specimen was subjected to cryogenic double tempering cycle.

2.4 Experiment setup

Turning was carried out on MTAB CNC lathe machine (Maxturn Plus+) of 5.5 kW equipped with Siemen’s Sinumerik 828D control. The experimental setup with measuring instruments related to turning of Inconel 718 is shown in Fig. 5. The cutting forces were measured using piezoelectric dynamometer (Make: Kistler, type: 9257B). The tool vibrations in transverse direction were measured with Industrial ICP® single-point accelerometer (Make: IMI sensors, model: 603C11). The output signals of dynamometer and accelerometer were acquired with National Instrument’s DAQ 9178 and stored in computer using Lab VIEW 2012. After completion of all the experiments, the surface roughness values of machined parts were measured with portable surface roughness tester (Make: Mitutoyo, SURFTEST SJ-410). Surface roughness (R a ) measurements were repeated at several points on circumference of the work piece. The values were very close to each other and their average R a is considered for further study. The tool flank wear was measured with the help of USB microscope (UM5 CAM). The values of flank wear at nose region (VBmax) were then noted for further analysis. The maximum temperature generated at tool-chip interface while machining was measured using a digital thermometer having range up to 1850 °C (MEXTECH IR 1800 with thermocouple). For getting accurate temperature, K-type thermocouple probe with range up to 1370 °C is attached at a distance of 0.5 mm from the cutting edge as recommended by Tazehkandi et al. [11].

Fig. 5
figure 5

Experimental setup (left) and measuring instrument in turning of Inconel 718 (right)

Full size image

2.5 Experiment design and results

The focus of the experiments was to examine the effect of cryogenic treatment of tool and minimum quantity lubrication in turning of Inconel 718, on various responses such as cutting force, tool vibration, cutting temperature, surface roughness and tool flank wear. RSM is used for the modelling and optimization of the machining process. The experiments were designed using the central composite design (CCD) of RSM [21].

The design consists of a factorial design or corner portion (2k) of cube, centre points and axial or star points. The corner portions of cube show eight experiments with 23 factorial design levels coded by − 1 and +1, situated at the vertices of central composite design (CCD). Repetitive six experiments show an added centre point to the CCD which helps to confirm the repeatability of measuring process. The additional axial points basically raise the number of levels to five where distance of the axial point from the design centre is \(\alpha = 2^{k/4}\). CCD including axial points with the designed value α is called circumscribed central composite design placed on the coordinate axis of the factorial segment at a supplementary distance of α = 1.682 from the centre. Due to rotatable design of central composite design, it is observed that, when α > 1, the run for every independent variable is at five levels and are denoted by − 1.682, − 1, 0, + 1, + 1.682 as shown in Fig. 6. Thus, a great deal of information is accumulated about the effects of the factors and their interactions.

Fig. 6
figure 6

Central composite design for factors

Full size image

Experimental design uses variation of three input parameters—cutting speed (v), feed rate (f) and depth of cut (d) whose five levels are coded. The coded and normal levels of the input variables are tabulated in Table 1. The outcome of design of experiment (DOE) shows 20 experiments along with six duplication of centre point according to the CCD, using RSM in MINITAB 17 software. The design matrix of input and measured values are presented in Table 2. The measured values of cutting force (Fc), tool vibration (As), cutting temperature (T), surface roughness (R a ), and tool flank wear (VBmax) for the four conditions used, i.e. DRY, DRY-CT, MQL and MQL-CT are listed in Table 2.

Table 1 Level of cutting parameters for the turning of Inconel 718
Full size table
Table 2 Design matrix and experimental result (DRY/DRY-CT/MQL/MQL-CT)
Full size table

3 RSM modelling

Response surface method determines relationship among independent variables and dependent variables [38, 39]. In RSM, the second-order model is generally used due to its simplicity, flexibility and ability to estimate the correct response surface parameters. The values of output variables are measured experimentally from the machining process and input–output relationship is determined using RSM methodology in MINITAB 17 software. The empirical relationships (prediction models) establish correlation between output parameters (Fc, As, T, R a and VBmax) and cutting parameters (v, f, d) in the turning of Inconel 718. By statistical analysis, the adequacy of developed models, for four machining conditions, was verified using analysis of variance (ANOVA) to determine the significant and non-significant parameters on the basis of p (probability) values, F values and correlation coefficient (R2). These are shown in Table 3a–d. The p value less than 0.05 specifies that model variables are significant and p values greater than 0.05 show the insignificant variables of model for confidence limit of 95% [38]. The larger values of F and smaller values of p (p < 0.05) show that models are significant. The R2 nearer to unity indicates best correlation between predicted data of model and experimental data.

Table 3 Summarized results of ANOVA for second-order polynomial prediction model
Full size table

3.1 Refinement of developed model using backward elimination method

To remove insignificant effects from developed models, the backward regression elimination method is used to find a best fit model. In this method, a test for significance is performed on the variables. The largest p value of the variable is eliminated based on value specified in predefined level. The modified (refined) models are obtained with significant variables and then used for further regression analysis. The modified models suitable for estimating Fc, As, T, R a and VBmax for all machining conditions on basis of fine R2 and R2 adj values are presented in Eqs. (1)–(20).

Refined models for dry turning with untreated inserts:

$$F_{\text{c}} = 783.8 - 4.822v + 604f \, + 146.9d \, \left( {R^{2} = 92.35} \right) \left( {R^{2} {\text{ adj}} = 90.92} \right)$$
(1)
$$A_{s} = 2.14 - 0.0295v + 16.4f - 0.80d + 0.000270v^{2} + 115.4f^{2} + 4.86d^{2} - 0.2492vf - 35.3fd\,\,\,\,\,\,\,\left( {R^{2} = 94.43} \right) \, \left( {R^{2} {\text{ adj}} = 90.38} \right)$$
(2)
$$T = 681.9 + 1.139v + 1019f + 100.2d - 0.00574v^{2} - 2296f^{2} + 5.25vf + 0.95vd - 670fd\,\,\,\,\,\,\left( {R^{2} = 98.64} \right) \left( {R^{2} {\text{ adj}} = 97.64} \right)$$
(3)
$$R_{a} = 6.466 - 0.04145v - 25.25f - 4.93d + 0.000211v^{2} + 69.5f^{2} + 2.520d^{2} + 11.80fd\,\,\,\,\,\,\,\left( {R^{2} = 93.08} \right) \left( {R^{2} {\text{ adj}} = 89.05} \right)$$
(4)
$${\text {VB}}_{\hbox{max} } = 0.129 + 0.00973v + 5.40f + 0.1708d - 0.000046v^{2 } - 15.25f^{2} \,\,\,\,\,\,\left( {R^{2} = 90.34} \right) \left( {R^{2} {\text{ adj}} = 86.84} \right)$$
(5)

Refined models for dry turning with cryo-treated inserts:

$$F_{\text{c}} = 1040.3 - 8.85v - 155f - 92.2d + 12.75vf + 2.95vd\,\,\,\,\,\,\,\,\,\,\left( {R^{2} = 97.12} \right) \left( {R^{2} {\text{ adj}} = 96.77} \right)$$
(6)
$$A_{\text{s}} = 6.458 - 0.0673v + 7.39f - 7.06d + 0.000282v^{2} + 5.27d^{2} \,\,\,\,\,\left( {R^{2} = 91.46} \right) \left( {R^{2} {\text{ adj}} = 88.41} \right)$$
(7)
$$T = 496.7 - 0.15v + 490f + 170.9d + 0.02770v^{2} \,\,\,\,\,\,\,\,\,\,\,\left( {R^{2} = 91.01} \right) \left( {R^{2} {\text{ adj}} = 88.62} \right)$$
(8)
$$R_{a} = 3.057 - 0.01104v - 11.93f - 2.31d + 20.40fd\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\left( {R^{2} = 73.18} \right) \left( {R^{2 \, } {\text{adj}} = 66.03} \right)$$
(9)
$${\text{VB}}_{\hbox{max} } = - 0.277 + 0.00979v + 6.28f + 0.906d - 0.000051v^{2} - 17.69f^{2} - 0.479d^{2} \,\,\,\,\,\,\,\,\,\,\,\,\,\left( {R^{2} = 92.19} \right) \left( {R^{2} {\text{ adj}} = 88.58} \right)$$
(10)

Refined models for turning with MQL using untreated inserts:

$$\,F_{\text{c}} = 759.4 - 4.637v + 562f + 76.8d\,\,\,\,\,\,\,\,\,\,\,\,\,\,\left( {R^{2} = 94.34} \right) \left( {R^{2} {\text{ adj}} = 93.34} \right)$$
(11)
$$A_{\text{s}} = 6.28 - 0.02717v - 18.1f - 6.35d + 87.1f^{2} + 4.86d^{2\,} \,\,\,\left( {R^{2} = 87.18} \right) \left( {R^{2} {\text{ adj}} = 82.60} \right)$$
(12)
$$T = 409.3 + 3.993v + 1138f + 75.4d - 0.01661v^{2} - 3208f^{2} \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\left( {R^{2} = 95.22} \right) \left( {R^{2} {\text{ adj}} = 93.51} \right)$$
(13)
$$R_{a} = 5.694 - 0.04103v - 25.16f - 4.93d + 0.000209v^{2} + 69.3f^{2} + 2.494d^{2} + 11.90fd\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\left( {R^{2} = 93.32} \right) \left( {R^{2} {\text{ adj}} = 89.43} \right)$$
(14)
$${\text {VB}}_{\hbox{max} } = -\,0.090 + 0.00896v + 5.27f + 0.1903d - 0.000041v^{2} - 14.02f^{2} \,\,\,\,\,\,\,\,\,\,\left( {R^{2} = 87.25} \right) \left( {R^{2} {\text{ adj}} = 82.70} \right)$$
(15)

Refined models for turning with MQL using cryo-treated inserts:

$$F_{\text{c}} = 937.6 - 8.79v - 160f - 87.9d + 12.67vf + 2.90vd\,\,\,\,\,\,\left( {R^{2} = 97.53} \right) \left( {R^{2} {\text{ adj}} = 96.64} \right)$$
(16)
$$A_{\text{s}} = 6.74 - 0.0713v - 21.7f - 4.92d + 0.000362v2 + 101.5f^{2} + 4.00d^{2} \,\,\,\,\,\,\,\,\,\,\,\,\left( {R^{2} = 90.43} \right) \left( {R^{2 \, } {\text{adj}} = 86.01} \right)$$
(17)
$$T = 386.9 + 2.285v + 416f + 69.4d\,\,\,\,\,\,\,\,\,\,\,\,\left( {R^{2} = 79.13} \right) \left( {R^{2} {\text{ adj}} = 75.21} \right)$$
(18)
$$R_{a} = 0.8465 - 0.01125v + 1.007f + 0.1661d + 0.000055v^{2} \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\left( {R^{2} = 89.09} \right) \left( {R^{2} {\text{ adj}} = 86.18} \right)$$
(19)
$${\text{VB}}_{\hbox{max} } = -\,0.341 + 0.01061v + 5.36f + 0.1618d - 0.000052v^{2} - 14.68f^{2} \,\,\,\,\,\,\,\,\,\,\,\left( {R^{2} = 87.79} \right) \left( {R^{2} {\text{ adj}} = 83.43} \right)$$
(20)

The ANOVA is used again to evaluate significance of modified regression models for p and F ratio values. The p value less than 0.05 and high F values of the modified models indicate that the models are extremely significant. The parameters v, f and d are found significant for all the modified models. Therefore, these models can produce better results compared to models with insignificant terms, when same ranges of cutting parameters are utilized in the modified model.

Using input data from Table 2 in Eqs. (1)–(20), the estimated response values are evaluated. The estimated values obtained by RSM modified models are compared with the measured responses. The comparison of estimation capability of RSM with experimental values based on R2 values, R2 adj values and percentage of mean absolute error (MAE) is carried out. The obtained percentages of mean absolute error (MAE) along with R2 values and R2 adj values are tabulated in Table 4 for the four machining conditions DRY, DRY-CT, MQL and MQL-CT, respectively, for individual responses. It is observed that for most of the responses MAE is less than 15%. Based on R2 value and MAE, it is found that these models are acceptable for further analysis.

Table 4 Estimated accuracies of modified models
Full size table

4 Pictorial representation of models

The performance of the modified models is enumerated with different responses such as Fc, As, T, R a and VBmax. Figures 7, 8, 9, 10 and 11 list the plots of these models as against cutting parameters v, f and d for different machining conditions—DRY, DRY-CT, MQL and MQL-CT. While plotting, one cutting parameter is varied and the others are kept constant at mid value. The performance models are plotted for evaluating the performance in the turning of Inconel 718 under different machining conditions.

Fig. 7
figure 7

Effect of cutting parameters on cutting force (Fc)

Full size image
Fig. 8
figure 8

Effect of cutting parameters on tool vibrations (As)

Full size image
Fig. 9
figure 9

Effect of cutting parameters on temperature (T)

Full size image
Fig. 10
figure 10

Effect of cutting parameters on surface roughness (R a )

Full size image
Fig. 11
figure 11

Effect of cutting parameters on tool wear (VBmax)

Full size image

4.1 Effect of cutting parameters on cutting force

Figure 7a represents the variation in cutting force for four cutting conditions against cutting speed, keeping feed and depth of cut constant at mid value. Similarly, Fig. 7b, c are plotted for variation of force with feed and depth of cut, respectively. In all the conditions, cutting speed is more significant parameter for the cutting force followed by feed rate. However, less significant variation is observed with respect to the depth of cut. These main effect plots clearly show that minimum Fc occurs at a higher v, lower f and d. However, it increases with increase in the f and d in all machining conditions. In DRY-CT condition, the cutting force required is slightly less than in DRY. Further, small reduction in cutting forces is observed for MQL compared to the DRY-CT condition due to the heat removal from the tool-chip edge and the lubrication between the cutting edge and the work piece. It is observed that amongst four conditions MQL-CT condition requires minimum cutting force.

4.2 Effect of cutting parameters on tool vibration

The vibration (acceleration) of the tool was measured with respect to the varying cutting parameters and results are shown in Fig. 8. There is a prominent decrease in tool vibrations with increasing v. However, for increasing f and d up to a certain threshold value (0.15 mm/rev and 0.75 mm, respectively) minimum vibrations are observed. Beyond this threshold value the tool vibrations increase rapidly in all conditions. In DRY condition more tool vibrations are observed. While, the decreasing trends of tool vibrations are shown in DRY-CT, MQL and MQL-CT conditions.

4.3 Effect of input parameters on cutting temperature

Increase of cutting speed indicates a substantial increase in the cutting temperature than the rise which occurs with increase in feed and depth of cut (Fig. 9a–c). In MQL-CT condition, cutting temperature reduces because of lubricity through the deformation zone mainly at tool-chip edge than DRY-CT and MQL. DRY condition shows highest temperature and MQL-CT shows lowest temperature.

4.4 Effect of cutting parameters on surface roughness

Figure 10a shows that there is a pronounced improvement in the surface roughness value with increasing cutting speed. However, for feed and depth of cut, the improvement in the surface roughness is found only up to a certain threshold value as visible in Fig. 10b, c, respectively. Beyond this value, the surface finish degrades. This behaviour is observed prominently for the DRY and MQL condition. It is seen that, MQL and MQL-CT conditions achieve the best result (0.5–0.7 R a for high v, low f and d) compared to the other conditions.

4.5 Effect of cutting parameters on tool wear

The variations of tool wear against v, f and d are shown in Fig. 11. The tool wear is similar for almost all the machining conditions. The effect of change of cutting speed influences the tool wear to a greater extent than f and d as observed in the Fig. 11a. There is variation in the tool wear values for the varying f and d for the different machining conditions as found in the Fig. 11b, c. The wear of the tool is the most severe in the DRY condition because of intense heat generation and material removal. The rate of material removal from tool flank face progressively reduces for different machining conditions, beginning from maximum for the DRY condition and minimum for the MQL-CT condition.

4.6 Effect of cutting conditions on machinability

It is observed that amongst all the machining conditions shown in Fig. 7a, the value of cutting force (Fc) is highest for the DRY condition at low cutting speed (v). It is known that Inconel 718 is hard to cut material. At high v, the reducing trend of the Fc is attributed to the low coefficient of thermal conductivity of Inconel 718, on account of which, the heat is built up in the work piece which causes it to soften [21, 40]. The temperature rise with cutting speed is clearly seen in Fig. 9a and this temperature rise is responsible for softening of work piece and reduction in cutting force.

It is seen from Fig. 7b that Fc increases as the feed (f) increases for all the machining conditions. The reason for this trend is that at high f, large amount of material comes in contact with tool in less time. Thus, high f is responsible for generating high friction.

According to Fig. 7c, depth of cut (d) shows less significant effect on Fc compared to effect of v and f amongst all machining conditions. Increasing d leads to increase in volume of material removal which only lengthens the contact but force/unit length remain stable [6]. As reported in literature [13, 41], more increase of d result in breaking and chipping of cutting tool which drastically increases R a and VBmax. It was observed that sometimes breakage and chipping of tool affect continuity of machining process.

As the v increases, the Fc is reduced (Fig. 7a). This affects the deflection of tool also. More Fc means more deflection and vice versa. Since, the tool vibration (As) is the effect of interaction between machining process and the machine tool, lesser tool force results into lesser As and, therefore, increase in speed (v) shows reduction in vibrations (Fig. 8a). As the tool vibrations are reduced with speed, a better surface finish is obtained with increase in speed (Fig. 10a). Similarly, the vibration change with feed and depth of cut (Fig. 8b, c) affects the surface roughness, R a (Fig. 10b, c).

In case of tool wear (Fig. 11), a direct correlation is seen with temperature (Fig. 9). With increase in all the cutting parameters, the temperature is showing an increasing trend and therefor, the tool wear also shows the same trend.

The highest temperature is observed in DRY condition which results in higher VBmax, whereas slightly less VBmax in DRY-CT condition. Further using MQL, the T and VBmax values are reduced. Because of cooling and lubrication at cutting edge, the reduction of T is demonstrated in Fig. 9. Thus, MQL-CT condition results in least amount of wear due to least cutting temperature.

The cryogenic treatment results in microstructure change which are responsible for the increased wear resistance of the cutting tool and helps to retain its sharpness even at high temperature. Improvement in hardness enhances wear resistance of the tool and higher thermal conductivity helps for rapid heat conduction in tool, reducing temperature. Using cryogenically treated tools, the cutting force, vibration and cutting temperature are observed to be less compared to untreated tool in DRY turning of Inconel 718. The combined effect of all these responses is observed on the reduction in the surface roughness and tool wear values.

Further using MQL, the cutting temperature and tool wear values are reduced. Because, MQL achieves dual motives of heat removal from the tool-chip edge and the lubrication between the cutting edge and the work piece. The air stream continuously cools down the cutting zone while the micro-droplets of lubricant ensure proper lubrication by providing smooth layer at tool-chip interface. Moreover, MQL reduces the environmental impact by significantly decreasing cutting fluid usage and eliminating the necessity of coolant treatment, disposal, etc.

The combined effect of MQL with cryogenically treated tool in turning of Inconel 718 results in minimum cutting force, vibration, cutting temperature and surface roughness. Besides, MQL-CT condition contributes to good tool life which offers better productivity among other cutting conditions. Overall MQL-CT gives the best possible reduction in all response parameters.

The goal in machining is achievement of good quality machined products at low cost with negligible damage to the environment and the human health. Looking into the economic aspect of machining, it is imperative that it can be improved by optimizing the machining parameters. Therefore, it was deemed to be necessary to determine an optimum value of the input cutting parameters for the desired output parameters.

5 Optimization of parameters using RSM

For turning of Inconel 718 combined optimization of the cutting parameters—v, f and d with the aim of minimization of Fc, As, T, R a and VBmax for DRY, DRY-CT, MQL and MQL-CT conditions are performed. RSM optimization analysis is performed for all modified models of four machining conditions using default values in MINITAB 17 software.

Multi-objective optimization is carried out in MINITAB 17 software which displays the global desirability solution, by default. Various points are picked up for the search of optimal values of variables in starting of optimization algorithm procedures. Local and global solutions are two solution types for the search. Out of all local solutions, only one best solution is the global solution and it is categorized as the best relation between independent variables for attaining the best possible response. RSM optimization method has various benefits over conventional experimental or optimization methods, where at a time, only one variable can be used. Also, RSM offers economically better method as a lesser number of experiments are required to get effects of main parameters and interaction among parameters. Therefore, in present work, response surface-based desirability approach [38, 42, 43] has been used for optimizing the response parameters in turning of Inconel 718. All the values of the response function are transformed into a dimensionless desirability value (d) that varies over the range 0 ≤ d ≤ 1. The value, d = 0, suggests that the response is completely unacceptable, and d = 1 suggests that the response is acceptable and desirable. The individual desirability values (d i ) are combined with geometric mean. Thus, the overall desirability objective function is obtained by Eq. (21).

$$D = \left( {\prod\limits_{i = 1}^{n} {d_{i}^{\text{ri}} } } \right)^{{\frac{1}{{\sum {\text{ri}} }}}}$$
(21)

where D is the overall desirability objective function. The value of r is the response importance relative to the other response parameters of the process. The high value of overall desirability D indicates the more desirable and the best function of the system. This indicates that the optimum value is reached. The optimum values of factors are determined from value of individual desired function that enhances the overall desirability value. Figure 12 shows the optimum cutting parameters with overall desirability values of 0.62, 0.70, 0.76 and 0.85 for DRY, DRY-CT, MQL and MQL-CT conditions, respectively. So, these responses are accepted. The values of inputs parameters are chosen for further experimentation.

Fig. 12
figure 12

RSM optimization: a DRY; b DRY-CT; c MQL; d MQL-CT

Full size image

5.1 Validation of results

Three confirmation experiments were performed for each condition using optimum values of cutting parameters obtained by RSM methodology for all response parameters together. The results obtained and percentage error are presented in Table 5. Thus, it is seen that the estimated values by modified Eqs. (1)–(20) correlate well with the experimental values.

Table 5 Comparisons and confirmation of optimization result
Full size table

It is observed from Table 5 that the errors for T, R a and VBmax are less than 10% whereas overall error is higher in case of vibration and this is natural because vibration contains the effect of machining process and the machine tool. It is also seen that MQL-CT condition gives best values of response parameters.

Economics of machining can be enhanced using optimum cutting parameters with proper machining conditions shown in Fig. 12 and Table 5. The main objective of this experimentation is to identify best machining conditions for minimizing surface finish, tool wear with maximum material removal rate. Therefore, to know the productivity for the used machining condition, it is essential to determine the material removal rate of workpiece (MRR) [44]. The optimum cutting parameters obtained from Fig. 12 are selected for getting the maximum MRR. These parameters would be useful to study MRR which can be calculated as, MRR (mm3/min) = 1000 × v × f × d.

5.2 Spider plot for confirmation experiment using optimum parameters

A spider plot is an excellent tool for having an overview by organizing complex information in a simple manner. It is graphical method of presenting multivariate facts in the form of 2D chart. Distinguishing feature of spider plot is use of separate scales for every parameter on the same plot [15, 45]. It can help in obtaining specified response for the stipulated condition. Therefore, this plot is selected for showing data of confirmation experiments with optimum conditions, under the four conditions used for machining of Inconel 718.

To get the spider diagram, a polygon is drawn with number of vertices equal to the number of parameters. It is required to show the effect of three cutting parameters on six responses. Therefore, a polygon having nine vertices is drawn. Each of the nine parameters (three input + six responses) are represented by lines starting from centre and joining each vertex of the polygon. The minimum value of each parameter is kept as zero and is at the centre; whereas, maximum values are at the vertices of polygon. The associations among inputs and responses are presented using data of Table 6. Parameter values are marked on the line joining centre to vertex by considering proportionate scale and lines are drawn to connect the points falling under same cutting conditions. Thus, the spider plot is generated as shown in Fig. 13.

Table 6 Optimum parameters and limiting values for spider plot
Full size table
Fig. 13
figure 13

Spider plot for optimum machining conditions

Full size image

As seen from the spider plot, the curve obtained for MQL-CT represents the best machining condition for the machining of Inconel 718, which is at moderate cutting speed (75 m/min), medium feed (0.15 mm/rev) and low depth of cut (0.35 mm) leading to low value of cutting force, tool vibration, cutting temperature, tool wear and surface roughness. Besides, MQL-CT condition contributes to more material removal rate which offers better productivity among other optimum conditions.

Thus, MQL-CT condition resulted in 25, 22 and 13% lowering of Fc, 33, 32 and 25% decrease of As, 32, 22 and 18% lesser T, 68, 47 and 34% drop in R a of the work piece, 44, 39 and 32% better tool life and 16, 14 and 0.4% improvement in MRR compared to that of DRY, DRY-CT and MQL conditions, respectively. Thus, MQL-CT condition results in the best machining of Inconel 718 with desired surface quality needed for high precision parts. Figure 14a–d demonstrates the surface topography of machined surface. The enhancement in surface texture is noticeable from 3D optical profilometer which reveals a better topography of workpiece surface machined with MQL-CT condition as seen in Fig. 14d.

Fig. 14
figure 14

Surface roughness at optimum parameters

Full size image

The flank wear at nose region is measured at a different cutting lengths using optimum parameters and the performance of tool is presented in Fig. 15. The corresponding tool wear at the end of each machining condition is shown in Fig. 16a–d for four machining conditions. Spider plot clearly shows higher and lower values of R a and VBmax which is evidenced in Figs. 14, 15 and 16. The high values of R a , VB and low MRR are observed in DRY machining. In DRY-CT condition, slight reduction in R a , VBmax and improvement in MRR is observed due to cryo treatment. The hardness of untreated insert was found to be 1474 VHN and that of cryo-treated insert was found to be 1576 VHN. Thus, the erosion of the material from the tool flank is reduced which also helped in slight improvement of surface finish. By employing MQL, there is reduction of all responses except MRR. Because of good lubricity between tool and chip, there is significant improvement in surface roughness. MQL results in immediate lowering of the tool-chip interface temperature value because of which the tendency of the flank wear at nose region is reduced. The low values of Fc, As, and T ultimately improve the R a value compared to other conditions. The reduction in T also implies reduction in the VBmax evidenced in Figs. 15 and 16. The least amount of wear is achieved on tool which deforms less with MQL-CT condition as seen in Fig. 16d. It is because, MQL spray provides the penetration of the lubricant within the tool-chip interface and helps for heat dissipation; cryo treatment improves wear resistance and provides harder tool edges. As a result, the performance of tool is enhanced which reflects on the quality of machined surface.

Fig. 15
figure 15

Performance of tools at optimum parameters in DRY, DRY-CT, MQL and MQL-CT conditions

Full size image
Fig. 16
figure 16

Variations of flank wear at nose region using optimum cutting parameters

Full size image

6 Conclusion

In the present study, turning of Inconel 718 was performed under four machining conditions. Models were developed for five response parameters, i.e. cutting force, vibration, temperature, surface roughness and tool wear. Analysis of variance (ANOVA) was performed on each model and the models were updated by removing insignificant terms. Later on, desirability approach was used to perform simultaneous optimization of all the responses. Predictions of the combined model were confirmed by confirmation experiments. On the basis of this experimental study, following conclusions are drawn.

  • Cryogenic treatment changed the microstructure of tool which improved hardness of tool by 7% along with improvement in electrical conductivity by 14.83%. Improvement in hardness enhances wear resistance of the tool and higher thermal conductivity helps for rapid heat conduction in tool reducing temperature and tool wear.

  • Analysis of variance is used to evaluate significant parameters of modified regression models for p and F ratio values. The p value less than 0.05 and high F values of the modified models indicate that the models are extremely significant. Moreover, it is observed that cutting speed is the most significant parameter for all the responses, followed by feed and depth of cut in most of the machining conditions.

  • The ANOVA of the prediction models developed by RSM for five response parameters (cutting force, vibration, temperature, surface roughness and tool wear) showed good association between measured and estimated values. This was indicated by the R2 (correlation coefficient) values. The modified regression models using RSM are having R2 values in range of 73.18–99. 72%. The R2 closer to unity indicates best correlation between predicted data of model and experimental data. Additionally, the adequacy of models was examined by percentage of mean absolute error (MAE). It was observed that the MAE was obtained as less than 15% in most of the machining conditions. Based on prediction accuracy, it is concluded that the modified models accurately predict the responses.

  • Comparing machining with untreated insert in DRY and MQL conditions, the use of MQL resulted in 14% reduction in force, 10% decrease of vibration, 16% lowering of temperature, 51% enhancement in surface finish, 17% improvement in tool life and 15% improvement in material removal rate which are helpful to increase the productivity.

  • The machining performance can be further improved using MQL with cryo-treated inserts. MQL prominently helps in improving the performance of tool due to better heat dissipation from cutting region and cryo treatment improved wear resistance. Comparing untreated and cryo-treated insert; the use of cryo-treated insert resulted in 4% reduction in force, 1.50% decrease of vibration, 12% lowering of temperature, 40% enhancement in surface finish, 8% improvement in tool life and 1.50% improvement in material removal rate. Hence, the characteristics of cryo-treated inserts reflect on the quality of machined surface and tool life.

  • Spider diagram was used to represent the best machining condition in turning of Inconel 718. The optimum condition is found for MQL-CT condition which is at moderate cutting speed (75 m/min), medium feed (0.15 mm/rev) and low depth of cut (0.32 mm). It can be further concluded that MQL-CT resulted in 25, 22 and 13% reduction in cutting force, 33, 32 and 25% decrease of vibration, 32, 22 and 18% lowering of temperature than machining with DRY, DRY-CT and MQL conditions, respectively. Similarly, the use of MQL-CT also resulted in 68, 47 and 34% enhancements in surface finish; 44, 39 and 32% improvements in tool life and 16, 14 and 0.4% improvements in material removal rate, respectively. Thus, MQL-CT is identified as the best method for improving the overall productivity of the process.

  • Hence, the use of cryogenically treated inserts with Minimum Quantity Lubrication is recommended for the turning of Inconel 718.